{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 50.0,
  "eval_steps": 800,
  "global_step": 666250,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004803001876172608,
      "grad_norm": 3.330345630645752,
      "learning_rate": 0.0001113525,
      "loss": 4.3152,
      "step": 64
    },
    {
      "epoch": 0.009606003752345216,
      "grad_norm": 2.5987207889556885,
      "learning_rate": 0.00022447249999999998,
      "loss": 3.5908,
      "step": 128
    },
    {
      "epoch": 0.014409005628517824,
      "grad_norm": 10.176867485046387,
      "learning_rate": 0.00033759249999999996,
      "loss": 3.3927,
      "step": 192
    },
    {
      "epoch": 0.01921200750469043,
      "grad_norm": 6.534875869750977,
      "learning_rate": 0.00045071249999999993,
      "loss": 3.3333,
      "step": 256
    },
    {
      "epoch": 0.02401500938086304,
      "grad_norm": 6.088456630706787,
      "learning_rate": 0.0005638325,
      "loss": 3.2928,
      "step": 320
    },
    {
      "epoch": 0.028818011257035647,
      "grad_norm": 6.937580108642578,
      "learning_rate": 0.0006769524999999999,
      "loss": 3.2901,
      "step": 384
    },
    {
      "epoch": 0.033621013133208255,
      "grad_norm": 6.744969844818115,
      "learning_rate": 0.0007900724999999999,
      "loss": 3.2589,
      "step": 448
    },
    {
      "epoch": 0.03842401500938086,
      "grad_norm": 2.2261719703674316,
      "learning_rate": 0.0009031925,
      "loss": 3.231,
      "step": 512
    },
    {
      "epoch": 0.04322701688555347,
      "grad_norm": 1.030404806137085,
      "learning_rate": 0.0010163124999999999,
      "loss": 3.2278,
      "step": 576
    },
    {
      "epoch": 0.04803001876172608,
      "grad_norm": 1.036293387413025,
      "learning_rate": 0.0011294324999999998,
      "loss": 3.272,
      "step": 640
    },
    {
      "epoch": 0.05283302063789869,
      "grad_norm": 1.1835274696350098,
      "learning_rate": 0.0012425525,
      "loss": 3.256,
      "step": 704
    },
    {
      "epoch": 0.057636022514071295,
      "grad_norm": 0.8378634452819824,
      "learning_rate": 0.0013556724999999998,
      "loss": 3.27,
      "step": 768
    },
    {
      "epoch": 0.0624390243902439,
      "grad_norm": 0.7602612972259521,
      "learning_rate": 0.0014687925,
      "loss": 3.2261,
      "step": 832
    },
    {
      "epoch": 0.06724202626641651,
      "grad_norm": 0.6387987732887268,
      "learning_rate": 0.0015819124999999997,
      "loss": 3.2153,
      "step": 896
    },
    {
      "epoch": 0.07204502814258912,
      "grad_norm": 0.4422095715999603,
      "learning_rate": 0.0016950325,
      "loss": 3.1975,
      "step": 960
    },
    {
      "epoch": 0.07684803001876173,
      "grad_norm": 0.39002183079719543,
      "learning_rate": 0.0018081524999999999,
      "loss": 3.1983,
      "step": 1024
    },
    {
      "epoch": 0.08165103189493433,
      "grad_norm": 5.926162242889404,
      "learning_rate": 0.0019212724999999996,
      "loss": 3.1763,
      "step": 1088
    },
    {
      "epoch": 0.08645403377110694,
      "grad_norm": 0.4173193871974945,
      "learning_rate": 0.0020343924999999996,
      "loss": 3.1833,
      "step": 1152
    },
    {
      "epoch": 0.09125703564727955,
      "grad_norm": 0.4136042594909668,
      "learning_rate": 0.0021475125,
      "loss": 3.1846,
      "step": 1216
    },
    {
      "epoch": 0.09606003752345216,
      "grad_norm": 0.39301183819770813,
      "learning_rate": 0.0022606324999999996,
      "loss": 3.1739,
      "step": 1280
    },
    {
      "epoch": 0.10086303939962477,
      "grad_norm": 0.4910842776298523,
      "learning_rate": 0.0023737525,
      "loss": 3.1614,
      "step": 1344
    },
    {
      "epoch": 0.10566604127579737,
      "grad_norm": 0.4039038121700287,
      "learning_rate": 0.0024868725,
      "loss": 3.1577,
      "step": 1408
    },
    {
      "epoch": 0.11046904315196998,
      "grad_norm": 0.3286585211753845,
      "learning_rate": 0.0025999925,
      "loss": 3.1458,
      "step": 1472
    },
    {
      "epoch": 0.11527204502814259,
      "grad_norm": 0.44095373153686523,
      "learning_rate": 0.0027131125,
      "loss": 3.155,
      "step": 1536
    },
    {
      "epoch": 0.1200750469043152,
      "grad_norm": 0.40613290667533875,
      "learning_rate": 0.0028262325,
      "loss": 3.1469,
      "step": 1600
    },
    {
      "epoch": 0.1248780487804878,
      "grad_norm": 0.4613141417503357,
      "learning_rate": 0.002828,
      "loss": 3.1392,
      "step": 1664
    },
    {
      "epoch": 0.1296810506566604,
      "grad_norm": 0.3758493661880493,
      "learning_rate": 0.002828,
      "loss": 3.1298,
      "step": 1728
    },
    {
      "epoch": 0.13448405253283302,
      "grad_norm": 0.32609787583351135,
      "learning_rate": 0.002828,
      "loss": 3.123,
      "step": 1792
    },
    {
      "epoch": 0.13928705440900563,
      "grad_norm": 0.4221761226654053,
      "learning_rate": 0.002828,
      "loss": 3.1076,
      "step": 1856
    },
    {
      "epoch": 0.14409005628517824,
      "grad_norm": 0.4372267425060272,
      "learning_rate": 0.002828,
      "loss": 3.098,
      "step": 1920
    },
    {
      "epoch": 0.14889305816135084,
      "grad_norm": 0.36804404854774475,
      "learning_rate": 0.002828,
      "loss": 3.0952,
      "step": 1984
    },
    {
      "epoch": 0.15369606003752345,
      "grad_norm": 0.314120888710022,
      "learning_rate": 0.002828,
      "loss": 3.0751,
      "step": 2048
    },
    {
      "epoch": 0.15849906191369606,
      "grad_norm": 0.3158409297466278,
      "learning_rate": 0.002828,
      "loss": 3.0574,
      "step": 2112
    },
    {
      "epoch": 0.16330206378986867,
      "grad_norm": 0.35668376088142395,
      "learning_rate": 0.002828,
      "loss": 3.0598,
      "step": 2176
    },
    {
      "epoch": 0.16810506566604128,
      "grad_norm": 0.3429064452648163,
      "learning_rate": 0.002828,
      "loss": 3.0554,
      "step": 2240
    },
    {
      "epoch": 0.17290806754221388,
      "grad_norm": 0.37981563806533813,
      "learning_rate": 0.002828,
      "loss": 3.0439,
      "step": 2304
    },
    {
      "epoch": 0.1777110694183865,
      "grad_norm": 0.45046043395996094,
      "learning_rate": 0.002828,
      "loss": 3.034,
      "step": 2368
    },
    {
      "epoch": 0.1825140712945591,
      "grad_norm": 0.30424681305885315,
      "learning_rate": 0.002828,
      "loss": 3.0408,
      "step": 2432
    },
    {
      "epoch": 0.1873170731707317,
      "grad_norm": 0.4374525845050812,
      "learning_rate": 0.002828,
      "loss": 3.0289,
      "step": 2496
    },
    {
      "epoch": 0.19212007504690432,
      "grad_norm": 0.4312361776828766,
      "learning_rate": 0.002828,
      "loss": 3.0252,
      "step": 2560
    },
    {
      "epoch": 0.19692307692307692,
      "grad_norm": 0.33109021186828613,
      "learning_rate": 0.002828,
      "loss": 3.0094,
      "step": 2624
    },
    {
      "epoch": 0.20172607879924953,
      "grad_norm": 0.4393901228904724,
      "learning_rate": 0.002828,
      "loss": 3.0021,
      "step": 2688
    },
    {
      "epoch": 0.20652908067542214,
      "grad_norm": 0.44241341948509216,
      "learning_rate": 0.002828,
      "loss": 3.0005,
      "step": 2752
    },
    {
      "epoch": 0.21133208255159475,
      "grad_norm": 0.36241745948791504,
      "learning_rate": 0.002828,
      "loss": 2.9939,
      "step": 2816
    },
    {
      "epoch": 0.21613508442776735,
      "grad_norm": 0.40780672430992126,
      "learning_rate": 0.002828,
      "loss": 2.9788,
      "step": 2880
    },
    {
      "epoch": 0.22093808630393996,
      "grad_norm": 0.3944590389728546,
      "learning_rate": 0.002828,
      "loss": 2.9854,
      "step": 2944
    },
    {
      "epoch": 0.22574108818011257,
      "grad_norm": 0.40449267625808716,
      "learning_rate": 0.002828,
      "loss": 2.9819,
      "step": 3008
    },
    {
      "epoch": 0.23054409005628518,
      "grad_norm": 0.37247487902641296,
      "learning_rate": 0.002828,
      "loss": 2.9827,
      "step": 3072
    },
    {
      "epoch": 0.23534709193245779,
      "grad_norm": 0.3732891082763672,
      "learning_rate": 0.002828,
      "loss": 2.9714,
      "step": 3136
    },
    {
      "epoch": 0.2401500938086304,
      "grad_norm": 0.3168690800666809,
      "learning_rate": 0.002828,
      "loss": 2.9649,
      "step": 3200
    },
    {
      "epoch": 0.244953095684803,
      "grad_norm": 0.32185083627700806,
      "learning_rate": 0.002828,
      "loss": 2.9607,
      "step": 3264
    },
    {
      "epoch": 0.2497560975609756,
      "grad_norm": 0.3293335437774658,
      "learning_rate": 0.002828,
      "loss": 2.9464,
      "step": 3328
    },
    {
      "epoch": 0.2545590994371482,
      "grad_norm": 0.39153945446014404,
      "learning_rate": 0.002828,
      "loss": 2.9513,
      "step": 3392
    },
    {
      "epoch": 0.2593621013133208,
      "grad_norm": 0.36884990334510803,
      "learning_rate": 0.002828,
      "loss": 2.9418,
      "step": 3456
    },
    {
      "epoch": 0.26416510318949343,
      "grad_norm": 0.39196011424064636,
      "learning_rate": 0.002828,
      "loss": 2.9407,
      "step": 3520
    },
    {
      "epoch": 0.26896810506566604,
      "grad_norm": 0.36011603474617004,
      "learning_rate": 0.002828,
      "loss": 2.9461,
      "step": 3584
    },
    {
      "epoch": 0.27377110694183865,
      "grad_norm": 0.3608081638813019,
      "learning_rate": 0.002828,
      "loss": 2.937,
      "step": 3648
    },
    {
      "epoch": 0.27857410881801126,
      "grad_norm": 0.3833774924278259,
      "learning_rate": 0.002828,
      "loss": 2.9254,
      "step": 3712
    },
    {
      "epoch": 0.28337711069418386,
      "grad_norm": 0.35225459933280945,
      "learning_rate": 0.002828,
      "loss": 2.9165,
      "step": 3776
    },
    {
      "epoch": 0.2881801125703565,
      "grad_norm": 0.39832860231399536,
      "learning_rate": 0.002828,
      "loss": 2.9259,
      "step": 3840
    },
    {
      "epoch": 0.2929831144465291,
      "grad_norm": 0.36834558844566345,
      "learning_rate": 0.002828,
      "loss": 2.9186,
      "step": 3904
    },
    {
      "epoch": 0.2977861163227017,
      "grad_norm": 0.3877101540565491,
      "learning_rate": 0.002828,
      "loss": 2.9107,
      "step": 3968
    },
    {
      "epoch": 0.3025891181988743,
      "grad_norm": 0.40037983655929565,
      "learning_rate": 0.002828,
      "loss": 2.9086,
      "step": 4032
    },
    {
      "epoch": 0.3073921200750469,
      "grad_norm": 0.35432353615760803,
      "learning_rate": 0.002828,
      "loss": 2.9039,
      "step": 4096
    },
    {
      "epoch": 0.3121951219512195,
      "grad_norm": 0.3740752935409546,
      "learning_rate": 0.002828,
      "loss": 2.8973,
      "step": 4160
    },
    {
      "epoch": 0.3169981238273921,
      "grad_norm": 0.3972289264202118,
      "learning_rate": 0.002828,
      "loss": 2.8868,
      "step": 4224
    },
    {
      "epoch": 0.3218011257035647,
      "grad_norm": 0.3818065822124481,
      "learning_rate": 0.002828,
      "loss": 2.8916,
      "step": 4288
    },
    {
      "epoch": 0.32660412757973734,
      "grad_norm": 0.31802886724472046,
      "learning_rate": 0.002828,
      "loss": 2.895,
      "step": 4352
    },
    {
      "epoch": 0.33140712945590994,
      "grad_norm": 0.3920498192310333,
      "learning_rate": 0.002828,
      "loss": 2.896,
      "step": 4416
    },
    {
      "epoch": 0.33621013133208255,
      "grad_norm": 0.42001602053642273,
      "learning_rate": 0.002828,
      "loss": 2.8757,
      "step": 4480
    },
    {
      "epoch": 0.34101313320825516,
      "grad_norm": 0.38037222623825073,
      "learning_rate": 0.002828,
      "loss": 2.8812,
      "step": 4544
    },
    {
      "epoch": 0.34581613508442777,
      "grad_norm": 0.6402748823165894,
      "learning_rate": 0.002828,
      "loss": 2.8741,
      "step": 4608
    },
    {
      "epoch": 0.3506191369606004,
      "grad_norm": 0.3265625536441803,
      "learning_rate": 0.002828,
      "loss": 2.8659,
      "step": 4672
    },
    {
      "epoch": 0.355422138836773,
      "grad_norm": 0.3389698565006256,
      "learning_rate": 0.002828,
      "loss": 2.863,
      "step": 4736
    },
    {
      "epoch": 0.3602251407129456,
      "grad_norm": 0.34922096133232117,
      "learning_rate": 0.002828,
      "loss": 2.8555,
      "step": 4800
    },
    {
      "epoch": 0.3650281425891182,
      "grad_norm": 0.370980441570282,
      "learning_rate": 0.002828,
      "loss": 2.8624,
      "step": 4864
    },
    {
      "epoch": 0.3698311444652908,
      "grad_norm": 0.3553221821784973,
      "learning_rate": 0.002828,
      "loss": 2.8573,
      "step": 4928
    },
    {
      "epoch": 0.3746341463414634,
      "grad_norm": 0.36796537041664124,
      "learning_rate": 0.002828,
      "loss": 2.8567,
      "step": 4992
    },
    {
      "epoch": 0.379437148217636,
      "grad_norm": 0.3615240752696991,
      "learning_rate": 0.002828,
      "loss": 2.8444,
      "step": 5056
    },
    {
      "epoch": 0.38424015009380863,
      "grad_norm": 0.4196101427078247,
      "learning_rate": 0.002828,
      "loss": 2.845,
      "step": 5120
    },
    {
      "epoch": 0.38904315196998124,
      "grad_norm": 0.334185928106308,
      "learning_rate": 0.002828,
      "loss": 2.8376,
      "step": 5184
    },
    {
      "epoch": 0.39384615384615385,
      "grad_norm": 0.30093756318092346,
      "learning_rate": 0.002828,
      "loss": 2.8302,
      "step": 5248
    },
    {
      "epoch": 0.39864915572232645,
      "grad_norm": 0.41615140438079834,
      "learning_rate": 0.002828,
      "loss": 2.8365,
      "step": 5312
    },
    {
      "epoch": 0.40345215759849906,
      "grad_norm": 0.38547712564468384,
      "learning_rate": 0.002828,
      "loss": 2.833,
      "step": 5376
    },
    {
      "epoch": 0.40825515947467167,
      "grad_norm": 0.336453378200531,
      "learning_rate": 0.002828,
      "loss": 2.8289,
      "step": 5440
    },
    {
      "epoch": 0.4130581613508443,
      "grad_norm": 0.33043336868286133,
      "learning_rate": 0.002828,
      "loss": 2.8154,
      "step": 5504
    },
    {
      "epoch": 0.4178611632270169,
      "grad_norm": 0.33151519298553467,
      "learning_rate": 0.002828,
      "loss": 2.8267,
      "step": 5568
    },
    {
      "epoch": 0.4226641651031895,
      "grad_norm": 0.29418498277664185,
      "learning_rate": 0.002828,
      "loss": 2.8167,
      "step": 5632
    },
    {
      "epoch": 0.4274671669793621,
      "grad_norm": 0.3507523536682129,
      "learning_rate": 0.002828,
      "loss": 2.8227,
      "step": 5696
    },
    {
      "epoch": 0.4322701688555347,
      "grad_norm": 0.36976736783981323,
      "learning_rate": 0.002828,
      "loss": 2.8087,
      "step": 5760
    },
    {
      "epoch": 0.4370731707317073,
      "grad_norm": 0.4142448604106903,
      "learning_rate": 0.002828,
      "loss": 2.8191,
      "step": 5824
    },
    {
      "epoch": 0.4418761726078799,
      "grad_norm": 0.3893688917160034,
      "learning_rate": 0.002828,
      "loss": 2.8032,
      "step": 5888
    },
    {
      "epoch": 0.44667917448405253,
      "grad_norm": 0.3025995194911957,
      "learning_rate": 0.002828,
      "loss": 2.8049,
      "step": 5952
    },
    {
      "epoch": 0.45148217636022514,
      "grad_norm": 0.3676198422908783,
      "learning_rate": 0.002828,
      "loss": 2.7976,
      "step": 6016
    },
    {
      "epoch": 0.45628517823639775,
      "grad_norm": 0.39022454619407654,
      "learning_rate": 0.002828,
      "loss": 2.796,
      "step": 6080
    },
    {
      "epoch": 0.46108818011257036,
      "grad_norm": 0.38986560702323914,
      "learning_rate": 0.002828,
      "loss": 2.791,
      "step": 6144
    },
    {
      "epoch": 0.46589118198874296,
      "grad_norm": 0.35879769921302795,
      "learning_rate": 0.002828,
      "loss": 2.7949,
      "step": 6208
    },
    {
      "epoch": 0.47069418386491557,
      "grad_norm": 0.44419315457344055,
      "learning_rate": 0.002828,
      "loss": 2.7862,
      "step": 6272
    },
    {
      "epoch": 0.4754971857410882,
      "grad_norm": 0.30884304642677307,
      "learning_rate": 0.002828,
      "loss": 2.7864,
      "step": 6336
    },
    {
      "epoch": 0.4803001876172608,
      "grad_norm": 0.542960524559021,
      "learning_rate": 0.002828,
      "loss": 2.7842,
      "step": 6400
    },
    {
      "epoch": 0.4851031894934334,
      "grad_norm": 0.39032405614852905,
      "learning_rate": 0.002828,
      "loss": 2.7798,
      "step": 6464
    },
    {
      "epoch": 0.489906191369606,
      "grad_norm": 0.3760650157928467,
      "learning_rate": 0.002828,
      "loss": 2.78,
      "step": 6528
    },
    {
      "epoch": 0.4947091932457786,
      "grad_norm": 0.33309632539749146,
      "learning_rate": 0.002828,
      "loss": 2.7741,
      "step": 6592
    },
    {
      "epoch": 0.4995121951219512,
      "grad_norm": 0.37640711665153503,
      "learning_rate": 0.002828,
      "loss": 2.7795,
      "step": 6656
    },
    {
      "epoch": 0.5043151969981239,
      "grad_norm": 0.36830273270606995,
      "learning_rate": 0.002828,
      "loss": 2.7596,
      "step": 6720
    },
    {
      "epoch": 0.5091181988742964,
      "grad_norm": 0.3751394748687744,
      "learning_rate": 0.002828,
      "loss": 2.761,
      "step": 6784
    },
    {
      "epoch": 0.5139212007504691,
      "grad_norm": 0.3472868800163269,
      "learning_rate": 0.002828,
      "loss": 2.7567,
      "step": 6848
    },
    {
      "epoch": 0.5187242026266417,
      "grad_norm": 0.3749905526638031,
      "learning_rate": 0.002828,
      "loss": 2.7654,
      "step": 6912
    },
    {
      "epoch": 0.5235272045028143,
      "grad_norm": 0.4672335982322693,
      "learning_rate": 0.002828,
      "loss": 2.7467,
      "step": 6976
    },
    {
      "epoch": 0.5283302063789869,
      "grad_norm": 0.30083194375038147,
      "learning_rate": 0.002828,
      "loss": 2.7596,
      "step": 7040
    },
    {
      "epoch": 0.5331332082551595,
      "grad_norm": 0.34232673048973083,
      "learning_rate": 0.002828,
      "loss": 2.7425,
      "step": 7104
    },
    {
      "epoch": 0.5379362101313321,
      "grad_norm": 0.42222973704338074,
      "learning_rate": 0.002828,
      "loss": 2.7486,
      "step": 7168
    },
    {
      "epoch": 0.5427392120075047,
      "grad_norm": 0.36008650064468384,
      "learning_rate": 0.002828,
      "loss": 2.7451,
      "step": 7232
    },
    {
      "epoch": 0.5475422138836773,
      "grad_norm": 0.34359127283096313,
      "learning_rate": 0.002828,
      "loss": 2.734,
      "step": 7296
    },
    {
      "epoch": 0.55234521575985,
      "grad_norm": 0.3953745663166046,
      "learning_rate": 0.002828,
      "loss": 2.7397,
      "step": 7360
    },
    {
      "epoch": 0.5571482176360225,
      "grad_norm": 0.36703094840049744,
      "learning_rate": 0.002828,
      "loss": 2.7313,
      "step": 7424
    },
    {
      "epoch": 0.5619512195121952,
      "grad_norm": 0.31787919998168945,
      "learning_rate": 0.002828,
      "loss": 2.7363,
      "step": 7488
    },
    {
      "epoch": 0.5667542213883677,
      "grad_norm": 0.31179967522621155,
      "learning_rate": 0.002828,
      "loss": 2.7236,
      "step": 7552
    },
    {
      "epoch": 0.5715572232645404,
      "grad_norm": 0.3990299105644226,
      "learning_rate": 0.002828,
      "loss": 2.7191,
      "step": 7616
    },
    {
      "epoch": 0.576360225140713,
      "grad_norm": 0.3776848018169403,
      "learning_rate": 0.002828,
      "loss": 2.7244,
      "step": 7680
    },
    {
      "epoch": 0.5811632270168856,
      "grad_norm": 0.36117562651634216,
      "learning_rate": 0.002828,
      "loss": 2.7131,
      "step": 7744
    },
    {
      "epoch": 0.5859662288930582,
      "grad_norm": 0.3219313323497772,
      "learning_rate": 0.002828,
      "loss": 2.7202,
      "step": 7808
    },
    {
      "epoch": 0.5907692307692308,
      "grad_norm": 0.4501495957374573,
      "learning_rate": 0.002828,
      "loss": 2.7115,
      "step": 7872
    },
    {
      "epoch": 0.5955722326454034,
      "grad_norm": 0.3939913809299469,
      "learning_rate": 0.002828,
      "loss": 2.7076,
      "step": 7936
    },
    {
      "epoch": 0.600375234521576,
      "grad_norm": 0.3244933485984802,
      "learning_rate": 0.002828,
      "loss": 2.7047,
      "step": 8000
    },
    {
      "epoch": 0.6051782363977486,
      "grad_norm": 0.3094891607761383,
      "learning_rate": 0.002828,
      "loss": 2.698,
      "step": 8064
    },
    {
      "epoch": 0.6099812382739213,
      "grad_norm": 0.3525580167770386,
      "learning_rate": 0.002828,
      "loss": 2.7056,
      "step": 8128
    },
    {
      "epoch": 0.6147842401500938,
      "grad_norm": 0.3058718144893646,
      "learning_rate": 0.002828,
      "loss": 2.6937,
      "step": 8192
    },
    {
      "epoch": 0.6195872420262665,
      "grad_norm": 0.31864726543426514,
      "learning_rate": 0.002828,
      "loss": 2.6935,
      "step": 8256
    },
    {
      "epoch": 0.624390243902439,
      "grad_norm": 0.3197256028652191,
      "learning_rate": 0.002828,
      "loss": 2.6981,
      "step": 8320
    },
    {
      "epoch": 0.6291932457786117,
      "grad_norm": 0.30954182147979736,
      "learning_rate": 0.002828,
      "loss": 2.705,
      "step": 8384
    },
    {
      "epoch": 0.6339962476547842,
      "grad_norm": 0.4144911468029022,
      "learning_rate": 0.002828,
      "loss": 2.6832,
      "step": 8448
    },
    {
      "epoch": 0.6387992495309569,
      "grad_norm": 0.34720951318740845,
      "learning_rate": 0.002828,
      "loss": 2.6858,
      "step": 8512
    },
    {
      "epoch": 0.6436022514071295,
      "grad_norm": 0.30545172095298767,
      "learning_rate": 0.002828,
      "loss": 2.6758,
      "step": 8576
    },
    {
      "epoch": 0.6484052532833021,
      "grad_norm": 0.3341416120529175,
      "learning_rate": 0.002828,
      "loss": 2.6673,
      "step": 8640
    },
    {
      "epoch": 0.6532082551594747,
      "grad_norm": 0.5191973447799683,
      "learning_rate": 0.002828,
      "loss": 2.6798,
      "step": 8704
    },
    {
      "epoch": 0.6580112570356473,
      "grad_norm": 0.44382575154304504,
      "learning_rate": 0.002828,
      "loss": 2.683,
      "step": 8768
    },
    {
      "epoch": 0.6628142589118199,
      "grad_norm": 0.45676809549331665,
      "learning_rate": 0.002828,
      "loss": 2.6731,
      "step": 8832
    },
    {
      "epoch": 0.6676172607879926,
      "grad_norm": 0.3542475700378418,
      "learning_rate": 0.002828,
      "loss": 2.6813,
      "step": 8896
    },
    {
      "epoch": 0.6724202626641651,
      "grad_norm": 0.3976110517978668,
      "learning_rate": 0.002828,
      "loss": 2.6714,
      "step": 8960
    },
    {
      "epoch": 0.6772232645403378,
      "grad_norm": 0.37194061279296875,
      "learning_rate": 0.002828,
      "loss": 2.6646,
      "step": 9024
    },
    {
      "epoch": 0.6820262664165103,
      "grad_norm": 0.4080849289894104,
      "learning_rate": 0.002828,
      "loss": 2.6638,
      "step": 9088
    },
    {
      "epoch": 0.686829268292683,
      "grad_norm": 0.3275296986103058,
      "learning_rate": 0.002828,
      "loss": 2.6643,
      "step": 9152
    },
    {
      "epoch": 0.6916322701688555,
      "grad_norm": 0.4300732910633087,
      "learning_rate": 0.002828,
      "loss": 2.6545,
      "step": 9216
    },
    {
      "epoch": 0.6964352720450282,
      "grad_norm": 0.528816282749176,
      "learning_rate": 0.002828,
      "loss": 2.6639,
      "step": 9280
    },
    {
      "epoch": 0.7012382739212007,
      "grad_norm": 0.39729437232017517,
      "learning_rate": 0.002828,
      "loss": 2.6669,
      "step": 9344
    },
    {
      "epoch": 0.7060412757973734,
      "grad_norm": 0.36177024245262146,
      "learning_rate": 0.002828,
      "loss": 2.6429,
      "step": 9408
    },
    {
      "epoch": 0.710844277673546,
      "grad_norm": 0.3488599956035614,
      "learning_rate": 0.002828,
      "loss": 2.6409,
      "step": 9472
    },
    {
      "epoch": 0.7156472795497186,
      "grad_norm": 0.361208438873291,
      "learning_rate": 0.002828,
      "loss": 2.6354,
      "step": 9536
    },
    {
      "epoch": 0.7204502814258912,
      "grad_norm": 0.3307696282863617,
      "learning_rate": 0.002828,
      "loss": 2.6398,
      "step": 9600
    },
    {
      "epoch": 0.7252532833020638,
      "grad_norm": 0.47409588098526,
      "learning_rate": 0.002828,
      "loss": 2.6492,
      "step": 9664
    },
    {
      "epoch": 0.7300562851782364,
      "grad_norm": 0.43482983112335205,
      "learning_rate": 0.002828,
      "loss": 2.6251,
      "step": 9728
    },
    {
      "epoch": 0.7348592870544091,
      "grad_norm": 0.43177512288093567,
      "learning_rate": 0.002828,
      "loss": 2.6277,
      "step": 9792
    },
    {
      "epoch": 0.7396622889305816,
      "grad_norm": 0.5830815434455872,
      "learning_rate": 0.002828,
      "loss": 2.6291,
      "step": 9856
    },
    {
      "epoch": 0.7444652908067543,
      "grad_norm": 0.42559024691581726,
      "learning_rate": 0.002828,
      "loss": 2.6248,
      "step": 9920
    },
    {
      "epoch": 0.7492682926829268,
      "grad_norm": 0.36572182178497314,
      "learning_rate": 0.002828,
      "loss": 2.6175,
      "step": 9984
    },
    {
      "epoch": 0.7540712945590995,
      "grad_norm": 0.42863738536834717,
      "learning_rate": 0.002828,
      "loss": 2.6249,
      "step": 10048
    },
    {
      "epoch": 0.758874296435272,
      "grad_norm": 0.34681934118270874,
      "learning_rate": 0.002828,
      "loss": 2.6207,
      "step": 10112
    },
    {
      "epoch": 0.7636772983114447,
      "grad_norm": 0.40332967042922974,
      "learning_rate": 0.002828,
      "loss": 2.6088,
      "step": 10176
    },
    {
      "epoch": 0.7684803001876173,
      "grad_norm": 0.47137463092803955,
      "learning_rate": 0.002828,
      "loss": 2.6109,
      "step": 10240
    },
    {
      "epoch": 0.7732833020637899,
      "grad_norm": 0.3324384093284607,
      "learning_rate": 0.002828,
      "loss": 2.6025,
      "step": 10304
    },
    {
      "epoch": 0.7780863039399625,
      "grad_norm": 0.3714103698730469,
      "learning_rate": 0.002828,
      "loss": 2.6038,
      "step": 10368
    },
    {
      "epoch": 0.7828893058161351,
      "grad_norm": 0.3684547543525696,
      "learning_rate": 0.002828,
      "loss": 2.6059,
      "step": 10432
    },
    {
      "epoch": 0.7876923076923077,
      "grad_norm": 0.3580617606639862,
      "learning_rate": 0.002828,
      "loss": 2.5972,
      "step": 10496
    },
    {
      "epoch": 0.7924953095684804,
      "grad_norm": 0.4132176339626312,
      "learning_rate": 0.002828,
      "loss": 2.5979,
      "step": 10560
    },
    {
      "epoch": 0.7972983114446529,
      "grad_norm": 0.4079800546169281,
      "learning_rate": 0.002828,
      "loss": 2.5923,
      "step": 10624
    },
    {
      "epoch": 0.8021013133208256,
      "grad_norm": 0.40170854330062866,
      "learning_rate": 0.002828,
      "loss": 2.5861,
      "step": 10688
    },
    {
      "epoch": 0.8069043151969981,
      "grad_norm": 0.4748755097389221,
      "learning_rate": 0.002828,
      "loss": 2.6031,
      "step": 10752
    },
    {
      "epoch": 0.8117073170731708,
      "grad_norm": 0.3806183338165283,
      "learning_rate": 0.002828,
      "loss": 2.5913,
      "step": 10816
    },
    {
      "epoch": 0.8165103189493433,
      "grad_norm": 0.32777532935142517,
      "learning_rate": 0.002828,
      "loss": 2.5937,
      "step": 10880
    },
    {
      "epoch": 0.821313320825516,
      "grad_norm": 0.4884773790836334,
      "learning_rate": 0.002828,
      "loss": 2.5851,
      "step": 10944
    },
    {
      "epoch": 0.8261163227016886,
      "grad_norm": 0.43175649642944336,
      "learning_rate": 0.002828,
      "loss": 2.5861,
      "step": 11008
    },
    {
      "epoch": 0.8309193245778612,
      "grad_norm": 0.44375500082969666,
      "learning_rate": 0.002828,
      "loss": 2.5823,
      "step": 11072
    },
    {
      "epoch": 0.8357223264540338,
      "grad_norm": 0.36503469944000244,
      "learning_rate": 0.002828,
      "loss": 2.579,
      "step": 11136
    },
    {
      "epoch": 0.8405253283302064,
      "grad_norm": 0.3493196368217468,
      "learning_rate": 0.002828,
      "loss": 2.5776,
      "step": 11200
    },
    {
      "epoch": 0.845328330206379,
      "grad_norm": 0.3593812584877014,
      "learning_rate": 0.002828,
      "loss": 2.5681,
      "step": 11264
    },
    {
      "epoch": 0.8501313320825516,
      "grad_norm": 0.4043927788734436,
      "learning_rate": 0.002828,
      "loss": 2.5766,
      "step": 11328
    },
    {
      "epoch": 0.8549343339587242,
      "grad_norm": 0.3805730938911438,
      "learning_rate": 0.002828,
      "loss": 2.5713,
      "step": 11392
    },
    {
      "epoch": 0.8597373358348969,
      "grad_norm": 0.40822461247444153,
      "learning_rate": 0.002828,
      "loss": 2.576,
      "step": 11456
    },
    {
      "epoch": 0.8645403377110694,
      "grad_norm": 0.3430253565311432,
      "learning_rate": 0.002828,
      "loss": 2.5669,
      "step": 11520
    },
    {
      "epoch": 0.8693433395872421,
      "grad_norm": 0.3665921688079834,
      "learning_rate": 0.002828,
      "loss": 2.5619,
      "step": 11584
    },
    {
      "epoch": 0.8741463414634146,
      "grad_norm": 0.3768637776374817,
      "learning_rate": 0.002828,
      "loss": 2.5617,
      "step": 11648
    },
    {
      "epoch": 0.8789493433395873,
      "grad_norm": 0.6709098219871521,
      "learning_rate": 0.002828,
      "loss": 2.5592,
      "step": 11712
    },
    {
      "epoch": 0.8837523452157598,
      "grad_norm": 0.37109729647636414,
      "learning_rate": 0.002828,
      "loss": 2.55,
      "step": 11776
    },
    {
      "epoch": 0.8885553470919325,
      "grad_norm": 0.35545358061790466,
      "learning_rate": 0.002828,
      "loss": 2.5601,
      "step": 11840
    },
    {
      "epoch": 0.8933583489681051,
      "grad_norm": 0.34493309259414673,
      "learning_rate": 0.002828,
      "loss": 2.5555,
      "step": 11904
    },
    {
      "epoch": 0.8981613508442777,
      "grad_norm": 0.35226738452911377,
      "learning_rate": 0.002828,
      "loss": 2.5464,
      "step": 11968
    },
    {
      "epoch": 0.9029643527204503,
      "grad_norm": 0.3626823425292969,
      "learning_rate": 0.002828,
      "loss": 2.5526,
      "step": 12032
    },
    {
      "epoch": 0.9077673545966229,
      "grad_norm": 0.4639281630516052,
      "learning_rate": 0.002828,
      "loss": 2.5535,
      "step": 12096
    },
    {
      "epoch": 0.9125703564727955,
      "grad_norm": 0.425073504447937,
      "learning_rate": 0.002828,
      "loss": 2.54,
      "step": 12160
    },
    {
      "epoch": 0.9173733583489682,
      "grad_norm": 0.4849206507205963,
      "learning_rate": 0.002828,
      "loss": 2.5415,
      "step": 12224
    },
    {
      "epoch": 0.9221763602251407,
      "grad_norm": 0.3517647385597229,
      "learning_rate": 0.002828,
      "loss": 2.5396,
      "step": 12288
    },
    {
      "epoch": 0.9269793621013134,
      "grad_norm": 0.4217440187931061,
      "learning_rate": 0.002828,
      "loss": 2.5353,
      "step": 12352
    },
    {
      "epoch": 0.9317823639774859,
      "grad_norm": 0.3862438499927521,
      "learning_rate": 0.002828,
      "loss": 2.532,
      "step": 12416
    },
    {
      "epoch": 0.9365853658536586,
      "grad_norm": 0.4026007056236267,
      "learning_rate": 0.002828,
      "loss": 2.5342,
      "step": 12480
    },
    {
      "epoch": 0.9413883677298311,
      "grad_norm": 0.3353049159049988,
      "learning_rate": 0.002828,
      "loss": 2.5303,
      "step": 12544
    },
    {
      "epoch": 0.9461913696060038,
      "grad_norm": 0.35357797145843506,
      "learning_rate": 0.002828,
      "loss": 2.5283,
      "step": 12608
    },
    {
      "epoch": 0.9509943714821764,
      "grad_norm": 0.35685861110687256,
      "learning_rate": 0.002828,
      "loss": 2.5331,
      "step": 12672
    },
    {
      "epoch": 0.955797373358349,
      "grad_norm": 0.36265361309051514,
      "learning_rate": 0.002828,
      "loss": 2.5306,
      "step": 12736
    },
    {
      "epoch": 0.9606003752345216,
      "grad_norm": 0.4119773805141449,
      "learning_rate": 0.002828,
      "loss": 2.5199,
      "step": 12800
    },
    {
      "epoch": 0.9654033771106942,
      "grad_norm": 0.3662680387496948,
      "learning_rate": 0.002828,
      "loss": 2.523,
      "step": 12864
    },
    {
      "epoch": 0.9702063789868668,
      "grad_norm": 0.3822716772556305,
      "learning_rate": 0.002828,
      "loss": 2.5186,
      "step": 12928
    },
    {
      "epoch": 0.9750093808630395,
      "grad_norm": 0.3412950038909912,
      "learning_rate": 0.002828,
      "loss": 2.5201,
      "step": 12992
    },
    {
      "epoch": 0.979812382739212,
      "grad_norm": 0.373353511095047,
      "learning_rate": 0.002828,
      "loss": 2.518,
      "step": 13056
    },
    {
      "epoch": 0.9846153846153847,
      "grad_norm": 10.112526893615723,
      "learning_rate": 0.002828,
      "loss": 2.5128,
      "step": 13120
    },
    {
      "epoch": 0.9894183864915572,
      "grad_norm": 0.36393383145332336,
      "learning_rate": 0.002828,
      "loss": 2.5249,
      "step": 13184
    },
    {
      "epoch": 0.9942213883677299,
      "grad_norm": 0.4176023006439209,
      "learning_rate": 0.002828,
      "loss": 2.5082,
      "step": 13248
    },
    {
      "epoch": 0.9990243902439024,
      "grad_norm": 0.4248984456062317,
      "learning_rate": 0.002828,
      "loss": 2.5106,
      "step": 13312
    },
    {
      "epoch": 1.003827392120075,
      "grad_norm": 0.3931824564933777,
      "learning_rate": 0.002828,
      "loss": 2.4797,
      "step": 13376
    },
    {
      "epoch": 1.0086303939962478,
      "grad_norm": 0.3742982745170593,
      "learning_rate": 0.002828,
      "loss": 2.4675,
      "step": 13440
    },
    {
      "epoch": 1.0134333958724202,
      "grad_norm": 0.4388613998889923,
      "learning_rate": 0.002828,
      "loss": 2.4746,
      "step": 13504
    },
    {
      "epoch": 1.0182363977485929,
      "grad_norm": 0.41458427906036377,
      "learning_rate": 0.002828,
      "loss": 2.4699,
      "step": 13568
    },
    {
      "epoch": 1.0230393996247655,
      "grad_norm": 0.3841855227947235,
      "learning_rate": 0.002828,
      "loss": 2.4675,
      "step": 13632
    },
    {
      "epoch": 1.027842401500938,
      "grad_norm": 0.43930500745773315,
      "learning_rate": 0.002828,
      "loss": 2.4725,
      "step": 13696
    },
    {
      "epoch": 1.0326454033771106,
      "grad_norm": 0.3687760531902313,
      "learning_rate": 0.002828,
      "loss": 2.4573,
      "step": 13760
    },
    {
      "epoch": 1.0374484052532833,
      "grad_norm": 0.3823833465576172,
      "learning_rate": 0.002828,
      "loss": 2.4592,
      "step": 13824
    },
    {
      "epoch": 1.042251407129456,
      "grad_norm": 0.40025222301483154,
      "learning_rate": 0.002828,
      "loss": 2.4713,
      "step": 13888
    },
    {
      "epoch": 1.0470544090056286,
      "grad_norm": 0.40790122747421265,
      "learning_rate": 0.002828,
      "loss": 2.4618,
      "step": 13952
    },
    {
      "epoch": 1.051857410881801,
      "grad_norm": 0.42718634009361267,
      "learning_rate": 0.002828,
      "loss": 2.4646,
      "step": 14016
    },
    {
      "epoch": 1.0566604127579737,
      "grad_norm": 0.3305782079696655,
      "learning_rate": 0.002828,
      "loss": 2.4563,
      "step": 14080
    },
    {
      "epoch": 1.0614634146341464,
      "grad_norm": 0.37126559019088745,
      "learning_rate": 0.002828,
      "loss": 2.4666,
      "step": 14144
    },
    {
      "epoch": 1.0662664165103188,
      "grad_norm": 0.414987176656723,
      "learning_rate": 0.002828,
      "loss": 2.4558,
      "step": 14208
    },
    {
      "epoch": 1.0710694183864915,
      "grad_norm": 0.45917075872421265,
      "learning_rate": 0.002828,
      "loss": 2.4604,
      "step": 14272
    },
    {
      "epoch": 1.0758724202626642,
      "grad_norm": 0.4362465739250183,
      "learning_rate": 0.002828,
      "loss": 2.4504,
      "step": 14336
    },
    {
      "epoch": 1.0806754221388368,
      "grad_norm": 0.40015289187431335,
      "learning_rate": 0.002828,
      "loss": 2.4501,
      "step": 14400
    },
    {
      "epoch": 1.0854784240150095,
      "grad_norm": 0.3781159818172455,
      "learning_rate": 0.002828,
      "loss": 2.4555,
      "step": 14464
    },
    {
      "epoch": 1.090281425891182,
      "grad_norm": 0.4165579676628113,
      "learning_rate": 0.002828,
      "loss": 2.4472,
      "step": 14528
    },
    {
      "epoch": 1.0950844277673546,
      "grad_norm": 0.4100767970085144,
      "learning_rate": 0.002828,
      "loss": 2.4545,
      "step": 14592
    },
    {
      "epoch": 1.0998874296435273,
      "grad_norm": 0.4211256504058838,
      "learning_rate": 0.002828,
      "loss": 2.4444,
      "step": 14656
    },
    {
      "epoch": 1.1046904315196997,
      "grad_norm": 0.390396386384964,
      "learning_rate": 0.002828,
      "loss": 2.4502,
      "step": 14720
    },
    {
      "epoch": 1.1094934333958724,
      "grad_norm": 0.3585355281829834,
      "learning_rate": 0.002828,
      "loss": 2.437,
      "step": 14784
    },
    {
      "epoch": 1.114296435272045,
      "grad_norm": 0.5148431062698364,
      "learning_rate": 0.002828,
      "loss": 2.4429,
      "step": 14848
    },
    {
      "epoch": 1.1190994371482177,
      "grad_norm": 0.44254639744758606,
      "learning_rate": 0.002828,
      "loss": 2.4393,
      "step": 14912
    },
    {
      "epoch": 1.1239024390243904,
      "grad_norm": 0.3710468113422394,
      "learning_rate": 0.002828,
      "loss": 2.4379,
      "step": 14976
    },
    {
      "epoch": 1.1287054409005628,
      "grad_norm": 0.41197285056114197,
      "learning_rate": 0.002828,
      "loss": 2.4387,
      "step": 15040
    },
    {
      "epoch": 1.1335084427767355,
      "grad_norm": 0.37512508034706116,
      "learning_rate": 0.002828,
      "loss": 2.4327,
      "step": 15104
    },
    {
      "epoch": 1.1383114446529081,
      "grad_norm": 0.4286038279533386,
      "learning_rate": 0.002828,
      "loss": 2.4314,
      "step": 15168
    },
    {
      "epoch": 1.1431144465290806,
      "grad_norm": 0.37446776032447815,
      "learning_rate": 0.002828,
      "loss": 2.4304,
      "step": 15232
    },
    {
      "epoch": 1.1479174484052532,
      "grad_norm": 0.4537597894668579,
      "learning_rate": 0.002828,
      "loss": 2.4239,
      "step": 15296
    },
    {
      "epoch": 1.152720450281426,
      "grad_norm": 0.36247050762176514,
      "learning_rate": 0.002828,
      "loss": 2.4284,
      "step": 15360
    },
    {
      "epoch": 1.1575234521575986,
      "grad_norm": 0.3772297501564026,
      "learning_rate": 0.002828,
      "loss": 2.4307,
      "step": 15424
    },
    {
      "epoch": 1.1623264540337712,
      "grad_norm": 0.3789200186729431,
      "learning_rate": 0.002828,
      "loss": 2.4261,
      "step": 15488
    },
    {
      "epoch": 1.1671294559099437,
      "grad_norm": 0.36870113015174866,
      "learning_rate": 0.002828,
      "loss": 2.4242,
      "step": 15552
    },
    {
      "epoch": 1.1719324577861163,
      "grad_norm": 0.37448298931121826,
      "learning_rate": 0.002828,
      "loss": 2.418,
      "step": 15616
    },
    {
      "epoch": 1.176735459662289,
      "grad_norm": 0.4384878873825073,
      "learning_rate": 0.002828,
      "loss": 2.4222,
      "step": 15680
    },
    {
      "epoch": 1.1815384615384614,
      "grad_norm": 0.37811148166656494,
      "learning_rate": 0.002828,
      "loss": 2.4187,
      "step": 15744
    },
    {
      "epoch": 1.186341463414634,
      "grad_norm": 0.4190385341644287,
      "learning_rate": 0.002828,
      "loss": 2.4221,
      "step": 15808
    },
    {
      "epoch": 1.1911444652908068,
      "grad_norm": 0.48885485529899597,
      "learning_rate": 0.002828,
      "loss": 2.4146,
      "step": 15872
    },
    {
      "epoch": 1.1959474671669794,
      "grad_norm": 0.42267611622810364,
      "learning_rate": 0.002828,
      "loss": 2.4245,
      "step": 15936
    },
    {
      "epoch": 1.200750469043152,
      "grad_norm": 0.3886626064777374,
      "learning_rate": 0.002828,
      "loss": 2.4283,
      "step": 16000
    },
    {
      "epoch": 1.2055534709193245,
      "grad_norm": 0.40871456265449524,
      "learning_rate": 0.002828,
      "loss": 2.4127,
      "step": 16064
    },
    {
      "epoch": 1.2103564727954972,
      "grad_norm": 0.46952739357948303,
      "learning_rate": 0.002828,
      "loss": 2.408,
      "step": 16128
    },
    {
      "epoch": 1.2151594746716698,
      "grad_norm": 0.41340023279190063,
      "learning_rate": 0.002828,
      "loss": 2.3977,
      "step": 16192
    },
    {
      "epoch": 1.2199624765478423,
      "grad_norm": 0.36176440119743347,
      "learning_rate": 0.002828,
      "loss": 2.405,
      "step": 16256
    },
    {
      "epoch": 1.224765478424015,
      "grad_norm": 0.4117899239063263,
      "learning_rate": 0.002828,
      "loss": 2.4011,
      "step": 16320
    },
    {
      "epoch": 1.2295684803001876,
      "grad_norm": 0.5039286613464355,
      "learning_rate": 0.002828,
      "loss": 2.4142,
      "step": 16384
    },
    {
      "epoch": 1.2343714821763603,
      "grad_norm": 0.3716677129268646,
      "learning_rate": 0.002828,
      "loss": 2.4098,
      "step": 16448
    },
    {
      "epoch": 1.239174484052533,
      "grad_norm": 0.42316168546676636,
      "learning_rate": 0.002828,
      "loss": 2.4004,
      "step": 16512
    },
    {
      "epoch": 1.2439774859287054,
      "grad_norm": 0.5081620216369629,
      "learning_rate": 0.002828,
      "loss": 2.3926,
      "step": 16576
    },
    {
      "epoch": 1.248780487804878,
      "grad_norm": 0.39409589767456055,
      "learning_rate": 0.002828,
      "loss": 2.3942,
      "step": 16640
    },
    {
      "epoch": 1.2535834896810507,
      "grad_norm": 0.38638824224472046,
      "learning_rate": 0.002828,
      "loss": 2.3936,
      "step": 16704
    },
    {
      "epoch": 1.2583864915572232,
      "grad_norm": 0.41918718814849854,
      "learning_rate": 0.002828,
      "loss": 2.3972,
      "step": 16768
    },
    {
      "epoch": 1.2631894934333958,
      "grad_norm": 0.3932395279407501,
      "learning_rate": 0.002828,
      "loss": 2.3983,
      "step": 16832
    },
    {
      "epoch": 1.2679924953095685,
      "grad_norm": 0.3787371814250946,
      "learning_rate": 0.002828,
      "loss": 2.3963,
      "step": 16896
    },
    {
      "epoch": 1.2727954971857411,
      "grad_norm": 0.40612953901290894,
      "learning_rate": 0.002828,
      "loss": 2.3825,
      "step": 16960
    },
    {
      "epoch": 1.2775984990619138,
      "grad_norm": 0.4243071675300598,
      "learning_rate": 0.002828,
      "loss": 2.3835,
      "step": 17024
    },
    {
      "epoch": 1.2824015009380862,
      "grad_norm": 0.4240303039550781,
      "learning_rate": 0.002828,
      "loss": 2.4023,
      "step": 17088
    },
    {
      "epoch": 1.287204502814259,
      "grad_norm": 0.4888259470462799,
      "learning_rate": 0.002828,
      "loss": 2.3918,
      "step": 17152
    },
    {
      "epoch": 1.2920075046904316,
      "grad_norm": 0.4678399860858917,
      "learning_rate": 0.002828,
      "loss": 2.389,
      "step": 17216
    },
    {
      "epoch": 1.296810506566604,
      "grad_norm": 0.38733649253845215,
      "learning_rate": 0.002828,
      "loss": 2.3898,
      "step": 17280
    },
    {
      "epoch": 1.3016135084427767,
      "grad_norm": 0.38587358593940735,
      "learning_rate": 0.002828,
      "loss": 2.3797,
      "step": 17344
    },
    {
      "epoch": 1.3064165103189493,
      "grad_norm": 0.39998751878738403,
      "learning_rate": 0.002828,
      "loss": 2.3907,
      "step": 17408
    },
    {
      "epoch": 1.311219512195122,
      "grad_norm": 0.36294978857040405,
      "learning_rate": 0.002828,
      "loss": 2.3822,
      "step": 17472
    },
    {
      "epoch": 1.3160225140712947,
      "grad_norm": 0.3924562633037567,
      "learning_rate": 0.002828,
      "loss": 2.3772,
      "step": 17536
    },
    {
      "epoch": 1.320825515947467,
      "grad_norm": 0.3837553560733795,
      "learning_rate": 0.002828,
      "loss": 2.3804,
      "step": 17600
    },
    {
      "epoch": 1.3256285178236398,
      "grad_norm": 0.38875913619995117,
      "learning_rate": 0.002828,
      "loss": 2.3747,
      "step": 17664
    },
    {
      "epoch": 1.3304315196998124,
      "grad_norm": 0.41738125681877136,
      "learning_rate": 0.002828,
      "loss": 2.3778,
      "step": 17728
    },
    {
      "epoch": 1.3352345215759849,
      "grad_norm": 0.3645491898059845,
      "learning_rate": 0.002828,
      "loss": 2.3731,
      "step": 17792
    },
    {
      "epoch": 1.3400375234521575,
      "grad_norm": 0.43829870223999023,
      "learning_rate": 0.002828,
      "loss": 2.3674,
      "step": 17856
    },
    {
      "epoch": 1.3448405253283302,
      "grad_norm": 0.3851640820503235,
      "learning_rate": 0.002828,
      "loss": 2.3753,
      "step": 17920
    },
    {
      "epoch": 1.3496435272045029,
      "grad_norm": 0.36147060990333557,
      "learning_rate": 0.002828,
      "loss": 2.3669,
      "step": 17984
    },
    {
      "epoch": 1.3544465290806755,
      "grad_norm": 0.42050638794898987,
      "learning_rate": 0.002828,
      "loss": 2.3693,
      "step": 18048
    },
    {
      "epoch": 1.359249530956848,
      "grad_norm": 0.3830699920654297,
      "learning_rate": 0.002828,
      "loss": 2.3668,
      "step": 18112
    },
    {
      "epoch": 1.3640525328330206,
      "grad_norm": 0.3830968737602234,
      "learning_rate": 0.002828,
      "loss": 2.3647,
      "step": 18176
    },
    {
      "epoch": 1.3688555347091933,
      "grad_norm": 0.3880060017108917,
      "learning_rate": 0.002828,
      "loss": 2.37,
      "step": 18240
    },
    {
      "epoch": 1.3736585365853657,
      "grad_norm": 0.45445796847343445,
      "learning_rate": 0.002828,
      "loss": 2.3594,
      "step": 18304
    },
    {
      "epoch": 1.3784615384615384,
      "grad_norm": 0.3750540316104889,
      "learning_rate": 0.002828,
      "loss": 2.3574,
      "step": 18368
    },
    {
      "epoch": 1.383264540337711,
      "grad_norm": 0.3783455193042755,
      "learning_rate": 0.002828,
      "loss": 2.3537,
      "step": 18432
    },
    {
      "epoch": 1.3880675422138837,
      "grad_norm": 0.40336528420448303,
      "learning_rate": 0.002828,
      "loss": 2.3672,
      "step": 18496
    },
    {
      "epoch": 1.3928705440900564,
      "grad_norm": 0.43220385909080505,
      "learning_rate": 0.002828,
      "loss": 2.3588,
      "step": 18560
    },
    {
      "epoch": 1.3976735459662288,
      "grad_norm": 0.4069630205631256,
      "learning_rate": 0.002828,
      "loss": 2.3619,
      "step": 18624
    },
    {
      "epoch": 1.4024765478424015,
      "grad_norm": 0.3866819441318512,
      "learning_rate": 0.002828,
      "loss": 2.3493,
      "step": 18688
    },
    {
      "epoch": 1.4072795497185742,
      "grad_norm": 0.3699668347835541,
      "learning_rate": 0.002828,
      "loss": 2.349,
      "step": 18752
    },
    {
      "epoch": 1.4120825515947466,
      "grad_norm": 0.377645879983902,
      "learning_rate": 0.002828,
      "loss": 2.3523,
      "step": 18816
    },
    {
      "epoch": 1.4168855534709193,
      "grad_norm": 0.36612892150878906,
      "learning_rate": 0.002828,
      "loss": 2.3542,
      "step": 18880
    },
    {
      "epoch": 1.421688555347092,
      "grad_norm": 0.385735422372818,
      "learning_rate": 0.002828,
      "loss": 2.3528,
      "step": 18944
    },
    {
      "epoch": 1.4264915572232646,
      "grad_norm": 0.4026818871498108,
      "learning_rate": 0.002828,
      "loss": 2.3478,
      "step": 19008
    },
    {
      "epoch": 1.4312945590994373,
      "grad_norm": 0.39212891459465027,
      "learning_rate": 0.002828,
      "loss": 2.348,
      "step": 19072
    },
    {
      "epoch": 1.4360975609756097,
      "grad_norm": 0.43533411622047424,
      "learning_rate": 0.002828,
      "loss": 2.3502,
      "step": 19136
    },
    {
      "epoch": 1.4409005628517824,
      "grad_norm": 0.4136466383934021,
      "learning_rate": 0.002828,
      "loss": 2.347,
      "step": 19200
    },
    {
      "epoch": 1.445703564727955,
      "grad_norm": 0.38349345326423645,
      "learning_rate": 0.002828,
      "loss": 2.3468,
      "step": 19264
    },
    {
      "epoch": 1.4505065666041275,
      "grad_norm": 0.42666760087013245,
      "learning_rate": 0.002828,
      "loss": 2.3306,
      "step": 19328
    },
    {
      "epoch": 1.4553095684803001,
      "grad_norm": 0.3926577866077423,
      "learning_rate": 0.002828,
      "loss": 2.3466,
      "step": 19392
    },
    {
      "epoch": 1.4601125703564728,
      "grad_norm": 0.3736414611339569,
      "learning_rate": 0.002828,
      "loss": 2.3425,
      "step": 19456
    },
    {
      "epoch": 1.4649155722326455,
      "grad_norm": 0.36343908309936523,
      "learning_rate": 0.002828,
      "loss": 2.3371,
      "step": 19520
    },
    {
      "epoch": 1.4697185741088181,
      "grad_norm": 0.380211740732193,
      "learning_rate": 0.002828,
      "loss": 2.34,
      "step": 19584
    },
    {
      "epoch": 1.4745215759849906,
      "grad_norm": 0.40481454133987427,
      "learning_rate": 0.002828,
      "loss": 2.332,
      "step": 19648
    },
    {
      "epoch": 1.4793245778611632,
      "grad_norm": 0.45368635654449463,
      "learning_rate": 0.002828,
      "loss": 2.3288,
      "step": 19712
    },
    {
      "epoch": 1.484127579737336,
      "grad_norm": 0.4029395580291748,
      "learning_rate": 0.002828,
      "loss": 2.3293,
      "step": 19776
    },
    {
      "epoch": 1.4889305816135083,
      "grad_norm": 0.3748946785926819,
      "learning_rate": 0.002828,
      "loss": 2.332,
      "step": 19840
    },
    {
      "epoch": 1.493733583489681,
      "grad_norm": 0.36640551686286926,
      "learning_rate": 0.002828,
      "loss": 2.3231,
      "step": 19904
    },
    {
      "epoch": 1.4985365853658537,
      "grad_norm": 0.4150533676147461,
      "learning_rate": 0.002828,
      "loss": 2.3284,
      "step": 19968
    },
    {
      "epoch": 1.5033395872420263,
      "grad_norm": 0.49730879068374634,
      "learning_rate": 0.002828,
      "loss": 2.3243,
      "step": 20032
    },
    {
      "epoch": 1.508142589118199,
      "grad_norm": 0.37675461173057556,
      "learning_rate": 0.002828,
      "loss": 2.3273,
      "step": 20096
    },
    {
      "epoch": 1.5129455909943714,
      "grad_norm": 0.3647516965866089,
      "learning_rate": 0.002828,
      "loss": 2.3308,
      "step": 20160
    },
    {
      "epoch": 1.517748592870544,
      "grad_norm": 1.2981253862380981,
      "learning_rate": 0.002828,
      "loss": 2.3265,
      "step": 20224
    },
    {
      "epoch": 1.5225515947467168,
      "grad_norm": 0.5044511556625366,
      "learning_rate": 0.002828,
      "loss": 2.3173,
      "step": 20288
    },
    {
      "epoch": 1.5273545966228892,
      "grad_norm": 0.3651883006095886,
      "learning_rate": 0.002828,
      "loss": 2.319,
      "step": 20352
    },
    {
      "epoch": 1.532157598499062,
      "grad_norm": 0.4419403076171875,
      "learning_rate": 0.002828,
      "loss": 2.3188,
      "step": 20416
    },
    {
      "epoch": 1.5369606003752345,
      "grad_norm": 0.38631224632263184,
      "learning_rate": 0.002828,
      "loss": 2.3189,
      "step": 20480
    },
    {
      "epoch": 1.5417636022514072,
      "grad_norm": 0.34725359082221985,
      "learning_rate": 0.002828,
      "loss": 2.3162,
      "step": 20544
    },
    {
      "epoch": 1.5465666041275798,
      "grad_norm": 0.3991786241531372,
      "learning_rate": 0.002828,
      "loss": 2.3154,
      "step": 20608
    },
    {
      "epoch": 1.5513696060037523,
      "grad_norm": 0.3595084846019745,
      "learning_rate": 0.002828,
      "loss": 2.3149,
      "step": 20672
    },
    {
      "epoch": 1.556172607879925,
      "grad_norm": 0.4021853804588318,
      "learning_rate": 0.002828,
      "loss": 2.3113,
      "step": 20736
    },
    {
      "epoch": 1.5609756097560976,
      "grad_norm": 0.3939075767993927,
      "learning_rate": 0.002828,
      "loss": 2.3187,
      "step": 20800
    },
    {
      "epoch": 1.56577861163227,
      "grad_norm": 0.3889540135860443,
      "learning_rate": 0.002828,
      "loss": 2.316,
      "step": 20864
    },
    {
      "epoch": 1.570581613508443,
      "grad_norm": 0.41366517543792725,
      "learning_rate": 0.002828,
      "loss": 2.3025,
      "step": 20928
    },
    {
      "epoch": 1.5753846153846154,
      "grad_norm": 0.37127187848091125,
      "learning_rate": 0.002828,
      "loss": 2.3049,
      "step": 20992
    },
    {
      "epoch": 1.580187617260788,
      "grad_norm": 0.4014946520328522,
      "learning_rate": 0.002828,
      "loss": 2.3025,
      "step": 21056
    },
    {
      "epoch": 1.5849906191369607,
      "grad_norm": 0.35794708132743835,
      "learning_rate": 0.002828,
      "loss": 2.3089,
      "step": 21120
    },
    {
      "epoch": 1.5897936210131332,
      "grad_norm": 0.3924767076969147,
      "learning_rate": 0.002828,
      "loss": 2.3042,
      "step": 21184
    },
    {
      "epoch": 1.5945966228893058,
      "grad_norm": 0.34789031744003296,
      "learning_rate": 0.002828,
      "loss": 2.3025,
      "step": 21248
    },
    {
      "epoch": 1.5993996247654785,
      "grad_norm": 0.37461933493614197,
      "learning_rate": 0.002828,
      "loss": 2.2977,
      "step": 21312
    },
    {
      "epoch": 1.604202626641651,
      "grad_norm": 0.40146076679229736,
      "learning_rate": 0.002828,
      "loss": 2.2997,
      "step": 21376
    },
    {
      "epoch": 1.6090056285178238,
      "grad_norm": 0.4080921411514282,
      "learning_rate": 0.002828,
      "loss": 2.3001,
      "step": 21440
    },
    {
      "epoch": 1.6138086303939962,
      "grad_norm": 0.40802744030952454,
      "learning_rate": 0.002828,
      "loss": 2.2964,
      "step": 21504
    },
    {
      "epoch": 1.618611632270169,
      "grad_norm": 0.420188307762146,
      "learning_rate": 0.002828,
      "loss": 2.2981,
      "step": 21568
    },
    {
      "epoch": 1.6234146341463416,
      "grad_norm": 0.40739214420318604,
      "learning_rate": 0.002828,
      "loss": 2.2994,
      "step": 21632
    },
    {
      "epoch": 1.628217636022514,
      "grad_norm": 0.41674676537513733,
      "learning_rate": 0.002828,
      "loss": 2.292,
      "step": 21696
    },
    {
      "epoch": 1.6330206378986867,
      "grad_norm": 0.41856762766838074,
      "learning_rate": 0.002828,
      "loss": 2.2941,
      "step": 21760
    },
    {
      "epoch": 1.6378236397748593,
      "grad_norm": 0.39763346314430237,
      "learning_rate": 0.002828,
      "loss": 2.2939,
      "step": 21824
    },
    {
      "epoch": 1.6426266416510318,
      "grad_norm": 0.3777034282684326,
      "learning_rate": 0.002828,
      "loss": 2.2995,
      "step": 21888
    },
    {
      "epoch": 1.6474296435272047,
      "grad_norm": 0.3617188036441803,
      "learning_rate": 0.002828,
      "loss": 2.2924,
      "step": 21952
    },
    {
      "epoch": 1.652232645403377,
      "grad_norm": 0.4504718482494354,
      "learning_rate": 0.002828,
      "loss": 2.2869,
      "step": 22016
    },
    {
      "epoch": 1.6570356472795498,
      "grad_norm": 0.37388357520103455,
      "learning_rate": 0.002828,
      "loss": 2.2973,
      "step": 22080
    },
    {
      "epoch": 1.6618386491557224,
      "grad_norm": 0.3807313144207001,
      "learning_rate": 0.002828,
      "loss": 2.2829,
      "step": 22144
    },
    {
      "epoch": 1.6666416510318949,
      "grad_norm": 0.4428509771823883,
      "learning_rate": 0.002828,
      "loss": 2.2933,
      "step": 22208
    },
    {
      "epoch": 1.6714446529080675,
      "grad_norm": 0.39028382301330566,
      "learning_rate": 0.002828,
      "loss": 2.2931,
      "step": 22272
    },
    {
      "epoch": 1.6762476547842402,
      "grad_norm": 0.482424259185791,
      "learning_rate": 0.002828,
      "loss": 2.2781,
      "step": 22336
    },
    {
      "epoch": 1.6810506566604126,
      "grad_norm": 0.39801299571990967,
      "learning_rate": 0.002828,
      "loss": 2.2722,
      "step": 22400
    },
    {
      "epoch": 1.6858536585365855,
      "grad_norm": 0.4351527988910675,
      "learning_rate": 0.002828,
      "loss": 2.2801,
      "step": 22464
    },
    {
      "epoch": 1.690656660412758,
      "grad_norm": 0.4509490430355072,
      "learning_rate": 0.002828,
      "loss": 2.2839,
      "step": 22528
    },
    {
      "epoch": 1.6954596622889306,
      "grad_norm": 0.35885152220726013,
      "learning_rate": 0.002828,
      "loss": 2.2783,
      "step": 22592
    },
    {
      "epoch": 1.7002626641651033,
      "grad_norm": 0.4146900177001953,
      "learning_rate": 0.002828,
      "loss": 2.2803,
      "step": 22656
    },
    {
      "epoch": 1.7050656660412757,
      "grad_norm": 0.40194573998451233,
      "learning_rate": 0.002828,
      "loss": 2.2807,
      "step": 22720
    },
    {
      "epoch": 1.7098686679174484,
      "grad_norm": 0.43570390343666077,
      "learning_rate": 0.002828,
      "loss": 2.2824,
      "step": 22784
    },
    {
      "epoch": 1.714671669793621,
      "grad_norm": 0.35558512806892395,
      "learning_rate": 0.002828,
      "loss": 2.2777,
      "step": 22848
    },
    {
      "epoch": 1.7194746716697935,
      "grad_norm": 0.3700902760028839,
      "learning_rate": 0.002828,
      "loss": 2.2802,
      "step": 22912
    },
    {
      "epoch": 1.7242776735459664,
      "grad_norm": 0.4287453591823578,
      "learning_rate": 0.002828,
      "loss": 2.267,
      "step": 22976
    },
    {
      "epoch": 1.7290806754221388,
      "grad_norm": 0.41843536496162415,
      "learning_rate": 0.002828,
      "loss": 2.2695,
      "step": 23040
    },
    {
      "epoch": 1.7338836772983115,
      "grad_norm": 0.3938317596912384,
      "learning_rate": 0.002828,
      "loss": 2.2765,
      "step": 23104
    },
    {
      "epoch": 1.7386866791744842,
      "grad_norm": 0.44625958800315857,
      "learning_rate": 0.002828,
      "loss": 2.2661,
      "step": 23168
    },
    {
      "epoch": 1.7434896810506566,
      "grad_norm": 0.4598078727722168,
      "learning_rate": 0.002828,
      "loss": 2.2565,
      "step": 23232
    },
    {
      "epoch": 1.7482926829268293,
      "grad_norm": 0.4126788377761841,
      "learning_rate": 0.002828,
      "loss": 2.2688,
      "step": 23296
    },
    {
      "epoch": 1.753095684803002,
      "grad_norm": 0.3801914155483246,
      "learning_rate": 0.002828,
      "loss": 2.2633,
      "step": 23360
    },
    {
      "epoch": 1.7578986866791744,
      "grad_norm": 0.4619985818862915,
      "learning_rate": 0.002828,
      "loss": 2.2666,
      "step": 23424
    },
    {
      "epoch": 1.7627016885553473,
      "grad_norm": 0.4068593680858612,
      "learning_rate": 0.002828,
      "loss": 2.2578,
      "step": 23488
    },
    {
      "epoch": 1.7675046904315197,
      "grad_norm": 0.36146870255470276,
      "learning_rate": 0.002828,
      "loss": 2.2641,
      "step": 23552
    },
    {
      "epoch": 1.7723076923076924,
      "grad_norm": 0.3995908200740814,
      "learning_rate": 0.002828,
      "loss": 2.2592,
      "step": 23616
    },
    {
      "epoch": 1.777110694183865,
      "grad_norm": 0.3970596492290497,
      "learning_rate": 0.002828,
      "loss": 2.2719,
      "step": 23680
    },
    {
      "epoch": 1.7819136960600375,
      "grad_norm": 0.4287073612213135,
      "learning_rate": 0.002828,
      "loss": 2.2566,
      "step": 23744
    },
    {
      "epoch": 1.7867166979362101,
      "grad_norm": 0.41250482201576233,
      "learning_rate": 0.002828,
      "loss": 2.2533,
      "step": 23808
    },
    {
      "epoch": 1.7915196998123828,
      "grad_norm": 0.411668062210083,
      "learning_rate": 0.002828,
      "loss": 2.2586,
      "step": 23872
    },
    {
      "epoch": 1.7963227016885552,
      "grad_norm": 0.4834740459918976,
      "learning_rate": 0.002828,
      "loss": 2.2488,
      "step": 23936
    },
    {
      "epoch": 1.8011257035647281,
      "grad_norm": 0.3624022603034973,
      "learning_rate": 0.002828,
      "loss": 2.2485,
      "step": 24000
    },
    {
      "epoch": 1.8059287054409006,
      "grad_norm": 0.36700454354286194,
      "learning_rate": 0.002828,
      "loss": 2.2516,
      "step": 24064
    },
    {
      "epoch": 1.8107317073170732,
      "grad_norm": 0.3666454255580902,
      "learning_rate": 0.002828,
      "loss": 2.2483,
      "step": 24128
    },
    {
      "epoch": 1.8155347091932459,
      "grad_norm": 0.4110506474971771,
      "learning_rate": 0.002828,
      "loss": 2.2539,
      "step": 24192
    },
    {
      "epoch": 1.8203377110694183,
      "grad_norm": 0.3604464530944824,
      "learning_rate": 0.002828,
      "loss": 2.2477,
      "step": 24256
    },
    {
      "epoch": 1.825140712945591,
      "grad_norm": 0.40807706117630005,
      "learning_rate": 0.002828,
      "loss": 2.2483,
      "step": 24320
    },
    {
      "epoch": 1.8299437148217637,
      "grad_norm": 0.3632533848285675,
      "learning_rate": 0.002828,
      "loss": 2.2496,
      "step": 24384
    },
    {
      "epoch": 1.834746716697936,
      "grad_norm": 0.38520562648773193,
      "learning_rate": 0.002828,
      "loss": 2.238,
      "step": 24448
    },
    {
      "epoch": 1.839549718574109,
      "grad_norm": 0.4228810966014862,
      "learning_rate": 0.002828,
      "loss": 2.2437,
      "step": 24512
    },
    {
      "epoch": 1.8443527204502814,
      "grad_norm": 0.4542325735092163,
      "learning_rate": 0.002828,
      "loss": 2.2392,
      "step": 24576
    },
    {
      "epoch": 1.849155722326454,
      "grad_norm": 0.37316882610321045,
      "learning_rate": 0.002828,
      "loss": 2.2414,
      "step": 24640
    },
    {
      "epoch": 1.8539587242026268,
      "grad_norm": 0.5505624413490295,
      "learning_rate": 0.002828,
      "loss": 2.2535,
      "step": 24704
    },
    {
      "epoch": 1.8587617260787992,
      "grad_norm": 0.4269484281539917,
      "learning_rate": 0.002828,
      "loss": 2.2403,
      "step": 24768
    },
    {
      "epoch": 1.8635647279549719,
      "grad_norm": 0.407760888338089,
      "learning_rate": 0.002828,
      "loss": 2.2399,
      "step": 24832
    },
    {
      "epoch": 1.8683677298311445,
      "grad_norm": 0.4192192554473877,
      "learning_rate": 0.002828,
      "loss": 2.2418,
      "step": 24896
    },
    {
      "epoch": 1.873170731707317,
      "grad_norm": 0.3924838602542877,
      "learning_rate": 0.002828,
      "loss": 2.2345,
      "step": 24960
    },
    {
      "epoch": 1.8779737335834898,
      "grad_norm": 0.3799656629562378,
      "learning_rate": 0.002828,
      "loss": 2.2281,
      "step": 25024
    },
    {
      "epoch": 1.8827767354596623,
      "grad_norm": 0.40570494532585144,
      "learning_rate": 0.002828,
      "loss": 2.2401,
      "step": 25088
    },
    {
      "epoch": 1.887579737335835,
      "grad_norm": 0.3898228704929352,
      "learning_rate": 0.002828,
      "loss": 2.2313,
      "step": 25152
    },
    {
      "epoch": 1.8923827392120076,
      "grad_norm": 0.393216073513031,
      "learning_rate": 0.002828,
      "loss": 2.2407,
      "step": 25216
    },
    {
      "epoch": 1.89718574108818,
      "grad_norm": 0.4247749149799347,
      "learning_rate": 0.002828,
      "loss": 2.2365,
      "step": 25280
    },
    {
      "epoch": 1.9019887429643527,
      "grad_norm": 0.4670035243034363,
      "learning_rate": 0.002828,
      "loss": 2.2345,
      "step": 25344
    },
    {
      "epoch": 1.9067917448405254,
      "grad_norm": 0.40336644649505615,
      "learning_rate": 0.002828,
      "loss": 2.2367,
      "step": 25408
    },
    {
      "epoch": 1.9115947467166978,
      "grad_norm": 0.48462921380996704,
      "learning_rate": 0.002828,
      "loss": 2.2239,
      "step": 25472
    },
    {
      "epoch": 1.9163977485928707,
      "grad_norm": 0.44047805666923523,
      "learning_rate": 0.002828,
      "loss": 2.2291,
      "step": 25536
    },
    {
      "epoch": 1.9212007504690432,
      "grad_norm": 0.4221409261226654,
      "learning_rate": 0.002828,
      "loss": 2.2331,
      "step": 25600
    },
    {
      "epoch": 1.9260037523452158,
      "grad_norm": 0.4272362291812897,
      "learning_rate": 0.002828,
      "loss": 2.2286,
      "step": 25664
    },
    {
      "epoch": 1.9308067542213885,
      "grad_norm": 0.4309645891189575,
      "learning_rate": 0.002828,
      "loss": 2.2208,
      "step": 25728
    },
    {
      "epoch": 1.935609756097561,
      "grad_norm": 0.4220867156982422,
      "learning_rate": 0.002828,
      "loss": 2.2257,
      "step": 25792
    },
    {
      "epoch": 1.9404127579737336,
      "grad_norm": 0.3765920102596283,
      "learning_rate": 0.002828,
      "loss": 2.2262,
      "step": 25856
    },
    {
      "epoch": 1.9452157598499062,
      "grad_norm": 0.44643986225128174,
      "learning_rate": 0.002828,
      "loss": 2.2225,
      "step": 25920
    },
    {
      "epoch": 1.9500187617260787,
      "grad_norm": 0.4022061824798584,
      "learning_rate": 0.002828,
      "loss": 2.2263,
      "step": 25984
    },
    {
      "epoch": 1.9548217636022516,
      "grad_norm": 0.3935778737068176,
      "learning_rate": 0.002828,
      "loss": 2.2187,
      "step": 26048
    },
    {
      "epoch": 1.959624765478424,
      "grad_norm": 0.3877500295639038,
      "learning_rate": 0.002828,
      "loss": 2.2231,
      "step": 26112
    },
    {
      "epoch": 1.9644277673545967,
      "grad_norm": 0.3891729712486267,
      "learning_rate": 0.002828,
      "loss": 2.219,
      "step": 26176
    },
    {
      "epoch": 1.9692307692307693,
      "grad_norm": 0.3616099953651428,
      "learning_rate": 0.002828,
      "loss": 2.2221,
      "step": 26240
    },
    {
      "epoch": 1.9740337711069418,
      "grad_norm": 0.3855280578136444,
      "learning_rate": 0.002828,
      "loss": 2.2288,
      "step": 26304
    },
    {
      "epoch": 1.9788367729831144,
      "grad_norm": 0.44039493799209595,
      "learning_rate": 0.002828,
      "loss": 2.2248,
      "step": 26368
    },
    {
      "epoch": 1.983639774859287,
      "grad_norm": 0.37217262387275696,
      "learning_rate": 0.002828,
      "loss": 2.2201,
      "step": 26432
    },
    {
      "epoch": 1.9884427767354595,
      "grad_norm": 0.3942553997039795,
      "learning_rate": 0.002828,
      "loss": 2.2087,
      "step": 26496
    },
    {
      "epoch": 1.9932457786116324,
      "grad_norm": 0.3975297808647156,
      "learning_rate": 0.002828,
      "loss": 2.2037,
      "step": 26560
    },
    {
      "epoch": 1.9980487804878049,
      "grad_norm": 0.39197394251823425,
      "learning_rate": 0.002828,
      "loss": 2.2114,
      "step": 26624
    },
    {
      "epoch": 2.0028517823639773,
      "grad_norm": 0.38722801208496094,
      "learning_rate": 0.002828,
      "loss": 2.1951,
      "step": 26688
    },
    {
      "epoch": 2.00765478424015,
      "grad_norm": 0.38619640469551086,
      "learning_rate": 0.002828,
      "loss": 2.1736,
      "step": 26752
    },
    {
      "epoch": 2.0124577861163226,
      "grad_norm": 0.49529945850372314,
      "learning_rate": 0.002828,
      "loss": 2.1734,
      "step": 26816
    },
    {
      "epoch": 2.0172607879924955,
      "grad_norm": 0.4199656844139099,
      "learning_rate": 0.002828,
      "loss": 2.1684,
      "step": 26880
    },
    {
      "epoch": 2.022063789868668,
      "grad_norm": 0.45820868015289307,
      "learning_rate": 0.002828,
      "loss": 2.1861,
      "step": 26944
    },
    {
      "epoch": 2.0268667917448404,
      "grad_norm": 0.4006725251674652,
      "learning_rate": 0.002828,
      "loss": 2.1743,
      "step": 27008
    },
    {
      "epoch": 2.0316697936210133,
      "grad_norm": 0.4596467614173889,
      "learning_rate": 0.002828,
      "loss": 2.1734,
      "step": 27072
    },
    {
      "epoch": 2.0364727954971857,
      "grad_norm": 0.38660213351249695,
      "learning_rate": 0.002828,
      "loss": 2.1673,
      "step": 27136
    },
    {
      "epoch": 2.041275797373358,
      "grad_norm": 0.44082361459732056,
      "learning_rate": 0.002828,
      "loss": 2.1747,
      "step": 27200
    },
    {
      "epoch": 2.046078799249531,
      "grad_norm": 0.3886605203151703,
      "learning_rate": 0.002828,
      "loss": 2.1771,
      "step": 27264
    },
    {
      "epoch": 2.0508818011257035,
      "grad_norm": 0.41386017203330994,
      "learning_rate": 0.002828,
      "loss": 2.1734,
      "step": 27328
    },
    {
      "epoch": 2.055684803001876,
      "grad_norm": 0.411478191614151,
      "learning_rate": 0.002828,
      "loss": 2.1682,
      "step": 27392
    },
    {
      "epoch": 2.060487804878049,
      "grad_norm": 0.47288912534713745,
      "learning_rate": 0.002828,
      "loss": 2.1703,
      "step": 27456
    },
    {
      "epoch": 2.0652908067542213,
      "grad_norm": 0.36384883522987366,
      "learning_rate": 0.002828,
      "loss": 2.1678,
      "step": 27520
    },
    {
      "epoch": 2.070093808630394,
      "grad_norm": 0.40636852383613586,
      "learning_rate": 0.002828,
      "loss": 2.1671,
      "step": 27584
    },
    {
      "epoch": 2.0748968105065666,
      "grad_norm": 0.4425170421600342,
      "learning_rate": 0.002828,
      "loss": 2.1727,
      "step": 27648
    },
    {
      "epoch": 2.079699812382739,
      "grad_norm": 0.48468896746635437,
      "learning_rate": 0.002828,
      "loss": 2.1678,
      "step": 27712
    },
    {
      "epoch": 2.084502814258912,
      "grad_norm": 0.40420570969581604,
      "learning_rate": 0.002828,
      "loss": 2.1763,
      "step": 27776
    },
    {
      "epoch": 2.0893058161350844,
      "grad_norm": 0.44314709305763245,
      "learning_rate": 0.002828,
      "loss": 2.166,
      "step": 27840
    },
    {
      "epoch": 2.0941088180112573,
      "grad_norm": 0.5187743306159973,
      "learning_rate": 0.002828,
      "loss": 2.1702,
      "step": 27904
    },
    {
      "epoch": 2.0989118198874297,
      "grad_norm": 0.4796048104763031,
      "learning_rate": 0.002828,
      "loss": 2.1618,
      "step": 27968
    },
    {
      "epoch": 2.103714821763602,
      "grad_norm": 0.43605130910873413,
      "learning_rate": 0.002828,
      "loss": 2.1622,
      "step": 28032
    },
    {
      "epoch": 2.108517823639775,
      "grad_norm": 0.4523628056049347,
      "learning_rate": 0.002828,
      "loss": 2.1633,
      "step": 28096
    },
    {
      "epoch": 2.1133208255159475,
      "grad_norm": 0.4183247983455658,
      "learning_rate": 0.002828,
      "loss": 2.1612,
      "step": 28160
    },
    {
      "epoch": 2.11812382739212,
      "grad_norm": 0.5113268494606018,
      "learning_rate": 0.002828,
      "loss": 2.1591,
      "step": 28224
    },
    {
      "epoch": 2.122926829268293,
      "grad_norm": 0.40837016701698303,
      "learning_rate": 0.002828,
      "loss": 2.1616,
      "step": 28288
    },
    {
      "epoch": 2.1277298311444652,
      "grad_norm": 0.40093889832496643,
      "learning_rate": 0.002828,
      "loss": 2.1672,
      "step": 28352
    },
    {
      "epoch": 2.1325328330206377,
      "grad_norm": 0.3988894820213318,
      "learning_rate": 0.002828,
      "loss": 2.1638,
      "step": 28416
    },
    {
      "epoch": 2.1373358348968106,
      "grad_norm": 0.42024731636047363,
      "learning_rate": 0.002828,
      "loss": 2.1544,
      "step": 28480
    },
    {
      "epoch": 2.142138836772983,
      "grad_norm": 0.38691264390945435,
      "learning_rate": 0.002828,
      "loss": 2.1649,
      "step": 28544
    },
    {
      "epoch": 2.146941838649156,
      "grad_norm": 0.41956332325935364,
      "learning_rate": 0.002828,
      "loss": 2.1586,
      "step": 28608
    },
    {
      "epoch": 2.1517448405253283,
      "grad_norm": 0.4035188555717468,
      "learning_rate": 0.002828,
      "loss": 2.165,
      "step": 28672
    },
    {
      "epoch": 2.1565478424015008,
      "grad_norm": 0.35282230377197266,
      "learning_rate": 0.002828,
      "loss": 2.145,
      "step": 28736
    },
    {
      "epoch": 2.1613508442776737,
      "grad_norm": 0.43618568778038025,
      "learning_rate": 0.002828,
      "loss": 2.1548,
      "step": 28800
    },
    {
      "epoch": 2.166153846153846,
      "grad_norm": 0.4310976564884186,
      "learning_rate": 0.002828,
      "loss": 2.1476,
      "step": 28864
    },
    {
      "epoch": 2.170956848030019,
      "grad_norm": 0.4475420415401459,
      "learning_rate": 0.002828,
      "loss": 2.1556,
      "step": 28928
    },
    {
      "epoch": 2.1757598499061914,
      "grad_norm": 0.4384845197200775,
      "learning_rate": 0.002828,
      "loss": 2.154,
      "step": 28992
    },
    {
      "epoch": 2.180562851782364,
      "grad_norm": 0.40141811966896057,
      "learning_rate": 0.002828,
      "loss": 2.1534,
      "step": 29056
    },
    {
      "epoch": 2.1853658536585368,
      "grad_norm": 0.3754780888557434,
      "learning_rate": 0.002828,
      "loss": 2.1488,
      "step": 29120
    },
    {
      "epoch": 2.190168855534709,
      "grad_norm": 0.40471306443214417,
      "learning_rate": 0.002828,
      "loss": 2.1514,
      "step": 29184
    },
    {
      "epoch": 2.1949718574108816,
      "grad_norm": 1.464024543762207,
      "learning_rate": 0.002828,
      "loss": 2.1525,
      "step": 29248
    },
    {
      "epoch": 2.1997748592870545,
      "grad_norm": 0.3818819522857666,
      "learning_rate": 0.002828,
      "loss": 2.1484,
      "step": 29312
    },
    {
      "epoch": 2.204577861163227,
      "grad_norm": 0.3688436448574066,
      "learning_rate": 0.002828,
      "loss": 2.1451,
      "step": 29376
    },
    {
      "epoch": 2.2093808630393994,
      "grad_norm": 0.4367921054363251,
      "learning_rate": 0.002828,
      "loss": 2.147,
      "step": 29440
    },
    {
      "epoch": 2.2141838649155723,
      "grad_norm": 0.3566763401031494,
      "learning_rate": 0.002828,
      "loss": 2.1476,
      "step": 29504
    },
    {
      "epoch": 2.2189868667917447,
      "grad_norm": 0.4481133222579956,
      "learning_rate": 0.002828,
      "loss": 2.1431,
      "step": 29568
    },
    {
      "epoch": 2.2237898686679176,
      "grad_norm": 0.44622039794921875,
      "learning_rate": 0.002828,
      "loss": 2.148,
      "step": 29632
    },
    {
      "epoch": 2.22859287054409,
      "grad_norm": 0.4857657253742218,
      "learning_rate": 0.002828,
      "loss": 2.1383,
      "step": 29696
    },
    {
      "epoch": 2.2333958724202625,
      "grad_norm": 0.41923773288726807,
      "learning_rate": 0.002828,
      "loss": 2.1443,
      "step": 29760
    },
    {
      "epoch": 2.2381988742964354,
      "grad_norm": 0.4176802933216095,
      "learning_rate": 0.002828,
      "loss": 2.1413,
      "step": 29824
    },
    {
      "epoch": 2.243001876172608,
      "grad_norm": 0.4086935520172119,
      "learning_rate": 0.002828,
      "loss": 2.1349,
      "step": 29888
    },
    {
      "epoch": 2.2478048780487807,
      "grad_norm": 0.40138566493988037,
      "learning_rate": 0.002828,
      "loss": 2.1481,
      "step": 29952
    },
    {
      "epoch": 2.252607879924953,
      "grad_norm": 0.393996000289917,
      "learning_rate": 0.002828,
      "loss": 2.1435,
      "step": 30016
    },
    {
      "epoch": 2.2574108818011256,
      "grad_norm": 0.3962005078792572,
      "learning_rate": 0.002828,
      "loss": 2.1345,
      "step": 30080
    },
    {
      "epoch": 2.2622138836772985,
      "grad_norm": 0.41648438572883606,
      "learning_rate": 0.002828,
      "loss": 2.1469,
      "step": 30144
    },
    {
      "epoch": 2.267016885553471,
      "grad_norm": 0.3810112774372101,
      "learning_rate": 0.002828,
      "loss": 2.1365,
      "step": 30208
    },
    {
      "epoch": 2.2718198874296434,
      "grad_norm": 0.4520975649356842,
      "learning_rate": 0.002828,
      "loss": 2.1419,
      "step": 30272
    },
    {
      "epoch": 2.2766228893058162,
      "grad_norm": 0.4406943917274475,
      "learning_rate": 0.002828,
      "loss": 2.1363,
      "step": 30336
    },
    {
      "epoch": 2.2814258911819887,
      "grad_norm": 0.4186633825302124,
      "learning_rate": 0.002828,
      "loss": 2.1406,
      "step": 30400
    },
    {
      "epoch": 2.286228893058161,
      "grad_norm": 0.6210718154907227,
      "learning_rate": 0.002828,
      "loss": 2.1333,
      "step": 30464
    },
    {
      "epoch": 2.291031894934334,
      "grad_norm": 0.4192369282245636,
      "learning_rate": 0.002828,
      "loss": 2.1396,
      "step": 30528
    },
    {
      "epoch": 2.2958348968105065,
      "grad_norm": 0.4297274053096771,
      "learning_rate": 0.002828,
      "loss": 2.1304,
      "step": 30592
    },
    {
      "epoch": 2.3006378986866793,
      "grad_norm": 0.4519890546798706,
      "learning_rate": 0.002828,
      "loss": 2.1356,
      "step": 30656
    },
    {
      "epoch": 2.305440900562852,
      "grad_norm": 0.3641507625579834,
      "learning_rate": 0.002828,
      "loss": 2.1211,
      "step": 30720
    },
    {
      "epoch": 2.310243902439024,
      "grad_norm": 0.4051010310649872,
      "learning_rate": 0.002828,
      "loss": 2.1318,
      "step": 30784
    },
    {
      "epoch": 2.315046904315197,
      "grad_norm": 0.43515390157699585,
      "learning_rate": 0.002828,
      "loss": 2.1277,
      "step": 30848
    },
    {
      "epoch": 2.3198499061913695,
      "grad_norm": 0.4015902280807495,
      "learning_rate": 0.002828,
      "loss": 2.1343,
      "step": 30912
    },
    {
      "epoch": 2.3246529080675424,
      "grad_norm": 0.41404440999031067,
      "learning_rate": 0.002828,
      "loss": 2.13,
      "step": 30976
    },
    {
      "epoch": 2.329455909943715,
      "grad_norm": 0.3820297122001648,
      "learning_rate": 0.002828,
      "loss": 2.1323,
      "step": 31040
    },
    {
      "epoch": 2.3342589118198873,
      "grad_norm": 0.4260556697845459,
      "learning_rate": 0.002828,
      "loss": 2.1355,
      "step": 31104
    },
    {
      "epoch": 2.33906191369606,
      "grad_norm": 0.38206708431243896,
      "learning_rate": 0.002828,
      "loss": 2.133,
      "step": 31168
    },
    {
      "epoch": 2.3438649155722326,
      "grad_norm": 0.3741287291049957,
      "learning_rate": 0.002828,
      "loss": 2.1312,
      "step": 31232
    },
    {
      "epoch": 2.348667917448405,
      "grad_norm": 0.3969305157661438,
      "learning_rate": 0.002828,
      "loss": 2.1328,
      "step": 31296
    },
    {
      "epoch": 2.353470919324578,
      "grad_norm": 0.3579777479171753,
      "learning_rate": 0.002828,
      "loss": 2.1292,
      "step": 31360
    },
    {
      "epoch": 2.3582739212007504,
      "grad_norm": 0.40797194838523865,
      "learning_rate": 0.002828,
      "loss": 2.1268,
      "step": 31424
    },
    {
      "epoch": 2.363076923076923,
      "grad_norm": 0.4283476173877716,
      "learning_rate": 0.002828,
      "loss": 2.1298,
      "step": 31488
    },
    {
      "epoch": 2.3678799249530957,
      "grad_norm": 0.4102086126804352,
      "learning_rate": 0.002828,
      "loss": 2.1269,
      "step": 31552
    },
    {
      "epoch": 2.372682926829268,
      "grad_norm": 0.47719141840934753,
      "learning_rate": 0.002828,
      "loss": 2.1286,
      "step": 31616
    },
    {
      "epoch": 2.377485928705441,
      "grad_norm": 0.4112071692943573,
      "learning_rate": 0.002828,
      "loss": 2.1385,
      "step": 31680
    },
    {
      "epoch": 2.3822889305816135,
      "grad_norm": 0.3762068450450897,
      "learning_rate": 0.002828,
      "loss": 2.1208,
      "step": 31744
    },
    {
      "epoch": 2.387091932457786,
      "grad_norm": 0.4260576665401459,
      "learning_rate": 0.002828,
      "loss": 2.1238,
      "step": 31808
    },
    {
      "epoch": 2.391894934333959,
      "grad_norm": 0.4393415153026581,
      "learning_rate": 0.002828,
      "loss": 2.121,
      "step": 31872
    },
    {
      "epoch": 2.3966979362101313,
      "grad_norm": 0.36091846227645874,
      "learning_rate": 0.002828,
      "loss": 2.1181,
      "step": 31936
    },
    {
      "epoch": 2.401500938086304,
      "grad_norm": 0.3533402681350708,
      "learning_rate": 0.002828,
      "loss": 2.1184,
      "step": 32000
    },
    {
      "epoch": 2.4063039399624766,
      "grad_norm": 0.3956640362739563,
      "learning_rate": 0.002828,
      "loss": 2.1242,
      "step": 32064
    },
    {
      "epoch": 2.411106941838649,
      "grad_norm": 0.40656957030296326,
      "learning_rate": 0.002828,
      "loss": 2.1145,
      "step": 32128
    },
    {
      "epoch": 2.415909943714822,
      "grad_norm": 0.3594007194042206,
      "learning_rate": 0.002828,
      "loss": 2.1179,
      "step": 32192
    },
    {
      "epoch": 2.4207129455909944,
      "grad_norm": 0.4054548740386963,
      "learning_rate": 0.002828,
      "loss": 2.1081,
      "step": 32256
    },
    {
      "epoch": 2.425515947467167,
      "grad_norm": 0.39665716886520386,
      "learning_rate": 0.002828,
      "loss": 2.1192,
      "step": 32320
    },
    {
      "epoch": 2.4303189493433397,
      "grad_norm": 0.39287346601486206,
      "learning_rate": 0.002828,
      "loss": 2.1178,
      "step": 32384
    },
    {
      "epoch": 2.435121951219512,
      "grad_norm": 0.4614468514919281,
      "learning_rate": 0.002828,
      "loss": 2.1123,
      "step": 32448
    },
    {
      "epoch": 2.4399249530956846,
      "grad_norm": 0.36579185724258423,
      "learning_rate": 0.002828,
      "loss": 2.1149,
      "step": 32512
    },
    {
      "epoch": 2.4447279549718575,
      "grad_norm": 0.41442158818244934,
      "learning_rate": 0.002828,
      "loss": 2.1095,
      "step": 32576
    },
    {
      "epoch": 2.44953095684803,
      "grad_norm": 0.4049338102340698,
      "learning_rate": 0.002828,
      "loss": 2.1155,
      "step": 32640
    },
    {
      "epoch": 2.454333958724203,
      "grad_norm": 0.3908383250236511,
      "learning_rate": 0.002828,
      "loss": 2.1099,
      "step": 32704
    },
    {
      "epoch": 2.4591369606003752,
      "grad_norm": 0.4169704020023346,
      "learning_rate": 0.002828,
      "loss": 2.1162,
      "step": 32768
    },
    {
      "epoch": 2.4639399624765477,
      "grad_norm": 0.3969012498855591,
      "learning_rate": 0.002828,
      "loss": 2.1074,
      "step": 32832
    },
    {
      "epoch": 2.4687429643527206,
      "grad_norm": 0.40591520071029663,
      "learning_rate": 0.002828,
      "loss": 2.1114,
      "step": 32896
    },
    {
      "epoch": 2.473545966228893,
      "grad_norm": 0.44612857699394226,
      "learning_rate": 0.002828,
      "loss": 2.1111,
      "step": 32960
    },
    {
      "epoch": 2.478348968105066,
      "grad_norm": 0.38000956177711487,
      "learning_rate": 0.002828,
      "loss": 2.1076,
      "step": 33024
    },
    {
      "epoch": 2.4831519699812383,
      "grad_norm": 0.4089069068431854,
      "learning_rate": 0.002828,
      "loss": 2.1048,
      "step": 33088
    },
    {
      "epoch": 2.4879549718574108,
      "grad_norm": 0.38385865092277527,
      "learning_rate": 0.002828,
      "loss": 2.1078,
      "step": 33152
    },
    {
      "epoch": 2.4927579737335837,
      "grad_norm": 0.40285399556159973,
      "learning_rate": 0.002828,
      "loss": 2.106,
      "step": 33216
    },
    {
      "epoch": 2.497560975609756,
      "grad_norm": 0.43141239881515503,
      "learning_rate": 0.002828,
      "loss": 2.1022,
      "step": 33280
    },
    {
      "epoch": 2.5023639774859285,
      "grad_norm": 0.43552160263061523,
      "learning_rate": 0.002828,
      "loss": 2.1015,
      "step": 33344
    },
    {
      "epoch": 2.5071669793621014,
      "grad_norm": 0.41757678985595703,
      "learning_rate": 0.002828,
      "loss": 2.106,
      "step": 33408
    },
    {
      "epoch": 2.511969981238274,
      "grad_norm": 0.3721197247505188,
      "learning_rate": 0.002828,
      "loss": 2.108,
      "step": 33472
    },
    {
      "epoch": 2.5167729831144463,
      "grad_norm": 0.4426725208759308,
      "learning_rate": 0.002828,
      "loss": 2.11,
      "step": 33536
    },
    {
      "epoch": 2.521575984990619,
      "grad_norm": 0.37126022577285767,
      "learning_rate": 0.002828,
      "loss": 2.0907,
      "step": 33600
    },
    {
      "epoch": 2.5263789868667916,
      "grad_norm": 0.42241954803466797,
      "learning_rate": 0.002828,
      "loss": 2.099,
      "step": 33664
    },
    {
      "epoch": 2.5311819887429645,
      "grad_norm": 0.40958234667778015,
      "learning_rate": 0.002828,
      "loss": 2.1024,
      "step": 33728
    },
    {
      "epoch": 2.535984990619137,
      "grad_norm": 0.36936697363853455,
      "learning_rate": 0.002828,
      "loss": 2.0945,
      "step": 33792
    },
    {
      "epoch": 2.54078799249531,
      "grad_norm": 0.44610148668289185,
      "learning_rate": 0.002828,
      "loss": 2.1003,
      "step": 33856
    },
    {
      "epoch": 2.5455909943714823,
      "grad_norm": 0.5091739296913147,
      "learning_rate": 0.002828,
      "loss": 2.1024,
      "step": 33920
    },
    {
      "epoch": 2.5503939962476547,
      "grad_norm": 0.40063703060150146,
      "learning_rate": 0.002828,
      "loss": 2.094,
      "step": 33984
    },
    {
      "epoch": 2.5551969981238276,
      "grad_norm": 0.41543325781822205,
      "learning_rate": 0.002828,
      "loss": 2.1005,
      "step": 34048
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.38303419947624207,
      "learning_rate": 0.002828,
      "loss": 2.0995,
      "step": 34112
    },
    {
      "epoch": 2.5648030018761725,
      "grad_norm": 0.3361968696117401,
      "learning_rate": 0.002828,
      "loss": 2.0978,
      "step": 34176
    },
    {
      "epoch": 2.5696060037523454,
      "grad_norm": 0.4174779951572418,
      "learning_rate": 0.002828,
      "loss": 2.0965,
      "step": 34240
    },
    {
      "epoch": 2.574409005628518,
      "grad_norm": 0.3994362950325012,
      "learning_rate": 0.002828,
      "loss": 2.0917,
      "step": 34304
    },
    {
      "epoch": 2.5792120075046903,
      "grad_norm": 0.41613155603408813,
      "learning_rate": 0.002828,
      "loss": 2.0848,
      "step": 34368
    },
    {
      "epoch": 2.584015009380863,
      "grad_norm": 0.4391821324825287,
      "learning_rate": 0.002828,
      "loss": 2.0962,
      "step": 34432
    },
    {
      "epoch": 2.5888180112570356,
      "grad_norm": 0.3478713035583496,
      "learning_rate": 0.002828,
      "loss": 2.0881,
      "step": 34496
    },
    {
      "epoch": 2.593621013133208,
      "grad_norm": 0.4065977931022644,
      "learning_rate": 0.002828,
      "loss": 2.092,
      "step": 34560
    },
    {
      "epoch": 2.598424015009381,
      "grad_norm": 0.4749963581562042,
      "learning_rate": 0.002828,
      "loss": 2.0899,
      "step": 34624
    },
    {
      "epoch": 2.6032270168855534,
      "grad_norm": 0.4545346796512604,
      "learning_rate": 0.002828,
      "loss": 2.097,
      "step": 34688
    },
    {
      "epoch": 2.6080300187617262,
      "grad_norm": 0.40314188599586487,
      "learning_rate": 0.002828,
      "loss": 2.0837,
      "step": 34752
    },
    {
      "epoch": 2.6128330206378987,
      "grad_norm": 0.3616882562637329,
      "learning_rate": 0.002828,
      "loss": 2.0867,
      "step": 34816
    },
    {
      "epoch": 2.6176360225140716,
      "grad_norm": 0.37748977541923523,
      "learning_rate": 0.002828,
      "loss": 2.0899,
      "step": 34880
    },
    {
      "epoch": 2.622439024390244,
      "grad_norm": 0.43397149443626404,
      "learning_rate": 0.002828,
      "loss": 2.0831,
      "step": 34944
    },
    {
      "epoch": 2.6272420262664165,
      "grad_norm": 0.41302821040153503,
      "learning_rate": 0.002828,
      "loss": 2.0822,
      "step": 35008
    },
    {
      "epoch": 2.6320450281425893,
      "grad_norm": 0.47421374917030334,
      "learning_rate": 0.002828,
      "loss": 2.0803,
      "step": 35072
    },
    {
      "epoch": 2.636848030018762,
      "grad_norm": 0.4405677020549774,
      "learning_rate": 0.002828,
      "loss": 2.0802,
      "step": 35136
    },
    {
      "epoch": 2.641651031894934,
      "grad_norm": 0.448106974363327,
      "learning_rate": 0.002828,
      "loss": 2.0778,
      "step": 35200
    },
    {
      "epoch": 2.646454033771107,
      "grad_norm": 0.4040149748325348,
      "learning_rate": 0.002828,
      "loss": 2.0831,
      "step": 35264
    },
    {
      "epoch": 2.6512570356472795,
      "grad_norm": 0.3847975730895996,
      "learning_rate": 0.002828,
      "loss": 2.08,
      "step": 35328
    },
    {
      "epoch": 2.656060037523452,
      "grad_norm": 0.4193227291107178,
      "learning_rate": 0.002828,
      "loss": 2.0845,
      "step": 35392
    },
    {
      "epoch": 2.660863039399625,
      "grad_norm": 0.4024184048175812,
      "learning_rate": 0.002828,
      "loss": 2.0831,
      "step": 35456
    },
    {
      "epoch": 2.6656660412757973,
      "grad_norm": 0.3550715744495392,
      "learning_rate": 0.002828,
      "loss": 2.077,
      "step": 35520
    },
    {
      "epoch": 2.6704690431519698,
      "grad_norm": 0.41777607798576355,
      "learning_rate": 0.002828,
      "loss": 2.0735,
      "step": 35584
    },
    {
      "epoch": 2.6752720450281426,
      "grad_norm": 0.4474610388278961,
      "learning_rate": 0.002828,
      "loss": 2.0759,
      "step": 35648
    },
    {
      "epoch": 2.680075046904315,
      "grad_norm": 0.41162994503974915,
      "learning_rate": 0.002828,
      "loss": 2.0713,
      "step": 35712
    },
    {
      "epoch": 2.684878048780488,
      "grad_norm": 0.4786086976528168,
      "learning_rate": 0.002828,
      "loss": 2.0813,
      "step": 35776
    },
    {
      "epoch": 2.6896810506566604,
      "grad_norm": 0.3829205334186554,
      "learning_rate": 0.002828,
      "loss": 2.0697,
      "step": 35840
    },
    {
      "epoch": 2.6944840525328333,
      "grad_norm": 0.3732524514198303,
      "learning_rate": 0.002828,
      "loss": 2.077,
      "step": 35904
    },
    {
      "epoch": 2.6992870544090057,
      "grad_norm": 0.40800660848617554,
      "learning_rate": 0.002828,
      "loss": 2.0806,
      "step": 35968
    },
    {
      "epoch": 2.704090056285178,
      "grad_norm": 0.4223944842815399,
      "learning_rate": 0.002828,
      "loss": 2.0789,
      "step": 36032
    },
    {
      "epoch": 2.708893058161351,
      "grad_norm": 0.4355471134185791,
      "learning_rate": 0.002828,
      "loss": 2.0814,
      "step": 36096
    },
    {
      "epoch": 2.7136960600375235,
      "grad_norm": 0.4148624539375305,
      "learning_rate": 0.002828,
      "loss": 2.0716,
      "step": 36160
    },
    {
      "epoch": 2.718499061913696,
      "grad_norm": 0.399810254573822,
      "learning_rate": 0.002828,
      "loss": 2.08,
      "step": 36224
    },
    {
      "epoch": 2.723302063789869,
      "grad_norm": 0.47194531559944153,
      "learning_rate": 0.002828,
      "loss": 2.0709,
      "step": 36288
    },
    {
      "epoch": 2.7281050656660413,
      "grad_norm": 0.4752259850502014,
      "learning_rate": 0.002828,
      "loss": 2.0634,
      "step": 36352
    },
    {
      "epoch": 2.7329080675422137,
      "grad_norm": 0.38671767711639404,
      "learning_rate": 0.002828,
      "loss": 2.0782,
      "step": 36416
    },
    {
      "epoch": 2.7377110694183866,
      "grad_norm": 0.38888657093048096,
      "learning_rate": 0.002828,
      "loss": 2.0725,
      "step": 36480
    },
    {
      "epoch": 2.742514071294559,
      "grad_norm": 0.4416655898094177,
      "learning_rate": 0.002828,
      "loss": 2.0659,
      "step": 36544
    },
    {
      "epoch": 2.7473170731707315,
      "grad_norm": 0.4418342113494873,
      "learning_rate": 0.002828,
      "loss": 2.0783,
      "step": 36608
    },
    {
      "epoch": 2.7521200750469044,
      "grad_norm": 0.40704232454299927,
      "learning_rate": 0.002828,
      "loss": 2.0693,
      "step": 36672
    },
    {
      "epoch": 2.756923076923077,
      "grad_norm": 0.3956950008869171,
      "learning_rate": 0.002828,
      "loss": 2.0622,
      "step": 36736
    },
    {
      "epoch": 2.7617260787992497,
      "grad_norm": 0.43156442046165466,
      "learning_rate": 0.002828,
      "loss": 2.0664,
      "step": 36800
    },
    {
      "epoch": 2.766529080675422,
      "grad_norm": 0.3915492296218872,
      "learning_rate": 0.002828,
      "loss": 2.0751,
      "step": 36864
    },
    {
      "epoch": 2.771332082551595,
      "grad_norm": 0.39085447788238525,
      "learning_rate": 0.002828,
      "loss": 2.0688,
      "step": 36928
    },
    {
      "epoch": 2.7761350844277675,
      "grad_norm": 0.39727815985679626,
      "learning_rate": 0.002828,
      "loss": 2.0623,
      "step": 36992
    },
    {
      "epoch": 2.78093808630394,
      "grad_norm": 0.36193764209747314,
      "learning_rate": 0.002828,
      "loss": 2.0653,
      "step": 37056
    },
    {
      "epoch": 2.785741088180113,
      "grad_norm": 0.39149948954582214,
      "learning_rate": 0.002828,
      "loss": 2.066,
      "step": 37120
    },
    {
      "epoch": 2.7905440900562852,
      "grad_norm": 0.3922441601753235,
      "learning_rate": 0.002828,
      "loss": 2.0651,
      "step": 37184
    },
    {
      "epoch": 2.7953470919324577,
      "grad_norm": 0.40434616804122925,
      "learning_rate": 0.002828,
      "loss": 2.0569,
      "step": 37248
    },
    {
      "epoch": 2.8001500938086306,
      "grad_norm": 0.4445114731788635,
      "learning_rate": 0.002828,
      "loss": 2.0641,
      "step": 37312
    },
    {
      "epoch": 2.804953095684803,
      "grad_norm": 0.4767908751964569,
      "learning_rate": 0.002828,
      "loss": 2.0639,
      "step": 37376
    },
    {
      "epoch": 2.8097560975609754,
      "grad_norm": 0.49127042293548584,
      "learning_rate": 0.002828,
      "loss": 2.0545,
      "step": 37440
    },
    {
      "epoch": 2.8145590994371483,
      "grad_norm": 0.4333418607711792,
      "learning_rate": 0.002828,
      "loss": 2.0632,
      "step": 37504
    },
    {
      "epoch": 2.8193621013133208,
      "grad_norm": 0.4018191993236542,
      "learning_rate": 0.002828,
      "loss": 2.0601,
      "step": 37568
    },
    {
      "epoch": 2.824165103189493,
      "grad_norm": 0.4417150318622589,
      "learning_rate": 0.002828,
      "loss": 2.0596,
      "step": 37632
    },
    {
      "epoch": 2.828968105065666,
      "grad_norm": 0.38089001178741455,
      "learning_rate": 0.002828,
      "loss": 2.0592,
      "step": 37696
    },
    {
      "epoch": 2.8337711069418385,
      "grad_norm": 0.47056955099105835,
      "learning_rate": 0.002828,
      "loss": 2.0534,
      "step": 37760
    },
    {
      "epoch": 2.8385741088180114,
      "grad_norm": 0.3839799165725708,
      "learning_rate": 0.002828,
      "loss": 2.0564,
      "step": 37824
    },
    {
      "epoch": 2.843377110694184,
      "grad_norm": 0.4095609784126282,
      "learning_rate": 0.002828,
      "loss": 2.0529,
      "step": 37888
    },
    {
      "epoch": 2.8481801125703567,
      "grad_norm": 0.435360848903656,
      "learning_rate": 0.002828,
      "loss": 2.0605,
      "step": 37952
    },
    {
      "epoch": 2.852983114446529,
      "grad_norm": 0.40065625309944153,
      "learning_rate": 0.002828,
      "loss": 2.0606,
      "step": 38016
    },
    {
      "epoch": 2.8577861163227016,
      "grad_norm": 0.42678216099739075,
      "learning_rate": 0.002828,
      "loss": 2.0544,
      "step": 38080
    },
    {
      "epoch": 2.8625891181988745,
      "grad_norm": 0.46134546399116516,
      "learning_rate": 0.002828,
      "loss": 2.0529,
      "step": 38144
    },
    {
      "epoch": 2.867392120075047,
      "grad_norm": 0.42689305543899536,
      "learning_rate": 0.002828,
      "loss": 2.0568,
      "step": 38208
    },
    {
      "epoch": 2.8721951219512194,
      "grad_norm": 0.4389091730117798,
      "learning_rate": 0.002828,
      "loss": 2.0466,
      "step": 38272
    },
    {
      "epoch": 2.8769981238273923,
      "grad_norm": 0.416259229183197,
      "learning_rate": 0.002828,
      "loss": 2.0555,
      "step": 38336
    },
    {
      "epoch": 2.8818011257035647,
      "grad_norm": 0.3812130093574524,
      "learning_rate": 0.002828,
      "loss": 2.0505,
      "step": 38400
    },
    {
      "epoch": 2.886604127579737,
      "grad_norm": 0.41849854588508606,
      "learning_rate": 0.002828,
      "loss": 2.0533,
      "step": 38464
    },
    {
      "epoch": 2.89140712945591,
      "grad_norm": 0.38751932978630066,
      "learning_rate": 0.002828,
      "loss": 2.047,
      "step": 38528
    },
    {
      "epoch": 2.8962101313320825,
      "grad_norm": 0.4492979347705841,
      "learning_rate": 0.002828,
      "loss": 2.0477,
      "step": 38592
    },
    {
      "epoch": 2.901013133208255,
      "grad_norm": 0.4301586449146271,
      "learning_rate": 0.002828,
      "loss": 2.0517,
      "step": 38656
    },
    {
      "epoch": 2.905816135084428,
      "grad_norm": 0.42884722352027893,
      "learning_rate": 0.002828,
      "loss": 2.0513,
      "step": 38720
    },
    {
      "epoch": 2.9106191369606003,
      "grad_norm": 0.43645617365837097,
      "learning_rate": 0.002828,
      "loss": 2.0469,
      "step": 38784
    },
    {
      "epoch": 2.915422138836773,
      "grad_norm": 0.41132786870002747,
      "learning_rate": 0.002828,
      "loss": 2.0514,
      "step": 38848
    },
    {
      "epoch": 2.9202251407129456,
      "grad_norm": 0.40186938643455505,
      "learning_rate": 0.002828,
      "loss": 2.0473,
      "step": 38912
    },
    {
      "epoch": 2.9250281425891185,
      "grad_norm": 0.43834152817726135,
      "learning_rate": 0.002828,
      "loss": 2.0431,
      "step": 38976
    },
    {
      "epoch": 2.929831144465291,
      "grad_norm": 0.42384204268455505,
      "learning_rate": 0.002828,
      "loss": 2.0423,
      "step": 39040
    },
    {
      "epoch": 2.9346341463414634,
      "grad_norm": 0.41747942566871643,
      "learning_rate": 0.002828,
      "loss": 2.0459,
      "step": 39104
    },
    {
      "epoch": 2.9394371482176362,
      "grad_norm": 0.4123360514640808,
      "learning_rate": 0.002828,
      "loss": 2.0454,
      "step": 39168
    },
    {
      "epoch": 2.9442401500938087,
      "grad_norm": 0.3995719254016876,
      "learning_rate": 0.002828,
      "loss": 2.0422,
      "step": 39232
    },
    {
      "epoch": 2.949043151969981,
      "grad_norm": 0.38762396574020386,
      "learning_rate": 0.002828,
      "loss": 2.0428,
      "step": 39296
    },
    {
      "epoch": 2.953846153846154,
      "grad_norm": 0.4130190312862396,
      "learning_rate": 0.002828,
      "loss": 2.0383,
      "step": 39360
    },
    {
      "epoch": 2.9586491557223265,
      "grad_norm": 0.401840478181839,
      "learning_rate": 0.002828,
      "loss": 2.037,
      "step": 39424
    },
    {
      "epoch": 2.963452157598499,
      "grad_norm": 0.44890210032463074,
      "learning_rate": 0.002828,
      "loss": 2.0477,
      "step": 39488
    },
    {
      "epoch": 2.968255159474672,
      "grad_norm": 0.4392335116863251,
      "learning_rate": 0.002828,
      "loss": 2.0329,
      "step": 39552
    },
    {
      "epoch": 2.973058161350844,
      "grad_norm": 0.37307026982307434,
      "learning_rate": 0.002828,
      "loss": 2.0361,
      "step": 39616
    },
    {
      "epoch": 2.9778611632270167,
      "grad_norm": 0.3836345970630646,
      "learning_rate": 0.002828,
      "loss": 2.0389,
      "step": 39680
    },
    {
      "epoch": 2.9826641651031895,
      "grad_norm": 0.43940141797065735,
      "learning_rate": 0.002828,
      "loss": 2.0424,
      "step": 39744
    },
    {
      "epoch": 2.987467166979362,
      "grad_norm": 0.4530318081378937,
      "learning_rate": 0.002828,
      "loss": 2.0361,
      "step": 39808
    },
    {
      "epoch": 2.992270168855535,
      "grad_norm": 0.4106103777885437,
      "learning_rate": 0.002828,
      "loss": 2.0349,
      "step": 39872
    },
    {
      "epoch": 2.9970731707317073,
      "grad_norm": 0.3956582546234131,
      "learning_rate": 0.002828,
      "loss": 2.0371,
      "step": 39936
    },
    {
      "epoch": 3.0018761726078798,
      "grad_norm": 0.5075650811195374,
      "learning_rate": 0.002828,
      "loss": 2.0121,
      "step": 40000
    },
    {
      "epoch": 3.0066791744840526,
      "grad_norm": 0.39176109433174133,
      "learning_rate": 0.002828,
      "loss": 1.9939,
      "step": 40064
    },
    {
      "epoch": 3.011482176360225,
      "grad_norm": 0.40185442566871643,
      "learning_rate": 0.002828,
      "loss": 1.9972,
      "step": 40128
    },
    {
      "epoch": 3.016285178236398,
      "grad_norm": 0.41981250047683716,
      "learning_rate": 0.002828,
      "loss": 1.9962,
      "step": 40192
    },
    {
      "epoch": 3.0210881801125704,
      "grad_norm": 0.42668989300727844,
      "learning_rate": 0.002828,
      "loss": 1.9907,
      "step": 40256
    },
    {
      "epoch": 3.025891181988743,
      "grad_norm": 0.42785272002220154,
      "learning_rate": 0.002828,
      "loss": 1.9959,
      "step": 40320
    },
    {
      "epoch": 3.0306941838649157,
      "grad_norm": 0.4478468596935272,
      "learning_rate": 0.002828,
      "loss": 1.9921,
      "step": 40384
    },
    {
      "epoch": 3.035497185741088,
      "grad_norm": 0.4365384578704834,
      "learning_rate": 0.002828,
      "loss": 1.9849,
      "step": 40448
    },
    {
      "epoch": 3.0403001876172606,
      "grad_norm": 0.36439043283462524,
      "learning_rate": 0.002828,
      "loss": 1.9975,
      "step": 40512
    },
    {
      "epoch": 3.0451031894934335,
      "grad_norm": 0.38191044330596924,
      "learning_rate": 0.002828,
      "loss": 1.9898,
      "step": 40576
    },
    {
      "epoch": 3.049906191369606,
      "grad_norm": 0.403347909450531,
      "learning_rate": 0.002828,
      "loss": 1.9979,
      "step": 40640
    },
    {
      "epoch": 3.0547091932457784,
      "grad_norm": 0.40450242161750793,
      "learning_rate": 0.002828,
      "loss": 2.0019,
      "step": 40704
    },
    {
      "epoch": 3.0595121951219513,
      "grad_norm": 0.5009217858314514,
      "learning_rate": 0.002828,
      "loss": 1.9905,
      "step": 40768
    },
    {
      "epoch": 3.0643151969981237,
      "grad_norm": 0.4922749400138855,
      "learning_rate": 0.002828,
      "loss": 1.9924,
      "step": 40832
    },
    {
      "epoch": 3.0691181988742966,
      "grad_norm": 0.39887261390686035,
      "learning_rate": 0.002828,
      "loss": 1.9911,
      "step": 40896
    },
    {
      "epoch": 3.073921200750469,
      "grad_norm": 0.5205056071281433,
      "learning_rate": 0.002828,
      "loss": 1.9878,
      "step": 40960
    },
    {
      "epoch": 3.0787242026266415,
      "grad_norm": 0.41092032194137573,
      "learning_rate": 0.002828,
      "loss": 1.9867,
      "step": 41024
    },
    {
      "epoch": 3.0835272045028144,
      "grad_norm": 0.41716116666793823,
      "learning_rate": 0.002828,
      "loss": 1.9856,
      "step": 41088
    },
    {
      "epoch": 3.088330206378987,
      "grad_norm": 0.44289645552635193,
      "learning_rate": 0.002828,
      "loss": 1.986,
      "step": 41152
    },
    {
      "epoch": 3.0931332082551597,
      "grad_norm": 0.3966487944126129,
      "learning_rate": 0.002828,
      "loss": 1.99,
      "step": 41216
    },
    {
      "epoch": 3.097936210131332,
      "grad_norm": 0.4425393044948578,
      "learning_rate": 0.002828,
      "loss": 1.9896,
      "step": 41280
    },
    {
      "epoch": 3.1027392120075046,
      "grad_norm": 0.40042001008987427,
      "learning_rate": 0.002828,
      "loss": 1.9902,
      "step": 41344
    },
    {
      "epoch": 3.1075422138836775,
      "grad_norm": 0.3969666361808777,
      "learning_rate": 0.002828,
      "loss": 1.989,
      "step": 41408
    },
    {
      "epoch": 3.11234521575985,
      "grad_norm": 0.41544854640960693,
      "learning_rate": 0.002828,
      "loss": 1.9933,
      "step": 41472
    },
    {
      "epoch": 3.1171482176360223,
      "grad_norm": 0.436469167470932,
      "learning_rate": 0.002828,
      "loss": 1.9848,
      "step": 41536
    },
    {
      "epoch": 3.1219512195121952,
      "grad_norm": 0.39815834164619446,
      "learning_rate": 0.002828,
      "loss": 1.9866,
      "step": 41600
    },
    {
      "epoch": 3.1267542213883677,
      "grad_norm": 0.39662250876426697,
      "learning_rate": 0.002828,
      "loss": 1.985,
      "step": 41664
    },
    {
      "epoch": 3.13155722326454,
      "grad_norm": 0.4261801838874817,
      "learning_rate": 0.002828,
      "loss": 1.991,
      "step": 41728
    },
    {
      "epoch": 3.136360225140713,
      "grad_norm": 0.4388674199581146,
      "learning_rate": 0.002828,
      "loss": 1.9894,
      "step": 41792
    },
    {
      "epoch": 3.1411632270168854,
      "grad_norm": 0.4637046456336975,
      "learning_rate": 0.002828,
      "loss": 1.9897,
      "step": 41856
    },
    {
      "epoch": 3.1459662288930583,
      "grad_norm": 0.49249354004859924,
      "learning_rate": 0.002828,
      "loss": 1.985,
      "step": 41920
    },
    {
      "epoch": 3.1507692307692308,
      "grad_norm": 0.40578174591064453,
      "learning_rate": 0.002828,
      "loss": 1.9785,
      "step": 41984
    },
    {
      "epoch": 3.155572232645403,
      "grad_norm": 0.46401965618133545,
      "learning_rate": 0.002828,
      "loss": 1.9894,
      "step": 42048
    },
    {
      "epoch": 3.160375234521576,
      "grad_norm": 0.4829576909542084,
      "learning_rate": 0.002828,
      "loss": 1.9867,
      "step": 42112
    },
    {
      "epoch": 3.1651782363977485,
      "grad_norm": 0.4457123875617981,
      "learning_rate": 0.002828,
      "loss": 1.9877,
      "step": 42176
    },
    {
      "epoch": 3.1699812382739214,
      "grad_norm": 0.44146353006362915,
      "learning_rate": 0.002828,
      "loss": 1.9919,
      "step": 42240
    },
    {
      "epoch": 3.174784240150094,
      "grad_norm": 0.4063395857810974,
      "learning_rate": 0.002828,
      "loss": 1.9886,
      "step": 42304
    },
    {
      "epoch": 3.1795872420262663,
      "grad_norm": 0.40236446261405945,
      "learning_rate": 0.002828,
      "loss": 1.9808,
      "step": 42368
    },
    {
      "epoch": 3.184390243902439,
      "grad_norm": 0.41492629051208496,
      "learning_rate": 0.002828,
      "loss": 1.9826,
      "step": 42432
    },
    {
      "epoch": 3.1891932457786116,
      "grad_norm": 0.39708060026168823,
      "learning_rate": 0.002828,
      "loss": 1.9853,
      "step": 42496
    },
    {
      "epoch": 3.193996247654784,
      "grad_norm": 0.427532821893692,
      "learning_rate": 0.002828,
      "loss": 1.9845,
      "step": 42560
    },
    {
      "epoch": 3.198799249530957,
      "grad_norm": 0.48030269145965576,
      "learning_rate": 0.002828,
      "loss": 1.9801,
      "step": 42624
    },
    {
      "epoch": 3.2036022514071294,
      "grad_norm": 0.4397580027580261,
      "learning_rate": 0.002828,
      "loss": 1.982,
      "step": 42688
    },
    {
      "epoch": 3.208405253283302,
      "grad_norm": 0.41475823521614075,
      "learning_rate": 0.002828,
      "loss": 1.9828,
      "step": 42752
    },
    {
      "epoch": 3.2132082551594747,
      "grad_norm": 0.4905751049518585,
      "learning_rate": 0.002828,
      "loss": 1.9821,
      "step": 42816
    },
    {
      "epoch": 3.218011257035647,
      "grad_norm": 0.46179690957069397,
      "learning_rate": 0.002828,
      "loss": 1.9842,
      "step": 42880
    },
    {
      "epoch": 3.22281425891182,
      "grad_norm": 0.38780882954597473,
      "learning_rate": 0.002828,
      "loss": 1.9819,
      "step": 42944
    },
    {
      "epoch": 3.2276172607879925,
      "grad_norm": 0.48357096314430237,
      "learning_rate": 0.002828,
      "loss": 1.9851,
      "step": 43008
    },
    {
      "epoch": 3.232420262664165,
      "grad_norm": 0.4514494836330414,
      "learning_rate": 0.002828,
      "loss": 1.9788,
      "step": 43072
    },
    {
      "epoch": 3.237223264540338,
      "grad_norm": 0.3997925817966461,
      "learning_rate": 0.002828,
      "loss": 1.9815,
      "step": 43136
    },
    {
      "epoch": 3.2420262664165103,
      "grad_norm": 0.47594690322875977,
      "learning_rate": 0.002828,
      "loss": 1.9809,
      "step": 43200
    },
    {
      "epoch": 3.246829268292683,
      "grad_norm": 0.41271838545799255,
      "learning_rate": 0.002828,
      "loss": 1.9821,
      "step": 43264
    },
    {
      "epoch": 3.2516322701688556,
      "grad_norm": 0.3931159973144531,
      "learning_rate": 0.002828,
      "loss": 1.9772,
      "step": 43328
    },
    {
      "epoch": 3.256435272045028,
      "grad_norm": 0.4065423011779785,
      "learning_rate": 0.002828,
      "loss": 1.9776,
      "step": 43392
    },
    {
      "epoch": 3.261238273921201,
      "grad_norm": 0.4076635241508484,
      "learning_rate": 0.002828,
      "loss": 1.9906,
      "step": 43456
    },
    {
      "epoch": 3.2660412757973734,
      "grad_norm": 0.39124780893325806,
      "learning_rate": 0.002828,
      "loss": 1.986,
      "step": 43520
    },
    {
      "epoch": 3.270844277673546,
      "grad_norm": 0.40072616934776306,
      "learning_rate": 0.002828,
      "loss": 1.978,
      "step": 43584
    },
    {
      "epoch": 3.2756472795497187,
      "grad_norm": 0.41515403985977173,
      "learning_rate": 0.002828,
      "loss": 1.9816,
      "step": 43648
    },
    {
      "epoch": 3.280450281425891,
      "grad_norm": 0.3668558895587921,
      "learning_rate": 0.002828,
      "loss": 1.9737,
      "step": 43712
    },
    {
      "epoch": 3.2852532833020636,
      "grad_norm": 0.4493162930011749,
      "learning_rate": 0.002828,
      "loss": 1.9814,
      "step": 43776
    },
    {
      "epoch": 3.2900562851782365,
      "grad_norm": 0.38466036319732666,
      "learning_rate": 0.002828,
      "loss": 1.9762,
      "step": 43840
    },
    {
      "epoch": 3.294859287054409,
      "grad_norm": 0.5075457692146301,
      "learning_rate": 0.002828,
      "loss": 1.9858,
      "step": 43904
    },
    {
      "epoch": 3.2996622889305818,
      "grad_norm": 0.5009188055992126,
      "learning_rate": 0.002828,
      "loss": 1.9826,
      "step": 43968
    },
    {
      "epoch": 3.304465290806754,
      "grad_norm": 0.45684710144996643,
      "learning_rate": 0.002828,
      "loss": 1.9658,
      "step": 44032
    },
    {
      "epoch": 3.3092682926829267,
      "grad_norm": 0.4678841531276703,
      "learning_rate": 0.002828,
      "loss": 1.9745,
      "step": 44096
    },
    {
      "epoch": 3.3140712945590995,
      "grad_norm": 0.44828855991363525,
      "learning_rate": 0.002828,
      "loss": 1.9794,
      "step": 44160
    },
    {
      "epoch": 3.318874296435272,
      "grad_norm": 0.4242972731590271,
      "learning_rate": 0.002828,
      "loss": 1.977,
      "step": 44224
    },
    {
      "epoch": 3.323677298311445,
      "grad_norm": 0.4554446339607239,
      "learning_rate": 0.002828,
      "loss": 1.9722,
      "step": 44288
    },
    {
      "epoch": 3.3284803001876173,
      "grad_norm": 0.3926396667957306,
      "learning_rate": 0.002828,
      "loss": 1.9777,
      "step": 44352
    },
    {
      "epoch": 3.3332833020637898,
      "grad_norm": 0.4520684778690338,
      "learning_rate": 0.002828,
      "loss": 1.9738,
      "step": 44416
    },
    {
      "epoch": 3.3380863039399626,
      "grad_norm": 0.3810643255710602,
      "learning_rate": 0.002828,
      "loss": 1.9692,
      "step": 44480
    },
    {
      "epoch": 3.342889305816135,
      "grad_norm": 0.3734522759914398,
      "learning_rate": 0.002828,
      "loss": 1.9693,
      "step": 44544
    },
    {
      "epoch": 3.3476923076923075,
      "grad_norm": 0.43206340074539185,
      "learning_rate": 0.002828,
      "loss": 1.9667,
      "step": 44608
    },
    {
      "epoch": 3.3524953095684804,
      "grad_norm": 0.4002898335456848,
      "learning_rate": 0.002828,
      "loss": 1.9713,
      "step": 44672
    },
    {
      "epoch": 3.357298311444653,
      "grad_norm": 0.40499213337898254,
      "learning_rate": 0.002828,
      "loss": 1.9679,
      "step": 44736
    },
    {
      "epoch": 3.3621013133208253,
      "grad_norm": 0.41439104080200195,
      "learning_rate": 0.002828,
      "loss": 1.9746,
      "step": 44800
    },
    {
      "epoch": 3.366904315196998,
      "grad_norm": 0.3716445565223694,
      "learning_rate": 0.002828,
      "loss": 1.977,
      "step": 44864
    },
    {
      "epoch": 3.3717073170731706,
      "grad_norm": 0.40978100895881653,
      "learning_rate": 0.002828,
      "loss": 1.9733,
      "step": 44928
    },
    {
      "epoch": 3.3765103189493435,
      "grad_norm": 0.4042293131351471,
      "learning_rate": 0.002828,
      "loss": 1.964,
      "step": 44992
    },
    {
      "epoch": 3.381313320825516,
      "grad_norm": 0.3775876462459564,
      "learning_rate": 0.002828,
      "loss": 1.9592,
      "step": 45056
    },
    {
      "epoch": 3.3861163227016884,
      "grad_norm": 0.40276941657066345,
      "learning_rate": 0.002828,
      "loss": 1.9675,
      "step": 45120
    },
    {
      "epoch": 3.3909193245778613,
      "grad_norm": 0.43238821625709534,
      "learning_rate": 0.002828,
      "loss": 1.9659,
      "step": 45184
    },
    {
      "epoch": 3.3957223264540337,
      "grad_norm": 0.43237367272377014,
      "learning_rate": 0.002828,
      "loss": 1.9658,
      "step": 45248
    },
    {
      "epoch": 3.4005253283302066,
      "grad_norm": 0.4257899522781372,
      "learning_rate": 0.002828,
      "loss": 1.9661,
      "step": 45312
    },
    {
      "epoch": 3.405328330206379,
      "grad_norm": 0.4422502815723419,
      "learning_rate": 0.002828,
      "loss": 1.9674,
      "step": 45376
    },
    {
      "epoch": 3.4101313320825515,
      "grad_norm": 0.49216824769973755,
      "learning_rate": 0.002828,
      "loss": 1.9677,
      "step": 45440
    },
    {
      "epoch": 3.4149343339587244,
      "grad_norm": 0.43938690423965454,
      "learning_rate": 0.002828,
      "loss": 1.964,
      "step": 45504
    },
    {
      "epoch": 3.419737335834897,
      "grad_norm": 0.4162237346172333,
      "learning_rate": 0.002828,
      "loss": 1.9675,
      "step": 45568
    },
    {
      "epoch": 3.4245403377110692,
      "grad_norm": 0.37918907403945923,
      "learning_rate": 0.002828,
      "loss": 1.9639,
      "step": 45632
    },
    {
      "epoch": 3.429343339587242,
      "grad_norm": 0.3653721809387207,
      "learning_rate": 0.002828,
      "loss": 1.9695,
      "step": 45696
    },
    {
      "epoch": 3.4341463414634146,
      "grad_norm": 0.4152611494064331,
      "learning_rate": 0.002828,
      "loss": 1.9628,
      "step": 45760
    },
    {
      "epoch": 3.438949343339587,
      "grad_norm": 0.42145082354545593,
      "learning_rate": 0.002828,
      "loss": 1.9626,
      "step": 45824
    },
    {
      "epoch": 3.44375234521576,
      "grad_norm": 0.43907630443573,
      "learning_rate": 0.002828,
      "loss": 1.9604,
      "step": 45888
    },
    {
      "epoch": 3.4485553470919323,
      "grad_norm": 0.4232199788093567,
      "learning_rate": 0.002828,
      "loss": 1.9663,
      "step": 45952
    },
    {
      "epoch": 3.4533583489681052,
      "grad_norm": 0.4069931209087372,
      "learning_rate": 0.002828,
      "loss": 1.9663,
      "step": 46016
    },
    {
      "epoch": 3.4581613508442777,
      "grad_norm": 0.42427578568458557,
      "learning_rate": 0.002828,
      "loss": 1.9562,
      "step": 46080
    },
    {
      "epoch": 3.46296435272045,
      "grad_norm": 0.417915016412735,
      "learning_rate": 0.002828,
      "loss": 1.9594,
      "step": 46144
    },
    {
      "epoch": 3.467767354596623,
      "grad_norm": 0.46104422211647034,
      "learning_rate": 0.002828,
      "loss": 1.9634,
      "step": 46208
    },
    {
      "epoch": 3.4725703564727954,
      "grad_norm": 0.5542020201683044,
      "learning_rate": 0.002828,
      "loss": 1.9643,
      "step": 46272
    },
    {
      "epoch": 3.4773733583489683,
      "grad_norm": 0.5005798935890198,
      "learning_rate": 0.002828,
      "loss": 1.9575,
      "step": 46336
    },
    {
      "epoch": 3.4821763602251408,
      "grad_norm": 0.41317811608314514,
      "learning_rate": 0.002828,
      "loss": 1.9564,
      "step": 46400
    },
    {
      "epoch": 3.486979362101313,
      "grad_norm": 0.4168444275856018,
      "learning_rate": 0.002828,
      "loss": 1.9619,
      "step": 46464
    },
    {
      "epoch": 3.491782363977486,
      "grad_norm": 0.4567578434944153,
      "learning_rate": 0.002828,
      "loss": 1.9597,
      "step": 46528
    },
    {
      "epoch": 3.4965853658536585,
      "grad_norm": 0.4371911287307739,
      "learning_rate": 0.002828,
      "loss": 1.9596,
      "step": 46592
    },
    {
      "epoch": 3.501388367729831,
      "grad_norm": 0.38273531198501587,
      "learning_rate": 0.002828,
      "loss": 1.9554,
      "step": 46656
    },
    {
      "epoch": 3.506191369606004,
      "grad_norm": 0.4254724681377411,
      "learning_rate": 0.002828,
      "loss": 1.9556,
      "step": 46720
    },
    {
      "epoch": 3.5109943714821763,
      "grad_norm": 0.39332327246665955,
      "learning_rate": 0.002828,
      "loss": 1.9555,
      "step": 46784
    },
    {
      "epoch": 3.5157973733583487,
      "grad_norm": 0.3772275745868683,
      "learning_rate": 0.002828,
      "loss": 1.9573,
      "step": 46848
    },
    {
      "epoch": 3.5206003752345216,
      "grad_norm": 0.4225102663040161,
      "learning_rate": 0.002828,
      "loss": 1.9581,
      "step": 46912
    },
    {
      "epoch": 3.525403377110694,
      "grad_norm": 0.4391845762729645,
      "learning_rate": 0.002828,
      "loss": 1.958,
      "step": 46976
    },
    {
      "epoch": 3.530206378986867,
      "grad_norm": 0.43624913692474365,
      "learning_rate": 0.002828,
      "loss": 1.9596,
      "step": 47040
    },
    {
      "epoch": 3.5350093808630394,
      "grad_norm": 0.4952937364578247,
      "learning_rate": 0.002828,
      "loss": 1.9511,
      "step": 47104
    },
    {
      "epoch": 3.5398123827392123,
      "grad_norm": 0.4626399576663971,
      "learning_rate": 0.002828,
      "loss": 1.9537,
      "step": 47168
    },
    {
      "epoch": 3.5446153846153847,
      "grad_norm": 0.40251588821411133,
      "learning_rate": 0.002828,
      "loss": 1.9512,
      "step": 47232
    },
    {
      "epoch": 3.549418386491557,
      "grad_norm": 0.4524631202220917,
      "learning_rate": 0.002828,
      "loss": 1.953,
      "step": 47296
    },
    {
      "epoch": 3.55422138836773,
      "grad_norm": 0.3918924629688263,
      "learning_rate": 0.002828,
      "loss": 1.9534,
      "step": 47360
    },
    {
      "epoch": 3.5590243902439025,
      "grad_norm": 0.3780909478664398,
      "learning_rate": 0.002828,
      "loss": 1.9525,
      "step": 47424
    },
    {
      "epoch": 3.563827392120075,
      "grad_norm": 0.47011178731918335,
      "learning_rate": 0.002828,
      "loss": 1.9526,
      "step": 47488
    },
    {
      "epoch": 3.568630393996248,
      "grad_norm": 0.5357784628868103,
      "learning_rate": 0.002828,
      "loss": 1.9562,
      "step": 47552
    },
    {
      "epoch": 3.5734333958724203,
      "grad_norm": 0.3771846890449524,
      "learning_rate": 0.002828,
      "loss": 1.9556,
      "step": 47616
    },
    {
      "epoch": 3.5782363977485927,
      "grad_norm": 0.44459056854248047,
      "learning_rate": 0.002828,
      "loss": 1.9459,
      "step": 47680
    },
    {
      "epoch": 3.5830393996247656,
      "grad_norm": 0.45114779472351074,
      "learning_rate": 0.002828,
      "loss": 1.9509,
      "step": 47744
    },
    {
      "epoch": 3.587842401500938,
      "grad_norm": 0.49684053659439087,
      "learning_rate": 0.002828,
      "loss": 1.9499,
      "step": 47808
    },
    {
      "epoch": 3.5926454033771105,
      "grad_norm": 0.5553718209266663,
      "learning_rate": 0.002828,
      "loss": 1.9526,
      "step": 47872
    },
    {
      "epoch": 3.5974484052532834,
      "grad_norm": 0.4265674650669098,
      "learning_rate": 0.002828,
      "loss": 1.9544,
      "step": 47936
    },
    {
      "epoch": 3.602251407129456,
      "grad_norm": 0.401682585477829,
      "learning_rate": 0.002828,
      "loss": 1.9458,
      "step": 48000
    },
    {
      "epoch": 3.6070544090056287,
      "grad_norm": 0.4206481873989105,
      "learning_rate": 0.002828,
      "loss": 1.9523,
      "step": 48064
    },
    {
      "epoch": 3.611857410881801,
      "grad_norm": 0.40604767203330994,
      "learning_rate": 0.002828,
      "loss": 1.9511,
      "step": 48128
    },
    {
      "epoch": 3.616660412757974,
      "grad_norm": 0.43780919909477234,
      "learning_rate": 0.002828,
      "loss": 1.9409,
      "step": 48192
    },
    {
      "epoch": 3.6214634146341464,
      "grad_norm": 0.46581652760505676,
      "learning_rate": 0.002828,
      "loss": 1.9493,
      "step": 48256
    },
    {
      "epoch": 3.626266416510319,
      "grad_norm": 0.4305538237094879,
      "learning_rate": 0.002828,
      "loss": 1.9489,
      "step": 48320
    },
    {
      "epoch": 3.6310694183864918,
      "grad_norm": 0.4123769998550415,
      "learning_rate": 0.002828,
      "loss": 1.9512,
      "step": 48384
    },
    {
      "epoch": 3.635872420262664,
      "grad_norm": 0.41718190908432007,
      "learning_rate": 0.002828,
      "loss": 1.9449,
      "step": 48448
    },
    {
      "epoch": 3.6406754221388367,
      "grad_norm": 0.4468398988246918,
      "learning_rate": 0.002828,
      "loss": 1.9395,
      "step": 48512
    },
    {
      "epoch": 3.6454784240150095,
      "grad_norm": 0.47508302330970764,
      "learning_rate": 0.002828,
      "loss": 1.9455,
      "step": 48576
    },
    {
      "epoch": 3.650281425891182,
      "grad_norm": 0.4177429676055908,
      "learning_rate": 0.002828,
      "loss": 1.9431,
      "step": 48640
    },
    {
      "epoch": 3.6550844277673544,
      "grad_norm": 0.44048476219177246,
      "learning_rate": 0.002828,
      "loss": 1.9477,
      "step": 48704
    },
    {
      "epoch": 3.6598874296435273,
      "grad_norm": 0.41455012559890747,
      "learning_rate": 0.002828,
      "loss": 1.9436,
      "step": 48768
    },
    {
      "epoch": 3.6646904315196998,
      "grad_norm": 0.39578694105148315,
      "learning_rate": 0.002828,
      "loss": 1.9437,
      "step": 48832
    },
    {
      "epoch": 3.669493433395872,
      "grad_norm": 0.45000991225242615,
      "learning_rate": 0.002828,
      "loss": 1.9433,
      "step": 48896
    },
    {
      "epoch": 3.674296435272045,
      "grad_norm": 0.43240562081336975,
      "learning_rate": 0.002828,
      "loss": 1.9445,
      "step": 48960
    },
    {
      "epoch": 3.6790994371482175,
      "grad_norm": 0.4270322322845459,
      "learning_rate": 0.002828,
      "loss": 1.9489,
      "step": 49024
    },
    {
      "epoch": 3.6839024390243904,
      "grad_norm": 0.3854438066482544,
      "learning_rate": 0.002828,
      "loss": 1.9316,
      "step": 49088
    },
    {
      "epoch": 3.688705440900563,
      "grad_norm": 0.4212285280227661,
      "learning_rate": 0.002828,
      "loss": 1.941,
      "step": 49152
    },
    {
      "epoch": 3.6935084427767357,
      "grad_norm": 0.38058602809906006,
      "learning_rate": 0.002828,
      "loss": 1.948,
      "step": 49216
    },
    {
      "epoch": 3.698311444652908,
      "grad_norm": 0.4383929669857025,
      "learning_rate": 0.002828,
      "loss": 1.9357,
      "step": 49280
    },
    {
      "epoch": 3.7031144465290806,
      "grad_norm": 0.42744216322898865,
      "learning_rate": 0.002828,
      "loss": 1.9379,
      "step": 49344
    },
    {
      "epoch": 3.7079174484052535,
      "grad_norm": 0.43913623690605164,
      "learning_rate": 0.002828,
      "loss": 1.94,
      "step": 49408
    },
    {
      "epoch": 3.712720450281426,
      "grad_norm": 0.3816799521446228,
      "learning_rate": 0.002828,
      "loss": 1.9438,
      "step": 49472
    },
    {
      "epoch": 3.7175234521575984,
      "grad_norm": 0.4266246259212494,
      "learning_rate": 0.002828,
      "loss": 1.9346,
      "step": 49536
    },
    {
      "epoch": 3.7223264540337713,
      "grad_norm": 0.4960864186286926,
      "learning_rate": 0.002828,
      "loss": 1.9402,
      "step": 49600
    },
    {
      "epoch": 3.7271294559099437,
      "grad_norm": 0.4896080493927002,
      "learning_rate": 0.002828,
      "loss": 1.935,
      "step": 49664
    },
    {
      "epoch": 3.731932457786116,
      "grad_norm": 0.4197131395339966,
      "learning_rate": 0.002828,
      "loss": 1.9363,
      "step": 49728
    },
    {
      "epoch": 3.736735459662289,
      "grad_norm": 0.4813455045223236,
      "learning_rate": 0.002828,
      "loss": 1.9419,
      "step": 49792
    },
    {
      "epoch": 3.7415384615384615,
      "grad_norm": 0.4002421796321869,
      "learning_rate": 0.002828,
      "loss": 1.941,
      "step": 49856
    },
    {
      "epoch": 3.746341463414634,
      "grad_norm": 0.37192049622535706,
      "learning_rate": 0.002828,
      "loss": 1.9327,
      "step": 49920
    },
    {
      "epoch": 3.751144465290807,
      "grad_norm": 0.39583730697631836,
      "learning_rate": 0.002828,
      "loss": 1.9378,
      "step": 49984
    },
    {
      "epoch": 3.7559474671669792,
      "grad_norm": 0.43674999475479126,
      "learning_rate": 0.002828,
      "loss": 1.9378,
      "step": 50048
    },
    {
      "epoch": 3.760750469043152,
      "grad_norm": 0.5882846117019653,
      "learning_rate": 0.002828,
      "loss": 1.9322,
      "step": 50112
    },
    {
      "epoch": 3.7655534709193246,
      "grad_norm": 0.448281466960907,
      "learning_rate": 0.002828,
      "loss": 1.9353,
      "step": 50176
    },
    {
      "epoch": 3.7703564727954975,
      "grad_norm": 0.4727672338485718,
      "learning_rate": 0.002828,
      "loss": 1.9298,
      "step": 50240
    },
    {
      "epoch": 3.77515947467167,
      "grad_norm": 0.4881625175476074,
      "learning_rate": 0.002828,
      "loss": 1.9377,
      "step": 50304
    },
    {
      "epoch": 3.7799624765478423,
      "grad_norm": 0.5209890007972717,
      "learning_rate": 0.002828,
      "loss": 1.933,
      "step": 50368
    },
    {
      "epoch": 3.7847654784240152,
      "grad_norm": 0.38057464361190796,
      "learning_rate": 0.002828,
      "loss": 1.9348,
      "step": 50432
    },
    {
      "epoch": 3.7895684803001877,
      "grad_norm": 0.4636854827404022,
      "learning_rate": 0.002828,
      "loss": 1.9339,
      "step": 50496
    },
    {
      "epoch": 3.79437148217636,
      "grad_norm": 0.48334306478500366,
      "learning_rate": 0.002828,
      "loss": 1.9333,
      "step": 50560
    },
    {
      "epoch": 3.799174484052533,
      "grad_norm": 0.3957020938396454,
      "learning_rate": 0.002828,
      "loss": 1.9392,
      "step": 50624
    },
    {
      "epoch": 3.8039774859287054,
      "grad_norm": 0.3829587996006012,
      "learning_rate": 0.002828,
      "loss": 1.9245,
      "step": 50688
    },
    {
      "epoch": 3.808780487804878,
      "grad_norm": 0.38526973128318787,
      "learning_rate": 0.002828,
      "loss": 1.9269,
      "step": 50752
    },
    {
      "epoch": 3.8135834896810508,
      "grad_norm": 0.3830317258834839,
      "learning_rate": 0.002828,
      "loss": 1.9258,
      "step": 50816
    },
    {
      "epoch": 3.818386491557223,
      "grad_norm": 0.39473289251327515,
      "learning_rate": 0.002828,
      "loss": 1.9269,
      "step": 50880
    },
    {
      "epoch": 3.8231894934333956,
      "grad_norm": 0.4700523614883423,
      "learning_rate": 0.002828,
      "loss": 1.925,
      "step": 50944
    },
    {
      "epoch": 3.8279924953095685,
      "grad_norm": 0.42843738198280334,
      "learning_rate": 0.002828,
      "loss": 1.9293,
      "step": 51008
    },
    {
      "epoch": 3.832795497185741,
      "grad_norm": 0.47291210293769836,
      "learning_rate": 0.002828,
      "loss": 1.9304,
      "step": 51072
    },
    {
      "epoch": 3.837598499061914,
      "grad_norm": 0.46181681752204895,
      "learning_rate": 0.002828,
      "loss": 1.9339,
      "step": 51136
    },
    {
      "epoch": 3.8424015009380863,
      "grad_norm": 0.4250589609146118,
      "learning_rate": 0.002828,
      "loss": 1.9271,
      "step": 51200
    },
    {
      "epoch": 3.847204502814259,
      "grad_norm": 0.45093217492103577,
      "learning_rate": 0.002828,
      "loss": 1.9299,
      "step": 51264
    },
    {
      "epoch": 3.8520075046904316,
      "grad_norm": 0.40182143449783325,
      "learning_rate": 0.002828,
      "loss": 1.9321,
      "step": 51328
    },
    {
      "epoch": 3.856810506566604,
      "grad_norm": 0.43783003091812134,
      "learning_rate": 0.002828,
      "loss": 1.9267,
      "step": 51392
    },
    {
      "epoch": 3.861613508442777,
      "grad_norm": 0.3968484103679657,
      "learning_rate": 0.002828,
      "loss": 1.9245,
      "step": 51456
    },
    {
      "epoch": 3.8664165103189494,
      "grad_norm": 0.4145183563232422,
      "learning_rate": 0.002828,
      "loss": 1.9171,
      "step": 51520
    },
    {
      "epoch": 3.871219512195122,
      "grad_norm": 0.42234665155410767,
      "learning_rate": 0.002828,
      "loss": 1.9269,
      "step": 51584
    },
    {
      "epoch": 3.8760225140712947,
      "grad_norm": 0.414061963558197,
      "learning_rate": 0.002828,
      "loss": 1.931,
      "step": 51648
    },
    {
      "epoch": 3.880825515947467,
      "grad_norm": 0.36635687947273254,
      "learning_rate": 0.002828,
      "loss": 1.9259,
      "step": 51712
    },
    {
      "epoch": 3.8856285178236396,
      "grad_norm": 0.37826499342918396,
      "learning_rate": 0.002828,
      "loss": 1.9234,
      "step": 51776
    },
    {
      "epoch": 3.8904315196998125,
      "grad_norm": 0.39356598258018494,
      "learning_rate": 0.002828,
      "loss": 1.925,
      "step": 51840
    },
    {
      "epoch": 3.895234521575985,
      "grad_norm": 0.49679431319236755,
      "learning_rate": 0.002828,
      "loss": 1.9167,
      "step": 51904
    },
    {
      "epoch": 3.9000375234521574,
      "grad_norm": 0.4129243493080139,
      "learning_rate": 0.002828,
      "loss": 1.9151,
      "step": 51968
    },
    {
      "epoch": 3.9048405253283303,
      "grad_norm": 0.4656813442707062,
      "learning_rate": 0.002828,
      "loss": 1.9203,
      "step": 52032
    },
    {
      "epoch": 3.9096435272045027,
      "grad_norm": 0.4859287440776825,
      "learning_rate": 0.002828,
      "loss": 1.9345,
      "step": 52096
    },
    {
      "epoch": 3.9144465290806756,
      "grad_norm": 0.4260033965110779,
      "learning_rate": 0.002828,
      "loss": 1.9259,
      "step": 52160
    },
    {
      "epoch": 3.919249530956848,
      "grad_norm": 0.4332883059978485,
      "learning_rate": 0.002828,
      "loss": 1.9214,
      "step": 52224
    },
    {
      "epoch": 3.924052532833021,
      "grad_norm": 0.418282687664032,
      "learning_rate": 0.002828,
      "loss": 1.9226,
      "step": 52288
    },
    {
      "epoch": 3.9288555347091934,
      "grad_norm": 0.38305866718292236,
      "learning_rate": 0.002828,
      "loss": 1.918,
      "step": 52352
    },
    {
      "epoch": 3.933658536585366,
      "grad_norm": 0.4469663202762604,
      "learning_rate": 0.002828,
      "loss": 1.9224,
      "step": 52416
    },
    {
      "epoch": 3.9384615384615387,
      "grad_norm": 0.4243173599243164,
      "learning_rate": 0.002828,
      "loss": 1.9261,
      "step": 52480
    },
    {
      "epoch": 3.943264540337711,
      "grad_norm": 0.4377906918525696,
      "learning_rate": 0.002828,
      "loss": 1.9203,
      "step": 52544
    },
    {
      "epoch": 3.9480675422138836,
      "grad_norm": 0.4146629571914673,
      "learning_rate": 0.002828,
      "loss": 1.9228,
      "step": 52608
    },
    {
      "epoch": 3.9528705440900564,
      "grad_norm": 0.41136953234672546,
      "learning_rate": 0.002828,
      "loss": 1.9203,
      "step": 52672
    },
    {
      "epoch": 3.957673545966229,
      "grad_norm": 0.39297160506248474,
      "learning_rate": 0.002828,
      "loss": 1.9237,
      "step": 52736
    },
    {
      "epoch": 3.9624765478424013,
      "grad_norm": 0.39389610290527344,
      "learning_rate": 0.002828,
      "loss": 1.9248,
      "step": 52800
    },
    {
      "epoch": 3.967279549718574,
      "grad_norm": 0.45044317841529846,
      "learning_rate": 0.002828,
      "loss": 1.9186,
      "step": 52864
    },
    {
      "epoch": 3.9720825515947467,
      "grad_norm": 0.4556477665901184,
      "learning_rate": 0.002828,
      "loss": 1.9136,
      "step": 52928
    },
    {
      "epoch": 3.976885553470919,
      "grad_norm": 0.41699573397636414,
      "learning_rate": 0.002828,
      "loss": 1.916,
      "step": 52992
    },
    {
      "epoch": 3.981688555347092,
      "grad_norm": 0.45948272943496704,
      "learning_rate": 0.002828,
      "loss": 1.9189,
      "step": 53056
    },
    {
      "epoch": 3.9864915572232644,
      "grad_norm": 0.3985957205295563,
      "learning_rate": 0.002828,
      "loss": 1.9218,
      "step": 53120
    },
    {
      "epoch": 3.9912945590994373,
      "grad_norm": 0.3850821852684021,
      "learning_rate": 0.002828,
      "loss": 1.9249,
      "step": 53184
    },
    {
      "epoch": 3.9960975609756098,
      "grad_norm": 0.4339929521083832,
      "learning_rate": 0.002828,
      "loss": 1.9169,
      "step": 53248
    },
    {
      "epoch": 4.000900562851783,
      "grad_norm": 0.42089322209358215,
      "learning_rate": 0.002828,
      "loss": 1.9067,
      "step": 53312
    },
    {
      "epoch": 4.005703564727955,
      "grad_norm": 0.4298423230648041,
      "learning_rate": 0.002828,
      "loss": 1.8761,
      "step": 53376
    },
    {
      "epoch": 4.0105065666041275,
      "grad_norm": 0.46022525429725647,
      "learning_rate": 0.002828,
      "loss": 1.8828,
      "step": 53440
    },
    {
      "epoch": 4.0153095684803,
      "grad_norm": 0.39294660091400146,
      "learning_rate": 0.002828,
      "loss": 1.8795,
      "step": 53504
    },
    {
      "epoch": 4.020112570356472,
      "grad_norm": 0.4576168656349182,
      "learning_rate": 0.002828,
      "loss": 1.8856,
      "step": 53568
    },
    {
      "epoch": 4.024915572232645,
      "grad_norm": 0.47902917861938477,
      "learning_rate": 0.002828,
      "loss": 1.8745,
      "step": 53632
    },
    {
      "epoch": 4.029718574108818,
      "grad_norm": 0.4290916621685028,
      "learning_rate": 0.002828,
      "loss": 1.8673,
      "step": 53696
    },
    {
      "epoch": 4.034521575984991,
      "grad_norm": 0.508795976638794,
      "learning_rate": 0.002828,
      "loss": 1.8712,
      "step": 53760
    },
    {
      "epoch": 4.039324577861163,
      "grad_norm": 0.4904632866382599,
      "learning_rate": 0.002828,
      "loss": 1.8711,
      "step": 53824
    },
    {
      "epoch": 4.044127579737336,
      "grad_norm": 0.4192737936973572,
      "learning_rate": 0.002828,
      "loss": 1.8804,
      "step": 53888
    },
    {
      "epoch": 4.048930581613509,
      "grad_norm": 0.4394719898700714,
      "learning_rate": 0.002828,
      "loss": 1.8748,
      "step": 53952
    },
    {
      "epoch": 4.053733583489681,
      "grad_norm": 0.4148879051208496,
      "learning_rate": 0.002828,
      "loss": 1.8713,
      "step": 54016
    },
    {
      "epoch": 4.058536585365854,
      "grad_norm": 0.4630700349807739,
      "learning_rate": 0.002828,
      "loss": 1.8691,
      "step": 54080
    },
    {
      "epoch": 4.063339587242027,
      "grad_norm": 0.43519315123558044,
      "learning_rate": 0.002828,
      "loss": 1.8795,
      "step": 54144
    },
    {
      "epoch": 4.068142589118199,
      "grad_norm": 0.49164682626724243,
      "learning_rate": 0.002828,
      "loss": 1.8829,
      "step": 54208
    },
    {
      "epoch": 4.0729455909943715,
      "grad_norm": 0.4109393358230591,
      "learning_rate": 0.002828,
      "loss": 1.8735,
      "step": 54272
    },
    {
      "epoch": 4.077748592870544,
      "grad_norm": 0.5678685903549194,
      "learning_rate": 0.002828,
      "loss": 1.8681,
      "step": 54336
    },
    {
      "epoch": 4.082551594746716,
      "grad_norm": 0.461561918258667,
      "learning_rate": 0.002828,
      "loss": 1.8678,
      "step": 54400
    },
    {
      "epoch": 4.087354596622889,
      "grad_norm": 0.7234866619110107,
      "learning_rate": 0.002828,
      "loss": 1.8763,
      "step": 54464
    },
    {
      "epoch": 4.092157598499062,
      "grad_norm": 0.4788929522037506,
      "learning_rate": 0.002828,
      "loss": 1.8791,
      "step": 54528
    },
    {
      "epoch": 4.096960600375234,
      "grad_norm": 0.40153634548187256,
      "learning_rate": 0.002828,
      "loss": 1.8763,
      "step": 54592
    },
    {
      "epoch": 4.101763602251407,
      "grad_norm": 0.4675711691379547,
      "learning_rate": 0.002828,
      "loss": 1.8755,
      "step": 54656
    },
    {
      "epoch": 4.10656660412758,
      "grad_norm": 0.40922197699546814,
      "learning_rate": 0.002828,
      "loss": 1.8688,
      "step": 54720
    },
    {
      "epoch": 4.111369606003752,
      "grad_norm": 0.46092039346694946,
      "learning_rate": 0.002828,
      "loss": 1.8722,
      "step": 54784
    },
    {
      "epoch": 4.116172607879925,
      "grad_norm": 0.41091591119766235,
      "learning_rate": 0.002828,
      "loss": 1.8691,
      "step": 54848
    },
    {
      "epoch": 4.120975609756098,
      "grad_norm": 0.39091047644615173,
      "learning_rate": 0.002828,
      "loss": 1.8752,
      "step": 54912
    },
    {
      "epoch": 4.1257786116322706,
      "grad_norm": 0.379625141620636,
      "learning_rate": 0.002828,
      "loss": 1.8691,
      "step": 54976
    },
    {
      "epoch": 4.1305816135084426,
      "grad_norm": 0.44625696539878845,
      "learning_rate": 0.002828,
      "loss": 1.878,
      "step": 55040
    },
    {
      "epoch": 4.135384615384615,
      "grad_norm": 0.4220277965068817,
      "learning_rate": 0.002828,
      "loss": 1.8701,
      "step": 55104
    },
    {
      "epoch": 4.140187617260788,
      "grad_norm": 0.5710861682891846,
      "learning_rate": 0.002828,
      "loss": 1.874,
      "step": 55168
    },
    {
      "epoch": 4.14499061913696,
      "grad_norm": 0.39823463559150696,
      "learning_rate": 0.002828,
      "loss": 1.8695,
      "step": 55232
    },
    {
      "epoch": 4.149793621013133,
      "grad_norm": 0.38708966970443726,
      "learning_rate": 0.002828,
      "loss": 1.8693,
      "step": 55296
    },
    {
      "epoch": 4.154596622889306,
      "grad_norm": 0.4767194390296936,
      "learning_rate": 0.002828,
      "loss": 1.8725,
      "step": 55360
    },
    {
      "epoch": 4.159399624765478,
      "grad_norm": 0.42039304971694946,
      "learning_rate": 0.002828,
      "loss": 1.8705,
      "step": 55424
    },
    {
      "epoch": 4.164202626641651,
      "grad_norm": 0.4277998208999634,
      "learning_rate": 0.002828,
      "loss": 1.8793,
      "step": 55488
    },
    {
      "epoch": 4.169005628517824,
      "grad_norm": 0.41498157382011414,
      "learning_rate": 0.002828,
      "loss": 1.8763,
      "step": 55552
    },
    {
      "epoch": 4.173808630393996,
      "grad_norm": 0.6043418049812317,
      "learning_rate": 0.002828,
      "loss": 1.8711,
      "step": 55616
    },
    {
      "epoch": 4.178611632270169,
      "grad_norm": 0.4799351692199707,
      "learning_rate": 0.002828,
      "loss": 1.8747,
      "step": 55680
    },
    {
      "epoch": 4.183414634146342,
      "grad_norm": 0.40535420179367065,
      "learning_rate": 0.002828,
      "loss": 1.8722,
      "step": 55744
    },
    {
      "epoch": 4.1882176360225145,
      "grad_norm": 0.4478679597377777,
      "learning_rate": 0.002828,
      "loss": 1.8682,
      "step": 55808
    },
    {
      "epoch": 4.1930206378986865,
      "grad_norm": 0.4330839216709137,
      "learning_rate": 0.002828,
      "loss": 1.864,
      "step": 55872
    },
    {
      "epoch": 4.197823639774859,
      "grad_norm": 0.45498356223106384,
      "learning_rate": 0.002828,
      "loss": 1.8723,
      "step": 55936
    },
    {
      "epoch": 4.202626641651032,
      "grad_norm": 0.5293490290641785,
      "learning_rate": 0.002828,
      "loss": 1.8685,
      "step": 56000
    },
    {
      "epoch": 4.207429643527204,
      "grad_norm": 0.7796311974525452,
      "learning_rate": 0.002828,
      "loss": 1.8708,
      "step": 56064
    },
    {
      "epoch": 4.212232645403377,
      "grad_norm": 0.4882802963256836,
      "learning_rate": 0.002828,
      "loss": 1.8755,
      "step": 56128
    },
    {
      "epoch": 4.21703564727955,
      "grad_norm": 0.44128352403640747,
      "learning_rate": 0.002828,
      "loss": 1.8751,
      "step": 56192
    },
    {
      "epoch": 4.221838649155722,
      "grad_norm": 0.41939055919647217,
      "learning_rate": 0.002828,
      "loss": 1.8772,
      "step": 56256
    },
    {
      "epoch": 4.226641651031895,
      "grad_norm": 0.4133007228374481,
      "learning_rate": 0.002828,
      "loss": 1.8759,
      "step": 56320
    },
    {
      "epoch": 4.231444652908068,
      "grad_norm": 0.466492235660553,
      "learning_rate": 0.002828,
      "loss": 1.8758,
      "step": 56384
    },
    {
      "epoch": 4.23624765478424,
      "grad_norm": 0.43265125155448914,
      "learning_rate": 0.002828,
      "loss": 1.868,
      "step": 56448
    },
    {
      "epoch": 4.241050656660413,
      "grad_norm": 0.4278468191623688,
      "learning_rate": 0.002828,
      "loss": 1.8696,
      "step": 56512
    },
    {
      "epoch": 4.245853658536586,
      "grad_norm": 0.43418794870376587,
      "learning_rate": 0.002828,
      "loss": 1.8714,
      "step": 56576
    },
    {
      "epoch": 4.250656660412758,
      "grad_norm": 0.4910091459751129,
      "learning_rate": 0.002828,
      "loss": 1.8633,
      "step": 56640
    },
    {
      "epoch": 4.2554596622889305,
      "grad_norm": 0.7604688405990601,
      "learning_rate": 0.002828,
      "loss": 1.8744,
      "step": 56704
    },
    {
      "epoch": 4.260262664165103,
      "grad_norm": 0.44893890619277954,
      "learning_rate": 0.002828,
      "loss": 1.8705,
      "step": 56768
    },
    {
      "epoch": 4.265065666041275,
      "grad_norm": 0.4631795585155487,
      "learning_rate": 0.002828,
      "loss": 1.8638,
      "step": 56832
    },
    {
      "epoch": 4.269868667917448,
      "grad_norm": 0.4360625147819519,
      "learning_rate": 0.002828,
      "loss": 1.8621,
      "step": 56896
    },
    {
      "epoch": 4.274671669793621,
      "grad_norm": 0.40779635310173035,
      "learning_rate": 0.002828,
      "loss": 1.8536,
      "step": 56960
    },
    {
      "epoch": 4.279474671669794,
      "grad_norm": 0.4392165243625641,
      "learning_rate": 0.002828,
      "loss": 1.8534,
      "step": 57024
    },
    {
      "epoch": 4.284277673545966,
      "grad_norm": 0.49357831478118896,
      "learning_rate": 0.002828,
      "loss": 1.8607,
      "step": 57088
    },
    {
      "epoch": 4.289080675422139,
      "grad_norm": 0.4300871193408966,
      "learning_rate": 0.002828,
      "loss": 1.8587,
      "step": 57152
    },
    {
      "epoch": 4.293883677298312,
      "grad_norm": 0.4610048532485962,
      "learning_rate": 0.002828,
      "loss": 1.8677,
      "step": 57216
    },
    {
      "epoch": 4.298686679174484,
      "grad_norm": 0.4305998682975769,
      "learning_rate": 0.002828,
      "loss": 1.8646,
      "step": 57280
    },
    {
      "epoch": 4.303489681050657,
      "grad_norm": 0.4064194858074188,
      "learning_rate": 0.002828,
      "loss": 1.864,
      "step": 57344
    },
    {
      "epoch": 4.3082926829268295,
      "grad_norm": 0.3984262943267822,
      "learning_rate": 0.002828,
      "loss": 1.8621,
      "step": 57408
    },
    {
      "epoch": 4.3130956848030015,
      "grad_norm": 0.4508727192878723,
      "learning_rate": 0.002828,
      "loss": 1.8579,
      "step": 57472
    },
    {
      "epoch": 4.317898686679174,
      "grad_norm": 0.4348895251750946,
      "learning_rate": 0.002828,
      "loss": 1.8553,
      "step": 57536
    },
    {
      "epoch": 4.322701688555347,
      "grad_norm": 0.3835434317588806,
      "learning_rate": 0.002828,
      "loss": 1.8568,
      "step": 57600
    },
    {
      "epoch": 4.327504690431519,
      "grad_norm": 0.429548054933548,
      "learning_rate": 0.002828,
      "loss": 1.8619,
      "step": 57664
    },
    {
      "epoch": 4.332307692307692,
      "grad_norm": 0.43585842847824097,
      "learning_rate": 0.002828,
      "loss": 1.87,
      "step": 57728
    },
    {
      "epoch": 4.337110694183865,
      "grad_norm": 0.4420074224472046,
      "learning_rate": 0.002828,
      "loss": 1.858,
      "step": 57792
    },
    {
      "epoch": 4.341913696060038,
      "grad_norm": 0.4094654321670532,
      "learning_rate": 0.002828,
      "loss": 1.8586,
      "step": 57856
    },
    {
      "epoch": 4.34671669793621,
      "grad_norm": 0.40466663241386414,
      "learning_rate": 0.002828,
      "loss": 1.8472,
      "step": 57920
    },
    {
      "epoch": 4.351519699812383,
      "grad_norm": 0.4832182228565216,
      "learning_rate": 0.002828,
      "loss": 1.8575,
      "step": 57984
    },
    {
      "epoch": 4.356322701688556,
      "grad_norm": 0.4997383654117584,
      "learning_rate": 0.002828,
      "loss": 1.8655,
      "step": 58048
    },
    {
      "epoch": 4.361125703564728,
      "grad_norm": 0.4515133500099182,
      "learning_rate": 0.002828,
      "loss": 1.8639,
      "step": 58112
    },
    {
      "epoch": 4.365928705440901,
      "grad_norm": 0.4245718717575073,
      "learning_rate": 0.002828,
      "loss": 1.8611,
      "step": 58176
    },
    {
      "epoch": 4.3707317073170735,
      "grad_norm": 0.42741069197654724,
      "learning_rate": 0.002828,
      "loss": 1.8614,
      "step": 58240
    },
    {
      "epoch": 4.3755347091932455,
      "grad_norm": 0.49281513690948486,
      "learning_rate": 0.002828,
      "loss": 1.8527,
      "step": 58304
    },
    {
      "epoch": 4.380337711069418,
      "grad_norm": 0.4547692537307739,
      "learning_rate": 0.002828,
      "loss": 1.8618,
      "step": 58368
    },
    {
      "epoch": 4.385140712945591,
      "grad_norm": 0.4504334628582001,
      "learning_rate": 0.002828,
      "loss": 1.8611,
      "step": 58432
    },
    {
      "epoch": 4.389943714821763,
      "grad_norm": 0.39381441473960876,
      "learning_rate": 0.002828,
      "loss": 1.8503,
      "step": 58496
    },
    {
      "epoch": 4.394746716697936,
      "grad_norm": 0.60453861951828,
      "learning_rate": 0.002828,
      "loss": 1.8597,
      "step": 58560
    },
    {
      "epoch": 4.399549718574109,
      "grad_norm": 0.45758405327796936,
      "learning_rate": 0.002828,
      "loss": 1.8675,
      "step": 58624
    },
    {
      "epoch": 4.404352720450281,
      "grad_norm": 0.3902152478694916,
      "learning_rate": 0.002828,
      "loss": 1.8561,
      "step": 58688
    },
    {
      "epoch": 4.409155722326454,
      "grad_norm": 0.457916796207428,
      "learning_rate": 0.002828,
      "loss": 1.8589,
      "step": 58752
    },
    {
      "epoch": 4.413958724202627,
      "grad_norm": 0.424445241689682,
      "learning_rate": 0.002828,
      "loss": 1.8548,
      "step": 58816
    },
    {
      "epoch": 4.418761726078799,
      "grad_norm": 0.4457383453845978,
      "learning_rate": 0.002828,
      "loss": 1.8553,
      "step": 58880
    },
    {
      "epoch": 4.423564727954972,
      "grad_norm": 1.47513747215271,
      "learning_rate": 0.002828,
      "loss": 1.8599,
      "step": 58944
    },
    {
      "epoch": 4.428367729831145,
      "grad_norm": 0.40880438685417175,
      "learning_rate": 0.002828,
      "loss": 1.8554,
      "step": 59008
    },
    {
      "epoch": 4.4331707317073175,
      "grad_norm": 0.4844193756580353,
      "learning_rate": 0.002828,
      "loss": 1.8558,
      "step": 59072
    },
    {
      "epoch": 4.4379737335834895,
      "grad_norm": 0.3893410861492157,
      "learning_rate": 0.002828,
      "loss": 1.8615,
      "step": 59136
    },
    {
      "epoch": 4.442776735459662,
      "grad_norm": 0.43397217988967896,
      "learning_rate": 0.002828,
      "loss": 1.8617,
      "step": 59200
    },
    {
      "epoch": 4.447579737335835,
      "grad_norm": 0.40611037611961365,
      "learning_rate": 0.002828,
      "loss": 1.86,
      "step": 59264
    },
    {
      "epoch": 4.452382739212007,
      "grad_norm": 0.42537084221839905,
      "learning_rate": 0.002828,
      "loss": 1.8555,
      "step": 59328
    },
    {
      "epoch": 4.45718574108818,
      "grad_norm": 0.39853808283805847,
      "learning_rate": 0.002828,
      "loss": 1.8637,
      "step": 59392
    },
    {
      "epoch": 4.461988742964353,
      "grad_norm": 0.43848931789398193,
      "learning_rate": 0.002828,
      "loss": 1.8505,
      "step": 59456
    },
    {
      "epoch": 4.466791744840525,
      "grad_norm": 0.4788341522216797,
      "learning_rate": 0.002828,
      "loss": 1.8477,
      "step": 59520
    },
    {
      "epoch": 4.471594746716698,
      "grad_norm": 0.42831316590309143,
      "learning_rate": 0.002828,
      "loss": 1.8514,
      "step": 59584
    },
    {
      "epoch": 4.476397748592871,
      "grad_norm": 0.51185542345047,
      "learning_rate": 0.002828,
      "loss": 1.8575,
      "step": 59648
    },
    {
      "epoch": 4.481200750469043,
      "grad_norm": 0.4338521361351013,
      "learning_rate": 0.002828,
      "loss": 1.8546,
      "step": 59712
    },
    {
      "epoch": 4.486003752345216,
      "grad_norm": 0.40350961685180664,
      "learning_rate": 0.002828,
      "loss": 1.8528,
      "step": 59776
    },
    {
      "epoch": 4.4908067542213885,
      "grad_norm": 0.433065801858902,
      "learning_rate": 0.002828,
      "loss": 1.8479,
      "step": 59840
    },
    {
      "epoch": 4.495609756097561,
      "grad_norm": 0.44782695174217224,
      "learning_rate": 0.002828,
      "loss": 1.8568,
      "step": 59904
    },
    {
      "epoch": 4.500412757973733,
      "grad_norm": 0.4756898283958435,
      "learning_rate": 0.002828,
      "loss": 1.8544,
      "step": 59968
    },
    {
      "epoch": 4.505215759849906,
      "grad_norm": 0.3805321455001831,
      "learning_rate": 0.002828,
      "loss": 1.8506,
      "step": 60032
    },
    {
      "epoch": 4.510018761726079,
      "grad_norm": 0.4259619414806366,
      "learning_rate": 0.002828,
      "loss": 1.8482,
      "step": 60096
    },
    {
      "epoch": 4.514821763602251,
      "grad_norm": 0.4335617125034332,
      "learning_rate": 0.002828,
      "loss": 1.8547,
      "step": 60160
    },
    {
      "epoch": 4.519624765478424,
      "grad_norm": 0.39122387766838074,
      "learning_rate": 0.002828,
      "loss": 1.8506,
      "step": 60224
    },
    {
      "epoch": 4.524427767354597,
      "grad_norm": 0.4266170859336853,
      "learning_rate": 0.002828,
      "loss": 1.8502,
      "step": 60288
    },
    {
      "epoch": 4.529230769230769,
      "grad_norm": 0.439899206161499,
      "learning_rate": 0.002828,
      "loss": 1.8572,
      "step": 60352
    },
    {
      "epoch": 4.534033771106942,
      "grad_norm": 0.4882529675960541,
      "learning_rate": 0.002828,
      "loss": 1.8493,
      "step": 60416
    },
    {
      "epoch": 4.538836772983115,
      "grad_norm": 0.44590744376182556,
      "learning_rate": 0.002828,
      "loss": 1.8406,
      "step": 60480
    },
    {
      "epoch": 4.543639774859287,
      "grad_norm": 0.4067928194999695,
      "learning_rate": 0.002828,
      "loss": 1.851,
      "step": 60544
    },
    {
      "epoch": 4.54844277673546,
      "grad_norm": 0.4017028510570526,
      "learning_rate": 0.002828,
      "loss": 1.8473,
      "step": 60608
    },
    {
      "epoch": 4.5532457786116325,
      "grad_norm": 0.42601388692855835,
      "learning_rate": 0.002828,
      "loss": 1.8501,
      "step": 60672
    },
    {
      "epoch": 4.558048780487805,
      "grad_norm": 0.43437841534614563,
      "learning_rate": 0.002828,
      "loss": 1.847,
      "step": 60736
    },
    {
      "epoch": 4.562851782363977,
      "grad_norm": 0.42863306403160095,
      "learning_rate": 0.002828,
      "loss": 1.8462,
      "step": 60800
    },
    {
      "epoch": 4.56765478424015,
      "grad_norm": 0.43381252884864807,
      "learning_rate": 0.002828,
      "loss": 1.8448,
      "step": 60864
    },
    {
      "epoch": 4.572457786116322,
      "grad_norm": 0.46240073442459106,
      "learning_rate": 0.002828,
      "loss": 1.852,
      "step": 60928
    },
    {
      "epoch": 4.577260787992495,
      "grad_norm": 0.4556284546852112,
      "learning_rate": 0.002828,
      "loss": 1.8522,
      "step": 60992
    },
    {
      "epoch": 4.582063789868668,
      "grad_norm": 0.41591572761535645,
      "learning_rate": 0.002828,
      "loss": 1.8448,
      "step": 61056
    },
    {
      "epoch": 4.586866791744841,
      "grad_norm": 0.44293197989463806,
      "learning_rate": 0.002828,
      "loss": 1.8434,
      "step": 61120
    },
    {
      "epoch": 4.591669793621013,
      "grad_norm": 0.42343223094940186,
      "learning_rate": 0.002828,
      "loss": 1.8484,
      "step": 61184
    },
    {
      "epoch": 4.596472795497186,
      "grad_norm": 0.3981548547744751,
      "learning_rate": 0.002828,
      "loss": 1.8507,
      "step": 61248
    },
    {
      "epoch": 4.601275797373359,
      "grad_norm": 0.47165608406066895,
      "learning_rate": 0.002828,
      "loss": 1.8514,
      "step": 61312
    },
    {
      "epoch": 4.606078799249531,
      "grad_norm": 0.4086002707481384,
      "learning_rate": 0.002828,
      "loss": 1.8517,
      "step": 61376
    },
    {
      "epoch": 4.610881801125704,
      "grad_norm": 0.5088809728622437,
      "learning_rate": 0.002828,
      "loss": 1.8492,
      "step": 61440
    },
    {
      "epoch": 4.6156848030018764,
      "grad_norm": 0.5022900104522705,
      "learning_rate": 0.002828,
      "loss": 1.8445,
      "step": 61504
    },
    {
      "epoch": 4.620487804878048,
      "grad_norm": 0.4150559604167938,
      "learning_rate": 0.002828,
      "loss": 1.8515,
      "step": 61568
    },
    {
      "epoch": 4.625290806754221,
      "grad_norm": 0.4678759276866913,
      "learning_rate": 0.002828,
      "loss": 1.8392,
      "step": 61632
    },
    {
      "epoch": 4.630093808630394,
      "grad_norm": 0.5109483003616333,
      "learning_rate": 0.002828,
      "loss": 1.8465,
      "step": 61696
    },
    {
      "epoch": 4.634896810506566,
      "grad_norm": 0.4642334580421448,
      "learning_rate": 0.002828,
      "loss": 1.8472,
      "step": 61760
    },
    {
      "epoch": 4.639699812382739,
      "grad_norm": 0.49772489070892334,
      "learning_rate": 0.002828,
      "loss": 1.8409,
      "step": 61824
    },
    {
      "epoch": 4.644502814258912,
      "grad_norm": 0.3974035680294037,
      "learning_rate": 0.002828,
      "loss": 1.8481,
      "step": 61888
    },
    {
      "epoch": 4.649305816135085,
      "grad_norm": 0.5001426935195923,
      "learning_rate": 0.002828,
      "loss": 1.843,
      "step": 61952
    },
    {
      "epoch": 4.654108818011257,
      "grad_norm": 0.44995561242103577,
      "learning_rate": 0.002828,
      "loss": 1.8444,
      "step": 62016
    },
    {
      "epoch": 4.65891181988743,
      "grad_norm": 0.42114222049713135,
      "learning_rate": 0.002828,
      "loss": 1.8455,
      "step": 62080
    },
    {
      "epoch": 4.663714821763603,
      "grad_norm": 0.4493741989135742,
      "learning_rate": 0.002828,
      "loss": 1.8366,
      "step": 62144
    },
    {
      "epoch": 4.668517823639775,
      "grad_norm": 0.48146486282348633,
      "learning_rate": 0.002828,
      "loss": 1.8477,
      "step": 62208
    },
    {
      "epoch": 4.6733208255159475,
      "grad_norm": 0.40148022770881653,
      "learning_rate": 0.002828,
      "loss": 1.8408,
      "step": 62272
    },
    {
      "epoch": 4.67812382739212,
      "grad_norm": 0.4373031556606293,
      "learning_rate": 0.002828,
      "loss": 1.8406,
      "step": 62336
    },
    {
      "epoch": 4.682926829268292,
      "grad_norm": 0.4787302315235138,
      "learning_rate": 0.002828,
      "loss": 1.8398,
      "step": 62400
    },
    {
      "epoch": 4.687729831144465,
      "grad_norm": 0.5098510384559631,
      "learning_rate": 0.002828,
      "loss": 1.847,
      "step": 62464
    },
    {
      "epoch": 4.692532833020638,
      "grad_norm": 0.5322007536888123,
      "learning_rate": 0.002828,
      "loss": 1.8393,
      "step": 62528
    },
    {
      "epoch": 4.69733583489681,
      "grad_norm": 0.4712188243865967,
      "learning_rate": 0.002828,
      "loss": 1.84,
      "step": 62592
    },
    {
      "epoch": 4.702138836772983,
      "grad_norm": 0.42479485273361206,
      "learning_rate": 0.002828,
      "loss": 1.8411,
      "step": 62656
    },
    {
      "epoch": 4.706941838649156,
      "grad_norm": 0.47006043791770935,
      "learning_rate": 0.002828,
      "loss": 1.8337,
      "step": 62720
    },
    {
      "epoch": 4.711744840525329,
      "grad_norm": 0.46571770310401917,
      "learning_rate": 0.002828,
      "loss": 1.8417,
      "step": 62784
    },
    {
      "epoch": 4.716547842401501,
      "grad_norm": 0.4172969162464142,
      "learning_rate": 0.002828,
      "loss": 1.838,
      "step": 62848
    },
    {
      "epoch": 4.721350844277674,
      "grad_norm": 0.4751930236816406,
      "learning_rate": 0.002828,
      "loss": 1.8304,
      "step": 62912
    },
    {
      "epoch": 4.726153846153846,
      "grad_norm": 0.42527517676353455,
      "learning_rate": 0.002828,
      "loss": 1.8387,
      "step": 62976
    },
    {
      "epoch": 4.730956848030019,
      "grad_norm": 0.43444591760635376,
      "learning_rate": 0.002828,
      "loss": 1.834,
      "step": 63040
    },
    {
      "epoch": 4.7357598499061915,
      "grad_norm": 0.4354824125766754,
      "learning_rate": 0.002828,
      "loss": 1.8398,
      "step": 63104
    },
    {
      "epoch": 4.740562851782364,
      "grad_norm": 0.4284718334674835,
      "learning_rate": 0.002828,
      "loss": 1.8385,
      "step": 63168
    },
    {
      "epoch": 4.745365853658536,
      "grad_norm": 0.5047585964202881,
      "learning_rate": 0.002828,
      "loss": 1.8361,
      "step": 63232
    },
    {
      "epoch": 4.750168855534709,
      "grad_norm": 0.41910332441329956,
      "learning_rate": 0.002828,
      "loss": 1.842,
      "step": 63296
    },
    {
      "epoch": 4.754971857410882,
      "grad_norm": 0.41034504771232605,
      "learning_rate": 0.002828,
      "loss": 1.8323,
      "step": 63360
    },
    {
      "epoch": 4.759774859287054,
      "grad_norm": 0.389961302280426,
      "learning_rate": 0.002828,
      "loss": 1.8378,
      "step": 63424
    },
    {
      "epoch": 4.764577861163227,
      "grad_norm": 0.4235137701034546,
      "learning_rate": 0.002828,
      "loss": 1.8403,
      "step": 63488
    },
    {
      "epoch": 4.7693808630394,
      "grad_norm": 0.4341602623462677,
      "learning_rate": 0.002828,
      "loss": 1.8316,
      "step": 63552
    },
    {
      "epoch": 4.774183864915572,
      "grad_norm": 0.42686939239501953,
      "learning_rate": 0.002828,
      "loss": 1.8335,
      "step": 63616
    },
    {
      "epoch": 4.778986866791745,
      "grad_norm": 0.41930991411209106,
      "learning_rate": 0.002828,
      "loss": 1.8312,
      "step": 63680
    },
    {
      "epoch": 4.783789868667918,
      "grad_norm": 0.4320056736469269,
      "learning_rate": 0.002828,
      "loss": 1.8255,
      "step": 63744
    },
    {
      "epoch": 4.78859287054409,
      "grad_norm": 0.4222176671028137,
      "learning_rate": 0.002828,
      "loss": 1.8367,
      "step": 63808
    },
    {
      "epoch": 4.7933958724202625,
      "grad_norm": 0.39090004563331604,
      "learning_rate": 0.002828,
      "loss": 1.836,
      "step": 63872
    },
    {
      "epoch": 4.798198874296435,
      "grad_norm": 0.5152971148490906,
      "learning_rate": 0.002828,
      "loss": 1.8315,
      "step": 63936
    },
    {
      "epoch": 4.803001876172608,
      "grad_norm": 0.3895571231842041,
      "learning_rate": 0.002828,
      "loss": 1.8388,
      "step": 64000
    },
    {
      "epoch": 4.80780487804878,
      "grad_norm": 0.4033324420452118,
      "learning_rate": 0.002828,
      "loss": 1.8329,
      "step": 64064
    },
    {
      "epoch": 4.812607879924953,
      "grad_norm": 0.43192601203918457,
      "learning_rate": 0.002828,
      "loss": 1.8303,
      "step": 64128
    },
    {
      "epoch": 4.817410881801126,
      "grad_norm": 0.4123119115829468,
      "learning_rate": 0.002828,
      "loss": 1.8365,
      "step": 64192
    },
    {
      "epoch": 4.822213883677298,
      "grad_norm": 0.394336998462677,
      "learning_rate": 0.002828,
      "loss": 1.8332,
      "step": 64256
    },
    {
      "epoch": 4.827016885553471,
      "grad_norm": 0.4631196856498718,
      "learning_rate": 0.002828,
      "loss": 1.8314,
      "step": 64320
    },
    {
      "epoch": 4.831819887429644,
      "grad_norm": 0.4436214566230774,
      "learning_rate": 0.002828,
      "loss": 1.8335,
      "step": 64384
    },
    {
      "epoch": 4.836622889305816,
      "grad_norm": 0.4646340608596802,
      "learning_rate": 0.002828,
      "loss": 1.8257,
      "step": 64448
    },
    {
      "epoch": 4.841425891181989,
      "grad_norm": 0.488765686750412,
      "learning_rate": 0.002828,
      "loss": 1.8328,
      "step": 64512
    },
    {
      "epoch": 4.846228893058162,
      "grad_norm": 0.5807002186775208,
      "learning_rate": 0.002828,
      "loss": 1.8302,
      "step": 64576
    },
    {
      "epoch": 4.851031894934334,
      "grad_norm": 0.41557422280311584,
      "learning_rate": 0.002828,
      "loss": 1.8274,
      "step": 64640
    },
    {
      "epoch": 4.8558348968105065,
      "grad_norm": 0.40899091958999634,
      "learning_rate": 0.002828,
      "loss": 1.8294,
      "step": 64704
    },
    {
      "epoch": 4.860637898686679,
      "grad_norm": 0.41941386461257935,
      "learning_rate": 0.002828,
      "loss": 1.8314,
      "step": 64768
    },
    {
      "epoch": 4.865440900562852,
      "grad_norm": 0.4324967861175537,
      "learning_rate": 0.002828,
      "loss": 1.8242,
      "step": 64832
    },
    {
      "epoch": 4.870243902439024,
      "grad_norm": 0.5122097730636597,
      "learning_rate": 0.002828,
      "loss": 1.8284,
      "step": 64896
    },
    {
      "epoch": 4.875046904315197,
      "grad_norm": 0.4794155955314636,
      "learning_rate": 0.002828,
      "loss": 1.8306,
      "step": 64960
    },
    {
      "epoch": 4.879849906191369,
      "grad_norm": 0.41596829891204834,
      "learning_rate": 0.002828,
      "loss": 1.8332,
      "step": 65024
    },
    {
      "epoch": 4.884652908067542,
      "grad_norm": 0.47452154755592346,
      "learning_rate": 0.002828,
      "loss": 1.8215,
      "step": 65088
    },
    {
      "epoch": 4.889455909943715,
      "grad_norm": 0.46592849493026733,
      "learning_rate": 0.002828,
      "loss": 1.8254,
      "step": 65152
    },
    {
      "epoch": 4.894258911819888,
      "grad_norm": 0.400258868932724,
      "learning_rate": 0.002828,
      "loss": 1.832,
      "step": 65216
    },
    {
      "epoch": 4.89906191369606,
      "grad_norm": 0.4501512050628662,
      "learning_rate": 0.002828,
      "loss": 1.8272,
      "step": 65280
    },
    {
      "epoch": 4.903864915572233,
      "grad_norm": 0.5389344096183777,
      "learning_rate": 0.002828,
      "loss": 1.8258,
      "step": 65344
    },
    {
      "epoch": 4.908667917448406,
      "grad_norm": 0.4736655354499817,
      "learning_rate": 0.002828,
      "loss": 1.8277,
      "step": 65408
    },
    {
      "epoch": 4.913470919324578,
      "grad_norm": 0.42256057262420654,
      "learning_rate": 0.002828,
      "loss": 1.8259,
      "step": 65472
    },
    {
      "epoch": 4.9182739212007505,
      "grad_norm": 0.45916080474853516,
      "learning_rate": 0.002828,
      "loss": 1.8317,
      "step": 65536
    },
    {
      "epoch": 4.923076923076923,
      "grad_norm": 0.3935365080833435,
      "learning_rate": 0.002828,
      "loss": 1.8253,
      "step": 65600
    },
    {
      "epoch": 4.927879924953095,
      "grad_norm": 0.4702571928501129,
      "learning_rate": 0.002828,
      "loss": 1.8194,
      "step": 65664
    },
    {
      "epoch": 4.932682926829268,
      "grad_norm": 0.4381246864795685,
      "learning_rate": 0.002828,
      "loss": 1.822,
      "step": 65728
    },
    {
      "epoch": 4.937485928705441,
      "grad_norm": 0.4881627857685089,
      "learning_rate": 0.002828,
      "loss": 1.8266,
      "step": 65792
    },
    {
      "epoch": 4.942288930581613,
      "grad_norm": 0.5015894770622253,
      "learning_rate": 0.002828,
      "loss": 1.8221,
      "step": 65856
    },
    {
      "epoch": 4.947091932457786,
      "grad_norm": 0.4536570608615875,
      "learning_rate": 0.002828,
      "loss": 1.8246,
      "step": 65920
    },
    {
      "epoch": 4.951894934333959,
      "grad_norm": 0.43508243560791016,
      "learning_rate": 0.002828,
      "loss": 1.8188,
      "step": 65984
    },
    {
      "epoch": 4.956697936210132,
      "grad_norm": 0.42308923602104187,
      "learning_rate": 0.002828,
      "loss": 1.8306,
      "step": 66048
    },
    {
      "epoch": 4.961500938086304,
      "grad_norm": 0.44210103154182434,
      "learning_rate": 0.002828,
      "loss": 1.8244,
      "step": 66112
    },
    {
      "epoch": 4.966303939962477,
      "grad_norm": 0.47712111473083496,
      "learning_rate": 0.002828,
      "loss": 1.8265,
      "step": 66176
    },
    {
      "epoch": 4.9711069418386495,
      "grad_norm": 0.4795370399951935,
      "learning_rate": 0.002828,
      "loss": 1.825,
      "step": 66240
    },
    {
      "epoch": 4.9759099437148215,
      "grad_norm": 0.3904446065425873,
      "learning_rate": 0.002828,
      "loss": 1.8251,
      "step": 66304
    },
    {
      "epoch": 4.980712945590994,
      "grad_norm": 0.4810478389263153,
      "learning_rate": 0.002828,
      "loss": 1.8258,
      "step": 66368
    },
    {
      "epoch": 4.985515947467167,
      "grad_norm": 0.39784982800483704,
      "learning_rate": 0.002828,
      "loss": 1.8232,
      "step": 66432
    },
    {
      "epoch": 4.990318949343339,
      "grad_norm": 0.4552132189273834,
      "learning_rate": 0.002828,
      "loss": 1.8203,
      "step": 66496
    },
    {
      "epoch": 4.995121951219512,
      "grad_norm": 0.44738858938217163,
      "learning_rate": 0.002828,
      "loss": 1.8241,
      "step": 66560
    },
    {
      "epoch": 4.999924953095685,
      "grad_norm": 0.41392263770103455,
      "learning_rate": 0.002828,
      "loss": 1.8214,
      "step": 66624
    },
    {
      "epoch": 5.004727954971857,
      "grad_norm": 0.4698507487773895,
      "learning_rate": 0.002828,
      "loss": 1.7802,
      "step": 66688
    },
    {
      "epoch": 5.00953095684803,
      "grad_norm": 0.45044583082199097,
      "learning_rate": 0.002828,
      "loss": 1.779,
      "step": 66752
    },
    {
      "epoch": 5.014333958724203,
      "grad_norm": 0.4659009277820587,
      "learning_rate": 0.002828,
      "loss": 1.7693,
      "step": 66816
    },
    {
      "epoch": 5.019136960600375,
      "grad_norm": 0.45445555448532104,
      "learning_rate": 0.002828,
      "loss": 1.7778,
      "step": 66880
    },
    {
      "epoch": 5.023939962476548,
      "grad_norm": 0.5755953192710876,
      "learning_rate": 0.002828,
      "loss": 1.783,
      "step": 66944
    },
    {
      "epoch": 5.028742964352721,
      "grad_norm": 0.46172744035720825,
      "learning_rate": 0.002828,
      "loss": 1.7779,
      "step": 67008
    },
    {
      "epoch": 5.0335459662288935,
      "grad_norm": 0.42822253704071045,
      "learning_rate": 0.002828,
      "loss": 1.7759,
      "step": 67072
    },
    {
      "epoch": 5.0383489681050655,
      "grad_norm": 0.40689054131507874,
      "learning_rate": 0.002828,
      "loss": 1.7818,
      "step": 67136
    },
    {
      "epoch": 5.043151969981238,
      "grad_norm": 0.4451557993888855,
      "learning_rate": 0.002828,
      "loss": 1.7802,
      "step": 67200
    },
    {
      "epoch": 5.047954971857411,
      "grad_norm": 0.4525567889213562,
      "learning_rate": 0.002828,
      "loss": 1.7752,
      "step": 67264
    },
    {
      "epoch": 5.052757973733583,
      "grad_norm": 0.40108218789100647,
      "learning_rate": 0.002828,
      "loss": 1.7732,
      "step": 67328
    },
    {
      "epoch": 5.057560975609756,
      "grad_norm": 0.42633527517318726,
      "learning_rate": 0.002828,
      "loss": 1.7771,
      "step": 67392
    },
    {
      "epoch": 5.062363977485929,
      "grad_norm": 0.46593084931373596,
      "learning_rate": 0.002828,
      "loss": 1.7815,
      "step": 67456
    },
    {
      "epoch": 5.067166979362101,
      "grad_norm": 0.5002214312553406,
      "learning_rate": 0.002828,
      "loss": 1.7734,
      "step": 67520
    },
    {
      "epoch": 5.071969981238274,
      "grad_norm": 0.5098074674606323,
      "learning_rate": 0.002828,
      "loss": 1.7714,
      "step": 67584
    },
    {
      "epoch": 5.076772983114447,
      "grad_norm": 0.5381320714950562,
      "learning_rate": 0.002828,
      "loss": 1.7808,
      "step": 67648
    },
    {
      "epoch": 5.081575984990619,
      "grad_norm": 0.43003377318382263,
      "learning_rate": 0.002828,
      "loss": 1.7773,
      "step": 67712
    },
    {
      "epoch": 5.086378986866792,
      "grad_norm": 0.41460487246513367,
      "learning_rate": 0.002828,
      "loss": 1.7736,
      "step": 67776
    },
    {
      "epoch": 5.091181988742965,
      "grad_norm": 0.42502009868621826,
      "learning_rate": 0.002828,
      "loss": 1.7799,
      "step": 67840
    },
    {
      "epoch": 5.095984990619137,
      "grad_norm": 0.40019145607948303,
      "learning_rate": 0.002828,
      "loss": 1.7843,
      "step": 67904
    },
    {
      "epoch": 5.1007879924953095,
      "grad_norm": 0.4934079349040985,
      "learning_rate": 0.002828,
      "loss": 1.7727,
      "step": 67968
    },
    {
      "epoch": 5.105590994371482,
      "grad_norm": 0.42615658044815063,
      "learning_rate": 0.002828,
      "loss": 1.7815,
      "step": 68032
    },
    {
      "epoch": 5.110393996247655,
      "grad_norm": 0.4133114814758301,
      "learning_rate": 0.002828,
      "loss": 1.7799,
      "step": 68096
    },
    {
      "epoch": 5.115196998123827,
      "grad_norm": 0.4989236891269684,
      "learning_rate": 0.002828,
      "loss": 1.7834,
      "step": 68160
    },
    {
      "epoch": 5.12,
      "grad_norm": 0.39801713824272156,
      "learning_rate": 0.002828,
      "loss": 1.7713,
      "step": 68224
    },
    {
      "epoch": 5.124803001876173,
      "grad_norm": 0.47219210863113403,
      "learning_rate": 0.002828,
      "loss": 1.7784,
      "step": 68288
    },
    {
      "epoch": 5.129606003752345,
      "grad_norm": 0.4931895434856415,
      "learning_rate": 0.002828,
      "loss": 1.775,
      "step": 68352
    },
    {
      "epoch": 5.134409005628518,
      "grad_norm": 0.4328326880931854,
      "learning_rate": 0.002828,
      "loss": 1.7913,
      "step": 68416
    },
    {
      "epoch": 5.139212007504691,
      "grad_norm": 0.5637125372886658,
      "learning_rate": 0.002828,
      "loss": 1.7767,
      "step": 68480
    },
    {
      "epoch": 5.144015009380863,
      "grad_norm": 0.4656321704387665,
      "learning_rate": 0.002828,
      "loss": 1.7729,
      "step": 68544
    },
    {
      "epoch": 5.148818011257036,
      "grad_norm": 0.46301645040512085,
      "learning_rate": 0.002828,
      "loss": 1.7714,
      "step": 68608
    },
    {
      "epoch": 5.1536210131332085,
      "grad_norm": 0.407530277967453,
      "learning_rate": 0.002828,
      "loss": 1.7778,
      "step": 68672
    },
    {
      "epoch": 5.1584240150093805,
      "grad_norm": 0.48896515369415283,
      "learning_rate": 0.002828,
      "loss": 1.773,
      "step": 68736
    },
    {
      "epoch": 5.163227016885553,
      "grad_norm": 0.5092155933380127,
      "learning_rate": 0.002828,
      "loss": 1.7818,
      "step": 68800
    },
    {
      "epoch": 5.168030018761726,
      "grad_norm": 0.40807560086250305,
      "learning_rate": 0.002828,
      "loss": 1.7747,
      "step": 68864
    },
    {
      "epoch": 5.172833020637898,
      "grad_norm": 0.4265471398830414,
      "learning_rate": 0.002828,
      "loss": 1.7801,
      "step": 68928
    },
    {
      "epoch": 5.177636022514071,
      "grad_norm": 0.4193212389945984,
      "learning_rate": 0.002828,
      "loss": 1.7779,
      "step": 68992
    },
    {
      "epoch": 5.182439024390244,
      "grad_norm": 0.4063675105571747,
      "learning_rate": 0.002828,
      "loss": 1.7769,
      "step": 69056
    },
    {
      "epoch": 5.187242026266416,
      "grad_norm": 0.4742323160171509,
      "learning_rate": 0.002828,
      "loss": 1.7748,
      "step": 69120
    },
    {
      "epoch": 5.192045028142589,
      "grad_norm": 0.4574708342552185,
      "learning_rate": 0.002828,
      "loss": 1.7728,
      "step": 69184
    },
    {
      "epoch": 5.196848030018762,
      "grad_norm": 0.3898203670978546,
      "learning_rate": 0.002828,
      "loss": 1.7788,
      "step": 69248
    },
    {
      "epoch": 5.201651031894935,
      "grad_norm": 0.48600926995277405,
      "learning_rate": 0.002828,
      "loss": 1.7767,
      "step": 69312
    },
    {
      "epoch": 5.206454033771107,
      "grad_norm": 0.5444703102111816,
      "learning_rate": 0.002828,
      "loss": 1.7778,
      "step": 69376
    },
    {
      "epoch": 5.21125703564728,
      "grad_norm": 0.54710453748703,
      "learning_rate": 0.002828,
      "loss": 1.7764,
      "step": 69440
    },
    {
      "epoch": 5.2160600375234525,
      "grad_norm": 0.5190069675445557,
      "learning_rate": 0.002828,
      "loss": 1.7731,
      "step": 69504
    },
    {
      "epoch": 5.2208630393996245,
      "grad_norm": 0.5375124216079712,
      "learning_rate": 0.002828,
      "loss": 1.7779,
      "step": 69568
    },
    {
      "epoch": 5.225666041275797,
      "grad_norm": 0.4914718568325043,
      "learning_rate": 0.002828,
      "loss": 1.7735,
      "step": 69632
    },
    {
      "epoch": 5.23046904315197,
      "grad_norm": 0.44196775555610657,
      "learning_rate": 0.002828,
      "loss": 1.7765,
      "step": 69696
    },
    {
      "epoch": 5.235272045028142,
      "grad_norm": 0.43454575538635254,
      "learning_rate": 0.002828,
      "loss": 1.7814,
      "step": 69760
    },
    {
      "epoch": 5.240075046904315,
      "grad_norm": 0.5103824138641357,
      "learning_rate": 0.002828,
      "loss": 1.779,
      "step": 69824
    },
    {
      "epoch": 5.244878048780488,
      "grad_norm": 0.396705687046051,
      "learning_rate": 0.002828,
      "loss": 1.7799,
      "step": 69888
    },
    {
      "epoch": 5.24968105065666,
      "grad_norm": 0.4350598156452179,
      "learning_rate": 0.002828,
      "loss": 1.7798,
      "step": 69952
    },
    {
      "epoch": 5.254484052532833,
      "grad_norm": 0.4352000653743744,
      "learning_rate": 0.002828,
      "loss": 1.7764,
      "step": 70016
    },
    {
      "epoch": 5.259287054409006,
      "grad_norm": 0.43230265378952026,
      "learning_rate": 0.002828,
      "loss": 1.7754,
      "step": 70080
    },
    {
      "epoch": 5.264090056285179,
      "grad_norm": 0.4460284113883972,
      "learning_rate": 0.002828,
      "loss": 1.7759,
      "step": 70144
    },
    {
      "epoch": 5.268893058161351,
      "grad_norm": 0.5028729438781738,
      "learning_rate": 0.002828,
      "loss": 1.7774,
      "step": 70208
    },
    {
      "epoch": 5.273696060037524,
      "grad_norm": 0.4400043189525604,
      "learning_rate": 0.002828,
      "loss": 1.7824,
      "step": 70272
    },
    {
      "epoch": 5.278499061913696,
      "grad_norm": 0.4092008173465729,
      "learning_rate": 0.002828,
      "loss": 1.7757,
      "step": 70336
    },
    {
      "epoch": 5.283302063789868,
      "grad_norm": 0.49051231145858765,
      "learning_rate": 0.002828,
      "loss": 1.7836,
      "step": 70400
    },
    {
      "epoch": 5.288105065666041,
      "grad_norm": 0.4183790981769562,
      "learning_rate": 0.002828,
      "loss": 1.7734,
      "step": 70464
    },
    {
      "epoch": 5.292908067542214,
      "grad_norm": 0.4172358512878418,
      "learning_rate": 0.002828,
      "loss": 1.7709,
      "step": 70528
    },
    {
      "epoch": 5.297711069418386,
      "grad_norm": 0.38371556997299194,
      "learning_rate": 0.002828,
      "loss": 1.773,
      "step": 70592
    },
    {
      "epoch": 5.302514071294559,
      "grad_norm": 0.47471094131469727,
      "learning_rate": 0.002828,
      "loss": 1.7812,
      "step": 70656
    },
    {
      "epoch": 5.307317073170732,
      "grad_norm": 0.4047795534133911,
      "learning_rate": 0.002828,
      "loss": 1.7621,
      "step": 70720
    },
    {
      "epoch": 5.312120075046904,
      "grad_norm": 0.4711602032184601,
      "learning_rate": 0.002828,
      "loss": 1.7662,
      "step": 70784
    },
    {
      "epoch": 5.316923076923077,
      "grad_norm": 0.41445788741111755,
      "learning_rate": 0.002828,
      "loss": 1.7776,
      "step": 70848
    },
    {
      "epoch": 5.32172607879925,
      "grad_norm": 0.40588343143463135,
      "learning_rate": 0.002828,
      "loss": 1.7704,
      "step": 70912
    },
    {
      "epoch": 5.326529080675422,
      "grad_norm": 0.4976954162120819,
      "learning_rate": 0.002828,
      "loss": 1.7772,
      "step": 70976
    },
    {
      "epoch": 5.331332082551595,
      "grad_norm": 0.47486379742622375,
      "learning_rate": 0.002828,
      "loss": 1.775,
      "step": 71040
    },
    {
      "epoch": 5.3361350844277675,
      "grad_norm": 0.41308653354644775,
      "learning_rate": 0.002828,
      "loss": 1.7709,
      "step": 71104
    },
    {
      "epoch": 5.3409380863039395,
      "grad_norm": 0.4700968861579895,
      "learning_rate": 0.002828,
      "loss": 1.7704,
      "step": 71168
    },
    {
      "epoch": 5.345741088180112,
      "grad_norm": 0.4668005108833313,
      "learning_rate": 0.002828,
      "loss": 1.7687,
      "step": 71232
    },
    {
      "epoch": 5.350544090056285,
      "grad_norm": 0.4304405152797699,
      "learning_rate": 0.002828,
      "loss": 1.7713,
      "step": 71296
    },
    {
      "epoch": 5.355347091932458,
      "grad_norm": 0.50732421875,
      "learning_rate": 0.002828,
      "loss": 1.7648,
      "step": 71360
    },
    {
      "epoch": 5.36015009380863,
      "grad_norm": 0.4894561469554901,
      "learning_rate": 0.002828,
      "loss": 1.7704,
      "step": 71424
    },
    {
      "epoch": 5.364953095684803,
      "grad_norm": 0.4268856942653656,
      "learning_rate": 0.002828,
      "loss": 1.7709,
      "step": 71488
    },
    {
      "epoch": 5.369756097560976,
      "grad_norm": 0.37423834204673767,
      "learning_rate": 0.002828,
      "loss": 1.7742,
      "step": 71552
    },
    {
      "epoch": 5.374559099437148,
      "grad_norm": 0.4616115987300873,
      "learning_rate": 0.002828,
      "loss": 1.7804,
      "step": 71616
    },
    {
      "epoch": 5.379362101313321,
      "grad_norm": 0.3891536593437195,
      "learning_rate": 0.002828,
      "loss": 1.7745,
      "step": 71680
    },
    {
      "epoch": 5.384165103189494,
      "grad_norm": 0.4057512581348419,
      "learning_rate": 0.002828,
      "loss": 1.7774,
      "step": 71744
    },
    {
      "epoch": 5.388968105065666,
      "grad_norm": 0.4673733115196228,
      "learning_rate": 0.002828,
      "loss": 1.7751,
      "step": 71808
    },
    {
      "epoch": 5.393771106941839,
      "grad_norm": 0.38505420088768005,
      "learning_rate": 0.002828,
      "loss": 1.7713,
      "step": 71872
    },
    {
      "epoch": 5.3985741088180115,
      "grad_norm": 0.4115614593029022,
      "learning_rate": 0.002828,
      "loss": 1.7743,
      "step": 71936
    },
    {
      "epoch": 5.4033771106941835,
      "grad_norm": 0.42474934458732605,
      "learning_rate": 0.002828,
      "loss": 1.7748,
      "step": 72000
    },
    {
      "epoch": 5.408180112570356,
      "grad_norm": 0.44842562079429626,
      "learning_rate": 0.002828,
      "loss": 1.7753,
      "step": 72064
    },
    {
      "epoch": 5.412983114446529,
      "grad_norm": 0.4199115037918091,
      "learning_rate": 0.002828,
      "loss": 1.7719,
      "step": 72128
    },
    {
      "epoch": 5.417786116322702,
      "grad_norm": 0.40089020133018494,
      "learning_rate": 0.002828,
      "loss": 1.7723,
      "step": 72192
    },
    {
      "epoch": 5.422589118198874,
      "grad_norm": 0.5181447863578796,
      "learning_rate": 0.002828,
      "loss": 1.7673,
      "step": 72256
    },
    {
      "epoch": 5.427392120075047,
      "grad_norm": 0.4892163574695587,
      "learning_rate": 0.002828,
      "loss": 1.7742,
      "step": 72320
    },
    {
      "epoch": 5.43219512195122,
      "grad_norm": 0.41054004430770874,
      "learning_rate": 0.002828,
      "loss": 1.7701,
      "step": 72384
    },
    {
      "epoch": 5.436998123827392,
      "grad_norm": 0.4574233293533325,
      "learning_rate": 0.002828,
      "loss": 1.7648,
      "step": 72448
    },
    {
      "epoch": 5.441801125703565,
      "grad_norm": 0.4245249629020691,
      "learning_rate": 0.002828,
      "loss": 1.7762,
      "step": 72512
    },
    {
      "epoch": 5.446604127579738,
      "grad_norm": 0.43858468532562256,
      "learning_rate": 0.002828,
      "loss": 1.7692,
      "step": 72576
    },
    {
      "epoch": 5.45140712945591,
      "grad_norm": 0.477273553609848,
      "learning_rate": 0.002828,
      "loss": 1.7723,
      "step": 72640
    },
    {
      "epoch": 5.4562101313320825,
      "grad_norm": 0.5352038145065308,
      "learning_rate": 0.002828,
      "loss": 1.7735,
      "step": 72704
    },
    {
      "epoch": 5.461013133208255,
      "grad_norm": 0.4089607894420624,
      "learning_rate": 0.002828,
      "loss": 1.7744,
      "step": 72768
    },
    {
      "epoch": 5.465816135084427,
      "grad_norm": 0.4176151752471924,
      "learning_rate": 0.002828,
      "loss": 1.7725,
      "step": 72832
    },
    {
      "epoch": 5.4706191369606,
      "grad_norm": 0.44598644971847534,
      "learning_rate": 0.002828,
      "loss": 1.7747,
      "step": 72896
    },
    {
      "epoch": 5.475422138836773,
      "grad_norm": 0.3903334438800812,
      "learning_rate": 0.002828,
      "loss": 1.7687,
      "step": 72960
    },
    {
      "epoch": 5.480225140712945,
      "grad_norm": 0.4536639451980591,
      "learning_rate": 0.002828,
      "loss": 1.7726,
      "step": 73024
    },
    {
      "epoch": 5.485028142589118,
      "grad_norm": 0.42450371384620667,
      "learning_rate": 0.002828,
      "loss": 1.7757,
      "step": 73088
    },
    {
      "epoch": 5.489831144465291,
      "grad_norm": 0.49686697125434875,
      "learning_rate": 0.002828,
      "loss": 1.7723,
      "step": 73152
    },
    {
      "epoch": 5.494634146341463,
      "grad_norm": 0.4569362699985504,
      "learning_rate": 0.002828,
      "loss": 1.7665,
      "step": 73216
    },
    {
      "epoch": 5.499437148217636,
      "grad_norm": 0.47700774669647217,
      "learning_rate": 0.002828,
      "loss": 1.7733,
      "step": 73280
    },
    {
      "epoch": 5.504240150093809,
      "grad_norm": 0.4036250412464142,
      "learning_rate": 0.002828,
      "loss": 1.7719,
      "step": 73344
    },
    {
      "epoch": 5.509043151969982,
      "grad_norm": 0.4487983286380768,
      "learning_rate": 0.002828,
      "loss": 1.766,
      "step": 73408
    },
    {
      "epoch": 5.513846153846154,
      "grad_norm": 0.519871175289154,
      "learning_rate": 0.002828,
      "loss": 1.7709,
      "step": 73472
    },
    {
      "epoch": 5.5186491557223265,
      "grad_norm": 0.4376335144042969,
      "learning_rate": 0.002828,
      "loss": 1.769,
      "step": 73536
    },
    {
      "epoch": 5.523452157598499,
      "grad_norm": 0.45950847864151,
      "learning_rate": 0.002828,
      "loss": 1.7753,
      "step": 73600
    },
    {
      "epoch": 5.528255159474671,
      "grad_norm": 0.47752419114112854,
      "learning_rate": 0.002828,
      "loss": 1.7612,
      "step": 73664
    },
    {
      "epoch": 5.533058161350844,
      "grad_norm": 0.4177493453025818,
      "learning_rate": 0.002828,
      "loss": 1.7689,
      "step": 73728
    },
    {
      "epoch": 5.537861163227017,
      "grad_norm": 0.427636981010437,
      "learning_rate": 0.002828,
      "loss": 1.7645,
      "step": 73792
    },
    {
      "epoch": 5.542664165103189,
      "grad_norm": 0.46148422360420227,
      "learning_rate": 0.002828,
      "loss": 1.7697,
      "step": 73856
    },
    {
      "epoch": 5.547467166979362,
      "grad_norm": 0.4212297797203064,
      "learning_rate": 0.002828,
      "loss": 1.7689,
      "step": 73920
    },
    {
      "epoch": 5.552270168855535,
      "grad_norm": 0.46947184205055237,
      "learning_rate": 0.002828,
      "loss": 1.7663,
      "step": 73984
    },
    {
      "epoch": 5.557073170731707,
      "grad_norm": 0.4970608353614807,
      "learning_rate": 0.002828,
      "loss": 1.7706,
      "step": 74048
    },
    {
      "epoch": 5.56187617260788,
      "grad_norm": 0.43422332406044006,
      "learning_rate": 0.002828,
      "loss": 1.7674,
      "step": 74112
    },
    {
      "epoch": 5.566679174484053,
      "grad_norm": 0.4593464732170105,
      "learning_rate": 0.002828,
      "loss": 1.7651,
      "step": 74176
    },
    {
      "epoch": 5.571482176360226,
      "grad_norm": 0.5521155595779419,
      "learning_rate": 0.002828,
      "loss": 1.7626,
      "step": 74240
    },
    {
      "epoch": 5.576285178236398,
      "grad_norm": 0.44747570157051086,
      "learning_rate": 0.002828,
      "loss": 1.7628,
      "step": 74304
    },
    {
      "epoch": 5.5810881801125705,
      "grad_norm": 0.4199005365371704,
      "learning_rate": 0.002828,
      "loss": 1.7654,
      "step": 74368
    },
    {
      "epoch": 5.585891181988743,
      "grad_norm": 0.4620073735713959,
      "learning_rate": 0.002828,
      "loss": 1.7596,
      "step": 74432
    },
    {
      "epoch": 5.590694183864915,
      "grad_norm": 0.4747791886329651,
      "learning_rate": 0.002828,
      "loss": 1.7681,
      "step": 74496
    },
    {
      "epoch": 5.595497185741088,
      "grad_norm": 0.39181622862815857,
      "learning_rate": 0.002828,
      "loss": 1.7659,
      "step": 74560
    },
    {
      "epoch": 5.600300187617261,
      "grad_norm": 0.48630768060684204,
      "learning_rate": 0.002828,
      "loss": 1.761,
      "step": 74624
    },
    {
      "epoch": 5.605103189493433,
      "grad_norm": 0.41863974928855896,
      "learning_rate": 0.002828,
      "loss": 1.764,
      "step": 74688
    },
    {
      "epoch": 5.609906191369606,
      "grad_norm": 0.4605071544647217,
      "learning_rate": 0.002828,
      "loss": 1.7615,
      "step": 74752
    },
    {
      "epoch": 5.614709193245779,
      "grad_norm": 0.4120197594165802,
      "learning_rate": 0.002828,
      "loss": 1.765,
      "step": 74816
    },
    {
      "epoch": 5.619512195121951,
      "grad_norm": 0.49022045731544495,
      "learning_rate": 0.002828,
      "loss": 1.7649,
      "step": 74880
    },
    {
      "epoch": 5.624315196998124,
      "grad_norm": 0.5220785140991211,
      "learning_rate": 0.002828,
      "loss": 1.7689,
      "step": 74944
    },
    {
      "epoch": 5.629118198874297,
      "grad_norm": 0.5221869945526123,
      "learning_rate": 0.002828,
      "loss": 1.7584,
      "step": 75008
    },
    {
      "epoch": 5.6339212007504695,
      "grad_norm": 0.44775456190109253,
      "learning_rate": 0.002828,
      "loss": 1.7608,
      "step": 75072
    },
    {
      "epoch": 5.6387242026266415,
      "grad_norm": 0.4067821502685547,
      "learning_rate": 0.002828,
      "loss": 1.7664,
      "step": 75136
    },
    {
      "epoch": 5.643527204502814,
      "grad_norm": 0.411039263010025,
      "learning_rate": 0.002828,
      "loss": 1.7673,
      "step": 75200
    },
    {
      "epoch": 5.648330206378986,
      "grad_norm": 0.45366838574409485,
      "learning_rate": 0.002828,
      "loss": 1.7668,
      "step": 75264
    },
    {
      "epoch": 5.653133208255159,
      "grad_norm": 0.4279555380344391,
      "learning_rate": 0.002828,
      "loss": 1.7585,
      "step": 75328
    },
    {
      "epoch": 5.657936210131332,
      "grad_norm": 0.41225770115852356,
      "learning_rate": 0.002828,
      "loss": 1.7647,
      "step": 75392
    },
    {
      "epoch": 5.662739212007505,
      "grad_norm": 0.4502100944519043,
      "learning_rate": 0.002828,
      "loss": 1.7556,
      "step": 75456
    },
    {
      "epoch": 5.667542213883677,
      "grad_norm": 0.4367251992225647,
      "learning_rate": 0.002828,
      "loss": 1.7626,
      "step": 75520
    },
    {
      "epoch": 5.67234521575985,
      "grad_norm": 0.3955664038658142,
      "learning_rate": 0.002828,
      "loss": 1.7611,
      "step": 75584
    },
    {
      "epoch": 5.677148217636023,
      "grad_norm": 0.4193713963031769,
      "learning_rate": 0.002828,
      "loss": 1.7591,
      "step": 75648
    },
    {
      "epoch": 5.681951219512195,
      "grad_norm": 0.4191085696220398,
      "learning_rate": 0.002828,
      "loss": 1.7553,
      "step": 75712
    },
    {
      "epoch": 5.686754221388368,
      "grad_norm": 0.47768324613571167,
      "learning_rate": 0.002828,
      "loss": 1.7623,
      "step": 75776
    },
    {
      "epoch": 5.691557223264541,
      "grad_norm": 0.46289485692977905,
      "learning_rate": 0.002828,
      "loss": 1.7692,
      "step": 75840
    },
    {
      "epoch": 5.696360225140713,
      "grad_norm": 0.45249009132385254,
      "learning_rate": 0.002828,
      "loss": 1.7575,
      "step": 75904
    },
    {
      "epoch": 5.7011632270168855,
      "grad_norm": 0.4678303003311157,
      "learning_rate": 0.002828,
      "loss": 1.755,
      "step": 75968
    },
    {
      "epoch": 5.705966228893058,
      "grad_norm": 0.49041491746902466,
      "learning_rate": 0.002828,
      "loss": 1.7617,
      "step": 76032
    },
    {
      "epoch": 5.71076923076923,
      "grad_norm": 0.47408047318458557,
      "learning_rate": 0.002828,
      "loss": 1.7593,
      "step": 76096
    },
    {
      "epoch": 5.715572232645403,
      "grad_norm": 0.5664355754852295,
      "learning_rate": 0.002828,
      "loss": 1.7649,
      "step": 76160
    },
    {
      "epoch": 5.720375234521576,
      "grad_norm": 0.4134531319141388,
      "learning_rate": 0.002828,
      "loss": 1.7619,
      "step": 76224
    },
    {
      "epoch": 5.725178236397749,
      "grad_norm": 0.4312486946582794,
      "learning_rate": 0.002828,
      "loss": 1.7547,
      "step": 76288
    },
    {
      "epoch": 5.729981238273921,
      "grad_norm": 0.41487306356430054,
      "learning_rate": 0.002828,
      "loss": 1.7501,
      "step": 76352
    },
    {
      "epoch": 5.734784240150094,
      "grad_norm": 0.4305776059627533,
      "learning_rate": 0.002828,
      "loss": 1.7502,
      "step": 76416
    },
    {
      "epoch": 5.739587242026267,
      "grad_norm": 0.4429924190044403,
      "learning_rate": 0.002828,
      "loss": 1.7554,
      "step": 76480
    },
    {
      "epoch": 5.744390243902439,
      "grad_norm": 0.47873786091804504,
      "learning_rate": 0.002828,
      "loss": 1.7535,
      "step": 76544
    },
    {
      "epoch": 5.749193245778612,
      "grad_norm": 0.4853677749633789,
      "learning_rate": 0.002828,
      "loss": 1.7514,
      "step": 76608
    },
    {
      "epoch": 5.753996247654785,
      "grad_norm": 0.43151888251304626,
      "learning_rate": 0.002828,
      "loss": 1.7548,
      "step": 76672
    },
    {
      "epoch": 5.758799249530957,
      "grad_norm": 0.5341268181800842,
      "learning_rate": 0.002828,
      "loss": 1.7626,
      "step": 76736
    },
    {
      "epoch": 5.7636022514071295,
      "grad_norm": 0.5068721175193787,
      "learning_rate": 0.002828,
      "loss": 1.7559,
      "step": 76800
    },
    {
      "epoch": 5.768405253283302,
      "grad_norm": 0.4304395318031311,
      "learning_rate": 0.002828,
      "loss": 1.7604,
      "step": 76864
    },
    {
      "epoch": 5.773208255159474,
      "grad_norm": 0.44748765230178833,
      "learning_rate": 0.002828,
      "loss": 1.7568,
      "step": 76928
    },
    {
      "epoch": 5.778011257035647,
      "grad_norm": 0.40714704990386963,
      "learning_rate": 0.002828,
      "loss": 1.7553,
      "step": 76992
    },
    {
      "epoch": 5.78281425891182,
      "grad_norm": 0.46348631381988525,
      "learning_rate": 0.002828,
      "loss": 1.7536,
      "step": 77056
    },
    {
      "epoch": 5.787617260787993,
      "grad_norm": 0.5729508996009827,
      "learning_rate": 0.002828,
      "loss": 1.7513,
      "step": 77120
    },
    {
      "epoch": 5.792420262664165,
      "grad_norm": 0.4666328728199005,
      "learning_rate": 0.002828,
      "loss": 1.7535,
      "step": 77184
    },
    {
      "epoch": 5.797223264540338,
      "grad_norm": 0.5445581674575806,
      "learning_rate": 0.002828,
      "loss": 1.7589,
      "step": 77248
    },
    {
      "epoch": 5.80202626641651,
      "grad_norm": 0.3767351508140564,
      "learning_rate": 0.002828,
      "loss": 1.7585,
      "step": 77312
    },
    {
      "epoch": 5.806829268292683,
      "grad_norm": 0.4318215548992157,
      "learning_rate": 0.002828,
      "loss": 1.76,
      "step": 77376
    },
    {
      "epoch": 5.811632270168856,
      "grad_norm": 0.43348705768585205,
      "learning_rate": 0.002828,
      "loss": 1.7468,
      "step": 77440
    },
    {
      "epoch": 5.8164352720450285,
      "grad_norm": 0.4296760559082031,
      "learning_rate": 0.002828,
      "loss": 1.7599,
      "step": 77504
    },
    {
      "epoch": 5.8212382739212005,
      "grad_norm": 0.42506301403045654,
      "learning_rate": 0.002828,
      "loss": 1.7577,
      "step": 77568
    },
    {
      "epoch": 5.826041275797373,
      "grad_norm": 0.4925733506679535,
      "learning_rate": 0.002828,
      "loss": 1.7571,
      "step": 77632
    },
    {
      "epoch": 5.830844277673546,
      "grad_norm": 0.4275703728199005,
      "learning_rate": 0.002828,
      "loss": 1.7515,
      "step": 77696
    },
    {
      "epoch": 5.835647279549718,
      "grad_norm": 0.4526989459991455,
      "learning_rate": 0.002828,
      "loss": 1.7631,
      "step": 77760
    },
    {
      "epoch": 5.840450281425891,
      "grad_norm": 0.39497390389442444,
      "learning_rate": 0.002828,
      "loss": 1.7548,
      "step": 77824
    },
    {
      "epoch": 5.845253283302064,
      "grad_norm": 0.39526093006134033,
      "learning_rate": 0.002828,
      "loss": 1.7547,
      "step": 77888
    },
    {
      "epoch": 5.850056285178236,
      "grad_norm": 0.4767382740974426,
      "learning_rate": 0.002828,
      "loss": 1.7515,
      "step": 77952
    },
    {
      "epoch": 5.854859287054409,
      "grad_norm": 0.36858007311820984,
      "learning_rate": 0.002828,
      "loss": 1.7564,
      "step": 78016
    },
    {
      "epoch": 5.859662288930582,
      "grad_norm": 0.5229625701904297,
      "learning_rate": 0.002828,
      "loss": 1.7523,
      "step": 78080
    },
    {
      "epoch": 5.864465290806754,
      "grad_norm": 0.46653860807418823,
      "learning_rate": 0.002828,
      "loss": 1.7581,
      "step": 78144
    },
    {
      "epoch": 5.869268292682927,
      "grad_norm": 0.4254682660102844,
      "learning_rate": 0.002828,
      "loss": 1.7524,
      "step": 78208
    },
    {
      "epoch": 5.8740712945591,
      "grad_norm": 0.4454876482486725,
      "learning_rate": 0.002828,
      "loss": 1.7572,
      "step": 78272
    },
    {
      "epoch": 5.8788742964352725,
      "grad_norm": 0.41046759486198425,
      "learning_rate": 0.002828,
      "loss": 1.7523,
      "step": 78336
    },
    {
      "epoch": 5.8836772983114445,
      "grad_norm": 0.4965238869190216,
      "learning_rate": 0.002828,
      "loss": 1.7567,
      "step": 78400
    },
    {
      "epoch": 5.888480300187617,
      "grad_norm": 0.4154772460460663,
      "learning_rate": 0.002828,
      "loss": 1.758,
      "step": 78464
    },
    {
      "epoch": 5.89328330206379,
      "grad_norm": 0.46626389026641846,
      "learning_rate": 0.002828,
      "loss": 1.7572,
      "step": 78528
    },
    {
      "epoch": 5.898086303939962,
      "grad_norm": 0.4979894161224365,
      "learning_rate": 0.002828,
      "loss": 1.7583,
      "step": 78592
    },
    {
      "epoch": 5.902889305816135,
      "grad_norm": 0.4381833076477051,
      "learning_rate": 0.002828,
      "loss": 1.7547,
      "step": 78656
    },
    {
      "epoch": 5.907692307692308,
      "grad_norm": 0.40586981177330017,
      "learning_rate": 0.002828,
      "loss": 1.7521,
      "step": 78720
    },
    {
      "epoch": 5.91249530956848,
      "grad_norm": 0.3989832401275635,
      "learning_rate": 0.002828,
      "loss": 1.7526,
      "step": 78784
    },
    {
      "epoch": 5.917298311444653,
      "grad_norm": 0.473142147064209,
      "learning_rate": 0.002828,
      "loss": 1.7523,
      "step": 78848
    },
    {
      "epoch": 5.922101313320826,
      "grad_norm": 0.4312800168991089,
      "learning_rate": 0.002828,
      "loss": 1.7458,
      "step": 78912
    },
    {
      "epoch": 5.926904315196998,
      "grad_norm": 0.48550522327423096,
      "learning_rate": 0.002828,
      "loss": 1.7527,
      "step": 78976
    },
    {
      "epoch": 5.931707317073171,
      "grad_norm": 0.41313230991363525,
      "learning_rate": 0.002828,
      "loss": 1.7584,
      "step": 79040
    },
    {
      "epoch": 5.936510318949344,
      "grad_norm": 0.4612511396408081,
      "learning_rate": 0.002828,
      "loss": 1.7542,
      "step": 79104
    },
    {
      "epoch": 5.941313320825516,
      "grad_norm": 0.46551233530044556,
      "learning_rate": 0.002828,
      "loss": 1.7577,
      "step": 79168
    },
    {
      "epoch": 5.946116322701688,
      "grad_norm": 0.4390702247619629,
      "learning_rate": 0.002828,
      "loss": 1.7494,
      "step": 79232
    },
    {
      "epoch": 5.950919324577861,
      "grad_norm": 0.4530523717403412,
      "learning_rate": 0.002828,
      "loss": 1.7448,
      "step": 79296
    },
    {
      "epoch": 5.955722326454033,
      "grad_norm": 0.5139921307563782,
      "learning_rate": 0.002828,
      "loss": 1.7472,
      "step": 79360
    },
    {
      "epoch": 5.960525328330206,
      "grad_norm": 0.4357989430427551,
      "learning_rate": 0.002828,
      "loss": 1.7522,
      "step": 79424
    },
    {
      "epoch": 5.965328330206379,
      "grad_norm": 0.46577298641204834,
      "learning_rate": 0.002828,
      "loss": 1.7445,
      "step": 79488
    },
    {
      "epoch": 5.970131332082552,
      "grad_norm": 0.5673984885215759,
      "learning_rate": 0.002828,
      "loss": 1.7479,
      "step": 79552
    },
    {
      "epoch": 5.974934333958724,
      "grad_norm": 0.41843244433403015,
      "learning_rate": 0.002828,
      "loss": 1.7528,
      "step": 79616
    },
    {
      "epoch": 5.979737335834897,
      "grad_norm": 0.5128591656684875,
      "learning_rate": 0.002828,
      "loss": 1.7487,
      "step": 79680
    },
    {
      "epoch": 5.98454033771107,
      "grad_norm": 0.4667825698852539,
      "learning_rate": 0.002828,
      "loss": 1.7429,
      "step": 79744
    },
    {
      "epoch": 5.989343339587242,
      "grad_norm": 0.41787078976631165,
      "learning_rate": 0.002828,
      "loss": 1.7463,
      "step": 79808
    },
    {
      "epoch": 5.994146341463415,
      "grad_norm": 0.43411746621131897,
      "learning_rate": 0.002828,
      "loss": 1.7477,
      "step": 79872
    },
    {
      "epoch": 5.9989493433395875,
      "grad_norm": 0.4941996932029724,
      "learning_rate": 0.002828,
      "loss": 1.7558,
      "step": 79936
    },
    {
      "epoch": 6.0037523452157595,
      "grad_norm": 0.48769477009773254,
      "learning_rate": 0.002828,
      "loss": 1.7094,
      "step": 80000
    },
    {
      "epoch": 6.008555347091932,
      "grad_norm": 0.47256094217300415,
      "learning_rate": 0.002828,
      "loss": 1.7016,
      "step": 80064
    },
    {
      "epoch": 6.013358348968105,
      "grad_norm": 0.4429653286933899,
      "learning_rate": 0.002828,
      "loss": 1.698,
      "step": 80128
    },
    {
      "epoch": 6.018161350844277,
      "grad_norm": 0.44992125034332275,
      "learning_rate": 0.002828,
      "loss": 1.697,
      "step": 80192
    },
    {
      "epoch": 6.02296435272045,
      "grad_norm": 0.4283198416233063,
      "learning_rate": 0.002828,
      "loss": 1.7042,
      "step": 80256
    },
    {
      "epoch": 6.027767354596623,
      "grad_norm": 0.41068553924560547,
      "learning_rate": 0.002828,
      "loss": 1.7081,
      "step": 80320
    },
    {
      "epoch": 6.032570356472796,
      "grad_norm": 0.4621987044811249,
      "learning_rate": 0.002828,
      "loss": 1.7038,
      "step": 80384
    },
    {
      "epoch": 6.037373358348968,
      "grad_norm": 0.43681883811950684,
      "learning_rate": 0.002828,
      "loss": 1.707,
      "step": 80448
    },
    {
      "epoch": 6.042176360225141,
      "grad_norm": 0.43538138270378113,
      "learning_rate": 0.002828,
      "loss": 1.7031,
      "step": 80512
    },
    {
      "epoch": 6.046979362101314,
      "grad_norm": 0.488596647977829,
      "learning_rate": 0.002828,
      "loss": 1.7071,
      "step": 80576
    },
    {
      "epoch": 6.051782363977486,
      "grad_norm": 0.48277080059051514,
      "learning_rate": 0.002828,
      "loss": 1.7054,
      "step": 80640
    },
    {
      "epoch": 6.056585365853659,
      "grad_norm": 0.5193591713905334,
      "learning_rate": 0.002828,
      "loss": 1.705,
      "step": 80704
    },
    {
      "epoch": 6.0613883677298315,
      "grad_norm": 0.42429229617118835,
      "learning_rate": 0.002828,
      "loss": 1.7042,
      "step": 80768
    },
    {
      "epoch": 6.0661913696060035,
      "grad_norm": 0.42066484689712524,
      "learning_rate": 0.002828,
      "loss": 1.705,
      "step": 80832
    },
    {
      "epoch": 6.070994371482176,
      "grad_norm": 0.5428970456123352,
      "learning_rate": 0.002828,
      "loss": 1.705,
      "step": 80896
    },
    {
      "epoch": 6.075797373358349,
      "grad_norm": 0.4695224463939667,
      "learning_rate": 0.002828,
      "loss": 1.7107,
      "step": 80960
    },
    {
      "epoch": 6.080600375234521,
      "grad_norm": 0.5089054703712463,
      "learning_rate": 0.002828,
      "loss": 1.7038,
      "step": 81024
    },
    {
      "epoch": 6.085403377110694,
      "grad_norm": 0.4825155735015869,
      "learning_rate": 0.002828,
      "loss": 1.7107,
      "step": 81088
    },
    {
      "epoch": 6.090206378986867,
      "grad_norm": 0.5433881282806396,
      "learning_rate": 0.002828,
      "loss": 1.7076,
      "step": 81152
    },
    {
      "epoch": 6.095009380863039,
      "grad_norm": 0.563622236251831,
      "learning_rate": 0.002828,
      "loss": 1.7025,
      "step": 81216
    },
    {
      "epoch": 6.099812382739212,
      "grad_norm": 0.4570015072822571,
      "learning_rate": 0.002828,
      "loss": 1.7039,
      "step": 81280
    },
    {
      "epoch": 6.104615384615385,
      "grad_norm": 0.6059426069259644,
      "learning_rate": 0.002828,
      "loss": 1.7065,
      "step": 81344
    },
    {
      "epoch": 6.109418386491557,
      "grad_norm": 0.46249598264694214,
      "learning_rate": 0.002828,
      "loss": 1.7057,
      "step": 81408
    },
    {
      "epoch": 6.11422138836773,
      "grad_norm": 0.4301151931285858,
      "learning_rate": 0.002828,
      "loss": 1.7054,
      "step": 81472
    },
    {
      "epoch": 6.1190243902439025,
      "grad_norm": 0.5160223245620728,
      "learning_rate": 0.002828,
      "loss": 1.7087,
      "step": 81536
    },
    {
      "epoch": 6.123827392120075,
      "grad_norm": 0.48462730646133423,
      "learning_rate": 0.002828,
      "loss": 1.7093,
      "step": 81600
    },
    {
      "epoch": 6.128630393996247,
      "grad_norm": 0.5068091750144958,
      "learning_rate": 0.002828,
      "loss": 1.7085,
      "step": 81664
    },
    {
      "epoch": 6.13343339587242,
      "grad_norm": 0.4427911341190338,
      "learning_rate": 0.002828,
      "loss": 1.6977,
      "step": 81728
    },
    {
      "epoch": 6.138236397748593,
      "grad_norm": 0.4816674292087555,
      "learning_rate": 0.002828,
      "loss": 1.7067,
      "step": 81792
    },
    {
      "epoch": 6.143039399624765,
      "grad_norm": 0.48784971237182617,
      "learning_rate": 0.002828,
      "loss": 1.7099,
      "step": 81856
    },
    {
      "epoch": 6.147842401500938,
      "grad_norm": 0.4686130881309509,
      "learning_rate": 0.002828,
      "loss": 1.7072,
      "step": 81920
    },
    {
      "epoch": 6.152645403377111,
      "grad_norm": 0.5386856198310852,
      "learning_rate": 0.002828,
      "loss": 1.7106,
      "step": 81984
    },
    {
      "epoch": 6.157448405253283,
      "grad_norm": 0.44840648770332336,
      "learning_rate": 0.002828,
      "loss": 1.7064,
      "step": 82048
    },
    {
      "epoch": 6.162251407129456,
      "grad_norm": 0.4645565450191498,
      "learning_rate": 0.002828,
      "loss": 1.706,
      "step": 82112
    },
    {
      "epoch": 6.167054409005629,
      "grad_norm": 0.6145454049110413,
      "learning_rate": 0.002828,
      "loss": 1.7072,
      "step": 82176
    },
    {
      "epoch": 6.171857410881801,
      "grad_norm": 0.526295006275177,
      "learning_rate": 0.002828,
      "loss": 1.7054,
      "step": 82240
    },
    {
      "epoch": 6.176660412757974,
      "grad_norm": 0.4715871810913086,
      "learning_rate": 0.002828,
      "loss": 1.7063,
      "step": 82304
    },
    {
      "epoch": 6.1814634146341465,
      "grad_norm": 0.48130881786346436,
      "learning_rate": 0.002828,
      "loss": 1.7061,
      "step": 82368
    },
    {
      "epoch": 6.186266416510319,
      "grad_norm": 0.4163391590118408,
      "learning_rate": 0.002828,
      "loss": 1.7079,
      "step": 82432
    },
    {
      "epoch": 6.191069418386491,
      "grad_norm": 0.509125292301178,
      "learning_rate": 0.002828,
      "loss": 1.7078,
      "step": 82496
    },
    {
      "epoch": 6.195872420262664,
      "grad_norm": 0.454570472240448,
      "learning_rate": 0.002828,
      "loss": 1.7058,
      "step": 82560
    },
    {
      "epoch": 6.200675422138837,
      "grad_norm": 0.40525802969932556,
      "learning_rate": 0.002828,
      "loss": 1.7054,
      "step": 82624
    },
    {
      "epoch": 6.205478424015009,
      "grad_norm": 0.44681093096733093,
      "learning_rate": 0.002828,
      "loss": 1.7154,
      "step": 82688
    },
    {
      "epoch": 6.210281425891182,
      "grad_norm": 0.4531172811985016,
      "learning_rate": 0.002828,
      "loss": 1.7073,
      "step": 82752
    },
    {
      "epoch": 6.215084427767355,
      "grad_norm": 0.42768582701683044,
      "learning_rate": 0.002828,
      "loss": 1.7033,
      "step": 82816
    },
    {
      "epoch": 6.219887429643527,
      "grad_norm": 0.49657461047172546,
      "learning_rate": 0.002828,
      "loss": 1.7067,
      "step": 82880
    },
    {
      "epoch": 6.2246904315197,
      "grad_norm": 0.45019227266311646,
      "learning_rate": 0.002828,
      "loss": 1.6968,
      "step": 82944
    },
    {
      "epoch": 6.229493433395873,
      "grad_norm": 0.41543009877204895,
      "learning_rate": 0.002828,
      "loss": 1.7031,
      "step": 83008
    },
    {
      "epoch": 6.234296435272045,
      "grad_norm": 0.44448015093803406,
      "learning_rate": 0.002828,
      "loss": 1.7034,
      "step": 83072
    },
    {
      "epoch": 6.239099437148218,
      "grad_norm": 0.49361652135849,
      "learning_rate": 0.002828,
      "loss": 1.7072,
      "step": 83136
    },
    {
      "epoch": 6.2439024390243905,
      "grad_norm": 0.4431740939617157,
      "learning_rate": 0.002828,
      "loss": 1.6989,
      "step": 83200
    },
    {
      "epoch": 6.2487054409005625,
      "grad_norm": 0.43959954380989075,
      "learning_rate": 0.002828,
      "loss": 1.7078,
      "step": 83264
    },
    {
      "epoch": 6.253508442776735,
      "grad_norm": 0.4439740777015686,
      "learning_rate": 0.002828,
      "loss": 1.7065,
      "step": 83328
    },
    {
      "epoch": 6.258311444652908,
      "grad_norm": 0.5072445273399353,
      "learning_rate": 0.002828,
      "loss": 1.7058,
      "step": 83392
    },
    {
      "epoch": 6.26311444652908,
      "grad_norm": 0.6075462102890015,
      "learning_rate": 0.002828,
      "loss": 1.7105,
      "step": 83456
    },
    {
      "epoch": 6.267917448405253,
      "grad_norm": 0.47057533264160156,
      "learning_rate": 0.002828,
      "loss": 1.7081,
      "step": 83520
    },
    {
      "epoch": 6.272720450281426,
      "grad_norm": 0.48063352704048157,
      "learning_rate": 0.002828,
      "loss": 1.7027,
      "step": 83584
    },
    {
      "epoch": 6.277523452157599,
      "grad_norm": 0.5260996222496033,
      "learning_rate": 0.002828,
      "loss": 1.7072,
      "step": 83648
    },
    {
      "epoch": 6.282326454033771,
      "grad_norm": 0.45026955008506775,
      "learning_rate": 0.002828,
      "loss": 1.7023,
      "step": 83712
    },
    {
      "epoch": 6.287129455909944,
      "grad_norm": 0.4766295552253723,
      "learning_rate": 0.002828,
      "loss": 1.7048,
      "step": 83776
    },
    {
      "epoch": 6.291932457786117,
      "grad_norm": 0.5029019713401794,
      "learning_rate": 0.002828,
      "loss": 1.7103,
      "step": 83840
    },
    {
      "epoch": 6.296735459662289,
      "grad_norm": 0.4043557643890381,
      "learning_rate": 0.002828,
      "loss": 1.7081,
      "step": 83904
    },
    {
      "epoch": 6.3015384615384615,
      "grad_norm": 0.4448949694633484,
      "learning_rate": 0.002828,
      "loss": 1.7056,
      "step": 83968
    },
    {
      "epoch": 6.306341463414634,
      "grad_norm": 0.44328081607818604,
      "learning_rate": 0.002828,
      "loss": 1.7021,
      "step": 84032
    },
    {
      "epoch": 6.311144465290806,
      "grad_norm": 0.5088468194007874,
      "learning_rate": 0.002828,
      "loss": 1.7025,
      "step": 84096
    },
    {
      "epoch": 6.315947467166979,
      "grad_norm": 0.5326889753341675,
      "learning_rate": 0.002828,
      "loss": 1.714,
      "step": 84160
    },
    {
      "epoch": 6.320750469043152,
      "grad_norm": 0.4468030631542206,
      "learning_rate": 0.002828,
      "loss": 1.7006,
      "step": 84224
    },
    {
      "epoch": 6.325553470919324,
      "grad_norm": 0.6114151477813721,
      "learning_rate": 0.002828,
      "loss": 1.7012,
      "step": 84288
    },
    {
      "epoch": 6.330356472795497,
      "grad_norm": 0.44562435150146484,
      "learning_rate": 0.002828,
      "loss": 1.7079,
      "step": 84352
    },
    {
      "epoch": 6.33515947467167,
      "grad_norm": 0.4240759313106537,
      "learning_rate": 0.002828,
      "loss": 1.7092,
      "step": 84416
    },
    {
      "epoch": 6.339962476547843,
      "grad_norm": 0.5037844777107239,
      "learning_rate": 0.002828,
      "loss": 1.7017,
      "step": 84480
    },
    {
      "epoch": 6.344765478424015,
      "grad_norm": 0.4932270348072052,
      "learning_rate": 0.002828,
      "loss": 1.7058,
      "step": 84544
    },
    {
      "epoch": 6.349568480300188,
      "grad_norm": 0.5098112225532532,
      "learning_rate": 0.002828,
      "loss": 1.7089,
      "step": 84608
    },
    {
      "epoch": 6.354371482176361,
      "grad_norm": 0.4815918207168579,
      "learning_rate": 0.002828,
      "loss": 1.6983,
      "step": 84672
    },
    {
      "epoch": 6.359174484052533,
      "grad_norm": 0.42424315214157104,
      "learning_rate": 0.002828,
      "loss": 1.7111,
      "step": 84736
    },
    {
      "epoch": 6.3639774859287055,
      "grad_norm": 0.4818706214427948,
      "learning_rate": 0.002828,
      "loss": 1.7007,
      "step": 84800
    },
    {
      "epoch": 6.368780487804878,
      "grad_norm": 0.4532407522201538,
      "learning_rate": 0.002828,
      "loss": 1.7026,
      "step": 84864
    },
    {
      "epoch": 6.37358348968105,
      "grad_norm": 0.4606661796569824,
      "learning_rate": 0.002828,
      "loss": 1.7033,
      "step": 84928
    },
    {
      "epoch": 6.378386491557223,
      "grad_norm": 0.5225577354431152,
      "learning_rate": 0.002828,
      "loss": 1.7058,
      "step": 84992
    },
    {
      "epoch": 6.383189493433396,
      "grad_norm": 0.6367613673210144,
      "learning_rate": 0.002828,
      "loss": 1.7025,
      "step": 85056
    },
    {
      "epoch": 6.387992495309568,
      "grad_norm": 0.48196592926979065,
      "learning_rate": 0.002828,
      "loss": 1.7055,
      "step": 85120
    },
    {
      "epoch": 6.392795497185741,
      "grad_norm": 0.48065000772476196,
      "learning_rate": 0.002828,
      "loss": 1.6962,
      "step": 85184
    },
    {
      "epoch": 6.397598499061914,
      "grad_norm": 0.5992358326911926,
      "learning_rate": 0.002828,
      "loss": 1.7008,
      "step": 85248
    },
    {
      "epoch": 6.402401500938086,
      "grad_norm": 0.4605671763420105,
      "learning_rate": 0.002828,
      "loss": 1.7023,
      "step": 85312
    },
    {
      "epoch": 6.407204502814259,
      "grad_norm": 0.42417392134666443,
      "learning_rate": 0.002828,
      "loss": 1.7101,
      "step": 85376
    },
    {
      "epoch": 6.412007504690432,
      "grad_norm": 0.48525306582450867,
      "learning_rate": 0.002828,
      "loss": 1.7131,
      "step": 85440
    },
    {
      "epoch": 6.416810506566604,
      "grad_norm": 0.4923460781574249,
      "learning_rate": 0.002828,
      "loss": 1.7043,
      "step": 85504
    },
    {
      "epoch": 6.421613508442777,
      "grad_norm": 0.47997698187828064,
      "learning_rate": 0.002828,
      "loss": 1.6941,
      "step": 85568
    },
    {
      "epoch": 6.4264165103189494,
      "grad_norm": 0.47466325759887695,
      "learning_rate": 0.002828,
      "loss": 1.7028,
      "step": 85632
    },
    {
      "epoch": 6.431219512195122,
      "grad_norm": 0.4309011399745941,
      "learning_rate": 0.002828,
      "loss": 1.703,
      "step": 85696
    },
    {
      "epoch": 6.436022514071294,
      "grad_norm": 0.44139745831489563,
      "learning_rate": 0.002828,
      "loss": 1.7037,
      "step": 85760
    },
    {
      "epoch": 6.440825515947467,
      "grad_norm": 0.47135940194129944,
      "learning_rate": 0.002828,
      "loss": 1.7003,
      "step": 85824
    },
    {
      "epoch": 6.44562851782364,
      "grad_norm": 0.40643376111984253,
      "learning_rate": 0.002828,
      "loss": 1.6999,
      "step": 85888
    },
    {
      "epoch": 6.450431519699812,
      "grad_norm": 0.4392285943031311,
      "learning_rate": 0.002828,
      "loss": 1.7047,
      "step": 85952
    },
    {
      "epoch": 6.455234521575985,
      "grad_norm": 0.4400002062320709,
      "learning_rate": 0.002828,
      "loss": 1.7026,
      "step": 86016
    },
    {
      "epoch": 6.460037523452158,
      "grad_norm": 0.45881539583206177,
      "learning_rate": 0.002828,
      "loss": 1.7121,
      "step": 86080
    },
    {
      "epoch": 6.46484052532833,
      "grad_norm": 0.47242626547813416,
      "learning_rate": 0.002828,
      "loss": 1.7069,
      "step": 86144
    },
    {
      "epoch": 6.469643527204503,
      "grad_norm": 0.4134158790111542,
      "learning_rate": 0.002828,
      "loss": 1.6968,
      "step": 86208
    },
    {
      "epoch": 6.474446529080676,
      "grad_norm": 0.45994988083839417,
      "learning_rate": 0.002828,
      "loss": 1.7053,
      "step": 86272
    },
    {
      "epoch": 6.479249530956848,
      "grad_norm": 0.43823009729385376,
      "learning_rate": 0.002828,
      "loss": 1.7046,
      "step": 86336
    },
    {
      "epoch": 6.4840525328330205,
      "grad_norm": 0.40943190455436707,
      "learning_rate": 0.002828,
      "loss": 1.6995,
      "step": 86400
    },
    {
      "epoch": 6.488855534709193,
      "grad_norm": 0.4266324043273926,
      "learning_rate": 0.002828,
      "loss": 1.6967,
      "step": 86464
    },
    {
      "epoch": 6.493658536585366,
      "grad_norm": 0.4427225887775421,
      "learning_rate": 0.002828,
      "loss": 1.7075,
      "step": 86528
    },
    {
      "epoch": 6.498461538461538,
      "grad_norm": 0.44521722197532654,
      "learning_rate": 0.002828,
      "loss": 1.7057,
      "step": 86592
    },
    {
      "epoch": 6.503264540337711,
      "grad_norm": 0.45472028851509094,
      "learning_rate": 0.002828,
      "loss": 1.7096,
      "step": 86656
    },
    {
      "epoch": 6.508067542213884,
      "grad_norm": 0.5230434536933899,
      "learning_rate": 0.002828,
      "loss": 1.7025,
      "step": 86720
    },
    {
      "epoch": 6.512870544090056,
      "grad_norm": 0.5306261777877808,
      "learning_rate": 0.002828,
      "loss": 1.698,
      "step": 86784
    },
    {
      "epoch": 6.517673545966229,
      "grad_norm": 0.427950918674469,
      "learning_rate": 0.002828,
      "loss": 1.7068,
      "step": 86848
    },
    {
      "epoch": 6.522476547842402,
      "grad_norm": 0.5199810862541199,
      "learning_rate": 0.002828,
      "loss": 1.6962,
      "step": 86912
    },
    {
      "epoch": 6.527279549718574,
      "grad_norm": 0.4530502259731293,
      "learning_rate": 0.002828,
      "loss": 1.698,
      "step": 86976
    },
    {
      "epoch": 6.532082551594747,
      "grad_norm": 0.5144789218902588,
      "learning_rate": 0.002828,
      "loss": 1.7048,
      "step": 87040
    },
    {
      "epoch": 6.53688555347092,
      "grad_norm": 0.47851622104644775,
      "learning_rate": 0.002828,
      "loss": 1.7021,
      "step": 87104
    },
    {
      "epoch": 6.541688555347092,
      "grad_norm": 0.47514185309410095,
      "learning_rate": 0.002828,
      "loss": 1.703,
      "step": 87168
    },
    {
      "epoch": 6.5464915572232645,
      "grad_norm": 0.5364302396774292,
      "learning_rate": 0.002828,
      "loss": 1.703,
      "step": 87232
    },
    {
      "epoch": 6.551294559099437,
      "grad_norm": 0.4827858507633209,
      "learning_rate": 0.002828,
      "loss": 1.6982,
      "step": 87296
    },
    {
      "epoch": 6.55609756097561,
      "grad_norm": 0.4004391133785248,
      "learning_rate": 0.002828,
      "loss": 1.6977,
      "step": 87360
    },
    {
      "epoch": 6.560900562851782,
      "grad_norm": 0.4460011422634125,
      "learning_rate": 0.002828,
      "loss": 1.6986,
      "step": 87424
    },
    {
      "epoch": 6.565703564727955,
      "grad_norm": 0.4100639820098877,
      "learning_rate": 0.002828,
      "loss": 1.6983,
      "step": 87488
    },
    {
      "epoch": 6.570506566604127,
      "grad_norm": 0.4702662527561188,
      "learning_rate": 0.002828,
      "loss": 1.7032,
      "step": 87552
    },
    {
      "epoch": 6.5753095684803,
      "grad_norm": 0.4798121750354767,
      "learning_rate": 0.002828,
      "loss": 1.7067,
      "step": 87616
    },
    {
      "epoch": 6.580112570356473,
      "grad_norm": 0.4707585871219635,
      "learning_rate": 0.002828,
      "loss": 1.698,
      "step": 87680
    },
    {
      "epoch": 6.584915572232646,
      "grad_norm": 0.4406650960445404,
      "learning_rate": 0.002828,
      "loss": 1.7022,
      "step": 87744
    },
    {
      "epoch": 6.589718574108818,
      "grad_norm": 0.48812413215637207,
      "learning_rate": 0.002828,
      "loss": 1.7042,
      "step": 87808
    },
    {
      "epoch": 6.594521575984991,
      "grad_norm": 0.41487032175064087,
      "learning_rate": 0.002828,
      "loss": 1.7009,
      "step": 87872
    },
    {
      "epoch": 6.5993245778611636,
      "grad_norm": 0.4759446382522583,
      "learning_rate": 0.002828,
      "loss": 1.7,
      "step": 87936
    },
    {
      "epoch": 6.6041275797373356,
      "grad_norm": 0.47563308477401733,
      "learning_rate": 0.002828,
      "loss": 1.7023,
      "step": 88000
    },
    {
      "epoch": 6.608930581613508,
      "grad_norm": 0.43401914834976196,
      "learning_rate": 0.002828,
      "loss": 1.6924,
      "step": 88064
    },
    {
      "epoch": 6.613733583489681,
      "grad_norm": 0.621225118637085,
      "learning_rate": 0.002828,
      "loss": 1.7028,
      "step": 88128
    },
    {
      "epoch": 6.618536585365853,
      "grad_norm": 0.4878978729248047,
      "learning_rate": 0.002828,
      "loss": 1.7029,
      "step": 88192
    },
    {
      "epoch": 6.623339587242026,
      "grad_norm": 0.4743395745754242,
      "learning_rate": 0.002828,
      "loss": 1.6918,
      "step": 88256
    },
    {
      "epoch": 6.628142589118199,
      "grad_norm": 0.4858582317829132,
      "learning_rate": 0.002828,
      "loss": 1.7026,
      "step": 88320
    },
    {
      "epoch": 6.632945590994371,
      "grad_norm": 0.4205172657966614,
      "learning_rate": 0.002828,
      "loss": 1.6956,
      "step": 88384
    },
    {
      "epoch": 6.637748592870544,
      "grad_norm": 0.4614770710468292,
      "learning_rate": 0.002828,
      "loss": 1.6934,
      "step": 88448
    },
    {
      "epoch": 6.642551594746717,
      "grad_norm": 0.42253434658050537,
      "learning_rate": 0.002828,
      "loss": 1.6968,
      "step": 88512
    },
    {
      "epoch": 6.64735459662289,
      "grad_norm": 0.4911728799343109,
      "learning_rate": 0.002828,
      "loss": 1.6969,
      "step": 88576
    },
    {
      "epoch": 6.652157598499062,
      "grad_norm": 0.48026782274246216,
      "learning_rate": 0.002828,
      "loss": 1.6965,
      "step": 88640
    },
    {
      "epoch": 6.656960600375235,
      "grad_norm": 0.449324369430542,
      "learning_rate": 0.002828,
      "loss": 1.6888,
      "step": 88704
    },
    {
      "epoch": 6.6617636022514075,
      "grad_norm": 0.4832703769207001,
      "learning_rate": 0.002828,
      "loss": 1.6987,
      "step": 88768
    },
    {
      "epoch": 6.6665666041275795,
      "grad_norm": 0.5017644166946411,
      "learning_rate": 0.002828,
      "loss": 1.7053,
      "step": 88832
    },
    {
      "epoch": 6.671369606003752,
      "grad_norm": 0.5028623342514038,
      "learning_rate": 0.002828,
      "loss": 1.6902,
      "step": 88896
    },
    {
      "epoch": 6.676172607879925,
      "grad_norm": 0.4900693893432617,
      "learning_rate": 0.002828,
      "loss": 1.6945,
      "step": 88960
    },
    {
      "epoch": 6.680975609756097,
      "grad_norm": 0.5036649703979492,
      "learning_rate": 0.002828,
      "loss": 1.6938,
      "step": 89024
    },
    {
      "epoch": 6.68577861163227,
      "grad_norm": 0.5400296449661255,
      "learning_rate": 0.002828,
      "loss": 1.6996,
      "step": 89088
    },
    {
      "epoch": 6.690581613508443,
      "grad_norm": 0.4391849935054779,
      "learning_rate": 0.002828,
      "loss": 1.6955,
      "step": 89152
    },
    {
      "epoch": 6.695384615384615,
      "grad_norm": 0.46695676445961,
      "learning_rate": 0.002828,
      "loss": 1.6952,
      "step": 89216
    },
    {
      "epoch": 6.700187617260788,
      "grad_norm": 0.5176795125007629,
      "learning_rate": 0.002828,
      "loss": 1.6982,
      "step": 89280
    },
    {
      "epoch": 6.704990619136961,
      "grad_norm": 0.4465230703353882,
      "learning_rate": 0.002828,
      "loss": 1.6922,
      "step": 89344
    },
    {
      "epoch": 6.709793621013134,
      "grad_norm": 0.7129762172698975,
      "learning_rate": 0.002828,
      "loss": 1.699,
      "step": 89408
    },
    {
      "epoch": 6.714596622889306,
      "grad_norm": 0.5269412994384766,
      "learning_rate": 0.002828,
      "loss": 1.6941,
      "step": 89472
    },
    {
      "epoch": 6.719399624765479,
      "grad_norm": 0.5359578728675842,
      "learning_rate": 0.002828,
      "loss": 1.6965,
      "step": 89536
    },
    {
      "epoch": 6.724202626641651,
      "grad_norm": 0.4400736093521118,
      "learning_rate": 0.002828,
      "loss": 1.6972,
      "step": 89600
    },
    {
      "epoch": 6.7290056285178235,
      "grad_norm": 0.4457342326641083,
      "learning_rate": 0.002828,
      "loss": 1.7058,
      "step": 89664
    },
    {
      "epoch": 6.733808630393996,
      "grad_norm": 0.5111649632453918,
      "learning_rate": 0.002828,
      "loss": 1.697,
      "step": 89728
    },
    {
      "epoch": 6.738611632270169,
      "grad_norm": 0.4815118908882141,
      "learning_rate": 0.002828,
      "loss": 1.6908,
      "step": 89792
    },
    {
      "epoch": 6.743414634146341,
      "grad_norm": 0.4404054284095764,
      "learning_rate": 0.002828,
      "loss": 1.6954,
      "step": 89856
    },
    {
      "epoch": 6.748217636022514,
      "grad_norm": 0.3914402425289154,
      "learning_rate": 0.002828,
      "loss": 1.6932,
      "step": 89920
    },
    {
      "epoch": 6.753020637898687,
      "grad_norm": 0.43506091833114624,
      "learning_rate": 0.002828,
      "loss": 1.6954,
      "step": 89984
    },
    {
      "epoch": 6.757823639774859,
      "grad_norm": 0.46764346957206726,
      "learning_rate": 0.002828,
      "loss": 1.6915,
      "step": 90048
    },
    {
      "epoch": 6.762626641651032,
      "grad_norm": 0.4379426836967468,
      "learning_rate": 0.002828,
      "loss": 1.6941,
      "step": 90112
    },
    {
      "epoch": 6.767429643527205,
      "grad_norm": 0.45644304156303406,
      "learning_rate": 0.002828,
      "loss": 1.6888,
      "step": 90176
    },
    {
      "epoch": 6.772232645403377,
      "grad_norm": 0.45382314920425415,
      "learning_rate": 0.002828,
      "loss": 1.6948,
      "step": 90240
    },
    {
      "epoch": 6.77703564727955,
      "grad_norm": 0.43029579520225525,
      "learning_rate": 0.002828,
      "loss": 1.6922,
      "step": 90304
    },
    {
      "epoch": 6.7818386491557225,
      "grad_norm": 0.47476446628570557,
      "learning_rate": 0.002828,
      "loss": 1.6933,
      "step": 90368
    },
    {
      "epoch": 6.7866416510318945,
      "grad_norm": 0.5024870038032532,
      "learning_rate": 0.002828,
      "loss": 1.6939,
      "step": 90432
    },
    {
      "epoch": 6.791444652908067,
      "grad_norm": 0.483139306306839,
      "learning_rate": 0.002828,
      "loss": 1.6956,
      "step": 90496
    },
    {
      "epoch": 6.79624765478424,
      "grad_norm": 0.4529098868370056,
      "learning_rate": 0.002828,
      "loss": 1.6897,
      "step": 90560
    },
    {
      "epoch": 6.801050656660413,
      "grad_norm": 0.4902280271053314,
      "learning_rate": 0.002828,
      "loss": 1.694,
      "step": 90624
    },
    {
      "epoch": 6.805853658536585,
      "grad_norm": 0.46275779604911804,
      "learning_rate": 0.002828,
      "loss": 1.6938,
      "step": 90688
    },
    {
      "epoch": 6.810656660412758,
      "grad_norm": 0.44121673703193665,
      "learning_rate": 0.002828,
      "loss": 1.6957,
      "step": 90752
    },
    {
      "epoch": 6.815459662288931,
      "grad_norm": 0.3859396278858185,
      "learning_rate": 0.002828,
      "loss": 1.6918,
      "step": 90816
    },
    {
      "epoch": 6.820262664165103,
      "grad_norm": 0.43591949343681335,
      "learning_rate": 0.002828,
      "loss": 1.6939,
      "step": 90880
    },
    {
      "epoch": 6.825065666041276,
      "grad_norm": 0.43440181016921997,
      "learning_rate": 0.002828,
      "loss": 1.6919,
      "step": 90944
    },
    {
      "epoch": 6.829868667917449,
      "grad_norm": 0.43394795060157776,
      "learning_rate": 0.002828,
      "loss": 1.6877,
      "step": 91008
    },
    {
      "epoch": 6.834671669793621,
      "grad_norm": 0.46919190883636475,
      "learning_rate": 0.002828,
      "loss": 1.687,
      "step": 91072
    },
    {
      "epoch": 6.839474671669794,
      "grad_norm": 0.4598900377750397,
      "learning_rate": 0.002828,
      "loss": 1.6905,
      "step": 91136
    },
    {
      "epoch": 6.8442776735459665,
      "grad_norm": 0.5242931246757507,
      "learning_rate": 0.002828,
      "loss": 1.6806,
      "step": 91200
    },
    {
      "epoch": 6.8490806754221385,
      "grad_norm": 0.43550997972488403,
      "learning_rate": 0.002828,
      "loss": 1.6966,
      "step": 91264
    },
    {
      "epoch": 6.853883677298311,
      "grad_norm": 0.5565023422241211,
      "learning_rate": 0.002828,
      "loss": 1.6903,
      "step": 91328
    },
    {
      "epoch": 6.858686679174484,
      "grad_norm": 0.4555063545703888,
      "learning_rate": 0.002828,
      "loss": 1.686,
      "step": 91392
    },
    {
      "epoch": 6.863489681050657,
      "grad_norm": 0.4962327480316162,
      "learning_rate": 0.002828,
      "loss": 1.69,
      "step": 91456
    },
    {
      "epoch": 6.868292682926829,
      "grad_norm": 0.3948613107204437,
      "learning_rate": 0.002828,
      "loss": 1.6863,
      "step": 91520
    },
    {
      "epoch": 6.873095684803002,
      "grad_norm": 0.4701470732688904,
      "learning_rate": 0.002828,
      "loss": 1.6849,
      "step": 91584
    },
    {
      "epoch": 6.877898686679174,
      "grad_norm": 0.5295090079307556,
      "learning_rate": 0.002828,
      "loss": 1.6957,
      "step": 91648
    },
    {
      "epoch": 6.882701688555347,
      "grad_norm": 0.5115393996238708,
      "learning_rate": 0.002828,
      "loss": 1.6951,
      "step": 91712
    },
    {
      "epoch": 6.88750469043152,
      "grad_norm": 0.4638088047504425,
      "learning_rate": 0.002828,
      "loss": 1.6871,
      "step": 91776
    },
    {
      "epoch": 6.892307692307693,
      "grad_norm": 0.46743154525756836,
      "learning_rate": 0.002828,
      "loss": 1.684,
      "step": 91840
    },
    {
      "epoch": 6.897110694183865,
      "grad_norm": 0.6080755591392517,
      "learning_rate": 0.002828,
      "loss": 1.6914,
      "step": 91904
    },
    {
      "epoch": 6.901913696060038,
      "grad_norm": 0.42485353350639343,
      "learning_rate": 0.002828,
      "loss": 1.6949,
      "step": 91968
    },
    {
      "epoch": 6.9067166979362105,
      "grad_norm": 0.45448240637779236,
      "learning_rate": 0.002828,
      "loss": 1.6943,
      "step": 92032
    },
    {
      "epoch": 6.9115196998123825,
      "grad_norm": 0.4901677072048187,
      "learning_rate": 0.002828,
      "loss": 1.6899,
      "step": 92096
    },
    {
      "epoch": 6.916322701688555,
      "grad_norm": 0.44569453597068787,
      "learning_rate": 0.002828,
      "loss": 1.6897,
      "step": 92160
    },
    {
      "epoch": 6.921125703564728,
      "grad_norm": 0.47087913751602173,
      "learning_rate": 0.002828,
      "loss": 1.6911,
      "step": 92224
    },
    {
      "epoch": 6.9259287054409,
      "grad_norm": 0.4655872881412506,
      "learning_rate": 0.002828,
      "loss": 1.6844,
      "step": 92288
    },
    {
      "epoch": 6.930731707317073,
      "grad_norm": 0.41938281059265137,
      "learning_rate": 0.002828,
      "loss": 1.689,
      "step": 92352
    },
    {
      "epoch": 6.935534709193246,
      "grad_norm": 0.4859105050563812,
      "learning_rate": 0.002828,
      "loss": 1.6864,
      "step": 92416
    },
    {
      "epoch": 6.940337711069418,
      "grad_norm": 0.5081743597984314,
      "learning_rate": 0.002828,
      "loss": 1.6937,
      "step": 92480
    },
    {
      "epoch": 6.945140712945591,
      "grad_norm": 0.5057895183563232,
      "learning_rate": 0.002828,
      "loss": 1.6844,
      "step": 92544
    },
    {
      "epoch": 6.949943714821764,
      "grad_norm": 0.4937455952167511,
      "learning_rate": 0.002828,
      "loss": 1.6875,
      "step": 92608
    },
    {
      "epoch": 6.954746716697937,
      "grad_norm": 0.4866175353527069,
      "learning_rate": 0.002828,
      "loss": 1.6933,
      "step": 92672
    },
    {
      "epoch": 6.959549718574109,
      "grad_norm": 0.5158609747886658,
      "learning_rate": 0.002828,
      "loss": 1.6874,
      "step": 92736
    },
    {
      "epoch": 6.9643527204502815,
      "grad_norm": 0.5342572331428528,
      "learning_rate": 0.002828,
      "loss": 1.6811,
      "step": 92800
    },
    {
      "epoch": 6.969155722326454,
      "grad_norm": 0.45014888048171997,
      "learning_rate": 0.002828,
      "loss": 1.6901,
      "step": 92864
    },
    {
      "epoch": 6.973958724202626,
      "grad_norm": 0.43926528096199036,
      "learning_rate": 0.002828,
      "loss": 1.6906,
      "step": 92928
    },
    {
      "epoch": 6.978761726078799,
      "grad_norm": 0.4542122781276703,
      "learning_rate": 0.002828,
      "loss": 1.6865,
      "step": 92992
    },
    {
      "epoch": 6.983564727954972,
      "grad_norm": 0.44546231627464294,
      "learning_rate": 0.002828,
      "loss": 1.6787,
      "step": 93056
    },
    {
      "epoch": 6.988367729831144,
      "grad_norm": 0.45137014985084534,
      "learning_rate": 0.002828,
      "loss": 1.6883,
      "step": 93120
    },
    {
      "epoch": 6.993170731707317,
      "grad_norm": 0.49241000413894653,
      "learning_rate": 0.002828,
      "loss": 1.6953,
      "step": 93184
    },
    {
      "epoch": 6.99797373358349,
      "grad_norm": 0.4922092854976654,
      "learning_rate": 0.002828,
      "loss": 1.6848,
      "step": 93248
    },
    {
      "epoch": 7.002776735459662,
      "grad_norm": 0.5105228424072266,
      "learning_rate": 0.002828,
      "loss": 1.6573,
      "step": 93312
    },
    {
      "epoch": 7.007579737335835,
      "grad_norm": 0.42724743485450745,
      "learning_rate": 0.002828,
      "loss": 1.6412,
      "step": 93376
    },
    {
      "epoch": 7.012382739212008,
      "grad_norm": 0.5028865933418274,
      "learning_rate": 0.002828,
      "loss": 1.6333,
      "step": 93440
    },
    {
      "epoch": 7.01718574108818,
      "grad_norm": 0.44145479798316956,
      "learning_rate": 0.002828,
      "loss": 1.6397,
      "step": 93504
    },
    {
      "epoch": 7.021988742964353,
      "grad_norm": 0.40729594230651855,
      "learning_rate": 0.002828,
      "loss": 1.6481,
      "step": 93568
    },
    {
      "epoch": 7.0267917448405255,
      "grad_norm": 0.4236595332622528,
      "learning_rate": 0.002828,
      "loss": 1.6464,
      "step": 93632
    },
    {
      "epoch": 7.031594746716698,
      "grad_norm": 0.4586251378059387,
      "learning_rate": 0.002828,
      "loss": 1.6449,
      "step": 93696
    },
    {
      "epoch": 7.03639774859287,
      "grad_norm": 0.44536247849464417,
      "learning_rate": 0.002828,
      "loss": 1.6446,
      "step": 93760
    },
    {
      "epoch": 7.041200750469043,
      "grad_norm": 0.49832186102867126,
      "learning_rate": 0.002828,
      "loss": 1.6446,
      "step": 93824
    },
    {
      "epoch": 7.046003752345216,
      "grad_norm": 0.49344515800476074,
      "learning_rate": 0.002828,
      "loss": 1.6442,
      "step": 93888
    },
    {
      "epoch": 7.050806754221388,
      "grad_norm": 0.4440404176712036,
      "learning_rate": 0.002828,
      "loss": 1.6487,
      "step": 93952
    },
    {
      "epoch": 7.055609756097561,
      "grad_norm": 0.5434808135032654,
      "learning_rate": 0.002828,
      "loss": 1.6508,
      "step": 94016
    },
    {
      "epoch": 7.060412757973734,
      "grad_norm": 0.4643784463405609,
      "learning_rate": 0.002828,
      "loss": 1.645,
      "step": 94080
    },
    {
      "epoch": 7.065215759849906,
      "grad_norm": 0.4471110999584198,
      "learning_rate": 0.002828,
      "loss": 1.6423,
      "step": 94144
    },
    {
      "epoch": 7.070018761726079,
      "grad_norm": 0.5672070980072021,
      "learning_rate": 0.002828,
      "loss": 1.6407,
      "step": 94208
    },
    {
      "epoch": 7.074821763602252,
      "grad_norm": 0.4709682762622833,
      "learning_rate": 0.002828,
      "loss": 1.6361,
      "step": 94272
    },
    {
      "epoch": 7.079624765478424,
      "grad_norm": 0.4074195921421051,
      "learning_rate": 0.002828,
      "loss": 1.6451,
      "step": 94336
    },
    {
      "epoch": 7.084427767354597,
      "grad_norm": 0.43681713938713074,
      "learning_rate": 0.002828,
      "loss": 1.6538,
      "step": 94400
    },
    {
      "epoch": 7.0892307692307694,
      "grad_norm": 0.47196078300476074,
      "learning_rate": 0.002828,
      "loss": 1.6479,
      "step": 94464
    },
    {
      "epoch": 7.094033771106941,
      "grad_norm": 0.3886224031448364,
      "learning_rate": 0.002828,
      "loss": 1.6552,
      "step": 94528
    },
    {
      "epoch": 7.098836772983114,
      "grad_norm": 0.47881880402565,
      "learning_rate": 0.002828,
      "loss": 1.6468,
      "step": 94592
    },
    {
      "epoch": 7.103639774859287,
      "grad_norm": 0.47139066457748413,
      "learning_rate": 0.002828,
      "loss": 1.6474,
      "step": 94656
    },
    {
      "epoch": 7.10844277673546,
      "grad_norm": 0.3880135118961334,
      "learning_rate": 0.002828,
      "loss": 1.6494,
      "step": 94720
    },
    {
      "epoch": 7.113245778611632,
      "grad_norm": 0.44806167483329773,
      "learning_rate": 0.002828,
      "loss": 1.6406,
      "step": 94784
    },
    {
      "epoch": 7.118048780487805,
      "grad_norm": 0.5234540104866028,
      "learning_rate": 0.002828,
      "loss": 1.6466,
      "step": 94848
    },
    {
      "epoch": 7.122851782363978,
      "grad_norm": 0.4456152021884918,
      "learning_rate": 0.002828,
      "loss": 1.6489,
      "step": 94912
    },
    {
      "epoch": 7.12765478424015,
      "grad_norm": 0.43684425950050354,
      "learning_rate": 0.002828,
      "loss": 1.6517,
      "step": 94976
    },
    {
      "epoch": 7.132457786116323,
      "grad_norm": 0.4596727788448334,
      "learning_rate": 0.002828,
      "loss": 1.6462,
      "step": 95040
    },
    {
      "epoch": 7.137260787992496,
      "grad_norm": 0.44558820128440857,
      "learning_rate": 0.002828,
      "loss": 1.6452,
      "step": 95104
    },
    {
      "epoch": 7.142063789868668,
      "grad_norm": 0.4541861116886139,
      "learning_rate": 0.002828,
      "loss": 1.6423,
      "step": 95168
    },
    {
      "epoch": 7.1468667917448405,
      "grad_norm": 0.6683451533317566,
      "learning_rate": 0.002828,
      "loss": 1.6444,
      "step": 95232
    },
    {
      "epoch": 7.151669793621013,
      "grad_norm": 0.5007615089416504,
      "learning_rate": 0.002828,
      "loss": 1.6455,
      "step": 95296
    },
    {
      "epoch": 7.156472795497185,
      "grad_norm": 0.5515764355659485,
      "learning_rate": 0.002828,
      "loss": 1.6496,
      "step": 95360
    },
    {
      "epoch": 7.161275797373358,
      "grad_norm": 0.460638165473938,
      "learning_rate": 0.002828,
      "loss": 1.6431,
      "step": 95424
    },
    {
      "epoch": 7.166078799249531,
      "grad_norm": 0.4183456003665924,
      "learning_rate": 0.002828,
      "loss": 1.6367,
      "step": 95488
    },
    {
      "epoch": 7.170881801125703,
      "grad_norm": 0.5132945775985718,
      "learning_rate": 0.002828,
      "loss": 1.6471,
      "step": 95552
    },
    {
      "epoch": 7.175684803001876,
      "grad_norm": 0.5141153931617737,
      "learning_rate": 0.002828,
      "loss": 1.6483,
      "step": 95616
    },
    {
      "epoch": 7.180487804878049,
      "grad_norm": 0.5110387802124023,
      "learning_rate": 0.002828,
      "loss": 1.647,
      "step": 95680
    },
    {
      "epoch": 7.185290806754221,
      "grad_norm": 0.5410206317901611,
      "learning_rate": 0.002828,
      "loss": 1.6478,
      "step": 95744
    },
    {
      "epoch": 7.190093808630394,
      "grad_norm": 0.43306368589401245,
      "learning_rate": 0.002828,
      "loss": 1.6453,
      "step": 95808
    },
    {
      "epoch": 7.194896810506567,
      "grad_norm": 0.4982653856277466,
      "learning_rate": 0.002828,
      "loss": 1.6492,
      "step": 95872
    },
    {
      "epoch": 7.19969981238274,
      "grad_norm": 0.47771182656288147,
      "learning_rate": 0.002828,
      "loss": 1.6522,
      "step": 95936
    },
    {
      "epoch": 7.204502814258912,
      "grad_norm": 0.5106894373893738,
      "learning_rate": 0.002828,
      "loss": 1.6449,
      "step": 96000
    },
    {
      "epoch": 7.2093058161350845,
      "grad_norm": 0.5163542628288269,
      "learning_rate": 0.002828,
      "loss": 1.646,
      "step": 96064
    },
    {
      "epoch": 7.214108818011257,
      "grad_norm": 0.48441585898399353,
      "learning_rate": 0.002828,
      "loss": 1.6508,
      "step": 96128
    },
    {
      "epoch": 7.218911819887429,
      "grad_norm": 0.4799892008304596,
      "learning_rate": 0.002828,
      "loss": 1.6429,
      "step": 96192
    },
    {
      "epoch": 7.223714821763602,
      "grad_norm": 0.48237788677215576,
      "learning_rate": 0.002828,
      "loss": 1.6502,
      "step": 96256
    },
    {
      "epoch": 7.228517823639775,
      "grad_norm": 0.43436750769615173,
      "learning_rate": 0.002828,
      "loss": 1.6486,
      "step": 96320
    },
    {
      "epoch": 7.233320825515947,
      "grad_norm": 0.48235654830932617,
      "learning_rate": 0.002828,
      "loss": 1.6516,
      "step": 96384
    },
    {
      "epoch": 7.23812382739212,
      "grad_norm": 0.4596686065196991,
      "learning_rate": 0.002828,
      "loss": 1.6482,
      "step": 96448
    },
    {
      "epoch": 7.242926829268293,
      "grad_norm": 0.509095311164856,
      "learning_rate": 0.002828,
      "loss": 1.6528,
      "step": 96512
    },
    {
      "epoch": 7.247729831144465,
      "grad_norm": 0.5236621499061584,
      "learning_rate": 0.002828,
      "loss": 1.6522,
      "step": 96576
    },
    {
      "epoch": 7.252532833020638,
      "grad_norm": 0.46608901023864746,
      "learning_rate": 0.002828,
      "loss": 1.6456,
      "step": 96640
    },
    {
      "epoch": 7.257335834896811,
      "grad_norm": 0.40260711312294006,
      "learning_rate": 0.002828,
      "loss": 1.6442,
      "step": 96704
    },
    {
      "epoch": 7.2621388367729836,
      "grad_norm": 0.4557930529117584,
      "learning_rate": 0.002828,
      "loss": 1.6447,
      "step": 96768
    },
    {
      "epoch": 7.2669418386491555,
      "grad_norm": 0.4857008457183838,
      "learning_rate": 0.002828,
      "loss": 1.645,
      "step": 96832
    },
    {
      "epoch": 7.271744840525328,
      "grad_norm": 0.47859156131744385,
      "learning_rate": 0.002828,
      "loss": 1.6446,
      "step": 96896
    },
    {
      "epoch": 7.276547842401501,
      "grad_norm": 0.4860116243362427,
      "learning_rate": 0.002828,
      "loss": 1.6451,
      "step": 96960
    },
    {
      "epoch": 7.281350844277673,
      "grad_norm": 0.47068920731544495,
      "learning_rate": 0.002828,
      "loss": 1.643,
      "step": 97024
    },
    {
      "epoch": 7.286153846153846,
      "grad_norm": 0.4627678394317627,
      "learning_rate": 0.002828,
      "loss": 1.6464,
      "step": 97088
    },
    {
      "epoch": 7.290956848030019,
      "grad_norm": 0.43063241243362427,
      "learning_rate": 0.002828,
      "loss": 1.6455,
      "step": 97152
    },
    {
      "epoch": 7.295759849906191,
      "grad_norm": 0.5483308434486389,
      "learning_rate": 0.002828,
      "loss": 1.6477,
      "step": 97216
    },
    {
      "epoch": 7.300562851782364,
      "grad_norm": 0.41557881236076355,
      "learning_rate": 0.002828,
      "loss": 1.6518,
      "step": 97280
    },
    {
      "epoch": 7.305365853658537,
      "grad_norm": 0.4560103416442871,
      "learning_rate": 0.002828,
      "loss": 1.6494,
      "step": 97344
    },
    {
      "epoch": 7.310168855534709,
      "grad_norm": 0.41694557666778564,
      "learning_rate": 0.002828,
      "loss": 1.6485,
      "step": 97408
    },
    {
      "epoch": 7.314971857410882,
      "grad_norm": 0.393020898103714,
      "learning_rate": 0.002828,
      "loss": 1.6483,
      "step": 97472
    },
    {
      "epoch": 7.319774859287055,
      "grad_norm": 0.4669319987297058,
      "learning_rate": 0.002828,
      "loss": 1.6502,
      "step": 97536
    },
    {
      "epoch": 7.324577861163227,
      "grad_norm": 0.4242900609970093,
      "learning_rate": 0.002828,
      "loss": 1.644,
      "step": 97600
    },
    {
      "epoch": 7.3293808630393995,
      "grad_norm": 0.4920247197151184,
      "learning_rate": 0.002828,
      "loss": 1.6453,
      "step": 97664
    },
    {
      "epoch": 7.334183864915572,
      "grad_norm": 0.46534204483032227,
      "learning_rate": 0.002828,
      "loss": 1.6506,
      "step": 97728
    },
    {
      "epoch": 7.338986866791744,
      "grad_norm": 0.4295295476913452,
      "learning_rate": 0.002828,
      "loss": 1.6429,
      "step": 97792
    },
    {
      "epoch": 7.343789868667917,
      "grad_norm": 0.6084052324295044,
      "learning_rate": 0.002828,
      "loss": 1.6414,
      "step": 97856
    },
    {
      "epoch": 7.34859287054409,
      "grad_norm": 0.5155935883522034,
      "learning_rate": 0.002828,
      "loss": 1.649,
      "step": 97920
    },
    {
      "epoch": 7.353395872420263,
      "grad_norm": 0.5151852369308472,
      "learning_rate": 0.002828,
      "loss": 1.6484,
      "step": 97984
    },
    {
      "epoch": 7.358198874296435,
      "grad_norm": 0.4509267210960388,
      "learning_rate": 0.002828,
      "loss": 1.6504,
      "step": 98048
    },
    {
      "epoch": 7.363001876172608,
      "grad_norm": 0.4989111125469208,
      "learning_rate": 0.002828,
      "loss": 1.6384,
      "step": 98112
    },
    {
      "epoch": 7.367804878048781,
      "grad_norm": 0.5407853722572327,
      "learning_rate": 0.002828,
      "loss": 1.6462,
      "step": 98176
    },
    {
      "epoch": 7.372607879924953,
      "grad_norm": 0.48210641741752625,
      "learning_rate": 0.002828,
      "loss": 1.6422,
      "step": 98240
    },
    {
      "epoch": 7.377410881801126,
      "grad_norm": 0.4497006833553314,
      "learning_rate": 0.002828,
      "loss": 1.6476,
      "step": 98304
    },
    {
      "epoch": 7.382213883677299,
      "grad_norm": 0.5384613871574402,
      "learning_rate": 0.002828,
      "loss": 1.65,
      "step": 98368
    },
    {
      "epoch": 7.387016885553471,
      "grad_norm": 0.45183315873146057,
      "learning_rate": 0.002828,
      "loss": 1.6489,
      "step": 98432
    },
    {
      "epoch": 7.3918198874296435,
      "grad_norm": 0.5084384083747864,
      "learning_rate": 0.002828,
      "loss": 1.647,
      "step": 98496
    },
    {
      "epoch": 7.396622889305816,
      "grad_norm": 0.4499737620353699,
      "learning_rate": 0.002828,
      "loss": 1.6491,
      "step": 98560
    },
    {
      "epoch": 7.401425891181988,
      "grad_norm": 0.46009787917137146,
      "learning_rate": 0.002828,
      "loss": 1.6419,
      "step": 98624
    },
    {
      "epoch": 7.406228893058161,
      "grad_norm": 0.45367369055747986,
      "learning_rate": 0.002828,
      "loss": 1.651,
      "step": 98688
    },
    {
      "epoch": 7.411031894934334,
      "grad_norm": 0.44495901465415955,
      "learning_rate": 0.002828,
      "loss": 1.647,
      "step": 98752
    },
    {
      "epoch": 7.415834896810507,
      "grad_norm": 0.5723180770874023,
      "learning_rate": 0.002828,
      "loss": 1.6486,
      "step": 98816
    },
    {
      "epoch": 7.420637898686679,
      "grad_norm": 0.4274926483631134,
      "learning_rate": 0.002828,
      "loss": 1.6558,
      "step": 98880
    },
    {
      "epoch": 7.425440900562852,
      "grad_norm": 0.4769154489040375,
      "learning_rate": 0.002828,
      "loss": 1.6476,
      "step": 98944
    },
    {
      "epoch": 7.430243902439025,
      "grad_norm": 0.46415236592292786,
      "learning_rate": 0.002828,
      "loss": 1.6446,
      "step": 99008
    },
    {
      "epoch": 7.435046904315197,
      "grad_norm": 0.4672410488128662,
      "learning_rate": 0.002828,
      "loss": 1.6532,
      "step": 99072
    },
    {
      "epoch": 7.43984990619137,
      "grad_norm": 0.4316319227218628,
      "learning_rate": 0.002828,
      "loss": 1.6459,
      "step": 99136
    },
    {
      "epoch": 7.4446529080675425,
      "grad_norm": 0.4864663779735565,
      "learning_rate": 0.002828,
      "loss": 1.6427,
      "step": 99200
    },
    {
      "epoch": 7.4494559099437145,
      "grad_norm": 0.4411129951477051,
      "learning_rate": 0.002828,
      "loss": 1.6439,
      "step": 99264
    },
    {
      "epoch": 7.454258911819887,
      "grad_norm": 0.4902966618537903,
      "learning_rate": 0.002828,
      "loss": 1.6392,
      "step": 99328
    },
    {
      "epoch": 7.45906191369606,
      "grad_norm": 0.5182162523269653,
      "learning_rate": 0.002828,
      "loss": 1.6546,
      "step": 99392
    },
    {
      "epoch": 7.463864915572232,
      "grad_norm": 0.44750234484672546,
      "learning_rate": 0.002828,
      "loss": 1.641,
      "step": 99456
    },
    {
      "epoch": 7.468667917448405,
      "grad_norm": 0.646702229976654,
      "learning_rate": 0.002828,
      "loss": 1.65,
      "step": 99520
    },
    {
      "epoch": 7.473470919324578,
      "grad_norm": 0.4905090630054474,
      "learning_rate": 0.002828,
      "loss": 1.6444,
      "step": 99584
    },
    {
      "epoch": 7.47827392120075,
      "grad_norm": 0.5883878469467163,
      "learning_rate": 0.002828,
      "loss": 1.6424,
      "step": 99648
    },
    {
      "epoch": 7.483076923076923,
      "grad_norm": 0.5606529116630554,
      "learning_rate": 0.002828,
      "loss": 1.6474,
      "step": 99712
    },
    {
      "epoch": 7.487879924953096,
      "grad_norm": 0.47397708892822266,
      "learning_rate": 0.002828,
      "loss": 1.6432,
      "step": 99776
    },
    {
      "epoch": 7.492682926829268,
      "grad_norm": 0.47367072105407715,
      "learning_rate": 0.002828,
      "loss": 1.6422,
      "step": 99840
    },
    {
      "epoch": 7.497485928705441,
      "grad_norm": 0.4972234070301056,
      "learning_rate": 0.002828,
      "loss": 1.6483,
      "step": 99904
    },
    {
      "epoch": 7.502288930581614,
      "grad_norm": 0.5520415306091309,
      "learning_rate": 0.002828,
      "loss": 1.6409,
      "step": 99968
    },
    {
      "epoch": 7.5070919324577865,
      "grad_norm": 0.403537392616272,
      "learning_rate": 0.002828,
      "loss": 1.6456,
      "step": 100032
    },
    {
      "epoch": 7.5118949343339585,
      "grad_norm": 0.5851362347602844,
      "learning_rate": 0.002828,
      "loss": 1.6429,
      "step": 100096
    },
    {
      "epoch": 7.516697936210131,
      "grad_norm": 0.4804452359676361,
      "learning_rate": 0.002828,
      "loss": 1.6398,
      "step": 100160
    },
    {
      "epoch": 7.521500938086304,
      "grad_norm": 0.4467880427837372,
      "learning_rate": 0.002828,
      "loss": 1.6457,
      "step": 100224
    },
    {
      "epoch": 7.526303939962476,
      "grad_norm": 0.5438836216926575,
      "learning_rate": 0.002828,
      "loss": 1.6468,
      "step": 100288
    },
    {
      "epoch": 7.531106941838649,
      "grad_norm": 0.4456494450569153,
      "learning_rate": 0.002828,
      "loss": 1.6462,
      "step": 100352
    },
    {
      "epoch": 7.535909943714822,
      "grad_norm": 0.4625204801559448,
      "learning_rate": 0.002828,
      "loss": 1.6462,
      "step": 100416
    },
    {
      "epoch": 7.540712945590994,
      "grad_norm": 0.45541369915008545,
      "learning_rate": 0.002828,
      "loss": 1.6545,
      "step": 100480
    },
    {
      "epoch": 7.545515947467167,
      "grad_norm": 0.4680885374546051,
      "learning_rate": 0.002828,
      "loss": 1.6479,
      "step": 100544
    },
    {
      "epoch": 7.55031894934334,
      "grad_norm": 0.5363187789916992,
      "learning_rate": 0.002828,
      "loss": 1.6457,
      "step": 100608
    },
    {
      "epoch": 7.555121951219512,
      "grad_norm": 0.4436447024345398,
      "learning_rate": 0.002828,
      "loss": 1.6489,
      "step": 100672
    },
    {
      "epoch": 7.559924953095685,
      "grad_norm": 0.4831717014312744,
      "learning_rate": 0.002828,
      "loss": 1.6452,
      "step": 100736
    },
    {
      "epoch": 7.564727954971858,
      "grad_norm": 0.587341845035553,
      "learning_rate": 0.002828,
      "loss": 1.6511,
      "step": 100800
    },
    {
      "epoch": 7.5695309568480305,
      "grad_norm": 0.5343487858772278,
      "learning_rate": 0.002828,
      "loss": 1.6419,
      "step": 100864
    },
    {
      "epoch": 7.5743339587242025,
      "grad_norm": 0.5156888365745544,
      "learning_rate": 0.002828,
      "loss": 1.6443,
      "step": 100928
    },
    {
      "epoch": 7.579136960600375,
      "grad_norm": 0.4895589053630829,
      "learning_rate": 0.002828,
      "loss": 1.6489,
      "step": 100992
    },
    {
      "epoch": 7.583939962476548,
      "grad_norm": 0.4382045269012451,
      "learning_rate": 0.002828,
      "loss": 1.6381,
      "step": 101056
    },
    {
      "epoch": 7.58874296435272,
      "grad_norm": 0.5956616401672363,
      "learning_rate": 0.002828,
      "loss": 1.643,
      "step": 101120
    },
    {
      "epoch": 7.593545966228893,
      "grad_norm": 0.496944785118103,
      "learning_rate": 0.002828,
      "loss": 1.6496,
      "step": 101184
    },
    {
      "epoch": 7.598348968105066,
      "grad_norm": 0.5628007054328918,
      "learning_rate": 0.002828,
      "loss": 1.644,
      "step": 101248
    },
    {
      "epoch": 7.603151969981238,
      "grad_norm": 0.45873934030532837,
      "learning_rate": 0.002828,
      "loss": 1.6426,
      "step": 101312
    },
    {
      "epoch": 7.607954971857411,
      "grad_norm": 0.5054008960723877,
      "learning_rate": 0.002828,
      "loss": 1.6446,
      "step": 101376
    },
    {
      "epoch": 7.612757973733584,
      "grad_norm": 0.45213744044303894,
      "learning_rate": 0.002828,
      "loss": 1.6423,
      "step": 101440
    },
    {
      "epoch": 7.617560975609756,
      "grad_norm": 0.45042672753334045,
      "learning_rate": 0.002828,
      "loss": 1.6449,
      "step": 101504
    },
    {
      "epoch": 7.622363977485929,
      "grad_norm": 0.4785134494304657,
      "learning_rate": 0.002828,
      "loss": 1.6434,
      "step": 101568
    },
    {
      "epoch": 7.6271669793621015,
      "grad_norm": 0.47091761231422424,
      "learning_rate": 0.002828,
      "loss": 1.6417,
      "step": 101632
    },
    {
      "epoch": 7.631969981238274,
      "grad_norm": 0.5239785313606262,
      "learning_rate": 0.002828,
      "loss": 1.642,
      "step": 101696
    },
    {
      "epoch": 7.636772983114446,
      "grad_norm": 0.47115275263786316,
      "learning_rate": 0.002828,
      "loss": 1.6402,
      "step": 101760
    },
    {
      "epoch": 7.641575984990619,
      "grad_norm": 0.4137537181377411,
      "learning_rate": 0.002828,
      "loss": 1.645,
      "step": 101824
    },
    {
      "epoch": 7.646378986866791,
      "grad_norm": 0.45282185077667236,
      "learning_rate": 0.002828,
      "loss": 1.6422,
      "step": 101888
    },
    {
      "epoch": 7.651181988742964,
      "grad_norm": 0.47771942615509033,
      "learning_rate": 0.002828,
      "loss": 1.6416,
      "step": 101952
    },
    {
      "epoch": 7.655984990619137,
      "grad_norm": 0.4466339945793152,
      "learning_rate": 0.002828,
      "loss": 1.6432,
      "step": 102016
    },
    {
      "epoch": 7.66078799249531,
      "grad_norm": 0.5430694818496704,
      "learning_rate": 0.002828,
      "loss": 1.6379,
      "step": 102080
    },
    {
      "epoch": 7.665590994371482,
      "grad_norm": 0.6273244619369507,
      "learning_rate": 0.002828,
      "loss": 1.6511,
      "step": 102144
    },
    {
      "epoch": 7.670393996247655,
      "grad_norm": 0.5001136660575867,
      "learning_rate": 0.002828,
      "loss": 1.6378,
      "step": 102208
    },
    {
      "epoch": 7.675196998123828,
      "grad_norm": 0.48289889097213745,
      "learning_rate": 0.002828,
      "loss": 1.6464,
      "step": 102272
    },
    {
      "epoch": 7.68,
      "grad_norm": 0.38933464884757996,
      "learning_rate": 0.002828,
      "loss": 1.64,
      "step": 102336
    },
    {
      "epoch": 7.684803001876173,
      "grad_norm": 0.4248032867908478,
      "learning_rate": 0.002828,
      "loss": 1.6437,
      "step": 102400
    },
    {
      "epoch": 7.6896060037523455,
      "grad_norm": 0.4562476575374603,
      "learning_rate": 0.002828,
      "loss": 1.6469,
      "step": 102464
    },
    {
      "epoch": 7.6944090056285175,
      "grad_norm": 0.44395706057548523,
      "learning_rate": 0.002828,
      "loss": 1.6411,
      "step": 102528
    },
    {
      "epoch": 7.69921200750469,
      "grad_norm": 0.43343424797058105,
      "learning_rate": 0.002828,
      "loss": 1.6431,
      "step": 102592
    },
    {
      "epoch": 7.704015009380863,
      "grad_norm": 0.5182087421417236,
      "learning_rate": 0.002828,
      "loss": 1.6481,
      "step": 102656
    },
    {
      "epoch": 7.708818011257035,
      "grad_norm": 0.4748702943325043,
      "learning_rate": 0.002828,
      "loss": 1.6454,
      "step": 102720
    },
    {
      "epoch": 7.713621013133208,
      "grad_norm": 0.46305951476097107,
      "learning_rate": 0.002828,
      "loss": 1.6428,
      "step": 102784
    },
    {
      "epoch": 7.718424015009381,
      "grad_norm": 0.4430948495864868,
      "learning_rate": 0.002828,
      "loss": 1.6378,
      "step": 102848
    },
    {
      "epoch": 7.723227016885554,
      "grad_norm": 0.5415197610855103,
      "learning_rate": 0.002828,
      "loss": 1.6456,
      "step": 102912
    },
    {
      "epoch": 7.728030018761726,
      "grad_norm": 0.4879511594772339,
      "learning_rate": 0.002828,
      "loss": 1.6418,
      "step": 102976
    },
    {
      "epoch": 7.732833020637899,
      "grad_norm": 0.4881740212440491,
      "learning_rate": 0.002828,
      "loss": 1.6409,
      "step": 103040
    },
    {
      "epoch": 7.737636022514072,
      "grad_norm": 0.4493964910507202,
      "learning_rate": 0.002828,
      "loss": 1.6496,
      "step": 103104
    },
    {
      "epoch": 7.742439024390244,
      "grad_norm": 0.44141024351119995,
      "learning_rate": 0.002828,
      "loss": 1.6432,
      "step": 103168
    },
    {
      "epoch": 7.747242026266417,
      "grad_norm": 0.5163127183914185,
      "learning_rate": 0.002828,
      "loss": 1.6386,
      "step": 103232
    },
    {
      "epoch": 7.7520450281425894,
      "grad_norm": 0.46502265334129333,
      "learning_rate": 0.002828,
      "loss": 1.6484,
      "step": 103296
    },
    {
      "epoch": 7.756848030018761,
      "grad_norm": 0.4692935347557068,
      "learning_rate": 0.002828,
      "loss": 1.6417,
      "step": 103360
    },
    {
      "epoch": 7.761651031894934,
      "grad_norm": 0.5229098796844482,
      "learning_rate": 0.002828,
      "loss": 1.6402,
      "step": 103424
    },
    {
      "epoch": 7.766454033771107,
      "grad_norm": 0.4577805697917938,
      "learning_rate": 0.002828,
      "loss": 1.6413,
      "step": 103488
    },
    {
      "epoch": 7.771257035647279,
      "grad_norm": 0.4625105559825897,
      "learning_rate": 0.002828,
      "loss": 1.64,
      "step": 103552
    },
    {
      "epoch": 7.776060037523452,
      "grad_norm": 0.4869496524333954,
      "learning_rate": 0.002828,
      "loss": 1.631,
      "step": 103616
    },
    {
      "epoch": 7.780863039399625,
      "grad_norm": 0.5049871802330017,
      "learning_rate": 0.002828,
      "loss": 1.637,
      "step": 103680
    },
    {
      "epoch": 7.785666041275798,
      "grad_norm": 0.4739224910736084,
      "learning_rate": 0.002828,
      "loss": 1.6349,
      "step": 103744
    },
    {
      "epoch": 7.79046904315197,
      "grad_norm": 0.5248587727546692,
      "learning_rate": 0.002828,
      "loss": 1.6416,
      "step": 103808
    },
    {
      "epoch": 7.795272045028143,
      "grad_norm": 0.5122642517089844,
      "learning_rate": 0.002828,
      "loss": 1.6353,
      "step": 103872
    },
    {
      "epoch": 7.800075046904315,
      "grad_norm": 0.5079776048660278,
      "learning_rate": 0.002828,
      "loss": 1.6377,
      "step": 103936
    },
    {
      "epoch": 7.804878048780488,
      "grad_norm": 0.5826370120048523,
      "learning_rate": 0.002828,
      "loss": 1.6399,
      "step": 104000
    },
    {
      "epoch": 7.8096810506566605,
      "grad_norm": 0.4368886351585388,
      "learning_rate": 0.002828,
      "loss": 1.641,
      "step": 104064
    },
    {
      "epoch": 7.814484052532833,
      "grad_norm": 0.4432503581047058,
      "learning_rate": 0.002828,
      "loss": 1.6411,
      "step": 104128
    },
    {
      "epoch": 7.819287054409005,
      "grad_norm": 0.4679797291755676,
      "learning_rate": 0.002828,
      "loss": 1.639,
      "step": 104192
    },
    {
      "epoch": 7.824090056285178,
      "grad_norm": 0.44275757670402527,
      "learning_rate": 0.002828,
      "loss": 1.6344,
      "step": 104256
    },
    {
      "epoch": 7.828893058161351,
      "grad_norm": 0.5065768957138062,
      "learning_rate": 0.002828,
      "loss": 1.6364,
      "step": 104320
    },
    {
      "epoch": 7.833696060037523,
      "grad_norm": 0.5551817417144775,
      "learning_rate": 0.002828,
      "loss": 1.6428,
      "step": 104384
    },
    {
      "epoch": 7.838499061913696,
      "grad_norm": 0.5234085321426392,
      "learning_rate": 0.002828,
      "loss": 1.6422,
      "step": 104448
    },
    {
      "epoch": 7.843302063789869,
      "grad_norm": 0.4474858045578003,
      "learning_rate": 0.002828,
      "loss": 1.6349,
      "step": 104512
    },
    {
      "epoch": 7.848105065666041,
      "grad_norm": 0.41089287400245667,
      "learning_rate": 0.002828,
      "loss": 1.6374,
      "step": 104576
    },
    {
      "epoch": 7.852908067542214,
      "grad_norm": 0.5417289733886719,
      "learning_rate": 0.002828,
      "loss": 1.6416,
      "step": 104640
    },
    {
      "epoch": 7.857711069418387,
      "grad_norm": 0.492750883102417,
      "learning_rate": 0.002828,
      "loss": 1.636,
      "step": 104704
    },
    {
      "epoch": 7.862514071294559,
      "grad_norm": 0.5007115602493286,
      "learning_rate": 0.002828,
      "loss": 1.637,
      "step": 104768
    },
    {
      "epoch": 7.867317073170732,
      "grad_norm": 0.42121168971061707,
      "learning_rate": 0.002828,
      "loss": 1.64,
      "step": 104832
    },
    {
      "epoch": 7.8721200750469045,
      "grad_norm": 0.4751914143562317,
      "learning_rate": 0.002828,
      "loss": 1.643,
      "step": 104896
    },
    {
      "epoch": 7.876923076923077,
      "grad_norm": 0.4676034450531006,
      "learning_rate": 0.002828,
      "loss": 1.6444,
      "step": 104960
    },
    {
      "epoch": 7.881726078799249,
      "grad_norm": 0.42461201548576355,
      "learning_rate": 0.002828,
      "loss": 1.6382,
      "step": 105024
    },
    {
      "epoch": 7.886529080675422,
      "grad_norm": 0.41529977321624756,
      "learning_rate": 0.002828,
      "loss": 1.6321,
      "step": 105088
    },
    {
      "epoch": 7.891332082551595,
      "grad_norm": 0.5157459378242493,
      "learning_rate": 0.002828,
      "loss": 1.6278,
      "step": 105152
    },
    {
      "epoch": 7.896135084427767,
      "grad_norm": 0.386042058467865,
      "learning_rate": 0.002828,
      "loss": 1.6296,
      "step": 105216
    },
    {
      "epoch": 7.90093808630394,
      "grad_norm": 0.5516283512115479,
      "learning_rate": 0.002828,
      "loss": 1.6367,
      "step": 105280
    },
    {
      "epoch": 7.905741088180113,
      "grad_norm": 0.4407348930835724,
      "learning_rate": 0.002828,
      "loss": 1.632,
      "step": 105344
    },
    {
      "epoch": 7.910544090056285,
      "grad_norm": 0.511715829372406,
      "learning_rate": 0.002828,
      "loss": 1.6401,
      "step": 105408
    },
    {
      "epoch": 7.915347091932458,
      "grad_norm": 0.5613473057746887,
      "learning_rate": 0.002828,
      "loss": 1.6327,
      "step": 105472
    },
    {
      "epoch": 7.920150093808631,
      "grad_norm": 0.47922655940055847,
      "learning_rate": 0.002828,
      "loss": 1.6299,
      "step": 105536
    },
    {
      "epoch": 7.924953095684803,
      "grad_norm": 0.5997971296310425,
      "learning_rate": 0.002828,
      "loss": 1.6349,
      "step": 105600
    },
    {
      "epoch": 7.9297560975609755,
      "grad_norm": 0.4698721766471863,
      "learning_rate": 0.002828,
      "loss": 1.6397,
      "step": 105664
    },
    {
      "epoch": 7.934559099437148,
      "grad_norm": 0.42611223459243774,
      "learning_rate": 0.002828,
      "loss": 1.6408,
      "step": 105728
    },
    {
      "epoch": 7.939362101313321,
      "grad_norm": 0.5407405495643616,
      "learning_rate": 0.002828,
      "loss": 1.6364,
      "step": 105792
    },
    {
      "epoch": 7.944165103189493,
      "grad_norm": 0.4046322703361511,
      "learning_rate": 0.002828,
      "loss": 1.6384,
      "step": 105856
    },
    {
      "epoch": 7.948968105065666,
      "grad_norm": 0.4765555262565613,
      "learning_rate": 0.002828,
      "loss": 1.6422,
      "step": 105920
    },
    {
      "epoch": 7.953771106941838,
      "grad_norm": 0.4602759778499603,
      "learning_rate": 0.002828,
      "loss": 1.6348,
      "step": 105984
    },
    {
      "epoch": 7.958574108818011,
      "grad_norm": 0.42289867997169495,
      "learning_rate": 0.002828,
      "loss": 1.6329,
      "step": 106048
    },
    {
      "epoch": 7.963377110694184,
      "grad_norm": 0.4758918285369873,
      "learning_rate": 0.002828,
      "loss": 1.6373,
      "step": 106112
    },
    {
      "epoch": 7.968180112570357,
      "grad_norm": 0.543738603591919,
      "learning_rate": 0.002828,
      "loss": 1.6276,
      "step": 106176
    },
    {
      "epoch": 7.972983114446529,
      "grad_norm": 0.5426182150840759,
      "learning_rate": 0.002828,
      "loss": 1.6377,
      "step": 106240
    },
    {
      "epoch": 7.977786116322702,
      "grad_norm": 0.3883894681930542,
      "learning_rate": 0.002828,
      "loss": 1.6428,
      "step": 106304
    },
    {
      "epoch": 7.982589118198875,
      "grad_norm": 0.49923983216285706,
      "learning_rate": 0.002828,
      "loss": 1.6389,
      "step": 106368
    },
    {
      "epoch": 7.987392120075047,
      "grad_norm": 0.5695561766624451,
      "learning_rate": 0.002828,
      "loss": 1.6374,
      "step": 106432
    },
    {
      "epoch": 7.9921951219512195,
      "grad_norm": 0.5229542255401611,
      "learning_rate": 0.002828,
      "loss": 1.6402,
      "step": 106496
    },
    {
      "epoch": 7.996998123827392,
      "grad_norm": 0.5376026630401611,
      "learning_rate": 0.002828,
      "loss": 1.6382,
      "step": 106560
    },
    {
      "epoch": 8.001801125703565,
      "grad_norm": 0.4897994101047516,
      "learning_rate": 0.002828,
      "loss": 1.6212,
      "step": 106624
    },
    {
      "epoch": 8.006604127579738,
      "grad_norm": 0.5158925652503967,
      "learning_rate": 0.002828,
      "loss": 1.5909,
      "step": 106688
    },
    {
      "epoch": 8.01140712945591,
      "grad_norm": 0.4795593023300171,
      "learning_rate": 0.002828,
      "loss": 1.5891,
      "step": 106752
    },
    {
      "epoch": 8.016210131332082,
      "grad_norm": 0.4325263500213623,
      "learning_rate": 0.002828,
      "loss": 1.591,
      "step": 106816
    },
    {
      "epoch": 8.021013133208255,
      "grad_norm": 0.41282597184181213,
      "learning_rate": 0.002828,
      "loss": 1.59,
      "step": 106880
    },
    {
      "epoch": 8.025816135084428,
      "grad_norm": 0.5297383666038513,
      "learning_rate": 0.002828,
      "loss": 1.5954,
      "step": 106944
    },
    {
      "epoch": 8.0306191369606,
      "grad_norm": 0.5734117031097412,
      "learning_rate": 0.002828,
      "loss": 1.5855,
      "step": 107008
    },
    {
      "epoch": 8.035422138836774,
      "grad_norm": 0.49331584572792053,
      "learning_rate": 0.002828,
      "loss": 1.591,
      "step": 107072
    },
    {
      "epoch": 8.040225140712945,
      "grad_norm": 0.4974007308483124,
      "learning_rate": 0.002828,
      "loss": 1.5948,
      "step": 107136
    },
    {
      "epoch": 8.045028142589118,
      "grad_norm": 0.5249443650245667,
      "learning_rate": 0.002828,
      "loss": 1.5982,
      "step": 107200
    },
    {
      "epoch": 8.04983114446529,
      "grad_norm": 0.42286255955696106,
      "learning_rate": 0.002828,
      "loss": 1.5985,
      "step": 107264
    },
    {
      "epoch": 8.054634146341463,
      "grad_norm": 0.4468154311180115,
      "learning_rate": 0.002828,
      "loss": 1.5902,
      "step": 107328
    },
    {
      "epoch": 8.059437148217636,
      "grad_norm": 0.483468621969223,
      "learning_rate": 0.002828,
      "loss": 1.5882,
      "step": 107392
    },
    {
      "epoch": 8.06424015009381,
      "grad_norm": 0.43714627623558044,
      "learning_rate": 0.002828,
      "loss": 1.59,
      "step": 107456
    },
    {
      "epoch": 8.069043151969982,
      "grad_norm": 0.46889758110046387,
      "learning_rate": 0.002828,
      "loss": 1.5907,
      "step": 107520
    },
    {
      "epoch": 8.073846153846153,
      "grad_norm": 0.43405240774154663,
      "learning_rate": 0.002828,
      "loss": 1.593,
      "step": 107584
    },
    {
      "epoch": 8.078649155722326,
      "grad_norm": 0.45629021525382996,
      "learning_rate": 0.002828,
      "loss": 1.5964,
      "step": 107648
    },
    {
      "epoch": 8.083452157598499,
      "grad_norm": 0.46646854281425476,
      "learning_rate": 0.002828,
      "loss": 1.5921,
      "step": 107712
    },
    {
      "epoch": 8.088255159474672,
      "grad_norm": 0.4769529104232788,
      "learning_rate": 0.002828,
      "loss": 1.5984,
      "step": 107776
    },
    {
      "epoch": 8.093058161350845,
      "grad_norm": 0.4490543305873871,
      "learning_rate": 0.002828,
      "loss": 1.6025,
      "step": 107840
    },
    {
      "epoch": 8.097861163227018,
      "grad_norm": 0.5665892958641052,
      "learning_rate": 0.002828,
      "loss": 1.5988,
      "step": 107904
    },
    {
      "epoch": 8.102664165103189,
      "grad_norm": 0.4945269227027893,
      "learning_rate": 0.002828,
      "loss": 1.5919,
      "step": 107968
    },
    {
      "epoch": 8.107467166979362,
      "grad_norm": 0.579629123210907,
      "learning_rate": 0.002828,
      "loss": 1.594,
      "step": 108032
    },
    {
      "epoch": 8.112270168855535,
      "grad_norm": 0.5322190523147583,
      "learning_rate": 0.002828,
      "loss": 1.5918,
      "step": 108096
    },
    {
      "epoch": 8.117073170731707,
      "grad_norm": 0.4702974259853363,
      "learning_rate": 0.002828,
      "loss": 1.5897,
      "step": 108160
    },
    {
      "epoch": 8.12187617260788,
      "grad_norm": 0.5027482509613037,
      "learning_rate": 0.002828,
      "loss": 1.5974,
      "step": 108224
    },
    {
      "epoch": 8.126679174484053,
      "grad_norm": 0.43558311462402344,
      "learning_rate": 0.002828,
      "loss": 1.5913,
      "step": 108288
    },
    {
      "epoch": 8.131482176360224,
      "grad_norm": 0.5288829803466797,
      "learning_rate": 0.002828,
      "loss": 1.5986,
      "step": 108352
    },
    {
      "epoch": 8.136285178236397,
      "grad_norm": 0.4630937874317169,
      "learning_rate": 0.002828,
      "loss": 1.599,
      "step": 108416
    },
    {
      "epoch": 8.14108818011257,
      "grad_norm": 0.44643405079841614,
      "learning_rate": 0.002828,
      "loss": 1.6024,
      "step": 108480
    },
    {
      "epoch": 8.145891181988743,
      "grad_norm": 0.4901682436466217,
      "learning_rate": 0.002828,
      "loss": 1.599,
      "step": 108544
    },
    {
      "epoch": 8.150694183864916,
      "grad_norm": 0.4672674536705017,
      "learning_rate": 0.002828,
      "loss": 1.5951,
      "step": 108608
    },
    {
      "epoch": 8.155497185741089,
      "grad_norm": 0.5036239624023438,
      "learning_rate": 0.002828,
      "loss": 1.5997,
      "step": 108672
    },
    {
      "epoch": 8.160300187617262,
      "grad_norm": 0.46045413613319397,
      "learning_rate": 0.002828,
      "loss": 1.5918,
      "step": 108736
    },
    {
      "epoch": 8.165103189493433,
      "grad_norm": 0.45920488238334656,
      "learning_rate": 0.002828,
      "loss": 1.597,
      "step": 108800
    },
    {
      "epoch": 8.169906191369606,
      "grad_norm": 0.5360845327377319,
      "learning_rate": 0.002828,
      "loss": 1.588,
      "step": 108864
    },
    {
      "epoch": 8.174709193245778,
      "grad_norm": 0.5119756460189819,
      "learning_rate": 0.002828,
      "loss": 1.5874,
      "step": 108928
    },
    {
      "epoch": 8.179512195121951,
      "grad_norm": 0.4600110948085785,
      "learning_rate": 0.002828,
      "loss": 1.6008,
      "step": 108992
    },
    {
      "epoch": 8.184315196998124,
      "grad_norm": 0.47242122888565063,
      "learning_rate": 0.002828,
      "loss": 1.6017,
      "step": 109056
    },
    {
      "epoch": 8.189118198874297,
      "grad_norm": 0.5298404097557068,
      "learning_rate": 0.002828,
      "loss": 1.595,
      "step": 109120
    },
    {
      "epoch": 8.193921200750468,
      "grad_norm": 0.5570024251937866,
      "learning_rate": 0.002828,
      "loss": 1.5949,
      "step": 109184
    },
    {
      "epoch": 8.198724202626641,
      "grad_norm": 0.46537068486213684,
      "learning_rate": 0.002828,
      "loss": 1.5901,
      "step": 109248
    },
    {
      "epoch": 8.203527204502814,
      "grad_norm": 0.5504530668258667,
      "learning_rate": 0.002828,
      "loss": 1.5961,
      "step": 109312
    },
    {
      "epoch": 8.208330206378987,
      "grad_norm": 0.5117718577384949,
      "learning_rate": 0.002828,
      "loss": 1.5986,
      "step": 109376
    },
    {
      "epoch": 8.21313320825516,
      "grad_norm": 0.45419740676879883,
      "learning_rate": 0.002828,
      "loss": 1.5995,
      "step": 109440
    },
    {
      "epoch": 8.217936210131333,
      "grad_norm": 0.5005888938903809,
      "learning_rate": 0.002828,
      "loss": 1.6025,
      "step": 109504
    },
    {
      "epoch": 8.222739212007504,
      "grad_norm": 0.5024480223655701,
      "learning_rate": 0.002828,
      "loss": 1.5961,
      "step": 109568
    },
    {
      "epoch": 8.227542213883677,
      "grad_norm": 0.5216082334518433,
      "learning_rate": 0.002828,
      "loss": 1.6023,
      "step": 109632
    },
    {
      "epoch": 8.23234521575985,
      "grad_norm": 0.4577321708202362,
      "learning_rate": 0.002828,
      "loss": 1.5976,
      "step": 109696
    },
    {
      "epoch": 8.237148217636022,
      "grad_norm": 0.5149946808815002,
      "learning_rate": 0.002828,
      "loss": 1.6038,
      "step": 109760
    },
    {
      "epoch": 8.241951219512195,
      "grad_norm": 0.43793120980262756,
      "learning_rate": 0.002828,
      "loss": 1.5975,
      "step": 109824
    },
    {
      "epoch": 8.246754221388368,
      "grad_norm": 0.45768973231315613,
      "learning_rate": 0.002828,
      "loss": 1.6025,
      "step": 109888
    },
    {
      "epoch": 8.251557223264541,
      "grad_norm": 0.5439128875732422,
      "learning_rate": 0.002828,
      "loss": 1.592,
      "step": 109952
    },
    {
      "epoch": 8.256360225140712,
      "grad_norm": 0.47533735632896423,
      "learning_rate": 0.002828,
      "loss": 1.6008,
      "step": 110016
    },
    {
      "epoch": 8.261163227016885,
      "grad_norm": 0.4820833206176758,
      "learning_rate": 0.002828,
      "loss": 1.5972,
      "step": 110080
    },
    {
      "epoch": 8.265966228893058,
      "grad_norm": 0.5155043005943298,
      "learning_rate": 0.002828,
      "loss": 1.5898,
      "step": 110144
    },
    {
      "epoch": 8.27076923076923,
      "grad_norm": 0.5212211012840271,
      "learning_rate": 0.002828,
      "loss": 1.5986,
      "step": 110208
    },
    {
      "epoch": 8.275572232645404,
      "grad_norm": 0.42652082443237305,
      "learning_rate": 0.002828,
      "loss": 1.6001,
      "step": 110272
    },
    {
      "epoch": 8.280375234521577,
      "grad_norm": 0.4728452265262604,
      "learning_rate": 0.002828,
      "loss": 1.5949,
      "step": 110336
    },
    {
      "epoch": 8.285178236397748,
      "grad_norm": 0.496060311794281,
      "learning_rate": 0.002828,
      "loss": 1.6013,
      "step": 110400
    },
    {
      "epoch": 8.28998123827392,
      "grad_norm": 0.4925978481769562,
      "learning_rate": 0.002828,
      "loss": 1.5957,
      "step": 110464
    },
    {
      "epoch": 8.294784240150094,
      "grad_norm": 0.4829269051551819,
      "learning_rate": 0.002828,
      "loss": 1.5999,
      "step": 110528
    },
    {
      "epoch": 8.299587242026266,
      "grad_norm": 0.509156346321106,
      "learning_rate": 0.002828,
      "loss": 1.5939,
      "step": 110592
    },
    {
      "epoch": 8.30439024390244,
      "grad_norm": 0.4902212619781494,
      "learning_rate": 0.002828,
      "loss": 1.5992,
      "step": 110656
    },
    {
      "epoch": 8.309193245778612,
      "grad_norm": 0.50597083568573,
      "learning_rate": 0.002828,
      "loss": 1.6017,
      "step": 110720
    },
    {
      "epoch": 8.313996247654785,
      "grad_norm": 0.519462525844574,
      "learning_rate": 0.002828,
      "loss": 1.5987,
      "step": 110784
    },
    {
      "epoch": 8.318799249530956,
      "grad_norm": 0.5620601773262024,
      "learning_rate": 0.002828,
      "loss": 1.5972,
      "step": 110848
    },
    {
      "epoch": 8.323602251407129,
      "grad_norm": 0.5269021391868591,
      "learning_rate": 0.002828,
      "loss": 1.6045,
      "step": 110912
    },
    {
      "epoch": 8.328405253283302,
      "grad_norm": 0.4557552635669708,
      "learning_rate": 0.002828,
      "loss": 1.5982,
      "step": 110976
    },
    {
      "epoch": 8.333208255159475,
      "grad_norm": 0.4556206464767456,
      "learning_rate": 0.002828,
      "loss": 1.5969,
      "step": 111040
    },
    {
      "epoch": 8.338011257035648,
      "grad_norm": 0.434209942817688,
      "learning_rate": 0.002828,
      "loss": 1.5939,
      "step": 111104
    },
    {
      "epoch": 8.34281425891182,
      "grad_norm": 0.44927170872688293,
      "learning_rate": 0.002828,
      "loss": 1.6014,
      "step": 111168
    },
    {
      "epoch": 8.347617260787992,
      "grad_norm": 0.5330567359924316,
      "learning_rate": 0.002828,
      "loss": 1.601,
      "step": 111232
    },
    {
      "epoch": 8.352420262664165,
      "grad_norm": 0.4359879195690155,
      "learning_rate": 0.002828,
      "loss": 1.6015,
      "step": 111296
    },
    {
      "epoch": 8.357223264540337,
      "grad_norm": 0.5834143161773682,
      "learning_rate": 0.002828,
      "loss": 1.5954,
      "step": 111360
    },
    {
      "epoch": 8.36202626641651,
      "grad_norm": 0.46430498361587524,
      "learning_rate": 0.002828,
      "loss": 1.5949,
      "step": 111424
    },
    {
      "epoch": 8.366829268292683,
      "grad_norm": 0.4292107820510864,
      "learning_rate": 0.002828,
      "loss": 1.5992,
      "step": 111488
    },
    {
      "epoch": 8.371632270168856,
      "grad_norm": 0.49093276262283325,
      "learning_rate": 0.002828,
      "loss": 1.5946,
      "step": 111552
    },
    {
      "epoch": 8.376435272045029,
      "grad_norm": 0.4749935567378998,
      "learning_rate": 0.002828,
      "loss": 1.5964,
      "step": 111616
    },
    {
      "epoch": 8.3812382739212,
      "grad_norm": 0.5645825862884521,
      "learning_rate": 0.002828,
      "loss": 1.5981,
      "step": 111680
    },
    {
      "epoch": 8.386041275797373,
      "grad_norm": 0.5568863749504089,
      "learning_rate": 0.002828,
      "loss": 1.6017,
      "step": 111744
    },
    {
      "epoch": 8.390844277673546,
      "grad_norm": 0.5183071494102478,
      "learning_rate": 0.002828,
      "loss": 1.5968,
      "step": 111808
    },
    {
      "epoch": 8.395647279549719,
      "grad_norm": 0.4693017601966858,
      "learning_rate": 0.002828,
      "loss": 1.5932,
      "step": 111872
    },
    {
      "epoch": 8.400450281425892,
      "grad_norm": 0.6434198617935181,
      "learning_rate": 0.002828,
      "loss": 1.5973,
      "step": 111936
    },
    {
      "epoch": 8.405253283302065,
      "grad_norm": 0.45251399278640747,
      "learning_rate": 0.002828,
      "loss": 1.5982,
      "step": 112000
    },
    {
      "epoch": 8.410056285178236,
      "grad_norm": 0.5282191634178162,
      "learning_rate": 0.002828,
      "loss": 1.5974,
      "step": 112064
    },
    {
      "epoch": 8.414859287054409,
      "grad_norm": 0.5637361407279968,
      "learning_rate": 0.002828,
      "loss": 1.6016,
      "step": 112128
    },
    {
      "epoch": 8.419662288930581,
      "grad_norm": 0.5421961545944214,
      "learning_rate": 0.002828,
      "loss": 1.5968,
      "step": 112192
    },
    {
      "epoch": 8.424465290806754,
      "grad_norm": 0.45800650119781494,
      "learning_rate": 0.002828,
      "loss": 1.6019,
      "step": 112256
    },
    {
      "epoch": 8.429268292682927,
      "grad_norm": 0.5147725939750671,
      "learning_rate": 0.002828,
      "loss": 1.5966,
      "step": 112320
    },
    {
      "epoch": 8.4340712945591,
      "grad_norm": 0.5175865888595581,
      "learning_rate": 0.002828,
      "loss": 1.6023,
      "step": 112384
    },
    {
      "epoch": 8.438874296435271,
      "grad_norm": 0.48384368419647217,
      "learning_rate": 0.002828,
      "loss": 1.5909,
      "step": 112448
    },
    {
      "epoch": 8.443677298311444,
      "grad_norm": 0.49091124534606934,
      "learning_rate": 0.002828,
      "loss": 1.5965,
      "step": 112512
    },
    {
      "epoch": 8.448480300187617,
      "grad_norm": 0.5016306042671204,
      "learning_rate": 0.002828,
      "loss": 1.592,
      "step": 112576
    },
    {
      "epoch": 8.45328330206379,
      "grad_norm": 0.49617043137550354,
      "learning_rate": 0.002828,
      "loss": 1.6,
      "step": 112640
    },
    {
      "epoch": 8.458086303939963,
      "grad_norm": 0.5295305848121643,
      "learning_rate": 0.002828,
      "loss": 1.6006,
      "step": 112704
    },
    {
      "epoch": 8.462889305816136,
      "grad_norm": 0.48764100670814514,
      "learning_rate": 0.002828,
      "loss": 1.5977,
      "step": 112768
    },
    {
      "epoch": 8.467692307692309,
      "grad_norm": 0.5215612053871155,
      "learning_rate": 0.002828,
      "loss": 1.5955,
      "step": 112832
    },
    {
      "epoch": 8.47249530956848,
      "grad_norm": 0.4538777768611908,
      "learning_rate": 0.002828,
      "loss": 1.5948,
      "step": 112896
    },
    {
      "epoch": 8.477298311444653,
      "grad_norm": 0.44952964782714844,
      "learning_rate": 0.002828,
      "loss": 1.5979,
      "step": 112960
    },
    {
      "epoch": 8.482101313320825,
      "grad_norm": 0.4811982214450836,
      "learning_rate": 0.002828,
      "loss": 1.594,
      "step": 113024
    },
    {
      "epoch": 8.486904315196998,
      "grad_norm": 0.4790002703666687,
      "learning_rate": 0.002828,
      "loss": 1.6019,
      "step": 113088
    },
    {
      "epoch": 8.491707317073171,
      "grad_norm": 0.475006639957428,
      "learning_rate": 0.002828,
      "loss": 1.597,
      "step": 113152
    },
    {
      "epoch": 8.496510318949344,
      "grad_norm": 0.5283963680267334,
      "learning_rate": 0.002828,
      "loss": 1.6016,
      "step": 113216
    },
    {
      "epoch": 8.501313320825515,
      "grad_norm": 0.4370746612548828,
      "learning_rate": 0.002828,
      "loss": 1.5985,
      "step": 113280
    },
    {
      "epoch": 8.506116322701688,
      "grad_norm": 0.4338639974594116,
      "learning_rate": 0.002828,
      "loss": 1.5985,
      "step": 113344
    },
    {
      "epoch": 8.510919324577861,
      "grad_norm": 0.47683221101760864,
      "learning_rate": 0.002828,
      "loss": 1.6004,
      "step": 113408
    },
    {
      "epoch": 8.515722326454034,
      "grad_norm": 0.5058532357215881,
      "learning_rate": 0.002828,
      "loss": 1.5979,
      "step": 113472
    },
    {
      "epoch": 8.520525328330207,
      "grad_norm": 0.5377301573753357,
      "learning_rate": 0.002828,
      "loss": 1.6062,
      "step": 113536
    },
    {
      "epoch": 8.52532833020638,
      "grad_norm": 0.4185563921928406,
      "learning_rate": 0.002828,
      "loss": 1.5993,
      "step": 113600
    },
    {
      "epoch": 8.53013133208255,
      "grad_norm": 0.48849865794181824,
      "learning_rate": 0.002828,
      "loss": 1.5992,
      "step": 113664
    },
    {
      "epoch": 8.534934333958724,
      "grad_norm": 0.5468032956123352,
      "learning_rate": 0.002828,
      "loss": 1.5979,
      "step": 113728
    },
    {
      "epoch": 8.539737335834896,
      "grad_norm": 0.45104995369911194,
      "learning_rate": 0.002828,
      "loss": 1.5952,
      "step": 113792
    },
    {
      "epoch": 8.54454033771107,
      "grad_norm": 0.4612830877304077,
      "learning_rate": 0.002828,
      "loss": 1.602,
      "step": 113856
    },
    {
      "epoch": 8.549343339587242,
      "grad_norm": 0.5112800598144531,
      "learning_rate": 0.002828,
      "loss": 1.5954,
      "step": 113920
    },
    {
      "epoch": 8.554146341463415,
      "grad_norm": 0.4625680446624756,
      "learning_rate": 0.002828,
      "loss": 1.5951,
      "step": 113984
    },
    {
      "epoch": 8.558949343339588,
      "grad_norm": 0.4750332236289978,
      "learning_rate": 0.002828,
      "loss": 1.5939,
      "step": 114048
    },
    {
      "epoch": 8.56375234521576,
      "grad_norm": 0.5651158690452576,
      "learning_rate": 0.002828,
      "loss": 1.5946,
      "step": 114112
    },
    {
      "epoch": 8.568555347091932,
      "grad_norm": 0.5241173505783081,
      "learning_rate": 0.002828,
      "loss": 1.5867,
      "step": 114176
    },
    {
      "epoch": 8.573358348968105,
      "grad_norm": 0.42227280139923096,
      "learning_rate": 0.002828,
      "loss": 1.5895,
      "step": 114240
    },
    {
      "epoch": 8.578161350844278,
      "grad_norm": 0.511060893535614,
      "learning_rate": 0.002828,
      "loss": 1.5994,
      "step": 114304
    },
    {
      "epoch": 8.58296435272045,
      "grad_norm": 0.49274709820747375,
      "learning_rate": 0.002828,
      "loss": 1.5916,
      "step": 114368
    },
    {
      "epoch": 8.587767354596624,
      "grad_norm": 0.5100725293159485,
      "learning_rate": 0.002828,
      "loss": 1.5998,
      "step": 114432
    },
    {
      "epoch": 8.592570356472795,
      "grad_norm": 0.5600299835205078,
      "learning_rate": 0.002828,
      "loss": 1.5997,
      "step": 114496
    },
    {
      "epoch": 8.597373358348968,
      "grad_norm": 0.4198225736618042,
      "learning_rate": 0.002828,
      "loss": 1.5999,
      "step": 114560
    },
    {
      "epoch": 8.60217636022514,
      "grad_norm": 0.520063579082489,
      "learning_rate": 0.002828,
      "loss": 1.595,
      "step": 114624
    },
    {
      "epoch": 8.606979362101313,
      "grad_norm": 0.41708284616470337,
      "learning_rate": 0.002828,
      "loss": 1.5995,
      "step": 114688
    },
    {
      "epoch": 8.611782363977486,
      "grad_norm": 0.4914540648460388,
      "learning_rate": 0.002828,
      "loss": 1.5882,
      "step": 114752
    },
    {
      "epoch": 8.616585365853659,
      "grad_norm": 0.451637327671051,
      "learning_rate": 0.002828,
      "loss": 1.5981,
      "step": 114816
    },
    {
      "epoch": 8.621388367729832,
      "grad_norm": 0.42104360461235046,
      "learning_rate": 0.002828,
      "loss": 1.5989,
      "step": 114880
    },
    {
      "epoch": 8.626191369606003,
      "grad_norm": 0.46855536103248596,
      "learning_rate": 0.002828,
      "loss": 1.5929,
      "step": 114944
    },
    {
      "epoch": 8.630994371482176,
      "grad_norm": 0.5319945216178894,
      "learning_rate": 0.002828,
      "loss": 1.5986,
      "step": 115008
    },
    {
      "epoch": 8.635797373358349,
      "grad_norm": 0.5827271342277527,
      "learning_rate": 0.002828,
      "loss": 1.5942,
      "step": 115072
    },
    {
      "epoch": 8.640600375234522,
      "grad_norm": 0.4484277069568634,
      "learning_rate": 0.002828,
      "loss": 1.5949,
      "step": 115136
    },
    {
      "epoch": 8.645403377110695,
      "grad_norm": 0.46110737323760986,
      "learning_rate": 0.002828,
      "loss": 1.6058,
      "step": 115200
    },
    {
      "epoch": 8.650206378986868,
      "grad_norm": 0.4264478385448456,
      "learning_rate": 0.002828,
      "loss": 1.5897,
      "step": 115264
    },
    {
      "epoch": 8.655009380863039,
      "grad_norm": 0.46964988112449646,
      "learning_rate": 0.002828,
      "loss": 1.6006,
      "step": 115328
    },
    {
      "epoch": 8.659812382739212,
      "grad_norm": 0.5994449853897095,
      "learning_rate": 0.002828,
      "loss": 1.5972,
      "step": 115392
    },
    {
      "epoch": 8.664615384615384,
      "grad_norm": 0.5398836135864258,
      "learning_rate": 0.002828,
      "loss": 1.5993,
      "step": 115456
    },
    {
      "epoch": 8.669418386491557,
      "grad_norm": 0.4390712082386017,
      "learning_rate": 0.002828,
      "loss": 1.5967,
      "step": 115520
    },
    {
      "epoch": 8.67422138836773,
      "grad_norm": 0.656513512134552,
      "learning_rate": 0.002828,
      "loss": 1.5974,
      "step": 115584
    },
    {
      "epoch": 8.679024390243903,
      "grad_norm": 0.4846341013908386,
      "learning_rate": 0.002828,
      "loss": 1.5978,
      "step": 115648
    },
    {
      "epoch": 8.683827392120076,
      "grad_norm": 0.5119456052780151,
      "learning_rate": 0.002828,
      "loss": 1.5852,
      "step": 115712
    },
    {
      "epoch": 8.688630393996247,
      "grad_norm": 0.40535804629325867,
      "learning_rate": 0.002828,
      "loss": 1.5917,
      "step": 115776
    },
    {
      "epoch": 8.69343339587242,
      "grad_norm": 0.49570637941360474,
      "learning_rate": 0.002828,
      "loss": 1.592,
      "step": 115840
    },
    {
      "epoch": 8.698236397748593,
      "grad_norm": 0.5806848406791687,
      "learning_rate": 0.002828,
      "loss": 1.5978,
      "step": 115904
    },
    {
      "epoch": 8.703039399624766,
      "grad_norm": 0.375515341758728,
      "learning_rate": 0.002828,
      "loss": 1.5944,
      "step": 115968
    },
    {
      "epoch": 8.707842401500939,
      "grad_norm": 0.4328213334083557,
      "learning_rate": 0.002828,
      "loss": 1.5898,
      "step": 116032
    },
    {
      "epoch": 8.712645403377111,
      "grad_norm": 0.4718143343925476,
      "learning_rate": 0.002828,
      "loss": 1.5986,
      "step": 116096
    },
    {
      "epoch": 8.717448405253283,
      "grad_norm": 0.5862587094306946,
      "learning_rate": 0.002828,
      "loss": 1.5922,
      "step": 116160
    },
    {
      "epoch": 8.722251407129455,
      "grad_norm": 0.4656484127044678,
      "learning_rate": 0.002828,
      "loss": 1.5934,
      "step": 116224
    },
    {
      "epoch": 8.727054409005628,
      "grad_norm": 0.541840672492981,
      "learning_rate": 0.002828,
      "loss": 1.5985,
      "step": 116288
    },
    {
      "epoch": 8.731857410881801,
      "grad_norm": 0.5951374769210815,
      "learning_rate": 0.002828,
      "loss": 1.5975,
      "step": 116352
    },
    {
      "epoch": 8.736660412757974,
      "grad_norm": 0.47397860884666443,
      "learning_rate": 0.002828,
      "loss": 1.5932,
      "step": 116416
    },
    {
      "epoch": 8.741463414634147,
      "grad_norm": 0.4907262325286865,
      "learning_rate": 0.002828,
      "loss": 1.5945,
      "step": 116480
    },
    {
      "epoch": 8.74626641651032,
      "grad_norm": 0.48710212111473083,
      "learning_rate": 0.002828,
      "loss": 1.5972,
      "step": 116544
    },
    {
      "epoch": 8.751069418386491,
      "grad_norm": 0.5130937099456787,
      "learning_rate": 0.002828,
      "loss": 1.5977,
      "step": 116608
    },
    {
      "epoch": 8.755872420262664,
      "grad_norm": 0.45630306005477905,
      "learning_rate": 0.002828,
      "loss": 1.597,
      "step": 116672
    },
    {
      "epoch": 8.760675422138837,
      "grad_norm": 0.44322487711906433,
      "learning_rate": 0.002828,
      "loss": 1.5982,
      "step": 116736
    },
    {
      "epoch": 8.76547842401501,
      "grad_norm": 0.7029752135276794,
      "learning_rate": 0.002828,
      "loss": 1.5926,
      "step": 116800
    },
    {
      "epoch": 8.770281425891183,
      "grad_norm": 0.5284256935119629,
      "learning_rate": 0.002828,
      "loss": 1.5958,
      "step": 116864
    },
    {
      "epoch": 8.775084427767355,
      "grad_norm": 0.5001572966575623,
      "learning_rate": 0.002828,
      "loss": 1.5955,
      "step": 116928
    },
    {
      "epoch": 8.779887429643527,
      "grad_norm": 0.5075249075889587,
      "learning_rate": 0.002828,
      "loss": 1.5963,
      "step": 116992
    },
    {
      "epoch": 8.7846904315197,
      "grad_norm": 0.4319472312927246,
      "learning_rate": 0.002828,
      "loss": 1.5997,
      "step": 117056
    },
    {
      "epoch": 8.789493433395872,
      "grad_norm": 0.501660168170929,
      "learning_rate": 0.002828,
      "loss": 1.5913,
      "step": 117120
    },
    {
      "epoch": 8.794296435272045,
      "grad_norm": 0.44797807931900024,
      "learning_rate": 0.002828,
      "loss": 1.5995,
      "step": 117184
    },
    {
      "epoch": 8.799099437148218,
      "grad_norm": 0.47948718070983887,
      "learning_rate": 0.002828,
      "loss": 1.5882,
      "step": 117248
    },
    {
      "epoch": 8.803902439024391,
      "grad_norm": 0.4581029713153839,
      "learning_rate": 0.002828,
      "loss": 1.5953,
      "step": 117312
    },
    {
      "epoch": 8.808705440900562,
      "grad_norm": 0.44964277744293213,
      "learning_rate": 0.002828,
      "loss": 1.5922,
      "step": 117376
    },
    {
      "epoch": 8.813508442776735,
      "grad_norm": 0.49619776010513306,
      "learning_rate": 0.002828,
      "loss": 1.5883,
      "step": 117440
    },
    {
      "epoch": 8.818311444652908,
      "grad_norm": 0.5036442279815674,
      "learning_rate": 0.002828,
      "loss": 1.5942,
      "step": 117504
    },
    {
      "epoch": 8.82311444652908,
      "grad_norm": 0.5327112078666687,
      "learning_rate": 0.002828,
      "loss": 1.5956,
      "step": 117568
    },
    {
      "epoch": 8.827917448405254,
      "grad_norm": 0.49304166436195374,
      "learning_rate": 0.002828,
      "loss": 1.5938,
      "step": 117632
    },
    {
      "epoch": 8.832720450281426,
      "grad_norm": 0.4130309820175171,
      "learning_rate": 0.002828,
      "loss": 1.5984,
      "step": 117696
    },
    {
      "epoch": 8.837523452157598,
      "grad_norm": 0.4643678367137909,
      "learning_rate": 0.002828,
      "loss": 1.6014,
      "step": 117760
    },
    {
      "epoch": 8.84232645403377,
      "grad_norm": 0.4973071813583374,
      "learning_rate": 0.002828,
      "loss": 1.5914,
      "step": 117824
    },
    {
      "epoch": 8.847129455909943,
      "grad_norm": 0.4967636466026306,
      "learning_rate": 0.002828,
      "loss": 1.5935,
      "step": 117888
    },
    {
      "epoch": 8.851932457786116,
      "grad_norm": 0.44560450315475464,
      "learning_rate": 0.002828,
      "loss": 1.5917,
      "step": 117952
    },
    {
      "epoch": 8.85673545966229,
      "grad_norm": 0.46391820907592773,
      "learning_rate": 0.002828,
      "loss": 1.5944,
      "step": 118016
    },
    {
      "epoch": 8.861538461538462,
      "grad_norm": 0.5275543928146362,
      "learning_rate": 0.002828,
      "loss": 1.6014,
      "step": 118080
    },
    {
      "epoch": 8.866341463414635,
      "grad_norm": 0.45606210827827454,
      "learning_rate": 0.002828,
      "loss": 1.5921,
      "step": 118144
    },
    {
      "epoch": 8.871144465290806,
      "grad_norm": 0.4854564964771271,
      "learning_rate": 0.002828,
      "loss": 1.5954,
      "step": 118208
    },
    {
      "epoch": 8.875947467166979,
      "grad_norm": 0.4536757171154022,
      "learning_rate": 0.002828,
      "loss": 1.5932,
      "step": 118272
    },
    {
      "epoch": 8.880750469043152,
      "grad_norm": 0.47313639521598816,
      "learning_rate": 0.002828,
      "loss": 1.5925,
      "step": 118336
    },
    {
      "epoch": 8.885553470919325,
      "grad_norm": 0.5030401945114136,
      "learning_rate": 0.002828,
      "loss": 1.5933,
      "step": 118400
    },
    {
      "epoch": 8.890356472795498,
      "grad_norm": 0.5216529369354248,
      "learning_rate": 0.002828,
      "loss": 1.5942,
      "step": 118464
    },
    {
      "epoch": 8.89515947467167,
      "grad_norm": 0.4898476302623749,
      "learning_rate": 0.002828,
      "loss": 1.5958,
      "step": 118528
    },
    {
      "epoch": 8.899962476547842,
      "grad_norm": 0.5011880397796631,
      "learning_rate": 0.002828,
      "loss": 1.5913,
      "step": 118592
    },
    {
      "epoch": 8.904765478424014,
      "grad_norm": 0.5021920800209045,
      "learning_rate": 0.002828,
      "loss": 1.59,
      "step": 118656
    },
    {
      "epoch": 8.909568480300187,
      "grad_norm": 0.48647958040237427,
      "learning_rate": 0.002828,
      "loss": 1.5982,
      "step": 118720
    },
    {
      "epoch": 8.91437148217636,
      "grad_norm": 0.5826468467712402,
      "learning_rate": 0.002828,
      "loss": 1.5871,
      "step": 118784
    },
    {
      "epoch": 8.919174484052533,
      "grad_norm": 0.49877214431762695,
      "learning_rate": 0.002828,
      "loss": 1.5945,
      "step": 118848
    },
    {
      "epoch": 8.923977485928706,
      "grad_norm": 0.6048728227615356,
      "learning_rate": 0.002828,
      "loss": 1.596,
      "step": 118912
    },
    {
      "epoch": 8.928780487804879,
      "grad_norm": 0.5304832458496094,
      "learning_rate": 0.002828,
      "loss": 1.5984,
      "step": 118976
    },
    {
      "epoch": 8.93358348968105,
      "grad_norm": 0.454377144575119,
      "learning_rate": 0.002828,
      "loss": 1.5853,
      "step": 119040
    },
    {
      "epoch": 8.938386491557223,
      "grad_norm": 0.509590208530426,
      "learning_rate": 0.002828,
      "loss": 1.5908,
      "step": 119104
    },
    {
      "epoch": 8.943189493433396,
      "grad_norm": 0.5293945074081421,
      "learning_rate": 0.002828,
      "loss": 1.592,
      "step": 119168
    },
    {
      "epoch": 8.947992495309569,
      "grad_norm": 0.4682133197784424,
      "learning_rate": 0.002828,
      "loss": 1.6001,
      "step": 119232
    },
    {
      "epoch": 8.952795497185742,
      "grad_norm": 0.5309058427810669,
      "learning_rate": 0.002828,
      "loss": 1.5966,
      "step": 119296
    },
    {
      "epoch": 8.957598499061914,
      "grad_norm": 0.4920218586921692,
      "learning_rate": 0.002828,
      "loss": 1.5979,
      "step": 119360
    },
    {
      "epoch": 8.962401500938086,
      "grad_norm": 0.4925355911254883,
      "learning_rate": 0.002828,
      "loss": 1.5893,
      "step": 119424
    },
    {
      "epoch": 8.967204502814258,
      "grad_norm": 0.4320846199989319,
      "learning_rate": 0.002828,
      "loss": 1.593,
      "step": 119488
    },
    {
      "epoch": 8.972007504690431,
      "grad_norm": 0.46012082695961,
      "learning_rate": 0.002828,
      "loss": 1.5946,
      "step": 119552
    },
    {
      "epoch": 8.976810506566604,
      "grad_norm": 0.47652629017829895,
      "learning_rate": 0.002828,
      "loss": 1.5938,
      "step": 119616
    },
    {
      "epoch": 8.981613508442777,
      "grad_norm": 0.5796999931335449,
      "learning_rate": 0.002828,
      "loss": 1.5926,
      "step": 119680
    },
    {
      "epoch": 8.98641651031895,
      "grad_norm": 0.41885271668434143,
      "learning_rate": 0.002828,
      "loss": 1.5879,
      "step": 119744
    },
    {
      "epoch": 8.991219512195123,
      "grad_norm": 0.43641573190689087,
      "learning_rate": 0.002828,
      "loss": 1.5838,
      "step": 119808
    },
    {
      "epoch": 8.996022514071294,
      "grad_norm": 0.4800099730491638,
      "learning_rate": 0.002828,
      "loss": 1.5821,
      "step": 119872
    },
    {
      "epoch": 9.000825515947467,
      "grad_norm": 0.49796926975250244,
      "learning_rate": 0.002828,
      "loss": 1.5808,
      "step": 119936
    },
    {
      "epoch": 9.00562851782364,
      "grad_norm": 0.5159605145454407,
      "learning_rate": 0.002828,
      "loss": 1.5448,
      "step": 120000
    },
    {
      "epoch": 9.010431519699813,
      "grad_norm": 0.48490002751350403,
      "learning_rate": 0.002828,
      "loss": 1.5543,
      "step": 120064
    },
    {
      "epoch": 9.015234521575985,
      "grad_norm": 0.45639586448669434,
      "learning_rate": 0.002828,
      "loss": 1.5462,
      "step": 120128
    },
    {
      "epoch": 9.020037523452158,
      "grad_norm": 0.4975225329399109,
      "learning_rate": 0.002828,
      "loss": 1.5506,
      "step": 120192
    },
    {
      "epoch": 9.02484052532833,
      "grad_norm": 0.4190879166126251,
      "learning_rate": 0.002828,
      "loss": 1.5452,
      "step": 120256
    },
    {
      "epoch": 9.029643527204502,
      "grad_norm": 0.42787736654281616,
      "learning_rate": 0.002828,
      "loss": 1.5464,
      "step": 120320
    },
    {
      "epoch": 9.034446529080675,
      "grad_norm": 0.5576650500297546,
      "learning_rate": 0.002828,
      "loss": 1.5484,
      "step": 120384
    },
    {
      "epoch": 9.039249530956848,
      "grad_norm": 0.4466994106769562,
      "learning_rate": 0.002828,
      "loss": 1.5424,
      "step": 120448
    },
    {
      "epoch": 9.044052532833021,
      "grad_norm": 0.5612980127334595,
      "learning_rate": 0.002828,
      "loss": 1.5489,
      "step": 120512
    },
    {
      "epoch": 9.048855534709194,
      "grad_norm": 0.48870643973350525,
      "learning_rate": 0.002828,
      "loss": 1.5475,
      "step": 120576
    },
    {
      "epoch": 9.053658536585365,
      "grad_norm": 0.47606512904167175,
      "learning_rate": 0.002828,
      "loss": 1.5477,
      "step": 120640
    },
    {
      "epoch": 9.058461538461538,
      "grad_norm": 0.4935826063156128,
      "learning_rate": 0.002828,
      "loss": 1.5471,
      "step": 120704
    },
    {
      "epoch": 9.06326454033771,
      "grad_norm": 0.49965599179267883,
      "learning_rate": 0.002828,
      "loss": 1.5419,
      "step": 120768
    },
    {
      "epoch": 9.068067542213884,
      "grad_norm": 0.4511246979236603,
      "learning_rate": 0.002828,
      "loss": 1.5498,
      "step": 120832
    },
    {
      "epoch": 9.072870544090057,
      "grad_norm": 0.4655289649963379,
      "learning_rate": 0.002828,
      "loss": 1.5493,
      "step": 120896
    },
    {
      "epoch": 9.07767354596623,
      "grad_norm": 0.48337608575820923,
      "learning_rate": 0.002828,
      "loss": 1.5566,
      "step": 120960
    },
    {
      "epoch": 9.082476547842402,
      "grad_norm": 0.47634145617485046,
      "learning_rate": 0.002828,
      "loss": 1.5547,
      "step": 121024
    },
    {
      "epoch": 9.087279549718573,
      "grad_norm": 0.5103233456611633,
      "learning_rate": 0.002828,
      "loss": 1.5515,
      "step": 121088
    },
    {
      "epoch": 9.092082551594746,
      "grad_norm": 0.47967302799224854,
      "learning_rate": 0.002828,
      "loss": 1.5536,
      "step": 121152
    },
    {
      "epoch": 9.09688555347092,
      "grad_norm": 0.4684443771839142,
      "learning_rate": 0.002828,
      "loss": 1.5507,
      "step": 121216
    },
    {
      "epoch": 9.101688555347092,
      "grad_norm": 0.513687014579773,
      "learning_rate": 0.002828,
      "loss": 1.5533,
      "step": 121280
    },
    {
      "epoch": 9.106491557223265,
      "grad_norm": 0.4781445264816284,
      "learning_rate": 0.002828,
      "loss": 1.5541,
      "step": 121344
    },
    {
      "epoch": 9.111294559099438,
      "grad_norm": 0.4334908723831177,
      "learning_rate": 0.002828,
      "loss": 1.5603,
      "step": 121408
    },
    {
      "epoch": 9.116097560975609,
      "grad_norm": 0.5581165552139282,
      "learning_rate": 0.002828,
      "loss": 1.5509,
      "step": 121472
    },
    {
      "epoch": 9.120900562851782,
      "grad_norm": 0.5921232104301453,
      "learning_rate": 0.002828,
      "loss": 1.5508,
      "step": 121536
    },
    {
      "epoch": 9.125703564727955,
      "grad_norm": 0.48528948426246643,
      "learning_rate": 0.002828,
      "loss": 1.5567,
      "step": 121600
    },
    {
      "epoch": 9.130506566604128,
      "grad_norm": 0.5249315500259399,
      "learning_rate": 0.002828,
      "loss": 1.5521,
      "step": 121664
    },
    {
      "epoch": 9.1353095684803,
      "grad_norm": 0.5910688638687134,
      "learning_rate": 0.002828,
      "loss": 1.5578,
      "step": 121728
    },
    {
      "epoch": 9.140112570356473,
      "grad_norm": 0.46241605281829834,
      "learning_rate": 0.002828,
      "loss": 1.5549,
      "step": 121792
    },
    {
      "epoch": 9.144915572232645,
      "grad_norm": 0.5226661562919617,
      "learning_rate": 0.002828,
      "loss": 1.551,
      "step": 121856
    },
    {
      "epoch": 9.149718574108817,
      "grad_norm": 0.4741944372653961,
      "learning_rate": 0.002828,
      "loss": 1.554,
      "step": 121920
    },
    {
      "epoch": 9.15452157598499,
      "grad_norm": 0.5220636129379272,
      "learning_rate": 0.002828,
      "loss": 1.555,
      "step": 121984
    },
    {
      "epoch": 9.159324577861163,
      "grad_norm": 0.5372878909111023,
      "learning_rate": 0.002828,
      "loss": 1.5505,
      "step": 122048
    },
    {
      "epoch": 9.164127579737336,
      "grad_norm": 0.5775820016860962,
      "learning_rate": 0.002828,
      "loss": 1.551,
      "step": 122112
    },
    {
      "epoch": 9.168930581613509,
      "grad_norm": 0.5545300841331482,
      "learning_rate": 0.002828,
      "loss": 1.5519,
      "step": 122176
    },
    {
      "epoch": 9.173733583489682,
      "grad_norm": 0.467183917760849,
      "learning_rate": 0.002828,
      "loss": 1.5539,
      "step": 122240
    },
    {
      "epoch": 9.178536585365853,
      "grad_norm": 0.4525934159755707,
      "learning_rate": 0.002828,
      "loss": 1.5543,
      "step": 122304
    },
    {
      "epoch": 9.183339587242026,
      "grad_norm": 0.4548516869544983,
      "learning_rate": 0.002828,
      "loss": 1.5475,
      "step": 122368
    },
    {
      "epoch": 9.188142589118199,
      "grad_norm": 0.5126798152923584,
      "learning_rate": 0.002828,
      "loss": 1.5585,
      "step": 122432
    },
    {
      "epoch": 9.192945590994372,
      "grad_norm": 0.4691389203071594,
      "learning_rate": 0.002828,
      "loss": 1.5574,
      "step": 122496
    },
    {
      "epoch": 9.197748592870544,
      "grad_norm": 0.4580743610858917,
      "learning_rate": 0.002828,
      "loss": 1.5553,
      "step": 122560
    },
    {
      "epoch": 9.202551594746717,
      "grad_norm": 0.5749825239181519,
      "learning_rate": 0.002828,
      "loss": 1.547,
      "step": 122624
    },
    {
      "epoch": 9.207354596622888,
      "grad_norm": 0.4706513285636902,
      "learning_rate": 0.002828,
      "loss": 1.5549,
      "step": 122688
    },
    {
      "epoch": 9.212157598499061,
      "grad_norm": 0.437346875667572,
      "learning_rate": 0.002828,
      "loss": 1.5565,
      "step": 122752
    },
    {
      "epoch": 9.216960600375234,
      "grad_norm": 0.5276687145233154,
      "learning_rate": 0.002828,
      "loss": 1.5549,
      "step": 122816
    },
    {
      "epoch": 9.221763602251407,
      "grad_norm": 0.49012863636016846,
      "learning_rate": 0.002828,
      "loss": 1.5511,
      "step": 122880
    },
    {
      "epoch": 9.22656660412758,
      "grad_norm": 0.5751110911369324,
      "learning_rate": 0.002828,
      "loss": 1.5493,
      "step": 122944
    },
    {
      "epoch": 9.231369606003753,
      "grad_norm": 0.48521631956100464,
      "learning_rate": 0.002828,
      "loss": 1.5529,
      "step": 123008
    },
    {
      "epoch": 9.236172607879926,
      "grad_norm": 0.5013362765312195,
      "learning_rate": 0.002828,
      "loss": 1.5571,
      "step": 123072
    },
    {
      "epoch": 9.240975609756097,
      "grad_norm": 0.4691832661628723,
      "learning_rate": 0.002828,
      "loss": 1.5601,
      "step": 123136
    },
    {
      "epoch": 9.24577861163227,
      "grad_norm": 0.41744399070739746,
      "learning_rate": 0.002828,
      "loss": 1.5555,
      "step": 123200
    },
    {
      "epoch": 9.250581613508443,
      "grad_norm": 0.5104173421859741,
      "learning_rate": 0.002828,
      "loss": 1.552,
      "step": 123264
    },
    {
      "epoch": 9.255384615384616,
      "grad_norm": 0.46932265162467957,
      "learning_rate": 0.002828,
      "loss": 1.5593,
      "step": 123328
    },
    {
      "epoch": 9.260187617260788,
      "grad_norm": 0.5633493065834045,
      "learning_rate": 0.002828,
      "loss": 1.5554,
      "step": 123392
    },
    {
      "epoch": 9.264990619136961,
      "grad_norm": 0.4229661226272583,
      "learning_rate": 0.002828,
      "loss": 1.5577,
      "step": 123456
    },
    {
      "epoch": 9.269793621013132,
      "grad_norm": 0.4221712648868561,
      "learning_rate": 0.002828,
      "loss": 1.5584,
      "step": 123520
    },
    {
      "epoch": 9.274596622889305,
      "grad_norm": 0.5246994495391846,
      "learning_rate": 0.002828,
      "loss": 1.56,
      "step": 123584
    },
    {
      "epoch": 9.279399624765478,
      "grad_norm": 0.49985742568969727,
      "learning_rate": 0.002828,
      "loss": 1.5539,
      "step": 123648
    },
    {
      "epoch": 9.284202626641651,
      "grad_norm": 0.529069721698761,
      "learning_rate": 0.002828,
      "loss": 1.559,
      "step": 123712
    },
    {
      "epoch": 9.289005628517824,
      "grad_norm": 0.5723833441734314,
      "learning_rate": 0.002828,
      "loss": 1.5469,
      "step": 123776
    },
    {
      "epoch": 9.293808630393997,
      "grad_norm": 0.4774170219898224,
      "learning_rate": 0.002828,
      "loss": 1.5562,
      "step": 123840
    },
    {
      "epoch": 9.29861163227017,
      "grad_norm": 0.4825045168399811,
      "learning_rate": 0.002828,
      "loss": 1.5616,
      "step": 123904
    },
    {
      "epoch": 9.30341463414634,
      "grad_norm": 0.49546265602111816,
      "learning_rate": 0.002828,
      "loss": 1.5528,
      "step": 123968
    },
    {
      "epoch": 9.308217636022514,
      "grad_norm": 0.48817411065101624,
      "learning_rate": 0.002828,
      "loss": 1.5557,
      "step": 124032
    },
    {
      "epoch": 9.313020637898687,
      "grad_norm": 0.4852955937385559,
      "learning_rate": 0.002828,
      "loss": 1.5556,
      "step": 124096
    },
    {
      "epoch": 9.31782363977486,
      "grad_norm": 0.605785071849823,
      "learning_rate": 0.002828,
      "loss": 1.557,
      "step": 124160
    },
    {
      "epoch": 9.322626641651032,
      "grad_norm": 0.5245658159255981,
      "learning_rate": 0.002828,
      "loss": 1.5571,
      "step": 124224
    },
    {
      "epoch": 9.327429643527205,
      "grad_norm": 0.4966484606266022,
      "learning_rate": 0.002828,
      "loss": 1.5491,
      "step": 124288
    },
    {
      "epoch": 9.332232645403376,
      "grad_norm": 0.5174903273582458,
      "learning_rate": 0.002828,
      "loss": 1.5584,
      "step": 124352
    },
    {
      "epoch": 9.33703564727955,
      "grad_norm": 0.5896949172019958,
      "learning_rate": 0.002828,
      "loss": 1.5501,
      "step": 124416
    },
    {
      "epoch": 9.341838649155722,
      "grad_norm": 0.5136840343475342,
      "learning_rate": 0.002828,
      "loss": 1.5584,
      "step": 124480
    },
    {
      "epoch": 9.346641651031895,
      "grad_norm": 0.4528697431087494,
      "learning_rate": 0.002828,
      "loss": 1.5581,
      "step": 124544
    },
    {
      "epoch": 9.351444652908068,
      "grad_norm": 0.6355164647102356,
      "learning_rate": 0.002828,
      "loss": 1.5514,
      "step": 124608
    },
    {
      "epoch": 9.35624765478424,
      "grad_norm": 0.5369568467140198,
      "learning_rate": 0.002828,
      "loss": 1.5552,
      "step": 124672
    },
    {
      "epoch": 9.361050656660412,
      "grad_norm": 0.4956488311290741,
      "learning_rate": 0.002828,
      "loss": 1.5487,
      "step": 124736
    },
    {
      "epoch": 9.365853658536585,
      "grad_norm": 0.5135923027992249,
      "learning_rate": 0.002828,
      "loss": 1.5601,
      "step": 124800
    },
    {
      "epoch": 9.370656660412758,
      "grad_norm": 0.4957681894302368,
      "learning_rate": 0.002828,
      "loss": 1.5616,
      "step": 124864
    },
    {
      "epoch": 9.37545966228893,
      "grad_norm": 0.48925191164016724,
      "learning_rate": 0.002828,
      "loss": 1.5578,
      "step": 124928
    },
    {
      "epoch": 9.380262664165103,
      "grad_norm": 0.6204767227172852,
      "learning_rate": 0.002828,
      "loss": 1.5657,
      "step": 124992
    },
    {
      "epoch": 9.385065666041276,
      "grad_norm": 0.49402710795402527,
      "learning_rate": 0.002828,
      "loss": 1.5544,
      "step": 125056
    },
    {
      "epoch": 9.38986866791745,
      "grad_norm": 0.4795926809310913,
      "learning_rate": 0.002828,
      "loss": 1.553,
      "step": 125120
    },
    {
      "epoch": 9.39467166979362,
      "grad_norm": 0.5284917950630188,
      "learning_rate": 0.002828,
      "loss": 1.555,
      "step": 125184
    },
    {
      "epoch": 9.399474671669793,
      "grad_norm": 0.47258827090263367,
      "learning_rate": 0.002828,
      "loss": 1.5555,
      "step": 125248
    },
    {
      "epoch": 9.404277673545966,
      "grad_norm": 0.4388088881969452,
      "learning_rate": 0.002828,
      "loss": 1.5554,
      "step": 125312
    },
    {
      "epoch": 9.409080675422139,
      "grad_norm": 0.5329185724258423,
      "learning_rate": 0.002828,
      "loss": 1.5545,
      "step": 125376
    },
    {
      "epoch": 9.413883677298312,
      "grad_norm": 0.47534602880477905,
      "learning_rate": 0.002828,
      "loss": 1.5517,
      "step": 125440
    },
    {
      "epoch": 9.418686679174485,
      "grad_norm": 0.5153233408927917,
      "learning_rate": 0.002828,
      "loss": 1.5531,
      "step": 125504
    },
    {
      "epoch": 9.423489681050656,
      "grad_norm": 0.49084314703941345,
      "learning_rate": 0.002828,
      "loss": 1.5623,
      "step": 125568
    },
    {
      "epoch": 9.428292682926829,
      "grad_norm": 0.46658453345298767,
      "learning_rate": 0.002828,
      "loss": 1.5582,
      "step": 125632
    },
    {
      "epoch": 9.433095684803002,
      "grad_norm": 0.5435781478881836,
      "learning_rate": 0.002828,
      "loss": 1.553,
      "step": 125696
    },
    {
      "epoch": 9.437898686679175,
      "grad_norm": 0.6206216812133789,
      "learning_rate": 0.002828,
      "loss": 1.5506,
      "step": 125760
    },
    {
      "epoch": 9.442701688555347,
      "grad_norm": 0.4667552411556244,
      "learning_rate": 0.002828,
      "loss": 1.5577,
      "step": 125824
    },
    {
      "epoch": 9.44750469043152,
      "grad_norm": 0.5482316613197327,
      "learning_rate": 0.002828,
      "loss": 1.5541,
      "step": 125888
    },
    {
      "epoch": 9.452307692307691,
      "grad_norm": 0.5336337685585022,
      "learning_rate": 0.002828,
      "loss": 1.5604,
      "step": 125952
    },
    {
      "epoch": 9.457110694183864,
      "grad_norm": 0.46397972106933594,
      "learning_rate": 0.002828,
      "loss": 1.5505,
      "step": 126016
    },
    {
      "epoch": 9.461913696060037,
      "grad_norm": 0.5540631413459778,
      "learning_rate": 0.002828,
      "loss": 1.5598,
      "step": 126080
    },
    {
      "epoch": 9.46671669793621,
      "grad_norm": 0.666083574295044,
      "learning_rate": 0.002828,
      "loss": 1.5603,
      "step": 126144
    },
    {
      "epoch": 9.471519699812383,
      "grad_norm": 0.5453064441680908,
      "learning_rate": 0.002828,
      "loss": 1.5511,
      "step": 126208
    },
    {
      "epoch": 9.476322701688556,
      "grad_norm": 0.5765575170516968,
      "learning_rate": 0.002828,
      "loss": 1.5521,
      "step": 126272
    },
    {
      "epoch": 9.481125703564729,
      "grad_norm": 0.4911395311355591,
      "learning_rate": 0.002828,
      "loss": 1.5536,
      "step": 126336
    },
    {
      "epoch": 9.4859287054409,
      "grad_norm": 0.5155625343322754,
      "learning_rate": 0.002828,
      "loss": 1.5509,
      "step": 126400
    },
    {
      "epoch": 9.490731707317073,
      "grad_norm": 0.4965815842151642,
      "learning_rate": 0.002828,
      "loss": 1.5475,
      "step": 126464
    },
    {
      "epoch": 9.495534709193246,
      "grad_norm": 0.54978346824646,
      "learning_rate": 0.002828,
      "loss": 1.5591,
      "step": 126528
    },
    {
      "epoch": 9.500337711069418,
      "grad_norm": 0.4493124485015869,
      "learning_rate": 0.002828,
      "loss": 1.5539,
      "step": 126592
    },
    {
      "epoch": 9.505140712945591,
      "grad_norm": 0.4517803490161896,
      "learning_rate": 0.002828,
      "loss": 1.5568,
      "step": 126656
    },
    {
      "epoch": 9.509943714821764,
      "grad_norm": 0.527080237865448,
      "learning_rate": 0.002828,
      "loss": 1.5565,
      "step": 126720
    },
    {
      "epoch": 9.514746716697935,
      "grad_norm": 0.41383326053619385,
      "learning_rate": 0.002828,
      "loss": 1.5559,
      "step": 126784
    },
    {
      "epoch": 9.519549718574108,
      "grad_norm": 0.5651464462280273,
      "learning_rate": 0.002828,
      "loss": 1.5596,
      "step": 126848
    },
    {
      "epoch": 9.524352720450281,
      "grad_norm": 0.4518251419067383,
      "learning_rate": 0.002828,
      "loss": 1.5542,
      "step": 126912
    },
    {
      "epoch": 9.529155722326454,
      "grad_norm": 0.49248403310775757,
      "learning_rate": 0.002828,
      "loss": 1.5586,
      "step": 126976
    },
    {
      "epoch": 9.533958724202627,
      "grad_norm": 0.48522064089775085,
      "learning_rate": 0.002828,
      "loss": 1.5526,
      "step": 127040
    },
    {
      "epoch": 9.5387617260788,
      "grad_norm": 0.5640375018119812,
      "learning_rate": 0.002828,
      "loss": 1.554,
      "step": 127104
    },
    {
      "epoch": 9.543564727954973,
      "grad_norm": 0.5300438404083252,
      "learning_rate": 0.002828,
      "loss": 1.5627,
      "step": 127168
    },
    {
      "epoch": 9.548367729831144,
      "grad_norm": 0.459332138299942,
      "learning_rate": 0.002828,
      "loss": 1.5527,
      "step": 127232
    },
    {
      "epoch": 9.553170731707317,
      "grad_norm": 0.47487667202949524,
      "learning_rate": 0.002828,
      "loss": 1.5533,
      "step": 127296
    },
    {
      "epoch": 9.55797373358349,
      "grad_norm": 0.4343540370464325,
      "learning_rate": 0.002828,
      "loss": 1.5515,
      "step": 127360
    },
    {
      "epoch": 9.562776735459662,
      "grad_norm": 0.4315250813961029,
      "learning_rate": 0.002828,
      "loss": 1.5608,
      "step": 127424
    },
    {
      "epoch": 9.567579737335835,
      "grad_norm": 0.4699438512325287,
      "learning_rate": 0.002828,
      "loss": 1.562,
      "step": 127488
    },
    {
      "epoch": 9.572382739212008,
      "grad_norm": 0.6131508350372314,
      "learning_rate": 0.002828,
      "loss": 1.5469,
      "step": 127552
    },
    {
      "epoch": 9.57718574108818,
      "grad_norm": 0.55040043592453,
      "learning_rate": 0.002828,
      "loss": 1.5565,
      "step": 127616
    },
    {
      "epoch": 9.581988742964352,
      "grad_norm": 0.5390975475311279,
      "learning_rate": 0.002828,
      "loss": 1.5547,
      "step": 127680
    },
    {
      "epoch": 9.586791744840525,
      "grad_norm": 0.49108564853668213,
      "learning_rate": 0.002828,
      "loss": 1.5581,
      "step": 127744
    },
    {
      "epoch": 9.591594746716698,
      "grad_norm": 0.47410160303115845,
      "learning_rate": 0.002828,
      "loss": 1.5608,
      "step": 127808
    },
    {
      "epoch": 9.59639774859287,
      "grad_norm": 0.4867013990879059,
      "learning_rate": 0.002828,
      "loss": 1.5515,
      "step": 127872
    },
    {
      "epoch": 9.601200750469044,
      "grad_norm": 0.5015280842781067,
      "learning_rate": 0.002828,
      "loss": 1.5522,
      "step": 127936
    },
    {
      "epoch": 9.606003752345217,
      "grad_norm": 0.4710654616355896,
      "learning_rate": 0.002828,
      "loss": 1.5614,
      "step": 128000
    },
    {
      "epoch": 9.610806754221388,
      "grad_norm": 0.5308662056922913,
      "learning_rate": 0.002828,
      "loss": 1.5507,
      "step": 128064
    },
    {
      "epoch": 9.61560975609756,
      "grad_norm": 0.5719583630561829,
      "learning_rate": 0.002828,
      "loss": 1.5561,
      "step": 128128
    },
    {
      "epoch": 9.620412757973734,
      "grad_norm": 0.5578813552856445,
      "learning_rate": 0.002828,
      "loss": 1.5684,
      "step": 128192
    },
    {
      "epoch": 9.625215759849906,
      "grad_norm": 0.5093487501144409,
      "learning_rate": 0.002828,
      "loss": 1.5595,
      "step": 128256
    },
    {
      "epoch": 9.63001876172608,
      "grad_norm": 0.4938162863254547,
      "learning_rate": 0.002828,
      "loss": 1.5508,
      "step": 128320
    },
    {
      "epoch": 9.634821763602252,
      "grad_norm": 0.5062044858932495,
      "learning_rate": 0.002828,
      "loss": 1.5566,
      "step": 128384
    },
    {
      "epoch": 9.639624765478423,
      "grad_norm": 0.48692384362220764,
      "learning_rate": 0.002828,
      "loss": 1.5557,
      "step": 128448
    },
    {
      "epoch": 9.644427767354596,
      "grad_norm": 0.4967643916606903,
      "learning_rate": 0.002828,
      "loss": 1.5494,
      "step": 128512
    },
    {
      "epoch": 9.649230769230769,
      "grad_norm": 0.442503422498703,
      "learning_rate": 0.002828,
      "loss": 1.5553,
      "step": 128576
    },
    {
      "epoch": 9.654033771106942,
      "grad_norm": 0.5926002860069275,
      "learning_rate": 0.002828,
      "loss": 1.5523,
      "step": 128640
    },
    {
      "epoch": 9.658836772983115,
      "grad_norm": 0.4846031367778778,
      "learning_rate": 0.002828,
      "loss": 1.5611,
      "step": 128704
    },
    {
      "epoch": 9.663639774859288,
      "grad_norm": 0.5495798587799072,
      "learning_rate": 0.002828,
      "loss": 1.5555,
      "step": 128768
    },
    {
      "epoch": 9.66844277673546,
      "grad_norm": 0.5806244611740112,
      "learning_rate": 0.002828,
      "loss": 1.5589,
      "step": 128832
    },
    {
      "epoch": 9.673245778611632,
      "grad_norm": 0.5770285725593567,
      "learning_rate": 0.002828,
      "loss": 1.5488,
      "step": 128896
    },
    {
      "epoch": 9.678048780487805,
      "grad_norm": 0.4988243877887726,
      "learning_rate": 0.002828,
      "loss": 1.5523,
      "step": 128960
    },
    {
      "epoch": 9.682851782363977,
      "grad_norm": 0.48712292313575745,
      "learning_rate": 0.002828,
      "loss": 1.5577,
      "step": 129024
    },
    {
      "epoch": 9.68765478424015,
      "grad_norm": 0.5018770694732666,
      "learning_rate": 0.002828,
      "loss": 1.5556,
      "step": 129088
    },
    {
      "epoch": 9.692457786116323,
      "grad_norm": 0.4957042336463928,
      "learning_rate": 0.002828,
      "loss": 1.5579,
      "step": 129152
    },
    {
      "epoch": 9.697260787992496,
      "grad_norm": 0.48829516768455505,
      "learning_rate": 0.002828,
      "loss": 1.5658,
      "step": 129216
    },
    {
      "epoch": 9.702063789868667,
      "grad_norm": 0.4693721532821655,
      "learning_rate": 0.002828,
      "loss": 1.5524,
      "step": 129280
    },
    {
      "epoch": 9.70686679174484,
      "grad_norm": 0.4855722486972809,
      "learning_rate": 0.002828,
      "loss": 1.5557,
      "step": 129344
    },
    {
      "epoch": 9.711669793621013,
      "grad_norm": 0.5685173869132996,
      "learning_rate": 0.002828,
      "loss": 1.5567,
      "step": 129408
    },
    {
      "epoch": 9.716472795497186,
      "grad_norm": 0.55943363904953,
      "learning_rate": 0.002828,
      "loss": 1.5562,
      "step": 129472
    },
    {
      "epoch": 9.721275797373359,
      "grad_norm": 0.36396172642707825,
      "learning_rate": 0.002828,
      "loss": 1.5611,
      "step": 129536
    },
    {
      "epoch": 9.726078799249532,
      "grad_norm": 0.45727115869522095,
      "learning_rate": 0.002828,
      "loss": 1.5502,
      "step": 129600
    },
    {
      "epoch": 9.730881801125703,
      "grad_norm": 0.5950860381126404,
      "learning_rate": 0.002828,
      "loss": 1.5565,
      "step": 129664
    },
    {
      "epoch": 9.735684803001876,
      "grad_norm": 0.5075332522392273,
      "learning_rate": 0.002828,
      "loss": 1.549,
      "step": 129728
    },
    {
      "epoch": 9.740487804878049,
      "grad_norm": 0.6805112361907959,
      "learning_rate": 0.002828,
      "loss": 1.5531,
      "step": 129792
    },
    {
      "epoch": 9.745290806754221,
      "grad_norm": 0.5490958094596863,
      "learning_rate": 0.002828,
      "loss": 1.5578,
      "step": 129856
    },
    {
      "epoch": 9.750093808630394,
      "grad_norm": 0.5078960657119751,
      "learning_rate": 0.002828,
      "loss": 1.5608,
      "step": 129920
    },
    {
      "epoch": 9.754896810506567,
      "grad_norm": 0.47650301456451416,
      "learning_rate": 0.002828,
      "loss": 1.554,
      "step": 129984
    },
    {
      "epoch": 9.759699812382738,
      "grad_norm": 0.5622501969337463,
      "learning_rate": 0.002828,
      "loss": 1.552,
      "step": 130048
    },
    {
      "epoch": 9.764502814258911,
      "grad_norm": 0.5386357307434082,
      "learning_rate": 0.002828,
      "loss": 1.5519,
      "step": 130112
    },
    {
      "epoch": 9.769305816135084,
      "grad_norm": 0.47315895557403564,
      "learning_rate": 0.002828,
      "loss": 1.547,
      "step": 130176
    },
    {
      "epoch": 9.774108818011257,
      "grad_norm": 0.40879422426223755,
      "learning_rate": 0.002828,
      "loss": 1.5522,
      "step": 130240
    },
    {
      "epoch": 9.77891181988743,
      "grad_norm": 0.44783979654312134,
      "learning_rate": 0.002828,
      "loss": 1.5561,
      "step": 130304
    },
    {
      "epoch": 9.783714821763603,
      "grad_norm": 0.6161860227584839,
      "learning_rate": 0.002828,
      "loss": 1.5581,
      "step": 130368
    },
    {
      "epoch": 9.788517823639776,
      "grad_norm": 0.5071606636047363,
      "learning_rate": 0.002828,
      "loss": 1.5562,
      "step": 130432
    },
    {
      "epoch": 9.793320825515947,
      "grad_norm": 0.5199759602546692,
      "learning_rate": 0.002828,
      "loss": 1.5561,
      "step": 130496
    },
    {
      "epoch": 9.79812382739212,
      "grad_norm": 0.6187331080436707,
      "learning_rate": 0.002828,
      "loss": 1.5519,
      "step": 130560
    },
    {
      "epoch": 9.802926829268293,
      "grad_norm": 0.5684900283813477,
      "learning_rate": 0.002828,
      "loss": 1.5551,
      "step": 130624
    },
    {
      "epoch": 9.807729831144465,
      "grad_norm": 0.5067030191421509,
      "learning_rate": 0.002828,
      "loss": 1.5493,
      "step": 130688
    },
    {
      "epoch": 9.812532833020638,
      "grad_norm": 0.46885350346565247,
      "learning_rate": 0.002828,
      "loss": 1.5558,
      "step": 130752
    },
    {
      "epoch": 9.817335834896811,
      "grad_norm": 0.6028760075569153,
      "learning_rate": 0.002828,
      "loss": 1.5552,
      "step": 130816
    },
    {
      "epoch": 9.822138836772982,
      "grad_norm": 0.5503089427947998,
      "learning_rate": 0.002828,
      "loss": 1.5589,
      "step": 130880
    },
    {
      "epoch": 9.826941838649155,
      "grad_norm": 0.5198109149932861,
      "learning_rate": 0.002828,
      "loss": 1.5586,
      "step": 130944
    },
    {
      "epoch": 9.831744840525328,
      "grad_norm": 0.4858454763889313,
      "learning_rate": 0.002828,
      "loss": 1.5538,
      "step": 131008
    },
    {
      "epoch": 9.836547842401501,
      "grad_norm": 0.4834841191768646,
      "learning_rate": 0.002828,
      "loss": 1.5546,
      "step": 131072
    },
    {
      "epoch": 9.841350844277674,
      "grad_norm": 0.5896739959716797,
      "learning_rate": 0.002828,
      "loss": 1.5529,
      "step": 131136
    },
    {
      "epoch": 9.846153846153847,
      "grad_norm": 0.5708956122398376,
      "learning_rate": 0.002828,
      "loss": 1.5568,
      "step": 131200
    },
    {
      "epoch": 9.85095684803002,
      "grad_norm": 0.5776032209396362,
      "learning_rate": 0.002828,
      "loss": 1.5532,
      "step": 131264
    },
    {
      "epoch": 9.85575984990619,
      "grad_norm": 0.4519674479961395,
      "learning_rate": 0.002828,
      "loss": 1.5519,
      "step": 131328
    },
    {
      "epoch": 9.860562851782364,
      "grad_norm": 0.45993855595588684,
      "learning_rate": 0.002828,
      "loss": 1.5575,
      "step": 131392
    },
    {
      "epoch": 9.865365853658536,
      "grad_norm": 0.42766380310058594,
      "learning_rate": 0.002828,
      "loss": 1.5468,
      "step": 131456
    },
    {
      "epoch": 9.87016885553471,
      "grad_norm": 0.424795925617218,
      "learning_rate": 0.002828,
      "loss": 1.5502,
      "step": 131520
    },
    {
      "epoch": 9.874971857410882,
      "grad_norm": 0.4582250714302063,
      "learning_rate": 0.002828,
      "loss": 1.5556,
      "step": 131584
    },
    {
      "epoch": 9.879774859287055,
      "grad_norm": 0.45927566289901733,
      "learning_rate": 0.002828,
      "loss": 1.5576,
      "step": 131648
    },
    {
      "epoch": 9.884577861163226,
      "grad_norm": 0.44731155037879944,
      "learning_rate": 0.002828,
      "loss": 1.5526,
      "step": 131712
    },
    {
      "epoch": 9.8893808630394,
      "grad_norm": 0.48120561242103577,
      "learning_rate": 0.002828,
      "loss": 1.5607,
      "step": 131776
    },
    {
      "epoch": 9.894183864915572,
      "grad_norm": 0.6588419675827026,
      "learning_rate": 0.002828,
      "loss": 1.5606,
      "step": 131840
    },
    {
      "epoch": 9.898986866791745,
      "grad_norm": 0.4711664617061615,
      "learning_rate": 0.002828,
      "loss": 1.5563,
      "step": 131904
    },
    {
      "epoch": 9.903789868667918,
      "grad_norm": 0.4643876552581787,
      "learning_rate": 0.002828,
      "loss": 1.5528,
      "step": 131968
    },
    {
      "epoch": 9.90859287054409,
      "grad_norm": 0.52628093957901,
      "learning_rate": 0.002828,
      "loss": 1.5528,
      "step": 132032
    },
    {
      "epoch": 9.913395872420264,
      "grad_norm": 0.5681013464927673,
      "learning_rate": 0.002828,
      "loss": 1.5451,
      "step": 132096
    },
    {
      "epoch": 9.918198874296435,
      "grad_norm": 0.6170089840888977,
      "learning_rate": 0.002828,
      "loss": 1.5538,
      "step": 132160
    },
    {
      "epoch": 9.923001876172608,
      "grad_norm": 0.5162079930305481,
      "learning_rate": 0.002828,
      "loss": 1.5472,
      "step": 132224
    },
    {
      "epoch": 9.92780487804878,
      "grad_norm": 0.46833181381225586,
      "learning_rate": 0.002828,
      "loss": 1.5568,
      "step": 132288
    },
    {
      "epoch": 9.932607879924953,
      "grad_norm": 0.44369474053382874,
      "learning_rate": 0.002828,
      "loss": 1.5613,
      "step": 132352
    },
    {
      "epoch": 9.937410881801126,
      "grad_norm": 0.5638606548309326,
      "learning_rate": 0.002828,
      "loss": 1.5561,
      "step": 132416
    },
    {
      "epoch": 9.942213883677299,
      "grad_norm": 0.4265860319137573,
      "learning_rate": 0.002828,
      "loss": 1.555,
      "step": 132480
    },
    {
      "epoch": 9.94701688555347,
      "grad_norm": 0.4962076246738434,
      "learning_rate": 0.002828,
      "loss": 1.551,
      "step": 132544
    },
    {
      "epoch": 9.951819887429643,
      "grad_norm": 0.45441246032714844,
      "learning_rate": 0.002828,
      "loss": 1.5524,
      "step": 132608
    },
    {
      "epoch": 9.956622889305816,
      "grad_norm": 0.5333769917488098,
      "learning_rate": 0.002828,
      "loss": 1.5521,
      "step": 132672
    },
    {
      "epoch": 9.961425891181989,
      "grad_norm": 0.4870472550392151,
      "learning_rate": 0.002828,
      "loss": 1.5489,
      "step": 132736
    },
    {
      "epoch": 9.966228893058162,
      "grad_norm": 0.4919857382774353,
      "learning_rate": 0.002828,
      "loss": 1.5503,
      "step": 132800
    },
    {
      "epoch": 9.971031894934335,
      "grad_norm": 0.43766430020332336,
      "learning_rate": 0.002828,
      "loss": 1.5521,
      "step": 132864
    },
    {
      "epoch": 9.975834896810508,
      "grad_norm": 0.435209721326828,
      "learning_rate": 0.002828,
      "loss": 1.5529,
      "step": 132928
    },
    {
      "epoch": 9.980637898686679,
      "grad_norm": 0.441018283367157,
      "learning_rate": 0.002828,
      "loss": 1.5558,
      "step": 132992
    },
    {
      "epoch": 9.985440900562851,
      "grad_norm": 0.41537073254585266,
      "learning_rate": 0.002828,
      "loss": 1.5537,
      "step": 133056
    },
    {
      "epoch": 9.990243902439024,
      "grad_norm": 0.4654955565929413,
      "learning_rate": 0.002828,
      "loss": 1.5511,
      "step": 133120
    },
    {
      "epoch": 9.995046904315197,
      "grad_norm": 0.5095586776733398,
      "learning_rate": 0.002828,
      "loss": 1.5585,
      "step": 133184
    },
    {
      "epoch": 9.99984990619137,
      "grad_norm": 0.49905091524124146,
      "learning_rate": 0.002828,
      "loss": 1.5502,
      "step": 133248
    },
    {
      "epoch": 10.004652908067543,
      "grad_norm": 0.479478657245636,
      "learning_rate": 0.002828,
      "loss": 1.5136,
      "step": 133312
    },
    {
      "epoch": 10.009455909943714,
      "grad_norm": 0.4345605671405792,
      "learning_rate": 0.002828,
      "loss": 1.513,
      "step": 133376
    },
    {
      "epoch": 10.014258911819887,
      "grad_norm": 0.5158377289772034,
      "learning_rate": 0.002828,
      "loss": 1.5104,
      "step": 133440
    },
    {
      "epoch": 10.01906191369606,
      "grad_norm": 0.49847567081451416,
      "learning_rate": 0.002828,
      "loss": 1.5042,
      "step": 133504
    },
    {
      "epoch": 10.023864915572233,
      "grad_norm": 0.49734270572662354,
      "learning_rate": 0.002828,
      "loss": 1.5085,
      "step": 133568
    },
    {
      "epoch": 10.028667917448406,
      "grad_norm": 0.5903614163398743,
      "learning_rate": 0.002828,
      "loss": 1.5079,
      "step": 133632
    },
    {
      "epoch": 10.033470919324579,
      "grad_norm": 0.4688267409801483,
      "learning_rate": 0.002828,
      "loss": 1.5104,
      "step": 133696
    },
    {
      "epoch": 10.03827392120075,
      "grad_norm": 0.5410575866699219,
      "learning_rate": 0.002828,
      "loss": 1.5101,
      "step": 133760
    },
    {
      "epoch": 10.043076923076923,
      "grad_norm": 0.5670648813247681,
      "learning_rate": 0.002828,
      "loss": 1.5066,
      "step": 133824
    },
    {
      "epoch": 10.047879924953095,
      "grad_norm": 0.544195830821991,
      "learning_rate": 0.002828,
      "loss": 1.5108,
      "step": 133888
    },
    {
      "epoch": 10.052682926829268,
      "grad_norm": 0.4335324168205261,
      "learning_rate": 0.002828,
      "loss": 1.5113,
      "step": 133952
    },
    {
      "epoch": 10.057485928705441,
      "grad_norm": 0.734622061252594,
      "learning_rate": 0.002828,
      "loss": 1.5098,
      "step": 134016
    },
    {
      "epoch": 10.062288930581614,
      "grad_norm": 0.5276984572410583,
      "learning_rate": 0.002828,
      "loss": 1.5151,
      "step": 134080
    },
    {
      "epoch": 10.067091932457787,
      "grad_norm": 0.5516037344932556,
      "learning_rate": 0.002828,
      "loss": 1.5087,
      "step": 134144
    },
    {
      "epoch": 10.071894934333958,
      "grad_norm": 0.4399818778038025,
      "learning_rate": 0.002828,
      "loss": 1.523,
      "step": 134208
    },
    {
      "epoch": 10.076697936210131,
      "grad_norm": 0.5881381630897522,
      "learning_rate": 0.002828,
      "loss": 1.5112,
      "step": 134272
    },
    {
      "epoch": 10.081500938086304,
      "grad_norm": 0.480007529258728,
      "learning_rate": 0.002828,
      "loss": 1.5039,
      "step": 134336
    },
    {
      "epoch": 10.086303939962477,
      "grad_norm": 0.6424974203109741,
      "learning_rate": 0.002828,
      "loss": 1.5081,
      "step": 134400
    },
    {
      "epoch": 10.09110694183865,
      "grad_norm": 0.5155991911888123,
      "learning_rate": 0.002828,
      "loss": 1.5121,
      "step": 134464
    },
    {
      "epoch": 10.095909943714823,
      "grad_norm": 0.48689907789230347,
      "learning_rate": 0.002828,
      "loss": 1.5095,
      "step": 134528
    },
    {
      "epoch": 10.100712945590994,
      "grad_norm": 0.5510069727897644,
      "learning_rate": 0.002828,
      "loss": 1.5156,
      "step": 134592
    },
    {
      "epoch": 10.105515947467167,
      "grad_norm": 0.49252885580062866,
      "learning_rate": 0.002828,
      "loss": 1.5132,
      "step": 134656
    },
    {
      "epoch": 10.11031894934334,
      "grad_norm": 0.6291511058807373,
      "learning_rate": 0.002828,
      "loss": 1.5103,
      "step": 134720
    },
    {
      "epoch": 10.115121951219512,
      "grad_norm": 0.4867250323295593,
      "learning_rate": 0.002828,
      "loss": 1.5185,
      "step": 134784
    },
    {
      "epoch": 10.119924953095685,
      "grad_norm": 0.5530567765235901,
      "learning_rate": 0.002828,
      "loss": 1.5067,
      "step": 134848
    },
    {
      "epoch": 10.124727954971858,
      "grad_norm": 0.43756282329559326,
      "learning_rate": 0.002828,
      "loss": 1.5142,
      "step": 134912
    },
    {
      "epoch": 10.12953095684803,
      "grad_norm": 0.5081878304481506,
      "learning_rate": 0.002828,
      "loss": 1.5127,
      "step": 134976
    },
    {
      "epoch": 10.134333958724202,
      "grad_norm": 0.5711334943771362,
      "learning_rate": 0.002828,
      "loss": 1.5125,
      "step": 135040
    },
    {
      "epoch": 10.139136960600375,
      "grad_norm": 0.5855487585067749,
      "learning_rate": 0.002828,
      "loss": 1.511,
      "step": 135104
    },
    {
      "epoch": 10.143939962476548,
      "grad_norm": 0.5145984888076782,
      "learning_rate": 0.002828,
      "loss": 1.5179,
      "step": 135168
    },
    {
      "epoch": 10.14874296435272,
      "grad_norm": 0.4353695511817932,
      "learning_rate": 0.002828,
      "loss": 1.5148,
      "step": 135232
    },
    {
      "epoch": 10.153545966228894,
      "grad_norm": 0.5010886788368225,
      "learning_rate": 0.002828,
      "loss": 1.5164,
      "step": 135296
    },
    {
      "epoch": 10.158348968105066,
      "grad_norm": 0.5228750109672546,
      "learning_rate": 0.002828,
      "loss": 1.523,
      "step": 135360
    },
    {
      "epoch": 10.163151969981238,
      "grad_norm": 0.6481117606163025,
      "learning_rate": 0.002828,
      "loss": 1.5176,
      "step": 135424
    },
    {
      "epoch": 10.16795497185741,
      "grad_norm": 0.4466358423233032,
      "learning_rate": 0.002828,
      "loss": 1.509,
      "step": 135488
    },
    {
      "epoch": 10.172757973733583,
      "grad_norm": 0.5442012548446655,
      "learning_rate": 0.002828,
      "loss": 1.5116,
      "step": 135552
    },
    {
      "epoch": 10.177560975609756,
      "grad_norm": 0.4774165451526642,
      "learning_rate": 0.002828,
      "loss": 1.5152,
      "step": 135616
    },
    {
      "epoch": 10.18236397748593,
      "grad_norm": 0.45034146308898926,
      "learning_rate": 0.002828,
      "loss": 1.5206,
      "step": 135680
    },
    {
      "epoch": 10.187166979362102,
      "grad_norm": 0.5528009533882141,
      "learning_rate": 0.002828,
      "loss": 1.5185,
      "step": 135744
    },
    {
      "epoch": 10.191969981238273,
      "grad_norm": 0.45600980520248413,
      "learning_rate": 0.002828,
      "loss": 1.5149,
      "step": 135808
    },
    {
      "epoch": 10.196772983114446,
      "grad_norm": 0.5091872811317444,
      "learning_rate": 0.002828,
      "loss": 1.5151,
      "step": 135872
    },
    {
      "epoch": 10.201575984990619,
      "grad_norm": 0.46785837411880493,
      "learning_rate": 0.002828,
      "loss": 1.5125,
      "step": 135936
    },
    {
      "epoch": 10.206378986866792,
      "grad_norm": 0.5324965119361877,
      "learning_rate": 0.002828,
      "loss": 1.516,
      "step": 136000
    },
    {
      "epoch": 10.211181988742965,
      "grad_norm": 0.5576740503311157,
      "learning_rate": 0.002828,
      "loss": 1.5118,
      "step": 136064
    },
    {
      "epoch": 10.215984990619138,
      "grad_norm": 0.5563480854034424,
      "learning_rate": 0.002828,
      "loss": 1.5173,
      "step": 136128
    },
    {
      "epoch": 10.22078799249531,
      "grad_norm": 0.6065492630004883,
      "learning_rate": 0.002828,
      "loss": 1.5169,
      "step": 136192
    },
    {
      "epoch": 10.225590994371482,
      "grad_norm": 0.5944182872772217,
      "learning_rate": 0.002828,
      "loss": 1.5136,
      "step": 136256
    },
    {
      "epoch": 10.230393996247654,
      "grad_norm": 0.48441362380981445,
      "learning_rate": 0.002828,
      "loss": 1.5211,
      "step": 136320
    },
    {
      "epoch": 10.235196998123827,
      "grad_norm": 0.4990592300891876,
      "learning_rate": 0.002828,
      "loss": 1.5162,
      "step": 136384
    },
    {
      "epoch": 10.24,
      "grad_norm": 0.5005013942718506,
      "learning_rate": 0.002828,
      "loss": 1.5155,
      "step": 136448
    },
    {
      "epoch": 10.244803001876173,
      "grad_norm": 0.4367721974849701,
      "learning_rate": 0.002828,
      "loss": 1.5178,
      "step": 136512
    },
    {
      "epoch": 10.249606003752346,
      "grad_norm": 0.6128390431404114,
      "learning_rate": 0.002828,
      "loss": 1.5206,
      "step": 136576
    },
    {
      "epoch": 10.254409005628517,
      "grad_norm": 0.48846036195755005,
      "learning_rate": 0.002828,
      "loss": 1.5129,
      "step": 136640
    },
    {
      "epoch": 10.25921200750469,
      "grad_norm": 0.5013264417648315,
      "learning_rate": 0.002828,
      "loss": 1.5154,
      "step": 136704
    },
    {
      "epoch": 10.264015009380863,
      "grad_norm": 0.5089060664176941,
      "learning_rate": 0.002828,
      "loss": 1.5242,
      "step": 136768
    },
    {
      "epoch": 10.268818011257036,
      "grad_norm": 0.4745241701602936,
      "learning_rate": 0.002828,
      "loss": 1.5215,
      "step": 136832
    },
    {
      "epoch": 10.273621013133209,
      "grad_norm": 0.553170382976532,
      "learning_rate": 0.002828,
      "loss": 1.5187,
      "step": 136896
    },
    {
      "epoch": 10.278424015009382,
      "grad_norm": 0.5633316040039062,
      "learning_rate": 0.002828,
      "loss": 1.5124,
      "step": 136960
    },
    {
      "epoch": 10.283227016885553,
      "grad_norm": 0.6235347390174866,
      "learning_rate": 0.002828,
      "loss": 1.5103,
      "step": 137024
    },
    {
      "epoch": 10.288030018761726,
      "grad_norm": 0.44016027450561523,
      "learning_rate": 0.002828,
      "loss": 1.5144,
      "step": 137088
    },
    {
      "epoch": 10.292833020637898,
      "grad_norm": 0.5390121340751648,
      "learning_rate": 0.002828,
      "loss": 1.525,
      "step": 137152
    },
    {
      "epoch": 10.297636022514071,
      "grad_norm": 0.5932199954986572,
      "learning_rate": 0.002828,
      "loss": 1.5268,
      "step": 137216
    },
    {
      "epoch": 10.302439024390244,
      "grad_norm": 0.46444085240364075,
      "learning_rate": 0.002828,
      "loss": 1.5165,
      "step": 137280
    },
    {
      "epoch": 10.307242026266417,
      "grad_norm": 0.46962660551071167,
      "learning_rate": 0.002828,
      "loss": 1.522,
      "step": 137344
    },
    {
      "epoch": 10.31204502814259,
      "grad_norm": 0.5389118790626526,
      "learning_rate": 0.002828,
      "loss": 1.5266,
      "step": 137408
    },
    {
      "epoch": 10.316848030018761,
      "grad_norm": 0.5998408198356628,
      "learning_rate": 0.002828,
      "loss": 1.523,
      "step": 137472
    },
    {
      "epoch": 10.321651031894934,
      "grad_norm": 0.5823659300804138,
      "learning_rate": 0.002828,
      "loss": 1.5149,
      "step": 137536
    },
    {
      "epoch": 10.326454033771107,
      "grad_norm": 0.5257498621940613,
      "learning_rate": 0.002828,
      "loss": 1.5203,
      "step": 137600
    },
    {
      "epoch": 10.33125703564728,
      "grad_norm": 0.5792316794395447,
      "learning_rate": 0.002828,
      "loss": 1.526,
      "step": 137664
    },
    {
      "epoch": 10.336060037523453,
      "grad_norm": 0.5107080936431885,
      "learning_rate": 0.002828,
      "loss": 1.5182,
      "step": 137728
    },
    {
      "epoch": 10.340863039399625,
      "grad_norm": 0.5128101110458374,
      "learning_rate": 0.002828,
      "loss": 1.5126,
      "step": 137792
    },
    {
      "epoch": 10.345666041275797,
      "grad_norm": 0.4674902558326721,
      "learning_rate": 0.002828,
      "loss": 1.5182,
      "step": 137856
    },
    {
      "epoch": 10.35046904315197,
      "grad_norm": 0.462354838848114,
      "learning_rate": 0.002828,
      "loss": 1.5234,
      "step": 137920
    },
    {
      "epoch": 10.355272045028142,
      "grad_norm": 0.543552815914154,
      "learning_rate": 0.002828,
      "loss": 1.5245,
      "step": 137984
    },
    {
      "epoch": 10.360075046904315,
      "grad_norm": 0.44141873717308044,
      "learning_rate": 0.002828,
      "loss": 1.5237,
      "step": 138048
    },
    {
      "epoch": 10.364878048780488,
      "grad_norm": 0.4777672588825226,
      "learning_rate": 0.002828,
      "loss": 1.5183,
      "step": 138112
    },
    {
      "epoch": 10.369681050656661,
      "grad_norm": 0.5415074229240417,
      "learning_rate": 0.002828,
      "loss": 1.5164,
      "step": 138176
    },
    {
      "epoch": 10.374484052532832,
      "grad_norm": 0.6049688458442688,
      "learning_rate": 0.002828,
      "loss": 1.524,
      "step": 138240
    },
    {
      "epoch": 10.379287054409005,
      "grad_norm": 0.46092769503593445,
      "learning_rate": 0.002828,
      "loss": 1.5212,
      "step": 138304
    },
    {
      "epoch": 10.384090056285178,
      "grad_norm": 0.5517581701278687,
      "learning_rate": 0.002828,
      "loss": 1.519,
      "step": 138368
    },
    {
      "epoch": 10.38889305816135,
      "grad_norm": 0.4884757399559021,
      "learning_rate": 0.002828,
      "loss": 1.5104,
      "step": 138432
    },
    {
      "epoch": 10.393696060037524,
      "grad_norm": 0.5153927803039551,
      "learning_rate": 0.002828,
      "loss": 1.5133,
      "step": 138496
    },
    {
      "epoch": 10.398499061913697,
      "grad_norm": 0.5135812163352966,
      "learning_rate": 0.002828,
      "loss": 1.5277,
      "step": 138560
    },
    {
      "epoch": 10.40330206378987,
      "grad_norm": 0.5352478623390198,
      "learning_rate": 0.002828,
      "loss": 1.5249,
      "step": 138624
    },
    {
      "epoch": 10.40810506566604,
      "grad_norm": 0.4779094457626343,
      "learning_rate": 0.002828,
      "loss": 1.5181,
      "step": 138688
    },
    {
      "epoch": 10.412908067542213,
      "grad_norm": 0.5232270956039429,
      "learning_rate": 0.002828,
      "loss": 1.5187,
      "step": 138752
    },
    {
      "epoch": 10.417711069418386,
      "grad_norm": 0.48955580592155457,
      "learning_rate": 0.002828,
      "loss": 1.5121,
      "step": 138816
    },
    {
      "epoch": 10.42251407129456,
      "grad_norm": 0.5007327795028687,
      "learning_rate": 0.002828,
      "loss": 1.5245,
      "step": 138880
    },
    {
      "epoch": 10.427317073170732,
      "grad_norm": 0.5203362107276917,
      "learning_rate": 0.002828,
      "loss": 1.5269,
      "step": 138944
    },
    {
      "epoch": 10.432120075046905,
      "grad_norm": 0.632523238658905,
      "learning_rate": 0.002828,
      "loss": 1.5158,
      "step": 139008
    },
    {
      "epoch": 10.436923076923076,
      "grad_norm": 0.4640112817287445,
      "learning_rate": 0.002828,
      "loss": 1.527,
      "step": 139072
    },
    {
      "epoch": 10.441726078799249,
      "grad_norm": 0.4999347925186157,
      "learning_rate": 0.002828,
      "loss": 1.506,
      "step": 139136
    },
    {
      "epoch": 10.446529080675422,
      "grad_norm": 0.44936737418174744,
      "learning_rate": 0.002828,
      "loss": 1.5176,
      "step": 139200
    },
    {
      "epoch": 10.451332082551595,
      "grad_norm": 0.4773159325122833,
      "learning_rate": 0.002828,
      "loss": 1.5199,
      "step": 139264
    },
    {
      "epoch": 10.456135084427768,
      "grad_norm": 0.4635850191116333,
      "learning_rate": 0.002828,
      "loss": 1.521,
      "step": 139328
    },
    {
      "epoch": 10.46093808630394,
      "grad_norm": 0.42555472254753113,
      "learning_rate": 0.002828,
      "loss": 1.5181,
      "step": 139392
    },
    {
      "epoch": 10.465741088180113,
      "grad_norm": 0.458814412355423,
      "learning_rate": 0.002828,
      "loss": 1.524,
      "step": 139456
    },
    {
      "epoch": 10.470544090056285,
      "grad_norm": 0.4896027743816376,
      "learning_rate": 0.002828,
      "loss": 1.5192,
      "step": 139520
    },
    {
      "epoch": 10.475347091932457,
      "grad_norm": 0.5083631873130798,
      "learning_rate": 0.002828,
      "loss": 1.5218,
      "step": 139584
    },
    {
      "epoch": 10.48015009380863,
      "grad_norm": 0.4863784909248352,
      "learning_rate": 0.002828,
      "loss": 1.5117,
      "step": 139648
    },
    {
      "epoch": 10.484953095684803,
      "grad_norm": 0.4946701228618622,
      "learning_rate": 0.002828,
      "loss": 1.5182,
      "step": 139712
    },
    {
      "epoch": 10.489756097560976,
      "grad_norm": 0.521332859992981,
      "learning_rate": 0.002828,
      "loss": 1.5196,
      "step": 139776
    },
    {
      "epoch": 10.494559099437149,
      "grad_norm": 0.5090921521186829,
      "learning_rate": 0.002828,
      "loss": 1.5168,
      "step": 139840
    },
    {
      "epoch": 10.49936210131332,
      "grad_norm": 0.5902573466300964,
      "learning_rate": 0.002828,
      "loss": 1.5195,
      "step": 139904
    },
    {
      "epoch": 10.504165103189493,
      "grad_norm": 0.5510809421539307,
      "learning_rate": 0.002828,
      "loss": 1.5182,
      "step": 139968
    },
    {
      "epoch": 10.508968105065666,
      "grad_norm": 0.5307651162147522,
      "learning_rate": 0.002828,
      "loss": 1.5237,
      "step": 140032
    },
    {
      "epoch": 10.513771106941839,
      "grad_norm": 0.6894339919090271,
      "learning_rate": 0.002828,
      "loss": 1.5199,
      "step": 140096
    },
    {
      "epoch": 10.518574108818012,
      "grad_norm": 0.48183873295783997,
      "learning_rate": 0.002828,
      "loss": 1.5268,
      "step": 140160
    },
    {
      "epoch": 10.523377110694184,
      "grad_norm": 0.5286323428153992,
      "learning_rate": 0.002828,
      "loss": 1.5252,
      "step": 140224
    },
    {
      "epoch": 10.528180112570357,
      "grad_norm": 0.44124990701675415,
      "learning_rate": 0.002828,
      "loss": 1.5244,
      "step": 140288
    },
    {
      "epoch": 10.532983114446528,
      "grad_norm": 0.5079673528671265,
      "learning_rate": 0.002828,
      "loss": 1.518,
      "step": 140352
    },
    {
      "epoch": 10.537786116322701,
      "grad_norm": 0.48387521505355835,
      "learning_rate": 0.002828,
      "loss": 1.5168,
      "step": 140416
    },
    {
      "epoch": 10.542589118198874,
      "grad_norm": 0.6227713823318481,
      "learning_rate": 0.002828,
      "loss": 1.5268,
      "step": 140480
    },
    {
      "epoch": 10.547392120075047,
      "grad_norm": 0.549760103225708,
      "learning_rate": 0.002828,
      "loss": 1.5183,
      "step": 140544
    },
    {
      "epoch": 10.55219512195122,
      "grad_norm": 0.5009521842002869,
      "learning_rate": 0.002828,
      "loss": 1.5187,
      "step": 140608
    },
    {
      "epoch": 10.556998123827393,
      "grad_norm": 0.5136734247207642,
      "learning_rate": 0.002828,
      "loss": 1.5133,
      "step": 140672
    },
    {
      "epoch": 10.561801125703564,
      "grad_norm": 0.5014452338218689,
      "learning_rate": 0.002828,
      "loss": 1.5174,
      "step": 140736
    },
    {
      "epoch": 10.566604127579737,
      "grad_norm": 0.4686838686466217,
      "learning_rate": 0.002828,
      "loss": 1.5221,
      "step": 140800
    },
    {
      "epoch": 10.57140712945591,
      "grad_norm": 0.49404841661453247,
      "learning_rate": 0.002828,
      "loss": 1.5252,
      "step": 140864
    },
    {
      "epoch": 10.576210131332083,
      "grad_norm": 0.5186532139778137,
      "learning_rate": 0.002828,
      "loss": 1.5168,
      "step": 140928
    },
    {
      "epoch": 10.581013133208256,
      "grad_norm": 0.578127384185791,
      "learning_rate": 0.002828,
      "loss": 1.5242,
      "step": 140992
    },
    {
      "epoch": 10.585816135084428,
      "grad_norm": 0.5328065156936646,
      "learning_rate": 0.002828,
      "loss": 1.5219,
      "step": 141056
    },
    {
      "epoch": 10.590619136960601,
      "grad_norm": 0.459264874458313,
      "learning_rate": 0.002828,
      "loss": 1.5199,
      "step": 141120
    },
    {
      "epoch": 10.595422138836772,
      "grad_norm": 0.5663778185844421,
      "learning_rate": 0.002828,
      "loss": 1.5162,
      "step": 141184
    },
    {
      "epoch": 10.600225140712945,
      "grad_norm": 0.5025619268417358,
      "learning_rate": 0.002828,
      "loss": 1.5182,
      "step": 141248
    },
    {
      "epoch": 10.605028142589118,
      "grad_norm": 0.4759722948074341,
      "learning_rate": 0.002828,
      "loss": 1.514,
      "step": 141312
    },
    {
      "epoch": 10.609831144465291,
      "grad_norm": 0.4879244267940521,
      "learning_rate": 0.002828,
      "loss": 1.5133,
      "step": 141376
    },
    {
      "epoch": 10.614634146341464,
      "grad_norm": 0.4883573055267334,
      "learning_rate": 0.002828,
      "loss": 1.5144,
      "step": 141440
    },
    {
      "epoch": 10.619437148217637,
      "grad_norm": 0.5533087253570557,
      "learning_rate": 0.002828,
      "loss": 1.5111,
      "step": 141504
    },
    {
      "epoch": 10.624240150093808,
      "grad_norm": 0.5033180117607117,
      "learning_rate": 0.002828,
      "loss": 1.5253,
      "step": 141568
    },
    {
      "epoch": 10.62904315196998,
      "grad_norm": 0.4731650948524475,
      "learning_rate": 0.002828,
      "loss": 1.5171,
      "step": 141632
    },
    {
      "epoch": 10.633846153846154,
      "grad_norm": 0.5617311596870422,
      "learning_rate": 0.002828,
      "loss": 1.5187,
      "step": 141696
    },
    {
      "epoch": 10.638649155722327,
      "grad_norm": 0.588597297668457,
      "learning_rate": 0.002828,
      "loss": 1.524,
      "step": 141760
    },
    {
      "epoch": 10.6434521575985,
      "grad_norm": 0.47357311844825745,
      "learning_rate": 0.002828,
      "loss": 1.5179,
      "step": 141824
    },
    {
      "epoch": 10.648255159474672,
      "grad_norm": 0.5507169365882874,
      "learning_rate": 0.002828,
      "loss": 1.5238,
      "step": 141888
    },
    {
      "epoch": 10.653058161350843,
      "grad_norm": 0.4835931658744812,
      "learning_rate": 0.002828,
      "loss": 1.5186,
      "step": 141952
    },
    {
      "epoch": 10.657861163227016,
      "grad_norm": 0.5200017690658569,
      "learning_rate": 0.002828,
      "loss": 1.5179,
      "step": 142016
    },
    {
      "epoch": 10.66266416510319,
      "grad_norm": 0.6572912931442261,
      "learning_rate": 0.002828,
      "loss": 1.5224,
      "step": 142080
    },
    {
      "epoch": 10.667467166979362,
      "grad_norm": 0.4692614674568176,
      "learning_rate": 0.002828,
      "loss": 1.5235,
      "step": 142144
    },
    {
      "epoch": 10.672270168855535,
      "grad_norm": 0.5642805695533752,
      "learning_rate": 0.002828,
      "loss": 1.5161,
      "step": 142208
    },
    {
      "epoch": 10.677073170731708,
      "grad_norm": 0.5048820972442627,
      "learning_rate": 0.002828,
      "loss": 1.5234,
      "step": 142272
    },
    {
      "epoch": 10.681876172607879,
      "grad_norm": 0.5080498456954956,
      "learning_rate": 0.002828,
      "loss": 1.5164,
      "step": 142336
    },
    {
      "epoch": 10.686679174484052,
      "grad_norm": 0.4397072494029999,
      "learning_rate": 0.002828,
      "loss": 1.5156,
      "step": 142400
    },
    {
      "epoch": 10.691482176360225,
      "grad_norm": 0.49026167392730713,
      "learning_rate": 0.002828,
      "loss": 1.5141,
      "step": 142464
    },
    {
      "epoch": 10.696285178236398,
      "grad_norm": 0.49835464358329773,
      "learning_rate": 0.002828,
      "loss": 1.5189,
      "step": 142528
    },
    {
      "epoch": 10.70108818011257,
      "grad_norm": 0.5250354409217834,
      "learning_rate": 0.002828,
      "loss": 1.5206,
      "step": 142592
    },
    {
      "epoch": 10.705891181988743,
      "grad_norm": 0.560287594795227,
      "learning_rate": 0.002828,
      "loss": 1.5118,
      "step": 142656
    },
    {
      "epoch": 10.710694183864916,
      "grad_norm": 0.47352853417396545,
      "learning_rate": 0.002828,
      "loss": 1.5204,
      "step": 142720
    },
    {
      "epoch": 10.715497185741087,
      "grad_norm": 0.49744275212287903,
      "learning_rate": 0.002828,
      "loss": 1.5195,
      "step": 142784
    },
    {
      "epoch": 10.72030018761726,
      "grad_norm": 0.5193085670471191,
      "learning_rate": 0.002828,
      "loss": 1.5181,
      "step": 142848
    },
    {
      "epoch": 10.725103189493433,
      "grad_norm": 0.5842470526695251,
      "learning_rate": 0.002828,
      "loss": 1.516,
      "step": 142912
    },
    {
      "epoch": 10.729906191369606,
      "grad_norm": 0.5229711532592773,
      "learning_rate": 0.002828,
      "loss": 1.5174,
      "step": 142976
    },
    {
      "epoch": 10.734709193245779,
      "grad_norm": 0.5207584500312805,
      "learning_rate": 0.002828,
      "loss": 1.5202,
      "step": 143040
    },
    {
      "epoch": 10.739512195121952,
      "grad_norm": 0.5027861595153809,
      "learning_rate": 0.002828,
      "loss": 1.5239,
      "step": 143104
    },
    {
      "epoch": 10.744315196998123,
      "grad_norm": 0.6836662292480469,
      "learning_rate": 0.002828,
      "loss": 1.5175,
      "step": 143168
    },
    {
      "epoch": 10.749118198874296,
      "grad_norm": 0.4938174784183502,
      "learning_rate": 0.002828,
      "loss": 1.518,
      "step": 143232
    },
    {
      "epoch": 10.753921200750469,
      "grad_norm": 0.49605420231819153,
      "learning_rate": 0.002828,
      "loss": 1.5169,
      "step": 143296
    },
    {
      "epoch": 10.758724202626642,
      "grad_norm": 0.6375707387924194,
      "learning_rate": 0.002828,
      "loss": 1.5079,
      "step": 143360
    },
    {
      "epoch": 10.763527204502815,
      "grad_norm": 0.40622422099113464,
      "learning_rate": 0.002828,
      "loss": 1.519,
      "step": 143424
    },
    {
      "epoch": 10.768330206378987,
      "grad_norm": 0.6180644631385803,
      "learning_rate": 0.002828,
      "loss": 1.5158,
      "step": 143488
    },
    {
      "epoch": 10.77313320825516,
      "grad_norm": 0.5086051821708679,
      "learning_rate": 0.002828,
      "loss": 1.5174,
      "step": 143552
    },
    {
      "epoch": 10.777936210131331,
      "grad_norm": 0.4496934711933136,
      "learning_rate": 0.002828,
      "loss": 1.5236,
      "step": 143616
    },
    {
      "epoch": 10.782739212007504,
      "grad_norm": 0.43942809104919434,
      "learning_rate": 0.002828,
      "loss": 1.5296,
      "step": 143680
    },
    {
      "epoch": 10.787542213883677,
      "grad_norm": 0.5430384874343872,
      "learning_rate": 0.002828,
      "loss": 1.5153,
      "step": 143744
    },
    {
      "epoch": 10.79234521575985,
      "grad_norm": 0.509769082069397,
      "learning_rate": 0.002828,
      "loss": 1.5215,
      "step": 143808
    },
    {
      "epoch": 10.797148217636023,
      "grad_norm": 0.4945880174636841,
      "learning_rate": 0.002828,
      "loss": 1.5199,
      "step": 143872
    },
    {
      "epoch": 10.801951219512196,
      "grad_norm": 0.5118721127510071,
      "learning_rate": 0.002828,
      "loss": 1.5202,
      "step": 143936
    },
    {
      "epoch": 10.806754221388367,
      "grad_norm": 0.4627864360809326,
      "learning_rate": 0.002828,
      "loss": 1.5179,
      "step": 144000
    },
    {
      "epoch": 10.81155722326454,
      "grad_norm": 0.6523723602294922,
      "learning_rate": 0.002828,
      "loss": 1.5245,
      "step": 144064
    },
    {
      "epoch": 10.816360225140713,
      "grad_norm": 0.5512601137161255,
      "learning_rate": 0.002828,
      "loss": 1.5178,
      "step": 144128
    },
    {
      "epoch": 10.821163227016886,
      "grad_norm": 0.5221168994903564,
      "learning_rate": 0.002828,
      "loss": 1.5207,
      "step": 144192
    },
    {
      "epoch": 10.825966228893058,
      "grad_norm": 0.4710821509361267,
      "learning_rate": 0.002828,
      "loss": 1.5303,
      "step": 144256
    },
    {
      "epoch": 10.830769230769231,
      "grad_norm": 0.5352916121482849,
      "learning_rate": 0.002828,
      "loss": 1.5189,
      "step": 144320
    },
    {
      "epoch": 10.835572232645404,
      "grad_norm": 0.44222092628479004,
      "learning_rate": 0.002828,
      "loss": 1.5245,
      "step": 144384
    },
    {
      "epoch": 10.840375234521575,
      "grad_norm": 0.5109274387359619,
      "learning_rate": 0.002828,
      "loss": 1.5162,
      "step": 144448
    },
    {
      "epoch": 10.845178236397748,
      "grad_norm": 0.6416229009628296,
      "learning_rate": 0.002828,
      "loss": 1.5259,
      "step": 144512
    },
    {
      "epoch": 10.849981238273921,
      "grad_norm": 0.46093955636024475,
      "learning_rate": 0.002828,
      "loss": 1.5198,
      "step": 144576
    },
    {
      "epoch": 10.854784240150094,
      "grad_norm": 0.6144540309906006,
      "learning_rate": 0.002828,
      "loss": 1.5149,
      "step": 144640
    },
    {
      "epoch": 10.859587242026267,
      "grad_norm": 0.47149455547332764,
      "learning_rate": 0.002828,
      "loss": 1.5196,
      "step": 144704
    },
    {
      "epoch": 10.86439024390244,
      "grad_norm": 0.5379576683044434,
      "learning_rate": 0.002828,
      "loss": 1.5149,
      "step": 144768
    },
    {
      "epoch": 10.869193245778611,
      "grad_norm": 0.4889363944530487,
      "learning_rate": 0.002828,
      "loss": 1.52,
      "step": 144832
    },
    {
      "epoch": 10.873996247654784,
      "grad_norm": 0.4816627502441406,
      "learning_rate": 0.002828,
      "loss": 1.5146,
      "step": 144896
    },
    {
      "epoch": 10.878799249530957,
      "grad_norm": 0.4828503131866455,
      "learning_rate": 0.002828,
      "loss": 1.5222,
      "step": 144960
    },
    {
      "epoch": 10.88360225140713,
      "grad_norm": 0.5127456188201904,
      "learning_rate": 0.002828,
      "loss": 1.5218,
      "step": 145024
    },
    {
      "epoch": 10.888405253283302,
      "grad_norm": 0.4628119170665741,
      "learning_rate": 0.002828,
      "loss": 1.5142,
      "step": 145088
    },
    {
      "epoch": 10.893208255159475,
      "grad_norm": 0.5557365417480469,
      "learning_rate": 0.002828,
      "loss": 1.5176,
      "step": 145152
    },
    {
      "epoch": 10.898011257035648,
      "grad_norm": 0.4339671730995178,
      "learning_rate": 0.002828,
      "loss": 1.5198,
      "step": 145216
    },
    {
      "epoch": 10.90281425891182,
      "grad_norm": 0.5440133213996887,
      "learning_rate": 0.002828,
      "loss": 1.5187,
      "step": 145280
    },
    {
      "epoch": 10.907617260787992,
      "grad_norm": 0.4955686628818512,
      "learning_rate": 0.002828,
      "loss": 1.5209,
      "step": 145344
    },
    {
      "epoch": 10.912420262664165,
      "grad_norm": 0.5579130053520203,
      "learning_rate": 0.002828,
      "loss": 1.5119,
      "step": 145408
    },
    {
      "epoch": 10.917223264540338,
      "grad_norm": 0.5804506540298462,
      "learning_rate": 0.002828,
      "loss": 1.5136,
      "step": 145472
    },
    {
      "epoch": 10.92202626641651,
      "grad_norm": 0.4968551993370056,
      "learning_rate": 0.002828,
      "loss": 1.5161,
      "step": 145536
    },
    {
      "epoch": 10.926829268292684,
      "grad_norm": 0.447307825088501,
      "learning_rate": 0.002828,
      "loss": 1.5191,
      "step": 145600
    },
    {
      "epoch": 10.931632270168855,
      "grad_norm": 0.5149531960487366,
      "learning_rate": 0.002828,
      "loss": 1.5232,
      "step": 145664
    },
    {
      "epoch": 10.936435272045028,
      "grad_norm": 0.44786104559898376,
      "learning_rate": 0.002828,
      "loss": 1.5176,
      "step": 145728
    },
    {
      "epoch": 10.9412382739212,
      "grad_norm": 0.4521365165710449,
      "learning_rate": 0.002828,
      "loss": 1.5164,
      "step": 145792
    },
    {
      "epoch": 10.946041275797374,
      "grad_norm": 0.5914323329925537,
      "learning_rate": 0.002828,
      "loss": 1.5193,
      "step": 145856
    },
    {
      "epoch": 10.950844277673546,
      "grad_norm": 0.6470093131065369,
      "learning_rate": 0.002828,
      "loss": 1.5125,
      "step": 145920
    },
    {
      "epoch": 10.95564727954972,
      "grad_norm": 0.3939136564731598,
      "learning_rate": 0.002828,
      "loss": 1.5165,
      "step": 145984
    },
    {
      "epoch": 10.96045028142589,
      "grad_norm": 0.6154167056083679,
      "learning_rate": 0.002828,
      "loss": 1.5115,
      "step": 146048
    },
    {
      "epoch": 10.965253283302063,
      "grad_norm": 0.5322780609130859,
      "learning_rate": 0.002828,
      "loss": 1.5189,
      "step": 146112
    },
    {
      "epoch": 10.970056285178236,
      "grad_norm": 0.5299228429794312,
      "learning_rate": 0.002828,
      "loss": 1.5166,
      "step": 146176
    },
    {
      "epoch": 10.974859287054409,
      "grad_norm": 0.4304593801498413,
      "learning_rate": 0.002828,
      "loss": 1.5258,
      "step": 146240
    },
    {
      "epoch": 10.979662288930582,
      "grad_norm": 0.5457515716552734,
      "learning_rate": 0.002828,
      "loss": 1.514,
      "step": 146304
    },
    {
      "epoch": 10.984465290806755,
      "grad_norm": 0.59228515625,
      "learning_rate": 0.002828,
      "loss": 1.5257,
      "step": 146368
    },
    {
      "epoch": 10.989268292682926,
      "grad_norm": 0.5043462514877319,
      "learning_rate": 0.002828,
      "loss": 1.5152,
      "step": 146432
    },
    {
      "epoch": 10.994071294559099,
      "grad_norm": 0.4221332371234894,
      "learning_rate": 0.002828,
      "loss": 1.5159,
      "step": 146496
    },
    {
      "epoch": 10.998874296435272,
      "grad_norm": 0.5075702667236328,
      "learning_rate": 0.002828,
      "loss": 1.5202,
      "step": 146560
    },
    {
      "epoch": 11.003677298311445,
      "grad_norm": 0.5219289064407349,
      "learning_rate": 0.002828,
      "loss": 1.4803,
      "step": 146624
    },
    {
      "epoch": 11.008480300187617,
      "grad_norm": 0.46624791622161865,
      "learning_rate": 0.002828,
      "loss": 1.4725,
      "step": 146688
    },
    {
      "epoch": 11.01328330206379,
      "grad_norm": 0.6117081642150879,
      "learning_rate": 0.002828,
      "loss": 1.4732,
      "step": 146752
    },
    {
      "epoch": 11.018086303939963,
      "grad_norm": 0.48853933811187744,
      "learning_rate": 0.002828,
      "loss": 1.4787,
      "step": 146816
    },
    {
      "epoch": 11.022889305816134,
      "grad_norm": 0.5432941913604736,
      "learning_rate": 0.002828,
      "loss": 1.4667,
      "step": 146880
    },
    {
      "epoch": 11.027692307692307,
      "grad_norm": 0.5087817311286926,
      "learning_rate": 0.002828,
      "loss": 1.4737,
      "step": 146944
    },
    {
      "epoch": 11.03249530956848,
      "grad_norm": 0.5708329677581787,
      "learning_rate": 0.002828,
      "loss": 1.4784,
      "step": 147008
    },
    {
      "epoch": 11.037298311444653,
      "grad_norm": 0.6395474672317505,
      "learning_rate": 0.002828,
      "loss": 1.4751,
      "step": 147072
    },
    {
      "epoch": 11.042101313320826,
      "grad_norm": 0.5034723877906799,
      "learning_rate": 0.002828,
      "loss": 1.4786,
      "step": 147136
    },
    {
      "epoch": 11.046904315196999,
      "grad_norm": 0.5139698386192322,
      "learning_rate": 0.002828,
      "loss": 1.4723,
      "step": 147200
    },
    {
      "epoch": 11.05170731707317,
      "grad_norm": 0.5037325024604797,
      "learning_rate": 0.002828,
      "loss": 1.4724,
      "step": 147264
    },
    {
      "epoch": 11.056510318949343,
      "grad_norm": 0.5322916507720947,
      "learning_rate": 0.002828,
      "loss": 1.4768,
      "step": 147328
    },
    {
      "epoch": 11.061313320825516,
      "grad_norm": 0.5106876492500305,
      "learning_rate": 0.002828,
      "loss": 1.4719,
      "step": 147392
    },
    {
      "epoch": 11.066116322701689,
      "grad_norm": 0.47849833965301514,
      "learning_rate": 0.002828,
      "loss": 1.4784,
      "step": 147456
    },
    {
      "epoch": 11.070919324577861,
      "grad_norm": 0.4921078383922577,
      "learning_rate": 0.002828,
      "loss": 1.4736,
      "step": 147520
    },
    {
      "epoch": 11.075722326454034,
      "grad_norm": 0.43421855568885803,
      "learning_rate": 0.002828,
      "loss": 1.4735,
      "step": 147584
    },
    {
      "epoch": 11.080525328330207,
      "grad_norm": 0.503207266330719,
      "learning_rate": 0.002828,
      "loss": 1.4761,
      "step": 147648
    },
    {
      "epoch": 11.085328330206378,
      "grad_norm": 0.5848730206489563,
      "learning_rate": 0.002828,
      "loss": 1.4719,
      "step": 147712
    },
    {
      "epoch": 11.090131332082551,
      "grad_norm": 0.5984218120574951,
      "learning_rate": 0.002828,
      "loss": 1.4796,
      "step": 147776
    },
    {
      "epoch": 11.094934333958724,
      "grad_norm": 0.4876102805137634,
      "learning_rate": 0.002828,
      "loss": 1.4795,
      "step": 147840
    },
    {
      "epoch": 11.099737335834897,
      "grad_norm": 0.4254842698574066,
      "learning_rate": 0.002828,
      "loss": 1.4785,
      "step": 147904
    },
    {
      "epoch": 11.10454033771107,
      "grad_norm": 0.5368649363517761,
      "learning_rate": 0.002828,
      "loss": 1.4803,
      "step": 147968
    },
    {
      "epoch": 11.109343339587243,
      "grad_norm": 0.4503213167190552,
      "learning_rate": 0.002828,
      "loss": 1.482,
      "step": 148032
    },
    {
      "epoch": 11.114146341463414,
      "grad_norm": 0.5051832795143127,
      "learning_rate": 0.002828,
      "loss": 1.4788,
      "step": 148096
    },
    {
      "epoch": 11.118949343339587,
      "grad_norm": 0.5128603577613831,
      "learning_rate": 0.002828,
      "loss": 1.4816,
      "step": 148160
    },
    {
      "epoch": 11.12375234521576,
      "grad_norm": 0.5454185605049133,
      "learning_rate": 0.002828,
      "loss": 1.4847,
      "step": 148224
    },
    {
      "epoch": 11.128555347091933,
      "grad_norm": 0.575675904750824,
      "learning_rate": 0.002828,
      "loss": 1.4874,
      "step": 148288
    },
    {
      "epoch": 11.133358348968105,
      "grad_norm": 0.49659401178359985,
      "learning_rate": 0.002828,
      "loss": 1.4844,
      "step": 148352
    },
    {
      "epoch": 11.138161350844278,
      "grad_norm": 0.4814682900905609,
      "learning_rate": 0.002828,
      "loss": 1.476,
      "step": 148416
    },
    {
      "epoch": 11.142964352720451,
      "grad_norm": 0.46637871861457825,
      "learning_rate": 0.002828,
      "loss": 1.4804,
      "step": 148480
    },
    {
      "epoch": 11.147767354596622,
      "grad_norm": 0.8030520677566528,
      "learning_rate": 0.002828,
      "loss": 1.4828,
      "step": 148544
    },
    {
      "epoch": 11.152570356472795,
      "grad_norm": 0.4283498525619507,
      "learning_rate": 0.002828,
      "loss": 1.4833,
      "step": 148608
    },
    {
      "epoch": 11.157373358348968,
      "grad_norm": 0.7140107154846191,
      "learning_rate": 0.002828,
      "loss": 1.4843,
      "step": 148672
    },
    {
      "epoch": 11.162176360225141,
      "grad_norm": 0.45211201906204224,
      "learning_rate": 0.002828,
      "loss": 1.4764,
      "step": 148736
    },
    {
      "epoch": 11.166979362101314,
      "grad_norm": 0.5594805479049683,
      "learning_rate": 0.002828,
      "loss": 1.4868,
      "step": 148800
    },
    {
      "epoch": 11.171782363977487,
      "grad_norm": 0.5397804975509644,
      "learning_rate": 0.002828,
      "loss": 1.4797,
      "step": 148864
    },
    {
      "epoch": 11.176585365853658,
      "grad_norm": 0.4983387887477875,
      "learning_rate": 0.002828,
      "loss": 1.4846,
      "step": 148928
    },
    {
      "epoch": 11.18138836772983,
      "grad_norm": 0.48610880970954895,
      "learning_rate": 0.002828,
      "loss": 1.4807,
      "step": 148992
    },
    {
      "epoch": 11.186191369606004,
      "grad_norm": 0.4960896968841553,
      "learning_rate": 0.002828,
      "loss": 1.4833,
      "step": 149056
    },
    {
      "epoch": 11.190994371482176,
      "grad_norm": 0.4940851628780365,
      "learning_rate": 0.002828,
      "loss": 1.4898,
      "step": 149120
    },
    {
      "epoch": 11.19579737335835,
      "grad_norm": 0.5122717618942261,
      "learning_rate": 0.002828,
      "loss": 1.483,
      "step": 149184
    },
    {
      "epoch": 11.200600375234522,
      "grad_norm": 0.4835565984249115,
      "learning_rate": 0.002828,
      "loss": 1.4833,
      "step": 149248
    },
    {
      "epoch": 11.205403377110693,
      "grad_norm": 0.5495861768722534,
      "learning_rate": 0.002828,
      "loss": 1.4825,
      "step": 149312
    },
    {
      "epoch": 11.210206378986866,
      "grad_norm": 0.4611824154853821,
      "learning_rate": 0.002828,
      "loss": 1.4819,
      "step": 149376
    },
    {
      "epoch": 11.21500938086304,
      "grad_norm": 0.47973930835723877,
      "learning_rate": 0.002828,
      "loss": 1.4839,
      "step": 149440
    },
    {
      "epoch": 11.219812382739212,
      "grad_norm": 0.5280267596244812,
      "learning_rate": 0.002828,
      "loss": 1.4813,
      "step": 149504
    },
    {
      "epoch": 11.224615384615385,
      "grad_norm": 0.5254800915718079,
      "learning_rate": 0.002828,
      "loss": 1.4843,
      "step": 149568
    },
    {
      "epoch": 11.229418386491558,
      "grad_norm": 0.5418986678123474,
      "learning_rate": 0.002828,
      "loss": 1.4887,
      "step": 149632
    },
    {
      "epoch": 11.23422138836773,
      "grad_norm": 0.522598147392273,
      "learning_rate": 0.002828,
      "loss": 1.4828,
      "step": 149696
    },
    {
      "epoch": 11.239024390243902,
      "grad_norm": 0.47169050574302673,
      "learning_rate": 0.002828,
      "loss": 1.4799,
      "step": 149760
    },
    {
      "epoch": 11.243827392120075,
      "grad_norm": 0.5565853714942932,
      "learning_rate": 0.002828,
      "loss": 1.4808,
      "step": 149824
    },
    {
      "epoch": 11.248630393996248,
      "grad_norm": 0.571573793888092,
      "learning_rate": 0.002828,
      "loss": 1.4843,
      "step": 149888
    },
    {
      "epoch": 11.25343339587242,
      "grad_norm": 0.5259160995483398,
      "learning_rate": 0.002828,
      "loss": 1.4854,
      "step": 149952
    },
    {
      "epoch": 11.258236397748593,
      "grad_norm": 0.4677402079105377,
      "learning_rate": 0.002828,
      "loss": 1.4822,
      "step": 150016
    },
    {
      "epoch": 11.263039399624766,
      "grad_norm": 0.5765787959098816,
      "learning_rate": 0.002828,
      "loss": 1.4877,
      "step": 150080
    },
    {
      "epoch": 11.267842401500937,
      "grad_norm": 0.6456829905509949,
      "learning_rate": 0.002828,
      "loss": 1.4806,
      "step": 150144
    },
    {
      "epoch": 11.27264540337711,
      "grad_norm": 0.4820854961872101,
      "learning_rate": 0.002828,
      "loss": 1.489,
      "step": 150208
    },
    {
      "epoch": 11.277448405253283,
      "grad_norm": 0.4561883807182312,
      "learning_rate": 0.002828,
      "loss": 1.4794,
      "step": 150272
    },
    {
      "epoch": 11.282251407129456,
      "grad_norm": 0.47787246108055115,
      "learning_rate": 0.002828,
      "loss": 1.4902,
      "step": 150336
    },
    {
      "epoch": 11.287054409005629,
      "grad_norm": 0.5143373012542725,
      "learning_rate": 0.002828,
      "loss": 1.4867,
      "step": 150400
    },
    {
      "epoch": 11.291857410881802,
      "grad_norm": 0.5364919304847717,
      "learning_rate": 0.002828,
      "loss": 1.4846,
      "step": 150464
    },
    {
      "epoch": 11.296660412757973,
      "grad_norm": 0.5826845169067383,
      "learning_rate": 0.002828,
      "loss": 1.4826,
      "step": 150528
    },
    {
      "epoch": 11.301463414634146,
      "grad_norm": 0.542300820350647,
      "learning_rate": 0.002828,
      "loss": 1.4848,
      "step": 150592
    },
    {
      "epoch": 11.306266416510319,
      "grad_norm": 0.4969736933708191,
      "learning_rate": 0.002828,
      "loss": 1.4897,
      "step": 150656
    },
    {
      "epoch": 11.311069418386491,
      "grad_norm": 0.4292810559272766,
      "learning_rate": 0.002828,
      "loss": 1.4799,
      "step": 150720
    },
    {
      "epoch": 11.315872420262664,
      "grad_norm": 0.5050080418586731,
      "learning_rate": 0.002828,
      "loss": 1.4905,
      "step": 150784
    },
    {
      "epoch": 11.320675422138837,
      "grad_norm": 0.45646074414253235,
      "learning_rate": 0.002828,
      "loss": 1.4905,
      "step": 150848
    },
    {
      "epoch": 11.32547842401501,
      "grad_norm": 0.4377295970916748,
      "learning_rate": 0.002828,
      "loss": 1.4776,
      "step": 150912
    },
    {
      "epoch": 11.330281425891181,
      "grad_norm": 0.6370419263839722,
      "learning_rate": 0.002828,
      "loss": 1.4773,
      "step": 150976
    },
    {
      "epoch": 11.335084427767354,
      "grad_norm": 0.48091229796409607,
      "learning_rate": 0.002828,
      "loss": 1.4845,
      "step": 151040
    },
    {
      "epoch": 11.339887429643527,
      "grad_norm": 0.5427354574203491,
      "learning_rate": 0.002828,
      "loss": 1.4856,
      "step": 151104
    },
    {
      "epoch": 11.3446904315197,
      "grad_norm": 0.5507506728172302,
      "learning_rate": 0.002828,
      "loss": 1.4907,
      "step": 151168
    },
    {
      "epoch": 11.349493433395873,
      "grad_norm": 0.5161782503128052,
      "learning_rate": 0.002828,
      "loss": 1.483,
      "step": 151232
    },
    {
      "epoch": 11.354296435272046,
      "grad_norm": 0.4885135293006897,
      "learning_rate": 0.002828,
      "loss": 1.4919,
      "step": 151296
    },
    {
      "epoch": 11.359099437148217,
      "grad_norm": 0.5884434580802917,
      "learning_rate": 0.002828,
      "loss": 1.4847,
      "step": 151360
    },
    {
      "epoch": 11.36390243902439,
      "grad_norm": 0.46827441453933716,
      "learning_rate": 0.002828,
      "loss": 1.4812,
      "step": 151424
    },
    {
      "epoch": 11.368705440900563,
      "grad_norm": 0.4739481508731842,
      "learning_rate": 0.002828,
      "loss": 1.4819,
      "step": 151488
    },
    {
      "epoch": 11.373508442776735,
      "grad_norm": 0.42353928089141846,
      "learning_rate": 0.002828,
      "loss": 1.4871,
      "step": 151552
    },
    {
      "epoch": 11.378311444652908,
      "grad_norm": 0.5274419784545898,
      "learning_rate": 0.002828,
      "loss": 1.4909,
      "step": 151616
    },
    {
      "epoch": 11.383114446529081,
      "grad_norm": 0.5639366507530212,
      "learning_rate": 0.002828,
      "loss": 1.4842,
      "step": 151680
    },
    {
      "epoch": 11.387917448405254,
      "grad_norm": 0.556846559047699,
      "learning_rate": 0.002828,
      "loss": 1.489,
      "step": 151744
    },
    {
      "epoch": 11.392720450281425,
      "grad_norm": 0.5005077123641968,
      "learning_rate": 0.002828,
      "loss": 1.4919,
      "step": 151808
    },
    {
      "epoch": 11.397523452157598,
      "grad_norm": 0.4254924952983856,
      "learning_rate": 0.002828,
      "loss": 1.493,
      "step": 151872
    },
    {
      "epoch": 11.402326454033771,
      "grad_norm": 0.5749356150627136,
      "learning_rate": 0.002828,
      "loss": 1.4806,
      "step": 151936
    },
    {
      "epoch": 11.407129455909944,
      "grad_norm": 0.5827454924583435,
      "learning_rate": 0.002828,
      "loss": 1.4886,
      "step": 152000
    },
    {
      "epoch": 11.411932457786117,
      "grad_norm": 0.5152413249015808,
      "learning_rate": 0.002828,
      "loss": 1.4867,
      "step": 152064
    },
    {
      "epoch": 11.41673545966229,
      "grad_norm": 0.5386618375778198,
      "learning_rate": 0.002828,
      "loss": 1.4914,
      "step": 152128
    },
    {
      "epoch": 11.42153846153846,
      "grad_norm": 0.4550732970237732,
      "learning_rate": 0.002828,
      "loss": 1.4839,
      "step": 152192
    },
    {
      "epoch": 11.426341463414634,
      "grad_norm": 0.45811569690704346,
      "learning_rate": 0.002828,
      "loss": 1.4843,
      "step": 152256
    },
    {
      "epoch": 11.431144465290807,
      "grad_norm": 0.6061787009239197,
      "learning_rate": 0.002828,
      "loss": 1.4828,
      "step": 152320
    },
    {
      "epoch": 11.43594746716698,
      "grad_norm": 0.54230797290802,
      "learning_rate": 0.002828,
      "loss": 1.4771,
      "step": 152384
    },
    {
      "epoch": 11.440750469043152,
      "grad_norm": 0.5468747019767761,
      "learning_rate": 0.002828,
      "loss": 1.4896,
      "step": 152448
    },
    {
      "epoch": 11.445553470919325,
      "grad_norm": 0.46672895550727844,
      "learning_rate": 0.002828,
      "loss": 1.4888,
      "step": 152512
    },
    {
      "epoch": 11.450356472795498,
      "grad_norm": 0.4434189796447754,
      "learning_rate": 0.002828,
      "loss": 1.4787,
      "step": 152576
    },
    {
      "epoch": 11.45515947467167,
      "grad_norm": 0.5849834084510803,
      "learning_rate": 0.002828,
      "loss": 1.4842,
      "step": 152640
    },
    {
      "epoch": 11.459962476547842,
      "grad_norm": 0.5004452466964722,
      "learning_rate": 0.002828,
      "loss": 1.4863,
      "step": 152704
    },
    {
      "epoch": 11.464765478424015,
      "grad_norm": 0.520357072353363,
      "learning_rate": 0.002828,
      "loss": 1.487,
      "step": 152768
    },
    {
      "epoch": 11.469568480300188,
      "grad_norm": 0.48489540815353394,
      "learning_rate": 0.002828,
      "loss": 1.4887,
      "step": 152832
    },
    {
      "epoch": 11.47437148217636,
      "grad_norm": 0.4426417052745819,
      "learning_rate": 0.002828,
      "loss": 1.4851,
      "step": 152896
    },
    {
      "epoch": 11.479174484052534,
      "grad_norm": 0.6546865701675415,
      "learning_rate": 0.002828,
      "loss": 1.4899,
      "step": 152960
    },
    {
      "epoch": 11.483977485928705,
      "grad_norm": 0.5141983032226562,
      "learning_rate": 0.002828,
      "loss": 1.49,
      "step": 153024
    },
    {
      "epoch": 11.488780487804878,
      "grad_norm": 0.6321448683738708,
      "learning_rate": 0.002828,
      "loss": 1.4817,
      "step": 153088
    },
    {
      "epoch": 11.49358348968105,
      "grad_norm": 0.5375384092330933,
      "learning_rate": 0.002828,
      "loss": 1.4815,
      "step": 153152
    },
    {
      "epoch": 11.498386491557223,
      "grad_norm": 0.5649176239967346,
      "learning_rate": 0.002828,
      "loss": 1.4851,
      "step": 153216
    },
    {
      "epoch": 11.503189493433396,
      "grad_norm": 0.5667992234230042,
      "learning_rate": 0.002828,
      "loss": 1.4843,
      "step": 153280
    },
    {
      "epoch": 11.50799249530957,
      "grad_norm": 0.4731646776199341,
      "learning_rate": 0.002828,
      "loss": 1.4861,
      "step": 153344
    },
    {
      "epoch": 11.512795497185742,
      "grad_norm": 0.4973836839199066,
      "learning_rate": 0.002828,
      "loss": 1.4858,
      "step": 153408
    },
    {
      "epoch": 11.517598499061913,
      "grad_norm": 0.4976014792919159,
      "learning_rate": 0.002828,
      "loss": 1.4814,
      "step": 153472
    },
    {
      "epoch": 11.522401500938086,
      "grad_norm": 0.48606857657432556,
      "learning_rate": 0.002828,
      "loss": 1.4873,
      "step": 153536
    },
    {
      "epoch": 11.527204502814259,
      "grad_norm": 0.5743147730827332,
      "learning_rate": 0.002828,
      "loss": 1.4874,
      "step": 153600
    },
    {
      "epoch": 11.532007504690432,
      "grad_norm": 0.5242615342140198,
      "learning_rate": 0.002828,
      "loss": 1.4842,
      "step": 153664
    },
    {
      "epoch": 11.536810506566605,
      "grad_norm": 0.5626909732818604,
      "learning_rate": 0.002828,
      "loss": 1.4839,
      "step": 153728
    },
    {
      "epoch": 11.541613508442778,
      "grad_norm": 0.4896009564399719,
      "learning_rate": 0.002828,
      "loss": 1.4885,
      "step": 153792
    },
    {
      "epoch": 11.546416510318949,
      "grad_norm": 0.5722072720527649,
      "learning_rate": 0.002828,
      "loss": 1.4835,
      "step": 153856
    },
    {
      "epoch": 11.551219512195122,
      "grad_norm": 0.4468698501586914,
      "learning_rate": 0.002828,
      "loss": 1.4905,
      "step": 153920
    },
    {
      "epoch": 11.556022514071294,
      "grad_norm": 0.46148356795310974,
      "learning_rate": 0.002828,
      "loss": 1.4859,
      "step": 153984
    },
    {
      "epoch": 11.560825515947467,
      "grad_norm": 0.5205918550491333,
      "learning_rate": 0.002828,
      "loss": 1.4945,
      "step": 154048
    },
    {
      "epoch": 11.56562851782364,
      "grad_norm": 0.5110222697257996,
      "learning_rate": 0.002828,
      "loss": 1.4885,
      "step": 154112
    },
    {
      "epoch": 11.570431519699813,
      "grad_norm": 0.5615659356117249,
      "learning_rate": 0.002828,
      "loss": 1.4814,
      "step": 154176
    },
    {
      "epoch": 11.575234521575984,
      "grad_norm": 0.5684924125671387,
      "learning_rate": 0.002828,
      "loss": 1.4904,
      "step": 154240
    },
    {
      "epoch": 11.580037523452157,
      "grad_norm": 0.5970752239227295,
      "learning_rate": 0.002828,
      "loss": 1.4937,
      "step": 154304
    },
    {
      "epoch": 11.58484052532833,
      "grad_norm": 0.5680916905403137,
      "learning_rate": 0.002828,
      "loss": 1.4936,
      "step": 154368
    },
    {
      "epoch": 11.589643527204503,
      "grad_norm": 0.6049686670303345,
      "learning_rate": 0.002828,
      "loss": 1.4866,
      "step": 154432
    },
    {
      "epoch": 11.594446529080676,
      "grad_norm": 0.5732573866844177,
      "learning_rate": 0.002828,
      "loss": 1.4913,
      "step": 154496
    },
    {
      "epoch": 11.599249530956849,
      "grad_norm": 0.5191873908042908,
      "learning_rate": 0.002828,
      "loss": 1.4863,
      "step": 154560
    },
    {
      "epoch": 11.60405253283302,
      "grad_norm": 0.4945391118526459,
      "learning_rate": 0.002828,
      "loss": 1.4907,
      "step": 154624
    },
    {
      "epoch": 11.608855534709193,
      "grad_norm": 0.5700604915618896,
      "learning_rate": 0.002828,
      "loss": 1.4872,
      "step": 154688
    },
    {
      "epoch": 11.613658536585366,
      "grad_norm": 0.5110488533973694,
      "learning_rate": 0.002828,
      "loss": 1.4877,
      "step": 154752
    },
    {
      "epoch": 11.618461538461538,
      "grad_norm": 0.5065351724624634,
      "learning_rate": 0.002828,
      "loss": 1.4922,
      "step": 154816
    },
    {
      "epoch": 11.623264540337711,
      "grad_norm": 0.4655206799507141,
      "learning_rate": 0.002828,
      "loss": 1.4875,
      "step": 154880
    },
    {
      "epoch": 11.628067542213884,
      "grad_norm": 0.5089915990829468,
      "learning_rate": 0.002828,
      "loss": 1.4876,
      "step": 154944
    },
    {
      "epoch": 11.632870544090057,
      "grad_norm": 0.48405689001083374,
      "learning_rate": 0.002828,
      "loss": 1.4834,
      "step": 155008
    },
    {
      "epoch": 11.637673545966228,
      "grad_norm": 0.49810877442359924,
      "learning_rate": 0.002828,
      "loss": 1.4911,
      "step": 155072
    },
    {
      "epoch": 11.642476547842401,
      "grad_norm": 0.5092597603797913,
      "learning_rate": 0.002828,
      "loss": 1.4919,
      "step": 155136
    },
    {
      "epoch": 11.647279549718574,
      "grad_norm": 0.5448451042175293,
      "learning_rate": 0.002828,
      "loss": 1.4886,
      "step": 155200
    },
    {
      "epoch": 11.652082551594747,
      "grad_norm": 0.5183640718460083,
      "learning_rate": 0.002828,
      "loss": 1.4883,
      "step": 155264
    },
    {
      "epoch": 11.65688555347092,
      "grad_norm": 0.47896742820739746,
      "learning_rate": 0.002828,
      "loss": 1.4827,
      "step": 155328
    },
    {
      "epoch": 11.661688555347093,
      "grad_norm": 0.4914989769458771,
      "learning_rate": 0.002828,
      "loss": 1.488,
      "step": 155392
    },
    {
      "epoch": 11.666491557223264,
      "grad_norm": 0.44482043385505676,
      "learning_rate": 0.002828,
      "loss": 1.4831,
      "step": 155456
    },
    {
      "epoch": 11.671294559099437,
      "grad_norm": 0.514543354511261,
      "learning_rate": 0.002828,
      "loss": 1.4967,
      "step": 155520
    },
    {
      "epoch": 11.67609756097561,
      "grad_norm": 0.4731675088405609,
      "learning_rate": 0.002828,
      "loss": 1.4877,
      "step": 155584
    },
    {
      "epoch": 11.680900562851782,
      "grad_norm": 0.602486252784729,
      "learning_rate": 0.002828,
      "loss": 1.4857,
      "step": 155648
    },
    {
      "epoch": 11.685703564727955,
      "grad_norm": 0.5217562913894653,
      "learning_rate": 0.002828,
      "loss": 1.4924,
      "step": 155712
    },
    {
      "epoch": 11.690506566604128,
      "grad_norm": 0.5473820567131042,
      "learning_rate": 0.002828,
      "loss": 1.4945,
      "step": 155776
    },
    {
      "epoch": 11.695309568480301,
      "grad_norm": 0.5123807787895203,
      "learning_rate": 0.002828,
      "loss": 1.4881,
      "step": 155840
    },
    {
      "epoch": 11.700112570356472,
      "grad_norm": 0.5650917291641235,
      "learning_rate": 0.002828,
      "loss": 1.4841,
      "step": 155904
    },
    {
      "epoch": 11.704915572232645,
      "grad_norm": 0.49404966831207275,
      "learning_rate": 0.002828,
      "loss": 1.4832,
      "step": 155968
    },
    {
      "epoch": 11.709718574108818,
      "grad_norm": 0.5131070613861084,
      "learning_rate": 0.002828,
      "loss": 1.4874,
      "step": 156032
    },
    {
      "epoch": 11.71452157598499,
      "grad_norm": 0.5612493753433228,
      "learning_rate": 0.002828,
      "loss": 1.4853,
      "step": 156096
    },
    {
      "epoch": 11.719324577861164,
      "grad_norm": 0.5529559850692749,
      "learning_rate": 0.002828,
      "loss": 1.4903,
      "step": 156160
    },
    {
      "epoch": 11.724127579737337,
      "grad_norm": 0.44944292306900024,
      "learning_rate": 0.002828,
      "loss": 1.4862,
      "step": 156224
    },
    {
      "epoch": 11.728930581613508,
      "grad_norm": 0.5775054693222046,
      "learning_rate": 0.002828,
      "loss": 1.4899,
      "step": 156288
    },
    {
      "epoch": 11.73373358348968,
      "grad_norm": 0.4601612985134125,
      "learning_rate": 0.002828,
      "loss": 1.4835,
      "step": 156352
    },
    {
      "epoch": 11.738536585365853,
      "grad_norm": 0.4491596817970276,
      "learning_rate": 0.002828,
      "loss": 1.4844,
      "step": 156416
    },
    {
      "epoch": 11.743339587242026,
      "grad_norm": 0.46096885204315186,
      "learning_rate": 0.002828,
      "loss": 1.486,
      "step": 156480
    },
    {
      "epoch": 11.7481425891182,
      "grad_norm": 0.5358552932739258,
      "learning_rate": 0.002828,
      "loss": 1.4839,
      "step": 156544
    },
    {
      "epoch": 11.752945590994372,
      "grad_norm": 0.6074651479721069,
      "learning_rate": 0.002828,
      "loss": 1.4822,
      "step": 156608
    },
    {
      "epoch": 11.757748592870545,
      "grad_norm": 0.5014597773551941,
      "learning_rate": 0.002828,
      "loss": 1.4812,
      "step": 156672
    },
    {
      "epoch": 11.762551594746716,
      "grad_norm": 0.41438910365104675,
      "learning_rate": 0.002828,
      "loss": 1.4874,
      "step": 156736
    },
    {
      "epoch": 11.767354596622889,
      "grad_norm": 0.5457605123519897,
      "learning_rate": 0.002828,
      "loss": 1.481,
      "step": 156800
    },
    {
      "epoch": 11.772157598499062,
      "grad_norm": 0.4685835540294647,
      "learning_rate": 0.002828,
      "loss": 1.4837,
      "step": 156864
    },
    {
      "epoch": 11.776960600375235,
      "grad_norm": 0.4858347177505493,
      "learning_rate": 0.002828,
      "loss": 1.4844,
      "step": 156928
    },
    {
      "epoch": 11.781763602251408,
      "grad_norm": 0.5601338148117065,
      "learning_rate": 0.002828,
      "loss": 1.489,
      "step": 156992
    },
    {
      "epoch": 11.78656660412758,
      "grad_norm": 0.4982069730758667,
      "learning_rate": 0.002828,
      "loss": 1.4902,
      "step": 157056
    },
    {
      "epoch": 11.791369606003752,
      "grad_norm": 0.5538622140884399,
      "learning_rate": 0.002828,
      "loss": 1.4853,
      "step": 157120
    },
    {
      "epoch": 11.796172607879924,
      "grad_norm": 0.5723830461502075,
      "learning_rate": 0.002828,
      "loss": 1.4867,
      "step": 157184
    },
    {
      "epoch": 11.800975609756097,
      "grad_norm": 0.4833488464355469,
      "learning_rate": 0.002828,
      "loss": 1.492,
      "step": 157248
    },
    {
      "epoch": 11.80577861163227,
      "grad_norm": 0.5311214923858643,
      "learning_rate": 0.002828,
      "loss": 1.4929,
      "step": 157312
    },
    {
      "epoch": 11.810581613508443,
      "grad_norm": 0.5109043121337891,
      "learning_rate": 0.002828,
      "loss": 1.4834,
      "step": 157376
    },
    {
      "epoch": 11.815384615384616,
      "grad_norm": 0.5096741318702698,
      "learning_rate": 0.002828,
      "loss": 1.4911,
      "step": 157440
    },
    {
      "epoch": 11.820187617260789,
      "grad_norm": 0.4858244061470032,
      "learning_rate": 0.002828,
      "loss": 1.485,
      "step": 157504
    },
    {
      "epoch": 11.82499061913696,
      "grad_norm": 0.4737776219844818,
      "learning_rate": 0.002828,
      "loss": 1.4922,
      "step": 157568
    },
    {
      "epoch": 11.829793621013133,
      "grad_norm": 0.5020055174827576,
      "learning_rate": 0.002828,
      "loss": 1.482,
      "step": 157632
    },
    {
      "epoch": 11.834596622889306,
      "grad_norm": 0.49587440490722656,
      "learning_rate": 0.002828,
      "loss": 1.4894,
      "step": 157696
    },
    {
      "epoch": 11.839399624765479,
      "grad_norm": 0.4922964870929718,
      "learning_rate": 0.002828,
      "loss": 1.4859,
      "step": 157760
    },
    {
      "epoch": 11.844202626641652,
      "grad_norm": 0.46961134672164917,
      "learning_rate": 0.002828,
      "loss": 1.4868,
      "step": 157824
    },
    {
      "epoch": 11.849005628517824,
      "grad_norm": 0.5366621613502502,
      "learning_rate": 0.002828,
      "loss": 1.4908,
      "step": 157888
    },
    {
      "epoch": 11.853808630393996,
      "grad_norm": 0.4588063955307007,
      "learning_rate": 0.002828,
      "loss": 1.4869,
      "step": 157952
    },
    {
      "epoch": 11.858611632270168,
      "grad_norm": 0.49376556277275085,
      "learning_rate": 0.002828,
      "loss": 1.4841,
      "step": 158016
    },
    {
      "epoch": 11.863414634146341,
      "grad_norm": 0.48326173424720764,
      "learning_rate": 0.002828,
      "loss": 1.4924,
      "step": 158080
    },
    {
      "epoch": 11.868217636022514,
      "grad_norm": 0.49264848232269287,
      "learning_rate": 0.002828,
      "loss": 1.4868,
      "step": 158144
    },
    {
      "epoch": 11.873020637898687,
      "grad_norm": 0.5427521467208862,
      "learning_rate": 0.002828,
      "loss": 1.4907,
      "step": 158208
    },
    {
      "epoch": 11.87782363977486,
      "grad_norm": 0.5619401335716248,
      "learning_rate": 0.002828,
      "loss": 1.4885,
      "step": 158272
    },
    {
      "epoch": 11.882626641651031,
      "grad_norm": 0.5136578679084778,
      "learning_rate": 0.002828,
      "loss": 1.4857,
      "step": 158336
    },
    {
      "epoch": 11.887429643527204,
      "grad_norm": 0.5775615572929382,
      "learning_rate": 0.002828,
      "loss": 1.4838,
      "step": 158400
    },
    {
      "epoch": 11.892232645403377,
      "grad_norm": 0.5155749917030334,
      "learning_rate": 0.002828,
      "loss": 1.4913,
      "step": 158464
    },
    {
      "epoch": 11.89703564727955,
      "grad_norm": 0.4589896500110626,
      "learning_rate": 0.002828,
      "loss": 1.4813,
      "step": 158528
    },
    {
      "epoch": 11.901838649155723,
      "grad_norm": 0.5712898969650269,
      "learning_rate": 0.002828,
      "loss": 1.4876,
      "step": 158592
    },
    {
      "epoch": 11.906641651031896,
      "grad_norm": 0.5223777890205383,
      "learning_rate": 0.002828,
      "loss": 1.4878,
      "step": 158656
    },
    {
      "epoch": 11.911444652908067,
      "grad_norm": 0.5290514230728149,
      "learning_rate": 0.002828,
      "loss": 1.4828,
      "step": 158720
    },
    {
      "epoch": 11.91624765478424,
      "grad_norm": 0.464539110660553,
      "learning_rate": 0.002828,
      "loss": 1.4851,
      "step": 158784
    },
    {
      "epoch": 11.921050656660412,
      "grad_norm": 0.5479804873466492,
      "learning_rate": 0.002828,
      "loss": 1.4893,
      "step": 158848
    },
    {
      "epoch": 11.925853658536585,
      "grad_norm": 0.5218388438224792,
      "learning_rate": 0.002828,
      "loss": 1.4898,
      "step": 158912
    },
    {
      "epoch": 11.930656660412758,
      "grad_norm": 0.5520194172859192,
      "learning_rate": 0.002828,
      "loss": 1.4862,
      "step": 158976
    },
    {
      "epoch": 11.935459662288931,
      "grad_norm": 0.6023911833763123,
      "learning_rate": 0.002828,
      "loss": 1.4897,
      "step": 159040
    },
    {
      "epoch": 11.940262664165104,
      "grad_norm": 0.522139847278595,
      "learning_rate": 0.002828,
      "loss": 1.4873,
      "step": 159104
    },
    {
      "epoch": 11.945065666041275,
      "grad_norm": 0.43739432096481323,
      "learning_rate": 0.002828,
      "loss": 1.4874,
      "step": 159168
    },
    {
      "epoch": 11.949868667917448,
      "grad_norm": 0.4542504847049713,
      "learning_rate": 0.002828,
      "loss": 1.4844,
      "step": 159232
    },
    {
      "epoch": 11.95467166979362,
      "grad_norm": 0.5269852876663208,
      "learning_rate": 0.002828,
      "loss": 1.4904,
      "step": 159296
    },
    {
      "epoch": 11.959474671669794,
      "grad_norm": 0.480561763048172,
      "learning_rate": 0.002828,
      "loss": 1.4884,
      "step": 159360
    },
    {
      "epoch": 11.964277673545967,
      "grad_norm": 0.47597286105155945,
      "learning_rate": 0.002828,
      "loss": 1.4833,
      "step": 159424
    },
    {
      "epoch": 11.96908067542214,
      "grad_norm": 0.49963515996932983,
      "learning_rate": 0.002828,
      "loss": 1.4844,
      "step": 159488
    },
    {
      "epoch": 11.97388367729831,
      "grad_norm": 0.4844549894332886,
      "learning_rate": 0.002828,
      "loss": 1.4933,
      "step": 159552
    },
    {
      "epoch": 11.978686679174483,
      "grad_norm": 0.4828104078769684,
      "learning_rate": 0.002828,
      "loss": 1.4902,
      "step": 159616
    },
    {
      "epoch": 11.983489681050656,
      "grad_norm": 0.5636357665061951,
      "learning_rate": 0.002828,
      "loss": 1.4877,
      "step": 159680
    },
    {
      "epoch": 11.98829268292683,
      "grad_norm": 0.5685691833496094,
      "learning_rate": 0.002828,
      "loss": 1.486,
      "step": 159744
    },
    {
      "epoch": 11.993095684803002,
      "grad_norm": 0.4913961887359619,
      "learning_rate": 0.002828,
      "loss": 1.4894,
      "step": 159808
    },
    {
      "epoch": 11.997898686679175,
      "grad_norm": 0.587879478931427,
      "learning_rate": 0.002828,
      "loss": 1.4858,
      "step": 159872
    },
    {
      "epoch": 12.002701688555348,
      "grad_norm": 0.7443265318870544,
      "learning_rate": 0.002828,
      "loss": 1.4605,
      "step": 159936
    },
    {
      "epoch": 12.007504690431519,
      "grad_norm": 0.5058794617652893,
      "learning_rate": 0.002828,
      "loss": 1.4435,
      "step": 160000
    },
    {
      "epoch": 12.012307692307692,
      "grad_norm": 0.4813808798789978,
      "learning_rate": 0.002828,
      "loss": 1.4437,
      "step": 160064
    },
    {
      "epoch": 12.017110694183865,
      "grad_norm": 0.5733544826507568,
      "learning_rate": 0.002828,
      "loss": 1.4447,
      "step": 160128
    },
    {
      "epoch": 12.021913696060038,
      "grad_norm": 0.48118847608566284,
      "learning_rate": 0.002828,
      "loss": 1.443,
      "step": 160192
    },
    {
      "epoch": 12.02671669793621,
      "grad_norm": 0.4636737108230591,
      "learning_rate": 0.002828,
      "loss": 1.4395,
      "step": 160256
    },
    {
      "epoch": 12.031519699812383,
      "grad_norm": 0.47300007939338684,
      "learning_rate": 0.002828,
      "loss": 1.4397,
      "step": 160320
    },
    {
      "epoch": 12.036322701688555,
      "grad_norm": 0.5260705351829529,
      "learning_rate": 0.002828,
      "loss": 1.4362,
      "step": 160384
    },
    {
      "epoch": 12.041125703564727,
      "grad_norm": 0.5173159241676331,
      "learning_rate": 0.002828,
      "loss": 1.447,
      "step": 160448
    },
    {
      "epoch": 12.0459287054409,
      "grad_norm": 0.49044883251190186,
      "learning_rate": 0.002828,
      "loss": 1.4443,
      "step": 160512
    },
    {
      "epoch": 12.050731707317073,
      "grad_norm": 0.5307776927947998,
      "learning_rate": 0.002828,
      "loss": 1.449,
      "step": 160576
    },
    {
      "epoch": 12.055534709193246,
      "grad_norm": 0.5528637170791626,
      "learning_rate": 0.002828,
      "loss": 1.4418,
      "step": 160640
    },
    {
      "epoch": 12.060337711069419,
      "grad_norm": 0.5058019161224365,
      "learning_rate": 0.002828,
      "loss": 1.4444,
      "step": 160704
    },
    {
      "epoch": 12.065140712945592,
      "grad_norm": 0.46592089533805847,
      "learning_rate": 0.002828,
      "loss": 1.4402,
      "step": 160768
    },
    {
      "epoch": 12.069943714821763,
      "grad_norm": 0.5039073824882507,
      "learning_rate": 0.002828,
      "loss": 1.4452,
      "step": 160832
    },
    {
      "epoch": 12.074746716697936,
      "grad_norm": 0.5121236443519592,
      "learning_rate": 0.002828,
      "loss": 1.4534,
      "step": 160896
    },
    {
      "epoch": 12.079549718574109,
      "grad_norm": 0.5076376795768738,
      "learning_rate": 0.002828,
      "loss": 1.4457,
      "step": 160960
    },
    {
      "epoch": 12.084352720450282,
      "grad_norm": 0.6144150495529175,
      "learning_rate": 0.002828,
      "loss": 1.4466,
      "step": 161024
    },
    {
      "epoch": 12.089155722326455,
      "grad_norm": 0.628743052482605,
      "learning_rate": 0.002828,
      "loss": 1.4506,
      "step": 161088
    },
    {
      "epoch": 12.093958724202627,
      "grad_norm": 0.4718782901763916,
      "learning_rate": 0.002828,
      "loss": 1.446,
      "step": 161152
    },
    {
      "epoch": 12.098761726078799,
      "grad_norm": 0.4742676615715027,
      "learning_rate": 0.002828,
      "loss": 1.4543,
      "step": 161216
    },
    {
      "epoch": 12.103564727954971,
      "grad_norm": 0.5861676931381226,
      "learning_rate": 0.002828,
      "loss": 1.4454,
      "step": 161280
    },
    {
      "epoch": 12.108367729831144,
      "grad_norm": 0.6501273512840271,
      "learning_rate": 0.002828,
      "loss": 1.4453,
      "step": 161344
    },
    {
      "epoch": 12.113170731707317,
      "grad_norm": 0.43199998140335083,
      "learning_rate": 0.002828,
      "loss": 1.4449,
      "step": 161408
    },
    {
      "epoch": 12.11797373358349,
      "grad_norm": 0.5195205807685852,
      "learning_rate": 0.002828,
      "loss": 1.4524,
      "step": 161472
    },
    {
      "epoch": 12.122776735459663,
      "grad_norm": 0.5002077221870422,
      "learning_rate": 0.002828,
      "loss": 1.4433,
      "step": 161536
    },
    {
      "epoch": 12.127579737335834,
      "grad_norm": 0.6138117909431458,
      "learning_rate": 0.002828,
      "loss": 1.4477,
      "step": 161600
    },
    {
      "epoch": 12.132382739212007,
      "grad_norm": 0.5133177638053894,
      "learning_rate": 0.002828,
      "loss": 1.4568,
      "step": 161664
    },
    {
      "epoch": 12.13718574108818,
      "grad_norm": 0.5995278358459473,
      "learning_rate": 0.002828,
      "loss": 1.4513,
      "step": 161728
    },
    {
      "epoch": 12.141988742964353,
      "grad_norm": 0.5219971537590027,
      "learning_rate": 0.002828,
      "loss": 1.4549,
      "step": 161792
    },
    {
      "epoch": 12.146791744840526,
      "grad_norm": 0.5502182245254517,
      "learning_rate": 0.002828,
      "loss": 1.4469,
      "step": 161856
    },
    {
      "epoch": 12.151594746716698,
      "grad_norm": 0.5348833799362183,
      "learning_rate": 0.002828,
      "loss": 1.4517,
      "step": 161920
    },
    {
      "epoch": 12.156397748592871,
      "grad_norm": 0.5264079570770264,
      "learning_rate": 0.002828,
      "loss": 1.4439,
      "step": 161984
    },
    {
      "epoch": 12.161200750469042,
      "grad_norm": 0.5887817740440369,
      "learning_rate": 0.002828,
      "loss": 1.449,
      "step": 162048
    },
    {
      "epoch": 12.166003752345215,
      "grad_norm": 0.5023970007896423,
      "learning_rate": 0.002828,
      "loss": 1.4507,
      "step": 162112
    },
    {
      "epoch": 12.170806754221388,
      "grad_norm": 0.5127968192100525,
      "learning_rate": 0.002828,
      "loss": 1.4531,
      "step": 162176
    },
    {
      "epoch": 12.175609756097561,
      "grad_norm": 0.5700290203094482,
      "learning_rate": 0.002828,
      "loss": 1.448,
      "step": 162240
    },
    {
      "epoch": 12.180412757973734,
      "grad_norm": 0.5987260937690735,
      "learning_rate": 0.002828,
      "loss": 1.457,
      "step": 162304
    },
    {
      "epoch": 12.185215759849907,
      "grad_norm": 0.528264582157135,
      "learning_rate": 0.002828,
      "loss": 1.4567,
      "step": 162368
    },
    {
      "epoch": 12.190018761726078,
      "grad_norm": 0.5861890316009521,
      "learning_rate": 0.002828,
      "loss": 1.4477,
      "step": 162432
    },
    {
      "epoch": 12.194821763602251,
      "grad_norm": 0.5477088689804077,
      "learning_rate": 0.002828,
      "loss": 1.4546,
      "step": 162496
    },
    {
      "epoch": 12.199624765478424,
      "grad_norm": 0.5453941822052002,
      "learning_rate": 0.002828,
      "loss": 1.4521,
      "step": 162560
    },
    {
      "epoch": 12.204427767354597,
      "grad_norm": 0.5315229892730713,
      "learning_rate": 0.002828,
      "loss": 1.4498,
      "step": 162624
    },
    {
      "epoch": 12.20923076923077,
      "grad_norm": 0.5340282917022705,
      "learning_rate": 0.002828,
      "loss": 1.4546,
      "step": 162688
    },
    {
      "epoch": 12.214033771106942,
      "grad_norm": 0.5422166585922241,
      "learning_rate": 0.002828,
      "loss": 1.4459,
      "step": 162752
    },
    {
      "epoch": 12.218836772983114,
      "grad_norm": 0.4889850914478302,
      "learning_rate": 0.002828,
      "loss": 1.4544,
      "step": 162816
    },
    {
      "epoch": 12.223639774859286,
      "grad_norm": 0.5909858345985413,
      "learning_rate": 0.002828,
      "loss": 1.4565,
      "step": 162880
    },
    {
      "epoch": 12.22844277673546,
      "grad_norm": 0.5490917563438416,
      "learning_rate": 0.002828,
      "loss": 1.4517,
      "step": 162944
    },
    {
      "epoch": 12.233245778611632,
      "grad_norm": 0.5376546382904053,
      "learning_rate": 0.002828,
      "loss": 1.4559,
      "step": 163008
    },
    {
      "epoch": 12.238048780487805,
      "grad_norm": 0.48789891600608826,
      "learning_rate": 0.002828,
      "loss": 1.4436,
      "step": 163072
    },
    {
      "epoch": 12.242851782363978,
      "grad_norm": 0.5413309335708618,
      "learning_rate": 0.002828,
      "loss": 1.4566,
      "step": 163136
    },
    {
      "epoch": 12.24765478424015,
      "grad_norm": 0.6094352006912231,
      "learning_rate": 0.002828,
      "loss": 1.4552,
      "step": 163200
    },
    {
      "epoch": 12.252457786116322,
      "grad_norm": 0.5055562853813171,
      "learning_rate": 0.002828,
      "loss": 1.4533,
      "step": 163264
    },
    {
      "epoch": 12.257260787992495,
      "grad_norm": 0.47938257455825806,
      "learning_rate": 0.002828,
      "loss": 1.4466,
      "step": 163328
    },
    {
      "epoch": 12.262063789868668,
      "grad_norm": 0.5385513305664062,
      "learning_rate": 0.002828,
      "loss": 1.4508,
      "step": 163392
    },
    {
      "epoch": 12.26686679174484,
      "grad_norm": 0.5873063802719116,
      "learning_rate": 0.002828,
      "loss": 1.4475,
      "step": 163456
    },
    {
      "epoch": 12.271669793621014,
      "grad_norm": 0.4774555265903473,
      "learning_rate": 0.002828,
      "loss": 1.458,
      "step": 163520
    },
    {
      "epoch": 12.276472795497186,
      "grad_norm": 0.5163844227790833,
      "learning_rate": 0.002828,
      "loss": 1.4489,
      "step": 163584
    },
    {
      "epoch": 12.281275797373358,
      "grad_norm": 0.5182616114616394,
      "learning_rate": 0.002828,
      "loss": 1.4518,
      "step": 163648
    },
    {
      "epoch": 12.28607879924953,
      "grad_norm": 0.49142399430274963,
      "learning_rate": 0.002828,
      "loss": 1.4539,
      "step": 163712
    },
    {
      "epoch": 12.290881801125703,
      "grad_norm": 0.5692216753959656,
      "learning_rate": 0.002828,
      "loss": 1.4644,
      "step": 163776
    },
    {
      "epoch": 12.295684803001876,
      "grad_norm": 0.46027058362960815,
      "learning_rate": 0.002828,
      "loss": 1.4474,
      "step": 163840
    },
    {
      "epoch": 12.300487804878049,
      "grad_norm": 0.46703729033470154,
      "learning_rate": 0.002828,
      "loss": 1.4508,
      "step": 163904
    },
    {
      "epoch": 12.305290806754222,
      "grad_norm": 0.500545859336853,
      "learning_rate": 0.002828,
      "loss": 1.455,
      "step": 163968
    },
    {
      "epoch": 12.310093808630395,
      "grad_norm": 0.5360391736030579,
      "learning_rate": 0.002828,
      "loss": 1.4587,
      "step": 164032
    },
    {
      "epoch": 12.314896810506566,
      "grad_norm": 0.4937238097190857,
      "learning_rate": 0.002828,
      "loss": 1.4573,
      "step": 164096
    },
    {
      "epoch": 12.319699812382739,
      "grad_norm": 0.5751468539237976,
      "learning_rate": 0.002828,
      "loss": 1.4583,
      "step": 164160
    },
    {
      "epoch": 12.324502814258912,
      "grad_norm": 0.5222389698028564,
      "learning_rate": 0.002828,
      "loss": 1.4541,
      "step": 164224
    },
    {
      "epoch": 12.329305816135085,
      "grad_norm": 0.4793798327445984,
      "learning_rate": 0.002828,
      "loss": 1.4617,
      "step": 164288
    },
    {
      "epoch": 12.334108818011257,
      "grad_norm": 0.5237858295440674,
      "learning_rate": 0.002828,
      "loss": 1.4543,
      "step": 164352
    },
    {
      "epoch": 12.33891181988743,
      "grad_norm": 0.5332927107810974,
      "learning_rate": 0.002828,
      "loss": 1.4529,
      "step": 164416
    },
    {
      "epoch": 12.343714821763601,
      "grad_norm": 0.5909544825553894,
      "learning_rate": 0.002828,
      "loss": 1.453,
      "step": 164480
    },
    {
      "epoch": 12.348517823639774,
      "grad_norm": 0.49265050888061523,
      "learning_rate": 0.002828,
      "loss": 1.456,
      "step": 164544
    },
    {
      "epoch": 12.353320825515947,
      "grad_norm": 0.5843788385391235,
      "learning_rate": 0.002828,
      "loss": 1.4516,
      "step": 164608
    },
    {
      "epoch": 12.35812382739212,
      "grad_norm": 0.5406148433685303,
      "learning_rate": 0.002828,
      "loss": 1.4583,
      "step": 164672
    },
    {
      "epoch": 12.362926829268293,
      "grad_norm": 0.5167796015739441,
      "learning_rate": 0.002828,
      "loss": 1.4493,
      "step": 164736
    },
    {
      "epoch": 12.367729831144466,
      "grad_norm": 0.48411092162132263,
      "learning_rate": 0.002828,
      "loss": 1.4511,
      "step": 164800
    },
    {
      "epoch": 12.372532833020639,
      "grad_norm": 0.44566309452056885,
      "learning_rate": 0.002828,
      "loss": 1.4623,
      "step": 164864
    },
    {
      "epoch": 12.37733583489681,
      "grad_norm": 0.5984368920326233,
      "learning_rate": 0.002828,
      "loss": 1.4604,
      "step": 164928
    },
    {
      "epoch": 12.382138836772983,
      "grad_norm": 0.5471071600914001,
      "learning_rate": 0.002828,
      "loss": 1.4584,
      "step": 164992
    },
    {
      "epoch": 12.386941838649156,
      "grad_norm": 0.5406402349472046,
      "learning_rate": 0.002828,
      "loss": 1.4611,
      "step": 165056
    },
    {
      "epoch": 12.391744840525329,
      "grad_norm": 0.4552817642688751,
      "learning_rate": 0.002828,
      "loss": 1.4553,
      "step": 165120
    },
    {
      "epoch": 12.396547842401501,
      "grad_norm": 0.569806694984436,
      "learning_rate": 0.002828,
      "loss": 1.4587,
      "step": 165184
    },
    {
      "epoch": 12.401350844277674,
      "grad_norm": 0.6869778037071228,
      "learning_rate": 0.002828,
      "loss": 1.4581,
      "step": 165248
    },
    {
      "epoch": 12.406153846153845,
      "grad_norm": 0.5332195162773132,
      "learning_rate": 0.002828,
      "loss": 1.4506,
      "step": 165312
    },
    {
      "epoch": 12.410956848030018,
      "grad_norm": 0.5285490155220032,
      "learning_rate": 0.002828,
      "loss": 1.4544,
      "step": 165376
    },
    {
      "epoch": 12.415759849906191,
      "grad_norm": 0.5308892726898193,
      "learning_rate": 0.002828,
      "loss": 1.4565,
      "step": 165440
    },
    {
      "epoch": 12.420562851782364,
      "grad_norm": 0.47045472264289856,
      "learning_rate": 0.002828,
      "loss": 1.458,
      "step": 165504
    },
    {
      "epoch": 12.425365853658537,
      "grad_norm": 0.41003862023353577,
      "learning_rate": 0.002828,
      "loss": 1.4614,
      "step": 165568
    },
    {
      "epoch": 12.43016885553471,
      "grad_norm": 0.48407503962516785,
      "learning_rate": 0.002828,
      "loss": 1.4541,
      "step": 165632
    },
    {
      "epoch": 12.434971857410881,
      "grad_norm": 0.5479405522346497,
      "learning_rate": 0.002828,
      "loss": 1.4584,
      "step": 165696
    },
    {
      "epoch": 12.439774859287054,
      "grad_norm": 0.4842681586742401,
      "learning_rate": 0.002828,
      "loss": 1.4513,
      "step": 165760
    },
    {
      "epoch": 12.444577861163227,
      "grad_norm": 0.5198821425437927,
      "learning_rate": 0.002828,
      "loss": 1.4579,
      "step": 165824
    },
    {
      "epoch": 12.4493808630394,
      "grad_norm": 0.5267767310142517,
      "learning_rate": 0.002828,
      "loss": 1.4562,
      "step": 165888
    },
    {
      "epoch": 12.454183864915572,
      "grad_norm": 0.5741159319877625,
      "learning_rate": 0.002828,
      "loss": 1.4495,
      "step": 165952
    },
    {
      "epoch": 12.458986866791745,
      "grad_norm": 0.5260004997253418,
      "learning_rate": 0.002828,
      "loss": 1.4529,
      "step": 166016
    },
    {
      "epoch": 12.463789868667918,
      "grad_norm": 0.4495656490325928,
      "learning_rate": 0.002828,
      "loss": 1.4629,
      "step": 166080
    },
    {
      "epoch": 12.46859287054409,
      "grad_norm": 0.5246948599815369,
      "learning_rate": 0.002828,
      "loss": 1.4578,
      "step": 166144
    },
    {
      "epoch": 12.473395872420262,
      "grad_norm": 0.6091532707214355,
      "learning_rate": 0.002828,
      "loss": 1.4579,
      "step": 166208
    },
    {
      "epoch": 12.478198874296435,
      "grad_norm": 0.4875415563583374,
      "learning_rate": 0.002828,
      "loss": 1.4587,
      "step": 166272
    },
    {
      "epoch": 12.483001876172608,
      "grad_norm": 0.507404625415802,
      "learning_rate": 0.002828,
      "loss": 1.4567,
      "step": 166336
    },
    {
      "epoch": 12.487804878048781,
      "grad_norm": 0.5386779308319092,
      "learning_rate": 0.002828,
      "loss": 1.4536,
      "step": 166400
    },
    {
      "epoch": 12.492607879924954,
      "grad_norm": 0.475137323141098,
      "learning_rate": 0.002828,
      "loss": 1.4544,
      "step": 166464
    },
    {
      "epoch": 12.497410881801125,
      "grad_norm": 0.49984148144721985,
      "learning_rate": 0.002828,
      "loss": 1.4582,
      "step": 166528
    },
    {
      "epoch": 12.502213883677298,
      "grad_norm": 0.4507026672363281,
      "learning_rate": 0.002828,
      "loss": 1.4583,
      "step": 166592
    },
    {
      "epoch": 12.50701688555347,
      "grad_norm": 0.49608319997787476,
      "learning_rate": 0.002828,
      "loss": 1.4612,
      "step": 166656
    },
    {
      "epoch": 12.511819887429644,
      "grad_norm": 0.5014333128929138,
      "learning_rate": 0.002828,
      "loss": 1.4565,
      "step": 166720
    },
    {
      "epoch": 12.516622889305816,
      "grad_norm": 0.45654651522636414,
      "learning_rate": 0.002828,
      "loss": 1.4641,
      "step": 166784
    },
    {
      "epoch": 12.52142589118199,
      "grad_norm": 0.5392569303512573,
      "learning_rate": 0.002828,
      "loss": 1.4652,
      "step": 166848
    },
    {
      "epoch": 12.52622889305816,
      "grad_norm": 0.5011345148086548,
      "learning_rate": 0.002828,
      "loss": 1.4621,
      "step": 166912
    },
    {
      "epoch": 12.531031894934333,
      "grad_norm": 0.5161344408988953,
      "learning_rate": 0.002828,
      "loss": 1.4655,
      "step": 166976
    },
    {
      "epoch": 12.535834896810506,
      "grad_norm": 0.7373325824737549,
      "learning_rate": 0.002828,
      "loss": 1.4647,
      "step": 167040
    },
    {
      "epoch": 12.540637898686679,
      "grad_norm": 0.5518906116485596,
      "learning_rate": 0.002828,
      "loss": 1.4565,
      "step": 167104
    },
    {
      "epoch": 12.545440900562852,
      "grad_norm": 0.4990812838077545,
      "learning_rate": 0.002828,
      "loss": 1.4629,
      "step": 167168
    },
    {
      "epoch": 12.550243902439025,
      "grad_norm": 0.4953933656215668,
      "learning_rate": 0.002828,
      "loss": 1.4595,
      "step": 167232
    },
    {
      "epoch": 12.555046904315198,
      "grad_norm": 0.4726664125919342,
      "learning_rate": 0.002828,
      "loss": 1.4622,
      "step": 167296
    },
    {
      "epoch": 12.559849906191369,
      "grad_norm": 0.4469330310821533,
      "learning_rate": 0.002828,
      "loss": 1.4604,
      "step": 167360
    },
    {
      "epoch": 12.564652908067542,
      "grad_norm": 0.48912984132766724,
      "learning_rate": 0.002828,
      "loss": 1.4563,
      "step": 167424
    },
    {
      "epoch": 12.569455909943715,
      "grad_norm": 0.6089437007904053,
      "learning_rate": 0.002828,
      "loss": 1.4596,
      "step": 167488
    },
    {
      "epoch": 12.574258911819888,
      "grad_norm": 0.4961417019367218,
      "learning_rate": 0.002828,
      "loss": 1.4605,
      "step": 167552
    },
    {
      "epoch": 12.57906191369606,
      "grad_norm": 0.4331898093223572,
      "learning_rate": 0.002828,
      "loss": 1.4646,
      "step": 167616
    },
    {
      "epoch": 12.583864915572233,
      "grad_norm": 0.515690803527832,
      "learning_rate": 0.002828,
      "loss": 1.4604,
      "step": 167680
    },
    {
      "epoch": 12.588667917448404,
      "grad_norm": 0.4997495412826538,
      "learning_rate": 0.002828,
      "loss": 1.4633,
      "step": 167744
    },
    {
      "epoch": 12.593470919324577,
      "grad_norm": 0.4562072157859802,
      "learning_rate": 0.002828,
      "loss": 1.4603,
      "step": 167808
    },
    {
      "epoch": 12.59827392120075,
      "grad_norm": 0.46387189626693726,
      "learning_rate": 0.002828,
      "loss": 1.4592,
      "step": 167872
    },
    {
      "epoch": 12.603076923076923,
      "grad_norm": 0.46868836879730225,
      "learning_rate": 0.002828,
      "loss": 1.4568,
      "step": 167936
    },
    {
      "epoch": 12.607879924953096,
      "grad_norm": 0.5087418556213379,
      "learning_rate": 0.002828,
      "loss": 1.4514,
      "step": 168000
    },
    {
      "epoch": 12.612682926829269,
      "grad_norm": 0.5555745959281921,
      "learning_rate": 0.002828,
      "loss": 1.4537,
      "step": 168064
    },
    {
      "epoch": 12.617485928705442,
      "grad_norm": 0.49841299653053284,
      "learning_rate": 0.002828,
      "loss": 1.4565,
      "step": 168128
    },
    {
      "epoch": 12.622288930581613,
      "grad_norm": 0.5797818899154663,
      "learning_rate": 0.002828,
      "loss": 1.4518,
      "step": 168192
    },
    {
      "epoch": 12.627091932457786,
      "grad_norm": 0.5726368427276611,
      "learning_rate": 0.002828,
      "loss": 1.4611,
      "step": 168256
    },
    {
      "epoch": 12.631894934333959,
      "grad_norm": 0.5542452335357666,
      "learning_rate": 0.002828,
      "loss": 1.4522,
      "step": 168320
    },
    {
      "epoch": 12.636697936210131,
      "grad_norm": 0.42682501673698425,
      "learning_rate": 0.002828,
      "loss": 1.456,
      "step": 168384
    },
    {
      "epoch": 12.641500938086304,
      "grad_norm": 0.4608093202114105,
      "learning_rate": 0.002828,
      "loss": 1.4614,
      "step": 168448
    },
    {
      "epoch": 12.646303939962477,
      "grad_norm": 0.6001752614974976,
      "learning_rate": 0.002828,
      "loss": 1.461,
      "step": 168512
    },
    {
      "epoch": 12.651106941838648,
      "grad_norm": 0.5241263508796692,
      "learning_rate": 0.002828,
      "loss": 1.4613,
      "step": 168576
    },
    {
      "epoch": 12.655909943714821,
      "grad_norm": 0.5729988813400269,
      "learning_rate": 0.002828,
      "loss": 1.4539,
      "step": 168640
    },
    {
      "epoch": 12.660712945590994,
      "grad_norm": 0.5749446153640747,
      "learning_rate": 0.002828,
      "loss": 1.4538,
      "step": 168704
    },
    {
      "epoch": 12.665515947467167,
      "grad_norm": 0.45133233070373535,
      "learning_rate": 0.002828,
      "loss": 1.4627,
      "step": 168768
    },
    {
      "epoch": 12.67031894934334,
      "grad_norm": 0.583003044128418,
      "learning_rate": 0.002828,
      "loss": 1.4638,
      "step": 168832
    },
    {
      "epoch": 12.675121951219513,
      "grad_norm": 0.5264136791229248,
      "learning_rate": 0.002828,
      "loss": 1.4579,
      "step": 168896
    },
    {
      "epoch": 12.679924953095686,
      "grad_norm": 0.5577759146690369,
      "learning_rate": 0.002828,
      "loss": 1.4613,
      "step": 168960
    },
    {
      "epoch": 12.684727954971857,
      "grad_norm": 0.5690147876739502,
      "learning_rate": 0.002828,
      "loss": 1.4583,
      "step": 169024
    },
    {
      "epoch": 12.68953095684803,
      "grad_norm": 0.5202603340148926,
      "learning_rate": 0.002828,
      "loss": 1.4605,
      "step": 169088
    },
    {
      "epoch": 12.694333958724203,
      "grad_norm": 0.6959776282310486,
      "learning_rate": 0.002828,
      "loss": 1.4665,
      "step": 169152
    },
    {
      "epoch": 12.699136960600375,
      "grad_norm": 0.5649484395980835,
      "learning_rate": 0.002828,
      "loss": 1.4547,
      "step": 169216
    },
    {
      "epoch": 12.703939962476548,
      "grad_norm": 0.5574206709861755,
      "learning_rate": 0.002828,
      "loss": 1.4602,
      "step": 169280
    },
    {
      "epoch": 12.708742964352721,
      "grad_norm": 0.4697055220603943,
      "learning_rate": 0.002828,
      "loss": 1.4622,
      "step": 169344
    },
    {
      "epoch": 12.713545966228892,
      "grad_norm": 0.5403327345848083,
      "learning_rate": 0.002828,
      "loss": 1.4602,
      "step": 169408
    },
    {
      "epoch": 12.718348968105065,
      "grad_norm": 0.5151571035385132,
      "learning_rate": 0.002828,
      "loss": 1.4611,
      "step": 169472
    },
    {
      "epoch": 12.723151969981238,
      "grad_norm": 0.48533812165260315,
      "learning_rate": 0.002828,
      "loss": 1.4651,
      "step": 169536
    },
    {
      "epoch": 12.727954971857411,
      "grad_norm": 0.5039843916893005,
      "learning_rate": 0.002828,
      "loss": 1.4582,
      "step": 169600
    },
    {
      "epoch": 12.732757973733584,
      "grad_norm": 0.5662795901298523,
      "learning_rate": 0.002828,
      "loss": 1.4576,
      "step": 169664
    },
    {
      "epoch": 12.737560975609757,
      "grad_norm": 0.568792998790741,
      "learning_rate": 0.002828,
      "loss": 1.4585,
      "step": 169728
    },
    {
      "epoch": 12.74236397748593,
      "grad_norm": 0.5497303009033203,
      "learning_rate": 0.002828,
      "loss": 1.4583,
      "step": 169792
    },
    {
      "epoch": 12.7471669793621,
      "grad_norm": 0.49786439538002014,
      "learning_rate": 0.002828,
      "loss": 1.4639,
      "step": 169856
    },
    {
      "epoch": 12.751969981238274,
      "grad_norm": 0.6249164342880249,
      "learning_rate": 0.002828,
      "loss": 1.4561,
      "step": 169920
    },
    {
      "epoch": 12.756772983114447,
      "grad_norm": 0.49063023924827576,
      "learning_rate": 0.002828,
      "loss": 1.4588,
      "step": 169984
    },
    {
      "epoch": 12.76157598499062,
      "grad_norm": 0.4647848904132843,
      "learning_rate": 0.002828,
      "loss": 1.4608,
      "step": 170048
    },
    {
      "epoch": 12.766378986866792,
      "grad_norm": 0.5198040008544922,
      "learning_rate": 0.002828,
      "loss": 1.4576,
      "step": 170112
    },
    {
      "epoch": 12.771181988742965,
      "grad_norm": 0.5746068358421326,
      "learning_rate": 0.002828,
      "loss": 1.4588,
      "step": 170176
    },
    {
      "epoch": 12.775984990619136,
      "grad_norm": 0.48485812544822693,
      "learning_rate": 0.002828,
      "loss": 1.4625,
      "step": 170240
    },
    {
      "epoch": 12.78078799249531,
      "grad_norm": 0.43870824575424194,
      "learning_rate": 0.002828,
      "loss": 1.4587,
      "step": 170304
    },
    {
      "epoch": 12.785590994371482,
      "grad_norm": 0.5836806893348694,
      "learning_rate": 0.002828,
      "loss": 1.4591,
      "step": 170368
    },
    {
      "epoch": 12.790393996247655,
      "grad_norm": 0.4936177432537079,
      "learning_rate": 0.002828,
      "loss": 1.4596,
      "step": 170432
    },
    {
      "epoch": 12.795196998123828,
      "grad_norm": 0.5804800987243652,
      "learning_rate": 0.002828,
      "loss": 1.4684,
      "step": 170496
    },
    {
      "epoch": 12.8,
      "grad_norm": 0.5017887353897095,
      "learning_rate": 0.002828,
      "loss": 1.459,
      "step": 170560
    },
    {
      "epoch": 12.804803001876172,
      "grad_norm": 0.5359672904014587,
      "learning_rate": 0.002828,
      "loss": 1.4555,
      "step": 170624
    },
    {
      "epoch": 12.809606003752345,
      "grad_norm": 0.592132031917572,
      "learning_rate": 0.002828,
      "loss": 1.4495,
      "step": 170688
    },
    {
      "epoch": 12.814409005628518,
      "grad_norm": 0.5940999388694763,
      "learning_rate": 0.002828,
      "loss": 1.451,
      "step": 170752
    },
    {
      "epoch": 12.81921200750469,
      "grad_norm": 0.4877603352069855,
      "learning_rate": 0.002828,
      "loss": 1.4562,
      "step": 170816
    },
    {
      "epoch": 12.824015009380863,
      "grad_norm": 0.5055379271507263,
      "learning_rate": 0.002828,
      "loss": 1.4593,
      "step": 170880
    },
    {
      "epoch": 12.828818011257036,
      "grad_norm": 0.5798386335372925,
      "learning_rate": 0.002828,
      "loss": 1.4549,
      "step": 170944
    },
    {
      "epoch": 12.833621013133207,
      "grad_norm": 0.5174988508224487,
      "learning_rate": 0.002828,
      "loss": 1.4614,
      "step": 171008
    },
    {
      "epoch": 12.83842401500938,
      "grad_norm": 0.5694937109947205,
      "learning_rate": 0.002828,
      "loss": 1.4595,
      "step": 171072
    },
    {
      "epoch": 12.843227016885553,
      "grad_norm": 0.5167633891105652,
      "learning_rate": 0.002828,
      "loss": 1.4569,
      "step": 171136
    },
    {
      "epoch": 12.848030018761726,
      "grad_norm": 0.5683608055114746,
      "learning_rate": 0.002828,
      "loss": 1.4556,
      "step": 171200
    },
    {
      "epoch": 12.852833020637899,
      "grad_norm": 0.5424241423606873,
      "learning_rate": 0.002828,
      "loss": 1.456,
      "step": 171264
    },
    {
      "epoch": 12.857636022514072,
      "grad_norm": 0.49074310064315796,
      "learning_rate": 0.002828,
      "loss": 1.4585,
      "step": 171328
    },
    {
      "epoch": 12.862439024390245,
      "grad_norm": 0.5989189743995667,
      "learning_rate": 0.002828,
      "loss": 1.4654,
      "step": 171392
    },
    {
      "epoch": 12.867242026266416,
      "grad_norm": 0.5615851879119873,
      "learning_rate": 0.002828,
      "loss": 1.4615,
      "step": 171456
    },
    {
      "epoch": 12.872045028142589,
      "grad_norm": 0.5346879959106445,
      "learning_rate": 0.002828,
      "loss": 1.4544,
      "step": 171520
    },
    {
      "epoch": 12.876848030018762,
      "grad_norm": 0.5329033136367798,
      "learning_rate": 0.002828,
      "loss": 1.4625,
      "step": 171584
    },
    {
      "epoch": 12.881651031894934,
      "grad_norm": 0.4970327317714691,
      "learning_rate": 0.002828,
      "loss": 1.459,
      "step": 171648
    },
    {
      "epoch": 12.886454033771107,
      "grad_norm": 0.5486003160476685,
      "learning_rate": 0.002828,
      "loss": 1.4543,
      "step": 171712
    },
    {
      "epoch": 12.89125703564728,
      "grad_norm": 0.4972151815891266,
      "learning_rate": 0.002828,
      "loss": 1.4608,
      "step": 171776
    },
    {
      "epoch": 12.896060037523451,
      "grad_norm": 0.4709850549697876,
      "learning_rate": 0.002828,
      "loss": 1.462,
      "step": 171840
    },
    {
      "epoch": 12.900863039399624,
      "grad_norm": 0.548951268196106,
      "learning_rate": 0.002828,
      "loss": 1.4526,
      "step": 171904
    },
    {
      "epoch": 12.905666041275797,
      "grad_norm": 0.6316720843315125,
      "learning_rate": 0.002828,
      "loss": 1.4548,
      "step": 171968
    },
    {
      "epoch": 12.91046904315197,
      "grad_norm": 0.711441159248352,
      "learning_rate": 0.002828,
      "loss": 1.4544,
      "step": 172032
    },
    {
      "epoch": 12.915272045028143,
      "grad_norm": 0.5292863845825195,
      "learning_rate": 0.002828,
      "loss": 1.4618,
      "step": 172096
    },
    {
      "epoch": 12.920075046904316,
      "grad_norm": 0.5618671178817749,
      "learning_rate": 0.002828,
      "loss": 1.4585,
      "step": 172160
    },
    {
      "epoch": 12.924878048780489,
      "grad_norm": 0.5491626262664795,
      "learning_rate": 0.002828,
      "loss": 1.4605,
      "step": 172224
    },
    {
      "epoch": 12.92968105065666,
      "grad_norm": 0.5500079393386841,
      "learning_rate": 0.002828,
      "loss": 1.4642,
      "step": 172288
    },
    {
      "epoch": 12.934484052532833,
      "grad_norm": 0.761467456817627,
      "learning_rate": 0.002828,
      "loss": 1.4611,
      "step": 172352
    },
    {
      "epoch": 12.939287054409006,
      "grad_norm": 0.4791053235530853,
      "learning_rate": 0.002828,
      "loss": 1.4521,
      "step": 172416
    },
    {
      "epoch": 12.944090056285178,
      "grad_norm": 0.5197027921676636,
      "learning_rate": 0.002828,
      "loss": 1.4569,
      "step": 172480
    },
    {
      "epoch": 12.948893058161351,
      "grad_norm": 0.4713915288448334,
      "learning_rate": 0.002828,
      "loss": 1.4586,
      "step": 172544
    },
    {
      "epoch": 12.953696060037524,
      "grad_norm": 0.5665622353553772,
      "learning_rate": 0.002828,
      "loss": 1.448,
      "step": 172608
    },
    {
      "epoch": 12.958499061913695,
      "grad_norm": 0.4697657823562622,
      "learning_rate": 0.002828,
      "loss": 1.4615,
      "step": 172672
    },
    {
      "epoch": 12.963302063789868,
      "grad_norm": 0.5036899447441101,
      "learning_rate": 0.002828,
      "loss": 1.4704,
      "step": 172736
    },
    {
      "epoch": 12.968105065666041,
      "grad_norm": 0.5290060639381409,
      "learning_rate": 0.002828,
      "loss": 1.4646,
      "step": 172800
    },
    {
      "epoch": 12.972908067542214,
      "grad_norm": 0.4270978271961212,
      "learning_rate": 0.002828,
      "loss": 1.4542,
      "step": 172864
    },
    {
      "epoch": 12.977711069418387,
      "grad_norm": 0.5034455060958862,
      "learning_rate": 0.002828,
      "loss": 1.4541,
      "step": 172928
    },
    {
      "epoch": 12.98251407129456,
      "grad_norm": 0.5570845603942871,
      "learning_rate": 0.002828,
      "loss": 1.4559,
      "step": 172992
    },
    {
      "epoch": 12.987317073170733,
      "grad_norm": 0.5905221700668335,
      "learning_rate": 0.002828,
      "loss": 1.4593,
      "step": 173056
    },
    {
      "epoch": 12.992120075046904,
      "grad_norm": 0.48616600036621094,
      "learning_rate": 0.002828,
      "loss": 1.462,
      "step": 173120
    },
    {
      "epoch": 12.996923076923077,
      "grad_norm": 0.5468747615814209,
      "learning_rate": 0.002828,
      "loss": 1.4559,
      "step": 173184
    },
    {
      "epoch": 13.00172607879925,
      "grad_norm": 0.5406197905540466,
      "learning_rate": 0.002828,
      "loss": 1.4456,
      "step": 173248
    },
    {
      "epoch": 13.006529080675422,
      "grad_norm": 0.6132946014404297,
      "learning_rate": 0.002828,
      "loss": 1.4222,
      "step": 173312
    },
    {
      "epoch": 13.011332082551595,
      "grad_norm": 0.6026862859725952,
      "learning_rate": 0.002828,
      "loss": 1.4131,
      "step": 173376
    },
    {
      "epoch": 13.016135084427768,
      "grad_norm": 0.49466168880462646,
      "learning_rate": 0.002828,
      "loss": 1.4189,
      "step": 173440
    },
    {
      "epoch": 13.02093808630394,
      "grad_norm": 0.4750056862831116,
      "learning_rate": 0.002828,
      "loss": 1.4215,
      "step": 173504
    },
    {
      "epoch": 13.025741088180112,
      "grad_norm": 0.6869794726371765,
      "learning_rate": 0.002828,
      "loss": 1.4197,
      "step": 173568
    },
    {
      "epoch": 13.030544090056285,
      "grad_norm": 0.5326526761054993,
      "learning_rate": 0.002828,
      "loss": 1.4168,
      "step": 173632
    },
    {
      "epoch": 13.035347091932458,
      "grad_norm": 0.5036327242851257,
      "learning_rate": 0.002828,
      "loss": 1.4156,
      "step": 173696
    },
    {
      "epoch": 13.04015009380863,
      "grad_norm": 0.5569616556167603,
      "learning_rate": 0.002828,
      "loss": 1.42,
      "step": 173760
    },
    {
      "epoch": 13.044953095684804,
      "grad_norm": 0.5766422152519226,
      "learning_rate": 0.002828,
      "loss": 1.4208,
      "step": 173824
    },
    {
      "epoch": 13.049756097560975,
      "grad_norm": 0.570945143699646,
      "learning_rate": 0.002828,
      "loss": 1.4199,
      "step": 173888
    },
    {
      "epoch": 13.054559099437148,
      "grad_norm": 0.4492458403110504,
      "learning_rate": 0.002828,
      "loss": 1.4163,
      "step": 173952
    },
    {
      "epoch": 13.05936210131332,
      "grad_norm": 0.5762324929237366,
      "learning_rate": 0.002828,
      "loss": 1.4224,
      "step": 174016
    },
    {
      "epoch": 13.064165103189493,
      "grad_norm": 0.5214763283729553,
      "learning_rate": 0.002828,
      "loss": 1.4176,
      "step": 174080
    },
    {
      "epoch": 13.068968105065666,
      "grad_norm": 0.49345290660858154,
      "learning_rate": 0.002828,
      "loss": 1.4207,
      "step": 174144
    },
    {
      "epoch": 13.07377110694184,
      "grad_norm": 0.5769919753074646,
      "learning_rate": 0.002828,
      "loss": 1.4256,
      "step": 174208
    },
    {
      "epoch": 13.078574108818012,
      "grad_norm": 0.5672862529754639,
      "learning_rate": 0.002828,
      "loss": 1.4269,
      "step": 174272
    },
    {
      "epoch": 13.083377110694183,
      "grad_norm": 0.6378344893455505,
      "learning_rate": 0.002828,
      "loss": 1.4198,
      "step": 174336
    },
    {
      "epoch": 13.088180112570356,
      "grad_norm": 0.5776243805885315,
      "learning_rate": 0.002828,
      "loss": 1.4198,
      "step": 174400
    },
    {
      "epoch": 13.092983114446529,
      "grad_norm": 0.5405304431915283,
      "learning_rate": 0.002828,
      "loss": 1.4225,
      "step": 174464
    },
    {
      "epoch": 13.097786116322702,
      "grad_norm": 0.47602930665016174,
      "learning_rate": 0.002828,
      "loss": 1.4244,
      "step": 174528
    },
    {
      "epoch": 13.102589118198875,
      "grad_norm": 0.554095447063446,
      "learning_rate": 0.002828,
      "loss": 1.4115,
      "step": 174592
    },
    {
      "epoch": 13.107392120075048,
      "grad_norm": 0.49326562881469727,
      "learning_rate": 0.002828,
      "loss": 1.4206,
      "step": 174656
    },
    {
      "epoch": 13.112195121951219,
      "grad_norm": 0.5742772221565247,
      "learning_rate": 0.002828,
      "loss": 1.4267,
      "step": 174720
    },
    {
      "epoch": 13.116998123827392,
      "grad_norm": 0.5664711594581604,
      "learning_rate": 0.002828,
      "loss": 1.4201,
      "step": 174784
    },
    {
      "epoch": 13.121801125703564,
      "grad_norm": 0.5422520041465759,
      "learning_rate": 0.002828,
      "loss": 1.4205,
      "step": 174848
    },
    {
      "epoch": 13.126604127579737,
      "grad_norm": 0.5431873798370361,
      "learning_rate": 0.002828,
      "loss": 1.4203,
      "step": 174912
    },
    {
      "epoch": 13.13140712945591,
      "grad_norm": 0.5788468718528748,
      "learning_rate": 0.002828,
      "loss": 1.4241,
      "step": 174976
    },
    {
      "epoch": 13.136210131332083,
      "grad_norm": 0.6146829724311829,
      "learning_rate": 0.002828,
      "loss": 1.4238,
      "step": 175040
    },
    {
      "epoch": 13.141013133208254,
      "grad_norm": 0.44153454899787903,
      "learning_rate": 0.002828,
      "loss": 1.4233,
      "step": 175104
    },
    {
      "epoch": 13.145816135084427,
      "grad_norm": 0.4862412214279175,
      "learning_rate": 0.002828,
      "loss": 1.4227,
      "step": 175168
    },
    {
      "epoch": 13.1506191369606,
      "grad_norm": 0.5295515060424805,
      "learning_rate": 0.002828,
      "loss": 1.4242,
      "step": 175232
    },
    {
      "epoch": 13.155422138836773,
      "grad_norm": 0.4826089143753052,
      "learning_rate": 0.002828,
      "loss": 1.4219,
      "step": 175296
    },
    {
      "epoch": 13.160225140712946,
      "grad_norm": 0.5136275291442871,
      "learning_rate": 0.002828,
      "loss": 1.421,
      "step": 175360
    },
    {
      "epoch": 13.165028142589119,
      "grad_norm": 0.5050927996635437,
      "learning_rate": 0.002828,
      "loss": 1.4283,
      "step": 175424
    },
    {
      "epoch": 13.169831144465292,
      "grad_norm": 0.4709009528160095,
      "learning_rate": 0.002828,
      "loss": 1.4244,
      "step": 175488
    },
    {
      "epoch": 13.174634146341463,
      "grad_norm": 0.5698823928833008,
      "learning_rate": 0.002828,
      "loss": 1.4242,
      "step": 175552
    },
    {
      "epoch": 13.179437148217636,
      "grad_norm": 0.5443013310432434,
      "learning_rate": 0.002828,
      "loss": 1.4245,
      "step": 175616
    },
    {
      "epoch": 13.184240150093808,
      "grad_norm": 0.5437474846839905,
      "learning_rate": 0.002828,
      "loss": 1.4211,
      "step": 175680
    },
    {
      "epoch": 13.189043151969981,
      "grad_norm": 0.5265381336212158,
      "learning_rate": 0.002828,
      "loss": 1.4228,
      "step": 175744
    },
    {
      "epoch": 13.193846153846154,
      "grad_norm": 0.5583203434944153,
      "learning_rate": 0.002828,
      "loss": 1.4257,
      "step": 175808
    },
    {
      "epoch": 13.198649155722327,
      "grad_norm": 0.5392881035804749,
      "learning_rate": 0.002828,
      "loss": 1.4245,
      "step": 175872
    },
    {
      "epoch": 13.203452157598498,
      "grad_norm": 0.5564644932746887,
      "learning_rate": 0.002828,
      "loss": 1.4226,
      "step": 175936
    },
    {
      "epoch": 13.208255159474671,
      "grad_norm": 0.5897397994995117,
      "learning_rate": 0.002828,
      "loss": 1.4283,
      "step": 176000
    },
    {
      "epoch": 13.213058161350844,
      "grad_norm": 0.48577988147735596,
      "learning_rate": 0.002828,
      "loss": 1.4227,
      "step": 176064
    },
    {
      "epoch": 13.217861163227017,
      "grad_norm": 0.5362162590026855,
      "learning_rate": 0.002828,
      "loss": 1.4212,
      "step": 176128
    },
    {
      "epoch": 13.22266416510319,
      "grad_norm": 0.4959189295768738,
      "learning_rate": 0.002828,
      "loss": 1.4266,
      "step": 176192
    },
    {
      "epoch": 13.227467166979363,
      "grad_norm": 0.4970269203186035,
      "learning_rate": 0.002828,
      "loss": 1.4181,
      "step": 176256
    },
    {
      "epoch": 13.232270168855536,
      "grad_norm": 0.6063593029975891,
      "learning_rate": 0.002828,
      "loss": 1.4228,
      "step": 176320
    },
    {
      "epoch": 13.237073170731707,
      "grad_norm": 0.5324926972389221,
      "learning_rate": 0.002828,
      "loss": 1.4233,
      "step": 176384
    },
    {
      "epoch": 13.24187617260788,
      "grad_norm": 0.6734857559204102,
      "learning_rate": 0.002828,
      "loss": 1.4236,
      "step": 176448
    },
    {
      "epoch": 13.246679174484052,
      "grad_norm": 0.5381393432617188,
      "learning_rate": 0.002828,
      "loss": 1.4191,
      "step": 176512
    },
    {
      "epoch": 13.251482176360225,
      "grad_norm": 0.6253378391265869,
      "learning_rate": 0.002828,
      "loss": 1.424,
      "step": 176576
    },
    {
      "epoch": 13.256285178236398,
      "grad_norm": 0.4672098755836487,
      "learning_rate": 0.002828,
      "loss": 1.4212,
      "step": 176640
    },
    {
      "epoch": 13.261088180112571,
      "grad_norm": 0.6410490870475769,
      "learning_rate": 0.002828,
      "loss": 1.4266,
      "step": 176704
    },
    {
      "epoch": 13.265891181988742,
      "grad_norm": 0.6024436354637146,
      "learning_rate": 0.002828,
      "loss": 1.4216,
      "step": 176768
    },
    {
      "epoch": 13.270694183864915,
      "grad_norm": 0.6217461228370667,
      "learning_rate": 0.002828,
      "loss": 1.4252,
      "step": 176832
    },
    {
      "epoch": 13.275497185741088,
      "grad_norm": 0.7444032430648804,
      "learning_rate": 0.002828,
      "loss": 1.43,
      "step": 176896
    },
    {
      "epoch": 13.28030018761726,
      "grad_norm": 0.5802358984947205,
      "learning_rate": 0.002828,
      "loss": 1.4287,
      "step": 176960
    },
    {
      "epoch": 13.285103189493434,
      "grad_norm": 0.4666067361831665,
      "learning_rate": 0.002828,
      "loss": 1.4229,
      "step": 177024
    },
    {
      "epoch": 13.289906191369607,
      "grad_norm": 0.4871880114078522,
      "learning_rate": 0.002828,
      "loss": 1.4292,
      "step": 177088
    },
    {
      "epoch": 13.29470919324578,
      "grad_norm": 0.5733477473258972,
      "learning_rate": 0.002828,
      "loss": 1.4227,
      "step": 177152
    },
    {
      "epoch": 13.29951219512195,
      "grad_norm": 0.6557471752166748,
      "learning_rate": 0.002828,
      "loss": 1.4365,
      "step": 177216
    },
    {
      "epoch": 13.304315196998123,
      "grad_norm": 0.478577584028244,
      "learning_rate": 0.002828,
      "loss": 1.4361,
      "step": 177280
    },
    {
      "epoch": 13.309118198874296,
      "grad_norm": 0.5637518167495728,
      "learning_rate": 0.002828,
      "loss": 1.4278,
      "step": 177344
    },
    {
      "epoch": 13.31392120075047,
      "grad_norm": 0.561852216720581,
      "learning_rate": 0.002828,
      "loss": 1.4278,
      "step": 177408
    },
    {
      "epoch": 13.318724202626642,
      "grad_norm": 0.5598611235618591,
      "learning_rate": 0.002828,
      "loss": 1.4292,
      "step": 177472
    },
    {
      "epoch": 13.323527204502815,
      "grad_norm": 0.5581865310668945,
      "learning_rate": 0.002828,
      "loss": 1.4317,
      "step": 177536
    },
    {
      "epoch": 13.328330206378986,
      "grad_norm": 0.42338356375694275,
      "learning_rate": 0.002828,
      "loss": 1.4261,
      "step": 177600
    },
    {
      "epoch": 13.333133208255159,
      "grad_norm": 0.49160394072532654,
      "learning_rate": 0.002828,
      "loss": 1.4244,
      "step": 177664
    },
    {
      "epoch": 13.337936210131332,
      "grad_norm": 0.4856988787651062,
      "learning_rate": 0.002828,
      "loss": 1.4298,
      "step": 177728
    },
    {
      "epoch": 13.342739212007505,
      "grad_norm": 0.547420859336853,
      "learning_rate": 0.002828,
      "loss": 1.4238,
      "step": 177792
    },
    {
      "epoch": 13.347542213883678,
      "grad_norm": 0.5926133990287781,
      "learning_rate": 0.002828,
      "loss": 1.4258,
      "step": 177856
    },
    {
      "epoch": 13.35234521575985,
      "grad_norm": 0.5286599397659302,
      "learning_rate": 0.002828,
      "loss": 1.425,
      "step": 177920
    },
    {
      "epoch": 13.357148217636022,
      "grad_norm": 0.4704970121383667,
      "learning_rate": 0.002828,
      "loss": 1.4322,
      "step": 177984
    },
    {
      "epoch": 13.361951219512195,
      "grad_norm": 0.6747083067893982,
      "learning_rate": 0.002828,
      "loss": 1.4266,
      "step": 178048
    },
    {
      "epoch": 13.366754221388367,
      "grad_norm": 0.4974871873855591,
      "learning_rate": 0.002828,
      "loss": 1.4228,
      "step": 178112
    },
    {
      "epoch": 13.37155722326454,
      "grad_norm": 0.51888108253479,
      "learning_rate": 0.002828,
      "loss": 1.4265,
      "step": 178176
    },
    {
      "epoch": 13.376360225140713,
      "grad_norm": 0.6136510968208313,
      "learning_rate": 0.002828,
      "loss": 1.4304,
      "step": 178240
    },
    {
      "epoch": 13.381163227016886,
      "grad_norm": 0.5147790908813477,
      "learning_rate": 0.002828,
      "loss": 1.4339,
      "step": 178304
    },
    {
      "epoch": 13.385966228893059,
      "grad_norm": 0.572532594203949,
      "learning_rate": 0.002828,
      "loss": 1.4254,
      "step": 178368
    },
    {
      "epoch": 13.39076923076923,
      "grad_norm": 0.4552408754825592,
      "learning_rate": 0.002828,
      "loss": 1.4265,
      "step": 178432
    },
    {
      "epoch": 13.395572232645403,
      "grad_norm": 0.5829170346260071,
      "learning_rate": 0.002828,
      "loss": 1.4266,
      "step": 178496
    },
    {
      "epoch": 13.400375234521576,
      "grad_norm": 0.43553870916366577,
      "learning_rate": 0.002828,
      "loss": 1.4339,
      "step": 178560
    },
    {
      "epoch": 13.405178236397749,
      "grad_norm": 0.5179685354232788,
      "learning_rate": 0.002828,
      "loss": 1.4263,
      "step": 178624
    },
    {
      "epoch": 13.409981238273922,
      "grad_norm": 0.5742409825325012,
      "learning_rate": 0.002828,
      "loss": 1.4303,
      "step": 178688
    },
    {
      "epoch": 13.414784240150095,
      "grad_norm": 0.6354867815971375,
      "learning_rate": 0.002828,
      "loss": 1.4311,
      "step": 178752
    },
    {
      "epoch": 13.419587242026266,
      "grad_norm": 0.5913352370262146,
      "learning_rate": 0.002828,
      "loss": 1.4259,
      "step": 178816
    },
    {
      "epoch": 13.424390243902439,
      "grad_norm": 0.47367367148399353,
      "learning_rate": 0.002828,
      "loss": 1.4281,
      "step": 178880
    },
    {
      "epoch": 13.429193245778611,
      "grad_norm": 0.5402623414993286,
      "learning_rate": 0.002828,
      "loss": 1.4281,
      "step": 178944
    },
    {
      "epoch": 13.433996247654784,
      "grad_norm": 0.5336705446243286,
      "learning_rate": 0.002828,
      "loss": 1.4314,
      "step": 179008
    },
    {
      "epoch": 13.438799249530957,
      "grad_norm": 0.5876074433326721,
      "learning_rate": 0.002828,
      "loss": 1.4333,
      "step": 179072
    },
    {
      "epoch": 13.44360225140713,
      "grad_norm": 0.5205850005149841,
      "learning_rate": 0.002828,
      "loss": 1.4237,
      "step": 179136
    },
    {
      "epoch": 13.448405253283301,
      "grad_norm": 0.5328284502029419,
      "learning_rate": 0.002828,
      "loss": 1.4304,
      "step": 179200
    },
    {
      "epoch": 13.453208255159474,
      "grad_norm": 0.5152733325958252,
      "learning_rate": 0.002828,
      "loss": 1.429,
      "step": 179264
    },
    {
      "epoch": 13.458011257035647,
      "grad_norm": 0.45550423860549927,
      "learning_rate": 0.002828,
      "loss": 1.4347,
      "step": 179328
    },
    {
      "epoch": 13.46281425891182,
      "grad_norm": 0.485805869102478,
      "learning_rate": 0.002828,
      "loss": 1.4312,
      "step": 179392
    },
    {
      "epoch": 13.467617260787993,
      "grad_norm": 0.5946416258811951,
      "learning_rate": 0.002828,
      "loss": 1.422,
      "step": 179456
    },
    {
      "epoch": 13.472420262664166,
      "grad_norm": 0.5384697318077087,
      "learning_rate": 0.002828,
      "loss": 1.4328,
      "step": 179520
    },
    {
      "epoch": 13.477223264540338,
      "grad_norm": 0.48482996225357056,
      "learning_rate": 0.002828,
      "loss": 1.4267,
      "step": 179584
    },
    {
      "epoch": 13.48202626641651,
      "grad_norm": 0.4668543040752411,
      "learning_rate": 0.002828,
      "loss": 1.43,
      "step": 179648
    },
    {
      "epoch": 13.486829268292682,
      "grad_norm": 0.508421003818512,
      "learning_rate": 0.002828,
      "loss": 1.4295,
      "step": 179712
    },
    {
      "epoch": 13.491632270168855,
      "grad_norm": 0.5609977841377258,
      "learning_rate": 0.002828,
      "loss": 1.4242,
      "step": 179776
    },
    {
      "epoch": 13.496435272045028,
      "grad_norm": 0.43659254908561707,
      "learning_rate": 0.002828,
      "loss": 1.4306,
      "step": 179840
    },
    {
      "epoch": 13.501238273921201,
      "grad_norm": 0.5313818454742432,
      "learning_rate": 0.002828,
      "loss": 1.4317,
      "step": 179904
    },
    {
      "epoch": 13.506041275797374,
      "grad_norm": 0.5664373636245728,
      "learning_rate": 0.002828,
      "loss": 1.4326,
      "step": 179968
    },
    {
      "epoch": 13.510844277673545,
      "grad_norm": 0.5259975790977478,
      "learning_rate": 0.002828,
      "loss": 1.4283,
      "step": 180032
    },
    {
      "epoch": 13.515647279549718,
      "grad_norm": 0.5671187043190002,
      "learning_rate": 0.002828,
      "loss": 1.4305,
      "step": 180096
    },
    {
      "epoch": 13.520450281425891,
      "grad_norm": 0.593250036239624,
      "learning_rate": 0.002828,
      "loss": 1.433,
      "step": 180160
    },
    {
      "epoch": 13.525253283302064,
      "grad_norm": 0.4607378840446472,
      "learning_rate": 0.002828,
      "loss": 1.4254,
      "step": 180224
    },
    {
      "epoch": 13.530056285178237,
      "grad_norm": 0.4907410144805908,
      "learning_rate": 0.002828,
      "loss": 1.4331,
      "step": 180288
    },
    {
      "epoch": 13.53485928705441,
      "grad_norm": 0.6493052840232849,
      "learning_rate": 0.002828,
      "loss": 1.4391,
      "step": 180352
    },
    {
      "epoch": 13.539662288930582,
      "grad_norm": 0.5539810657501221,
      "learning_rate": 0.002828,
      "loss": 1.4323,
      "step": 180416
    },
    {
      "epoch": 13.544465290806754,
      "grad_norm": 0.5795511603355408,
      "learning_rate": 0.002828,
      "loss": 1.4378,
      "step": 180480
    },
    {
      "epoch": 13.549268292682926,
      "grad_norm": 0.5561606287956238,
      "learning_rate": 0.002828,
      "loss": 1.4298,
      "step": 180544
    },
    {
      "epoch": 13.5540712945591,
      "grad_norm": 0.49703845381736755,
      "learning_rate": 0.002828,
      "loss": 1.433,
      "step": 180608
    },
    {
      "epoch": 13.558874296435272,
      "grad_norm": 0.5289865136146545,
      "learning_rate": 0.002828,
      "loss": 1.4323,
      "step": 180672
    },
    {
      "epoch": 13.563677298311445,
      "grad_norm": 0.588064432144165,
      "learning_rate": 0.002828,
      "loss": 1.4278,
      "step": 180736
    },
    {
      "epoch": 13.568480300187618,
      "grad_norm": 0.5216802954673767,
      "learning_rate": 0.002828,
      "loss": 1.4262,
      "step": 180800
    },
    {
      "epoch": 13.573283302063789,
      "grad_norm": 0.6461803317070007,
      "learning_rate": 0.002828,
      "loss": 1.4399,
      "step": 180864
    },
    {
      "epoch": 13.578086303939962,
      "grad_norm": 0.4628434479236603,
      "learning_rate": 0.002828,
      "loss": 1.4294,
      "step": 180928
    },
    {
      "epoch": 13.582889305816135,
      "grad_norm": 0.630901038646698,
      "learning_rate": 0.002828,
      "loss": 1.4251,
      "step": 180992
    },
    {
      "epoch": 13.587692307692308,
      "grad_norm": 0.6294755339622498,
      "learning_rate": 0.002828,
      "loss": 1.4268,
      "step": 181056
    },
    {
      "epoch": 13.59249530956848,
      "grad_norm": 0.5399813055992126,
      "learning_rate": 0.002828,
      "loss": 1.4293,
      "step": 181120
    },
    {
      "epoch": 13.597298311444654,
      "grad_norm": 0.589293360710144,
      "learning_rate": 0.002828,
      "loss": 1.4352,
      "step": 181184
    },
    {
      "epoch": 13.602101313320826,
      "grad_norm": 0.5084601640701294,
      "learning_rate": 0.002828,
      "loss": 1.4342,
      "step": 181248
    },
    {
      "epoch": 13.606904315196998,
      "grad_norm": 0.5964998006820679,
      "learning_rate": 0.002828,
      "loss": 1.4355,
      "step": 181312
    },
    {
      "epoch": 13.61170731707317,
      "grad_norm": 0.48000743985176086,
      "learning_rate": 0.002828,
      "loss": 1.4298,
      "step": 181376
    },
    {
      "epoch": 13.616510318949343,
      "grad_norm": 0.5122075080871582,
      "learning_rate": 0.002828,
      "loss": 1.435,
      "step": 181440
    },
    {
      "epoch": 13.621313320825516,
      "grad_norm": 0.511771559715271,
      "learning_rate": 0.002828,
      "loss": 1.4284,
      "step": 181504
    },
    {
      "epoch": 13.626116322701689,
      "grad_norm": 0.5523917078971863,
      "learning_rate": 0.002828,
      "loss": 1.4267,
      "step": 181568
    },
    {
      "epoch": 13.630919324577862,
      "grad_norm": 0.48201820254325867,
      "learning_rate": 0.002828,
      "loss": 1.4286,
      "step": 181632
    },
    {
      "epoch": 13.635722326454033,
      "grad_norm": 0.5382052659988403,
      "learning_rate": 0.002828,
      "loss": 1.4371,
      "step": 181696
    },
    {
      "epoch": 13.640525328330206,
      "grad_norm": 0.5262125134468079,
      "learning_rate": 0.002828,
      "loss": 1.4329,
      "step": 181760
    },
    {
      "epoch": 13.645328330206379,
      "grad_norm": 0.5679051280021667,
      "learning_rate": 0.002828,
      "loss": 1.4376,
      "step": 181824
    },
    {
      "epoch": 13.650131332082552,
      "grad_norm": 0.5577001571655273,
      "learning_rate": 0.002828,
      "loss": 1.4318,
      "step": 181888
    },
    {
      "epoch": 13.654934333958725,
      "grad_norm": 0.6490796804428101,
      "learning_rate": 0.002828,
      "loss": 1.4339,
      "step": 181952
    },
    {
      "epoch": 13.659737335834897,
      "grad_norm": 0.5554207563400269,
      "learning_rate": 0.002828,
      "loss": 1.4373,
      "step": 182016
    },
    {
      "epoch": 13.66454033771107,
      "grad_norm": 0.4835376739501953,
      "learning_rate": 0.002828,
      "loss": 1.4353,
      "step": 182080
    },
    {
      "epoch": 13.669343339587241,
      "grad_norm": 0.4857877194881439,
      "learning_rate": 0.002828,
      "loss": 1.4319,
      "step": 182144
    },
    {
      "epoch": 13.674146341463414,
      "grad_norm": 0.5001787543296814,
      "learning_rate": 0.002828,
      "loss": 1.4292,
      "step": 182208
    },
    {
      "epoch": 13.678949343339587,
      "grad_norm": 0.6918062567710876,
      "learning_rate": 0.002828,
      "loss": 1.4337,
      "step": 182272
    },
    {
      "epoch": 13.68375234521576,
      "grad_norm": 0.6048857569694519,
      "learning_rate": 0.002828,
      "loss": 1.4298,
      "step": 182336
    },
    {
      "epoch": 13.688555347091933,
      "grad_norm": 0.5380560755729675,
      "learning_rate": 0.002828,
      "loss": 1.4282,
      "step": 182400
    },
    {
      "epoch": 13.693358348968106,
      "grad_norm": 0.534903883934021,
      "learning_rate": 0.002828,
      "loss": 1.4303,
      "step": 182464
    },
    {
      "epoch": 13.698161350844277,
      "grad_norm": 0.5099212527275085,
      "learning_rate": 0.002828,
      "loss": 1.4273,
      "step": 182528
    },
    {
      "epoch": 13.70296435272045,
      "grad_norm": 0.5878075957298279,
      "learning_rate": 0.002828,
      "loss": 1.4342,
      "step": 182592
    },
    {
      "epoch": 13.707767354596623,
      "grad_norm": 0.5337736010551453,
      "learning_rate": 0.002828,
      "loss": 1.4289,
      "step": 182656
    },
    {
      "epoch": 13.712570356472796,
      "grad_norm": 0.5323655605316162,
      "learning_rate": 0.002828,
      "loss": 1.4341,
      "step": 182720
    },
    {
      "epoch": 13.717373358348969,
      "grad_norm": 0.4101346433162689,
      "learning_rate": 0.002828,
      "loss": 1.4336,
      "step": 182784
    },
    {
      "epoch": 13.722176360225141,
      "grad_norm": 0.6183751225471497,
      "learning_rate": 0.002828,
      "loss": 1.4309,
      "step": 182848
    },
    {
      "epoch": 13.726979362101313,
      "grad_norm": 0.6076233983039856,
      "learning_rate": 0.002828,
      "loss": 1.4309,
      "step": 182912
    },
    {
      "epoch": 13.731782363977485,
      "grad_norm": 0.5721030831336975,
      "learning_rate": 0.002828,
      "loss": 1.4325,
      "step": 182976
    },
    {
      "epoch": 13.736585365853658,
      "grad_norm": 0.55230712890625,
      "learning_rate": 0.002828,
      "loss": 1.4294,
      "step": 183040
    },
    {
      "epoch": 13.741388367729831,
      "grad_norm": 0.5440775752067566,
      "learning_rate": 0.002828,
      "loss": 1.4313,
      "step": 183104
    },
    {
      "epoch": 13.746191369606004,
      "grad_norm": 0.5208132266998291,
      "learning_rate": 0.002828,
      "loss": 1.431,
      "step": 183168
    },
    {
      "epoch": 13.750994371482177,
      "grad_norm": 0.5345131158828735,
      "learning_rate": 0.002828,
      "loss": 1.4301,
      "step": 183232
    },
    {
      "epoch": 13.755797373358348,
      "grad_norm": 0.5960522890090942,
      "learning_rate": 0.002828,
      "loss": 1.4357,
      "step": 183296
    },
    {
      "epoch": 13.760600375234521,
      "grad_norm": 0.5363312363624573,
      "learning_rate": 0.002828,
      "loss": 1.4419,
      "step": 183360
    },
    {
      "epoch": 13.765403377110694,
      "grad_norm": 0.5879179835319519,
      "learning_rate": 0.002828,
      "loss": 1.4368,
      "step": 183424
    },
    {
      "epoch": 13.770206378986867,
      "grad_norm": 0.45576685667037964,
      "learning_rate": 0.002828,
      "loss": 1.4352,
      "step": 183488
    },
    {
      "epoch": 13.77500938086304,
      "grad_norm": 0.5763787031173706,
      "learning_rate": 0.002828,
      "loss": 1.4365,
      "step": 183552
    },
    {
      "epoch": 13.779812382739212,
      "grad_norm": 0.5253689885139465,
      "learning_rate": 0.002828,
      "loss": 1.4321,
      "step": 183616
    },
    {
      "epoch": 13.784615384615385,
      "grad_norm": 0.5026938915252686,
      "learning_rate": 0.002828,
      "loss": 1.4333,
      "step": 183680
    },
    {
      "epoch": 13.789418386491556,
      "grad_norm": 0.505780816078186,
      "learning_rate": 0.002828,
      "loss": 1.4363,
      "step": 183744
    },
    {
      "epoch": 13.79422138836773,
      "grad_norm": 0.530052900314331,
      "learning_rate": 0.002828,
      "loss": 1.4337,
      "step": 183808
    },
    {
      "epoch": 13.799024390243902,
      "grad_norm": 0.526617169380188,
      "learning_rate": 0.002828,
      "loss": 1.428,
      "step": 183872
    },
    {
      "epoch": 13.803827392120075,
      "grad_norm": 0.4842779040336609,
      "learning_rate": 0.002828,
      "loss": 1.432,
      "step": 183936
    },
    {
      "epoch": 13.808630393996248,
      "grad_norm": 0.4398646056652069,
      "learning_rate": 0.002828,
      "loss": 1.4313,
      "step": 184000
    },
    {
      "epoch": 13.813433395872421,
      "grad_norm": 0.6402506828308105,
      "learning_rate": 0.002828,
      "loss": 1.4344,
      "step": 184064
    },
    {
      "epoch": 13.818236397748592,
      "grad_norm": 0.4699048101902008,
      "learning_rate": 0.002828,
      "loss": 1.4377,
      "step": 184128
    },
    {
      "epoch": 13.823039399624765,
      "grad_norm": 0.5081595182418823,
      "learning_rate": 0.002828,
      "loss": 1.4373,
      "step": 184192
    },
    {
      "epoch": 13.827842401500938,
      "grad_norm": 0.46454253792762756,
      "learning_rate": 0.002828,
      "loss": 1.4318,
      "step": 184256
    },
    {
      "epoch": 13.83264540337711,
      "grad_norm": 0.6264491677284241,
      "learning_rate": 0.002828,
      "loss": 1.426,
      "step": 184320
    },
    {
      "epoch": 13.837448405253284,
      "grad_norm": 0.6079069972038269,
      "learning_rate": 0.002828,
      "loss": 1.4284,
      "step": 184384
    },
    {
      "epoch": 13.842251407129456,
      "grad_norm": 0.516166627407074,
      "learning_rate": 0.002828,
      "loss": 1.4334,
      "step": 184448
    },
    {
      "epoch": 13.84705440900563,
      "grad_norm": 0.4872256815433502,
      "learning_rate": 0.002828,
      "loss": 1.431,
      "step": 184512
    },
    {
      "epoch": 13.8518574108818,
      "grad_norm": 0.5415956974029541,
      "learning_rate": 0.002828,
      "loss": 1.4312,
      "step": 184576
    },
    {
      "epoch": 13.856660412757973,
      "grad_norm": 0.5147574543952942,
      "learning_rate": 0.002828,
      "loss": 1.43,
      "step": 184640
    },
    {
      "epoch": 13.861463414634146,
      "grad_norm": 0.5008430480957031,
      "learning_rate": 0.002828,
      "loss": 1.4332,
      "step": 184704
    },
    {
      "epoch": 13.866266416510319,
      "grad_norm": 0.5969945192337036,
      "learning_rate": 0.002828,
      "loss": 1.434,
      "step": 184768
    },
    {
      "epoch": 13.871069418386492,
      "grad_norm": 0.44767871499061584,
      "learning_rate": 0.002828,
      "loss": 1.4314,
      "step": 184832
    },
    {
      "epoch": 13.875872420262665,
      "grad_norm": 0.6381478905677795,
      "learning_rate": 0.002828,
      "loss": 1.4285,
      "step": 184896
    },
    {
      "epoch": 13.880675422138836,
      "grad_norm": 0.47069770097732544,
      "learning_rate": 0.002828,
      "loss": 1.4372,
      "step": 184960
    },
    {
      "epoch": 13.885478424015009,
      "grad_norm": 0.5484511256217957,
      "learning_rate": 0.002828,
      "loss": 1.4317,
      "step": 185024
    },
    {
      "epoch": 13.890281425891182,
      "grad_norm": 0.576411783695221,
      "learning_rate": 0.002828,
      "loss": 1.4271,
      "step": 185088
    },
    {
      "epoch": 13.895084427767355,
      "grad_norm": 0.5928360223770142,
      "learning_rate": 0.002828,
      "loss": 1.4376,
      "step": 185152
    },
    {
      "epoch": 13.899887429643528,
      "grad_norm": 0.7210304141044617,
      "learning_rate": 0.002828,
      "loss": 1.4365,
      "step": 185216
    },
    {
      "epoch": 13.9046904315197,
      "grad_norm": 0.5716149806976318,
      "learning_rate": 0.002828,
      "loss": 1.4312,
      "step": 185280
    },
    {
      "epoch": 13.909493433395873,
      "grad_norm": 0.5631930232048035,
      "learning_rate": 0.002828,
      "loss": 1.4337,
      "step": 185344
    },
    {
      "epoch": 13.914296435272044,
      "grad_norm": 0.4883274435997009,
      "learning_rate": 0.002828,
      "loss": 1.4279,
      "step": 185408
    },
    {
      "epoch": 13.919099437148217,
      "grad_norm": 0.5907841920852661,
      "learning_rate": 0.002828,
      "loss": 1.4298,
      "step": 185472
    },
    {
      "epoch": 13.92390243902439,
      "grad_norm": 0.7262232899665833,
      "learning_rate": 0.002828,
      "loss": 1.4352,
      "step": 185536
    },
    {
      "epoch": 13.928705440900563,
      "grad_norm": 0.5713350176811218,
      "learning_rate": 0.002828,
      "loss": 1.4302,
      "step": 185600
    },
    {
      "epoch": 13.933508442776736,
      "grad_norm": 0.48538756370544434,
      "learning_rate": 0.002828,
      "loss": 1.434,
      "step": 185664
    },
    {
      "epoch": 13.938311444652909,
      "grad_norm": 0.610945463180542,
      "learning_rate": 0.002828,
      "loss": 1.4289,
      "step": 185728
    },
    {
      "epoch": 13.94311444652908,
      "grad_norm": 0.5916759967803955,
      "learning_rate": 0.002828,
      "loss": 1.4358,
      "step": 185792
    },
    {
      "epoch": 13.947917448405253,
      "grad_norm": 0.49742409586906433,
      "learning_rate": 0.002828,
      "loss": 1.4264,
      "step": 185856
    },
    {
      "epoch": 13.952720450281426,
      "grad_norm": 0.5827100872993469,
      "learning_rate": 0.002828,
      "loss": 1.4301,
      "step": 185920
    },
    {
      "epoch": 13.957523452157599,
      "grad_norm": 0.517087459564209,
      "learning_rate": 0.002828,
      "loss": 1.4312,
      "step": 185984
    },
    {
      "epoch": 13.962326454033771,
      "grad_norm": 0.5403559803962708,
      "learning_rate": 0.002828,
      "loss": 1.4332,
      "step": 186048
    },
    {
      "epoch": 13.967129455909944,
      "grad_norm": 0.4724648594856262,
      "learning_rate": 0.002828,
      "loss": 1.4357,
      "step": 186112
    },
    {
      "epoch": 13.971932457786117,
      "grad_norm": 0.5587869882583618,
      "learning_rate": 0.002828,
      "loss": 1.4331,
      "step": 186176
    },
    {
      "epoch": 13.976735459662288,
      "grad_norm": 0.45337358117103577,
      "learning_rate": 0.002828,
      "loss": 1.4296,
      "step": 186240
    },
    {
      "epoch": 13.981538461538461,
      "grad_norm": 0.515040934085846,
      "learning_rate": 0.002828,
      "loss": 1.4357,
      "step": 186304
    },
    {
      "epoch": 13.986341463414634,
      "grad_norm": 0.5681731700897217,
      "learning_rate": 0.002828,
      "loss": 1.4379,
      "step": 186368
    },
    {
      "epoch": 13.991144465290807,
      "grad_norm": 0.6006734371185303,
      "learning_rate": 0.002828,
      "loss": 1.4316,
      "step": 186432
    },
    {
      "epoch": 13.99594746716698,
      "grad_norm": 0.6272892951965332,
      "learning_rate": 0.002828,
      "loss": 1.4394,
      "step": 186496
    },
    {
      "epoch": 14.000750469043153,
      "grad_norm": 0.4696020185947418,
      "learning_rate": 0.002828,
      "loss": 1.4281,
      "step": 186560
    },
    {
      "epoch": 14.005553470919324,
      "grad_norm": 0.5587984919548035,
      "learning_rate": 0.002828,
      "loss": 1.3876,
      "step": 186624
    },
    {
      "epoch": 14.010356472795497,
      "grad_norm": 0.514490008354187,
      "learning_rate": 0.002828,
      "loss": 1.391,
      "step": 186688
    },
    {
      "epoch": 14.01515947467167,
      "grad_norm": 0.4929649233818054,
      "learning_rate": 0.002828,
      "loss": 1.3858,
      "step": 186752
    },
    {
      "epoch": 14.019962476547843,
      "grad_norm": 0.552597165107727,
      "learning_rate": 0.002828,
      "loss": 1.3846,
      "step": 186816
    },
    {
      "epoch": 14.024765478424015,
      "grad_norm": 0.6769406199455261,
      "learning_rate": 0.002828,
      "loss": 1.3929,
      "step": 186880
    },
    {
      "epoch": 14.029568480300188,
      "grad_norm": 0.5288282632827759,
      "learning_rate": 0.002828,
      "loss": 1.3921,
      "step": 186944
    },
    {
      "epoch": 14.03437148217636,
      "grad_norm": 0.4571775496006012,
      "learning_rate": 0.002828,
      "loss": 1.3965,
      "step": 187008
    },
    {
      "epoch": 14.039174484052532,
      "grad_norm": 0.6377585530281067,
      "learning_rate": 0.002828,
      "loss": 1.3923,
      "step": 187072
    },
    {
      "epoch": 14.043977485928705,
      "grad_norm": 0.5901743769645691,
      "learning_rate": 0.002828,
      "loss": 1.3912,
      "step": 187136
    },
    {
      "epoch": 14.048780487804878,
      "grad_norm": 0.45205187797546387,
      "learning_rate": 0.002828,
      "loss": 1.3922,
      "step": 187200
    },
    {
      "epoch": 14.053583489681051,
      "grad_norm": 0.6338102221488953,
      "learning_rate": 0.002828,
      "loss": 1.3913,
      "step": 187264
    },
    {
      "epoch": 14.058386491557224,
      "grad_norm": 0.515400230884552,
      "learning_rate": 0.002828,
      "loss": 1.3877,
      "step": 187328
    },
    {
      "epoch": 14.063189493433397,
      "grad_norm": 0.5111676454544067,
      "learning_rate": 0.002828,
      "loss": 1.39,
      "step": 187392
    },
    {
      "epoch": 14.067992495309568,
      "grad_norm": 0.5326048731803894,
      "learning_rate": 0.002828,
      "loss": 1.396,
      "step": 187456
    },
    {
      "epoch": 14.07279549718574,
      "grad_norm": 0.5814757347106934,
      "learning_rate": 0.002828,
      "loss": 1.3915,
      "step": 187520
    },
    {
      "epoch": 14.077598499061914,
      "grad_norm": 0.5279295444488525,
      "learning_rate": 0.002828,
      "loss": 1.3944,
      "step": 187584
    },
    {
      "epoch": 14.082401500938087,
      "grad_norm": 0.5335688591003418,
      "learning_rate": 0.002828,
      "loss": 1.4015,
      "step": 187648
    },
    {
      "epoch": 14.08720450281426,
      "grad_norm": 0.5352903008460999,
      "learning_rate": 0.002828,
      "loss": 1.3978,
      "step": 187712
    },
    {
      "epoch": 14.092007504690432,
      "grad_norm": 0.5011479258537292,
      "learning_rate": 0.002828,
      "loss": 1.3978,
      "step": 187776
    },
    {
      "epoch": 14.096810506566603,
      "grad_norm": 0.504731297492981,
      "learning_rate": 0.002828,
      "loss": 1.3983,
      "step": 187840
    },
    {
      "epoch": 14.101613508442776,
      "grad_norm": 0.5467804670333862,
      "learning_rate": 0.002828,
      "loss": 1.3979,
      "step": 187904
    },
    {
      "epoch": 14.10641651031895,
      "grad_norm": 0.5368660688400269,
      "learning_rate": 0.002828,
      "loss": 1.4006,
      "step": 187968
    },
    {
      "epoch": 14.111219512195122,
      "grad_norm": 0.5375622510910034,
      "learning_rate": 0.002828,
      "loss": 1.3932,
      "step": 188032
    },
    {
      "epoch": 14.116022514071295,
      "grad_norm": 0.5747446417808533,
      "learning_rate": 0.002828,
      "loss": 1.3975,
      "step": 188096
    },
    {
      "epoch": 14.120825515947468,
      "grad_norm": 0.4997574985027313,
      "learning_rate": 0.002828,
      "loss": 1.3893,
      "step": 188160
    },
    {
      "epoch": 14.125628517823639,
      "grad_norm": 0.5358232259750366,
      "learning_rate": 0.002828,
      "loss": 1.3987,
      "step": 188224
    },
    {
      "epoch": 14.130431519699812,
      "grad_norm": 0.5235414505004883,
      "learning_rate": 0.002828,
      "loss": 1.3933,
      "step": 188288
    },
    {
      "epoch": 14.135234521575985,
      "grad_norm": 0.5142853260040283,
      "learning_rate": 0.002828,
      "loss": 1.3996,
      "step": 188352
    },
    {
      "epoch": 14.140037523452158,
      "grad_norm": 0.5567483901977539,
      "learning_rate": 0.002828,
      "loss": 1.3967,
      "step": 188416
    },
    {
      "epoch": 14.14484052532833,
      "grad_norm": 0.590877890586853,
      "learning_rate": 0.002828,
      "loss": 1.4016,
      "step": 188480
    },
    {
      "epoch": 14.149643527204503,
      "grad_norm": 0.5319374203681946,
      "learning_rate": 0.002828,
      "loss": 1.3925,
      "step": 188544
    },
    {
      "epoch": 14.154446529080676,
      "grad_norm": 0.5865071415901184,
      "learning_rate": 0.002828,
      "loss": 1.3952,
      "step": 188608
    },
    {
      "epoch": 14.159249530956847,
      "grad_norm": 0.5776681303977966,
      "learning_rate": 0.002828,
      "loss": 1.3993,
      "step": 188672
    },
    {
      "epoch": 14.16405253283302,
      "grad_norm": 0.49375268816947937,
      "learning_rate": 0.002828,
      "loss": 1.3935,
      "step": 188736
    },
    {
      "epoch": 14.168855534709193,
      "grad_norm": 0.5858403444290161,
      "learning_rate": 0.002828,
      "loss": 1.4046,
      "step": 188800
    },
    {
      "epoch": 14.173658536585366,
      "grad_norm": 0.4694221615791321,
      "learning_rate": 0.002828,
      "loss": 1.3969,
      "step": 188864
    },
    {
      "epoch": 14.178461538461539,
      "grad_norm": 0.6087458729743958,
      "learning_rate": 0.002828,
      "loss": 1.3959,
      "step": 188928
    },
    {
      "epoch": 14.183264540337712,
      "grad_norm": 0.46161791682243347,
      "learning_rate": 0.002828,
      "loss": 1.4003,
      "step": 188992
    },
    {
      "epoch": 14.188067542213883,
      "grad_norm": 0.5340674519538879,
      "learning_rate": 0.002828,
      "loss": 1.4042,
      "step": 189056
    },
    {
      "epoch": 14.192870544090056,
      "grad_norm": 0.504485011100769,
      "learning_rate": 0.002828,
      "loss": 1.4034,
      "step": 189120
    },
    {
      "epoch": 14.197673545966229,
      "grad_norm": 0.584781289100647,
      "learning_rate": 0.002828,
      "loss": 1.3935,
      "step": 189184
    },
    {
      "epoch": 14.202476547842402,
      "grad_norm": 0.5133593082427979,
      "learning_rate": 0.002828,
      "loss": 1.3983,
      "step": 189248
    },
    {
      "epoch": 14.207279549718574,
      "grad_norm": 0.5022109746932983,
      "learning_rate": 0.002828,
      "loss": 1.4013,
      "step": 189312
    },
    {
      "epoch": 14.212082551594747,
      "grad_norm": 0.44956791400909424,
      "learning_rate": 0.002828,
      "loss": 1.4033,
      "step": 189376
    },
    {
      "epoch": 14.21688555347092,
      "grad_norm": 0.5756507515907288,
      "learning_rate": 0.002828,
      "loss": 1.4017,
      "step": 189440
    },
    {
      "epoch": 14.221688555347091,
      "grad_norm": 0.6916122436523438,
      "learning_rate": 0.002828,
      "loss": 1.4018,
      "step": 189504
    },
    {
      "epoch": 14.226491557223264,
      "grad_norm": 0.532734751701355,
      "learning_rate": 0.002828,
      "loss": 1.4015,
      "step": 189568
    },
    {
      "epoch": 14.231294559099437,
      "grad_norm": 0.726618230342865,
      "learning_rate": 0.002828,
      "loss": 1.3956,
      "step": 189632
    },
    {
      "epoch": 14.23609756097561,
      "grad_norm": 0.530405580997467,
      "learning_rate": 0.002828,
      "loss": 1.407,
      "step": 189696
    },
    {
      "epoch": 14.240900562851783,
      "grad_norm": 0.4780811369419098,
      "learning_rate": 0.002828,
      "loss": 1.4016,
      "step": 189760
    },
    {
      "epoch": 14.245703564727956,
      "grad_norm": 0.5396596193313599,
      "learning_rate": 0.002828,
      "loss": 1.403,
      "step": 189824
    },
    {
      "epoch": 14.250506566604127,
      "grad_norm": 0.4692460298538208,
      "learning_rate": 0.002828,
      "loss": 1.3945,
      "step": 189888
    },
    {
      "epoch": 14.2553095684803,
      "grad_norm": 0.4458451271057129,
      "learning_rate": 0.002828,
      "loss": 1.3985,
      "step": 189952
    },
    {
      "epoch": 14.260112570356473,
      "grad_norm": 0.5502991080284119,
      "learning_rate": 0.002828,
      "loss": 1.4037,
      "step": 190016
    },
    {
      "epoch": 14.264915572232646,
      "grad_norm": 0.5964510440826416,
      "learning_rate": 0.002828,
      "loss": 1.4016,
      "step": 190080
    },
    {
      "epoch": 14.269718574108818,
      "grad_norm": 0.49159181118011475,
      "learning_rate": 0.002828,
      "loss": 1.4036,
      "step": 190144
    },
    {
      "epoch": 14.274521575984991,
      "grad_norm": 0.4755879044532776,
      "learning_rate": 0.002828,
      "loss": 1.405,
      "step": 190208
    },
    {
      "epoch": 14.279324577861162,
      "grad_norm": 0.6329382061958313,
      "learning_rate": 0.002828,
      "loss": 1.3975,
      "step": 190272
    },
    {
      "epoch": 14.284127579737335,
      "grad_norm": 0.5941036343574524,
      "learning_rate": 0.002828,
      "loss": 1.3992,
      "step": 190336
    },
    {
      "epoch": 14.288930581613508,
      "grad_norm": 0.4472516179084778,
      "learning_rate": 0.002828,
      "loss": 1.402,
      "step": 190400
    },
    {
      "epoch": 14.293733583489681,
      "grad_norm": 0.614414393901825,
      "learning_rate": 0.002828,
      "loss": 1.4067,
      "step": 190464
    },
    {
      "epoch": 14.298536585365854,
      "grad_norm": 0.469532310962677,
      "learning_rate": 0.002828,
      "loss": 1.4017,
      "step": 190528
    },
    {
      "epoch": 14.303339587242027,
      "grad_norm": 0.49450796842575073,
      "learning_rate": 0.002828,
      "loss": 1.4022,
      "step": 190592
    },
    {
      "epoch": 14.3081425891182,
      "grad_norm": 0.6684780716896057,
      "learning_rate": 0.002828,
      "loss": 1.4056,
      "step": 190656
    },
    {
      "epoch": 14.31294559099437,
      "grad_norm": 0.5274812579154968,
      "learning_rate": 0.002828,
      "loss": 1.409,
      "step": 190720
    },
    {
      "epoch": 14.317748592870544,
      "grad_norm": 0.5079588890075684,
      "learning_rate": 0.002828,
      "loss": 1.3962,
      "step": 190784
    },
    {
      "epoch": 14.322551594746717,
      "grad_norm": 0.5359573364257812,
      "learning_rate": 0.002828,
      "loss": 1.4018,
      "step": 190848
    },
    {
      "epoch": 14.32735459662289,
      "grad_norm": 0.5036746859550476,
      "learning_rate": 0.002828,
      "loss": 1.3986,
      "step": 190912
    },
    {
      "epoch": 14.332157598499062,
      "grad_norm": 0.510995626449585,
      "learning_rate": 0.002828,
      "loss": 1.4031,
      "step": 190976
    },
    {
      "epoch": 14.336960600375235,
      "grad_norm": 0.48344898223876953,
      "learning_rate": 0.002828,
      "loss": 1.4071,
      "step": 191040
    },
    {
      "epoch": 14.341763602251406,
      "grad_norm": 0.5662881731987,
      "learning_rate": 0.002828,
      "loss": 1.4005,
      "step": 191104
    },
    {
      "epoch": 14.34656660412758,
      "grad_norm": 0.5350774526596069,
      "learning_rate": 0.002828,
      "loss": 1.402,
      "step": 191168
    },
    {
      "epoch": 14.351369606003752,
      "grad_norm": 0.5573024749755859,
      "learning_rate": 0.002828,
      "loss": 1.3971,
      "step": 191232
    },
    {
      "epoch": 14.356172607879925,
      "grad_norm": 0.5997970700263977,
      "learning_rate": 0.002828,
      "loss": 1.4008,
      "step": 191296
    },
    {
      "epoch": 14.360975609756098,
      "grad_norm": 0.5876054763793945,
      "learning_rate": 0.002828,
      "loss": 1.404,
      "step": 191360
    },
    {
      "epoch": 14.36577861163227,
      "grad_norm": 0.5237852931022644,
      "learning_rate": 0.002828,
      "loss": 1.4115,
      "step": 191424
    },
    {
      "epoch": 14.370581613508442,
      "grad_norm": 0.5114300847053528,
      "learning_rate": 0.002828,
      "loss": 1.405,
      "step": 191488
    },
    {
      "epoch": 14.375384615384615,
      "grad_norm": 0.5123426914215088,
      "learning_rate": 0.002828,
      "loss": 1.3975,
      "step": 191552
    },
    {
      "epoch": 14.380187617260788,
      "grad_norm": 0.5161942839622498,
      "learning_rate": 0.002828,
      "loss": 1.4076,
      "step": 191616
    },
    {
      "epoch": 14.38499061913696,
      "grad_norm": 0.5509490370750427,
      "learning_rate": 0.002828,
      "loss": 1.4038,
      "step": 191680
    },
    {
      "epoch": 14.389793621013133,
      "grad_norm": 0.49601686000823975,
      "learning_rate": 0.002828,
      "loss": 1.4021,
      "step": 191744
    },
    {
      "epoch": 14.394596622889306,
      "grad_norm": 0.46365320682525635,
      "learning_rate": 0.002828,
      "loss": 1.4075,
      "step": 191808
    },
    {
      "epoch": 14.39939962476548,
      "grad_norm": 0.5520146489143372,
      "learning_rate": 0.002828,
      "loss": 1.4054,
      "step": 191872
    },
    {
      "epoch": 14.40420262664165,
      "grad_norm": 0.49274471402168274,
      "learning_rate": 0.002828,
      "loss": 1.4005,
      "step": 191936
    },
    {
      "epoch": 14.409005628517823,
      "grad_norm": 0.6408708095550537,
      "learning_rate": 0.002828,
      "loss": 1.3995,
      "step": 192000
    },
    {
      "epoch": 14.413808630393996,
      "grad_norm": 0.5490612983703613,
      "learning_rate": 0.002828,
      "loss": 1.4086,
      "step": 192064
    },
    {
      "epoch": 14.418611632270169,
      "grad_norm": 0.5333877205848694,
      "learning_rate": 0.002828,
      "loss": 1.4091,
      "step": 192128
    },
    {
      "epoch": 14.423414634146342,
      "grad_norm": 0.5989339351654053,
      "learning_rate": 0.002828,
      "loss": 1.4026,
      "step": 192192
    },
    {
      "epoch": 14.428217636022515,
      "grad_norm": 0.5512287020683289,
      "learning_rate": 0.002828,
      "loss": 1.4088,
      "step": 192256
    },
    {
      "epoch": 14.433020637898686,
      "grad_norm": 0.5418707132339478,
      "learning_rate": 0.002828,
      "loss": 1.4048,
      "step": 192320
    },
    {
      "epoch": 14.437823639774859,
      "grad_norm": 0.5657520890235901,
      "learning_rate": 0.002828,
      "loss": 1.4074,
      "step": 192384
    },
    {
      "epoch": 14.442626641651032,
      "grad_norm": 0.5935457348823547,
      "learning_rate": 0.002828,
      "loss": 1.4041,
      "step": 192448
    },
    {
      "epoch": 14.447429643527204,
      "grad_norm": 0.5439548492431641,
      "learning_rate": 0.002828,
      "loss": 1.4035,
      "step": 192512
    },
    {
      "epoch": 14.452232645403377,
      "grad_norm": 0.5025527477264404,
      "learning_rate": 0.002828,
      "loss": 1.4072,
      "step": 192576
    },
    {
      "epoch": 14.45703564727955,
      "grad_norm": 0.46855494379997253,
      "learning_rate": 0.002828,
      "loss": 1.4017,
      "step": 192640
    },
    {
      "epoch": 14.461838649155723,
      "grad_norm": 0.5161790251731873,
      "learning_rate": 0.002828,
      "loss": 1.4117,
      "step": 192704
    },
    {
      "epoch": 14.466641651031894,
      "grad_norm": 0.52967369556427,
      "learning_rate": 0.002828,
      "loss": 1.4033,
      "step": 192768
    },
    {
      "epoch": 14.471444652908067,
      "grad_norm": 0.5561485290527344,
      "learning_rate": 0.002828,
      "loss": 1.4057,
      "step": 192832
    },
    {
      "epoch": 14.47624765478424,
      "grad_norm": 0.5013352632522583,
      "learning_rate": 0.002828,
      "loss": 1.4089,
      "step": 192896
    },
    {
      "epoch": 14.481050656660413,
      "grad_norm": 0.4954836070537567,
      "learning_rate": 0.002828,
      "loss": 1.4081,
      "step": 192960
    },
    {
      "epoch": 14.485853658536586,
      "grad_norm": 0.4955345690250397,
      "learning_rate": 0.002828,
      "loss": 1.4092,
      "step": 193024
    },
    {
      "epoch": 14.490656660412759,
      "grad_norm": 0.5120537877082825,
      "learning_rate": 0.002828,
      "loss": 1.4062,
      "step": 193088
    },
    {
      "epoch": 14.49545966228893,
      "grad_norm": 0.5533945560455322,
      "learning_rate": 0.002828,
      "loss": 1.4134,
      "step": 193152
    },
    {
      "epoch": 14.500262664165103,
      "grad_norm": 0.5256770849227905,
      "learning_rate": 0.002828,
      "loss": 1.4035,
      "step": 193216
    },
    {
      "epoch": 14.505065666041276,
      "grad_norm": 0.5609396696090698,
      "learning_rate": 0.002828,
      "loss": 1.4016,
      "step": 193280
    },
    {
      "epoch": 14.509868667917448,
      "grad_norm": 0.6170324683189392,
      "learning_rate": 0.002828,
      "loss": 1.4034,
      "step": 193344
    },
    {
      "epoch": 14.514671669793621,
      "grad_norm": 0.5539875626564026,
      "learning_rate": 0.002828,
      "loss": 1.4052,
      "step": 193408
    },
    {
      "epoch": 14.519474671669794,
      "grad_norm": 0.5213680267333984,
      "learning_rate": 0.002828,
      "loss": 1.4072,
      "step": 193472
    },
    {
      "epoch": 14.524277673545967,
      "grad_norm": 0.5134106278419495,
      "learning_rate": 0.002828,
      "loss": 1.4126,
      "step": 193536
    },
    {
      "epoch": 14.529080675422138,
      "grad_norm": 0.5207715630531311,
      "learning_rate": 0.002828,
      "loss": 1.4099,
      "step": 193600
    },
    {
      "epoch": 14.533883677298311,
      "grad_norm": 0.6132842302322388,
      "learning_rate": 0.002828,
      "loss": 1.4017,
      "step": 193664
    },
    {
      "epoch": 14.538686679174484,
      "grad_norm": 0.5139353275299072,
      "learning_rate": 0.002828,
      "loss": 1.4039,
      "step": 193728
    },
    {
      "epoch": 14.543489681050657,
      "grad_norm": 0.5206514000892639,
      "learning_rate": 0.002828,
      "loss": 1.4017,
      "step": 193792
    },
    {
      "epoch": 14.54829268292683,
      "grad_norm": 0.5245344638824463,
      "learning_rate": 0.002828,
      "loss": 1.4036,
      "step": 193856
    },
    {
      "epoch": 14.553095684803003,
      "grad_norm": 0.48017024993896484,
      "learning_rate": 0.002828,
      "loss": 1.4082,
      "step": 193920
    },
    {
      "epoch": 14.557898686679174,
      "grad_norm": 0.5209543704986572,
      "learning_rate": 0.002828,
      "loss": 1.4007,
      "step": 193984
    },
    {
      "epoch": 14.562701688555347,
      "grad_norm": 0.6338755488395691,
      "learning_rate": 0.002828,
      "loss": 1.4064,
      "step": 194048
    },
    {
      "epoch": 14.56750469043152,
      "grad_norm": 0.5111967325210571,
      "learning_rate": 0.002828,
      "loss": 1.3982,
      "step": 194112
    },
    {
      "epoch": 14.572307692307692,
      "grad_norm": 0.5508695244789124,
      "learning_rate": 0.002828,
      "loss": 1.409,
      "step": 194176
    },
    {
      "epoch": 14.577110694183865,
      "grad_norm": 0.549004077911377,
      "learning_rate": 0.002828,
      "loss": 1.4103,
      "step": 194240
    },
    {
      "epoch": 14.581913696060038,
      "grad_norm": 0.4958650767803192,
      "learning_rate": 0.002828,
      "loss": 1.4077,
      "step": 194304
    },
    {
      "epoch": 14.586716697936211,
      "grad_norm": 0.5928217172622681,
      "learning_rate": 0.002828,
      "loss": 1.4053,
      "step": 194368
    },
    {
      "epoch": 14.591519699812382,
      "grad_norm": 0.5710644125938416,
      "learning_rate": 0.002828,
      "loss": 1.4077,
      "step": 194432
    },
    {
      "epoch": 14.596322701688555,
      "grad_norm": 0.5788741111755371,
      "learning_rate": 0.002828,
      "loss": 1.4032,
      "step": 194496
    },
    {
      "epoch": 14.601125703564728,
      "grad_norm": 0.4544178247451782,
      "learning_rate": 0.002828,
      "loss": 1.4017,
      "step": 194560
    },
    {
      "epoch": 14.6059287054409,
      "grad_norm": 0.5514257550239563,
      "learning_rate": 0.002828,
      "loss": 1.4081,
      "step": 194624
    },
    {
      "epoch": 14.610731707317074,
      "grad_norm": 0.48939162492752075,
      "learning_rate": 0.002828,
      "loss": 1.4081,
      "step": 194688
    },
    {
      "epoch": 14.615534709193247,
      "grad_norm": 0.7316038608551025,
      "learning_rate": 0.002828,
      "loss": 1.4049,
      "step": 194752
    },
    {
      "epoch": 14.620337711069418,
      "grad_norm": 0.5802710652351379,
      "learning_rate": 0.002828,
      "loss": 1.4035,
      "step": 194816
    },
    {
      "epoch": 14.62514071294559,
      "grad_norm": 0.5377091765403748,
      "learning_rate": 0.002828,
      "loss": 1.4026,
      "step": 194880
    },
    {
      "epoch": 14.629943714821763,
      "grad_norm": 0.5343953371047974,
      "learning_rate": 0.002828,
      "loss": 1.4099,
      "step": 194944
    },
    {
      "epoch": 14.634746716697936,
      "grad_norm": 0.5183101296424866,
      "learning_rate": 0.002828,
      "loss": 1.4083,
      "step": 195008
    },
    {
      "epoch": 14.63954971857411,
      "grad_norm": 0.48518890142440796,
      "learning_rate": 0.002828,
      "loss": 1.4111,
      "step": 195072
    },
    {
      "epoch": 14.644352720450282,
      "grad_norm": 0.5483662486076355,
      "learning_rate": 0.002828,
      "loss": 1.4027,
      "step": 195136
    },
    {
      "epoch": 14.649155722326453,
      "grad_norm": 0.5991758108139038,
      "learning_rate": 0.002828,
      "loss": 1.4041,
      "step": 195200
    },
    {
      "epoch": 14.653958724202626,
      "grad_norm": 0.554240345954895,
      "learning_rate": 0.002828,
      "loss": 1.4085,
      "step": 195264
    },
    {
      "epoch": 14.658761726078799,
      "grad_norm": 0.5651766061782837,
      "learning_rate": 0.002828,
      "loss": 1.4124,
      "step": 195328
    },
    {
      "epoch": 14.663564727954972,
      "grad_norm": 0.5138752460479736,
      "learning_rate": 0.002828,
      "loss": 1.4122,
      "step": 195392
    },
    {
      "epoch": 14.668367729831145,
      "grad_norm": 0.5441687703132629,
      "learning_rate": 0.002828,
      "loss": 1.4135,
      "step": 195456
    },
    {
      "epoch": 14.673170731707318,
      "grad_norm": 0.5138674974441528,
      "learning_rate": 0.002828,
      "loss": 1.4086,
      "step": 195520
    },
    {
      "epoch": 14.677973733583489,
      "grad_norm": 0.5577060580253601,
      "learning_rate": 0.002828,
      "loss": 1.4129,
      "step": 195584
    },
    {
      "epoch": 14.682776735459662,
      "grad_norm": 0.6189695000648499,
      "learning_rate": 0.002828,
      "loss": 1.4085,
      "step": 195648
    },
    {
      "epoch": 14.687579737335835,
      "grad_norm": 0.5921385884284973,
      "learning_rate": 0.002828,
      "loss": 1.4108,
      "step": 195712
    },
    {
      "epoch": 14.692382739212007,
      "grad_norm": 0.577090859413147,
      "learning_rate": 0.002828,
      "loss": 1.4071,
      "step": 195776
    },
    {
      "epoch": 14.69718574108818,
      "grad_norm": 0.553539514541626,
      "learning_rate": 0.002828,
      "loss": 1.3994,
      "step": 195840
    },
    {
      "epoch": 14.701988742964353,
      "grad_norm": 0.4840497374534607,
      "learning_rate": 0.002828,
      "loss": 1.4099,
      "step": 195904
    },
    {
      "epoch": 14.706791744840526,
      "grad_norm": 0.5382407307624817,
      "learning_rate": 0.002828,
      "loss": 1.4073,
      "step": 195968
    },
    {
      "epoch": 14.711594746716697,
      "grad_norm": 0.5615788102149963,
      "learning_rate": 0.002828,
      "loss": 1.4057,
      "step": 196032
    },
    {
      "epoch": 14.71639774859287,
      "grad_norm": 0.46225443482398987,
      "learning_rate": 0.002828,
      "loss": 1.4049,
      "step": 196096
    },
    {
      "epoch": 14.721200750469043,
      "grad_norm": 0.49434852600097656,
      "learning_rate": 0.002828,
      "loss": 1.4065,
      "step": 196160
    },
    {
      "epoch": 14.726003752345216,
      "grad_norm": 0.7275422811508179,
      "learning_rate": 0.002828,
      "loss": 1.4059,
      "step": 196224
    },
    {
      "epoch": 14.730806754221389,
      "grad_norm": 0.542582094669342,
      "learning_rate": 0.002828,
      "loss": 1.4071,
      "step": 196288
    },
    {
      "epoch": 14.735609756097562,
      "grad_norm": 0.4883684515953064,
      "learning_rate": 0.002828,
      "loss": 1.4058,
      "step": 196352
    },
    {
      "epoch": 14.740412757973733,
      "grad_norm": 0.5530311465263367,
      "learning_rate": 0.002828,
      "loss": 1.406,
      "step": 196416
    },
    {
      "epoch": 14.745215759849906,
      "grad_norm": 0.5233826041221619,
      "learning_rate": 0.002828,
      "loss": 1.4044,
      "step": 196480
    },
    {
      "epoch": 14.750018761726079,
      "grad_norm": 0.5350394248962402,
      "learning_rate": 0.002828,
      "loss": 1.4089,
      "step": 196544
    },
    {
      "epoch": 14.754821763602251,
      "grad_norm": 0.550305962562561,
      "learning_rate": 0.002828,
      "loss": 1.3991,
      "step": 196608
    },
    {
      "epoch": 14.759624765478424,
      "grad_norm": 0.5499565005302429,
      "learning_rate": 0.002828,
      "loss": 1.4074,
      "step": 196672
    },
    {
      "epoch": 14.764427767354597,
      "grad_norm": 0.6032944917678833,
      "learning_rate": 0.002828,
      "loss": 1.407,
      "step": 196736
    },
    {
      "epoch": 14.76923076923077,
      "grad_norm": 0.45191773772239685,
      "learning_rate": 0.002828,
      "loss": 1.4066,
      "step": 196800
    },
    {
      "epoch": 14.774033771106941,
      "grad_norm": 0.7020035982131958,
      "learning_rate": 0.002828,
      "loss": 1.4012,
      "step": 196864
    },
    {
      "epoch": 14.778836772983114,
      "grad_norm": 0.5450897216796875,
      "learning_rate": 0.002828,
      "loss": 1.4012,
      "step": 196928
    },
    {
      "epoch": 14.783639774859287,
      "grad_norm": 0.5871745347976685,
      "learning_rate": 0.002828,
      "loss": 1.4064,
      "step": 196992
    },
    {
      "epoch": 14.78844277673546,
      "grad_norm": 0.5858376622200012,
      "learning_rate": 0.002828,
      "loss": 1.4134,
      "step": 197056
    },
    {
      "epoch": 14.793245778611633,
      "grad_norm": 0.4942261576652527,
      "learning_rate": 0.002828,
      "loss": 1.405,
      "step": 197120
    },
    {
      "epoch": 14.798048780487806,
      "grad_norm": 0.4891422390937805,
      "learning_rate": 0.002828,
      "loss": 1.4166,
      "step": 197184
    },
    {
      "epoch": 14.802851782363977,
      "grad_norm": 0.5212449431419373,
      "learning_rate": 0.002828,
      "loss": 1.4097,
      "step": 197248
    },
    {
      "epoch": 14.80765478424015,
      "grad_norm": 0.48919591307640076,
      "learning_rate": 0.002828,
      "loss": 1.4023,
      "step": 197312
    },
    {
      "epoch": 14.812457786116322,
      "grad_norm": 0.5116396546363831,
      "learning_rate": 0.002828,
      "loss": 1.4029,
      "step": 197376
    },
    {
      "epoch": 14.817260787992495,
      "grad_norm": 0.5862394571304321,
      "learning_rate": 0.002828,
      "loss": 1.4091,
      "step": 197440
    },
    {
      "epoch": 14.822063789868668,
      "grad_norm": 0.6540351510047913,
      "learning_rate": 0.002828,
      "loss": 1.4074,
      "step": 197504
    },
    {
      "epoch": 14.826866791744841,
      "grad_norm": 0.5462371110916138,
      "learning_rate": 0.002828,
      "loss": 1.4067,
      "step": 197568
    },
    {
      "epoch": 14.831669793621014,
      "grad_norm": 0.5242631435394287,
      "learning_rate": 0.002828,
      "loss": 1.4058,
      "step": 197632
    },
    {
      "epoch": 14.836472795497185,
      "grad_norm": 0.6098849177360535,
      "learning_rate": 0.002828,
      "loss": 1.4088,
      "step": 197696
    },
    {
      "epoch": 14.841275797373358,
      "grad_norm": 0.601551353931427,
      "learning_rate": 0.002828,
      "loss": 1.4041,
      "step": 197760
    },
    {
      "epoch": 14.846078799249531,
      "grad_norm": 0.6099250912666321,
      "learning_rate": 0.002828,
      "loss": 1.4057,
      "step": 197824
    },
    {
      "epoch": 14.850881801125704,
      "grad_norm": 0.4936770796775818,
      "learning_rate": 0.002828,
      "loss": 1.4117,
      "step": 197888
    },
    {
      "epoch": 14.855684803001877,
      "grad_norm": 0.5036704540252686,
      "learning_rate": 0.002828,
      "loss": 1.4132,
      "step": 197952
    },
    {
      "epoch": 14.86048780487805,
      "grad_norm": 0.5548554062843323,
      "learning_rate": 0.002828,
      "loss": 1.4105,
      "step": 198016
    },
    {
      "epoch": 14.86529080675422,
      "grad_norm": 0.5694041848182678,
      "learning_rate": 0.002828,
      "loss": 1.4101,
      "step": 198080
    },
    {
      "epoch": 14.870093808630394,
      "grad_norm": 0.5238790512084961,
      "learning_rate": 0.002828,
      "loss": 1.416,
      "step": 198144
    },
    {
      "epoch": 14.874896810506566,
      "grad_norm": 0.5360506176948547,
      "learning_rate": 0.002828,
      "loss": 1.4096,
      "step": 198208
    },
    {
      "epoch": 14.87969981238274,
      "grad_norm": 0.6266505122184753,
      "learning_rate": 0.002828,
      "loss": 1.4102,
      "step": 198272
    },
    {
      "epoch": 14.884502814258912,
      "grad_norm": 0.5030892491340637,
      "learning_rate": 0.002828,
      "loss": 1.4052,
      "step": 198336
    },
    {
      "epoch": 14.889305816135085,
      "grad_norm": 0.48328977823257446,
      "learning_rate": 0.002828,
      "loss": 1.4052,
      "step": 198400
    },
    {
      "epoch": 14.894108818011258,
      "grad_norm": 0.567842960357666,
      "learning_rate": 0.002828,
      "loss": 1.4113,
      "step": 198464
    },
    {
      "epoch": 14.898911819887429,
      "grad_norm": 0.507936418056488,
      "learning_rate": 0.002828,
      "loss": 1.4117,
      "step": 198528
    },
    {
      "epoch": 14.903714821763602,
      "grad_norm": 0.46350201964378357,
      "learning_rate": 0.002828,
      "loss": 1.4081,
      "step": 198592
    },
    {
      "epoch": 14.908517823639775,
      "grad_norm": 0.550236165523529,
      "learning_rate": 0.002828,
      "loss": 1.4115,
      "step": 198656
    },
    {
      "epoch": 14.913320825515948,
      "grad_norm": 0.5407876372337341,
      "learning_rate": 0.002828,
      "loss": 1.415,
      "step": 198720
    },
    {
      "epoch": 14.91812382739212,
      "grad_norm": 0.4663850665092468,
      "learning_rate": 0.002828,
      "loss": 1.4109,
      "step": 198784
    },
    {
      "epoch": 14.922926829268294,
      "grad_norm": 0.5446844100952148,
      "learning_rate": 0.002828,
      "loss": 1.4141,
      "step": 198848
    },
    {
      "epoch": 14.927729831144465,
      "grad_norm": 0.517081081867218,
      "learning_rate": 0.002828,
      "loss": 1.4122,
      "step": 198912
    },
    {
      "epoch": 14.932532833020637,
      "grad_norm": 0.5174572467803955,
      "learning_rate": 0.002828,
      "loss": 1.411,
      "step": 198976
    },
    {
      "epoch": 14.93733583489681,
      "grad_norm": 0.5665841698646545,
      "learning_rate": 0.002828,
      "loss": 1.4124,
      "step": 199040
    },
    {
      "epoch": 14.942138836772983,
      "grad_norm": 0.6959680318832397,
      "learning_rate": 0.002828,
      "loss": 1.4055,
      "step": 199104
    },
    {
      "epoch": 14.946941838649156,
      "grad_norm": 0.7196769714355469,
      "learning_rate": 0.002828,
      "loss": 1.41,
      "step": 199168
    },
    {
      "epoch": 14.951744840525329,
      "grad_norm": 0.5885917544364929,
      "learning_rate": 0.002828,
      "loss": 1.41,
      "step": 199232
    },
    {
      "epoch": 14.9565478424015,
      "grad_norm": 0.5237954258918762,
      "learning_rate": 0.002828,
      "loss": 1.4133,
      "step": 199296
    },
    {
      "epoch": 14.961350844277673,
      "grad_norm": 0.5811425447463989,
      "learning_rate": 0.002828,
      "loss": 1.4104,
      "step": 199360
    },
    {
      "epoch": 14.966153846153846,
      "grad_norm": 0.4897754192352295,
      "learning_rate": 0.002828,
      "loss": 1.4087,
      "step": 199424
    },
    {
      "epoch": 14.970956848030019,
      "grad_norm": 0.6255586743354797,
      "learning_rate": 0.002828,
      "loss": 1.4179,
      "step": 199488
    },
    {
      "epoch": 14.975759849906192,
      "grad_norm": 0.617106556892395,
      "learning_rate": 0.002828,
      "loss": 1.4058,
      "step": 199552
    },
    {
      "epoch": 14.980562851782365,
      "grad_norm": 0.5452525019645691,
      "learning_rate": 0.002828,
      "loss": 1.4179,
      "step": 199616
    },
    {
      "epoch": 14.985365853658536,
      "grad_norm": 0.6155555248260498,
      "learning_rate": 0.002828,
      "loss": 1.4104,
      "step": 199680
    },
    {
      "epoch": 14.990168855534709,
      "grad_norm": 0.4624778628349304,
      "learning_rate": 0.002828,
      "loss": 1.4117,
      "step": 199744
    },
    {
      "epoch": 14.994971857410881,
      "grad_norm": 0.5621700286865234,
      "learning_rate": 0.002828,
      "loss": 1.4115,
      "step": 199808
    },
    {
      "epoch": 14.999774859287054,
      "grad_norm": 0.5766751766204834,
      "learning_rate": 0.002828,
      "loss": 1.4116,
      "step": 199872
    },
    {
      "epoch": 15.004577861163227,
      "grad_norm": 0.5539935231208801,
      "learning_rate": 0.002828,
      "loss": 1.3687,
      "step": 199936
    },
    {
      "epoch": 15.0093808630394,
      "grad_norm": 0.518322765827179,
      "learning_rate": 0.002828,
      "loss": 1.3684,
      "step": 200000
    },
    {
      "epoch": 15.014183864915573,
      "grad_norm": 0.48596128821372986,
      "learning_rate": 0.002828,
      "loss": 1.3641,
      "step": 200064
    },
    {
      "epoch": 15.018986866791744,
      "grad_norm": 0.46796655654907227,
      "learning_rate": 0.002828,
      "loss": 1.3666,
      "step": 200128
    },
    {
      "epoch": 15.023789868667917,
      "grad_norm": 0.5862635374069214,
      "learning_rate": 0.002828,
      "loss": 1.3703,
      "step": 200192
    },
    {
      "epoch": 15.02859287054409,
      "grad_norm": 0.48129528760910034,
      "learning_rate": 0.002828,
      "loss": 1.3698,
      "step": 200256
    },
    {
      "epoch": 15.033395872420263,
      "grad_norm": 0.5526949167251587,
      "learning_rate": 0.002828,
      "loss": 1.3714,
      "step": 200320
    },
    {
      "epoch": 15.038198874296436,
      "grad_norm": 0.6339066028594971,
      "learning_rate": 0.002828,
      "loss": 1.3652,
      "step": 200384
    },
    {
      "epoch": 15.043001876172609,
      "grad_norm": 0.7033270597457886,
      "learning_rate": 0.002828,
      "loss": 1.3748,
      "step": 200448
    },
    {
      "epoch": 15.04780487804878,
      "grad_norm": 0.5913888216018677,
      "learning_rate": 0.002828,
      "loss": 1.3612,
      "step": 200512
    },
    {
      "epoch": 15.052607879924953,
      "grad_norm": 0.6152012944221497,
      "learning_rate": 0.002828,
      "loss": 1.369,
      "step": 200576
    },
    {
      "epoch": 15.057410881801125,
      "grad_norm": 0.5023623704910278,
      "learning_rate": 0.002828,
      "loss": 1.3675,
      "step": 200640
    },
    {
      "epoch": 15.062213883677298,
      "grad_norm": 0.5588781833648682,
      "learning_rate": 0.002828,
      "loss": 1.3697,
      "step": 200704
    },
    {
      "epoch": 15.067016885553471,
      "grad_norm": 0.5159551501274109,
      "learning_rate": 0.002828,
      "loss": 1.3702,
      "step": 200768
    },
    {
      "epoch": 15.071819887429644,
      "grad_norm": 0.6186797022819519,
      "learning_rate": 0.002828,
      "loss": 1.373,
      "step": 200832
    },
    {
      "epoch": 15.076622889305817,
      "grad_norm": 0.5377756357192993,
      "learning_rate": 0.002828,
      "loss": 1.3685,
      "step": 200896
    },
    {
      "epoch": 15.081425891181988,
      "grad_norm": 0.679719865322113,
      "learning_rate": 0.002828,
      "loss": 1.3656,
      "step": 200960
    },
    {
      "epoch": 15.086228893058161,
      "grad_norm": 0.47345390915870667,
      "learning_rate": 0.002828,
      "loss": 1.3731,
      "step": 201024
    },
    {
      "epoch": 15.091031894934334,
      "grad_norm": 0.6200671792030334,
      "learning_rate": 0.002828,
      "loss": 1.3692,
      "step": 201088
    },
    {
      "epoch": 15.095834896810507,
      "grad_norm": 0.5481675267219543,
      "learning_rate": 0.002828,
      "loss": 1.3709,
      "step": 201152
    },
    {
      "epoch": 15.10063789868668,
      "grad_norm": 0.5293875336647034,
      "learning_rate": 0.002828,
      "loss": 1.376,
      "step": 201216
    },
    {
      "epoch": 15.105440900562852,
      "grad_norm": 0.4326186776161194,
      "learning_rate": 0.002828,
      "loss": 1.3772,
      "step": 201280
    },
    {
      "epoch": 15.110243902439024,
      "grad_norm": 0.7246432900428772,
      "learning_rate": 0.002828,
      "loss": 1.3748,
      "step": 201344
    },
    {
      "epoch": 15.115046904315196,
      "grad_norm": 0.5310239195823669,
      "learning_rate": 0.002828,
      "loss": 1.3749,
      "step": 201408
    },
    {
      "epoch": 15.11984990619137,
      "grad_norm": 0.48077937960624695,
      "learning_rate": 0.002828,
      "loss": 1.3682,
      "step": 201472
    },
    {
      "epoch": 15.124652908067542,
      "grad_norm": 0.5190368890762329,
      "learning_rate": 0.002828,
      "loss": 1.3728,
      "step": 201536
    },
    {
      "epoch": 15.129455909943715,
      "grad_norm": 0.6393243074417114,
      "learning_rate": 0.002828,
      "loss": 1.3781,
      "step": 201600
    },
    {
      "epoch": 15.134258911819888,
      "grad_norm": 0.5104050636291504,
      "learning_rate": 0.002828,
      "loss": 1.3743,
      "step": 201664
    },
    {
      "epoch": 15.139061913696061,
      "grad_norm": 0.46626007556915283,
      "learning_rate": 0.002828,
      "loss": 1.3776,
      "step": 201728
    },
    {
      "epoch": 15.143864915572232,
      "grad_norm": 0.5677950978279114,
      "learning_rate": 0.002828,
      "loss": 1.3767,
      "step": 201792
    },
    {
      "epoch": 15.148667917448405,
      "grad_norm": 0.47358930110931396,
      "learning_rate": 0.002828,
      "loss": 1.3763,
      "step": 201856
    },
    {
      "epoch": 15.153470919324578,
      "grad_norm": 0.6243211030960083,
      "learning_rate": 0.002828,
      "loss": 1.3805,
      "step": 201920
    },
    {
      "epoch": 15.15827392120075,
      "grad_norm": 0.6380032896995544,
      "learning_rate": 0.002828,
      "loss": 1.3793,
      "step": 201984
    },
    {
      "epoch": 15.163076923076924,
      "grad_norm": 0.4932658076286316,
      "learning_rate": 0.002828,
      "loss": 1.3752,
      "step": 202048
    },
    {
      "epoch": 15.167879924953096,
      "grad_norm": 0.5148807764053345,
      "learning_rate": 0.002828,
      "loss": 1.3764,
      "step": 202112
    },
    {
      "epoch": 15.172682926829268,
      "grad_norm": 0.5843935608863831,
      "learning_rate": 0.002828,
      "loss": 1.3772,
      "step": 202176
    },
    {
      "epoch": 15.17748592870544,
      "grad_norm": 0.46910738945007324,
      "learning_rate": 0.002828,
      "loss": 1.3737,
      "step": 202240
    },
    {
      "epoch": 15.182288930581613,
      "grad_norm": 0.683540403842926,
      "learning_rate": 0.002828,
      "loss": 1.3769,
      "step": 202304
    },
    {
      "epoch": 15.187091932457786,
      "grad_norm": 0.5317997932434082,
      "learning_rate": 0.002828,
      "loss": 1.3818,
      "step": 202368
    },
    {
      "epoch": 15.191894934333959,
      "grad_norm": 0.55516117811203,
      "learning_rate": 0.002828,
      "loss": 1.3796,
      "step": 202432
    },
    {
      "epoch": 15.196697936210132,
      "grad_norm": 0.5858634114265442,
      "learning_rate": 0.002828,
      "loss": 1.3733,
      "step": 202496
    },
    {
      "epoch": 15.201500938086303,
      "grad_norm": 0.5299066305160522,
      "learning_rate": 0.002828,
      "loss": 1.3725,
      "step": 202560
    },
    {
      "epoch": 15.206303939962476,
      "grad_norm": 0.5147692561149597,
      "learning_rate": 0.002828,
      "loss": 1.3785,
      "step": 202624
    },
    {
      "epoch": 15.211106941838649,
      "grad_norm": 0.5491654276847839,
      "learning_rate": 0.002828,
      "loss": 1.3741,
      "step": 202688
    },
    {
      "epoch": 15.215909943714822,
      "grad_norm": 0.6498444080352783,
      "learning_rate": 0.002828,
      "loss": 1.38,
      "step": 202752
    },
    {
      "epoch": 15.220712945590995,
      "grad_norm": 0.4774509072303772,
      "learning_rate": 0.002828,
      "loss": 1.3837,
      "step": 202816
    },
    {
      "epoch": 15.225515947467168,
      "grad_norm": 0.46530207991600037,
      "learning_rate": 0.002828,
      "loss": 1.3756,
      "step": 202880
    },
    {
      "epoch": 15.23031894934334,
      "grad_norm": 0.47992247343063354,
      "learning_rate": 0.002828,
      "loss": 1.3722,
      "step": 202944
    },
    {
      "epoch": 15.235121951219512,
      "grad_norm": 0.6161879897117615,
      "learning_rate": 0.002828,
      "loss": 1.3792,
      "step": 203008
    },
    {
      "epoch": 15.239924953095684,
      "grad_norm": 0.5422763228416443,
      "learning_rate": 0.002828,
      "loss": 1.3738,
      "step": 203072
    },
    {
      "epoch": 15.244727954971857,
      "grad_norm": 0.4786650240421295,
      "learning_rate": 0.002828,
      "loss": 1.3798,
      "step": 203136
    },
    {
      "epoch": 15.24953095684803,
      "grad_norm": 0.5520129799842834,
      "learning_rate": 0.002828,
      "loss": 1.3786,
      "step": 203200
    },
    {
      "epoch": 15.254333958724203,
      "grad_norm": 0.5212474465370178,
      "learning_rate": 0.002828,
      "loss": 1.381,
      "step": 203264
    },
    {
      "epoch": 15.259136960600376,
      "grad_norm": 0.5647684931755066,
      "learning_rate": 0.002828,
      "loss": 1.3776,
      "step": 203328
    },
    {
      "epoch": 15.263939962476547,
      "grad_norm": 0.5040179491043091,
      "learning_rate": 0.002828,
      "loss": 1.3732,
      "step": 203392
    },
    {
      "epoch": 15.26874296435272,
      "grad_norm": 0.5457432866096497,
      "learning_rate": 0.002828,
      "loss": 1.3776,
      "step": 203456
    },
    {
      "epoch": 15.273545966228893,
      "grad_norm": 0.5505269765853882,
      "learning_rate": 0.002828,
      "loss": 1.3809,
      "step": 203520
    },
    {
      "epoch": 15.278348968105066,
      "grad_norm": 0.5592472553253174,
      "learning_rate": 0.002828,
      "loss": 1.3862,
      "step": 203584
    },
    {
      "epoch": 15.283151969981239,
      "grad_norm": 0.5437265634536743,
      "learning_rate": 0.002828,
      "loss": 1.3823,
      "step": 203648
    },
    {
      "epoch": 15.287954971857411,
      "grad_norm": 0.5756497383117676,
      "learning_rate": 0.002828,
      "loss": 1.3781,
      "step": 203712
    },
    {
      "epoch": 15.292757973733583,
      "grad_norm": 0.5972731709480286,
      "learning_rate": 0.002828,
      "loss": 1.3715,
      "step": 203776
    },
    {
      "epoch": 15.297560975609755,
      "grad_norm": 0.4744695723056793,
      "learning_rate": 0.002828,
      "loss": 1.3743,
      "step": 203840
    },
    {
      "epoch": 15.302363977485928,
      "grad_norm": 0.4611794352531433,
      "learning_rate": 0.002828,
      "loss": 1.3796,
      "step": 203904
    },
    {
      "epoch": 15.307166979362101,
      "grad_norm": 0.5892916917800903,
      "learning_rate": 0.002828,
      "loss": 1.3834,
      "step": 203968
    },
    {
      "epoch": 15.311969981238274,
      "grad_norm": 0.5543408989906311,
      "learning_rate": 0.002828,
      "loss": 1.3856,
      "step": 204032
    },
    {
      "epoch": 15.316772983114447,
      "grad_norm": 0.5828499794006348,
      "learning_rate": 0.002828,
      "loss": 1.3767,
      "step": 204096
    },
    {
      "epoch": 15.32157598499062,
      "grad_norm": 0.5696132779121399,
      "learning_rate": 0.002828,
      "loss": 1.3778,
      "step": 204160
    },
    {
      "epoch": 15.326378986866791,
      "grad_norm": 0.5638845562934875,
      "learning_rate": 0.002828,
      "loss": 1.382,
      "step": 204224
    },
    {
      "epoch": 15.331181988742964,
      "grad_norm": 0.49652546644210815,
      "learning_rate": 0.002828,
      "loss": 1.378,
      "step": 204288
    },
    {
      "epoch": 15.335984990619137,
      "grad_norm": 0.6045218706130981,
      "learning_rate": 0.002828,
      "loss": 1.3744,
      "step": 204352
    },
    {
      "epoch": 15.34078799249531,
      "grad_norm": 0.459846556186676,
      "learning_rate": 0.002828,
      "loss": 1.3842,
      "step": 204416
    },
    {
      "epoch": 15.345590994371483,
      "grad_norm": 0.5356844067573547,
      "learning_rate": 0.002828,
      "loss": 1.3824,
      "step": 204480
    },
    {
      "epoch": 15.350393996247655,
      "grad_norm": 0.565377950668335,
      "learning_rate": 0.002828,
      "loss": 1.3852,
      "step": 204544
    },
    {
      "epoch": 15.355196998123827,
      "grad_norm": 0.4540599584579468,
      "learning_rate": 0.002828,
      "loss": 1.3803,
      "step": 204608
    },
    {
      "epoch": 15.36,
      "grad_norm": 0.5405008792877197,
      "learning_rate": 0.002828,
      "loss": 1.3843,
      "step": 204672
    },
    {
      "epoch": 15.364803001876172,
      "grad_norm": 0.5116636753082275,
      "learning_rate": 0.002828,
      "loss": 1.376,
      "step": 204736
    },
    {
      "epoch": 15.369606003752345,
      "grad_norm": 0.5005480647087097,
      "learning_rate": 0.002828,
      "loss": 1.3815,
      "step": 204800
    },
    {
      "epoch": 15.374409005628518,
      "grad_norm": 0.5235862731933594,
      "learning_rate": 0.002828,
      "loss": 1.3831,
      "step": 204864
    },
    {
      "epoch": 15.379212007504691,
      "grad_norm": 0.5118006467819214,
      "learning_rate": 0.002828,
      "loss": 1.381,
      "step": 204928
    },
    {
      "epoch": 15.384015009380864,
      "grad_norm": 0.43683668971061707,
      "learning_rate": 0.002828,
      "loss": 1.3904,
      "step": 204992
    },
    {
      "epoch": 15.388818011257035,
      "grad_norm": 0.49627912044525146,
      "learning_rate": 0.002828,
      "loss": 1.3912,
      "step": 205056
    },
    {
      "epoch": 15.393621013133208,
      "grad_norm": 0.535225510597229,
      "learning_rate": 0.002828,
      "loss": 1.3858,
      "step": 205120
    },
    {
      "epoch": 15.39842401500938,
      "grad_norm": 0.565176248550415,
      "learning_rate": 0.002828,
      "loss": 1.3785,
      "step": 205184
    },
    {
      "epoch": 15.403227016885554,
      "grad_norm": 0.5722188949584961,
      "learning_rate": 0.002828,
      "loss": 1.3792,
      "step": 205248
    },
    {
      "epoch": 15.408030018761727,
      "grad_norm": 0.5208520889282227,
      "learning_rate": 0.002828,
      "loss": 1.3817,
      "step": 205312
    },
    {
      "epoch": 15.4128330206379,
      "grad_norm": 0.5423282980918884,
      "learning_rate": 0.002828,
      "loss": 1.3837,
      "step": 205376
    },
    {
      "epoch": 15.41763602251407,
      "grad_norm": 0.538154125213623,
      "learning_rate": 0.002828,
      "loss": 1.3803,
      "step": 205440
    },
    {
      "epoch": 15.422439024390243,
      "grad_norm": 0.6185964941978455,
      "learning_rate": 0.002828,
      "loss": 1.378,
      "step": 205504
    },
    {
      "epoch": 15.427242026266416,
      "grad_norm": 0.4758653938770294,
      "learning_rate": 0.002828,
      "loss": 1.3789,
      "step": 205568
    },
    {
      "epoch": 15.43204502814259,
      "grad_norm": 0.5673360824584961,
      "learning_rate": 0.002828,
      "loss": 1.3804,
      "step": 205632
    },
    {
      "epoch": 15.436848030018762,
      "grad_norm": 0.5129354000091553,
      "learning_rate": 0.002828,
      "loss": 1.3832,
      "step": 205696
    },
    {
      "epoch": 15.441651031894935,
      "grad_norm": 0.507262110710144,
      "learning_rate": 0.002828,
      "loss": 1.3807,
      "step": 205760
    },
    {
      "epoch": 15.446454033771108,
      "grad_norm": 0.5329039692878723,
      "learning_rate": 0.002828,
      "loss": 1.3827,
      "step": 205824
    },
    {
      "epoch": 15.451257035647279,
      "grad_norm": 0.48899421095848083,
      "learning_rate": 0.002828,
      "loss": 1.3803,
      "step": 205888
    },
    {
      "epoch": 15.456060037523452,
      "grad_norm": 0.5405635833740234,
      "learning_rate": 0.002828,
      "loss": 1.3748,
      "step": 205952
    },
    {
      "epoch": 15.460863039399625,
      "grad_norm": 0.506642758846283,
      "learning_rate": 0.002828,
      "loss": 1.384,
      "step": 206016
    },
    {
      "epoch": 15.465666041275798,
      "grad_norm": 0.4873145818710327,
      "learning_rate": 0.002828,
      "loss": 1.38,
      "step": 206080
    },
    {
      "epoch": 15.47046904315197,
      "grad_norm": 0.5708457231521606,
      "learning_rate": 0.002828,
      "loss": 1.3843,
      "step": 206144
    },
    {
      "epoch": 15.475272045028143,
      "grad_norm": 0.6255208849906921,
      "learning_rate": 0.002828,
      "loss": 1.3762,
      "step": 206208
    },
    {
      "epoch": 15.480075046904314,
      "grad_norm": 0.6195931434631348,
      "learning_rate": 0.002828,
      "loss": 1.383,
      "step": 206272
    },
    {
      "epoch": 15.484878048780487,
      "grad_norm": 0.611381471157074,
      "learning_rate": 0.002828,
      "loss": 1.3834,
      "step": 206336
    },
    {
      "epoch": 15.48968105065666,
      "grad_norm": 0.5173780918121338,
      "learning_rate": 0.002828,
      "loss": 1.3825,
      "step": 206400
    },
    {
      "epoch": 15.494484052532833,
      "grad_norm": 0.5662381052970886,
      "learning_rate": 0.002828,
      "loss": 1.382,
      "step": 206464
    },
    {
      "epoch": 15.499287054409006,
      "grad_norm": 0.6043580174446106,
      "learning_rate": 0.002828,
      "loss": 1.3785,
      "step": 206528
    },
    {
      "epoch": 15.504090056285179,
      "grad_norm": 0.5214434266090393,
      "learning_rate": 0.002828,
      "loss": 1.3803,
      "step": 206592
    },
    {
      "epoch": 15.508893058161352,
      "grad_norm": 0.5894580483436584,
      "learning_rate": 0.002828,
      "loss": 1.3859,
      "step": 206656
    },
    {
      "epoch": 15.513696060037523,
      "grad_norm": 0.5345766544342041,
      "learning_rate": 0.002828,
      "loss": 1.386,
      "step": 206720
    },
    {
      "epoch": 15.518499061913696,
      "grad_norm": 0.5783444046974182,
      "learning_rate": 0.002828,
      "loss": 1.3781,
      "step": 206784
    },
    {
      "epoch": 15.523302063789869,
      "grad_norm": 0.5113909244537354,
      "learning_rate": 0.002828,
      "loss": 1.3815,
      "step": 206848
    },
    {
      "epoch": 15.528105065666042,
      "grad_norm": 0.6762928366661072,
      "learning_rate": 0.002828,
      "loss": 1.3899,
      "step": 206912
    },
    {
      "epoch": 15.532908067542214,
      "grad_norm": 0.5763272047042847,
      "learning_rate": 0.002828,
      "loss": 1.3844,
      "step": 206976
    },
    {
      "epoch": 15.537711069418387,
      "grad_norm": 0.4625857174396515,
      "learning_rate": 0.002828,
      "loss": 1.3819,
      "step": 207040
    },
    {
      "epoch": 15.542514071294558,
      "grad_norm": 0.5895203948020935,
      "learning_rate": 0.002828,
      "loss": 1.3839,
      "step": 207104
    },
    {
      "epoch": 15.547317073170731,
      "grad_norm": 0.5495424270629883,
      "learning_rate": 0.002828,
      "loss": 1.3886,
      "step": 207168
    },
    {
      "epoch": 15.552120075046904,
      "grad_norm": 0.5420121550559998,
      "learning_rate": 0.002828,
      "loss": 1.3844,
      "step": 207232
    },
    {
      "epoch": 15.556923076923077,
      "grad_norm": 0.585555374622345,
      "learning_rate": 0.002828,
      "loss": 1.3834,
      "step": 207296
    },
    {
      "epoch": 15.56172607879925,
      "grad_norm": 0.5260416269302368,
      "learning_rate": 0.002828,
      "loss": 1.3823,
      "step": 207360
    },
    {
      "epoch": 15.566529080675423,
      "grad_norm": 0.7156786322593689,
      "learning_rate": 0.002828,
      "loss": 1.3841,
      "step": 207424
    },
    {
      "epoch": 15.571332082551594,
      "grad_norm": 0.4893474578857422,
      "learning_rate": 0.002828,
      "loss": 1.38,
      "step": 207488
    },
    {
      "epoch": 15.576135084427767,
      "grad_norm": 0.5773012638092041,
      "learning_rate": 0.002828,
      "loss": 1.3887,
      "step": 207552
    },
    {
      "epoch": 15.58093808630394,
      "grad_norm": 0.5502187013626099,
      "learning_rate": 0.002828,
      "loss": 1.3877,
      "step": 207616
    },
    {
      "epoch": 15.585741088180113,
      "grad_norm": 0.4841638505458832,
      "learning_rate": 0.002828,
      "loss": 1.3835,
      "step": 207680
    },
    {
      "epoch": 15.590544090056285,
      "grad_norm": 0.6301225423812866,
      "learning_rate": 0.002828,
      "loss": 1.3782,
      "step": 207744
    },
    {
      "epoch": 15.595347091932458,
      "grad_norm": 0.5887924432754517,
      "learning_rate": 0.002828,
      "loss": 1.3792,
      "step": 207808
    },
    {
      "epoch": 15.60015009380863,
      "grad_norm": 0.5262457728385925,
      "learning_rate": 0.002828,
      "loss": 1.3898,
      "step": 207872
    },
    {
      "epoch": 15.604953095684802,
      "grad_norm": 0.5991269946098328,
      "learning_rate": 0.002828,
      "loss": 1.3809,
      "step": 207936
    },
    {
      "epoch": 15.609756097560975,
      "grad_norm": 0.556826651096344,
      "learning_rate": 0.002828,
      "loss": 1.3833,
      "step": 208000
    },
    {
      "epoch": 15.614559099437148,
      "grad_norm": 0.6832888722419739,
      "learning_rate": 0.002828,
      "loss": 1.388,
      "step": 208064
    },
    {
      "epoch": 15.619362101313321,
      "grad_norm": 0.5342592000961304,
      "learning_rate": 0.002828,
      "loss": 1.3851,
      "step": 208128
    },
    {
      "epoch": 15.624165103189494,
      "grad_norm": 0.5431777834892273,
      "learning_rate": 0.002828,
      "loss": 1.3914,
      "step": 208192
    },
    {
      "epoch": 15.628968105065667,
      "grad_norm": 0.5456152558326721,
      "learning_rate": 0.002828,
      "loss": 1.3799,
      "step": 208256
    },
    {
      "epoch": 15.633771106941838,
      "grad_norm": 0.5626450181007385,
      "learning_rate": 0.002828,
      "loss": 1.386,
      "step": 208320
    },
    {
      "epoch": 15.63857410881801,
      "grad_norm": 0.5906189680099487,
      "learning_rate": 0.002828,
      "loss": 1.3872,
      "step": 208384
    },
    {
      "epoch": 15.643377110694184,
      "grad_norm": 0.6024333834648132,
      "learning_rate": 0.002828,
      "loss": 1.39,
      "step": 208448
    },
    {
      "epoch": 15.648180112570357,
      "grad_norm": 0.5341130495071411,
      "learning_rate": 0.002828,
      "loss": 1.3866,
      "step": 208512
    },
    {
      "epoch": 15.65298311444653,
      "grad_norm": 0.6072880625724792,
      "learning_rate": 0.002828,
      "loss": 1.3868,
      "step": 208576
    },
    {
      "epoch": 15.657786116322702,
      "grad_norm": 0.5600910782814026,
      "learning_rate": 0.002828,
      "loss": 1.3864,
      "step": 208640
    },
    {
      "epoch": 15.662589118198873,
      "grad_norm": 0.502774178981781,
      "learning_rate": 0.002828,
      "loss": 1.3875,
      "step": 208704
    },
    {
      "epoch": 15.667392120075046,
      "grad_norm": 0.6341469883918762,
      "learning_rate": 0.002828,
      "loss": 1.3903,
      "step": 208768
    },
    {
      "epoch": 15.67219512195122,
      "grad_norm": 0.5548173189163208,
      "learning_rate": 0.002828,
      "loss": 1.3833,
      "step": 208832
    },
    {
      "epoch": 15.676998123827392,
      "grad_norm": 0.4528850018978119,
      "learning_rate": 0.002828,
      "loss": 1.3881,
      "step": 208896
    },
    {
      "epoch": 15.681801125703565,
      "grad_norm": 0.6027002930641174,
      "learning_rate": 0.002828,
      "loss": 1.3802,
      "step": 208960
    },
    {
      "epoch": 15.686604127579738,
      "grad_norm": 0.5031800866127014,
      "learning_rate": 0.002828,
      "loss": 1.3938,
      "step": 209024
    },
    {
      "epoch": 15.69140712945591,
      "grad_norm": 0.650576114654541,
      "learning_rate": 0.002828,
      "loss": 1.3892,
      "step": 209088
    },
    {
      "epoch": 15.696210131332082,
      "grad_norm": 0.5973570346832275,
      "learning_rate": 0.002828,
      "loss": 1.386,
      "step": 209152
    },
    {
      "epoch": 15.701013133208255,
      "grad_norm": 0.5872775316238403,
      "learning_rate": 0.002828,
      "loss": 1.3855,
      "step": 209216
    },
    {
      "epoch": 15.705816135084428,
      "grad_norm": 0.6111817955970764,
      "learning_rate": 0.002828,
      "loss": 1.3825,
      "step": 209280
    },
    {
      "epoch": 15.7106191369606,
      "grad_norm": 0.5026399493217468,
      "learning_rate": 0.002828,
      "loss": 1.3855,
      "step": 209344
    },
    {
      "epoch": 15.715422138836773,
      "grad_norm": 0.5055294036865234,
      "learning_rate": 0.002828,
      "loss": 1.3843,
      "step": 209408
    },
    {
      "epoch": 15.720225140712946,
      "grad_norm": 0.510923445224762,
      "learning_rate": 0.002828,
      "loss": 1.3939,
      "step": 209472
    },
    {
      "epoch": 15.725028142589117,
      "grad_norm": 0.4886623024940491,
      "learning_rate": 0.002828,
      "loss": 1.3862,
      "step": 209536
    },
    {
      "epoch": 15.72983114446529,
      "grad_norm": 0.4989912211894989,
      "learning_rate": 0.002828,
      "loss": 1.375,
      "step": 209600
    },
    {
      "epoch": 15.734634146341463,
      "grad_norm": 0.5838048458099365,
      "learning_rate": 0.002828,
      "loss": 1.3845,
      "step": 209664
    },
    {
      "epoch": 15.739437148217636,
      "grad_norm": 0.5743147730827332,
      "learning_rate": 0.002828,
      "loss": 1.3854,
      "step": 209728
    },
    {
      "epoch": 15.744240150093809,
      "grad_norm": 0.5653936862945557,
      "learning_rate": 0.002828,
      "loss": 1.3794,
      "step": 209792
    },
    {
      "epoch": 15.749043151969982,
      "grad_norm": 0.5322737097740173,
      "learning_rate": 0.002828,
      "loss": 1.3866,
      "step": 209856
    },
    {
      "epoch": 15.753846153846155,
      "grad_norm": 0.5013566017150879,
      "learning_rate": 0.002828,
      "loss": 1.3853,
      "step": 209920
    },
    {
      "epoch": 15.758649155722326,
      "grad_norm": 0.5475387573242188,
      "learning_rate": 0.002828,
      "loss": 1.3605,
      "step": 209984
    },
    {
      "epoch": 15.763452157598499,
      "grad_norm": 0.5118324160575867,
      "learning_rate": 0.002828,
      "loss": 1.3574,
      "step": 210048
    },
    {
      "epoch": 15.768255159474672,
      "grad_norm": 0.5450679659843445,
      "learning_rate": 0.002828,
      "loss": 1.3605,
      "step": 210112
    },
    {
      "epoch": 15.773058161350844,
      "grad_norm": 0.5049896240234375,
      "learning_rate": 0.002828,
      "loss": 1.3572,
      "step": 210176
    },
    {
      "epoch": 15.777861163227017,
      "grad_norm": 0.6178387403488159,
      "learning_rate": 0.002828,
      "loss": 1.3619,
      "step": 210240
    },
    {
      "epoch": 15.78266416510319,
      "grad_norm": 0.5072968602180481,
      "learning_rate": 0.002828,
      "loss": 1.363,
      "step": 210304
    },
    {
      "epoch": 15.787467166979361,
      "grad_norm": 0.5722081661224365,
      "learning_rate": 0.002828,
      "loss": 1.3542,
      "step": 210368
    },
    {
      "epoch": 15.792270168855534,
      "grad_norm": 0.5220540165901184,
      "learning_rate": 0.002828,
      "loss": 1.365,
      "step": 210432
    },
    {
      "epoch": 15.797073170731707,
      "grad_norm": 0.4980974793434143,
      "learning_rate": 0.002828,
      "loss": 1.3593,
      "step": 210496
    },
    {
      "epoch": 15.80187617260788,
      "grad_norm": 0.5043259263038635,
      "learning_rate": 0.002828,
      "loss": 1.359,
      "step": 210560
    },
    {
      "epoch": 15.806679174484053,
      "grad_norm": 0.5757930278778076,
      "learning_rate": 0.002828,
      "loss": 1.3663,
      "step": 210624
    },
    {
      "epoch": 15.811482176360226,
      "grad_norm": 0.5797905921936035,
      "learning_rate": 0.002828,
      "loss": 1.3611,
      "step": 210688
    },
    {
      "epoch": 15.816285178236399,
      "grad_norm": 0.535588800907135,
      "learning_rate": 0.002828,
      "loss": 1.3659,
      "step": 210752
    },
    {
      "epoch": 15.82108818011257,
      "grad_norm": 0.5623985528945923,
      "learning_rate": 0.002828,
      "loss": 1.3637,
      "step": 210816
    },
    {
      "epoch": 15.825891181988743,
      "grad_norm": 0.587024986743927,
      "learning_rate": 0.002828,
      "loss": 1.366,
      "step": 210880
    },
    {
      "epoch": 15.830694183864916,
      "grad_norm": 0.5084807276725769,
      "learning_rate": 0.002828,
      "loss": 1.3615,
      "step": 210944
    },
    {
      "epoch": 15.835497185741088,
      "grad_norm": 0.5876424908638,
      "learning_rate": 0.002828,
      "loss": 1.3655,
      "step": 211008
    },
    {
      "epoch": 15.840300187617261,
      "grad_norm": 0.5620876550674438,
      "learning_rate": 0.002828,
      "loss": 1.3646,
      "step": 211072
    },
    {
      "epoch": 15.845103189493434,
      "grad_norm": 0.4719368815422058,
      "learning_rate": 0.002828,
      "loss": 1.3611,
      "step": 211136
    },
    {
      "epoch": 15.849906191369605,
      "grad_norm": 0.5494318008422852,
      "learning_rate": 0.002828,
      "loss": 1.368,
      "step": 211200
    },
    {
      "epoch": 15.854709193245778,
      "grad_norm": 0.4704643785953522,
      "learning_rate": 0.002828,
      "loss": 1.3666,
      "step": 211264
    },
    {
      "epoch": 15.859512195121951,
      "grad_norm": 0.588175892829895,
      "learning_rate": 0.002828,
      "loss": 1.3696,
      "step": 211328
    },
    {
      "epoch": 15.864315196998124,
      "grad_norm": 0.6054316759109497,
      "learning_rate": 0.002828,
      "loss": 1.3661,
      "step": 211392
    },
    {
      "epoch": 15.869118198874297,
      "grad_norm": 0.5715833902359009,
      "learning_rate": 0.002828,
      "loss": 1.3627,
      "step": 211456
    },
    {
      "epoch": 15.87392120075047,
      "grad_norm": 0.5927461981773376,
      "learning_rate": 0.002828,
      "loss": 1.3678,
      "step": 211520
    },
    {
      "epoch": 15.87872420262664,
      "grad_norm": 0.6130070090293884,
      "learning_rate": 0.002828,
      "loss": 1.3689,
      "step": 211584
    },
    {
      "epoch": 15.883527204502814,
      "grad_norm": 0.5040857791900635,
      "learning_rate": 0.002828,
      "loss": 1.3591,
      "step": 211648
    },
    {
      "epoch": 15.888330206378987,
      "grad_norm": 0.5852608680725098,
      "learning_rate": 0.002828,
      "loss": 1.3669,
      "step": 211712
    },
    {
      "epoch": 15.89313320825516,
      "grad_norm": 0.5842130184173584,
      "learning_rate": 0.002828,
      "loss": 1.3697,
      "step": 211776
    },
    {
      "epoch": 15.897936210131332,
      "grad_norm": 0.5647704005241394,
      "learning_rate": 0.002828,
      "loss": 1.3679,
      "step": 211840
    },
    {
      "epoch": 15.902739212007505,
      "grad_norm": 0.5339858531951904,
      "learning_rate": 0.002828,
      "loss": 1.3733,
      "step": 211904
    },
    {
      "epoch": 15.907542213883676,
      "grad_norm": 0.6253660321235657,
      "learning_rate": 0.002828,
      "loss": 1.3775,
      "step": 211968
    },
    {
      "epoch": 15.91234521575985,
      "grad_norm": 0.5627161860466003,
      "learning_rate": 0.002828,
      "loss": 1.363,
      "step": 212032
    },
    {
      "epoch": 15.917148217636022,
      "grad_norm": 0.508140504360199,
      "learning_rate": 0.002828,
      "loss": 1.3695,
      "step": 212096
    },
    {
      "epoch": 15.921951219512195,
      "grad_norm": 0.5700469017028809,
      "learning_rate": 0.002828,
      "loss": 1.3674,
      "step": 212160
    },
    {
      "epoch": 15.926754221388368,
      "grad_norm": 0.6160951256752014,
      "learning_rate": 0.002828,
      "loss": 1.3687,
      "step": 212224
    },
    {
      "epoch": 15.93155722326454,
      "grad_norm": 0.4808539152145386,
      "learning_rate": 0.002828,
      "loss": 1.3681,
      "step": 212288
    },
    {
      "epoch": 15.936360225140714,
      "grad_norm": 0.5854923129081726,
      "learning_rate": 0.002828,
      "loss": 1.3682,
      "step": 212352
    },
    {
      "epoch": 15.941163227016885,
      "grad_norm": 0.4344835877418518,
      "learning_rate": 0.002828,
      "loss": 1.3648,
      "step": 212416
    },
    {
      "epoch": 15.945966228893058,
      "grad_norm": 0.6398991346359253,
      "learning_rate": 0.002828,
      "loss": 1.3665,
      "step": 212480
    },
    {
      "epoch": 15.95076923076923,
      "grad_norm": 0.5153722763061523,
      "learning_rate": 0.002828,
      "loss": 1.3748,
      "step": 212544
    },
    {
      "epoch": 15.955572232645403,
      "grad_norm": 0.6487797498703003,
      "learning_rate": 0.002828,
      "loss": 1.3728,
      "step": 212608
    },
    {
      "epoch": 15.960375234521576,
      "grad_norm": 0.6281577348709106,
      "learning_rate": 0.002828,
      "loss": 1.3655,
      "step": 212672
    },
    {
      "epoch": 15.96517823639775,
      "grad_norm": 0.6675734519958496,
      "learning_rate": 0.002828,
      "loss": 1.3732,
      "step": 212736
    },
    {
      "epoch": 15.96998123827392,
      "grad_norm": 0.46980762481689453,
      "learning_rate": 0.002828,
      "loss": 1.3667,
      "step": 212800
    },
    {
      "epoch": 15.974784240150093,
      "grad_norm": 0.5400161147117615,
      "learning_rate": 0.002828,
      "loss": 1.3732,
      "step": 212864
    },
    {
      "epoch": 15.979587242026266,
      "grad_norm": 0.5441579222679138,
      "learning_rate": 0.002828,
      "loss": 1.3732,
      "step": 212928
    },
    {
      "epoch": 15.984390243902439,
      "grad_norm": 0.48642051219940186,
      "learning_rate": 0.002828,
      "loss": 1.3668,
      "step": 212992
    },
    {
      "epoch": 15.989193245778612,
      "grad_norm": 0.4595506489276886,
      "learning_rate": 0.002828,
      "loss": 1.3701,
      "step": 213056
    },
    {
      "epoch": 15.993996247654785,
      "grad_norm": 0.5701270699501038,
      "learning_rate": 0.002828,
      "loss": 1.3703,
      "step": 213120
    },
    {
      "epoch": 15.998799249530958,
      "grad_norm": 0.48854881525039673,
      "learning_rate": 0.002828,
      "loss": 1.3723,
      "step": 213184
    },
    {
      "epoch": 16.00360225140713,
      "grad_norm": 0.6080936193466187,
      "learning_rate": 0.002828,
      "loss": 1.3561,
      "step": 213248
    },
    {
      "epoch": 16.008405253283303,
      "grad_norm": 0.5640507936477661,
      "learning_rate": 0.002828,
      "loss": 1.3516,
      "step": 213312
    },
    {
      "epoch": 16.013208255159476,
      "grad_norm": 0.5358575582504272,
      "learning_rate": 0.002828,
      "loss": 1.3571,
      "step": 213376
    },
    {
      "epoch": 16.018011257035646,
      "grad_norm": 0.5076690316200256,
      "learning_rate": 0.002828,
      "loss": 1.3547,
      "step": 213440
    },
    {
      "epoch": 16.02281425891182,
      "grad_norm": 0.5925807356834412,
      "learning_rate": 0.002828,
      "loss": 1.3598,
      "step": 213504
    },
    {
      "epoch": 16.02761726078799,
      "grad_norm": 0.5533409714698792,
      "learning_rate": 0.002828,
      "loss": 1.3533,
      "step": 213568
    },
    {
      "epoch": 16.032420262664164,
      "grad_norm": 0.507273256778717,
      "learning_rate": 0.002828,
      "loss": 1.3521,
      "step": 213632
    },
    {
      "epoch": 16.037223264540337,
      "grad_norm": 0.5224199891090393,
      "learning_rate": 0.002828,
      "loss": 1.3587,
      "step": 213696
    },
    {
      "epoch": 16.04202626641651,
      "grad_norm": 0.5440003275871277,
      "learning_rate": 0.002828,
      "loss": 1.354,
      "step": 213760
    },
    {
      "epoch": 16.046829268292683,
      "grad_norm": 0.5843242406845093,
      "learning_rate": 0.002828,
      "loss": 1.364,
      "step": 213824
    },
    {
      "epoch": 16.051632270168856,
      "grad_norm": 0.5096895694732666,
      "learning_rate": 0.002828,
      "loss": 1.3611,
      "step": 213888
    },
    {
      "epoch": 16.05643527204503,
      "grad_norm": 0.6344131827354431,
      "learning_rate": 0.002828,
      "loss": 1.3617,
      "step": 213952
    },
    {
      "epoch": 16.0612382739212,
      "grad_norm": 0.6061226725578308,
      "learning_rate": 0.002828,
      "loss": 1.3586,
      "step": 214016
    },
    {
      "epoch": 16.066041275797375,
      "grad_norm": 0.5750422477722168,
      "learning_rate": 0.002828,
      "loss": 1.3576,
      "step": 214080
    },
    {
      "epoch": 16.070844277673547,
      "grad_norm": 0.495776891708374,
      "learning_rate": 0.002828,
      "loss": 1.3573,
      "step": 214144
    },
    {
      "epoch": 16.07564727954972,
      "grad_norm": 0.4955301582813263,
      "learning_rate": 0.002828,
      "loss": 1.3589,
      "step": 214208
    },
    {
      "epoch": 16.08045028142589,
      "grad_norm": 0.6393030881881714,
      "learning_rate": 0.002828,
      "loss": 1.3622,
      "step": 214272
    },
    {
      "epoch": 16.085253283302062,
      "grad_norm": 0.5954785346984863,
      "learning_rate": 0.002828,
      "loss": 1.3577,
      "step": 214336
    },
    {
      "epoch": 16.090056285178235,
      "grad_norm": 0.5928305983543396,
      "learning_rate": 0.002828,
      "loss": 1.3551,
      "step": 214400
    },
    {
      "epoch": 16.09485928705441,
      "grad_norm": 0.5459696650505066,
      "learning_rate": 0.002828,
      "loss": 1.3643,
      "step": 214464
    },
    {
      "epoch": 16.09966228893058,
      "grad_norm": 0.7050192952156067,
      "learning_rate": 0.002828,
      "loss": 1.3586,
      "step": 214528
    },
    {
      "epoch": 16.104465290806754,
      "grad_norm": 0.5199483633041382,
      "learning_rate": 0.002828,
      "loss": 1.3611,
      "step": 214592
    },
    {
      "epoch": 16.109268292682927,
      "grad_norm": 0.4743548333644867,
      "learning_rate": 0.002828,
      "loss": 1.3612,
      "step": 214656
    },
    {
      "epoch": 16.1140712945591,
      "grad_norm": 0.5115127563476562,
      "learning_rate": 0.002828,
      "loss": 1.3634,
      "step": 214720
    },
    {
      "epoch": 16.118874296435273,
      "grad_norm": 0.5872589945793152,
      "learning_rate": 0.002828,
      "loss": 1.3626,
      "step": 214784
    },
    {
      "epoch": 16.123677298311446,
      "grad_norm": 0.5724947452545166,
      "learning_rate": 0.002828,
      "loss": 1.3645,
      "step": 214848
    },
    {
      "epoch": 16.12848030018762,
      "grad_norm": 0.6491893529891968,
      "learning_rate": 0.002828,
      "loss": 1.3583,
      "step": 214912
    },
    {
      "epoch": 16.13328330206379,
      "grad_norm": 0.579578697681427,
      "learning_rate": 0.002828,
      "loss": 1.3671,
      "step": 214976
    },
    {
      "epoch": 16.138086303939964,
      "grad_norm": 0.5547657608985901,
      "learning_rate": 0.002828,
      "loss": 1.3614,
      "step": 215040
    },
    {
      "epoch": 16.142889305816134,
      "grad_norm": 0.6079813241958618,
      "learning_rate": 0.002828,
      "loss": 1.3577,
      "step": 215104
    },
    {
      "epoch": 16.147692307692306,
      "grad_norm": 0.6912538409233093,
      "learning_rate": 0.002828,
      "loss": 1.3616,
      "step": 215168
    },
    {
      "epoch": 16.15249530956848,
      "grad_norm": 0.5580403208732605,
      "learning_rate": 0.002828,
      "loss": 1.3637,
      "step": 215232
    },
    {
      "epoch": 16.157298311444652,
      "grad_norm": 0.6073527932167053,
      "learning_rate": 0.002828,
      "loss": 1.3685,
      "step": 215296
    },
    {
      "epoch": 16.162101313320825,
      "grad_norm": 0.5496214032173157,
      "learning_rate": 0.002828,
      "loss": 1.3663,
      "step": 215360
    },
    {
      "epoch": 16.166904315196998,
      "grad_norm": 0.47354716062545776,
      "learning_rate": 0.002828,
      "loss": 1.3676,
      "step": 215424
    },
    {
      "epoch": 16.17170731707317,
      "grad_norm": 0.607663094997406,
      "learning_rate": 0.002828,
      "loss": 1.3666,
      "step": 215488
    },
    {
      "epoch": 16.176510318949344,
      "grad_norm": 0.5812382102012634,
      "learning_rate": 0.002828,
      "loss": 1.3652,
      "step": 215552
    },
    {
      "epoch": 16.181313320825517,
      "grad_norm": 0.713192880153656,
      "learning_rate": 0.002828,
      "loss": 1.3691,
      "step": 215616
    },
    {
      "epoch": 16.18611632270169,
      "grad_norm": 0.5395285487174988,
      "learning_rate": 0.002828,
      "loss": 1.3642,
      "step": 215680
    },
    {
      "epoch": 16.190919324577862,
      "grad_norm": 0.5327445268630981,
      "learning_rate": 0.002828,
      "loss": 1.3648,
      "step": 215744
    },
    {
      "epoch": 16.195722326454035,
      "grad_norm": 0.6013232469558716,
      "learning_rate": 0.002828,
      "loss": 1.3677,
      "step": 215808
    },
    {
      "epoch": 16.200525328330205,
      "grad_norm": 0.5524231195449829,
      "learning_rate": 0.002828,
      "loss": 1.3631,
      "step": 215872
    },
    {
      "epoch": 16.205328330206378,
      "grad_norm": 0.4762788712978363,
      "learning_rate": 0.002828,
      "loss": 1.3605,
      "step": 215936
    },
    {
      "epoch": 16.21013133208255,
      "grad_norm": 0.4953647553920746,
      "learning_rate": 0.002828,
      "loss": 1.3656,
      "step": 216000
    },
    {
      "epoch": 16.214934333958723,
      "grad_norm": 0.5233112573623657,
      "learning_rate": 0.002828,
      "loss": 1.3685,
      "step": 216064
    },
    {
      "epoch": 16.219737335834896,
      "grad_norm": 0.5747987031936646,
      "learning_rate": 0.002828,
      "loss": 1.3609,
      "step": 216128
    },
    {
      "epoch": 16.22454033771107,
      "grad_norm": 0.5410612225532532,
      "learning_rate": 0.002828,
      "loss": 1.364,
      "step": 216192
    },
    {
      "epoch": 16.229343339587242,
      "grad_norm": 0.5992555022239685,
      "learning_rate": 0.002828,
      "loss": 1.3719,
      "step": 216256
    },
    {
      "epoch": 16.234146341463415,
      "grad_norm": 0.6096734404563904,
      "learning_rate": 0.002828,
      "loss": 1.3628,
      "step": 216320
    },
    {
      "epoch": 16.238949343339588,
      "grad_norm": 0.5461161732673645,
      "learning_rate": 0.002828,
      "loss": 1.3621,
      "step": 216384
    },
    {
      "epoch": 16.24375234521576,
      "grad_norm": 0.552280068397522,
      "learning_rate": 0.002828,
      "loss": 1.3694,
      "step": 216448
    },
    {
      "epoch": 16.248555347091933,
      "grad_norm": 0.6835575103759766,
      "learning_rate": 0.002828,
      "loss": 1.3676,
      "step": 216512
    },
    {
      "epoch": 16.253358348968106,
      "grad_norm": 0.5092332363128662,
      "learning_rate": 0.002828,
      "loss": 1.3701,
      "step": 216576
    },
    {
      "epoch": 16.25816135084428,
      "grad_norm": 0.582518458366394,
      "learning_rate": 0.002828,
      "loss": 1.3665,
      "step": 216640
    },
    {
      "epoch": 16.26296435272045,
      "grad_norm": 0.5306048393249512,
      "learning_rate": 0.002828,
      "loss": 1.3657,
      "step": 216704
    },
    {
      "epoch": 16.26776735459662,
      "grad_norm": 0.6176237463951111,
      "learning_rate": 0.002828,
      "loss": 1.3717,
      "step": 216768
    },
    {
      "epoch": 16.272570356472794,
      "grad_norm": 0.5232059359550476,
      "learning_rate": 0.002828,
      "loss": 1.3675,
      "step": 216832
    },
    {
      "epoch": 16.277373358348967,
      "grad_norm": 0.6212632060050964,
      "learning_rate": 0.002828,
      "loss": 1.3663,
      "step": 216896
    },
    {
      "epoch": 16.28217636022514,
      "grad_norm": 0.4637361168861389,
      "learning_rate": 0.002828,
      "loss": 1.3652,
      "step": 216960
    },
    {
      "epoch": 16.286979362101313,
      "grad_norm": 0.5681953430175781,
      "learning_rate": 0.002828,
      "loss": 1.3757,
      "step": 217024
    },
    {
      "epoch": 16.291782363977486,
      "grad_norm": 0.5358713865280151,
      "learning_rate": 0.002828,
      "loss": 1.3647,
      "step": 217088
    },
    {
      "epoch": 16.29658536585366,
      "grad_norm": 0.5948871970176697,
      "learning_rate": 0.002828,
      "loss": 1.365,
      "step": 217152
    },
    {
      "epoch": 16.30138836772983,
      "grad_norm": 0.553391695022583,
      "learning_rate": 0.002828,
      "loss": 1.3698,
      "step": 217216
    },
    {
      "epoch": 16.306191369606005,
      "grad_norm": 0.6753780245780945,
      "learning_rate": 0.002828,
      "loss": 1.3693,
      "step": 217280
    },
    {
      "epoch": 16.310994371482177,
      "grad_norm": 0.5137534737586975,
      "learning_rate": 0.002828,
      "loss": 1.3696,
      "step": 217344
    },
    {
      "epoch": 16.31579737335835,
      "grad_norm": 0.5785669684410095,
      "learning_rate": 0.002828,
      "loss": 1.3699,
      "step": 217408
    },
    {
      "epoch": 16.320600375234523,
      "grad_norm": 0.45743459463119507,
      "learning_rate": 0.002828,
      "loss": 1.3691,
      "step": 217472
    },
    {
      "epoch": 16.325403377110693,
      "grad_norm": 0.6033856272697449,
      "learning_rate": 0.002828,
      "loss": 1.3697,
      "step": 217536
    },
    {
      "epoch": 16.330206378986865,
      "grad_norm": 0.5758998394012451,
      "learning_rate": 0.002828,
      "loss": 1.3687,
      "step": 217600
    },
    {
      "epoch": 16.33500938086304,
      "grad_norm": 0.5502091646194458,
      "learning_rate": 0.002828,
      "loss": 1.3664,
      "step": 217664
    },
    {
      "epoch": 16.33981238273921,
      "grad_norm": 0.530497670173645,
      "learning_rate": 0.002828,
      "loss": 1.3658,
      "step": 217728
    },
    {
      "epoch": 16.344615384615384,
      "grad_norm": 0.5255142450332642,
      "learning_rate": 0.002828,
      "loss": 1.3703,
      "step": 217792
    },
    {
      "epoch": 16.349418386491557,
      "grad_norm": 0.49928170442581177,
      "learning_rate": 0.002828,
      "loss": 1.3662,
      "step": 217856
    },
    {
      "epoch": 16.35422138836773,
      "grad_norm": 0.6465878486633301,
      "learning_rate": 0.002828,
      "loss": 1.3705,
      "step": 217920
    },
    {
      "epoch": 16.359024390243903,
      "grad_norm": 0.5199236869812012,
      "learning_rate": 0.002828,
      "loss": 1.3671,
      "step": 217984
    },
    {
      "epoch": 16.363827392120076,
      "grad_norm": 0.4599873721599579,
      "learning_rate": 0.002828,
      "loss": 1.3764,
      "step": 218048
    },
    {
      "epoch": 16.36863039399625,
      "grad_norm": 0.5065380930900574,
      "learning_rate": 0.002828,
      "loss": 1.3682,
      "step": 218112
    },
    {
      "epoch": 16.37343339587242,
      "grad_norm": 0.5988659262657166,
      "learning_rate": 0.002828,
      "loss": 1.3698,
      "step": 218176
    },
    {
      "epoch": 16.378236397748594,
      "grad_norm": 0.6505783796310425,
      "learning_rate": 0.002828,
      "loss": 1.366,
      "step": 218240
    },
    {
      "epoch": 16.383039399624767,
      "grad_norm": 0.5402714014053345,
      "learning_rate": 0.002828,
      "loss": 1.3703,
      "step": 218304
    },
    {
      "epoch": 16.387842401500937,
      "grad_norm": 0.5108478665351868,
      "learning_rate": 0.002828,
      "loss": 1.368,
      "step": 218368
    },
    {
      "epoch": 16.39264540337711,
      "grad_norm": 0.5588182210922241,
      "learning_rate": 0.002828,
      "loss": 1.3725,
      "step": 218432
    },
    {
      "epoch": 16.397448405253282,
      "grad_norm": 0.5530375838279724,
      "learning_rate": 0.002828,
      "loss": 1.3701,
      "step": 218496
    },
    {
      "epoch": 16.402251407129455,
      "grad_norm": 0.5329497456550598,
      "learning_rate": 0.002828,
      "loss": 1.3756,
      "step": 218560
    },
    {
      "epoch": 16.407054409005628,
      "grad_norm": 0.6167603731155396,
      "learning_rate": 0.002828,
      "loss": 1.3802,
      "step": 218624
    },
    {
      "epoch": 16.4118574108818,
      "grad_norm": 0.5216876268386841,
      "learning_rate": 0.002828,
      "loss": 1.3763,
      "step": 218688
    },
    {
      "epoch": 16.416660412757974,
      "grad_norm": 0.5277157425880432,
      "learning_rate": 0.002828,
      "loss": 1.3644,
      "step": 218752
    },
    {
      "epoch": 16.421463414634147,
      "grad_norm": 0.5166988372802734,
      "learning_rate": 0.002828,
      "loss": 1.3719,
      "step": 218816
    },
    {
      "epoch": 16.42626641651032,
      "grad_norm": 0.5149502754211426,
      "learning_rate": 0.002828,
      "loss": 1.3742,
      "step": 218880
    },
    {
      "epoch": 16.431069418386492,
      "grad_norm": 0.48137810826301575,
      "learning_rate": 0.002828,
      "loss": 1.3733,
      "step": 218944
    },
    {
      "epoch": 16.435872420262665,
      "grad_norm": 0.5344496369361877,
      "learning_rate": 0.002828,
      "loss": 1.3716,
      "step": 219008
    },
    {
      "epoch": 16.44067542213884,
      "grad_norm": 0.5200846791267395,
      "learning_rate": 0.002828,
      "loss": 1.3789,
      "step": 219072
    },
    {
      "epoch": 16.445478424015008,
      "grad_norm": 0.6055989861488342,
      "learning_rate": 0.002828,
      "loss": 1.3712,
      "step": 219136
    },
    {
      "epoch": 16.45028142589118,
      "grad_norm": 0.5656676888465881,
      "learning_rate": 0.002828,
      "loss": 1.369,
      "step": 219200
    },
    {
      "epoch": 16.455084427767353,
      "grad_norm": 0.5064542293548584,
      "learning_rate": 0.002828,
      "loss": 1.3709,
      "step": 219264
    },
    {
      "epoch": 16.459887429643526,
      "grad_norm": 0.49400073289871216,
      "learning_rate": 0.002828,
      "loss": 1.3781,
      "step": 219328
    },
    {
      "epoch": 16.4646904315197,
      "grad_norm": 0.5486621260643005,
      "learning_rate": 0.002828,
      "loss": 1.3748,
      "step": 219392
    },
    {
      "epoch": 16.469493433395872,
      "grad_norm": 0.5221492052078247,
      "learning_rate": 0.002828,
      "loss": 1.3722,
      "step": 219456
    },
    {
      "epoch": 16.474296435272045,
      "grad_norm": 0.6707462668418884,
      "learning_rate": 0.002828,
      "loss": 1.3674,
      "step": 219520
    },
    {
      "epoch": 16.479099437148218,
      "grad_norm": 0.5669146776199341,
      "learning_rate": 0.002828,
      "loss": 1.3685,
      "step": 219584
    },
    {
      "epoch": 16.48390243902439,
      "grad_norm": 0.5900082588195801,
      "learning_rate": 0.002828,
      "loss": 1.3724,
      "step": 219648
    },
    {
      "epoch": 16.488705440900564,
      "grad_norm": 0.4715104103088379,
      "learning_rate": 0.002828,
      "loss": 1.3763,
      "step": 219712
    },
    {
      "epoch": 16.493508442776736,
      "grad_norm": 0.6620226502418518,
      "learning_rate": 0.002828,
      "loss": 1.3711,
      "step": 219776
    },
    {
      "epoch": 16.49831144465291,
      "grad_norm": 0.5841343402862549,
      "learning_rate": 0.002828,
      "loss": 1.374,
      "step": 219840
    },
    {
      "epoch": 16.503114446529082,
      "grad_norm": 0.5079573392868042,
      "learning_rate": 0.002828,
      "loss": 1.3735,
      "step": 219904
    },
    {
      "epoch": 16.50791744840525,
      "grad_norm": 0.6196205019950867,
      "learning_rate": 0.002828,
      "loss": 1.3734,
      "step": 219968
    },
    {
      "epoch": 16.512720450281424,
      "grad_norm": 0.5488569736480713,
      "learning_rate": 0.002828,
      "loss": 1.377,
      "step": 220032
    },
    {
      "epoch": 16.517523452157597,
      "grad_norm": 0.5166944861412048,
      "learning_rate": 0.002828,
      "loss": 1.3781,
      "step": 220096
    },
    {
      "epoch": 16.52232645403377,
      "grad_norm": 0.5003791451454163,
      "learning_rate": 0.002828,
      "loss": 1.3772,
      "step": 220160
    },
    {
      "epoch": 16.527129455909943,
      "grad_norm": 0.5646136403083801,
      "learning_rate": 0.002828,
      "loss": 1.374,
      "step": 220224
    },
    {
      "epoch": 16.531932457786116,
      "grad_norm": 0.4585341811180115,
      "learning_rate": 0.002828,
      "loss": 1.3738,
      "step": 220288
    },
    {
      "epoch": 16.53673545966229,
      "grad_norm": 0.5085334181785583,
      "learning_rate": 0.002828,
      "loss": 1.3713,
      "step": 220352
    },
    {
      "epoch": 16.54153846153846,
      "grad_norm": 0.6526809930801392,
      "learning_rate": 0.002828,
      "loss": 1.3746,
      "step": 220416
    },
    {
      "epoch": 16.546341463414635,
      "grad_norm": 0.5648365020751953,
      "learning_rate": 0.002828,
      "loss": 1.3766,
      "step": 220480
    },
    {
      "epoch": 16.551144465290808,
      "grad_norm": 0.6174665689468384,
      "learning_rate": 0.002828,
      "loss": 1.3688,
      "step": 220544
    },
    {
      "epoch": 16.55594746716698,
      "grad_norm": 0.5618147253990173,
      "learning_rate": 0.002828,
      "loss": 1.3718,
      "step": 220608
    },
    {
      "epoch": 16.560750469043153,
      "grad_norm": 0.5534228682518005,
      "learning_rate": 0.002828,
      "loss": 1.3731,
      "step": 220672
    },
    {
      "epoch": 16.565553470919326,
      "grad_norm": 0.5594824552536011,
      "learning_rate": 0.002828,
      "loss": 1.3691,
      "step": 220736
    },
    {
      "epoch": 16.570356472795496,
      "grad_norm": 0.48453760147094727,
      "learning_rate": 0.002828,
      "loss": 1.3733,
      "step": 220800
    },
    {
      "epoch": 16.57515947467167,
      "grad_norm": 0.58493572473526,
      "learning_rate": 0.002828,
      "loss": 1.3763,
      "step": 220864
    },
    {
      "epoch": 16.57996247654784,
      "grad_norm": 0.5082259774208069,
      "learning_rate": 0.002828,
      "loss": 1.368,
      "step": 220928
    },
    {
      "epoch": 16.584765478424014,
      "grad_norm": 0.6410751342773438,
      "learning_rate": 0.002828,
      "loss": 1.3764,
      "step": 220992
    },
    {
      "epoch": 16.589568480300187,
      "grad_norm": 0.6781150698661804,
      "learning_rate": 0.002828,
      "loss": 1.3693,
      "step": 221056
    },
    {
      "epoch": 16.59437148217636,
      "grad_norm": 0.5025544762611389,
      "learning_rate": 0.002828,
      "loss": 1.3664,
      "step": 221120
    },
    {
      "epoch": 16.599174484052533,
      "grad_norm": 0.48140037059783936,
      "learning_rate": 0.002828,
      "loss": 1.3721,
      "step": 221184
    },
    {
      "epoch": 16.603977485928706,
      "grad_norm": 0.6274412274360657,
      "learning_rate": 0.002828,
      "loss": 1.3723,
      "step": 221248
    },
    {
      "epoch": 16.60878048780488,
      "grad_norm": 0.5692063570022583,
      "learning_rate": 0.002828,
      "loss": 1.3746,
      "step": 221312
    },
    {
      "epoch": 16.61358348968105,
      "grad_norm": 0.5218042135238647,
      "learning_rate": 0.002828,
      "loss": 1.371,
      "step": 221376
    },
    {
      "epoch": 16.618386491557224,
      "grad_norm": 0.5552325248718262,
      "learning_rate": 0.002828,
      "loss": 1.3761,
      "step": 221440
    },
    {
      "epoch": 16.623189493433397,
      "grad_norm": 0.5654879212379456,
      "learning_rate": 0.002828,
      "loss": 1.3727,
      "step": 221504
    },
    {
      "epoch": 16.62799249530957,
      "grad_norm": 0.5669631958007812,
      "learning_rate": 0.002828,
      "loss": 1.3835,
      "step": 221568
    },
    {
      "epoch": 16.63279549718574,
      "grad_norm": 0.669795572757721,
      "learning_rate": 0.002828,
      "loss": 1.3811,
      "step": 221632
    },
    {
      "epoch": 16.637598499061912,
      "grad_norm": 0.5432254076004028,
      "learning_rate": 0.002828,
      "loss": 1.3763,
      "step": 221696
    },
    {
      "epoch": 16.642401500938085,
      "grad_norm": 0.5432413220405579,
      "learning_rate": 0.002828,
      "loss": 1.3776,
      "step": 221760
    },
    {
      "epoch": 16.647204502814258,
      "grad_norm": 0.42493462562561035,
      "learning_rate": 0.002828,
      "loss": 1.3737,
      "step": 221824
    },
    {
      "epoch": 16.65200750469043,
      "grad_norm": 0.6230249404907227,
      "learning_rate": 0.002828,
      "loss": 1.3781,
      "step": 221888
    },
    {
      "epoch": 16.656810506566604,
      "grad_norm": 0.5844151973724365,
      "learning_rate": 0.002828,
      "loss": 1.3718,
      "step": 221952
    },
    {
      "epoch": 16.661613508442777,
      "grad_norm": 0.5197720527648926,
      "learning_rate": 0.002828,
      "loss": 1.3752,
      "step": 222016
    },
    {
      "epoch": 16.66641651031895,
      "grad_norm": 0.5851918458938599,
      "learning_rate": 0.002828,
      "loss": 1.3775,
      "step": 222080
    },
    {
      "epoch": 16.671219512195123,
      "grad_norm": 0.5978190898895264,
      "learning_rate": 0.002828,
      "loss": 1.3683,
      "step": 222144
    },
    {
      "epoch": 16.676022514071295,
      "grad_norm": 0.5710034370422363,
      "learning_rate": 0.002828,
      "loss": 1.3705,
      "step": 222208
    },
    {
      "epoch": 16.68082551594747,
      "grad_norm": 0.55964195728302,
      "learning_rate": 0.002828,
      "loss": 1.3699,
      "step": 222272
    },
    {
      "epoch": 16.68562851782364,
      "grad_norm": 0.6255779266357422,
      "learning_rate": 0.002828,
      "loss": 1.3712,
      "step": 222336
    },
    {
      "epoch": 16.690431519699814,
      "grad_norm": 0.5409938097000122,
      "learning_rate": 0.002828,
      "loss": 1.3798,
      "step": 222400
    },
    {
      "epoch": 16.695234521575983,
      "grad_norm": 0.5851971507072449,
      "learning_rate": 0.002828,
      "loss": 1.3754,
      "step": 222464
    },
    {
      "epoch": 16.700037523452156,
      "grad_norm": 0.5326768159866333,
      "learning_rate": 0.002828,
      "loss": 1.3707,
      "step": 222528
    },
    {
      "epoch": 16.70484052532833,
      "grad_norm": 0.5734680891036987,
      "learning_rate": 0.002828,
      "loss": 1.3704,
      "step": 222592
    },
    {
      "epoch": 16.709643527204502,
      "grad_norm": 0.5416348576545715,
      "learning_rate": 0.002828,
      "loss": 1.3777,
      "step": 222656
    },
    {
      "epoch": 16.714446529080675,
      "grad_norm": 0.49465447664260864,
      "learning_rate": 0.002828,
      "loss": 1.3753,
      "step": 222720
    },
    {
      "epoch": 16.719249530956848,
      "grad_norm": 0.4981372654438019,
      "learning_rate": 0.002828,
      "loss": 1.3759,
      "step": 222784
    },
    {
      "epoch": 16.72405253283302,
      "grad_norm": 0.5880029201507568,
      "learning_rate": 0.002828,
      "loss": 1.3668,
      "step": 222848
    },
    {
      "epoch": 16.728855534709194,
      "grad_norm": 0.5089569091796875,
      "learning_rate": 0.002828,
      "loss": 1.375,
      "step": 222912
    },
    {
      "epoch": 16.733658536585367,
      "grad_norm": 0.576672375202179,
      "learning_rate": 0.002828,
      "loss": 1.373,
      "step": 222976
    },
    {
      "epoch": 16.73846153846154,
      "grad_norm": 0.592309296131134,
      "learning_rate": 0.002828,
      "loss": 1.371,
      "step": 223040
    },
    {
      "epoch": 16.743264540337712,
      "grad_norm": 0.5511286854743958,
      "learning_rate": 0.002828,
      "loss": 1.3758,
      "step": 223104
    },
    {
      "epoch": 16.748067542213885,
      "grad_norm": 0.5695531368255615,
      "learning_rate": 0.002828,
      "loss": 1.3713,
      "step": 223168
    },
    {
      "epoch": 16.752870544090058,
      "grad_norm": 0.6317942142486572,
      "learning_rate": 0.002828,
      "loss": 1.3765,
      "step": 223232
    },
    {
      "epoch": 16.757673545966227,
      "grad_norm": 0.5613177418708801,
      "learning_rate": 0.002828,
      "loss": 1.3727,
      "step": 223296
    },
    {
      "epoch": 16.7624765478424,
      "grad_norm": 0.6550566554069519,
      "learning_rate": 0.002828,
      "loss": 1.3751,
      "step": 223360
    },
    {
      "epoch": 16.767279549718573,
      "grad_norm": 0.559780478477478,
      "learning_rate": 0.002828,
      "loss": 1.375,
      "step": 223424
    },
    {
      "epoch": 16.772082551594746,
      "grad_norm": 0.5176964402198792,
      "learning_rate": 0.002828,
      "loss": 1.3682,
      "step": 223488
    },
    {
      "epoch": 16.77688555347092,
      "grad_norm": 0.5380821824073792,
      "learning_rate": 0.002828,
      "loss": 1.3715,
      "step": 223552
    },
    {
      "epoch": 16.781688555347092,
      "grad_norm": 0.5392264723777771,
      "learning_rate": 0.002828,
      "loss": 1.3785,
      "step": 223616
    },
    {
      "epoch": 16.786491557223265,
      "grad_norm": 0.5847867727279663,
      "learning_rate": 0.002828,
      "loss": 1.3732,
      "step": 223680
    },
    {
      "epoch": 16.791294559099438,
      "grad_norm": 0.564135730266571,
      "learning_rate": 0.002828,
      "loss": 1.3753,
      "step": 223744
    },
    {
      "epoch": 16.79609756097561,
      "grad_norm": 0.5908150672912598,
      "learning_rate": 0.002828,
      "loss": 1.3724,
      "step": 223808
    },
    {
      "epoch": 16.800900562851783,
      "grad_norm": 0.5761885643005371,
      "learning_rate": 0.002828,
      "loss": 1.3764,
      "step": 223872
    },
    {
      "epoch": 16.805703564727956,
      "grad_norm": 0.5740852355957031,
      "learning_rate": 0.002828,
      "loss": 1.3798,
      "step": 223936
    },
    {
      "epoch": 16.81050656660413,
      "grad_norm": 0.49333980679512024,
      "learning_rate": 0.002828,
      "loss": 1.3788,
      "step": 224000
    },
    {
      "epoch": 16.815309568480302,
      "grad_norm": 0.4524737000465393,
      "learning_rate": 0.002828,
      "loss": 1.3672,
      "step": 224064
    },
    {
      "epoch": 16.82011257035647,
      "grad_norm": 0.5872974991798401,
      "learning_rate": 0.002828,
      "loss": 1.3792,
      "step": 224128
    },
    {
      "epoch": 16.824915572232644,
      "grad_norm": 0.6633861660957336,
      "learning_rate": 0.002828,
      "loss": 1.3736,
      "step": 224192
    },
    {
      "epoch": 16.829718574108817,
      "grad_norm": 0.7240063548088074,
      "learning_rate": 0.002828,
      "loss": 1.3703,
      "step": 224256
    },
    {
      "epoch": 16.83452157598499,
      "grad_norm": 0.5940957069396973,
      "learning_rate": 0.002828,
      "loss": 1.3753,
      "step": 224320
    },
    {
      "epoch": 16.839324577861163,
      "grad_norm": 0.5730873346328735,
      "learning_rate": 0.002828,
      "loss": 1.3767,
      "step": 224384
    },
    {
      "epoch": 16.844127579737336,
      "grad_norm": 0.6729320883750916,
      "learning_rate": 0.002828,
      "loss": 1.3753,
      "step": 224448
    },
    {
      "epoch": 16.84893058161351,
      "grad_norm": 0.5936250686645508,
      "learning_rate": 0.002828,
      "loss": 1.3726,
      "step": 224512
    },
    {
      "epoch": 16.85373358348968,
      "grad_norm": 0.546510636806488,
      "learning_rate": 0.002828,
      "loss": 1.3723,
      "step": 224576
    },
    {
      "epoch": 16.858536585365854,
      "grad_norm": 0.6515787839889526,
      "learning_rate": 0.002828,
      "loss": 1.3709,
      "step": 224640
    },
    {
      "epoch": 16.863339587242027,
      "grad_norm": 0.455196350812912,
      "learning_rate": 0.002828,
      "loss": 1.3698,
      "step": 224704
    },
    {
      "epoch": 16.8681425891182,
      "grad_norm": 0.498130202293396,
      "learning_rate": 0.002828,
      "loss": 1.3744,
      "step": 224768
    },
    {
      "epoch": 16.872945590994373,
      "grad_norm": 0.4296855926513672,
      "learning_rate": 0.002828,
      "loss": 1.3725,
      "step": 224832
    },
    {
      "epoch": 16.877748592870542,
      "grad_norm": 0.47755885124206543,
      "learning_rate": 0.002828,
      "loss": 1.3778,
      "step": 224896
    },
    {
      "epoch": 16.882551594746715,
      "grad_norm": 0.5465676188468933,
      "learning_rate": 0.002828,
      "loss": 1.3782,
      "step": 224960
    },
    {
      "epoch": 16.887354596622888,
      "grad_norm": 0.6344484686851501,
      "learning_rate": 0.002828,
      "loss": 1.3742,
      "step": 225024
    },
    {
      "epoch": 16.89215759849906,
      "grad_norm": 0.5083233714103699,
      "learning_rate": 0.002828,
      "loss": 1.3718,
      "step": 225088
    },
    {
      "epoch": 16.896960600375234,
      "grad_norm": 0.5428247451782227,
      "learning_rate": 0.002828,
      "loss": 1.375,
      "step": 225152
    },
    {
      "epoch": 16.901763602251407,
      "grad_norm": 0.47932419180870056,
      "learning_rate": 0.002828,
      "loss": 1.3739,
      "step": 225216
    },
    {
      "epoch": 16.90656660412758,
      "grad_norm": 0.534848153591156,
      "learning_rate": 0.002828,
      "loss": 1.3779,
      "step": 225280
    },
    {
      "epoch": 16.911369606003753,
      "grad_norm": 0.5262565612792969,
      "learning_rate": 0.002828,
      "loss": 1.3792,
      "step": 225344
    },
    {
      "epoch": 16.916172607879925,
      "grad_norm": 0.5471115112304688,
      "learning_rate": 0.002828,
      "loss": 1.3732,
      "step": 225408
    },
    {
      "epoch": 16.9209756097561,
      "grad_norm": 0.5170307755470276,
      "learning_rate": 0.002828,
      "loss": 1.3742,
      "step": 225472
    },
    {
      "epoch": 16.92577861163227,
      "grad_norm": 0.491405189037323,
      "learning_rate": 0.002828,
      "loss": 1.3755,
      "step": 225536
    },
    {
      "epoch": 16.930581613508444,
      "grad_norm": 0.5076206922531128,
      "learning_rate": 0.002828,
      "loss": 1.3717,
      "step": 225600
    },
    {
      "epoch": 16.935384615384617,
      "grad_norm": 0.6649317741394043,
      "learning_rate": 0.002828,
      "loss": 1.3787,
      "step": 225664
    },
    {
      "epoch": 16.940187617260786,
      "grad_norm": 0.5338141918182373,
      "learning_rate": 0.002828,
      "loss": 1.3786,
      "step": 225728
    },
    {
      "epoch": 16.94499061913696,
      "grad_norm": 0.4969607889652252,
      "learning_rate": 0.002828,
      "loss": 1.3772,
      "step": 225792
    },
    {
      "epoch": 16.949793621013132,
      "grad_norm": 0.6936911344528198,
      "learning_rate": 0.002828,
      "loss": 1.3713,
      "step": 225856
    },
    {
      "epoch": 16.954596622889305,
      "grad_norm": 0.5464868545532227,
      "learning_rate": 0.002828,
      "loss": 1.3798,
      "step": 225920
    },
    {
      "epoch": 16.959399624765478,
      "grad_norm": 0.4734916687011719,
      "learning_rate": 0.002828,
      "loss": 1.3737,
      "step": 225984
    },
    {
      "epoch": 16.96420262664165,
      "grad_norm": 0.5360217094421387,
      "learning_rate": 0.002828,
      "loss": 1.3718,
      "step": 226048
    },
    {
      "epoch": 16.969005628517824,
      "grad_norm": 0.5119659304618835,
      "learning_rate": 0.002828,
      "loss": 1.3771,
      "step": 226112
    },
    {
      "epoch": 16.973808630393997,
      "grad_norm": 0.6373065114021301,
      "learning_rate": 0.002828,
      "loss": 1.3751,
      "step": 226176
    },
    {
      "epoch": 16.97861163227017,
      "grad_norm": 0.6539568305015564,
      "learning_rate": 0.002828,
      "loss": 1.3789,
      "step": 226240
    },
    {
      "epoch": 16.983414634146342,
      "grad_norm": 0.5388275384902954,
      "learning_rate": 0.002828,
      "loss": 1.3793,
      "step": 226304
    },
    {
      "epoch": 16.988217636022515,
      "grad_norm": 0.5029416680335999,
      "learning_rate": 0.002828,
      "loss": 1.3769,
      "step": 226368
    },
    {
      "epoch": 16.993020637898688,
      "grad_norm": 0.6130713820457458,
      "learning_rate": 0.002828,
      "loss": 1.3824,
      "step": 226432
    },
    {
      "epoch": 16.99782363977486,
      "grad_norm": 0.5726588368415833,
      "learning_rate": 0.002828,
      "loss": 1.3749,
      "step": 226496
    },
    {
      "epoch": 17.00262664165103,
      "grad_norm": 0.5122669339179993,
      "learning_rate": 0.002828,
      "loss": 1.3472,
      "step": 226560
    },
    {
      "epoch": 17.007429643527203,
      "grad_norm": 0.5689617395401001,
      "learning_rate": 0.002828,
      "loss": 1.3311,
      "step": 226624
    },
    {
      "epoch": 17.012232645403376,
      "grad_norm": 0.4785527288913727,
      "learning_rate": 0.002828,
      "loss": 1.3295,
      "step": 226688
    },
    {
      "epoch": 17.01703564727955,
      "grad_norm": 0.48141196370124817,
      "learning_rate": 0.002828,
      "loss": 1.3372,
      "step": 226752
    },
    {
      "epoch": 17.021838649155722,
      "grad_norm": 0.5237526893615723,
      "learning_rate": 0.002828,
      "loss": 1.3331,
      "step": 226816
    },
    {
      "epoch": 17.026641651031895,
      "grad_norm": 0.6010721325874329,
      "learning_rate": 0.002828,
      "loss": 1.3416,
      "step": 226880
    },
    {
      "epoch": 17.031444652908068,
      "grad_norm": 0.5771501660346985,
      "learning_rate": 0.002828,
      "loss": 1.3296,
      "step": 226944
    },
    {
      "epoch": 17.03624765478424,
      "grad_norm": 0.48024001717567444,
      "learning_rate": 0.002828,
      "loss": 1.333,
      "step": 227008
    },
    {
      "epoch": 17.041050656660413,
      "grad_norm": 0.6010895371437073,
      "learning_rate": 0.002828,
      "loss": 1.3413,
      "step": 227072
    },
    {
      "epoch": 17.045853658536586,
      "grad_norm": 0.49624505639076233,
      "learning_rate": 0.002828,
      "loss": 1.339,
      "step": 227136
    },
    {
      "epoch": 17.05065666041276,
      "grad_norm": 0.6421620845794678,
      "learning_rate": 0.002828,
      "loss": 1.341,
      "step": 227200
    },
    {
      "epoch": 17.055459662288932,
      "grad_norm": 0.5180822014808655,
      "learning_rate": 0.002828,
      "loss": 1.3375,
      "step": 227264
    },
    {
      "epoch": 17.060262664165105,
      "grad_norm": 0.48310354351997375,
      "learning_rate": 0.002828,
      "loss": 1.3388,
      "step": 227328
    },
    {
      "epoch": 17.065065666041274,
      "grad_norm": 0.5890333652496338,
      "learning_rate": 0.002828,
      "loss": 1.3361,
      "step": 227392
    },
    {
      "epoch": 17.069868667917447,
      "grad_norm": 0.5011729001998901,
      "learning_rate": 0.002828,
      "loss": 1.3412,
      "step": 227456
    },
    {
      "epoch": 17.07467166979362,
      "grad_norm": 0.5238128304481506,
      "learning_rate": 0.002828,
      "loss": 1.3366,
      "step": 227520
    },
    {
      "epoch": 17.079474671669793,
      "grad_norm": 0.5375518798828125,
      "learning_rate": 0.002828,
      "loss": 1.3405,
      "step": 227584
    },
    {
      "epoch": 17.084277673545966,
      "grad_norm": 0.646587610244751,
      "learning_rate": 0.002828,
      "loss": 1.3415,
      "step": 227648
    },
    {
      "epoch": 17.08908067542214,
      "grad_norm": 0.538013756275177,
      "learning_rate": 0.002828,
      "loss": 1.3367,
      "step": 227712
    },
    {
      "epoch": 17.09388367729831,
      "grad_norm": 0.493001788854599,
      "learning_rate": 0.002828,
      "loss": 1.3374,
      "step": 227776
    },
    {
      "epoch": 17.098686679174484,
      "grad_norm": 0.6065325140953064,
      "learning_rate": 0.002828,
      "loss": 1.3353,
      "step": 227840
    },
    {
      "epoch": 17.103489681050657,
      "grad_norm": 0.6094088554382324,
      "learning_rate": 0.002828,
      "loss": 1.341,
      "step": 227904
    },
    {
      "epoch": 17.10829268292683,
      "grad_norm": 0.555350661277771,
      "learning_rate": 0.002828,
      "loss": 1.3364,
      "step": 227968
    },
    {
      "epoch": 17.113095684803003,
      "grad_norm": 0.5394548177719116,
      "learning_rate": 0.002828,
      "loss": 1.342,
      "step": 228032
    },
    {
      "epoch": 17.117898686679176,
      "grad_norm": 0.5966508984565735,
      "learning_rate": 0.002828,
      "loss": 1.3394,
      "step": 228096
    },
    {
      "epoch": 17.122701688555345,
      "grad_norm": 0.49855026602745056,
      "learning_rate": 0.002828,
      "loss": 1.3375,
      "step": 228160
    },
    {
      "epoch": 17.12750469043152,
      "grad_norm": 0.5403494834899902,
      "learning_rate": 0.002828,
      "loss": 1.3414,
      "step": 228224
    },
    {
      "epoch": 17.13230769230769,
      "grad_norm": 0.5960873365402222,
      "learning_rate": 0.002828,
      "loss": 1.3391,
      "step": 228288
    },
    {
      "epoch": 17.137110694183864,
      "grad_norm": 0.5608639717102051,
      "learning_rate": 0.002828,
      "loss": 1.3459,
      "step": 228352
    },
    {
      "epoch": 17.141913696060037,
      "grad_norm": 0.5412024855613708,
      "learning_rate": 0.002828,
      "loss": 1.3461,
      "step": 228416
    },
    {
      "epoch": 17.14671669793621,
      "grad_norm": 0.5174167156219482,
      "learning_rate": 0.002828,
      "loss": 1.3387,
      "step": 228480
    },
    {
      "epoch": 17.151519699812383,
      "grad_norm": 0.4773479104042053,
      "learning_rate": 0.002828,
      "loss": 1.346,
      "step": 228544
    },
    {
      "epoch": 17.156322701688556,
      "grad_norm": 0.5367104411125183,
      "learning_rate": 0.002828,
      "loss": 1.3468,
      "step": 228608
    },
    {
      "epoch": 17.16112570356473,
      "grad_norm": 0.5563543438911438,
      "learning_rate": 0.002828,
      "loss": 1.3451,
      "step": 228672
    },
    {
      "epoch": 17.1659287054409,
      "grad_norm": 0.5474345684051514,
      "learning_rate": 0.002828,
      "loss": 1.3351,
      "step": 228736
    },
    {
      "epoch": 17.170731707317074,
      "grad_norm": 0.621706485748291,
      "learning_rate": 0.002828,
      "loss": 1.3382,
      "step": 228800
    },
    {
      "epoch": 17.175534709193247,
      "grad_norm": 0.6461145281791687,
      "learning_rate": 0.002828,
      "loss": 1.3426,
      "step": 228864
    },
    {
      "epoch": 17.18033771106942,
      "grad_norm": 0.6269495487213135,
      "learning_rate": 0.002828,
      "loss": 1.3445,
      "step": 228928
    },
    {
      "epoch": 17.18514071294559,
      "grad_norm": 0.5621188282966614,
      "learning_rate": 0.002828,
      "loss": 1.3442,
      "step": 228992
    },
    {
      "epoch": 17.189943714821762,
      "grad_norm": 0.544793426990509,
      "learning_rate": 0.002828,
      "loss": 1.3443,
      "step": 229056
    },
    {
      "epoch": 17.194746716697935,
      "grad_norm": 0.5203141570091248,
      "learning_rate": 0.002828,
      "loss": 1.3445,
      "step": 229120
    },
    {
      "epoch": 17.199549718574108,
      "grad_norm": 0.6426690816879272,
      "learning_rate": 0.002828,
      "loss": 1.3463,
      "step": 229184
    },
    {
      "epoch": 17.20435272045028,
      "grad_norm": 0.6133685111999512,
      "learning_rate": 0.002828,
      "loss": 1.3511,
      "step": 229248
    },
    {
      "epoch": 17.209155722326454,
      "grad_norm": 0.5746537446975708,
      "learning_rate": 0.002828,
      "loss": 1.3431,
      "step": 229312
    },
    {
      "epoch": 17.213958724202627,
      "grad_norm": 0.6326211094856262,
      "learning_rate": 0.002828,
      "loss": 1.3479,
      "step": 229376
    },
    {
      "epoch": 17.2187617260788,
      "grad_norm": 0.545437216758728,
      "learning_rate": 0.002828,
      "loss": 1.338,
      "step": 229440
    },
    {
      "epoch": 17.223564727954972,
      "grad_norm": 0.5582890510559082,
      "learning_rate": 0.002828,
      "loss": 1.3433,
      "step": 229504
    },
    {
      "epoch": 17.228367729831145,
      "grad_norm": 0.4580340087413788,
      "learning_rate": 0.002828,
      "loss": 1.3476,
      "step": 229568
    },
    {
      "epoch": 17.233170731707318,
      "grad_norm": 0.6911355257034302,
      "learning_rate": 0.002828,
      "loss": 1.3526,
      "step": 229632
    },
    {
      "epoch": 17.23797373358349,
      "grad_norm": 0.5566486120223999,
      "learning_rate": 0.002828,
      "loss": 1.3482,
      "step": 229696
    },
    {
      "epoch": 17.242776735459664,
      "grad_norm": 0.6304554343223572,
      "learning_rate": 0.002828,
      "loss": 1.3476,
      "step": 229760
    },
    {
      "epoch": 17.247579737335833,
      "grad_norm": 0.6848873496055603,
      "learning_rate": 0.002828,
      "loss": 1.3424,
      "step": 229824
    },
    {
      "epoch": 17.252382739212006,
      "grad_norm": 0.5782124996185303,
      "learning_rate": 0.002828,
      "loss": 1.3502,
      "step": 229888
    },
    {
      "epoch": 17.25718574108818,
      "grad_norm": 0.533871054649353,
      "learning_rate": 0.002828,
      "loss": 1.3418,
      "step": 229952
    },
    {
      "epoch": 17.261988742964352,
      "grad_norm": 0.636326253414154,
      "learning_rate": 0.002828,
      "loss": 1.3464,
      "step": 230016
    },
    {
      "epoch": 17.266791744840525,
      "grad_norm": 0.5879301428794861,
      "learning_rate": 0.002828,
      "loss": 1.3457,
      "step": 230080
    },
    {
      "epoch": 17.271594746716698,
      "grad_norm": 0.6339291334152222,
      "learning_rate": 0.002828,
      "loss": 1.3487,
      "step": 230144
    },
    {
      "epoch": 17.27639774859287,
      "grad_norm": 0.5430083274841309,
      "learning_rate": 0.002828,
      "loss": 1.3452,
      "step": 230208
    },
    {
      "epoch": 17.281200750469043,
      "grad_norm": 0.498355895280838,
      "learning_rate": 0.002828,
      "loss": 1.3491,
      "step": 230272
    },
    {
      "epoch": 17.286003752345216,
      "grad_norm": 0.5435526371002197,
      "learning_rate": 0.002828,
      "loss": 1.3493,
      "step": 230336
    },
    {
      "epoch": 17.29080675422139,
      "grad_norm": 0.5432043671607971,
      "learning_rate": 0.002828,
      "loss": 1.3465,
      "step": 230400
    },
    {
      "epoch": 17.295609756097562,
      "grad_norm": 0.5218422412872314,
      "learning_rate": 0.002828,
      "loss": 1.3509,
      "step": 230464
    },
    {
      "epoch": 17.300412757973735,
      "grad_norm": 0.5061742663383484,
      "learning_rate": 0.002828,
      "loss": 1.3441,
      "step": 230528
    },
    {
      "epoch": 17.305215759849908,
      "grad_norm": 0.5243954062461853,
      "learning_rate": 0.002828,
      "loss": 1.3497,
      "step": 230592
    },
    {
      "epoch": 17.310018761726077,
      "grad_norm": 0.6132803559303284,
      "learning_rate": 0.002828,
      "loss": 1.3467,
      "step": 230656
    },
    {
      "epoch": 17.31482176360225,
      "grad_norm": 0.5467346906661987,
      "learning_rate": 0.002828,
      "loss": 1.3525,
      "step": 230720
    },
    {
      "epoch": 17.319624765478423,
      "grad_norm": 0.5703427791595459,
      "learning_rate": 0.002828,
      "loss": 1.3509,
      "step": 230784
    },
    {
      "epoch": 17.324427767354596,
      "grad_norm": 0.6172657012939453,
      "learning_rate": 0.002828,
      "loss": 1.3464,
      "step": 230848
    },
    {
      "epoch": 17.32923076923077,
      "grad_norm": 0.5590468645095825,
      "learning_rate": 0.002828,
      "loss": 1.3447,
      "step": 230912
    },
    {
      "epoch": 17.33403377110694,
      "grad_norm": 0.6089560985565186,
      "learning_rate": 0.002828,
      "loss": 1.3444,
      "step": 230976
    },
    {
      "epoch": 17.338836772983115,
      "grad_norm": 0.6214098334312439,
      "learning_rate": 0.002828,
      "loss": 1.3541,
      "step": 231040
    },
    {
      "epoch": 17.343639774859287,
      "grad_norm": 0.48207151889801025,
      "learning_rate": 0.002828,
      "loss": 1.3516,
      "step": 231104
    },
    {
      "epoch": 17.34844277673546,
      "grad_norm": 0.5069649815559387,
      "learning_rate": 0.002828,
      "loss": 1.3422,
      "step": 231168
    },
    {
      "epoch": 17.353245778611633,
      "grad_norm": 0.6617603302001953,
      "learning_rate": 0.002828,
      "loss": 1.345,
      "step": 231232
    },
    {
      "epoch": 17.358048780487806,
      "grad_norm": 0.5475003123283386,
      "learning_rate": 0.002828,
      "loss": 1.3446,
      "step": 231296
    },
    {
      "epoch": 17.36285178236398,
      "grad_norm": 0.495044469833374,
      "learning_rate": 0.002828,
      "loss": 1.3514,
      "step": 231360
    },
    {
      "epoch": 17.367654784240152,
      "grad_norm": 0.646589994430542,
      "learning_rate": 0.002828,
      "loss": 1.3472,
      "step": 231424
    },
    {
      "epoch": 17.37245778611632,
      "grad_norm": 0.6345487236976624,
      "learning_rate": 0.002828,
      "loss": 1.3473,
      "step": 231488
    },
    {
      "epoch": 17.377260787992494,
      "grad_norm": 0.6994035243988037,
      "learning_rate": 0.002828,
      "loss": 1.348,
      "step": 231552
    },
    {
      "epoch": 17.382063789868667,
      "grad_norm": 0.7058942317962646,
      "learning_rate": 0.002828,
      "loss": 1.348,
      "step": 231616
    },
    {
      "epoch": 17.38686679174484,
      "grad_norm": 0.5912184715270996,
      "learning_rate": 0.002828,
      "loss": 1.3523,
      "step": 231680
    },
    {
      "epoch": 17.391669793621013,
      "grad_norm": 0.4432578384876251,
      "learning_rate": 0.002828,
      "loss": 1.3516,
      "step": 231744
    },
    {
      "epoch": 17.396472795497186,
      "grad_norm": 0.5171277523040771,
      "learning_rate": 0.002828,
      "loss": 1.3544,
      "step": 231808
    },
    {
      "epoch": 17.40127579737336,
      "grad_norm": 0.6040003299713135,
      "learning_rate": 0.002828,
      "loss": 1.3496,
      "step": 231872
    },
    {
      "epoch": 17.40607879924953,
      "grad_norm": 0.5984319448471069,
      "learning_rate": 0.002828,
      "loss": 1.3529,
      "step": 231936
    },
    {
      "epoch": 17.410881801125704,
      "grad_norm": 0.5507851839065552,
      "learning_rate": 0.002828,
      "loss": 1.3544,
      "step": 232000
    },
    {
      "epoch": 17.415684803001877,
      "grad_norm": 0.5893108248710632,
      "learning_rate": 0.002828,
      "loss": 1.347,
      "step": 232064
    },
    {
      "epoch": 17.42048780487805,
      "grad_norm": 0.6688479781150818,
      "learning_rate": 0.002828,
      "loss": 1.352,
      "step": 232128
    },
    {
      "epoch": 17.425290806754223,
      "grad_norm": 0.5244783759117126,
      "learning_rate": 0.002828,
      "loss": 1.3554,
      "step": 232192
    },
    {
      "epoch": 17.430093808630392,
      "grad_norm": 0.5949782133102417,
      "learning_rate": 0.002828,
      "loss": 1.3511,
      "step": 232256
    },
    {
      "epoch": 17.434896810506565,
      "grad_norm": 0.6664429903030396,
      "learning_rate": 0.002828,
      "loss": 1.3498,
      "step": 232320
    },
    {
      "epoch": 17.439699812382738,
      "grad_norm": 0.48198601603507996,
      "learning_rate": 0.002828,
      "loss": 1.3457,
      "step": 232384
    },
    {
      "epoch": 17.44450281425891,
      "grad_norm": 0.5703681111335754,
      "learning_rate": 0.002828,
      "loss": 1.3515,
      "step": 232448
    },
    {
      "epoch": 17.449305816135084,
      "grad_norm": 0.6069356799125671,
      "learning_rate": 0.002828,
      "loss": 1.3586,
      "step": 232512
    },
    {
      "epoch": 17.454108818011257,
      "grad_norm": 0.5629026889801025,
      "learning_rate": 0.002828,
      "loss": 1.3515,
      "step": 232576
    },
    {
      "epoch": 17.45891181988743,
      "grad_norm": 0.6319453716278076,
      "learning_rate": 0.002828,
      "loss": 1.3477,
      "step": 232640
    },
    {
      "epoch": 17.463714821763602,
      "grad_norm": 0.5908174514770508,
      "learning_rate": 0.002828,
      "loss": 1.3521,
      "step": 232704
    },
    {
      "epoch": 17.468517823639775,
      "grad_norm": 0.5290798544883728,
      "learning_rate": 0.002828,
      "loss": 1.3528,
      "step": 232768
    },
    {
      "epoch": 17.47332082551595,
      "grad_norm": 0.553308367729187,
      "learning_rate": 0.002828,
      "loss": 1.3519,
      "step": 232832
    },
    {
      "epoch": 17.47812382739212,
      "grad_norm": 0.5727306008338928,
      "learning_rate": 0.002828,
      "loss": 1.3467,
      "step": 232896
    },
    {
      "epoch": 17.482926829268294,
      "grad_norm": 0.5325832962989807,
      "learning_rate": 0.002828,
      "loss": 1.3567,
      "step": 232960
    },
    {
      "epoch": 17.487729831144467,
      "grad_norm": 0.5459145307540894,
      "learning_rate": 0.002828,
      "loss": 1.3495,
      "step": 233024
    },
    {
      "epoch": 17.492532833020636,
      "grad_norm": 0.5476085543632507,
      "learning_rate": 0.002828,
      "loss": 1.3514,
      "step": 233088
    },
    {
      "epoch": 17.49733583489681,
      "grad_norm": 0.5742253661155701,
      "learning_rate": 0.002828,
      "loss": 1.3503,
      "step": 233152
    },
    {
      "epoch": 17.502138836772982,
      "grad_norm": 0.6163190007209778,
      "learning_rate": 0.002828,
      "loss": 1.3525,
      "step": 233216
    },
    {
      "epoch": 17.506941838649155,
      "grad_norm": 0.5045285224914551,
      "learning_rate": 0.002828,
      "loss": 1.3545,
      "step": 233280
    },
    {
      "epoch": 17.511744840525328,
      "grad_norm": 0.5438716411590576,
      "learning_rate": 0.002828,
      "loss": 1.3506,
      "step": 233344
    },
    {
      "epoch": 17.5165478424015,
      "grad_norm": 0.6147487163543701,
      "learning_rate": 0.002828,
      "loss": 1.3532,
      "step": 233408
    },
    {
      "epoch": 17.521350844277674,
      "grad_norm": 0.5019384026527405,
      "learning_rate": 0.002828,
      "loss": 1.3551,
      "step": 233472
    },
    {
      "epoch": 17.526153846153846,
      "grad_norm": 0.5296437740325928,
      "learning_rate": 0.002828,
      "loss": 1.3573,
      "step": 233536
    },
    {
      "epoch": 17.53095684803002,
      "grad_norm": 0.689786970615387,
      "learning_rate": 0.002828,
      "loss": 1.3553,
      "step": 233600
    },
    {
      "epoch": 17.535759849906192,
      "grad_norm": 0.5649539232254028,
      "learning_rate": 0.002828,
      "loss": 1.3569,
      "step": 233664
    },
    {
      "epoch": 17.540562851782365,
      "grad_norm": 0.563384473323822,
      "learning_rate": 0.002828,
      "loss": 1.3527,
      "step": 233728
    },
    {
      "epoch": 17.545365853658538,
      "grad_norm": 0.6230480670928955,
      "learning_rate": 0.002828,
      "loss": 1.348,
      "step": 233792
    },
    {
      "epoch": 17.55016885553471,
      "grad_norm": 0.5748286247253418,
      "learning_rate": 0.002828,
      "loss": 1.3464,
      "step": 233856
    },
    {
      "epoch": 17.55497185741088,
      "grad_norm": 0.6217999458312988,
      "learning_rate": 0.002828,
      "loss": 1.3488,
      "step": 233920
    },
    {
      "epoch": 17.559774859287053,
      "grad_norm": 0.5565934777259827,
      "learning_rate": 0.002828,
      "loss": 1.3505,
      "step": 233984
    },
    {
      "epoch": 17.564577861163226,
      "grad_norm": 0.5826540589332581,
      "learning_rate": 0.002828,
      "loss": 1.3547,
      "step": 234048
    },
    {
      "epoch": 17.5693808630394,
      "grad_norm": 0.6381579637527466,
      "learning_rate": 0.002828,
      "loss": 1.3519,
      "step": 234112
    },
    {
      "epoch": 17.57418386491557,
      "grad_norm": 0.5593814849853516,
      "learning_rate": 0.002828,
      "loss": 1.3553,
      "step": 234176
    },
    {
      "epoch": 17.578986866791745,
      "grad_norm": 0.5353369116783142,
      "learning_rate": 0.002828,
      "loss": 1.3504,
      "step": 234240
    },
    {
      "epoch": 17.583789868667917,
      "grad_norm": 0.5172046422958374,
      "learning_rate": 0.002828,
      "loss": 1.354,
      "step": 234304
    },
    {
      "epoch": 17.58859287054409,
      "grad_norm": 0.5404736995697021,
      "learning_rate": 0.002828,
      "loss": 1.3529,
      "step": 234368
    },
    {
      "epoch": 17.593395872420263,
      "grad_norm": 0.5068063139915466,
      "learning_rate": 0.002828,
      "loss": 1.3487,
      "step": 234432
    },
    {
      "epoch": 17.598198874296436,
      "grad_norm": 0.600818932056427,
      "learning_rate": 0.002828,
      "loss": 1.3605,
      "step": 234496
    },
    {
      "epoch": 17.60300187617261,
      "grad_norm": 0.5050070285797119,
      "learning_rate": 0.002828,
      "loss": 1.3499,
      "step": 234560
    },
    {
      "epoch": 17.607804878048782,
      "grad_norm": 0.5006924271583557,
      "learning_rate": 0.002828,
      "loss": 1.3574,
      "step": 234624
    },
    {
      "epoch": 17.612607879924955,
      "grad_norm": 0.5756498575210571,
      "learning_rate": 0.002828,
      "loss": 1.356,
      "step": 234688
    },
    {
      "epoch": 17.617410881801124,
      "grad_norm": 0.5910568237304688,
      "learning_rate": 0.002828,
      "loss": 1.3516,
      "step": 234752
    },
    {
      "epoch": 17.622213883677297,
      "grad_norm": 0.48506519198417664,
      "learning_rate": 0.002828,
      "loss": 1.3545,
      "step": 234816
    },
    {
      "epoch": 17.62701688555347,
      "grad_norm": 1.732113242149353,
      "learning_rate": 0.002828,
      "loss": 1.3485,
      "step": 234880
    },
    {
      "epoch": 17.631819887429643,
      "grad_norm": 0.4970155954360962,
      "learning_rate": 0.002828,
      "loss": 1.3588,
      "step": 234944
    },
    {
      "epoch": 17.636622889305816,
      "grad_norm": 0.5651705265045166,
      "learning_rate": 0.002828,
      "loss": 1.3495,
      "step": 235008
    },
    {
      "epoch": 17.64142589118199,
      "grad_norm": 0.5904951095581055,
      "learning_rate": 0.002828,
      "loss": 1.3566,
      "step": 235072
    },
    {
      "epoch": 17.64622889305816,
      "grad_norm": 0.5296029448509216,
      "learning_rate": 0.002828,
      "loss": 1.3471,
      "step": 235136
    },
    {
      "epoch": 17.651031894934334,
      "grad_norm": 0.5207241177558899,
      "learning_rate": 0.002828,
      "loss": 1.3547,
      "step": 235200
    },
    {
      "epoch": 17.655834896810507,
      "grad_norm": 0.6143551468849182,
      "learning_rate": 0.002828,
      "loss": 1.3537,
      "step": 235264
    },
    {
      "epoch": 17.66063789868668,
      "grad_norm": 0.6017210483551025,
      "learning_rate": 0.002828,
      "loss": 1.3512,
      "step": 235328
    },
    {
      "epoch": 17.665440900562853,
      "grad_norm": 0.4639129340648651,
      "learning_rate": 0.002828,
      "loss": 1.3461,
      "step": 235392
    },
    {
      "epoch": 17.670243902439026,
      "grad_norm": 0.5471293330192566,
      "learning_rate": 0.002828,
      "loss": 1.349,
      "step": 235456
    },
    {
      "epoch": 17.675046904315195,
      "grad_norm": 0.5401851534843445,
      "learning_rate": 0.002828,
      "loss": 1.3588,
      "step": 235520
    },
    {
      "epoch": 17.679849906191368,
      "grad_norm": 0.4593111574649811,
      "learning_rate": 0.002828,
      "loss": 1.3465,
      "step": 235584
    },
    {
      "epoch": 17.68465290806754,
      "grad_norm": 0.5718096494674683,
      "learning_rate": 0.002828,
      "loss": 1.3565,
      "step": 235648
    },
    {
      "epoch": 17.689455909943714,
      "grad_norm": 0.5341073274612427,
      "learning_rate": 0.002828,
      "loss": 1.3534,
      "step": 235712
    },
    {
      "epoch": 17.694258911819887,
      "grad_norm": 0.5088701248168945,
      "learning_rate": 0.002828,
      "loss": 1.3553,
      "step": 235776
    },
    {
      "epoch": 17.69906191369606,
      "grad_norm": 0.5865604877471924,
      "learning_rate": 0.002828,
      "loss": 1.3529,
      "step": 235840
    },
    {
      "epoch": 17.703864915572233,
      "grad_norm": 0.5808840990066528,
      "learning_rate": 0.002828,
      "loss": 1.354,
      "step": 235904
    },
    {
      "epoch": 17.708667917448405,
      "grad_norm": 0.5625920295715332,
      "learning_rate": 0.002828,
      "loss": 1.3522,
      "step": 235968
    },
    {
      "epoch": 17.71347091932458,
      "grad_norm": 0.6199504733085632,
      "learning_rate": 0.002828,
      "loss": 1.3521,
      "step": 236032
    },
    {
      "epoch": 17.71827392120075,
      "grad_norm": 0.4788922071456909,
      "learning_rate": 0.002828,
      "loss": 1.3543,
      "step": 236096
    },
    {
      "epoch": 17.723076923076924,
      "grad_norm": 0.6806089282035828,
      "learning_rate": 0.002828,
      "loss": 1.3548,
      "step": 236160
    },
    {
      "epoch": 17.727879924953097,
      "grad_norm": 0.5504467487335205,
      "learning_rate": 0.002828,
      "loss": 1.3534,
      "step": 236224
    },
    {
      "epoch": 17.73268292682927,
      "grad_norm": 0.5322136878967285,
      "learning_rate": 0.002828,
      "loss": 1.3538,
      "step": 236288
    },
    {
      "epoch": 17.73748592870544,
      "grad_norm": 0.5951442122459412,
      "learning_rate": 0.002828,
      "loss": 1.3485,
      "step": 236352
    },
    {
      "epoch": 17.742288930581612,
      "grad_norm": 0.5594306588172913,
      "learning_rate": 0.002828,
      "loss": 1.3579,
      "step": 236416
    },
    {
      "epoch": 17.747091932457785,
      "grad_norm": 0.6795274615287781,
      "learning_rate": 0.002828,
      "loss": 1.3487,
      "step": 236480
    },
    {
      "epoch": 17.751894934333958,
      "grad_norm": 0.4794277846813202,
      "learning_rate": 0.002828,
      "loss": 1.3562,
      "step": 236544
    },
    {
      "epoch": 17.75669793621013,
      "grad_norm": 0.5802697539329529,
      "learning_rate": 0.002828,
      "loss": 1.3489,
      "step": 236608
    },
    {
      "epoch": 17.761500938086304,
      "grad_norm": 0.4832291305065155,
      "learning_rate": 0.002828,
      "loss": 1.3501,
      "step": 236672
    },
    {
      "epoch": 17.766303939962476,
      "grad_norm": 0.5023680925369263,
      "learning_rate": 0.002828,
      "loss": 1.3534,
      "step": 236736
    },
    {
      "epoch": 17.77110694183865,
      "grad_norm": 0.6139066815376282,
      "learning_rate": 0.002828,
      "loss": 1.3501,
      "step": 236800
    },
    {
      "epoch": 17.775909943714822,
      "grad_norm": 0.5624128580093384,
      "learning_rate": 0.002828,
      "loss": 1.3564,
      "step": 236864
    },
    {
      "epoch": 17.780712945590995,
      "grad_norm": 0.584254801273346,
      "learning_rate": 0.002828,
      "loss": 1.3519,
      "step": 236928
    },
    {
      "epoch": 17.785515947467168,
      "grad_norm": 0.549300491809845,
      "learning_rate": 0.002828,
      "loss": 1.3556,
      "step": 236992
    },
    {
      "epoch": 17.79031894934334,
      "grad_norm": 0.590458333492279,
      "learning_rate": 0.002828,
      "loss": 1.3543,
      "step": 237056
    },
    {
      "epoch": 17.795121951219514,
      "grad_norm": 0.5119503140449524,
      "learning_rate": 0.002828,
      "loss": 1.3551,
      "step": 237120
    },
    {
      "epoch": 17.799924953095683,
      "grad_norm": 0.5858562588691711,
      "learning_rate": 0.002828,
      "loss": 1.3494,
      "step": 237184
    },
    {
      "epoch": 17.804727954971856,
      "grad_norm": 0.4904852509498596,
      "learning_rate": 0.002828,
      "loss": 1.3517,
      "step": 237248
    },
    {
      "epoch": 17.80953095684803,
      "grad_norm": 0.5314335823059082,
      "learning_rate": 0.002828,
      "loss": 1.3565,
      "step": 237312
    },
    {
      "epoch": 17.814333958724202,
      "grad_norm": 0.5037870407104492,
      "learning_rate": 0.002828,
      "loss": 1.3538,
      "step": 237376
    },
    {
      "epoch": 17.819136960600375,
      "grad_norm": 0.6075197458267212,
      "learning_rate": 0.002828,
      "loss": 1.3534,
      "step": 237440
    },
    {
      "epoch": 17.823939962476548,
      "grad_norm": 0.5014629364013672,
      "learning_rate": 0.002828,
      "loss": 1.3571,
      "step": 237504
    },
    {
      "epoch": 17.82874296435272,
      "grad_norm": 0.655502438545227,
      "learning_rate": 0.002828,
      "loss": 1.3561,
      "step": 237568
    },
    {
      "epoch": 17.833545966228893,
      "grad_norm": 0.5249149203300476,
      "learning_rate": 0.002828,
      "loss": 1.3469,
      "step": 237632
    },
    {
      "epoch": 17.838348968105066,
      "grad_norm": 0.5390370488166809,
      "learning_rate": 0.002828,
      "loss": 1.3502,
      "step": 237696
    },
    {
      "epoch": 17.84315196998124,
      "grad_norm": 0.5452220439910889,
      "learning_rate": 0.002828,
      "loss": 1.3536,
      "step": 237760
    },
    {
      "epoch": 17.847954971857412,
      "grad_norm": 0.545632541179657,
      "learning_rate": 0.002828,
      "loss": 1.3593,
      "step": 237824
    },
    {
      "epoch": 17.852757973733585,
      "grad_norm": 0.6663748025894165,
      "learning_rate": 0.002828,
      "loss": 1.3566,
      "step": 237888
    },
    {
      "epoch": 17.857560975609758,
      "grad_norm": 0.5412925481796265,
      "learning_rate": 0.002828,
      "loss": 1.3586,
      "step": 237952
    },
    {
      "epoch": 17.862363977485927,
      "grad_norm": 0.5036992430686951,
      "learning_rate": 0.002828,
      "loss": 1.358,
      "step": 238016
    },
    {
      "epoch": 17.8671669793621,
      "grad_norm": 0.6046959161758423,
      "learning_rate": 0.002828,
      "loss": 1.3565,
      "step": 238080
    },
    {
      "epoch": 17.871969981238273,
      "grad_norm": 0.520389199256897,
      "learning_rate": 0.002828,
      "loss": 1.3528,
      "step": 238144
    },
    {
      "epoch": 17.876772983114446,
      "grad_norm": 0.561380922794342,
      "learning_rate": 0.002828,
      "loss": 1.353,
      "step": 238208
    },
    {
      "epoch": 17.88157598499062,
      "grad_norm": 0.5111840963363647,
      "learning_rate": 0.002828,
      "loss": 1.356,
      "step": 238272
    },
    {
      "epoch": 17.88637898686679,
      "grad_norm": 0.5611425638198853,
      "learning_rate": 0.002828,
      "loss": 1.3502,
      "step": 238336
    },
    {
      "epoch": 17.891181988742964,
      "grad_norm": 0.5181549191474915,
      "learning_rate": 0.002828,
      "loss": 1.3537,
      "step": 238400
    },
    {
      "epoch": 17.895984990619137,
      "grad_norm": 0.5285221338272095,
      "learning_rate": 0.002828,
      "loss": 1.3543,
      "step": 238464
    },
    {
      "epoch": 17.90078799249531,
      "grad_norm": 0.5775920152664185,
      "learning_rate": 0.002828,
      "loss": 1.3565,
      "step": 238528
    },
    {
      "epoch": 17.905590994371483,
      "grad_norm": 0.5790485143661499,
      "learning_rate": 0.002828,
      "loss": 1.3532,
      "step": 238592
    },
    {
      "epoch": 17.910393996247656,
      "grad_norm": 0.5824828743934631,
      "learning_rate": 0.002828,
      "loss": 1.3488,
      "step": 238656
    },
    {
      "epoch": 17.91519699812383,
      "grad_norm": 0.5291104912757874,
      "learning_rate": 0.002828,
      "loss": 1.3598,
      "step": 238720
    },
    {
      "epoch": 17.92,
      "grad_norm": 0.5468594431877136,
      "learning_rate": 0.002828,
      "loss": 1.3536,
      "step": 238784
    },
    {
      "epoch": 17.92480300187617,
      "grad_norm": 0.5960270762443542,
      "learning_rate": 0.002828,
      "loss": 1.3548,
      "step": 238848
    },
    {
      "epoch": 17.929606003752344,
      "grad_norm": 0.514122724533081,
      "learning_rate": 0.002828,
      "loss": 1.3614,
      "step": 238912
    },
    {
      "epoch": 17.934409005628517,
      "grad_norm": 0.5227863192558289,
      "learning_rate": 0.002828,
      "loss": 1.3611,
      "step": 238976
    },
    {
      "epoch": 17.93921200750469,
      "grad_norm": 0.6620153784751892,
      "learning_rate": 0.002828,
      "loss": 1.3597,
      "step": 239040
    },
    {
      "epoch": 17.944015009380863,
      "grad_norm": 0.5719172954559326,
      "learning_rate": 0.002828,
      "loss": 1.3542,
      "step": 239104
    },
    {
      "epoch": 17.948818011257035,
      "grad_norm": 0.5674269795417786,
      "learning_rate": 0.002828,
      "loss": 1.3608,
      "step": 239168
    },
    {
      "epoch": 17.95362101313321,
      "grad_norm": 0.5276614427566528,
      "learning_rate": 0.002828,
      "loss": 1.3565,
      "step": 239232
    },
    {
      "epoch": 17.95842401500938,
      "grad_norm": 0.5094176530838013,
      "learning_rate": 0.002828,
      "loss": 1.3565,
      "step": 239296
    },
    {
      "epoch": 17.963227016885554,
      "grad_norm": 0.6211153864860535,
      "learning_rate": 0.002828,
      "loss": 1.3575,
      "step": 239360
    },
    {
      "epoch": 17.968030018761727,
      "grad_norm": 0.47066572308540344,
      "learning_rate": 0.002828,
      "loss": 1.3576,
      "step": 239424
    },
    {
      "epoch": 17.9728330206379,
      "grad_norm": 0.5774534940719604,
      "learning_rate": 0.002828,
      "loss": 1.3522,
      "step": 239488
    },
    {
      "epoch": 17.977636022514073,
      "grad_norm": 0.7815809845924377,
      "learning_rate": 0.002828,
      "loss": 1.3599,
      "step": 239552
    },
    {
      "epoch": 17.982439024390246,
      "grad_norm": 0.5297728776931763,
      "learning_rate": 0.002828,
      "loss": 1.3558,
      "step": 239616
    },
    {
      "epoch": 17.987242026266415,
      "grad_norm": 0.7529789209365845,
      "learning_rate": 0.002828,
      "loss": 1.3534,
      "step": 239680
    },
    {
      "epoch": 17.992045028142588,
      "grad_norm": 0.5998978018760681,
      "learning_rate": 0.002828,
      "loss": 1.3526,
      "step": 239744
    },
    {
      "epoch": 17.99684803001876,
      "grad_norm": 0.5187614560127258,
      "learning_rate": 0.002828,
      "loss": 1.3577,
      "step": 239808
    },
    {
      "epoch": 18.001651031894934,
      "grad_norm": 0.6527639031410217,
      "learning_rate": 0.002828,
      "loss": 1.3353,
      "step": 239872
    },
    {
      "epoch": 18.006454033771107,
      "grad_norm": 0.5733532309532166,
      "learning_rate": 0.002828,
      "loss": 1.3174,
      "step": 239936
    },
    {
      "epoch": 18.01125703564728,
      "grad_norm": 0.5666289925575256,
      "learning_rate": 0.002828,
      "loss": 1.3088,
      "step": 240000
    },
    {
      "epoch": 18.016060037523452,
      "grad_norm": 0.6003232598304749,
      "learning_rate": 0.002828,
      "loss": 1.317,
      "step": 240064
    },
    {
      "epoch": 18.020863039399625,
      "grad_norm": 0.5648309588432312,
      "learning_rate": 0.002828,
      "loss": 1.3115,
      "step": 240128
    },
    {
      "epoch": 18.025666041275798,
      "grad_norm": 0.5369406342506409,
      "learning_rate": 0.002828,
      "loss": 1.3172,
      "step": 240192
    },
    {
      "epoch": 18.03046904315197,
      "grad_norm": 0.5485430955886841,
      "learning_rate": 0.002828,
      "loss": 1.3124,
      "step": 240256
    },
    {
      "epoch": 18.035272045028144,
      "grad_norm": 0.5280949473381042,
      "learning_rate": 0.002828,
      "loss": 1.3177,
      "step": 240320
    },
    {
      "epoch": 18.040075046904317,
      "grad_norm": 0.5924586057662964,
      "learning_rate": 0.002828,
      "loss": 1.3154,
      "step": 240384
    },
    {
      "epoch": 18.044878048780486,
      "grad_norm": 0.4871281683444977,
      "learning_rate": 0.002828,
      "loss": 1.311,
      "step": 240448
    },
    {
      "epoch": 18.04968105065666,
      "grad_norm": 0.5817774534225464,
      "learning_rate": 0.002828,
      "loss": 1.3153,
      "step": 240512
    },
    {
      "epoch": 18.054484052532832,
      "grad_norm": 0.5690784454345703,
      "learning_rate": 0.002828,
      "loss": 1.3202,
      "step": 240576
    },
    {
      "epoch": 18.059287054409005,
      "grad_norm": 0.601919412612915,
      "learning_rate": 0.002828,
      "loss": 1.3211,
      "step": 240640
    },
    {
      "epoch": 18.064090056285178,
      "grad_norm": 0.5639865398406982,
      "learning_rate": 0.002828,
      "loss": 1.322,
      "step": 240704
    },
    {
      "epoch": 18.06889305816135,
      "grad_norm": 0.5849490165710449,
      "learning_rate": 0.002828,
      "loss": 1.3189,
      "step": 240768
    },
    {
      "epoch": 18.073696060037523,
      "grad_norm": 0.5788787007331848,
      "learning_rate": 0.002828,
      "loss": 1.3182,
      "step": 240832
    },
    {
      "epoch": 18.078499061913696,
      "grad_norm": 0.5909736156463623,
      "learning_rate": 0.002828,
      "loss": 1.3202,
      "step": 240896
    },
    {
      "epoch": 18.08330206378987,
      "grad_norm": 0.6812686324119568,
      "learning_rate": 0.002828,
      "loss": 1.3215,
      "step": 240960
    },
    {
      "epoch": 18.088105065666042,
      "grad_norm": 0.6109818816184998,
      "learning_rate": 0.002828,
      "loss": 1.3196,
      "step": 241024
    },
    {
      "epoch": 18.092908067542215,
      "grad_norm": 0.5453113317489624,
      "learning_rate": 0.002828,
      "loss": 1.322,
      "step": 241088
    },
    {
      "epoch": 18.097711069418388,
      "grad_norm": 0.5813498497009277,
      "learning_rate": 0.002828,
      "loss": 1.3173,
      "step": 241152
    },
    {
      "epoch": 18.10251407129456,
      "grad_norm": 0.5807278156280518,
      "learning_rate": 0.002828,
      "loss": 1.3244,
      "step": 241216
    },
    {
      "epoch": 18.10731707317073,
      "grad_norm": 0.5079891085624695,
      "learning_rate": 0.002828,
      "loss": 1.3103,
      "step": 241280
    },
    {
      "epoch": 18.112120075046903,
      "grad_norm": 0.6419689059257507,
      "learning_rate": 0.002828,
      "loss": 1.3181,
      "step": 241344
    },
    {
      "epoch": 18.116923076923076,
      "grad_norm": 0.5580244660377502,
      "learning_rate": 0.002828,
      "loss": 1.3225,
      "step": 241408
    },
    {
      "epoch": 18.12172607879925,
      "grad_norm": 0.574207067489624,
      "learning_rate": 0.002828,
      "loss": 1.3197,
      "step": 241472
    },
    {
      "epoch": 18.12652908067542,
      "grad_norm": 0.580829918384552,
      "learning_rate": 0.002828,
      "loss": 1.3214,
      "step": 241536
    },
    {
      "epoch": 18.131332082551594,
      "grad_norm": 0.5580022931098938,
      "learning_rate": 0.002828,
      "loss": 1.3252,
      "step": 241600
    },
    {
      "epoch": 18.136135084427767,
      "grad_norm": 0.5125178098678589,
      "learning_rate": 0.002828,
      "loss": 1.329,
      "step": 241664
    },
    {
      "epoch": 18.14093808630394,
      "grad_norm": 0.5466482639312744,
      "learning_rate": 0.002828,
      "loss": 1.3277,
      "step": 241728
    },
    {
      "epoch": 18.145741088180113,
      "grad_norm": 0.4944307804107666,
      "learning_rate": 0.002828,
      "loss": 1.3267,
      "step": 241792
    },
    {
      "epoch": 18.150544090056286,
      "grad_norm": 0.5637335777282715,
      "learning_rate": 0.002828,
      "loss": 1.3253,
      "step": 241856
    },
    {
      "epoch": 18.15534709193246,
      "grad_norm": 0.787543535232544,
      "learning_rate": 0.002828,
      "loss": 1.322,
      "step": 241920
    },
    {
      "epoch": 18.16015009380863,
      "grad_norm": 0.534782886505127,
      "learning_rate": 0.002828,
      "loss": 1.3198,
      "step": 241984
    },
    {
      "epoch": 18.164953095684805,
      "grad_norm": 0.5230289101600647,
      "learning_rate": 0.002828,
      "loss": 1.324,
      "step": 242048
    },
    {
      "epoch": 18.169756097560974,
      "grad_norm": 0.5719302296638489,
      "learning_rate": 0.002828,
      "loss": 1.3228,
      "step": 242112
    },
    {
      "epoch": 18.174559099437147,
      "grad_norm": 0.4946596920490265,
      "learning_rate": 0.002828,
      "loss": 1.3259,
      "step": 242176
    },
    {
      "epoch": 18.17936210131332,
      "grad_norm": 0.5574563145637512,
      "learning_rate": 0.002828,
      "loss": 1.322,
      "step": 242240
    },
    {
      "epoch": 18.184165103189493,
      "grad_norm": 0.6062040328979492,
      "learning_rate": 0.002828,
      "loss": 1.3249,
      "step": 242304
    },
    {
      "epoch": 18.188968105065666,
      "grad_norm": 0.5341858267784119,
      "learning_rate": 0.002828,
      "loss": 1.3193,
      "step": 242368
    },
    {
      "epoch": 18.19377110694184,
      "grad_norm": 0.7191112041473389,
      "learning_rate": 0.002828,
      "loss": 1.3215,
      "step": 242432
    },
    {
      "epoch": 18.19857410881801,
      "grad_norm": 0.5244624018669128,
      "learning_rate": 0.002828,
      "loss": 1.3217,
      "step": 242496
    },
    {
      "epoch": 18.203377110694184,
      "grad_norm": 0.5120688676834106,
      "learning_rate": 0.002828,
      "loss": 1.3297,
      "step": 242560
    },
    {
      "epoch": 18.208180112570357,
      "grad_norm": 0.49170535802841187,
      "learning_rate": 0.002828,
      "loss": 1.3244,
      "step": 242624
    },
    {
      "epoch": 18.21298311444653,
      "grad_norm": 0.554652750492096,
      "learning_rate": 0.002828,
      "loss": 1.3202,
      "step": 242688
    },
    {
      "epoch": 18.217786116322703,
      "grad_norm": 0.573849618434906,
      "learning_rate": 0.002828,
      "loss": 1.3267,
      "step": 242752
    },
    {
      "epoch": 18.222589118198876,
      "grad_norm": 0.49524179100990295,
      "learning_rate": 0.002828,
      "loss": 1.3331,
      "step": 242816
    },
    {
      "epoch": 18.22739212007505,
      "grad_norm": 0.485775351524353,
      "learning_rate": 0.002828,
      "loss": 1.3226,
      "step": 242880
    },
    {
      "epoch": 18.232195121951218,
      "grad_norm": 0.6272198557853699,
      "learning_rate": 0.002828,
      "loss": 1.3285,
      "step": 242944
    },
    {
      "epoch": 18.23699812382739,
      "grad_norm": 0.554471492767334,
      "learning_rate": 0.002828,
      "loss": 1.3237,
      "step": 243008
    },
    {
      "epoch": 18.241801125703564,
      "grad_norm": 0.5620614290237427,
      "learning_rate": 0.002828,
      "loss": 1.3283,
      "step": 243072
    },
    {
      "epoch": 18.246604127579737,
      "grad_norm": 0.5371513366699219,
      "learning_rate": 0.002828,
      "loss": 1.3298,
      "step": 243136
    },
    {
      "epoch": 18.25140712945591,
      "grad_norm": 0.6535647511482239,
      "learning_rate": 0.002828,
      "loss": 1.3258,
      "step": 243200
    },
    {
      "epoch": 18.256210131332082,
      "grad_norm": 0.5744162201881409,
      "learning_rate": 0.002828,
      "loss": 1.3286,
      "step": 243264
    },
    {
      "epoch": 18.261013133208255,
      "grad_norm": 0.5868451595306396,
      "learning_rate": 0.002828,
      "loss": 1.3273,
      "step": 243328
    },
    {
      "epoch": 18.265816135084428,
      "grad_norm": 0.46351656317710876,
      "learning_rate": 0.002828,
      "loss": 1.3271,
      "step": 243392
    },
    {
      "epoch": 18.2706191369606,
      "grad_norm": 0.5453354716300964,
      "learning_rate": 0.002828,
      "loss": 1.3263,
      "step": 243456
    },
    {
      "epoch": 18.275422138836774,
      "grad_norm": 0.5502650737762451,
      "learning_rate": 0.002828,
      "loss": 1.328,
      "step": 243520
    },
    {
      "epoch": 18.280225140712947,
      "grad_norm": 0.6256674528121948,
      "learning_rate": 0.002828,
      "loss": 1.3327,
      "step": 243584
    },
    {
      "epoch": 18.28502814258912,
      "grad_norm": 0.6549529433250427,
      "learning_rate": 0.002828,
      "loss": 1.3266,
      "step": 243648
    },
    {
      "epoch": 18.28983114446529,
      "grad_norm": 0.6387646198272705,
      "learning_rate": 0.002828,
      "loss": 1.3337,
      "step": 243712
    },
    {
      "epoch": 18.294634146341462,
      "grad_norm": 0.529467761516571,
      "learning_rate": 0.002828,
      "loss": 1.3333,
      "step": 243776
    },
    {
      "epoch": 18.299437148217635,
      "grad_norm": 0.6114325523376465,
      "learning_rate": 0.002828,
      "loss": 1.3263,
      "step": 243840
    },
    {
      "epoch": 18.304240150093808,
      "grad_norm": 0.5541909337043762,
      "learning_rate": 0.002828,
      "loss": 1.3254,
      "step": 243904
    },
    {
      "epoch": 18.30904315196998,
      "grad_norm": 0.5705113410949707,
      "learning_rate": 0.002828,
      "loss": 1.3202,
      "step": 243968
    },
    {
      "epoch": 18.313846153846153,
      "grad_norm": 0.4778616428375244,
      "learning_rate": 0.002828,
      "loss": 1.3278,
      "step": 244032
    },
    {
      "epoch": 18.318649155722326,
      "grad_norm": 0.489709734916687,
      "learning_rate": 0.002828,
      "loss": 1.3245,
      "step": 244096
    },
    {
      "epoch": 18.3234521575985,
      "grad_norm": 0.5244088172912598,
      "learning_rate": 0.002828,
      "loss": 1.3312,
      "step": 244160
    },
    {
      "epoch": 18.328255159474672,
      "grad_norm": 0.6958155632019043,
      "learning_rate": 0.002828,
      "loss": 1.3255,
      "step": 244224
    },
    {
      "epoch": 18.333058161350845,
      "grad_norm": 0.646683394908905,
      "learning_rate": 0.002828,
      "loss": 1.3244,
      "step": 244288
    },
    {
      "epoch": 18.337861163227018,
      "grad_norm": 0.6917694807052612,
      "learning_rate": 0.002828,
      "loss": 1.3287,
      "step": 244352
    },
    {
      "epoch": 18.34266416510319,
      "grad_norm": 0.6619091629981995,
      "learning_rate": 0.002828,
      "loss": 1.325,
      "step": 244416
    },
    {
      "epoch": 18.347467166979364,
      "grad_norm": 0.6327647566795349,
      "learning_rate": 0.002828,
      "loss": 1.3373,
      "step": 244480
    },
    {
      "epoch": 18.352270168855533,
      "grad_norm": 0.537758469581604,
      "learning_rate": 0.002828,
      "loss": 1.33,
      "step": 244544
    },
    {
      "epoch": 18.357073170731706,
      "grad_norm": 0.5703405737876892,
      "learning_rate": 0.002828,
      "loss": 1.3276,
      "step": 244608
    },
    {
      "epoch": 18.36187617260788,
      "grad_norm": 0.5090919733047485,
      "learning_rate": 0.002828,
      "loss": 1.3284,
      "step": 244672
    },
    {
      "epoch": 18.36667917448405,
      "grad_norm": 0.7423458695411682,
      "learning_rate": 0.002828,
      "loss": 1.3236,
      "step": 244736
    },
    {
      "epoch": 18.371482176360225,
      "grad_norm": 0.5132669806480408,
      "learning_rate": 0.002828,
      "loss": 1.3334,
      "step": 244800
    },
    {
      "epoch": 18.376285178236397,
      "grad_norm": 0.5711482763290405,
      "learning_rate": 0.002828,
      "loss": 1.3332,
      "step": 244864
    },
    {
      "epoch": 18.38108818011257,
      "grad_norm": 0.5306158065795898,
      "learning_rate": 0.002828,
      "loss": 1.3336,
      "step": 244928
    },
    {
      "epoch": 18.385891181988743,
      "grad_norm": 0.5341092348098755,
      "learning_rate": 0.002828,
      "loss": 1.3287,
      "step": 244992
    },
    {
      "epoch": 18.390694183864916,
      "grad_norm": 0.562984049320221,
      "learning_rate": 0.002828,
      "loss": 1.3245,
      "step": 245056
    },
    {
      "epoch": 18.39549718574109,
      "grad_norm": 0.5212814807891846,
      "learning_rate": 0.002828,
      "loss": 1.3239,
      "step": 245120
    },
    {
      "epoch": 18.400300187617262,
      "grad_norm": 0.6818906664848328,
      "learning_rate": 0.002828,
      "loss": 1.3314,
      "step": 245184
    },
    {
      "epoch": 18.405103189493435,
      "grad_norm": 0.4990367889404297,
      "learning_rate": 0.002828,
      "loss": 1.329,
      "step": 245248
    },
    {
      "epoch": 18.409906191369608,
      "grad_norm": 0.5012723803520203,
      "learning_rate": 0.002828,
      "loss": 1.3373,
      "step": 245312
    },
    {
      "epoch": 18.414709193245777,
      "grad_norm": 0.6447637677192688,
      "learning_rate": 0.002828,
      "loss": 1.3288,
      "step": 245376
    },
    {
      "epoch": 18.41951219512195,
      "grad_norm": 0.6037337779998779,
      "learning_rate": 0.002828,
      "loss": 1.3304,
      "step": 245440
    },
    {
      "epoch": 18.424315196998123,
      "grad_norm": 0.6380529999732971,
      "learning_rate": 0.002828,
      "loss": 1.3302,
      "step": 245504
    },
    {
      "epoch": 18.429118198874296,
      "grad_norm": 0.6336204409599304,
      "learning_rate": 0.002828,
      "loss": 1.3312,
      "step": 245568
    },
    {
      "epoch": 18.43392120075047,
      "grad_norm": 0.5741140842437744,
      "learning_rate": 0.002828,
      "loss": 1.3269,
      "step": 245632
    },
    {
      "epoch": 18.43872420262664,
      "grad_norm": 0.4480578899383545,
      "learning_rate": 0.002828,
      "loss": 1.3392,
      "step": 245696
    },
    {
      "epoch": 18.443527204502814,
      "grad_norm": 0.5134573578834534,
      "learning_rate": 0.002828,
      "loss": 1.3299,
      "step": 245760
    },
    {
      "epoch": 18.448330206378987,
      "grad_norm": 0.5798774361610413,
      "learning_rate": 0.002828,
      "loss": 1.3284,
      "step": 245824
    },
    {
      "epoch": 18.45313320825516,
      "grad_norm": 0.49131253361701965,
      "learning_rate": 0.002828,
      "loss": 1.3329,
      "step": 245888
    },
    {
      "epoch": 18.457936210131333,
      "grad_norm": 0.64532870054245,
      "learning_rate": 0.002828,
      "loss": 1.333,
      "step": 245952
    },
    {
      "epoch": 18.462739212007506,
      "grad_norm": 0.5741368532180786,
      "learning_rate": 0.002828,
      "loss": 1.3314,
      "step": 246016
    },
    {
      "epoch": 18.46754221388368,
      "grad_norm": 0.5318816304206848,
      "learning_rate": 0.002828,
      "loss": 1.3324,
      "step": 246080
    },
    {
      "epoch": 18.47234521575985,
      "grad_norm": 0.49761515855789185,
      "learning_rate": 0.002828,
      "loss": 1.33,
      "step": 246144
    },
    {
      "epoch": 18.47714821763602,
      "grad_norm": 0.5801960825920105,
      "learning_rate": 0.002828,
      "loss": 1.3322,
      "step": 246208
    },
    {
      "epoch": 18.481951219512194,
      "grad_norm": 0.6103347539901733,
      "learning_rate": 0.002828,
      "loss": 1.3373,
      "step": 246272
    },
    {
      "epoch": 18.486754221388367,
      "grad_norm": 0.6255427598953247,
      "learning_rate": 0.002828,
      "loss": 1.3294,
      "step": 246336
    },
    {
      "epoch": 18.49155722326454,
      "grad_norm": 0.6342930197715759,
      "learning_rate": 0.002828,
      "loss": 1.3299,
      "step": 246400
    },
    {
      "epoch": 18.496360225140712,
      "grad_norm": 0.5457690358161926,
      "learning_rate": 0.002828,
      "loss": 1.3235,
      "step": 246464
    },
    {
      "epoch": 18.501163227016885,
      "grad_norm": 0.4921017289161682,
      "learning_rate": 0.002828,
      "loss": 1.3298,
      "step": 246528
    },
    {
      "epoch": 18.505966228893058,
      "grad_norm": 0.520806074142456,
      "learning_rate": 0.002828,
      "loss": 1.3272,
      "step": 246592
    },
    {
      "epoch": 18.51076923076923,
      "grad_norm": 0.49572330713272095,
      "learning_rate": 0.002828,
      "loss": 1.3307,
      "step": 246656
    },
    {
      "epoch": 18.515572232645404,
      "grad_norm": 0.6336793303489685,
      "learning_rate": 0.002828,
      "loss": 1.3293,
      "step": 246720
    },
    {
      "epoch": 18.520375234521577,
      "grad_norm": 0.5762158036231995,
      "learning_rate": 0.002828,
      "loss": 1.329,
      "step": 246784
    },
    {
      "epoch": 18.52517823639775,
      "grad_norm": 0.5626549124717712,
      "learning_rate": 0.002828,
      "loss": 1.3354,
      "step": 246848
    },
    {
      "epoch": 18.529981238273923,
      "grad_norm": 0.5641196966171265,
      "learning_rate": 0.002828,
      "loss": 1.3277,
      "step": 246912
    },
    {
      "epoch": 18.534784240150096,
      "grad_norm": 0.7018244862556458,
      "learning_rate": 0.002828,
      "loss": 1.3281,
      "step": 246976
    },
    {
      "epoch": 18.539587242026265,
      "grad_norm": 0.582954466342926,
      "learning_rate": 0.002828,
      "loss": 1.333,
      "step": 247040
    },
    {
      "epoch": 18.544390243902438,
      "grad_norm": 0.5603383183479309,
      "learning_rate": 0.002828,
      "loss": 1.3335,
      "step": 247104
    },
    {
      "epoch": 18.54919324577861,
      "grad_norm": 0.6978287696838379,
      "learning_rate": 0.002828,
      "loss": 1.3328,
      "step": 247168
    },
    {
      "epoch": 18.553996247654784,
      "grad_norm": 0.6165763139724731,
      "learning_rate": 0.002828,
      "loss": 1.3345,
      "step": 247232
    },
    {
      "epoch": 18.558799249530956,
      "grad_norm": 0.5360153913497925,
      "learning_rate": 0.002828,
      "loss": 1.3302,
      "step": 247296
    },
    {
      "epoch": 18.56360225140713,
      "grad_norm": 0.6076267957687378,
      "learning_rate": 0.002828,
      "loss": 1.3295,
      "step": 247360
    },
    {
      "epoch": 18.568405253283302,
      "grad_norm": 0.6428861021995544,
      "learning_rate": 0.002828,
      "loss": 1.3255,
      "step": 247424
    },
    {
      "epoch": 18.573208255159475,
      "grad_norm": 0.6469327211380005,
      "learning_rate": 0.002828,
      "loss": 1.3348,
      "step": 247488
    },
    {
      "epoch": 18.578011257035648,
      "grad_norm": 0.49372467398643494,
      "learning_rate": 0.002828,
      "loss": 1.3343,
      "step": 247552
    },
    {
      "epoch": 18.58281425891182,
      "grad_norm": 0.5334155559539795,
      "learning_rate": 0.002828,
      "loss": 1.3337,
      "step": 247616
    },
    {
      "epoch": 18.587617260787994,
      "grad_norm": 0.589820384979248,
      "learning_rate": 0.002828,
      "loss": 1.3345,
      "step": 247680
    },
    {
      "epoch": 18.592420262664167,
      "grad_norm": 0.5126547813415527,
      "learning_rate": 0.002828,
      "loss": 1.3337,
      "step": 247744
    },
    {
      "epoch": 18.59722326454034,
      "grad_norm": 0.5486066937446594,
      "learning_rate": 0.002828,
      "loss": 1.3365,
      "step": 247808
    },
    {
      "epoch": 18.60202626641651,
      "grad_norm": 0.48571765422821045,
      "learning_rate": 0.002828,
      "loss": 1.3337,
      "step": 247872
    },
    {
      "epoch": 18.60682926829268,
      "grad_norm": 0.5378714799880981,
      "learning_rate": 0.002828,
      "loss": 1.3348,
      "step": 247936
    },
    {
      "epoch": 18.611632270168855,
      "grad_norm": 0.5203801393508911,
      "learning_rate": 0.002828,
      "loss": 1.3332,
      "step": 248000
    },
    {
      "epoch": 18.616435272045027,
      "grad_norm": 0.5799272060394287,
      "learning_rate": 0.002828,
      "loss": 1.3319,
      "step": 248064
    },
    {
      "epoch": 18.6212382739212,
      "grad_norm": 0.5430501103401184,
      "learning_rate": 0.002828,
      "loss": 1.3387,
      "step": 248128
    },
    {
      "epoch": 18.626041275797373,
      "grad_norm": 0.5245668888092041,
      "learning_rate": 0.002828,
      "loss": 1.3368,
      "step": 248192
    },
    {
      "epoch": 18.630844277673546,
      "grad_norm": 0.5500134229660034,
      "learning_rate": 0.002828,
      "loss": 1.3378,
      "step": 248256
    },
    {
      "epoch": 18.63564727954972,
      "grad_norm": 0.6872992515563965,
      "learning_rate": 0.002828,
      "loss": 1.3332,
      "step": 248320
    },
    {
      "epoch": 18.640450281425892,
      "grad_norm": 0.510874330997467,
      "learning_rate": 0.002828,
      "loss": 1.3332,
      "step": 248384
    },
    {
      "epoch": 18.645253283302065,
      "grad_norm": 0.5345406532287598,
      "learning_rate": 0.002828,
      "loss": 1.3368,
      "step": 248448
    },
    {
      "epoch": 18.650056285178238,
      "grad_norm": 0.5812759399414062,
      "learning_rate": 0.002828,
      "loss": 1.3316,
      "step": 248512
    },
    {
      "epoch": 18.65485928705441,
      "grad_norm": 0.5518085360527039,
      "learning_rate": 0.002828,
      "loss": 1.3394,
      "step": 248576
    },
    {
      "epoch": 18.659662288930583,
      "grad_norm": 0.6124371886253357,
      "learning_rate": 0.002828,
      "loss": 1.3317,
      "step": 248640
    },
    {
      "epoch": 18.664465290806753,
      "grad_norm": 0.5146282911300659,
      "learning_rate": 0.002828,
      "loss": 1.331,
      "step": 248704
    },
    {
      "epoch": 18.669268292682926,
      "grad_norm": 0.5922949314117432,
      "learning_rate": 0.002828,
      "loss": 1.3372,
      "step": 248768
    },
    {
      "epoch": 18.6740712945591,
      "grad_norm": 0.5442377924919128,
      "learning_rate": 0.002828,
      "loss": 1.334,
      "step": 248832
    },
    {
      "epoch": 18.67887429643527,
      "grad_norm": 0.561581552028656,
      "learning_rate": 0.002828,
      "loss": 1.3375,
      "step": 248896
    },
    {
      "epoch": 18.683677298311444,
      "grad_norm": 0.546955406665802,
      "learning_rate": 0.002828,
      "loss": 1.3296,
      "step": 248960
    },
    {
      "epoch": 18.688480300187617,
      "grad_norm": 0.531005322933197,
      "learning_rate": 0.002828,
      "loss": 1.3318,
      "step": 249024
    },
    {
      "epoch": 18.69328330206379,
      "grad_norm": 0.5529969334602356,
      "learning_rate": 0.002828,
      "loss": 1.3347,
      "step": 249088
    },
    {
      "epoch": 18.698086303939963,
      "grad_norm": 0.6071882843971252,
      "learning_rate": 0.002828,
      "loss": 1.3339,
      "step": 249152
    },
    {
      "epoch": 18.702889305816136,
      "grad_norm": 0.5620871782302856,
      "learning_rate": 0.002828,
      "loss": 1.3388,
      "step": 249216
    },
    {
      "epoch": 18.70769230769231,
      "grad_norm": 0.6317915320396423,
      "learning_rate": 0.002828,
      "loss": 1.34,
      "step": 249280
    },
    {
      "epoch": 18.71249530956848,
      "grad_norm": 0.5283500552177429,
      "learning_rate": 0.002828,
      "loss": 1.3445,
      "step": 249344
    },
    {
      "epoch": 18.717298311444655,
      "grad_norm": 0.5978621244430542,
      "learning_rate": 0.002828,
      "loss": 1.3301,
      "step": 249408
    },
    {
      "epoch": 18.722101313320824,
      "grad_norm": 0.5326447486877441,
      "learning_rate": 0.002828,
      "loss": 1.3356,
      "step": 249472
    },
    {
      "epoch": 18.726904315196997,
      "grad_norm": 0.49449408054351807,
      "learning_rate": 0.002828,
      "loss": 1.3377,
      "step": 249536
    },
    {
      "epoch": 18.73170731707317,
      "grad_norm": 0.5880950689315796,
      "learning_rate": 0.002828,
      "loss": 1.332,
      "step": 249600
    },
    {
      "epoch": 18.736510318949342,
      "grad_norm": 0.513163685798645,
      "learning_rate": 0.002828,
      "loss": 1.3346,
      "step": 249664
    },
    {
      "epoch": 18.741313320825515,
      "grad_norm": 0.5993733406066895,
      "learning_rate": 0.002828,
      "loss": 1.3402,
      "step": 249728
    },
    {
      "epoch": 18.74611632270169,
      "grad_norm": 0.581052839756012,
      "learning_rate": 0.002828,
      "loss": 1.3343,
      "step": 249792
    },
    {
      "epoch": 18.75091932457786,
      "grad_norm": 0.5721963047981262,
      "learning_rate": 0.002828,
      "loss": 1.3405,
      "step": 249856
    },
    {
      "epoch": 18.755722326454034,
      "grad_norm": 0.612507700920105,
      "learning_rate": 0.002828,
      "loss": 1.3327,
      "step": 249920
    },
    {
      "epoch": 18.760525328330207,
      "grad_norm": 0.6289708614349365,
      "learning_rate": 0.002828,
      "loss": 1.3423,
      "step": 249984
    },
    {
      "epoch": 18.76532833020638,
      "grad_norm": 0.5146111845970154,
      "learning_rate": 0.002828,
      "loss": 1.3331,
      "step": 250048
    },
    {
      "epoch": 18.770131332082553,
      "grad_norm": 0.5346642136573792,
      "learning_rate": 0.002828,
      "loss": 1.3326,
      "step": 250112
    },
    {
      "epoch": 18.774934333958726,
      "grad_norm": 0.4899759292602539,
      "learning_rate": 0.002828,
      "loss": 1.3313,
      "step": 250176
    },
    {
      "epoch": 18.7797373358349,
      "grad_norm": 0.5708340406417847,
      "learning_rate": 0.002828,
      "loss": 1.3411,
      "step": 250240
    },
    {
      "epoch": 18.784540337711068,
      "grad_norm": 0.5400072932243347,
      "learning_rate": 0.002828,
      "loss": 1.3393,
      "step": 250304
    },
    {
      "epoch": 18.78934333958724,
      "grad_norm": 0.5824398398399353,
      "learning_rate": 0.002828,
      "loss": 1.3345,
      "step": 250368
    },
    {
      "epoch": 18.794146341463414,
      "grad_norm": 0.6986384987831116,
      "learning_rate": 0.002828,
      "loss": 1.3321,
      "step": 250432
    },
    {
      "epoch": 18.798949343339586,
      "grad_norm": 0.6101935505867004,
      "learning_rate": 0.002828,
      "loss": 1.3393,
      "step": 250496
    },
    {
      "epoch": 18.80375234521576,
      "grad_norm": 0.6558166146278381,
      "learning_rate": 0.002828,
      "loss": 1.336,
      "step": 250560
    },
    {
      "epoch": 18.808555347091932,
      "grad_norm": 0.526673436164856,
      "learning_rate": 0.002828,
      "loss": 1.3294,
      "step": 250624
    },
    {
      "epoch": 18.813358348968105,
      "grad_norm": 0.48416274785995483,
      "learning_rate": 0.002828,
      "loss": 1.3446,
      "step": 250688
    },
    {
      "epoch": 18.818161350844278,
      "grad_norm": 0.5560637712478638,
      "learning_rate": 0.002828,
      "loss": 1.3373,
      "step": 250752
    },
    {
      "epoch": 18.82296435272045,
      "grad_norm": 0.572029173374176,
      "learning_rate": 0.002828,
      "loss": 1.3367,
      "step": 250816
    },
    {
      "epoch": 18.827767354596624,
      "grad_norm": 0.4713887572288513,
      "learning_rate": 0.002828,
      "loss": 1.3418,
      "step": 250880
    },
    {
      "epoch": 18.832570356472797,
      "grad_norm": 0.5924293398857117,
      "learning_rate": 0.002828,
      "loss": 1.332,
      "step": 250944
    },
    {
      "epoch": 18.83737335834897,
      "grad_norm": 0.6764249205589294,
      "learning_rate": 0.002828,
      "loss": 1.3407,
      "step": 251008
    },
    {
      "epoch": 18.842176360225142,
      "grad_norm": 0.5857234001159668,
      "learning_rate": 0.002828,
      "loss": 1.3359,
      "step": 251072
    },
    {
      "epoch": 18.84697936210131,
      "grad_norm": 0.6040510535240173,
      "learning_rate": 0.002828,
      "loss": 1.3356,
      "step": 251136
    },
    {
      "epoch": 18.851782363977485,
      "grad_norm": 0.49172133207321167,
      "learning_rate": 0.002828,
      "loss": 1.3361,
      "step": 251200
    },
    {
      "epoch": 18.856585365853658,
      "grad_norm": 0.5720678567886353,
      "learning_rate": 0.002828,
      "loss": 1.3394,
      "step": 251264
    },
    {
      "epoch": 18.86138836772983,
      "grad_norm": 0.6578644514083862,
      "learning_rate": 0.002828,
      "loss": 1.3347,
      "step": 251328
    },
    {
      "epoch": 18.866191369606003,
      "grad_norm": 0.5127058029174805,
      "learning_rate": 0.002828,
      "loss": 1.3417,
      "step": 251392
    },
    {
      "epoch": 18.870994371482176,
      "grad_norm": 0.5229156613349915,
      "learning_rate": 0.002828,
      "loss": 1.3362,
      "step": 251456
    },
    {
      "epoch": 18.87579737335835,
      "grad_norm": 0.5593118667602539,
      "learning_rate": 0.002828,
      "loss": 1.3402,
      "step": 251520
    },
    {
      "epoch": 18.880600375234522,
      "grad_norm": 0.648236870765686,
      "learning_rate": 0.002828,
      "loss": 1.336,
      "step": 251584
    },
    {
      "epoch": 18.885403377110695,
      "grad_norm": 0.584019660949707,
      "learning_rate": 0.002828,
      "loss": 1.3407,
      "step": 251648
    },
    {
      "epoch": 18.890206378986868,
      "grad_norm": 0.5781084299087524,
      "learning_rate": 0.002828,
      "loss": 1.3381,
      "step": 251712
    },
    {
      "epoch": 18.89500938086304,
      "grad_norm": 0.5639113783836365,
      "learning_rate": 0.002828,
      "loss": 1.3416,
      "step": 251776
    },
    {
      "epoch": 18.899812382739213,
      "grad_norm": 0.5335075259208679,
      "learning_rate": 0.002828,
      "loss": 1.3373,
      "step": 251840
    },
    {
      "epoch": 18.904615384615383,
      "grad_norm": 0.5257242321968079,
      "learning_rate": 0.002828,
      "loss": 1.3389,
      "step": 251904
    },
    {
      "epoch": 18.909418386491556,
      "grad_norm": 0.5944527387619019,
      "learning_rate": 0.002828,
      "loss": 1.335,
      "step": 251968
    },
    {
      "epoch": 18.91422138836773,
      "grad_norm": 0.5301898121833801,
      "learning_rate": 0.002828,
      "loss": 1.3346,
      "step": 252032
    },
    {
      "epoch": 18.9190243902439,
      "grad_norm": 0.5499849915504456,
      "learning_rate": 0.002828,
      "loss": 1.3382,
      "step": 252096
    },
    {
      "epoch": 18.923827392120074,
      "grad_norm": 0.5929718017578125,
      "learning_rate": 0.002828,
      "loss": 1.3324,
      "step": 252160
    },
    {
      "epoch": 18.928630393996247,
      "grad_norm": 0.6182397603988647,
      "learning_rate": 0.002828,
      "loss": 1.3444,
      "step": 252224
    },
    {
      "epoch": 18.93343339587242,
      "grad_norm": 0.5975897908210754,
      "learning_rate": 0.002828,
      "loss": 1.3331,
      "step": 252288
    },
    {
      "epoch": 18.938236397748593,
      "grad_norm": 0.5303201675415039,
      "learning_rate": 0.002828,
      "loss": 1.3354,
      "step": 252352
    },
    {
      "epoch": 18.943039399624766,
      "grad_norm": 0.5473834872245789,
      "learning_rate": 0.002828,
      "loss": 1.3406,
      "step": 252416
    },
    {
      "epoch": 18.94784240150094,
      "grad_norm": 0.5609472990036011,
      "learning_rate": 0.002828,
      "loss": 1.3362,
      "step": 252480
    },
    {
      "epoch": 18.95264540337711,
      "grad_norm": 0.6121054291725159,
      "learning_rate": 0.002828,
      "loss": 1.341,
      "step": 252544
    },
    {
      "epoch": 18.957448405253285,
      "grad_norm": 0.5182227492332458,
      "learning_rate": 0.002828,
      "loss": 1.3422,
      "step": 252608
    },
    {
      "epoch": 18.962251407129457,
      "grad_norm": 0.5840742588043213,
      "learning_rate": 0.002828,
      "loss": 1.3372,
      "step": 252672
    },
    {
      "epoch": 18.967054409005627,
      "grad_norm": 0.5088547468185425,
      "learning_rate": 0.002828,
      "loss": 1.34,
      "step": 252736
    },
    {
      "epoch": 18.9718574108818,
      "grad_norm": 0.63606196641922,
      "learning_rate": 0.002828,
      "loss": 1.3368,
      "step": 252800
    },
    {
      "epoch": 18.976660412757973,
      "grad_norm": 0.5369198322296143,
      "learning_rate": 0.002828,
      "loss": 1.3366,
      "step": 252864
    },
    {
      "epoch": 18.981463414634145,
      "grad_norm": 0.5701738595962524,
      "learning_rate": 0.002828,
      "loss": 1.3416,
      "step": 252928
    },
    {
      "epoch": 18.98626641651032,
      "grad_norm": 0.7686113119125366,
      "learning_rate": 0.002828,
      "loss": 1.3326,
      "step": 252992
    },
    {
      "epoch": 18.99106941838649,
      "grad_norm": 0.5373234152793884,
      "learning_rate": 0.002828,
      "loss": 1.3331,
      "step": 253056
    },
    {
      "epoch": 18.995872420262664,
      "grad_norm": 0.5551041960716248,
      "learning_rate": 0.002828,
      "loss": 1.3396,
      "step": 253120
    },
    {
      "epoch": 19.000675422138837,
      "grad_norm": 0.47208505868911743,
      "learning_rate": 0.002828,
      "loss": 1.3379,
      "step": 253184
    },
    {
      "epoch": 19.00547842401501,
      "grad_norm": 0.6245517730712891,
      "learning_rate": 0.002828,
      "loss": 1.2966,
      "step": 253248
    },
    {
      "epoch": 19.010281425891183,
      "grad_norm": 0.5484210252761841,
      "learning_rate": 0.002828,
      "loss": 1.3004,
      "step": 253312
    },
    {
      "epoch": 19.015084427767356,
      "grad_norm": 0.583123505115509,
      "learning_rate": 0.002828,
      "loss": 1.2976,
      "step": 253376
    },
    {
      "epoch": 19.01988742964353,
      "grad_norm": 0.5009977221488953,
      "learning_rate": 0.002828,
      "loss": 1.3014,
      "step": 253440
    },
    {
      "epoch": 19.0246904315197,
      "grad_norm": 0.5530526638031006,
      "learning_rate": 0.002828,
      "loss": 1.2933,
      "step": 253504
    },
    {
      "epoch": 19.02949343339587,
      "grad_norm": 0.5873714089393616,
      "learning_rate": 0.002828,
      "loss": 1.2982,
      "step": 253568
    },
    {
      "epoch": 19.034296435272044,
      "grad_norm": 0.6122865080833435,
      "learning_rate": 0.002828,
      "loss": 1.2976,
      "step": 253632
    },
    {
      "epoch": 19.039099437148217,
      "grad_norm": 0.5970596671104431,
      "learning_rate": 0.002828,
      "loss": 1.3003,
      "step": 253696
    },
    {
      "epoch": 19.04390243902439,
      "grad_norm": 0.5157353281974792,
      "learning_rate": 0.002828,
      "loss": 1.3001,
      "step": 253760
    },
    {
      "epoch": 19.048705440900562,
      "grad_norm": 0.5586423873901367,
      "learning_rate": 0.002828,
      "loss": 1.2937,
      "step": 253824
    },
    {
      "epoch": 19.053508442776735,
      "grad_norm": 0.588364839553833,
      "learning_rate": 0.002828,
      "loss": 1.2976,
      "step": 253888
    },
    {
      "epoch": 19.058311444652908,
      "grad_norm": 0.6247104406356812,
      "learning_rate": 0.002828,
      "loss": 1.2913,
      "step": 253952
    },
    {
      "epoch": 19.06311444652908,
      "grad_norm": 0.6078224182128906,
      "learning_rate": 0.002828,
      "loss": 1.3058,
      "step": 254016
    },
    {
      "epoch": 19.067917448405254,
      "grad_norm": 0.5876947045326233,
      "learning_rate": 0.002828,
      "loss": 1.3009,
      "step": 254080
    },
    {
      "epoch": 19.072720450281427,
      "grad_norm": 0.5504317879676819,
      "learning_rate": 0.002828,
      "loss": 1.3055,
      "step": 254144
    },
    {
      "epoch": 19.0775234521576,
      "grad_norm": 0.5758258104324341,
      "learning_rate": 0.002828,
      "loss": 1.3036,
      "step": 254208
    },
    {
      "epoch": 19.082326454033772,
      "grad_norm": 0.6489672064781189,
      "learning_rate": 0.002828,
      "loss": 1.3009,
      "step": 254272
    },
    {
      "epoch": 19.087129455909945,
      "grad_norm": 0.41989511251449585,
      "learning_rate": 0.002828,
      "loss": 1.3018,
      "step": 254336
    },
    {
      "epoch": 19.091932457786115,
      "grad_norm": 0.48466578125953674,
      "learning_rate": 0.002828,
      "loss": 1.3025,
      "step": 254400
    },
    {
      "epoch": 19.096735459662288,
      "grad_norm": 0.5625917315483093,
      "learning_rate": 0.002828,
      "loss": 1.3011,
      "step": 254464
    },
    {
      "epoch": 19.10153846153846,
      "grad_norm": 0.7944159507751465,
      "learning_rate": 0.002828,
      "loss": 1.3037,
      "step": 254528
    },
    {
      "epoch": 19.106341463414633,
      "grad_norm": 0.612187385559082,
      "learning_rate": 0.002828,
      "loss": 1.3056,
      "step": 254592
    },
    {
      "epoch": 19.111144465290806,
      "grad_norm": 0.48978328704833984,
      "learning_rate": 0.002828,
      "loss": 1.3029,
      "step": 254656
    },
    {
      "epoch": 19.11594746716698,
      "grad_norm": 0.5934069156646729,
      "learning_rate": 0.002828,
      "loss": 1.3038,
      "step": 254720
    },
    {
      "epoch": 19.120750469043152,
      "grad_norm": 0.5898623466491699,
      "learning_rate": 0.002828,
      "loss": 1.2984,
      "step": 254784
    },
    {
      "epoch": 19.125553470919325,
      "grad_norm": 0.5596930384635925,
      "learning_rate": 0.002828,
      "loss": 1.3055,
      "step": 254848
    },
    {
      "epoch": 19.130356472795498,
      "grad_norm": 0.6557482481002808,
      "learning_rate": 0.002828,
      "loss": 1.3019,
      "step": 254912
    },
    {
      "epoch": 19.13515947467167,
      "grad_norm": 0.5334518551826477,
      "learning_rate": 0.002828,
      "loss": 1.3057,
      "step": 254976
    },
    {
      "epoch": 19.139962476547844,
      "grad_norm": 0.49692779779434204,
      "learning_rate": 0.002828,
      "loss": 1.3074,
      "step": 255040
    },
    {
      "epoch": 19.144765478424016,
      "grad_norm": 0.5714998245239258,
      "learning_rate": 0.002828,
      "loss": 1.3107,
      "step": 255104
    },
    {
      "epoch": 19.14956848030019,
      "grad_norm": 0.5583386421203613,
      "learning_rate": 0.002828,
      "loss": 1.303,
      "step": 255168
    },
    {
      "epoch": 19.15437148217636,
      "grad_norm": 0.5013801455497742,
      "learning_rate": 0.002828,
      "loss": 1.301,
      "step": 255232
    },
    {
      "epoch": 19.15917448405253,
      "grad_norm": 0.5694747567176819,
      "learning_rate": 0.002828,
      "loss": 1.3118,
      "step": 255296
    },
    {
      "epoch": 19.163977485928704,
      "grad_norm": 0.5212017893791199,
      "learning_rate": 0.002828,
      "loss": 1.3072,
      "step": 255360
    },
    {
      "epoch": 19.168780487804877,
      "grad_norm": 0.6951311230659485,
      "learning_rate": 0.002828,
      "loss": 1.3139,
      "step": 255424
    },
    {
      "epoch": 19.17358348968105,
      "grad_norm": 0.6076920628547668,
      "learning_rate": 0.002828,
      "loss": 1.3021,
      "step": 255488
    },
    {
      "epoch": 19.178386491557223,
      "grad_norm": 0.6477721929550171,
      "learning_rate": 0.002828,
      "loss": 1.3085,
      "step": 255552
    },
    {
      "epoch": 19.183189493433396,
      "grad_norm": 0.5557982325553894,
      "learning_rate": 0.002828,
      "loss": 1.3043,
      "step": 255616
    },
    {
      "epoch": 19.18799249530957,
      "grad_norm": 0.6123504638671875,
      "learning_rate": 0.002828,
      "loss": 1.3,
      "step": 255680
    },
    {
      "epoch": 19.19279549718574,
      "grad_norm": 0.5342661142349243,
      "learning_rate": 0.002828,
      "loss": 1.3066,
      "step": 255744
    },
    {
      "epoch": 19.197598499061915,
      "grad_norm": 0.5250298976898193,
      "learning_rate": 0.002828,
      "loss": 1.3057,
      "step": 255808
    },
    {
      "epoch": 19.202401500938088,
      "grad_norm": 0.577377438545227,
      "learning_rate": 0.002828,
      "loss": 1.3101,
      "step": 255872
    },
    {
      "epoch": 19.20720450281426,
      "grad_norm": 0.47325846552848816,
      "learning_rate": 0.002828,
      "loss": 1.3117,
      "step": 255936
    },
    {
      "epoch": 19.212007504690433,
      "grad_norm": 0.5749341249465942,
      "learning_rate": 0.002828,
      "loss": 1.3073,
      "step": 256000
    },
    {
      "epoch": 19.216810506566603,
      "grad_norm": 0.6236361861228943,
      "learning_rate": 0.002828,
      "loss": 1.299,
      "step": 256064
    },
    {
      "epoch": 19.221613508442775,
      "grad_norm": 0.5049059987068176,
      "learning_rate": 0.002828,
      "loss": 1.3105,
      "step": 256128
    },
    {
      "epoch": 19.22641651031895,
      "grad_norm": 0.5014175772666931,
      "learning_rate": 0.002828,
      "loss": 1.3136,
      "step": 256192
    },
    {
      "epoch": 19.23121951219512,
      "grad_norm": 0.6570005416870117,
      "learning_rate": 0.002828,
      "loss": 1.3021,
      "step": 256256
    },
    {
      "epoch": 19.236022514071294,
      "grad_norm": 0.5768321752548218,
      "learning_rate": 0.002828,
      "loss": 1.3102,
      "step": 256320
    },
    {
      "epoch": 19.240825515947467,
      "grad_norm": 0.5796010494232178,
      "learning_rate": 0.002828,
      "loss": 1.3108,
      "step": 256384
    },
    {
      "epoch": 19.24562851782364,
      "grad_norm": 0.5883336663246155,
      "learning_rate": 0.002828,
      "loss": 1.3029,
      "step": 256448
    },
    {
      "epoch": 19.250431519699813,
      "grad_norm": 0.5516217947006226,
      "learning_rate": 0.002828,
      "loss": 1.3038,
      "step": 256512
    },
    {
      "epoch": 19.255234521575986,
      "grad_norm": 0.5372753143310547,
      "learning_rate": 0.002828,
      "loss": 1.3082,
      "step": 256576
    },
    {
      "epoch": 19.26003752345216,
      "grad_norm": 0.5117861032485962,
      "learning_rate": 0.002828,
      "loss": 1.3106,
      "step": 256640
    },
    {
      "epoch": 19.26484052532833,
      "grad_norm": 0.5282958149909973,
      "learning_rate": 0.002828,
      "loss": 1.3051,
      "step": 256704
    },
    {
      "epoch": 19.269643527204504,
      "grad_norm": 0.5638288855552673,
      "learning_rate": 0.002828,
      "loss": 1.3112,
      "step": 256768
    },
    {
      "epoch": 19.274446529080674,
      "grad_norm": 0.5715838074684143,
      "learning_rate": 0.002828,
      "loss": 1.3119,
      "step": 256832
    },
    {
      "epoch": 19.279249530956847,
      "grad_norm": 0.5761364698410034,
      "learning_rate": 0.002828,
      "loss": 1.3069,
      "step": 256896
    },
    {
      "epoch": 19.28405253283302,
      "grad_norm": 0.5988823771476746,
      "learning_rate": 0.002828,
      "loss": 1.3132,
      "step": 256960
    },
    {
      "epoch": 19.288855534709192,
      "grad_norm": 0.7384783029556274,
      "learning_rate": 0.002828,
      "loss": 1.3094,
      "step": 257024
    },
    {
      "epoch": 19.293658536585365,
      "grad_norm": 0.5268698930740356,
      "learning_rate": 0.002828,
      "loss": 1.3098,
      "step": 257088
    },
    {
      "epoch": 19.298461538461538,
      "grad_norm": 0.6224947571754456,
      "learning_rate": 0.002828,
      "loss": 1.3122,
      "step": 257152
    },
    {
      "epoch": 19.30326454033771,
      "grad_norm": 0.5584701895713806,
      "learning_rate": 0.002828,
      "loss": 1.3127,
      "step": 257216
    },
    {
      "epoch": 19.308067542213884,
      "grad_norm": 0.5066078305244446,
      "learning_rate": 0.002828,
      "loss": 1.3117,
      "step": 257280
    },
    {
      "epoch": 19.312870544090057,
      "grad_norm": 0.6613951325416565,
      "learning_rate": 0.002828,
      "loss": 1.3134,
      "step": 257344
    },
    {
      "epoch": 19.31767354596623,
      "grad_norm": 0.4861260652542114,
      "learning_rate": 0.002828,
      "loss": 1.3143,
      "step": 257408
    },
    {
      "epoch": 19.322476547842403,
      "grad_norm": 0.5384355783462524,
      "learning_rate": 0.002828,
      "loss": 1.3144,
      "step": 257472
    },
    {
      "epoch": 19.327279549718575,
      "grad_norm": 0.6605150699615479,
      "learning_rate": 0.002828,
      "loss": 1.309,
      "step": 257536
    },
    {
      "epoch": 19.33208255159475,
      "grad_norm": 0.5852476358413696,
      "learning_rate": 0.002828,
      "loss": 1.3083,
      "step": 257600
    },
    {
      "epoch": 19.336885553470918,
      "grad_norm": 0.6060090065002441,
      "learning_rate": 0.002828,
      "loss": 1.3171,
      "step": 257664
    },
    {
      "epoch": 19.34168855534709,
      "grad_norm": 0.5727629661560059,
      "learning_rate": 0.002828,
      "loss": 1.3078,
      "step": 257728
    },
    {
      "epoch": 19.346491557223263,
      "grad_norm": 0.7055615782737732,
      "learning_rate": 0.002828,
      "loss": 1.3177,
      "step": 257792
    },
    {
      "epoch": 19.351294559099436,
      "grad_norm": 0.5629110932350159,
      "learning_rate": 0.002828,
      "loss": 1.3113,
      "step": 257856
    },
    {
      "epoch": 19.35609756097561,
      "grad_norm": 0.5396313667297363,
      "learning_rate": 0.002828,
      "loss": 1.313,
      "step": 257920
    },
    {
      "epoch": 19.360900562851782,
      "grad_norm": 0.5892098546028137,
      "learning_rate": 0.002828,
      "loss": 1.3139,
      "step": 257984
    },
    {
      "epoch": 19.365703564727955,
      "grad_norm": 0.5580728650093079,
      "learning_rate": 0.002828,
      "loss": 1.3034,
      "step": 258048
    },
    {
      "epoch": 19.370506566604128,
      "grad_norm": 0.561872124671936,
      "learning_rate": 0.002828,
      "loss": 1.3134,
      "step": 258112
    },
    {
      "epoch": 19.3753095684803,
      "grad_norm": 0.4941599667072296,
      "learning_rate": 0.002828,
      "loss": 1.3119,
      "step": 258176
    },
    {
      "epoch": 19.380112570356474,
      "grad_norm": 0.6258257627487183,
      "learning_rate": 0.002828,
      "loss": 1.3164,
      "step": 258240
    },
    {
      "epoch": 19.384915572232646,
      "grad_norm": 0.5165227651596069,
      "learning_rate": 0.002828,
      "loss": 1.3111,
      "step": 258304
    },
    {
      "epoch": 19.38971857410882,
      "grad_norm": 0.6330660581588745,
      "learning_rate": 0.002828,
      "loss": 1.3124,
      "step": 258368
    },
    {
      "epoch": 19.394521575984992,
      "grad_norm": 0.6326627135276794,
      "learning_rate": 0.002828,
      "loss": 1.3093,
      "step": 258432
    },
    {
      "epoch": 19.39932457786116,
      "grad_norm": 0.5849427580833435,
      "learning_rate": 0.002828,
      "loss": 1.3093,
      "step": 258496
    },
    {
      "epoch": 19.404127579737334,
      "grad_norm": 0.566590428352356,
      "learning_rate": 0.002828,
      "loss": 1.3096,
      "step": 258560
    },
    {
      "epoch": 19.408930581613507,
      "grad_norm": 0.5885806679725647,
      "learning_rate": 0.002828,
      "loss": 1.3137,
      "step": 258624
    },
    {
      "epoch": 19.41373358348968,
      "grad_norm": 0.5910857319831848,
      "learning_rate": 0.002828,
      "loss": 1.3125,
      "step": 258688
    },
    {
      "epoch": 19.418536585365853,
      "grad_norm": 0.6003797650337219,
      "learning_rate": 0.002828,
      "loss": 1.3155,
      "step": 258752
    },
    {
      "epoch": 19.423339587242026,
      "grad_norm": 0.5352999567985535,
      "learning_rate": 0.002828,
      "loss": 1.3114,
      "step": 258816
    },
    {
      "epoch": 19.4281425891182,
      "grad_norm": 0.5617457032203674,
      "learning_rate": 0.002828,
      "loss": 1.3167,
      "step": 258880
    },
    {
      "epoch": 19.432945590994372,
      "grad_norm": 0.6122784614562988,
      "learning_rate": 0.002828,
      "loss": 1.308,
      "step": 258944
    },
    {
      "epoch": 19.437748592870545,
      "grad_norm": 0.5012685656547546,
      "learning_rate": 0.002828,
      "loss": 1.3183,
      "step": 259008
    },
    {
      "epoch": 19.442551594746718,
      "grad_norm": 0.6228601336479187,
      "learning_rate": 0.002828,
      "loss": 1.312,
      "step": 259072
    },
    {
      "epoch": 19.44735459662289,
      "grad_norm": 0.6101039052009583,
      "learning_rate": 0.002828,
      "loss": 1.3033,
      "step": 259136
    },
    {
      "epoch": 19.452157598499063,
      "grad_norm": 0.5090866088867188,
      "learning_rate": 0.002828,
      "loss": 1.3085,
      "step": 259200
    },
    {
      "epoch": 19.456960600375236,
      "grad_norm": 0.5848556756973267,
      "learning_rate": 0.002828,
      "loss": 1.3176,
      "step": 259264
    },
    {
      "epoch": 19.461763602251406,
      "grad_norm": 0.6140968799591064,
      "learning_rate": 0.002828,
      "loss": 1.3206,
      "step": 259328
    },
    {
      "epoch": 19.46656660412758,
      "grad_norm": 0.6449792385101318,
      "learning_rate": 0.002828,
      "loss": 1.319,
      "step": 259392
    },
    {
      "epoch": 19.47136960600375,
      "grad_norm": 0.5546607375144958,
      "learning_rate": 0.002828,
      "loss": 1.3096,
      "step": 259456
    },
    {
      "epoch": 19.476172607879924,
      "grad_norm": 0.6009331345558167,
      "learning_rate": 0.002828,
      "loss": 1.3136,
      "step": 259520
    },
    {
      "epoch": 19.480975609756097,
      "grad_norm": 0.6098905801773071,
      "learning_rate": 0.002828,
      "loss": 1.3142,
      "step": 259584
    },
    {
      "epoch": 19.48577861163227,
      "grad_norm": 0.5157196521759033,
      "learning_rate": 0.002828,
      "loss": 1.3093,
      "step": 259648
    },
    {
      "epoch": 19.490581613508443,
      "grad_norm": 0.6030533313751221,
      "learning_rate": 0.002828,
      "loss": 1.3119,
      "step": 259712
    },
    {
      "epoch": 19.495384615384616,
      "grad_norm": 0.6334670186042786,
      "learning_rate": 0.002828,
      "loss": 1.3147,
      "step": 259776
    },
    {
      "epoch": 19.50018761726079,
      "grad_norm": 0.6147441267967224,
      "learning_rate": 0.002828,
      "loss": 1.3148,
      "step": 259840
    },
    {
      "epoch": 19.50499061913696,
      "grad_norm": 0.49021661281585693,
      "learning_rate": 0.002828,
      "loss": 1.3183,
      "step": 259904
    },
    {
      "epoch": 19.509793621013134,
      "grad_norm": 0.540419340133667,
      "learning_rate": 0.002828,
      "loss": 1.3155,
      "step": 259968
    },
    {
      "epoch": 19.514596622889307,
      "grad_norm": 0.5634638667106628,
      "learning_rate": 0.002828,
      "loss": 1.3161,
      "step": 260032
    },
    {
      "epoch": 19.519399624765477,
      "grad_norm": 0.6089213490486145,
      "learning_rate": 0.002828,
      "loss": 1.3147,
      "step": 260096
    },
    {
      "epoch": 19.52420262664165,
      "grad_norm": 0.488808274269104,
      "learning_rate": 0.002828,
      "loss": 1.3174,
      "step": 260160
    },
    {
      "epoch": 19.529005628517822,
      "grad_norm": 0.5810720324516296,
      "learning_rate": 0.002828,
      "loss": 1.3131,
      "step": 260224
    },
    {
      "epoch": 19.533808630393995,
      "grad_norm": 0.5251587629318237,
      "learning_rate": 0.002828,
      "loss": 1.3177,
      "step": 260288
    },
    {
      "epoch": 19.538611632270168,
      "grad_norm": 0.5949933528900146,
      "learning_rate": 0.002828,
      "loss": 1.3114,
      "step": 260352
    },
    {
      "epoch": 19.54341463414634,
      "grad_norm": 0.5253098607063293,
      "learning_rate": 0.002828,
      "loss": 1.3075,
      "step": 260416
    },
    {
      "epoch": 19.548217636022514,
      "grad_norm": 0.4630920886993408,
      "learning_rate": 0.002828,
      "loss": 1.3176,
      "step": 260480
    },
    {
      "epoch": 19.553020637898687,
      "grad_norm": 0.5550661683082581,
      "learning_rate": 0.002828,
      "loss": 1.317,
      "step": 260544
    },
    {
      "epoch": 19.55782363977486,
      "grad_norm": 0.50571608543396,
      "learning_rate": 0.002828,
      "loss": 1.3135,
      "step": 260608
    },
    {
      "epoch": 19.562626641651033,
      "grad_norm": 0.5230061411857605,
      "learning_rate": 0.002828,
      "loss": 1.318,
      "step": 260672
    },
    {
      "epoch": 19.567429643527205,
      "grad_norm": 0.4942905306816101,
      "learning_rate": 0.002828,
      "loss": 1.3102,
      "step": 260736
    },
    {
      "epoch": 19.57223264540338,
      "grad_norm": 0.5436031222343445,
      "learning_rate": 0.002828,
      "loss": 1.322,
      "step": 260800
    },
    {
      "epoch": 19.57703564727955,
      "grad_norm": 0.6137192845344543,
      "learning_rate": 0.002828,
      "loss": 1.3267,
      "step": 260864
    },
    {
      "epoch": 19.58183864915572,
      "grad_norm": 0.5399419665336609,
      "learning_rate": 0.002828,
      "loss": 1.3147,
      "step": 260928
    },
    {
      "epoch": 19.586641651031893,
      "grad_norm": 0.6851577758789062,
      "learning_rate": 0.002828,
      "loss": 1.312,
      "step": 260992
    },
    {
      "epoch": 19.591444652908066,
      "grad_norm": 0.5793424844741821,
      "learning_rate": 0.002828,
      "loss": 1.3104,
      "step": 261056
    },
    {
      "epoch": 19.59624765478424,
      "grad_norm": 0.486041784286499,
      "learning_rate": 0.002828,
      "loss": 1.3192,
      "step": 261120
    },
    {
      "epoch": 19.601050656660412,
      "grad_norm": 0.5898070335388184,
      "learning_rate": 0.002828,
      "loss": 1.3165,
      "step": 261184
    },
    {
      "epoch": 19.605853658536585,
      "grad_norm": 0.611981987953186,
      "learning_rate": 0.002828,
      "loss": 1.3169,
      "step": 261248
    },
    {
      "epoch": 19.610656660412758,
      "grad_norm": 0.5300493240356445,
      "learning_rate": 0.002828,
      "loss": 1.3128,
      "step": 261312
    },
    {
      "epoch": 19.61545966228893,
      "grad_norm": 0.6088967323303223,
      "learning_rate": 0.002828,
      "loss": 1.3145,
      "step": 261376
    },
    {
      "epoch": 19.620262664165104,
      "grad_norm": 0.5334705114364624,
      "learning_rate": 0.002828,
      "loss": 1.3172,
      "step": 261440
    },
    {
      "epoch": 19.625065666041277,
      "grad_norm": 0.5719045400619507,
      "learning_rate": 0.002828,
      "loss": 1.3176,
      "step": 261504
    },
    {
      "epoch": 19.62986866791745,
      "grad_norm": 0.5838398337364197,
      "learning_rate": 0.002828,
      "loss": 1.3199,
      "step": 261568
    },
    {
      "epoch": 19.634671669793622,
      "grad_norm": 0.565754234790802,
      "learning_rate": 0.002828,
      "loss": 1.3113,
      "step": 261632
    },
    {
      "epoch": 19.639474671669795,
      "grad_norm": 0.5282318592071533,
      "learning_rate": 0.002828,
      "loss": 1.319,
      "step": 261696
    },
    {
      "epoch": 19.644277673545965,
      "grad_norm": 0.5384307503700256,
      "learning_rate": 0.002828,
      "loss": 1.3143,
      "step": 261760
    },
    {
      "epoch": 19.649080675422137,
      "grad_norm": 0.5330702662467957,
      "learning_rate": 0.002828,
      "loss": 1.3165,
      "step": 261824
    },
    {
      "epoch": 19.65388367729831,
      "grad_norm": 0.521190345287323,
      "learning_rate": 0.002828,
      "loss": 1.3153,
      "step": 261888
    },
    {
      "epoch": 19.658686679174483,
      "grad_norm": 0.6237691640853882,
      "learning_rate": 0.002828,
      "loss": 1.3095,
      "step": 261952
    },
    {
      "epoch": 19.663489681050656,
      "grad_norm": 0.6528506278991699,
      "learning_rate": 0.002828,
      "loss": 1.3205,
      "step": 262016
    },
    {
      "epoch": 19.66829268292683,
      "grad_norm": 0.5690321922302246,
      "learning_rate": 0.002828,
      "loss": 1.3233,
      "step": 262080
    },
    {
      "epoch": 19.673095684803002,
      "grad_norm": 0.5753093957901001,
      "learning_rate": 0.002828,
      "loss": 1.3184,
      "step": 262144
    },
    {
      "epoch": 19.677898686679175,
      "grad_norm": 0.5182336568832397,
      "learning_rate": 0.002828,
      "loss": 1.3149,
      "step": 262208
    },
    {
      "epoch": 19.682701688555348,
      "grad_norm": 0.946252167224884,
      "learning_rate": 0.002828,
      "loss": 1.32,
      "step": 262272
    },
    {
      "epoch": 19.68750469043152,
      "grad_norm": 0.7170292735099792,
      "learning_rate": 0.002828,
      "loss": 1.3173,
      "step": 262336
    },
    {
      "epoch": 19.692307692307693,
      "grad_norm": 0.5056231021881104,
      "learning_rate": 0.002828,
      "loss": 1.3166,
      "step": 262400
    },
    {
      "epoch": 19.697110694183866,
      "grad_norm": 0.5264489054679871,
      "learning_rate": 0.002828,
      "loss": 1.3173,
      "step": 262464
    },
    {
      "epoch": 19.70191369606004,
      "grad_norm": 0.5630133152008057,
      "learning_rate": 0.002828,
      "loss": 1.3196,
      "step": 262528
    },
    {
      "epoch": 19.70671669793621,
      "grad_norm": 0.5217622518539429,
      "learning_rate": 0.002828,
      "loss": 1.3186,
      "step": 262592
    },
    {
      "epoch": 19.71151969981238,
      "grad_norm": 0.5659202337265015,
      "learning_rate": 0.002828,
      "loss": 1.3193,
      "step": 262656
    },
    {
      "epoch": 19.716322701688554,
      "grad_norm": 0.5758469104766846,
      "learning_rate": 0.002828,
      "loss": 1.3157,
      "step": 262720
    },
    {
      "epoch": 19.721125703564727,
      "grad_norm": 0.5595651865005493,
      "learning_rate": 0.002828,
      "loss": 1.3198,
      "step": 262784
    },
    {
      "epoch": 19.7259287054409,
      "grad_norm": 0.5688280463218689,
      "learning_rate": 0.002828,
      "loss": 1.3193,
      "step": 262848
    },
    {
      "epoch": 19.730731707317073,
      "grad_norm": 0.5543525815010071,
      "learning_rate": 0.002828,
      "loss": 1.3144,
      "step": 262912
    },
    {
      "epoch": 19.735534709193246,
      "grad_norm": 0.5669763684272766,
      "learning_rate": 0.002828,
      "loss": 1.3183,
      "step": 262976
    },
    {
      "epoch": 19.74033771106942,
      "grad_norm": 0.628093957901001,
      "learning_rate": 0.002828,
      "loss": 1.3183,
      "step": 263040
    },
    {
      "epoch": 19.74514071294559,
      "grad_norm": 0.5473799705505371,
      "learning_rate": 0.002828,
      "loss": 1.3212,
      "step": 263104
    },
    {
      "epoch": 19.749943714821764,
      "grad_norm": 0.5699611306190491,
      "learning_rate": 0.002828,
      "loss": 1.3144,
      "step": 263168
    },
    {
      "epoch": 19.754746716697937,
      "grad_norm": 0.5728660821914673,
      "learning_rate": 0.002828,
      "loss": 1.3159,
      "step": 263232
    },
    {
      "epoch": 19.75954971857411,
      "grad_norm": 0.5876243114471436,
      "learning_rate": 0.002828,
      "loss": 1.3225,
      "step": 263296
    },
    {
      "epoch": 19.764352720450283,
      "grad_norm": 0.6909279227256775,
      "learning_rate": 0.002828,
      "loss": 1.3139,
      "step": 263360
    },
    {
      "epoch": 19.769155722326452,
      "grad_norm": 0.5972185730934143,
      "learning_rate": 0.002828,
      "loss": 1.3166,
      "step": 263424
    },
    {
      "epoch": 19.773958724202625,
      "grad_norm": 0.5375490784645081,
      "learning_rate": 0.002828,
      "loss": 1.317,
      "step": 263488
    },
    {
      "epoch": 19.7787617260788,
      "grad_norm": 0.5809566378593445,
      "learning_rate": 0.002828,
      "loss": 1.3194,
      "step": 263552
    },
    {
      "epoch": 19.78356472795497,
      "grad_norm": 0.5196545124053955,
      "learning_rate": 0.002828,
      "loss": 1.3144,
      "step": 263616
    },
    {
      "epoch": 19.788367729831144,
      "grad_norm": 0.5775570869445801,
      "learning_rate": 0.002828,
      "loss": 1.3206,
      "step": 263680
    },
    {
      "epoch": 19.793170731707317,
      "grad_norm": 0.6974487900733948,
      "learning_rate": 0.002828,
      "loss": 1.3193,
      "step": 263744
    },
    {
      "epoch": 19.79797373358349,
      "grad_norm": 0.6145743727684021,
      "learning_rate": 0.002828,
      "loss": 1.3178,
      "step": 263808
    },
    {
      "epoch": 19.802776735459663,
      "grad_norm": 0.5650912523269653,
      "learning_rate": 0.002828,
      "loss": 1.3139,
      "step": 263872
    },
    {
      "epoch": 19.807579737335836,
      "grad_norm": 0.5052109956741333,
      "learning_rate": 0.002828,
      "loss": 1.314,
      "step": 263936
    },
    {
      "epoch": 19.81238273921201,
      "grad_norm": 0.5860249996185303,
      "learning_rate": 0.002828,
      "loss": 1.3167,
      "step": 264000
    },
    {
      "epoch": 19.81718574108818,
      "grad_norm": 0.5640794038772583,
      "learning_rate": 0.002828,
      "loss": 1.3223,
      "step": 264064
    },
    {
      "epoch": 19.821988742964354,
      "grad_norm": 0.6080434322357178,
      "learning_rate": 0.002828,
      "loss": 1.3229,
      "step": 264128
    },
    {
      "epoch": 19.826791744840527,
      "grad_norm": 0.48237085342407227,
      "learning_rate": 0.002828,
      "loss": 1.322,
      "step": 264192
    },
    {
      "epoch": 19.831594746716696,
      "grad_norm": 0.6327677965164185,
      "learning_rate": 0.002828,
      "loss": 1.321,
      "step": 264256
    },
    {
      "epoch": 19.83639774859287,
      "grad_norm": 0.6129979491233826,
      "learning_rate": 0.002828,
      "loss": 1.3129,
      "step": 264320
    },
    {
      "epoch": 19.841200750469042,
      "grad_norm": 0.5425165891647339,
      "learning_rate": 0.002828,
      "loss": 1.3244,
      "step": 264384
    },
    {
      "epoch": 19.846003752345215,
      "grad_norm": 0.47874894738197327,
      "learning_rate": 0.002828,
      "loss": 1.3237,
      "step": 264448
    },
    {
      "epoch": 19.850806754221388,
      "grad_norm": 0.635398805141449,
      "learning_rate": 0.002828,
      "loss": 1.3207,
      "step": 264512
    },
    {
      "epoch": 19.85560975609756,
      "grad_norm": 0.5302428603172302,
      "learning_rate": 0.002828,
      "loss": 1.3237,
      "step": 264576
    },
    {
      "epoch": 19.860412757973734,
      "grad_norm": 0.5885991454124451,
      "learning_rate": 0.002828,
      "loss": 1.3187,
      "step": 264640
    },
    {
      "epoch": 19.865215759849907,
      "grad_norm": 0.6125088334083557,
      "learning_rate": 0.002828,
      "loss": 1.3198,
      "step": 264704
    },
    {
      "epoch": 19.87001876172608,
      "grad_norm": 0.5166810154914856,
      "learning_rate": 0.002828,
      "loss": 1.318,
      "step": 264768
    },
    {
      "epoch": 19.874821763602252,
      "grad_norm": 0.5309271216392517,
      "learning_rate": 0.002828,
      "loss": 1.3206,
      "step": 264832
    },
    {
      "epoch": 19.879624765478425,
      "grad_norm": 0.5516149997711182,
      "learning_rate": 0.002828,
      "loss": 1.3206,
      "step": 264896
    },
    {
      "epoch": 19.884427767354598,
      "grad_norm": 0.4545881450176239,
      "learning_rate": 0.002828,
      "loss": 1.3209,
      "step": 264960
    },
    {
      "epoch": 19.88923076923077,
      "grad_norm": 0.7351852059364319,
      "learning_rate": 0.002828,
      "loss": 1.3182,
      "step": 265024
    },
    {
      "epoch": 19.89403377110694,
      "grad_norm": 0.5521069169044495,
      "learning_rate": 0.002828,
      "loss": 1.3113,
      "step": 265088
    },
    {
      "epoch": 19.898836772983113,
      "grad_norm": 0.5605167746543884,
      "learning_rate": 0.002828,
      "loss": 1.3185,
      "step": 265152
    },
    {
      "epoch": 19.903639774859286,
      "grad_norm": 0.4302228093147278,
      "learning_rate": 0.002828,
      "loss": 1.3168,
      "step": 265216
    },
    {
      "epoch": 19.90844277673546,
      "grad_norm": 0.6508972644805908,
      "learning_rate": 0.002828,
      "loss": 1.3318,
      "step": 265280
    },
    {
      "epoch": 19.913245778611632,
      "grad_norm": 0.5926183462142944,
      "learning_rate": 0.002828,
      "loss": 1.3169,
      "step": 265344
    },
    {
      "epoch": 19.918048780487805,
      "grad_norm": 0.6462618708610535,
      "learning_rate": 0.002828,
      "loss": 1.3203,
      "step": 265408
    },
    {
      "epoch": 19.922851782363978,
      "grad_norm": 0.5538685321807861,
      "learning_rate": 0.002828,
      "loss": 1.3143,
      "step": 265472
    },
    {
      "epoch": 19.92765478424015,
      "grad_norm": 0.5557839870452881,
      "learning_rate": 0.002828,
      "loss": 1.3187,
      "step": 265536
    },
    {
      "epoch": 19.932457786116323,
      "grad_norm": 0.5402202606201172,
      "learning_rate": 0.002828,
      "loss": 1.3221,
      "step": 265600
    },
    {
      "epoch": 19.937260787992496,
      "grad_norm": 0.6323610544204712,
      "learning_rate": 0.002828,
      "loss": 1.3179,
      "step": 265664
    },
    {
      "epoch": 19.94206378986867,
      "grad_norm": 0.5772566795349121,
      "learning_rate": 0.002828,
      "loss": 1.3186,
      "step": 265728
    },
    {
      "epoch": 19.946866791744842,
      "grad_norm": 0.5509674549102783,
      "learning_rate": 0.002828,
      "loss": 1.3251,
      "step": 265792
    },
    {
      "epoch": 19.95166979362101,
      "grad_norm": 0.5037369728088379,
      "learning_rate": 0.002828,
      "loss": 1.3192,
      "step": 265856
    },
    {
      "epoch": 19.956472795497184,
      "grad_norm": 0.6050003170967102,
      "learning_rate": 0.002828,
      "loss": 1.3198,
      "step": 265920
    },
    {
      "epoch": 19.961275797373357,
      "grad_norm": 0.519004225730896,
      "learning_rate": 0.002828,
      "loss": 1.3132,
      "step": 265984
    },
    {
      "epoch": 19.96607879924953,
      "grad_norm": 0.5484166741371155,
      "learning_rate": 0.002828,
      "loss": 1.3227,
      "step": 266048
    },
    {
      "epoch": 19.970881801125703,
      "grad_norm": 0.5241848826408386,
      "learning_rate": 0.002828,
      "loss": 1.321,
      "step": 266112
    },
    {
      "epoch": 19.975684803001876,
      "grad_norm": 0.5934039950370789,
      "learning_rate": 0.002828,
      "loss": 1.3175,
      "step": 266176
    },
    {
      "epoch": 19.98048780487805,
      "grad_norm": 0.5047505497932434,
      "learning_rate": 0.002828,
      "loss": 1.3216,
      "step": 266240
    },
    {
      "epoch": 19.98529080675422,
      "grad_norm": 0.5770825743675232,
      "learning_rate": 0.002828,
      "loss": 1.3225,
      "step": 266304
    },
    {
      "epoch": 19.990093808630395,
      "grad_norm": 0.6241472959518433,
      "learning_rate": 0.002828,
      "loss": 1.3177,
      "step": 266368
    },
    {
      "epoch": 19.994896810506567,
      "grad_norm": 0.6075835227966309,
      "learning_rate": 0.002828,
      "loss": 1.3214,
      "step": 266432
    },
    {
      "epoch": 19.99969981238274,
      "grad_norm": 0.559272050857544,
      "learning_rate": 0.002828,
      "loss": 1.3189,
      "step": 266496
    },
    {
      "epoch": 20.004502814258913,
      "grad_norm": 0.6129955053329468,
      "learning_rate": 0.002828,
      "loss": 1.2869,
      "step": 266560
    },
    {
      "epoch": 20.009305816135086,
      "grad_norm": 0.5431690812110901,
      "learning_rate": 0.002828,
      "loss": 1.281,
      "step": 266624
    },
    {
      "epoch": 20.014108818011255,
      "grad_norm": 0.657706618309021,
      "learning_rate": 0.002828,
      "loss": 1.2773,
      "step": 266688
    },
    {
      "epoch": 20.01891181988743,
      "grad_norm": 0.5088411569595337,
      "learning_rate": 0.002828,
      "loss": 1.2845,
      "step": 266752
    },
    {
      "epoch": 20.0237148217636,
      "grad_norm": 0.5734243392944336,
      "learning_rate": 0.002828,
      "loss": 1.2781,
      "step": 266816
    },
    {
      "epoch": 20.028517823639774,
      "grad_norm": 0.5754413604736328,
      "learning_rate": 0.002828,
      "loss": 1.2766,
      "step": 266880
    },
    {
      "epoch": 20.033320825515947,
      "grad_norm": 0.5661938190460205,
      "learning_rate": 0.002828,
      "loss": 1.287,
      "step": 266944
    },
    {
      "epoch": 20.03812382739212,
      "grad_norm": 0.6118998527526855,
      "learning_rate": 0.002828,
      "loss": 1.2851,
      "step": 267008
    },
    {
      "epoch": 20.042926829268293,
      "grad_norm": 0.5828491449356079,
      "learning_rate": 0.002828,
      "loss": 1.2842,
      "step": 267072
    },
    {
      "epoch": 20.047729831144466,
      "grad_norm": 0.6424586176872253,
      "learning_rate": 0.002828,
      "loss": 1.2801,
      "step": 267136
    },
    {
      "epoch": 20.05253283302064,
      "grad_norm": 0.5585246086120605,
      "learning_rate": 0.002828,
      "loss": 1.2882,
      "step": 267200
    },
    {
      "epoch": 20.05733583489681,
      "grad_norm": 0.6090246438980103,
      "learning_rate": 0.002828,
      "loss": 1.2753,
      "step": 267264
    },
    {
      "epoch": 20.062138836772984,
      "grad_norm": 0.5552029609680176,
      "learning_rate": 0.002828,
      "loss": 1.2748,
      "step": 267328
    },
    {
      "epoch": 20.066941838649157,
      "grad_norm": 0.588688850402832,
      "learning_rate": 0.002828,
      "loss": 1.2835,
      "step": 267392
    },
    {
      "epoch": 20.07174484052533,
      "grad_norm": 0.537607729434967,
      "learning_rate": 0.002828,
      "loss": 1.2879,
      "step": 267456
    },
    {
      "epoch": 20.0765478424015,
      "grad_norm": 0.5644827485084534,
      "learning_rate": 0.002828,
      "loss": 1.2814,
      "step": 267520
    },
    {
      "epoch": 20.081350844277672,
      "grad_norm": 0.631341278553009,
      "learning_rate": 0.002828,
      "loss": 1.2826,
      "step": 267584
    },
    {
      "epoch": 20.086153846153845,
      "grad_norm": 0.493162602186203,
      "learning_rate": 0.002828,
      "loss": 1.2781,
      "step": 267648
    },
    {
      "epoch": 20.090956848030018,
      "grad_norm": 0.5350943803787231,
      "learning_rate": 0.002828,
      "loss": 1.2857,
      "step": 267712
    },
    {
      "epoch": 20.09575984990619,
      "grad_norm": 0.6216751337051392,
      "learning_rate": 0.002828,
      "loss": 1.284,
      "step": 267776
    },
    {
      "epoch": 20.100562851782364,
      "grad_norm": 0.5709570050239563,
      "learning_rate": 0.002828,
      "loss": 1.2817,
      "step": 267840
    },
    {
      "epoch": 20.105365853658537,
      "grad_norm": 0.5807507634162903,
      "learning_rate": 0.002828,
      "loss": 1.2902,
      "step": 267904
    },
    {
      "epoch": 20.11016885553471,
      "grad_norm": 0.632093071937561,
      "learning_rate": 0.002828,
      "loss": 1.2926,
      "step": 267968
    },
    {
      "epoch": 20.114971857410882,
      "grad_norm": 0.5635599493980408,
      "learning_rate": 0.002828,
      "loss": 1.2876,
      "step": 268032
    },
    {
      "epoch": 20.119774859287055,
      "grad_norm": 0.547120213508606,
      "learning_rate": 0.002828,
      "loss": 1.2883,
      "step": 268096
    },
    {
      "epoch": 20.124577861163228,
      "grad_norm": 0.47603321075439453,
      "learning_rate": 0.002828,
      "loss": 1.2875,
      "step": 268160
    },
    {
      "epoch": 20.1293808630394,
      "grad_norm": 0.5889759063720703,
      "learning_rate": 0.002828,
      "loss": 1.2913,
      "step": 268224
    },
    {
      "epoch": 20.134183864915574,
      "grad_norm": 0.5772056579589844,
      "learning_rate": 0.002828,
      "loss": 1.287,
      "step": 268288
    },
    {
      "epoch": 20.138986866791743,
      "grad_norm": 0.6541629433631897,
      "learning_rate": 0.002828,
      "loss": 1.2917,
      "step": 268352
    },
    {
      "epoch": 20.143789868667916,
      "grad_norm": 0.5309247970581055,
      "learning_rate": 0.002828,
      "loss": 1.2911,
      "step": 268416
    },
    {
      "epoch": 20.14859287054409,
      "grad_norm": 0.5737690329551697,
      "learning_rate": 0.002828,
      "loss": 1.2924,
      "step": 268480
    },
    {
      "epoch": 20.153395872420262,
      "grad_norm": 0.5364387035369873,
      "learning_rate": 0.002828,
      "loss": 1.2874,
      "step": 268544
    },
    {
      "epoch": 20.158198874296435,
      "grad_norm": 0.5387442111968994,
      "learning_rate": 0.002828,
      "loss": 1.286,
      "step": 268608
    },
    {
      "epoch": 20.163001876172608,
      "grad_norm": 0.5343860983848572,
      "learning_rate": 0.002828,
      "loss": 1.2869,
      "step": 268672
    },
    {
      "epoch": 20.16780487804878,
      "grad_norm": 0.46021518111228943,
      "learning_rate": 0.002828,
      "loss": 1.293,
      "step": 268736
    },
    {
      "epoch": 20.172607879924954,
      "grad_norm": 0.6175848841667175,
      "learning_rate": 0.002828,
      "loss": 1.2891,
      "step": 268800
    },
    {
      "epoch": 20.177410881801126,
      "grad_norm": 0.5815446376800537,
      "learning_rate": 0.002828,
      "loss": 1.2934,
      "step": 268864
    },
    {
      "epoch": 20.1822138836773,
      "grad_norm": 0.689041018486023,
      "learning_rate": 0.002828,
      "loss": 1.2935,
      "step": 268928
    },
    {
      "epoch": 20.187016885553472,
      "grad_norm": 0.6535454392433167,
      "learning_rate": 0.002828,
      "loss": 1.2897,
      "step": 268992
    },
    {
      "epoch": 20.191819887429645,
      "grad_norm": 0.5424286723136902,
      "learning_rate": 0.002828,
      "loss": 1.2889,
      "step": 269056
    },
    {
      "epoch": 20.196622889305814,
      "grad_norm": 0.6392897367477417,
      "learning_rate": 0.002828,
      "loss": 1.2892,
      "step": 269120
    },
    {
      "epoch": 20.201425891181987,
      "grad_norm": 0.5682271122932434,
      "learning_rate": 0.002828,
      "loss": 1.2906,
      "step": 269184
    },
    {
      "epoch": 20.20622889305816,
      "grad_norm": 0.5338545441627502,
      "learning_rate": 0.002828,
      "loss": 1.2911,
      "step": 269248
    },
    {
      "epoch": 20.211031894934333,
      "grad_norm": 0.5464316606521606,
      "learning_rate": 0.002828,
      "loss": 1.2963,
      "step": 269312
    },
    {
      "epoch": 20.215834896810506,
      "grad_norm": 0.5830249786376953,
      "learning_rate": 0.002828,
      "loss": 1.2947,
      "step": 269376
    },
    {
      "epoch": 20.22063789868668,
      "grad_norm": 0.5794913172721863,
      "learning_rate": 0.002828,
      "loss": 1.2867,
      "step": 269440
    },
    {
      "epoch": 20.22544090056285,
      "grad_norm": 0.5707951784133911,
      "learning_rate": 0.002828,
      "loss": 1.2867,
      "step": 269504
    },
    {
      "epoch": 20.230243902439025,
      "grad_norm": 0.5334992408752441,
      "learning_rate": 0.002828,
      "loss": 1.2926,
      "step": 269568
    },
    {
      "epoch": 20.235046904315197,
      "grad_norm": 0.5544065833091736,
      "learning_rate": 0.002828,
      "loss": 1.2916,
      "step": 269632
    },
    {
      "epoch": 20.23984990619137,
      "grad_norm": 0.5692053437232971,
      "learning_rate": 0.002828,
      "loss": 1.2893,
      "step": 269696
    },
    {
      "epoch": 20.244652908067543,
      "grad_norm": 0.45951566100120544,
      "learning_rate": 0.002828,
      "loss": 1.2872,
      "step": 269760
    },
    {
      "epoch": 20.249455909943716,
      "grad_norm": 0.6202203631401062,
      "learning_rate": 0.002828,
      "loss": 1.2919,
      "step": 269824
    },
    {
      "epoch": 20.25425891181989,
      "grad_norm": 0.546671986579895,
      "learning_rate": 0.002828,
      "loss": 1.2948,
      "step": 269888
    },
    {
      "epoch": 20.25906191369606,
      "grad_norm": 0.5328788161277771,
      "learning_rate": 0.002828,
      "loss": 1.2969,
      "step": 269952
    },
    {
      "epoch": 20.26386491557223,
      "grad_norm": 0.5520268678665161,
      "learning_rate": 0.002828,
      "loss": 1.2964,
      "step": 270016
    },
    {
      "epoch": 20.268667917448404,
      "grad_norm": 0.5719852447509766,
      "learning_rate": 0.002828,
      "loss": 1.2941,
      "step": 270080
    },
    {
      "epoch": 20.273470919324577,
      "grad_norm": 0.5243651270866394,
      "learning_rate": 0.002828,
      "loss": 1.2941,
      "step": 270144
    },
    {
      "epoch": 20.27827392120075,
      "grad_norm": 0.5593424439430237,
      "learning_rate": 0.002828,
      "loss": 1.2931,
      "step": 270208
    },
    {
      "epoch": 20.283076923076923,
      "grad_norm": 0.5311948657035828,
      "learning_rate": 0.002828,
      "loss": 1.2953,
      "step": 270272
    },
    {
      "epoch": 20.287879924953096,
      "grad_norm": 0.7635405659675598,
      "learning_rate": 0.002828,
      "loss": 1.2937,
      "step": 270336
    },
    {
      "epoch": 20.29268292682927,
      "grad_norm": 0.5601173639297485,
      "learning_rate": 0.002828,
      "loss": 1.2888,
      "step": 270400
    },
    {
      "epoch": 20.29748592870544,
      "grad_norm": 0.5598116517066956,
      "learning_rate": 0.002828,
      "loss": 1.2926,
      "step": 270464
    },
    {
      "epoch": 20.302288930581614,
      "grad_norm": 0.551178514957428,
      "learning_rate": 0.002828,
      "loss": 1.2932,
      "step": 270528
    },
    {
      "epoch": 20.307091932457787,
      "grad_norm": 0.5252653956413269,
      "learning_rate": 0.002828,
      "loss": 1.2907,
      "step": 270592
    },
    {
      "epoch": 20.31189493433396,
      "grad_norm": 0.617731511592865,
      "learning_rate": 0.002828,
      "loss": 1.2957,
      "step": 270656
    },
    {
      "epoch": 20.316697936210133,
      "grad_norm": 0.5604806542396545,
      "learning_rate": 0.002828,
      "loss": 1.2946,
      "step": 270720
    },
    {
      "epoch": 20.321500938086302,
      "grad_norm": 0.5689629316329956,
      "learning_rate": 0.002828,
      "loss": 1.2946,
      "step": 270784
    },
    {
      "epoch": 20.326303939962475,
      "grad_norm": 0.5235245227813721,
      "learning_rate": 0.002828,
      "loss": 1.2964,
      "step": 270848
    },
    {
      "epoch": 20.331106941838648,
      "grad_norm": 0.5162010788917542,
      "learning_rate": 0.002828,
      "loss": 1.2939,
      "step": 270912
    },
    {
      "epoch": 20.33590994371482,
      "grad_norm": 0.4949166178703308,
      "learning_rate": 0.002828,
      "loss": 1.3024,
      "step": 270976
    },
    {
      "epoch": 20.340712945590994,
      "grad_norm": 0.5738046765327454,
      "learning_rate": 0.002828,
      "loss": 1.2948,
      "step": 271040
    },
    {
      "epoch": 20.345515947467167,
      "grad_norm": 0.5544030070304871,
      "learning_rate": 0.002828,
      "loss": 1.296,
      "step": 271104
    },
    {
      "epoch": 20.35031894934334,
      "grad_norm": 0.5360429883003235,
      "learning_rate": 0.002828,
      "loss": 1.2923,
      "step": 271168
    },
    {
      "epoch": 20.355121951219513,
      "grad_norm": 0.5099622011184692,
      "learning_rate": 0.002828,
      "loss": 1.3032,
      "step": 271232
    },
    {
      "epoch": 20.359924953095685,
      "grad_norm": 0.5253172516822815,
      "learning_rate": 0.002828,
      "loss": 1.29,
      "step": 271296
    },
    {
      "epoch": 20.36472795497186,
      "grad_norm": 0.5624486207962036,
      "learning_rate": 0.002828,
      "loss": 1.2991,
      "step": 271360
    },
    {
      "epoch": 20.36953095684803,
      "grad_norm": 0.575293242931366,
      "learning_rate": 0.002828,
      "loss": 1.297,
      "step": 271424
    },
    {
      "epoch": 20.374333958724204,
      "grad_norm": 0.6318356990814209,
      "learning_rate": 0.002828,
      "loss": 1.2949,
      "step": 271488
    },
    {
      "epoch": 20.379136960600377,
      "grad_norm": 0.5846620798110962,
      "learning_rate": 0.002828,
      "loss": 1.2944,
      "step": 271552
    },
    {
      "epoch": 20.383939962476546,
      "grad_norm": 0.7515684962272644,
      "learning_rate": 0.002828,
      "loss": 1.295,
      "step": 271616
    },
    {
      "epoch": 20.38874296435272,
      "grad_norm": 0.6497512459754944,
      "learning_rate": 0.002828,
      "loss": 1.2903,
      "step": 271680
    },
    {
      "epoch": 20.393545966228892,
      "grad_norm": 0.5049740076065063,
      "learning_rate": 0.002828,
      "loss": 1.3037,
      "step": 271744
    },
    {
      "epoch": 20.398348968105065,
      "grad_norm": 0.48297742009162903,
      "learning_rate": 0.002828,
      "loss": 1.3004,
      "step": 271808
    },
    {
      "epoch": 20.403151969981238,
      "grad_norm": 0.6171135902404785,
      "learning_rate": 0.002828,
      "loss": 1.294,
      "step": 271872
    },
    {
      "epoch": 20.40795497185741,
      "grad_norm": 0.5428125858306885,
      "learning_rate": 0.002828,
      "loss": 1.2954,
      "step": 271936
    },
    {
      "epoch": 20.412757973733584,
      "grad_norm": 0.5896263718605042,
      "learning_rate": 0.002828,
      "loss": 1.2969,
      "step": 272000
    },
    {
      "epoch": 20.417560975609756,
      "grad_norm": 0.6516025066375732,
      "learning_rate": 0.002828,
      "loss": 1.2992,
      "step": 272064
    },
    {
      "epoch": 20.42236397748593,
      "grad_norm": 0.5841122269630432,
      "learning_rate": 0.002828,
      "loss": 1.2928,
      "step": 272128
    },
    {
      "epoch": 20.427166979362102,
      "grad_norm": 0.6658204197883606,
      "learning_rate": 0.002828,
      "loss": 1.3003,
      "step": 272192
    },
    {
      "epoch": 20.431969981238275,
      "grad_norm": 0.7099432349205017,
      "learning_rate": 0.002828,
      "loss": 1.2908,
      "step": 272256
    },
    {
      "epoch": 20.436772983114448,
      "grad_norm": 0.6290414333343506,
      "learning_rate": 0.002828,
      "loss": 1.2968,
      "step": 272320
    },
    {
      "epoch": 20.44157598499062,
      "grad_norm": 0.6891517639160156,
      "learning_rate": 0.002828,
      "loss": 1.2992,
      "step": 272384
    },
    {
      "epoch": 20.44637898686679,
      "grad_norm": 0.5390793085098267,
      "learning_rate": 0.002828,
      "loss": 1.2989,
      "step": 272448
    },
    {
      "epoch": 20.451181988742963,
      "grad_norm": 0.5205430388450623,
      "learning_rate": 0.002828,
      "loss": 1.293,
      "step": 272512
    },
    {
      "epoch": 20.455984990619136,
      "grad_norm": 0.618954598903656,
      "learning_rate": 0.002828,
      "loss": 1.295,
      "step": 272576
    },
    {
      "epoch": 20.46078799249531,
      "grad_norm": 0.5977427959442139,
      "learning_rate": 0.002828,
      "loss": 1.2957,
      "step": 272640
    },
    {
      "epoch": 20.46559099437148,
      "grad_norm": 0.48537740111351013,
      "learning_rate": 0.002828,
      "loss": 1.2998,
      "step": 272704
    },
    {
      "epoch": 20.470393996247655,
      "grad_norm": 0.6163172125816345,
      "learning_rate": 0.002828,
      "loss": 1.2984,
      "step": 272768
    },
    {
      "epoch": 20.475196998123828,
      "grad_norm": 0.6726917028427124,
      "learning_rate": 0.002828,
      "loss": 1.3049,
      "step": 272832
    },
    {
      "epoch": 20.48,
      "grad_norm": 0.7405112385749817,
      "learning_rate": 0.002828,
      "loss": 1.2933,
      "step": 272896
    },
    {
      "epoch": 20.484803001876173,
      "grad_norm": 0.6282209753990173,
      "learning_rate": 0.002828,
      "loss": 1.2972,
      "step": 272960
    },
    {
      "epoch": 20.489606003752346,
      "grad_norm": 0.5530390739440918,
      "learning_rate": 0.002828,
      "loss": 1.2978,
      "step": 273024
    },
    {
      "epoch": 20.49440900562852,
      "grad_norm": 0.5814974308013916,
      "learning_rate": 0.002828,
      "loss": 1.2933,
      "step": 273088
    },
    {
      "epoch": 20.499212007504692,
      "grad_norm": 0.6190447211265564,
      "learning_rate": 0.002828,
      "loss": 1.3,
      "step": 273152
    },
    {
      "epoch": 20.504015009380865,
      "grad_norm": 0.5674455761909485,
      "learning_rate": 0.002828,
      "loss": 1.2985,
      "step": 273216
    },
    {
      "epoch": 20.508818011257034,
      "grad_norm": 0.6419714689254761,
      "learning_rate": 0.002828,
      "loss": 1.2984,
      "step": 273280
    },
    {
      "epoch": 20.513621013133207,
      "grad_norm": 0.5519217252731323,
      "learning_rate": 0.002828,
      "loss": 1.2982,
      "step": 273344
    },
    {
      "epoch": 20.51842401500938,
      "grad_norm": 0.46819642186164856,
      "learning_rate": 0.002828,
      "loss": 1.3034,
      "step": 273408
    },
    {
      "epoch": 20.523227016885553,
      "grad_norm": 0.5590828061103821,
      "learning_rate": 0.002828,
      "loss": 1.2998,
      "step": 273472
    },
    {
      "epoch": 20.528030018761726,
      "grad_norm": 0.6636494994163513,
      "learning_rate": 0.002828,
      "loss": 1.2943,
      "step": 273536
    },
    {
      "epoch": 20.5328330206379,
      "grad_norm": 0.613201916217804,
      "learning_rate": 0.002828,
      "loss": 1.2974,
      "step": 273600
    },
    {
      "epoch": 20.53763602251407,
      "grad_norm": 0.6075567007064819,
      "learning_rate": 0.002828,
      "loss": 1.2946,
      "step": 273664
    },
    {
      "epoch": 20.542439024390244,
      "grad_norm": 0.5289536714553833,
      "learning_rate": 0.002828,
      "loss": 1.2995,
      "step": 273728
    },
    {
      "epoch": 20.547242026266417,
      "grad_norm": 0.5214759111404419,
      "learning_rate": 0.002828,
      "loss": 1.3006,
      "step": 273792
    },
    {
      "epoch": 20.55204502814259,
      "grad_norm": 0.5907725095748901,
      "learning_rate": 0.002828,
      "loss": 1.301,
      "step": 273856
    },
    {
      "epoch": 20.556848030018763,
      "grad_norm": 0.4892508387565613,
      "learning_rate": 0.002828,
      "loss": 1.3025,
      "step": 273920
    },
    {
      "epoch": 20.561651031894936,
      "grad_norm": 0.6342883110046387,
      "learning_rate": 0.002828,
      "loss": 1.297,
      "step": 273984
    },
    {
      "epoch": 20.566454033771105,
      "grad_norm": 0.7062511444091797,
      "learning_rate": 0.002828,
      "loss": 1.3077,
      "step": 274048
    },
    {
      "epoch": 20.571257035647278,
      "grad_norm": 0.5731887221336365,
      "learning_rate": 0.002828,
      "loss": 1.2989,
      "step": 274112
    },
    {
      "epoch": 20.57606003752345,
      "grad_norm": 0.5922696590423584,
      "learning_rate": 0.002828,
      "loss": 1.2957,
      "step": 274176
    },
    {
      "epoch": 20.580863039399624,
      "grad_norm": 0.6939809322357178,
      "learning_rate": 0.002828,
      "loss": 1.2998,
      "step": 274240
    },
    {
      "epoch": 20.585666041275797,
      "grad_norm": 0.6065263152122498,
      "learning_rate": 0.002828,
      "loss": 1.297,
      "step": 274304
    },
    {
      "epoch": 20.59046904315197,
      "grad_norm": 0.4768809378147125,
      "learning_rate": 0.002828,
      "loss": 1.3048,
      "step": 274368
    },
    {
      "epoch": 20.595272045028143,
      "grad_norm": 0.6705830097198486,
      "learning_rate": 0.002828,
      "loss": 1.2991,
      "step": 274432
    },
    {
      "epoch": 20.600075046904315,
      "grad_norm": 0.6444434523582458,
      "learning_rate": 0.002828,
      "loss": 1.2978,
      "step": 274496
    },
    {
      "epoch": 20.60487804878049,
      "grad_norm": 0.5016328692436218,
      "learning_rate": 0.002828,
      "loss": 1.2974,
      "step": 274560
    },
    {
      "epoch": 20.60968105065666,
      "grad_norm": 0.5974975824356079,
      "learning_rate": 0.002828,
      "loss": 1.297,
      "step": 274624
    },
    {
      "epoch": 20.614484052532834,
      "grad_norm": 0.4709956645965576,
      "learning_rate": 0.002828,
      "loss": 1.3012,
      "step": 274688
    },
    {
      "epoch": 20.619287054409007,
      "grad_norm": 0.7240418195724487,
      "learning_rate": 0.002828,
      "loss": 1.3036,
      "step": 274752
    },
    {
      "epoch": 20.62409005628518,
      "grad_norm": 0.5455818772315979,
      "learning_rate": 0.002828,
      "loss": 1.3052,
      "step": 274816
    },
    {
      "epoch": 20.62889305816135,
      "grad_norm": 0.6497557759284973,
      "learning_rate": 0.002828,
      "loss": 1.3011,
      "step": 274880
    },
    {
      "epoch": 20.633696060037522,
      "grad_norm": 0.623468816280365,
      "learning_rate": 0.002828,
      "loss": 1.2998,
      "step": 274944
    },
    {
      "epoch": 20.638499061913695,
      "grad_norm": 0.5101485252380371,
      "learning_rate": 0.002828,
      "loss": 1.2985,
      "step": 275008
    },
    {
      "epoch": 20.643302063789868,
      "grad_norm": 0.6895449757575989,
      "learning_rate": 0.002828,
      "loss": 1.2969,
      "step": 275072
    },
    {
      "epoch": 20.64810506566604,
      "grad_norm": 0.6263419985771179,
      "learning_rate": 0.002828,
      "loss": 1.302,
      "step": 275136
    },
    {
      "epoch": 20.652908067542214,
      "grad_norm": 0.46888402104377747,
      "learning_rate": 0.002828,
      "loss": 1.2973,
      "step": 275200
    },
    {
      "epoch": 20.657711069418387,
      "grad_norm": 0.5644204616546631,
      "learning_rate": 0.002828,
      "loss": 1.3054,
      "step": 275264
    },
    {
      "epoch": 20.66251407129456,
      "grad_norm": 0.4956626892089844,
      "learning_rate": 0.002828,
      "loss": 1.3011,
      "step": 275328
    },
    {
      "epoch": 20.667317073170732,
      "grad_norm": 0.5327547192573547,
      "learning_rate": 0.002828,
      "loss": 1.3006,
      "step": 275392
    },
    {
      "epoch": 20.672120075046905,
      "grad_norm": 0.5446391105651855,
      "learning_rate": 0.002828,
      "loss": 1.3011,
      "step": 275456
    },
    {
      "epoch": 20.676923076923078,
      "grad_norm": 0.5319408178329468,
      "learning_rate": 0.002828,
      "loss": 1.3028,
      "step": 275520
    },
    {
      "epoch": 20.68172607879925,
      "grad_norm": 0.6537469625473022,
      "learning_rate": 0.002828,
      "loss": 1.2999,
      "step": 275584
    },
    {
      "epoch": 20.686529080675424,
      "grad_norm": 0.5841550827026367,
      "learning_rate": 0.002828,
      "loss": 1.3002,
      "step": 275648
    },
    {
      "epoch": 20.691332082551593,
      "grad_norm": 0.6798041462898254,
      "learning_rate": 0.002828,
      "loss": 1.3048,
      "step": 275712
    },
    {
      "epoch": 20.696135084427766,
      "grad_norm": 0.604843020439148,
      "learning_rate": 0.002828,
      "loss": 1.3019,
      "step": 275776
    },
    {
      "epoch": 20.70093808630394,
      "grad_norm": 0.5991361737251282,
      "learning_rate": 0.002828,
      "loss": 1.2993,
      "step": 275840
    },
    {
      "epoch": 20.705741088180112,
      "grad_norm": 0.5777614116668701,
      "learning_rate": 0.002828,
      "loss": 1.2961,
      "step": 275904
    },
    {
      "epoch": 20.710544090056285,
      "grad_norm": 0.5120457410812378,
      "learning_rate": 0.002828,
      "loss": 1.3075,
      "step": 275968
    },
    {
      "epoch": 20.715347091932458,
      "grad_norm": 0.5318966507911682,
      "learning_rate": 0.002828,
      "loss": 1.3067,
      "step": 276032
    },
    {
      "epoch": 20.72015009380863,
      "grad_norm": 0.5237390995025635,
      "learning_rate": 0.002828,
      "loss": 1.2978,
      "step": 276096
    },
    {
      "epoch": 20.724953095684803,
      "grad_norm": 0.4869547486305237,
      "learning_rate": 0.002828,
      "loss": 1.3004,
      "step": 276160
    },
    {
      "epoch": 20.729756097560976,
      "grad_norm": 0.51389080286026,
      "learning_rate": 0.002828,
      "loss": 1.3011,
      "step": 276224
    },
    {
      "epoch": 20.73455909943715,
      "grad_norm": 0.5911182761192322,
      "learning_rate": 0.002828,
      "loss": 1.2999,
      "step": 276288
    },
    {
      "epoch": 20.739362101313322,
      "grad_norm": 0.641406774520874,
      "learning_rate": 0.002828,
      "loss": 1.3015,
      "step": 276352
    },
    {
      "epoch": 20.744165103189495,
      "grad_norm": 0.5013548731803894,
      "learning_rate": 0.002828,
      "loss": 1.3015,
      "step": 276416
    },
    {
      "epoch": 20.748968105065664,
      "grad_norm": 0.6204653978347778,
      "learning_rate": 0.002828,
      "loss": 1.3002,
      "step": 276480
    },
    {
      "epoch": 20.753771106941837,
      "grad_norm": 0.49453988671302795,
      "learning_rate": 0.002828,
      "loss": 1.3045,
      "step": 276544
    },
    {
      "epoch": 20.75857410881801,
      "grad_norm": 0.5399415493011475,
      "learning_rate": 0.002828,
      "loss": 1.3039,
      "step": 276608
    },
    {
      "epoch": 20.763377110694183,
      "grad_norm": 0.5224660634994507,
      "learning_rate": 0.002828,
      "loss": 1.3077,
      "step": 276672
    },
    {
      "epoch": 20.768180112570356,
      "grad_norm": 0.5413986444473267,
      "learning_rate": 0.002828,
      "loss": 1.3068,
      "step": 276736
    },
    {
      "epoch": 20.77298311444653,
      "grad_norm": 0.6324653029441833,
      "learning_rate": 0.002828,
      "loss": 1.3022,
      "step": 276800
    },
    {
      "epoch": 20.7777861163227,
      "grad_norm": 0.6608572602272034,
      "learning_rate": 0.002828,
      "loss": 1.3042,
      "step": 276864
    },
    {
      "epoch": 20.782589118198874,
      "grad_norm": 0.5002397298812866,
      "learning_rate": 0.002828,
      "loss": 1.3065,
      "step": 276928
    },
    {
      "epoch": 20.787392120075047,
      "grad_norm": 0.5986420512199402,
      "learning_rate": 0.002828,
      "loss": 1.3046,
      "step": 276992
    },
    {
      "epoch": 20.79219512195122,
      "grad_norm": 0.5150978565216064,
      "learning_rate": 0.002828,
      "loss": 1.2971,
      "step": 277056
    },
    {
      "epoch": 20.796998123827393,
      "grad_norm": 0.7404240369796753,
      "learning_rate": 0.002828,
      "loss": 1.309,
      "step": 277120
    },
    {
      "epoch": 20.801801125703566,
      "grad_norm": 0.5593386888504028,
      "learning_rate": 0.002828,
      "loss": 1.3049,
      "step": 277184
    },
    {
      "epoch": 20.80660412757974,
      "grad_norm": 0.595770001411438,
      "learning_rate": 0.002828,
      "loss": 1.3039,
      "step": 277248
    },
    {
      "epoch": 20.811407129455908,
      "grad_norm": 0.7926905751228333,
      "learning_rate": 0.002828,
      "loss": 1.3033,
      "step": 277312
    },
    {
      "epoch": 20.81621013133208,
      "grad_norm": 0.5238022804260254,
      "learning_rate": 0.002828,
      "loss": 1.299,
      "step": 277376
    },
    {
      "epoch": 20.821013133208254,
      "grad_norm": 0.5705653429031372,
      "learning_rate": 0.002828,
      "loss": 1.3016,
      "step": 277440
    },
    {
      "epoch": 20.825816135084427,
      "grad_norm": 0.6585155725479126,
      "learning_rate": 0.002828,
      "loss": 1.3044,
      "step": 277504
    },
    {
      "epoch": 20.8306191369606,
      "grad_norm": 0.48894691467285156,
      "learning_rate": 0.002828,
      "loss": 1.3049,
      "step": 277568
    },
    {
      "epoch": 20.835422138836773,
      "grad_norm": 0.5562286972999573,
      "learning_rate": 0.002828,
      "loss": 1.2995,
      "step": 277632
    },
    {
      "epoch": 20.840225140712946,
      "grad_norm": 0.5397721529006958,
      "learning_rate": 0.002828,
      "loss": 1.3069,
      "step": 277696
    },
    {
      "epoch": 20.84502814258912,
      "grad_norm": 0.7238554954528809,
      "learning_rate": 0.002828,
      "loss": 1.3091,
      "step": 277760
    },
    {
      "epoch": 20.84983114446529,
      "grad_norm": 0.5862271785736084,
      "learning_rate": 0.002828,
      "loss": 1.3019,
      "step": 277824
    },
    {
      "epoch": 20.854634146341464,
      "grad_norm": 0.5831578969955444,
      "learning_rate": 0.002828,
      "loss": 1.3017,
      "step": 277888
    },
    {
      "epoch": 20.859437148217637,
      "grad_norm": 0.46079304814338684,
      "learning_rate": 0.002828,
      "loss": 1.3035,
      "step": 277952
    },
    {
      "epoch": 20.86424015009381,
      "grad_norm": 0.555037796497345,
      "learning_rate": 0.002828,
      "loss": 1.3041,
      "step": 278016
    },
    {
      "epoch": 20.869043151969983,
      "grad_norm": 0.5918899178504944,
      "learning_rate": 0.002828,
      "loss": 1.3068,
      "step": 278080
    },
    {
      "epoch": 20.873846153846152,
      "grad_norm": 0.4550643265247345,
      "learning_rate": 0.002828,
      "loss": 1.3038,
      "step": 278144
    },
    {
      "epoch": 20.878649155722325,
      "grad_norm": 0.7102483510971069,
      "learning_rate": 0.002828,
      "loss": 1.2991,
      "step": 278208
    },
    {
      "epoch": 20.883452157598498,
      "grad_norm": 0.543178141117096,
      "learning_rate": 0.002828,
      "loss": 1.3017,
      "step": 278272
    },
    {
      "epoch": 20.88825515947467,
      "grad_norm": 0.6087631583213806,
      "learning_rate": 0.002828,
      "loss": 1.3039,
      "step": 278336
    },
    {
      "epoch": 20.893058161350844,
      "grad_norm": 0.5347623229026794,
      "learning_rate": 0.002828,
      "loss": 1.3048,
      "step": 278400
    },
    {
      "epoch": 20.897861163227017,
      "grad_norm": 0.496459037065506,
      "learning_rate": 0.002828,
      "loss": 1.306,
      "step": 278464
    },
    {
      "epoch": 20.90266416510319,
      "grad_norm": 0.5459811687469482,
      "learning_rate": 0.002828,
      "loss": 1.3012,
      "step": 278528
    },
    {
      "epoch": 20.907467166979362,
      "grad_norm": 0.5910803079605103,
      "learning_rate": 0.002828,
      "loss": 1.2998,
      "step": 278592
    },
    {
      "epoch": 20.912270168855535,
      "grad_norm": 0.575896143913269,
      "learning_rate": 0.002828,
      "loss": 1.3012,
      "step": 278656
    },
    {
      "epoch": 20.917073170731708,
      "grad_norm": 0.5138533115386963,
      "learning_rate": 0.002828,
      "loss": 1.2963,
      "step": 278720
    },
    {
      "epoch": 20.92187617260788,
      "grad_norm": 0.6679477095603943,
      "learning_rate": 0.002828,
      "loss": 1.3056,
      "step": 278784
    },
    {
      "epoch": 20.926679174484054,
      "grad_norm": 0.5711171627044678,
      "learning_rate": 0.002828,
      "loss": 1.3042,
      "step": 278848
    },
    {
      "epoch": 20.931482176360227,
      "grad_norm": 0.5481835603713989,
      "learning_rate": 0.002828,
      "loss": 1.3011,
      "step": 278912
    },
    {
      "epoch": 20.936285178236396,
      "grad_norm": 0.6149061918258667,
      "learning_rate": 0.002828,
      "loss": 1.3,
      "step": 278976
    },
    {
      "epoch": 20.94108818011257,
      "grad_norm": 0.5665428042411804,
      "learning_rate": 0.002828,
      "loss": 1.3016,
      "step": 279040
    },
    {
      "epoch": 20.945891181988742,
      "grad_norm": 0.5922060608863831,
      "learning_rate": 0.002828,
      "loss": 1.3035,
      "step": 279104
    },
    {
      "epoch": 20.950694183864915,
      "grad_norm": 0.5712735056877136,
      "learning_rate": 0.002828,
      "loss": 1.3049,
      "step": 279168
    },
    {
      "epoch": 20.955497185741088,
      "grad_norm": 0.49438712000846863,
      "learning_rate": 0.002828,
      "loss": 1.2999,
      "step": 279232
    },
    {
      "epoch": 20.96030018761726,
      "grad_norm": 0.5500780344009399,
      "learning_rate": 0.002828,
      "loss": 1.3024,
      "step": 279296
    },
    {
      "epoch": 20.965103189493433,
      "grad_norm": 0.41028276085853577,
      "learning_rate": 0.002828,
      "loss": 1.3077,
      "step": 279360
    },
    {
      "epoch": 20.969906191369606,
      "grad_norm": 0.5960568785667419,
      "learning_rate": 0.002828,
      "loss": 1.3016,
      "step": 279424
    },
    {
      "epoch": 20.97470919324578,
      "grad_norm": 0.5779933929443359,
      "learning_rate": 0.002828,
      "loss": 1.3052,
      "step": 279488
    },
    {
      "epoch": 20.979512195121952,
      "grad_norm": 0.5713076591491699,
      "learning_rate": 0.002828,
      "loss": 1.3042,
      "step": 279552
    },
    {
      "epoch": 20.984315196998125,
      "grad_norm": 0.657752513885498,
      "learning_rate": 0.002828,
      "loss": 1.3039,
      "step": 279616
    },
    {
      "epoch": 20.989118198874298,
      "grad_norm": 0.6048377156257629,
      "learning_rate": 0.002828,
      "loss": 1.3052,
      "step": 279680
    },
    {
      "epoch": 20.99392120075047,
      "grad_norm": 0.5630027651786804,
      "learning_rate": 0.002828,
      "loss": 1.3045,
      "step": 279744
    },
    {
      "epoch": 20.99872420262664,
      "grad_norm": 0.5261644721031189,
      "learning_rate": 0.002828,
      "loss": 1.3037,
      "step": 279808
    },
    {
      "epoch": 21.003527204502813,
      "grad_norm": 0.7338108420372009,
      "learning_rate": 0.002828,
      "loss": 1.2755,
      "step": 279872
    },
    {
      "epoch": 21.008330206378986,
      "grad_norm": 0.5403333902359009,
      "learning_rate": 0.002828,
      "loss": 1.267,
      "step": 279936
    },
    {
      "epoch": 21.01313320825516,
      "grad_norm": 0.568750262260437,
      "learning_rate": 0.002828,
      "loss": 1.2625,
      "step": 280000
    },
    {
      "epoch": 21.01793621013133,
      "grad_norm": 0.5088416934013367,
      "learning_rate": 0.002828,
      "loss": 1.2696,
      "step": 280064
    },
    {
      "epoch": 21.022739212007505,
      "grad_norm": 0.5926600098609924,
      "learning_rate": 0.002828,
      "loss": 1.2638,
      "step": 280128
    },
    {
      "epoch": 21.027542213883677,
      "grad_norm": 0.5535401105880737,
      "learning_rate": 0.002828,
      "loss": 1.2676,
      "step": 280192
    },
    {
      "epoch": 21.03234521575985,
      "grad_norm": 0.5679246187210083,
      "learning_rate": 0.002828,
      "loss": 1.2622,
      "step": 280256
    },
    {
      "epoch": 21.037148217636023,
      "grad_norm": 0.6422357559204102,
      "learning_rate": 0.002828,
      "loss": 1.267,
      "step": 280320
    },
    {
      "epoch": 21.041951219512196,
      "grad_norm": 0.5129832625389099,
      "learning_rate": 0.002828,
      "loss": 1.268,
      "step": 280384
    },
    {
      "epoch": 21.04675422138837,
      "grad_norm": 0.6040861010551453,
      "learning_rate": 0.002828,
      "loss": 1.2665,
      "step": 280448
    },
    {
      "epoch": 21.051557223264542,
      "grad_norm": 0.6140467524528503,
      "learning_rate": 0.002828,
      "loss": 1.2644,
      "step": 280512
    },
    {
      "epoch": 21.056360225140715,
      "grad_norm": 0.5909779667854309,
      "learning_rate": 0.002828,
      "loss": 1.2658,
      "step": 280576
    },
    {
      "epoch": 21.061163227016884,
      "grad_norm": 0.5795265436172485,
      "learning_rate": 0.002828,
      "loss": 1.27,
      "step": 280640
    },
    {
      "epoch": 21.065966228893057,
      "grad_norm": 0.6152122616767883,
      "learning_rate": 0.002828,
      "loss": 1.2637,
      "step": 280704
    },
    {
      "epoch": 21.07076923076923,
      "grad_norm": 0.5055559277534485,
      "learning_rate": 0.002828,
      "loss": 1.2677,
      "step": 280768
    },
    {
      "epoch": 21.075572232645403,
      "grad_norm": 0.5432459115982056,
      "learning_rate": 0.002828,
      "loss": 1.2681,
      "step": 280832
    },
    {
      "epoch": 21.080375234521576,
      "grad_norm": 0.6607730388641357,
      "learning_rate": 0.002828,
      "loss": 1.2727,
      "step": 280896
    },
    {
      "epoch": 21.08517823639775,
      "grad_norm": 0.5829780697822571,
      "learning_rate": 0.002828,
      "loss": 1.2725,
      "step": 280960
    },
    {
      "epoch": 21.08998123827392,
      "grad_norm": 0.6266777515411377,
      "learning_rate": 0.002828,
      "loss": 1.2665,
      "step": 281024
    },
    {
      "epoch": 21.094784240150094,
      "grad_norm": 0.5578786730766296,
      "learning_rate": 0.002828,
      "loss": 1.2754,
      "step": 281088
    },
    {
      "epoch": 21.099587242026267,
      "grad_norm": 0.5670081973075867,
      "learning_rate": 0.002828,
      "loss": 1.2744,
      "step": 281152
    },
    {
      "epoch": 21.10439024390244,
      "grad_norm": 0.717337429523468,
      "learning_rate": 0.002828,
      "loss": 1.2719,
      "step": 281216
    },
    {
      "epoch": 21.109193245778613,
      "grad_norm": 0.7886514067649841,
      "learning_rate": 0.002828,
      "loss": 1.272,
      "step": 281280
    },
    {
      "epoch": 21.113996247654786,
      "grad_norm": 0.6627486944198608,
      "learning_rate": 0.002828,
      "loss": 1.2662,
      "step": 281344
    },
    {
      "epoch": 21.118799249530955,
      "grad_norm": 0.5494076609611511,
      "learning_rate": 0.002828,
      "loss": 1.2688,
      "step": 281408
    },
    {
      "epoch": 21.123602251407128,
      "grad_norm": 0.48781818151474,
      "learning_rate": 0.002828,
      "loss": 1.2748,
      "step": 281472
    },
    {
      "epoch": 21.1284052532833,
      "grad_norm": 0.6375535726547241,
      "learning_rate": 0.002828,
      "loss": 1.265,
      "step": 281536
    },
    {
      "epoch": 21.133208255159474,
      "grad_norm": 0.5140436887741089,
      "learning_rate": 0.002828,
      "loss": 1.2674,
      "step": 281600
    },
    {
      "epoch": 21.138011257035647,
      "grad_norm": 0.7258280515670776,
      "learning_rate": 0.002828,
      "loss": 1.2759,
      "step": 281664
    },
    {
      "epoch": 21.14281425891182,
      "grad_norm": 0.567965567111969,
      "learning_rate": 0.002828,
      "loss": 1.2768,
      "step": 281728
    },
    {
      "epoch": 21.147617260787992,
      "grad_norm": 0.5151071548461914,
      "learning_rate": 0.002828,
      "loss": 1.2714,
      "step": 281792
    },
    {
      "epoch": 21.152420262664165,
      "grad_norm": 0.5506834983825684,
      "learning_rate": 0.002828,
      "loss": 1.273,
      "step": 281856
    },
    {
      "epoch": 21.157223264540338,
      "grad_norm": 0.5140776038169861,
      "learning_rate": 0.002828,
      "loss": 1.2743,
      "step": 281920
    },
    {
      "epoch": 21.16202626641651,
      "grad_norm": 0.5231831669807434,
      "learning_rate": 0.002828,
      "loss": 1.2739,
      "step": 281984
    },
    {
      "epoch": 21.166829268292684,
      "grad_norm": 0.5400099754333496,
      "learning_rate": 0.002828,
      "loss": 1.2755,
      "step": 282048
    },
    {
      "epoch": 21.171632270168857,
      "grad_norm": 0.6175333261489868,
      "learning_rate": 0.002828,
      "loss": 1.2722,
      "step": 282112
    },
    {
      "epoch": 21.17643527204503,
      "grad_norm": 0.4814399182796478,
      "learning_rate": 0.002828,
      "loss": 1.2777,
      "step": 282176
    },
    {
      "epoch": 21.1812382739212,
      "grad_norm": 0.5004304051399231,
      "learning_rate": 0.002828,
      "loss": 1.2718,
      "step": 282240
    },
    {
      "epoch": 21.186041275797372,
      "grad_norm": 0.6475741267204285,
      "learning_rate": 0.002828,
      "loss": 1.2723,
      "step": 282304
    },
    {
      "epoch": 21.190844277673545,
      "grad_norm": 0.5444566607475281,
      "learning_rate": 0.002828,
      "loss": 1.2799,
      "step": 282368
    },
    {
      "epoch": 21.195647279549718,
      "grad_norm": 0.6097396016120911,
      "learning_rate": 0.002828,
      "loss": 1.2755,
      "step": 282432
    },
    {
      "epoch": 21.20045028142589,
      "grad_norm": 0.6205595135688782,
      "learning_rate": 0.002828,
      "loss": 1.2723,
      "step": 282496
    },
    {
      "epoch": 21.205253283302063,
      "grad_norm": 0.5144677758216858,
      "learning_rate": 0.002828,
      "loss": 1.276,
      "step": 282560
    },
    {
      "epoch": 21.210056285178236,
      "grad_norm": 0.5892837047576904,
      "learning_rate": 0.002828,
      "loss": 1.2726,
      "step": 282624
    },
    {
      "epoch": 21.21485928705441,
      "grad_norm": 0.576680064201355,
      "learning_rate": 0.002828,
      "loss": 1.274,
      "step": 282688
    },
    {
      "epoch": 21.219662288930582,
      "grad_norm": 0.503209114074707,
      "learning_rate": 0.002828,
      "loss": 1.2746,
      "step": 282752
    },
    {
      "epoch": 21.224465290806755,
      "grad_norm": 0.5236272215843201,
      "learning_rate": 0.002828,
      "loss": 1.2779,
      "step": 282816
    },
    {
      "epoch": 21.229268292682928,
      "grad_norm": 0.6208475828170776,
      "learning_rate": 0.002828,
      "loss": 1.2789,
      "step": 282880
    },
    {
      "epoch": 21.2340712945591,
      "grad_norm": 0.5424129962921143,
      "learning_rate": 0.002828,
      "loss": 1.2768,
      "step": 282944
    },
    {
      "epoch": 21.238874296435274,
      "grad_norm": 0.6895538568496704,
      "learning_rate": 0.002828,
      "loss": 1.2798,
      "step": 283008
    },
    {
      "epoch": 21.243677298311443,
      "grad_norm": 0.5749019384384155,
      "learning_rate": 0.002828,
      "loss": 1.2832,
      "step": 283072
    },
    {
      "epoch": 21.248480300187616,
      "grad_norm": 0.6259738206863403,
      "learning_rate": 0.002828,
      "loss": 1.272,
      "step": 283136
    },
    {
      "epoch": 21.25328330206379,
      "grad_norm": 0.4793575406074524,
      "learning_rate": 0.002828,
      "loss": 1.2798,
      "step": 283200
    },
    {
      "epoch": 21.25808630393996,
      "grad_norm": 0.5138023495674133,
      "learning_rate": 0.002828,
      "loss": 1.2779,
      "step": 283264
    },
    {
      "epoch": 21.262889305816135,
      "grad_norm": 0.49802491068840027,
      "learning_rate": 0.002828,
      "loss": 1.2779,
      "step": 283328
    },
    {
      "epoch": 21.267692307692307,
      "grad_norm": 0.6311704516410828,
      "learning_rate": 0.002828,
      "loss": 1.2791,
      "step": 283392
    },
    {
      "epoch": 21.27249530956848,
      "grad_norm": 0.5918208956718445,
      "learning_rate": 0.002828,
      "loss": 1.2738,
      "step": 283456
    },
    {
      "epoch": 21.277298311444653,
      "grad_norm": 0.5432458519935608,
      "learning_rate": 0.002828,
      "loss": 1.2774,
      "step": 283520
    },
    {
      "epoch": 21.282101313320826,
      "grad_norm": 0.5453625917434692,
      "learning_rate": 0.002828,
      "loss": 1.2743,
      "step": 283584
    },
    {
      "epoch": 21.286904315197,
      "grad_norm": 0.577239990234375,
      "learning_rate": 0.002828,
      "loss": 1.2859,
      "step": 283648
    },
    {
      "epoch": 21.291707317073172,
      "grad_norm": 0.5318759679794312,
      "learning_rate": 0.002828,
      "loss": 1.2773,
      "step": 283712
    },
    {
      "epoch": 21.296510318949345,
      "grad_norm": 0.5675327181816101,
      "learning_rate": 0.002828,
      "loss": 1.2796,
      "step": 283776
    },
    {
      "epoch": 21.301313320825518,
      "grad_norm": 0.5135879516601562,
      "learning_rate": 0.002828,
      "loss": 1.2753,
      "step": 283840
    },
    {
      "epoch": 21.306116322701687,
      "grad_norm": 0.6239807605743408,
      "learning_rate": 0.002828,
      "loss": 1.2774,
      "step": 283904
    },
    {
      "epoch": 21.31091932457786,
      "grad_norm": 0.5842599272727966,
      "learning_rate": 0.002828,
      "loss": 1.2775,
      "step": 283968
    },
    {
      "epoch": 21.315722326454033,
      "grad_norm": 0.6358805894851685,
      "learning_rate": 0.002828,
      "loss": 1.2707,
      "step": 284032
    },
    {
      "epoch": 21.320525328330206,
      "grad_norm": 0.6516125202178955,
      "learning_rate": 0.002828,
      "loss": 1.2791,
      "step": 284096
    },
    {
      "epoch": 21.32532833020638,
      "grad_norm": 0.5607619285583496,
      "learning_rate": 0.002828,
      "loss": 1.2773,
      "step": 284160
    },
    {
      "epoch": 21.33013133208255,
      "grad_norm": 0.596806526184082,
      "learning_rate": 0.002828,
      "loss": 1.2762,
      "step": 284224
    },
    {
      "epoch": 21.334934333958724,
      "grad_norm": 0.6293962001800537,
      "learning_rate": 0.002828,
      "loss": 1.277,
      "step": 284288
    },
    {
      "epoch": 21.339737335834897,
      "grad_norm": 0.5905919075012207,
      "learning_rate": 0.002828,
      "loss": 1.2821,
      "step": 284352
    },
    {
      "epoch": 21.34454033771107,
      "grad_norm": 0.6652618646621704,
      "learning_rate": 0.002828,
      "loss": 1.2787,
      "step": 284416
    },
    {
      "epoch": 21.349343339587243,
      "grad_norm": 0.648979663848877,
      "learning_rate": 0.002828,
      "loss": 1.2781,
      "step": 284480
    },
    {
      "epoch": 21.354146341463416,
      "grad_norm": 0.5890728831291199,
      "learning_rate": 0.002828,
      "loss": 1.2785,
      "step": 284544
    },
    {
      "epoch": 21.35894934333959,
      "grad_norm": 0.7403590083122253,
      "learning_rate": 0.002828,
      "loss": 1.285,
      "step": 284608
    },
    {
      "epoch": 21.363752345215758,
      "grad_norm": 0.5386162996292114,
      "learning_rate": 0.002828,
      "loss": 1.2775,
      "step": 284672
    },
    {
      "epoch": 21.36855534709193,
      "grad_norm": 0.5761041641235352,
      "learning_rate": 0.002828,
      "loss": 1.2713,
      "step": 284736
    },
    {
      "epoch": 21.373358348968104,
      "grad_norm": 0.6262449622154236,
      "learning_rate": 0.002828,
      "loss": 1.2774,
      "step": 284800
    },
    {
      "epoch": 21.378161350844277,
      "grad_norm": 0.5526410937309265,
      "learning_rate": 0.002828,
      "loss": 1.2798,
      "step": 284864
    },
    {
      "epoch": 21.38296435272045,
      "grad_norm": 0.4797821044921875,
      "learning_rate": 0.002828,
      "loss": 1.2822,
      "step": 284928
    },
    {
      "epoch": 21.387767354596622,
      "grad_norm": 0.5310882925987244,
      "learning_rate": 0.002828,
      "loss": 1.2833,
      "step": 284992
    },
    {
      "epoch": 21.392570356472795,
      "grad_norm": 0.5559850931167603,
      "learning_rate": 0.002828,
      "loss": 1.2823,
      "step": 285056
    },
    {
      "epoch": 21.39737335834897,
      "grad_norm": 0.6088487505912781,
      "learning_rate": 0.002828,
      "loss": 1.2767,
      "step": 285120
    },
    {
      "epoch": 21.40217636022514,
      "grad_norm": 0.6352992057800293,
      "learning_rate": 0.002828,
      "loss": 1.2831,
      "step": 285184
    },
    {
      "epoch": 21.406979362101314,
      "grad_norm": 0.5523979067802429,
      "learning_rate": 0.002828,
      "loss": 1.286,
      "step": 285248
    },
    {
      "epoch": 21.411782363977487,
      "grad_norm": 0.5660821795463562,
      "learning_rate": 0.002828,
      "loss": 1.2825,
      "step": 285312
    },
    {
      "epoch": 21.41658536585366,
      "grad_norm": 0.5910194516181946,
      "learning_rate": 0.002828,
      "loss": 1.2803,
      "step": 285376
    },
    {
      "epoch": 21.421388367729833,
      "grad_norm": 0.6662206649780273,
      "learning_rate": 0.002828,
      "loss": 1.2854,
      "step": 285440
    },
    {
      "epoch": 21.426191369606002,
      "grad_norm": 0.6278867125511169,
      "learning_rate": 0.002828,
      "loss": 1.2768,
      "step": 285504
    },
    {
      "epoch": 21.430994371482175,
      "grad_norm": 0.4952567219734192,
      "learning_rate": 0.002828,
      "loss": 1.2753,
      "step": 285568
    },
    {
      "epoch": 21.435797373358348,
      "grad_norm": 0.5748569369316101,
      "learning_rate": 0.002828,
      "loss": 1.2827,
      "step": 285632
    },
    {
      "epoch": 21.44060037523452,
      "grad_norm": 0.6461130976676941,
      "learning_rate": 0.002828,
      "loss": 1.2861,
      "step": 285696
    },
    {
      "epoch": 21.445403377110694,
      "grad_norm": 0.5584768652915955,
      "learning_rate": 0.002828,
      "loss": 1.2838,
      "step": 285760
    },
    {
      "epoch": 21.450206378986866,
      "grad_norm": 0.5767645239830017,
      "learning_rate": 0.002828,
      "loss": 1.2748,
      "step": 285824
    },
    {
      "epoch": 21.45500938086304,
      "grad_norm": 0.5557846426963806,
      "learning_rate": 0.002828,
      "loss": 1.2789,
      "step": 285888
    },
    {
      "epoch": 21.459812382739212,
      "grad_norm": 0.6409792304039001,
      "learning_rate": 0.002828,
      "loss": 1.2827,
      "step": 285952
    },
    {
      "epoch": 21.464615384615385,
      "grad_norm": 0.5421139597892761,
      "learning_rate": 0.002828,
      "loss": 1.2761,
      "step": 286016
    },
    {
      "epoch": 21.469418386491558,
      "grad_norm": 0.5195848941802979,
      "learning_rate": 0.002828,
      "loss": 1.2768,
      "step": 286080
    },
    {
      "epoch": 21.47422138836773,
      "grad_norm": 0.7797949314117432,
      "learning_rate": 0.002828,
      "loss": 1.2837,
      "step": 286144
    },
    {
      "epoch": 21.479024390243904,
      "grad_norm": 0.544447124004364,
      "learning_rate": 0.002828,
      "loss": 1.2879,
      "step": 286208
    },
    {
      "epoch": 21.483827392120077,
      "grad_norm": 0.6963238716125488,
      "learning_rate": 0.002828,
      "loss": 1.2911,
      "step": 286272
    },
    {
      "epoch": 21.488630393996246,
      "grad_norm": 0.6099432110786438,
      "learning_rate": 0.002828,
      "loss": 1.2825,
      "step": 286336
    },
    {
      "epoch": 21.49343339587242,
      "grad_norm": 0.5661420822143555,
      "learning_rate": 0.002828,
      "loss": 1.2758,
      "step": 286400
    },
    {
      "epoch": 21.49823639774859,
      "grad_norm": 0.5302500128746033,
      "learning_rate": 0.002828,
      "loss": 1.2813,
      "step": 286464
    },
    {
      "epoch": 21.503039399624765,
      "grad_norm": 0.5726025104522705,
      "learning_rate": 0.002828,
      "loss": 1.2836,
      "step": 286528
    },
    {
      "epoch": 21.507842401500938,
      "grad_norm": 0.6002224683761597,
      "learning_rate": 0.002828,
      "loss": 1.2858,
      "step": 286592
    },
    {
      "epoch": 21.51264540337711,
      "grad_norm": 0.5370309948921204,
      "learning_rate": 0.002828,
      "loss": 1.282,
      "step": 286656
    },
    {
      "epoch": 21.517448405253283,
      "grad_norm": 0.5378057360649109,
      "learning_rate": 0.002828,
      "loss": 1.2857,
      "step": 286720
    },
    {
      "epoch": 21.522251407129456,
      "grad_norm": 0.5748156309127808,
      "learning_rate": 0.002828,
      "loss": 1.2805,
      "step": 286784
    },
    {
      "epoch": 21.52705440900563,
      "grad_norm": 0.6204039454460144,
      "learning_rate": 0.002828,
      "loss": 1.2848,
      "step": 286848
    },
    {
      "epoch": 21.531857410881802,
      "grad_norm": 0.5553511381149292,
      "learning_rate": 0.002828,
      "loss": 1.285,
      "step": 286912
    },
    {
      "epoch": 21.536660412757975,
      "grad_norm": 0.5893772840499878,
      "learning_rate": 0.002828,
      "loss": 1.287,
      "step": 286976
    },
    {
      "epoch": 21.541463414634148,
      "grad_norm": 0.6876904964447021,
      "learning_rate": 0.002828,
      "loss": 1.2822,
      "step": 287040
    },
    {
      "epoch": 21.54626641651032,
      "grad_norm": 0.5917910933494568,
      "learning_rate": 0.002828,
      "loss": 1.2772,
      "step": 287104
    },
    {
      "epoch": 21.55106941838649,
      "grad_norm": 0.6226091980934143,
      "learning_rate": 0.002828,
      "loss": 1.2824,
      "step": 287168
    },
    {
      "epoch": 21.555872420262663,
      "grad_norm": 0.5301666259765625,
      "learning_rate": 0.002828,
      "loss": 1.2803,
      "step": 287232
    },
    {
      "epoch": 21.560675422138836,
      "grad_norm": 0.6062954664230347,
      "learning_rate": 0.002828,
      "loss": 1.2887,
      "step": 287296
    },
    {
      "epoch": 21.56547842401501,
      "grad_norm": 0.514739990234375,
      "learning_rate": 0.002828,
      "loss": 1.2813,
      "step": 287360
    },
    {
      "epoch": 21.57028142589118,
      "grad_norm": 0.5676846504211426,
      "learning_rate": 0.002828,
      "loss": 1.2752,
      "step": 287424
    },
    {
      "epoch": 21.575084427767354,
      "grad_norm": 0.5088371634483337,
      "learning_rate": 0.002828,
      "loss": 1.2806,
      "step": 287488
    },
    {
      "epoch": 21.579887429643527,
      "grad_norm": 0.5662075281143188,
      "learning_rate": 0.002828,
      "loss": 1.2808,
      "step": 287552
    },
    {
      "epoch": 21.5846904315197,
      "grad_norm": 0.5890974402427673,
      "learning_rate": 0.002828,
      "loss": 1.2865,
      "step": 287616
    },
    {
      "epoch": 21.589493433395873,
      "grad_norm": 0.7173524498939514,
      "learning_rate": 0.002828,
      "loss": 1.2864,
      "step": 287680
    },
    {
      "epoch": 21.594296435272046,
      "grad_norm": 0.5397960543632507,
      "learning_rate": 0.002828,
      "loss": 1.2835,
      "step": 287744
    },
    {
      "epoch": 21.59909943714822,
      "grad_norm": 0.6206925511360168,
      "learning_rate": 0.002828,
      "loss": 1.2801,
      "step": 287808
    },
    {
      "epoch": 21.60390243902439,
      "grad_norm": 0.5514209270477295,
      "learning_rate": 0.002828,
      "loss": 1.2858,
      "step": 287872
    },
    {
      "epoch": 21.608705440900565,
      "grad_norm": 0.4978121817111969,
      "learning_rate": 0.002828,
      "loss": 1.281,
      "step": 287936
    },
    {
      "epoch": 21.613508442776734,
      "grad_norm": 0.555472195148468,
      "learning_rate": 0.002828,
      "loss": 1.2894,
      "step": 288000
    },
    {
      "epoch": 21.618311444652907,
      "grad_norm": 0.6257559061050415,
      "learning_rate": 0.002828,
      "loss": 1.2929,
      "step": 288064
    },
    {
      "epoch": 21.62311444652908,
      "grad_norm": 0.5221234560012817,
      "learning_rate": 0.002828,
      "loss": 1.2959,
      "step": 288128
    },
    {
      "epoch": 21.627917448405253,
      "grad_norm": 0.6040545105934143,
      "learning_rate": 0.002828,
      "loss": 1.2855,
      "step": 288192
    },
    {
      "epoch": 21.632720450281425,
      "grad_norm": 0.6095830202102661,
      "learning_rate": 0.002828,
      "loss": 1.2895,
      "step": 288256
    },
    {
      "epoch": 21.6375234521576,
      "grad_norm": 0.6214110851287842,
      "learning_rate": 0.002828,
      "loss": 1.2862,
      "step": 288320
    },
    {
      "epoch": 21.64232645403377,
      "grad_norm": 0.6199238300323486,
      "learning_rate": 0.002828,
      "loss": 1.2877,
      "step": 288384
    },
    {
      "epoch": 21.647129455909944,
      "grad_norm": 0.637213945388794,
      "learning_rate": 0.002828,
      "loss": 1.2838,
      "step": 288448
    },
    {
      "epoch": 21.651932457786117,
      "grad_norm": 0.6265941262245178,
      "learning_rate": 0.002828,
      "loss": 1.2866,
      "step": 288512
    },
    {
      "epoch": 21.65673545966229,
      "grad_norm": 0.6128517985343933,
      "learning_rate": 0.002828,
      "loss": 1.2806,
      "step": 288576
    },
    {
      "epoch": 21.661538461538463,
      "grad_norm": 0.6448500752449036,
      "learning_rate": 0.002828,
      "loss": 1.2811,
      "step": 288640
    },
    {
      "epoch": 21.666341463414636,
      "grad_norm": 0.5811263918876648,
      "learning_rate": 0.002828,
      "loss": 1.2828,
      "step": 288704
    },
    {
      "epoch": 21.67114446529081,
      "grad_norm": 0.607732892036438,
      "learning_rate": 0.002828,
      "loss": 1.2848,
      "step": 288768
    },
    {
      "epoch": 21.675947467166978,
      "grad_norm": 0.50178462266922,
      "learning_rate": 0.002828,
      "loss": 1.2832,
      "step": 288832
    },
    {
      "epoch": 21.68075046904315,
      "grad_norm": 0.4868558347225189,
      "learning_rate": 0.002828,
      "loss": 1.28,
      "step": 288896
    },
    {
      "epoch": 21.685553470919324,
      "grad_norm": 0.5406298041343689,
      "learning_rate": 0.002828,
      "loss": 1.2916,
      "step": 288960
    },
    {
      "epoch": 21.690356472795497,
      "grad_norm": 0.5491721034049988,
      "learning_rate": 0.002828,
      "loss": 1.2882,
      "step": 289024
    },
    {
      "epoch": 21.69515947467167,
      "grad_norm": 0.5529964566230774,
      "learning_rate": 0.002828,
      "loss": 1.2872,
      "step": 289088
    },
    {
      "epoch": 21.699962476547842,
      "grad_norm": 0.5563608407974243,
      "learning_rate": 0.002828,
      "loss": 1.2884,
      "step": 289152
    },
    {
      "epoch": 21.704765478424015,
      "grad_norm": 0.5522491931915283,
      "learning_rate": 0.002828,
      "loss": 1.279,
      "step": 289216
    },
    {
      "epoch": 21.709568480300188,
      "grad_norm": 0.5772581696510315,
      "learning_rate": 0.002828,
      "loss": 1.2892,
      "step": 289280
    },
    {
      "epoch": 21.71437148217636,
      "grad_norm": 0.5620320439338684,
      "learning_rate": 0.002828,
      "loss": 1.2852,
      "step": 289344
    },
    {
      "epoch": 21.719174484052534,
      "grad_norm": 0.5383132100105286,
      "learning_rate": 0.002828,
      "loss": 1.2871,
      "step": 289408
    },
    {
      "epoch": 21.723977485928707,
      "grad_norm": 0.6416934132575989,
      "learning_rate": 0.002828,
      "loss": 1.2857,
      "step": 289472
    },
    {
      "epoch": 21.72878048780488,
      "grad_norm": 0.6494410037994385,
      "learning_rate": 0.002828,
      "loss": 1.2897,
      "step": 289536
    },
    {
      "epoch": 21.733583489681052,
      "grad_norm": 0.5881087183952332,
      "learning_rate": 0.002828,
      "loss": 1.2848,
      "step": 289600
    },
    {
      "epoch": 21.738386491557222,
      "grad_norm": 0.4864734411239624,
      "learning_rate": 0.002828,
      "loss": 1.2854,
      "step": 289664
    },
    {
      "epoch": 21.743189493433395,
      "grad_norm": 0.4938899874687195,
      "learning_rate": 0.002828,
      "loss": 1.2883,
      "step": 289728
    },
    {
      "epoch": 21.747992495309568,
      "grad_norm": 0.5875544548034668,
      "learning_rate": 0.002828,
      "loss": 1.2829,
      "step": 289792
    },
    {
      "epoch": 21.75279549718574,
      "grad_norm": 0.48364803194999695,
      "learning_rate": 0.002828,
      "loss": 1.2861,
      "step": 289856
    },
    {
      "epoch": 21.757598499061913,
      "grad_norm": 0.5670167207717896,
      "learning_rate": 0.002828,
      "loss": 1.2867,
      "step": 289920
    },
    {
      "epoch": 21.762401500938086,
      "grad_norm": 0.5402014851570129,
      "learning_rate": 0.002828,
      "loss": 1.2809,
      "step": 289984
    },
    {
      "epoch": 21.76720450281426,
      "grad_norm": 0.6188728213310242,
      "learning_rate": 0.002828,
      "loss": 1.2861,
      "step": 290048
    },
    {
      "epoch": 21.772007504690432,
      "grad_norm": 0.4771275222301483,
      "learning_rate": 0.002828,
      "loss": 1.2866,
      "step": 290112
    },
    {
      "epoch": 21.776810506566605,
      "grad_norm": 0.5720518827438354,
      "learning_rate": 0.002828,
      "loss": 1.2884,
      "step": 290176
    },
    {
      "epoch": 21.781613508442778,
      "grad_norm": 0.6083287596702576,
      "learning_rate": 0.002828,
      "loss": 1.2871,
      "step": 290240
    },
    {
      "epoch": 21.78641651031895,
      "grad_norm": 0.7133959531784058,
      "learning_rate": 0.002828,
      "loss": 1.2878,
      "step": 290304
    },
    {
      "epoch": 21.791219512195124,
      "grad_norm": 0.6493473052978516,
      "learning_rate": 0.002828,
      "loss": 1.2842,
      "step": 290368
    },
    {
      "epoch": 21.796022514071293,
      "grad_norm": 0.6049429774284363,
      "learning_rate": 0.002828,
      "loss": 1.2845,
      "step": 290432
    },
    {
      "epoch": 21.800825515947466,
      "grad_norm": 0.559673011302948,
      "learning_rate": 0.002828,
      "loss": 1.2919,
      "step": 290496
    },
    {
      "epoch": 21.80562851782364,
      "grad_norm": 0.5735464692115784,
      "learning_rate": 0.002828,
      "loss": 1.2887,
      "step": 290560
    },
    {
      "epoch": 21.81043151969981,
      "grad_norm": 0.633755087852478,
      "learning_rate": 0.002828,
      "loss": 1.2823,
      "step": 290624
    },
    {
      "epoch": 21.815234521575984,
      "grad_norm": 0.5874369740486145,
      "learning_rate": 0.002828,
      "loss": 1.2963,
      "step": 290688
    },
    {
      "epoch": 21.820037523452157,
      "grad_norm": 0.890895426273346,
      "learning_rate": 0.002828,
      "loss": 1.2936,
      "step": 290752
    },
    {
      "epoch": 21.82484052532833,
      "grad_norm": 0.7190248370170593,
      "learning_rate": 0.002828,
      "loss": 1.2889,
      "step": 290816
    },
    {
      "epoch": 21.829643527204503,
      "grad_norm": 0.5849973559379578,
      "learning_rate": 0.002828,
      "loss": 1.2882,
      "step": 290880
    },
    {
      "epoch": 21.834446529080676,
      "grad_norm": 0.5190068483352661,
      "learning_rate": 0.002828,
      "loss": 1.2789,
      "step": 290944
    },
    {
      "epoch": 21.83924953095685,
      "grad_norm": 0.5481669306755066,
      "learning_rate": 0.002828,
      "loss": 1.2915,
      "step": 291008
    },
    {
      "epoch": 21.84405253283302,
      "grad_norm": 0.5439419746398926,
      "learning_rate": 0.002828,
      "loss": 1.2856,
      "step": 291072
    },
    {
      "epoch": 21.848855534709195,
      "grad_norm": 0.6238653063774109,
      "learning_rate": 0.002828,
      "loss": 1.2911,
      "step": 291136
    },
    {
      "epoch": 21.853658536585368,
      "grad_norm": 0.4976770877838135,
      "learning_rate": 0.002828,
      "loss": 1.2947,
      "step": 291200
    },
    {
      "epoch": 21.858461538461537,
      "grad_norm": 0.578702449798584,
      "learning_rate": 0.002828,
      "loss": 1.2854,
      "step": 291264
    },
    {
      "epoch": 21.86326454033771,
      "grad_norm": 0.6257563829421997,
      "learning_rate": 0.002828,
      "loss": 1.2936,
      "step": 291328
    },
    {
      "epoch": 21.868067542213883,
      "grad_norm": 0.5133591294288635,
      "learning_rate": 0.002828,
      "loss": 1.2864,
      "step": 291392
    },
    {
      "epoch": 21.872870544090055,
      "grad_norm": 0.500873327255249,
      "learning_rate": 0.002828,
      "loss": 1.2851,
      "step": 291456
    },
    {
      "epoch": 21.87767354596623,
      "grad_norm": 0.7154791951179504,
      "learning_rate": 0.002828,
      "loss": 1.2836,
      "step": 291520
    },
    {
      "epoch": 21.8824765478424,
      "grad_norm": 0.5378066897392273,
      "learning_rate": 0.002828,
      "loss": 1.283,
      "step": 291584
    },
    {
      "epoch": 21.887279549718574,
      "grad_norm": 0.5295055508613586,
      "learning_rate": 0.002828,
      "loss": 1.2891,
      "step": 291648
    },
    {
      "epoch": 21.892082551594747,
      "grad_norm": 0.49576306343078613,
      "learning_rate": 0.002828,
      "loss": 1.2863,
      "step": 291712
    },
    {
      "epoch": 21.89688555347092,
      "grad_norm": 0.5278056859970093,
      "learning_rate": 0.002828,
      "loss": 1.29,
      "step": 291776
    },
    {
      "epoch": 21.901688555347093,
      "grad_norm": 0.639045000076294,
      "learning_rate": 0.002828,
      "loss": 1.287,
      "step": 291840
    },
    {
      "epoch": 21.906491557223266,
      "grad_norm": 0.5763165950775146,
      "learning_rate": 0.002828,
      "loss": 1.2885,
      "step": 291904
    },
    {
      "epoch": 21.91129455909944,
      "grad_norm": 0.7740374207496643,
      "learning_rate": 0.002828,
      "loss": 1.2925,
      "step": 291968
    },
    {
      "epoch": 21.91609756097561,
      "grad_norm": 0.4997633993625641,
      "learning_rate": 0.002828,
      "loss": 1.2869,
      "step": 292032
    },
    {
      "epoch": 21.92090056285178,
      "grad_norm": 0.5590546131134033,
      "learning_rate": 0.002828,
      "loss": 1.2802,
      "step": 292096
    },
    {
      "epoch": 21.925703564727954,
      "grad_norm": 0.4461499750614166,
      "learning_rate": 0.002828,
      "loss": 1.2903,
      "step": 292160
    },
    {
      "epoch": 21.930506566604127,
      "grad_norm": 0.550574541091919,
      "learning_rate": 0.002828,
      "loss": 1.2864,
      "step": 292224
    },
    {
      "epoch": 21.9353095684803,
      "grad_norm": 0.6598172783851624,
      "learning_rate": 0.002828,
      "loss": 1.2873,
      "step": 292288
    },
    {
      "epoch": 21.940112570356472,
      "grad_norm": 0.6326315999031067,
      "learning_rate": 0.002828,
      "loss": 1.2884,
      "step": 292352
    },
    {
      "epoch": 21.944915572232645,
      "grad_norm": 0.5537628531455994,
      "learning_rate": 0.002828,
      "loss": 1.2879,
      "step": 292416
    },
    {
      "epoch": 21.949718574108818,
      "grad_norm": 0.4792967736721039,
      "learning_rate": 0.002828,
      "loss": 1.2877,
      "step": 292480
    },
    {
      "epoch": 21.95452157598499,
      "grad_norm": 0.529313325881958,
      "learning_rate": 0.002828,
      "loss": 1.2802,
      "step": 292544
    },
    {
      "epoch": 21.959324577861164,
      "grad_norm": 0.48615750670433044,
      "learning_rate": 0.002828,
      "loss": 1.2925,
      "step": 292608
    },
    {
      "epoch": 21.964127579737337,
      "grad_norm": 0.5281648635864258,
      "learning_rate": 0.002828,
      "loss": 1.289,
      "step": 292672
    },
    {
      "epoch": 21.96893058161351,
      "grad_norm": 0.5400027632713318,
      "learning_rate": 0.002828,
      "loss": 1.2874,
      "step": 292736
    },
    {
      "epoch": 21.973733583489683,
      "grad_norm": 0.5279366374015808,
      "learning_rate": 0.002828,
      "loss": 1.2917,
      "step": 292800
    },
    {
      "epoch": 21.978536585365852,
      "grad_norm": 0.5460038185119629,
      "learning_rate": 0.002828,
      "loss": 1.2986,
      "step": 292864
    },
    {
      "epoch": 21.983339587242025,
      "grad_norm": 0.6252233982086182,
      "learning_rate": 0.002828,
      "loss": 1.2908,
      "step": 292928
    },
    {
      "epoch": 21.988142589118198,
      "grad_norm": 0.6752356290817261,
      "learning_rate": 0.002828,
      "loss": 1.2835,
      "step": 292992
    },
    {
      "epoch": 21.99294559099437,
      "grad_norm": 0.6158551573753357,
      "learning_rate": 0.002828,
      "loss": 1.2886,
      "step": 293056
    },
    {
      "epoch": 21.997748592870543,
      "grad_norm": 0.6006995439529419,
      "learning_rate": 0.002828,
      "loss": 1.2917,
      "step": 293120
    },
    {
      "epoch": 22.002551594746716,
      "grad_norm": 0.5210825204849243,
      "learning_rate": 0.002828,
      "loss": 1.2682,
      "step": 293184
    },
    {
      "epoch": 22.00735459662289,
      "grad_norm": 0.655004620552063,
      "learning_rate": 0.002828,
      "loss": 1.2534,
      "step": 293248
    },
    {
      "epoch": 22.012157598499062,
      "grad_norm": 0.46713322401046753,
      "learning_rate": 0.002828,
      "loss": 1.2493,
      "step": 293312
    },
    {
      "epoch": 22.016960600375235,
      "grad_norm": 0.6327230334281921,
      "learning_rate": 0.002828,
      "loss": 1.2514,
      "step": 293376
    },
    {
      "epoch": 22.021763602251408,
      "grad_norm": 0.5137706995010376,
      "learning_rate": 0.002828,
      "loss": 1.2518,
      "step": 293440
    },
    {
      "epoch": 22.02656660412758,
      "grad_norm": 0.5789818167686462,
      "learning_rate": 0.002828,
      "loss": 1.2493,
      "step": 293504
    },
    {
      "epoch": 22.031369606003754,
      "grad_norm": 0.5683451294898987,
      "learning_rate": 0.002828,
      "loss": 1.2462,
      "step": 293568
    },
    {
      "epoch": 22.036172607879926,
      "grad_norm": 0.6083554029464722,
      "learning_rate": 0.002828,
      "loss": 1.2404,
      "step": 293632
    },
    {
      "epoch": 22.040975609756096,
      "grad_norm": 0.5888787508010864,
      "learning_rate": 0.002828,
      "loss": 1.2517,
      "step": 293696
    },
    {
      "epoch": 22.04577861163227,
      "grad_norm": 0.6548123955726624,
      "learning_rate": 0.002828,
      "loss": 1.2501,
      "step": 293760
    },
    {
      "epoch": 22.05058161350844,
      "grad_norm": 0.6625087857246399,
      "learning_rate": 0.002828,
      "loss": 1.2545,
      "step": 293824
    },
    {
      "epoch": 22.055384615384614,
      "grad_norm": 0.6816575527191162,
      "learning_rate": 0.002828,
      "loss": 1.2508,
      "step": 293888
    },
    {
      "epoch": 22.060187617260787,
      "grad_norm": 0.6531043648719788,
      "learning_rate": 0.002828,
      "loss": 1.2507,
      "step": 293952
    },
    {
      "epoch": 22.06499061913696,
      "grad_norm": 0.5314988493919373,
      "learning_rate": 0.002828,
      "loss": 1.2502,
      "step": 294016
    },
    {
      "epoch": 22.069793621013133,
      "grad_norm": 0.5735332369804382,
      "learning_rate": 0.002828,
      "loss": 1.257,
      "step": 294080
    },
    {
      "epoch": 22.074596622889306,
      "grad_norm": 0.6446841955184937,
      "learning_rate": 0.002828,
      "loss": 1.2588,
      "step": 294144
    },
    {
      "epoch": 22.07939962476548,
      "grad_norm": 0.5817360281944275,
      "learning_rate": 0.002828,
      "loss": 1.255,
      "step": 294208
    },
    {
      "epoch": 22.084202626641652,
      "grad_norm": 0.5101385712623596,
      "learning_rate": 0.002828,
      "loss": 1.2557,
      "step": 294272
    },
    {
      "epoch": 22.089005628517825,
      "grad_norm": 0.5546039342880249,
      "learning_rate": 0.002828,
      "loss": 1.254,
      "step": 294336
    },
    {
      "epoch": 22.093808630393998,
      "grad_norm": 0.5416100025177002,
      "learning_rate": 0.002828,
      "loss": 1.256,
      "step": 294400
    },
    {
      "epoch": 22.09861163227017,
      "grad_norm": 0.5552825331687927,
      "learning_rate": 0.002828,
      "loss": 1.2531,
      "step": 294464
    },
    {
      "epoch": 22.10341463414634,
      "grad_norm": 0.5232527256011963,
      "learning_rate": 0.002828,
      "loss": 1.2585,
      "step": 294528
    },
    {
      "epoch": 22.108217636022513,
      "grad_norm": 0.5346928834915161,
      "learning_rate": 0.002828,
      "loss": 1.2496,
      "step": 294592
    },
    {
      "epoch": 22.113020637898686,
      "grad_norm": 0.5848618149757385,
      "learning_rate": 0.002828,
      "loss": 1.2561,
      "step": 294656
    },
    {
      "epoch": 22.11782363977486,
      "grad_norm": 0.5828258991241455,
      "learning_rate": 0.002828,
      "loss": 1.2582,
      "step": 294720
    },
    {
      "epoch": 22.12262664165103,
      "grad_norm": 0.5497180223464966,
      "learning_rate": 0.002828,
      "loss": 1.2537,
      "step": 294784
    },
    {
      "epoch": 22.127429643527204,
      "grad_norm": 0.5274826884269714,
      "learning_rate": 0.002828,
      "loss": 1.2553,
      "step": 294848
    },
    {
      "epoch": 22.132232645403377,
      "grad_norm": 0.5913788676261902,
      "learning_rate": 0.002828,
      "loss": 1.2567,
      "step": 294912
    },
    {
      "epoch": 22.13703564727955,
      "grad_norm": 0.5481519103050232,
      "learning_rate": 0.002828,
      "loss": 1.2555,
      "step": 294976
    },
    {
      "epoch": 22.141838649155723,
      "grad_norm": 0.625358521938324,
      "learning_rate": 0.002828,
      "loss": 1.2606,
      "step": 295040
    },
    {
      "epoch": 22.146641651031896,
      "grad_norm": 0.5830749869346619,
      "learning_rate": 0.002828,
      "loss": 1.2586,
      "step": 295104
    },
    {
      "epoch": 22.15144465290807,
      "grad_norm": 0.5845195651054382,
      "learning_rate": 0.002828,
      "loss": 1.2582,
      "step": 295168
    },
    {
      "epoch": 22.15624765478424,
      "grad_norm": 0.5070827007293701,
      "learning_rate": 0.002828,
      "loss": 1.2565,
      "step": 295232
    },
    {
      "epoch": 22.161050656660414,
      "grad_norm": 0.5942825675010681,
      "learning_rate": 0.002828,
      "loss": 1.2552,
      "step": 295296
    },
    {
      "epoch": 22.165853658536584,
      "grad_norm": 0.4849054515361786,
      "learning_rate": 0.002828,
      "loss": 1.2574,
      "step": 295360
    },
    {
      "epoch": 22.170656660412757,
      "grad_norm": 0.630711019039154,
      "learning_rate": 0.002828,
      "loss": 1.2635,
      "step": 295424
    },
    {
      "epoch": 22.17545966228893,
      "grad_norm": 0.6968967914581299,
      "learning_rate": 0.002828,
      "loss": 1.2608,
      "step": 295488
    },
    {
      "epoch": 22.180262664165102,
      "grad_norm": 0.5041672587394714,
      "learning_rate": 0.002828,
      "loss": 1.2606,
      "step": 295552
    },
    {
      "epoch": 22.185065666041275,
      "grad_norm": 0.49958130717277527,
      "learning_rate": 0.002828,
      "loss": 1.2538,
      "step": 295616
    },
    {
      "epoch": 22.189868667917448,
      "grad_norm": 0.6439744234085083,
      "learning_rate": 0.002828,
      "loss": 1.262,
      "step": 295680
    },
    {
      "epoch": 22.19467166979362,
      "grad_norm": 0.6664990186691284,
      "learning_rate": 0.002828,
      "loss": 1.2547,
      "step": 295744
    },
    {
      "epoch": 22.199474671669794,
      "grad_norm": 0.6299195885658264,
      "learning_rate": 0.002828,
      "loss": 1.2589,
      "step": 295808
    },
    {
      "epoch": 22.204277673545967,
      "grad_norm": 0.581447422504425,
      "learning_rate": 0.002828,
      "loss": 1.2592,
      "step": 295872
    },
    {
      "epoch": 22.20908067542214,
      "grad_norm": 0.5462125539779663,
      "learning_rate": 0.002828,
      "loss": 1.2623,
      "step": 295936
    },
    {
      "epoch": 22.213883677298313,
      "grad_norm": 0.5986690521240234,
      "learning_rate": 0.002828,
      "loss": 1.2597,
      "step": 296000
    },
    {
      "epoch": 22.218686679174485,
      "grad_norm": 0.5245385766029358,
      "learning_rate": 0.002828,
      "loss": 1.2536,
      "step": 296064
    },
    {
      "epoch": 22.22348968105066,
      "grad_norm": 0.7947522401809692,
      "learning_rate": 0.002828,
      "loss": 1.2589,
      "step": 296128
    },
    {
      "epoch": 22.228292682926828,
      "grad_norm": 0.5781105756759644,
      "learning_rate": 0.002828,
      "loss": 1.2614,
      "step": 296192
    },
    {
      "epoch": 22.233095684803,
      "grad_norm": 0.5033446550369263,
      "learning_rate": 0.002828,
      "loss": 1.2594,
      "step": 296256
    },
    {
      "epoch": 22.237898686679173,
      "grad_norm": 0.5990842580795288,
      "learning_rate": 0.002828,
      "loss": 1.2634,
      "step": 296320
    },
    {
      "epoch": 22.242701688555346,
      "grad_norm": 0.549052357673645,
      "learning_rate": 0.002828,
      "loss": 1.2574,
      "step": 296384
    },
    {
      "epoch": 22.24750469043152,
      "grad_norm": 0.4947829842567444,
      "learning_rate": 0.002828,
      "loss": 1.2612,
      "step": 296448
    },
    {
      "epoch": 22.252307692307692,
      "grad_norm": 0.5878117084503174,
      "learning_rate": 0.002828,
      "loss": 1.2578,
      "step": 296512
    },
    {
      "epoch": 22.257110694183865,
      "grad_norm": 0.5507844686508179,
      "learning_rate": 0.002828,
      "loss": 1.261,
      "step": 296576
    },
    {
      "epoch": 22.261913696060038,
      "grad_norm": 0.5388092398643494,
      "learning_rate": 0.002828,
      "loss": 1.2651,
      "step": 296640
    },
    {
      "epoch": 22.26671669793621,
      "grad_norm": 0.46271833777427673,
      "learning_rate": 0.002828,
      "loss": 1.2599,
      "step": 296704
    },
    {
      "epoch": 22.271519699812384,
      "grad_norm": 0.7457178235054016,
      "learning_rate": 0.002828,
      "loss": 1.2606,
      "step": 296768
    },
    {
      "epoch": 22.276322701688557,
      "grad_norm": 0.6202179789543152,
      "learning_rate": 0.002828,
      "loss": 1.2644,
      "step": 296832
    },
    {
      "epoch": 22.28112570356473,
      "grad_norm": 0.5775248408317566,
      "learning_rate": 0.002828,
      "loss": 1.2643,
      "step": 296896
    },
    {
      "epoch": 22.285928705440902,
      "grad_norm": 0.7320228219032288,
      "learning_rate": 0.002828,
      "loss": 1.2631,
      "step": 296960
    },
    {
      "epoch": 22.29073170731707,
      "grad_norm": 0.6378146409988403,
      "learning_rate": 0.002828,
      "loss": 1.2634,
      "step": 297024
    },
    {
      "epoch": 22.295534709193245,
      "grad_norm": 0.5487433075904846,
      "learning_rate": 0.002828,
      "loss": 1.2545,
      "step": 297088
    },
    {
      "epoch": 22.300337711069417,
      "grad_norm": 0.5953262448310852,
      "learning_rate": 0.002828,
      "loss": 1.265,
      "step": 297152
    },
    {
      "epoch": 22.30514071294559,
      "grad_norm": 0.6441254615783691,
      "learning_rate": 0.002828,
      "loss": 1.2678,
      "step": 297216
    },
    {
      "epoch": 22.309943714821763,
      "grad_norm": 0.6376492381095886,
      "learning_rate": 0.002828,
      "loss": 1.2668,
      "step": 297280
    },
    {
      "epoch": 22.314746716697936,
      "grad_norm": 0.535435676574707,
      "learning_rate": 0.002828,
      "loss": 1.2593,
      "step": 297344
    },
    {
      "epoch": 22.31954971857411,
      "grad_norm": 0.6632246971130371,
      "learning_rate": 0.002828,
      "loss": 1.2677,
      "step": 297408
    },
    {
      "epoch": 22.324352720450282,
      "grad_norm": 0.46085768938064575,
      "learning_rate": 0.002828,
      "loss": 1.2647,
      "step": 297472
    },
    {
      "epoch": 22.329155722326455,
      "grad_norm": 0.5245596170425415,
      "learning_rate": 0.002828,
      "loss": 1.2665,
      "step": 297536
    },
    {
      "epoch": 22.333958724202628,
      "grad_norm": 0.6162623167037964,
      "learning_rate": 0.002828,
      "loss": 1.2651,
      "step": 297600
    },
    {
      "epoch": 22.3387617260788,
      "grad_norm": 0.5637834072113037,
      "learning_rate": 0.002828,
      "loss": 1.2705,
      "step": 297664
    },
    {
      "epoch": 22.343564727954973,
      "grad_norm": 0.575354814529419,
      "learning_rate": 0.002828,
      "loss": 1.262,
      "step": 297728
    },
    {
      "epoch": 22.348367729831143,
      "grad_norm": 0.4700267016887665,
      "learning_rate": 0.002828,
      "loss": 1.2641,
      "step": 297792
    },
    {
      "epoch": 22.353170731707316,
      "grad_norm": 0.6297546029090881,
      "learning_rate": 0.002828,
      "loss": 1.2629,
      "step": 297856
    },
    {
      "epoch": 22.35797373358349,
      "grad_norm": 0.5787842273712158,
      "learning_rate": 0.002828,
      "loss": 1.2688,
      "step": 297920
    },
    {
      "epoch": 22.36277673545966,
      "grad_norm": 0.5737005472183228,
      "learning_rate": 0.002828,
      "loss": 1.2678,
      "step": 297984
    },
    {
      "epoch": 22.367579737335834,
      "grad_norm": 0.5520718693733215,
      "learning_rate": 0.002828,
      "loss": 1.2641,
      "step": 298048
    },
    {
      "epoch": 22.372382739212007,
      "grad_norm": 0.5324095487594604,
      "learning_rate": 0.002828,
      "loss": 1.2664,
      "step": 298112
    },
    {
      "epoch": 22.37718574108818,
      "grad_norm": 0.6231042742729187,
      "learning_rate": 0.002828,
      "loss": 1.2626,
      "step": 298176
    },
    {
      "epoch": 22.381988742964353,
      "grad_norm": 0.5680738091468811,
      "learning_rate": 0.002828,
      "loss": 1.2663,
      "step": 298240
    },
    {
      "epoch": 22.386791744840526,
      "grad_norm": 0.5304076671600342,
      "learning_rate": 0.002828,
      "loss": 1.2431,
      "step": 298304
    },
    {
      "epoch": 22.3915947467167,
      "grad_norm": 0.4264357089996338,
      "learning_rate": 0.002828,
      "loss": 1.2415,
      "step": 298368
    },
    {
      "epoch": 22.39639774859287,
      "grad_norm": 0.5016809105873108,
      "learning_rate": 0.002828,
      "loss": 1.248,
      "step": 298432
    },
    {
      "epoch": 22.401200750469044,
      "grad_norm": 0.5329827666282654,
      "learning_rate": 0.002828,
      "loss": 1.2457,
      "step": 298496
    },
    {
      "epoch": 22.406003752345217,
      "grad_norm": 0.5466082692146301,
      "learning_rate": 0.002828,
      "loss": 1.2464,
      "step": 298560
    },
    {
      "epoch": 22.410806754221387,
      "grad_norm": 0.5093466639518738,
      "learning_rate": 0.002828,
      "loss": 1.2508,
      "step": 298624
    },
    {
      "epoch": 22.41560975609756,
      "grad_norm": 0.5469045639038086,
      "learning_rate": 0.002828,
      "loss": 1.247,
      "step": 298688
    },
    {
      "epoch": 22.420412757973732,
      "grad_norm": 0.5645148158073425,
      "learning_rate": 0.002828,
      "loss": 1.2553,
      "step": 298752
    },
    {
      "epoch": 22.425215759849905,
      "grad_norm": 0.6576793193817139,
      "learning_rate": 0.002828,
      "loss": 1.2499,
      "step": 298816
    },
    {
      "epoch": 22.43001876172608,
      "grad_norm": 0.726359486579895,
      "learning_rate": 0.002828,
      "loss": 1.2541,
      "step": 298880
    },
    {
      "epoch": 22.43482176360225,
      "grad_norm": 0.6088193655014038,
      "learning_rate": 0.002828,
      "loss": 1.2569,
      "step": 298944
    },
    {
      "epoch": 22.439624765478424,
      "grad_norm": 0.6478526592254639,
      "learning_rate": 0.002828,
      "loss": 1.2528,
      "step": 299008
    },
    {
      "epoch": 22.444427767354597,
      "grad_norm": 0.6103188991546631,
      "learning_rate": 0.002828,
      "loss": 1.2513,
      "step": 299072
    },
    {
      "epoch": 22.44923076923077,
      "grad_norm": 0.5749058127403259,
      "learning_rate": 0.002828,
      "loss": 1.2569,
      "step": 299136
    },
    {
      "epoch": 22.454033771106943,
      "grad_norm": 0.5394167900085449,
      "learning_rate": 0.002828,
      "loss": 1.25,
      "step": 299200
    },
    {
      "epoch": 22.458836772983116,
      "grad_norm": 0.650351881980896,
      "learning_rate": 0.002828,
      "loss": 1.2553,
      "step": 299264
    },
    {
      "epoch": 22.46363977485929,
      "grad_norm": 0.5926783680915833,
      "learning_rate": 0.002828,
      "loss": 1.2544,
      "step": 299328
    },
    {
      "epoch": 22.46844277673546,
      "grad_norm": 0.6527881622314453,
      "learning_rate": 0.002828,
      "loss": 1.2531,
      "step": 299392
    },
    {
      "epoch": 22.47324577861163,
      "grad_norm": 0.6134124994277954,
      "learning_rate": 0.002828,
      "loss": 1.2544,
      "step": 299456
    },
    {
      "epoch": 22.478048780487804,
      "grad_norm": 0.5028894543647766,
      "learning_rate": 0.002828,
      "loss": 1.253,
      "step": 299520
    },
    {
      "epoch": 22.482851782363976,
      "grad_norm": 0.6587144136428833,
      "learning_rate": 0.002828,
      "loss": 1.2562,
      "step": 299584
    },
    {
      "epoch": 22.48765478424015,
      "grad_norm": 0.6202356219291687,
      "learning_rate": 0.002828,
      "loss": 1.2548,
      "step": 299648
    },
    {
      "epoch": 22.492457786116322,
      "grad_norm": 0.5491941571235657,
      "learning_rate": 0.002828,
      "loss": 1.2548,
      "step": 299712
    },
    {
      "epoch": 22.497260787992495,
      "grad_norm": 0.5050047039985657,
      "learning_rate": 0.002828,
      "loss": 1.2542,
      "step": 299776
    },
    {
      "epoch": 22.502063789868668,
      "grad_norm": 0.5200276970863342,
      "learning_rate": 0.002828,
      "loss": 1.2547,
      "step": 299840
    },
    {
      "epoch": 22.50686679174484,
      "grad_norm": 0.5613254308700562,
      "learning_rate": 0.002828,
      "loss": 1.2535,
      "step": 299904
    },
    {
      "epoch": 22.511669793621014,
      "grad_norm": 0.574245274066925,
      "learning_rate": 0.002828,
      "loss": 1.2503,
      "step": 299968
    },
    {
      "epoch": 22.516472795497187,
      "grad_norm": 0.6551904678344727,
      "learning_rate": 0.002828,
      "loss": 1.259,
      "step": 300032
    },
    {
      "epoch": 22.52127579737336,
      "grad_norm": 0.7811374664306641,
      "learning_rate": 0.002828,
      "loss": 1.2488,
      "step": 300096
    },
    {
      "epoch": 22.526078799249532,
      "grad_norm": 0.5673550367355347,
      "learning_rate": 0.002828,
      "loss": 1.2572,
      "step": 300160
    },
    {
      "epoch": 22.530881801125705,
      "grad_norm": 0.6635274887084961,
      "learning_rate": 0.002828,
      "loss": 1.2563,
      "step": 300224
    },
    {
      "epoch": 22.535684803001875,
      "grad_norm": 0.702856719493866,
      "learning_rate": 0.002828,
      "loss": 1.2637,
      "step": 300288
    },
    {
      "epoch": 22.540487804878047,
      "grad_norm": 0.571111798286438,
      "learning_rate": 0.002828,
      "loss": 1.2551,
      "step": 300352
    },
    {
      "epoch": 22.54529080675422,
      "grad_norm": 0.5393403172492981,
      "learning_rate": 0.002828,
      "loss": 1.2558,
      "step": 300416
    },
    {
      "epoch": 22.550093808630393,
      "grad_norm": 0.5712943077087402,
      "learning_rate": 0.002828,
      "loss": 1.2518,
      "step": 300480
    },
    {
      "epoch": 22.554896810506566,
      "grad_norm": 0.5832908749580383,
      "learning_rate": 0.002828,
      "loss": 1.2515,
      "step": 300544
    },
    {
      "epoch": 22.55969981238274,
      "grad_norm": 0.7355929613113403,
      "learning_rate": 0.002828,
      "loss": 1.2559,
      "step": 300608
    },
    {
      "epoch": 22.564502814258912,
      "grad_norm": 0.5435024499893188,
      "learning_rate": 0.002828,
      "loss": 1.2595,
      "step": 300672
    },
    {
      "epoch": 22.569305816135085,
      "grad_norm": 0.49227020144462585,
      "learning_rate": 0.002828,
      "loss": 1.2578,
      "step": 300736
    },
    {
      "epoch": 22.574108818011258,
      "grad_norm": 0.5402520895004272,
      "learning_rate": 0.002828,
      "loss": 1.2539,
      "step": 300800
    },
    {
      "epoch": 22.57891181988743,
      "grad_norm": 0.6322745084762573,
      "learning_rate": 0.002828,
      "loss": 1.2583,
      "step": 300864
    },
    {
      "epoch": 22.583714821763603,
      "grad_norm": 0.5023521184921265,
      "learning_rate": 0.002828,
      "loss": 1.2526,
      "step": 300928
    },
    {
      "epoch": 22.588517823639776,
      "grad_norm": 0.5673877000808716,
      "learning_rate": 0.002828,
      "loss": 1.2597,
      "step": 300992
    },
    {
      "epoch": 22.593320825515946,
      "grad_norm": 0.6613234281539917,
      "learning_rate": 0.002828,
      "loss": 1.2573,
      "step": 301056
    },
    {
      "epoch": 22.59812382739212,
      "grad_norm": 0.7267638444900513,
      "learning_rate": 0.002828,
      "loss": 1.2582,
      "step": 301120
    },
    {
      "epoch": 22.60292682926829,
      "grad_norm": 0.5590457916259766,
      "learning_rate": 0.002828,
      "loss": 1.2539,
      "step": 301184
    },
    {
      "epoch": 22.607729831144464,
      "grad_norm": 0.7561535239219666,
      "learning_rate": 0.002828,
      "loss": 1.2589,
      "step": 301248
    },
    {
      "epoch": 22.612532833020637,
      "grad_norm": 0.7155293226242065,
      "learning_rate": 0.002828,
      "loss": 1.2557,
      "step": 301312
    },
    {
      "epoch": 22.61733583489681,
      "grad_norm": 0.5476259589195251,
      "learning_rate": 0.002828,
      "loss": 1.2555,
      "step": 301376
    },
    {
      "epoch": 22.622138836772983,
      "grad_norm": 0.6327958703041077,
      "learning_rate": 0.002828,
      "loss": 1.2592,
      "step": 301440
    },
    {
      "epoch": 22.626941838649156,
      "grad_norm": 0.6359103322029114,
      "learning_rate": 0.002828,
      "loss": 1.2686,
      "step": 301504
    },
    {
      "epoch": 22.63174484052533,
      "grad_norm": 0.5994049906730652,
      "learning_rate": 0.002828,
      "loss": 1.2623,
      "step": 301568
    },
    {
      "epoch": 22.6365478424015,
      "grad_norm": 0.5330276489257812,
      "learning_rate": 0.002828,
      "loss": 1.2582,
      "step": 301632
    },
    {
      "epoch": 22.641350844277675,
      "grad_norm": 0.5444290637969971,
      "learning_rate": 0.002828,
      "loss": 1.2632,
      "step": 301696
    },
    {
      "epoch": 22.646153846153847,
      "grad_norm": 0.5603839159011841,
      "learning_rate": 0.002828,
      "loss": 1.2537,
      "step": 301760
    },
    {
      "epoch": 22.65095684803002,
      "grad_norm": 0.5644785165786743,
      "learning_rate": 0.002828,
      "loss": 1.2584,
      "step": 301824
    },
    {
      "epoch": 22.65575984990619,
      "grad_norm": 0.5198463201522827,
      "learning_rate": 0.002828,
      "loss": 1.266,
      "step": 301888
    },
    {
      "epoch": 22.660562851782363,
      "grad_norm": 0.4774089455604553,
      "learning_rate": 0.002828,
      "loss": 1.2642,
      "step": 301952
    },
    {
      "epoch": 22.665365853658535,
      "grad_norm": 0.5863246321678162,
      "learning_rate": 0.002828,
      "loss": 1.2603,
      "step": 302016
    },
    {
      "epoch": 22.67016885553471,
      "grad_norm": 0.6068688631057739,
      "learning_rate": 0.002828,
      "loss": 1.2679,
      "step": 302080
    },
    {
      "epoch": 22.67497185741088,
      "grad_norm": 0.6282370686531067,
      "learning_rate": 0.002828,
      "loss": 1.2609,
      "step": 302144
    },
    {
      "epoch": 22.679774859287054,
      "grad_norm": 0.6089537739753723,
      "learning_rate": 0.002828,
      "loss": 1.2618,
      "step": 302208
    },
    {
      "epoch": 22.684577861163227,
      "grad_norm": 0.4584980607032776,
      "learning_rate": 0.002828,
      "loss": 1.2587,
      "step": 302272
    },
    {
      "epoch": 22.6893808630394,
      "grad_norm": 0.6628486514091492,
      "learning_rate": 0.002828,
      "loss": 1.2617,
      "step": 302336
    },
    {
      "epoch": 22.694183864915573,
      "grad_norm": 0.5792520642280579,
      "learning_rate": 0.002828,
      "loss": 1.2596,
      "step": 302400
    },
    {
      "epoch": 22.698986866791746,
      "grad_norm": 0.6892536878585815,
      "learning_rate": 0.002828,
      "loss": 1.2658,
      "step": 302464
    },
    {
      "epoch": 22.70378986866792,
      "grad_norm": 0.6139593720436096,
      "learning_rate": 0.002828,
      "loss": 1.258,
      "step": 302528
    },
    {
      "epoch": 22.70859287054409,
      "grad_norm": 0.7230052947998047,
      "learning_rate": 0.002828,
      "loss": 1.2583,
      "step": 302592
    },
    {
      "epoch": 22.713395872420264,
      "grad_norm": 0.65574049949646,
      "learning_rate": 0.002828,
      "loss": 1.2598,
      "step": 302656
    },
    {
      "epoch": 22.718198874296434,
      "grad_norm": 0.5445919036865234,
      "learning_rate": 0.002828,
      "loss": 1.2613,
      "step": 302720
    },
    {
      "epoch": 22.723001876172606,
      "grad_norm": 0.5585668087005615,
      "learning_rate": 0.002828,
      "loss": 1.2664,
      "step": 302784
    },
    {
      "epoch": 22.72780487804878,
      "grad_norm": 0.46071845293045044,
      "learning_rate": 0.002828,
      "loss": 1.2616,
      "step": 302848
    },
    {
      "epoch": 22.732607879924952,
      "grad_norm": 0.5296932458877563,
      "learning_rate": 0.002828,
      "loss": 1.2626,
      "step": 302912
    },
    {
      "epoch": 22.737410881801125,
      "grad_norm": 0.592583954334259,
      "learning_rate": 0.002828,
      "loss": 1.2579,
      "step": 302976
    },
    {
      "epoch": 22.742213883677298,
      "grad_norm": 0.4957088828086853,
      "learning_rate": 0.002828,
      "loss": 1.2663,
      "step": 303040
    },
    {
      "epoch": 22.74701688555347,
      "grad_norm": 0.6246522068977356,
      "learning_rate": 0.002828,
      "loss": 1.2614,
      "step": 303104
    },
    {
      "epoch": 22.751819887429644,
      "grad_norm": 0.6576674580574036,
      "learning_rate": 0.002828,
      "loss": 1.2598,
      "step": 303168
    },
    {
      "epoch": 22.756622889305817,
      "grad_norm": 0.48002710938453674,
      "learning_rate": 0.002828,
      "loss": 1.2548,
      "step": 303232
    },
    {
      "epoch": 22.76142589118199,
      "grad_norm": 0.6484505534172058,
      "learning_rate": 0.002828,
      "loss": 1.2603,
      "step": 303296
    },
    {
      "epoch": 22.766228893058162,
      "grad_norm": 0.5288729071617126,
      "learning_rate": 0.002828,
      "loss": 1.2497,
      "step": 303360
    },
    {
      "epoch": 22.771031894934335,
      "grad_norm": 0.595810055732727,
      "learning_rate": 0.002828,
      "loss": 1.2533,
      "step": 303424
    },
    {
      "epoch": 22.775834896810508,
      "grad_norm": 0.5582146048545837,
      "learning_rate": 0.002828,
      "loss": 1.2586,
      "step": 303488
    },
    {
      "epoch": 22.780637898686678,
      "grad_norm": 0.6199321746826172,
      "learning_rate": 0.002828,
      "loss": 1.252,
      "step": 303552
    },
    {
      "epoch": 22.78544090056285,
      "grad_norm": 0.5382596254348755,
      "learning_rate": 0.002828,
      "loss": 1.2543,
      "step": 303616
    },
    {
      "epoch": 22.790243902439023,
      "grad_norm": 0.5364497303962708,
      "learning_rate": 0.002828,
      "loss": 1.253,
      "step": 303680
    },
    {
      "epoch": 22.795046904315196,
      "grad_norm": 0.6804397106170654,
      "learning_rate": 0.002828,
      "loss": 1.2511,
      "step": 303744
    },
    {
      "epoch": 22.79984990619137,
      "grad_norm": 0.6165252923965454,
      "learning_rate": 0.002828,
      "loss": 1.2578,
      "step": 303808
    },
    {
      "epoch": 22.804652908067542,
      "grad_norm": 0.6188555359840393,
      "learning_rate": 0.002828,
      "loss": 1.2545,
      "step": 303872
    },
    {
      "epoch": 22.809455909943715,
      "grad_norm": 0.5402896404266357,
      "learning_rate": 0.002828,
      "loss": 1.257,
      "step": 303936
    },
    {
      "epoch": 22.814258911819888,
      "grad_norm": 0.5374215245246887,
      "learning_rate": 0.002828,
      "loss": 1.2541,
      "step": 304000
    },
    {
      "epoch": 22.81906191369606,
      "grad_norm": 0.6006323099136353,
      "learning_rate": 0.002828,
      "loss": 1.2603,
      "step": 304064
    },
    {
      "epoch": 22.823864915572234,
      "grad_norm": 0.570731520652771,
      "learning_rate": 0.002828,
      "loss": 1.2523,
      "step": 304128
    },
    {
      "epoch": 22.828667917448406,
      "grad_norm": 0.7125750184059143,
      "learning_rate": 0.002828,
      "loss": 1.2607,
      "step": 304192
    },
    {
      "epoch": 22.83347091932458,
      "grad_norm": 0.4166480004787445,
      "learning_rate": 0.002828,
      "loss": 1.254,
      "step": 304256
    },
    {
      "epoch": 22.838273921200752,
      "grad_norm": 0.6014498472213745,
      "learning_rate": 0.002828,
      "loss": 1.2516,
      "step": 304320
    },
    {
      "epoch": 22.84307692307692,
      "grad_norm": 0.6451423764228821,
      "learning_rate": 0.002828,
      "loss": 1.2588,
      "step": 304384
    },
    {
      "epoch": 22.847879924953094,
      "grad_norm": 0.5513681769371033,
      "learning_rate": 0.002828,
      "loss": 1.2577,
      "step": 304448
    },
    {
      "epoch": 22.852682926829267,
      "grad_norm": 0.5224577188491821,
      "learning_rate": 0.002828,
      "loss": 1.257,
      "step": 304512
    },
    {
      "epoch": 22.85748592870544,
      "grad_norm": 0.7520785331726074,
      "learning_rate": 0.002828,
      "loss": 1.2573,
      "step": 304576
    },
    {
      "epoch": 22.862288930581613,
      "grad_norm": 0.7311927080154419,
      "learning_rate": 0.002828,
      "loss": 1.2556,
      "step": 304640
    },
    {
      "epoch": 22.867091932457786,
      "grad_norm": 0.6609572768211365,
      "learning_rate": 0.002828,
      "loss": 1.2586,
      "step": 304704
    },
    {
      "epoch": 22.87189493433396,
      "grad_norm": 0.5938644409179688,
      "learning_rate": 0.002828,
      "loss": 1.2518,
      "step": 304768
    },
    {
      "epoch": 22.87669793621013,
      "grad_norm": 0.7628321647644043,
      "learning_rate": 0.002828,
      "loss": 1.2622,
      "step": 304832
    },
    {
      "epoch": 22.881500938086305,
      "grad_norm": 0.5340333580970764,
      "learning_rate": 0.002828,
      "loss": 1.2495,
      "step": 304896
    },
    {
      "epoch": 22.886303939962477,
      "grad_norm": 0.6140053868293762,
      "learning_rate": 0.002828,
      "loss": 1.2565,
      "step": 304960
    },
    {
      "epoch": 22.89110694183865,
      "grad_norm": 0.53498774766922,
      "learning_rate": 0.002828,
      "loss": 1.2614,
      "step": 305024
    },
    {
      "epoch": 22.895909943714823,
      "grad_norm": 0.5342109203338623,
      "learning_rate": 0.002828,
      "loss": 1.2576,
      "step": 305088
    },
    {
      "epoch": 22.900712945590996,
      "grad_norm": 0.4914964735507965,
      "learning_rate": 0.002828,
      "loss": 1.2587,
      "step": 305152
    },
    {
      "epoch": 22.905515947467165,
      "grad_norm": 0.6862971186637878,
      "learning_rate": 0.002828,
      "loss": 1.2657,
      "step": 305216
    },
    {
      "epoch": 22.91031894934334,
      "grad_norm": 0.5172854065895081,
      "learning_rate": 0.002828,
      "loss": 1.2579,
      "step": 305280
    },
    {
      "epoch": 22.91512195121951,
      "grad_norm": 0.6534024477005005,
      "learning_rate": 0.002828,
      "loss": 1.2566,
      "step": 305344
    },
    {
      "epoch": 22.919924953095684,
      "grad_norm": 0.5077025890350342,
      "learning_rate": 0.002828,
      "loss": 1.2571,
      "step": 305408
    },
    {
      "epoch": 22.924727954971857,
      "grad_norm": 0.5739777684211731,
      "learning_rate": 0.002828,
      "loss": 1.2576,
      "step": 305472
    },
    {
      "epoch": 22.92953095684803,
      "grad_norm": 0.5454936623573303,
      "learning_rate": 0.002828,
      "loss": 1.2545,
      "step": 305536
    },
    {
      "epoch": 22.934333958724203,
      "grad_norm": 0.5912908315658569,
      "learning_rate": 0.002828,
      "loss": 1.2565,
      "step": 305600
    },
    {
      "epoch": 22.939136960600376,
      "grad_norm": 0.7086852788925171,
      "learning_rate": 0.002828,
      "loss": 1.2585,
      "step": 305664
    },
    {
      "epoch": 22.94393996247655,
      "grad_norm": 0.6017676591873169,
      "learning_rate": 0.002828,
      "loss": 1.2558,
      "step": 305728
    },
    {
      "epoch": 22.94874296435272,
      "grad_norm": 0.4917507767677307,
      "learning_rate": 0.002828,
      "loss": 1.2609,
      "step": 305792
    },
    {
      "epoch": 22.953545966228894,
      "grad_norm": 0.6148297786712646,
      "learning_rate": 0.002828,
      "loss": 1.2641,
      "step": 305856
    },
    {
      "epoch": 22.958348968105067,
      "grad_norm": 0.4830070436000824,
      "learning_rate": 0.002828,
      "loss": 1.2533,
      "step": 305920
    },
    {
      "epoch": 22.96315196998124,
      "grad_norm": 0.6182305216789246,
      "learning_rate": 0.002828,
      "loss": 1.2576,
      "step": 305984
    },
    {
      "epoch": 22.96795497185741,
      "grad_norm": 0.628701388835907,
      "learning_rate": 0.002828,
      "loss": 1.2589,
      "step": 306048
    },
    {
      "epoch": 22.972757973733582,
      "grad_norm": 0.586304247379303,
      "learning_rate": 0.002828,
      "loss": 1.2603,
      "step": 306112
    },
    {
      "epoch": 22.977560975609755,
      "grad_norm": 0.5976091027259827,
      "learning_rate": 0.002828,
      "loss": 1.2561,
      "step": 306176
    },
    {
      "epoch": 22.982363977485928,
      "grad_norm": 0.6377904415130615,
      "learning_rate": 0.002828,
      "loss": 1.2572,
      "step": 306240
    },
    {
      "epoch": 22.9871669793621,
      "grad_norm": 0.6798734664916992,
      "learning_rate": 0.002828,
      "loss": 1.2614,
      "step": 306304
    },
    {
      "epoch": 22.991969981238274,
      "grad_norm": 0.6504051685333252,
      "learning_rate": 0.002828,
      "loss": 1.2579,
      "step": 306368
    },
    {
      "epoch": 22.996772983114447,
      "grad_norm": 0.6908726096153259,
      "learning_rate": 0.002828,
      "loss": 1.2567,
      "step": 306432
    },
    {
      "epoch": 23.00157598499062,
      "grad_norm": 0.54974764585495,
      "learning_rate": 0.002828,
      "loss": 1.2522,
      "step": 306496
    },
    {
      "epoch": 23.006378986866793,
      "grad_norm": 0.5098195672035217,
      "learning_rate": 0.002828,
      "loss": 1.24,
      "step": 306560
    },
    {
      "epoch": 23.011181988742965,
      "grad_norm": 0.5870168805122375,
      "learning_rate": 0.002828,
      "loss": 1.2337,
      "step": 306624
    },
    {
      "epoch": 23.01598499061914,
      "grad_norm": 0.6402837634086609,
      "learning_rate": 0.002828,
      "loss": 1.2408,
      "step": 306688
    },
    {
      "epoch": 23.02078799249531,
      "grad_norm": 0.680777907371521,
      "learning_rate": 0.002828,
      "loss": 1.2426,
      "step": 306752
    },
    {
      "epoch": 23.02559099437148,
      "grad_norm": 0.69080650806427,
      "learning_rate": 0.002828,
      "loss": 1.2435,
      "step": 306816
    },
    {
      "epoch": 23.030393996247653,
      "grad_norm": 0.5605624914169312,
      "learning_rate": 0.002828,
      "loss": 1.2378,
      "step": 306880
    },
    {
      "epoch": 23.035196998123826,
      "grad_norm": 0.5387503504753113,
      "learning_rate": 0.002828,
      "loss": 1.2391,
      "step": 306944
    },
    {
      "epoch": 23.04,
      "grad_norm": 0.6666533946990967,
      "learning_rate": 0.002828,
      "loss": 1.2399,
      "step": 307008
    },
    {
      "epoch": 23.044803001876172,
      "grad_norm": 0.6080389022827148,
      "learning_rate": 0.002828,
      "loss": 1.2466,
      "step": 307072
    },
    {
      "epoch": 23.049606003752345,
      "grad_norm": 0.562734842300415,
      "learning_rate": 0.002828,
      "loss": 1.2389,
      "step": 307136
    },
    {
      "epoch": 23.054409005628518,
      "grad_norm": 0.6044827103614807,
      "learning_rate": 0.002828,
      "loss": 1.2388,
      "step": 307200
    },
    {
      "epoch": 23.05921200750469,
      "grad_norm": 0.6313736438751221,
      "learning_rate": 0.002828,
      "loss": 1.2342,
      "step": 307264
    },
    {
      "epoch": 23.064015009380864,
      "grad_norm": 0.5507990717887878,
      "learning_rate": 0.002828,
      "loss": 1.2448,
      "step": 307328
    },
    {
      "epoch": 23.068818011257036,
      "grad_norm": 0.49053919315338135,
      "learning_rate": 0.002828,
      "loss": 1.2453,
      "step": 307392
    },
    {
      "epoch": 23.07362101313321,
      "grad_norm": 0.5089336633682251,
      "learning_rate": 0.002828,
      "loss": 1.237,
      "step": 307456
    },
    {
      "epoch": 23.078424015009382,
      "grad_norm": 0.5594099164009094,
      "learning_rate": 0.002828,
      "loss": 1.2404,
      "step": 307520
    },
    {
      "epoch": 23.083227016885555,
      "grad_norm": 0.6270295977592468,
      "learning_rate": 0.002828,
      "loss": 1.2442,
      "step": 307584
    },
    {
      "epoch": 23.088030018761724,
      "grad_norm": 0.5999992489814758,
      "learning_rate": 0.002828,
      "loss": 1.2422,
      "step": 307648
    },
    {
      "epoch": 23.092833020637897,
      "grad_norm": 0.6353987455368042,
      "learning_rate": 0.002828,
      "loss": 1.241,
      "step": 307712
    },
    {
      "epoch": 23.09763602251407,
      "grad_norm": 0.6352692246437073,
      "learning_rate": 0.002828,
      "loss": 1.2475,
      "step": 307776
    },
    {
      "epoch": 23.102439024390243,
      "grad_norm": 0.4999005198478699,
      "learning_rate": 0.002828,
      "loss": 1.2432,
      "step": 307840
    },
    {
      "epoch": 23.107242026266416,
      "grad_norm": 0.5238738656044006,
      "learning_rate": 0.002828,
      "loss": 1.2434,
      "step": 307904
    },
    {
      "epoch": 23.11204502814259,
      "grad_norm": 0.46657606959342957,
      "learning_rate": 0.002828,
      "loss": 1.2468,
      "step": 307968
    },
    {
      "epoch": 23.11684803001876,
      "grad_norm": 0.7083847522735596,
      "learning_rate": 0.002828,
      "loss": 1.2546,
      "step": 308032
    },
    {
      "epoch": 23.121651031894935,
      "grad_norm": 0.7005468606948853,
      "learning_rate": 0.002828,
      "loss": 1.2384,
      "step": 308096
    },
    {
      "epoch": 23.126454033771108,
      "grad_norm": 0.5426920652389526,
      "learning_rate": 0.002828,
      "loss": 1.2453,
      "step": 308160
    },
    {
      "epoch": 23.13125703564728,
      "grad_norm": 0.5448518395423889,
      "learning_rate": 0.002828,
      "loss": 1.2519,
      "step": 308224
    },
    {
      "epoch": 23.136060037523453,
      "grad_norm": 0.6012153625488281,
      "learning_rate": 0.002828,
      "loss": 1.2515,
      "step": 308288
    },
    {
      "epoch": 23.140863039399626,
      "grad_norm": 0.5151079893112183,
      "learning_rate": 0.002828,
      "loss": 1.2365,
      "step": 308352
    },
    {
      "epoch": 23.1456660412758,
      "grad_norm": 0.704038143157959,
      "learning_rate": 0.002828,
      "loss": 1.2418,
      "step": 308416
    },
    {
      "epoch": 23.15046904315197,
      "grad_norm": 0.6568431854248047,
      "learning_rate": 0.002828,
      "loss": 1.2478,
      "step": 308480
    },
    {
      "epoch": 23.15527204502814,
      "grad_norm": 0.6518442034721375,
      "learning_rate": 0.002828,
      "loss": 1.2481,
      "step": 308544
    },
    {
      "epoch": 23.160075046904314,
      "grad_norm": 0.5150387287139893,
      "learning_rate": 0.002828,
      "loss": 1.2459,
      "step": 308608
    },
    {
      "epoch": 23.164878048780487,
      "grad_norm": 0.5261139273643494,
      "learning_rate": 0.002828,
      "loss": 1.2471,
      "step": 308672
    },
    {
      "epoch": 23.16968105065666,
      "grad_norm": 0.6531349420547485,
      "learning_rate": 0.002828,
      "loss": 1.2472,
      "step": 308736
    },
    {
      "epoch": 23.174484052532833,
      "grad_norm": 0.5209482312202454,
      "learning_rate": 0.002828,
      "loss": 1.2441,
      "step": 308800
    },
    {
      "epoch": 23.179287054409006,
      "grad_norm": 0.5978618860244751,
      "learning_rate": 0.002828,
      "loss": 1.252,
      "step": 308864
    },
    {
      "epoch": 23.18409005628518,
      "grad_norm": 0.5009728670120239,
      "learning_rate": 0.002828,
      "loss": 1.2424,
      "step": 308928
    },
    {
      "epoch": 23.18889305816135,
      "grad_norm": 0.5005267262458801,
      "learning_rate": 0.002828,
      "loss": 1.2517,
      "step": 308992
    },
    {
      "epoch": 23.193696060037524,
      "grad_norm": 0.5890987515449524,
      "learning_rate": 0.002828,
      "loss": 1.2473,
      "step": 309056
    },
    {
      "epoch": 23.198499061913697,
      "grad_norm": 0.5844420194625854,
      "learning_rate": 0.002828,
      "loss": 1.2487,
      "step": 309120
    },
    {
      "epoch": 23.20330206378987,
      "grad_norm": 0.6215989589691162,
      "learning_rate": 0.002828,
      "loss": 1.248,
      "step": 309184
    },
    {
      "epoch": 23.208105065666043,
      "grad_norm": 0.6483440399169922,
      "learning_rate": 0.002828,
      "loss": 1.2436,
      "step": 309248
    },
    {
      "epoch": 23.212908067542212,
      "grad_norm": 0.773412823677063,
      "learning_rate": 0.002828,
      "loss": 1.2488,
      "step": 309312
    },
    {
      "epoch": 23.217711069418385,
      "grad_norm": 0.6054885387420654,
      "learning_rate": 0.002828,
      "loss": 1.2511,
      "step": 309376
    },
    {
      "epoch": 23.222514071294558,
      "grad_norm": 0.6024729013442993,
      "learning_rate": 0.002828,
      "loss": 1.2474,
      "step": 309440
    },
    {
      "epoch": 23.22731707317073,
      "grad_norm": 0.6172342896461487,
      "learning_rate": 0.002828,
      "loss": 1.2559,
      "step": 309504
    },
    {
      "epoch": 23.232120075046904,
      "grad_norm": 0.5660899877548218,
      "learning_rate": 0.002828,
      "loss": 1.2453,
      "step": 309568
    },
    {
      "epoch": 23.236923076923077,
      "grad_norm": 0.5336431860923767,
      "learning_rate": 0.002828,
      "loss": 1.2432,
      "step": 309632
    },
    {
      "epoch": 23.24172607879925,
      "grad_norm": 0.5058355927467346,
      "learning_rate": 0.002828,
      "loss": 1.2503,
      "step": 309696
    },
    {
      "epoch": 23.246529080675423,
      "grad_norm": 0.6016878485679626,
      "learning_rate": 0.002828,
      "loss": 1.2588,
      "step": 309760
    },
    {
      "epoch": 23.251332082551595,
      "grad_norm": 0.5316280722618103,
      "learning_rate": 0.002828,
      "loss": 1.2511,
      "step": 309824
    },
    {
      "epoch": 23.25613508442777,
      "grad_norm": 0.6197585463523865,
      "learning_rate": 0.002828,
      "loss": 1.2472,
      "step": 309888
    },
    {
      "epoch": 23.26093808630394,
      "grad_norm": 0.5783255100250244,
      "learning_rate": 0.002828,
      "loss": 1.2519,
      "step": 309952
    },
    {
      "epoch": 23.265741088180114,
      "grad_norm": 0.6217781901359558,
      "learning_rate": 0.002828,
      "loss": 1.2505,
      "step": 310016
    },
    {
      "epoch": 23.270544090056283,
      "grad_norm": 0.5191296339035034,
      "learning_rate": 0.002828,
      "loss": 1.2459,
      "step": 310080
    },
    {
      "epoch": 23.275347091932456,
      "grad_norm": 0.6412822008132935,
      "learning_rate": 0.002828,
      "loss": 1.2503,
      "step": 310144
    },
    {
      "epoch": 23.28015009380863,
      "grad_norm": 0.5884774327278137,
      "learning_rate": 0.002828,
      "loss": 1.2491,
      "step": 310208
    },
    {
      "epoch": 23.284953095684802,
      "grad_norm": 0.621462345123291,
      "learning_rate": 0.002828,
      "loss": 1.2527,
      "step": 310272
    },
    {
      "epoch": 23.289756097560975,
      "grad_norm": 0.5567516684532166,
      "learning_rate": 0.002828,
      "loss": 1.2462,
      "step": 310336
    },
    {
      "epoch": 23.294559099437148,
      "grad_norm": 0.5859234929084778,
      "learning_rate": 0.002828,
      "loss": 1.2466,
      "step": 310400
    },
    {
      "epoch": 23.29936210131332,
      "grad_norm": 0.5633601546287537,
      "learning_rate": 0.002828,
      "loss": 1.2518,
      "step": 310464
    },
    {
      "epoch": 23.304165103189494,
      "grad_norm": 0.6056027412414551,
      "learning_rate": 0.002828,
      "loss": 1.249,
      "step": 310528
    },
    {
      "epoch": 23.308968105065667,
      "grad_norm": 0.6314893960952759,
      "learning_rate": 0.002828,
      "loss": 1.2521,
      "step": 310592
    },
    {
      "epoch": 23.31377110694184,
      "grad_norm": 0.47842320799827576,
      "learning_rate": 0.002828,
      "loss": 1.2545,
      "step": 310656
    },
    {
      "epoch": 23.318574108818012,
      "grad_norm": 0.5698492527008057,
      "learning_rate": 0.002828,
      "loss": 1.2455,
      "step": 310720
    },
    {
      "epoch": 23.323377110694185,
      "grad_norm": 0.6335123777389526,
      "learning_rate": 0.002828,
      "loss": 1.2493,
      "step": 310784
    },
    {
      "epoch": 23.328180112570358,
      "grad_norm": 0.7336736917495728,
      "learning_rate": 0.002828,
      "loss": 1.244,
      "step": 310848
    },
    {
      "epoch": 23.332983114446527,
      "grad_norm": 0.5209872126579285,
      "learning_rate": 0.002828,
      "loss": 1.2477,
      "step": 310912
    },
    {
      "epoch": 23.3377861163227,
      "grad_norm": 0.5684576630592346,
      "learning_rate": 0.002828,
      "loss": 1.2531,
      "step": 310976
    },
    {
      "epoch": 23.342589118198873,
      "grad_norm": 0.5723300576210022,
      "learning_rate": 0.002828,
      "loss": 1.2519,
      "step": 311040
    },
    {
      "epoch": 23.347392120075046,
      "grad_norm": 0.48642081022262573,
      "learning_rate": 0.002828,
      "loss": 1.2462,
      "step": 311104
    },
    {
      "epoch": 23.35219512195122,
      "grad_norm": 0.5746228098869324,
      "learning_rate": 0.002828,
      "loss": 1.2518,
      "step": 311168
    },
    {
      "epoch": 23.356998123827392,
      "grad_norm": 0.6249984502792358,
      "learning_rate": 0.002828,
      "loss": 1.249,
      "step": 311232
    },
    {
      "epoch": 23.361801125703565,
      "grad_norm": 0.4942440092563629,
      "learning_rate": 0.002828,
      "loss": 1.252,
      "step": 311296
    },
    {
      "epoch": 23.366604127579738,
      "grad_norm": 0.610084056854248,
      "learning_rate": 0.002828,
      "loss": 1.2458,
      "step": 311360
    },
    {
      "epoch": 23.37140712945591,
      "grad_norm": 0.5461624264717102,
      "learning_rate": 0.002828,
      "loss": 1.2532,
      "step": 311424
    },
    {
      "epoch": 23.376210131332083,
      "grad_norm": 0.5444890856742859,
      "learning_rate": 0.002828,
      "loss": 1.2524,
      "step": 311488
    },
    {
      "epoch": 23.381013133208256,
      "grad_norm": 0.5665737986564636,
      "learning_rate": 0.002828,
      "loss": 1.2473,
      "step": 311552
    },
    {
      "epoch": 23.38581613508443,
      "grad_norm": 0.6717258095741272,
      "learning_rate": 0.002828,
      "loss": 1.2475,
      "step": 311616
    },
    {
      "epoch": 23.390619136960602,
      "grad_norm": 0.6941819787025452,
      "learning_rate": 0.002828,
      "loss": 1.2523,
      "step": 311680
    },
    {
      "epoch": 23.39542213883677,
      "grad_norm": 0.6986182332038879,
      "learning_rate": 0.002828,
      "loss": 1.2545,
      "step": 311744
    },
    {
      "epoch": 23.400225140712944,
      "grad_norm": 0.5212686657905579,
      "learning_rate": 0.002828,
      "loss": 1.2437,
      "step": 311808
    },
    {
      "epoch": 23.405028142589117,
      "grad_norm": 0.5907437205314636,
      "learning_rate": 0.002828,
      "loss": 1.251,
      "step": 311872
    },
    {
      "epoch": 23.40983114446529,
      "grad_norm": 0.5176777243614197,
      "learning_rate": 0.002828,
      "loss": 1.2488,
      "step": 311936
    },
    {
      "epoch": 23.414634146341463,
      "grad_norm": 0.5791288614273071,
      "learning_rate": 0.002828,
      "loss": 1.2541,
      "step": 312000
    },
    {
      "epoch": 23.419437148217636,
      "grad_norm": 0.6610795855522156,
      "learning_rate": 0.002828,
      "loss": 1.2544,
      "step": 312064
    },
    {
      "epoch": 23.42424015009381,
      "grad_norm": 0.5516941547393799,
      "learning_rate": 0.002828,
      "loss": 1.2451,
      "step": 312128
    },
    {
      "epoch": 23.42904315196998,
      "grad_norm": 0.5505845546722412,
      "learning_rate": 0.002828,
      "loss": 1.2558,
      "step": 312192
    },
    {
      "epoch": 23.433846153846154,
      "grad_norm": 0.6624215841293335,
      "learning_rate": 0.002828,
      "loss": 1.2516,
      "step": 312256
    },
    {
      "epoch": 23.438649155722327,
      "grad_norm": 0.560444176197052,
      "learning_rate": 0.002828,
      "loss": 1.2552,
      "step": 312320
    },
    {
      "epoch": 23.4434521575985,
      "grad_norm": 0.5208444595336914,
      "learning_rate": 0.002828,
      "loss": 1.2489,
      "step": 312384
    },
    {
      "epoch": 23.448255159474673,
      "grad_norm": 0.6262831091880798,
      "learning_rate": 0.002828,
      "loss": 1.2564,
      "step": 312448
    },
    {
      "epoch": 23.453058161350846,
      "grad_norm": 0.5905798673629761,
      "learning_rate": 0.002828,
      "loss": 1.2508,
      "step": 312512
    },
    {
      "epoch": 23.457861163227015,
      "grad_norm": 0.6314270496368408,
      "learning_rate": 0.002828,
      "loss": 1.2546,
      "step": 312576
    },
    {
      "epoch": 23.462664165103188,
      "grad_norm": 0.5014112591743469,
      "learning_rate": 0.002828,
      "loss": 1.2592,
      "step": 312640
    },
    {
      "epoch": 23.46746716697936,
      "grad_norm": 0.5422669053077698,
      "learning_rate": 0.002828,
      "loss": 1.2542,
      "step": 312704
    },
    {
      "epoch": 23.472270168855534,
      "grad_norm": 0.6367536187171936,
      "learning_rate": 0.002828,
      "loss": 1.2526,
      "step": 312768
    },
    {
      "epoch": 23.477073170731707,
      "grad_norm": 0.4678763151168823,
      "learning_rate": 0.002828,
      "loss": 1.2562,
      "step": 312832
    },
    {
      "epoch": 23.48187617260788,
      "grad_norm": 0.5110854506492615,
      "learning_rate": 0.002828,
      "loss": 1.2489,
      "step": 312896
    },
    {
      "epoch": 23.486679174484053,
      "grad_norm": 0.6120699644088745,
      "learning_rate": 0.002828,
      "loss": 1.2528,
      "step": 312960
    },
    {
      "epoch": 23.491482176360226,
      "grad_norm": 0.5177861452102661,
      "learning_rate": 0.002828,
      "loss": 1.2532,
      "step": 313024
    },
    {
      "epoch": 23.4962851782364,
      "grad_norm": 0.5200073719024658,
      "learning_rate": 0.002828,
      "loss": 1.2581,
      "step": 313088
    },
    {
      "epoch": 23.50108818011257,
      "grad_norm": 0.5814197063446045,
      "learning_rate": 0.002828,
      "loss": 1.2497,
      "step": 313152
    },
    {
      "epoch": 23.505891181988744,
      "grad_norm": 0.5872126221656799,
      "learning_rate": 0.002828,
      "loss": 1.2538,
      "step": 313216
    },
    {
      "epoch": 23.510694183864917,
      "grad_norm": 0.7856559157371521,
      "learning_rate": 0.002828,
      "loss": 1.2546,
      "step": 313280
    },
    {
      "epoch": 23.51549718574109,
      "grad_norm": 0.4535023868083954,
      "learning_rate": 0.002828,
      "loss": 1.2582,
      "step": 313344
    },
    {
      "epoch": 23.52030018761726,
      "grad_norm": 0.5431272387504578,
      "learning_rate": 0.002828,
      "loss": 1.2562,
      "step": 313408
    },
    {
      "epoch": 23.525103189493432,
      "grad_norm": 0.6125060319900513,
      "learning_rate": 0.002828,
      "loss": 1.2514,
      "step": 313472
    },
    {
      "epoch": 23.529906191369605,
      "grad_norm": 0.6155609488487244,
      "learning_rate": 0.002828,
      "loss": 1.2592,
      "step": 313536
    },
    {
      "epoch": 23.534709193245778,
      "grad_norm": 0.7311649322509766,
      "learning_rate": 0.002828,
      "loss": 1.2579,
      "step": 313600
    },
    {
      "epoch": 23.53951219512195,
      "grad_norm": 0.5599619746208191,
      "learning_rate": 0.002828,
      "loss": 1.2623,
      "step": 313664
    },
    {
      "epoch": 23.544315196998124,
      "grad_norm": 0.5777077078819275,
      "learning_rate": 0.002828,
      "loss": 1.2585,
      "step": 313728
    },
    {
      "epoch": 23.549118198874297,
      "grad_norm": 0.5461658239364624,
      "learning_rate": 0.002828,
      "loss": 1.2617,
      "step": 313792
    },
    {
      "epoch": 23.55392120075047,
      "grad_norm": 0.5293754935264587,
      "learning_rate": 0.002828,
      "loss": 1.2525,
      "step": 313856
    },
    {
      "epoch": 23.558724202626642,
      "grad_norm": 0.5593359470367432,
      "learning_rate": 0.002828,
      "loss": 1.2644,
      "step": 313920
    },
    {
      "epoch": 23.563527204502815,
      "grad_norm": 0.6732170581817627,
      "learning_rate": 0.002828,
      "loss": 1.2615,
      "step": 313984
    },
    {
      "epoch": 23.568330206378988,
      "grad_norm": 0.6268848776817322,
      "learning_rate": 0.002828,
      "loss": 1.257,
      "step": 314048
    },
    {
      "epoch": 23.57313320825516,
      "grad_norm": 0.5705792307853699,
      "learning_rate": 0.002828,
      "loss": 1.2537,
      "step": 314112
    },
    {
      "epoch": 23.577936210131334,
      "grad_norm": 0.5307909846305847,
      "learning_rate": 0.002828,
      "loss": 1.2529,
      "step": 314176
    },
    {
      "epoch": 23.582739212007503,
      "grad_norm": 0.7117211222648621,
      "learning_rate": 0.002828,
      "loss": 1.2586,
      "step": 314240
    },
    {
      "epoch": 23.587542213883676,
      "grad_norm": 0.5914830565452576,
      "learning_rate": 0.002828,
      "loss": 1.2587,
      "step": 314304
    },
    {
      "epoch": 23.59234521575985,
      "grad_norm": 0.5162680149078369,
      "learning_rate": 0.002828,
      "loss": 1.2541,
      "step": 314368
    },
    {
      "epoch": 23.597148217636022,
      "grad_norm": 0.5690885186195374,
      "learning_rate": 0.002828,
      "loss": 1.2557,
      "step": 314432
    },
    {
      "epoch": 23.601951219512195,
      "grad_norm": 0.45376938581466675,
      "learning_rate": 0.002828,
      "loss": 1.2629,
      "step": 314496
    },
    {
      "epoch": 23.606754221388368,
      "grad_norm": 0.6359778642654419,
      "learning_rate": 0.002828,
      "loss": 1.2536,
      "step": 314560
    },
    {
      "epoch": 23.61155722326454,
      "grad_norm": 0.6286948919296265,
      "learning_rate": 0.002828,
      "loss": 1.2577,
      "step": 314624
    },
    {
      "epoch": 23.616360225140713,
      "grad_norm": 0.6775930523872375,
      "learning_rate": 0.002828,
      "loss": 1.2615,
      "step": 314688
    },
    {
      "epoch": 23.621163227016886,
      "grad_norm": 0.6079651713371277,
      "learning_rate": 0.002828,
      "loss": 1.2559,
      "step": 314752
    },
    {
      "epoch": 23.62596622889306,
      "grad_norm": 0.6116461157798767,
      "learning_rate": 0.002828,
      "loss": 1.261,
      "step": 314816
    },
    {
      "epoch": 23.630769230769232,
      "grad_norm": 0.6348520517349243,
      "learning_rate": 0.002828,
      "loss": 1.2597,
      "step": 314880
    },
    {
      "epoch": 23.635572232645405,
      "grad_norm": 0.6520925164222717,
      "learning_rate": 0.002828,
      "loss": 1.256,
      "step": 314944
    },
    {
      "epoch": 23.640375234521574,
      "grad_norm": 0.5682945847511292,
      "learning_rate": 0.002828,
      "loss": 1.2589,
      "step": 315008
    },
    {
      "epoch": 23.645178236397747,
      "grad_norm": 0.6923990249633789,
      "learning_rate": 0.002828,
      "loss": 1.2578,
      "step": 315072
    },
    {
      "epoch": 23.64998123827392,
      "grad_norm": 0.5347400903701782,
      "learning_rate": 0.002828,
      "loss": 1.25,
      "step": 315136
    },
    {
      "epoch": 23.654784240150093,
      "grad_norm": 0.6759274005889893,
      "learning_rate": 0.002828,
      "loss": 1.2533,
      "step": 315200
    },
    {
      "epoch": 23.659587242026266,
      "grad_norm": 0.587490439414978,
      "learning_rate": 0.002828,
      "loss": 1.2617,
      "step": 315264
    },
    {
      "epoch": 23.66439024390244,
      "grad_norm": 0.5441737174987793,
      "learning_rate": 0.002828,
      "loss": 1.2507,
      "step": 315328
    },
    {
      "epoch": 23.66919324577861,
      "grad_norm": 0.5269403457641602,
      "learning_rate": 0.002828,
      "loss": 1.2596,
      "step": 315392
    },
    {
      "epoch": 23.673996247654784,
      "grad_norm": 0.5322028398513794,
      "learning_rate": 0.002828,
      "loss": 1.256,
      "step": 315456
    },
    {
      "epoch": 23.678799249530957,
      "grad_norm": 0.493210107088089,
      "learning_rate": 0.002828,
      "loss": 1.2605,
      "step": 315520
    },
    {
      "epoch": 23.68360225140713,
      "grad_norm": 0.5588217377662659,
      "learning_rate": 0.002828,
      "loss": 1.2501,
      "step": 315584
    },
    {
      "epoch": 23.688405253283303,
      "grad_norm": 0.49217018485069275,
      "learning_rate": 0.002828,
      "loss": 1.2596,
      "step": 315648
    },
    {
      "epoch": 23.693208255159476,
      "grad_norm": 0.7225490808486938,
      "learning_rate": 0.002828,
      "loss": 1.256,
      "step": 315712
    },
    {
      "epoch": 23.69801125703565,
      "grad_norm": 0.5031096935272217,
      "learning_rate": 0.002828,
      "loss": 1.2539,
      "step": 315776
    },
    {
      "epoch": 23.70281425891182,
      "grad_norm": 0.502692461013794,
      "learning_rate": 0.002828,
      "loss": 1.2592,
      "step": 315840
    },
    {
      "epoch": 23.70761726078799,
      "grad_norm": 0.6043107509613037,
      "learning_rate": 0.002828,
      "loss": 1.2516,
      "step": 315904
    },
    {
      "epoch": 23.712420262664164,
      "grad_norm": 0.5847479104995728,
      "learning_rate": 0.002828,
      "loss": 1.2596,
      "step": 315968
    },
    {
      "epoch": 23.717223264540337,
      "grad_norm": 0.5990855693817139,
      "learning_rate": 0.002828,
      "loss": 1.2573,
      "step": 316032
    },
    {
      "epoch": 23.72202626641651,
      "grad_norm": 0.5547415614128113,
      "learning_rate": 0.002828,
      "loss": 1.2683,
      "step": 316096
    },
    {
      "epoch": 23.726829268292683,
      "grad_norm": 0.5649647116661072,
      "learning_rate": 0.002828,
      "loss": 1.2652,
      "step": 316160
    },
    {
      "epoch": 23.731632270168856,
      "grad_norm": 0.5840405225753784,
      "learning_rate": 0.002828,
      "loss": 1.2621,
      "step": 316224
    },
    {
      "epoch": 23.73643527204503,
      "grad_norm": 0.5046823620796204,
      "learning_rate": 0.002828,
      "loss": 1.2634,
      "step": 316288
    },
    {
      "epoch": 23.7412382739212,
      "grad_norm": 0.6047646403312683,
      "learning_rate": 0.002828,
      "loss": 1.2597,
      "step": 316352
    },
    {
      "epoch": 23.746041275797374,
      "grad_norm": 0.7113666534423828,
      "learning_rate": 0.002828,
      "loss": 1.2622,
      "step": 316416
    },
    {
      "epoch": 23.750844277673547,
      "grad_norm": 0.6494774222373962,
      "learning_rate": 0.002828,
      "loss": 1.2525,
      "step": 316480
    },
    {
      "epoch": 23.75564727954972,
      "grad_norm": 0.5977435111999512,
      "learning_rate": 0.002828,
      "loss": 1.2635,
      "step": 316544
    },
    {
      "epoch": 23.760450281425893,
      "grad_norm": 0.5716019868850708,
      "learning_rate": 0.002828,
      "loss": 1.2613,
      "step": 316608
    },
    {
      "epoch": 23.765253283302062,
      "grad_norm": 0.5523091554641724,
      "learning_rate": 0.002828,
      "loss": 1.2553,
      "step": 316672
    },
    {
      "epoch": 23.770056285178235,
      "grad_norm": 0.5492827296257019,
      "learning_rate": 0.002828,
      "loss": 1.2604,
      "step": 316736
    },
    {
      "epoch": 23.774859287054408,
      "grad_norm": 0.5364034175872803,
      "learning_rate": 0.002828,
      "loss": 1.2621,
      "step": 316800
    },
    {
      "epoch": 23.77966228893058,
      "grad_norm": 0.5898091197013855,
      "learning_rate": 0.002828,
      "loss": 1.2528,
      "step": 316864
    },
    {
      "epoch": 23.784465290806754,
      "grad_norm": 0.5669413208961487,
      "learning_rate": 0.002828,
      "loss": 1.2593,
      "step": 316928
    },
    {
      "epoch": 23.789268292682927,
      "grad_norm": 0.600856602191925,
      "learning_rate": 0.002828,
      "loss": 1.2639,
      "step": 316992
    },
    {
      "epoch": 23.7940712945591,
      "grad_norm": 0.47627902030944824,
      "learning_rate": 0.002828,
      "loss": 1.2625,
      "step": 317056
    },
    {
      "epoch": 23.798874296435272,
      "grad_norm": 0.5335204005241394,
      "learning_rate": 0.002828,
      "loss": 1.2665,
      "step": 317120
    },
    {
      "epoch": 23.803677298311445,
      "grad_norm": 0.622696042060852,
      "learning_rate": 0.002828,
      "loss": 1.2573,
      "step": 317184
    },
    {
      "epoch": 23.808480300187618,
      "grad_norm": 0.6902386546134949,
      "learning_rate": 0.002828,
      "loss": 1.2627,
      "step": 317248
    },
    {
      "epoch": 23.81328330206379,
      "grad_norm": 0.579950213432312,
      "learning_rate": 0.002828,
      "loss": 1.2636,
      "step": 317312
    },
    {
      "epoch": 23.818086303939964,
      "grad_norm": 0.5069259405136108,
      "learning_rate": 0.002828,
      "loss": 1.2591,
      "step": 317376
    },
    {
      "epoch": 23.822889305816133,
      "grad_norm": 0.8760371804237366,
      "learning_rate": 0.002828,
      "loss": 1.2586,
      "step": 317440
    },
    {
      "epoch": 23.827692307692306,
      "grad_norm": 0.6794724464416504,
      "learning_rate": 0.002828,
      "loss": 1.2548,
      "step": 317504
    },
    {
      "epoch": 23.83249530956848,
      "grad_norm": 0.5844535231590271,
      "learning_rate": 0.002828,
      "loss": 1.2602,
      "step": 317568
    },
    {
      "epoch": 23.837298311444652,
      "grad_norm": 0.5497626662254333,
      "learning_rate": 0.002828,
      "loss": 1.2613,
      "step": 317632
    },
    {
      "epoch": 23.842101313320825,
      "grad_norm": 0.5619154572486877,
      "learning_rate": 0.002828,
      "loss": 1.2647,
      "step": 317696
    },
    {
      "epoch": 23.846904315196998,
      "grad_norm": 0.49801504611968994,
      "learning_rate": 0.002828,
      "loss": 1.2678,
      "step": 317760
    },
    {
      "epoch": 23.85170731707317,
      "grad_norm": 0.5336838364601135,
      "learning_rate": 0.002828,
      "loss": 1.264,
      "step": 317824
    },
    {
      "epoch": 23.856510318949343,
      "grad_norm": 0.5173972845077515,
      "learning_rate": 0.002828,
      "loss": 1.2602,
      "step": 317888
    },
    {
      "epoch": 23.861313320825516,
      "grad_norm": 0.5220203399658203,
      "learning_rate": 0.002828,
      "loss": 1.2601,
      "step": 317952
    },
    {
      "epoch": 23.86611632270169,
      "grad_norm": 0.5974367260932922,
      "learning_rate": 0.002828,
      "loss": 1.2599,
      "step": 318016
    },
    {
      "epoch": 23.870919324577862,
      "grad_norm": 0.6067824363708496,
      "learning_rate": 0.002828,
      "loss": 1.2618,
      "step": 318080
    },
    {
      "epoch": 23.875722326454035,
      "grad_norm": 0.6112476587295532,
      "learning_rate": 0.002828,
      "loss": 1.261,
      "step": 318144
    },
    {
      "epoch": 23.880525328330208,
      "grad_norm": 0.5728985071182251,
      "learning_rate": 0.002828,
      "loss": 1.2524,
      "step": 318208
    },
    {
      "epoch": 23.885328330206377,
      "grad_norm": 0.5144209265708923,
      "learning_rate": 0.002828,
      "loss": 1.2581,
      "step": 318272
    },
    {
      "epoch": 23.89013133208255,
      "grad_norm": 0.5003145337104797,
      "learning_rate": 0.002828,
      "loss": 1.2571,
      "step": 318336
    },
    {
      "epoch": 23.894934333958723,
      "grad_norm": 0.5286693572998047,
      "learning_rate": 0.002828,
      "loss": 1.2614,
      "step": 318400
    },
    {
      "epoch": 23.899737335834896,
      "grad_norm": 0.6406574249267578,
      "learning_rate": 0.002828,
      "loss": 1.2615,
      "step": 318464
    },
    {
      "epoch": 23.90454033771107,
      "grad_norm": 0.6060843467712402,
      "learning_rate": 0.002828,
      "loss": 1.2617,
      "step": 318528
    },
    {
      "epoch": 23.90934333958724,
      "grad_norm": 0.5950831174850464,
      "learning_rate": 0.002828,
      "loss": 1.2646,
      "step": 318592
    },
    {
      "epoch": 23.914146341463415,
      "grad_norm": 0.6237718462944031,
      "learning_rate": 0.002828,
      "loss": 1.2632,
      "step": 318656
    },
    {
      "epoch": 23.918949343339587,
      "grad_norm": 0.6165586709976196,
      "learning_rate": 0.002828,
      "loss": 1.2601,
      "step": 318720
    },
    {
      "epoch": 23.92375234521576,
      "grad_norm": 0.6163604259490967,
      "learning_rate": 0.002828,
      "loss": 1.2625,
      "step": 318784
    },
    {
      "epoch": 23.928555347091933,
      "grad_norm": 0.4493919014930725,
      "learning_rate": 0.002828,
      "loss": 1.2594,
      "step": 318848
    },
    {
      "epoch": 23.933358348968106,
      "grad_norm": 0.5826929211616516,
      "learning_rate": 0.002828,
      "loss": 1.2648,
      "step": 318912
    },
    {
      "epoch": 23.93816135084428,
      "grad_norm": 0.5376659631729126,
      "learning_rate": 0.002828,
      "loss": 1.2665,
      "step": 318976
    },
    {
      "epoch": 23.942964352720452,
      "grad_norm": 0.6579872369766235,
      "learning_rate": 0.002828,
      "loss": 1.2586,
      "step": 319040
    },
    {
      "epoch": 23.94776735459662,
      "grad_norm": 0.5282186269760132,
      "learning_rate": 0.002828,
      "loss": 1.2665,
      "step": 319104
    },
    {
      "epoch": 23.952570356472794,
      "grad_norm": 0.666273295879364,
      "learning_rate": 0.002828,
      "loss": 1.2608,
      "step": 319168
    },
    {
      "epoch": 23.957373358348967,
      "grad_norm": 0.5335906147956848,
      "learning_rate": 0.002828,
      "loss": 1.2585,
      "step": 319232
    },
    {
      "epoch": 23.96217636022514,
      "grad_norm": 0.576663613319397,
      "learning_rate": 0.002828,
      "loss": 1.26,
      "step": 319296
    },
    {
      "epoch": 23.966979362101313,
      "grad_norm": 0.565240204334259,
      "learning_rate": 0.002828,
      "loss": 1.2635,
      "step": 319360
    },
    {
      "epoch": 23.971782363977486,
      "grad_norm": 0.5416858196258545,
      "learning_rate": 0.002828,
      "loss": 1.262,
      "step": 319424
    },
    {
      "epoch": 23.97658536585366,
      "grad_norm": 0.6308193206787109,
      "learning_rate": 0.002828,
      "loss": 1.259,
      "step": 319488
    },
    {
      "epoch": 23.98138836772983,
      "grad_norm": 0.6142536401748657,
      "learning_rate": 0.002828,
      "loss": 1.2629,
      "step": 319552
    },
    {
      "epoch": 23.986191369606004,
      "grad_norm": 0.5130268335342407,
      "learning_rate": 0.002828,
      "loss": 1.2638,
      "step": 319616
    },
    {
      "epoch": 23.990994371482177,
      "grad_norm": 0.5770696401596069,
      "learning_rate": 0.002828,
      "loss": 1.2651,
      "step": 319680
    },
    {
      "epoch": 23.99579737335835,
      "grad_norm": 0.5990517139434814,
      "learning_rate": 0.002828,
      "loss": 1.2555,
      "step": 319744
    },
    {
      "epoch": 24.000600375234523,
      "grad_norm": 0.6668261885643005,
      "learning_rate": 0.002828,
      "loss": 1.2589,
      "step": 319808
    },
    {
      "epoch": 24.005403377110696,
      "grad_norm": 0.511762797832489,
      "learning_rate": 0.002828,
      "loss": 1.2231,
      "step": 319872
    },
    {
      "epoch": 24.010206378986865,
      "grad_norm": 0.5417555570602417,
      "learning_rate": 0.002828,
      "loss": 1.2193,
      "step": 319936
    },
    {
      "epoch": 24.015009380863038,
      "grad_norm": 0.5588164925575256,
      "learning_rate": 0.002828,
      "loss": 1.2193,
      "step": 320000
    },
    {
      "epoch": 24.01981238273921,
      "grad_norm": 0.6115003824234009,
      "learning_rate": 0.002828,
      "loss": 1.2227,
      "step": 320064
    },
    {
      "epoch": 24.024615384615384,
      "grad_norm": 0.5852314829826355,
      "learning_rate": 0.002828,
      "loss": 1.2229,
      "step": 320128
    },
    {
      "epoch": 24.029418386491557,
      "grad_norm": 0.5571886301040649,
      "learning_rate": 0.002828,
      "loss": 1.2246,
      "step": 320192
    },
    {
      "epoch": 24.03422138836773,
      "grad_norm": 0.5183240175247192,
      "learning_rate": 0.002828,
      "loss": 1.2263,
      "step": 320256
    },
    {
      "epoch": 24.039024390243902,
      "grad_norm": 0.5464686751365662,
      "learning_rate": 0.002828,
      "loss": 1.224,
      "step": 320320
    },
    {
      "epoch": 24.043827392120075,
      "grad_norm": 0.6346117258071899,
      "learning_rate": 0.002828,
      "loss": 1.2244,
      "step": 320384
    },
    {
      "epoch": 24.04863039399625,
      "grad_norm": 0.5489941835403442,
      "learning_rate": 0.002828,
      "loss": 1.2294,
      "step": 320448
    },
    {
      "epoch": 24.05343339587242,
      "grad_norm": 0.5453773140907288,
      "learning_rate": 0.002828,
      "loss": 1.224,
      "step": 320512
    },
    {
      "epoch": 24.058236397748594,
      "grad_norm": 0.5428609251976013,
      "learning_rate": 0.002828,
      "loss": 1.226,
      "step": 320576
    },
    {
      "epoch": 24.063039399624767,
      "grad_norm": 0.5705650448799133,
      "learning_rate": 0.002828,
      "loss": 1.2266,
      "step": 320640
    },
    {
      "epoch": 24.06784240150094,
      "grad_norm": 0.6298850774765015,
      "learning_rate": 0.002828,
      "loss": 1.2264,
      "step": 320704
    },
    {
      "epoch": 24.07264540337711,
      "grad_norm": 0.6152244210243225,
      "learning_rate": 0.002828,
      "loss": 1.2287,
      "step": 320768
    },
    {
      "epoch": 24.077448405253282,
      "grad_norm": 0.47920313477516174,
      "learning_rate": 0.002828,
      "loss": 1.2313,
      "step": 320832
    },
    {
      "epoch": 24.082251407129455,
      "grad_norm": 0.6054471731185913,
      "learning_rate": 0.002828,
      "loss": 1.2287,
      "step": 320896
    },
    {
      "epoch": 24.087054409005628,
      "grad_norm": 0.6222940683364868,
      "learning_rate": 0.002828,
      "loss": 1.2284,
      "step": 320960
    },
    {
      "epoch": 24.0918574108818,
      "grad_norm": 0.6720530390739441,
      "learning_rate": 0.002828,
      "loss": 1.2318,
      "step": 321024
    },
    {
      "epoch": 24.096660412757974,
      "grad_norm": 0.5883446335792542,
      "learning_rate": 0.002828,
      "loss": 1.2312,
      "step": 321088
    },
    {
      "epoch": 24.101463414634146,
      "grad_norm": 0.5007783770561218,
      "learning_rate": 0.002828,
      "loss": 1.2261,
      "step": 321152
    },
    {
      "epoch": 24.10626641651032,
      "grad_norm": 0.56673264503479,
      "learning_rate": 0.002828,
      "loss": 1.2274,
      "step": 321216
    },
    {
      "epoch": 24.111069418386492,
      "grad_norm": 0.5843421816825867,
      "learning_rate": 0.002828,
      "loss": 1.2324,
      "step": 321280
    },
    {
      "epoch": 24.115872420262665,
      "grad_norm": 0.5719348192214966,
      "learning_rate": 0.002828,
      "loss": 1.2279,
      "step": 321344
    },
    {
      "epoch": 24.120675422138838,
      "grad_norm": 0.5110663771629333,
      "learning_rate": 0.002828,
      "loss": 1.2298,
      "step": 321408
    },
    {
      "epoch": 24.12547842401501,
      "grad_norm": 0.5339451432228088,
      "learning_rate": 0.002828,
      "loss": 1.2315,
      "step": 321472
    },
    {
      "epoch": 24.130281425891184,
      "grad_norm": 0.574852705001831,
      "learning_rate": 0.002828,
      "loss": 1.2359,
      "step": 321536
    },
    {
      "epoch": 24.135084427767353,
      "grad_norm": 0.5415715575218201,
      "learning_rate": 0.002828,
      "loss": 1.2232,
      "step": 321600
    },
    {
      "epoch": 24.139887429643526,
      "grad_norm": 0.5436941385269165,
      "learning_rate": 0.002828,
      "loss": 1.2306,
      "step": 321664
    },
    {
      "epoch": 24.1446904315197,
      "grad_norm": 0.6057150959968567,
      "learning_rate": 0.002828,
      "loss": 1.2322,
      "step": 321728
    },
    {
      "epoch": 24.14949343339587,
      "grad_norm": 0.5295004844665527,
      "learning_rate": 0.002828,
      "loss": 1.2261,
      "step": 321792
    },
    {
      "epoch": 24.154296435272045,
      "grad_norm": 0.573469340801239,
      "learning_rate": 0.002828,
      "loss": 1.23,
      "step": 321856
    },
    {
      "epoch": 24.159099437148218,
      "grad_norm": 0.5898007750511169,
      "learning_rate": 0.002828,
      "loss": 1.2311,
      "step": 321920
    },
    {
      "epoch": 24.16390243902439,
      "grad_norm": 0.6019816994667053,
      "learning_rate": 0.002828,
      "loss": 1.233,
      "step": 321984
    },
    {
      "epoch": 24.168705440900563,
      "grad_norm": 0.7136012315750122,
      "learning_rate": 0.002828,
      "loss": 1.2301,
      "step": 322048
    },
    {
      "epoch": 24.173508442776736,
      "grad_norm": 0.7115374803543091,
      "learning_rate": 0.002828,
      "loss": 1.2339,
      "step": 322112
    },
    {
      "epoch": 24.17831144465291,
      "grad_norm": 0.6407468914985657,
      "learning_rate": 0.002828,
      "loss": 1.2349,
      "step": 322176
    },
    {
      "epoch": 24.183114446529082,
      "grad_norm": 0.718950629234314,
      "learning_rate": 0.002828,
      "loss": 1.227,
      "step": 322240
    },
    {
      "epoch": 24.187917448405255,
      "grad_norm": 0.4775286912918091,
      "learning_rate": 0.002828,
      "loss": 1.2332,
      "step": 322304
    },
    {
      "epoch": 24.192720450281424,
      "grad_norm": 0.6076687574386597,
      "learning_rate": 0.002828,
      "loss": 1.2331,
      "step": 322368
    },
    {
      "epoch": 24.197523452157597,
      "grad_norm": 0.5814255475997925,
      "learning_rate": 0.002828,
      "loss": 1.2329,
      "step": 322432
    },
    {
      "epoch": 24.20232645403377,
      "grad_norm": 0.5358974933624268,
      "learning_rate": 0.002828,
      "loss": 1.2366,
      "step": 322496
    },
    {
      "epoch": 24.207129455909943,
      "grad_norm": 0.5103739500045776,
      "learning_rate": 0.002828,
      "loss": 1.23,
      "step": 322560
    },
    {
      "epoch": 24.211932457786116,
      "grad_norm": 0.570276141166687,
      "learning_rate": 0.002828,
      "loss": 1.2315,
      "step": 322624
    },
    {
      "epoch": 24.21673545966229,
      "grad_norm": 0.6150188446044922,
      "learning_rate": 0.002828,
      "loss": 1.2246,
      "step": 322688
    },
    {
      "epoch": 24.22153846153846,
      "grad_norm": 0.4611063301563263,
      "learning_rate": 0.002828,
      "loss": 1.2341,
      "step": 322752
    },
    {
      "epoch": 24.226341463414634,
      "grad_norm": 0.6687694191932678,
      "learning_rate": 0.002828,
      "loss": 1.227,
      "step": 322816
    },
    {
      "epoch": 24.231144465290807,
      "grad_norm": 0.522638201713562,
      "learning_rate": 0.002828,
      "loss": 1.235,
      "step": 322880
    },
    {
      "epoch": 24.23594746716698,
      "grad_norm": 0.6681767106056213,
      "learning_rate": 0.002828,
      "loss": 1.2383,
      "step": 322944
    },
    {
      "epoch": 24.240750469043153,
      "grad_norm": 0.6990424394607544,
      "learning_rate": 0.002828,
      "loss": 1.232,
      "step": 323008
    },
    {
      "epoch": 24.245553470919326,
      "grad_norm": 0.635747492313385,
      "learning_rate": 0.002828,
      "loss": 1.2327,
      "step": 323072
    },
    {
      "epoch": 24.2503564727955,
      "grad_norm": 0.603436291217804,
      "learning_rate": 0.002828,
      "loss": 1.238,
      "step": 323136
    },
    {
      "epoch": 24.255159474671668,
      "grad_norm": 0.603233277797699,
      "learning_rate": 0.002828,
      "loss": 1.23,
      "step": 323200
    },
    {
      "epoch": 24.25996247654784,
      "grad_norm": 0.5681001543998718,
      "learning_rate": 0.002828,
      "loss": 1.2315,
      "step": 323264
    },
    {
      "epoch": 24.264765478424014,
      "grad_norm": 0.5758222937583923,
      "learning_rate": 0.002828,
      "loss": 1.2319,
      "step": 323328
    },
    {
      "epoch": 24.269568480300187,
      "grad_norm": 0.6049182415008545,
      "learning_rate": 0.002828,
      "loss": 1.2345,
      "step": 323392
    },
    {
      "epoch": 24.27437148217636,
      "grad_norm": 0.5548405051231384,
      "learning_rate": 0.002828,
      "loss": 1.2415,
      "step": 323456
    },
    {
      "epoch": 24.279174484052533,
      "grad_norm": 0.48350638151168823,
      "learning_rate": 0.002828,
      "loss": 1.2378,
      "step": 323520
    },
    {
      "epoch": 24.283977485928705,
      "grad_norm": 0.5970533490180969,
      "learning_rate": 0.002828,
      "loss": 1.2375,
      "step": 323584
    },
    {
      "epoch": 24.28878048780488,
      "grad_norm": 0.5122383236885071,
      "learning_rate": 0.002828,
      "loss": 1.2393,
      "step": 323648
    },
    {
      "epoch": 24.29358348968105,
      "grad_norm": 0.5411232113838196,
      "learning_rate": 0.002828,
      "loss": 1.2321,
      "step": 323712
    },
    {
      "epoch": 24.298386491557224,
      "grad_norm": 0.6688981056213379,
      "learning_rate": 0.002828,
      "loss": 1.2389,
      "step": 323776
    },
    {
      "epoch": 24.303189493433397,
      "grad_norm": 0.594946563243866,
      "learning_rate": 0.002828,
      "loss": 1.2364,
      "step": 323840
    },
    {
      "epoch": 24.30799249530957,
      "grad_norm": 0.5973834991455078,
      "learning_rate": 0.002828,
      "loss": 1.2354,
      "step": 323904
    },
    {
      "epoch": 24.312795497185743,
      "grad_norm": 0.6772333383560181,
      "learning_rate": 0.002828,
      "loss": 1.2408,
      "step": 323968
    },
    {
      "epoch": 24.317598499061912,
      "grad_norm": 0.5157610774040222,
      "learning_rate": 0.002828,
      "loss": 1.2302,
      "step": 324032
    },
    {
      "epoch": 24.322401500938085,
      "grad_norm": 0.6570464968681335,
      "learning_rate": 0.002828,
      "loss": 1.235,
      "step": 324096
    },
    {
      "epoch": 24.327204502814258,
      "grad_norm": 0.6352835297584534,
      "learning_rate": 0.002828,
      "loss": 1.2415,
      "step": 324160
    },
    {
      "epoch": 24.33200750469043,
      "grad_norm": 0.504051148891449,
      "learning_rate": 0.002828,
      "loss": 1.2316,
      "step": 324224
    },
    {
      "epoch": 24.336810506566604,
      "grad_norm": 0.5722164511680603,
      "learning_rate": 0.002828,
      "loss": 1.2435,
      "step": 324288
    },
    {
      "epoch": 24.341613508442776,
      "grad_norm": 0.6468031406402588,
      "learning_rate": 0.002828,
      "loss": 1.2351,
      "step": 324352
    },
    {
      "epoch": 24.34641651031895,
      "grad_norm": 0.4418530762195587,
      "learning_rate": 0.002828,
      "loss": 1.2363,
      "step": 324416
    },
    {
      "epoch": 24.351219512195122,
      "grad_norm": 0.49656835198402405,
      "learning_rate": 0.002828,
      "loss": 1.2377,
      "step": 324480
    },
    {
      "epoch": 24.356022514071295,
      "grad_norm": 0.6155210733413696,
      "learning_rate": 0.002828,
      "loss": 1.2396,
      "step": 324544
    },
    {
      "epoch": 24.360825515947468,
      "grad_norm": 0.49608752131462097,
      "learning_rate": 0.002828,
      "loss": 1.24,
      "step": 324608
    },
    {
      "epoch": 24.36562851782364,
      "grad_norm": 0.46983015537261963,
      "learning_rate": 0.002828,
      "loss": 1.2425,
      "step": 324672
    },
    {
      "epoch": 24.370431519699814,
      "grad_norm": 0.6393495798110962,
      "learning_rate": 0.002828,
      "loss": 1.2359,
      "step": 324736
    },
    {
      "epoch": 24.375234521575987,
      "grad_norm": 0.6129648089408875,
      "learning_rate": 0.002828,
      "loss": 1.238,
      "step": 324800
    },
    {
      "epoch": 24.380037523452156,
      "grad_norm": 0.7626722455024719,
      "learning_rate": 0.002828,
      "loss": 1.2364,
      "step": 324864
    },
    {
      "epoch": 24.38484052532833,
      "grad_norm": 0.4975043535232544,
      "learning_rate": 0.002828,
      "loss": 1.2371,
      "step": 324928
    },
    {
      "epoch": 24.389643527204502,
      "grad_norm": 0.6254850625991821,
      "learning_rate": 0.002828,
      "loss": 1.2428,
      "step": 324992
    },
    {
      "epoch": 24.394446529080675,
      "grad_norm": 0.5704091787338257,
      "learning_rate": 0.002828,
      "loss": 1.2426,
      "step": 325056
    },
    {
      "epoch": 24.399249530956848,
      "grad_norm": 0.5449159145355225,
      "learning_rate": 0.002828,
      "loss": 1.2353,
      "step": 325120
    },
    {
      "epoch": 24.40405253283302,
      "grad_norm": 0.5547662377357483,
      "learning_rate": 0.002828,
      "loss": 1.2367,
      "step": 325184
    },
    {
      "epoch": 24.408855534709193,
      "grad_norm": 0.7620545029640198,
      "learning_rate": 0.002828,
      "loss": 1.2379,
      "step": 325248
    },
    {
      "epoch": 24.413658536585366,
      "grad_norm": 0.5453926920890808,
      "learning_rate": 0.002828,
      "loss": 1.2407,
      "step": 325312
    },
    {
      "epoch": 24.41846153846154,
      "grad_norm": 0.6322466135025024,
      "learning_rate": 0.002828,
      "loss": 1.2394,
      "step": 325376
    },
    {
      "epoch": 24.423264540337712,
      "grad_norm": 0.47954753041267395,
      "learning_rate": 0.002828,
      "loss": 1.2482,
      "step": 325440
    },
    {
      "epoch": 24.428067542213885,
      "grad_norm": 0.5336704850196838,
      "learning_rate": 0.002828,
      "loss": 1.2374,
      "step": 325504
    },
    {
      "epoch": 24.432870544090058,
      "grad_norm": 0.6320717334747314,
      "learning_rate": 0.002828,
      "loss": 1.2398,
      "step": 325568
    },
    {
      "epoch": 24.437673545966227,
      "grad_norm": 0.540543258190155,
      "learning_rate": 0.002828,
      "loss": 1.2385,
      "step": 325632
    },
    {
      "epoch": 24.4424765478424,
      "grad_norm": 0.5124441981315613,
      "learning_rate": 0.002828,
      "loss": 1.2398,
      "step": 325696
    },
    {
      "epoch": 24.447279549718573,
      "grad_norm": 0.6076022386550903,
      "learning_rate": 0.002828,
      "loss": 1.2363,
      "step": 325760
    },
    {
      "epoch": 24.452082551594746,
      "grad_norm": 0.4827095568180084,
      "learning_rate": 0.002828,
      "loss": 1.2444,
      "step": 325824
    },
    {
      "epoch": 24.45688555347092,
      "grad_norm": 0.6274040937423706,
      "learning_rate": 0.002828,
      "loss": 1.2423,
      "step": 325888
    },
    {
      "epoch": 24.46168855534709,
      "grad_norm": 0.49718886613845825,
      "learning_rate": 0.002828,
      "loss": 1.2441,
      "step": 325952
    },
    {
      "epoch": 24.466491557223264,
      "grad_norm": 0.5871378779411316,
      "learning_rate": 0.002828,
      "loss": 1.2401,
      "step": 326016
    },
    {
      "epoch": 24.471294559099437,
      "grad_norm": 0.6377558708190918,
      "learning_rate": 0.002828,
      "loss": 1.2387,
      "step": 326080
    },
    {
      "epoch": 24.47609756097561,
      "grad_norm": 0.6675707697868347,
      "learning_rate": 0.002828,
      "loss": 1.2369,
      "step": 326144
    },
    {
      "epoch": 24.480900562851783,
      "grad_norm": 0.5000851154327393,
      "learning_rate": 0.002828,
      "loss": 1.2403,
      "step": 326208
    },
    {
      "epoch": 24.485703564727956,
      "grad_norm": 0.6159429550170898,
      "learning_rate": 0.002828,
      "loss": 1.2411,
      "step": 326272
    },
    {
      "epoch": 24.49050656660413,
      "grad_norm": 0.5147876143455505,
      "learning_rate": 0.002828,
      "loss": 1.2425,
      "step": 326336
    },
    {
      "epoch": 24.4953095684803,
      "grad_norm": 0.625784695148468,
      "learning_rate": 0.002828,
      "loss": 1.2449,
      "step": 326400
    },
    {
      "epoch": 24.50011257035647,
      "grad_norm": 0.5830493569374084,
      "learning_rate": 0.002828,
      "loss": 1.2395,
      "step": 326464
    },
    {
      "epoch": 24.504915572232644,
      "grad_norm": 0.5036903023719788,
      "learning_rate": 0.002828,
      "loss": 1.237,
      "step": 326528
    },
    {
      "epoch": 24.509718574108817,
      "grad_norm": 0.5821319222450256,
      "learning_rate": 0.002828,
      "loss": 1.2448,
      "step": 326592
    },
    {
      "epoch": 24.51452157598499,
      "grad_norm": 0.4889194667339325,
      "learning_rate": 0.002828,
      "loss": 1.2449,
      "step": 326656
    },
    {
      "epoch": 24.519324577861163,
      "grad_norm": 0.5718082189559937,
      "learning_rate": 0.002828,
      "loss": 1.2452,
      "step": 326720
    },
    {
      "epoch": 24.524127579737335,
      "grad_norm": 0.5081624984741211,
      "learning_rate": 0.002828,
      "loss": 1.24,
      "step": 326784
    },
    {
      "epoch": 24.52893058161351,
      "grad_norm": 0.5082750916481018,
      "learning_rate": 0.002828,
      "loss": 1.2424,
      "step": 326848
    },
    {
      "epoch": 24.53373358348968,
      "grad_norm": 0.592952311038971,
      "learning_rate": 0.002828,
      "loss": 1.2394,
      "step": 326912
    },
    {
      "epoch": 24.538536585365854,
      "grad_norm": 0.6645016670227051,
      "learning_rate": 0.002828,
      "loss": 1.2419,
      "step": 326976
    },
    {
      "epoch": 24.543339587242027,
      "grad_norm": 0.5564192533493042,
      "learning_rate": 0.002828,
      "loss": 1.2457,
      "step": 327040
    },
    {
      "epoch": 24.5481425891182,
      "grad_norm": 0.6197208166122437,
      "learning_rate": 0.002828,
      "loss": 1.2441,
      "step": 327104
    },
    {
      "epoch": 24.552945590994373,
      "grad_norm": 0.6238077878952026,
      "learning_rate": 0.002828,
      "loss": 1.2384,
      "step": 327168
    },
    {
      "epoch": 24.557748592870546,
      "grad_norm": 0.5843613743782043,
      "learning_rate": 0.002828,
      "loss": 1.2416,
      "step": 327232
    },
    {
      "epoch": 24.562551594746715,
      "grad_norm": 0.6086361408233643,
      "learning_rate": 0.002828,
      "loss": 1.2429,
      "step": 327296
    },
    {
      "epoch": 24.567354596622888,
      "grad_norm": 0.5316898822784424,
      "learning_rate": 0.002828,
      "loss": 1.2424,
      "step": 327360
    },
    {
      "epoch": 24.57215759849906,
      "grad_norm": 0.6011339426040649,
      "learning_rate": 0.002828,
      "loss": 1.2401,
      "step": 327424
    },
    {
      "epoch": 24.576960600375234,
      "grad_norm": 0.6036356687545776,
      "learning_rate": 0.002828,
      "loss": 1.2383,
      "step": 327488
    },
    {
      "epoch": 24.581763602251407,
      "grad_norm": 0.5892400145530701,
      "learning_rate": 0.002828,
      "loss": 1.2411,
      "step": 327552
    },
    {
      "epoch": 24.58656660412758,
      "grad_norm": 0.5821129083633423,
      "learning_rate": 0.002828,
      "loss": 1.2485,
      "step": 327616
    },
    {
      "epoch": 24.591369606003752,
      "grad_norm": 0.570139467716217,
      "learning_rate": 0.002828,
      "loss": 1.2372,
      "step": 327680
    },
    {
      "epoch": 24.596172607879925,
      "grad_norm": 0.54395991563797,
      "learning_rate": 0.002828,
      "loss": 1.2416,
      "step": 327744
    },
    {
      "epoch": 24.600975609756098,
      "grad_norm": 0.5703491568565369,
      "learning_rate": 0.002828,
      "loss": 1.2434,
      "step": 327808
    },
    {
      "epoch": 24.60577861163227,
      "grad_norm": 0.5862521529197693,
      "learning_rate": 0.002828,
      "loss": 1.2428,
      "step": 327872
    },
    {
      "epoch": 24.610581613508444,
      "grad_norm": 0.5634772777557373,
      "learning_rate": 0.002828,
      "loss": 1.2417,
      "step": 327936
    },
    {
      "epoch": 24.615384615384617,
      "grad_norm": 0.5159964561462402,
      "learning_rate": 0.002828,
      "loss": 1.2453,
      "step": 328000
    },
    {
      "epoch": 24.62018761726079,
      "grad_norm": 0.5875333547592163,
      "learning_rate": 0.002828,
      "loss": 1.247,
      "step": 328064
    },
    {
      "epoch": 24.62499061913696,
      "grad_norm": 0.6296870708465576,
      "learning_rate": 0.002828,
      "loss": 1.2458,
      "step": 328128
    },
    {
      "epoch": 24.629793621013132,
      "grad_norm": 0.5884017944335938,
      "learning_rate": 0.002828,
      "loss": 1.253,
      "step": 328192
    },
    {
      "epoch": 24.634596622889305,
      "grad_norm": 0.749247670173645,
      "learning_rate": 0.002828,
      "loss": 1.2331,
      "step": 328256
    },
    {
      "epoch": 24.639399624765478,
      "grad_norm": 0.5577207803726196,
      "learning_rate": 0.002828,
      "loss": 1.2441,
      "step": 328320
    },
    {
      "epoch": 24.64420262664165,
      "grad_norm": 0.6233530044555664,
      "learning_rate": 0.002828,
      "loss": 1.2486,
      "step": 328384
    },
    {
      "epoch": 24.649005628517823,
      "grad_norm": 0.5876480340957642,
      "learning_rate": 0.002828,
      "loss": 1.2415,
      "step": 328448
    },
    {
      "epoch": 24.653808630393996,
      "grad_norm": 0.6091268062591553,
      "learning_rate": 0.002828,
      "loss": 1.2382,
      "step": 328512
    },
    {
      "epoch": 24.65861163227017,
      "grad_norm": 0.5153234004974365,
      "learning_rate": 0.002828,
      "loss": 1.246,
      "step": 328576
    },
    {
      "epoch": 24.663414634146342,
      "grad_norm": 0.6070051193237305,
      "learning_rate": 0.002828,
      "loss": 1.249,
      "step": 328640
    },
    {
      "epoch": 24.668217636022515,
      "grad_norm": 0.6957892179489136,
      "learning_rate": 0.002828,
      "loss": 1.2457,
      "step": 328704
    },
    {
      "epoch": 24.673020637898688,
      "grad_norm": 0.6002076268196106,
      "learning_rate": 0.002828,
      "loss": 1.2389,
      "step": 328768
    },
    {
      "epoch": 24.67782363977486,
      "grad_norm": 0.5522453188896179,
      "learning_rate": 0.002828,
      "loss": 1.2462,
      "step": 328832
    },
    {
      "epoch": 24.682626641651034,
      "grad_norm": 0.6673900485038757,
      "learning_rate": 0.002828,
      "loss": 1.2483,
      "step": 328896
    },
    {
      "epoch": 24.687429643527203,
      "grad_norm": 0.5456265807151794,
      "learning_rate": 0.002828,
      "loss": 1.2353,
      "step": 328960
    },
    {
      "epoch": 24.692232645403376,
      "grad_norm": 0.6234123110771179,
      "learning_rate": 0.002828,
      "loss": 1.2536,
      "step": 329024
    },
    {
      "epoch": 24.69703564727955,
      "grad_norm": 0.5306026339530945,
      "learning_rate": 0.002828,
      "loss": 1.2464,
      "step": 329088
    },
    {
      "epoch": 24.70183864915572,
      "grad_norm": 0.5644907355308533,
      "learning_rate": 0.002828,
      "loss": 1.2485,
      "step": 329152
    },
    {
      "epoch": 24.706641651031894,
      "grad_norm": 0.5596889853477478,
      "learning_rate": 0.002828,
      "loss": 1.2375,
      "step": 329216
    },
    {
      "epoch": 24.711444652908067,
      "grad_norm": 0.5232831835746765,
      "learning_rate": 0.002828,
      "loss": 1.2446,
      "step": 329280
    },
    {
      "epoch": 24.71624765478424,
      "grad_norm": 0.557348906993866,
      "learning_rate": 0.002828,
      "loss": 1.25,
      "step": 329344
    },
    {
      "epoch": 24.721050656660413,
      "grad_norm": 0.5692952871322632,
      "learning_rate": 0.002828,
      "loss": 1.2447,
      "step": 329408
    },
    {
      "epoch": 24.725853658536586,
      "grad_norm": 0.5626512169837952,
      "learning_rate": 0.002828,
      "loss": 1.2452,
      "step": 329472
    },
    {
      "epoch": 24.73065666041276,
      "grad_norm": 0.5778016448020935,
      "learning_rate": 0.002828,
      "loss": 1.2429,
      "step": 329536
    },
    {
      "epoch": 24.735459662288932,
      "grad_norm": 0.6249626278877258,
      "learning_rate": 0.002828,
      "loss": 1.2449,
      "step": 329600
    },
    {
      "epoch": 24.740262664165105,
      "grad_norm": 0.6157994270324707,
      "learning_rate": 0.002828,
      "loss": 1.2446,
      "step": 329664
    },
    {
      "epoch": 24.745065666041278,
      "grad_norm": 0.6171292662620544,
      "learning_rate": 0.002828,
      "loss": 1.2384,
      "step": 329728
    },
    {
      "epoch": 24.749868667917447,
      "grad_norm": 0.5539014935493469,
      "learning_rate": 0.002828,
      "loss": 1.2455,
      "step": 329792
    },
    {
      "epoch": 24.75467166979362,
      "grad_norm": 0.5088152885437012,
      "learning_rate": 0.002828,
      "loss": 1.251,
      "step": 329856
    },
    {
      "epoch": 24.759474671669793,
      "grad_norm": 0.6428780555725098,
      "learning_rate": 0.002828,
      "loss": 1.2421,
      "step": 329920
    },
    {
      "epoch": 24.764277673545966,
      "grad_norm": 0.6033629179000854,
      "learning_rate": 0.002828,
      "loss": 1.2461,
      "step": 329984
    },
    {
      "epoch": 24.76908067542214,
      "grad_norm": 0.6996398568153381,
      "learning_rate": 0.002828,
      "loss": 1.2451,
      "step": 330048
    },
    {
      "epoch": 24.77388367729831,
      "grad_norm": 0.5007488131523132,
      "learning_rate": 0.002828,
      "loss": 1.247,
      "step": 330112
    },
    {
      "epoch": 24.778686679174484,
      "grad_norm": 0.6352538466453552,
      "learning_rate": 0.002828,
      "loss": 1.2454,
      "step": 330176
    },
    {
      "epoch": 24.783489681050657,
      "grad_norm": 0.6288419961929321,
      "learning_rate": 0.002828,
      "loss": 1.2469,
      "step": 330240
    },
    {
      "epoch": 24.78829268292683,
      "grad_norm": 0.5228093266487122,
      "learning_rate": 0.002828,
      "loss": 1.2371,
      "step": 330304
    },
    {
      "epoch": 24.793095684803003,
      "grad_norm": 0.5313441753387451,
      "learning_rate": 0.002828,
      "loss": 1.2481,
      "step": 330368
    },
    {
      "epoch": 24.797898686679176,
      "grad_norm": 0.5033171772956848,
      "learning_rate": 0.002828,
      "loss": 1.2441,
      "step": 330432
    },
    {
      "epoch": 24.80270168855535,
      "grad_norm": 0.6732704043388367,
      "learning_rate": 0.002828,
      "loss": 1.2424,
      "step": 330496
    },
    {
      "epoch": 24.80750469043152,
      "grad_norm": 0.6865556240081787,
      "learning_rate": 0.002828,
      "loss": 1.2479,
      "step": 330560
    },
    {
      "epoch": 24.81230769230769,
      "grad_norm": 0.6815218925476074,
      "learning_rate": 0.002828,
      "loss": 1.2452,
      "step": 330624
    },
    {
      "epoch": 24.817110694183864,
      "grad_norm": 0.5610247254371643,
      "learning_rate": 0.002828,
      "loss": 1.2476,
      "step": 330688
    },
    {
      "epoch": 24.821913696060037,
      "grad_norm": 0.5337264537811279,
      "learning_rate": 0.002828,
      "loss": 1.2421,
      "step": 330752
    },
    {
      "epoch": 24.82671669793621,
      "grad_norm": 0.6259786486625671,
      "learning_rate": 0.002828,
      "loss": 1.2493,
      "step": 330816
    },
    {
      "epoch": 24.831519699812382,
      "grad_norm": 0.5326517820358276,
      "learning_rate": 0.002828,
      "loss": 1.2528,
      "step": 330880
    },
    {
      "epoch": 24.836322701688555,
      "grad_norm": 0.6553822159767151,
      "learning_rate": 0.002828,
      "loss": 1.2551,
      "step": 330944
    },
    {
      "epoch": 24.841125703564728,
      "grad_norm": 0.47039973735809326,
      "learning_rate": 0.002828,
      "loss": 1.2456,
      "step": 331008
    },
    {
      "epoch": 24.8459287054409,
      "grad_norm": 0.5827360153198242,
      "learning_rate": 0.002828,
      "loss": 1.2433,
      "step": 331072
    },
    {
      "epoch": 24.850731707317074,
      "grad_norm": 0.660505473613739,
      "learning_rate": 0.002828,
      "loss": 1.2538,
      "step": 331136
    },
    {
      "epoch": 24.855534709193247,
      "grad_norm": 0.5613062977790833,
      "learning_rate": 0.002828,
      "loss": 1.2443,
      "step": 331200
    },
    {
      "epoch": 24.86033771106942,
      "grad_norm": 0.6975498795509338,
      "learning_rate": 0.002828,
      "loss": 1.2495,
      "step": 331264
    },
    {
      "epoch": 24.865140712945593,
      "grad_norm": 0.6232232451438904,
      "learning_rate": 0.002828,
      "loss": 1.2478,
      "step": 331328
    },
    {
      "epoch": 24.869943714821762,
      "grad_norm": 0.6001636385917664,
      "learning_rate": 0.002828,
      "loss": 1.2479,
      "step": 331392
    },
    {
      "epoch": 24.874746716697935,
      "grad_norm": 0.596153199672699,
      "learning_rate": 0.002828,
      "loss": 1.2536,
      "step": 331456
    },
    {
      "epoch": 24.879549718574108,
      "grad_norm": 0.6497877836227417,
      "learning_rate": 0.002828,
      "loss": 1.2482,
      "step": 331520
    },
    {
      "epoch": 24.88435272045028,
      "grad_norm": 0.5464568734169006,
      "learning_rate": 0.002828,
      "loss": 1.2473,
      "step": 331584
    },
    {
      "epoch": 24.889155722326453,
      "grad_norm": 0.5703956484794617,
      "learning_rate": 0.002828,
      "loss": 1.2567,
      "step": 331648
    },
    {
      "epoch": 24.893958724202626,
      "grad_norm": 0.5373716354370117,
      "learning_rate": 0.002828,
      "loss": 1.2563,
      "step": 331712
    },
    {
      "epoch": 24.8987617260788,
      "grad_norm": 0.5806722640991211,
      "learning_rate": 0.002828,
      "loss": 1.2453,
      "step": 331776
    },
    {
      "epoch": 24.903564727954972,
      "grad_norm": 0.6477004289627075,
      "learning_rate": 0.002828,
      "loss": 1.2567,
      "step": 331840
    },
    {
      "epoch": 24.908367729831145,
      "grad_norm": 0.6425051689147949,
      "learning_rate": 0.002828,
      "loss": 1.2456,
      "step": 331904
    },
    {
      "epoch": 24.913170731707318,
      "grad_norm": 0.46345874667167664,
      "learning_rate": 0.002828,
      "loss": 1.2482,
      "step": 331968
    },
    {
      "epoch": 24.91797373358349,
      "grad_norm": 0.6676378846168518,
      "learning_rate": 0.002828,
      "loss": 1.2454,
      "step": 332032
    },
    {
      "epoch": 24.922776735459664,
      "grad_norm": 0.5868575572967529,
      "learning_rate": 0.002828,
      "loss": 1.2476,
      "step": 332096
    },
    {
      "epoch": 24.927579737335837,
      "grad_norm": 0.5779089331626892,
      "learning_rate": 0.002828,
      "loss": 1.25,
      "step": 332160
    },
    {
      "epoch": 24.932382739212006,
      "grad_norm": 0.5757585763931274,
      "learning_rate": 0.002828,
      "loss": 1.2473,
      "step": 332224
    },
    {
      "epoch": 24.93718574108818,
      "grad_norm": 0.6109635233879089,
      "learning_rate": 0.002828,
      "loss": 1.2469,
      "step": 332288
    },
    {
      "epoch": 24.94198874296435,
      "grad_norm": 0.5922868847846985,
      "learning_rate": 0.002828,
      "loss": 1.253,
      "step": 332352
    },
    {
      "epoch": 24.946791744840525,
      "grad_norm": 0.5897438526153564,
      "learning_rate": 0.002828,
      "loss": 1.2477,
      "step": 332416
    },
    {
      "epoch": 24.951594746716697,
      "grad_norm": 0.54740971326828,
      "learning_rate": 0.002828,
      "loss": 1.2475,
      "step": 332480
    },
    {
      "epoch": 24.95639774859287,
      "grad_norm": 0.6911497712135315,
      "learning_rate": 0.002828,
      "loss": 1.2538,
      "step": 332544
    },
    {
      "epoch": 24.961200750469043,
      "grad_norm": 0.6406940221786499,
      "learning_rate": 0.002828,
      "loss": 1.2447,
      "step": 332608
    },
    {
      "epoch": 24.966003752345216,
      "grad_norm": 0.6405665874481201,
      "learning_rate": 0.002828,
      "loss": 1.2461,
      "step": 332672
    },
    {
      "epoch": 24.97080675422139,
      "grad_norm": 0.6722344160079956,
      "learning_rate": 0.002828,
      "loss": 1.2477,
      "step": 332736
    },
    {
      "epoch": 24.975609756097562,
      "grad_norm": 0.6267404556274414,
      "learning_rate": 0.002828,
      "loss": 1.2528,
      "step": 332800
    },
    {
      "epoch": 24.980412757973735,
      "grad_norm": 0.6228541135787964,
      "learning_rate": 0.002828,
      "loss": 1.2497,
      "step": 332864
    },
    {
      "epoch": 24.985215759849908,
      "grad_norm": 0.5420615673065186,
      "learning_rate": 0.002828,
      "loss": 1.2507,
      "step": 332928
    },
    {
      "epoch": 24.99001876172608,
      "grad_norm": 0.5710049271583557,
      "learning_rate": 0.002828,
      "loss": 1.2493,
      "step": 332992
    },
    {
      "epoch": 24.99482176360225,
      "grad_norm": 0.6155769228935242,
      "learning_rate": 0.002828,
      "loss": 1.25,
      "step": 333056
    },
    {
      "epoch": 24.999624765478423,
      "grad_norm": 0.6931199431419373,
      "learning_rate": 0.002828,
      "loss": 1.2476,
      "step": 333120
    },
    {
      "epoch": 25.004427767354596,
      "grad_norm": 0.5726258158683777,
      "learning_rate": 0.002828,
      "loss": 1.215,
      "step": 333184
    },
    {
      "epoch": 25.00923076923077,
      "grad_norm": 0.6955495476722717,
      "learning_rate": 0.002828,
      "loss": 1.2106,
      "step": 333248
    },
    {
      "epoch": 25.01403377110694,
      "grad_norm": 0.5848275423049927,
      "learning_rate": 0.002828,
      "loss": 1.2055,
      "step": 333312
    },
    {
      "epoch": 25.018836772983114,
      "grad_norm": 0.5955660939216614,
      "learning_rate": 0.002828,
      "loss": 1.2115,
      "step": 333376
    },
    {
      "epoch": 25.023639774859287,
      "grad_norm": 0.7096522450447083,
      "learning_rate": 0.002828,
      "loss": 1.2125,
      "step": 333440
    },
    {
      "epoch": 25.02844277673546,
      "grad_norm": 0.6311249136924744,
      "learning_rate": 0.002828,
      "loss": 1.214,
      "step": 333504
    },
    {
      "epoch": 25.033245778611633,
      "grad_norm": 0.5731686949729919,
      "learning_rate": 0.002828,
      "loss": 1.2164,
      "step": 333568
    },
    {
      "epoch": 25.038048780487806,
      "grad_norm": 0.5741409063339233,
      "learning_rate": 0.002828,
      "loss": 1.2083,
      "step": 333632
    },
    {
      "epoch": 25.04285178236398,
      "grad_norm": 0.6162007451057434,
      "learning_rate": 0.002828,
      "loss": 1.2087,
      "step": 333696
    },
    {
      "epoch": 25.04765478424015,
      "grad_norm": 0.5544477701187134,
      "learning_rate": 0.002828,
      "loss": 1.2106,
      "step": 333760
    },
    {
      "epoch": 25.052457786116324,
      "grad_norm": 0.5872480273246765,
      "learning_rate": 0.002828,
      "loss": 1.2141,
      "step": 333824
    },
    {
      "epoch": 25.057260787992494,
      "grad_norm": 0.575803279876709,
      "learning_rate": 0.002828,
      "loss": 1.2121,
      "step": 333888
    },
    {
      "epoch": 25.062063789868667,
      "grad_norm": 0.5969261527061462,
      "learning_rate": 0.002828,
      "loss": 1.2123,
      "step": 333952
    },
    {
      "epoch": 25.06686679174484,
      "grad_norm": 0.5465327501296997,
      "learning_rate": 0.002828,
      "loss": 1.2156,
      "step": 334016
    },
    {
      "epoch": 25.071669793621012,
      "grad_norm": 0.619968056678772,
      "learning_rate": 0.002828,
      "loss": 1.2128,
      "step": 334080
    },
    {
      "epoch": 25.076472795497185,
      "grad_norm": 0.5793817639350891,
      "learning_rate": 0.002828,
      "loss": 1.2147,
      "step": 334144
    },
    {
      "epoch": 25.081275797373358,
      "grad_norm": 0.6986529231071472,
      "learning_rate": 0.002828,
      "loss": 1.216,
      "step": 334208
    },
    {
      "epoch": 25.08607879924953,
      "grad_norm": 0.657088041305542,
      "learning_rate": 0.002828,
      "loss": 1.2194,
      "step": 334272
    },
    {
      "epoch": 25.090881801125704,
      "grad_norm": 0.5881126523017883,
      "learning_rate": 0.002828,
      "loss": 1.2164,
      "step": 334336
    },
    {
      "epoch": 25.095684803001877,
      "grad_norm": 0.7006486654281616,
      "learning_rate": 0.002828,
      "loss": 1.2176,
      "step": 334400
    },
    {
      "epoch": 25.10048780487805,
      "grad_norm": 0.659379243850708,
      "learning_rate": 0.002828,
      "loss": 1.2179,
      "step": 334464
    },
    {
      "epoch": 25.105290806754223,
      "grad_norm": 0.6826364994049072,
      "learning_rate": 0.002828,
      "loss": 1.2147,
      "step": 334528
    },
    {
      "epoch": 25.110093808630396,
      "grad_norm": 0.48427775502204895,
      "learning_rate": 0.002828,
      "loss": 1.2108,
      "step": 334592
    },
    {
      "epoch": 25.114896810506565,
      "grad_norm": 0.6580747961997986,
      "learning_rate": 0.002828,
      "loss": 1.216,
      "step": 334656
    },
    {
      "epoch": 25.119699812382738,
      "grad_norm": 0.639634370803833,
      "learning_rate": 0.002828,
      "loss": 1.2165,
      "step": 334720
    },
    {
      "epoch": 25.12450281425891,
      "grad_norm": 0.6104490160942078,
      "learning_rate": 0.002828,
      "loss": 1.2168,
      "step": 334784
    },
    {
      "epoch": 25.129305816135084,
      "grad_norm": 0.4889427423477173,
      "learning_rate": 0.002828,
      "loss": 1.2198,
      "step": 334848
    },
    {
      "epoch": 25.134108818011256,
      "grad_norm": 0.722379207611084,
      "learning_rate": 0.002828,
      "loss": 1.2154,
      "step": 334912
    },
    {
      "epoch": 25.13891181988743,
      "grad_norm": 0.6009499430656433,
      "learning_rate": 0.002828,
      "loss": 1.2233,
      "step": 334976
    },
    {
      "epoch": 25.143714821763602,
      "grad_norm": 0.6094868183135986,
      "learning_rate": 0.002828,
      "loss": 1.2134,
      "step": 335040
    },
    {
      "epoch": 25.148517823639775,
      "grad_norm": 0.5987445712089539,
      "learning_rate": 0.002828,
      "loss": 1.2195,
      "step": 335104
    },
    {
      "epoch": 25.153320825515948,
      "grad_norm": 0.5814436674118042,
      "learning_rate": 0.002828,
      "loss": 1.2146,
      "step": 335168
    },
    {
      "epoch": 25.15812382739212,
      "grad_norm": 0.7082455158233643,
      "learning_rate": 0.002828,
      "loss": 1.2162,
      "step": 335232
    },
    {
      "epoch": 25.162926829268294,
      "grad_norm": 0.6125720143318176,
      "learning_rate": 0.002828,
      "loss": 1.2214,
      "step": 335296
    },
    {
      "epoch": 25.167729831144467,
      "grad_norm": 0.715304970741272,
      "learning_rate": 0.002828,
      "loss": 1.2233,
      "step": 335360
    },
    {
      "epoch": 25.17253283302064,
      "grad_norm": 0.5754349231719971,
      "learning_rate": 0.002828,
      "loss": 1.2147,
      "step": 335424
    },
    {
      "epoch": 25.17733583489681,
      "grad_norm": 0.5818027853965759,
      "learning_rate": 0.002828,
      "loss": 1.2178,
      "step": 335488
    },
    {
      "epoch": 25.18213883677298,
      "grad_norm": 0.5796423554420471,
      "learning_rate": 0.002828,
      "loss": 1.2182,
      "step": 335552
    },
    {
      "epoch": 25.186941838649155,
      "grad_norm": 0.5709074139595032,
      "learning_rate": 0.002828,
      "loss": 1.2223,
      "step": 335616
    },
    {
      "epoch": 25.191744840525327,
      "grad_norm": 0.5902493596076965,
      "learning_rate": 0.002828,
      "loss": 1.222,
      "step": 335680
    },
    {
      "epoch": 25.1965478424015,
      "grad_norm": 0.6953119039535522,
      "learning_rate": 0.002828,
      "loss": 1.2207,
      "step": 335744
    },
    {
      "epoch": 25.201350844277673,
      "grad_norm": 0.6451302766799927,
      "learning_rate": 0.002828,
      "loss": 1.22,
      "step": 335808
    },
    {
      "epoch": 25.206153846153846,
      "grad_norm": 0.6847954392433167,
      "learning_rate": 0.002828,
      "loss": 1.2209,
      "step": 335872
    },
    {
      "epoch": 25.21095684803002,
      "grad_norm": 0.5568794012069702,
      "learning_rate": 0.002828,
      "loss": 1.2244,
      "step": 335936
    },
    {
      "epoch": 25.215759849906192,
      "grad_norm": 0.4961014986038208,
      "learning_rate": 0.002828,
      "loss": 1.2113,
      "step": 336000
    },
    {
      "epoch": 25.220562851782365,
      "grad_norm": 0.5914636850357056,
      "learning_rate": 0.002828,
      "loss": 1.2225,
      "step": 336064
    },
    {
      "epoch": 25.225365853658538,
      "grad_norm": 0.7324196100234985,
      "learning_rate": 0.002828,
      "loss": 1.2229,
      "step": 336128
    },
    {
      "epoch": 25.23016885553471,
      "grad_norm": 0.5581563711166382,
      "learning_rate": 0.002828,
      "loss": 1.2198,
      "step": 336192
    },
    {
      "epoch": 25.234971857410883,
      "grad_norm": 0.5884627103805542,
      "learning_rate": 0.002828,
      "loss": 1.2228,
      "step": 336256
    },
    {
      "epoch": 25.239774859287053,
      "grad_norm": 0.6624916195869446,
      "learning_rate": 0.002828,
      "loss": 1.2297,
      "step": 336320
    },
    {
      "epoch": 25.244577861163226,
      "grad_norm": 0.6953412890434265,
      "learning_rate": 0.002828,
      "loss": 1.2244,
      "step": 336384
    },
    {
      "epoch": 25.2493808630394,
      "grad_norm": 0.5076711773872375,
      "learning_rate": 0.002828,
      "loss": 1.2222,
      "step": 336448
    },
    {
      "epoch": 25.25418386491557,
      "grad_norm": 0.6712566614151001,
      "learning_rate": 0.002828,
      "loss": 1.2276,
      "step": 336512
    },
    {
      "epoch": 25.258986866791744,
      "grad_norm": 0.5484094023704529,
      "learning_rate": 0.002828,
      "loss": 1.2217,
      "step": 336576
    },
    {
      "epoch": 25.263789868667917,
      "grad_norm": 0.5387206673622131,
      "learning_rate": 0.002828,
      "loss": 1.2231,
      "step": 336640
    },
    {
      "epoch": 25.26859287054409,
      "grad_norm": 0.47983431816101074,
      "learning_rate": 0.002828,
      "loss": 1.2184,
      "step": 336704
    },
    {
      "epoch": 25.273395872420263,
      "grad_norm": 0.6589762568473816,
      "learning_rate": 0.002828,
      "loss": 1.2182,
      "step": 336768
    },
    {
      "epoch": 25.278198874296436,
      "grad_norm": 0.7007054686546326,
      "learning_rate": 0.002828,
      "loss": 1.2193,
      "step": 336832
    },
    {
      "epoch": 25.28300187617261,
      "grad_norm": 0.6451464891433716,
      "learning_rate": 0.002828,
      "loss": 1.2238,
      "step": 336896
    },
    {
      "epoch": 25.28780487804878,
      "grad_norm": 0.5933736562728882,
      "learning_rate": 0.002828,
      "loss": 1.2259,
      "step": 336960
    },
    {
      "epoch": 25.292607879924955,
      "grad_norm": 0.5318650603294373,
      "learning_rate": 0.002828,
      "loss": 1.2267,
      "step": 337024
    },
    {
      "epoch": 25.297410881801127,
      "grad_norm": 0.5615575313568115,
      "learning_rate": 0.002828,
      "loss": 1.2262,
      "step": 337088
    },
    {
      "epoch": 25.302213883677297,
      "grad_norm": 0.5231067538261414,
      "learning_rate": 0.002828,
      "loss": 1.2302,
      "step": 337152
    },
    {
      "epoch": 25.30701688555347,
      "grad_norm": 0.5721030235290527,
      "learning_rate": 0.002828,
      "loss": 1.2228,
      "step": 337216
    },
    {
      "epoch": 25.311819887429643,
      "grad_norm": 0.8417955040931702,
      "learning_rate": 0.002828,
      "loss": 1.219,
      "step": 337280
    },
    {
      "epoch": 25.316622889305815,
      "grad_norm": 0.6137963533401489,
      "learning_rate": 0.002828,
      "loss": 1.2214,
      "step": 337344
    },
    {
      "epoch": 25.32142589118199,
      "grad_norm": 0.6264809370040894,
      "learning_rate": 0.002828,
      "loss": 1.2256,
      "step": 337408
    },
    {
      "epoch": 25.32622889305816,
      "grad_norm": 0.561613142490387,
      "learning_rate": 0.002828,
      "loss": 1.2259,
      "step": 337472
    },
    {
      "epoch": 25.331031894934334,
      "grad_norm": 0.5886028409004211,
      "learning_rate": 0.002828,
      "loss": 1.2226,
      "step": 337536
    },
    {
      "epoch": 25.335834896810507,
      "grad_norm": 0.6436465382575989,
      "learning_rate": 0.002828,
      "loss": 1.2268,
      "step": 337600
    },
    {
      "epoch": 25.34063789868668,
      "grad_norm": 0.6189551949501038,
      "learning_rate": 0.002828,
      "loss": 1.2292,
      "step": 337664
    },
    {
      "epoch": 25.345440900562853,
      "grad_norm": 0.5378521680831909,
      "learning_rate": 0.002828,
      "loss": 1.2247,
      "step": 337728
    },
    {
      "epoch": 25.350243902439026,
      "grad_norm": 0.5442854762077332,
      "learning_rate": 0.002828,
      "loss": 1.2267,
      "step": 337792
    },
    {
      "epoch": 25.3550469043152,
      "grad_norm": 0.5752109885215759,
      "learning_rate": 0.002828,
      "loss": 1.2325,
      "step": 337856
    },
    {
      "epoch": 25.35984990619137,
      "grad_norm": 0.5554222464561462,
      "learning_rate": 0.002828,
      "loss": 1.2236,
      "step": 337920
    },
    {
      "epoch": 25.36465290806754,
      "grad_norm": 0.675406813621521,
      "learning_rate": 0.002828,
      "loss": 1.2296,
      "step": 337984
    },
    {
      "epoch": 25.369455909943714,
      "grad_norm": 0.5926669239997864,
      "learning_rate": 0.002828,
      "loss": 1.221,
      "step": 338048
    },
    {
      "epoch": 25.374258911819886,
      "grad_norm": 0.4972551763057709,
      "learning_rate": 0.002828,
      "loss": 1.2233,
      "step": 338112
    },
    {
      "epoch": 25.37906191369606,
      "grad_norm": 0.6431121230125427,
      "learning_rate": 0.002828,
      "loss": 1.2251,
      "step": 338176
    },
    {
      "epoch": 25.383864915572232,
      "grad_norm": 0.5158210396766663,
      "learning_rate": 0.002828,
      "loss": 1.2228,
      "step": 338240
    },
    {
      "epoch": 25.388667917448405,
      "grad_norm": 0.5621346235275269,
      "learning_rate": 0.002828,
      "loss": 1.2302,
      "step": 338304
    },
    {
      "epoch": 25.393470919324578,
      "grad_norm": 0.6656622886657715,
      "learning_rate": 0.002828,
      "loss": 1.2243,
      "step": 338368
    },
    {
      "epoch": 25.39827392120075,
      "grad_norm": 0.5919707417488098,
      "learning_rate": 0.002828,
      "loss": 1.2254,
      "step": 338432
    },
    {
      "epoch": 25.403076923076924,
      "grad_norm": 0.5798816084861755,
      "learning_rate": 0.002828,
      "loss": 1.225,
      "step": 338496
    },
    {
      "epoch": 25.407879924953097,
      "grad_norm": 0.6159927248954773,
      "learning_rate": 0.002828,
      "loss": 1.2238,
      "step": 338560
    },
    {
      "epoch": 25.41268292682927,
      "grad_norm": 0.6071495413780212,
      "learning_rate": 0.002828,
      "loss": 1.2212,
      "step": 338624
    },
    {
      "epoch": 25.417485928705442,
      "grad_norm": 0.4696201682090759,
      "learning_rate": 0.002828,
      "loss": 1.2276,
      "step": 338688
    },
    {
      "epoch": 25.42228893058161,
      "grad_norm": 0.6017554402351379,
      "learning_rate": 0.002828,
      "loss": 1.2243,
      "step": 338752
    },
    {
      "epoch": 25.427091932457785,
      "grad_norm": 0.532129168510437,
      "learning_rate": 0.002828,
      "loss": 1.221,
      "step": 338816
    },
    {
      "epoch": 25.431894934333958,
      "grad_norm": 0.5939502120018005,
      "learning_rate": 0.002828,
      "loss": 1.2281,
      "step": 338880
    },
    {
      "epoch": 25.43669793621013,
      "grad_norm": 0.5088008046150208,
      "learning_rate": 0.002828,
      "loss": 1.2344,
      "step": 338944
    },
    {
      "epoch": 25.441500938086303,
      "grad_norm": 0.7171235680580139,
      "learning_rate": 0.002828,
      "loss": 1.2259,
      "step": 339008
    },
    {
      "epoch": 25.446303939962476,
      "grad_norm": 0.6164196133613586,
      "learning_rate": 0.002828,
      "loss": 1.2281,
      "step": 339072
    },
    {
      "epoch": 25.45110694183865,
      "grad_norm": 0.6111053228378296,
      "learning_rate": 0.002828,
      "loss": 1.2263,
      "step": 339136
    },
    {
      "epoch": 25.455909943714822,
      "grad_norm": 0.6192201375961304,
      "learning_rate": 0.002828,
      "loss": 1.23,
      "step": 339200
    },
    {
      "epoch": 25.460712945590995,
      "grad_norm": 0.7079026699066162,
      "learning_rate": 0.002828,
      "loss": 1.2239,
      "step": 339264
    },
    {
      "epoch": 25.465515947467168,
      "grad_norm": 0.630245566368103,
      "learning_rate": 0.002828,
      "loss": 1.2308,
      "step": 339328
    },
    {
      "epoch": 25.47031894934334,
      "grad_norm": 0.6225885152816772,
      "learning_rate": 0.002828,
      "loss": 1.2343,
      "step": 339392
    },
    {
      "epoch": 25.475121951219514,
      "grad_norm": 0.779857873916626,
      "learning_rate": 0.002828,
      "loss": 1.231,
      "step": 339456
    },
    {
      "epoch": 25.479924953095686,
      "grad_norm": 0.534664511680603,
      "learning_rate": 0.002828,
      "loss": 1.2294,
      "step": 339520
    },
    {
      "epoch": 25.484727954971856,
      "grad_norm": 0.6669053435325623,
      "learning_rate": 0.002828,
      "loss": 1.2285,
      "step": 339584
    },
    {
      "epoch": 25.48953095684803,
      "grad_norm": 0.6443493366241455,
      "learning_rate": 0.002828,
      "loss": 1.2205,
      "step": 339648
    },
    {
      "epoch": 25.4943339587242,
      "grad_norm": 0.5567485690116882,
      "learning_rate": 0.002828,
      "loss": 1.224,
      "step": 339712
    },
    {
      "epoch": 25.499136960600374,
      "grad_norm": 0.6056983470916748,
      "learning_rate": 0.002828,
      "loss": 1.2273,
      "step": 339776
    },
    {
      "epoch": 25.503939962476547,
      "grad_norm": 0.6651111245155334,
      "learning_rate": 0.002828,
      "loss": 1.2341,
      "step": 339840
    },
    {
      "epoch": 25.50874296435272,
      "grad_norm": 0.6043602824211121,
      "learning_rate": 0.002828,
      "loss": 1.2263,
      "step": 339904
    },
    {
      "epoch": 25.513545966228893,
      "grad_norm": 0.49276939034461975,
      "learning_rate": 0.002828,
      "loss": 1.2311,
      "step": 339968
    },
    {
      "epoch": 25.518348968105066,
      "grad_norm": 0.6172980070114136,
      "learning_rate": 0.002828,
      "loss": 1.2292,
      "step": 340032
    },
    {
      "epoch": 25.52315196998124,
      "grad_norm": 0.6005577445030212,
      "learning_rate": 0.002828,
      "loss": 1.2288,
      "step": 340096
    },
    {
      "epoch": 25.52795497185741,
      "grad_norm": 0.6682016253471375,
      "learning_rate": 0.002828,
      "loss": 1.2272,
      "step": 340160
    },
    {
      "epoch": 25.532757973733585,
      "grad_norm": 0.5602551698684692,
      "learning_rate": 0.002828,
      "loss": 1.2324,
      "step": 340224
    },
    {
      "epoch": 25.537560975609757,
      "grad_norm": 0.5643219351768494,
      "learning_rate": 0.002828,
      "loss": 1.2286,
      "step": 340288
    },
    {
      "epoch": 25.54236397748593,
      "grad_norm": 0.5434116721153259,
      "learning_rate": 0.002828,
      "loss": 1.2283,
      "step": 340352
    },
    {
      "epoch": 25.5471669793621,
      "grad_norm": 0.7124276161193848,
      "learning_rate": 0.002828,
      "loss": 1.2331,
      "step": 340416
    },
    {
      "epoch": 25.551969981238273,
      "grad_norm": 0.5400499105453491,
      "learning_rate": 0.002828,
      "loss": 1.2309,
      "step": 340480
    },
    {
      "epoch": 25.556772983114445,
      "grad_norm": 0.5698312520980835,
      "learning_rate": 0.002828,
      "loss": 1.2292,
      "step": 340544
    },
    {
      "epoch": 25.56157598499062,
      "grad_norm": 0.523460865020752,
      "learning_rate": 0.002828,
      "loss": 1.2223,
      "step": 340608
    },
    {
      "epoch": 25.56637898686679,
      "grad_norm": 0.5728746056556702,
      "learning_rate": 0.002828,
      "loss": 1.2371,
      "step": 340672
    },
    {
      "epoch": 25.571181988742964,
      "grad_norm": 0.553048849105835,
      "learning_rate": 0.002828,
      "loss": 1.2279,
      "step": 340736
    },
    {
      "epoch": 25.575984990619137,
      "grad_norm": 0.6705506443977356,
      "learning_rate": 0.002828,
      "loss": 1.2263,
      "step": 340800
    },
    {
      "epoch": 25.58078799249531,
      "grad_norm": 0.6559802293777466,
      "learning_rate": 0.002828,
      "loss": 1.2218,
      "step": 340864
    },
    {
      "epoch": 25.585590994371483,
      "grad_norm": 0.5637137293815613,
      "learning_rate": 0.002828,
      "loss": 1.2322,
      "step": 340928
    },
    {
      "epoch": 25.590393996247656,
      "grad_norm": 0.6659891605377197,
      "learning_rate": 0.002828,
      "loss": 1.2308,
      "step": 340992
    },
    {
      "epoch": 25.59519699812383,
      "grad_norm": 0.622877836227417,
      "learning_rate": 0.002828,
      "loss": 1.233,
      "step": 341056
    },
    {
      "epoch": 25.6,
      "grad_norm": 0.5427113771438599,
      "learning_rate": 0.002828,
      "loss": 1.2264,
      "step": 341120
    },
    {
      "epoch": 25.604803001876174,
      "grad_norm": 0.5061426758766174,
      "learning_rate": 0.002828,
      "loss": 1.2317,
      "step": 341184
    },
    {
      "epoch": 25.609606003752344,
      "grad_norm": 0.5866330862045288,
      "learning_rate": 0.002828,
      "loss": 1.2317,
      "step": 341248
    },
    {
      "epoch": 25.614409005628517,
      "grad_norm": 0.5532949566841125,
      "learning_rate": 0.002828,
      "loss": 1.2328,
      "step": 341312
    },
    {
      "epoch": 25.61921200750469,
      "grad_norm": 0.6040782332420349,
      "learning_rate": 0.002828,
      "loss": 1.2357,
      "step": 341376
    },
    {
      "epoch": 25.624015009380862,
      "grad_norm": 0.6957950592041016,
      "learning_rate": 0.002828,
      "loss": 1.2309,
      "step": 341440
    },
    {
      "epoch": 25.628818011257035,
      "grad_norm": 0.6486824750900269,
      "learning_rate": 0.002828,
      "loss": 1.2275,
      "step": 341504
    },
    {
      "epoch": 25.633621013133208,
      "grad_norm": 0.6482639908790588,
      "learning_rate": 0.002828,
      "loss": 1.2311,
      "step": 341568
    },
    {
      "epoch": 25.63842401500938,
      "grad_norm": 0.5687380433082581,
      "learning_rate": 0.002828,
      "loss": 1.2278,
      "step": 341632
    },
    {
      "epoch": 25.643227016885554,
      "grad_norm": 0.511089026927948,
      "learning_rate": 0.002828,
      "loss": 1.2295,
      "step": 341696
    },
    {
      "epoch": 25.648030018761727,
      "grad_norm": 0.6156114935874939,
      "learning_rate": 0.002828,
      "loss": 1.2332,
      "step": 341760
    },
    {
      "epoch": 25.6528330206379,
      "grad_norm": 0.5956178903579712,
      "learning_rate": 0.002828,
      "loss": 1.2286,
      "step": 341824
    },
    {
      "epoch": 25.657636022514072,
      "grad_norm": 0.5628892183303833,
      "learning_rate": 0.002828,
      "loss": 1.2298,
      "step": 341888
    },
    {
      "epoch": 25.662439024390245,
      "grad_norm": 0.6388689279556274,
      "learning_rate": 0.002828,
      "loss": 1.2276,
      "step": 341952
    },
    {
      "epoch": 25.667242026266415,
      "grad_norm": 0.577284038066864,
      "learning_rate": 0.002828,
      "loss": 1.2365,
      "step": 342016
    },
    {
      "epoch": 25.672045028142588,
      "grad_norm": 0.6047957539558411,
      "learning_rate": 0.002828,
      "loss": 1.232,
      "step": 342080
    },
    {
      "epoch": 25.67684803001876,
      "grad_norm": 0.6236744523048401,
      "learning_rate": 0.002828,
      "loss": 1.2375,
      "step": 342144
    },
    {
      "epoch": 25.681651031894933,
      "grad_norm": 0.6257603764533997,
      "learning_rate": 0.002828,
      "loss": 1.2312,
      "step": 342208
    },
    {
      "epoch": 25.686454033771106,
      "grad_norm": 0.5153629183769226,
      "learning_rate": 0.002828,
      "loss": 1.231,
      "step": 342272
    },
    {
      "epoch": 25.69125703564728,
      "grad_norm": 0.6125085949897766,
      "learning_rate": 0.002828,
      "loss": 1.2329,
      "step": 342336
    },
    {
      "epoch": 25.696060037523452,
      "grad_norm": 0.6297896504402161,
      "learning_rate": 0.002828,
      "loss": 1.2385,
      "step": 342400
    },
    {
      "epoch": 25.700863039399625,
      "grad_norm": 0.6260985732078552,
      "learning_rate": 0.002828,
      "loss": 1.2366,
      "step": 342464
    },
    {
      "epoch": 25.705666041275798,
      "grad_norm": 0.5179916620254517,
      "learning_rate": 0.002828,
      "loss": 1.2289,
      "step": 342528
    },
    {
      "epoch": 25.71046904315197,
      "grad_norm": 0.63902747631073,
      "learning_rate": 0.002828,
      "loss": 1.2319,
      "step": 342592
    },
    {
      "epoch": 25.715272045028144,
      "grad_norm": 0.5296642184257507,
      "learning_rate": 0.002828,
      "loss": 1.2257,
      "step": 342656
    },
    {
      "epoch": 25.720075046904316,
      "grad_norm": 0.5687287449836731,
      "learning_rate": 0.002828,
      "loss": 1.2313,
      "step": 342720
    },
    {
      "epoch": 25.72487804878049,
      "grad_norm": 0.5681202411651611,
      "learning_rate": 0.002828,
      "loss": 1.2356,
      "step": 342784
    },
    {
      "epoch": 25.72968105065666,
      "grad_norm": 0.6682190895080566,
      "learning_rate": 0.002828,
      "loss": 1.2351,
      "step": 342848
    },
    {
      "epoch": 25.73448405253283,
      "grad_norm": 0.6136249303817749,
      "learning_rate": 0.002828,
      "loss": 1.2342,
      "step": 342912
    },
    {
      "epoch": 25.739287054409004,
      "grad_norm": 0.6873012781143188,
      "learning_rate": 0.002828,
      "loss": 1.2331,
      "step": 342976
    },
    {
      "epoch": 25.744090056285177,
      "grad_norm": 0.5340631008148193,
      "learning_rate": 0.002828,
      "loss": 1.2297,
      "step": 343040
    },
    {
      "epoch": 25.74889305816135,
      "grad_norm": 0.698490560054779,
      "learning_rate": 0.002828,
      "loss": 1.2368,
      "step": 343104
    },
    {
      "epoch": 25.753696060037523,
      "grad_norm": 0.6508604884147644,
      "learning_rate": 0.002828,
      "loss": 1.2317,
      "step": 343168
    },
    {
      "epoch": 25.758499061913696,
      "grad_norm": 0.6807830333709717,
      "learning_rate": 0.002828,
      "loss": 1.2306,
      "step": 343232
    },
    {
      "epoch": 25.76330206378987,
      "grad_norm": 0.6143683195114136,
      "learning_rate": 0.002828,
      "loss": 1.2398,
      "step": 343296
    },
    {
      "epoch": 25.76810506566604,
      "grad_norm": 0.7183651924133301,
      "learning_rate": 0.002828,
      "loss": 1.2347,
      "step": 343360
    },
    {
      "epoch": 25.772908067542215,
      "grad_norm": 0.5654355883598328,
      "learning_rate": 0.002828,
      "loss": 1.2409,
      "step": 343424
    },
    {
      "epoch": 25.777711069418388,
      "grad_norm": 0.5893970727920532,
      "learning_rate": 0.002828,
      "loss": 1.2288,
      "step": 343488
    },
    {
      "epoch": 25.78251407129456,
      "grad_norm": 0.688660204410553,
      "learning_rate": 0.002828,
      "loss": 1.2375,
      "step": 343552
    },
    {
      "epoch": 25.787317073170733,
      "grad_norm": 0.5644953846931458,
      "learning_rate": 0.002828,
      "loss": 1.2311,
      "step": 343616
    },
    {
      "epoch": 25.792120075046903,
      "grad_norm": 0.6274914145469666,
      "learning_rate": 0.002828,
      "loss": 1.2347,
      "step": 343680
    },
    {
      "epoch": 25.796923076923076,
      "grad_norm": 0.48792901635169983,
      "learning_rate": 0.002828,
      "loss": 1.2397,
      "step": 343744
    },
    {
      "epoch": 25.80172607879925,
      "grad_norm": 0.6912416219711304,
      "learning_rate": 0.002828,
      "loss": 1.2295,
      "step": 343808
    },
    {
      "epoch": 25.80652908067542,
      "grad_norm": 0.6125044226646423,
      "learning_rate": 0.002828,
      "loss": 1.2404,
      "step": 343872
    },
    {
      "epoch": 25.811332082551594,
      "grad_norm": 0.5817704200744629,
      "learning_rate": 0.002828,
      "loss": 1.2366,
      "step": 343936
    },
    {
      "epoch": 25.816135084427767,
      "grad_norm": 0.653104841709137,
      "learning_rate": 0.002828,
      "loss": 1.2352,
      "step": 344000
    },
    {
      "epoch": 25.82093808630394,
      "grad_norm": 0.646251380443573,
      "learning_rate": 0.002828,
      "loss": 1.2314,
      "step": 344064
    },
    {
      "epoch": 25.825741088180113,
      "grad_norm": 0.5538818836212158,
      "learning_rate": 0.002828,
      "loss": 1.2318,
      "step": 344128
    },
    {
      "epoch": 25.830544090056286,
      "grad_norm": 0.6054178476333618,
      "learning_rate": 0.002828,
      "loss": 1.2324,
      "step": 344192
    },
    {
      "epoch": 25.83534709193246,
      "grad_norm": 0.4697195887565613,
      "learning_rate": 0.002828,
      "loss": 1.2404,
      "step": 344256
    },
    {
      "epoch": 25.84015009380863,
      "grad_norm": 0.6623765230178833,
      "learning_rate": 0.002828,
      "loss": 1.2313,
      "step": 344320
    },
    {
      "epoch": 25.844953095684804,
      "grad_norm": 0.5547646284103394,
      "learning_rate": 0.002828,
      "loss": 1.2306,
      "step": 344384
    },
    {
      "epoch": 25.849756097560977,
      "grad_norm": 0.6155876517295837,
      "learning_rate": 0.002828,
      "loss": 1.2322,
      "step": 344448
    },
    {
      "epoch": 25.854559099437147,
      "grad_norm": 0.6721112132072449,
      "learning_rate": 0.002828,
      "loss": 1.2421,
      "step": 344512
    },
    {
      "epoch": 25.85936210131332,
      "grad_norm": 0.7572811245918274,
      "learning_rate": 0.002828,
      "loss": 1.2304,
      "step": 344576
    },
    {
      "epoch": 25.864165103189492,
      "grad_norm": 0.48257964849472046,
      "learning_rate": 0.002828,
      "loss": 1.2321,
      "step": 344640
    },
    {
      "epoch": 25.868968105065665,
      "grad_norm": 0.5753004550933838,
      "learning_rate": 0.002828,
      "loss": 1.2361,
      "step": 344704
    },
    {
      "epoch": 25.873771106941838,
      "grad_norm": 0.6127859950065613,
      "learning_rate": 0.002828,
      "loss": 1.2339,
      "step": 344768
    },
    {
      "epoch": 25.87857410881801,
      "grad_norm": 0.6566331386566162,
      "learning_rate": 0.002828,
      "loss": 1.238,
      "step": 344832
    },
    {
      "epoch": 25.883377110694184,
      "grad_norm": 0.49689486622810364,
      "learning_rate": 0.002828,
      "loss": 1.2299,
      "step": 344896
    },
    {
      "epoch": 25.888180112570357,
      "grad_norm": 0.9033483266830444,
      "learning_rate": 0.002828,
      "loss": 1.2346,
      "step": 344960
    },
    {
      "epoch": 25.89298311444653,
      "grad_norm": 0.5268735885620117,
      "learning_rate": 0.002828,
      "loss": 1.2379,
      "step": 345024
    },
    {
      "epoch": 25.897786116322703,
      "grad_norm": 0.5659050941467285,
      "learning_rate": 0.002828,
      "loss": 1.239,
      "step": 345088
    },
    {
      "epoch": 25.902589118198875,
      "grad_norm": 0.700505793094635,
      "learning_rate": 0.002828,
      "loss": 1.2352,
      "step": 345152
    },
    {
      "epoch": 25.90739212007505,
      "grad_norm": 0.5935922265052795,
      "learning_rate": 0.002828,
      "loss": 1.2401,
      "step": 345216
    },
    {
      "epoch": 25.91219512195122,
      "grad_norm": 0.6356076002120972,
      "learning_rate": 0.002828,
      "loss": 1.2369,
      "step": 345280
    },
    {
      "epoch": 25.91699812382739,
      "grad_norm": 0.7625199556350708,
      "learning_rate": 0.002828,
      "loss": 1.234,
      "step": 345344
    },
    {
      "epoch": 25.921801125703563,
      "grad_norm": 0.5876018404960632,
      "learning_rate": 0.002828,
      "loss": 1.2397,
      "step": 345408
    },
    {
      "epoch": 25.926604127579736,
      "grad_norm": 0.5337401628494263,
      "learning_rate": 0.002828,
      "loss": 1.235,
      "step": 345472
    },
    {
      "epoch": 25.93140712945591,
      "grad_norm": 0.597531259059906,
      "learning_rate": 0.002828,
      "loss": 1.2314,
      "step": 345536
    },
    {
      "epoch": 25.936210131332082,
      "grad_norm": 0.5823056101799011,
      "learning_rate": 0.002828,
      "loss": 1.2322,
      "step": 345600
    },
    {
      "epoch": 25.941013133208255,
      "grad_norm": 0.5407066345214844,
      "learning_rate": 0.002828,
      "loss": 1.2338,
      "step": 345664
    },
    {
      "epoch": 25.945816135084428,
      "grad_norm": 0.687560498714447,
      "learning_rate": 0.002828,
      "loss": 1.2347,
      "step": 345728
    },
    {
      "epoch": 25.9506191369606,
      "grad_norm": 0.5314828753471375,
      "learning_rate": 0.002828,
      "loss": 1.2438,
      "step": 345792
    },
    {
      "epoch": 25.955422138836774,
      "grad_norm": 0.7035368084907532,
      "learning_rate": 0.002828,
      "loss": 1.238,
      "step": 345856
    },
    {
      "epoch": 25.960225140712947,
      "grad_norm": 0.5470126867294312,
      "learning_rate": 0.002828,
      "loss": 1.2313,
      "step": 345920
    },
    {
      "epoch": 25.96502814258912,
      "grad_norm": 0.5416370630264282,
      "learning_rate": 0.002828,
      "loss": 1.2382,
      "step": 345984
    },
    {
      "epoch": 25.969831144465292,
      "grad_norm": 0.6480735540390015,
      "learning_rate": 0.002828,
      "loss": 1.2387,
      "step": 346048
    },
    {
      "epoch": 25.974634146341465,
      "grad_norm": 0.5238390564918518,
      "learning_rate": 0.002828,
      "loss": 1.2367,
      "step": 346112
    },
    {
      "epoch": 25.979437148217635,
      "grad_norm": 0.5900247693061829,
      "learning_rate": 0.002828,
      "loss": 1.2376,
      "step": 346176
    },
    {
      "epoch": 25.984240150093807,
      "grad_norm": 0.6199198365211487,
      "learning_rate": 0.002828,
      "loss": 1.2373,
      "step": 346240
    },
    {
      "epoch": 25.98904315196998,
      "grad_norm": 0.6323431730270386,
      "learning_rate": 0.002828,
      "loss": 1.2355,
      "step": 346304
    },
    {
      "epoch": 25.993846153846153,
      "grad_norm": 0.5909257531166077,
      "learning_rate": 0.002828,
      "loss": 1.2339,
      "step": 346368
    },
    {
      "epoch": 25.998649155722326,
      "grad_norm": 0.5273489356040955,
      "learning_rate": 0.002828,
      "loss": 1.2345,
      "step": 346432
    },
    {
      "epoch": 26.0034521575985,
      "grad_norm": 0.5202985405921936,
      "learning_rate": 0.002828,
      "loss": 1.2114,
      "step": 346496
    },
    {
      "epoch": 26.008255159474672,
      "grad_norm": 0.5007950663566589,
      "learning_rate": 0.002828,
      "loss": 1.1978,
      "step": 346560
    },
    {
      "epoch": 26.013058161350845,
      "grad_norm": 0.5772574543952942,
      "learning_rate": 0.002828,
      "loss": 1.1976,
      "step": 346624
    },
    {
      "epoch": 26.017861163227018,
      "grad_norm": 0.5691853165626526,
      "learning_rate": 0.002828,
      "loss": 1.1911,
      "step": 346688
    },
    {
      "epoch": 26.02266416510319,
      "grad_norm": 0.5564082264900208,
      "learning_rate": 0.002828,
      "loss": 1.1955,
      "step": 346752
    },
    {
      "epoch": 26.027467166979363,
      "grad_norm": 0.6861823797225952,
      "learning_rate": 0.002828,
      "loss": 1.2017,
      "step": 346816
    },
    {
      "epoch": 26.032270168855536,
      "grad_norm": 0.6327756643295288,
      "learning_rate": 0.002828,
      "loss": 1.202,
      "step": 346880
    },
    {
      "epoch": 26.037073170731706,
      "grad_norm": 0.5592343211174011,
      "learning_rate": 0.002828,
      "loss": 1.2033,
      "step": 346944
    },
    {
      "epoch": 26.04187617260788,
      "grad_norm": 0.6427003145217896,
      "learning_rate": 0.002828,
      "loss": 1.1967,
      "step": 347008
    },
    {
      "epoch": 26.04667917448405,
      "grad_norm": 0.550666093826294,
      "learning_rate": 0.002828,
      "loss": 1.2017,
      "step": 347072
    },
    {
      "epoch": 26.051482176360224,
      "grad_norm": 0.5372848510742188,
      "learning_rate": 0.002828,
      "loss": 1.2099,
      "step": 347136
    },
    {
      "epoch": 26.056285178236397,
      "grad_norm": 0.6634334325790405,
      "learning_rate": 0.002828,
      "loss": 1.2021,
      "step": 347200
    },
    {
      "epoch": 26.06108818011257,
      "grad_norm": 0.6219868659973145,
      "learning_rate": 0.002828,
      "loss": 1.2004,
      "step": 347264
    },
    {
      "epoch": 26.065891181988743,
      "grad_norm": 0.5100303888320923,
      "learning_rate": 0.002828,
      "loss": 1.1993,
      "step": 347328
    },
    {
      "epoch": 26.070694183864916,
      "grad_norm": 0.6207034587860107,
      "learning_rate": 0.002828,
      "loss": 1.1937,
      "step": 347392
    },
    {
      "epoch": 26.07549718574109,
      "grad_norm": 0.6422325968742371,
      "learning_rate": 0.002828,
      "loss": 1.2051,
      "step": 347456
    },
    {
      "epoch": 26.08030018761726,
      "grad_norm": 0.5773105025291443,
      "learning_rate": 0.002828,
      "loss": 1.2094,
      "step": 347520
    },
    {
      "epoch": 26.085103189493434,
      "grad_norm": 0.5325859785079956,
      "learning_rate": 0.002828,
      "loss": 1.2028,
      "step": 347584
    },
    {
      "epoch": 26.089906191369607,
      "grad_norm": 0.6021090149879456,
      "learning_rate": 0.002828,
      "loss": 1.2003,
      "step": 347648
    },
    {
      "epoch": 26.09470919324578,
      "grad_norm": 0.6171033978462219,
      "learning_rate": 0.002828,
      "loss": 1.2009,
      "step": 347712
    },
    {
      "epoch": 26.09951219512195,
      "grad_norm": 0.5657275915145874,
      "learning_rate": 0.002828,
      "loss": 1.2007,
      "step": 347776
    },
    {
      "epoch": 26.104315196998122,
      "grad_norm": 0.6622403860092163,
      "learning_rate": 0.002828,
      "loss": 1.2071,
      "step": 347840
    },
    {
      "epoch": 26.109118198874295,
      "grad_norm": 0.6320151090621948,
      "learning_rate": 0.002828,
      "loss": 1.2067,
      "step": 347904
    },
    {
      "epoch": 26.113921200750468,
      "grad_norm": 0.5358821749687195,
      "learning_rate": 0.002828,
      "loss": 1.2033,
      "step": 347968
    },
    {
      "epoch": 26.11872420262664,
      "grad_norm": 0.6047529578208923,
      "learning_rate": 0.002828,
      "loss": 1.2053,
      "step": 348032
    },
    {
      "epoch": 26.123527204502814,
      "grad_norm": 0.5269867777824402,
      "learning_rate": 0.002828,
      "loss": 1.204,
      "step": 348096
    },
    {
      "epoch": 26.128330206378987,
      "grad_norm": 0.5756949782371521,
      "learning_rate": 0.002828,
      "loss": 1.2025,
      "step": 348160
    },
    {
      "epoch": 26.13313320825516,
      "grad_norm": 0.5575070977210999,
      "learning_rate": 0.002828,
      "loss": 1.209,
      "step": 348224
    },
    {
      "epoch": 26.137936210131333,
      "grad_norm": 0.6042987704277039,
      "learning_rate": 0.002828,
      "loss": 1.2052,
      "step": 348288
    },
    {
      "epoch": 26.142739212007506,
      "grad_norm": 0.6796466708183289,
      "learning_rate": 0.002828,
      "loss": 1.2014,
      "step": 348352
    },
    {
      "epoch": 26.14754221388368,
      "grad_norm": 0.6714834570884705,
      "learning_rate": 0.002828,
      "loss": 1.2021,
      "step": 348416
    },
    {
      "epoch": 26.15234521575985,
      "grad_norm": 0.6910606026649475,
      "learning_rate": 0.002828,
      "loss": 1.2045,
      "step": 348480
    },
    {
      "epoch": 26.157148217636024,
      "grad_norm": 0.515529990196228,
      "learning_rate": 0.002828,
      "loss": 1.2079,
      "step": 348544
    },
    {
      "epoch": 26.161951219512193,
      "grad_norm": 0.6046344041824341,
      "learning_rate": 0.002828,
      "loss": 1.2038,
      "step": 348608
    },
    {
      "epoch": 26.166754221388366,
      "grad_norm": 0.5959147810935974,
      "learning_rate": 0.002828,
      "loss": 1.2098,
      "step": 348672
    },
    {
      "epoch": 26.17155722326454,
      "grad_norm": 0.5565488934516907,
      "learning_rate": 0.002828,
      "loss": 1.2106,
      "step": 348736
    },
    {
      "epoch": 26.176360225140712,
      "grad_norm": 0.5920954346656799,
      "learning_rate": 0.002828,
      "loss": 1.2069,
      "step": 348800
    },
    {
      "epoch": 26.181163227016885,
      "grad_norm": 0.5918201208114624,
      "learning_rate": 0.002828,
      "loss": 1.2023,
      "step": 348864
    },
    {
      "epoch": 26.185966228893058,
      "grad_norm": 0.579826831817627,
      "learning_rate": 0.002828,
      "loss": 1.2045,
      "step": 348928
    },
    {
      "epoch": 26.19076923076923,
      "grad_norm": 0.6079787611961365,
      "learning_rate": 0.002828,
      "loss": 1.2042,
      "step": 348992
    },
    {
      "epoch": 26.195572232645404,
      "grad_norm": 0.5046894550323486,
      "learning_rate": 0.002828,
      "loss": 1.2053,
      "step": 349056
    },
    {
      "epoch": 26.200375234521577,
      "grad_norm": 0.6525089740753174,
      "learning_rate": 0.002828,
      "loss": 1.207,
      "step": 349120
    },
    {
      "epoch": 26.20517823639775,
      "grad_norm": 0.5801729559898376,
      "learning_rate": 0.002828,
      "loss": 1.2055,
      "step": 349184
    },
    {
      "epoch": 26.209981238273922,
      "grad_norm": 0.625745415687561,
      "learning_rate": 0.002828,
      "loss": 1.2069,
      "step": 349248
    },
    {
      "epoch": 26.214784240150095,
      "grad_norm": 0.6147674918174744,
      "learning_rate": 0.002828,
      "loss": 1.2063,
      "step": 349312
    },
    {
      "epoch": 26.219587242026268,
      "grad_norm": 0.5819293856620789,
      "learning_rate": 0.002828,
      "loss": 1.211,
      "step": 349376
    },
    {
      "epoch": 26.224390243902437,
      "grad_norm": 0.5987066626548767,
      "learning_rate": 0.002828,
      "loss": 1.2159,
      "step": 349440
    },
    {
      "epoch": 26.22919324577861,
      "grad_norm": 0.5650080442428589,
      "learning_rate": 0.002828,
      "loss": 1.2152,
      "step": 349504
    },
    {
      "epoch": 26.233996247654783,
      "grad_norm": 0.5981659889221191,
      "learning_rate": 0.002828,
      "loss": 1.2103,
      "step": 349568
    },
    {
      "epoch": 26.238799249530956,
      "grad_norm": 0.5904810428619385,
      "learning_rate": 0.002828,
      "loss": 1.2043,
      "step": 349632
    },
    {
      "epoch": 26.24360225140713,
      "grad_norm": 0.5519508123397827,
      "learning_rate": 0.002828,
      "loss": 1.2107,
      "step": 349696
    },
    {
      "epoch": 26.248405253283302,
      "grad_norm": 0.5517117381095886,
      "learning_rate": 0.002828,
      "loss": 1.2064,
      "step": 349760
    },
    {
      "epoch": 26.253208255159475,
      "grad_norm": 0.5335156321525574,
      "learning_rate": 0.002828,
      "loss": 1.2143,
      "step": 349824
    },
    {
      "epoch": 26.258011257035648,
      "grad_norm": 0.5944948792457581,
      "learning_rate": 0.002828,
      "loss": 1.2093,
      "step": 349888
    },
    {
      "epoch": 26.26281425891182,
      "grad_norm": 0.5173745155334473,
      "learning_rate": 0.002828,
      "loss": 1.2091,
      "step": 349952
    },
    {
      "epoch": 26.267617260787993,
      "grad_norm": 0.5639364123344421,
      "learning_rate": 0.002828,
      "loss": 1.2113,
      "step": 350016
    },
    {
      "epoch": 26.272420262664166,
      "grad_norm": 0.508617103099823,
      "learning_rate": 0.002828,
      "loss": 1.2089,
      "step": 350080
    },
    {
      "epoch": 26.27722326454034,
      "grad_norm": 0.4807698428630829,
      "learning_rate": 0.002828,
      "loss": 1.2057,
      "step": 350144
    },
    {
      "epoch": 26.28202626641651,
      "grad_norm": 0.5463166236877441,
      "learning_rate": 0.002828,
      "loss": 1.2077,
      "step": 350208
    },
    {
      "epoch": 26.28682926829268,
      "grad_norm": 0.6454668641090393,
      "learning_rate": 0.002828,
      "loss": 1.2096,
      "step": 350272
    },
    {
      "epoch": 26.291632270168854,
      "grad_norm": 0.6266613602638245,
      "learning_rate": 0.002828,
      "loss": 1.2051,
      "step": 350336
    },
    {
      "epoch": 26.296435272045027,
      "grad_norm": 0.6026892066001892,
      "learning_rate": 0.002828,
      "loss": 1.2124,
      "step": 350400
    },
    {
      "epoch": 26.3012382739212,
      "grad_norm": 0.6331421732902527,
      "learning_rate": 0.002828,
      "loss": 1.2122,
      "step": 350464
    },
    {
      "epoch": 26.306041275797373,
      "grad_norm": 0.5798826217651367,
      "learning_rate": 0.002828,
      "loss": 1.2106,
      "step": 350528
    },
    {
      "epoch": 26.310844277673546,
      "grad_norm": 0.527355432510376,
      "learning_rate": 0.002828,
      "loss": 1.2091,
      "step": 350592
    },
    {
      "epoch": 26.31564727954972,
      "grad_norm": 0.5960228443145752,
      "learning_rate": 0.002828,
      "loss": 1.2136,
      "step": 350656
    },
    {
      "epoch": 26.32045028142589,
      "grad_norm": 0.667033851146698,
      "learning_rate": 0.002828,
      "loss": 1.2088,
      "step": 350720
    },
    {
      "epoch": 26.325253283302064,
      "grad_norm": 0.6654618978500366,
      "learning_rate": 0.002828,
      "loss": 1.2161,
      "step": 350784
    },
    {
      "epoch": 26.330056285178237,
      "grad_norm": 0.5624558925628662,
      "learning_rate": 0.002828,
      "loss": 1.2076,
      "step": 350848
    },
    {
      "epoch": 26.33485928705441,
      "grad_norm": 0.534230649471283,
      "learning_rate": 0.002828,
      "loss": 1.2166,
      "step": 350912
    },
    {
      "epoch": 26.339662288930583,
      "grad_norm": 0.6490930914878845,
      "learning_rate": 0.002828,
      "loss": 1.2119,
      "step": 350976
    },
    {
      "epoch": 26.344465290806752,
      "grad_norm": 0.70958411693573,
      "learning_rate": 0.002828,
      "loss": 1.2099,
      "step": 351040
    },
    {
      "epoch": 26.349268292682925,
      "grad_norm": 0.6444900035858154,
      "learning_rate": 0.002828,
      "loss": 1.2166,
      "step": 351104
    },
    {
      "epoch": 26.3540712945591,
      "grad_norm": 0.5082390904426575,
      "learning_rate": 0.002828,
      "loss": 1.2157,
      "step": 351168
    },
    {
      "epoch": 26.35887429643527,
      "grad_norm": 0.46947455406188965,
      "learning_rate": 0.002828,
      "loss": 1.2133,
      "step": 351232
    },
    {
      "epoch": 26.363677298311444,
      "grad_norm": 0.58272385597229,
      "learning_rate": 0.002828,
      "loss": 1.2162,
      "step": 351296
    },
    {
      "epoch": 26.368480300187617,
      "grad_norm": 0.6278541684150696,
      "learning_rate": 0.002828,
      "loss": 1.2255,
      "step": 351360
    },
    {
      "epoch": 26.37328330206379,
      "grad_norm": 0.7564519047737122,
      "learning_rate": 0.002828,
      "loss": 1.2239,
      "step": 351424
    },
    {
      "epoch": 26.378086303939963,
      "grad_norm": 0.6092026829719543,
      "learning_rate": 0.002828,
      "loss": 1.2554,
      "step": 351488
    },
    {
      "epoch": 26.382889305816136,
      "grad_norm": 0.5624083280563354,
      "learning_rate": 0.002828,
      "loss": 1.2196,
      "step": 351552
    },
    {
      "epoch": 26.38769230769231,
      "grad_norm": 0.6422922611236572,
      "learning_rate": 0.002828,
      "loss": 1.2153,
      "step": 351616
    },
    {
      "epoch": 26.39249530956848,
      "grad_norm": 0.7231433391571045,
      "learning_rate": 0.002828,
      "loss": 1.2128,
      "step": 351680
    },
    {
      "epoch": 26.397298311444654,
      "grad_norm": 0.6518374681472778,
      "learning_rate": 0.002828,
      "loss": 1.213,
      "step": 351744
    },
    {
      "epoch": 26.402101313320827,
      "grad_norm": 0.5458804965019226,
      "learning_rate": 0.002828,
      "loss": 1.2119,
      "step": 351808
    },
    {
      "epoch": 26.406904315196996,
      "grad_norm": 0.5119909048080444,
      "learning_rate": 0.002828,
      "loss": 1.2137,
      "step": 351872
    },
    {
      "epoch": 26.41170731707317,
      "grad_norm": 0.5168522000312805,
      "learning_rate": 0.002828,
      "loss": 1.2135,
      "step": 351936
    },
    {
      "epoch": 26.416510318949342,
      "grad_norm": 0.6378693580627441,
      "learning_rate": 0.002828,
      "loss": 1.2189,
      "step": 352000
    },
    {
      "epoch": 26.421313320825515,
      "grad_norm": 0.6508474946022034,
      "learning_rate": 0.002828,
      "loss": 1.2135,
      "step": 352064
    },
    {
      "epoch": 26.426116322701688,
      "grad_norm": 0.7472125291824341,
      "learning_rate": 0.002828,
      "loss": 1.222,
      "step": 352128
    },
    {
      "epoch": 26.43091932457786,
      "grad_norm": 0.644286572933197,
      "learning_rate": 0.002828,
      "loss": 1.2105,
      "step": 352192
    },
    {
      "epoch": 26.435722326454034,
      "grad_norm": 0.5401178598403931,
      "learning_rate": 0.002828,
      "loss": 1.2194,
      "step": 352256
    },
    {
      "epoch": 26.440525328330207,
      "grad_norm": 0.6656849384307861,
      "learning_rate": 0.002828,
      "loss": 1.2116,
      "step": 352320
    },
    {
      "epoch": 26.44532833020638,
      "grad_norm": 0.8256285190582275,
      "learning_rate": 0.002828,
      "loss": 1.2161,
      "step": 352384
    },
    {
      "epoch": 26.450131332082552,
      "grad_norm": 0.6020409464836121,
      "learning_rate": 0.002828,
      "loss": 1.2163,
      "step": 352448
    },
    {
      "epoch": 26.454934333958725,
      "grad_norm": 0.550679624080658,
      "learning_rate": 0.002828,
      "loss": 1.2161,
      "step": 352512
    },
    {
      "epoch": 26.459737335834898,
      "grad_norm": 0.6722231507301331,
      "learning_rate": 0.002828,
      "loss": 1.2166,
      "step": 352576
    },
    {
      "epoch": 26.46454033771107,
      "grad_norm": 0.6431364417076111,
      "learning_rate": 0.002828,
      "loss": 1.2214,
      "step": 352640
    },
    {
      "epoch": 26.46934333958724,
      "grad_norm": 0.5253560543060303,
      "learning_rate": 0.002828,
      "loss": 1.219,
      "step": 352704
    },
    {
      "epoch": 26.474146341463413,
      "grad_norm": 0.6368989944458008,
      "learning_rate": 0.002828,
      "loss": 1.2195,
      "step": 352768
    },
    {
      "epoch": 26.478949343339586,
      "grad_norm": 0.5993379950523376,
      "learning_rate": 0.002828,
      "loss": 1.2115,
      "step": 352832
    },
    {
      "epoch": 26.48375234521576,
      "grad_norm": 0.493409126996994,
      "learning_rate": 0.002828,
      "loss": 1.2249,
      "step": 352896
    },
    {
      "epoch": 26.488555347091932,
      "grad_norm": 0.6184572577476501,
      "learning_rate": 0.002828,
      "loss": 1.218,
      "step": 352960
    },
    {
      "epoch": 26.493358348968105,
      "grad_norm": 0.592602014541626,
      "learning_rate": 0.002828,
      "loss": 1.2259,
      "step": 353024
    },
    {
      "epoch": 26.498161350844278,
      "grad_norm": 0.6765268445014954,
      "learning_rate": 0.002828,
      "loss": 1.2167,
      "step": 353088
    },
    {
      "epoch": 26.50296435272045,
      "grad_norm": 0.651823103427887,
      "learning_rate": 0.002828,
      "loss": 1.219,
      "step": 353152
    },
    {
      "epoch": 26.507767354596623,
      "grad_norm": 0.5332434177398682,
      "learning_rate": 0.002828,
      "loss": 1.2175,
      "step": 353216
    },
    {
      "epoch": 26.512570356472796,
      "grad_norm": 0.5774806141853333,
      "learning_rate": 0.002828,
      "loss": 1.2194,
      "step": 353280
    },
    {
      "epoch": 26.51737335834897,
      "grad_norm": 0.5372803807258606,
      "learning_rate": 0.002828,
      "loss": 1.2159,
      "step": 353344
    },
    {
      "epoch": 26.522176360225142,
      "grad_norm": 0.6259738206863403,
      "learning_rate": 0.002828,
      "loss": 1.2207,
      "step": 353408
    },
    {
      "epoch": 26.526979362101315,
      "grad_norm": 0.5487435460090637,
      "learning_rate": 0.002828,
      "loss": 1.2166,
      "step": 353472
    },
    {
      "epoch": 26.531782363977484,
      "grad_norm": 0.6260308027267456,
      "learning_rate": 0.002828,
      "loss": 1.2161,
      "step": 353536
    },
    {
      "epoch": 26.536585365853657,
      "grad_norm": 0.581946611404419,
      "learning_rate": 0.002828,
      "loss": 1.2206,
      "step": 353600
    },
    {
      "epoch": 26.54138836772983,
      "grad_norm": 0.6206014752388,
      "learning_rate": 0.002828,
      "loss": 1.2178,
      "step": 353664
    },
    {
      "epoch": 26.546191369606003,
      "grad_norm": 0.624851644039154,
      "learning_rate": 0.002828,
      "loss": 1.2206,
      "step": 353728
    },
    {
      "epoch": 26.550994371482176,
      "grad_norm": 0.6026180982589722,
      "learning_rate": 0.002828,
      "loss": 1.2178,
      "step": 353792
    },
    {
      "epoch": 26.55579737335835,
      "grad_norm": 0.6708505153656006,
      "learning_rate": 0.002828,
      "loss": 1.2139,
      "step": 353856
    },
    {
      "epoch": 26.56060037523452,
      "grad_norm": 0.6612686514854431,
      "learning_rate": 0.002828,
      "loss": 1.2192,
      "step": 353920
    },
    {
      "epoch": 26.565403377110695,
      "grad_norm": 0.6385770440101624,
      "learning_rate": 0.002828,
      "loss": 1.2189,
      "step": 353984
    },
    {
      "epoch": 26.570206378986867,
      "grad_norm": 0.7035972476005554,
      "learning_rate": 0.002828,
      "loss": 1.2212,
      "step": 354048
    },
    {
      "epoch": 26.57500938086304,
      "grad_norm": 0.4659041166305542,
      "learning_rate": 0.002828,
      "loss": 1.2139,
      "step": 354112
    },
    {
      "epoch": 26.579812382739213,
      "grad_norm": 0.4893842339515686,
      "learning_rate": 0.002828,
      "loss": 1.219,
      "step": 354176
    },
    {
      "epoch": 26.584615384615386,
      "grad_norm": 0.5250750780105591,
      "learning_rate": 0.002828,
      "loss": 1.2261,
      "step": 354240
    },
    {
      "epoch": 26.58941838649156,
      "grad_norm": 0.5908548831939697,
      "learning_rate": 0.002828,
      "loss": 1.2216,
      "step": 354304
    },
    {
      "epoch": 26.59422138836773,
      "grad_norm": 0.5817123651504517,
      "learning_rate": 0.002828,
      "loss": 1.2183,
      "step": 354368
    },
    {
      "epoch": 26.5990243902439,
      "grad_norm": 0.5596290826797485,
      "learning_rate": 0.002828,
      "loss": 1.2177,
      "step": 354432
    },
    {
      "epoch": 26.603827392120074,
      "grad_norm": 0.7440441846847534,
      "learning_rate": 0.002828,
      "loss": 1.2171,
      "step": 354496
    },
    {
      "epoch": 26.608630393996247,
      "grad_norm": 0.6214221119880676,
      "learning_rate": 0.002828,
      "loss": 1.2158,
      "step": 354560
    },
    {
      "epoch": 26.61343339587242,
      "grad_norm": 0.5204899907112122,
      "learning_rate": 0.002828,
      "loss": 1.2196,
      "step": 354624
    },
    {
      "epoch": 26.618236397748593,
      "grad_norm": 0.6049349904060364,
      "learning_rate": 0.002828,
      "loss": 1.2217,
      "step": 354688
    },
    {
      "epoch": 26.623039399624766,
      "grad_norm": 0.5005108118057251,
      "learning_rate": 0.002828,
      "loss": 1.2208,
      "step": 354752
    },
    {
      "epoch": 26.62784240150094,
      "grad_norm": 0.6195684671401978,
      "learning_rate": 0.002828,
      "loss": 1.2126,
      "step": 354816
    },
    {
      "epoch": 26.63264540337711,
      "grad_norm": 0.6185719966888428,
      "learning_rate": 0.002828,
      "loss": 1.2211,
      "step": 354880
    },
    {
      "epoch": 26.637448405253284,
      "grad_norm": 0.6542859673500061,
      "learning_rate": 0.002828,
      "loss": 1.2183,
      "step": 354944
    },
    {
      "epoch": 26.642251407129457,
      "grad_norm": 0.73667973279953,
      "learning_rate": 0.002828,
      "loss": 1.2255,
      "step": 355008
    },
    {
      "epoch": 26.64705440900563,
      "grad_norm": 0.6478698253631592,
      "learning_rate": 0.002828,
      "loss": 1.2181,
      "step": 355072
    },
    {
      "epoch": 26.651857410881803,
      "grad_norm": 0.5741866230964661,
      "learning_rate": 0.002828,
      "loss": 1.218,
      "step": 355136
    },
    {
      "epoch": 26.656660412757972,
      "grad_norm": 0.5622037649154663,
      "learning_rate": 0.002828,
      "loss": 1.2163,
      "step": 355200
    },
    {
      "epoch": 26.661463414634145,
      "grad_norm": 0.5420361161231995,
      "learning_rate": 0.002828,
      "loss": 1.2218,
      "step": 355264
    },
    {
      "epoch": 26.666266416510318,
      "grad_norm": 0.6684010624885559,
      "learning_rate": 0.002828,
      "loss": 1.2209,
      "step": 355328
    },
    {
      "epoch": 26.67106941838649,
      "grad_norm": 0.6673370003700256,
      "learning_rate": 0.002828,
      "loss": 1.2175,
      "step": 355392
    },
    {
      "epoch": 26.675872420262664,
      "grad_norm": 0.5472583770751953,
      "learning_rate": 0.002828,
      "loss": 1.2205,
      "step": 355456
    },
    {
      "epoch": 26.680675422138837,
      "grad_norm": 0.5732004642486572,
      "learning_rate": 0.002828,
      "loss": 1.2233,
      "step": 355520
    },
    {
      "epoch": 26.68547842401501,
      "grad_norm": 0.5972156524658203,
      "learning_rate": 0.002828,
      "loss": 1.2203,
      "step": 355584
    },
    {
      "epoch": 26.690281425891182,
      "grad_norm": 0.6077661514282227,
      "learning_rate": 0.002828,
      "loss": 1.2213,
      "step": 355648
    },
    {
      "epoch": 26.695084427767355,
      "grad_norm": 0.5697489380836487,
      "learning_rate": 0.002828,
      "loss": 1.2214,
      "step": 355712
    },
    {
      "epoch": 26.69988742964353,
      "grad_norm": 0.5756834745407104,
      "learning_rate": 0.002828,
      "loss": 1.2201,
      "step": 355776
    },
    {
      "epoch": 26.7046904315197,
      "grad_norm": 0.5190218687057495,
      "learning_rate": 0.002828,
      "loss": 1.2224,
      "step": 355840
    },
    {
      "epoch": 26.709493433395874,
      "grad_norm": 0.5287311673164368,
      "learning_rate": 0.002828,
      "loss": 1.2254,
      "step": 355904
    },
    {
      "epoch": 26.714296435272043,
      "grad_norm": 0.6516658663749695,
      "learning_rate": 0.002828,
      "loss": 1.2202,
      "step": 355968
    },
    {
      "epoch": 26.719099437148216,
      "grad_norm": 0.5656246542930603,
      "learning_rate": 0.002828,
      "loss": 1.2218,
      "step": 356032
    },
    {
      "epoch": 26.72390243902439,
      "grad_norm": 0.5156927704811096,
      "learning_rate": 0.002828,
      "loss": 1.2163,
      "step": 356096
    },
    {
      "epoch": 26.728705440900562,
      "grad_norm": 0.645133912563324,
      "learning_rate": 0.002828,
      "loss": 1.2275,
      "step": 356160
    },
    {
      "epoch": 26.733508442776735,
      "grad_norm": 0.5428419709205627,
      "learning_rate": 0.002828,
      "loss": 1.224,
      "step": 356224
    },
    {
      "epoch": 26.738311444652908,
      "grad_norm": 0.6389603018760681,
      "learning_rate": 0.002828,
      "loss": 1.2184,
      "step": 356288
    },
    {
      "epoch": 26.74311444652908,
      "grad_norm": 0.5498131513595581,
      "learning_rate": 0.002828,
      "loss": 1.2227,
      "step": 356352
    },
    {
      "epoch": 26.747917448405254,
      "grad_norm": 0.6554180383682251,
      "learning_rate": 0.002828,
      "loss": 1.2233,
      "step": 356416
    },
    {
      "epoch": 26.752720450281426,
      "grad_norm": 0.69988614320755,
      "learning_rate": 0.002828,
      "loss": 1.2159,
      "step": 356480
    },
    {
      "epoch": 26.7575234521576,
      "grad_norm": 0.6135717630386353,
      "learning_rate": 0.002828,
      "loss": 1.2236,
      "step": 356544
    },
    {
      "epoch": 26.762326454033772,
      "grad_norm": 0.5877432227134705,
      "learning_rate": 0.002828,
      "loss": 1.2185,
      "step": 356608
    },
    {
      "epoch": 26.767129455909945,
      "grad_norm": 0.6867513060569763,
      "learning_rate": 0.002828,
      "loss": 1.222,
      "step": 356672
    },
    {
      "epoch": 26.771932457786118,
      "grad_norm": 0.6261159181594849,
      "learning_rate": 0.002828,
      "loss": 1.2247,
      "step": 356736
    },
    {
      "epoch": 26.776735459662287,
      "grad_norm": 0.6596559286117554,
      "learning_rate": 0.002828,
      "loss": 1.223,
      "step": 356800
    },
    {
      "epoch": 26.78153846153846,
      "grad_norm": 0.5095260143280029,
      "learning_rate": 0.002828,
      "loss": 1.22,
      "step": 356864
    },
    {
      "epoch": 26.786341463414633,
      "grad_norm": 0.7851542830467224,
      "learning_rate": 0.002828,
      "loss": 1.217,
      "step": 356928
    },
    {
      "epoch": 26.791144465290806,
      "grad_norm": 0.6265200972557068,
      "learning_rate": 0.002828,
      "loss": 1.2242,
      "step": 356992
    },
    {
      "epoch": 26.79594746716698,
      "grad_norm": 0.5440147519111633,
      "learning_rate": 0.002828,
      "loss": 1.2189,
      "step": 357056
    },
    {
      "epoch": 26.80075046904315,
      "grad_norm": 0.5513255596160889,
      "learning_rate": 0.002828,
      "loss": 1.221,
      "step": 357120
    },
    {
      "epoch": 26.805553470919325,
      "grad_norm": 0.5585968494415283,
      "learning_rate": 0.002828,
      "loss": 1.2275,
      "step": 357184
    },
    {
      "epoch": 26.810356472795497,
      "grad_norm": 0.5784460306167603,
      "learning_rate": 0.002828,
      "loss": 1.2221,
      "step": 357248
    },
    {
      "epoch": 26.81515947467167,
      "grad_norm": 0.6278428435325623,
      "learning_rate": 0.002828,
      "loss": 1.2271,
      "step": 357312
    },
    {
      "epoch": 26.819962476547843,
      "grad_norm": 0.6241024136543274,
      "learning_rate": 0.002828,
      "loss": 1.2237,
      "step": 357376
    },
    {
      "epoch": 26.824765478424016,
      "grad_norm": 0.5838163495063782,
      "learning_rate": 0.002828,
      "loss": 1.2276,
      "step": 357440
    },
    {
      "epoch": 26.82956848030019,
      "grad_norm": 0.49887073040008545,
      "learning_rate": 0.002828,
      "loss": 1.2245,
      "step": 357504
    },
    {
      "epoch": 26.834371482176362,
      "grad_norm": 0.5843515992164612,
      "learning_rate": 0.002828,
      "loss": 1.2187,
      "step": 357568
    },
    {
      "epoch": 26.83917448405253,
      "grad_norm": 0.5945191979408264,
      "learning_rate": 0.002828,
      "loss": 1.2238,
      "step": 357632
    },
    {
      "epoch": 26.843977485928704,
      "grad_norm": 0.5643002986907959,
      "learning_rate": 0.002828,
      "loss": 1.2173,
      "step": 357696
    },
    {
      "epoch": 26.848780487804877,
      "grad_norm": 0.545021116733551,
      "learning_rate": 0.002828,
      "loss": 1.2252,
      "step": 357760
    },
    {
      "epoch": 26.85358348968105,
      "grad_norm": 0.5684725642204285,
      "learning_rate": 0.002828,
      "loss": 1.2253,
      "step": 357824
    },
    {
      "epoch": 26.858386491557223,
      "grad_norm": 0.6127748489379883,
      "learning_rate": 0.002828,
      "loss": 1.2191,
      "step": 357888
    },
    {
      "epoch": 26.863189493433396,
      "grad_norm": 0.6032697558403015,
      "learning_rate": 0.002828,
      "loss": 1.217,
      "step": 357952
    },
    {
      "epoch": 26.86799249530957,
      "grad_norm": 0.6626061797142029,
      "learning_rate": 0.002828,
      "loss": 1.2218,
      "step": 358016
    },
    {
      "epoch": 26.87279549718574,
      "grad_norm": 0.7547987699508667,
      "learning_rate": 0.002828,
      "loss": 1.2178,
      "step": 358080
    },
    {
      "epoch": 26.877598499061914,
      "grad_norm": 0.5466580390930176,
      "learning_rate": 0.002828,
      "loss": 1.2196,
      "step": 358144
    },
    {
      "epoch": 26.882401500938087,
      "grad_norm": 0.7023701667785645,
      "learning_rate": 0.002828,
      "loss": 1.2242,
      "step": 358208
    },
    {
      "epoch": 26.88720450281426,
      "grad_norm": 0.6215884685516357,
      "learning_rate": 0.002828,
      "loss": 1.2253,
      "step": 358272
    },
    {
      "epoch": 26.892007504690433,
      "grad_norm": 0.621918797492981,
      "learning_rate": 0.002828,
      "loss": 1.2227,
      "step": 358336
    },
    {
      "epoch": 26.896810506566602,
      "grad_norm": 0.5851648449897766,
      "learning_rate": 0.002828,
      "loss": 1.2261,
      "step": 358400
    },
    {
      "epoch": 26.901613508442775,
      "grad_norm": 0.6201433539390564,
      "learning_rate": 0.002828,
      "loss": 1.2241,
      "step": 358464
    },
    {
      "epoch": 26.906416510318948,
      "grad_norm": 0.5449443459510803,
      "learning_rate": 0.002828,
      "loss": 1.2227,
      "step": 358528
    },
    {
      "epoch": 26.91121951219512,
      "grad_norm": 0.5317158102989197,
      "learning_rate": 0.002828,
      "loss": 1.2207,
      "step": 358592
    },
    {
      "epoch": 26.916022514071294,
      "grad_norm": 0.6267401576042175,
      "learning_rate": 0.002828,
      "loss": 1.2224,
      "step": 358656
    },
    {
      "epoch": 26.920825515947467,
      "grad_norm": 0.5952174663543701,
      "learning_rate": 0.002828,
      "loss": 1.2229,
      "step": 358720
    },
    {
      "epoch": 26.92562851782364,
      "grad_norm": 0.6656439304351807,
      "learning_rate": 0.002828,
      "loss": 1.2215,
      "step": 358784
    },
    {
      "epoch": 26.930431519699813,
      "grad_norm": 0.5638568997383118,
      "learning_rate": 0.002828,
      "loss": 1.2289,
      "step": 358848
    },
    {
      "epoch": 26.935234521575985,
      "grad_norm": 0.5787448883056641,
      "learning_rate": 0.002828,
      "loss": 1.2243,
      "step": 358912
    },
    {
      "epoch": 26.94003752345216,
      "grad_norm": 0.7306926846504211,
      "learning_rate": 0.002828,
      "loss": 1.2243,
      "step": 358976
    },
    {
      "epoch": 26.94484052532833,
      "grad_norm": 0.5893208980560303,
      "learning_rate": 0.002828,
      "loss": 1.2222,
      "step": 359040
    },
    {
      "epoch": 26.949643527204504,
      "grad_norm": 0.5581895709037781,
      "learning_rate": 0.002828,
      "loss": 1.2264,
      "step": 359104
    },
    {
      "epoch": 26.954446529080677,
      "grad_norm": 0.548004686832428,
      "learning_rate": 0.002828,
      "loss": 1.2236,
      "step": 359168
    },
    {
      "epoch": 26.959249530956846,
      "grad_norm": 0.5354708433151245,
      "learning_rate": 0.002828,
      "loss": 1.2243,
      "step": 359232
    },
    {
      "epoch": 26.96405253283302,
      "grad_norm": 0.5066986083984375,
      "learning_rate": 0.002828,
      "loss": 1.2257,
      "step": 359296
    },
    {
      "epoch": 26.968855534709192,
      "grad_norm": 0.5515348315238953,
      "learning_rate": 0.002828,
      "loss": 1.2286,
      "step": 359360
    },
    {
      "epoch": 26.973658536585365,
      "grad_norm": 0.5831103920936584,
      "learning_rate": 0.002828,
      "loss": 1.2265,
      "step": 359424
    },
    {
      "epoch": 26.978461538461538,
      "grad_norm": 0.5766531229019165,
      "learning_rate": 0.002828,
      "loss": 1.2284,
      "step": 359488
    },
    {
      "epoch": 26.98326454033771,
      "grad_norm": 0.7886342406272888,
      "learning_rate": 0.002828,
      "loss": 1.2273,
      "step": 359552
    },
    {
      "epoch": 26.988067542213884,
      "grad_norm": 0.6386691927909851,
      "learning_rate": 0.002828,
      "loss": 1.2314,
      "step": 359616
    },
    {
      "epoch": 26.992870544090056,
      "grad_norm": 0.5953747630119324,
      "learning_rate": 0.002828,
      "loss": 1.2205,
      "step": 359680
    },
    {
      "epoch": 26.99767354596623,
      "grad_norm": 0.5155026316642761,
      "learning_rate": 0.002828,
      "loss": 1.2194,
      "step": 359744
    },
    {
      "epoch": 27.002476547842402,
      "grad_norm": 0.6157639026641846,
      "learning_rate": 0.002828,
      "loss": 1.2097,
      "step": 359808
    },
    {
      "epoch": 27.007279549718575,
      "grad_norm": 0.5952342748641968,
      "learning_rate": 0.002828,
      "loss": 1.1807,
      "step": 359872
    },
    {
      "epoch": 27.012082551594748,
      "grad_norm": 0.6238614916801453,
      "learning_rate": 0.002828,
      "loss": 1.1813,
      "step": 359936
    },
    {
      "epoch": 27.01688555347092,
      "grad_norm": 0.5040086507797241,
      "learning_rate": 0.002828,
      "loss": 1.1838,
      "step": 360000
    },
    {
      "epoch": 27.02168855534709,
      "grad_norm": 0.5880621671676636,
      "learning_rate": 0.002828,
      "loss": 1.1862,
      "step": 360064
    },
    {
      "epoch": 27.026491557223263,
      "grad_norm": 0.5540125966072083,
      "learning_rate": 0.002828,
      "loss": 1.1836,
      "step": 360128
    },
    {
      "epoch": 27.031294559099436,
      "grad_norm": 0.7543290853500366,
      "learning_rate": 0.002828,
      "loss": 1.1844,
      "step": 360192
    },
    {
      "epoch": 27.03609756097561,
      "grad_norm": 0.5563064217567444,
      "learning_rate": 0.002828,
      "loss": 1.1951,
      "step": 360256
    },
    {
      "epoch": 27.040900562851782,
      "grad_norm": 0.5678032040596008,
      "learning_rate": 0.002828,
      "loss": 1.1882,
      "step": 360320
    },
    {
      "epoch": 27.045703564727955,
      "grad_norm": 0.5668981671333313,
      "learning_rate": 0.002828,
      "loss": 1.1883,
      "step": 360384
    },
    {
      "epoch": 27.050506566604128,
      "grad_norm": 0.6456194519996643,
      "learning_rate": 0.002828,
      "loss": 1.188,
      "step": 360448
    },
    {
      "epoch": 27.0553095684803,
      "grad_norm": 0.5644674897193909,
      "learning_rate": 0.002828,
      "loss": 1.1907,
      "step": 360512
    },
    {
      "epoch": 27.060112570356473,
      "grad_norm": 0.6194739937782288,
      "learning_rate": 0.002828,
      "loss": 1.19,
      "step": 360576
    },
    {
      "epoch": 27.064915572232646,
      "grad_norm": 0.584814190864563,
      "learning_rate": 0.002828,
      "loss": 1.1868,
      "step": 360640
    },
    {
      "epoch": 27.06971857410882,
      "grad_norm": 0.5569928288459778,
      "learning_rate": 0.002828,
      "loss": 1.1862,
      "step": 360704
    },
    {
      "epoch": 27.074521575984992,
      "grad_norm": 0.5890145301818848,
      "learning_rate": 0.002828,
      "loss": 1.1916,
      "step": 360768
    },
    {
      "epoch": 27.079324577861165,
      "grad_norm": 0.5473900437355042,
      "learning_rate": 0.002828,
      "loss": 1.1914,
      "step": 360832
    },
    {
      "epoch": 27.084127579737334,
      "grad_norm": 0.6628814339637756,
      "learning_rate": 0.002828,
      "loss": 1.1914,
      "step": 360896
    },
    {
      "epoch": 27.088930581613507,
      "grad_norm": 0.6406671404838562,
      "learning_rate": 0.002828,
      "loss": 1.1932,
      "step": 360960
    },
    {
      "epoch": 27.09373358348968,
      "grad_norm": 0.5623260140419006,
      "learning_rate": 0.002828,
      "loss": 1.1938,
      "step": 361024
    },
    {
      "epoch": 27.098536585365853,
      "grad_norm": 0.7211869359016418,
      "learning_rate": 0.002828,
      "loss": 1.1985,
      "step": 361088
    },
    {
      "epoch": 27.103339587242026,
      "grad_norm": 0.5265310406684875,
      "learning_rate": 0.002828,
      "loss": 1.191,
      "step": 361152
    },
    {
      "epoch": 27.1081425891182,
      "grad_norm": 0.7146878242492676,
      "learning_rate": 0.002828,
      "loss": 1.1942,
      "step": 361216
    },
    {
      "epoch": 27.11294559099437,
      "grad_norm": 0.6107980012893677,
      "learning_rate": 0.002828,
      "loss": 1.193,
      "step": 361280
    },
    {
      "epoch": 27.117748592870544,
      "grad_norm": 0.6449308395385742,
      "learning_rate": 0.002828,
      "loss": 1.1963,
      "step": 361344
    },
    {
      "epoch": 27.122551594746717,
      "grad_norm": 0.6099646091461182,
      "learning_rate": 0.002828,
      "loss": 1.1949,
      "step": 361408
    },
    {
      "epoch": 27.12735459662289,
      "grad_norm": 0.7447180151939392,
      "learning_rate": 0.002828,
      "loss": 1.1938,
      "step": 361472
    },
    {
      "epoch": 27.132157598499063,
      "grad_norm": 0.7580665349960327,
      "learning_rate": 0.002828,
      "loss": 1.1975,
      "step": 361536
    },
    {
      "epoch": 27.136960600375236,
      "grad_norm": 0.5835822820663452,
      "learning_rate": 0.002828,
      "loss": 1.1875,
      "step": 361600
    },
    {
      "epoch": 27.14176360225141,
      "grad_norm": 0.5681918859481812,
      "learning_rate": 0.002828,
      "loss": 1.1966,
      "step": 361664
    },
    {
      "epoch": 27.146566604127578,
      "grad_norm": 0.5537122488021851,
      "learning_rate": 0.002828,
      "loss": 1.1916,
      "step": 361728
    },
    {
      "epoch": 27.15136960600375,
      "grad_norm": 0.5107552409172058,
      "learning_rate": 0.002828,
      "loss": 1.1883,
      "step": 361792
    },
    {
      "epoch": 27.156172607879924,
      "grad_norm": 0.6284750699996948,
      "learning_rate": 0.002828,
      "loss": 1.196,
      "step": 361856
    },
    {
      "epoch": 27.160975609756097,
      "grad_norm": 0.7075376510620117,
      "learning_rate": 0.002828,
      "loss": 1.1987,
      "step": 361920
    },
    {
      "epoch": 27.16577861163227,
      "grad_norm": 0.7268292903900146,
      "learning_rate": 0.002828,
      "loss": 1.1958,
      "step": 361984
    },
    {
      "epoch": 27.170581613508443,
      "grad_norm": 0.6370378136634827,
      "learning_rate": 0.002828,
      "loss": 1.1899,
      "step": 362048
    },
    {
      "epoch": 27.175384615384615,
      "grad_norm": 0.5429769158363342,
      "learning_rate": 0.002828,
      "loss": 1.1987,
      "step": 362112
    },
    {
      "epoch": 27.18018761726079,
      "grad_norm": 0.569195568561554,
      "learning_rate": 0.002828,
      "loss": 1.1927,
      "step": 362176
    },
    {
      "epoch": 27.18499061913696,
      "grad_norm": 0.6059727668762207,
      "learning_rate": 0.002828,
      "loss": 1.1929,
      "step": 362240
    },
    {
      "epoch": 27.189793621013134,
      "grad_norm": 0.6149117946624756,
      "learning_rate": 0.002828,
      "loss": 1.1942,
      "step": 362304
    },
    {
      "epoch": 27.194596622889307,
      "grad_norm": 0.5530393719673157,
      "learning_rate": 0.002828,
      "loss": 1.1926,
      "step": 362368
    },
    {
      "epoch": 27.19939962476548,
      "grad_norm": 0.5254247188568115,
      "learning_rate": 0.002828,
      "loss": 1.1984,
      "step": 362432
    },
    {
      "epoch": 27.204202626641653,
      "grad_norm": 0.5869778990745544,
      "learning_rate": 0.002828,
      "loss": 1.1975,
      "step": 362496
    },
    {
      "epoch": 27.209005628517822,
      "grad_norm": 0.6287544369697571,
      "learning_rate": 0.002828,
      "loss": 1.1919,
      "step": 362560
    },
    {
      "epoch": 27.213808630393995,
      "grad_norm": 0.6365551948547363,
      "learning_rate": 0.002828,
      "loss": 1.1961,
      "step": 362624
    },
    {
      "epoch": 27.218611632270168,
      "grad_norm": 0.501448929309845,
      "learning_rate": 0.002828,
      "loss": 1.2058,
      "step": 362688
    },
    {
      "epoch": 27.22341463414634,
      "grad_norm": 0.5506671667098999,
      "learning_rate": 0.002828,
      "loss": 1.2008,
      "step": 362752
    },
    {
      "epoch": 27.228217636022514,
      "grad_norm": 0.6569094657897949,
      "learning_rate": 0.002828,
      "loss": 1.2003,
      "step": 362816
    },
    {
      "epoch": 27.233020637898687,
      "grad_norm": 0.5557685494422913,
      "learning_rate": 0.002828,
      "loss": 1.1955,
      "step": 362880
    },
    {
      "epoch": 27.23782363977486,
      "grad_norm": 0.6757054328918457,
      "learning_rate": 0.002828,
      "loss": 1.1967,
      "step": 362944
    },
    {
      "epoch": 27.242626641651032,
      "grad_norm": 0.5856550335884094,
      "learning_rate": 0.002828,
      "loss": 1.197,
      "step": 363008
    },
    {
      "epoch": 27.247429643527205,
      "grad_norm": 0.5449078679084778,
      "learning_rate": 0.002828,
      "loss": 1.1995,
      "step": 363072
    },
    {
      "epoch": 27.252232645403378,
      "grad_norm": 0.5528790354728699,
      "learning_rate": 0.002828,
      "loss": 1.2017,
      "step": 363136
    },
    {
      "epoch": 27.25703564727955,
      "grad_norm": 0.5473564863204956,
      "learning_rate": 0.002828,
      "loss": 1.1953,
      "step": 363200
    },
    {
      "epoch": 27.261838649155724,
      "grad_norm": 0.6807570457458496,
      "learning_rate": 0.002828,
      "loss": 1.2011,
      "step": 363264
    },
    {
      "epoch": 27.266641651031893,
      "grad_norm": 0.5755122303962708,
      "learning_rate": 0.002828,
      "loss": 1.201,
      "step": 363328
    },
    {
      "epoch": 27.271444652908066,
      "grad_norm": 0.6368680000305176,
      "learning_rate": 0.002828,
      "loss": 1.2027,
      "step": 363392
    },
    {
      "epoch": 27.27624765478424,
      "grad_norm": 0.555194079875946,
      "learning_rate": 0.002828,
      "loss": 1.2019,
      "step": 363456
    },
    {
      "epoch": 27.281050656660412,
      "grad_norm": 0.6025437712669373,
      "learning_rate": 0.002828,
      "loss": 1.1962,
      "step": 363520
    },
    {
      "epoch": 27.285853658536585,
      "grad_norm": 0.5380324721336365,
      "learning_rate": 0.002828,
      "loss": 1.196,
      "step": 363584
    },
    {
      "epoch": 27.290656660412758,
      "grad_norm": 0.6038591265678406,
      "learning_rate": 0.002828,
      "loss": 1.2011,
      "step": 363648
    },
    {
      "epoch": 27.29545966228893,
      "grad_norm": 0.5660392642021179,
      "learning_rate": 0.002828,
      "loss": 1.2017,
      "step": 363712
    },
    {
      "epoch": 27.300262664165103,
      "grad_norm": 0.590695858001709,
      "learning_rate": 0.002828,
      "loss": 1.1945,
      "step": 363776
    },
    {
      "epoch": 27.305065666041276,
      "grad_norm": 0.6372555494308472,
      "learning_rate": 0.002828,
      "loss": 1.2016,
      "step": 363840
    },
    {
      "epoch": 27.30986866791745,
      "grad_norm": 0.6135676503181458,
      "learning_rate": 0.002828,
      "loss": 1.1995,
      "step": 363904
    },
    {
      "epoch": 27.314671669793622,
      "grad_norm": 0.5026392340660095,
      "learning_rate": 0.002828,
      "loss": 1.2005,
      "step": 363968
    },
    {
      "epoch": 27.319474671669795,
      "grad_norm": 0.5486312508583069,
      "learning_rate": 0.002828,
      "loss": 1.2019,
      "step": 364032
    },
    {
      "epoch": 27.324277673545968,
      "grad_norm": 0.5584508776664734,
      "learning_rate": 0.002828,
      "loss": 1.2015,
      "step": 364096
    },
    {
      "epoch": 27.329080675422137,
      "grad_norm": 0.5603387951850891,
      "learning_rate": 0.002828,
      "loss": 1.2008,
      "step": 364160
    },
    {
      "epoch": 27.33388367729831,
      "grad_norm": 0.6798414587974548,
      "learning_rate": 0.002828,
      "loss": 1.1984,
      "step": 364224
    },
    {
      "epoch": 27.338686679174483,
      "grad_norm": 0.5990169048309326,
      "learning_rate": 0.002828,
      "loss": 1.2001,
      "step": 364288
    },
    {
      "epoch": 27.343489681050656,
      "grad_norm": 0.7844681739807129,
      "learning_rate": 0.002828,
      "loss": 1.1981,
      "step": 364352
    },
    {
      "epoch": 27.34829268292683,
      "grad_norm": 0.6317253708839417,
      "learning_rate": 0.002828,
      "loss": 1.2035,
      "step": 364416
    },
    {
      "epoch": 27.353095684803,
      "grad_norm": 0.5308550596237183,
      "learning_rate": 0.002828,
      "loss": 1.2005,
      "step": 364480
    },
    {
      "epoch": 27.357898686679174,
      "grad_norm": 0.4954928755760193,
      "learning_rate": 0.002828,
      "loss": 1.195,
      "step": 364544
    },
    {
      "epoch": 27.362701688555347,
      "grad_norm": 0.5386978387832642,
      "learning_rate": 0.002828,
      "loss": 1.203,
      "step": 364608
    },
    {
      "epoch": 27.36750469043152,
      "grad_norm": 0.5351986885070801,
      "learning_rate": 0.002828,
      "loss": 1.2036,
      "step": 364672
    },
    {
      "epoch": 27.372307692307693,
      "grad_norm": 0.5314650535583496,
      "learning_rate": 0.002828,
      "loss": 1.202,
      "step": 364736
    },
    {
      "epoch": 27.377110694183866,
      "grad_norm": 0.5204005837440491,
      "learning_rate": 0.002828,
      "loss": 1.202,
      "step": 364800
    },
    {
      "epoch": 27.38191369606004,
      "grad_norm": 0.6783261299133301,
      "learning_rate": 0.002828,
      "loss": 1.2035,
      "step": 364864
    },
    {
      "epoch": 27.38671669793621,
      "grad_norm": 0.678269624710083,
      "learning_rate": 0.002828,
      "loss": 1.2031,
      "step": 364928
    },
    {
      "epoch": 27.39151969981238,
      "grad_norm": 0.7005301713943481,
      "learning_rate": 0.002828,
      "loss": 1.2084,
      "step": 364992
    },
    {
      "epoch": 27.396322701688554,
      "grad_norm": 0.6017382740974426,
      "learning_rate": 0.002828,
      "loss": 1.2083,
      "step": 365056
    },
    {
      "epoch": 27.401125703564727,
      "grad_norm": 0.5842321515083313,
      "learning_rate": 0.002828,
      "loss": 1.2031,
      "step": 365120
    },
    {
      "epoch": 27.4059287054409,
      "grad_norm": 0.5812395811080933,
      "learning_rate": 0.002828,
      "loss": 1.2075,
      "step": 365184
    },
    {
      "epoch": 27.410731707317073,
      "grad_norm": 0.487456351518631,
      "learning_rate": 0.002828,
      "loss": 1.2042,
      "step": 365248
    },
    {
      "epoch": 27.415534709193246,
      "grad_norm": 0.5946784019470215,
      "learning_rate": 0.002828,
      "loss": 1.2007,
      "step": 365312
    },
    {
      "epoch": 27.42033771106942,
      "grad_norm": 0.6484962105751038,
      "learning_rate": 0.002828,
      "loss": 1.2049,
      "step": 365376
    },
    {
      "epoch": 27.42514071294559,
      "grad_norm": 0.5717827081680298,
      "learning_rate": 0.002828,
      "loss": 1.2045,
      "step": 365440
    },
    {
      "epoch": 27.429943714821764,
      "grad_norm": 0.5931734442710876,
      "learning_rate": 0.002828,
      "loss": 1.2047,
      "step": 365504
    },
    {
      "epoch": 27.434746716697937,
      "grad_norm": 0.5433608889579773,
      "learning_rate": 0.002828,
      "loss": 1.2075,
      "step": 365568
    },
    {
      "epoch": 27.43954971857411,
      "grad_norm": 0.5237091779708862,
      "learning_rate": 0.002828,
      "loss": 1.1959,
      "step": 365632
    },
    {
      "epoch": 27.444352720450283,
      "grad_norm": 0.5132204294204712,
      "learning_rate": 0.002828,
      "loss": 1.2055,
      "step": 365696
    },
    {
      "epoch": 27.449155722326456,
      "grad_norm": 0.7822837233543396,
      "learning_rate": 0.002828,
      "loss": 1.2055,
      "step": 365760
    },
    {
      "epoch": 27.453958724202625,
      "grad_norm": 0.4899812638759613,
      "learning_rate": 0.002828,
      "loss": 1.199,
      "step": 365824
    },
    {
      "epoch": 27.458761726078798,
      "grad_norm": 0.5946176648139954,
      "learning_rate": 0.002828,
      "loss": 1.1991,
      "step": 365888
    },
    {
      "epoch": 27.46356472795497,
      "grad_norm": 0.7705643177032471,
      "learning_rate": 0.002828,
      "loss": 1.208,
      "step": 365952
    },
    {
      "epoch": 27.468367729831144,
      "grad_norm": 0.5611868500709534,
      "learning_rate": 0.002828,
      "loss": 1.2011,
      "step": 366016
    },
    {
      "epoch": 27.473170731707317,
      "grad_norm": 0.6456145644187927,
      "learning_rate": 0.002828,
      "loss": 1.2046,
      "step": 366080
    },
    {
      "epoch": 27.47797373358349,
      "grad_norm": 0.5337197184562683,
      "learning_rate": 0.002828,
      "loss": 1.2025,
      "step": 366144
    },
    {
      "epoch": 27.482776735459662,
      "grad_norm": 0.6826463937759399,
      "learning_rate": 0.002828,
      "loss": 1.201,
      "step": 366208
    },
    {
      "epoch": 27.487579737335835,
      "grad_norm": 0.5261040329933167,
      "learning_rate": 0.002828,
      "loss": 1.2137,
      "step": 366272
    },
    {
      "epoch": 27.492382739212008,
      "grad_norm": 0.5058457255363464,
      "learning_rate": 0.002828,
      "loss": 1.207,
      "step": 366336
    },
    {
      "epoch": 27.49718574108818,
      "grad_norm": 0.6075417995452881,
      "learning_rate": 0.002828,
      "loss": 1.2058,
      "step": 366400
    },
    {
      "epoch": 27.501988742964354,
      "grad_norm": 0.5280016660690308,
      "learning_rate": 0.002828,
      "loss": 1.1983,
      "step": 366464
    },
    {
      "epoch": 27.506791744840527,
      "grad_norm": 0.6989396810531616,
      "learning_rate": 0.002828,
      "loss": 1.2057,
      "step": 366528
    },
    {
      "epoch": 27.511594746716696,
      "grad_norm": 0.5592902302742004,
      "learning_rate": 0.002828,
      "loss": 1.2045,
      "step": 366592
    },
    {
      "epoch": 27.51639774859287,
      "grad_norm": 0.6710501909255981,
      "learning_rate": 0.002828,
      "loss": 1.2068,
      "step": 366656
    },
    {
      "epoch": 27.521200750469042,
      "grad_norm": 0.590663492679596,
      "learning_rate": 0.002828,
      "loss": 1.2,
      "step": 366720
    },
    {
      "epoch": 27.526003752345215,
      "grad_norm": 0.5750949382781982,
      "learning_rate": 0.002828,
      "loss": 1.2102,
      "step": 366784
    },
    {
      "epoch": 27.530806754221388,
      "grad_norm": 0.592196524143219,
      "learning_rate": 0.002828,
      "loss": 1.2072,
      "step": 366848
    },
    {
      "epoch": 27.53560975609756,
      "grad_norm": 0.5525135397911072,
      "learning_rate": 0.002828,
      "loss": 1.2089,
      "step": 366912
    },
    {
      "epoch": 27.540412757973733,
      "grad_norm": 0.5719900131225586,
      "learning_rate": 0.002828,
      "loss": 1.2087,
      "step": 366976
    },
    {
      "epoch": 27.545215759849906,
      "grad_norm": 0.5599880218505859,
      "learning_rate": 0.002828,
      "loss": 1.2078,
      "step": 367040
    },
    {
      "epoch": 27.55001876172608,
      "grad_norm": 0.7876613736152649,
      "learning_rate": 0.002828,
      "loss": 1.2033,
      "step": 367104
    },
    {
      "epoch": 27.554821763602252,
      "grad_norm": 0.5568932294845581,
      "learning_rate": 0.002828,
      "loss": 1.2016,
      "step": 367168
    },
    {
      "epoch": 27.559624765478425,
      "grad_norm": 0.5835604667663574,
      "learning_rate": 0.002828,
      "loss": 1.2044,
      "step": 367232
    },
    {
      "epoch": 27.564427767354598,
      "grad_norm": 0.4952414631843567,
      "learning_rate": 0.002828,
      "loss": 1.2037,
      "step": 367296
    },
    {
      "epoch": 27.56923076923077,
      "grad_norm": 0.5456660389900208,
      "learning_rate": 0.002828,
      "loss": 1.198,
      "step": 367360
    },
    {
      "epoch": 27.57403377110694,
      "grad_norm": 0.5977687835693359,
      "learning_rate": 0.002828,
      "loss": 1.2115,
      "step": 367424
    },
    {
      "epoch": 27.578836772983113,
      "grad_norm": 0.5785269141197205,
      "learning_rate": 0.002828,
      "loss": 1.2126,
      "step": 367488
    },
    {
      "epoch": 27.583639774859286,
      "grad_norm": 0.6125280857086182,
      "learning_rate": 0.002828,
      "loss": 1.2084,
      "step": 367552
    },
    {
      "epoch": 27.58844277673546,
      "grad_norm": 0.5674769878387451,
      "learning_rate": 0.002828,
      "loss": 1.2078,
      "step": 367616
    },
    {
      "epoch": 27.59324577861163,
      "grad_norm": 0.5072876214981079,
      "learning_rate": 0.002828,
      "loss": 1.21,
      "step": 367680
    },
    {
      "epoch": 27.598048780487805,
      "grad_norm": 0.7092969417572021,
      "learning_rate": 0.002828,
      "loss": 1.2029,
      "step": 367744
    },
    {
      "epoch": 27.602851782363977,
      "grad_norm": 0.5517896413803101,
      "learning_rate": 0.002828,
      "loss": 1.2095,
      "step": 367808
    },
    {
      "epoch": 27.60765478424015,
      "grad_norm": 0.6972664594650269,
      "learning_rate": 0.002828,
      "loss": 1.2072,
      "step": 367872
    },
    {
      "epoch": 27.612457786116323,
      "grad_norm": 0.6762186288833618,
      "learning_rate": 0.002828,
      "loss": 1.2081,
      "step": 367936
    },
    {
      "epoch": 27.617260787992496,
      "grad_norm": 0.637596070766449,
      "learning_rate": 0.002828,
      "loss": 1.211,
      "step": 368000
    },
    {
      "epoch": 27.62206378986867,
      "grad_norm": 0.49644848704338074,
      "learning_rate": 0.002828,
      "loss": 1.2035,
      "step": 368064
    },
    {
      "epoch": 27.626866791744842,
      "grad_norm": 0.6612474322319031,
      "learning_rate": 0.002828,
      "loss": 1.2083,
      "step": 368128
    },
    {
      "epoch": 27.631669793621015,
      "grad_norm": 0.5150904655456543,
      "learning_rate": 0.002828,
      "loss": 1.2104,
      "step": 368192
    },
    {
      "epoch": 27.636472795497184,
      "grad_norm": 0.7277690768241882,
      "learning_rate": 0.002828,
      "loss": 1.2086,
      "step": 368256
    },
    {
      "epoch": 27.641275797373357,
      "grad_norm": 0.62581467628479,
      "learning_rate": 0.002828,
      "loss": 1.2106,
      "step": 368320
    },
    {
      "epoch": 27.64607879924953,
      "grad_norm": 0.7047035098075867,
      "learning_rate": 0.002828,
      "loss": 1.213,
      "step": 368384
    },
    {
      "epoch": 27.650881801125703,
      "grad_norm": 0.5907717347145081,
      "learning_rate": 0.002828,
      "loss": 1.2033,
      "step": 368448
    },
    {
      "epoch": 27.655684803001876,
      "grad_norm": 0.6000102162361145,
      "learning_rate": 0.002828,
      "loss": 1.2151,
      "step": 368512
    },
    {
      "epoch": 27.66048780487805,
      "grad_norm": 0.6553179025650024,
      "learning_rate": 0.002828,
      "loss": 1.2069,
      "step": 368576
    },
    {
      "epoch": 27.66529080675422,
      "grad_norm": 0.623139500617981,
      "learning_rate": 0.002828,
      "loss": 1.2105,
      "step": 368640
    },
    {
      "epoch": 27.670093808630394,
      "grad_norm": 0.6486320495605469,
      "learning_rate": 0.002828,
      "loss": 1.2103,
      "step": 368704
    },
    {
      "epoch": 27.674896810506567,
      "grad_norm": 0.5863367915153503,
      "learning_rate": 0.002828,
      "loss": 1.2074,
      "step": 368768
    },
    {
      "epoch": 27.67969981238274,
      "grad_norm": 0.5867400765419006,
      "learning_rate": 0.002828,
      "loss": 1.2076,
      "step": 368832
    },
    {
      "epoch": 27.684502814258913,
      "grad_norm": 0.6032609939575195,
      "learning_rate": 0.002828,
      "loss": 1.2093,
      "step": 368896
    },
    {
      "epoch": 27.689305816135086,
      "grad_norm": 0.5408675074577332,
      "learning_rate": 0.002828,
      "loss": 1.2117,
      "step": 368960
    },
    {
      "epoch": 27.69410881801126,
      "grad_norm": 0.7075996994972229,
      "learning_rate": 0.002828,
      "loss": 1.209,
      "step": 369024
    },
    {
      "epoch": 27.698911819887428,
      "grad_norm": 0.5513624548912048,
      "learning_rate": 0.002828,
      "loss": 1.2056,
      "step": 369088
    },
    {
      "epoch": 27.7037148217636,
      "grad_norm": 0.5250154137611389,
      "learning_rate": 0.002828,
      "loss": 1.2144,
      "step": 369152
    },
    {
      "epoch": 27.708517823639774,
      "grad_norm": 0.5961335301399231,
      "learning_rate": 0.002828,
      "loss": 1.2186,
      "step": 369216
    },
    {
      "epoch": 27.713320825515947,
      "grad_norm": 0.626379132270813,
      "learning_rate": 0.002828,
      "loss": 1.21,
      "step": 369280
    },
    {
      "epoch": 27.71812382739212,
      "grad_norm": 0.5551455616950989,
      "learning_rate": 0.002828,
      "loss": 1.204,
      "step": 369344
    },
    {
      "epoch": 27.722926829268292,
      "grad_norm": 0.5567622780799866,
      "learning_rate": 0.002828,
      "loss": 1.2046,
      "step": 369408
    },
    {
      "epoch": 27.727729831144465,
      "grad_norm": 0.6794117093086243,
      "learning_rate": 0.002828,
      "loss": 1.2087,
      "step": 369472
    },
    {
      "epoch": 27.732532833020638,
      "grad_norm": 0.6635667681694031,
      "learning_rate": 0.002828,
      "loss": 1.2106,
      "step": 369536
    },
    {
      "epoch": 27.73733583489681,
      "grad_norm": 0.5562478303909302,
      "learning_rate": 0.002828,
      "loss": 1.2108,
      "step": 369600
    },
    {
      "epoch": 27.742138836772984,
      "grad_norm": 0.6434085369110107,
      "learning_rate": 0.002828,
      "loss": 1.2013,
      "step": 369664
    },
    {
      "epoch": 27.746941838649157,
      "grad_norm": 0.7914767861366272,
      "learning_rate": 0.002828,
      "loss": 1.214,
      "step": 369728
    },
    {
      "epoch": 27.75174484052533,
      "grad_norm": 0.723877489566803,
      "learning_rate": 0.002828,
      "loss": 1.2139,
      "step": 369792
    },
    {
      "epoch": 27.756547842401503,
      "grad_norm": 0.5417585372924805,
      "learning_rate": 0.002828,
      "loss": 1.2147,
      "step": 369856
    },
    {
      "epoch": 27.761350844277672,
      "grad_norm": 0.5280811190605164,
      "learning_rate": 0.002828,
      "loss": 1.2093,
      "step": 369920
    },
    {
      "epoch": 27.766153846153845,
      "grad_norm": 0.5942422151565552,
      "learning_rate": 0.002828,
      "loss": 1.2131,
      "step": 369984
    },
    {
      "epoch": 27.770956848030018,
      "grad_norm": 0.5912095904350281,
      "learning_rate": 0.002828,
      "loss": 1.209,
      "step": 370048
    },
    {
      "epoch": 27.77575984990619,
      "grad_norm": 0.5693986415863037,
      "learning_rate": 0.002828,
      "loss": 1.2122,
      "step": 370112
    },
    {
      "epoch": 27.780562851782364,
      "grad_norm": 0.5609849691390991,
      "learning_rate": 0.002828,
      "loss": 1.2099,
      "step": 370176
    },
    {
      "epoch": 27.785365853658536,
      "grad_norm": 0.5266005992889404,
      "learning_rate": 0.002828,
      "loss": 1.2074,
      "step": 370240
    },
    {
      "epoch": 27.79016885553471,
      "grad_norm": 0.5555135011672974,
      "learning_rate": 0.002828,
      "loss": 1.218,
      "step": 370304
    },
    {
      "epoch": 27.794971857410882,
      "grad_norm": 0.6156359314918518,
      "learning_rate": 0.002828,
      "loss": 1.2109,
      "step": 370368
    },
    {
      "epoch": 27.799774859287055,
      "grad_norm": 0.5655944347381592,
      "learning_rate": 0.002828,
      "loss": 1.2076,
      "step": 370432
    },
    {
      "epoch": 27.804577861163228,
      "grad_norm": 0.4762551486492157,
      "learning_rate": 0.002828,
      "loss": 1.2109,
      "step": 370496
    },
    {
      "epoch": 27.8093808630394,
      "grad_norm": 0.5622091889381409,
      "learning_rate": 0.002828,
      "loss": 1.2161,
      "step": 370560
    },
    {
      "epoch": 27.814183864915574,
      "grad_norm": 0.6137254238128662,
      "learning_rate": 0.002828,
      "loss": 1.2119,
      "step": 370624
    },
    {
      "epoch": 27.818986866791747,
      "grad_norm": 0.6522278785705566,
      "learning_rate": 0.002828,
      "loss": 1.2092,
      "step": 370688
    },
    {
      "epoch": 27.823789868667916,
      "grad_norm": 0.683634340763092,
      "learning_rate": 0.002828,
      "loss": 1.2099,
      "step": 370752
    },
    {
      "epoch": 27.82859287054409,
      "grad_norm": 0.5084274411201477,
      "learning_rate": 0.002828,
      "loss": 1.2023,
      "step": 370816
    },
    {
      "epoch": 27.83339587242026,
      "grad_norm": 0.6315329074859619,
      "learning_rate": 0.002828,
      "loss": 1.2136,
      "step": 370880
    },
    {
      "epoch": 27.838198874296435,
      "grad_norm": 0.5127155780792236,
      "learning_rate": 0.002828,
      "loss": 1.2131,
      "step": 370944
    },
    {
      "epoch": 27.843001876172607,
      "grad_norm": 0.5103921890258789,
      "learning_rate": 0.002828,
      "loss": 1.2106,
      "step": 371008
    },
    {
      "epoch": 27.84780487804878,
      "grad_norm": 0.5688643455505371,
      "learning_rate": 0.002828,
      "loss": 1.2162,
      "step": 371072
    },
    {
      "epoch": 27.852607879924953,
      "grad_norm": 0.5613607168197632,
      "learning_rate": 0.002828,
      "loss": 1.2147,
      "step": 371136
    },
    {
      "epoch": 27.857410881801126,
      "grad_norm": 0.677702784538269,
      "learning_rate": 0.002828,
      "loss": 1.212,
      "step": 371200
    },
    {
      "epoch": 27.8622138836773,
      "grad_norm": 0.686955451965332,
      "learning_rate": 0.002828,
      "loss": 1.2093,
      "step": 371264
    },
    {
      "epoch": 27.867016885553472,
      "grad_norm": 0.6308797597885132,
      "learning_rate": 0.002828,
      "loss": 1.2121,
      "step": 371328
    },
    {
      "epoch": 27.871819887429645,
      "grad_norm": 0.5449283123016357,
      "learning_rate": 0.002828,
      "loss": 1.2182,
      "step": 371392
    },
    {
      "epoch": 27.876622889305818,
      "grad_norm": 0.6226169466972351,
      "learning_rate": 0.002828,
      "loss": 1.208,
      "step": 371456
    },
    {
      "epoch": 27.88142589118199,
      "grad_norm": 0.6177122592926025,
      "learning_rate": 0.002828,
      "loss": 1.2113,
      "step": 371520
    },
    {
      "epoch": 27.88622889305816,
      "grad_norm": 0.6782193183898926,
      "learning_rate": 0.002828,
      "loss": 1.2102,
      "step": 371584
    },
    {
      "epoch": 27.891031894934333,
      "grad_norm": 0.5172768831253052,
      "learning_rate": 0.002828,
      "loss": 1.207,
      "step": 371648
    },
    {
      "epoch": 27.895834896810506,
      "grad_norm": 0.5725329518318176,
      "learning_rate": 0.002828,
      "loss": 1.2118,
      "step": 371712
    },
    {
      "epoch": 27.90063789868668,
      "grad_norm": 0.5436069369316101,
      "learning_rate": 0.002828,
      "loss": 1.2157,
      "step": 371776
    },
    {
      "epoch": 27.90544090056285,
      "grad_norm": 0.5015886425971985,
      "learning_rate": 0.002828,
      "loss": 1.2092,
      "step": 371840
    },
    {
      "epoch": 27.910243902439024,
      "grad_norm": 0.6245830655097961,
      "learning_rate": 0.002828,
      "loss": 1.2167,
      "step": 371904
    },
    {
      "epoch": 27.915046904315197,
      "grad_norm": 0.548487663269043,
      "learning_rate": 0.002828,
      "loss": 1.2126,
      "step": 371968
    },
    {
      "epoch": 27.91984990619137,
      "grad_norm": 0.5911235809326172,
      "learning_rate": 0.002828,
      "loss": 1.2155,
      "step": 372032
    },
    {
      "epoch": 27.924652908067543,
      "grad_norm": 0.5632354021072388,
      "learning_rate": 0.002828,
      "loss": 1.2134,
      "step": 372096
    },
    {
      "epoch": 27.929455909943716,
      "grad_norm": 0.6129593849182129,
      "learning_rate": 0.002828,
      "loss": 1.2135,
      "step": 372160
    },
    {
      "epoch": 27.93425891181989,
      "grad_norm": 0.6318002939224243,
      "learning_rate": 0.002828,
      "loss": 1.2128,
      "step": 372224
    },
    {
      "epoch": 27.93906191369606,
      "grad_norm": 0.5950889587402344,
      "learning_rate": 0.002828,
      "loss": 1.2122,
      "step": 372288
    },
    {
      "epoch": 27.94386491557223,
      "grad_norm": 0.6795059442520142,
      "learning_rate": 0.002828,
      "loss": 1.2124,
      "step": 372352
    },
    {
      "epoch": 27.948667917448404,
      "grad_norm": 0.5208553075790405,
      "learning_rate": 0.002828,
      "loss": 1.2081,
      "step": 372416
    },
    {
      "epoch": 27.953470919324577,
      "grad_norm": 0.6146217584609985,
      "learning_rate": 0.002828,
      "loss": 1.2164,
      "step": 372480
    },
    {
      "epoch": 27.95827392120075,
      "grad_norm": 0.5160702466964722,
      "learning_rate": 0.002828,
      "loss": 1.2105,
      "step": 372544
    },
    {
      "epoch": 27.963076923076922,
      "grad_norm": 0.5801197290420532,
      "learning_rate": 0.002828,
      "loss": 1.2157,
      "step": 372608
    },
    {
      "epoch": 27.967879924953095,
      "grad_norm": 0.6412590146064758,
      "learning_rate": 0.002828,
      "loss": 1.2167,
      "step": 372672
    },
    {
      "epoch": 27.97268292682927,
      "grad_norm": 0.5855358242988586,
      "learning_rate": 0.002828,
      "loss": 1.2114,
      "step": 372736
    },
    {
      "epoch": 27.97748592870544,
      "grad_norm": 0.6856404542922974,
      "learning_rate": 0.002828,
      "loss": 1.2062,
      "step": 372800
    },
    {
      "epoch": 27.982288930581614,
      "grad_norm": 0.5524616837501526,
      "learning_rate": 0.002828,
      "loss": 1.2135,
      "step": 372864
    },
    {
      "epoch": 27.987091932457787,
      "grad_norm": 0.5954063534736633,
      "learning_rate": 0.002828,
      "loss": 1.2121,
      "step": 372928
    },
    {
      "epoch": 27.99189493433396,
      "grad_norm": 0.6046544909477234,
      "learning_rate": 0.002828,
      "loss": 1.2097,
      "step": 372992
    },
    {
      "epoch": 27.996697936210133,
      "grad_norm": 0.6298834681510925,
      "learning_rate": 0.002828,
      "loss": 1.2152,
      "step": 373056
    },
    {
      "epoch": 28.001500938086306,
      "grad_norm": 0.617692232131958,
      "learning_rate": 0.002828,
      "loss": 1.1993,
      "step": 373120
    },
    {
      "epoch": 28.006303939962475,
      "grad_norm": 0.5413621068000793,
      "learning_rate": 0.002828,
      "loss": 1.1748,
      "step": 373184
    },
    {
      "epoch": 28.011106941838648,
      "grad_norm": 0.6742262840270996,
      "learning_rate": 0.002828,
      "loss": 1.1814,
      "step": 373248
    },
    {
      "epoch": 28.01590994371482,
      "grad_norm": 0.5467806458473206,
      "learning_rate": 0.002828,
      "loss": 1.1843,
      "step": 373312
    },
    {
      "epoch": 28.020712945590994,
      "grad_norm": 0.578707754611969,
      "learning_rate": 0.002828,
      "loss": 1.1711,
      "step": 373376
    },
    {
      "epoch": 28.025515947467166,
      "grad_norm": 0.5677569508552551,
      "learning_rate": 0.002828,
      "loss": 1.1773,
      "step": 373440
    },
    {
      "epoch": 28.03031894934334,
      "grad_norm": 0.5994061231613159,
      "learning_rate": 0.002828,
      "loss": 1.1742,
      "step": 373504
    },
    {
      "epoch": 28.035121951219512,
      "grad_norm": 0.5310932993888855,
      "learning_rate": 0.002828,
      "loss": 1.1836,
      "step": 373568
    },
    {
      "epoch": 28.039924953095685,
      "grad_norm": 0.7003358006477356,
      "learning_rate": 0.002828,
      "loss": 1.1692,
      "step": 373632
    },
    {
      "epoch": 28.044727954971858,
      "grad_norm": 0.6028397083282471,
      "learning_rate": 0.002828,
      "loss": 1.1781,
      "step": 373696
    },
    {
      "epoch": 28.04953095684803,
      "grad_norm": 0.6095808148384094,
      "learning_rate": 0.002828,
      "loss": 1.1811,
      "step": 373760
    },
    {
      "epoch": 28.054333958724204,
      "grad_norm": 0.49688461422920227,
      "learning_rate": 0.002828,
      "loss": 1.1813,
      "step": 373824
    },
    {
      "epoch": 28.059136960600377,
      "grad_norm": 0.7206149697303772,
      "learning_rate": 0.002828,
      "loss": 1.1772,
      "step": 373888
    },
    {
      "epoch": 28.06393996247655,
      "grad_norm": 0.5841058492660522,
      "learning_rate": 0.002828,
      "loss": 1.1827,
      "step": 373952
    },
    {
      "epoch": 28.06874296435272,
      "grad_norm": 0.7009063363075256,
      "learning_rate": 0.002828,
      "loss": 1.1806,
      "step": 374016
    },
    {
      "epoch": 28.07354596622889,
      "grad_norm": 0.562809407711029,
      "learning_rate": 0.002828,
      "loss": 1.1794,
      "step": 374080
    },
    {
      "epoch": 28.078348968105065,
      "grad_norm": 0.6682745218276978,
      "learning_rate": 0.002828,
      "loss": 1.1766,
      "step": 374144
    },
    {
      "epoch": 28.083151969981238,
      "grad_norm": 0.6088096499443054,
      "learning_rate": 0.002828,
      "loss": 1.1808,
      "step": 374208
    },
    {
      "epoch": 28.08795497185741,
      "grad_norm": 0.640619695186615,
      "learning_rate": 0.002828,
      "loss": 1.1807,
      "step": 374272
    },
    {
      "epoch": 28.092757973733583,
      "grad_norm": 0.6709964871406555,
      "learning_rate": 0.002828,
      "loss": 1.1827,
      "step": 374336
    },
    {
      "epoch": 28.097560975609756,
      "grad_norm": 0.7114203572273254,
      "learning_rate": 0.002828,
      "loss": 1.1807,
      "step": 374400
    },
    {
      "epoch": 28.10236397748593,
      "grad_norm": 0.5254417657852173,
      "learning_rate": 0.002828,
      "loss": 1.1875,
      "step": 374464
    },
    {
      "epoch": 28.107166979362102,
      "grad_norm": 0.5190957188606262,
      "learning_rate": 0.002828,
      "loss": 1.1773,
      "step": 374528
    },
    {
      "epoch": 28.111969981238275,
      "grad_norm": 0.6001734137535095,
      "learning_rate": 0.002828,
      "loss": 1.1761,
      "step": 374592
    },
    {
      "epoch": 28.116772983114448,
      "grad_norm": 0.7337915897369385,
      "learning_rate": 0.002828,
      "loss": 1.1826,
      "step": 374656
    },
    {
      "epoch": 28.12157598499062,
      "grad_norm": 0.5772360563278198,
      "learning_rate": 0.002828,
      "loss": 1.1814,
      "step": 374720
    },
    {
      "epoch": 28.126378986866793,
      "grad_norm": 0.8065518140792847,
      "learning_rate": 0.002828,
      "loss": 1.1846,
      "step": 374784
    },
    {
      "epoch": 28.131181988742963,
      "grad_norm": 0.7258828282356262,
      "learning_rate": 0.002828,
      "loss": 1.1804,
      "step": 374848
    },
    {
      "epoch": 28.135984990619136,
      "grad_norm": 0.624515950679779,
      "learning_rate": 0.002828,
      "loss": 1.1781,
      "step": 374912
    },
    {
      "epoch": 28.14078799249531,
      "grad_norm": 0.611997127532959,
      "learning_rate": 0.002828,
      "loss": 1.1832,
      "step": 374976
    },
    {
      "epoch": 28.14559099437148,
      "grad_norm": 0.5260868668556213,
      "learning_rate": 0.002828,
      "loss": 1.1804,
      "step": 375040
    },
    {
      "epoch": 28.150393996247654,
      "grad_norm": 0.6274808049201965,
      "learning_rate": 0.002828,
      "loss": 1.185,
      "step": 375104
    },
    {
      "epoch": 28.155196998123827,
      "grad_norm": 0.6832979917526245,
      "learning_rate": 0.002828,
      "loss": 1.1808,
      "step": 375168
    },
    {
      "epoch": 28.16,
      "grad_norm": 0.5438277125358582,
      "learning_rate": 0.002828,
      "loss": 1.1827,
      "step": 375232
    },
    {
      "epoch": 28.164803001876173,
      "grad_norm": 0.5315305590629578,
      "learning_rate": 0.002828,
      "loss": 1.1837,
      "step": 375296
    },
    {
      "epoch": 28.169606003752346,
      "grad_norm": 0.5696422457695007,
      "learning_rate": 0.002828,
      "loss": 1.1862,
      "step": 375360
    },
    {
      "epoch": 28.17440900562852,
      "grad_norm": 0.5869351625442505,
      "learning_rate": 0.002828,
      "loss": 1.185,
      "step": 375424
    },
    {
      "epoch": 28.17921200750469,
      "grad_norm": 0.5588424205780029,
      "learning_rate": 0.002828,
      "loss": 1.1894,
      "step": 375488
    },
    {
      "epoch": 28.184015009380865,
      "grad_norm": 0.512924313545227,
      "learning_rate": 0.002828,
      "loss": 1.1883,
      "step": 375552
    },
    {
      "epoch": 28.188818011257034,
      "grad_norm": 0.6198523044586182,
      "learning_rate": 0.002828,
      "loss": 1.1939,
      "step": 375616
    },
    {
      "epoch": 28.193621013133207,
      "grad_norm": 0.766189455986023,
      "learning_rate": 0.002828,
      "loss": 1.1935,
      "step": 375680
    },
    {
      "epoch": 28.19842401500938,
      "grad_norm": 0.48696544766426086,
      "learning_rate": 0.002828,
      "loss": 1.1843,
      "step": 375744
    },
    {
      "epoch": 28.203227016885553,
      "grad_norm": 0.5477736592292786,
      "learning_rate": 0.002828,
      "loss": 1.1829,
      "step": 375808
    },
    {
      "epoch": 28.208030018761725,
      "grad_norm": 0.6221043467521667,
      "learning_rate": 0.002828,
      "loss": 1.1885,
      "step": 375872
    },
    {
      "epoch": 28.2128330206379,
      "grad_norm": 0.5639182329177856,
      "learning_rate": 0.002828,
      "loss": 1.1826,
      "step": 375936
    },
    {
      "epoch": 28.21763602251407,
      "grad_norm": 0.6532436609268188,
      "learning_rate": 0.002828,
      "loss": 1.1884,
      "step": 376000
    },
    {
      "epoch": 28.222439024390244,
      "grad_norm": 0.5616109371185303,
      "learning_rate": 0.002828,
      "loss": 1.1903,
      "step": 376064
    },
    {
      "epoch": 28.227242026266417,
      "grad_norm": 0.600661039352417,
      "learning_rate": 0.002828,
      "loss": 1.1866,
      "step": 376128
    },
    {
      "epoch": 28.23204502814259,
      "grad_norm": 0.652381420135498,
      "learning_rate": 0.002828,
      "loss": 1.1857,
      "step": 376192
    },
    {
      "epoch": 28.236848030018763,
      "grad_norm": 0.7465491890907288,
      "learning_rate": 0.002828,
      "loss": 1.188,
      "step": 376256
    },
    {
      "epoch": 28.241651031894936,
      "grad_norm": 0.6826187372207642,
      "learning_rate": 0.002828,
      "loss": 1.189,
      "step": 376320
    },
    {
      "epoch": 28.24645403377111,
      "grad_norm": 0.5452209711074829,
      "learning_rate": 0.002828,
      "loss": 1.185,
      "step": 376384
    },
    {
      "epoch": 28.251257035647278,
      "grad_norm": 0.6479519605636597,
      "learning_rate": 0.002828,
      "loss": 1.1894,
      "step": 376448
    },
    {
      "epoch": 28.25606003752345,
      "grad_norm": 0.7790426015853882,
      "learning_rate": 0.002828,
      "loss": 1.1889,
      "step": 376512
    },
    {
      "epoch": 28.260863039399624,
      "grad_norm": 0.5609104633331299,
      "learning_rate": 0.002828,
      "loss": 1.1846,
      "step": 376576
    },
    {
      "epoch": 28.265666041275797,
      "grad_norm": 0.612287700176239,
      "learning_rate": 0.002828,
      "loss": 1.1909,
      "step": 376640
    },
    {
      "epoch": 28.27046904315197,
      "grad_norm": 0.5369693636894226,
      "learning_rate": 0.002828,
      "loss": 1.1915,
      "step": 376704
    },
    {
      "epoch": 28.275272045028142,
      "grad_norm": 0.5666752457618713,
      "learning_rate": 0.002828,
      "loss": 1.1853,
      "step": 376768
    },
    {
      "epoch": 28.280075046904315,
      "grad_norm": 0.5566587448120117,
      "learning_rate": 0.002828,
      "loss": 1.1899,
      "step": 376832
    },
    {
      "epoch": 28.284878048780488,
      "grad_norm": 0.6081478595733643,
      "learning_rate": 0.002828,
      "loss": 1.1903,
      "step": 376896
    },
    {
      "epoch": 28.28968105065666,
      "grad_norm": 0.5504257082939148,
      "learning_rate": 0.002828,
      "loss": 1.195,
      "step": 376960
    },
    {
      "epoch": 28.294484052532834,
      "grad_norm": 0.5835635662078857,
      "learning_rate": 0.002828,
      "loss": 1.1859,
      "step": 377024
    },
    {
      "epoch": 28.299287054409007,
      "grad_norm": 0.5684508681297302,
      "learning_rate": 0.002828,
      "loss": 1.1919,
      "step": 377088
    },
    {
      "epoch": 28.30409005628518,
      "grad_norm": 0.6690579056739807,
      "learning_rate": 0.002828,
      "loss": 1.1893,
      "step": 377152
    },
    {
      "epoch": 28.308893058161352,
      "grad_norm": 0.6685102581977844,
      "learning_rate": 0.002828,
      "loss": 1.1895,
      "step": 377216
    },
    {
      "epoch": 28.313696060037522,
      "grad_norm": 0.5561646819114685,
      "learning_rate": 0.002828,
      "loss": 1.1918,
      "step": 377280
    },
    {
      "epoch": 28.318499061913695,
      "grad_norm": 0.47592681646347046,
      "learning_rate": 0.002828,
      "loss": 1.1851,
      "step": 377344
    },
    {
      "epoch": 28.323302063789868,
      "grad_norm": 0.5404749512672424,
      "learning_rate": 0.002828,
      "loss": 1.1914,
      "step": 377408
    },
    {
      "epoch": 28.32810506566604,
      "grad_norm": 0.5822522640228271,
      "learning_rate": 0.002828,
      "loss": 1.1938,
      "step": 377472
    },
    {
      "epoch": 28.332908067542213,
      "grad_norm": 0.5612675547599792,
      "learning_rate": 0.002828,
      "loss": 1.19,
      "step": 377536
    },
    {
      "epoch": 28.337711069418386,
      "grad_norm": 0.5781949758529663,
      "learning_rate": 0.002828,
      "loss": 1.1933,
      "step": 377600
    },
    {
      "epoch": 28.34251407129456,
      "grad_norm": 0.6771215200424194,
      "learning_rate": 0.002828,
      "loss": 1.1916,
      "step": 377664
    },
    {
      "epoch": 28.347317073170732,
      "grad_norm": 0.5959840416908264,
      "learning_rate": 0.002828,
      "loss": 1.1925,
      "step": 377728
    },
    {
      "epoch": 28.352120075046905,
      "grad_norm": 0.5531420111656189,
      "learning_rate": 0.002828,
      "loss": 1.1904,
      "step": 377792
    },
    {
      "epoch": 28.356923076923078,
      "grad_norm": 0.638388454914093,
      "learning_rate": 0.002828,
      "loss": 1.187,
      "step": 377856
    },
    {
      "epoch": 28.36172607879925,
      "grad_norm": 0.6081584095954895,
      "learning_rate": 0.002828,
      "loss": 1.184,
      "step": 377920
    },
    {
      "epoch": 28.366529080675424,
      "grad_norm": 0.6064649820327759,
      "learning_rate": 0.002828,
      "loss": 1.1913,
      "step": 377984
    },
    {
      "epoch": 28.371332082551596,
      "grad_norm": 0.48571982979774475,
      "learning_rate": 0.002828,
      "loss": 1.1862,
      "step": 378048
    },
    {
      "epoch": 28.376135084427766,
      "grad_norm": 0.5620942115783691,
      "learning_rate": 0.002828,
      "loss": 1.1951,
      "step": 378112
    },
    {
      "epoch": 28.38093808630394,
      "grad_norm": 0.674046516418457,
      "learning_rate": 0.002828,
      "loss": 1.1904,
      "step": 378176
    },
    {
      "epoch": 28.38574108818011,
      "grad_norm": 0.5276687145233154,
      "learning_rate": 0.002828,
      "loss": 1.1923,
      "step": 378240
    },
    {
      "epoch": 28.390544090056284,
      "grad_norm": 0.5791585445404053,
      "learning_rate": 0.002828,
      "loss": 1.1916,
      "step": 378304
    },
    {
      "epoch": 28.395347091932457,
      "grad_norm": 0.6942904591560364,
      "learning_rate": 0.002828,
      "loss": 1.1876,
      "step": 378368
    },
    {
      "epoch": 28.40015009380863,
      "grad_norm": 0.5368022918701172,
      "learning_rate": 0.002828,
      "loss": 1.1934,
      "step": 378432
    },
    {
      "epoch": 28.404953095684803,
      "grad_norm": 0.5756661295890808,
      "learning_rate": 0.002828,
      "loss": 1.1904,
      "step": 378496
    },
    {
      "epoch": 28.409756097560976,
      "grad_norm": 0.5052250027656555,
      "learning_rate": 0.002828,
      "loss": 1.1951,
      "step": 378560
    },
    {
      "epoch": 28.41455909943715,
      "grad_norm": 0.5766377449035645,
      "learning_rate": 0.002828,
      "loss": 1.1868,
      "step": 378624
    },
    {
      "epoch": 28.41936210131332,
      "grad_norm": 0.5014985799789429,
      "learning_rate": 0.002828,
      "loss": 1.193,
      "step": 378688
    },
    {
      "epoch": 28.424165103189495,
      "grad_norm": 0.5464946627616882,
      "learning_rate": 0.002828,
      "loss": 1.1926,
      "step": 378752
    },
    {
      "epoch": 28.428968105065668,
      "grad_norm": 0.7980459928512573,
      "learning_rate": 0.002828,
      "loss": 1.1949,
      "step": 378816
    },
    {
      "epoch": 28.43377110694184,
      "grad_norm": 0.5736128687858582,
      "learning_rate": 0.002828,
      "loss": 1.1987,
      "step": 378880
    },
    {
      "epoch": 28.43857410881801,
      "grad_norm": 0.6616801619529724,
      "learning_rate": 0.002828,
      "loss": 1.1913,
      "step": 378944
    },
    {
      "epoch": 28.443377110694183,
      "grad_norm": 0.587303638458252,
      "learning_rate": 0.002828,
      "loss": 1.1915,
      "step": 379008
    },
    {
      "epoch": 28.448180112570356,
      "grad_norm": 0.6374856233596802,
      "learning_rate": 0.002828,
      "loss": 1.1945,
      "step": 379072
    },
    {
      "epoch": 28.45298311444653,
      "grad_norm": 0.5934728384017944,
      "learning_rate": 0.002828,
      "loss": 1.1957,
      "step": 379136
    },
    {
      "epoch": 28.4577861163227,
      "grad_norm": 0.6261712312698364,
      "learning_rate": 0.002828,
      "loss": 1.187,
      "step": 379200
    },
    {
      "epoch": 28.462589118198874,
      "grad_norm": 0.6582685708999634,
      "learning_rate": 0.002828,
      "loss": 1.1984,
      "step": 379264
    },
    {
      "epoch": 28.467392120075047,
      "grad_norm": 0.6693552732467651,
      "learning_rate": 0.002828,
      "loss": 1.1932,
      "step": 379328
    },
    {
      "epoch": 28.47219512195122,
      "grad_norm": 0.5348833799362183,
      "learning_rate": 0.002828,
      "loss": 1.1854,
      "step": 379392
    },
    {
      "epoch": 28.476998123827393,
      "grad_norm": 0.5958510637283325,
      "learning_rate": 0.002828,
      "loss": 1.1896,
      "step": 379456
    },
    {
      "epoch": 28.481801125703566,
      "grad_norm": 0.5670205354690552,
      "learning_rate": 0.002828,
      "loss": 1.1949,
      "step": 379520
    },
    {
      "epoch": 28.48660412757974,
      "grad_norm": 0.5278819799423218,
      "learning_rate": 0.002828,
      "loss": 1.196,
      "step": 379584
    },
    {
      "epoch": 28.49140712945591,
      "grad_norm": 0.5390208959579468,
      "learning_rate": 0.002828,
      "loss": 1.1921,
      "step": 379648
    },
    {
      "epoch": 28.49621013133208,
      "grad_norm": 0.5674098134040833,
      "learning_rate": 0.002828,
      "loss": 1.1896,
      "step": 379712
    },
    {
      "epoch": 28.501013133208254,
      "grad_norm": 0.5298190116882324,
      "learning_rate": 0.002828,
      "loss": 1.1914,
      "step": 379776
    },
    {
      "epoch": 28.505816135084427,
      "grad_norm": 0.6678189039230347,
      "learning_rate": 0.002828,
      "loss": 1.1932,
      "step": 379840
    },
    {
      "epoch": 28.5106191369606,
      "grad_norm": 0.5374751091003418,
      "learning_rate": 0.002828,
      "loss": 1.1878,
      "step": 379904
    },
    {
      "epoch": 28.515422138836772,
      "grad_norm": 0.5720387697219849,
      "learning_rate": 0.002828,
      "loss": 1.1934,
      "step": 379968
    },
    {
      "epoch": 28.520225140712945,
      "grad_norm": 0.6150330901145935,
      "learning_rate": 0.002828,
      "loss": 1.1957,
      "step": 380032
    },
    {
      "epoch": 28.525028142589118,
      "grad_norm": 0.6207433938980103,
      "learning_rate": 0.002828,
      "loss": 1.1949,
      "step": 380096
    },
    {
      "epoch": 28.52983114446529,
      "grad_norm": 0.7584127187728882,
      "learning_rate": 0.002828,
      "loss": 1.1899,
      "step": 380160
    },
    {
      "epoch": 28.534634146341464,
      "grad_norm": 0.5574619174003601,
      "learning_rate": 0.002828,
      "loss": 1.197,
      "step": 380224
    },
    {
      "epoch": 28.539437148217637,
      "grad_norm": 0.6886953115463257,
      "learning_rate": 0.002828,
      "loss": 1.189,
      "step": 380288
    },
    {
      "epoch": 28.54424015009381,
      "grad_norm": 0.6965809464454651,
      "learning_rate": 0.002828,
      "loss": 1.1957,
      "step": 380352
    },
    {
      "epoch": 28.549043151969983,
      "grad_norm": 0.6613625288009644,
      "learning_rate": 0.002828,
      "loss": 1.2169,
      "step": 380416
    },
    {
      "epoch": 28.553846153846155,
      "grad_norm": 0.7155604958534241,
      "learning_rate": 0.002828,
      "loss": 1.199,
      "step": 380480
    },
    {
      "epoch": 28.558649155722325,
      "grad_norm": 0.7222745418548584,
      "learning_rate": 0.002828,
      "loss": 1.1983,
      "step": 380544
    },
    {
      "epoch": 28.563452157598498,
      "grad_norm": 0.5591253042221069,
      "learning_rate": 0.002828,
      "loss": 1.202,
      "step": 380608
    },
    {
      "epoch": 28.56825515947467,
      "grad_norm": 0.7596259117126465,
      "learning_rate": 0.002828,
      "loss": 1.2022,
      "step": 380672
    },
    {
      "epoch": 28.573058161350843,
      "grad_norm": 0.5827973484992981,
      "learning_rate": 0.002828,
      "loss": 1.1981,
      "step": 380736
    },
    {
      "epoch": 28.577861163227016,
      "grad_norm": 0.6316086649894714,
      "learning_rate": 0.002828,
      "loss": 1.1995,
      "step": 380800
    },
    {
      "epoch": 28.58266416510319,
      "grad_norm": 0.5625211000442505,
      "learning_rate": 0.002828,
      "loss": 1.1939,
      "step": 380864
    },
    {
      "epoch": 28.587467166979362,
      "grad_norm": 0.6687682271003723,
      "learning_rate": 0.002828,
      "loss": 1.1999,
      "step": 380928
    },
    {
      "epoch": 28.592270168855535,
      "grad_norm": 0.6526170969009399,
      "learning_rate": 0.002828,
      "loss": 1.1969,
      "step": 380992
    },
    {
      "epoch": 28.597073170731708,
      "grad_norm": 0.6328181028366089,
      "learning_rate": 0.002828,
      "loss": 1.1953,
      "step": 381056
    },
    {
      "epoch": 28.60187617260788,
      "grad_norm": 0.8068972826004028,
      "learning_rate": 0.002828,
      "loss": 1.1942,
      "step": 381120
    },
    {
      "epoch": 28.606679174484054,
      "grad_norm": 0.45512983202934265,
      "learning_rate": 0.002828,
      "loss": 1.2051,
      "step": 381184
    },
    {
      "epoch": 28.611482176360227,
      "grad_norm": 0.5556621551513672,
      "learning_rate": 0.002828,
      "loss": 1.1977,
      "step": 381248
    },
    {
      "epoch": 28.6162851782364,
      "grad_norm": 0.6210782527923584,
      "learning_rate": 0.002828,
      "loss": 1.1963,
      "step": 381312
    },
    {
      "epoch": 28.62108818011257,
      "grad_norm": 0.5571366548538208,
      "learning_rate": 0.002828,
      "loss": 1.1948,
      "step": 381376
    },
    {
      "epoch": 28.62589118198874,
      "grad_norm": 0.6097261309623718,
      "learning_rate": 0.002828,
      "loss": 1.1938,
      "step": 381440
    },
    {
      "epoch": 28.630694183864914,
      "grad_norm": 0.5357059240341187,
      "learning_rate": 0.002828,
      "loss": 1.2026,
      "step": 381504
    },
    {
      "epoch": 28.635497185741087,
      "grad_norm": 0.5957450866699219,
      "learning_rate": 0.002828,
      "loss": 1.1954,
      "step": 381568
    },
    {
      "epoch": 28.64030018761726,
      "grad_norm": 0.6477972269058228,
      "learning_rate": 0.002828,
      "loss": 1.1953,
      "step": 381632
    },
    {
      "epoch": 28.645103189493433,
      "grad_norm": 0.5469118356704712,
      "learning_rate": 0.002828,
      "loss": 1.1959,
      "step": 381696
    },
    {
      "epoch": 28.649906191369606,
      "grad_norm": 0.5682870745658875,
      "learning_rate": 0.002828,
      "loss": 1.1966,
      "step": 381760
    },
    {
      "epoch": 28.65470919324578,
      "grad_norm": 0.5928640365600586,
      "learning_rate": 0.002828,
      "loss": 1.1945,
      "step": 381824
    },
    {
      "epoch": 28.659512195121952,
      "grad_norm": 0.6658616662025452,
      "learning_rate": 0.002828,
      "loss": 1.1969,
      "step": 381888
    },
    {
      "epoch": 28.664315196998125,
      "grad_norm": 0.733881413936615,
      "learning_rate": 0.002828,
      "loss": 1.2032,
      "step": 381952
    },
    {
      "epoch": 28.669118198874298,
      "grad_norm": 0.5383415222167969,
      "learning_rate": 0.002828,
      "loss": 1.1996,
      "step": 382016
    },
    {
      "epoch": 28.67392120075047,
      "grad_norm": 0.6444189548492432,
      "learning_rate": 0.002828,
      "loss": 1.1998,
      "step": 382080
    },
    {
      "epoch": 28.678724202626643,
      "grad_norm": 0.5679410696029663,
      "learning_rate": 0.002828,
      "loss": 1.2054,
      "step": 382144
    },
    {
      "epoch": 28.683527204502813,
      "grad_norm": 0.7500342726707458,
      "learning_rate": 0.002828,
      "loss": 1.1948,
      "step": 382208
    },
    {
      "epoch": 28.688330206378986,
      "grad_norm": 0.5816766619682312,
      "learning_rate": 0.002828,
      "loss": 1.1937,
      "step": 382272
    },
    {
      "epoch": 28.69313320825516,
      "grad_norm": 0.7372063398361206,
      "learning_rate": 0.002828,
      "loss": 1.2074,
      "step": 382336
    },
    {
      "epoch": 28.69793621013133,
      "grad_norm": 0.5986203551292419,
      "learning_rate": 0.002828,
      "loss": 1.1959,
      "step": 382400
    },
    {
      "epoch": 28.702739212007504,
      "grad_norm": 0.7330045104026794,
      "learning_rate": 0.002828,
      "loss": 1.1967,
      "step": 382464
    },
    {
      "epoch": 28.707542213883677,
      "grad_norm": 0.5793071985244751,
      "learning_rate": 0.002828,
      "loss": 1.1961,
      "step": 382528
    },
    {
      "epoch": 28.71234521575985,
      "grad_norm": 0.6789469718933105,
      "learning_rate": 0.002828,
      "loss": 1.1986,
      "step": 382592
    },
    {
      "epoch": 28.717148217636023,
      "grad_norm": 0.6991356015205383,
      "learning_rate": 0.002828,
      "loss": 1.2007,
      "step": 382656
    },
    {
      "epoch": 28.721951219512196,
      "grad_norm": 0.4840135872364044,
      "learning_rate": 0.002828,
      "loss": 1.1987,
      "step": 382720
    },
    {
      "epoch": 28.72675422138837,
      "grad_norm": 0.5444275736808777,
      "learning_rate": 0.002828,
      "loss": 1.1973,
      "step": 382784
    },
    {
      "epoch": 28.73155722326454,
      "grad_norm": 0.6061873435974121,
      "learning_rate": 0.002828,
      "loss": 1.1999,
      "step": 382848
    },
    {
      "epoch": 28.736360225140714,
      "grad_norm": 0.5796545743942261,
      "learning_rate": 0.002828,
      "loss": 1.1985,
      "step": 382912
    },
    {
      "epoch": 28.741163227016884,
      "grad_norm": 0.5784839391708374,
      "learning_rate": 0.002828,
      "loss": 1.1993,
      "step": 382976
    },
    {
      "epoch": 28.745966228893057,
      "grad_norm": 0.6885812282562256,
      "learning_rate": 0.002828,
      "loss": 1.1995,
      "step": 383040
    },
    {
      "epoch": 28.75076923076923,
      "grad_norm": 0.6657517552375793,
      "learning_rate": 0.002828,
      "loss": 1.1992,
      "step": 383104
    },
    {
      "epoch": 28.755572232645402,
      "grad_norm": 0.5068328380584717,
      "learning_rate": 0.002828,
      "loss": 1.1941,
      "step": 383168
    },
    {
      "epoch": 28.760375234521575,
      "grad_norm": 0.5651753544807434,
      "learning_rate": 0.002828,
      "loss": 1.2008,
      "step": 383232
    },
    {
      "epoch": 28.765178236397748,
      "grad_norm": 0.5559189319610596,
      "learning_rate": 0.002828,
      "loss": 1.1995,
      "step": 383296
    },
    {
      "epoch": 28.76998123827392,
      "grad_norm": 0.6028053760528564,
      "learning_rate": 0.002828,
      "loss": 1.1993,
      "step": 383360
    },
    {
      "epoch": 28.774784240150094,
      "grad_norm": 0.6071565747261047,
      "learning_rate": 0.002828,
      "loss": 1.1979,
      "step": 383424
    },
    {
      "epoch": 28.779587242026267,
      "grad_norm": 0.6035954356193542,
      "learning_rate": 0.002828,
      "loss": 1.1982,
      "step": 383488
    },
    {
      "epoch": 28.78439024390244,
      "grad_norm": 0.6009169816970825,
      "learning_rate": 0.002828,
      "loss": 1.2044,
      "step": 383552
    },
    {
      "epoch": 28.789193245778613,
      "grad_norm": 0.5744941830635071,
      "learning_rate": 0.002828,
      "loss": 1.2034,
      "step": 383616
    },
    {
      "epoch": 28.793996247654785,
      "grad_norm": 0.748096764087677,
      "learning_rate": 0.002828,
      "loss": 1.2062,
      "step": 383680
    },
    {
      "epoch": 28.79879924953096,
      "grad_norm": 0.5846696496009827,
      "learning_rate": 0.002828,
      "loss": 1.2016,
      "step": 383744
    },
    {
      "epoch": 28.803602251407128,
      "grad_norm": 0.6742605566978455,
      "learning_rate": 0.002828,
      "loss": 1.1999,
      "step": 383808
    },
    {
      "epoch": 28.8084052532833,
      "grad_norm": 0.5580283999443054,
      "learning_rate": 0.002828,
      "loss": 1.1971,
      "step": 383872
    },
    {
      "epoch": 28.813208255159473,
      "grad_norm": 0.6639071106910706,
      "learning_rate": 0.002828,
      "loss": 1.2022,
      "step": 383936
    },
    {
      "epoch": 28.818011257035646,
      "grad_norm": 0.5421018600463867,
      "learning_rate": 0.002828,
      "loss": 1.2019,
      "step": 384000
    },
    {
      "epoch": 28.82281425891182,
      "grad_norm": 0.6587681174278259,
      "learning_rate": 0.002828,
      "loss": 1.2022,
      "step": 384064
    },
    {
      "epoch": 28.827617260787992,
      "grad_norm": 0.5655040144920349,
      "learning_rate": 0.002828,
      "loss": 1.202,
      "step": 384128
    },
    {
      "epoch": 28.832420262664165,
      "grad_norm": 0.6196697354316711,
      "learning_rate": 0.002828,
      "loss": 1.2007,
      "step": 384192
    },
    {
      "epoch": 28.837223264540338,
      "grad_norm": 0.716361939907074,
      "learning_rate": 0.002828,
      "loss": 1.192,
      "step": 384256
    },
    {
      "epoch": 28.84202626641651,
      "grad_norm": 0.5832474827766418,
      "learning_rate": 0.002828,
      "loss": 1.1963,
      "step": 384320
    },
    {
      "epoch": 28.846829268292684,
      "grad_norm": 0.5728902220726013,
      "learning_rate": 0.002828,
      "loss": 1.2075,
      "step": 384384
    },
    {
      "epoch": 28.851632270168857,
      "grad_norm": 0.6255494952201843,
      "learning_rate": 0.002828,
      "loss": 1.1992,
      "step": 384448
    },
    {
      "epoch": 28.85643527204503,
      "grad_norm": 0.5721420645713806,
      "learning_rate": 0.002828,
      "loss": 1.2027,
      "step": 384512
    },
    {
      "epoch": 28.861238273921202,
      "grad_norm": 0.594143807888031,
      "learning_rate": 0.002828,
      "loss": 1.2043,
      "step": 384576
    },
    {
      "epoch": 28.86604127579737,
      "grad_norm": 0.6351729035377502,
      "learning_rate": 0.002828,
      "loss": 1.2026,
      "step": 384640
    },
    {
      "epoch": 28.870844277673545,
      "grad_norm": 0.5315951704978943,
      "learning_rate": 0.002828,
      "loss": 1.2042,
      "step": 384704
    },
    {
      "epoch": 28.875647279549717,
      "grad_norm": 0.6394718885421753,
      "learning_rate": 0.002828,
      "loss": 1.2012,
      "step": 384768
    },
    {
      "epoch": 28.88045028142589,
      "grad_norm": 0.6304981708526611,
      "learning_rate": 0.002828,
      "loss": 1.2001,
      "step": 384832
    },
    {
      "epoch": 28.885253283302063,
      "grad_norm": 0.6619015336036682,
      "learning_rate": 0.002828,
      "loss": 1.2032,
      "step": 384896
    },
    {
      "epoch": 28.890056285178236,
      "grad_norm": 0.6189813017845154,
      "learning_rate": 0.002828,
      "loss": 1.2033,
      "step": 384960
    },
    {
      "epoch": 28.89485928705441,
      "grad_norm": 0.5804136395454407,
      "learning_rate": 0.002828,
      "loss": 1.2039,
      "step": 385024
    },
    {
      "epoch": 28.899662288930582,
      "grad_norm": 0.5444638133049011,
      "learning_rate": 0.002828,
      "loss": 1.2017,
      "step": 385088
    },
    {
      "epoch": 28.904465290806755,
      "grad_norm": 0.6017876863479614,
      "learning_rate": 0.002828,
      "loss": 1.2043,
      "step": 385152
    },
    {
      "epoch": 28.909268292682928,
      "grad_norm": 0.5880628824234009,
      "learning_rate": 0.002828,
      "loss": 1.2009,
      "step": 385216
    },
    {
      "epoch": 28.9140712945591,
      "grad_norm": 0.5595766305923462,
      "learning_rate": 0.002828,
      "loss": 1.1918,
      "step": 385280
    },
    {
      "epoch": 28.918874296435273,
      "grad_norm": 0.5918551087379456,
      "learning_rate": 0.002828,
      "loss": 1.2037,
      "step": 385344
    },
    {
      "epoch": 28.923677298311446,
      "grad_norm": 0.5559405088424683,
      "learning_rate": 0.002828,
      "loss": 1.2015,
      "step": 385408
    },
    {
      "epoch": 28.928480300187616,
      "grad_norm": 0.4949721693992615,
      "learning_rate": 0.002828,
      "loss": 1.2027,
      "step": 385472
    },
    {
      "epoch": 28.93328330206379,
      "grad_norm": 0.6168365478515625,
      "learning_rate": 0.002828,
      "loss": 1.2004,
      "step": 385536
    },
    {
      "epoch": 28.93808630393996,
      "grad_norm": 0.7354480624198914,
      "learning_rate": 0.002828,
      "loss": 1.2064,
      "step": 385600
    },
    {
      "epoch": 28.942889305816134,
      "grad_norm": 0.5195580720901489,
      "learning_rate": 0.002828,
      "loss": 1.1999,
      "step": 385664
    },
    {
      "epoch": 28.947692307692307,
      "grad_norm": 0.6821034550666809,
      "learning_rate": 0.002828,
      "loss": 1.2013,
      "step": 385728
    },
    {
      "epoch": 28.95249530956848,
      "grad_norm": 0.5596335530281067,
      "learning_rate": 0.002828,
      "loss": 1.1978,
      "step": 385792
    },
    {
      "epoch": 28.957298311444653,
      "grad_norm": 0.6192792057991028,
      "learning_rate": 0.002828,
      "loss": 1.1977,
      "step": 385856
    },
    {
      "epoch": 28.962101313320826,
      "grad_norm": 0.6030036211013794,
      "learning_rate": 0.002828,
      "loss": 1.2033,
      "step": 385920
    },
    {
      "epoch": 28.966904315197,
      "grad_norm": 0.728481650352478,
      "learning_rate": 0.002828,
      "loss": 1.2045,
      "step": 385984
    },
    {
      "epoch": 28.97170731707317,
      "grad_norm": 0.5132448077201843,
      "learning_rate": 0.002828,
      "loss": 1.1986,
      "step": 386048
    },
    {
      "epoch": 28.976510318949344,
      "grad_norm": 0.6252719163894653,
      "learning_rate": 0.002828,
      "loss": 1.1988,
      "step": 386112
    },
    {
      "epoch": 28.981313320825517,
      "grad_norm": 0.5599966645240784,
      "learning_rate": 0.002828,
      "loss": 1.1967,
      "step": 386176
    },
    {
      "epoch": 28.98611632270169,
      "grad_norm": 0.6355006098747253,
      "learning_rate": 0.002828,
      "loss": 1.205,
      "step": 386240
    },
    {
      "epoch": 28.99091932457786,
      "grad_norm": 0.5937554240226746,
      "learning_rate": 0.002828,
      "loss": 1.206,
      "step": 386304
    },
    {
      "epoch": 28.995722326454032,
      "grad_norm": 0.5503448843955994,
      "learning_rate": 0.002828,
      "loss": 1.2013,
      "step": 386368
    },
    {
      "epoch": 29.000525328330205,
      "grad_norm": 0.6377638578414917,
      "learning_rate": 0.002828,
      "loss": 1.1941,
      "step": 386432
    },
    {
      "epoch": 29.00532833020638,
      "grad_norm": 0.6582196950912476,
      "learning_rate": 0.002828,
      "loss": 1.1697,
      "step": 386496
    },
    {
      "epoch": 29.01013133208255,
      "grad_norm": 0.6100175380706787,
      "learning_rate": 0.002828,
      "loss": 1.1675,
      "step": 386560
    },
    {
      "epoch": 29.014934333958724,
      "grad_norm": 0.5015468001365662,
      "learning_rate": 0.002828,
      "loss": 1.1648,
      "step": 386624
    },
    {
      "epoch": 29.019737335834897,
      "grad_norm": 0.6561148166656494,
      "learning_rate": 0.002828,
      "loss": 1.1636,
      "step": 386688
    },
    {
      "epoch": 29.02454033771107,
      "grad_norm": 0.6683859825134277,
      "learning_rate": 0.002828,
      "loss": 1.1636,
      "step": 386752
    },
    {
      "epoch": 29.029343339587243,
      "grad_norm": 0.5666760206222534,
      "learning_rate": 0.002828,
      "loss": 1.1625,
      "step": 386816
    },
    {
      "epoch": 29.034146341463416,
      "grad_norm": 0.5546724796295166,
      "learning_rate": 0.002828,
      "loss": 1.1639,
      "step": 386880
    },
    {
      "epoch": 29.03894934333959,
      "grad_norm": 0.603721022605896,
      "learning_rate": 0.002828,
      "loss": 1.1633,
      "step": 386944
    },
    {
      "epoch": 29.04375234521576,
      "grad_norm": 0.6631825566291809,
      "learning_rate": 0.002828,
      "loss": 1.1628,
      "step": 387008
    },
    {
      "epoch": 29.048555347091934,
      "grad_norm": 0.575922429561615,
      "learning_rate": 0.002828,
      "loss": 1.1694,
      "step": 387072
    },
    {
      "epoch": 29.053358348968104,
      "grad_norm": 0.562066912651062,
      "learning_rate": 0.002828,
      "loss": 1.173,
      "step": 387136
    },
    {
      "epoch": 29.058161350844276,
      "grad_norm": 0.6301705837249756,
      "learning_rate": 0.002828,
      "loss": 1.1676,
      "step": 387200
    },
    {
      "epoch": 29.06296435272045,
      "grad_norm": 0.6013914942741394,
      "learning_rate": 0.002828,
      "loss": 1.1736,
      "step": 387264
    },
    {
      "epoch": 29.067767354596622,
      "grad_norm": 0.6656597256660461,
      "learning_rate": 0.002828,
      "loss": 1.1721,
      "step": 387328
    },
    {
      "epoch": 29.072570356472795,
      "grad_norm": 0.5685966610908508,
      "learning_rate": 0.002828,
      "loss": 1.1694,
      "step": 387392
    },
    {
      "epoch": 29.077373358348968,
      "grad_norm": 0.5013291835784912,
      "learning_rate": 0.002828,
      "loss": 1.1677,
      "step": 387456
    },
    {
      "epoch": 29.08217636022514,
      "grad_norm": 0.5750505924224854,
      "learning_rate": 0.002828,
      "loss": 1.1678,
      "step": 387520
    },
    {
      "epoch": 29.086979362101314,
      "grad_norm": 0.560761570930481,
      "learning_rate": 0.002828,
      "loss": 1.1659,
      "step": 387584
    },
    {
      "epoch": 29.091782363977487,
      "grad_norm": 0.5667386651039124,
      "learning_rate": 0.002828,
      "loss": 1.1646,
      "step": 387648
    },
    {
      "epoch": 29.09658536585366,
      "grad_norm": 0.5522923469543457,
      "learning_rate": 0.002828,
      "loss": 1.1777,
      "step": 387712
    },
    {
      "epoch": 29.101388367729832,
      "grad_norm": 0.599494457244873,
      "learning_rate": 0.002828,
      "loss": 1.1697,
      "step": 387776
    },
    {
      "epoch": 29.106191369606005,
      "grad_norm": 0.593332052230835,
      "learning_rate": 0.002828,
      "loss": 1.1694,
      "step": 387840
    },
    {
      "epoch": 29.110994371482175,
      "grad_norm": 0.603303849697113,
      "learning_rate": 0.002828,
      "loss": 1.1776,
      "step": 387904
    },
    {
      "epoch": 29.115797373358347,
      "grad_norm": 0.5618759989738464,
      "learning_rate": 0.002828,
      "loss": 1.1654,
      "step": 387968
    },
    {
      "epoch": 29.12060037523452,
      "grad_norm": 0.6696698665618896,
      "learning_rate": 0.002828,
      "loss": 1.1697,
      "step": 388032
    },
    {
      "epoch": 29.125403377110693,
      "grad_norm": 0.5471336841583252,
      "learning_rate": 0.002828,
      "loss": 1.1711,
      "step": 388096
    },
    {
      "epoch": 29.130206378986866,
      "grad_norm": 0.6245209574699402,
      "learning_rate": 0.002828,
      "loss": 1.1734,
      "step": 388160
    },
    {
      "epoch": 29.13500938086304,
      "grad_norm": 0.5719093084335327,
      "learning_rate": 0.002828,
      "loss": 1.1667,
      "step": 388224
    },
    {
      "epoch": 29.139812382739212,
      "grad_norm": 0.6765281558036804,
      "learning_rate": 0.002828,
      "loss": 1.1736,
      "step": 388288
    },
    {
      "epoch": 29.144615384615385,
      "grad_norm": 0.642324686050415,
      "learning_rate": 0.002828,
      "loss": 1.1729,
      "step": 388352
    },
    {
      "epoch": 29.149418386491558,
      "grad_norm": 0.5967279076576233,
      "learning_rate": 0.002828,
      "loss": 1.171,
      "step": 388416
    },
    {
      "epoch": 29.15422138836773,
      "grad_norm": 0.5580810904502869,
      "learning_rate": 0.002828,
      "loss": 1.1735,
      "step": 388480
    },
    {
      "epoch": 29.159024390243903,
      "grad_norm": 0.5501552224159241,
      "learning_rate": 0.002828,
      "loss": 1.1742,
      "step": 388544
    },
    {
      "epoch": 29.163827392120076,
      "grad_norm": 0.4875175356864929,
      "learning_rate": 0.002828,
      "loss": 1.171,
      "step": 388608
    },
    {
      "epoch": 29.16863039399625,
      "grad_norm": 0.4810082018375397,
      "learning_rate": 0.002828,
      "loss": 1.1753,
      "step": 388672
    },
    {
      "epoch": 29.17343339587242,
      "grad_norm": 0.6213483214378357,
      "learning_rate": 0.002828,
      "loss": 1.1722,
      "step": 388736
    },
    {
      "epoch": 29.17823639774859,
      "grad_norm": 0.6881386637687683,
      "learning_rate": 0.002828,
      "loss": 1.1712,
      "step": 388800
    },
    {
      "epoch": 29.183039399624764,
      "grad_norm": 0.5882981419563293,
      "learning_rate": 0.002828,
      "loss": 1.1747,
      "step": 388864
    },
    {
      "epoch": 29.187842401500937,
      "grad_norm": 0.8282897472381592,
      "learning_rate": 0.002828,
      "loss": 1.1726,
      "step": 388928
    },
    {
      "epoch": 29.19264540337711,
      "grad_norm": 0.6687741875648499,
      "learning_rate": 0.002828,
      "loss": 1.1707,
      "step": 388992
    },
    {
      "epoch": 29.197448405253283,
      "grad_norm": 0.6514523029327393,
      "learning_rate": 0.002828,
      "loss": 1.1734,
      "step": 389056
    },
    {
      "epoch": 29.202251407129456,
      "grad_norm": 0.7071826457977295,
      "learning_rate": 0.002828,
      "loss": 1.1726,
      "step": 389120
    },
    {
      "epoch": 29.20705440900563,
      "grad_norm": 0.539300262928009,
      "learning_rate": 0.002828,
      "loss": 1.1787,
      "step": 389184
    },
    {
      "epoch": 29.2118574108818,
      "grad_norm": 0.5657750368118286,
      "learning_rate": 0.002828,
      "loss": 1.1711,
      "step": 389248
    },
    {
      "epoch": 29.216660412757975,
      "grad_norm": 0.6181819438934326,
      "learning_rate": 0.002828,
      "loss": 1.1764,
      "step": 389312
    },
    {
      "epoch": 29.221463414634147,
      "grad_norm": 0.529534101486206,
      "learning_rate": 0.002828,
      "loss": 1.1675,
      "step": 389376
    },
    {
      "epoch": 29.22626641651032,
      "grad_norm": 0.5814141631126404,
      "learning_rate": 0.002828,
      "loss": 1.1776,
      "step": 389440
    },
    {
      "epoch": 29.231069418386493,
      "grad_norm": 0.5800368189811707,
      "learning_rate": 0.002828,
      "loss": 1.1763,
      "step": 389504
    },
    {
      "epoch": 29.235872420262663,
      "grad_norm": 0.591761589050293,
      "learning_rate": 0.002828,
      "loss": 1.1738,
      "step": 389568
    },
    {
      "epoch": 29.240675422138835,
      "grad_norm": 0.5492026805877686,
      "learning_rate": 0.002828,
      "loss": 1.1831,
      "step": 389632
    },
    {
      "epoch": 29.24547842401501,
      "grad_norm": 0.5412047505378723,
      "learning_rate": 0.002828,
      "loss": 1.1769,
      "step": 389696
    },
    {
      "epoch": 29.25028142589118,
      "grad_norm": 0.6620146036148071,
      "learning_rate": 0.002828,
      "loss": 1.1749,
      "step": 389760
    },
    {
      "epoch": 29.255084427767354,
      "grad_norm": 0.5703447461128235,
      "learning_rate": 0.002828,
      "loss": 1.1765,
      "step": 389824
    },
    {
      "epoch": 29.259887429643527,
      "grad_norm": 0.6181512475013733,
      "learning_rate": 0.002828,
      "loss": 1.1729,
      "step": 389888
    },
    {
      "epoch": 29.2646904315197,
      "grad_norm": 0.6570577621459961,
      "learning_rate": 0.002828,
      "loss": 1.1766,
      "step": 389952
    },
    {
      "epoch": 29.269493433395873,
      "grad_norm": 0.58260577917099,
      "learning_rate": 0.002828,
      "loss": 1.1764,
      "step": 390016
    },
    {
      "epoch": 29.274296435272046,
      "grad_norm": 0.5680594444274902,
      "learning_rate": 0.002828,
      "loss": 1.1803,
      "step": 390080
    },
    {
      "epoch": 29.27909943714822,
      "grad_norm": 0.5765116214752197,
      "learning_rate": 0.002828,
      "loss": 1.1812,
      "step": 390144
    },
    {
      "epoch": 29.28390243902439,
      "grad_norm": 0.6817169189453125,
      "learning_rate": 0.002828,
      "loss": 1.1792,
      "step": 390208
    },
    {
      "epoch": 29.288705440900564,
      "grad_norm": 0.5913732647895813,
      "learning_rate": 0.002828,
      "loss": 1.1774,
      "step": 390272
    },
    {
      "epoch": 29.293508442776737,
      "grad_norm": 0.5463665127754211,
      "learning_rate": 0.002828,
      "loss": 1.1787,
      "step": 390336
    },
    {
      "epoch": 29.298311444652906,
      "grad_norm": 0.5897245407104492,
      "learning_rate": 0.002828,
      "loss": 1.1752,
      "step": 390400
    },
    {
      "epoch": 29.30311444652908,
      "grad_norm": 0.5375869870185852,
      "learning_rate": 0.002828,
      "loss": 1.1772,
      "step": 390464
    },
    {
      "epoch": 29.307917448405252,
      "grad_norm": 0.5869905948638916,
      "learning_rate": 0.002828,
      "loss": 1.1808,
      "step": 390528
    },
    {
      "epoch": 29.312720450281425,
      "grad_norm": 0.6224765777587891,
      "learning_rate": 0.002828,
      "loss": 1.1729,
      "step": 390592
    },
    {
      "epoch": 29.317523452157598,
      "grad_norm": 0.5576164126396179,
      "learning_rate": 0.002828,
      "loss": 1.1754,
      "step": 390656
    },
    {
      "epoch": 29.32232645403377,
      "grad_norm": 0.6380188465118408,
      "learning_rate": 0.002828,
      "loss": 1.1786,
      "step": 390720
    },
    {
      "epoch": 29.327129455909944,
      "grad_norm": 0.5919608473777771,
      "learning_rate": 0.002828,
      "loss": 1.1842,
      "step": 390784
    },
    {
      "epoch": 29.331932457786117,
      "grad_norm": 0.5888688564300537,
      "learning_rate": 0.002828,
      "loss": 1.181,
      "step": 390848
    },
    {
      "epoch": 29.33673545966229,
      "grad_norm": 0.5569725632667542,
      "learning_rate": 0.002828,
      "loss": 1.1782,
      "step": 390912
    },
    {
      "epoch": 29.341538461538462,
      "grad_norm": 0.5338151454925537,
      "learning_rate": 0.002828,
      "loss": 1.1805,
      "step": 390976
    },
    {
      "epoch": 29.346341463414635,
      "grad_norm": 0.5496394038200378,
      "learning_rate": 0.002828,
      "loss": 1.1758,
      "step": 391040
    },
    {
      "epoch": 29.35114446529081,
      "grad_norm": 0.646307110786438,
      "learning_rate": 0.002828,
      "loss": 1.1824,
      "step": 391104
    },
    {
      "epoch": 29.355947467166978,
      "grad_norm": 0.5882683992385864,
      "learning_rate": 0.002828,
      "loss": 1.178,
      "step": 391168
    },
    {
      "epoch": 29.36075046904315,
      "grad_norm": 0.5194476842880249,
      "learning_rate": 0.002828,
      "loss": 1.1809,
      "step": 391232
    },
    {
      "epoch": 29.365553470919323,
      "grad_norm": 0.7695441842079163,
      "learning_rate": 0.002828,
      "loss": 1.1834,
      "step": 391296
    },
    {
      "epoch": 29.370356472795496,
      "grad_norm": 0.5390833020210266,
      "learning_rate": 0.002828,
      "loss": 1.1781,
      "step": 391360
    },
    {
      "epoch": 29.37515947467167,
      "grad_norm": 0.5380575656890869,
      "learning_rate": 0.002828,
      "loss": 1.1775,
      "step": 391424
    },
    {
      "epoch": 29.379962476547842,
      "grad_norm": 0.5972825884819031,
      "learning_rate": 0.002828,
      "loss": 1.1859,
      "step": 391488
    },
    {
      "epoch": 29.384765478424015,
      "grad_norm": 0.5862144827842712,
      "learning_rate": 0.002828,
      "loss": 1.1772,
      "step": 391552
    },
    {
      "epoch": 29.389568480300188,
      "grad_norm": 0.5644845962524414,
      "learning_rate": 0.002828,
      "loss": 1.1845,
      "step": 391616
    },
    {
      "epoch": 29.39437148217636,
      "grad_norm": 0.6240797638893127,
      "learning_rate": 0.002828,
      "loss": 1.1843,
      "step": 391680
    },
    {
      "epoch": 29.399174484052534,
      "grad_norm": 0.5433744192123413,
      "learning_rate": 0.002828,
      "loss": 1.1805,
      "step": 391744
    },
    {
      "epoch": 29.403977485928706,
      "grad_norm": 0.6091045141220093,
      "learning_rate": 0.002828,
      "loss": 1.1845,
      "step": 391808
    },
    {
      "epoch": 29.40878048780488,
      "grad_norm": 0.6940107941627502,
      "learning_rate": 0.002828,
      "loss": 1.187,
      "step": 391872
    },
    {
      "epoch": 29.413583489681052,
      "grad_norm": 0.6596461534500122,
      "learning_rate": 0.002828,
      "loss": 1.1863,
      "step": 391936
    },
    {
      "epoch": 29.41838649155722,
      "grad_norm": 0.6700625419616699,
      "learning_rate": 0.002828,
      "loss": 1.1876,
      "step": 392000
    },
    {
      "epoch": 29.423189493433394,
      "grad_norm": 0.5847828388214111,
      "learning_rate": 0.002828,
      "loss": 1.1808,
      "step": 392064
    },
    {
      "epoch": 29.427992495309567,
      "grad_norm": 0.6482068300247192,
      "learning_rate": 0.002828,
      "loss": 1.1774,
      "step": 392128
    },
    {
      "epoch": 29.43279549718574,
      "grad_norm": 0.6086841821670532,
      "learning_rate": 0.002828,
      "loss": 1.1828,
      "step": 392192
    },
    {
      "epoch": 29.437598499061913,
      "grad_norm": 0.606248140335083,
      "learning_rate": 0.002828,
      "loss": 1.1879,
      "step": 392256
    },
    {
      "epoch": 29.442401500938086,
      "grad_norm": 0.62447589635849,
      "learning_rate": 0.002828,
      "loss": 1.18,
      "step": 392320
    },
    {
      "epoch": 29.44720450281426,
      "grad_norm": 0.6819410920143127,
      "learning_rate": 0.002828,
      "loss": 1.1808,
      "step": 392384
    },
    {
      "epoch": 29.45200750469043,
      "grad_norm": 0.6249415278434753,
      "learning_rate": 0.002828,
      "loss": 1.1813,
      "step": 392448
    },
    {
      "epoch": 29.456810506566605,
      "grad_norm": 0.5132665038108826,
      "learning_rate": 0.002828,
      "loss": 1.1814,
      "step": 392512
    },
    {
      "epoch": 29.461613508442777,
      "grad_norm": 0.6131729483604431,
      "learning_rate": 0.002828,
      "loss": 1.1816,
      "step": 392576
    },
    {
      "epoch": 29.46641651031895,
      "grad_norm": 0.8811164498329163,
      "learning_rate": 0.002828,
      "loss": 1.1885,
      "step": 392640
    },
    {
      "epoch": 29.471219512195123,
      "grad_norm": 0.6464258432388306,
      "learning_rate": 0.002828,
      "loss": 1.1874,
      "step": 392704
    },
    {
      "epoch": 29.476022514071296,
      "grad_norm": 0.49052509665489197,
      "learning_rate": 0.002828,
      "loss": 1.1847,
      "step": 392768
    },
    {
      "epoch": 29.480825515947465,
      "grad_norm": 0.6686356067657471,
      "learning_rate": 0.002828,
      "loss": 1.1796,
      "step": 392832
    },
    {
      "epoch": 29.48562851782364,
      "grad_norm": 0.6112715005874634,
      "learning_rate": 0.002828,
      "loss": 1.1853,
      "step": 392896
    },
    {
      "epoch": 29.49043151969981,
      "grad_norm": 0.6032724976539612,
      "learning_rate": 0.002828,
      "loss": 1.1844,
      "step": 392960
    },
    {
      "epoch": 29.495234521575984,
      "grad_norm": 0.5601452589035034,
      "learning_rate": 0.002828,
      "loss": 1.1815,
      "step": 393024
    },
    {
      "epoch": 29.500037523452157,
      "grad_norm": 0.6651715636253357,
      "learning_rate": 0.002828,
      "loss": 1.1806,
      "step": 393088
    },
    {
      "epoch": 29.50484052532833,
      "grad_norm": 0.6076048016548157,
      "learning_rate": 0.002828,
      "loss": 1.1802,
      "step": 393152
    },
    {
      "epoch": 29.509643527204503,
      "grad_norm": 0.5798808336257935,
      "learning_rate": 0.002828,
      "loss": 1.1856,
      "step": 393216
    },
    {
      "epoch": 29.514446529080676,
      "grad_norm": 0.5465596318244934,
      "learning_rate": 0.002828,
      "loss": 1.1876,
      "step": 393280
    },
    {
      "epoch": 29.51924953095685,
      "grad_norm": 0.5441251397132874,
      "learning_rate": 0.002828,
      "loss": 1.1848,
      "step": 393344
    },
    {
      "epoch": 29.52405253283302,
      "grad_norm": 0.5961697101593018,
      "learning_rate": 0.002828,
      "loss": 1.182,
      "step": 393408
    },
    {
      "epoch": 29.528855534709194,
      "grad_norm": 0.7228647470474243,
      "learning_rate": 0.002828,
      "loss": 1.1837,
      "step": 393472
    },
    {
      "epoch": 29.533658536585367,
      "grad_norm": 0.633174479007721,
      "learning_rate": 0.002828,
      "loss": 1.1857,
      "step": 393536
    },
    {
      "epoch": 29.53846153846154,
      "grad_norm": 0.5586833357810974,
      "learning_rate": 0.002828,
      "loss": 1.1828,
      "step": 393600
    },
    {
      "epoch": 29.54326454033771,
      "grad_norm": 0.5021085143089294,
      "learning_rate": 0.002828,
      "loss": 1.1836,
      "step": 393664
    },
    {
      "epoch": 29.548067542213882,
      "grad_norm": 0.6552308201789856,
      "learning_rate": 0.002828,
      "loss": 1.1857,
      "step": 393728
    },
    {
      "epoch": 29.552870544090055,
      "grad_norm": 0.5525025725364685,
      "learning_rate": 0.002828,
      "loss": 1.1926,
      "step": 393792
    },
    {
      "epoch": 29.557673545966228,
      "grad_norm": 0.7086669206619263,
      "learning_rate": 0.002828,
      "loss": 1.1865,
      "step": 393856
    },
    {
      "epoch": 29.5624765478424,
      "grad_norm": 0.47913986444473267,
      "learning_rate": 0.002828,
      "loss": 1.1828,
      "step": 393920
    },
    {
      "epoch": 29.567279549718574,
      "grad_norm": 0.636206865310669,
      "learning_rate": 0.002828,
      "loss": 1.1842,
      "step": 393984
    },
    {
      "epoch": 29.572082551594747,
      "grad_norm": 0.5519427061080933,
      "learning_rate": 0.002828,
      "loss": 1.1885,
      "step": 394048
    },
    {
      "epoch": 29.57688555347092,
      "grad_norm": 0.4739179015159607,
      "learning_rate": 0.002828,
      "loss": 1.1836,
      "step": 394112
    },
    {
      "epoch": 29.581688555347093,
      "grad_norm": 0.5761741399765015,
      "learning_rate": 0.002828,
      "loss": 1.1823,
      "step": 394176
    },
    {
      "epoch": 29.586491557223265,
      "grad_norm": 0.6126164793968201,
      "learning_rate": 0.002828,
      "loss": 1.1857,
      "step": 394240
    },
    {
      "epoch": 29.59129455909944,
      "grad_norm": 0.491203635931015,
      "learning_rate": 0.002828,
      "loss": 1.1867,
      "step": 394304
    },
    {
      "epoch": 29.59609756097561,
      "grad_norm": 0.5796656608581543,
      "learning_rate": 0.002828,
      "loss": 1.1893,
      "step": 394368
    },
    {
      "epoch": 29.600900562851784,
      "grad_norm": 0.5100173950195312,
      "learning_rate": 0.002828,
      "loss": 1.1805,
      "step": 394432
    },
    {
      "epoch": 29.605703564727953,
      "grad_norm": 0.659576952457428,
      "learning_rate": 0.002828,
      "loss": 1.1874,
      "step": 394496
    },
    {
      "epoch": 29.610506566604126,
      "grad_norm": 0.5324804186820984,
      "learning_rate": 0.002828,
      "loss": 1.1889,
      "step": 394560
    },
    {
      "epoch": 29.6153095684803,
      "grad_norm": 0.5801253914833069,
      "learning_rate": 0.002828,
      "loss": 1.1865,
      "step": 394624
    },
    {
      "epoch": 29.620112570356472,
      "grad_norm": 0.5561210513114929,
      "learning_rate": 0.002828,
      "loss": 1.1859,
      "step": 394688
    },
    {
      "epoch": 29.624915572232645,
      "grad_norm": 0.49353739619255066,
      "learning_rate": 0.002828,
      "loss": 1.1835,
      "step": 394752
    },
    {
      "epoch": 29.629718574108818,
      "grad_norm": 0.648216962814331,
      "learning_rate": 0.002828,
      "loss": 1.1848,
      "step": 394816
    },
    {
      "epoch": 29.63452157598499,
      "grad_norm": 0.5576372146606445,
      "learning_rate": 0.002828,
      "loss": 1.1854,
      "step": 394880
    },
    {
      "epoch": 29.639324577861164,
      "grad_norm": 0.5799958109855652,
      "learning_rate": 0.002828,
      "loss": 1.184,
      "step": 394944
    },
    {
      "epoch": 29.644127579737336,
      "grad_norm": 0.5746650099754333,
      "learning_rate": 0.002828,
      "loss": 1.1855,
      "step": 395008
    },
    {
      "epoch": 29.64893058161351,
      "grad_norm": 0.6273934245109558,
      "learning_rate": 0.002828,
      "loss": 1.1866,
      "step": 395072
    },
    {
      "epoch": 29.653733583489682,
      "grad_norm": 0.5673152804374695,
      "learning_rate": 0.002828,
      "loss": 1.189,
      "step": 395136
    },
    {
      "epoch": 29.658536585365855,
      "grad_norm": 0.48399415612220764,
      "learning_rate": 0.002828,
      "loss": 1.1844,
      "step": 395200
    },
    {
      "epoch": 29.663339587242028,
      "grad_norm": 0.5884276032447815,
      "learning_rate": 0.002828,
      "loss": 1.19,
      "step": 395264
    },
    {
      "epoch": 29.668142589118197,
      "grad_norm": 0.5149835348129272,
      "learning_rate": 0.002828,
      "loss": 1.1916,
      "step": 395328
    },
    {
      "epoch": 29.67294559099437,
      "grad_norm": 0.5658831596374512,
      "learning_rate": 0.002828,
      "loss": 1.1876,
      "step": 395392
    },
    {
      "epoch": 29.677748592870543,
      "grad_norm": 0.5463730692863464,
      "learning_rate": 0.002828,
      "loss": 1.1936,
      "step": 395456
    },
    {
      "epoch": 29.682551594746716,
      "grad_norm": 0.8649245500564575,
      "learning_rate": 0.002828,
      "loss": 1.1888,
      "step": 395520
    },
    {
      "epoch": 29.68735459662289,
      "grad_norm": 0.6761359572410583,
      "learning_rate": 0.002828,
      "loss": 1.1836,
      "step": 395584
    },
    {
      "epoch": 29.692157598499062,
      "grad_norm": 0.5373187065124512,
      "learning_rate": 0.002828,
      "loss": 1.189,
      "step": 395648
    },
    {
      "epoch": 29.696960600375235,
      "grad_norm": 0.5609176158905029,
      "learning_rate": 0.002828,
      "loss": 1.1842,
      "step": 395712
    },
    {
      "epoch": 29.701763602251408,
      "grad_norm": 0.6463390588760376,
      "learning_rate": 0.002828,
      "loss": 1.1854,
      "step": 395776
    },
    {
      "epoch": 29.70656660412758,
      "grad_norm": 0.6921478509902954,
      "learning_rate": 0.002828,
      "loss": 1.1868,
      "step": 395840
    },
    {
      "epoch": 29.711369606003753,
      "grad_norm": 0.531934916973114,
      "learning_rate": 0.002828,
      "loss": 1.1914,
      "step": 395904
    },
    {
      "epoch": 29.716172607879926,
      "grad_norm": 0.6882066130638123,
      "learning_rate": 0.002828,
      "loss": 1.1874,
      "step": 395968
    },
    {
      "epoch": 29.7209756097561,
      "grad_norm": 0.5847033858299255,
      "learning_rate": 0.002828,
      "loss": 1.1921,
      "step": 396032
    },
    {
      "epoch": 29.725778611632272,
      "grad_norm": 0.557515025138855,
      "learning_rate": 0.002828,
      "loss": 1.1892,
      "step": 396096
    },
    {
      "epoch": 29.73058161350844,
      "grad_norm": 0.599223792552948,
      "learning_rate": 0.002828,
      "loss": 1.1873,
      "step": 396160
    },
    {
      "epoch": 29.735384615384614,
      "grad_norm": 0.6043716073036194,
      "learning_rate": 0.002828,
      "loss": 1.187,
      "step": 396224
    },
    {
      "epoch": 29.740187617260787,
      "grad_norm": 0.66204833984375,
      "learning_rate": 0.002828,
      "loss": 1.1891,
      "step": 396288
    },
    {
      "epoch": 29.74499061913696,
      "grad_norm": 0.5589165091514587,
      "learning_rate": 0.002828,
      "loss": 1.1914,
      "step": 396352
    },
    {
      "epoch": 29.749793621013133,
      "grad_norm": 0.46428748965263367,
      "learning_rate": 0.002828,
      "loss": 1.1847,
      "step": 396416
    },
    {
      "epoch": 29.754596622889306,
      "grad_norm": 0.5608261823654175,
      "learning_rate": 0.002828,
      "loss": 1.1884,
      "step": 396480
    },
    {
      "epoch": 29.75939962476548,
      "grad_norm": 0.694983184337616,
      "learning_rate": 0.002828,
      "loss": 1.1918,
      "step": 396544
    },
    {
      "epoch": 29.76420262664165,
      "grad_norm": 0.5633040070533752,
      "learning_rate": 0.002828,
      "loss": 1.1923,
      "step": 396608
    },
    {
      "epoch": 29.769005628517824,
      "grad_norm": 0.49123290181159973,
      "learning_rate": 0.002828,
      "loss": 1.1911,
      "step": 396672
    },
    {
      "epoch": 29.773808630393997,
      "grad_norm": 0.7319791316986084,
      "learning_rate": 0.002828,
      "loss": 1.1924,
      "step": 396736
    },
    {
      "epoch": 29.77861163227017,
      "grad_norm": 0.5839659571647644,
      "learning_rate": 0.002828,
      "loss": 1.1833,
      "step": 396800
    },
    {
      "epoch": 29.783414634146343,
      "grad_norm": 0.5470877885818481,
      "learning_rate": 0.002828,
      "loss": 1.1867,
      "step": 396864
    },
    {
      "epoch": 29.788217636022512,
      "grad_norm": 0.7589606642723083,
      "learning_rate": 0.002828,
      "loss": 1.1882,
      "step": 396928
    },
    {
      "epoch": 29.793020637898685,
      "grad_norm": 0.6475275158882141,
      "learning_rate": 0.002828,
      "loss": 1.1856,
      "step": 396992
    },
    {
      "epoch": 29.797823639774858,
      "grad_norm": 0.5769539475440979,
      "learning_rate": 0.002828,
      "loss": 1.1959,
      "step": 397056
    },
    {
      "epoch": 29.80262664165103,
      "grad_norm": 0.6455031633377075,
      "learning_rate": 0.002828,
      "loss": 1.1894,
      "step": 397120
    },
    {
      "epoch": 29.807429643527204,
      "grad_norm": 0.5357875227928162,
      "learning_rate": 0.002828,
      "loss": 1.1878,
      "step": 397184
    },
    {
      "epoch": 29.812232645403377,
      "grad_norm": 0.5677160620689392,
      "learning_rate": 0.002828,
      "loss": 1.1872,
      "step": 397248
    },
    {
      "epoch": 29.81703564727955,
      "grad_norm": 0.6778916716575623,
      "learning_rate": 0.002828,
      "loss": 1.1899,
      "step": 397312
    },
    {
      "epoch": 29.821838649155723,
      "grad_norm": 0.4837956130504608,
      "learning_rate": 0.002828,
      "loss": 1.1831,
      "step": 397376
    },
    {
      "epoch": 29.826641651031895,
      "grad_norm": 0.536710262298584,
      "learning_rate": 0.002828,
      "loss": 1.1884,
      "step": 397440
    },
    {
      "epoch": 29.83144465290807,
      "grad_norm": 0.5414986610412598,
      "learning_rate": 0.002828,
      "loss": 1.1848,
      "step": 397504
    },
    {
      "epoch": 29.83624765478424,
      "grad_norm": 0.5258774757385254,
      "learning_rate": 0.002828,
      "loss": 1.1907,
      "step": 397568
    },
    {
      "epoch": 29.841050656660414,
      "grad_norm": 0.5413457751274109,
      "learning_rate": 0.002828,
      "loss": 1.1883,
      "step": 397632
    },
    {
      "epoch": 29.845853658536587,
      "grad_norm": 0.5534824728965759,
      "learning_rate": 0.002828,
      "loss": 1.1866,
      "step": 397696
    },
    {
      "epoch": 29.850656660412756,
      "grad_norm": 0.5617156624794006,
      "learning_rate": 0.002828,
      "loss": 1.1848,
      "step": 397760
    },
    {
      "epoch": 29.85545966228893,
      "grad_norm": 0.6044380068778992,
      "learning_rate": 0.002828,
      "loss": 1.1914,
      "step": 397824
    },
    {
      "epoch": 29.860262664165102,
      "grad_norm": 0.6871965527534485,
      "learning_rate": 0.002828,
      "loss": 1.1887,
      "step": 397888
    },
    {
      "epoch": 29.865065666041275,
      "grad_norm": 0.6516048312187195,
      "learning_rate": 0.002828,
      "loss": 1.1877,
      "step": 397952
    },
    {
      "epoch": 29.869868667917448,
      "grad_norm": 0.5782538652420044,
      "learning_rate": 0.002828,
      "loss": 1.191,
      "step": 398016
    },
    {
      "epoch": 29.87467166979362,
      "grad_norm": 0.6860178709030151,
      "learning_rate": 0.002828,
      "loss": 1.1884,
      "step": 398080
    },
    {
      "epoch": 29.879474671669794,
      "grad_norm": 0.5459174513816833,
      "learning_rate": 0.002828,
      "loss": 1.1922,
      "step": 398144
    },
    {
      "epoch": 29.884277673545967,
      "grad_norm": 0.6087077260017395,
      "learning_rate": 0.002828,
      "loss": 1.1944,
      "step": 398208
    },
    {
      "epoch": 29.88908067542214,
      "grad_norm": 0.5355796813964844,
      "learning_rate": 0.002828,
      "loss": 1.1925,
      "step": 398272
    },
    {
      "epoch": 29.893883677298312,
      "grad_norm": 0.6955886483192444,
      "learning_rate": 0.002828,
      "loss": 1.1919,
      "step": 398336
    },
    {
      "epoch": 29.898686679174485,
      "grad_norm": 0.5552802085876465,
      "learning_rate": 0.002828,
      "loss": 1.1857,
      "step": 398400
    },
    {
      "epoch": 29.903489681050658,
      "grad_norm": 0.5979511141777039,
      "learning_rate": 0.002828,
      "loss": 1.1863,
      "step": 398464
    },
    {
      "epoch": 29.90829268292683,
      "grad_norm": 0.6473926305770874,
      "learning_rate": 0.002828,
      "loss": 1.1908,
      "step": 398528
    },
    {
      "epoch": 29.913095684803,
      "grad_norm": 0.5830625891685486,
      "learning_rate": 0.002828,
      "loss": 1.1893,
      "step": 398592
    },
    {
      "epoch": 29.917898686679173,
      "grad_norm": 0.6030213832855225,
      "learning_rate": 0.002828,
      "loss": 1.1897,
      "step": 398656
    },
    {
      "epoch": 29.922701688555346,
      "grad_norm": 0.6415476202964783,
      "learning_rate": 0.002828,
      "loss": 1.1884,
      "step": 398720
    },
    {
      "epoch": 29.92750469043152,
      "grad_norm": 0.5384945273399353,
      "learning_rate": 0.002828,
      "loss": 1.1921,
      "step": 398784
    },
    {
      "epoch": 29.932307692307692,
      "grad_norm": 0.5820983648300171,
      "learning_rate": 0.002828,
      "loss": 1.1942,
      "step": 398848
    },
    {
      "epoch": 29.937110694183865,
      "grad_norm": 0.545009195804596,
      "learning_rate": 0.002828,
      "loss": 1.1913,
      "step": 398912
    },
    {
      "epoch": 29.941913696060038,
      "grad_norm": 0.6134216785430908,
      "learning_rate": 0.002828,
      "loss": 1.1897,
      "step": 398976
    },
    {
      "epoch": 29.94671669793621,
      "grad_norm": 0.6903364658355713,
      "learning_rate": 0.002828,
      "loss": 1.1927,
      "step": 399040
    },
    {
      "epoch": 29.951519699812383,
      "grad_norm": 0.5660452246665955,
      "learning_rate": 0.002828,
      "loss": 1.1906,
      "step": 399104
    },
    {
      "epoch": 29.956322701688556,
      "grad_norm": 0.6166103482246399,
      "learning_rate": 0.002828,
      "loss": 1.1873,
      "step": 399168
    },
    {
      "epoch": 29.96112570356473,
      "grad_norm": 0.5798050761222839,
      "learning_rate": 0.002828,
      "loss": 1.1923,
      "step": 399232
    },
    {
      "epoch": 29.965928705440902,
      "grad_norm": 0.630953848361969,
      "learning_rate": 0.002828,
      "loss": 1.1904,
      "step": 399296
    },
    {
      "epoch": 29.97073170731707,
      "grad_norm": 0.66792893409729,
      "learning_rate": 0.002828,
      "loss": 1.1901,
      "step": 399360
    },
    {
      "epoch": 29.975534709193244,
      "grad_norm": 0.6293218731880188,
      "learning_rate": 0.002828,
      "loss": 1.195,
      "step": 399424
    },
    {
      "epoch": 29.980337711069417,
      "grad_norm": 0.5722078680992126,
      "learning_rate": 0.002828,
      "loss": 1.1934,
      "step": 399488
    },
    {
      "epoch": 29.98514071294559,
      "grad_norm": 0.6023445129394531,
      "learning_rate": 0.002828,
      "loss": 1.1904,
      "step": 399552
    },
    {
      "epoch": 29.989943714821763,
      "grad_norm": 0.6318193674087524,
      "learning_rate": 0.002828,
      "loss": 1.1968,
      "step": 399616
    },
    {
      "epoch": 29.994746716697936,
      "grad_norm": 0.6327502131462097,
      "learning_rate": 0.002828,
      "loss": 1.1902,
      "step": 399680
    },
    {
      "epoch": 29.99954971857411,
      "grad_norm": 0.5372485518455505,
      "learning_rate": 0.002828,
      "loss": 1.1938,
      "step": 399744
    },
    {
      "epoch": 30.00435272045028,
      "grad_norm": 0.5668807029724121,
      "learning_rate": 0.002828,
      "loss": 1.1554,
      "step": 399808
    },
    {
      "epoch": 30.009155722326454,
      "grad_norm": 0.6089488863945007,
      "learning_rate": 0.002828,
      "loss": 1.155,
      "step": 399872
    },
    {
      "epoch": 30.013958724202627,
      "grad_norm": 0.5145072937011719,
      "learning_rate": 0.002828,
      "loss": 1.1566,
      "step": 399936
    },
    {
      "epoch": 30.0187617260788,
      "grad_norm": 0.5616568326950073,
      "learning_rate": 0.002828,
      "loss": 1.1564,
      "step": 400000
    },
    {
      "epoch": 30.023564727954973,
      "grad_norm": 0.5550433993339539,
      "learning_rate": 0.002828,
      "loss": 1.1566,
      "step": 400064
    },
    {
      "epoch": 30.028367729831146,
      "grad_norm": 0.543992280960083,
      "learning_rate": 0.002828,
      "loss": 1.154,
      "step": 400128
    },
    {
      "epoch": 30.033170731707315,
      "grad_norm": 0.5426164269447327,
      "learning_rate": 0.002828,
      "loss": 1.1551,
      "step": 400192
    },
    {
      "epoch": 30.037973733583488,
      "grad_norm": 0.7619397044181824,
      "learning_rate": 0.002828,
      "loss": 1.1549,
      "step": 400256
    },
    {
      "epoch": 30.04277673545966,
      "grad_norm": 0.5529201030731201,
      "learning_rate": 0.002828,
      "loss": 1.1537,
      "step": 400320
    },
    {
      "epoch": 30.047579737335834,
      "grad_norm": 0.6061589121818542,
      "learning_rate": 0.002828,
      "loss": 1.1577,
      "step": 400384
    },
    {
      "epoch": 30.052382739212007,
      "grad_norm": 0.8556471467018127,
      "learning_rate": 0.002828,
      "loss": 1.1606,
      "step": 400448
    },
    {
      "epoch": 30.05718574108818,
      "grad_norm": 0.5779027342796326,
      "learning_rate": 0.002828,
      "loss": 1.1617,
      "step": 400512
    },
    {
      "epoch": 30.061988742964353,
      "grad_norm": 0.6338462829589844,
      "learning_rate": 0.002828,
      "loss": 1.1603,
      "step": 400576
    },
    {
      "epoch": 30.066791744840526,
      "grad_norm": 0.5658112168312073,
      "learning_rate": 0.002828,
      "loss": 1.1573,
      "step": 400640
    },
    {
      "epoch": 30.0715947467167,
      "grad_norm": 0.6836487054824829,
      "learning_rate": 0.002828,
      "loss": 1.1625,
      "step": 400704
    },
    {
      "epoch": 30.07639774859287,
      "grad_norm": 0.5295142531394958,
      "learning_rate": 0.002828,
      "loss": 1.1641,
      "step": 400768
    },
    {
      "epoch": 30.081200750469044,
      "grad_norm": 0.8519558906555176,
      "learning_rate": 0.002828,
      "loss": 1.1587,
      "step": 400832
    },
    {
      "epoch": 30.086003752345217,
      "grad_norm": 0.5779620409011841,
      "learning_rate": 0.002828,
      "loss": 1.1607,
      "step": 400896
    },
    {
      "epoch": 30.09080675422139,
      "grad_norm": 0.628088653087616,
      "learning_rate": 0.002828,
      "loss": 1.1626,
      "step": 400960
    },
    {
      "epoch": 30.09560975609756,
      "grad_norm": 0.69316166639328,
      "learning_rate": 0.002828,
      "loss": 1.1602,
      "step": 401024
    },
    {
      "epoch": 30.100412757973732,
      "grad_norm": 0.5918229222297668,
      "learning_rate": 0.002828,
      "loss": 1.1635,
      "step": 401088
    },
    {
      "epoch": 30.105215759849905,
      "grad_norm": 0.6181484460830688,
      "learning_rate": 0.002828,
      "loss": 1.1621,
      "step": 401152
    },
    {
      "epoch": 30.110018761726078,
      "grad_norm": 0.6306777000427246,
      "learning_rate": 0.002828,
      "loss": 1.1561,
      "step": 401216
    },
    {
      "epoch": 30.11482176360225,
      "grad_norm": 0.6380670666694641,
      "learning_rate": 0.002828,
      "loss": 1.1625,
      "step": 401280
    },
    {
      "epoch": 30.119624765478424,
      "grad_norm": 0.5497437119483948,
      "learning_rate": 0.002828,
      "loss": 1.1793,
      "step": 401344
    },
    {
      "epoch": 30.124427767354597,
      "grad_norm": 0.6903517246246338,
      "learning_rate": 0.002828,
      "loss": 1.166,
      "step": 401408
    },
    {
      "epoch": 30.12923076923077,
      "grad_norm": 0.610875129699707,
      "learning_rate": 0.002828,
      "loss": 1.163,
      "step": 401472
    },
    {
      "epoch": 30.134033771106942,
      "grad_norm": 0.5418033599853516,
      "learning_rate": 0.002828,
      "loss": 1.1626,
      "step": 401536
    },
    {
      "epoch": 30.138836772983115,
      "grad_norm": 0.7114241719245911,
      "learning_rate": 0.002828,
      "loss": 1.1668,
      "step": 401600
    },
    {
      "epoch": 30.143639774859288,
      "grad_norm": 0.49383074045181274,
      "learning_rate": 0.002828,
      "loss": 1.1636,
      "step": 401664
    },
    {
      "epoch": 30.14844277673546,
      "grad_norm": 0.6972509026527405,
      "learning_rate": 0.002828,
      "loss": 1.1642,
      "step": 401728
    },
    {
      "epoch": 30.153245778611634,
      "grad_norm": 0.5523052215576172,
      "learning_rate": 0.002828,
      "loss": 1.1636,
      "step": 401792
    },
    {
      "epoch": 30.158048780487803,
      "grad_norm": 0.5856896638870239,
      "learning_rate": 0.002828,
      "loss": 1.1632,
      "step": 401856
    },
    {
      "epoch": 30.162851782363976,
      "grad_norm": 0.6693354249000549,
      "learning_rate": 0.002828,
      "loss": 1.1665,
      "step": 401920
    },
    {
      "epoch": 30.16765478424015,
      "grad_norm": 0.5271797180175781,
      "learning_rate": 0.002828,
      "loss": 1.162,
      "step": 401984
    },
    {
      "epoch": 30.172457786116322,
      "grad_norm": 0.6253278255462646,
      "learning_rate": 0.002828,
      "loss": 1.1624,
      "step": 402048
    },
    {
      "epoch": 30.177260787992495,
      "grad_norm": 0.6121106743812561,
      "learning_rate": 0.002828,
      "loss": 1.1641,
      "step": 402112
    },
    {
      "epoch": 30.182063789868668,
      "grad_norm": 0.5015893578529358,
      "learning_rate": 0.002828,
      "loss": 1.1685,
      "step": 402176
    },
    {
      "epoch": 30.18686679174484,
      "grad_norm": 0.5743915438652039,
      "learning_rate": 0.002828,
      "loss": 1.1663,
      "step": 402240
    },
    {
      "epoch": 30.191669793621013,
      "grad_norm": 0.543053150177002,
      "learning_rate": 0.002828,
      "loss": 1.1644,
      "step": 402304
    },
    {
      "epoch": 30.196472795497186,
      "grad_norm": 0.6323408484458923,
      "learning_rate": 0.002828,
      "loss": 1.1635,
      "step": 402368
    },
    {
      "epoch": 30.20127579737336,
      "grad_norm": 0.5255658626556396,
      "learning_rate": 0.002828,
      "loss": 1.1688,
      "step": 402432
    },
    {
      "epoch": 30.206078799249532,
      "grad_norm": 0.5964052677154541,
      "learning_rate": 0.002828,
      "loss": 1.1654,
      "step": 402496
    },
    {
      "epoch": 30.210881801125705,
      "grad_norm": 0.6892465353012085,
      "learning_rate": 0.002828,
      "loss": 1.1659,
      "step": 402560
    },
    {
      "epoch": 30.215684803001878,
      "grad_norm": 0.5554350018501282,
      "learning_rate": 0.002828,
      "loss": 1.171,
      "step": 402624
    },
    {
      "epoch": 30.220487804878047,
      "grad_norm": 0.5373419523239136,
      "learning_rate": 0.002828,
      "loss": 1.1719,
      "step": 402688
    },
    {
      "epoch": 30.22529080675422,
      "grad_norm": 0.6016690731048584,
      "learning_rate": 0.002828,
      "loss": 1.1673,
      "step": 402752
    },
    {
      "epoch": 30.230093808630393,
      "grad_norm": 0.6829864978790283,
      "learning_rate": 0.002828,
      "loss": 1.1661,
      "step": 402816
    },
    {
      "epoch": 30.234896810506566,
      "grad_norm": 0.5701805353164673,
      "learning_rate": 0.002828,
      "loss": 1.1665,
      "step": 402880
    },
    {
      "epoch": 30.23969981238274,
      "grad_norm": 0.7625254988670349,
      "learning_rate": 0.002828,
      "loss": 1.1678,
      "step": 402944
    },
    {
      "epoch": 30.24450281425891,
      "grad_norm": 0.6311930418014526,
      "learning_rate": 0.002828,
      "loss": 1.1668,
      "step": 403008
    },
    {
      "epoch": 30.249305816135085,
      "grad_norm": 0.6099370121955872,
      "learning_rate": 0.002828,
      "loss": 1.1676,
      "step": 403072
    },
    {
      "epoch": 30.254108818011257,
      "grad_norm": 0.6538092494010925,
      "learning_rate": 0.002828,
      "loss": 1.1697,
      "step": 403136
    },
    {
      "epoch": 30.25891181988743,
      "grad_norm": 0.6979556679725647,
      "learning_rate": 0.002828,
      "loss": 1.1688,
      "step": 403200
    },
    {
      "epoch": 30.263714821763603,
      "grad_norm": 0.5625706911087036,
      "learning_rate": 0.002828,
      "loss": 1.169,
      "step": 403264
    },
    {
      "epoch": 30.268517823639776,
      "grad_norm": 0.7455666065216064,
      "learning_rate": 0.002828,
      "loss": 1.1688,
      "step": 403328
    },
    {
      "epoch": 30.27332082551595,
      "grad_norm": 0.5044279098510742,
      "learning_rate": 0.002828,
      "loss": 1.1712,
      "step": 403392
    },
    {
      "epoch": 30.278123827392122,
      "grad_norm": 0.6426975727081299,
      "learning_rate": 0.002828,
      "loss": 1.1709,
      "step": 403456
    },
    {
      "epoch": 30.28292682926829,
      "grad_norm": 0.6818196773529053,
      "learning_rate": 0.002828,
      "loss": 1.1684,
      "step": 403520
    },
    {
      "epoch": 30.287729831144464,
      "grad_norm": 0.5548245310783386,
      "learning_rate": 0.002828,
      "loss": 1.1676,
      "step": 403584
    },
    {
      "epoch": 30.292532833020637,
      "grad_norm": 0.5054613351821899,
      "learning_rate": 0.002828,
      "loss": 1.1725,
      "step": 403648
    },
    {
      "epoch": 30.29733583489681,
      "grad_norm": 0.6327176094055176,
      "learning_rate": 0.002828,
      "loss": 1.1651,
      "step": 403712
    },
    {
      "epoch": 30.302138836772983,
      "grad_norm": 0.6453624367713928,
      "learning_rate": 0.002828,
      "loss": 1.163,
      "step": 403776
    },
    {
      "epoch": 30.306941838649156,
      "grad_norm": 0.6579575538635254,
      "learning_rate": 0.002828,
      "loss": 1.1666,
      "step": 403840
    },
    {
      "epoch": 30.31174484052533,
      "grad_norm": 0.6196261644363403,
      "learning_rate": 0.002828,
      "loss": 1.1617,
      "step": 403904
    },
    {
      "epoch": 30.3165478424015,
      "grad_norm": 0.6769335865974426,
      "learning_rate": 0.002828,
      "loss": 1.1679,
      "step": 403968
    },
    {
      "epoch": 30.321350844277674,
      "grad_norm": 0.542854905128479,
      "learning_rate": 0.002828,
      "loss": 1.1681,
      "step": 404032
    },
    {
      "epoch": 30.326153846153847,
      "grad_norm": 0.6319664716720581,
      "learning_rate": 0.002828,
      "loss": 1.1684,
      "step": 404096
    },
    {
      "epoch": 30.33095684803002,
      "grad_norm": 0.629665732383728,
      "learning_rate": 0.002828,
      "loss": 1.174,
      "step": 404160
    },
    {
      "epoch": 30.335759849906193,
      "grad_norm": 0.6037463545799255,
      "learning_rate": 0.002828,
      "loss": 1.17,
      "step": 404224
    },
    {
      "epoch": 30.340562851782362,
      "grad_norm": 0.532941997051239,
      "learning_rate": 0.002828,
      "loss": 1.1655,
      "step": 404288
    },
    {
      "epoch": 30.345365853658535,
      "grad_norm": 0.5925180912017822,
      "learning_rate": 0.002828,
      "loss": 1.1652,
      "step": 404352
    },
    {
      "epoch": 30.350168855534708,
      "grad_norm": 0.5588472485542297,
      "learning_rate": 0.002828,
      "loss": 1.1725,
      "step": 404416
    },
    {
      "epoch": 30.35497185741088,
      "grad_norm": 0.6447106003761292,
      "learning_rate": 0.002828,
      "loss": 1.1721,
      "step": 404480
    },
    {
      "epoch": 30.359774859287054,
      "grad_norm": 0.5808579325675964,
      "learning_rate": 0.002828,
      "loss": 1.1677,
      "step": 404544
    },
    {
      "epoch": 30.364577861163227,
      "grad_norm": 0.5732340812683105,
      "learning_rate": 0.002828,
      "loss": 1.1697,
      "step": 404608
    },
    {
      "epoch": 30.3693808630394,
      "grad_norm": 0.5909567475318909,
      "learning_rate": 0.002828,
      "loss": 1.1743,
      "step": 404672
    },
    {
      "epoch": 30.374183864915572,
      "grad_norm": 0.5653650164604187,
      "learning_rate": 0.002828,
      "loss": 1.1726,
      "step": 404736
    },
    {
      "epoch": 30.378986866791745,
      "grad_norm": 0.61961430311203,
      "learning_rate": 0.002828,
      "loss": 1.175,
      "step": 404800
    },
    {
      "epoch": 30.383789868667918,
      "grad_norm": 0.6074270009994507,
      "learning_rate": 0.002828,
      "loss": 1.1674,
      "step": 404864
    },
    {
      "epoch": 30.38859287054409,
      "grad_norm": 0.5093051195144653,
      "learning_rate": 0.002828,
      "loss": 1.1762,
      "step": 404928
    },
    {
      "epoch": 30.393395872420264,
      "grad_norm": 0.6120220422744751,
      "learning_rate": 0.002828,
      "loss": 1.1694,
      "step": 404992
    },
    {
      "epoch": 30.398198874296437,
      "grad_norm": 0.6272151470184326,
      "learning_rate": 0.002828,
      "loss": 1.1674,
      "step": 405056
    },
    {
      "epoch": 30.403001876172606,
      "grad_norm": 0.5591142773628235,
      "learning_rate": 0.002828,
      "loss": 1.1673,
      "step": 405120
    },
    {
      "epoch": 30.40780487804878,
      "grad_norm": 0.6053633689880371,
      "learning_rate": 0.002828,
      "loss": 1.1729,
      "step": 405184
    },
    {
      "epoch": 30.412607879924952,
      "grad_norm": 0.5584940314292908,
      "learning_rate": 0.002828,
      "loss": 1.1755,
      "step": 405248
    },
    {
      "epoch": 30.417410881801125,
      "grad_norm": 0.5795779824256897,
      "learning_rate": 0.002828,
      "loss": 1.1747,
      "step": 405312
    },
    {
      "epoch": 30.422213883677298,
      "grad_norm": 0.700214684009552,
      "learning_rate": 0.002828,
      "loss": 1.1737,
      "step": 405376
    },
    {
      "epoch": 30.42701688555347,
      "grad_norm": 0.6408358216285706,
      "learning_rate": 0.002828,
      "loss": 1.1706,
      "step": 405440
    },
    {
      "epoch": 30.431819887429644,
      "grad_norm": 0.6774629950523376,
      "learning_rate": 0.002828,
      "loss": 1.1704,
      "step": 405504
    },
    {
      "epoch": 30.436622889305816,
      "grad_norm": 0.5555515289306641,
      "learning_rate": 0.002828,
      "loss": 1.1658,
      "step": 405568
    },
    {
      "epoch": 30.44142589118199,
      "grad_norm": 0.6035341620445251,
      "learning_rate": 0.002828,
      "loss": 1.1712,
      "step": 405632
    },
    {
      "epoch": 30.446228893058162,
      "grad_norm": 0.7900217771530151,
      "learning_rate": 0.002828,
      "loss": 1.1751,
      "step": 405696
    },
    {
      "epoch": 30.451031894934335,
      "grad_norm": 0.6393078565597534,
      "learning_rate": 0.002828,
      "loss": 1.1708,
      "step": 405760
    },
    {
      "epoch": 30.455834896810508,
      "grad_norm": 0.7335425615310669,
      "learning_rate": 0.002828,
      "loss": 1.1701,
      "step": 405824
    },
    {
      "epoch": 30.46063789868668,
      "grad_norm": 0.5384011268615723,
      "learning_rate": 0.002828,
      "loss": 1.1769,
      "step": 405888
    },
    {
      "epoch": 30.46544090056285,
      "grad_norm": 0.6363877058029175,
      "learning_rate": 0.002828,
      "loss": 1.1712,
      "step": 405952
    },
    {
      "epoch": 30.470243902439023,
      "grad_norm": 0.657049834728241,
      "learning_rate": 0.002828,
      "loss": 1.1776,
      "step": 406016
    },
    {
      "epoch": 30.475046904315196,
      "grad_norm": 0.6141687631607056,
      "learning_rate": 0.002828,
      "loss": 1.172,
      "step": 406080
    },
    {
      "epoch": 30.47984990619137,
      "grad_norm": 0.6134195327758789,
      "learning_rate": 0.002828,
      "loss": 1.1708,
      "step": 406144
    },
    {
      "epoch": 30.48465290806754,
      "grad_norm": 0.5732262134552002,
      "learning_rate": 0.002828,
      "loss": 1.1659,
      "step": 406208
    },
    {
      "epoch": 30.489455909943715,
      "grad_norm": 0.6324205994606018,
      "learning_rate": 0.002828,
      "loss": 1.1776,
      "step": 406272
    },
    {
      "epoch": 30.494258911819887,
      "grad_norm": 0.6956483125686646,
      "learning_rate": 0.002828,
      "loss": 1.1718,
      "step": 406336
    },
    {
      "epoch": 30.49906191369606,
      "grad_norm": 0.5355058908462524,
      "learning_rate": 0.002828,
      "loss": 1.1809,
      "step": 406400
    },
    {
      "epoch": 30.503864915572233,
      "grad_norm": 0.6604107022285461,
      "learning_rate": 0.002828,
      "loss": 1.1728,
      "step": 406464
    },
    {
      "epoch": 30.508667917448406,
      "grad_norm": 0.6063787937164307,
      "learning_rate": 0.002828,
      "loss": 1.1769,
      "step": 406528
    },
    {
      "epoch": 30.51347091932458,
      "grad_norm": 0.6198680996894836,
      "learning_rate": 0.002828,
      "loss": 1.1718,
      "step": 406592
    },
    {
      "epoch": 30.518273921200752,
      "grad_norm": 0.5494663715362549,
      "learning_rate": 0.002828,
      "loss": 1.1737,
      "step": 406656
    },
    {
      "epoch": 30.523076923076925,
      "grad_norm": 0.5736897587776184,
      "learning_rate": 0.002828,
      "loss": 1.1763,
      "step": 406720
    },
    {
      "epoch": 30.527879924953094,
      "grad_norm": 0.541961133480072,
      "learning_rate": 0.002828,
      "loss": 1.1742,
      "step": 406784
    },
    {
      "epoch": 30.532682926829267,
      "grad_norm": 0.6026626229286194,
      "learning_rate": 0.002828,
      "loss": 1.1762,
      "step": 406848
    },
    {
      "epoch": 30.53748592870544,
      "grad_norm": 0.5529982447624207,
      "learning_rate": 0.002828,
      "loss": 1.1736,
      "step": 406912
    },
    {
      "epoch": 30.542288930581613,
      "grad_norm": 0.7588163018226624,
      "learning_rate": 0.002828,
      "loss": 1.1778,
      "step": 406976
    },
    {
      "epoch": 30.547091932457786,
      "grad_norm": 0.5233981013298035,
      "learning_rate": 0.002828,
      "loss": 1.1703,
      "step": 407040
    },
    {
      "epoch": 30.55189493433396,
      "grad_norm": 0.5268771052360535,
      "learning_rate": 0.002828,
      "loss": 1.1731,
      "step": 407104
    },
    {
      "epoch": 30.55669793621013,
      "grad_norm": 0.6000686883926392,
      "learning_rate": 0.002828,
      "loss": 1.1786,
      "step": 407168
    },
    {
      "epoch": 30.561500938086304,
      "grad_norm": 0.6501026749610901,
      "learning_rate": 0.002828,
      "loss": 1.1776,
      "step": 407232
    },
    {
      "epoch": 30.566303939962477,
      "grad_norm": 0.5419455766677856,
      "learning_rate": 0.002828,
      "loss": 1.1815,
      "step": 407296
    },
    {
      "epoch": 30.57110694183865,
      "grad_norm": 0.526297390460968,
      "learning_rate": 0.002828,
      "loss": 1.1766,
      "step": 407360
    },
    {
      "epoch": 30.575909943714823,
      "grad_norm": 0.5357848405838013,
      "learning_rate": 0.002828,
      "loss": 1.1807,
      "step": 407424
    },
    {
      "epoch": 30.580712945590996,
      "grad_norm": 0.6130989789962769,
      "learning_rate": 0.002828,
      "loss": 1.1686,
      "step": 407488
    },
    {
      "epoch": 30.585515947467165,
      "grad_norm": 0.5149111747741699,
      "learning_rate": 0.002828,
      "loss": 1.167,
      "step": 407552
    },
    {
      "epoch": 30.590318949343338,
      "grad_norm": 0.7189837098121643,
      "learning_rate": 0.002828,
      "loss": 1.1808,
      "step": 407616
    },
    {
      "epoch": 30.59512195121951,
      "grad_norm": 0.5723125338554382,
      "learning_rate": 0.002828,
      "loss": 1.1721,
      "step": 407680
    },
    {
      "epoch": 30.599924953095684,
      "grad_norm": 0.6711981892585754,
      "learning_rate": 0.002828,
      "loss": 1.1741,
      "step": 407744
    },
    {
      "epoch": 30.604727954971857,
      "grad_norm": 0.5369583964347839,
      "learning_rate": 0.002828,
      "loss": 1.1755,
      "step": 407808
    },
    {
      "epoch": 30.60953095684803,
      "grad_norm": 0.5850250124931335,
      "learning_rate": 0.002828,
      "loss": 1.1781,
      "step": 407872
    },
    {
      "epoch": 30.614333958724202,
      "grad_norm": 0.601307213306427,
      "learning_rate": 0.002828,
      "loss": 1.1744,
      "step": 407936
    },
    {
      "epoch": 30.619136960600375,
      "grad_norm": 0.7238708734512329,
      "learning_rate": 0.002828,
      "loss": 1.1754,
      "step": 408000
    },
    {
      "epoch": 30.62393996247655,
      "grad_norm": 0.5604078769683838,
      "learning_rate": 0.002828,
      "loss": 1.1721,
      "step": 408064
    },
    {
      "epoch": 30.62874296435272,
      "grad_norm": 0.5621109008789062,
      "learning_rate": 0.002828,
      "loss": 1.1711,
      "step": 408128
    },
    {
      "epoch": 30.633545966228894,
      "grad_norm": 0.5630921721458435,
      "learning_rate": 0.002828,
      "loss": 1.1746,
      "step": 408192
    },
    {
      "epoch": 30.638348968105067,
      "grad_norm": 0.6420670747756958,
      "learning_rate": 0.002828,
      "loss": 1.1797,
      "step": 408256
    },
    {
      "epoch": 30.64315196998124,
      "grad_norm": 0.6493772864341736,
      "learning_rate": 0.002828,
      "loss": 1.1765,
      "step": 408320
    },
    {
      "epoch": 30.64795497185741,
      "grad_norm": 0.6907517910003662,
      "learning_rate": 0.002828,
      "loss": 1.1759,
      "step": 408384
    },
    {
      "epoch": 30.652757973733582,
      "grad_norm": 0.5132107734680176,
      "learning_rate": 0.002828,
      "loss": 1.1737,
      "step": 408448
    },
    {
      "epoch": 30.657560975609755,
      "grad_norm": 0.5009642839431763,
      "learning_rate": 0.002828,
      "loss": 1.1737,
      "step": 408512
    },
    {
      "epoch": 30.662363977485928,
      "grad_norm": 0.7794398069381714,
      "learning_rate": 0.002828,
      "loss": 1.1716,
      "step": 408576
    },
    {
      "epoch": 30.6671669793621,
      "grad_norm": 0.6804834604263306,
      "learning_rate": 0.002828,
      "loss": 1.1823,
      "step": 408640
    },
    {
      "epoch": 30.671969981238274,
      "grad_norm": 0.5983238220214844,
      "learning_rate": 0.002828,
      "loss": 1.1755,
      "step": 408704
    },
    {
      "epoch": 30.676772983114446,
      "grad_norm": 0.5429848432540894,
      "learning_rate": 0.002828,
      "loss": 1.1818,
      "step": 408768
    },
    {
      "epoch": 30.68157598499062,
      "grad_norm": 0.6377673149108887,
      "learning_rate": 0.002828,
      "loss": 1.1768,
      "step": 408832
    },
    {
      "epoch": 30.686378986866792,
      "grad_norm": 0.6683006286621094,
      "learning_rate": 0.002828,
      "loss": 1.1743,
      "step": 408896
    },
    {
      "epoch": 30.691181988742965,
      "grad_norm": 0.5785638689994812,
      "learning_rate": 0.002828,
      "loss": 1.1761,
      "step": 408960
    },
    {
      "epoch": 30.695984990619138,
      "grad_norm": 0.6333790421485901,
      "learning_rate": 0.002828,
      "loss": 1.1742,
      "step": 409024
    },
    {
      "epoch": 30.70078799249531,
      "grad_norm": 0.6007692217826843,
      "learning_rate": 0.002828,
      "loss": 1.1727,
      "step": 409088
    },
    {
      "epoch": 30.705590994371484,
      "grad_norm": 0.7093082070350647,
      "learning_rate": 0.002828,
      "loss": 1.1825,
      "step": 409152
    },
    {
      "epoch": 30.710393996247653,
      "grad_norm": 0.7057070136070251,
      "learning_rate": 0.002828,
      "loss": 1.1773,
      "step": 409216
    },
    {
      "epoch": 30.715196998123826,
      "grad_norm": 0.5125381946563721,
      "learning_rate": 0.002828,
      "loss": 1.1755,
      "step": 409280
    },
    {
      "epoch": 30.72,
      "grad_norm": 0.5635402202606201,
      "learning_rate": 0.002828,
      "loss": 1.1755,
      "step": 409344
    },
    {
      "epoch": 30.72480300187617,
      "grad_norm": 0.5870218873023987,
      "learning_rate": 0.002828,
      "loss": 1.179,
      "step": 409408
    },
    {
      "epoch": 30.729606003752345,
      "grad_norm": 0.5780309438705444,
      "learning_rate": 0.002828,
      "loss": 1.1775,
      "step": 409472
    },
    {
      "epoch": 30.734409005628518,
      "grad_norm": 0.6460204124450684,
      "learning_rate": 0.002828,
      "loss": 1.1762,
      "step": 409536
    },
    {
      "epoch": 30.73921200750469,
      "grad_norm": 0.5737305283546448,
      "learning_rate": 0.002828,
      "loss": 1.1709,
      "step": 409600
    },
    {
      "epoch": 30.744015009380863,
      "grad_norm": 0.5802620649337769,
      "learning_rate": 0.002828,
      "loss": 1.1795,
      "step": 409664
    },
    {
      "epoch": 30.748818011257036,
      "grad_norm": 0.587925910949707,
      "learning_rate": 0.002828,
      "loss": 1.1777,
      "step": 409728
    },
    {
      "epoch": 30.75362101313321,
      "grad_norm": 0.6866670846939087,
      "learning_rate": 0.002828,
      "loss": 1.1783,
      "step": 409792
    },
    {
      "epoch": 30.758424015009382,
      "grad_norm": 0.5596933364868164,
      "learning_rate": 0.002828,
      "loss": 1.1789,
      "step": 409856
    },
    {
      "epoch": 30.763227016885555,
      "grad_norm": 0.5398388504981995,
      "learning_rate": 0.002828,
      "loss": 1.177,
      "step": 409920
    },
    {
      "epoch": 30.768030018761728,
      "grad_norm": 0.5532910823822021,
      "learning_rate": 0.002828,
      "loss": 1.1841,
      "step": 409984
    },
    {
      "epoch": 30.772833020637897,
      "grad_norm": 0.6639517545700073,
      "learning_rate": 0.002828,
      "loss": 1.1726,
      "step": 410048
    },
    {
      "epoch": 30.77763602251407,
      "grad_norm": 0.5759304761886597,
      "learning_rate": 0.002828,
      "loss": 1.1795,
      "step": 410112
    },
    {
      "epoch": 30.782439024390243,
      "grad_norm": 0.5365297198295593,
      "learning_rate": 0.002828,
      "loss": 1.1781,
      "step": 410176
    },
    {
      "epoch": 30.787242026266416,
      "grad_norm": 0.6630024313926697,
      "learning_rate": 0.002828,
      "loss": 1.1813,
      "step": 410240
    },
    {
      "epoch": 30.79204502814259,
      "grad_norm": 0.6431249380111694,
      "learning_rate": 0.002828,
      "loss": 1.1851,
      "step": 410304
    },
    {
      "epoch": 30.79684803001876,
      "grad_norm": 0.6468825936317444,
      "learning_rate": 0.002828,
      "loss": 1.1761,
      "step": 410368
    },
    {
      "epoch": 30.801651031894934,
      "grad_norm": 0.572039008140564,
      "learning_rate": 0.002828,
      "loss": 1.1785,
      "step": 410432
    },
    {
      "epoch": 30.806454033771107,
      "grad_norm": 0.5533271431922913,
      "learning_rate": 0.002828,
      "loss": 1.1801,
      "step": 410496
    },
    {
      "epoch": 30.81125703564728,
      "grad_norm": 0.545091450214386,
      "learning_rate": 0.002828,
      "loss": 1.1834,
      "step": 410560
    },
    {
      "epoch": 30.816060037523453,
      "grad_norm": 0.6108741164207458,
      "learning_rate": 0.002828,
      "loss": 1.1827,
      "step": 410624
    },
    {
      "epoch": 30.820863039399626,
      "grad_norm": 0.5509781241416931,
      "learning_rate": 0.002828,
      "loss": 1.1791,
      "step": 410688
    },
    {
      "epoch": 30.8256660412758,
      "grad_norm": 0.5328975915908813,
      "learning_rate": 0.002828,
      "loss": 1.18,
      "step": 410752
    },
    {
      "epoch": 30.83046904315197,
      "grad_norm": 0.6925010085105896,
      "learning_rate": 0.002828,
      "loss": 1.179,
      "step": 410816
    },
    {
      "epoch": 30.83527204502814,
      "grad_norm": 0.5600910782814026,
      "learning_rate": 0.002828,
      "loss": 1.1772,
      "step": 410880
    },
    {
      "epoch": 30.840075046904314,
      "grad_norm": 0.5724281668663025,
      "learning_rate": 0.002828,
      "loss": 1.1798,
      "step": 410944
    },
    {
      "epoch": 30.844878048780487,
      "grad_norm": 0.5661110281944275,
      "learning_rate": 0.002828,
      "loss": 1.1794,
      "step": 411008
    },
    {
      "epoch": 30.84968105065666,
      "grad_norm": 0.6667885184288025,
      "learning_rate": 0.002828,
      "loss": 1.1812,
      "step": 411072
    },
    {
      "epoch": 30.854484052532833,
      "grad_norm": 0.5475946068763733,
      "learning_rate": 0.002828,
      "loss": 1.181,
      "step": 411136
    },
    {
      "epoch": 30.859287054409005,
      "grad_norm": 0.6985030770301819,
      "learning_rate": 0.002828,
      "loss": 1.1822,
      "step": 411200
    },
    {
      "epoch": 30.86409005628518,
      "grad_norm": 0.6968584656715393,
      "learning_rate": 0.002828,
      "loss": 1.1807,
      "step": 411264
    },
    {
      "epoch": 30.86889305816135,
      "grad_norm": 0.5080946087837219,
      "learning_rate": 0.002828,
      "loss": 1.1755,
      "step": 411328
    },
    {
      "epoch": 30.873696060037524,
      "grad_norm": 0.5723351836204529,
      "learning_rate": 0.002828,
      "loss": 1.1865,
      "step": 411392
    },
    {
      "epoch": 30.878499061913697,
      "grad_norm": 0.6020140051841736,
      "learning_rate": 0.002828,
      "loss": 1.1793,
      "step": 411456
    },
    {
      "epoch": 30.88330206378987,
      "grad_norm": 0.5678929686546326,
      "learning_rate": 0.002828,
      "loss": 1.1733,
      "step": 411520
    },
    {
      "epoch": 30.888105065666043,
      "grad_norm": 0.6971235275268555,
      "learning_rate": 0.002828,
      "loss": 1.1807,
      "step": 411584
    },
    {
      "epoch": 30.892908067542216,
      "grad_norm": 0.52821284532547,
      "learning_rate": 0.002828,
      "loss": 1.1818,
      "step": 411648
    },
    {
      "epoch": 30.897711069418385,
      "grad_norm": 0.6598101854324341,
      "learning_rate": 0.002828,
      "loss": 1.1825,
      "step": 411712
    },
    {
      "epoch": 30.902514071294558,
      "grad_norm": 0.6706920266151428,
      "learning_rate": 0.002828,
      "loss": 1.1868,
      "step": 411776
    },
    {
      "epoch": 30.90731707317073,
      "grad_norm": 0.5354040265083313,
      "learning_rate": 0.002828,
      "loss": 1.1813,
      "step": 411840
    },
    {
      "epoch": 30.912120075046904,
      "grad_norm": 0.5959027409553528,
      "learning_rate": 0.002828,
      "loss": 1.177,
      "step": 411904
    },
    {
      "epoch": 30.916923076923077,
      "grad_norm": 0.6717599034309387,
      "learning_rate": 0.002828,
      "loss": 1.1796,
      "step": 411968
    },
    {
      "epoch": 30.92172607879925,
      "grad_norm": 0.5805749297142029,
      "learning_rate": 0.002828,
      "loss": 1.1784,
      "step": 412032
    },
    {
      "epoch": 30.926529080675422,
      "grad_norm": 0.5120179057121277,
      "learning_rate": 0.002828,
      "loss": 1.1768,
      "step": 412096
    },
    {
      "epoch": 30.931332082551595,
      "grad_norm": 0.6287373304367065,
      "learning_rate": 0.002828,
      "loss": 1.1753,
      "step": 412160
    },
    {
      "epoch": 30.936135084427768,
      "grad_norm": 0.5255283117294312,
      "learning_rate": 0.002828,
      "loss": 1.1804,
      "step": 412224
    },
    {
      "epoch": 30.94093808630394,
      "grad_norm": 0.8690053224563599,
      "learning_rate": 0.002828,
      "loss": 1.187,
      "step": 412288
    },
    {
      "epoch": 30.945741088180114,
      "grad_norm": 0.5661391615867615,
      "learning_rate": 0.002828,
      "loss": 1.1774,
      "step": 412352
    },
    {
      "epoch": 30.950544090056287,
      "grad_norm": 0.6966512799263,
      "learning_rate": 0.002828,
      "loss": 1.1839,
      "step": 412416
    },
    {
      "epoch": 30.95534709193246,
      "grad_norm": 0.6669551134109497,
      "learning_rate": 0.002828,
      "loss": 1.1816,
      "step": 412480
    },
    {
      "epoch": 30.96015009380863,
      "grad_norm": 0.5155887603759766,
      "learning_rate": 0.002828,
      "loss": 1.186,
      "step": 412544
    },
    {
      "epoch": 30.964953095684802,
      "grad_norm": 0.514497697353363,
      "learning_rate": 0.002828,
      "loss": 1.1802,
      "step": 412608
    },
    {
      "epoch": 30.969756097560975,
      "grad_norm": 0.6499031782150269,
      "learning_rate": 0.002828,
      "loss": 1.1843,
      "step": 412672
    },
    {
      "epoch": 30.974559099437148,
      "grad_norm": 0.6161714792251587,
      "learning_rate": 0.002828,
      "loss": 1.1853,
      "step": 412736
    },
    {
      "epoch": 30.97936210131332,
      "grad_norm": 0.6288555860519409,
      "learning_rate": 0.002828,
      "loss": 1.1846,
      "step": 412800
    },
    {
      "epoch": 30.984165103189493,
      "grad_norm": 0.634675145149231,
      "learning_rate": 0.002828,
      "loss": 1.1807,
      "step": 412864
    },
    {
      "epoch": 30.988968105065666,
      "grad_norm": 0.5580902695655823,
      "learning_rate": 0.002828,
      "loss": 1.1842,
      "step": 412928
    },
    {
      "epoch": 30.99377110694184,
      "grad_norm": 0.5608106851577759,
      "learning_rate": 0.002828,
      "loss": 1.1795,
      "step": 412992
    },
    {
      "epoch": 30.998574108818012,
      "grad_norm": 0.6324924826622009,
      "learning_rate": 0.002828,
      "loss": 1.1744,
      "step": 413056
    },
    {
      "epoch": 31.003377110694185,
      "grad_norm": 0.7145226001739502,
      "learning_rate": 0.002828,
      "loss": 1.1578,
      "step": 413120
    },
    {
      "epoch": 31.008180112570358,
      "grad_norm": 0.6056089401245117,
      "learning_rate": 0.002828,
      "loss": 1.1462,
      "step": 413184
    },
    {
      "epoch": 31.01298311444653,
      "grad_norm": 0.7890845537185669,
      "learning_rate": 0.002828,
      "loss": 1.1467,
      "step": 413248
    },
    {
      "epoch": 31.0177861163227,
      "grad_norm": 0.6414325833320618,
      "learning_rate": 0.002828,
      "loss": 1.1444,
      "step": 413312
    },
    {
      "epoch": 31.022589118198873,
      "grad_norm": 0.6264815926551819,
      "learning_rate": 0.002828,
      "loss": 1.1489,
      "step": 413376
    },
    {
      "epoch": 31.027392120075046,
      "grad_norm": 0.5662343502044678,
      "learning_rate": 0.002828,
      "loss": 1.1404,
      "step": 413440
    },
    {
      "epoch": 31.03219512195122,
      "grad_norm": 0.5298193097114563,
      "learning_rate": 0.002828,
      "loss": 1.1442,
      "step": 413504
    },
    {
      "epoch": 31.03699812382739,
      "grad_norm": 0.5486260652542114,
      "learning_rate": 0.002828,
      "loss": 1.1468,
      "step": 413568
    },
    {
      "epoch": 31.041801125703564,
      "grad_norm": 0.6614625453948975,
      "learning_rate": 0.002828,
      "loss": 1.1462,
      "step": 413632
    },
    {
      "epoch": 31.046604127579737,
      "grad_norm": 0.7219079732894897,
      "learning_rate": 0.002828,
      "loss": 1.1469,
      "step": 413696
    },
    {
      "epoch": 31.05140712945591,
      "grad_norm": 0.7738757133483887,
      "learning_rate": 0.002828,
      "loss": 1.1488,
      "step": 413760
    },
    {
      "epoch": 31.056210131332083,
      "grad_norm": 0.7506507635116577,
      "learning_rate": 0.002828,
      "loss": 1.1512,
      "step": 413824
    },
    {
      "epoch": 31.061013133208256,
      "grad_norm": 0.6067076325416565,
      "learning_rate": 0.002828,
      "loss": 1.148,
      "step": 413888
    },
    {
      "epoch": 31.06581613508443,
      "grad_norm": 0.5615883469581604,
      "learning_rate": 0.002828,
      "loss": 1.1435,
      "step": 413952
    },
    {
      "epoch": 31.0706191369606,
      "grad_norm": 0.516376793384552,
      "learning_rate": 0.002828,
      "loss": 1.1453,
      "step": 414016
    },
    {
      "epoch": 31.075422138836775,
      "grad_norm": 0.47539058327674866,
      "learning_rate": 0.002828,
      "loss": 1.1508,
      "step": 414080
    },
    {
      "epoch": 31.080225140712944,
      "grad_norm": 0.7129861116409302,
      "learning_rate": 0.002828,
      "loss": 1.1499,
      "step": 414144
    },
    {
      "epoch": 31.085028142589117,
      "grad_norm": 0.6293662786483765,
      "learning_rate": 0.002828,
      "loss": 1.1505,
      "step": 414208
    },
    {
      "epoch": 31.08983114446529,
      "grad_norm": 0.6361925005912781,
      "learning_rate": 0.002828,
      "loss": 1.1526,
      "step": 414272
    },
    {
      "epoch": 31.094634146341463,
      "grad_norm": 0.6082159876823425,
      "learning_rate": 0.002828,
      "loss": 1.149,
      "step": 414336
    },
    {
      "epoch": 31.099437148217635,
      "grad_norm": 0.7075167298316956,
      "learning_rate": 0.002828,
      "loss": 1.1476,
      "step": 414400
    },
    {
      "epoch": 31.10424015009381,
      "grad_norm": 0.5679833889007568,
      "learning_rate": 0.002828,
      "loss": 1.1533,
      "step": 414464
    },
    {
      "epoch": 31.10904315196998,
      "grad_norm": 0.6799837350845337,
      "learning_rate": 0.002828,
      "loss": 1.1529,
      "step": 414528
    },
    {
      "epoch": 31.113846153846154,
      "grad_norm": 0.5135023593902588,
      "learning_rate": 0.002828,
      "loss": 1.155,
      "step": 414592
    },
    {
      "epoch": 31.118649155722327,
      "grad_norm": 0.5953816175460815,
      "learning_rate": 0.002828,
      "loss": 1.1558,
      "step": 414656
    },
    {
      "epoch": 31.1234521575985,
      "grad_norm": 0.6822885274887085,
      "learning_rate": 0.002828,
      "loss": 1.1483,
      "step": 414720
    },
    {
      "epoch": 31.128255159474673,
      "grad_norm": 0.5267248153686523,
      "learning_rate": 0.002828,
      "loss": 1.1501,
      "step": 414784
    },
    {
      "epoch": 31.133058161350846,
      "grad_norm": 0.6198325753211975,
      "learning_rate": 0.002828,
      "loss": 1.1484,
      "step": 414848
    },
    {
      "epoch": 31.13786116322702,
      "grad_norm": 0.6195210814476013,
      "learning_rate": 0.002828,
      "loss": 1.1517,
      "step": 414912
    },
    {
      "epoch": 31.142664165103188,
      "grad_norm": 0.7327703237533569,
      "learning_rate": 0.002828,
      "loss": 1.1582,
      "step": 414976
    },
    {
      "epoch": 31.14746716697936,
      "grad_norm": 0.603537917137146,
      "learning_rate": 0.002828,
      "loss": 1.1537,
      "step": 415040
    },
    {
      "epoch": 31.152270168855534,
      "grad_norm": 0.5337952971458435,
      "learning_rate": 0.002828,
      "loss": 1.1559,
      "step": 415104
    },
    {
      "epoch": 31.157073170731707,
      "grad_norm": 0.6618182063102722,
      "learning_rate": 0.002828,
      "loss": 1.1501,
      "step": 415168
    },
    {
      "epoch": 31.16187617260788,
      "grad_norm": 0.7155896425247192,
      "learning_rate": 0.002828,
      "loss": 1.1517,
      "step": 415232
    },
    {
      "epoch": 31.166679174484052,
      "grad_norm": 0.6300784945487976,
      "learning_rate": 0.002828,
      "loss": 1.1539,
      "step": 415296
    },
    {
      "epoch": 31.171482176360225,
      "grad_norm": 0.5707547068595886,
      "learning_rate": 0.002828,
      "loss": 1.1552,
      "step": 415360
    },
    {
      "epoch": 31.176285178236398,
      "grad_norm": 0.6165001392364502,
      "learning_rate": 0.002828,
      "loss": 1.147,
      "step": 415424
    },
    {
      "epoch": 31.18108818011257,
      "grad_norm": 0.6371972560882568,
      "learning_rate": 0.002828,
      "loss": 1.1593,
      "step": 415488
    },
    {
      "epoch": 31.185891181988744,
      "grad_norm": 0.5942570567131042,
      "learning_rate": 0.002828,
      "loss": 1.1571,
      "step": 415552
    },
    {
      "epoch": 31.190694183864917,
      "grad_norm": 0.5444065928459167,
      "learning_rate": 0.002828,
      "loss": 1.1561,
      "step": 415616
    },
    {
      "epoch": 31.19549718574109,
      "grad_norm": 0.7190663814544678,
      "learning_rate": 0.002828,
      "loss": 1.1551,
      "step": 415680
    },
    {
      "epoch": 31.200300187617263,
      "grad_norm": 0.5388247966766357,
      "learning_rate": 0.002828,
      "loss": 1.1554,
      "step": 415744
    },
    {
      "epoch": 31.205103189493432,
      "grad_norm": 0.8031442761421204,
      "learning_rate": 0.002828,
      "loss": 1.1585,
      "step": 415808
    },
    {
      "epoch": 31.209906191369605,
      "grad_norm": 0.6138423681259155,
      "learning_rate": 0.002828,
      "loss": 1.1591,
      "step": 415872
    },
    {
      "epoch": 31.214709193245778,
      "grad_norm": 0.5942168831825256,
      "learning_rate": 0.002828,
      "loss": 1.1572,
      "step": 415936
    },
    {
      "epoch": 31.21951219512195,
      "grad_norm": 0.6421748399734497,
      "learning_rate": 0.002828,
      "loss": 1.1552,
      "step": 416000
    },
    {
      "epoch": 31.224315196998123,
      "grad_norm": 0.6277354955673218,
      "learning_rate": 0.002828,
      "loss": 1.1539,
      "step": 416064
    },
    {
      "epoch": 31.229118198874296,
      "grad_norm": 0.5966684818267822,
      "learning_rate": 0.002828,
      "loss": 1.1542,
      "step": 416128
    },
    {
      "epoch": 31.23392120075047,
      "grad_norm": 0.628309428691864,
      "learning_rate": 0.002828,
      "loss": 1.1583,
      "step": 416192
    },
    {
      "epoch": 31.238724202626642,
      "grad_norm": 0.6220654845237732,
      "learning_rate": 0.002828,
      "loss": 1.1593,
      "step": 416256
    },
    {
      "epoch": 31.243527204502815,
      "grad_norm": 0.6429674029350281,
      "learning_rate": 0.002828,
      "loss": 1.1585,
      "step": 416320
    },
    {
      "epoch": 31.248330206378988,
      "grad_norm": 0.6579111814498901,
      "learning_rate": 0.002828,
      "loss": 1.154,
      "step": 416384
    },
    {
      "epoch": 31.25313320825516,
      "grad_norm": 0.6795548796653748,
      "learning_rate": 0.002828,
      "loss": 1.1572,
      "step": 416448
    },
    {
      "epoch": 31.257936210131334,
      "grad_norm": 0.6224991083145142,
      "learning_rate": 0.002828,
      "loss": 1.1569,
      "step": 416512
    },
    {
      "epoch": 31.262739212007503,
      "grad_norm": 0.526008129119873,
      "learning_rate": 0.002828,
      "loss": 1.1628,
      "step": 416576
    },
    {
      "epoch": 31.267542213883676,
      "grad_norm": 0.5240540504455566,
      "learning_rate": 0.002828,
      "loss": 1.1584,
      "step": 416640
    },
    {
      "epoch": 31.27234521575985,
      "grad_norm": 0.5980592370033264,
      "learning_rate": 0.002828,
      "loss": 1.1564,
      "step": 416704
    },
    {
      "epoch": 31.27714821763602,
      "grad_norm": 0.5666719079017639,
      "learning_rate": 0.002828,
      "loss": 1.1586,
      "step": 416768
    },
    {
      "epoch": 31.281951219512194,
      "grad_norm": 0.6986644268035889,
      "learning_rate": 0.002828,
      "loss": 1.1644,
      "step": 416832
    },
    {
      "epoch": 31.286754221388367,
      "grad_norm": 0.6172641515731812,
      "learning_rate": 0.002828,
      "loss": 1.1594,
      "step": 416896
    },
    {
      "epoch": 31.29155722326454,
      "grad_norm": 0.5929469466209412,
      "learning_rate": 0.002828,
      "loss": 1.1602,
      "step": 416960
    },
    {
      "epoch": 31.296360225140713,
      "grad_norm": 0.7132891416549683,
      "learning_rate": 0.002828,
      "loss": 1.1586,
      "step": 417024
    },
    {
      "epoch": 31.301163227016886,
      "grad_norm": 0.5946763753890991,
      "learning_rate": 0.002828,
      "loss": 1.1586,
      "step": 417088
    },
    {
      "epoch": 31.30596622889306,
      "grad_norm": 0.6350845098495483,
      "learning_rate": 0.002828,
      "loss": 1.1545,
      "step": 417152
    },
    {
      "epoch": 31.310769230769232,
      "grad_norm": 0.567474365234375,
      "learning_rate": 0.002828,
      "loss": 1.1594,
      "step": 417216
    },
    {
      "epoch": 31.315572232645405,
      "grad_norm": 0.6074239015579224,
      "learning_rate": 0.002828,
      "loss": 1.1551,
      "step": 417280
    },
    {
      "epoch": 31.320375234521578,
      "grad_norm": 0.7145832777023315,
      "learning_rate": 0.002828,
      "loss": 1.1566,
      "step": 417344
    },
    {
      "epoch": 31.325178236397747,
      "grad_norm": 0.6822763085365295,
      "learning_rate": 0.002828,
      "loss": 1.1582,
      "step": 417408
    },
    {
      "epoch": 31.32998123827392,
      "grad_norm": 0.6444886326789856,
      "learning_rate": 0.002828,
      "loss": 1.1576,
      "step": 417472
    },
    {
      "epoch": 31.334784240150093,
      "grad_norm": 0.6175189018249512,
      "learning_rate": 0.002828,
      "loss": 1.1586,
      "step": 417536
    },
    {
      "epoch": 31.339587242026266,
      "grad_norm": 0.6543102264404297,
      "learning_rate": 0.002828,
      "loss": 1.16,
      "step": 417600
    },
    {
      "epoch": 31.34439024390244,
      "grad_norm": 0.6924708485603333,
      "learning_rate": 0.002828,
      "loss": 1.1637,
      "step": 417664
    },
    {
      "epoch": 31.34919324577861,
      "grad_norm": 0.5551199913024902,
      "learning_rate": 0.002828,
      "loss": 1.1611,
      "step": 417728
    },
    {
      "epoch": 31.353996247654784,
      "grad_norm": 0.532099723815918,
      "learning_rate": 0.002828,
      "loss": 1.1605,
      "step": 417792
    },
    {
      "epoch": 31.358799249530957,
      "grad_norm": 0.5526866912841797,
      "learning_rate": 0.002828,
      "loss": 1.1582,
      "step": 417856
    },
    {
      "epoch": 31.36360225140713,
      "grad_norm": 0.5917322039604187,
      "learning_rate": 0.002828,
      "loss": 1.1591,
      "step": 417920
    },
    {
      "epoch": 31.368405253283303,
      "grad_norm": 0.5270063281059265,
      "learning_rate": 0.002828,
      "loss": 1.1663,
      "step": 417984
    },
    {
      "epoch": 31.373208255159476,
      "grad_norm": 0.50337153673172,
      "learning_rate": 0.002828,
      "loss": 1.1562,
      "step": 418048
    },
    {
      "epoch": 31.37801125703565,
      "grad_norm": 0.6552616953849792,
      "learning_rate": 0.002828,
      "loss": 1.162,
      "step": 418112
    },
    {
      "epoch": 31.38281425891182,
      "grad_norm": 0.6119304299354553,
      "learning_rate": 0.002828,
      "loss": 1.1607,
      "step": 418176
    },
    {
      "epoch": 31.38761726078799,
      "grad_norm": 0.5546829700469971,
      "learning_rate": 0.002828,
      "loss": 1.1641,
      "step": 418240
    },
    {
      "epoch": 31.392420262664164,
      "grad_norm": 0.606814444065094,
      "learning_rate": 0.002828,
      "loss": 1.167,
      "step": 418304
    },
    {
      "epoch": 31.397223264540337,
      "grad_norm": 0.6050477027893066,
      "learning_rate": 0.002828,
      "loss": 1.1601,
      "step": 418368
    },
    {
      "epoch": 31.40202626641651,
      "grad_norm": 0.6472161412239075,
      "learning_rate": 0.002828,
      "loss": 1.1654,
      "step": 418432
    },
    {
      "epoch": 31.406829268292682,
      "grad_norm": 0.6984646320343018,
      "learning_rate": 0.002828,
      "loss": 1.1628,
      "step": 418496
    },
    {
      "epoch": 31.411632270168855,
      "grad_norm": 0.758280873298645,
      "learning_rate": 0.002828,
      "loss": 1.1588,
      "step": 418560
    },
    {
      "epoch": 31.416435272045028,
      "grad_norm": 0.5866057872772217,
      "learning_rate": 0.002828,
      "loss": 1.161,
      "step": 418624
    },
    {
      "epoch": 31.4212382739212,
      "grad_norm": 0.6438357830047607,
      "learning_rate": 0.002828,
      "loss": 1.1589,
      "step": 418688
    },
    {
      "epoch": 31.426041275797374,
      "grad_norm": 0.5375080108642578,
      "learning_rate": 0.002828,
      "loss": 1.1602,
      "step": 418752
    },
    {
      "epoch": 31.430844277673547,
      "grad_norm": 0.6227983832359314,
      "learning_rate": 0.002828,
      "loss": 1.1646,
      "step": 418816
    },
    {
      "epoch": 31.43564727954972,
      "grad_norm": 0.5298811197280884,
      "learning_rate": 0.002828,
      "loss": 1.1617,
      "step": 418880
    },
    {
      "epoch": 31.440450281425893,
      "grad_norm": 0.5807226300239563,
      "learning_rate": 0.002828,
      "loss": 1.1638,
      "step": 418944
    },
    {
      "epoch": 31.445253283302065,
      "grad_norm": 0.5875961184501648,
      "learning_rate": 0.002828,
      "loss": 1.1649,
      "step": 419008
    },
    {
      "epoch": 31.450056285178235,
      "grad_norm": 0.5816643834114075,
      "learning_rate": 0.002828,
      "loss": 1.1615,
      "step": 419072
    },
    {
      "epoch": 31.454859287054408,
      "grad_norm": 0.7433393001556396,
      "learning_rate": 0.002828,
      "loss": 1.1664,
      "step": 419136
    },
    {
      "epoch": 31.45966228893058,
      "grad_norm": 0.6241632699966431,
      "learning_rate": 0.002828,
      "loss": 1.1659,
      "step": 419200
    },
    {
      "epoch": 31.464465290806753,
      "grad_norm": 0.5977398157119751,
      "learning_rate": 0.002828,
      "loss": 1.1657,
      "step": 419264
    },
    {
      "epoch": 31.469268292682926,
      "grad_norm": 0.5252711176872253,
      "learning_rate": 0.002828,
      "loss": 1.1614,
      "step": 419328
    },
    {
      "epoch": 31.4740712945591,
      "grad_norm": 0.533634603023529,
      "learning_rate": 0.002828,
      "loss": 1.1633,
      "step": 419392
    },
    {
      "epoch": 31.478874296435272,
      "grad_norm": 0.4986468553543091,
      "learning_rate": 0.002828,
      "loss": 1.1644,
      "step": 419456
    },
    {
      "epoch": 31.483677298311445,
      "grad_norm": 0.5706108808517456,
      "learning_rate": 0.002828,
      "loss": 1.1648,
      "step": 419520
    },
    {
      "epoch": 31.488480300187618,
      "grad_norm": 0.48681211471557617,
      "learning_rate": 0.002828,
      "loss": 1.1606,
      "step": 419584
    },
    {
      "epoch": 31.49328330206379,
      "grad_norm": 0.5788038372993469,
      "learning_rate": 0.002828,
      "loss": 1.1622,
      "step": 419648
    },
    {
      "epoch": 31.498086303939964,
      "grad_norm": 0.5305057764053345,
      "learning_rate": 0.002828,
      "loss": 1.1648,
      "step": 419712
    },
    {
      "epoch": 31.502889305816137,
      "grad_norm": 0.5407353639602661,
      "learning_rate": 0.002828,
      "loss": 1.1661,
      "step": 419776
    },
    {
      "epoch": 31.50769230769231,
      "grad_norm": 0.677531898021698,
      "learning_rate": 0.002828,
      "loss": 1.1607,
      "step": 419840
    },
    {
      "epoch": 31.51249530956848,
      "grad_norm": 0.6447296738624573,
      "learning_rate": 0.002828,
      "loss": 1.1635,
      "step": 419904
    },
    {
      "epoch": 31.51729831144465,
      "grad_norm": 0.6101669073104858,
      "learning_rate": 0.002828,
      "loss": 1.1591,
      "step": 419968
    },
    {
      "epoch": 31.522101313320825,
      "grad_norm": 0.5507171154022217,
      "learning_rate": 0.002828,
      "loss": 1.16,
      "step": 420032
    },
    {
      "epoch": 31.526904315196997,
      "grad_norm": 0.6411391496658325,
      "learning_rate": 0.002828,
      "loss": 1.1638,
      "step": 420096
    },
    {
      "epoch": 31.53170731707317,
      "grad_norm": 0.4897032380104065,
      "learning_rate": 0.002828,
      "loss": 1.1684,
      "step": 420160
    },
    {
      "epoch": 31.536510318949343,
      "grad_norm": 0.5630632042884827,
      "learning_rate": 0.002828,
      "loss": 1.1631,
      "step": 420224
    },
    {
      "epoch": 31.541313320825516,
      "grad_norm": 0.6872511506080627,
      "learning_rate": 0.002828,
      "loss": 1.1586,
      "step": 420288
    },
    {
      "epoch": 31.54611632270169,
      "grad_norm": 0.6808540225028992,
      "learning_rate": 0.002828,
      "loss": 1.1662,
      "step": 420352
    },
    {
      "epoch": 31.550919324577862,
      "grad_norm": 0.5341716408729553,
      "learning_rate": 0.002828,
      "loss": 1.1692,
      "step": 420416
    },
    {
      "epoch": 31.555722326454035,
      "grad_norm": 0.7518808245658875,
      "learning_rate": 0.002828,
      "loss": 1.1694,
      "step": 420480
    },
    {
      "epoch": 31.560525328330208,
      "grad_norm": 0.5088610649108887,
      "learning_rate": 0.002828,
      "loss": 1.1644,
      "step": 420544
    },
    {
      "epoch": 31.56532833020638,
      "grad_norm": 0.6243874430656433,
      "learning_rate": 0.002828,
      "loss": 1.1691,
      "step": 420608
    },
    {
      "epoch": 31.570131332082553,
      "grad_norm": 0.5158883929252625,
      "learning_rate": 0.002828,
      "loss": 1.1606,
      "step": 420672
    },
    {
      "epoch": 31.574934333958723,
      "grad_norm": 0.7455403208732605,
      "learning_rate": 0.002828,
      "loss": 1.165,
      "step": 420736
    },
    {
      "epoch": 31.579737335834896,
      "grad_norm": 0.5810747146606445,
      "learning_rate": 0.002828,
      "loss": 1.1674,
      "step": 420800
    },
    {
      "epoch": 31.58454033771107,
      "grad_norm": 0.6923263072967529,
      "learning_rate": 0.002828,
      "loss": 1.1597,
      "step": 420864
    },
    {
      "epoch": 31.58934333958724,
      "grad_norm": 0.669924259185791,
      "learning_rate": 0.002828,
      "loss": 1.1676,
      "step": 420928
    },
    {
      "epoch": 31.594146341463414,
      "grad_norm": 0.5412556529045105,
      "learning_rate": 0.002828,
      "loss": 1.1688,
      "step": 420992
    },
    {
      "epoch": 31.598949343339587,
      "grad_norm": 0.5750775933265686,
      "learning_rate": 0.002828,
      "loss": 1.1656,
      "step": 421056
    },
    {
      "epoch": 31.60375234521576,
      "grad_norm": 0.613204836845398,
      "learning_rate": 0.002828,
      "loss": 1.1648,
      "step": 421120
    },
    {
      "epoch": 31.608555347091933,
      "grad_norm": 0.6003854870796204,
      "learning_rate": 0.002828,
      "loss": 1.169,
      "step": 421184
    },
    {
      "epoch": 31.613358348968106,
      "grad_norm": 0.5949692130088806,
      "learning_rate": 0.002828,
      "loss": 1.1648,
      "step": 421248
    },
    {
      "epoch": 31.61816135084428,
      "grad_norm": 0.5880799293518066,
      "learning_rate": 0.002828,
      "loss": 1.1724,
      "step": 421312
    },
    {
      "epoch": 31.62296435272045,
      "grad_norm": 0.603401780128479,
      "learning_rate": 0.002828,
      "loss": 1.1664,
      "step": 421376
    },
    {
      "epoch": 31.627767354596624,
      "grad_norm": 0.5906250476837158,
      "learning_rate": 0.002828,
      "loss": 1.1645,
      "step": 421440
    },
    {
      "epoch": 31.632570356472794,
      "grad_norm": 0.6233527660369873,
      "learning_rate": 0.002828,
      "loss": 1.1674,
      "step": 421504
    },
    {
      "epoch": 31.637373358348967,
      "grad_norm": 0.6136730313301086,
      "learning_rate": 0.002828,
      "loss": 1.1586,
      "step": 421568
    },
    {
      "epoch": 31.64217636022514,
      "grad_norm": 0.6427733302116394,
      "learning_rate": 0.002828,
      "loss": 1.1695,
      "step": 421632
    },
    {
      "epoch": 31.646979362101312,
      "grad_norm": 0.5708802342414856,
      "learning_rate": 0.002828,
      "loss": 1.1654,
      "step": 421696
    },
    {
      "epoch": 31.651782363977485,
      "grad_norm": 0.7320801615715027,
      "learning_rate": 0.002828,
      "loss": 1.166,
      "step": 421760
    },
    {
      "epoch": 31.65658536585366,
      "grad_norm": 0.7255459427833557,
      "learning_rate": 0.002828,
      "loss": 1.1658,
      "step": 421824
    },
    {
      "epoch": 31.66138836772983,
      "grad_norm": 0.4449161887168884,
      "learning_rate": 0.002828,
      "loss": 1.1683,
      "step": 421888
    },
    {
      "epoch": 31.666191369606004,
      "grad_norm": 0.573015034198761,
      "learning_rate": 0.002828,
      "loss": 1.1687,
      "step": 421952
    },
    {
      "epoch": 31.670994371482177,
      "grad_norm": 0.566383421421051,
      "learning_rate": 0.002828,
      "loss": 1.1694,
      "step": 422016
    },
    {
      "epoch": 31.67579737335835,
      "grad_norm": 0.7386601567268372,
      "learning_rate": 0.002828,
      "loss": 1.174,
      "step": 422080
    },
    {
      "epoch": 31.680600375234523,
      "grad_norm": 0.6502916216850281,
      "learning_rate": 0.002828,
      "loss": 1.1616,
      "step": 422144
    },
    {
      "epoch": 31.685403377110696,
      "grad_norm": 0.6516982316970825,
      "learning_rate": 0.002828,
      "loss": 1.1654,
      "step": 422208
    },
    {
      "epoch": 31.69020637898687,
      "grad_norm": 0.5580388307571411,
      "learning_rate": 0.002828,
      "loss": 1.169,
      "step": 422272
    },
    {
      "epoch": 31.695009380863038,
      "grad_norm": 0.6566148996353149,
      "learning_rate": 0.002828,
      "loss": 1.1675,
      "step": 422336
    },
    {
      "epoch": 31.69981238273921,
      "grad_norm": 0.6050575375556946,
      "learning_rate": 0.002828,
      "loss": 1.1767,
      "step": 422400
    },
    {
      "epoch": 31.704615384615384,
      "grad_norm": 0.5732092261314392,
      "learning_rate": 0.002828,
      "loss": 1.1729,
      "step": 422464
    },
    {
      "epoch": 31.709418386491556,
      "grad_norm": 0.49055150151252747,
      "learning_rate": 0.002828,
      "loss": 1.1637,
      "step": 422528
    },
    {
      "epoch": 31.71422138836773,
      "grad_norm": 0.6409571766853333,
      "learning_rate": 0.002828,
      "loss": 1.1689,
      "step": 422592
    },
    {
      "epoch": 31.719024390243902,
      "grad_norm": 0.6670295596122742,
      "learning_rate": 0.002828,
      "loss": 1.166,
      "step": 422656
    },
    {
      "epoch": 31.723827392120075,
      "grad_norm": 0.5073591470718384,
      "learning_rate": 0.002828,
      "loss": 1.167,
      "step": 422720
    },
    {
      "epoch": 31.728630393996248,
      "grad_norm": 0.5917232632637024,
      "learning_rate": 0.002828,
      "loss": 1.1681,
      "step": 422784
    },
    {
      "epoch": 31.73343339587242,
      "grad_norm": 0.6723231077194214,
      "learning_rate": 0.002828,
      "loss": 1.164,
      "step": 422848
    },
    {
      "epoch": 31.738236397748594,
      "grad_norm": 0.619998037815094,
      "learning_rate": 0.002828,
      "loss": 1.1729,
      "step": 422912
    },
    {
      "epoch": 31.743039399624767,
      "grad_norm": 0.6515606045722961,
      "learning_rate": 0.002828,
      "loss": 1.1738,
      "step": 422976
    },
    {
      "epoch": 31.74784240150094,
      "grad_norm": 0.5862688422203064,
      "learning_rate": 0.002828,
      "loss": 1.1721,
      "step": 423040
    },
    {
      "epoch": 31.752645403377112,
      "grad_norm": 0.5546661615371704,
      "learning_rate": 0.002828,
      "loss": 1.1683,
      "step": 423104
    },
    {
      "epoch": 31.75744840525328,
      "grad_norm": 0.6349488496780396,
      "learning_rate": 0.002828,
      "loss": 1.1702,
      "step": 423168
    },
    {
      "epoch": 31.762251407129455,
      "grad_norm": 0.7520371675491333,
      "learning_rate": 0.002828,
      "loss": 1.1723,
      "step": 423232
    },
    {
      "epoch": 31.767054409005627,
      "grad_norm": 0.7457068562507629,
      "learning_rate": 0.002828,
      "loss": 1.1683,
      "step": 423296
    },
    {
      "epoch": 31.7718574108818,
      "grad_norm": 0.5763024091720581,
      "learning_rate": 0.002828,
      "loss": 1.1661,
      "step": 423360
    },
    {
      "epoch": 31.776660412757973,
      "grad_norm": 0.7857934832572937,
      "learning_rate": 0.002828,
      "loss": 1.1661,
      "step": 423424
    },
    {
      "epoch": 31.781463414634146,
      "grad_norm": 0.6475039720535278,
      "learning_rate": 0.002828,
      "loss": 1.172,
      "step": 423488
    },
    {
      "epoch": 31.78626641651032,
      "grad_norm": 0.6206485033035278,
      "learning_rate": 0.002828,
      "loss": 1.1691,
      "step": 423552
    },
    {
      "epoch": 31.791069418386492,
      "grad_norm": 0.5739595293998718,
      "learning_rate": 0.002828,
      "loss": 1.1657,
      "step": 423616
    },
    {
      "epoch": 31.795872420262665,
      "grad_norm": 0.6397718787193298,
      "learning_rate": 0.002828,
      "loss": 1.1719,
      "step": 423680
    },
    {
      "epoch": 31.800675422138838,
      "grad_norm": 0.6197733879089355,
      "learning_rate": 0.002828,
      "loss": 1.1668,
      "step": 423744
    },
    {
      "epoch": 31.80547842401501,
      "grad_norm": 0.6080317497253418,
      "learning_rate": 0.002828,
      "loss": 1.1711,
      "step": 423808
    },
    {
      "epoch": 31.810281425891183,
      "grad_norm": 0.684962272644043,
      "learning_rate": 0.002828,
      "loss": 1.1699,
      "step": 423872
    },
    {
      "epoch": 31.815084427767353,
      "grad_norm": 0.7387970685958862,
      "learning_rate": 0.002828,
      "loss": 1.1674,
      "step": 423936
    },
    {
      "epoch": 31.819887429643526,
      "grad_norm": 0.5995039343833923,
      "learning_rate": 0.002828,
      "loss": 1.1687,
      "step": 424000
    },
    {
      "epoch": 31.8246904315197,
      "grad_norm": 0.5478743314743042,
      "learning_rate": 0.002828,
      "loss": 1.1738,
      "step": 424064
    },
    {
      "epoch": 31.82949343339587,
      "grad_norm": 0.4925873577594757,
      "learning_rate": 0.002828,
      "loss": 1.1717,
      "step": 424128
    },
    {
      "epoch": 31.834296435272044,
      "grad_norm": 0.5601804256439209,
      "learning_rate": 0.002828,
      "loss": 1.1698,
      "step": 424192
    },
    {
      "epoch": 31.839099437148217,
      "grad_norm": 0.5623326897621155,
      "learning_rate": 0.002828,
      "loss": 1.1676,
      "step": 424256
    },
    {
      "epoch": 31.84390243902439,
      "grad_norm": 0.5536679029464722,
      "learning_rate": 0.002828,
      "loss": 1.1691,
      "step": 424320
    },
    {
      "epoch": 31.848705440900563,
      "grad_norm": 0.6321052312850952,
      "learning_rate": 0.002828,
      "loss": 1.1715,
      "step": 424384
    },
    {
      "epoch": 31.853508442776736,
      "grad_norm": 0.6257811784744263,
      "learning_rate": 0.002828,
      "loss": 1.1735,
      "step": 424448
    },
    {
      "epoch": 31.85831144465291,
      "grad_norm": 0.5375912189483643,
      "learning_rate": 0.002828,
      "loss": 1.1686,
      "step": 424512
    },
    {
      "epoch": 31.86311444652908,
      "grad_norm": 0.6057097315788269,
      "learning_rate": 0.002828,
      "loss": 1.175,
      "step": 424576
    },
    {
      "epoch": 31.867917448405255,
      "grad_norm": 0.6763109564781189,
      "learning_rate": 0.002828,
      "loss": 1.1708,
      "step": 424640
    },
    {
      "epoch": 31.872720450281427,
      "grad_norm": 0.5548925399780273,
      "learning_rate": 0.002828,
      "loss": 1.1662,
      "step": 424704
    },
    {
      "epoch": 31.877523452157597,
      "grad_norm": 0.65684974193573,
      "learning_rate": 0.002828,
      "loss": 1.1728,
      "step": 424768
    },
    {
      "epoch": 31.88232645403377,
      "grad_norm": 0.5662223100662231,
      "learning_rate": 0.002828,
      "loss": 1.169,
      "step": 424832
    },
    {
      "epoch": 31.887129455909943,
      "grad_norm": 0.591471791267395,
      "learning_rate": 0.002828,
      "loss": 1.1711,
      "step": 424896
    },
    {
      "epoch": 31.891932457786115,
      "grad_norm": 0.5270155668258667,
      "learning_rate": 0.002828,
      "loss": 1.171,
      "step": 424960
    },
    {
      "epoch": 31.89673545966229,
      "grad_norm": 0.6331884860992432,
      "learning_rate": 0.002828,
      "loss": 1.1762,
      "step": 425024
    },
    {
      "epoch": 31.90153846153846,
      "grad_norm": 0.6494824886322021,
      "learning_rate": 0.002828,
      "loss": 1.1682,
      "step": 425088
    },
    {
      "epoch": 31.906341463414634,
      "grad_norm": 0.5662149786949158,
      "learning_rate": 0.002828,
      "loss": 1.1738,
      "step": 425152
    },
    {
      "epoch": 31.911144465290807,
      "grad_norm": 0.6335403919219971,
      "learning_rate": 0.002828,
      "loss": 1.1717,
      "step": 425216
    },
    {
      "epoch": 31.91594746716698,
      "grad_norm": 0.5693516135215759,
      "learning_rate": 0.002828,
      "loss": 1.1689,
      "step": 425280
    },
    {
      "epoch": 31.920750469043153,
      "grad_norm": 0.7408801317214966,
      "learning_rate": 0.002828,
      "loss": 1.1772,
      "step": 425344
    },
    {
      "epoch": 31.925553470919326,
      "grad_norm": 0.5731872320175171,
      "learning_rate": 0.002828,
      "loss": 1.1721,
      "step": 425408
    },
    {
      "epoch": 31.9303564727955,
      "grad_norm": 0.7666354775428772,
      "learning_rate": 0.002828,
      "loss": 1.1711,
      "step": 425472
    },
    {
      "epoch": 31.93515947467167,
      "grad_norm": 0.5870010256767273,
      "learning_rate": 0.002828,
      "loss": 1.1761,
      "step": 425536
    },
    {
      "epoch": 31.93996247654784,
      "grad_norm": 0.6777177453041077,
      "learning_rate": 0.002828,
      "loss": 1.1723,
      "step": 425600
    },
    {
      "epoch": 31.944765478424014,
      "grad_norm": 0.613329291343689,
      "learning_rate": 0.002828,
      "loss": 1.1746,
      "step": 425664
    },
    {
      "epoch": 31.949568480300186,
      "grad_norm": 0.5409830212593079,
      "learning_rate": 0.002828,
      "loss": 1.1727,
      "step": 425728
    },
    {
      "epoch": 31.95437148217636,
      "grad_norm": 0.5840930938720703,
      "learning_rate": 0.002828,
      "loss": 1.1731,
      "step": 425792
    },
    {
      "epoch": 31.959174484052532,
      "grad_norm": 0.5402800440788269,
      "learning_rate": 0.002828,
      "loss": 1.1773,
      "step": 425856
    },
    {
      "epoch": 31.963977485928705,
      "grad_norm": 0.7048522233963013,
      "learning_rate": 0.002828,
      "loss": 1.1682,
      "step": 425920
    },
    {
      "epoch": 31.968780487804878,
      "grad_norm": 0.6770626306533813,
      "learning_rate": 0.002828,
      "loss": 1.1751,
      "step": 425984
    },
    {
      "epoch": 31.97358348968105,
      "grad_norm": 0.6093770265579224,
      "learning_rate": 0.002828,
      "loss": 1.1734,
      "step": 426048
    },
    {
      "epoch": 31.978386491557224,
      "grad_norm": 0.6553013920783997,
      "learning_rate": 0.002828,
      "loss": 1.1693,
      "step": 426112
    },
    {
      "epoch": 31.983189493433397,
      "grad_norm": 0.6486321091651917,
      "learning_rate": 0.002828,
      "loss": 1.1685,
      "step": 426176
    },
    {
      "epoch": 31.98799249530957,
      "grad_norm": 0.697055459022522,
      "learning_rate": 0.002828,
      "loss": 1.1714,
      "step": 426240
    },
    {
      "epoch": 31.992795497185742,
      "grad_norm": 0.6191198825836182,
      "learning_rate": 0.002828,
      "loss": 1.1763,
      "step": 426304
    },
    {
      "epoch": 31.997598499061915,
      "grad_norm": 0.5218004584312439,
      "learning_rate": 0.002828,
      "loss": 1.1689,
      "step": 426368
    },
    {
      "epoch": 32.002401500938085,
      "grad_norm": 0.5775057077407837,
      "learning_rate": 0.002828,
      "loss": 1.1579,
      "step": 426432
    },
    {
      "epoch": 32.00720450281426,
      "grad_norm": 0.5523258447647095,
      "learning_rate": 0.002828,
      "loss": 1.1342,
      "step": 426496
    },
    {
      "epoch": 32.01200750469043,
      "grad_norm": 0.6225584149360657,
      "learning_rate": 0.002828,
      "loss": 1.1318,
      "step": 426560
    },
    {
      "epoch": 32.01681050656661,
      "grad_norm": 0.5929083824157715,
      "learning_rate": 0.002828,
      "loss": 1.1372,
      "step": 426624
    },
    {
      "epoch": 32.021613508442776,
      "grad_norm": 0.6784092783927917,
      "learning_rate": 0.002828,
      "loss": 1.1344,
      "step": 426688
    },
    {
      "epoch": 32.02641651031895,
      "grad_norm": 0.546075701713562,
      "learning_rate": 0.002828,
      "loss": 1.1379,
      "step": 426752
    },
    {
      "epoch": 32.03121951219512,
      "grad_norm": 0.5539538860321045,
      "learning_rate": 0.002828,
      "loss": 1.1418,
      "step": 426816
    },
    {
      "epoch": 32.03602251407129,
      "grad_norm": 0.6780215501785278,
      "learning_rate": 0.002828,
      "loss": 1.1385,
      "step": 426880
    },
    {
      "epoch": 32.04082551594747,
      "grad_norm": 0.5612590312957764,
      "learning_rate": 0.002828,
      "loss": 1.134,
      "step": 426944
    },
    {
      "epoch": 32.04562851782364,
      "grad_norm": 0.49800795316696167,
      "learning_rate": 0.002828,
      "loss": 1.1353,
      "step": 427008
    },
    {
      "epoch": 32.05043151969981,
      "grad_norm": 0.6973561644554138,
      "learning_rate": 0.002828,
      "loss": 1.1387,
      "step": 427072
    },
    {
      "epoch": 32.05523452157598,
      "grad_norm": 0.8178017735481262,
      "learning_rate": 0.002828,
      "loss": 1.1467,
      "step": 427136
    },
    {
      "epoch": 32.06003752345216,
      "grad_norm": 0.6194310784339905,
      "learning_rate": 0.002828,
      "loss": 1.1386,
      "step": 427200
    },
    {
      "epoch": 32.06484052532833,
      "grad_norm": 0.5829865336418152,
      "learning_rate": 0.002828,
      "loss": 1.1411,
      "step": 427264
    },
    {
      "epoch": 32.069643527204505,
      "grad_norm": 0.493563175201416,
      "learning_rate": 0.002828,
      "loss": 1.1388,
      "step": 427328
    },
    {
      "epoch": 32.074446529080674,
      "grad_norm": 0.5824282765388489,
      "learning_rate": 0.002828,
      "loss": 1.1466,
      "step": 427392
    },
    {
      "epoch": 32.07924953095685,
      "grad_norm": 0.6629090905189514,
      "learning_rate": 0.002828,
      "loss": 1.1325,
      "step": 427456
    },
    {
      "epoch": 32.08405253283302,
      "grad_norm": 0.8915190696716309,
      "learning_rate": 0.002828,
      "loss": 1.1488,
      "step": 427520
    },
    {
      "epoch": 32.0888555347092,
      "grad_norm": 0.7942197918891907,
      "learning_rate": 0.002828,
      "loss": 1.1477,
      "step": 427584
    },
    {
      "epoch": 32.093658536585366,
      "grad_norm": 0.6769029498100281,
      "learning_rate": 0.002828,
      "loss": 1.1387,
      "step": 427648
    },
    {
      "epoch": 32.098461538461535,
      "grad_norm": 0.6665157079696655,
      "learning_rate": 0.002828,
      "loss": 1.1466,
      "step": 427712
    },
    {
      "epoch": 32.10326454033771,
      "grad_norm": 0.6797373294830322,
      "learning_rate": 0.002828,
      "loss": 1.1446,
      "step": 427776
    },
    {
      "epoch": 32.10806754221388,
      "grad_norm": 0.6158092617988586,
      "learning_rate": 0.002828,
      "loss": 1.1391,
      "step": 427840
    },
    {
      "epoch": 32.11287054409006,
      "grad_norm": 0.6532845497131348,
      "learning_rate": 0.002828,
      "loss": 1.1387,
      "step": 427904
    },
    {
      "epoch": 32.11767354596623,
      "grad_norm": 0.5372705459594727,
      "learning_rate": 0.002828,
      "loss": 1.1425,
      "step": 427968
    },
    {
      "epoch": 32.1224765478424,
      "grad_norm": 0.5978798866271973,
      "learning_rate": 0.002828,
      "loss": 1.145,
      "step": 428032
    },
    {
      "epoch": 32.12727954971857,
      "grad_norm": 0.6882296800613403,
      "learning_rate": 0.002828,
      "loss": 1.1387,
      "step": 428096
    },
    {
      "epoch": 32.13208255159475,
      "grad_norm": 0.5775601267814636,
      "learning_rate": 0.002828,
      "loss": 1.138,
      "step": 428160
    },
    {
      "epoch": 32.13688555347092,
      "grad_norm": 0.46795588731765747,
      "learning_rate": 0.002828,
      "loss": 1.1444,
      "step": 428224
    },
    {
      "epoch": 32.141688555347095,
      "grad_norm": 0.5983672142028809,
      "learning_rate": 0.002828,
      "loss": 1.1428,
      "step": 428288
    },
    {
      "epoch": 32.146491557223264,
      "grad_norm": 0.5804497599601746,
      "learning_rate": 0.002828,
      "loss": 1.1385,
      "step": 428352
    },
    {
      "epoch": 32.15129455909944,
      "grad_norm": 0.6478627920150757,
      "learning_rate": 0.002828,
      "loss": 1.1418,
      "step": 428416
    },
    {
      "epoch": 32.15609756097561,
      "grad_norm": 0.6652305722236633,
      "learning_rate": 0.002828,
      "loss": 1.1436,
      "step": 428480
    },
    {
      "epoch": 32.16090056285178,
      "grad_norm": 0.7055870890617371,
      "learning_rate": 0.002828,
      "loss": 1.1455,
      "step": 428544
    },
    {
      "epoch": 32.165703564727956,
      "grad_norm": 0.6123886704444885,
      "learning_rate": 0.002828,
      "loss": 1.1439,
      "step": 428608
    },
    {
      "epoch": 32.170506566604125,
      "grad_norm": 0.7516705393791199,
      "learning_rate": 0.002828,
      "loss": 1.1475,
      "step": 428672
    },
    {
      "epoch": 32.1753095684803,
      "grad_norm": 0.5789108276367188,
      "learning_rate": 0.002828,
      "loss": 1.1496,
      "step": 428736
    },
    {
      "epoch": 32.18011257035647,
      "grad_norm": 0.5870668292045593,
      "learning_rate": 0.002828,
      "loss": 1.144,
      "step": 428800
    },
    {
      "epoch": 32.18491557223265,
      "grad_norm": 0.6455457210540771,
      "learning_rate": 0.002828,
      "loss": 1.1413,
      "step": 428864
    },
    {
      "epoch": 32.18971857410882,
      "grad_norm": 0.6160818934440613,
      "learning_rate": 0.002828,
      "loss": 1.141,
      "step": 428928
    },
    {
      "epoch": 32.19452157598499,
      "grad_norm": 0.559694230556488,
      "learning_rate": 0.002828,
      "loss": 1.1412,
      "step": 428992
    },
    {
      "epoch": 32.19932457786116,
      "grad_norm": 0.587887167930603,
      "learning_rate": 0.002828,
      "loss": 1.1478,
      "step": 429056
    },
    {
      "epoch": 32.20412757973734,
      "grad_norm": 0.6603989005088806,
      "learning_rate": 0.002828,
      "loss": 1.1464,
      "step": 429120
    },
    {
      "epoch": 32.20893058161351,
      "grad_norm": 0.539109468460083,
      "learning_rate": 0.002828,
      "loss": 1.1494,
      "step": 429184
    },
    {
      "epoch": 32.213733583489685,
      "grad_norm": 0.6352860927581787,
      "learning_rate": 0.002828,
      "loss": 1.1468,
      "step": 429248
    },
    {
      "epoch": 32.218536585365854,
      "grad_norm": 0.6434819102287292,
      "learning_rate": 0.002828,
      "loss": 1.1466,
      "step": 429312
    },
    {
      "epoch": 32.22333958724202,
      "grad_norm": 0.5428691506385803,
      "learning_rate": 0.002828,
      "loss": 1.1452,
      "step": 429376
    },
    {
      "epoch": 32.2281425891182,
      "grad_norm": 0.6501672267913818,
      "learning_rate": 0.002828,
      "loss": 1.1452,
      "step": 429440
    },
    {
      "epoch": 32.23294559099437,
      "grad_norm": 0.5786007642745972,
      "learning_rate": 0.002828,
      "loss": 1.1456,
      "step": 429504
    },
    {
      "epoch": 32.237748592870545,
      "grad_norm": 0.5150612592697144,
      "learning_rate": 0.002828,
      "loss": 1.1513,
      "step": 429568
    },
    {
      "epoch": 32.242551594746715,
      "grad_norm": 0.502147912979126,
      "learning_rate": 0.002828,
      "loss": 1.1423,
      "step": 429632
    },
    {
      "epoch": 32.24735459662289,
      "grad_norm": 0.6412995457649231,
      "learning_rate": 0.002828,
      "loss": 1.1512,
      "step": 429696
    },
    {
      "epoch": 32.25215759849906,
      "grad_norm": 0.5578099489212036,
      "learning_rate": 0.002828,
      "loss": 1.1474,
      "step": 429760
    },
    {
      "epoch": 32.25696060037524,
      "grad_norm": 0.7341820597648621,
      "learning_rate": 0.002828,
      "loss": 1.1477,
      "step": 429824
    },
    {
      "epoch": 32.261763602251406,
      "grad_norm": 0.586214005947113,
      "learning_rate": 0.002828,
      "loss": 1.1449,
      "step": 429888
    },
    {
      "epoch": 32.26656660412758,
      "grad_norm": 0.6538716554641724,
      "learning_rate": 0.002828,
      "loss": 1.1443,
      "step": 429952
    },
    {
      "epoch": 32.27136960600375,
      "grad_norm": 0.6261554956436157,
      "learning_rate": 0.002828,
      "loss": 1.1498,
      "step": 430016
    },
    {
      "epoch": 32.27617260787993,
      "grad_norm": 0.561710000038147,
      "learning_rate": 0.002828,
      "loss": 1.1503,
      "step": 430080
    },
    {
      "epoch": 32.2809756097561,
      "grad_norm": 0.5323227643966675,
      "learning_rate": 0.002828,
      "loss": 1.148,
      "step": 430144
    },
    {
      "epoch": 32.28577861163227,
      "grad_norm": 0.6599904298782349,
      "learning_rate": 0.002828,
      "loss": 1.1521,
      "step": 430208
    },
    {
      "epoch": 32.290581613508444,
      "grad_norm": 0.7270208597183228,
      "learning_rate": 0.002828,
      "loss": 1.1475,
      "step": 430272
    },
    {
      "epoch": 32.29538461538461,
      "grad_norm": 0.4618419110774994,
      "learning_rate": 0.002828,
      "loss": 1.1474,
      "step": 430336
    },
    {
      "epoch": 32.30018761726079,
      "grad_norm": 0.640720784664154,
      "learning_rate": 0.002828,
      "loss": 1.1492,
      "step": 430400
    },
    {
      "epoch": 32.30499061913696,
      "grad_norm": 0.6126275658607483,
      "learning_rate": 0.002828,
      "loss": 1.1491,
      "step": 430464
    },
    {
      "epoch": 32.309793621013135,
      "grad_norm": 0.5292582511901855,
      "learning_rate": 0.002828,
      "loss": 1.1567,
      "step": 430528
    },
    {
      "epoch": 32.314596622889304,
      "grad_norm": 0.6568554043769836,
      "learning_rate": 0.002828,
      "loss": 1.1491,
      "step": 430592
    },
    {
      "epoch": 32.31939962476548,
      "grad_norm": 0.6243530511856079,
      "learning_rate": 0.002828,
      "loss": 1.1482,
      "step": 430656
    },
    {
      "epoch": 32.32420262664165,
      "grad_norm": 0.6516066789627075,
      "learning_rate": 0.002828,
      "loss": 1.1551,
      "step": 430720
    },
    {
      "epoch": 32.32900562851783,
      "grad_norm": 0.5971904993057251,
      "learning_rate": 0.002828,
      "loss": 1.1523,
      "step": 430784
    },
    {
      "epoch": 32.333808630393996,
      "grad_norm": 0.606874942779541,
      "learning_rate": 0.002828,
      "loss": 1.1477,
      "step": 430848
    },
    {
      "epoch": 32.33861163227017,
      "grad_norm": 0.7398237586021423,
      "learning_rate": 0.002828,
      "loss": 1.1501,
      "step": 430912
    },
    {
      "epoch": 32.34341463414634,
      "grad_norm": 0.6455931663513184,
      "learning_rate": 0.002828,
      "loss": 1.1486,
      "step": 430976
    },
    {
      "epoch": 32.34821763602251,
      "grad_norm": 0.6093574166297913,
      "learning_rate": 0.002828,
      "loss": 1.1534,
      "step": 431040
    },
    {
      "epoch": 32.35302063789869,
      "grad_norm": 0.5911879539489746,
      "learning_rate": 0.002828,
      "loss": 1.1492,
      "step": 431104
    },
    {
      "epoch": 32.35782363977486,
      "grad_norm": 0.6537988781929016,
      "learning_rate": 0.002828,
      "loss": 1.1508,
      "step": 431168
    },
    {
      "epoch": 32.36262664165103,
      "grad_norm": 0.6023768782615662,
      "learning_rate": 0.002828,
      "loss": 1.1543,
      "step": 431232
    },
    {
      "epoch": 32.3674296435272,
      "grad_norm": 0.5989535450935364,
      "learning_rate": 0.002828,
      "loss": 1.1476,
      "step": 431296
    },
    {
      "epoch": 32.37223264540338,
      "grad_norm": 0.5429506301879883,
      "learning_rate": 0.002828,
      "loss": 1.1467,
      "step": 431360
    },
    {
      "epoch": 32.37703564727955,
      "grad_norm": 0.5513024926185608,
      "learning_rate": 0.002828,
      "loss": 1.1446,
      "step": 431424
    },
    {
      "epoch": 32.381838649155725,
      "grad_norm": 0.6115459203720093,
      "learning_rate": 0.002828,
      "loss": 1.1523,
      "step": 431488
    },
    {
      "epoch": 32.386641651031894,
      "grad_norm": 0.7115570306777954,
      "learning_rate": 0.002828,
      "loss": 1.1481,
      "step": 431552
    },
    {
      "epoch": 32.39144465290807,
      "grad_norm": 0.561923086643219,
      "learning_rate": 0.002828,
      "loss": 1.1505,
      "step": 431616
    },
    {
      "epoch": 32.39624765478424,
      "grad_norm": 0.6152815818786621,
      "learning_rate": 0.002828,
      "loss": 1.1485,
      "step": 431680
    },
    {
      "epoch": 32.40105065666041,
      "grad_norm": 0.5495036840438843,
      "learning_rate": 0.002828,
      "loss": 1.1506,
      "step": 431744
    },
    {
      "epoch": 32.405853658536586,
      "grad_norm": 0.6558094620704651,
      "learning_rate": 0.002828,
      "loss": 1.1489,
      "step": 431808
    },
    {
      "epoch": 32.410656660412755,
      "grad_norm": 0.6343786120414734,
      "learning_rate": 0.002828,
      "loss": 1.1564,
      "step": 431872
    },
    {
      "epoch": 32.41545966228893,
      "grad_norm": 0.5835247039794922,
      "learning_rate": 0.002828,
      "loss": 1.1553,
      "step": 431936
    },
    {
      "epoch": 32.4202626641651,
      "grad_norm": 0.6473934054374695,
      "learning_rate": 0.002828,
      "loss": 1.1518,
      "step": 432000
    },
    {
      "epoch": 32.42506566604128,
      "grad_norm": 0.5743913650512695,
      "learning_rate": 0.002828,
      "loss": 1.1523,
      "step": 432064
    },
    {
      "epoch": 32.42986866791745,
      "grad_norm": 0.9068625569343567,
      "learning_rate": 0.002828,
      "loss": 1.1512,
      "step": 432128
    },
    {
      "epoch": 32.43467166979362,
      "grad_norm": 0.6137439608573914,
      "learning_rate": 0.002828,
      "loss": 1.1539,
      "step": 432192
    },
    {
      "epoch": 32.43947467166979,
      "grad_norm": 0.5453054904937744,
      "learning_rate": 0.002828,
      "loss": 1.1545,
      "step": 432256
    },
    {
      "epoch": 32.44427767354597,
      "grad_norm": 0.8272791504859924,
      "learning_rate": 0.002828,
      "loss": 1.1496,
      "step": 432320
    },
    {
      "epoch": 32.44908067542214,
      "grad_norm": 0.5712248682975769,
      "learning_rate": 0.002828,
      "loss": 1.1575,
      "step": 432384
    },
    {
      "epoch": 32.453883677298315,
      "grad_norm": 0.5890712141990662,
      "learning_rate": 0.002828,
      "loss": 1.1478,
      "step": 432448
    },
    {
      "epoch": 32.458686679174484,
      "grad_norm": 0.6274373531341553,
      "learning_rate": 0.002828,
      "loss": 1.1567,
      "step": 432512
    },
    {
      "epoch": 32.46348968105065,
      "grad_norm": 0.5753583908081055,
      "learning_rate": 0.002828,
      "loss": 1.15,
      "step": 432576
    },
    {
      "epoch": 32.46829268292683,
      "grad_norm": 0.6169336438179016,
      "learning_rate": 0.002828,
      "loss": 1.1548,
      "step": 432640
    },
    {
      "epoch": 32.473095684803,
      "grad_norm": 0.5989509224891663,
      "learning_rate": 0.002828,
      "loss": 1.1593,
      "step": 432704
    },
    {
      "epoch": 32.477898686679175,
      "grad_norm": 0.4917387068271637,
      "learning_rate": 0.002828,
      "loss": 1.1532,
      "step": 432768
    },
    {
      "epoch": 32.482701688555345,
      "grad_norm": 0.7819995284080505,
      "learning_rate": 0.002828,
      "loss": 1.1521,
      "step": 432832
    },
    {
      "epoch": 32.48750469043152,
      "grad_norm": 0.6098030805587769,
      "learning_rate": 0.002828,
      "loss": 1.1583,
      "step": 432896
    },
    {
      "epoch": 32.49230769230769,
      "grad_norm": 0.5828316807746887,
      "learning_rate": 0.002828,
      "loss": 1.1552,
      "step": 432960
    },
    {
      "epoch": 32.49711069418387,
      "grad_norm": 0.6648019552230835,
      "learning_rate": 0.002828,
      "loss": 1.1509,
      "step": 433024
    },
    {
      "epoch": 32.501913696060036,
      "grad_norm": 0.6286805272102356,
      "learning_rate": 0.002828,
      "loss": 1.1499,
      "step": 433088
    },
    {
      "epoch": 32.50671669793621,
      "grad_norm": 0.5933751463890076,
      "learning_rate": 0.002828,
      "loss": 1.1579,
      "step": 433152
    },
    {
      "epoch": 32.51151969981238,
      "grad_norm": 0.6346881985664368,
      "learning_rate": 0.002828,
      "loss": 1.1556,
      "step": 433216
    },
    {
      "epoch": 32.51632270168856,
      "grad_norm": 0.5340132713317871,
      "learning_rate": 0.002828,
      "loss": 1.1561,
      "step": 433280
    },
    {
      "epoch": 32.52112570356473,
      "grad_norm": 0.5881491899490356,
      "learning_rate": 0.002828,
      "loss": 1.1546,
      "step": 433344
    },
    {
      "epoch": 32.5259287054409,
      "grad_norm": 0.7711880207061768,
      "learning_rate": 0.002828,
      "loss": 1.1495,
      "step": 433408
    },
    {
      "epoch": 32.530731707317074,
      "grad_norm": 0.5821669101715088,
      "learning_rate": 0.002828,
      "loss": 1.1585,
      "step": 433472
    },
    {
      "epoch": 32.53553470919324,
      "grad_norm": 0.5389909148216248,
      "learning_rate": 0.002828,
      "loss": 1.1549,
      "step": 433536
    },
    {
      "epoch": 32.54033771106942,
      "grad_norm": 0.6722609996795654,
      "learning_rate": 0.002828,
      "loss": 1.1579,
      "step": 433600
    },
    {
      "epoch": 32.54514071294559,
      "grad_norm": 0.6273888349533081,
      "learning_rate": 0.002828,
      "loss": 1.1573,
      "step": 433664
    },
    {
      "epoch": 32.549943714821765,
      "grad_norm": 0.5908598303794861,
      "learning_rate": 0.002828,
      "loss": 1.1499,
      "step": 433728
    },
    {
      "epoch": 32.554746716697935,
      "grad_norm": 0.5363947153091431,
      "learning_rate": 0.002828,
      "loss": 1.1575,
      "step": 433792
    },
    {
      "epoch": 32.55954971857411,
      "grad_norm": 0.6329306364059448,
      "learning_rate": 0.002828,
      "loss": 1.1615,
      "step": 433856
    },
    {
      "epoch": 32.56435272045028,
      "grad_norm": 0.6444524526596069,
      "learning_rate": 0.002828,
      "loss": 1.1544,
      "step": 433920
    },
    {
      "epoch": 32.56915572232646,
      "grad_norm": 0.606938898563385,
      "learning_rate": 0.002828,
      "loss": 1.1586,
      "step": 433984
    },
    {
      "epoch": 32.573958724202626,
      "grad_norm": 0.6989884376525879,
      "learning_rate": 0.002828,
      "loss": 1.1526,
      "step": 434048
    },
    {
      "epoch": 32.5787617260788,
      "grad_norm": 0.6586000323295593,
      "learning_rate": 0.002828,
      "loss": 1.1515,
      "step": 434112
    },
    {
      "epoch": 32.58356472795497,
      "grad_norm": 0.591772198677063,
      "learning_rate": 0.002828,
      "loss": 1.1585,
      "step": 434176
    },
    {
      "epoch": 32.58836772983114,
      "grad_norm": 0.5981971025466919,
      "learning_rate": 0.002828,
      "loss": 1.1605,
      "step": 434240
    },
    {
      "epoch": 32.59317073170732,
      "grad_norm": 0.6120181679725647,
      "learning_rate": 0.002828,
      "loss": 1.1584,
      "step": 434304
    },
    {
      "epoch": 32.59797373358349,
      "grad_norm": 0.5320422649383545,
      "learning_rate": 0.002828,
      "loss": 1.1561,
      "step": 434368
    },
    {
      "epoch": 32.60277673545966,
      "grad_norm": 0.6150845885276794,
      "learning_rate": 0.002828,
      "loss": 1.1532,
      "step": 434432
    },
    {
      "epoch": 32.60757973733583,
      "grad_norm": 0.4970727860927582,
      "learning_rate": 0.002828,
      "loss": 1.1556,
      "step": 434496
    },
    {
      "epoch": 32.61238273921201,
      "grad_norm": 0.6972767114639282,
      "learning_rate": 0.002828,
      "loss": 1.1542,
      "step": 434560
    },
    {
      "epoch": 32.61718574108818,
      "grad_norm": 0.5277301669120789,
      "learning_rate": 0.002828,
      "loss": 1.1609,
      "step": 434624
    },
    {
      "epoch": 32.621988742964355,
      "grad_norm": 0.5520920157432556,
      "learning_rate": 0.002828,
      "loss": 1.155,
      "step": 434688
    },
    {
      "epoch": 32.626791744840524,
      "grad_norm": 0.645092785358429,
      "learning_rate": 0.002828,
      "loss": 1.1557,
      "step": 434752
    },
    {
      "epoch": 32.6315947467167,
      "grad_norm": 0.5922141075134277,
      "learning_rate": 0.002828,
      "loss": 1.1569,
      "step": 434816
    },
    {
      "epoch": 32.63639774859287,
      "grad_norm": 0.5581287741661072,
      "learning_rate": 0.002828,
      "loss": 1.1563,
      "step": 434880
    },
    {
      "epoch": 32.64120075046905,
      "grad_norm": 0.5025448799133301,
      "learning_rate": 0.002828,
      "loss": 1.1583,
      "step": 434944
    },
    {
      "epoch": 32.646003752345216,
      "grad_norm": 0.8015186786651611,
      "learning_rate": 0.002828,
      "loss": 1.1562,
      "step": 435008
    },
    {
      "epoch": 32.650806754221385,
      "grad_norm": 0.6109132170677185,
      "learning_rate": 0.002828,
      "loss": 1.1543,
      "step": 435072
    },
    {
      "epoch": 32.65560975609756,
      "grad_norm": 0.6086292266845703,
      "learning_rate": 0.002828,
      "loss": 1.1568,
      "step": 435136
    },
    {
      "epoch": 32.66041275797373,
      "grad_norm": 0.637120246887207,
      "learning_rate": 0.002828,
      "loss": 1.1556,
      "step": 435200
    },
    {
      "epoch": 32.66521575984991,
      "grad_norm": 0.6186965107917786,
      "learning_rate": 0.002828,
      "loss": 1.1586,
      "step": 435264
    },
    {
      "epoch": 32.67001876172608,
      "grad_norm": 0.7160683274269104,
      "learning_rate": 0.002828,
      "loss": 1.1552,
      "step": 435328
    },
    {
      "epoch": 32.67482176360225,
      "grad_norm": 0.5873770117759705,
      "learning_rate": 0.002828,
      "loss": 1.1558,
      "step": 435392
    },
    {
      "epoch": 32.67962476547842,
      "grad_norm": 0.6060091257095337,
      "learning_rate": 0.002828,
      "loss": 1.1648,
      "step": 435456
    },
    {
      "epoch": 32.6844277673546,
      "grad_norm": 0.682604193687439,
      "learning_rate": 0.002828,
      "loss": 1.163,
      "step": 435520
    },
    {
      "epoch": 32.68923076923077,
      "grad_norm": 0.5475385189056396,
      "learning_rate": 0.002828,
      "loss": 1.1619,
      "step": 435584
    },
    {
      "epoch": 32.694033771106945,
      "grad_norm": 0.6687578558921814,
      "learning_rate": 0.002828,
      "loss": 1.1584,
      "step": 435648
    },
    {
      "epoch": 32.698836772983114,
      "grad_norm": 0.5442762970924377,
      "learning_rate": 0.002828,
      "loss": 1.1594,
      "step": 435712
    },
    {
      "epoch": 32.70363977485929,
      "grad_norm": 0.6415442824363708,
      "learning_rate": 0.002828,
      "loss": 1.1621,
      "step": 435776
    },
    {
      "epoch": 32.70844277673546,
      "grad_norm": 0.5130046606063843,
      "learning_rate": 0.002828,
      "loss": 1.1586,
      "step": 435840
    },
    {
      "epoch": 32.71324577861163,
      "grad_norm": 0.610876202583313,
      "learning_rate": 0.002828,
      "loss": 1.1588,
      "step": 435904
    },
    {
      "epoch": 32.718048780487806,
      "grad_norm": 0.5902680158615112,
      "learning_rate": 0.002828,
      "loss": 1.1635,
      "step": 435968
    },
    {
      "epoch": 32.722851782363975,
      "grad_norm": 0.5602878332138062,
      "learning_rate": 0.002828,
      "loss": 1.1582,
      "step": 436032
    },
    {
      "epoch": 32.72765478424015,
      "grad_norm": 0.6569159626960754,
      "learning_rate": 0.002828,
      "loss": 1.1593,
      "step": 436096
    },
    {
      "epoch": 32.73245778611632,
      "grad_norm": 0.7306807637214661,
      "learning_rate": 0.002828,
      "loss": 1.1594,
      "step": 436160
    },
    {
      "epoch": 32.7372607879925,
      "grad_norm": 0.6899413466453552,
      "learning_rate": 0.002828,
      "loss": 1.1562,
      "step": 436224
    },
    {
      "epoch": 32.742063789868666,
      "grad_norm": 0.5341850519180298,
      "learning_rate": 0.002828,
      "loss": 1.1588,
      "step": 436288
    },
    {
      "epoch": 32.74686679174484,
      "grad_norm": 0.5809001326560974,
      "learning_rate": 0.002828,
      "loss": 1.1602,
      "step": 436352
    },
    {
      "epoch": 32.75166979362101,
      "grad_norm": 0.5650394558906555,
      "learning_rate": 0.002828,
      "loss": 1.158,
      "step": 436416
    },
    {
      "epoch": 32.75647279549719,
      "grad_norm": 0.5740302801132202,
      "learning_rate": 0.002828,
      "loss": 1.1595,
      "step": 436480
    },
    {
      "epoch": 32.76127579737336,
      "grad_norm": 0.57270747423172,
      "learning_rate": 0.002828,
      "loss": 1.1572,
      "step": 436544
    },
    {
      "epoch": 32.766078799249534,
      "grad_norm": 0.568086564540863,
      "learning_rate": 0.002828,
      "loss": 1.1609,
      "step": 436608
    },
    {
      "epoch": 32.770881801125704,
      "grad_norm": 0.6031185984611511,
      "learning_rate": 0.002828,
      "loss": 1.1559,
      "step": 436672
    },
    {
      "epoch": 32.77568480300187,
      "grad_norm": 0.6868975162506104,
      "learning_rate": 0.002828,
      "loss": 1.1593,
      "step": 436736
    },
    {
      "epoch": 32.78048780487805,
      "grad_norm": 0.6145572662353516,
      "learning_rate": 0.002828,
      "loss": 1.166,
      "step": 436800
    },
    {
      "epoch": 32.78529080675422,
      "grad_norm": 0.5906713604927063,
      "learning_rate": 0.002828,
      "loss": 1.1612,
      "step": 436864
    },
    {
      "epoch": 32.790093808630395,
      "grad_norm": 0.584947943687439,
      "learning_rate": 0.002828,
      "loss": 1.1581,
      "step": 436928
    },
    {
      "epoch": 32.794896810506565,
      "grad_norm": 0.6431328654289246,
      "learning_rate": 0.002828,
      "loss": 1.1649,
      "step": 436992
    },
    {
      "epoch": 32.79969981238274,
      "grad_norm": 0.5539917945861816,
      "learning_rate": 0.002828,
      "loss": 1.1606,
      "step": 437056
    },
    {
      "epoch": 32.80450281425891,
      "grad_norm": 0.4958864450454712,
      "learning_rate": 0.002828,
      "loss": 1.1635,
      "step": 437120
    },
    {
      "epoch": 32.80930581613509,
      "grad_norm": 0.6840506196022034,
      "learning_rate": 0.002828,
      "loss": 1.1619,
      "step": 437184
    },
    {
      "epoch": 32.814108818011256,
      "grad_norm": 0.6359443664550781,
      "learning_rate": 0.002828,
      "loss": 1.1603,
      "step": 437248
    },
    {
      "epoch": 32.81891181988743,
      "grad_norm": 0.578838050365448,
      "learning_rate": 0.002828,
      "loss": 1.1568,
      "step": 437312
    },
    {
      "epoch": 32.8237148217636,
      "grad_norm": 0.5920944809913635,
      "learning_rate": 0.002828,
      "loss": 1.1673,
      "step": 437376
    },
    {
      "epoch": 32.82851782363978,
      "grad_norm": 0.5457953214645386,
      "learning_rate": 0.002828,
      "loss": 1.1608,
      "step": 437440
    },
    {
      "epoch": 32.83332082551595,
      "grad_norm": 0.878715455532074,
      "learning_rate": 0.002828,
      "loss": 1.161,
      "step": 437504
    },
    {
      "epoch": 32.83812382739212,
      "grad_norm": 0.5791822671890259,
      "learning_rate": 0.002828,
      "loss": 1.1599,
      "step": 437568
    },
    {
      "epoch": 32.84292682926829,
      "grad_norm": 0.5815683603286743,
      "learning_rate": 0.002828,
      "loss": 1.1599,
      "step": 437632
    },
    {
      "epoch": 32.84772983114446,
      "grad_norm": 0.5945708751678467,
      "learning_rate": 0.002828,
      "loss": 1.1608,
      "step": 437696
    },
    {
      "epoch": 32.85253283302064,
      "grad_norm": 0.5922000408172607,
      "learning_rate": 0.002828,
      "loss": 1.1616,
      "step": 437760
    },
    {
      "epoch": 32.85733583489681,
      "grad_norm": 0.568234384059906,
      "learning_rate": 0.002828,
      "loss": 1.1621,
      "step": 437824
    },
    {
      "epoch": 32.862138836772985,
      "grad_norm": 0.7399768829345703,
      "learning_rate": 0.002828,
      "loss": 1.1614,
      "step": 437888
    },
    {
      "epoch": 32.866941838649154,
      "grad_norm": 0.4981638193130493,
      "learning_rate": 0.002828,
      "loss": 1.1622,
      "step": 437952
    },
    {
      "epoch": 32.87174484052533,
      "grad_norm": 0.5751581192016602,
      "learning_rate": 0.002828,
      "loss": 1.1604,
      "step": 438016
    },
    {
      "epoch": 32.8765478424015,
      "grad_norm": 0.5195406675338745,
      "learning_rate": 0.002828,
      "loss": 1.1633,
      "step": 438080
    },
    {
      "epoch": 32.88135084427768,
      "grad_norm": 0.49524155259132385,
      "learning_rate": 0.002828,
      "loss": 1.1667,
      "step": 438144
    },
    {
      "epoch": 32.886153846153846,
      "grad_norm": 0.684440553188324,
      "learning_rate": 0.002828,
      "loss": 1.16,
      "step": 438208
    },
    {
      "epoch": 32.890956848030015,
      "grad_norm": 0.6392771005630493,
      "learning_rate": 0.002828,
      "loss": 1.1637,
      "step": 438272
    },
    {
      "epoch": 32.89575984990619,
      "grad_norm": 0.6096252799034119,
      "learning_rate": 0.002828,
      "loss": 1.1601,
      "step": 438336
    },
    {
      "epoch": 32.90056285178236,
      "grad_norm": 0.6157074570655823,
      "learning_rate": 0.002828,
      "loss": 1.1674,
      "step": 438400
    },
    {
      "epoch": 32.90536585365854,
      "grad_norm": 0.5888133645057678,
      "learning_rate": 0.002828,
      "loss": 1.1647,
      "step": 438464
    },
    {
      "epoch": 32.91016885553471,
      "grad_norm": 0.7054190635681152,
      "learning_rate": 0.002828,
      "loss": 1.1604,
      "step": 438528
    },
    {
      "epoch": 32.91497185741088,
      "grad_norm": 0.6485116481781006,
      "learning_rate": 0.002828,
      "loss": 1.1671,
      "step": 438592
    },
    {
      "epoch": 32.91977485928705,
      "grad_norm": 0.5710200071334839,
      "learning_rate": 0.002828,
      "loss": 1.1681,
      "step": 438656
    },
    {
      "epoch": 32.92457786116323,
      "grad_norm": 0.6301112771034241,
      "learning_rate": 0.002828,
      "loss": 1.1568,
      "step": 438720
    },
    {
      "epoch": 32.9293808630394,
      "grad_norm": 0.5653761625289917,
      "learning_rate": 0.002828,
      "loss": 1.1616,
      "step": 438784
    },
    {
      "epoch": 32.934183864915575,
      "grad_norm": 0.6624953150749207,
      "learning_rate": 0.002828,
      "loss": 1.1616,
      "step": 438848
    },
    {
      "epoch": 32.938986866791744,
      "grad_norm": 0.5718125104904175,
      "learning_rate": 0.002828,
      "loss": 1.162,
      "step": 438912
    },
    {
      "epoch": 32.94378986866792,
      "grad_norm": 0.5164399147033691,
      "learning_rate": 0.002828,
      "loss": 1.1634,
      "step": 438976
    },
    {
      "epoch": 32.94859287054409,
      "grad_norm": 0.46567627787590027,
      "learning_rate": 0.002828,
      "loss": 1.1591,
      "step": 439040
    },
    {
      "epoch": 32.95339587242026,
      "grad_norm": 0.5681803226470947,
      "learning_rate": 0.002828,
      "loss": 1.1603,
      "step": 439104
    },
    {
      "epoch": 32.958198874296436,
      "grad_norm": 0.6604852080345154,
      "learning_rate": 0.002828,
      "loss": 1.1637,
      "step": 439168
    },
    {
      "epoch": 32.963001876172605,
      "grad_norm": 0.5455286502838135,
      "learning_rate": 0.002828,
      "loss": 1.1566,
      "step": 439232
    },
    {
      "epoch": 32.96780487804878,
      "grad_norm": 0.555708646774292,
      "learning_rate": 0.002828,
      "loss": 1.1659,
      "step": 439296
    },
    {
      "epoch": 32.97260787992495,
      "grad_norm": 0.698499321937561,
      "learning_rate": 0.002828,
      "loss": 1.16,
      "step": 439360
    },
    {
      "epoch": 32.97741088180113,
      "grad_norm": 0.5202065110206604,
      "learning_rate": 0.002828,
      "loss": 1.1676,
      "step": 439424
    },
    {
      "epoch": 32.9822138836773,
      "grad_norm": 0.5934895873069763,
      "learning_rate": 0.002828,
      "loss": 1.1661,
      "step": 439488
    },
    {
      "epoch": 32.98701688555347,
      "grad_norm": 0.5989176034927368,
      "learning_rate": 0.002828,
      "loss": 1.1609,
      "step": 439552
    },
    {
      "epoch": 32.99181988742964,
      "grad_norm": 0.5574471950531006,
      "learning_rate": 0.002828,
      "loss": 1.1624,
      "step": 439616
    },
    {
      "epoch": 32.99662288930582,
      "grad_norm": 0.6035043597221375,
      "learning_rate": 0.002828,
      "loss": 1.1581,
      "step": 439680
    },
    {
      "epoch": 33.00142589118199,
      "grad_norm": 0.6401668787002563,
      "learning_rate": 0.002828,
      "loss": 1.1551,
      "step": 439744
    },
    {
      "epoch": 33.006228893058164,
      "grad_norm": 0.6571485996246338,
      "learning_rate": 0.002828,
      "loss": 1.125,
      "step": 439808
    },
    {
      "epoch": 33.011031894934334,
      "grad_norm": 0.6424153447151184,
      "learning_rate": 0.002828,
      "loss": 1.1318,
      "step": 439872
    },
    {
      "epoch": 33.0158348968105,
      "grad_norm": 0.5356622338294983,
      "learning_rate": 0.002828,
      "loss": 1.1314,
      "step": 439936
    },
    {
      "epoch": 33.02063789868668,
      "grad_norm": 0.6436302065849304,
      "learning_rate": 0.002828,
      "loss": 1.1284,
      "step": 440000
    },
    {
      "epoch": 33.02544090056285,
      "grad_norm": 0.63233482837677,
      "learning_rate": 0.002828,
      "loss": 1.1314,
      "step": 440064
    },
    {
      "epoch": 33.030243902439025,
      "grad_norm": 0.5510907173156738,
      "learning_rate": 0.002828,
      "loss": 1.1284,
      "step": 440128
    },
    {
      "epoch": 33.035046904315195,
      "grad_norm": 0.590936541557312,
      "learning_rate": 0.002828,
      "loss": 1.1246,
      "step": 440192
    },
    {
      "epoch": 33.03984990619137,
      "grad_norm": 0.5466500520706177,
      "learning_rate": 0.002828,
      "loss": 1.1267,
      "step": 440256
    },
    {
      "epoch": 33.04465290806754,
      "grad_norm": 0.6178146004676819,
      "learning_rate": 0.002828,
      "loss": 1.1288,
      "step": 440320
    },
    {
      "epoch": 33.04945590994372,
      "grad_norm": 0.6533997058868408,
      "learning_rate": 0.002828,
      "loss": 1.1275,
      "step": 440384
    },
    {
      "epoch": 33.054258911819886,
      "grad_norm": 0.7475684285163879,
      "learning_rate": 0.002828,
      "loss": 1.132,
      "step": 440448
    },
    {
      "epoch": 33.05906191369606,
      "grad_norm": 0.6964446902275085,
      "learning_rate": 0.002828,
      "loss": 1.1329,
      "step": 440512
    },
    {
      "epoch": 33.06386491557223,
      "grad_norm": 0.6572120785713196,
      "learning_rate": 0.002828,
      "loss": 1.1313,
      "step": 440576
    },
    {
      "epoch": 33.06866791744841,
      "grad_norm": 0.6430488228797913,
      "learning_rate": 0.002828,
      "loss": 1.1312,
      "step": 440640
    },
    {
      "epoch": 33.07347091932458,
      "grad_norm": 0.5044344663619995,
      "learning_rate": 0.002828,
      "loss": 1.1287,
      "step": 440704
    },
    {
      "epoch": 33.07827392120075,
      "grad_norm": 0.7026501297950745,
      "learning_rate": 0.002828,
      "loss": 1.1306,
      "step": 440768
    },
    {
      "epoch": 33.08307692307692,
      "grad_norm": 0.5623020529747009,
      "learning_rate": 0.002828,
      "loss": 1.1241,
      "step": 440832
    },
    {
      "epoch": 33.08787992495309,
      "grad_norm": 0.7098539471626282,
      "learning_rate": 0.002828,
      "loss": 1.1364,
      "step": 440896
    },
    {
      "epoch": 33.09268292682927,
      "grad_norm": 0.6493317484855652,
      "learning_rate": 0.002828,
      "loss": 1.1341,
      "step": 440960
    },
    {
      "epoch": 33.09748592870544,
      "grad_norm": 0.4738115072250366,
      "learning_rate": 0.002828,
      "loss": 1.1348,
      "step": 441024
    },
    {
      "epoch": 33.102288930581615,
      "grad_norm": 0.6053628325462341,
      "learning_rate": 0.002828,
      "loss": 1.1364,
      "step": 441088
    },
    {
      "epoch": 33.107091932457784,
      "grad_norm": 0.6097379326820374,
      "learning_rate": 0.002828,
      "loss": 1.1236,
      "step": 441152
    },
    {
      "epoch": 33.11189493433396,
      "grad_norm": 0.650699257850647,
      "learning_rate": 0.002828,
      "loss": 1.1299,
      "step": 441216
    },
    {
      "epoch": 33.11669793621013,
      "grad_norm": 0.5362073183059692,
      "learning_rate": 0.002828,
      "loss": 1.1331,
      "step": 441280
    },
    {
      "epoch": 33.12150093808631,
      "grad_norm": 0.6099264621734619,
      "learning_rate": 0.002828,
      "loss": 1.1291,
      "step": 441344
    },
    {
      "epoch": 33.126303939962476,
      "grad_norm": 0.5848670601844788,
      "learning_rate": 0.002828,
      "loss": 1.1305,
      "step": 441408
    },
    {
      "epoch": 33.13110694183865,
      "grad_norm": 0.668308675289154,
      "learning_rate": 0.002828,
      "loss": 1.1348,
      "step": 441472
    },
    {
      "epoch": 33.13590994371482,
      "grad_norm": 0.6490448117256165,
      "learning_rate": 0.002828,
      "loss": 1.1398,
      "step": 441536
    },
    {
      "epoch": 33.14071294559099,
      "grad_norm": 0.5751400589942932,
      "learning_rate": 0.002828,
      "loss": 1.1296,
      "step": 441600
    },
    {
      "epoch": 33.14551594746717,
      "grad_norm": 0.6624541282653809,
      "learning_rate": 0.002828,
      "loss": 1.1303,
      "step": 441664
    },
    {
      "epoch": 33.15031894934334,
      "grad_norm": 0.5531780123710632,
      "learning_rate": 0.002828,
      "loss": 1.1377,
      "step": 441728
    },
    {
      "epoch": 33.15512195121951,
      "grad_norm": 0.5842620730400085,
      "learning_rate": 0.002828,
      "loss": 1.1351,
      "step": 441792
    },
    {
      "epoch": 33.15992495309568,
      "grad_norm": 0.6528531908988953,
      "learning_rate": 0.002828,
      "loss": 1.1372,
      "step": 441856
    },
    {
      "epoch": 33.16472795497186,
      "grad_norm": 0.6758348941802979,
      "learning_rate": 0.002828,
      "loss": 1.1381,
      "step": 441920
    },
    {
      "epoch": 33.16953095684803,
      "grad_norm": 0.7188329100608826,
      "learning_rate": 0.002828,
      "loss": 1.1354,
      "step": 441984
    },
    {
      "epoch": 33.174333958724205,
      "grad_norm": 0.8271748423576355,
      "learning_rate": 0.002828,
      "loss": 1.1332,
      "step": 442048
    },
    {
      "epoch": 33.179136960600374,
      "grad_norm": 0.6577697992324829,
      "learning_rate": 0.002828,
      "loss": 1.136,
      "step": 442112
    },
    {
      "epoch": 33.18393996247655,
      "grad_norm": 0.5746137499809265,
      "learning_rate": 0.002828,
      "loss": 1.1386,
      "step": 442176
    },
    {
      "epoch": 33.18874296435272,
      "grad_norm": 0.6213378310203552,
      "learning_rate": 0.002828,
      "loss": 1.1384,
      "step": 442240
    },
    {
      "epoch": 33.193545966228896,
      "grad_norm": 0.5850825309753418,
      "learning_rate": 0.002828,
      "loss": 1.1349,
      "step": 442304
    },
    {
      "epoch": 33.198348968105066,
      "grad_norm": 0.525640606880188,
      "learning_rate": 0.002828,
      "loss": 1.1373,
      "step": 442368
    },
    {
      "epoch": 33.203151969981235,
      "grad_norm": 0.575154721736908,
      "learning_rate": 0.002828,
      "loss": 1.1338,
      "step": 442432
    },
    {
      "epoch": 33.20795497185741,
      "grad_norm": 0.5524837374687195,
      "learning_rate": 0.002828,
      "loss": 1.1409,
      "step": 442496
    },
    {
      "epoch": 33.21275797373358,
      "grad_norm": 0.568781852722168,
      "learning_rate": 0.002828,
      "loss": 1.1369,
      "step": 442560
    },
    {
      "epoch": 33.21756097560976,
      "grad_norm": 0.7200085520744324,
      "learning_rate": 0.002828,
      "loss": 1.1332,
      "step": 442624
    },
    {
      "epoch": 33.22236397748593,
      "grad_norm": 0.564530611038208,
      "learning_rate": 0.002828,
      "loss": 1.142,
      "step": 442688
    },
    {
      "epoch": 33.2271669793621,
      "grad_norm": 0.7461503744125366,
      "learning_rate": 0.002828,
      "loss": 1.1355,
      "step": 442752
    },
    {
      "epoch": 33.23196998123827,
      "grad_norm": 0.564323365688324,
      "learning_rate": 0.002828,
      "loss": 1.1374,
      "step": 442816
    },
    {
      "epoch": 33.23677298311445,
      "grad_norm": 0.595663845539093,
      "learning_rate": 0.002828,
      "loss": 1.1365,
      "step": 442880
    },
    {
      "epoch": 33.24157598499062,
      "grad_norm": 0.7581208348274231,
      "learning_rate": 0.002828,
      "loss": 1.1421,
      "step": 442944
    },
    {
      "epoch": 33.246378986866794,
      "grad_norm": 0.581870973110199,
      "learning_rate": 0.002828,
      "loss": 1.1392,
      "step": 443008
    },
    {
      "epoch": 33.251181988742964,
      "grad_norm": 0.7423933148384094,
      "learning_rate": 0.002828,
      "loss": 1.1411,
      "step": 443072
    },
    {
      "epoch": 33.25598499061914,
      "grad_norm": 0.496860146522522,
      "learning_rate": 0.002828,
      "loss": 1.1322,
      "step": 443136
    },
    {
      "epoch": 33.26078799249531,
      "grad_norm": 0.5430870056152344,
      "learning_rate": 0.002828,
      "loss": 1.1357,
      "step": 443200
    },
    {
      "epoch": 33.26559099437148,
      "grad_norm": 0.5885310769081116,
      "learning_rate": 0.002828,
      "loss": 1.1435,
      "step": 443264
    },
    {
      "epoch": 33.270393996247655,
      "grad_norm": 0.5474235415458679,
      "learning_rate": 0.002828,
      "loss": 1.1409,
      "step": 443328
    },
    {
      "epoch": 33.275196998123825,
      "grad_norm": 0.5819973349571228,
      "learning_rate": 0.002828,
      "loss": 1.1409,
      "step": 443392
    },
    {
      "epoch": 33.28,
      "grad_norm": 0.5733329653739929,
      "learning_rate": 0.002828,
      "loss": 1.1401,
      "step": 443456
    },
    {
      "epoch": 33.28480300187617,
      "grad_norm": 0.5910017490386963,
      "learning_rate": 0.002828,
      "loss": 1.1366,
      "step": 443520
    },
    {
      "epoch": 33.28960600375235,
      "grad_norm": 0.6527971029281616,
      "learning_rate": 0.002828,
      "loss": 1.1372,
      "step": 443584
    },
    {
      "epoch": 33.294409005628516,
      "grad_norm": 0.5934697389602661,
      "learning_rate": 0.002828,
      "loss": 1.1426,
      "step": 443648
    },
    {
      "epoch": 33.29921200750469,
      "grad_norm": 0.6255312561988831,
      "learning_rate": 0.002828,
      "loss": 1.1405,
      "step": 443712
    },
    {
      "epoch": 33.30401500938086,
      "grad_norm": 0.657773494720459,
      "learning_rate": 0.002828,
      "loss": 1.1419,
      "step": 443776
    },
    {
      "epoch": 33.30881801125704,
      "grad_norm": 0.6253570914268494,
      "learning_rate": 0.002828,
      "loss": 1.1394,
      "step": 443840
    },
    {
      "epoch": 33.31362101313321,
      "grad_norm": 0.7471127510070801,
      "learning_rate": 0.002828,
      "loss": 1.1382,
      "step": 443904
    },
    {
      "epoch": 33.318424015009384,
      "grad_norm": 0.6400924921035767,
      "learning_rate": 0.002828,
      "loss": 1.1447,
      "step": 443968
    },
    {
      "epoch": 33.323227016885554,
      "grad_norm": 0.5876902937889099,
      "learning_rate": 0.002828,
      "loss": 1.137,
      "step": 444032
    },
    {
      "epoch": 33.32803001876172,
      "grad_norm": 0.566333532333374,
      "learning_rate": 0.002828,
      "loss": 1.139,
      "step": 444096
    },
    {
      "epoch": 33.3328330206379,
      "grad_norm": 0.4966079294681549,
      "learning_rate": 0.002828,
      "loss": 1.1432,
      "step": 444160
    },
    {
      "epoch": 33.33763602251407,
      "grad_norm": 0.5877018570899963,
      "learning_rate": 0.002828,
      "loss": 1.1398,
      "step": 444224
    },
    {
      "epoch": 33.342439024390245,
      "grad_norm": 0.5185602307319641,
      "learning_rate": 0.002828,
      "loss": 1.139,
      "step": 444288
    },
    {
      "epoch": 33.347242026266414,
      "grad_norm": 0.8135582804679871,
      "learning_rate": 0.002828,
      "loss": 1.1387,
      "step": 444352
    },
    {
      "epoch": 33.35204502814259,
      "grad_norm": 0.6387743353843689,
      "learning_rate": 0.002828,
      "loss": 1.1381,
      "step": 444416
    },
    {
      "epoch": 33.35684803001876,
      "grad_norm": 0.5668075084686279,
      "learning_rate": 0.002828,
      "loss": 1.1435,
      "step": 444480
    },
    {
      "epoch": 33.36165103189494,
      "grad_norm": 0.7903865575790405,
      "learning_rate": 0.002828,
      "loss": 1.14,
      "step": 444544
    },
    {
      "epoch": 33.366454033771106,
      "grad_norm": 0.6860894560813904,
      "learning_rate": 0.002828,
      "loss": 1.1449,
      "step": 444608
    },
    {
      "epoch": 33.37125703564728,
      "grad_norm": 0.6163083910942078,
      "learning_rate": 0.002828,
      "loss": 1.1408,
      "step": 444672
    },
    {
      "epoch": 33.37606003752345,
      "grad_norm": 0.5252005457878113,
      "learning_rate": 0.002828,
      "loss": 1.1445,
      "step": 444736
    },
    {
      "epoch": 33.38086303939963,
      "grad_norm": 0.6299996972084045,
      "learning_rate": 0.002828,
      "loss": 1.1385,
      "step": 444800
    },
    {
      "epoch": 33.3856660412758,
      "grad_norm": 0.6038787961006165,
      "learning_rate": 0.002828,
      "loss": 1.1419,
      "step": 444864
    },
    {
      "epoch": 33.39046904315197,
      "grad_norm": 0.6690816879272461,
      "learning_rate": 0.002828,
      "loss": 1.1429,
      "step": 444928
    },
    {
      "epoch": 33.39527204502814,
      "grad_norm": 0.4716529846191406,
      "learning_rate": 0.002828,
      "loss": 1.1408,
      "step": 444992
    },
    {
      "epoch": 33.40007504690431,
      "grad_norm": 0.5922883749008179,
      "learning_rate": 0.002828,
      "loss": 1.1458,
      "step": 445056
    },
    {
      "epoch": 33.40487804878049,
      "grad_norm": 0.6195250153541565,
      "learning_rate": 0.002828,
      "loss": 1.1429,
      "step": 445120
    },
    {
      "epoch": 33.40968105065666,
      "grad_norm": 0.6857700943946838,
      "learning_rate": 0.002828,
      "loss": 1.1406,
      "step": 445184
    },
    {
      "epoch": 33.414484052532835,
      "grad_norm": 0.5511137843132019,
      "learning_rate": 0.002828,
      "loss": 1.1419,
      "step": 445248
    },
    {
      "epoch": 33.419287054409004,
      "grad_norm": 0.6851285099983215,
      "learning_rate": 0.002828,
      "loss": 1.1423,
      "step": 445312
    },
    {
      "epoch": 33.42409005628518,
      "grad_norm": 0.7443287372589111,
      "learning_rate": 0.002828,
      "loss": 1.1462,
      "step": 445376
    },
    {
      "epoch": 33.42889305816135,
      "grad_norm": 0.5800443887710571,
      "learning_rate": 0.002828,
      "loss": 1.1457,
      "step": 445440
    },
    {
      "epoch": 33.433696060037526,
      "grad_norm": 0.5672863125801086,
      "learning_rate": 0.002828,
      "loss": 1.1464,
      "step": 445504
    },
    {
      "epoch": 33.438499061913696,
      "grad_norm": 0.5842262506484985,
      "learning_rate": 0.002828,
      "loss": 1.1483,
      "step": 445568
    },
    {
      "epoch": 33.44330206378987,
      "grad_norm": 0.5936047434806824,
      "learning_rate": 0.002828,
      "loss": 1.1433,
      "step": 445632
    },
    {
      "epoch": 33.44810506566604,
      "grad_norm": 0.6675282120704651,
      "learning_rate": 0.002828,
      "loss": 1.1482,
      "step": 445696
    },
    {
      "epoch": 33.45290806754221,
      "grad_norm": 0.6432201266288757,
      "learning_rate": 0.002828,
      "loss": 1.1464,
      "step": 445760
    },
    {
      "epoch": 33.45771106941839,
      "grad_norm": 0.5785965323448181,
      "learning_rate": 0.002828,
      "loss": 1.1421,
      "step": 445824
    },
    {
      "epoch": 33.46251407129456,
      "grad_norm": 0.6529351472854614,
      "learning_rate": 0.002828,
      "loss": 1.1452,
      "step": 445888
    },
    {
      "epoch": 33.46731707317073,
      "grad_norm": 0.6261816620826721,
      "learning_rate": 0.002828,
      "loss": 1.1392,
      "step": 445952
    },
    {
      "epoch": 33.4721200750469,
      "grad_norm": 0.6078612804412842,
      "learning_rate": 0.002828,
      "loss": 1.1529,
      "step": 446016
    },
    {
      "epoch": 33.47692307692308,
      "grad_norm": 0.7155763506889343,
      "learning_rate": 0.002828,
      "loss": 1.1392,
      "step": 446080
    },
    {
      "epoch": 33.48172607879925,
      "grad_norm": 0.6430063843727112,
      "learning_rate": 0.002828,
      "loss": 1.1464,
      "step": 446144
    },
    {
      "epoch": 33.486529080675425,
      "grad_norm": 0.5534988641738892,
      "learning_rate": 0.002828,
      "loss": 1.1417,
      "step": 446208
    },
    {
      "epoch": 33.491332082551594,
      "grad_norm": 0.5888988971710205,
      "learning_rate": 0.002828,
      "loss": 1.1525,
      "step": 446272
    },
    {
      "epoch": 33.49613508442777,
      "grad_norm": 0.5352635383605957,
      "learning_rate": 0.002828,
      "loss": 1.145,
      "step": 446336
    },
    {
      "epoch": 33.50093808630394,
      "grad_norm": 0.4819788336753845,
      "learning_rate": 0.002828,
      "loss": 1.1435,
      "step": 446400
    },
    {
      "epoch": 33.505741088180116,
      "grad_norm": 0.6127516627311707,
      "learning_rate": 0.002828,
      "loss": 1.1517,
      "step": 446464
    },
    {
      "epoch": 33.510544090056285,
      "grad_norm": 0.6050450801849365,
      "learning_rate": 0.002828,
      "loss": 1.1435,
      "step": 446528
    },
    {
      "epoch": 33.515347091932455,
      "grad_norm": 0.4720316231250763,
      "learning_rate": 0.002828,
      "loss": 1.1456,
      "step": 446592
    },
    {
      "epoch": 33.52015009380863,
      "grad_norm": 0.5383008718490601,
      "learning_rate": 0.002828,
      "loss": 1.1454,
      "step": 446656
    },
    {
      "epoch": 33.5249530956848,
      "grad_norm": 0.6708769798278809,
      "learning_rate": 0.002828,
      "loss": 1.1396,
      "step": 446720
    },
    {
      "epoch": 33.52975609756098,
      "grad_norm": 0.5709279775619507,
      "learning_rate": 0.002828,
      "loss": 1.1475,
      "step": 446784
    },
    {
      "epoch": 33.534559099437146,
      "grad_norm": 0.6338633298873901,
      "learning_rate": 0.002828,
      "loss": 1.1509,
      "step": 446848
    },
    {
      "epoch": 33.53936210131332,
      "grad_norm": 0.6444924473762512,
      "learning_rate": 0.002828,
      "loss": 1.1445,
      "step": 446912
    },
    {
      "epoch": 33.54416510318949,
      "grad_norm": 0.5632554888725281,
      "learning_rate": 0.002828,
      "loss": 1.1482,
      "step": 446976
    },
    {
      "epoch": 33.54896810506567,
      "grad_norm": 0.578456461429596,
      "learning_rate": 0.002828,
      "loss": 1.1448,
      "step": 447040
    },
    {
      "epoch": 33.55377110694184,
      "grad_norm": 0.6251580715179443,
      "learning_rate": 0.002828,
      "loss": 1.1452,
      "step": 447104
    },
    {
      "epoch": 33.558574108818014,
      "grad_norm": 0.6657046675682068,
      "learning_rate": 0.002828,
      "loss": 1.1459,
      "step": 447168
    },
    {
      "epoch": 33.563377110694184,
      "grad_norm": 0.7276448011398315,
      "learning_rate": 0.002828,
      "loss": 1.1472,
      "step": 447232
    },
    {
      "epoch": 33.56818011257036,
      "grad_norm": 0.612055778503418,
      "learning_rate": 0.002828,
      "loss": 1.1442,
      "step": 447296
    },
    {
      "epoch": 33.57298311444653,
      "grad_norm": 0.5227575898170471,
      "learning_rate": 0.002828,
      "loss": 1.1482,
      "step": 447360
    },
    {
      "epoch": 33.5777861163227,
      "grad_norm": 0.6529867053031921,
      "learning_rate": 0.002828,
      "loss": 1.144,
      "step": 447424
    },
    {
      "epoch": 33.582589118198875,
      "grad_norm": 0.5356407165527344,
      "learning_rate": 0.002828,
      "loss": 1.1446,
      "step": 447488
    },
    {
      "epoch": 33.587392120075044,
      "grad_norm": 0.5991299748420715,
      "learning_rate": 0.002828,
      "loss": 1.1451,
      "step": 447552
    },
    {
      "epoch": 33.59219512195122,
      "grad_norm": 0.6475511193275452,
      "learning_rate": 0.002828,
      "loss": 1.1528,
      "step": 447616
    },
    {
      "epoch": 33.59699812382739,
      "grad_norm": 0.6805687546730042,
      "learning_rate": 0.002828,
      "loss": 1.1452,
      "step": 447680
    },
    {
      "epoch": 33.60180112570357,
      "grad_norm": 0.5567482709884644,
      "learning_rate": 0.002828,
      "loss": 1.151,
      "step": 447744
    },
    {
      "epoch": 33.606604127579736,
      "grad_norm": 0.7172110080718994,
      "learning_rate": 0.002828,
      "loss": 1.1463,
      "step": 447808
    },
    {
      "epoch": 33.61140712945591,
      "grad_norm": 0.6430109739303589,
      "learning_rate": 0.002828,
      "loss": 1.1466,
      "step": 447872
    },
    {
      "epoch": 33.61621013133208,
      "grad_norm": 0.525716245174408,
      "learning_rate": 0.002828,
      "loss": 1.1508,
      "step": 447936
    },
    {
      "epoch": 33.62101313320826,
      "grad_norm": 0.6650141477584839,
      "learning_rate": 0.002828,
      "loss": 1.1528,
      "step": 448000
    },
    {
      "epoch": 33.62581613508443,
      "grad_norm": 0.5755172371864319,
      "learning_rate": 0.002828,
      "loss": 1.1478,
      "step": 448064
    },
    {
      "epoch": 33.6306191369606,
      "grad_norm": 0.6707451343536377,
      "learning_rate": 0.002828,
      "loss": 1.1486,
      "step": 448128
    },
    {
      "epoch": 33.63542213883677,
      "grad_norm": 0.5326623916625977,
      "learning_rate": 0.002828,
      "loss": 1.1545,
      "step": 448192
    },
    {
      "epoch": 33.64022514071294,
      "grad_norm": 0.7620773911476135,
      "learning_rate": 0.002828,
      "loss": 1.1525,
      "step": 448256
    },
    {
      "epoch": 33.64502814258912,
      "grad_norm": 0.7047601342201233,
      "learning_rate": 0.002828,
      "loss": 1.1534,
      "step": 448320
    },
    {
      "epoch": 33.64983114446529,
      "grad_norm": 0.6242048740386963,
      "learning_rate": 0.002828,
      "loss": 1.1518,
      "step": 448384
    },
    {
      "epoch": 33.654634146341465,
      "grad_norm": 0.6024118065834045,
      "learning_rate": 0.002828,
      "loss": 1.1533,
      "step": 448448
    },
    {
      "epoch": 33.659437148217634,
      "grad_norm": 0.593969464302063,
      "learning_rate": 0.002828,
      "loss": 1.1434,
      "step": 448512
    },
    {
      "epoch": 33.66424015009381,
      "grad_norm": 0.6917033195495605,
      "learning_rate": 0.002828,
      "loss": 1.1489,
      "step": 448576
    },
    {
      "epoch": 33.66904315196998,
      "grad_norm": 0.6661319136619568,
      "learning_rate": 0.002828,
      "loss": 1.1468,
      "step": 448640
    },
    {
      "epoch": 33.673846153846156,
      "grad_norm": 0.5408964157104492,
      "learning_rate": 0.002828,
      "loss": 1.1471,
      "step": 448704
    },
    {
      "epoch": 33.678649155722326,
      "grad_norm": 0.7085400819778442,
      "learning_rate": 0.002828,
      "loss": 1.1569,
      "step": 448768
    },
    {
      "epoch": 33.6834521575985,
      "grad_norm": 0.6100766658782959,
      "learning_rate": 0.002828,
      "loss": 1.1529,
      "step": 448832
    },
    {
      "epoch": 33.68825515947467,
      "grad_norm": 0.7409677505493164,
      "learning_rate": 0.002828,
      "loss": 1.1514,
      "step": 448896
    },
    {
      "epoch": 33.69305816135084,
      "grad_norm": 0.5764803886413574,
      "learning_rate": 0.002828,
      "loss": 1.1519,
      "step": 448960
    },
    {
      "epoch": 33.69786116322702,
      "grad_norm": 0.6267707943916321,
      "learning_rate": 0.002828,
      "loss": 1.1465,
      "step": 449024
    },
    {
      "epoch": 33.70266416510319,
      "grad_norm": 0.6436747312545776,
      "learning_rate": 0.002828,
      "loss": 1.154,
      "step": 449088
    },
    {
      "epoch": 33.70746716697936,
      "grad_norm": 0.5536535382270813,
      "learning_rate": 0.002828,
      "loss": 1.1471,
      "step": 449152
    },
    {
      "epoch": 33.71227016885553,
      "grad_norm": 0.554177463054657,
      "learning_rate": 0.002828,
      "loss": 1.1482,
      "step": 449216
    },
    {
      "epoch": 33.71707317073171,
      "grad_norm": 0.5661305785179138,
      "learning_rate": 0.002828,
      "loss": 1.1524,
      "step": 449280
    },
    {
      "epoch": 33.72187617260788,
      "grad_norm": 0.6556863188743591,
      "learning_rate": 0.002828,
      "loss": 1.1497,
      "step": 449344
    },
    {
      "epoch": 33.726679174484055,
      "grad_norm": 0.6361242532730103,
      "learning_rate": 0.002828,
      "loss": 1.1536,
      "step": 449408
    },
    {
      "epoch": 33.731482176360224,
      "grad_norm": 0.5205981731414795,
      "learning_rate": 0.002828,
      "loss": 1.1532,
      "step": 449472
    },
    {
      "epoch": 33.7362851782364,
      "grad_norm": 0.6734727025032043,
      "learning_rate": 0.002828,
      "loss": 1.1533,
      "step": 449536
    },
    {
      "epoch": 33.74108818011257,
      "grad_norm": 0.6392199397087097,
      "learning_rate": 0.002828,
      "loss": 1.1498,
      "step": 449600
    },
    {
      "epoch": 33.745891181988746,
      "grad_norm": 0.6905595660209656,
      "learning_rate": 0.002828,
      "loss": 1.147,
      "step": 449664
    },
    {
      "epoch": 33.750694183864915,
      "grad_norm": 0.6617002487182617,
      "learning_rate": 0.002828,
      "loss": 1.1502,
      "step": 449728
    },
    {
      "epoch": 33.755497185741085,
      "grad_norm": 0.5721489191055298,
      "learning_rate": 0.002828,
      "loss": 1.152,
      "step": 449792
    },
    {
      "epoch": 33.76030018761726,
      "grad_norm": 0.6114367842674255,
      "learning_rate": 0.002828,
      "loss": 1.1549,
      "step": 449856
    },
    {
      "epoch": 33.76510318949343,
      "grad_norm": 0.5258558988571167,
      "learning_rate": 0.002828,
      "loss": 1.1524,
      "step": 449920
    },
    {
      "epoch": 33.76990619136961,
      "grad_norm": 0.5574025511741638,
      "learning_rate": 0.002828,
      "loss": 1.1475,
      "step": 449984
    },
    {
      "epoch": 33.774709193245776,
      "grad_norm": 0.544082760810852,
      "learning_rate": 0.002828,
      "loss": 1.1556,
      "step": 450048
    },
    {
      "epoch": 33.77951219512195,
      "grad_norm": 0.6477503776550293,
      "learning_rate": 0.002828,
      "loss": 1.1496,
      "step": 450112
    },
    {
      "epoch": 33.78431519699812,
      "grad_norm": 0.5774228572845459,
      "learning_rate": 0.002828,
      "loss": 1.1572,
      "step": 450176
    },
    {
      "epoch": 33.7891181988743,
      "grad_norm": 0.633441686630249,
      "learning_rate": 0.002828,
      "loss": 1.1545,
      "step": 450240
    },
    {
      "epoch": 33.79392120075047,
      "grad_norm": 0.7636280655860901,
      "learning_rate": 0.002828,
      "loss": 1.1517,
      "step": 450304
    },
    {
      "epoch": 33.798724202626644,
      "grad_norm": 0.5494087934494019,
      "learning_rate": 0.002828,
      "loss": 1.1514,
      "step": 450368
    },
    {
      "epoch": 33.803527204502814,
      "grad_norm": 0.654360294342041,
      "learning_rate": 0.002828,
      "loss": 1.1582,
      "step": 450432
    },
    {
      "epoch": 33.80833020637899,
      "grad_norm": 0.6439893245697021,
      "learning_rate": 0.002828,
      "loss": 1.156,
      "step": 450496
    },
    {
      "epoch": 33.81313320825516,
      "grad_norm": 0.6434905529022217,
      "learning_rate": 0.002828,
      "loss": 1.1461,
      "step": 450560
    },
    {
      "epoch": 33.81793621013133,
      "grad_norm": 0.5695158839225769,
      "learning_rate": 0.002828,
      "loss": 1.1493,
      "step": 450624
    },
    {
      "epoch": 33.822739212007505,
      "grad_norm": 0.6693202257156372,
      "learning_rate": 0.002828,
      "loss": 1.1543,
      "step": 450688
    },
    {
      "epoch": 33.827542213883675,
      "grad_norm": 0.5362828969955444,
      "learning_rate": 0.002828,
      "loss": 1.1537,
      "step": 450752
    },
    {
      "epoch": 33.83234521575985,
      "grad_norm": 0.5842569470405579,
      "learning_rate": 0.002828,
      "loss": 1.1573,
      "step": 450816
    },
    {
      "epoch": 33.83714821763602,
      "grad_norm": 0.8023243546485901,
      "learning_rate": 0.002828,
      "loss": 1.1573,
      "step": 450880
    },
    {
      "epoch": 33.8419512195122,
      "grad_norm": 0.6493667364120483,
      "learning_rate": 0.002828,
      "loss": 1.1456,
      "step": 450944
    },
    {
      "epoch": 33.846754221388366,
      "grad_norm": 0.6712051630020142,
      "learning_rate": 0.002828,
      "loss": 1.1491,
      "step": 451008
    },
    {
      "epoch": 33.85155722326454,
      "grad_norm": 0.7749685645103455,
      "learning_rate": 0.002828,
      "loss": 1.1523,
      "step": 451072
    },
    {
      "epoch": 33.85636022514071,
      "grad_norm": 0.5475408434867859,
      "learning_rate": 0.002828,
      "loss": 1.1497,
      "step": 451136
    },
    {
      "epoch": 33.86116322701689,
      "grad_norm": 0.570763885974884,
      "learning_rate": 0.002828,
      "loss": 1.1498,
      "step": 451200
    },
    {
      "epoch": 33.86596622889306,
      "grad_norm": 0.6396635174751282,
      "learning_rate": 0.002828,
      "loss": 1.1529,
      "step": 451264
    },
    {
      "epoch": 33.870769230769234,
      "grad_norm": 0.6525161862373352,
      "learning_rate": 0.002828,
      "loss": 1.1569,
      "step": 451328
    },
    {
      "epoch": 33.8755722326454,
      "grad_norm": 0.6440078020095825,
      "learning_rate": 0.002828,
      "loss": 1.1531,
      "step": 451392
    },
    {
      "epoch": 33.88037523452157,
      "grad_norm": 0.557655930519104,
      "learning_rate": 0.002828,
      "loss": 1.1514,
      "step": 451456
    },
    {
      "epoch": 33.88517823639775,
      "grad_norm": 0.6325395107269287,
      "learning_rate": 0.002828,
      "loss": 1.1484,
      "step": 451520
    },
    {
      "epoch": 33.88998123827392,
      "grad_norm": 0.6961609721183777,
      "learning_rate": 0.002828,
      "loss": 1.1562,
      "step": 451584
    },
    {
      "epoch": 33.894784240150095,
      "grad_norm": 0.6224822402000427,
      "learning_rate": 0.002828,
      "loss": 1.1533,
      "step": 451648
    },
    {
      "epoch": 33.899587242026264,
      "grad_norm": 0.5128984451293945,
      "learning_rate": 0.002828,
      "loss": 1.1521,
      "step": 451712
    },
    {
      "epoch": 33.90439024390244,
      "grad_norm": 0.5620936155319214,
      "learning_rate": 0.002828,
      "loss": 1.7302,
      "step": 451776
    },
    {
      "epoch": 33.90919324577861,
      "grad_norm": 0.5801410675048828,
      "learning_rate": 0.002828,
      "loss": 3.5526,
      "step": 451840
    },
    {
      "epoch": 33.91399624765479,
      "grad_norm": 0.5933526158332825,
      "learning_rate": 0.002828,
      "loss": 3.1783,
      "step": 451904
    },
    {
      "epoch": 33.918799249530956,
      "grad_norm": 0.6869750618934631,
      "learning_rate": 0.002828,
      "loss": 3.4,
      "step": 451968
    },
    {
      "epoch": 33.92360225140713,
      "grad_norm": 0.5913759469985962,
      "learning_rate": 0.002828,
      "loss": 3.3694,
      "step": 452032
    },
    {
      "epoch": 33.9284052532833,
      "grad_norm": 0.5191683173179626,
      "learning_rate": 0.002828,
      "loss": 3.4183,
      "step": 452096
    },
    {
      "epoch": 33.93320825515948,
      "grad_norm": 0.5402680039405823,
      "learning_rate": 0.002828,
      "loss": 3.5793,
      "step": 452160
    },
    {
      "epoch": 33.93801125703565,
      "grad_norm": 0.5815765857696533,
      "learning_rate": 0.002828,
      "loss": 3.2655,
      "step": 452224
    },
    {
      "epoch": 33.94281425891182,
      "grad_norm": 0.5683310627937317,
      "learning_rate": 0.002828,
      "loss": 3.0861,
      "step": 452288
    },
    {
      "epoch": 33.94761726078799,
      "grad_norm": 0.6940186619758606,
      "learning_rate": 0.002828,
      "loss": 2.9187,
      "step": 452352
    },
    {
      "epoch": 33.95242026266416,
      "grad_norm": 0.5648021697998047,
      "learning_rate": 0.002828,
      "loss": 2.7537,
      "step": 452416
    },
    {
      "epoch": 33.95722326454034,
      "grad_norm": 0.6573771834373474,
      "learning_rate": 0.002828,
      "loss": 2.71,
      "step": 452480
    },
    {
      "epoch": 33.96202626641651,
      "grad_norm": 0.645450234413147,
      "learning_rate": 0.002828,
      "loss": 2.6809,
      "step": 452544
    },
    {
      "epoch": 33.966829268292685,
      "grad_norm": 0.5694723725318909,
      "learning_rate": 0.002828,
      "loss": 2.8897,
      "step": 452608
    },
    {
      "epoch": 33.971632270168854,
      "grad_norm": 0.6184071898460388,
      "learning_rate": 0.002828,
      "loss": 2.9641,
      "step": 452672
    },
    {
      "epoch": 33.97643527204503,
      "grad_norm": 0.7220840454101562,
      "learning_rate": 0.002828,
      "loss": 2.8128,
      "step": 452736
    },
    {
      "epoch": 33.9812382739212,
      "grad_norm": 0.7192930579185486,
      "learning_rate": 0.002828,
      "loss": 2.7682,
      "step": 452800
    },
    {
      "epoch": 33.986041275797376,
      "grad_norm": 0.7508559226989746,
      "learning_rate": 0.002828,
      "loss": 2.768,
      "step": 452864
    },
    {
      "epoch": 33.990844277673546,
      "grad_norm": 0.5012671947479248,
      "learning_rate": 0.002828,
      "loss": 2.6946,
      "step": 452928
    },
    {
      "epoch": 33.99564727954972,
      "grad_norm": 0.5871996283531189,
      "learning_rate": 0.002828,
      "loss": 2.6686,
      "step": 452992
    },
    {
      "epoch": 34.00045028142589,
      "grad_norm": 0.6014786958694458,
      "learning_rate": 0.002828,
      "loss": 2.6525,
      "step": 453056
    },
    {
      "epoch": 34.00525328330206,
      "grad_norm": 0.6529600024223328,
      "learning_rate": 0.002828,
      "loss": 2.618,
      "step": 453120
    },
    {
      "epoch": 34.01005628517824,
      "grad_norm": 0.6394699215888977,
      "learning_rate": 0.002828,
      "loss": 2.6139,
      "step": 453184
    },
    {
      "epoch": 34.014859287054406,
      "grad_norm": 0.6919693350791931,
      "learning_rate": 0.002828,
      "loss": 2.5977,
      "step": 453248
    },
    {
      "epoch": 34.01966228893058,
      "grad_norm": 0.5869269967079163,
      "learning_rate": 0.002828,
      "loss": 2.579,
      "step": 453312
    },
    {
      "epoch": 34.02446529080675,
      "grad_norm": 0.6233799457550049,
      "learning_rate": 0.002828,
      "loss": 2.575,
      "step": 453376
    },
    {
      "epoch": 34.02926829268293,
      "grad_norm": 0.7056044340133667,
      "learning_rate": 0.002828,
      "loss": 2.568,
      "step": 453440
    },
    {
      "epoch": 34.0340712945591,
      "grad_norm": 0.6083732843399048,
      "learning_rate": 0.002828,
      "loss": 2.5443,
      "step": 453504
    },
    {
      "epoch": 34.038874296435274,
      "grad_norm": 0.6978457570075989,
      "learning_rate": 0.002828,
      "loss": 2.5126,
      "step": 453568
    },
    {
      "epoch": 34.043677298311444,
      "grad_norm": 0.6388948559761047,
      "learning_rate": 0.002828,
      "loss": 2.5081,
      "step": 453632
    },
    {
      "epoch": 34.04848030018762,
      "grad_norm": 0.5888407230377197,
      "learning_rate": 0.002828,
      "loss": 2.4963,
      "step": 453696
    },
    {
      "epoch": 34.05328330206379,
      "grad_norm": 0.5932326912879944,
      "learning_rate": 0.002828,
      "loss": 2.4559,
      "step": 453760
    },
    {
      "epoch": 34.058086303939966,
      "grad_norm": 0.7458741664886475,
      "learning_rate": 0.002828,
      "loss": 2.4359,
      "step": 453824
    },
    {
      "epoch": 34.062889305816135,
      "grad_norm": 0.5495181679725647,
      "learning_rate": 0.002828,
      "loss": 2.3964,
      "step": 453888
    },
    {
      "epoch": 34.067692307692305,
      "grad_norm": 0.5372458696365356,
      "learning_rate": 0.002828,
      "loss": 2.4007,
      "step": 453952
    },
    {
      "epoch": 34.07249530956848,
      "grad_norm": 0.525921106338501,
      "learning_rate": 0.002828,
      "loss": 2.3913,
      "step": 454016
    },
    {
      "epoch": 34.07729831144465,
      "grad_norm": 0.6002963781356812,
      "learning_rate": 0.002828,
      "loss": 2.2537,
      "step": 454080
    },
    {
      "epoch": 34.08210131332083,
      "grad_norm": 0.6559569239616394,
      "learning_rate": 0.002828,
      "loss": 2.1257,
      "step": 454144
    },
    {
      "epoch": 34.086904315196996,
      "grad_norm": 0.6380961537361145,
      "learning_rate": 0.002828,
      "loss": 2.0642,
      "step": 454208
    },
    {
      "epoch": 34.09170731707317,
      "grad_norm": 0.6877055168151855,
      "learning_rate": 0.002828,
      "loss": 2.0349,
      "step": 454272
    },
    {
      "epoch": 34.09651031894934,
      "grad_norm": 0.6221262216567993,
      "learning_rate": 0.002828,
      "loss": 1.9857,
      "step": 454336
    },
    {
      "epoch": 34.10131332082552,
      "grad_norm": 0.7388865351676941,
      "learning_rate": 0.002828,
      "loss": 1.9721,
      "step": 454400
    },
    {
      "epoch": 34.10611632270169,
      "grad_norm": 0.5388085246086121,
      "learning_rate": 0.002828,
      "loss": 1.9403,
      "step": 454464
    },
    {
      "epoch": 34.110919324577864,
      "grad_norm": 0.4879409074783325,
      "learning_rate": 0.002828,
      "loss": 1.9614,
      "step": 454528
    },
    {
      "epoch": 34.11572232645403,
      "grad_norm": 0.5843081474304199,
      "learning_rate": 0.002828,
      "loss": 1.9817,
      "step": 454592
    },
    {
      "epoch": 34.12052532833021,
      "grad_norm": 0.671530544757843,
      "learning_rate": 0.002828,
      "loss": 2.1895,
      "step": 454656
    },
    {
      "epoch": 34.12532833020638,
      "grad_norm": 0.6688728928565979,
      "learning_rate": 0.002828,
      "loss": 2.1792,
      "step": 454720
    },
    {
      "epoch": 34.13013133208255,
      "grad_norm": 0.524191677570343,
      "learning_rate": 0.002828,
      "loss": 2.1088,
      "step": 454784
    },
    {
      "epoch": 34.134934333958725,
      "grad_norm": 0.5124364495277405,
      "learning_rate": 0.002828,
      "loss": 2.0359,
      "step": 454848
    },
    {
      "epoch": 34.139737335834894,
      "grad_norm": 0.5780380964279175,
      "learning_rate": 0.002828,
      "loss": 1.9583,
      "step": 454912
    },
    {
      "epoch": 34.14454033771107,
      "grad_norm": 0.5620104074478149,
      "learning_rate": 0.002828,
      "loss": 1.9113,
      "step": 454976
    },
    {
      "epoch": 34.14934333958724,
      "grad_norm": 0.7543662786483765,
      "learning_rate": 0.002828,
      "loss": 1.9022,
      "step": 455040
    },
    {
      "epoch": 34.15414634146342,
      "grad_norm": 0.6978630423545837,
      "learning_rate": 0.002828,
      "loss": 1.8686,
      "step": 455104
    },
    {
      "epoch": 34.158949343339586,
      "grad_norm": 0.6158016920089722,
      "learning_rate": 0.002828,
      "loss": 1.833,
      "step": 455168
    },
    {
      "epoch": 34.16375234521576,
      "grad_norm": 0.5737127065658569,
      "learning_rate": 0.002828,
      "loss": 1.8432,
      "step": 455232
    },
    {
      "epoch": 34.16855534709193,
      "grad_norm": 0.5505557060241699,
      "learning_rate": 0.002828,
      "loss": 1.7996,
      "step": 455296
    },
    {
      "epoch": 34.17335834896811,
      "grad_norm": 0.7005407214164734,
      "learning_rate": 0.002828,
      "loss": 1.7848,
      "step": 455360
    },
    {
      "epoch": 34.17816135084428,
      "grad_norm": 0.6658907532691956,
      "learning_rate": 0.002828,
      "loss": 1.7505,
      "step": 455424
    },
    {
      "epoch": 34.18296435272045,
      "grad_norm": 0.6440961360931396,
      "learning_rate": 0.002828,
      "loss": 1.7453,
      "step": 455488
    },
    {
      "epoch": 34.18776735459662,
      "grad_norm": 0.5199069380760193,
      "learning_rate": 0.002828,
      "loss": 1.721,
      "step": 455552
    },
    {
      "epoch": 34.19257035647279,
      "grad_norm": 0.6788700222969055,
      "learning_rate": 0.002828,
      "loss": 1.7196,
      "step": 455616
    },
    {
      "epoch": 34.19737335834897,
      "grad_norm": 0.6521267294883728,
      "learning_rate": 0.002828,
      "loss": 1.676,
      "step": 455680
    },
    {
      "epoch": 34.20217636022514,
      "grad_norm": 0.5314558148384094,
      "learning_rate": 0.002828,
      "loss": 1.678,
      "step": 455744
    },
    {
      "epoch": 34.206979362101315,
      "grad_norm": 0.8381668925285339,
      "learning_rate": 0.002828,
      "loss": 1.6474,
      "step": 455808
    },
    {
      "epoch": 34.211782363977484,
      "grad_norm": 0.6929973363876343,
      "learning_rate": 0.002828,
      "loss": 1.6476,
      "step": 455872
    },
    {
      "epoch": 34.21658536585366,
      "grad_norm": 0.594032347202301,
      "learning_rate": 0.002828,
      "loss": 1.6283,
      "step": 455936
    },
    {
      "epoch": 34.22138836772983,
      "grad_norm": 0.7464234828948975,
      "learning_rate": 0.002828,
      "loss": 1.6262,
      "step": 456000
    },
    {
      "epoch": 34.226191369606006,
      "grad_norm": 0.6788607239723206,
      "learning_rate": 0.002828,
      "loss": 1.6083,
      "step": 456064
    },
    {
      "epoch": 34.230994371482176,
      "grad_norm": 0.6504068970680237,
      "learning_rate": 0.002828,
      "loss": 1.6146,
      "step": 456128
    },
    {
      "epoch": 34.23579737335835,
      "grad_norm": 0.6420795917510986,
      "learning_rate": 0.002828,
      "loss": 1.6126,
      "step": 456192
    },
    {
      "epoch": 34.24060037523452,
      "grad_norm": 0.6302408576011658,
      "learning_rate": 0.002828,
      "loss": 1.5947,
      "step": 456256
    },
    {
      "epoch": 34.24540337711069,
      "grad_norm": 0.7099950909614563,
      "learning_rate": 0.002828,
      "loss": 1.5792,
      "step": 456320
    },
    {
      "epoch": 34.25020637898687,
      "grad_norm": 0.6178498268127441,
      "learning_rate": 0.002828,
      "loss": 1.5584,
      "step": 456384
    },
    {
      "epoch": 34.25500938086304,
      "grad_norm": 0.6391167640686035,
      "learning_rate": 0.002828,
      "loss": 1.5621,
      "step": 456448
    },
    {
      "epoch": 34.25981238273921,
      "grad_norm": 0.48477670550346375,
      "learning_rate": 0.002828,
      "loss": 1.547,
      "step": 456512
    },
    {
      "epoch": 34.26461538461538,
      "grad_norm": 0.5507023334503174,
      "learning_rate": 0.002828,
      "loss": 1.5239,
      "step": 456576
    },
    {
      "epoch": 34.26941838649156,
      "grad_norm": 0.6492827534675598,
      "learning_rate": 0.002828,
      "loss": 1.5009,
      "step": 456640
    },
    {
      "epoch": 34.27422138836773,
      "grad_norm": 0.5808483362197876,
      "learning_rate": 0.002828,
      "loss": 1.747,
      "step": 456704
    },
    {
      "epoch": 34.279024390243904,
      "grad_norm": 0.7006205916404724,
      "learning_rate": 0.002828,
      "loss": 1.5044,
      "step": 456768
    },
    {
      "epoch": 34.283827392120074,
      "grad_norm": 0.5767326951026917,
      "learning_rate": 0.002828,
      "loss": 1.4884,
      "step": 456832
    },
    {
      "epoch": 34.28863039399625,
      "grad_norm": 0.5823490619659424,
      "learning_rate": 0.002828,
      "loss": 1.5015,
      "step": 456896
    },
    {
      "epoch": 34.29343339587242,
      "grad_norm": 0.6484927535057068,
      "learning_rate": 0.002828,
      "loss": 1.4816,
      "step": 456960
    },
    {
      "epoch": 34.298236397748596,
      "grad_norm": 0.5902293920516968,
      "learning_rate": 0.002828,
      "loss": 1.4473,
      "step": 457024
    },
    {
      "epoch": 34.303039399624765,
      "grad_norm": 0.6143577694892883,
      "learning_rate": 0.002828,
      "loss": 1.4276,
      "step": 457088
    },
    {
      "epoch": 34.307842401500935,
      "grad_norm": 0.6154422760009766,
      "learning_rate": 0.002828,
      "loss": 1.642,
      "step": 457152
    },
    {
      "epoch": 34.31264540337711,
      "grad_norm": 0.5513522624969482,
      "learning_rate": 0.002828,
      "loss": 1.4159,
      "step": 457216
    },
    {
      "epoch": 34.31744840525328,
      "grad_norm": 0.6686850190162659,
      "learning_rate": 0.002828,
      "loss": 1.4497,
      "step": 457280
    },
    {
      "epoch": 34.32225140712946,
      "grad_norm": 0.5761899948120117,
      "learning_rate": 0.002828,
      "loss": 1.3841,
      "step": 457344
    },
    {
      "epoch": 34.327054409005626,
      "grad_norm": 0.6581642031669617,
      "learning_rate": 0.002828,
      "loss": 1.3786,
      "step": 457408
    },
    {
      "epoch": 34.3318574108818,
      "grad_norm": 0.6536948680877686,
      "learning_rate": 0.002828,
      "loss": 1.4748,
      "step": 457472
    },
    {
      "epoch": 34.33666041275797,
      "grad_norm": 0.6962592601776123,
      "learning_rate": 0.002828,
      "loss": 1.5362,
      "step": 457536
    },
    {
      "epoch": 34.34146341463415,
      "grad_norm": 0.6023150086402893,
      "learning_rate": 0.002828,
      "loss": 1.3787,
      "step": 457600
    },
    {
      "epoch": 34.34626641651032,
      "grad_norm": 0.654251754283905,
      "learning_rate": 0.002828,
      "loss": 1.4322,
      "step": 457664
    },
    {
      "epoch": 34.351069418386494,
      "grad_norm": 0.5367180109024048,
      "learning_rate": 0.002828,
      "loss": 1.6064,
      "step": 457728
    },
    {
      "epoch": 34.35587242026266,
      "grad_norm": 0.5423946976661682,
      "learning_rate": 0.002828,
      "loss": 1.3929,
      "step": 457792
    },
    {
      "epoch": 34.36067542213884,
      "grad_norm": 0.5561156272888184,
      "learning_rate": 0.002828,
      "loss": 1.33,
      "step": 457856
    },
    {
      "epoch": 34.36547842401501,
      "grad_norm": 0.5509253740310669,
      "learning_rate": 0.002828,
      "loss": 1.3037,
      "step": 457920
    },
    {
      "epoch": 34.37028142589118,
      "grad_norm": 0.6430239677429199,
      "learning_rate": 0.002828,
      "loss": 1.2854,
      "step": 457984
    },
    {
      "epoch": 34.375084427767355,
      "grad_norm": 0.5822476148605347,
      "learning_rate": 0.002828,
      "loss": 1.2722,
      "step": 458048
    },
    {
      "epoch": 34.379887429643524,
      "grad_norm": 0.6652078032493591,
      "learning_rate": 0.002828,
      "loss": 1.2584,
      "step": 458112
    },
    {
      "epoch": 34.3846904315197,
      "grad_norm": 0.5696972608566284,
      "learning_rate": 0.002828,
      "loss": 1.2443,
      "step": 458176
    },
    {
      "epoch": 34.38949343339587,
      "grad_norm": 0.5622748732566833,
      "learning_rate": 0.002828,
      "loss": 1.237,
      "step": 458240
    },
    {
      "epoch": 34.39429643527205,
      "grad_norm": 0.5725105404853821,
      "learning_rate": 0.002828,
      "loss": 1.2314,
      "step": 458304
    },
    {
      "epoch": 34.399099437148216,
      "grad_norm": 0.6276898980140686,
      "learning_rate": 0.002828,
      "loss": 1.2195,
      "step": 458368
    },
    {
      "epoch": 34.40390243902439,
      "grad_norm": 0.6529742479324341,
      "learning_rate": 0.002828,
      "loss": 1.2157,
      "step": 458432
    },
    {
      "epoch": 34.40870544090056,
      "grad_norm": 0.5537835955619812,
      "learning_rate": 0.002828,
      "loss": 1.2123,
      "step": 458496
    },
    {
      "epoch": 34.41350844277674,
      "grad_norm": 0.4857408404350281,
      "learning_rate": 0.002828,
      "loss": 1.1988,
      "step": 458560
    },
    {
      "epoch": 34.41831144465291,
      "grad_norm": 0.6616053581237793,
      "learning_rate": 0.002828,
      "loss": 1.2019,
      "step": 458624
    },
    {
      "epoch": 34.423114446529084,
      "grad_norm": 0.6578753590583801,
      "learning_rate": 0.002828,
      "loss": 1.1991,
      "step": 458688
    },
    {
      "epoch": 34.42791744840525,
      "grad_norm": 0.5510697364807129,
      "learning_rate": 0.002828,
      "loss": 1.196,
      "step": 458752
    },
    {
      "epoch": 34.43272045028142,
      "grad_norm": 0.7690936326980591,
      "learning_rate": 0.002828,
      "loss": 1.1898,
      "step": 458816
    },
    {
      "epoch": 34.4375234521576,
      "grad_norm": 0.6071969270706177,
      "learning_rate": 0.002828,
      "loss": 1.1845,
      "step": 458880
    },
    {
      "epoch": 34.44232645403377,
      "grad_norm": 0.773906409740448,
      "learning_rate": 0.002828,
      "loss": 1.1851,
      "step": 458944
    },
    {
      "epoch": 34.447129455909945,
      "grad_norm": 0.5726880431175232,
      "learning_rate": 0.002828,
      "loss": 1.184,
      "step": 459008
    },
    {
      "epoch": 34.451932457786114,
      "grad_norm": 0.5598574876785278,
      "learning_rate": 0.002828,
      "loss": 1.1842,
      "step": 459072
    },
    {
      "epoch": 34.45673545966229,
      "grad_norm": 0.6979678273200989,
      "learning_rate": 0.002828,
      "loss": 1.1765,
      "step": 459136
    },
    {
      "epoch": 34.46153846153846,
      "grad_norm": 0.6438438296318054,
      "learning_rate": 0.002828,
      "loss": 1.1765,
      "step": 459200
    },
    {
      "epoch": 34.466341463414636,
      "grad_norm": 0.6009986400604248,
      "learning_rate": 0.002828,
      "loss": 1.1695,
      "step": 459264
    },
    {
      "epoch": 34.471144465290806,
      "grad_norm": 0.8207782506942749,
      "learning_rate": 0.002828,
      "loss": 1.1736,
      "step": 459328
    },
    {
      "epoch": 34.47594746716698,
      "grad_norm": 0.5096902251243591,
      "learning_rate": 0.002828,
      "loss": 1.1669,
      "step": 459392
    },
    {
      "epoch": 34.48075046904315,
      "grad_norm": 0.6237781643867493,
      "learning_rate": 0.002828,
      "loss": 1.1659,
      "step": 459456
    },
    {
      "epoch": 34.48555347091933,
      "grad_norm": 0.66024249792099,
      "learning_rate": 0.002828,
      "loss": 1.1657,
      "step": 459520
    },
    {
      "epoch": 34.4903564727955,
      "grad_norm": 0.5944492816925049,
      "learning_rate": 0.002828,
      "loss": 1.1671,
      "step": 459584
    },
    {
      "epoch": 34.49515947467167,
      "grad_norm": 0.5822668075561523,
      "learning_rate": 0.002828,
      "loss": 1.1607,
      "step": 459648
    },
    {
      "epoch": 34.49996247654784,
      "grad_norm": 0.6067987680435181,
      "learning_rate": 0.002828,
      "loss": 1.1577,
      "step": 459712
    },
    {
      "epoch": 34.50476547842401,
      "grad_norm": 0.5925935506820679,
      "learning_rate": 0.002828,
      "loss": 1.1623,
      "step": 459776
    },
    {
      "epoch": 34.50956848030019,
      "grad_norm": 0.6033923029899597,
      "learning_rate": 0.002828,
      "loss": 1.1598,
      "step": 459840
    },
    {
      "epoch": 34.51437148217636,
      "grad_norm": 0.5351348519325256,
      "learning_rate": 0.002828,
      "loss": 1.16,
      "step": 459904
    },
    {
      "epoch": 34.519174484052535,
      "grad_norm": 0.5987980365753174,
      "learning_rate": 0.002828,
      "loss": 1.156,
      "step": 459968
    },
    {
      "epoch": 34.523977485928704,
      "grad_norm": 0.5701244473457336,
      "learning_rate": 0.002828,
      "loss": 1.1553,
      "step": 460032
    },
    {
      "epoch": 34.52878048780488,
      "grad_norm": 0.5841987729072571,
      "learning_rate": 0.002828,
      "loss": 1.1597,
      "step": 460096
    },
    {
      "epoch": 34.53358348968105,
      "grad_norm": 0.5855494737625122,
      "learning_rate": 0.002828,
      "loss": 1.1575,
      "step": 460160
    },
    {
      "epoch": 34.538386491557226,
      "grad_norm": 0.6749581694602966,
      "learning_rate": 0.002828,
      "loss": 1.1601,
      "step": 460224
    },
    {
      "epoch": 34.543189493433395,
      "grad_norm": 0.6047836542129517,
      "learning_rate": 0.002828,
      "loss": 1.1576,
      "step": 460288
    },
    {
      "epoch": 34.54799249530957,
      "grad_norm": 0.6513249278068542,
      "learning_rate": 0.002828,
      "loss": 1.1546,
      "step": 460352
    },
    {
      "epoch": 34.55279549718574,
      "grad_norm": 0.7231667637825012,
      "learning_rate": 0.002828,
      "loss": 1.1619,
      "step": 460416
    },
    {
      "epoch": 34.55759849906191,
      "grad_norm": 0.7058590054512024,
      "learning_rate": 0.002828,
      "loss": 1.1547,
      "step": 460480
    },
    {
      "epoch": 34.56240150093809,
      "grad_norm": 0.6364291310310364,
      "learning_rate": 0.002828,
      "loss": 1.1552,
      "step": 460544
    },
    {
      "epoch": 34.567204502814256,
      "grad_norm": 0.6043879389762878,
      "learning_rate": 0.002828,
      "loss": 1.159,
      "step": 460608
    },
    {
      "epoch": 34.57200750469043,
      "grad_norm": 0.5241459012031555,
      "learning_rate": 0.002828,
      "loss": 1.1589,
      "step": 460672
    },
    {
      "epoch": 34.5768105065666,
      "grad_norm": 0.5184655785560608,
      "learning_rate": 0.002828,
      "loss": 1.1575,
      "step": 460736
    },
    {
      "epoch": 34.58161350844278,
      "grad_norm": 0.6597776412963867,
      "learning_rate": 0.002828,
      "loss": 1.154,
      "step": 460800
    },
    {
      "epoch": 34.58641651031895,
      "grad_norm": 0.5717831254005432,
      "learning_rate": 0.002828,
      "loss": 1.1548,
      "step": 460864
    },
    {
      "epoch": 34.591219512195124,
      "grad_norm": 0.7088619470596313,
      "learning_rate": 0.002828,
      "loss": 1.1565,
      "step": 460928
    },
    {
      "epoch": 34.596022514071294,
      "grad_norm": 0.6347498893737793,
      "learning_rate": 0.002828,
      "loss": 1.1573,
      "step": 460992
    },
    {
      "epoch": 34.60082551594747,
      "grad_norm": 0.5776677131652832,
      "learning_rate": 0.002828,
      "loss": 1.1565,
      "step": 461056
    },
    {
      "epoch": 34.60562851782364,
      "grad_norm": 0.5810860395431519,
      "learning_rate": 0.002828,
      "loss": 1.1603,
      "step": 461120
    },
    {
      "epoch": 34.610431519699816,
      "grad_norm": 0.6792919039726257,
      "learning_rate": 0.002828,
      "loss": 1.1483,
      "step": 461184
    },
    {
      "epoch": 34.615234521575985,
      "grad_norm": 0.6665756702423096,
      "learning_rate": 0.002828,
      "loss": 1.1556,
      "step": 461248
    },
    {
      "epoch": 34.620037523452154,
      "grad_norm": 0.5993114709854126,
      "learning_rate": 0.002828,
      "loss": 1.1533,
      "step": 461312
    },
    {
      "epoch": 34.62484052532833,
      "grad_norm": 0.7352228164672852,
      "learning_rate": 0.002828,
      "loss": 1.1542,
      "step": 461376
    },
    {
      "epoch": 34.6296435272045,
      "grad_norm": 0.5764060020446777,
      "learning_rate": 0.002828,
      "loss": 1.1563,
      "step": 461440
    },
    {
      "epoch": 34.63444652908068,
      "grad_norm": 0.5434015989303589,
      "learning_rate": 0.002828,
      "loss": 1.1488,
      "step": 461504
    },
    {
      "epoch": 34.639249530956846,
      "grad_norm": 0.5354040265083313,
      "learning_rate": 0.002828,
      "loss": 1.1533,
      "step": 461568
    },
    {
      "epoch": 34.64405253283302,
      "grad_norm": 0.618698000907898,
      "learning_rate": 0.002828,
      "loss": 1.1528,
      "step": 461632
    },
    {
      "epoch": 34.64885553470919,
      "grad_norm": 0.6807321906089783,
      "learning_rate": 0.002828,
      "loss": 1.1542,
      "step": 461696
    },
    {
      "epoch": 34.65365853658537,
      "grad_norm": 0.5973524451255798,
      "learning_rate": 0.002828,
      "loss": 1.1537,
      "step": 461760
    },
    {
      "epoch": 34.65846153846154,
      "grad_norm": 0.6482577919960022,
      "learning_rate": 0.002828,
      "loss": 1.1475,
      "step": 461824
    },
    {
      "epoch": 34.663264540337714,
      "grad_norm": 0.694492518901825,
      "learning_rate": 0.002828,
      "loss": 1.1541,
      "step": 461888
    },
    {
      "epoch": 34.66806754221388,
      "grad_norm": 0.5939033031463623,
      "learning_rate": 0.002828,
      "loss": 1.1472,
      "step": 461952
    },
    {
      "epoch": 34.67287054409006,
      "grad_norm": 0.8614955544471741,
      "learning_rate": 0.002828,
      "loss": 1.1524,
      "step": 462016
    },
    {
      "epoch": 34.67767354596623,
      "grad_norm": 0.6040512919425964,
      "learning_rate": 0.002828,
      "loss": 1.152,
      "step": 462080
    },
    {
      "epoch": 34.6824765478424,
      "grad_norm": 0.5883114337921143,
      "learning_rate": 0.002828,
      "loss": 1.1536,
      "step": 462144
    },
    {
      "epoch": 34.687279549718575,
      "grad_norm": 0.5902653336524963,
      "learning_rate": 0.002828,
      "loss": 1.1511,
      "step": 462208
    },
    {
      "epoch": 34.692082551594744,
      "grad_norm": 0.6126294136047363,
      "learning_rate": 0.002828,
      "loss": 1.1528,
      "step": 462272
    },
    {
      "epoch": 34.69688555347092,
      "grad_norm": 0.5560796856880188,
      "learning_rate": 0.002828,
      "loss": 1.1508,
      "step": 462336
    },
    {
      "epoch": 34.70168855534709,
      "grad_norm": 0.6218116879463196,
      "learning_rate": 0.002828,
      "loss": 1.1464,
      "step": 462400
    },
    {
      "epoch": 34.706491557223266,
      "grad_norm": 0.5556285381317139,
      "learning_rate": 0.002828,
      "loss": 1.1526,
      "step": 462464
    },
    {
      "epoch": 34.711294559099436,
      "grad_norm": 0.7265997529029846,
      "learning_rate": 0.002828,
      "loss": 1.1567,
      "step": 462528
    },
    {
      "epoch": 34.71609756097561,
      "grad_norm": 0.5715520977973938,
      "learning_rate": 0.002828,
      "loss": 1.1498,
      "step": 462592
    },
    {
      "epoch": 34.72090056285178,
      "grad_norm": 0.5536825656890869,
      "learning_rate": 0.002828,
      "loss": 1.1602,
      "step": 462656
    },
    {
      "epoch": 34.72570356472796,
      "grad_norm": 0.552040696144104,
      "learning_rate": 0.002828,
      "loss": 1.1536,
      "step": 462720
    },
    {
      "epoch": 34.73050656660413,
      "grad_norm": 0.6238213181495667,
      "learning_rate": 0.002828,
      "loss": 1.153,
      "step": 462784
    },
    {
      "epoch": 34.735309568480304,
      "grad_norm": 0.6593725085258484,
      "learning_rate": 0.002828,
      "loss": 1.1496,
      "step": 462848
    },
    {
      "epoch": 34.74011257035647,
      "grad_norm": 0.7645888924598694,
      "learning_rate": 0.002828,
      "loss": 1.156,
      "step": 462912
    },
    {
      "epoch": 34.74491557223264,
      "grad_norm": 0.5748801827430725,
      "learning_rate": 0.002828,
      "loss": 1.1562,
      "step": 462976
    },
    {
      "epoch": 34.74971857410882,
      "grad_norm": 0.7016138434410095,
      "learning_rate": 0.002828,
      "loss": 1.1582,
      "step": 463040
    },
    {
      "epoch": 34.75452157598499,
      "grad_norm": 0.6343497037887573,
      "learning_rate": 0.002828,
      "loss": 1.1568,
      "step": 463104
    },
    {
      "epoch": 34.759324577861165,
      "grad_norm": 0.5537910461425781,
      "learning_rate": 0.002828,
      "loss": 1.1516,
      "step": 463168
    },
    {
      "epoch": 34.764127579737334,
      "grad_norm": 0.6629467606544495,
      "learning_rate": 0.002828,
      "loss": 1.1533,
      "step": 463232
    },
    {
      "epoch": 34.76893058161351,
      "grad_norm": 0.6417297124862671,
      "learning_rate": 0.002828,
      "loss": 1.1521,
      "step": 463296
    },
    {
      "epoch": 34.77373358348968,
      "grad_norm": 0.5607069134712219,
      "learning_rate": 0.002828,
      "loss": 1.1536,
      "step": 463360
    },
    {
      "epoch": 34.778536585365856,
      "grad_norm": 0.5720988512039185,
      "learning_rate": 0.002828,
      "loss": 1.151,
      "step": 463424
    },
    {
      "epoch": 34.783339587242025,
      "grad_norm": 0.6062473058700562,
      "learning_rate": 0.002828,
      "loss": 1.151,
      "step": 463488
    },
    {
      "epoch": 34.7881425891182,
      "grad_norm": 0.6105201244354248,
      "learning_rate": 0.002828,
      "loss": 1.151,
      "step": 463552
    },
    {
      "epoch": 34.79294559099437,
      "grad_norm": 0.6517950892448425,
      "learning_rate": 0.002828,
      "loss": 1.1578,
      "step": 463616
    },
    {
      "epoch": 34.79774859287055,
      "grad_norm": 0.5077842473983765,
      "learning_rate": 0.002828,
      "loss": 1.1524,
      "step": 463680
    },
    {
      "epoch": 34.80255159474672,
      "grad_norm": 0.4883643686771393,
      "learning_rate": 0.002828,
      "loss": 1.15,
      "step": 463744
    },
    {
      "epoch": 34.807354596622886,
      "grad_norm": 0.5117125511169434,
      "learning_rate": 0.002828,
      "loss": 1.1525,
      "step": 463808
    },
    {
      "epoch": 34.81215759849906,
      "grad_norm": 0.510983943939209,
      "learning_rate": 0.002828,
      "loss": 1.1483,
      "step": 463872
    },
    {
      "epoch": 34.81696060037523,
      "grad_norm": 0.5725462436676025,
      "learning_rate": 0.002828,
      "loss": 1.1537,
      "step": 463936
    },
    {
      "epoch": 34.82176360225141,
      "grad_norm": 0.7000669836997986,
      "learning_rate": 0.002828,
      "loss": 1.1543,
      "step": 464000
    },
    {
      "epoch": 34.82656660412758,
      "grad_norm": 0.597078800201416,
      "learning_rate": 0.002828,
      "loss": 1.153,
      "step": 464064
    },
    {
      "epoch": 34.831369606003754,
      "grad_norm": 0.5912140607833862,
      "learning_rate": 0.002828,
      "loss": 1.1547,
      "step": 464128
    },
    {
      "epoch": 34.836172607879924,
      "grad_norm": 0.6521306037902832,
      "learning_rate": 0.002828,
      "loss": 1.1561,
      "step": 464192
    },
    {
      "epoch": 34.8409756097561,
      "grad_norm": 0.5406016707420349,
      "learning_rate": 0.002828,
      "loss": 1.1598,
      "step": 464256
    },
    {
      "epoch": 34.84577861163227,
      "grad_norm": 0.6859225034713745,
      "learning_rate": 0.002828,
      "loss": 1.1541,
      "step": 464320
    },
    {
      "epoch": 34.850581613508446,
      "grad_norm": 0.5705737471580505,
      "learning_rate": 0.002828,
      "loss": 1.1581,
      "step": 464384
    },
    {
      "epoch": 34.855384615384615,
      "grad_norm": 0.6631976366043091,
      "learning_rate": 0.002828,
      "loss": 1.1534,
      "step": 464448
    },
    {
      "epoch": 34.860187617260785,
      "grad_norm": 0.6074179410934448,
      "learning_rate": 0.002828,
      "loss": 1.1521,
      "step": 464512
    },
    {
      "epoch": 34.86499061913696,
      "grad_norm": 0.5356570482254028,
      "learning_rate": 0.002828,
      "loss": 1.1497,
      "step": 464576
    },
    {
      "epoch": 34.86979362101313,
      "grad_norm": 0.760819137096405,
      "learning_rate": 0.002828,
      "loss": 1.1546,
      "step": 464640
    },
    {
      "epoch": 34.87459662288931,
      "grad_norm": 0.6953458189964294,
      "learning_rate": 0.002828,
      "loss": 1.146,
      "step": 464704
    },
    {
      "epoch": 34.879399624765476,
      "grad_norm": 0.5345589518547058,
      "learning_rate": 0.002828,
      "loss": 1.1503,
      "step": 464768
    },
    {
      "epoch": 34.88420262664165,
      "grad_norm": 0.6950819492340088,
      "learning_rate": 0.002828,
      "loss": 1.1518,
      "step": 464832
    },
    {
      "epoch": 34.88900562851782,
      "grad_norm": 0.6367828249931335,
      "learning_rate": 0.002828,
      "loss": 1.1519,
      "step": 464896
    },
    {
      "epoch": 34.893808630394,
      "grad_norm": 0.5749104619026184,
      "learning_rate": 0.002828,
      "loss": 1.1461,
      "step": 464960
    },
    {
      "epoch": 34.89861163227017,
      "grad_norm": 0.5483949184417725,
      "learning_rate": 0.002828,
      "loss": 1.1538,
      "step": 465024
    },
    {
      "epoch": 34.903414634146344,
      "grad_norm": 0.6119317412376404,
      "learning_rate": 0.002828,
      "loss": 1.1482,
      "step": 465088
    },
    {
      "epoch": 34.90821763602251,
      "grad_norm": 0.6510235071182251,
      "learning_rate": 0.002828,
      "loss": 1.1559,
      "step": 465152
    },
    {
      "epoch": 34.91302063789869,
      "grad_norm": 0.5535789728164673,
      "learning_rate": 0.002828,
      "loss": 1.1517,
      "step": 465216
    },
    {
      "epoch": 34.91782363977486,
      "grad_norm": 0.5926808714866638,
      "learning_rate": 0.002828,
      "loss": 1.1458,
      "step": 465280
    },
    {
      "epoch": 34.92262664165103,
      "grad_norm": 0.6453589200973511,
      "learning_rate": 0.002828,
      "loss": 1.1538,
      "step": 465344
    },
    {
      "epoch": 34.927429643527205,
      "grad_norm": 0.5200554132461548,
      "learning_rate": 0.002828,
      "loss": 1.1465,
      "step": 465408
    },
    {
      "epoch": 34.932232645403374,
      "grad_norm": 0.572009801864624,
      "learning_rate": 0.002828,
      "loss": 1.1535,
      "step": 465472
    },
    {
      "epoch": 34.93703564727955,
      "grad_norm": 0.5859612822532654,
      "learning_rate": 0.002828,
      "loss": 1.1583,
      "step": 465536
    },
    {
      "epoch": 34.94183864915572,
      "grad_norm": 0.8858081698417664,
      "learning_rate": 0.002828,
      "loss": 1.1491,
      "step": 465600
    },
    {
      "epoch": 34.9466416510319,
      "grad_norm": 0.5900371074676514,
      "learning_rate": 0.002828,
      "loss": 1.152,
      "step": 465664
    },
    {
      "epoch": 34.951444652908066,
      "grad_norm": 0.6436994075775146,
      "learning_rate": 0.002828,
      "loss": 1.1503,
      "step": 465728
    },
    {
      "epoch": 34.95624765478424,
      "grad_norm": 0.5482416152954102,
      "learning_rate": 0.002828,
      "loss": 1.1544,
      "step": 465792
    },
    {
      "epoch": 34.96105065666041,
      "grad_norm": 0.6615215539932251,
      "learning_rate": 0.002828,
      "loss": 1.1507,
      "step": 465856
    },
    {
      "epoch": 34.96585365853659,
      "grad_norm": 0.5764489769935608,
      "learning_rate": 0.002828,
      "loss": 1.1527,
      "step": 465920
    },
    {
      "epoch": 34.97065666041276,
      "grad_norm": 0.7269230484962463,
      "learning_rate": 0.002828,
      "loss": 1.1546,
      "step": 465984
    },
    {
      "epoch": 34.975459662288934,
      "grad_norm": 0.5440137982368469,
      "learning_rate": 0.002828,
      "loss": 1.1511,
      "step": 466048
    },
    {
      "epoch": 34.9802626641651,
      "grad_norm": 0.5844058990478516,
      "learning_rate": 0.002828,
      "loss": 1.1552,
      "step": 466112
    },
    {
      "epoch": 34.98506566604127,
      "grad_norm": 0.5198141932487488,
      "learning_rate": 0.002828,
      "loss": 1.1506,
      "step": 466176
    },
    {
      "epoch": 34.98986866791745,
      "grad_norm": 0.620516300201416,
      "learning_rate": 0.002828,
      "loss": 1.1557,
      "step": 466240
    },
    {
      "epoch": 34.99467166979362,
      "grad_norm": 0.6901420950889587,
      "learning_rate": 0.002828,
      "loss": 1.1532,
      "step": 466304
    },
    {
      "epoch": 34.999474671669795,
      "grad_norm": 0.6293230056762695,
      "learning_rate": 0.002828,
      "loss": 1.1526,
      "step": 466368
    },
    {
      "epoch": 35.004277673545964,
      "grad_norm": 0.6638535261154175,
      "learning_rate": 0.002828,
      "loss": 1.1266,
      "step": 466432
    },
    {
      "epoch": 35.00908067542214,
      "grad_norm": 0.5445776581764221,
      "learning_rate": 0.002828,
      "loss": 1.123,
      "step": 466496
    },
    {
      "epoch": 35.01388367729831,
      "grad_norm": 0.808910071849823,
      "learning_rate": 0.002828,
      "loss": 1.1231,
      "step": 466560
    },
    {
      "epoch": 35.018686679174486,
      "grad_norm": 0.6956142783164978,
      "learning_rate": 0.002828,
      "loss": 1.1235,
      "step": 466624
    },
    {
      "epoch": 35.023489681050656,
      "grad_norm": 0.6124756932258606,
      "learning_rate": 0.002828,
      "loss": 1.1242,
      "step": 466688
    },
    {
      "epoch": 35.02829268292683,
      "grad_norm": 0.5752196311950684,
      "learning_rate": 0.002828,
      "loss": 1.1175,
      "step": 466752
    },
    {
      "epoch": 35.033095684803,
      "grad_norm": 0.5530253052711487,
      "learning_rate": 0.002828,
      "loss": 1.1306,
      "step": 466816
    },
    {
      "epoch": 35.03789868667918,
      "grad_norm": 0.6303296685218811,
      "learning_rate": 0.002828,
      "loss": 1.1255,
      "step": 466880
    },
    {
      "epoch": 35.04270168855535,
      "grad_norm": 0.6001397967338562,
      "learning_rate": 0.002828,
      "loss": 1.117,
      "step": 466944
    },
    {
      "epoch": 35.047504690431516,
      "grad_norm": 0.5856416821479797,
      "learning_rate": 0.002828,
      "loss": 1.122,
      "step": 467008
    },
    {
      "epoch": 35.05230769230769,
      "grad_norm": 0.6043058633804321,
      "learning_rate": 0.002828,
      "loss": 1.1188,
      "step": 467072
    },
    {
      "epoch": 35.05711069418386,
      "grad_norm": 0.5569366812705994,
      "learning_rate": 0.002828,
      "loss": 1.1222,
      "step": 467136
    },
    {
      "epoch": 35.06191369606004,
      "grad_norm": 0.5817255973815918,
      "learning_rate": 0.002828,
      "loss": 1.1191,
      "step": 467200
    },
    {
      "epoch": 35.06671669793621,
      "grad_norm": 0.5985368490219116,
      "learning_rate": 0.002828,
      "loss": 1.1223,
      "step": 467264
    },
    {
      "epoch": 35.071519699812384,
      "grad_norm": 0.5214247107505798,
      "learning_rate": 0.002828,
      "loss": 1.1193,
      "step": 467328
    },
    {
      "epoch": 35.076322701688554,
      "grad_norm": 0.5910423994064331,
      "learning_rate": 0.002828,
      "loss": 1.1285,
      "step": 467392
    },
    {
      "epoch": 35.08112570356473,
      "grad_norm": 0.5970675945281982,
      "learning_rate": 0.002828,
      "loss": 1.1254,
      "step": 467456
    },
    {
      "epoch": 35.0859287054409,
      "grad_norm": 0.6935176849365234,
      "learning_rate": 0.002828,
      "loss": 1.1224,
      "step": 467520
    },
    {
      "epoch": 35.090731707317076,
      "grad_norm": 0.5811248421669006,
      "learning_rate": 0.002828,
      "loss": 1.1273,
      "step": 467584
    },
    {
      "epoch": 35.095534709193245,
      "grad_norm": 0.606397807598114,
      "learning_rate": 0.002828,
      "loss": 1.13,
      "step": 467648
    },
    {
      "epoch": 35.10033771106942,
      "grad_norm": 0.6233459711074829,
      "learning_rate": 0.002828,
      "loss": 1.1203,
      "step": 467712
    },
    {
      "epoch": 35.10514071294559,
      "grad_norm": 0.4995225965976715,
      "learning_rate": 0.002828,
      "loss": 1.1293,
      "step": 467776
    },
    {
      "epoch": 35.10994371482176,
      "grad_norm": 0.7771652340888977,
      "learning_rate": 0.002828,
      "loss": 1.1224,
      "step": 467840
    },
    {
      "epoch": 35.11474671669794,
      "grad_norm": 0.5408387780189514,
      "learning_rate": 0.002828,
      "loss": 1.1228,
      "step": 467904
    },
    {
      "epoch": 35.119549718574106,
      "grad_norm": 0.6495393514633179,
      "learning_rate": 0.002828,
      "loss": 1.1244,
      "step": 467968
    },
    {
      "epoch": 35.12435272045028,
      "grad_norm": 0.5671504139900208,
      "learning_rate": 0.002828,
      "loss": 1.1217,
      "step": 468032
    },
    {
      "epoch": 35.12915572232645,
      "grad_norm": 0.6291579008102417,
      "learning_rate": 0.002828,
      "loss": 1.123,
      "step": 468096
    },
    {
      "epoch": 35.13395872420263,
      "grad_norm": 0.5787367820739746,
      "learning_rate": 0.002828,
      "loss": 1.1198,
      "step": 468160
    },
    {
      "epoch": 35.1387617260788,
      "grad_norm": 0.6570515036582947,
      "learning_rate": 0.002828,
      "loss": 1.1217,
      "step": 468224
    },
    {
      "epoch": 35.143564727954974,
      "grad_norm": 0.6233621835708618,
      "learning_rate": 0.002828,
      "loss": 1.1312,
      "step": 468288
    },
    {
      "epoch": 35.14836772983114,
      "grad_norm": 0.6841063499450684,
      "learning_rate": 0.002828,
      "loss": 1.1202,
      "step": 468352
    },
    {
      "epoch": 35.15317073170732,
      "grad_norm": 0.5697184801101685,
      "learning_rate": 0.002828,
      "loss": 1.1258,
      "step": 468416
    },
    {
      "epoch": 35.15797373358349,
      "grad_norm": 0.6447716355323792,
      "learning_rate": 0.002828,
      "loss": 1.124,
      "step": 468480
    },
    {
      "epoch": 35.162776735459666,
      "grad_norm": 0.530463457107544,
      "learning_rate": 0.002828,
      "loss": 1.1245,
      "step": 468544
    },
    {
      "epoch": 35.167579737335835,
      "grad_norm": 0.6837313175201416,
      "learning_rate": 0.002828,
      "loss": 1.1258,
      "step": 468608
    },
    {
      "epoch": 35.172382739212004,
      "grad_norm": 0.7450433373451233,
      "learning_rate": 0.002828,
      "loss": 1.1297,
      "step": 468672
    },
    {
      "epoch": 35.17718574108818,
      "grad_norm": 0.6392331719398499,
      "learning_rate": 0.002828,
      "loss": 1.1263,
      "step": 468736
    },
    {
      "epoch": 35.18198874296435,
      "grad_norm": 0.5583149194717407,
      "learning_rate": 0.002828,
      "loss": 1.1284,
      "step": 468800
    },
    {
      "epoch": 35.18679174484053,
      "grad_norm": 0.43744319677352905,
      "learning_rate": 0.002828,
      "loss": 1.1268,
      "step": 468864
    },
    {
      "epoch": 35.191594746716696,
      "grad_norm": 0.5928747057914734,
      "learning_rate": 0.002828,
      "loss": 1.1272,
      "step": 468928
    },
    {
      "epoch": 35.19639774859287,
      "grad_norm": 0.5616432428359985,
      "learning_rate": 0.002828,
      "loss": 1.1273,
      "step": 468992
    },
    {
      "epoch": 35.20120075046904,
      "grad_norm": 0.5244531631469727,
      "learning_rate": 0.002828,
      "loss": 1.1274,
      "step": 469056
    },
    {
      "epoch": 35.20600375234522,
      "grad_norm": 0.5104504823684692,
      "learning_rate": 0.002828,
      "loss": 1.1269,
      "step": 469120
    },
    {
      "epoch": 35.21080675422139,
      "grad_norm": 0.5875263214111328,
      "learning_rate": 0.002828,
      "loss": 1.1352,
      "step": 469184
    },
    {
      "epoch": 35.215609756097564,
      "grad_norm": 0.6020212769508362,
      "learning_rate": 0.002828,
      "loss": 1.1231,
      "step": 469248
    },
    {
      "epoch": 35.22041275797373,
      "grad_norm": 0.7060278058052063,
      "learning_rate": 0.002828,
      "loss": 1.1249,
      "step": 469312
    },
    {
      "epoch": 35.22521575984991,
      "grad_norm": 0.5829773545265198,
      "learning_rate": 0.002828,
      "loss": 1.1286,
      "step": 469376
    },
    {
      "epoch": 35.23001876172608,
      "grad_norm": 0.6745613813400269,
      "learning_rate": 0.002828,
      "loss": 1.1309,
      "step": 469440
    },
    {
      "epoch": 35.23482176360225,
      "grad_norm": 0.6295201778411865,
      "learning_rate": 0.002828,
      "loss": 1.1232,
      "step": 469504
    },
    {
      "epoch": 35.239624765478425,
      "grad_norm": 0.6248756647109985,
      "learning_rate": 0.002828,
      "loss": 1.126,
      "step": 469568
    },
    {
      "epoch": 35.244427767354594,
      "grad_norm": 0.5959972739219666,
      "learning_rate": 0.002828,
      "loss": 1.1299,
      "step": 469632
    },
    {
      "epoch": 35.24923076923077,
      "grad_norm": 0.6190743446350098,
      "learning_rate": 0.002828,
      "loss": 1.1289,
      "step": 469696
    },
    {
      "epoch": 35.25403377110694,
      "grad_norm": 0.5316118001937866,
      "learning_rate": 0.002828,
      "loss": 1.1294,
      "step": 469760
    },
    {
      "epoch": 35.258836772983116,
      "grad_norm": 0.582175076007843,
      "learning_rate": 0.002828,
      "loss": 1.1314,
      "step": 469824
    },
    {
      "epoch": 35.263639774859286,
      "grad_norm": 0.6277863383293152,
      "learning_rate": 0.002828,
      "loss": 1.1315,
      "step": 469888
    },
    {
      "epoch": 35.26844277673546,
      "grad_norm": 0.5997483730316162,
      "learning_rate": 0.002828,
      "loss": 1.1332,
      "step": 469952
    },
    {
      "epoch": 35.27324577861163,
      "grad_norm": 0.6677803993225098,
      "learning_rate": 0.002828,
      "loss": 1.1248,
      "step": 470016
    },
    {
      "epoch": 35.27804878048781,
      "grad_norm": 0.5580044984817505,
      "learning_rate": 0.002828,
      "loss": 1.1291,
      "step": 470080
    },
    {
      "epoch": 35.28285178236398,
      "grad_norm": 0.65473872423172,
      "learning_rate": 0.002828,
      "loss": 1.1255,
      "step": 470144
    },
    {
      "epoch": 35.287654784240154,
      "grad_norm": 0.5278262495994568,
      "learning_rate": 0.002828,
      "loss": 1.1294,
      "step": 470208
    },
    {
      "epoch": 35.29245778611632,
      "grad_norm": 0.6099348664283752,
      "learning_rate": 0.002828,
      "loss": 1.133,
      "step": 470272
    },
    {
      "epoch": 35.29726078799249,
      "grad_norm": 0.5697486400604248,
      "learning_rate": 0.002828,
      "loss": 1.1358,
      "step": 470336
    },
    {
      "epoch": 35.30206378986867,
      "grad_norm": 0.6079903841018677,
      "learning_rate": 0.002828,
      "loss": 1.1344,
      "step": 470400
    },
    {
      "epoch": 35.30686679174484,
      "grad_norm": 0.6247881650924683,
      "learning_rate": 0.002828,
      "loss": 1.1288,
      "step": 470464
    },
    {
      "epoch": 35.311669793621014,
      "grad_norm": 0.5848674774169922,
      "learning_rate": 0.002828,
      "loss": 1.137,
      "step": 470528
    },
    {
      "epoch": 35.316472795497184,
      "grad_norm": 0.6638665199279785,
      "learning_rate": 0.002828,
      "loss": 1.1321,
      "step": 470592
    },
    {
      "epoch": 35.32127579737336,
      "grad_norm": 0.5754868388175964,
      "learning_rate": 0.002828,
      "loss": 1.1298,
      "step": 470656
    },
    {
      "epoch": 35.32607879924953,
      "grad_norm": 0.646364688873291,
      "learning_rate": 0.002828,
      "loss": 1.1319,
      "step": 470720
    },
    {
      "epoch": 35.330881801125706,
      "grad_norm": 0.6823537349700928,
      "learning_rate": 0.002828,
      "loss": 1.1314,
      "step": 470784
    },
    {
      "epoch": 35.335684803001875,
      "grad_norm": 0.5892389416694641,
      "learning_rate": 0.002828,
      "loss": 1.1346,
      "step": 470848
    },
    {
      "epoch": 35.34048780487805,
      "grad_norm": 0.5694969892501831,
      "learning_rate": 0.002828,
      "loss": 1.138,
      "step": 470912
    },
    {
      "epoch": 35.34529080675422,
      "grad_norm": 0.552528977394104,
      "learning_rate": 0.002828,
      "loss": 1.1331,
      "step": 470976
    },
    {
      "epoch": 35.3500938086304,
      "grad_norm": 0.6115056872367859,
      "learning_rate": 0.002828,
      "loss": 1.1331,
      "step": 471040
    },
    {
      "epoch": 35.35489681050657,
      "grad_norm": 0.5536540150642395,
      "learning_rate": 0.002828,
      "loss": 1.1308,
      "step": 471104
    },
    {
      "epoch": 35.359699812382736,
      "grad_norm": 0.570509135723114,
      "learning_rate": 0.002828,
      "loss": 1.1326,
      "step": 471168
    },
    {
      "epoch": 35.36450281425891,
      "grad_norm": 0.5459985733032227,
      "learning_rate": 0.002828,
      "loss": 1.1352,
      "step": 471232
    },
    {
      "epoch": 35.36930581613508,
      "grad_norm": 0.5294592380523682,
      "learning_rate": 0.002828,
      "loss": 1.1307,
      "step": 471296
    },
    {
      "epoch": 35.37410881801126,
      "grad_norm": 0.5800413489341736,
      "learning_rate": 0.002828,
      "loss": 1.1371,
      "step": 471360
    },
    {
      "epoch": 35.37891181988743,
      "grad_norm": 0.568485677242279,
      "learning_rate": 0.002828,
      "loss": 1.1393,
      "step": 471424
    },
    {
      "epoch": 35.383714821763604,
      "grad_norm": 0.7035385370254517,
      "learning_rate": 0.002828,
      "loss": 1.1338,
      "step": 471488
    },
    {
      "epoch": 35.38851782363977,
      "grad_norm": 0.738166868686676,
      "learning_rate": 0.002828,
      "loss": 1.1348,
      "step": 471552
    },
    {
      "epoch": 35.39332082551595,
      "grad_norm": 0.6015270352363586,
      "learning_rate": 0.002828,
      "loss": 1.1301,
      "step": 471616
    },
    {
      "epoch": 35.39812382739212,
      "grad_norm": 0.6614118218421936,
      "learning_rate": 0.002828,
      "loss": 1.1314,
      "step": 471680
    },
    {
      "epoch": 35.402926829268296,
      "grad_norm": 0.5099043846130371,
      "learning_rate": 0.002828,
      "loss": 1.1349,
      "step": 471744
    },
    {
      "epoch": 35.407729831144465,
      "grad_norm": 0.5569406747817993,
      "learning_rate": 0.002828,
      "loss": 1.1332,
      "step": 471808
    },
    {
      "epoch": 35.412532833020634,
      "grad_norm": 0.6680443286895752,
      "learning_rate": 0.002828,
      "loss": 1.1324,
      "step": 471872
    },
    {
      "epoch": 35.41733583489681,
      "grad_norm": 0.5892511010169983,
      "learning_rate": 0.002828,
      "loss": 1.134,
      "step": 471936
    },
    {
      "epoch": 35.42213883677298,
      "grad_norm": 0.6710988879203796,
      "learning_rate": 0.002828,
      "loss": 1.131,
      "step": 472000
    },
    {
      "epoch": 35.42694183864916,
      "grad_norm": 0.5715301036834717,
      "learning_rate": 0.002828,
      "loss": 1.1377,
      "step": 472064
    },
    {
      "epoch": 35.431744840525326,
      "grad_norm": 0.5255868434906006,
      "learning_rate": 0.002828,
      "loss": 1.1351,
      "step": 472128
    },
    {
      "epoch": 35.4365478424015,
      "grad_norm": 0.6189907789230347,
      "learning_rate": 0.002828,
      "loss": 1.1353,
      "step": 472192
    },
    {
      "epoch": 35.44135084427767,
      "grad_norm": 0.6305795311927795,
      "learning_rate": 0.002828,
      "loss": 1.1371,
      "step": 472256
    },
    {
      "epoch": 35.44615384615385,
      "grad_norm": 0.576610267162323,
      "learning_rate": 0.002828,
      "loss": 1.129,
      "step": 472320
    },
    {
      "epoch": 35.45095684803002,
      "grad_norm": 0.5561624765396118,
      "learning_rate": 0.002828,
      "loss": 1.129,
      "step": 472384
    },
    {
      "epoch": 35.455759849906194,
      "grad_norm": 0.6723029613494873,
      "learning_rate": 0.002828,
      "loss": 1.14,
      "step": 472448
    },
    {
      "epoch": 35.46056285178236,
      "grad_norm": 0.6434258222579956,
      "learning_rate": 0.002828,
      "loss": 1.1314,
      "step": 472512
    },
    {
      "epoch": 35.46536585365854,
      "grad_norm": 0.5116338729858398,
      "learning_rate": 0.002828,
      "loss": 1.1372,
      "step": 472576
    },
    {
      "epoch": 35.47016885553471,
      "grad_norm": 0.700920820236206,
      "learning_rate": 0.002828,
      "loss": 1.1396,
      "step": 472640
    },
    {
      "epoch": 35.47497185741088,
      "grad_norm": 0.5516610741615295,
      "learning_rate": 0.002828,
      "loss": 1.1342,
      "step": 472704
    },
    {
      "epoch": 35.479774859287055,
      "grad_norm": 0.6950984597206116,
      "learning_rate": 0.002828,
      "loss": 1.1338,
      "step": 472768
    },
    {
      "epoch": 35.484577861163224,
      "grad_norm": 0.6530833840370178,
      "learning_rate": 0.002828,
      "loss": 1.1363,
      "step": 472832
    },
    {
      "epoch": 35.4893808630394,
      "grad_norm": 0.6566064357757568,
      "learning_rate": 0.002828,
      "loss": 1.1324,
      "step": 472896
    },
    {
      "epoch": 35.49418386491557,
      "grad_norm": 0.5705472230911255,
      "learning_rate": 0.002828,
      "loss": 1.1372,
      "step": 472960
    },
    {
      "epoch": 35.498986866791746,
      "grad_norm": 0.6241821646690369,
      "learning_rate": 0.002828,
      "loss": 1.1318,
      "step": 473024
    },
    {
      "epoch": 35.503789868667916,
      "grad_norm": 0.5050911903381348,
      "learning_rate": 0.002828,
      "loss": 1.1373,
      "step": 473088
    },
    {
      "epoch": 35.50859287054409,
      "grad_norm": 0.6013885140419006,
      "learning_rate": 0.002828,
      "loss": 1.1356,
      "step": 473152
    },
    {
      "epoch": 35.51339587242026,
      "grad_norm": 0.6184156537055969,
      "learning_rate": 0.002828,
      "loss": 1.1333,
      "step": 473216
    },
    {
      "epoch": 35.51819887429644,
      "grad_norm": 0.5862095355987549,
      "learning_rate": 0.002828,
      "loss": 1.1352,
      "step": 473280
    },
    {
      "epoch": 35.52300187617261,
      "grad_norm": 0.6102173328399658,
      "learning_rate": 0.002828,
      "loss": 1.1343,
      "step": 473344
    },
    {
      "epoch": 35.527804878048784,
      "grad_norm": 0.6712579131126404,
      "learning_rate": 0.002828,
      "loss": 1.1341,
      "step": 473408
    },
    {
      "epoch": 35.53260787992495,
      "grad_norm": 0.7177531123161316,
      "learning_rate": 0.002828,
      "loss": 1.1356,
      "step": 473472
    },
    {
      "epoch": 35.53741088180112,
      "grad_norm": 0.6322616934776306,
      "learning_rate": 0.002828,
      "loss": 1.137,
      "step": 473536
    },
    {
      "epoch": 35.5422138836773,
      "grad_norm": 0.6838127374649048,
      "learning_rate": 0.002828,
      "loss": 1.1364,
      "step": 473600
    },
    {
      "epoch": 35.54701688555347,
      "grad_norm": 0.637467086315155,
      "learning_rate": 0.002828,
      "loss": 1.133,
      "step": 473664
    },
    {
      "epoch": 35.551819887429644,
      "grad_norm": 0.7239699363708496,
      "learning_rate": 0.002828,
      "loss": 1.1338,
      "step": 473728
    },
    {
      "epoch": 35.556622889305814,
      "grad_norm": 0.7126774787902832,
      "learning_rate": 0.002828,
      "loss": 1.1416,
      "step": 473792
    },
    {
      "epoch": 35.56142589118199,
      "grad_norm": 0.5230181813240051,
      "learning_rate": 0.002828,
      "loss": 1.1319,
      "step": 473856
    },
    {
      "epoch": 35.56622889305816,
      "grad_norm": 0.5874216556549072,
      "learning_rate": 0.002828,
      "loss": 1.1354,
      "step": 473920
    },
    {
      "epoch": 35.571031894934336,
      "grad_norm": 0.5958366990089417,
      "learning_rate": 0.002828,
      "loss": 1.1398,
      "step": 473984
    },
    {
      "epoch": 35.575834896810505,
      "grad_norm": 0.5132750868797302,
      "learning_rate": 0.002828,
      "loss": 1.1373,
      "step": 474048
    },
    {
      "epoch": 35.58063789868668,
      "grad_norm": 0.6392728686332703,
      "learning_rate": 0.002828,
      "loss": 1.1335,
      "step": 474112
    },
    {
      "epoch": 35.58544090056285,
      "grad_norm": 0.5683247447013855,
      "learning_rate": 0.002828,
      "loss": 1.1366,
      "step": 474176
    },
    {
      "epoch": 35.59024390243903,
      "grad_norm": 0.6130415797233582,
      "learning_rate": 0.002828,
      "loss": 1.1404,
      "step": 474240
    },
    {
      "epoch": 35.5950469043152,
      "grad_norm": 0.668921947479248,
      "learning_rate": 0.002828,
      "loss": 1.1421,
      "step": 474304
    },
    {
      "epoch": 35.599849906191366,
      "grad_norm": 0.6637597680091858,
      "learning_rate": 0.002828,
      "loss": 1.1379,
      "step": 474368
    },
    {
      "epoch": 35.60465290806754,
      "grad_norm": 0.5993471145629883,
      "learning_rate": 0.002828,
      "loss": 1.1381,
      "step": 474432
    },
    {
      "epoch": 35.60945590994371,
      "grad_norm": 0.6369923949241638,
      "learning_rate": 0.002828,
      "loss": 1.1388,
      "step": 474496
    },
    {
      "epoch": 35.61425891181989,
      "grad_norm": 0.5497013330459595,
      "learning_rate": 0.002828,
      "loss": 1.1306,
      "step": 474560
    },
    {
      "epoch": 35.61906191369606,
      "grad_norm": 0.617692768573761,
      "learning_rate": 0.002828,
      "loss": 1.1387,
      "step": 474624
    },
    {
      "epoch": 35.623864915572234,
      "grad_norm": 0.6321608424186707,
      "learning_rate": 0.002828,
      "loss": 1.1352,
      "step": 474688
    },
    {
      "epoch": 35.628667917448404,
      "grad_norm": 0.6047719717025757,
      "learning_rate": 0.002828,
      "loss": 1.1346,
      "step": 474752
    },
    {
      "epoch": 35.63347091932458,
      "grad_norm": 0.5528061985969543,
      "learning_rate": 0.002828,
      "loss": 1.1356,
      "step": 474816
    },
    {
      "epoch": 35.63827392120075,
      "grad_norm": 0.5263925790786743,
      "learning_rate": 0.002828,
      "loss": 1.1321,
      "step": 474880
    },
    {
      "epoch": 35.643076923076926,
      "grad_norm": 0.639483630657196,
      "learning_rate": 0.002828,
      "loss": 1.1418,
      "step": 474944
    },
    {
      "epoch": 35.647879924953095,
      "grad_norm": 0.605707049369812,
      "learning_rate": 0.002828,
      "loss": 1.1416,
      "step": 475008
    },
    {
      "epoch": 35.65268292682927,
      "grad_norm": 0.6229245662689209,
      "learning_rate": 0.002828,
      "loss": 1.1386,
      "step": 475072
    },
    {
      "epoch": 35.65748592870544,
      "grad_norm": 0.552807092666626,
      "learning_rate": 0.002828,
      "loss": 1.1362,
      "step": 475136
    },
    {
      "epoch": 35.66228893058161,
      "grad_norm": 0.5757431983947754,
      "learning_rate": 0.002828,
      "loss": 1.1323,
      "step": 475200
    },
    {
      "epoch": 35.66709193245779,
      "grad_norm": 0.7068957686424255,
      "learning_rate": 0.002828,
      "loss": 1.133,
      "step": 475264
    },
    {
      "epoch": 35.671894934333956,
      "grad_norm": 0.7497088313102722,
      "learning_rate": 0.002828,
      "loss": 1.1407,
      "step": 475328
    },
    {
      "epoch": 35.67669793621013,
      "grad_norm": 0.6466912627220154,
      "learning_rate": 0.002828,
      "loss": 1.1357,
      "step": 475392
    },
    {
      "epoch": 35.6815009380863,
      "grad_norm": 0.5262551307678223,
      "learning_rate": 0.002828,
      "loss": 1.1395,
      "step": 475456
    },
    {
      "epoch": 35.68630393996248,
      "grad_norm": 0.6165720820426941,
      "learning_rate": 0.002828,
      "loss": 1.1347,
      "step": 475520
    },
    {
      "epoch": 35.69110694183865,
      "grad_norm": 0.703349232673645,
      "learning_rate": 0.002828,
      "loss": 1.1536,
      "step": 475584
    },
    {
      "epoch": 35.695909943714824,
      "grad_norm": 47066.40234375,
      "learning_rate": 0.002828,
      "loss": 3.8045,
      "step": 475648
    },
    {
      "epoch": 35.70071294559099,
      "grad_norm": 21974.30859375,
      "learning_rate": 0.002828,
      "loss": 4.3187,
      "step": 475712
    },
    {
      "epoch": 35.70551594746717,
      "grad_norm": 2190.40478515625,
      "learning_rate": 0.002828,
      "loss": 4.4774,
      "step": 475776
    },
    {
      "epoch": 35.71031894934334,
      "grad_norm": 4.141016483306885,
      "learning_rate": 0.002828,
      "loss": 3.0715,
      "step": 475840
    },
    {
      "epoch": 35.715121951219515,
      "grad_norm": 3.2081074714660645,
      "learning_rate": 0.002828,
      "loss": 2.208,
      "step": 475904
    },
    {
      "epoch": 35.719924953095685,
      "grad_norm": 918.6317138671875,
      "learning_rate": 0.002828,
      "loss": 1.388,
      "step": 475968
    },
    {
      "epoch": 35.724727954971854,
      "grad_norm": 1.1493455171585083,
      "learning_rate": 0.002828,
      "loss": 1.2846,
      "step": 476032
    },
    {
      "epoch": 35.72953095684803,
      "grad_norm": 0.8047353625297546,
      "learning_rate": 0.002828,
      "loss": 1.1635,
      "step": 476096
    },
    {
      "epoch": 35.7343339587242,
      "grad_norm": 0.9038922190666199,
      "learning_rate": 0.002828,
      "loss": 1.1629,
      "step": 476160
    },
    {
      "epoch": 35.739136960600376,
      "grad_norm": 0.8342848420143127,
      "learning_rate": 0.002828,
      "loss": 1.1686,
      "step": 476224
    },
    {
      "epoch": 35.743939962476546,
      "grad_norm": 1.1942050457000732,
      "learning_rate": 0.002828,
      "loss": 1.1753,
      "step": 476288
    },
    {
      "epoch": 35.74874296435272,
      "grad_norm": 0.6982195377349854,
      "learning_rate": 0.002828,
      "loss": 1.1598,
      "step": 476352
    },
    {
      "epoch": 35.75354596622889,
      "grad_norm": 0.6133076548576355,
      "learning_rate": 0.002828,
      "loss": 1.1525,
      "step": 476416
    },
    {
      "epoch": 35.75834896810507,
      "grad_norm": 0.6019357442855835,
      "learning_rate": 0.002828,
      "loss": 1.1508,
      "step": 476480
    },
    {
      "epoch": 35.76315196998124,
      "grad_norm": 0.6552913188934326,
      "learning_rate": 0.002828,
      "loss": 1.1447,
      "step": 476544
    },
    {
      "epoch": 35.767954971857414,
      "grad_norm": 0.6798120737075806,
      "learning_rate": 0.002828,
      "loss": 1.1493,
      "step": 476608
    },
    {
      "epoch": 35.77275797373358,
      "grad_norm": 0.6229259967803955,
      "learning_rate": 0.002828,
      "loss": 1.1458,
      "step": 476672
    },
    {
      "epoch": 35.77756097560976,
      "grad_norm": 0.5518473386764526,
      "learning_rate": 0.002828,
      "loss": 1.1417,
      "step": 476736
    },
    {
      "epoch": 35.78236397748593,
      "grad_norm": 0.581292986869812,
      "learning_rate": 0.002828,
      "loss": 1.1444,
      "step": 476800
    },
    {
      "epoch": 35.7871669793621,
      "grad_norm": 0.6113260388374329,
      "learning_rate": 0.002828,
      "loss": 1.141,
      "step": 476864
    },
    {
      "epoch": 35.791969981238275,
      "grad_norm": 0.7037771940231323,
      "learning_rate": 0.002828,
      "loss": 1.1447,
      "step": 476928
    },
    {
      "epoch": 35.796772983114444,
      "grad_norm": 0.6758065223693848,
      "learning_rate": 0.002828,
      "loss": 1.1448,
      "step": 476992
    },
    {
      "epoch": 35.80157598499062,
      "grad_norm": 0.5714001655578613,
      "learning_rate": 0.002828,
      "loss": 1.1448,
      "step": 477056
    },
    {
      "epoch": 35.80637898686679,
      "grad_norm": 0.5781056880950928,
      "learning_rate": 0.002828,
      "loss": 1.1459,
      "step": 477120
    },
    {
      "epoch": 35.811181988742966,
      "grad_norm": 0.6759575605392456,
      "learning_rate": 0.002828,
      "loss": 1.1475,
      "step": 477184
    },
    {
      "epoch": 35.815984990619135,
      "grad_norm": 0.5613266825675964,
      "learning_rate": 0.002828,
      "loss": 1.1441,
      "step": 477248
    },
    {
      "epoch": 35.82078799249531,
      "grad_norm": 0.7072948813438416,
      "learning_rate": 0.002828,
      "loss": 1.1454,
      "step": 477312
    },
    {
      "epoch": 35.82559099437148,
      "grad_norm": 0.6761579513549805,
      "learning_rate": 0.002828,
      "loss": 1.1443,
      "step": 477376
    },
    {
      "epoch": 35.83039399624766,
      "grad_norm": 0.6502017974853516,
      "learning_rate": 0.002828,
      "loss": 1.1464,
      "step": 477440
    },
    {
      "epoch": 35.83519699812383,
      "grad_norm": 0.6976175904273987,
      "learning_rate": 0.002828,
      "loss": 1.1447,
      "step": 477504
    },
    {
      "epoch": 35.84,
      "grad_norm": 0.6039485335350037,
      "learning_rate": 0.002828,
      "loss": 1.1422,
      "step": 477568
    },
    {
      "epoch": 35.84480300187617,
      "grad_norm": 0.670085608959198,
      "learning_rate": 0.002828,
      "loss": 1.1412,
      "step": 477632
    },
    {
      "epoch": 35.84960600375234,
      "grad_norm": 0.6866872906684875,
      "learning_rate": 0.002828,
      "loss": 1.1491,
      "step": 477696
    },
    {
      "epoch": 35.85440900562852,
      "grad_norm": 0.6738069653511047,
      "learning_rate": 0.002828,
      "loss": 1.1477,
      "step": 477760
    },
    {
      "epoch": 35.85921200750469,
      "grad_norm": 0.597515344619751,
      "learning_rate": 0.002828,
      "loss": 1.1473,
      "step": 477824
    },
    {
      "epoch": 35.864015009380864,
      "grad_norm": 0.6964823603630066,
      "learning_rate": 0.002828,
      "loss": 1.1397,
      "step": 477888
    },
    {
      "epoch": 35.868818011257034,
      "grad_norm": 0.5846383571624756,
      "learning_rate": 0.002828,
      "loss": 1.1452,
      "step": 477952
    },
    {
      "epoch": 35.87362101313321,
      "grad_norm": 0.7641522884368896,
      "learning_rate": 0.002828,
      "loss": 1.1456,
      "step": 478016
    },
    {
      "epoch": 35.87842401500938,
      "grad_norm": 0.6246045231819153,
      "learning_rate": 0.002828,
      "loss": 1.1456,
      "step": 478080
    },
    {
      "epoch": 35.883227016885556,
      "grad_norm": 0.5942494869232178,
      "learning_rate": 0.002828,
      "loss": 1.1473,
      "step": 478144
    },
    {
      "epoch": 35.888030018761725,
      "grad_norm": 0.6786660552024841,
      "learning_rate": 0.002828,
      "loss": 1.1421,
      "step": 478208
    },
    {
      "epoch": 35.8928330206379,
      "grad_norm": 0.612209677696228,
      "learning_rate": 0.002828,
      "loss": 1.1488,
      "step": 478272
    },
    {
      "epoch": 35.89763602251407,
      "grad_norm": 0.7059167623519897,
      "learning_rate": 0.002828,
      "loss": 1.1428,
      "step": 478336
    },
    {
      "epoch": 35.90243902439025,
      "grad_norm": 0.48968270421028137,
      "learning_rate": 0.002828,
      "loss": 1.1393,
      "step": 478400
    },
    {
      "epoch": 35.90724202626642,
      "grad_norm": 0.5701474547386169,
      "learning_rate": 0.002828,
      "loss": 1.1466,
      "step": 478464
    },
    {
      "epoch": 35.912045028142586,
      "grad_norm": 0.6382841467857361,
      "learning_rate": 0.002828,
      "loss": 1.1427,
      "step": 478528
    },
    {
      "epoch": 35.91684803001876,
      "grad_norm": 0.7465910911560059,
      "learning_rate": 0.002828,
      "loss": 1.1348,
      "step": 478592
    },
    {
      "epoch": 35.92165103189493,
      "grad_norm": 0.6528146266937256,
      "learning_rate": 0.002828,
      "loss": 1.1522,
      "step": 478656
    },
    {
      "epoch": 35.92645403377111,
      "grad_norm": 0.577727198600769,
      "learning_rate": 0.002828,
      "loss": 1.1452,
      "step": 478720
    },
    {
      "epoch": 35.93125703564728,
      "grad_norm": 0.5171624422073364,
      "learning_rate": 0.002828,
      "loss": 1.1414,
      "step": 478784
    },
    {
      "epoch": 35.936060037523454,
      "grad_norm": 0.6009478569030762,
      "learning_rate": 0.002828,
      "loss": 1.1455,
      "step": 478848
    },
    {
      "epoch": 35.94086303939962,
      "grad_norm": 0.7087094783782959,
      "learning_rate": 0.002828,
      "loss": 1.142,
      "step": 478912
    },
    {
      "epoch": 35.9456660412758,
      "grad_norm": 0.5902978777885437,
      "learning_rate": 0.002828,
      "loss": 1.1488,
      "step": 478976
    },
    {
      "epoch": 35.95046904315197,
      "grad_norm": 0.6435022950172424,
      "learning_rate": 0.002828,
      "loss": 1.1403,
      "step": 479040
    },
    {
      "epoch": 35.955272045028146,
      "grad_norm": 0.5205735564231873,
      "learning_rate": 0.002828,
      "loss": 1.1398,
      "step": 479104
    },
    {
      "epoch": 35.960075046904315,
      "grad_norm": 0.5627501010894775,
      "learning_rate": 0.002828,
      "loss": 1.1452,
      "step": 479168
    },
    {
      "epoch": 35.96487804878049,
      "grad_norm": 0.5993406176567078,
      "learning_rate": 0.002828,
      "loss": 1.1441,
      "step": 479232
    },
    {
      "epoch": 35.96968105065666,
      "grad_norm": 0.6258234977722168,
      "learning_rate": 0.002828,
      "loss": 1.1438,
      "step": 479296
    },
    {
      "epoch": 35.97448405253283,
      "grad_norm": 0.7097495794296265,
      "learning_rate": 0.002828,
      "loss": 1.1469,
      "step": 479360
    },
    {
      "epoch": 35.979287054409006,
      "grad_norm": 0.5994864106178284,
      "learning_rate": 0.002828,
      "loss": 1.1457,
      "step": 479424
    },
    {
      "epoch": 35.984090056285176,
      "grad_norm": 0.7201527953147888,
      "learning_rate": 0.002828,
      "loss": 1.1439,
      "step": 479488
    },
    {
      "epoch": 35.98889305816135,
      "grad_norm": 0.5710475444793701,
      "learning_rate": 0.002828,
      "loss": 1.1458,
      "step": 479552
    },
    {
      "epoch": 35.99369606003752,
      "grad_norm": 0.7653760313987732,
      "learning_rate": 0.002828,
      "loss": 1.1467,
      "step": 479616
    },
    {
      "epoch": 35.9984990619137,
      "grad_norm": 0.7502647638320923,
      "learning_rate": 0.002828,
      "loss": 1.1445,
      "step": 479680
    },
    {
      "epoch": 36.00330206378987,
      "grad_norm": 0.6153002977371216,
      "learning_rate": 0.002828,
      "loss": 1.1176,
      "step": 479744
    },
    {
      "epoch": 36.008105065666044,
      "grad_norm": 0.733729362487793,
      "learning_rate": 0.002828,
      "loss": 1.1042,
      "step": 479808
    },
    {
      "epoch": 36.01290806754221,
      "grad_norm": 0.5874631404876709,
      "learning_rate": 0.002828,
      "loss": 1.1063,
      "step": 479872
    },
    {
      "epoch": 36.01771106941839,
      "grad_norm": 0.6294484734535217,
      "learning_rate": 0.002828,
      "loss": 1.1132,
      "step": 479936
    },
    {
      "epoch": 36.02251407129456,
      "grad_norm": 0.8180893063545227,
      "learning_rate": 0.002828,
      "loss": 1.108,
      "step": 480000
    },
    {
      "epoch": 36.02731707317073,
      "grad_norm": 0.6071234941482544,
      "learning_rate": 0.002828,
      "loss": 1.1157,
      "step": 480064
    },
    {
      "epoch": 36.032120075046905,
      "grad_norm": 0.8069936633110046,
      "learning_rate": 0.002828,
      "loss": 1.1082,
      "step": 480128
    },
    {
      "epoch": 36.036923076923074,
      "grad_norm": 0.6223371028900146,
      "learning_rate": 0.002828,
      "loss": 1.109,
      "step": 480192
    },
    {
      "epoch": 36.04172607879925,
      "grad_norm": 0.6613013744354248,
      "learning_rate": 0.002828,
      "loss": 1.1019,
      "step": 480256
    },
    {
      "epoch": 36.04652908067542,
      "grad_norm": 0.5982505083084106,
      "learning_rate": 0.002828,
      "loss": 1.1134,
      "step": 480320
    },
    {
      "epoch": 36.051332082551596,
      "grad_norm": 0.5876198410987854,
      "learning_rate": 0.002828,
      "loss": 1.1139,
      "step": 480384
    },
    {
      "epoch": 36.056135084427765,
      "grad_norm": 0.5883331894874573,
      "learning_rate": 0.002828,
      "loss": 1.1104,
      "step": 480448
    },
    {
      "epoch": 36.06093808630394,
      "grad_norm": 0.6238200664520264,
      "learning_rate": 0.002828,
      "loss": 1.1122,
      "step": 480512
    },
    {
      "epoch": 36.06574108818011,
      "grad_norm": 0.6554744839668274,
      "learning_rate": 0.002828,
      "loss": 1.11,
      "step": 480576
    },
    {
      "epoch": 36.07054409005629,
      "grad_norm": 0.7005370855331421,
      "learning_rate": 0.002828,
      "loss": 1.1063,
      "step": 480640
    },
    {
      "epoch": 36.07534709193246,
      "grad_norm": 0.5803971886634827,
      "learning_rate": 0.002828,
      "loss": 1.1154,
      "step": 480704
    },
    {
      "epoch": 36.08015009380863,
      "grad_norm": 0.7155123353004456,
      "learning_rate": 0.002828,
      "loss": 1.112,
      "step": 480768
    },
    {
      "epoch": 36.0849530956848,
      "grad_norm": 0.605573832988739,
      "learning_rate": 0.002828,
      "loss": 1.1157,
      "step": 480832
    },
    {
      "epoch": 36.08975609756097,
      "grad_norm": 0.587078332901001,
      "learning_rate": 0.002828,
      "loss": 1.1125,
      "step": 480896
    },
    {
      "epoch": 36.09455909943715,
      "grad_norm": 0.6320798993110657,
      "learning_rate": 0.002828,
      "loss": 1.1114,
      "step": 480960
    },
    {
      "epoch": 36.09936210131332,
      "grad_norm": 0.6711087822914124,
      "learning_rate": 0.002828,
      "loss": 1.1164,
      "step": 481024
    },
    {
      "epoch": 36.104165103189494,
      "grad_norm": 0.721170961856842,
      "learning_rate": 0.002828,
      "loss": 1.112,
      "step": 481088
    },
    {
      "epoch": 36.108968105065664,
      "grad_norm": 0.5629286170005798,
      "learning_rate": 0.002828,
      "loss": 1.1185,
      "step": 481152
    },
    {
      "epoch": 36.11377110694184,
      "grad_norm": 0.5569593906402588,
      "learning_rate": 0.002828,
      "loss": 1.1147,
      "step": 481216
    },
    {
      "epoch": 36.11857410881801,
      "grad_norm": 0.636684000492096,
      "learning_rate": 0.002828,
      "loss": 1.1149,
      "step": 481280
    },
    {
      "epoch": 36.123377110694186,
      "grad_norm": 0.5977029204368591,
      "learning_rate": 0.002828,
      "loss": 1.1142,
      "step": 481344
    },
    {
      "epoch": 36.128180112570355,
      "grad_norm": 0.6242080330848694,
      "learning_rate": 0.002828,
      "loss": 1.1133,
      "step": 481408
    },
    {
      "epoch": 36.13298311444653,
      "grad_norm": 0.6300401091575623,
      "learning_rate": 0.002828,
      "loss": 1.1175,
      "step": 481472
    },
    {
      "epoch": 36.1377861163227,
      "grad_norm": 0.6650278568267822,
      "learning_rate": 0.002828,
      "loss": 1.1174,
      "step": 481536
    },
    {
      "epoch": 36.14258911819888,
      "grad_norm": 0.5920589566230774,
      "learning_rate": 0.002828,
      "loss": 1.118,
      "step": 481600
    },
    {
      "epoch": 36.14739212007505,
      "grad_norm": 0.623343288898468,
      "learning_rate": 0.002828,
      "loss": 1.1124,
      "step": 481664
    },
    {
      "epoch": 36.152195121951216,
      "grad_norm": 0.7111536860466003,
      "learning_rate": 0.002828,
      "loss": 1.1205,
      "step": 481728
    },
    {
      "epoch": 36.15699812382739,
      "grad_norm": 0.6605375409126282,
      "learning_rate": 0.002828,
      "loss": 1.1181,
      "step": 481792
    },
    {
      "epoch": 36.16180112570356,
      "grad_norm": 0.6241599321365356,
      "learning_rate": 0.002828,
      "loss": 1.1124,
      "step": 481856
    },
    {
      "epoch": 36.16660412757974,
      "grad_norm": 0.5739508271217346,
      "learning_rate": 0.002828,
      "loss": 1.1213,
      "step": 481920
    },
    {
      "epoch": 36.17140712945591,
      "grad_norm": 0.5860132575035095,
      "learning_rate": 0.002828,
      "loss": 1.114,
      "step": 481984
    },
    {
      "epoch": 36.176210131332084,
      "grad_norm": 0.6558253169059753,
      "learning_rate": 0.002828,
      "loss": 1.1196,
      "step": 482048
    },
    {
      "epoch": 36.18101313320825,
      "grad_norm": 0.6566668152809143,
      "learning_rate": 0.002828,
      "loss": 1.1229,
      "step": 482112
    },
    {
      "epoch": 36.18581613508443,
      "grad_norm": 0.6323477625846863,
      "learning_rate": 0.002828,
      "loss": 1.113,
      "step": 482176
    },
    {
      "epoch": 36.1906191369606,
      "grad_norm": 0.6525058746337891,
      "learning_rate": 0.002828,
      "loss": 1.1159,
      "step": 482240
    },
    {
      "epoch": 36.195422138836776,
      "grad_norm": 0.6623163223266602,
      "learning_rate": 0.002828,
      "loss": 1.1177,
      "step": 482304
    },
    {
      "epoch": 36.200225140712945,
      "grad_norm": 0.683592677116394,
      "learning_rate": 0.002828,
      "loss": 1.1155,
      "step": 482368
    },
    {
      "epoch": 36.20502814258912,
      "grad_norm": 0.7265550494194031,
      "learning_rate": 0.002828,
      "loss": 1.1182,
      "step": 482432
    },
    {
      "epoch": 36.20983114446529,
      "grad_norm": 0.5851932168006897,
      "learning_rate": 0.002828,
      "loss": 1.1199,
      "step": 482496
    },
    {
      "epoch": 36.21463414634146,
      "grad_norm": 0.6824109554290771,
      "learning_rate": 0.002828,
      "loss": 1.1244,
      "step": 482560
    },
    {
      "epoch": 36.21943714821764,
      "grad_norm": 0.6103503704071045,
      "learning_rate": 0.002828,
      "loss": 1.1169,
      "step": 482624
    },
    {
      "epoch": 36.224240150093806,
      "grad_norm": 0.5993927717208862,
      "learning_rate": 0.002828,
      "loss": 1.1143,
      "step": 482688
    },
    {
      "epoch": 36.22904315196998,
      "grad_norm": 0.601505696773529,
      "learning_rate": 0.002828,
      "loss": 1.1209,
      "step": 482752
    },
    {
      "epoch": 36.23384615384615,
      "grad_norm": 0.6602485179901123,
      "learning_rate": 0.002828,
      "loss": 1.1153,
      "step": 482816
    },
    {
      "epoch": 36.23864915572233,
      "grad_norm": 0.7924669981002808,
      "learning_rate": 0.002828,
      "loss": 1.1218,
      "step": 482880
    },
    {
      "epoch": 36.2434521575985,
      "grad_norm": 0.6222769021987915,
      "learning_rate": 0.002828,
      "loss": 1.1201,
      "step": 482944
    },
    {
      "epoch": 36.248255159474674,
      "grad_norm": 0.5969858765602112,
      "learning_rate": 0.002828,
      "loss": 1.1201,
      "step": 483008
    },
    {
      "epoch": 36.25305816135084,
      "grad_norm": 0.6420055031776428,
      "learning_rate": 0.002828,
      "loss": 1.1219,
      "step": 483072
    },
    {
      "epoch": 36.25786116322702,
      "grad_norm": 0.6040505170822144,
      "learning_rate": 0.002828,
      "loss": 1.1151,
      "step": 483136
    },
    {
      "epoch": 36.26266416510319,
      "grad_norm": 0.5828750133514404,
      "learning_rate": 0.002828,
      "loss": 1.1249,
      "step": 483200
    },
    {
      "epoch": 36.267467166979365,
      "grad_norm": 0.5604116320610046,
      "learning_rate": 0.002828,
      "loss": 1.1217,
      "step": 483264
    },
    {
      "epoch": 36.272270168855535,
      "grad_norm": 0.7253416776657104,
      "learning_rate": 0.002828,
      "loss": 1.1243,
      "step": 483328
    },
    {
      "epoch": 36.277073170731704,
      "grad_norm": 0.538339376449585,
      "learning_rate": 0.002828,
      "loss": 1.1237,
      "step": 483392
    },
    {
      "epoch": 36.28187617260788,
      "grad_norm": 0.6550213098526001,
      "learning_rate": 0.002828,
      "loss": 1.1175,
      "step": 483456
    },
    {
      "epoch": 36.28667917448405,
      "grad_norm": 0.6307083964347839,
      "learning_rate": 0.002828,
      "loss": 1.1153,
      "step": 483520
    },
    {
      "epoch": 36.291482176360226,
      "grad_norm": 0.7109634280204773,
      "learning_rate": 0.002828,
      "loss": 1.1255,
      "step": 483584
    },
    {
      "epoch": 36.296285178236396,
      "grad_norm": 0.560472309589386,
      "learning_rate": 0.002828,
      "loss": 1.1171,
      "step": 483648
    },
    {
      "epoch": 36.30108818011257,
      "grad_norm": 0.5119628310203552,
      "learning_rate": 0.002828,
      "loss": 1.1189,
      "step": 483712
    },
    {
      "epoch": 36.30589118198874,
      "grad_norm": 0.66200852394104,
      "learning_rate": 0.002828,
      "loss": 1.1222,
      "step": 483776
    },
    {
      "epoch": 36.31069418386492,
      "grad_norm": 0.6191007494926453,
      "learning_rate": 0.002828,
      "loss": 1.1173,
      "step": 483840
    },
    {
      "epoch": 36.31549718574109,
      "grad_norm": 0.7778931260108948,
      "learning_rate": 0.002828,
      "loss": 1.1213,
      "step": 483904
    },
    {
      "epoch": 36.32030018761726,
      "grad_norm": 0.6234067678451538,
      "learning_rate": 0.002828,
      "loss": 1.1219,
      "step": 483968
    },
    {
      "epoch": 36.32510318949343,
      "grad_norm": 0.5811113119125366,
      "learning_rate": 0.002828,
      "loss": 1.1212,
      "step": 484032
    },
    {
      "epoch": 36.32990619136961,
      "grad_norm": 0.5138593912124634,
      "learning_rate": 0.002828,
      "loss": 1.125,
      "step": 484096
    },
    {
      "epoch": 36.33470919324578,
      "grad_norm": 0.5733650922775269,
      "learning_rate": 0.002828,
      "loss": 1.127,
      "step": 484160
    },
    {
      "epoch": 36.33951219512195,
      "grad_norm": 0.634680449962616,
      "learning_rate": 0.002828,
      "loss": 1.1238,
      "step": 484224
    },
    {
      "epoch": 36.344315196998124,
      "grad_norm": 0.6485024690628052,
      "learning_rate": 0.002828,
      "loss": 1.1217,
      "step": 484288
    },
    {
      "epoch": 36.349118198874294,
      "grad_norm": 0.6225195527076721,
      "learning_rate": 0.002828,
      "loss": 1.1254,
      "step": 484352
    },
    {
      "epoch": 36.35392120075047,
      "grad_norm": 0.6214507222175598,
      "learning_rate": 0.002828,
      "loss": 1.123,
      "step": 484416
    },
    {
      "epoch": 36.35872420262664,
      "grad_norm": 0.49921342730522156,
      "learning_rate": 0.002828,
      "loss": 1.1211,
      "step": 484480
    },
    {
      "epoch": 36.363527204502816,
      "grad_norm": 0.6985149383544922,
      "learning_rate": 0.002828,
      "loss": 1.1162,
      "step": 484544
    },
    {
      "epoch": 36.368330206378985,
      "grad_norm": 0.7828275561332703,
      "learning_rate": 0.002828,
      "loss": 1.132,
      "step": 484608
    },
    {
      "epoch": 36.37313320825516,
      "grad_norm": 0.6335474252700806,
      "learning_rate": 0.002828,
      "loss": 1.1208,
      "step": 484672
    },
    {
      "epoch": 36.37793621013133,
      "grad_norm": 0.6620512008666992,
      "learning_rate": 0.002828,
      "loss": 1.1188,
      "step": 484736
    },
    {
      "epoch": 36.38273921200751,
      "grad_norm": 0.48815280199050903,
      "learning_rate": 0.002828,
      "loss": 1.1223,
      "step": 484800
    },
    {
      "epoch": 36.38754221388368,
      "grad_norm": 0.6472222208976746,
      "learning_rate": 0.002828,
      "loss": 1.1208,
      "step": 484864
    },
    {
      "epoch": 36.39234521575985,
      "grad_norm": 0.5891759991645813,
      "learning_rate": 0.002828,
      "loss": 1.1215,
      "step": 484928
    },
    {
      "epoch": 36.39714821763602,
      "grad_norm": 0.7245331406593323,
      "learning_rate": 0.002828,
      "loss": 1.123,
      "step": 484992
    },
    {
      "epoch": 36.40195121951219,
      "grad_norm": 0.6620111465454102,
      "learning_rate": 0.002828,
      "loss": 1.1267,
      "step": 485056
    },
    {
      "epoch": 36.40675422138837,
      "grad_norm": 0.7744014263153076,
      "learning_rate": 0.002828,
      "loss": 1.1178,
      "step": 485120
    },
    {
      "epoch": 36.41155722326454,
      "grad_norm": 0.5176191329956055,
      "learning_rate": 0.002828,
      "loss": 1.1264,
      "step": 485184
    },
    {
      "epoch": 36.416360225140714,
      "grad_norm": 0.7624377608299255,
      "learning_rate": 0.002828,
      "loss": 1.1319,
      "step": 485248
    },
    {
      "epoch": 36.42116322701688,
      "grad_norm": 0.5986029505729675,
      "learning_rate": 0.002828,
      "loss": 1.118,
      "step": 485312
    },
    {
      "epoch": 36.42596622889306,
      "grad_norm": 0.5113912224769592,
      "learning_rate": 0.002828,
      "loss": 1.1189,
      "step": 485376
    },
    {
      "epoch": 36.43076923076923,
      "grad_norm": 0.6814943552017212,
      "learning_rate": 0.002828,
      "loss": 1.1253,
      "step": 485440
    },
    {
      "epoch": 36.435572232645406,
      "grad_norm": 0.682895839214325,
      "learning_rate": 0.002828,
      "loss": 1.125,
      "step": 485504
    },
    {
      "epoch": 36.440375234521575,
      "grad_norm": 0.5333254337310791,
      "learning_rate": 0.002828,
      "loss": 1.13,
      "step": 485568
    },
    {
      "epoch": 36.44517823639775,
      "grad_norm": 0.6200302839279175,
      "learning_rate": 0.002828,
      "loss": 1.1251,
      "step": 485632
    },
    {
      "epoch": 36.44998123827392,
      "grad_norm": 0.5972156524658203,
      "learning_rate": 0.002828,
      "loss": 1.1321,
      "step": 485696
    },
    {
      "epoch": 36.4547842401501,
      "grad_norm": 0.60141521692276,
      "learning_rate": 0.002828,
      "loss": 1.1285,
      "step": 485760
    },
    {
      "epoch": 36.45958724202627,
      "grad_norm": 0.47839000821113586,
      "learning_rate": 0.002828,
      "loss": 1.1252,
      "step": 485824
    },
    {
      "epoch": 36.464390243902436,
      "grad_norm": 0.6346657276153564,
      "learning_rate": 0.002828,
      "loss": 1.1298,
      "step": 485888
    },
    {
      "epoch": 36.46919324577861,
      "grad_norm": 0.6371756792068481,
      "learning_rate": 0.002828,
      "loss": 1.1202,
      "step": 485952
    },
    {
      "epoch": 36.47399624765478,
      "grad_norm": 0.5818215608596802,
      "learning_rate": 0.002828,
      "loss": 1.122,
      "step": 486016
    },
    {
      "epoch": 36.47879924953096,
      "grad_norm": 0.6121278405189514,
      "learning_rate": 0.002828,
      "loss": 1.1273,
      "step": 486080
    },
    {
      "epoch": 36.48360225140713,
      "grad_norm": 0.5700222849845886,
      "learning_rate": 0.002828,
      "loss": 1.1233,
      "step": 486144
    },
    {
      "epoch": 36.488405253283304,
      "grad_norm": 0.7344767451286316,
      "learning_rate": 0.002828,
      "loss": 1.1247,
      "step": 486208
    },
    {
      "epoch": 36.49320825515947,
      "grad_norm": 0.657575786113739,
      "learning_rate": 0.002828,
      "loss": 1.1269,
      "step": 486272
    },
    {
      "epoch": 36.49801125703565,
      "grad_norm": 0.574347972869873,
      "learning_rate": 0.002828,
      "loss": 1.1265,
      "step": 486336
    },
    {
      "epoch": 36.50281425891182,
      "grad_norm": 0.6713154911994934,
      "learning_rate": 0.002828,
      "loss": 1.1197,
      "step": 486400
    },
    {
      "epoch": 36.507617260787995,
      "grad_norm": 0.5219811201095581,
      "learning_rate": 0.002828,
      "loss": 1.1242,
      "step": 486464
    },
    {
      "epoch": 36.512420262664165,
      "grad_norm": 0.6195260286331177,
      "learning_rate": 0.002828,
      "loss": 1.1247,
      "step": 486528
    },
    {
      "epoch": 36.51722326454034,
      "grad_norm": 0.560436487197876,
      "learning_rate": 0.002828,
      "loss": 1.1284,
      "step": 486592
    },
    {
      "epoch": 36.52202626641651,
      "grad_norm": 0.6903839111328125,
      "learning_rate": 0.002828,
      "loss": 1.1195,
      "step": 486656
    },
    {
      "epoch": 36.52682926829268,
      "grad_norm": 0.606224536895752,
      "learning_rate": 0.002828,
      "loss": 1.1243,
      "step": 486720
    },
    {
      "epoch": 36.531632270168856,
      "grad_norm": 0.5438897609710693,
      "learning_rate": 0.002828,
      "loss": 1.1307,
      "step": 486784
    },
    {
      "epoch": 36.536435272045026,
      "grad_norm": 0.6882777214050293,
      "learning_rate": 0.002828,
      "loss": 1.1291,
      "step": 486848
    },
    {
      "epoch": 36.5412382739212,
      "grad_norm": 0.6011608839035034,
      "learning_rate": 0.002828,
      "loss": 1.1219,
      "step": 486912
    },
    {
      "epoch": 36.54604127579737,
      "grad_norm": 0.6009193658828735,
      "learning_rate": 0.002828,
      "loss": 1.1277,
      "step": 486976
    },
    {
      "epoch": 36.55084427767355,
      "grad_norm": 0.6801005601882935,
      "learning_rate": 0.002828,
      "loss": 1.1304,
      "step": 487040
    },
    {
      "epoch": 36.55564727954972,
      "grad_norm": 0.6301913261413574,
      "learning_rate": 0.002828,
      "loss": 1.1281,
      "step": 487104
    },
    {
      "epoch": 36.560450281425894,
      "grad_norm": 0.5682470798492432,
      "learning_rate": 0.002828,
      "loss": 1.1217,
      "step": 487168
    },
    {
      "epoch": 36.56525328330206,
      "grad_norm": 0.6990990042686462,
      "learning_rate": 0.002828,
      "loss": 1.1291,
      "step": 487232
    },
    {
      "epoch": 36.57005628517824,
      "grad_norm": 0.6065164804458618,
      "learning_rate": 0.002828,
      "loss": 1.1249,
      "step": 487296
    },
    {
      "epoch": 36.57485928705441,
      "grad_norm": 0.6101195216178894,
      "learning_rate": 0.002828,
      "loss": 1.128,
      "step": 487360
    },
    {
      "epoch": 36.57966228893058,
      "grad_norm": 0.6691819429397583,
      "learning_rate": 0.002828,
      "loss": 1.1252,
      "step": 487424
    },
    {
      "epoch": 36.584465290806754,
      "grad_norm": 0.7639368772506714,
      "learning_rate": 0.002828,
      "loss": 1.1233,
      "step": 487488
    },
    {
      "epoch": 36.589268292682924,
      "grad_norm": 0.8708202242851257,
      "learning_rate": 0.002828,
      "loss": 1.1313,
      "step": 487552
    },
    {
      "epoch": 36.5940712945591,
      "grad_norm": 0.5969586968421936,
      "learning_rate": 0.002828,
      "loss": 1.1267,
      "step": 487616
    },
    {
      "epoch": 36.59887429643527,
      "grad_norm": 0.542630672454834,
      "learning_rate": 0.002828,
      "loss": 1.1269,
      "step": 487680
    },
    {
      "epoch": 36.603677298311446,
      "grad_norm": 0.5667511224746704,
      "learning_rate": 0.002828,
      "loss": 1.1309,
      "step": 487744
    },
    {
      "epoch": 36.608480300187615,
      "grad_norm": 0.6424358487129211,
      "learning_rate": 0.002828,
      "loss": 1.1264,
      "step": 487808
    },
    {
      "epoch": 36.61328330206379,
      "grad_norm": 0.5762342810630798,
      "learning_rate": 0.002828,
      "loss": 1.1292,
      "step": 487872
    },
    {
      "epoch": 36.61808630393996,
      "grad_norm": 0.5913277268409729,
      "learning_rate": 0.002828,
      "loss": 1.1322,
      "step": 487936
    },
    {
      "epoch": 36.62288930581614,
      "grad_norm": 0.7270256280899048,
      "learning_rate": 0.002828,
      "loss": 1.1356,
      "step": 488000
    },
    {
      "epoch": 36.62769230769231,
      "grad_norm": 0.7120741605758667,
      "learning_rate": 0.002828,
      "loss": 1.1286,
      "step": 488064
    },
    {
      "epoch": 36.63249530956848,
      "grad_norm": 0.5873661637306213,
      "learning_rate": 0.002828,
      "loss": 1.1281,
      "step": 488128
    },
    {
      "epoch": 36.63729831144465,
      "grad_norm": 0.5965089797973633,
      "learning_rate": 0.002828,
      "loss": 1.1288,
      "step": 488192
    },
    {
      "epoch": 36.64210131332082,
      "grad_norm": 0.5685919523239136,
      "learning_rate": 0.002828,
      "loss": 1.1266,
      "step": 488256
    },
    {
      "epoch": 36.646904315197,
      "grad_norm": 0.5000462532043457,
      "learning_rate": 0.002828,
      "loss": 1.1342,
      "step": 488320
    },
    {
      "epoch": 36.65170731707317,
      "grad_norm": 0.6275484561920166,
      "learning_rate": 0.002828,
      "loss": 1.13,
      "step": 488384
    },
    {
      "epoch": 36.656510318949344,
      "grad_norm": 0.6414892077445984,
      "learning_rate": 0.002828,
      "loss": 1.1309,
      "step": 488448
    },
    {
      "epoch": 36.66131332082551,
      "grad_norm": 0.632480800151825,
      "learning_rate": 0.002828,
      "loss": 1.1315,
      "step": 488512
    },
    {
      "epoch": 36.66611632270169,
      "grad_norm": 0.6146697998046875,
      "learning_rate": 0.002828,
      "loss": 1.1311,
      "step": 488576
    },
    {
      "epoch": 36.67091932457786,
      "grad_norm": 0.6127843856811523,
      "learning_rate": 0.002828,
      "loss": 1.1307,
      "step": 488640
    },
    {
      "epoch": 36.675722326454036,
      "grad_norm": 0.6762889623641968,
      "learning_rate": 0.002828,
      "loss": 1.128,
      "step": 488704
    },
    {
      "epoch": 36.680525328330205,
      "grad_norm": 0.5495359301567078,
      "learning_rate": 0.002828,
      "loss": 1.1334,
      "step": 488768
    },
    {
      "epoch": 36.68532833020638,
      "grad_norm": 0.5311427116394043,
      "learning_rate": 0.002828,
      "loss": 1.1291,
      "step": 488832
    },
    {
      "epoch": 36.69013133208255,
      "grad_norm": 0.6519775986671448,
      "learning_rate": 0.002828,
      "loss": 1.1324,
      "step": 488896
    },
    {
      "epoch": 36.69493433395873,
      "grad_norm": 0.5537697076797485,
      "learning_rate": 0.002828,
      "loss": 1.1287,
      "step": 488960
    },
    {
      "epoch": 36.6997373358349,
      "grad_norm": 0.6321004629135132,
      "learning_rate": 0.002828,
      "loss": 1.1281,
      "step": 489024
    },
    {
      "epoch": 36.704540337711066,
      "grad_norm": 0.583900511264801,
      "learning_rate": 0.002828,
      "loss": 1.123,
      "step": 489088
    },
    {
      "epoch": 36.70934333958724,
      "grad_norm": 0.6706740856170654,
      "learning_rate": 0.002828,
      "loss": 1.1321,
      "step": 489152
    },
    {
      "epoch": 36.71414634146341,
      "grad_norm": 0.5570067763328552,
      "learning_rate": 0.002828,
      "loss": 1.132,
      "step": 489216
    },
    {
      "epoch": 36.71894934333959,
      "grad_norm": 0.747728705406189,
      "learning_rate": 0.002828,
      "loss": 1.1271,
      "step": 489280
    },
    {
      "epoch": 36.72375234521576,
      "grad_norm": 0.6183828115463257,
      "learning_rate": 0.002828,
      "loss": 1.1283,
      "step": 489344
    },
    {
      "epoch": 36.728555347091934,
      "grad_norm": 0.7345849871635437,
      "learning_rate": 0.002828,
      "loss": 1.1314,
      "step": 489408
    },
    {
      "epoch": 36.7333583489681,
      "grad_norm": 0.6629622578620911,
      "learning_rate": 0.002828,
      "loss": 1.1324,
      "step": 489472
    },
    {
      "epoch": 36.73816135084428,
      "grad_norm": 0.5575123429298401,
      "learning_rate": 0.002828,
      "loss": 1.131,
      "step": 489536
    },
    {
      "epoch": 36.74296435272045,
      "grad_norm": 0.5553168654441833,
      "learning_rate": 0.002828,
      "loss": 1.1322,
      "step": 489600
    },
    {
      "epoch": 36.747767354596625,
      "grad_norm": 0.7154178023338318,
      "learning_rate": 0.002828,
      "loss": 1.1329,
      "step": 489664
    },
    {
      "epoch": 36.752570356472795,
      "grad_norm": 0.6160262823104858,
      "learning_rate": 0.002828,
      "loss": 1.1305,
      "step": 489728
    },
    {
      "epoch": 36.75737335834897,
      "grad_norm": 0.577056348323822,
      "learning_rate": 0.002828,
      "loss": 1.1311,
      "step": 489792
    },
    {
      "epoch": 36.76217636022514,
      "grad_norm": 0.5922368168830872,
      "learning_rate": 0.002828,
      "loss": 1.1301,
      "step": 489856
    },
    {
      "epoch": 36.76697936210131,
      "grad_norm": 0.603999674320221,
      "learning_rate": 0.002828,
      "loss": 1.1282,
      "step": 489920
    },
    {
      "epoch": 36.771782363977486,
      "grad_norm": 0.5781303644180298,
      "learning_rate": 0.002828,
      "loss": 1.13,
      "step": 489984
    },
    {
      "epoch": 36.776585365853656,
      "grad_norm": 0.6862877607345581,
      "learning_rate": 0.002828,
      "loss": 1.1338,
      "step": 490048
    },
    {
      "epoch": 36.78138836772983,
      "grad_norm": 0.5813214182853699,
      "learning_rate": 0.002828,
      "loss": 1.1285,
      "step": 490112
    },
    {
      "epoch": 36.786191369606,
      "grad_norm": 0.6811426877975464,
      "learning_rate": 0.002828,
      "loss": 1.132,
      "step": 490176
    },
    {
      "epoch": 36.79099437148218,
      "grad_norm": 0.5649698376655579,
      "learning_rate": 0.002828,
      "loss": 1.1298,
      "step": 490240
    },
    {
      "epoch": 36.79579737335835,
      "grad_norm": 0.49407100677490234,
      "learning_rate": 0.002828,
      "loss": 1.1315,
      "step": 490304
    },
    {
      "epoch": 36.800600375234524,
      "grad_norm": 0.6909792423248291,
      "learning_rate": 0.002828,
      "loss": 1.1343,
      "step": 490368
    },
    {
      "epoch": 36.80540337711069,
      "grad_norm": 0.5602936744689941,
      "learning_rate": 0.002828,
      "loss": 1.1293,
      "step": 490432
    },
    {
      "epoch": 36.81020637898687,
      "grad_norm": 0.7587369680404663,
      "learning_rate": 0.002828,
      "loss": 1.1348,
      "step": 490496
    },
    {
      "epoch": 36.81500938086304,
      "grad_norm": 0.5359601378440857,
      "learning_rate": 0.002828,
      "loss": 1.1338,
      "step": 490560
    },
    {
      "epoch": 36.819812382739215,
      "grad_norm": 0.671832263469696,
      "learning_rate": 0.002828,
      "loss": 1.1335,
      "step": 490624
    },
    {
      "epoch": 36.824615384615385,
      "grad_norm": 0.5710498690605164,
      "learning_rate": 0.002828,
      "loss": 1.1339,
      "step": 490688
    },
    {
      "epoch": 36.829418386491554,
      "grad_norm": 0.6783512234687805,
      "learning_rate": 0.002828,
      "loss": 1.132,
      "step": 490752
    },
    {
      "epoch": 36.83422138836773,
      "grad_norm": 0.5253806114196777,
      "learning_rate": 0.002828,
      "loss": 1.1286,
      "step": 490816
    },
    {
      "epoch": 36.8390243902439,
      "grad_norm": 0.6593382954597473,
      "learning_rate": 0.002828,
      "loss": 1.1347,
      "step": 490880
    },
    {
      "epoch": 36.843827392120076,
      "grad_norm": 0.6432192325592041,
      "learning_rate": 0.002828,
      "loss": 1.1328,
      "step": 490944
    },
    {
      "epoch": 36.848630393996245,
      "grad_norm": 0.5609253644943237,
      "learning_rate": 0.002828,
      "loss": 1.1332,
      "step": 491008
    },
    {
      "epoch": 36.85343339587242,
      "grad_norm": 0.5702568292617798,
      "learning_rate": 0.002828,
      "loss": 1.1332,
      "step": 491072
    },
    {
      "epoch": 36.85823639774859,
      "grad_norm": 0.5661723613739014,
      "learning_rate": 0.002828,
      "loss": 1.1316,
      "step": 491136
    },
    {
      "epoch": 36.86303939962477,
      "grad_norm": 0.48676881194114685,
      "learning_rate": 0.002828,
      "loss": 1.1342,
      "step": 491200
    },
    {
      "epoch": 36.86784240150094,
      "grad_norm": 0.5914056897163391,
      "learning_rate": 0.002828,
      "loss": 1.1321,
      "step": 491264
    },
    {
      "epoch": 36.87264540337711,
      "grad_norm": 0.6446478962898254,
      "learning_rate": 0.002828,
      "loss": 1.128,
      "step": 491328
    },
    {
      "epoch": 36.87744840525328,
      "grad_norm": 0.6042805314064026,
      "learning_rate": 0.002828,
      "loss": 1.1299,
      "step": 491392
    },
    {
      "epoch": 36.88225140712946,
      "grad_norm": 0.5406030416488647,
      "learning_rate": 0.002828,
      "loss": 1.1354,
      "step": 491456
    },
    {
      "epoch": 36.88705440900563,
      "grad_norm": 0.5631481409072876,
      "learning_rate": 0.002828,
      "loss": 1.1374,
      "step": 491520
    },
    {
      "epoch": 36.8918574108818,
      "grad_norm": 0.6143375039100647,
      "learning_rate": 0.002828,
      "loss": 1.1303,
      "step": 491584
    },
    {
      "epoch": 36.896660412757974,
      "grad_norm": 0.5791195631027222,
      "learning_rate": 0.002828,
      "loss": 1.1346,
      "step": 491648
    },
    {
      "epoch": 36.901463414634144,
      "grad_norm": 0.6553552746772766,
      "learning_rate": 0.002828,
      "loss": 1.134,
      "step": 491712
    },
    {
      "epoch": 36.90626641651032,
      "grad_norm": 0.8238469958305359,
      "learning_rate": 0.002828,
      "loss": 1.1335,
      "step": 491776
    },
    {
      "epoch": 36.91106941838649,
      "grad_norm": 0.7293280959129333,
      "learning_rate": 0.002828,
      "loss": 1.1338,
      "step": 491840
    },
    {
      "epoch": 36.915872420262666,
      "grad_norm": 0.5596233010292053,
      "learning_rate": 0.002828,
      "loss": 1.1318,
      "step": 491904
    },
    {
      "epoch": 36.920675422138835,
      "grad_norm": 0.7237329483032227,
      "learning_rate": 0.002828,
      "loss": 1.1327,
      "step": 491968
    },
    {
      "epoch": 36.92547842401501,
      "grad_norm": 0.5710912346839905,
      "learning_rate": 0.002828,
      "loss": 1.1365,
      "step": 492032
    },
    {
      "epoch": 36.93028142589118,
      "grad_norm": 0.7024117708206177,
      "learning_rate": 0.002828,
      "loss": 1.1295,
      "step": 492096
    },
    {
      "epoch": 36.93508442776736,
      "grad_norm": 0.629672110080719,
      "learning_rate": 0.002828,
      "loss": 1.1367,
      "step": 492160
    },
    {
      "epoch": 36.93988742964353,
      "grad_norm": 0.7030770182609558,
      "learning_rate": 0.002828,
      "loss": 1.1313,
      "step": 492224
    },
    {
      "epoch": 36.9446904315197,
      "grad_norm": 0.6092023253440857,
      "learning_rate": 0.002828,
      "loss": 1.1352,
      "step": 492288
    },
    {
      "epoch": 36.94949343339587,
      "grad_norm": 0.5727062225341797,
      "learning_rate": 0.002828,
      "loss": 1.1333,
      "step": 492352
    },
    {
      "epoch": 36.95429643527204,
      "grad_norm": 0.9285120368003845,
      "learning_rate": 0.002828,
      "loss": 1.13,
      "step": 492416
    },
    {
      "epoch": 36.95909943714822,
      "grad_norm": 0.5199771523475647,
      "learning_rate": 0.002828,
      "loss": 1.1327,
      "step": 492480
    },
    {
      "epoch": 36.96390243902439,
      "grad_norm": 0.6424311995506287,
      "learning_rate": 0.002828,
      "loss": 1.1323,
      "step": 492544
    },
    {
      "epoch": 36.968705440900564,
      "grad_norm": 0.6531059145927429,
      "learning_rate": 0.002828,
      "loss": 1.1371,
      "step": 492608
    },
    {
      "epoch": 36.97350844277673,
      "grad_norm": 0.6560933589935303,
      "learning_rate": 0.002828,
      "loss": 1.1343,
      "step": 492672
    },
    {
      "epoch": 36.97831144465291,
      "grad_norm": 0.6605238914489746,
      "learning_rate": 0.002828,
      "loss": 1.1349,
      "step": 492736
    },
    {
      "epoch": 36.98311444652908,
      "grad_norm": 0.6459187865257263,
      "learning_rate": 0.002828,
      "loss": 1.1414,
      "step": 492800
    },
    {
      "epoch": 36.987917448405256,
      "grad_norm": 0.7478291392326355,
      "learning_rate": 0.002828,
      "loss": 1.1371,
      "step": 492864
    },
    {
      "epoch": 36.992720450281425,
      "grad_norm": 0.5982804894447327,
      "learning_rate": 0.002828,
      "loss": 1.141,
      "step": 492928
    },
    {
      "epoch": 36.9975234521576,
      "grad_norm": 0.6685630083084106,
      "learning_rate": 0.002828,
      "loss": 1.1354,
      "step": 492992
    },
    {
      "epoch": 37.00232645403377,
      "grad_norm": 0.6969371438026428,
      "learning_rate": 0.002828,
      "loss": 1.1187,
      "step": 493056
    },
    {
      "epoch": 37.00712945590995,
      "grad_norm": 0.6465162634849548,
      "learning_rate": 0.002828,
      "loss": 1.0988,
      "step": 493120
    },
    {
      "epoch": 37.011932457786116,
      "grad_norm": 0.5774773955345154,
      "learning_rate": 0.002828,
      "loss": 1.0939,
      "step": 493184
    },
    {
      "epoch": 37.016735459662286,
      "grad_norm": 0.6074837446212769,
      "learning_rate": 0.002828,
      "loss": 1.1,
      "step": 493248
    },
    {
      "epoch": 37.02153846153846,
      "grad_norm": 0.5828660726547241,
      "learning_rate": 0.002828,
      "loss": 1.0984,
      "step": 493312
    },
    {
      "epoch": 37.02634146341463,
      "grad_norm": 0.5609144568443298,
      "learning_rate": 0.002828,
      "loss": 1.0944,
      "step": 493376
    },
    {
      "epoch": 37.03114446529081,
      "grad_norm": 0.6178881525993347,
      "learning_rate": 0.002828,
      "loss": 1.0958,
      "step": 493440
    },
    {
      "epoch": 37.03594746716698,
      "grad_norm": 0.661352276802063,
      "learning_rate": 0.002828,
      "loss": 1.1001,
      "step": 493504
    },
    {
      "epoch": 37.040750469043154,
      "grad_norm": 0.5430305004119873,
      "learning_rate": 0.002828,
      "loss": 1.105,
      "step": 493568
    },
    {
      "epoch": 37.04555347091932,
      "grad_norm": 0.5815340280532837,
      "learning_rate": 0.002828,
      "loss": 1.1003,
      "step": 493632
    },
    {
      "epoch": 37.0503564727955,
      "grad_norm": 0.6391292214393616,
      "learning_rate": 0.002828,
      "loss": 1.0955,
      "step": 493696
    },
    {
      "epoch": 37.05515947467167,
      "grad_norm": 0.5859202742576599,
      "learning_rate": 0.002828,
      "loss": 1.0967,
      "step": 493760
    },
    {
      "epoch": 37.059962476547845,
      "grad_norm": 0.6187096238136292,
      "learning_rate": 0.002828,
      "loss": 1.101,
      "step": 493824
    },
    {
      "epoch": 37.064765478424015,
      "grad_norm": 0.659162163734436,
      "learning_rate": 0.002828,
      "loss": 1.1019,
      "step": 493888
    },
    {
      "epoch": 37.06956848030019,
      "grad_norm": 0.6119021773338318,
      "learning_rate": 0.002828,
      "loss": 1.1004,
      "step": 493952
    },
    {
      "epoch": 37.07437148217636,
      "grad_norm": 0.7687475681304932,
      "learning_rate": 0.002828,
      "loss": 1.1025,
      "step": 494016
    },
    {
      "epoch": 37.07917448405253,
      "grad_norm": 0.622445821762085,
      "learning_rate": 0.002828,
      "loss": 1.1039,
      "step": 494080
    },
    {
      "epoch": 37.083977485928706,
      "grad_norm": 0.6039100885391235,
      "learning_rate": 0.002828,
      "loss": 1.0992,
      "step": 494144
    },
    {
      "epoch": 37.088780487804875,
      "grad_norm": 0.6412420272827148,
      "learning_rate": 0.002828,
      "loss": 1.1027,
      "step": 494208
    },
    {
      "epoch": 37.09358348968105,
      "grad_norm": 0.8018775582313538,
      "learning_rate": 0.002828,
      "loss": 1.1036,
      "step": 494272
    },
    {
      "epoch": 37.09838649155722,
      "grad_norm": 0.5880022644996643,
      "learning_rate": 0.002828,
      "loss": 1.1047,
      "step": 494336
    },
    {
      "epoch": 37.1031894934334,
      "grad_norm": 0.6250966191291809,
      "learning_rate": 0.002828,
      "loss": 1.0999,
      "step": 494400
    },
    {
      "epoch": 37.10799249530957,
      "grad_norm": 0.5100603103637695,
      "learning_rate": 0.002828,
      "loss": 1.095,
      "step": 494464
    },
    {
      "epoch": 37.11279549718574,
      "grad_norm": 0.494398832321167,
      "learning_rate": 0.002828,
      "loss": 1.1086,
      "step": 494528
    },
    {
      "epoch": 37.11759849906191,
      "grad_norm": 0.5320671200752258,
      "learning_rate": 0.002828,
      "loss": 1.1065,
      "step": 494592
    },
    {
      "epoch": 37.12240150093809,
      "grad_norm": 0.5626203417778015,
      "learning_rate": 0.002828,
      "loss": 1.1079,
      "step": 494656
    },
    {
      "epoch": 37.12720450281426,
      "grad_norm": 0.6252749562263489,
      "learning_rate": 0.002828,
      "loss": 1.1043,
      "step": 494720
    },
    {
      "epoch": 37.132007504690435,
      "grad_norm": 0.6196590662002563,
      "learning_rate": 0.002828,
      "loss": 1.1061,
      "step": 494784
    },
    {
      "epoch": 37.136810506566604,
      "grad_norm": 0.5838425159454346,
      "learning_rate": 0.002828,
      "loss": 1.1026,
      "step": 494848
    },
    {
      "epoch": 37.141613508442774,
      "grad_norm": 0.6245355606079102,
      "learning_rate": 0.002828,
      "loss": 1.1104,
      "step": 494912
    },
    {
      "epoch": 37.14641651031895,
      "grad_norm": 0.5560353994369507,
      "learning_rate": 0.002828,
      "loss": 1.106,
      "step": 494976
    },
    {
      "epoch": 37.15121951219512,
      "grad_norm": 0.6064783930778503,
      "learning_rate": 0.002828,
      "loss": 1.1084,
      "step": 495040
    },
    {
      "epoch": 37.156022514071296,
      "grad_norm": 0.6612809300422668,
      "learning_rate": 0.002828,
      "loss": 1.1043,
      "step": 495104
    },
    {
      "epoch": 37.160825515947465,
      "grad_norm": 0.6238330006599426,
      "learning_rate": 0.002828,
      "loss": 1.1094,
      "step": 495168
    },
    {
      "epoch": 37.16562851782364,
      "grad_norm": 0.6410529613494873,
      "learning_rate": 0.002828,
      "loss": 1.0994,
      "step": 495232
    },
    {
      "epoch": 37.17043151969981,
      "grad_norm": 0.6090975403785706,
      "learning_rate": 0.002828,
      "loss": 1.1082,
      "step": 495296
    },
    {
      "epoch": 37.17523452157599,
      "grad_norm": 0.6249683499336243,
      "learning_rate": 0.002828,
      "loss": 1.1113,
      "step": 495360
    },
    {
      "epoch": 37.18003752345216,
      "grad_norm": 0.6652005314826965,
      "learning_rate": 0.002828,
      "loss": 1.1086,
      "step": 495424
    },
    {
      "epoch": 37.18484052532833,
      "grad_norm": 0.73992520570755,
      "learning_rate": 0.002828,
      "loss": 1.1009,
      "step": 495488
    },
    {
      "epoch": 37.1896435272045,
      "grad_norm": 0.7139239311218262,
      "learning_rate": 0.002828,
      "loss": 1.1062,
      "step": 495552
    },
    {
      "epoch": 37.19444652908068,
      "grad_norm": 0.7656526565551758,
      "learning_rate": 0.002828,
      "loss": 1.1103,
      "step": 495616
    },
    {
      "epoch": 37.19924953095685,
      "grad_norm": 0.6346558928489685,
      "learning_rate": 0.002828,
      "loss": 1.1107,
      "step": 495680
    },
    {
      "epoch": 37.20405253283302,
      "grad_norm": 0.5977003574371338,
      "learning_rate": 0.002828,
      "loss": 1.1101,
      "step": 495744
    },
    {
      "epoch": 37.208855534709194,
      "grad_norm": 0.6296799182891846,
      "learning_rate": 0.002828,
      "loss": 1.107,
      "step": 495808
    },
    {
      "epoch": 37.21365853658536,
      "grad_norm": 0.6340168118476868,
      "learning_rate": 0.002828,
      "loss": 1.1033,
      "step": 495872
    },
    {
      "epoch": 37.21846153846154,
      "grad_norm": 0.7288328409194946,
      "learning_rate": 0.002828,
      "loss": 1.1083,
      "step": 495936
    },
    {
      "epoch": 37.22326454033771,
      "grad_norm": 0.6049259901046753,
      "learning_rate": 0.002828,
      "loss": 1.1039,
      "step": 496000
    },
    {
      "epoch": 37.228067542213886,
      "grad_norm": 0.592760443687439,
      "learning_rate": 0.002828,
      "loss": 1.1101,
      "step": 496064
    },
    {
      "epoch": 37.232870544090055,
      "grad_norm": 0.5616378784179688,
      "learning_rate": 0.002828,
      "loss": 1.1076,
      "step": 496128
    },
    {
      "epoch": 37.23767354596623,
      "grad_norm": 0.45006880164146423,
      "learning_rate": 0.002828,
      "loss": 1.1095,
      "step": 496192
    },
    {
      "epoch": 37.2424765478424,
      "grad_norm": 0.6839893460273743,
      "learning_rate": 0.002828,
      "loss": 1.1145,
      "step": 496256
    },
    {
      "epoch": 37.24727954971858,
      "grad_norm": 0.591373085975647,
      "learning_rate": 0.002828,
      "loss": 1.1079,
      "step": 496320
    },
    {
      "epoch": 37.25208255159475,
      "grad_norm": 0.7545192837715149,
      "learning_rate": 0.002828,
      "loss": 1.1057,
      "step": 496384
    },
    {
      "epoch": 37.25688555347092,
      "grad_norm": 0.5493077635765076,
      "learning_rate": 0.002828,
      "loss": 1.1125,
      "step": 496448
    },
    {
      "epoch": 37.26168855534709,
      "grad_norm": 0.8801015615463257,
      "learning_rate": 0.002828,
      "loss": 1.108,
      "step": 496512
    },
    {
      "epoch": 37.26649155722326,
      "grad_norm": 0.5462446808815002,
      "learning_rate": 0.002828,
      "loss": 1.1068,
      "step": 496576
    },
    {
      "epoch": 37.27129455909944,
      "grad_norm": 0.5896430611610413,
      "learning_rate": 0.002828,
      "loss": 1.1084,
      "step": 496640
    },
    {
      "epoch": 37.27609756097561,
      "grad_norm": 0.591148853302002,
      "learning_rate": 0.002828,
      "loss": 1.1141,
      "step": 496704
    },
    {
      "epoch": 37.280900562851784,
      "grad_norm": 0.6122753620147705,
      "learning_rate": 0.002828,
      "loss": 1.1117,
      "step": 496768
    },
    {
      "epoch": 37.28570356472795,
      "grad_norm": 0.7818361520767212,
      "learning_rate": 0.002828,
      "loss": 1.1154,
      "step": 496832
    },
    {
      "epoch": 37.29050656660413,
      "grad_norm": 0.5743238925933838,
      "learning_rate": 0.002828,
      "loss": 1.1104,
      "step": 496896
    },
    {
      "epoch": 37.2953095684803,
      "grad_norm": 0.5294697284698486,
      "learning_rate": 0.002828,
      "loss": 1.1169,
      "step": 496960
    },
    {
      "epoch": 37.300112570356475,
      "grad_norm": 0.7010504603385925,
      "learning_rate": 0.002828,
      "loss": 1.1158,
      "step": 497024
    },
    {
      "epoch": 37.304915572232645,
      "grad_norm": 0.6668693423271179,
      "learning_rate": 0.002828,
      "loss": 1.1098,
      "step": 497088
    },
    {
      "epoch": 37.30971857410882,
      "grad_norm": 0.6958891153335571,
      "learning_rate": 0.002828,
      "loss": 1.1093,
      "step": 497152
    },
    {
      "epoch": 37.31452157598499,
      "grad_norm": 0.5669229030609131,
      "learning_rate": 0.002828,
      "loss": 1.111,
      "step": 497216
    },
    {
      "epoch": 37.31932457786116,
      "grad_norm": 0.6313174962997437,
      "learning_rate": 0.002828,
      "loss": 1.1121,
      "step": 497280
    },
    {
      "epoch": 37.324127579737336,
      "grad_norm": 0.5094839930534363,
      "learning_rate": 0.002828,
      "loss": 1.1148,
      "step": 497344
    },
    {
      "epoch": 37.328930581613506,
      "grad_norm": 0.7712111473083496,
      "learning_rate": 0.002828,
      "loss": 1.1102,
      "step": 497408
    },
    {
      "epoch": 37.33373358348968,
      "grad_norm": 0.7262586355209351,
      "learning_rate": 0.002828,
      "loss": 1.1141,
      "step": 497472
    },
    {
      "epoch": 37.33853658536585,
      "grad_norm": 0.619162917137146,
      "learning_rate": 0.002828,
      "loss": 1.1117,
      "step": 497536
    },
    {
      "epoch": 37.34333958724203,
      "grad_norm": 0.634042501449585,
      "learning_rate": 0.002828,
      "loss": 1.1135,
      "step": 497600
    },
    {
      "epoch": 37.3481425891182,
      "grad_norm": 0.6961790919303894,
      "learning_rate": 0.002828,
      "loss": 1.1155,
      "step": 497664
    },
    {
      "epoch": 37.35294559099437,
      "grad_norm": 0.724677324295044,
      "learning_rate": 0.002828,
      "loss": 1.1122,
      "step": 497728
    },
    {
      "epoch": 37.35774859287054,
      "grad_norm": 0.6303979158401489,
      "learning_rate": 0.002828,
      "loss": 1.1105,
      "step": 497792
    },
    {
      "epoch": 37.36255159474672,
      "grad_norm": 0.5575435161590576,
      "learning_rate": 0.002828,
      "loss": 1.1127,
      "step": 497856
    },
    {
      "epoch": 37.36735459662289,
      "grad_norm": 0.6775777339935303,
      "learning_rate": 0.002828,
      "loss": 1.1157,
      "step": 497920
    },
    {
      "epoch": 37.372157598499065,
      "grad_norm": 0.4811544120311737,
      "learning_rate": 0.002828,
      "loss": 1.1143,
      "step": 497984
    },
    {
      "epoch": 37.376960600375234,
      "grad_norm": 0.6227823495864868,
      "learning_rate": 0.002828,
      "loss": 1.1153,
      "step": 498048
    },
    {
      "epoch": 37.381763602251404,
      "grad_norm": 0.6934810876846313,
      "learning_rate": 0.002828,
      "loss": 1.1181,
      "step": 498112
    },
    {
      "epoch": 37.38656660412758,
      "grad_norm": 0.600058913230896,
      "learning_rate": 0.002828,
      "loss": 1.1118,
      "step": 498176
    },
    {
      "epoch": 37.39136960600375,
      "grad_norm": 0.5795884132385254,
      "learning_rate": 0.002828,
      "loss": 1.1218,
      "step": 498240
    },
    {
      "epoch": 37.396172607879926,
      "grad_norm": 0.5940358638763428,
      "learning_rate": 0.002828,
      "loss": 1.1086,
      "step": 498304
    },
    {
      "epoch": 37.400975609756095,
      "grad_norm": 0.5886350870132446,
      "learning_rate": 0.002828,
      "loss": 1.1114,
      "step": 498368
    },
    {
      "epoch": 37.40577861163227,
      "grad_norm": 0.6366506814956665,
      "learning_rate": 0.002828,
      "loss": 1.1171,
      "step": 498432
    },
    {
      "epoch": 37.41058161350844,
      "grad_norm": 0.6446359753608704,
      "learning_rate": 0.002828,
      "loss": 1.1209,
      "step": 498496
    },
    {
      "epoch": 37.41538461538462,
      "grad_norm": 0.6479003429412842,
      "learning_rate": 0.002828,
      "loss": 1.1141,
      "step": 498560
    },
    {
      "epoch": 37.42018761726079,
      "grad_norm": 0.5944030284881592,
      "learning_rate": 0.002828,
      "loss": 1.1162,
      "step": 498624
    },
    {
      "epoch": 37.42499061913696,
      "grad_norm": 0.6356111168861389,
      "learning_rate": 0.002828,
      "loss": 1.1199,
      "step": 498688
    },
    {
      "epoch": 37.42979362101313,
      "grad_norm": 0.7332292795181274,
      "learning_rate": 0.002828,
      "loss": 1.1138,
      "step": 498752
    },
    {
      "epoch": 37.43459662288931,
      "grad_norm": 0.6613455414772034,
      "learning_rate": 0.002828,
      "loss": 1.1253,
      "step": 498816
    },
    {
      "epoch": 37.43939962476548,
      "grad_norm": 0.5517838001251221,
      "learning_rate": 0.002828,
      "loss": 1.1151,
      "step": 498880
    },
    {
      "epoch": 37.44420262664165,
      "grad_norm": 0.5878711342811584,
      "learning_rate": 0.002828,
      "loss": 1.1155,
      "step": 498944
    },
    {
      "epoch": 37.449005628517824,
      "grad_norm": 0.6955589652061462,
      "learning_rate": 0.002828,
      "loss": 1.1114,
      "step": 499008
    },
    {
      "epoch": 37.45380863039399,
      "grad_norm": 0.7319133281707764,
      "learning_rate": 0.002828,
      "loss": 1.1204,
      "step": 499072
    },
    {
      "epoch": 37.45861163227017,
      "grad_norm": 0.5313762426376343,
      "learning_rate": 0.002828,
      "loss": 1.1194,
      "step": 499136
    },
    {
      "epoch": 37.46341463414634,
      "grad_norm": 0.5796071290969849,
      "learning_rate": 0.002828,
      "loss": 1.121,
      "step": 499200
    },
    {
      "epoch": 37.468217636022516,
      "grad_norm": 0.632054328918457,
      "learning_rate": 0.002828,
      "loss": 1.1115,
      "step": 499264
    },
    {
      "epoch": 37.473020637898685,
      "grad_norm": 0.6562007069587708,
      "learning_rate": 0.002828,
      "loss": 1.1174,
      "step": 499328
    },
    {
      "epoch": 37.47782363977486,
      "grad_norm": 0.7263187170028687,
      "learning_rate": 0.002828,
      "loss": 1.124,
      "step": 499392
    },
    {
      "epoch": 37.48262664165103,
      "grad_norm": 0.7176118493080139,
      "learning_rate": 0.002828,
      "loss": 1.115,
      "step": 499456
    },
    {
      "epoch": 37.48742964352721,
      "grad_norm": 0.7256899476051331,
      "learning_rate": 0.002828,
      "loss": 1.1184,
      "step": 499520
    },
    {
      "epoch": 37.49223264540338,
      "grad_norm": 0.6485072374343872,
      "learning_rate": 0.002828,
      "loss": 1.119,
      "step": 499584
    },
    {
      "epoch": 37.49703564727955,
      "grad_norm": 0.589574933052063,
      "learning_rate": 0.002828,
      "loss": 1.1155,
      "step": 499648
    },
    {
      "epoch": 37.50183864915572,
      "grad_norm": 0.5531770586967468,
      "learning_rate": 0.002828,
      "loss": 1.1154,
      "step": 499712
    },
    {
      "epoch": 37.50664165103189,
      "grad_norm": 0.7608128786087036,
      "learning_rate": 0.002828,
      "loss": 1.1173,
      "step": 499776
    },
    {
      "epoch": 37.51144465290807,
      "grad_norm": 0.7428147196769714,
      "learning_rate": 0.002828,
      "loss": 1.1147,
      "step": 499840
    },
    {
      "epoch": 37.51624765478424,
      "grad_norm": 0.6598548889160156,
      "learning_rate": 0.002828,
      "loss": 1.1154,
      "step": 499904
    },
    {
      "epoch": 37.521050656660414,
      "grad_norm": 0.6032675504684448,
      "learning_rate": 0.002828,
      "loss": 1.1154,
      "step": 499968
    },
    {
      "epoch": 37.52585365853658,
      "grad_norm": 0.6908695101737976,
      "learning_rate": 0.002828,
      "loss": 1.1208,
      "step": 500032
    },
    {
      "epoch": 37.53065666041276,
      "grad_norm": 0.5637995600700378,
      "learning_rate": 0.002828,
      "loss": 1.1177,
      "step": 500096
    },
    {
      "epoch": 37.53545966228893,
      "grad_norm": 0.6846232414245605,
      "learning_rate": 0.002828,
      "loss": 1.1136,
      "step": 500160
    },
    {
      "epoch": 37.540262664165105,
      "grad_norm": 0.6805240511894226,
      "learning_rate": 0.002828,
      "loss": 1.1135,
      "step": 500224
    },
    {
      "epoch": 37.545065666041275,
      "grad_norm": 0.6433038711547852,
      "learning_rate": 0.002828,
      "loss": 1.1211,
      "step": 500288
    },
    {
      "epoch": 37.54986866791745,
      "grad_norm": 0.6407705545425415,
      "learning_rate": 0.002828,
      "loss": 1.1197,
      "step": 500352
    },
    {
      "epoch": 37.55467166979362,
      "grad_norm": 0.5954273343086243,
      "learning_rate": 0.002828,
      "loss": 1.1181,
      "step": 500416
    },
    {
      "epoch": 37.5594746716698,
      "grad_norm": 0.5801562666893005,
      "learning_rate": 0.002828,
      "loss": 1.1257,
      "step": 500480
    },
    {
      "epoch": 37.564277673545966,
      "grad_norm": 0.7970672249794006,
      "learning_rate": 0.002828,
      "loss": 1.1191,
      "step": 500544
    },
    {
      "epoch": 37.569080675422136,
      "grad_norm": 0.5519067049026489,
      "learning_rate": 0.002828,
      "loss": 1.1225,
      "step": 500608
    },
    {
      "epoch": 37.57388367729831,
      "grad_norm": 0.623494029045105,
      "learning_rate": 0.002828,
      "loss": 1.1178,
      "step": 500672
    },
    {
      "epoch": 37.57868667917448,
      "grad_norm": 0.5677382349967957,
      "learning_rate": 0.002828,
      "loss": 1.1205,
      "step": 500736
    },
    {
      "epoch": 37.58348968105066,
      "grad_norm": 0.6610050797462463,
      "learning_rate": 0.002828,
      "loss": 1.119,
      "step": 500800
    },
    {
      "epoch": 37.58829268292683,
      "grad_norm": 0.6180018782615662,
      "learning_rate": 0.002828,
      "loss": 1.1178,
      "step": 500864
    },
    {
      "epoch": 37.593095684803004,
      "grad_norm": 0.6084687113761902,
      "learning_rate": 0.002828,
      "loss": 1.1167,
      "step": 500928
    },
    {
      "epoch": 37.59789868667917,
      "grad_norm": 0.6760867834091187,
      "learning_rate": 0.002828,
      "loss": 1.1169,
      "step": 500992
    },
    {
      "epoch": 37.60270168855535,
      "grad_norm": 0.7423652410507202,
      "learning_rate": 0.002828,
      "loss": 1.1234,
      "step": 501056
    },
    {
      "epoch": 37.60750469043152,
      "grad_norm": 0.6940181255340576,
      "learning_rate": 0.002828,
      "loss": 1.1214,
      "step": 501120
    },
    {
      "epoch": 37.612307692307695,
      "grad_norm": 0.5058683156967163,
      "learning_rate": 0.002828,
      "loss": 1.1197,
      "step": 501184
    },
    {
      "epoch": 37.617110694183864,
      "grad_norm": 0.6302307844161987,
      "learning_rate": 0.002828,
      "loss": 1.1183,
      "step": 501248
    },
    {
      "epoch": 37.62191369606004,
      "grad_norm": 0.6181077361106873,
      "learning_rate": 0.002828,
      "loss": 1.1226,
      "step": 501312
    },
    {
      "epoch": 37.62671669793621,
      "grad_norm": 0.527346134185791,
      "learning_rate": 0.002828,
      "loss": 1.1219,
      "step": 501376
    },
    {
      "epoch": 37.63151969981238,
      "grad_norm": 0.5519901514053345,
      "learning_rate": 0.002828,
      "loss": 1.1215,
      "step": 501440
    },
    {
      "epoch": 37.636322701688556,
      "grad_norm": 0.6316769123077393,
      "learning_rate": 0.002828,
      "loss": 1.122,
      "step": 501504
    },
    {
      "epoch": 37.641125703564725,
      "grad_norm": 0.6549479961395264,
      "learning_rate": 0.002828,
      "loss": 1.1235,
      "step": 501568
    },
    {
      "epoch": 37.6459287054409,
      "grad_norm": 0.579494297504425,
      "learning_rate": 0.002828,
      "loss": 1.1224,
      "step": 501632
    },
    {
      "epoch": 37.65073170731707,
      "grad_norm": 0.6767232418060303,
      "learning_rate": 0.002828,
      "loss": 1.1196,
      "step": 501696
    },
    {
      "epoch": 37.65553470919325,
      "grad_norm": 0.5806661248207092,
      "learning_rate": 0.002828,
      "loss": 1.1173,
      "step": 501760
    },
    {
      "epoch": 37.66033771106942,
      "grad_norm": 0.6679486036300659,
      "learning_rate": 0.002828,
      "loss": 1.1177,
      "step": 501824
    },
    {
      "epoch": 37.66514071294559,
      "grad_norm": 0.7422885298728943,
      "learning_rate": 0.002828,
      "loss": 1.1167,
      "step": 501888
    },
    {
      "epoch": 37.66994371482176,
      "grad_norm": 0.663256824016571,
      "learning_rate": 0.002828,
      "loss": 1.1221,
      "step": 501952
    },
    {
      "epoch": 37.67474671669794,
      "grad_norm": 0.5884062647819519,
      "learning_rate": 0.002828,
      "loss": 1.1179,
      "step": 502016
    },
    {
      "epoch": 37.67954971857411,
      "grad_norm": 0.5877790451049805,
      "learning_rate": 0.002828,
      "loss": 1.116,
      "step": 502080
    },
    {
      "epoch": 37.684352720450285,
      "grad_norm": 0.5684241652488708,
      "learning_rate": 0.002828,
      "loss": 1.1154,
      "step": 502144
    },
    {
      "epoch": 37.689155722326454,
      "grad_norm": 0.46695685386657715,
      "learning_rate": 0.002828,
      "loss": 1.121,
      "step": 502208
    },
    {
      "epoch": 37.69395872420262,
      "grad_norm": 0.6337296962738037,
      "learning_rate": 0.002828,
      "loss": 1.122,
      "step": 502272
    },
    {
      "epoch": 37.6987617260788,
      "grad_norm": 0.6555072665214539,
      "learning_rate": 0.002828,
      "loss": 1.1239,
      "step": 502336
    },
    {
      "epoch": 37.70356472795497,
      "grad_norm": 0.6246955394744873,
      "learning_rate": 0.002828,
      "loss": 1.1192,
      "step": 502400
    },
    {
      "epoch": 37.708367729831146,
      "grad_norm": 0.6092773079872131,
      "learning_rate": 0.002828,
      "loss": 1.125,
      "step": 502464
    },
    {
      "epoch": 37.713170731707315,
      "grad_norm": 0.5604978203773499,
      "learning_rate": 0.002828,
      "loss": 1.1228,
      "step": 502528
    },
    {
      "epoch": 37.71797373358349,
      "grad_norm": 0.6719847321510315,
      "learning_rate": 0.002828,
      "loss": 1.1253,
      "step": 502592
    },
    {
      "epoch": 37.72277673545966,
      "grad_norm": 0.5437619686126709,
      "learning_rate": 0.002828,
      "loss": 1.1236,
      "step": 502656
    },
    {
      "epoch": 37.72757973733584,
      "grad_norm": 0.6136580109596252,
      "learning_rate": 0.002828,
      "loss": 1.1218,
      "step": 502720
    },
    {
      "epoch": 37.73238273921201,
      "grad_norm": 0.5170011520385742,
      "learning_rate": 0.002828,
      "loss": 1.1196,
      "step": 502784
    },
    {
      "epoch": 37.73718574108818,
      "grad_norm": 0.619965136051178,
      "learning_rate": 0.002828,
      "loss": 1.1244,
      "step": 502848
    },
    {
      "epoch": 37.74198874296435,
      "grad_norm": 0.6624310612678528,
      "learning_rate": 0.002828,
      "loss": 1.1222,
      "step": 502912
    },
    {
      "epoch": 37.74679174484053,
      "grad_norm": 0.651011049747467,
      "learning_rate": 0.002828,
      "loss": 1.1227,
      "step": 502976
    },
    {
      "epoch": 37.7515947467167,
      "grad_norm": 0.7418103218078613,
      "learning_rate": 0.002828,
      "loss": 1.1168,
      "step": 503040
    },
    {
      "epoch": 37.75639774859287,
      "grad_norm": 0.5328953266143799,
      "learning_rate": 0.002828,
      "loss": 1.125,
      "step": 503104
    },
    {
      "epoch": 37.761200750469044,
      "grad_norm": 0.5455359816551208,
      "learning_rate": 0.002828,
      "loss": 1.1166,
      "step": 503168
    },
    {
      "epoch": 37.76600375234521,
      "grad_norm": 0.5440381765365601,
      "learning_rate": 0.002828,
      "loss": 1.1239,
      "step": 503232
    },
    {
      "epoch": 37.77080675422139,
      "grad_norm": 0.7047117948532104,
      "learning_rate": 0.002828,
      "loss": 1.1265,
      "step": 503296
    },
    {
      "epoch": 37.77560975609756,
      "grad_norm": 0.6114978790283203,
      "learning_rate": 0.002828,
      "loss": 1.126,
      "step": 503360
    },
    {
      "epoch": 37.780412757973735,
      "grad_norm": 0.5973011255264282,
      "learning_rate": 0.002828,
      "loss": 1.1196,
      "step": 503424
    },
    {
      "epoch": 37.785215759849905,
      "grad_norm": 0.7089464664459229,
      "learning_rate": 0.002828,
      "loss": 1.1223,
      "step": 503488
    },
    {
      "epoch": 37.79001876172608,
      "grad_norm": 0.5899783372879028,
      "learning_rate": 0.002828,
      "loss": 1.1222,
      "step": 503552
    },
    {
      "epoch": 37.79482176360225,
      "grad_norm": 0.5558510422706604,
      "learning_rate": 0.002828,
      "loss": 1.1183,
      "step": 503616
    },
    {
      "epoch": 37.79962476547843,
      "grad_norm": 0.5553831458091736,
      "learning_rate": 0.002828,
      "loss": 1.1214,
      "step": 503680
    },
    {
      "epoch": 37.804427767354596,
      "grad_norm": 0.58292555809021,
      "learning_rate": 0.002828,
      "loss": 1.1272,
      "step": 503744
    },
    {
      "epoch": 37.809230769230766,
      "grad_norm": 0.6653010249137878,
      "learning_rate": 0.002828,
      "loss": 1.1235,
      "step": 503808
    },
    {
      "epoch": 37.81403377110694,
      "grad_norm": 0.6157090067863464,
      "learning_rate": 0.002828,
      "loss": 1.1226,
      "step": 503872
    },
    {
      "epoch": 37.81883677298311,
      "grad_norm": 0.7392660975456238,
      "learning_rate": 0.002828,
      "loss": 1.1201,
      "step": 503936
    },
    {
      "epoch": 37.82363977485929,
      "grad_norm": 0.5767449140548706,
      "learning_rate": 0.002828,
      "loss": 1.123,
      "step": 504000
    },
    {
      "epoch": 37.82844277673546,
      "grad_norm": 0.7190343141555786,
      "learning_rate": 0.002828,
      "loss": 1.1232,
      "step": 504064
    },
    {
      "epoch": 37.833245778611634,
      "grad_norm": 0.5598681569099426,
      "learning_rate": 0.002828,
      "loss": 1.1228,
      "step": 504128
    },
    {
      "epoch": 37.8380487804878,
      "grad_norm": 0.5747976899147034,
      "learning_rate": 0.002828,
      "loss": 1.1229,
      "step": 504192
    },
    {
      "epoch": 37.84285178236398,
      "grad_norm": 0.542398989200592,
      "learning_rate": 0.002828,
      "loss": 1.1238,
      "step": 504256
    },
    {
      "epoch": 37.84765478424015,
      "grad_norm": 0.5907443165779114,
      "learning_rate": 0.002828,
      "loss": 1.1252,
      "step": 504320
    },
    {
      "epoch": 37.852457786116325,
      "grad_norm": 0.5543791055679321,
      "learning_rate": 0.002828,
      "loss": 1.1284,
      "step": 504384
    },
    {
      "epoch": 37.857260787992494,
      "grad_norm": 0.6768468618392944,
      "learning_rate": 0.002828,
      "loss": 1.1181,
      "step": 504448
    },
    {
      "epoch": 37.86206378986867,
      "grad_norm": 0.7407353520393372,
      "learning_rate": 0.002828,
      "loss": 1.1261,
      "step": 504512
    },
    {
      "epoch": 37.86686679174484,
      "grad_norm": 0.5630684494972229,
      "learning_rate": 0.002828,
      "loss": 1.1292,
      "step": 504576
    },
    {
      "epoch": 37.87166979362101,
      "grad_norm": 0.5335458517074585,
      "learning_rate": 0.002828,
      "loss": 1.1274,
      "step": 504640
    },
    {
      "epoch": 37.876472795497186,
      "grad_norm": 0.6631928086280823,
      "learning_rate": 0.002828,
      "loss": 1.1204,
      "step": 504704
    },
    {
      "epoch": 37.881275797373355,
      "grad_norm": 0.7281734943389893,
      "learning_rate": 0.002828,
      "loss": 1.122,
      "step": 504768
    },
    {
      "epoch": 37.88607879924953,
      "grad_norm": 0.5659152269363403,
      "learning_rate": 0.002828,
      "loss": 1.1308,
      "step": 504832
    },
    {
      "epoch": 37.8908818011257,
      "grad_norm": 0.4987253248691559,
      "learning_rate": 0.002828,
      "loss": 1.1241,
      "step": 504896
    },
    {
      "epoch": 37.89568480300188,
      "grad_norm": 0.6153334379196167,
      "learning_rate": 0.002828,
      "loss": 1.124,
      "step": 504960
    },
    {
      "epoch": 37.90048780487805,
      "grad_norm": 0.5674315094947815,
      "learning_rate": 0.002828,
      "loss": 1.1288,
      "step": 505024
    },
    {
      "epoch": 37.90529080675422,
      "grad_norm": 0.5308176875114441,
      "learning_rate": 0.002828,
      "loss": 1.131,
      "step": 505088
    },
    {
      "epoch": 37.91009380863039,
      "grad_norm": 0.6265832781791687,
      "learning_rate": 0.002828,
      "loss": 1.1251,
      "step": 505152
    },
    {
      "epoch": 37.91489681050657,
      "grad_norm": 0.6296497583389282,
      "learning_rate": 0.002828,
      "loss": 1.1206,
      "step": 505216
    },
    {
      "epoch": 37.91969981238274,
      "grad_norm": 0.6319650411605835,
      "learning_rate": 0.002828,
      "loss": 1.1263,
      "step": 505280
    },
    {
      "epoch": 37.924502814258915,
      "grad_norm": 0.6511676907539368,
      "learning_rate": 0.002828,
      "loss": 1.1226,
      "step": 505344
    },
    {
      "epoch": 37.929305816135084,
      "grad_norm": 0.6702365279197693,
      "learning_rate": 0.002828,
      "loss": 1.1238,
      "step": 505408
    },
    {
      "epoch": 37.934108818011254,
      "grad_norm": 0.558114767074585,
      "learning_rate": 0.002828,
      "loss": 1.1272,
      "step": 505472
    },
    {
      "epoch": 37.93891181988743,
      "grad_norm": 0.6984198689460754,
      "learning_rate": 0.002828,
      "loss": 1.123,
      "step": 505536
    },
    {
      "epoch": 37.9437148217636,
      "grad_norm": 0.6288992166519165,
      "learning_rate": 0.002828,
      "loss": 1.125,
      "step": 505600
    },
    {
      "epoch": 37.948517823639776,
      "grad_norm": 0.5294339060783386,
      "learning_rate": 0.002828,
      "loss": 1.1236,
      "step": 505664
    },
    {
      "epoch": 37.953320825515945,
      "grad_norm": 0.5702909827232361,
      "learning_rate": 0.002828,
      "loss": 1.1221,
      "step": 505728
    },
    {
      "epoch": 37.95812382739212,
      "grad_norm": 0.6646183133125305,
      "learning_rate": 0.002828,
      "loss": 1.1326,
      "step": 505792
    },
    {
      "epoch": 37.96292682926829,
      "grad_norm": 0.5951704978942871,
      "learning_rate": 0.002828,
      "loss": 1.1261,
      "step": 505856
    },
    {
      "epoch": 37.96772983114447,
      "grad_norm": 0.8041133284568787,
      "learning_rate": 0.002828,
      "loss": 1.1215,
      "step": 505920
    },
    {
      "epoch": 37.97253283302064,
      "grad_norm": 0.7284034490585327,
      "learning_rate": 0.002828,
      "loss": 1.1254,
      "step": 505984
    },
    {
      "epoch": 37.97733583489681,
      "grad_norm": 0.6413600444793701,
      "learning_rate": 0.002828,
      "loss": 1.1193,
      "step": 506048
    },
    {
      "epoch": 37.98213883677298,
      "grad_norm": 0.6052785515785217,
      "learning_rate": 0.002828,
      "loss": 1.1267,
      "step": 506112
    },
    {
      "epoch": 37.98694183864916,
      "grad_norm": 0.7194735407829285,
      "learning_rate": 0.002828,
      "loss": 1.1246,
      "step": 506176
    },
    {
      "epoch": 37.99174484052533,
      "grad_norm": 0.5225487351417542,
      "learning_rate": 0.002828,
      "loss": 1.1266,
      "step": 506240
    },
    {
      "epoch": 37.9965478424015,
      "grad_norm": 0.6789017915725708,
      "learning_rate": 0.002828,
      "loss": 1.1258,
      "step": 506304
    },
    {
      "epoch": 38.001350844277674,
      "grad_norm": 0.6713529229164124,
      "learning_rate": 0.002828,
      "loss": 1.1187,
      "step": 506368
    },
    {
      "epoch": 38.00615384615384,
      "grad_norm": 0.594455897808075,
      "learning_rate": 0.002828,
      "loss": 1.0846,
      "step": 506432
    },
    {
      "epoch": 38.01095684803002,
      "grad_norm": 0.5679947137832642,
      "learning_rate": 0.002828,
      "loss": 1.0902,
      "step": 506496
    },
    {
      "epoch": 38.01575984990619,
      "grad_norm": 0.6373549103736877,
      "learning_rate": 0.002828,
      "loss": 1.0958,
      "step": 506560
    },
    {
      "epoch": 38.020562851782365,
      "grad_norm": 0.603205144405365,
      "learning_rate": 0.002828,
      "loss": 1.0907,
      "step": 506624
    },
    {
      "epoch": 38.025365853658535,
      "grad_norm": 0.5866041779518127,
      "learning_rate": 0.002828,
      "loss": 1.0973,
      "step": 506688
    },
    {
      "epoch": 38.03016885553471,
      "grad_norm": 0.7726960182189941,
      "learning_rate": 0.002828,
      "loss": 1.0933,
      "step": 506752
    },
    {
      "epoch": 38.03497185741088,
      "grad_norm": 0.6744563579559326,
      "learning_rate": 0.002828,
      "loss": 1.0959,
      "step": 506816
    },
    {
      "epoch": 38.03977485928706,
      "grad_norm": 0.6458048224449158,
      "learning_rate": 0.002828,
      "loss": 1.0906,
      "step": 506880
    },
    {
      "epoch": 38.044577861163226,
      "grad_norm": 0.6350529193878174,
      "learning_rate": 0.002828,
      "loss": 1.0889,
      "step": 506944
    },
    {
      "epoch": 38.0493808630394,
      "grad_norm": 0.5482154488563538,
      "learning_rate": 0.002828,
      "loss": 1.0905,
      "step": 507008
    },
    {
      "epoch": 38.05418386491557,
      "grad_norm": 0.5511150360107422,
      "learning_rate": 0.002828,
      "loss": 1.0931,
      "step": 507072
    },
    {
      "epoch": 38.05898686679174,
      "grad_norm": 0.5821910500526428,
      "learning_rate": 0.002828,
      "loss": 1.0922,
      "step": 507136
    },
    {
      "epoch": 38.06378986866792,
      "grad_norm": 0.5219399929046631,
      "learning_rate": 0.002828,
      "loss": 1.0928,
      "step": 507200
    },
    {
      "epoch": 38.06859287054409,
      "grad_norm": 0.7586058378219604,
      "learning_rate": 0.002828,
      "loss": 1.0944,
      "step": 507264
    },
    {
      "epoch": 38.073395872420264,
      "grad_norm": 0.6015982627868652,
      "learning_rate": 0.002828,
      "loss": 1.0933,
      "step": 507328
    },
    {
      "epoch": 38.07819887429643,
      "grad_norm": 0.598664402961731,
      "learning_rate": 0.002828,
      "loss": 1.0922,
      "step": 507392
    },
    {
      "epoch": 38.08300187617261,
      "grad_norm": 0.5428532361984253,
      "learning_rate": 0.002828,
      "loss": 1.091,
      "step": 507456
    },
    {
      "epoch": 38.08780487804878,
      "grad_norm": 0.6144164204597473,
      "learning_rate": 0.002828,
      "loss": 1.0968,
      "step": 507520
    },
    {
      "epoch": 38.092607879924955,
      "grad_norm": 0.5404438376426697,
      "learning_rate": 0.002828,
      "loss": 1.0952,
      "step": 507584
    },
    {
      "epoch": 38.097410881801125,
      "grad_norm": 0.6267830729484558,
      "learning_rate": 0.002828,
      "loss": 1.0978,
      "step": 507648
    },
    {
      "epoch": 38.1022138836773,
      "grad_norm": 0.6026227474212646,
      "learning_rate": 0.002828,
      "loss": 1.0913,
      "step": 507712
    },
    {
      "epoch": 38.10701688555347,
      "grad_norm": 0.6021787524223328,
      "learning_rate": 0.002828,
      "loss": 1.0947,
      "step": 507776
    },
    {
      "epoch": 38.11181988742965,
      "grad_norm": 0.7789412140846252,
      "learning_rate": 0.002828,
      "loss": 1.0922,
      "step": 507840
    },
    {
      "epoch": 38.116622889305816,
      "grad_norm": 0.663597583770752,
      "learning_rate": 0.002828,
      "loss": 1.0943,
      "step": 507904
    },
    {
      "epoch": 38.121425891181985,
      "grad_norm": 0.6256552338600159,
      "learning_rate": 0.002828,
      "loss": 1.0924,
      "step": 507968
    },
    {
      "epoch": 38.12622889305816,
      "grad_norm": 0.6621292233467102,
      "learning_rate": 0.002828,
      "loss": 1.0954,
      "step": 508032
    },
    {
      "epoch": 38.13103189493433,
      "grad_norm": 0.6057916283607483,
      "learning_rate": 0.002828,
      "loss": 1.0977,
      "step": 508096
    },
    {
      "epoch": 38.13583489681051,
      "grad_norm": 0.606999397277832,
      "learning_rate": 0.002828,
      "loss": 1.0945,
      "step": 508160
    },
    {
      "epoch": 38.14063789868668,
      "grad_norm": 0.5733497142791748,
      "learning_rate": 0.002828,
      "loss": 1.0998,
      "step": 508224
    },
    {
      "epoch": 38.14544090056285,
      "grad_norm": 0.5235196352005005,
      "learning_rate": 0.002828,
      "loss": 1.1009,
      "step": 508288
    },
    {
      "epoch": 38.15024390243902,
      "grad_norm": 0.688124418258667,
      "learning_rate": 0.002828,
      "loss": 1.0945,
      "step": 508352
    },
    {
      "epoch": 38.1550469043152,
      "grad_norm": 0.5677928328514099,
      "learning_rate": 0.002828,
      "loss": 1.0985,
      "step": 508416
    },
    {
      "epoch": 38.15984990619137,
      "grad_norm": 0.7261043190956116,
      "learning_rate": 0.002828,
      "loss": 1.1007,
      "step": 508480
    },
    {
      "epoch": 38.164652908067545,
      "grad_norm": 0.6233505606651306,
      "learning_rate": 0.002828,
      "loss": 1.0984,
      "step": 508544
    },
    {
      "epoch": 38.169455909943714,
      "grad_norm": 0.5829945206642151,
      "learning_rate": 0.002828,
      "loss": 1.0973,
      "step": 508608
    },
    {
      "epoch": 38.17425891181989,
      "grad_norm": 0.6414790749549866,
      "learning_rate": 0.002828,
      "loss": 1.0946,
      "step": 508672
    },
    {
      "epoch": 38.17906191369606,
      "grad_norm": 0.5896680951118469,
      "learning_rate": 0.002828,
      "loss": 1.0977,
      "step": 508736
    },
    {
      "epoch": 38.18386491557223,
      "grad_norm": 0.8097200989723206,
      "learning_rate": 0.002828,
      "loss": 1.1029,
      "step": 508800
    },
    {
      "epoch": 38.188667917448406,
      "grad_norm": 0.5933405160903931,
      "learning_rate": 0.002828,
      "loss": 1.1028,
      "step": 508864
    },
    {
      "epoch": 38.193470919324575,
      "grad_norm": 0.5823057889938354,
      "learning_rate": 0.002828,
      "loss": 1.0985,
      "step": 508928
    },
    {
      "epoch": 38.19827392120075,
      "grad_norm": 0.6513922810554504,
      "learning_rate": 0.002828,
      "loss": 1.1007,
      "step": 508992
    },
    {
      "epoch": 38.20307692307692,
      "grad_norm": 0.5925971865653992,
      "learning_rate": 0.002828,
      "loss": 1.1033,
      "step": 509056
    },
    {
      "epoch": 38.2078799249531,
      "grad_norm": 0.5212792158126831,
      "learning_rate": 0.002828,
      "loss": 1.0937,
      "step": 509120
    },
    {
      "epoch": 38.21268292682927,
      "grad_norm": 0.6763430833816528,
      "learning_rate": 0.002828,
      "loss": 1.1048,
      "step": 509184
    },
    {
      "epoch": 38.21748592870544,
      "grad_norm": 0.6594332456588745,
      "learning_rate": 0.002828,
      "loss": 1.0959,
      "step": 509248
    },
    {
      "epoch": 38.22228893058161,
      "grad_norm": 0.4742533266544342,
      "learning_rate": 0.002828,
      "loss": 1.0996,
      "step": 509312
    },
    {
      "epoch": 38.22709193245779,
      "grad_norm": 0.6435297131538391,
      "learning_rate": 0.002828,
      "loss": 1.1035,
      "step": 509376
    },
    {
      "epoch": 38.23189493433396,
      "grad_norm": 0.6899946928024292,
      "learning_rate": 0.002828,
      "loss": 1.1011,
      "step": 509440
    },
    {
      "epoch": 38.236697936210135,
      "grad_norm": 0.8009315729141235,
      "learning_rate": 0.002828,
      "loss": 1.0996,
      "step": 509504
    },
    {
      "epoch": 38.241500938086304,
      "grad_norm": 0.7253291606903076,
      "learning_rate": 0.002828,
      "loss": 1.1016,
      "step": 509568
    },
    {
      "epoch": 38.24630393996247,
      "grad_norm": 0.7756628394126892,
      "learning_rate": 0.002828,
      "loss": 1.105,
      "step": 509632
    },
    {
      "epoch": 38.25110694183865,
      "grad_norm": 0.6615393757820129,
      "learning_rate": 0.002828,
      "loss": 1.0979,
      "step": 509696
    },
    {
      "epoch": 38.25590994371482,
      "grad_norm": 0.5859424471855164,
      "learning_rate": 0.002828,
      "loss": 1.0973,
      "step": 509760
    },
    {
      "epoch": 38.260712945590996,
      "grad_norm": 0.5020551681518555,
      "learning_rate": 0.002828,
      "loss": 1.1073,
      "step": 509824
    },
    {
      "epoch": 38.265515947467165,
      "grad_norm": 0.6972935795783997,
      "learning_rate": 0.002828,
      "loss": 1.1034,
      "step": 509888
    },
    {
      "epoch": 38.27031894934334,
      "grad_norm": 0.5921815633773804,
      "learning_rate": 0.002828,
      "loss": 1.1032,
      "step": 509952
    },
    {
      "epoch": 38.27512195121951,
      "grad_norm": 0.6624758243560791,
      "learning_rate": 0.002828,
      "loss": 1.1025,
      "step": 510016
    },
    {
      "epoch": 38.27992495309569,
      "grad_norm": 0.6163641810417175,
      "learning_rate": 0.002828,
      "loss": 1.1061,
      "step": 510080
    },
    {
      "epoch": 38.284727954971856,
      "grad_norm": 0.5077274441719055,
      "learning_rate": 0.002828,
      "loss": 1.1049,
      "step": 510144
    },
    {
      "epoch": 38.28953095684803,
      "grad_norm": 0.6114723682403564,
      "learning_rate": 0.002828,
      "loss": 1.1052,
      "step": 510208
    },
    {
      "epoch": 38.2943339587242,
      "grad_norm": 0.5323904156684875,
      "learning_rate": 0.002828,
      "loss": 1.1044,
      "step": 510272
    },
    {
      "epoch": 38.29913696060038,
      "grad_norm": 0.526978075504303,
      "learning_rate": 0.002828,
      "loss": 1.1013,
      "step": 510336
    },
    {
      "epoch": 38.30393996247655,
      "grad_norm": 0.6368070244789124,
      "learning_rate": 0.002828,
      "loss": 1.1009,
      "step": 510400
    },
    {
      "epoch": 38.30874296435272,
      "grad_norm": 0.5327396988868713,
      "learning_rate": 0.002828,
      "loss": 1.1024,
      "step": 510464
    },
    {
      "epoch": 38.313545966228894,
      "grad_norm": 0.7854524850845337,
      "learning_rate": 0.002828,
      "loss": 1.1085,
      "step": 510528
    },
    {
      "epoch": 38.31834896810506,
      "grad_norm": 0.5681763291358948,
      "learning_rate": 0.002828,
      "loss": 1.1038,
      "step": 510592
    },
    {
      "epoch": 38.32315196998124,
      "grad_norm": 0.5100849866867065,
      "learning_rate": 0.002828,
      "loss": 1.104,
      "step": 510656
    },
    {
      "epoch": 38.32795497185741,
      "grad_norm": 0.5657297968864441,
      "learning_rate": 0.002828,
      "loss": 1.1042,
      "step": 510720
    },
    {
      "epoch": 38.332757973733585,
      "grad_norm": 0.6012672781944275,
      "learning_rate": 0.002828,
      "loss": 1.1009,
      "step": 510784
    },
    {
      "epoch": 38.337560975609755,
      "grad_norm": 0.5699858665466309,
      "learning_rate": 0.002828,
      "loss": 1.11,
      "step": 510848
    },
    {
      "epoch": 38.34236397748593,
      "grad_norm": 0.6629965305328369,
      "learning_rate": 0.002828,
      "loss": 1.1089,
      "step": 510912
    },
    {
      "epoch": 38.3471669793621,
      "grad_norm": 0.5883797407150269,
      "learning_rate": 0.002828,
      "loss": 1.1072,
      "step": 510976
    },
    {
      "epoch": 38.35196998123828,
      "grad_norm": 0.5491295456886292,
      "learning_rate": 0.002828,
      "loss": 1.1028,
      "step": 511040
    },
    {
      "epoch": 38.356772983114446,
      "grad_norm": 0.5589887499809265,
      "learning_rate": 0.002828,
      "loss": 1.1025,
      "step": 511104
    },
    {
      "epoch": 38.36157598499062,
      "grad_norm": 0.7199347019195557,
      "learning_rate": 0.002828,
      "loss": 1.1031,
      "step": 511168
    },
    {
      "epoch": 38.36637898686679,
      "grad_norm": 0.5486679673194885,
      "learning_rate": 0.002828,
      "loss": 1.1099,
      "step": 511232
    },
    {
      "epoch": 38.37118198874296,
      "grad_norm": 0.6592404246330261,
      "learning_rate": 0.002828,
      "loss": 1.1115,
      "step": 511296
    },
    {
      "epoch": 38.37598499061914,
      "grad_norm": 0.7273109555244446,
      "learning_rate": 0.002828,
      "loss": 1.1117,
      "step": 511360
    },
    {
      "epoch": 38.38078799249531,
      "grad_norm": 0.6115809679031372,
      "learning_rate": 0.002828,
      "loss": 1.1043,
      "step": 511424
    },
    {
      "epoch": 38.38559099437148,
      "grad_norm": 0.688369870185852,
      "learning_rate": 0.002828,
      "loss": 1.1059,
      "step": 511488
    },
    {
      "epoch": 38.39039399624765,
      "grad_norm": 0.6574243307113647,
      "learning_rate": 0.002828,
      "loss": 1.1098,
      "step": 511552
    },
    {
      "epoch": 38.39519699812383,
      "grad_norm": 0.6283542513847351,
      "learning_rate": 0.002828,
      "loss": 1.1073,
      "step": 511616
    },
    {
      "epoch": 38.4,
      "grad_norm": 0.6580241918563843,
      "learning_rate": 0.002828,
      "loss": 1.1053,
      "step": 511680
    },
    {
      "epoch": 38.404803001876175,
      "grad_norm": 0.6294876337051392,
      "learning_rate": 0.002828,
      "loss": 1.078,
      "step": 511744
    },
    {
      "epoch": 38.409606003752344,
      "grad_norm": 0.6697729825973511,
      "learning_rate": 0.002828,
      "loss": 1.0859,
      "step": 511808
    },
    {
      "epoch": 38.41440900562852,
      "grad_norm": 0.587795615196228,
      "learning_rate": 0.002828,
      "loss": 1.081,
      "step": 511872
    },
    {
      "epoch": 38.41921200750469,
      "grad_norm": 0.5539612174034119,
      "learning_rate": 0.002828,
      "loss": 1.0887,
      "step": 511936
    },
    {
      "epoch": 38.42401500938087,
      "grad_norm": 0.5980706810951233,
      "learning_rate": 0.002828,
      "loss": 1.0868,
      "step": 512000
    },
    {
      "epoch": 38.428818011257036,
      "grad_norm": 0.5685859322547913,
      "learning_rate": 0.002828,
      "loss": 1.0872,
      "step": 512064
    },
    {
      "epoch": 38.433621013133205,
      "grad_norm": 0.8710797429084778,
      "learning_rate": 0.002828,
      "loss": 1.0896,
      "step": 512128
    },
    {
      "epoch": 38.43842401500938,
      "grad_norm": 0.5642542243003845,
      "learning_rate": 0.002828,
      "loss": 1.0899,
      "step": 512192
    },
    {
      "epoch": 38.44322701688555,
      "grad_norm": 0.7766580581665039,
      "learning_rate": 0.002828,
      "loss": 1.0905,
      "step": 512256
    },
    {
      "epoch": 38.44803001876173,
      "grad_norm": 0.5817084908485413,
      "learning_rate": 0.002828,
      "loss": 1.0894,
      "step": 512320
    },
    {
      "epoch": 38.4528330206379,
      "grad_norm": 0.6450785994529724,
      "learning_rate": 0.002828,
      "loss": 1.0845,
      "step": 512384
    },
    {
      "epoch": 38.45763602251407,
      "grad_norm": 0.6442525386810303,
      "learning_rate": 0.002828,
      "loss": 1.0891,
      "step": 512448
    },
    {
      "epoch": 38.46243902439024,
      "grad_norm": 0.7165563106536865,
      "learning_rate": 0.002828,
      "loss": 1.0873,
      "step": 512512
    },
    {
      "epoch": 38.46724202626642,
      "grad_norm": 0.6603183150291443,
      "learning_rate": 0.002828,
      "loss": 1.0875,
      "step": 512576
    },
    {
      "epoch": 38.47204502814259,
      "grad_norm": 0.5756245255470276,
      "learning_rate": 0.002828,
      "loss": 1.0897,
      "step": 512640
    },
    {
      "epoch": 38.476848030018765,
      "grad_norm": 0.5640691518783569,
      "learning_rate": 0.002828,
      "loss": 1.0868,
      "step": 512704
    },
    {
      "epoch": 38.481651031894934,
      "grad_norm": 0.5975730419158936,
      "learning_rate": 0.002828,
      "loss": 1.0905,
      "step": 512768
    },
    {
      "epoch": 38.48645403377111,
      "grad_norm": 0.7235714197158813,
      "learning_rate": 0.002828,
      "loss": 1.0891,
      "step": 512832
    },
    {
      "epoch": 38.49125703564728,
      "grad_norm": 0.691107988357544,
      "learning_rate": 0.002828,
      "loss": 1.0892,
      "step": 512896
    },
    {
      "epoch": 38.49606003752345,
      "grad_norm": 0.657853364944458,
      "learning_rate": 0.002828,
      "loss": 1.0908,
      "step": 512960
    },
    {
      "epoch": 38.500863039399626,
      "grad_norm": 0.5610063672065735,
      "learning_rate": 0.002828,
      "loss": 1.0891,
      "step": 513024
    },
    {
      "epoch": 38.505666041275795,
      "grad_norm": 0.5976006388664246,
      "learning_rate": 0.002828,
      "loss": 1.088,
      "step": 513088
    },
    {
      "epoch": 38.51046904315197,
      "grad_norm": 0.4549904465675354,
      "learning_rate": 0.002828,
      "loss": 1.0861,
      "step": 513152
    },
    {
      "epoch": 38.51527204502814,
      "grad_norm": 0.7005747556686401,
      "learning_rate": 0.002828,
      "loss": 1.0903,
      "step": 513216
    },
    {
      "epoch": 38.52007504690432,
      "grad_norm": 0.7073675990104675,
      "learning_rate": 0.002828,
      "loss": 1.0905,
      "step": 513280
    },
    {
      "epoch": 38.52487804878049,
      "grad_norm": 0.5825762152671814,
      "learning_rate": 0.002828,
      "loss": 1.0887,
      "step": 513344
    },
    {
      "epoch": 38.52968105065666,
      "grad_norm": 0.6172112822532654,
      "learning_rate": 0.002828,
      "loss": 1.0941,
      "step": 513408
    },
    {
      "epoch": 38.53448405253283,
      "grad_norm": 0.7240038514137268,
      "learning_rate": 0.002828,
      "loss": 1.0974,
      "step": 513472
    },
    {
      "epoch": 38.53928705440901,
      "grad_norm": 0.687610924243927,
      "learning_rate": 0.002828,
      "loss": 1.0917,
      "step": 513536
    },
    {
      "epoch": 38.54409005628518,
      "grad_norm": 0.5815373063087463,
      "learning_rate": 0.002828,
      "loss": 1.0937,
      "step": 513600
    },
    {
      "epoch": 38.54889305816135,
      "grad_norm": 0.5881292223930359,
      "learning_rate": 0.002828,
      "loss": 1.088,
      "step": 513664
    },
    {
      "epoch": 38.553696060037524,
      "grad_norm": 0.5459499359130859,
      "learning_rate": 0.002828,
      "loss": 1.0878,
      "step": 513728
    },
    {
      "epoch": 38.55849906191369,
      "grad_norm": 0.5804657340049744,
      "learning_rate": 0.002828,
      "loss": 1.0973,
      "step": 513792
    },
    {
      "epoch": 38.56330206378987,
      "grad_norm": 0.8041890859603882,
      "learning_rate": 0.002828,
      "loss": 1.0948,
      "step": 513856
    },
    {
      "epoch": 38.56810506566604,
      "grad_norm": 0.5507553815841675,
      "learning_rate": 0.002828,
      "loss": 1.0945,
      "step": 513920
    },
    {
      "epoch": 38.572908067542215,
      "grad_norm": 0.5750390291213989,
      "learning_rate": 0.002828,
      "loss": 1.0874,
      "step": 513984
    },
    {
      "epoch": 38.577711069418385,
      "grad_norm": 0.7084633708000183,
      "learning_rate": 0.002828,
      "loss": 1.0955,
      "step": 514048
    },
    {
      "epoch": 38.58251407129456,
      "grad_norm": 0.538191020488739,
      "learning_rate": 0.002828,
      "loss": 1.0871,
      "step": 514112
    },
    {
      "epoch": 38.58731707317073,
      "grad_norm": 0.5860592126846313,
      "learning_rate": 0.002828,
      "loss": 1.0946,
      "step": 514176
    },
    {
      "epoch": 38.59212007504691,
      "grad_norm": 0.7087304592132568,
      "learning_rate": 0.002828,
      "loss": 1.0938,
      "step": 514240
    },
    {
      "epoch": 38.596923076923076,
      "grad_norm": 0.736056923866272,
      "learning_rate": 0.002828,
      "loss": 1.0931,
      "step": 514304
    },
    {
      "epoch": 38.60172607879925,
      "grad_norm": 0.5794273614883423,
      "learning_rate": 0.002828,
      "loss": 1.092,
      "step": 514368
    },
    {
      "epoch": 38.60652908067542,
      "grad_norm": 0.5700469613075256,
      "learning_rate": 0.002828,
      "loss": 1.0928,
      "step": 514432
    },
    {
      "epoch": 38.61133208255159,
      "grad_norm": 0.8039986491203308,
      "learning_rate": 0.002828,
      "loss": 1.0953,
      "step": 514496
    },
    {
      "epoch": 38.61613508442777,
      "grad_norm": 0.5886857509613037,
      "learning_rate": 0.002828,
      "loss": 1.0937,
      "step": 514560
    },
    {
      "epoch": 38.62093808630394,
      "grad_norm": 0.70937579870224,
      "learning_rate": 0.002828,
      "loss": 1.0886,
      "step": 514624
    },
    {
      "epoch": 38.62574108818011,
      "grad_norm": 0.5150787830352783,
      "learning_rate": 0.002828,
      "loss": 1.0991,
      "step": 514688
    },
    {
      "epoch": 38.63054409005628,
      "grad_norm": 0.5741467475891113,
      "learning_rate": 0.002828,
      "loss": 1.1033,
      "step": 514752
    },
    {
      "epoch": 38.63534709193246,
      "grad_norm": 0.6546338796615601,
      "learning_rate": 0.002828,
      "loss": 1.0949,
      "step": 514816
    },
    {
      "epoch": 38.64015009380863,
      "grad_norm": 0.6613404750823975,
      "learning_rate": 0.002828,
      "loss": 1.0983,
      "step": 514880
    },
    {
      "epoch": 38.644953095684805,
      "grad_norm": 0.6713902950286865,
      "learning_rate": 0.002828,
      "loss": 1.0915,
      "step": 514944
    },
    {
      "epoch": 38.649756097560974,
      "grad_norm": 0.701934814453125,
      "learning_rate": 0.002828,
      "loss": 1.0891,
      "step": 515008
    },
    {
      "epoch": 38.65455909943715,
      "grad_norm": 0.7185537815093994,
      "learning_rate": 0.002828,
      "loss": 1.1002,
      "step": 515072
    },
    {
      "epoch": 38.65936210131332,
      "grad_norm": 0.7131350040435791,
      "learning_rate": 0.002828,
      "loss": 1.0986,
      "step": 515136
    },
    {
      "epoch": 38.6641651031895,
      "grad_norm": 0.5673279762268066,
      "learning_rate": 0.002828,
      "loss": 1.0947,
      "step": 515200
    },
    {
      "epoch": 38.668968105065666,
      "grad_norm": 0.8174933791160583,
      "learning_rate": 0.002828,
      "loss": 1.1058,
      "step": 515264
    },
    {
      "epoch": 38.673771106941835,
      "grad_norm": 0.6347920298576355,
      "learning_rate": 0.002828,
      "loss": 1.0964,
      "step": 515328
    },
    {
      "epoch": 38.67857410881801,
      "grad_norm": 0.630473792552948,
      "learning_rate": 0.002828,
      "loss": 1.1003,
      "step": 515392
    },
    {
      "epoch": 38.68337711069418,
      "grad_norm": 0.7728782296180725,
      "learning_rate": 0.002828,
      "loss": 1.0962,
      "step": 515456
    },
    {
      "epoch": 38.68818011257036,
      "grad_norm": 0.6685201525688171,
      "learning_rate": 0.002828,
      "loss": 1.0965,
      "step": 515520
    },
    {
      "epoch": 38.69298311444653,
      "grad_norm": 0.5990980267524719,
      "learning_rate": 0.002828,
      "loss": 1.0971,
      "step": 515584
    },
    {
      "epoch": 38.6977861163227,
      "grad_norm": 0.6950483918190002,
      "learning_rate": 0.002828,
      "loss": 1.1007,
      "step": 515648
    },
    {
      "epoch": 38.70258911819887,
      "grad_norm": 0.6069904565811157,
      "learning_rate": 0.002828,
      "loss": 1.0995,
      "step": 515712
    },
    {
      "epoch": 38.70739212007505,
      "grad_norm": 0.6790332794189453,
      "learning_rate": 0.002828,
      "loss": 1.0931,
      "step": 515776
    },
    {
      "epoch": 38.71219512195122,
      "grad_norm": 0.6225919127464294,
      "learning_rate": 0.002828,
      "loss": 1.0946,
      "step": 515840
    },
    {
      "epoch": 38.716998123827395,
      "grad_norm": 0.5538632869720459,
      "learning_rate": 0.002828,
      "loss": 1.0975,
      "step": 515904
    },
    {
      "epoch": 38.721801125703564,
      "grad_norm": 0.5822528600692749,
      "learning_rate": 0.002828,
      "loss": 1.1019,
      "step": 515968
    },
    {
      "epoch": 38.72660412757974,
      "grad_norm": 0.5993365049362183,
      "learning_rate": 0.002828,
      "loss": 1.099,
      "step": 516032
    },
    {
      "epoch": 38.73140712945591,
      "grad_norm": 0.5491153001785278,
      "learning_rate": 0.002828,
      "loss": 1.0961,
      "step": 516096
    },
    {
      "epoch": 38.73621013133208,
      "grad_norm": 0.5945103168487549,
      "learning_rate": 0.002828,
      "loss": 1.0929,
      "step": 516160
    },
    {
      "epoch": 38.741013133208256,
      "grad_norm": 0.7646733522415161,
      "learning_rate": 0.002828,
      "loss": 1.1043,
      "step": 516224
    },
    {
      "epoch": 38.745816135084425,
      "grad_norm": 0.5670483112335205,
      "learning_rate": 0.002828,
      "loss": 1.0979,
      "step": 516288
    },
    {
      "epoch": 38.7506191369606,
      "grad_norm": 0.5190681219100952,
      "learning_rate": 0.002828,
      "loss": 1.0952,
      "step": 516352
    },
    {
      "epoch": 38.75542213883677,
      "grad_norm": 0.4930049777030945,
      "learning_rate": 0.002828,
      "loss": 1.0968,
      "step": 516416
    },
    {
      "epoch": 38.76022514071295,
      "grad_norm": 0.5032718777656555,
      "learning_rate": 0.002828,
      "loss": 1.1002,
      "step": 516480
    },
    {
      "epoch": 38.76502814258912,
      "grad_norm": 0.6901083588600159,
      "learning_rate": 0.002828,
      "loss": 1.0912,
      "step": 516544
    },
    {
      "epoch": 38.76983114446529,
      "grad_norm": 0.6587644219398499,
      "learning_rate": 0.002828,
      "loss": 1.0992,
      "step": 516608
    },
    {
      "epoch": 38.77463414634146,
      "grad_norm": 0.5171419382095337,
      "learning_rate": 0.002828,
      "loss": 1.0938,
      "step": 516672
    },
    {
      "epoch": 38.77943714821764,
      "grad_norm": 0.5529362559318542,
      "learning_rate": 0.002828,
      "loss": 1.095,
      "step": 516736
    },
    {
      "epoch": 38.78424015009381,
      "grad_norm": 0.5704380869865417,
      "learning_rate": 0.002828,
      "loss": 1.0978,
      "step": 516800
    },
    {
      "epoch": 38.789043151969985,
      "grad_norm": 0.555026650428772,
      "learning_rate": 0.002828,
      "loss": 1.0913,
      "step": 516864
    },
    {
      "epoch": 38.793846153846154,
      "grad_norm": 0.5735583305358887,
      "learning_rate": 0.002828,
      "loss": 1.0955,
      "step": 516928
    },
    {
      "epoch": 38.79864915572232,
      "grad_norm": 0.625392496585846,
      "learning_rate": 0.002828,
      "loss": 1.0995,
      "step": 516992
    },
    {
      "epoch": 38.8034521575985,
      "grad_norm": 0.6034018397331238,
      "learning_rate": 0.002828,
      "loss": 1.0937,
      "step": 517056
    },
    {
      "epoch": 38.80825515947467,
      "grad_norm": 0.5921566486358643,
      "learning_rate": 0.002828,
      "loss": 1.0998,
      "step": 517120
    },
    {
      "epoch": 38.813058161350845,
      "grad_norm": 0.5109006762504578,
      "learning_rate": 0.002828,
      "loss": 1.0951,
      "step": 517184
    },
    {
      "epoch": 38.817861163227015,
      "grad_norm": 0.6649883985519409,
      "learning_rate": 0.002828,
      "loss": 1.102,
      "step": 517248
    },
    {
      "epoch": 38.82266416510319,
      "grad_norm": 0.662285566329956,
      "learning_rate": 0.002828,
      "loss": 1.0959,
      "step": 517312
    },
    {
      "epoch": 38.82746716697936,
      "grad_norm": 0.5535122156143188,
      "learning_rate": 0.002828,
      "loss": 1.1035,
      "step": 517376
    },
    {
      "epoch": 38.83227016885554,
      "grad_norm": 0.5791530013084412,
      "learning_rate": 0.002828,
      "loss": 1.0936,
      "step": 517440
    },
    {
      "epoch": 38.837073170731706,
      "grad_norm": 0.5605199337005615,
      "learning_rate": 0.002828,
      "loss": 1.0933,
      "step": 517504
    },
    {
      "epoch": 38.84187617260788,
      "grad_norm": 0.6626695394515991,
      "learning_rate": 0.002828,
      "loss": 1.103,
      "step": 517568
    },
    {
      "epoch": 38.84667917448405,
      "grad_norm": 0.5840564966201782,
      "learning_rate": 0.002828,
      "loss": 1.0962,
      "step": 517632
    },
    {
      "epoch": 38.85148217636023,
      "grad_norm": 0.771919846534729,
      "learning_rate": 0.002828,
      "loss": 1.1029,
      "step": 517696
    },
    {
      "epoch": 38.8562851782364,
      "grad_norm": 0.6716858148574829,
      "learning_rate": 0.002828,
      "loss": 1.0984,
      "step": 517760
    },
    {
      "epoch": 38.86108818011257,
      "grad_norm": 0.7528976202011108,
      "learning_rate": 0.002828,
      "loss": 1.1002,
      "step": 517824
    },
    {
      "epoch": 38.865891181988744,
      "grad_norm": 0.6416760087013245,
      "learning_rate": 0.002828,
      "loss": 1.0986,
      "step": 517888
    },
    {
      "epoch": 38.87069418386491,
      "grad_norm": 0.8085233569145203,
      "learning_rate": 0.002828,
      "loss": 1.0977,
      "step": 517952
    },
    {
      "epoch": 38.87549718574109,
      "grad_norm": 0.6201658844947815,
      "learning_rate": 0.002828,
      "loss": 1.101,
      "step": 518016
    },
    {
      "epoch": 38.88030018761726,
      "grad_norm": 0.6434842944145203,
      "learning_rate": 0.002828,
      "loss": 1.0945,
      "step": 518080
    },
    {
      "epoch": 38.885103189493435,
      "grad_norm": 0.6296826601028442,
      "learning_rate": 0.002828,
      "loss": 1.0953,
      "step": 518144
    },
    {
      "epoch": 38.889906191369604,
      "grad_norm": 0.6131699681282043,
      "learning_rate": 0.002828,
      "loss": 1.1066,
      "step": 518208
    },
    {
      "epoch": 38.89470919324578,
      "grad_norm": 0.6386839151382446,
      "learning_rate": 0.002828,
      "loss": 1.1016,
      "step": 518272
    },
    {
      "epoch": 38.89951219512195,
      "grad_norm": 0.5856118202209473,
      "learning_rate": 0.002828,
      "loss": 1.1013,
      "step": 518336
    },
    {
      "epoch": 38.90431519699813,
      "grad_norm": 0.627488374710083,
      "learning_rate": 0.002828,
      "loss": 1.1064,
      "step": 518400
    },
    {
      "epoch": 38.909118198874296,
      "grad_norm": 0.5808975696563721,
      "learning_rate": 0.002828,
      "loss": 1.1035,
      "step": 518464
    },
    {
      "epoch": 38.91392120075047,
      "grad_norm": 0.6418662667274475,
      "learning_rate": 0.002828,
      "loss": 1.0998,
      "step": 518528
    },
    {
      "epoch": 38.91872420262664,
      "grad_norm": 0.7283346652984619,
      "learning_rate": 0.002828,
      "loss": 1.0988,
      "step": 518592
    },
    {
      "epoch": 38.92352720450281,
      "grad_norm": 0.5792503952980042,
      "learning_rate": 0.002828,
      "loss": 1.1016,
      "step": 518656
    },
    {
      "epoch": 38.92833020637899,
      "grad_norm": 0.6570231318473816,
      "learning_rate": 0.002828,
      "loss": 1.0969,
      "step": 518720
    },
    {
      "epoch": 38.93313320825516,
      "grad_norm": 0.5340087413787842,
      "learning_rate": 0.002828,
      "loss": 1.1022,
      "step": 518784
    },
    {
      "epoch": 38.93793621013133,
      "grad_norm": 0.6886125802993774,
      "learning_rate": 0.002828,
      "loss": 1.0984,
      "step": 518848
    },
    {
      "epoch": 38.9427392120075,
      "grad_norm": 0.6742810010910034,
      "learning_rate": 0.002828,
      "loss": 1.0961,
      "step": 518912
    },
    {
      "epoch": 38.94754221388368,
      "grad_norm": 0.5848747491836548,
      "learning_rate": 0.002828,
      "loss": 1.1027,
      "step": 518976
    },
    {
      "epoch": 38.95234521575985,
      "grad_norm": 0.6648167967796326,
      "learning_rate": 0.002828,
      "loss": 1.108,
      "step": 519040
    },
    {
      "epoch": 38.957148217636025,
      "grad_norm": 0.5892636775970459,
      "learning_rate": 0.002828,
      "loss": 1.0965,
      "step": 519104
    },
    {
      "epoch": 38.961951219512194,
      "grad_norm": 0.5304135680198669,
      "learning_rate": 0.002828,
      "loss": 1.0996,
      "step": 519168
    },
    {
      "epoch": 38.96675422138837,
      "grad_norm": 0.5350622534751892,
      "learning_rate": 0.002828,
      "loss": 1.1015,
      "step": 519232
    },
    {
      "epoch": 38.97155722326454,
      "grad_norm": 0.6611760258674622,
      "learning_rate": 0.002828,
      "loss": 1.1017,
      "step": 519296
    },
    {
      "epoch": 38.976360225140716,
      "grad_norm": 0.6533018350601196,
      "learning_rate": 0.002828,
      "loss": 1.1043,
      "step": 519360
    },
    {
      "epoch": 38.981163227016886,
      "grad_norm": 0.6382613182067871,
      "learning_rate": 0.002828,
      "loss": 1.1024,
      "step": 519424
    },
    {
      "epoch": 38.985966228893055,
      "grad_norm": 0.6975395083427429,
      "learning_rate": 0.002828,
      "loss": 1.0993,
      "step": 519488
    },
    {
      "epoch": 38.99076923076923,
      "grad_norm": 0.5687751770019531,
      "learning_rate": 0.002828,
      "loss": 1.0989,
      "step": 519552
    },
    {
      "epoch": 38.9955722326454,
      "grad_norm": 0.5590277314186096,
      "learning_rate": 0.002828,
      "loss": 1.1026,
      "step": 519616
    },
    {
      "epoch": 39.00037523452158,
      "grad_norm": 0.6050010323524475,
      "learning_rate": 0.002828,
      "loss": 1.101,
      "step": 519680
    },
    {
      "epoch": 39.00517823639775,
      "grad_norm": 0.5483032464981079,
      "learning_rate": 0.002828,
      "loss": 1.0784,
      "step": 519744
    },
    {
      "epoch": 39.00998123827392,
      "grad_norm": 0.6067426800727844,
      "learning_rate": 0.002828,
      "loss": 1.081,
      "step": 519808
    },
    {
      "epoch": 39.01478424015009,
      "grad_norm": 0.6473919749259949,
      "learning_rate": 0.002828,
      "loss": 1.0773,
      "step": 519872
    },
    {
      "epoch": 39.01958724202627,
      "grad_norm": 0.6021099090576172,
      "learning_rate": 0.002828,
      "loss": 1.0786,
      "step": 519936
    },
    {
      "epoch": 39.02439024390244,
      "grad_norm": 0.5758794546127319,
      "learning_rate": 0.002828,
      "loss": 1.0821,
      "step": 520000
    },
    {
      "epoch": 39.029193245778615,
      "grad_norm": 0.5767053961753845,
      "learning_rate": 0.002828,
      "loss": 1.0775,
      "step": 520064
    },
    {
      "epoch": 39.033996247654784,
      "grad_norm": 0.6255131959915161,
      "learning_rate": 0.002828,
      "loss": 1.0813,
      "step": 520128
    },
    {
      "epoch": 39.03879924953096,
      "grad_norm": 0.6981056928634644,
      "learning_rate": 0.002828,
      "loss": 1.0787,
      "step": 520192
    },
    {
      "epoch": 39.04360225140713,
      "grad_norm": 0.557799756526947,
      "learning_rate": 0.002828,
      "loss": 1.0811,
      "step": 520256
    },
    {
      "epoch": 39.0484052532833,
      "grad_norm": 0.68275386095047,
      "learning_rate": 0.002828,
      "loss": 1.0839,
      "step": 520320
    },
    {
      "epoch": 39.053208255159475,
      "grad_norm": 0.6308878064155579,
      "learning_rate": 0.002828,
      "loss": 1.0849,
      "step": 520384
    },
    {
      "epoch": 39.058011257035645,
      "grad_norm": 0.5470284819602966,
      "learning_rate": 0.002828,
      "loss": 1.0865,
      "step": 520448
    },
    {
      "epoch": 39.06281425891182,
      "grad_norm": 0.677872359752655,
      "learning_rate": 0.002828,
      "loss": 1.0839,
      "step": 520512
    },
    {
      "epoch": 39.06761726078799,
      "grad_norm": 0.6441925168037415,
      "learning_rate": 0.002828,
      "loss": 1.0787,
      "step": 520576
    },
    {
      "epoch": 39.07242026266417,
      "grad_norm": 0.5155608654022217,
      "learning_rate": 0.002828,
      "loss": 1.0854,
      "step": 520640
    },
    {
      "epoch": 39.077223264540336,
      "grad_norm": 0.6666632294654846,
      "learning_rate": 0.002828,
      "loss": 1.0848,
      "step": 520704
    },
    {
      "epoch": 39.08202626641651,
      "grad_norm": 0.7269572615623474,
      "learning_rate": 0.002828,
      "loss": 1.0867,
      "step": 520768
    },
    {
      "epoch": 39.08682926829268,
      "grad_norm": 0.6180398464202881,
      "learning_rate": 0.002828,
      "loss": 1.0839,
      "step": 520832
    },
    {
      "epoch": 39.09163227016886,
      "grad_norm": 0.5263913869857788,
      "learning_rate": 0.002828,
      "loss": 1.0861,
      "step": 520896
    },
    {
      "epoch": 39.09643527204503,
      "grad_norm": 0.7088015079498291,
      "learning_rate": 0.002828,
      "loss": 1.089,
      "step": 520960
    },
    {
      "epoch": 39.1012382739212,
      "grad_norm": 0.7093296647071838,
      "learning_rate": 0.002828,
      "loss": 1.0882,
      "step": 521024
    },
    {
      "epoch": 39.106041275797374,
      "grad_norm": 0.6128780841827393,
      "learning_rate": 0.002828,
      "loss": 1.0845,
      "step": 521088
    },
    {
      "epoch": 39.11084427767354,
      "grad_norm": 0.6755836606025696,
      "learning_rate": 0.002828,
      "loss": 1.0873,
      "step": 521152
    },
    {
      "epoch": 39.11564727954972,
      "grad_norm": 0.616104245185852,
      "learning_rate": 0.002828,
      "loss": 1.0891,
      "step": 521216
    },
    {
      "epoch": 39.12045028142589,
      "grad_norm": 0.6191969513893127,
      "learning_rate": 0.002828,
      "loss": 1.0901,
      "step": 521280
    },
    {
      "epoch": 39.125253283302065,
      "grad_norm": 0.4933353066444397,
      "learning_rate": 0.002828,
      "loss": 1.0842,
      "step": 521344
    },
    {
      "epoch": 39.130056285178235,
      "grad_norm": 0.5262903571128845,
      "learning_rate": 0.002828,
      "loss": 1.0935,
      "step": 521408
    },
    {
      "epoch": 39.13485928705441,
      "grad_norm": 0.7534499168395996,
      "learning_rate": 0.002828,
      "loss": 1.0906,
      "step": 521472
    },
    {
      "epoch": 39.13966228893058,
      "grad_norm": 0.6676933765411377,
      "learning_rate": 0.002828,
      "loss": 1.0896,
      "step": 521536
    },
    {
      "epoch": 39.14446529080676,
      "grad_norm": 0.5996463894844055,
      "learning_rate": 0.002828,
      "loss": 1.0909,
      "step": 521600
    },
    {
      "epoch": 39.149268292682926,
      "grad_norm": 0.566777229309082,
      "learning_rate": 0.002828,
      "loss": 1.083,
      "step": 521664
    },
    {
      "epoch": 39.1540712945591,
      "grad_norm": 0.560036301612854,
      "learning_rate": 0.002828,
      "loss": 1.0947,
      "step": 521728
    },
    {
      "epoch": 39.15887429643527,
      "grad_norm": 0.49662432074546814,
      "learning_rate": 0.002828,
      "loss": 1.0925,
      "step": 521792
    },
    {
      "epoch": 39.16367729831144,
      "grad_norm": 0.6516875624656677,
      "learning_rate": 0.002828,
      "loss": 1.0967,
      "step": 521856
    },
    {
      "epoch": 39.16848030018762,
      "grad_norm": 0.6191356778144836,
      "learning_rate": 0.002828,
      "loss": 1.0933,
      "step": 521920
    },
    {
      "epoch": 39.17328330206379,
      "grad_norm": 0.8029376268386841,
      "learning_rate": 0.002828,
      "loss": 1.0939,
      "step": 521984
    },
    {
      "epoch": 39.17808630393996,
      "grad_norm": 0.7398802638053894,
      "learning_rate": 0.002828,
      "loss": 1.0931,
      "step": 522048
    },
    {
      "epoch": 39.18288930581613,
      "grad_norm": 0.5711380839347839,
      "learning_rate": 0.002828,
      "loss": 1.0872,
      "step": 522112
    },
    {
      "epoch": 39.18769230769231,
      "grad_norm": 0.5394037365913391,
      "learning_rate": 0.002828,
      "loss": 1.0941,
      "step": 522176
    },
    {
      "epoch": 39.19249530956848,
      "grad_norm": 0.7376382350921631,
      "learning_rate": 0.002828,
      "loss": 1.0896,
      "step": 522240
    },
    {
      "epoch": 39.197298311444655,
      "grad_norm": 0.6365683078765869,
      "learning_rate": 0.002828,
      "loss": 1.0917,
      "step": 522304
    },
    {
      "epoch": 39.202101313320824,
      "grad_norm": 0.7121477127075195,
      "learning_rate": 0.002828,
      "loss": 1.0907,
      "step": 522368
    },
    {
      "epoch": 39.206904315197,
      "grad_norm": 0.5471106171607971,
      "learning_rate": 0.002828,
      "loss": 1.0913,
      "step": 522432
    },
    {
      "epoch": 39.21170731707317,
      "grad_norm": 0.7246633172035217,
      "learning_rate": 0.002828,
      "loss": 1.0991,
      "step": 522496
    },
    {
      "epoch": 39.21651031894935,
      "grad_norm": 0.6447710990905762,
      "learning_rate": 0.002828,
      "loss": 1.0955,
      "step": 522560
    },
    {
      "epoch": 39.221313320825516,
      "grad_norm": 0.6338580250740051,
      "learning_rate": 0.002828,
      "loss": 1.088,
      "step": 522624
    },
    {
      "epoch": 39.226116322701685,
      "grad_norm": 0.539563775062561,
      "learning_rate": 0.002828,
      "loss": 1.0885,
      "step": 522688
    },
    {
      "epoch": 39.23091932457786,
      "grad_norm": 0.6259217262268066,
      "learning_rate": 0.002828,
      "loss": 1.0964,
      "step": 522752
    },
    {
      "epoch": 39.23572232645403,
      "grad_norm": 0.658747673034668,
      "learning_rate": 0.002828,
      "loss": 1.0877,
      "step": 522816
    },
    {
      "epoch": 39.24052532833021,
      "grad_norm": 0.9338593482971191,
      "learning_rate": 0.002828,
      "loss": 1.0912,
      "step": 522880
    },
    {
      "epoch": 39.24532833020638,
      "grad_norm": 0.6139676570892334,
      "learning_rate": 0.002828,
      "loss": 1.0891,
      "step": 522944
    },
    {
      "epoch": 39.25013133208255,
      "grad_norm": 0.5953230857849121,
      "learning_rate": 0.002828,
      "loss": 1.0909,
      "step": 523008
    },
    {
      "epoch": 39.25493433395872,
      "grad_norm": 0.607876718044281,
      "learning_rate": 0.002828,
      "loss": 1.0846,
      "step": 523072
    },
    {
      "epoch": 39.2597373358349,
      "grad_norm": 0.6370177268981934,
      "learning_rate": 0.002828,
      "loss": 1.0923,
      "step": 523136
    },
    {
      "epoch": 39.26454033771107,
      "grad_norm": 0.5965909957885742,
      "learning_rate": 0.002828,
      "loss": 1.0905,
      "step": 523200
    },
    {
      "epoch": 39.269343339587245,
      "grad_norm": 0.5890311002731323,
      "learning_rate": 0.002828,
      "loss": 1.0908,
      "step": 523264
    },
    {
      "epoch": 39.274146341463414,
      "grad_norm": 0.5888711214065552,
      "learning_rate": 0.002828,
      "loss": 1.0938,
      "step": 523328
    },
    {
      "epoch": 39.27894934333959,
      "grad_norm": 0.634347140789032,
      "learning_rate": 0.002828,
      "loss": 1.0916,
      "step": 523392
    },
    {
      "epoch": 39.28375234521576,
      "grad_norm": 0.6978724002838135,
      "learning_rate": 0.002828,
      "loss": 1.0919,
      "step": 523456
    },
    {
      "epoch": 39.28855534709193,
      "grad_norm": 0.5859372615814209,
      "learning_rate": 0.002828,
      "loss": 1.0921,
      "step": 523520
    },
    {
      "epoch": 39.293358348968106,
      "grad_norm": 0.6379390954971313,
      "learning_rate": 0.002828,
      "loss": 1.0935,
      "step": 523584
    },
    {
      "epoch": 39.298161350844275,
      "grad_norm": 0.6283619403839111,
      "learning_rate": 0.002828,
      "loss": 1.0962,
      "step": 523648
    },
    {
      "epoch": 39.30296435272045,
      "grad_norm": 0.7185676693916321,
      "learning_rate": 0.002828,
      "loss": 1.0929,
      "step": 523712
    },
    {
      "epoch": 39.30776735459662,
      "grad_norm": 0.5743740200996399,
      "learning_rate": 0.002828,
      "loss": 1.0989,
      "step": 523776
    },
    {
      "epoch": 39.3125703564728,
      "grad_norm": 0.5320876836776733,
      "learning_rate": 0.002828,
      "loss": 1.0939,
      "step": 523840
    },
    {
      "epoch": 39.317373358348966,
      "grad_norm": 0.6361159682273865,
      "learning_rate": 0.002828,
      "loss": 1.0963,
      "step": 523904
    },
    {
      "epoch": 39.32217636022514,
      "grad_norm": 0.6137511730194092,
      "learning_rate": 0.002828,
      "loss": 1.0893,
      "step": 523968
    },
    {
      "epoch": 39.32697936210131,
      "grad_norm": 0.5668014883995056,
      "learning_rate": 0.002828,
      "loss": 1.0992,
      "step": 524032
    },
    {
      "epoch": 39.33178236397749,
      "grad_norm": 0.8398164510726929,
      "learning_rate": 0.002828,
      "loss": 1.0974,
      "step": 524096
    },
    {
      "epoch": 39.33658536585366,
      "grad_norm": 0.6183857917785645,
      "learning_rate": 0.002828,
      "loss": 1.0904,
      "step": 524160
    },
    {
      "epoch": 39.341388367729834,
      "grad_norm": 0.5952325463294983,
      "learning_rate": 0.002828,
      "loss": 1.0937,
      "step": 524224
    },
    {
      "epoch": 39.346191369606004,
      "grad_norm": 0.6355870366096497,
      "learning_rate": 0.002828,
      "loss": 1.0983,
      "step": 524288
    },
    {
      "epoch": 39.35099437148217,
      "grad_norm": 0.6765314936637878,
      "learning_rate": 0.002828,
      "loss": 1.0876,
      "step": 524352
    },
    {
      "epoch": 39.35579737335835,
      "grad_norm": 0.6968716979026794,
      "learning_rate": 0.002828,
      "loss": 1.095,
      "step": 524416
    },
    {
      "epoch": 39.36060037523452,
      "grad_norm": 0.6691702008247375,
      "learning_rate": 0.002828,
      "loss": 1.0964,
      "step": 524480
    },
    {
      "epoch": 39.365403377110695,
      "grad_norm": 0.5527083873748779,
      "learning_rate": 0.002828,
      "loss": 1.0981,
      "step": 524544
    },
    {
      "epoch": 39.370206378986865,
      "grad_norm": 0.6122605800628662,
      "learning_rate": 0.002828,
      "loss": 1.0969,
      "step": 524608
    },
    {
      "epoch": 39.37500938086304,
      "grad_norm": 0.5258229374885559,
      "learning_rate": 0.002828,
      "loss": 1.0961,
      "step": 524672
    },
    {
      "epoch": 39.37981238273921,
      "grad_norm": 0.7176069021224976,
      "learning_rate": 0.002828,
      "loss": 1.0971,
      "step": 524736
    },
    {
      "epoch": 39.38461538461539,
      "grad_norm": 0.544308066368103,
      "learning_rate": 0.002828,
      "loss": 1.0941,
      "step": 524800
    },
    {
      "epoch": 39.389418386491556,
      "grad_norm": 0.6387256383895874,
      "learning_rate": 0.002828,
      "loss": 1.0963,
      "step": 524864
    },
    {
      "epoch": 39.39422138836773,
      "grad_norm": 0.6066888570785522,
      "learning_rate": 0.002828,
      "loss": 1.0975,
      "step": 524928
    },
    {
      "epoch": 39.3990243902439,
      "grad_norm": 0.6804490089416504,
      "learning_rate": 0.002828,
      "loss": 1.1012,
      "step": 524992
    },
    {
      "epoch": 39.40382739212008,
      "grad_norm": 0.6256688237190247,
      "learning_rate": 0.002828,
      "loss": 1.0984,
      "step": 525056
    },
    {
      "epoch": 39.40863039399625,
      "grad_norm": 0.5704160332679749,
      "learning_rate": 0.002828,
      "loss": 1.0965,
      "step": 525120
    },
    {
      "epoch": 39.41343339587242,
      "grad_norm": 0.643605649471283,
      "learning_rate": 0.002828,
      "loss": 1.0993,
      "step": 525184
    },
    {
      "epoch": 39.41823639774859,
      "grad_norm": 0.691459059715271,
      "learning_rate": 0.002828,
      "loss": 1.0889,
      "step": 525248
    },
    {
      "epoch": 39.42303939962476,
      "grad_norm": 0.5297619104385376,
      "learning_rate": 0.002828,
      "loss": 1.0973,
      "step": 525312
    },
    {
      "epoch": 39.42784240150094,
      "grad_norm": 0.650749921798706,
      "learning_rate": 0.002828,
      "loss": 1.1024,
      "step": 525376
    },
    {
      "epoch": 39.43264540337711,
      "grad_norm": 0.6719233393669128,
      "learning_rate": 0.002828,
      "loss": 1.1037,
      "step": 525440
    },
    {
      "epoch": 39.437448405253285,
      "grad_norm": 0.7150486707687378,
      "learning_rate": 0.002828,
      "loss": 1.0935,
      "step": 525504
    },
    {
      "epoch": 39.442251407129454,
      "grad_norm": 0.5051080584526062,
      "learning_rate": 0.002828,
      "loss": 1.1025,
      "step": 525568
    },
    {
      "epoch": 39.44705440900563,
      "grad_norm": 0.5855237245559692,
      "learning_rate": 0.002828,
      "loss": 1.0889,
      "step": 525632
    },
    {
      "epoch": 39.4518574108818,
      "grad_norm": 0.561397135257721,
      "learning_rate": 0.002828,
      "loss": 1.0973,
      "step": 525696
    },
    {
      "epoch": 39.45666041275798,
      "grad_norm": 0.6438418626785278,
      "learning_rate": 0.002828,
      "loss": 1.104,
      "step": 525760
    },
    {
      "epoch": 39.461463414634146,
      "grad_norm": 0.6153573989868164,
      "learning_rate": 0.002828,
      "loss": 1.0989,
      "step": 525824
    },
    {
      "epoch": 39.46626641651032,
      "grad_norm": 0.6531253457069397,
      "learning_rate": 0.002828,
      "loss": 1.0977,
      "step": 525888
    },
    {
      "epoch": 39.47106941838649,
      "grad_norm": 0.7286359667778015,
      "learning_rate": 0.002828,
      "loss": 1.0986,
      "step": 525952
    },
    {
      "epoch": 39.47587242026266,
      "grad_norm": 0.5893198847770691,
      "learning_rate": 0.002828,
      "loss": 1.101,
      "step": 526016
    },
    {
      "epoch": 39.48067542213884,
      "grad_norm": 0.5942003726959229,
      "learning_rate": 0.002828,
      "loss": 1.1037,
      "step": 526080
    },
    {
      "epoch": 39.48547842401501,
      "grad_norm": 0.5884833335876465,
      "learning_rate": 0.002828,
      "loss": 1.0984,
      "step": 526144
    },
    {
      "epoch": 39.49028142589118,
      "grad_norm": 0.6396442651748657,
      "learning_rate": 0.002828,
      "loss": 1.0984,
      "step": 526208
    },
    {
      "epoch": 39.49508442776735,
      "grad_norm": 0.664161205291748,
      "learning_rate": 0.002828,
      "loss": 1.103,
      "step": 526272
    },
    {
      "epoch": 39.49988742964353,
      "grad_norm": 0.6068063974380493,
      "learning_rate": 0.002828,
      "loss": 1.098,
      "step": 526336
    },
    {
      "epoch": 39.5046904315197,
      "grad_norm": 0.6167985200881958,
      "learning_rate": 0.002828,
      "loss": 1.1029,
      "step": 526400
    },
    {
      "epoch": 39.509493433395875,
      "grad_norm": 0.6410574316978455,
      "learning_rate": 0.002828,
      "loss": 1.1036,
      "step": 526464
    },
    {
      "epoch": 39.514296435272044,
      "grad_norm": 0.5695111751556396,
      "learning_rate": 0.002828,
      "loss": 1.1009,
      "step": 526528
    },
    {
      "epoch": 39.51909943714822,
      "grad_norm": 0.8604126572608948,
      "learning_rate": 0.002828,
      "loss": 1.1033,
      "step": 526592
    },
    {
      "epoch": 39.52390243902439,
      "grad_norm": 0.6506824493408203,
      "learning_rate": 0.002828,
      "loss": 1.0995,
      "step": 526656
    },
    {
      "epoch": 39.528705440900566,
      "grad_norm": 0.5646496415138245,
      "learning_rate": 0.002828,
      "loss": 1.0983,
      "step": 526720
    },
    {
      "epoch": 39.533508442776736,
      "grad_norm": 0.7363172769546509,
      "learning_rate": 0.002828,
      "loss": 1.1057,
      "step": 526784
    },
    {
      "epoch": 39.538311444652905,
      "grad_norm": 0.598517894744873,
      "learning_rate": 0.002828,
      "loss": 1.1013,
      "step": 526848
    },
    {
      "epoch": 39.54311444652908,
      "grad_norm": 0.8225451111793518,
      "learning_rate": 0.002828,
      "loss": 1.1021,
      "step": 526912
    },
    {
      "epoch": 39.54791744840525,
      "grad_norm": 0.5679796934127808,
      "learning_rate": 0.002828,
      "loss": 1.0973,
      "step": 526976
    },
    {
      "epoch": 39.55272045028143,
      "grad_norm": 0.5201491117477417,
      "learning_rate": 0.002828,
      "loss": 1.099,
      "step": 527040
    },
    {
      "epoch": 39.5575234521576,
      "grad_norm": 0.6762983202934265,
      "learning_rate": 0.002828,
      "loss": 1.1049,
      "step": 527104
    },
    {
      "epoch": 39.56232645403377,
      "grad_norm": 0.7534833550453186,
      "learning_rate": 0.002828,
      "loss": 1.104,
      "step": 527168
    },
    {
      "epoch": 39.56712945590994,
      "grad_norm": 0.6385214328765869,
      "learning_rate": 0.002828,
      "loss": 1.1013,
      "step": 527232
    },
    {
      "epoch": 39.57193245778612,
      "grad_norm": 0.59105384349823,
      "learning_rate": 0.002828,
      "loss": 1.1059,
      "step": 527296
    },
    {
      "epoch": 39.57673545966229,
      "grad_norm": 0.6427984237670898,
      "learning_rate": 0.002828,
      "loss": 1.1064,
      "step": 527360
    },
    {
      "epoch": 39.581538461538464,
      "grad_norm": 0.7744491696357727,
      "learning_rate": 0.002828,
      "loss": 1.1046,
      "step": 527424
    },
    {
      "epoch": 39.586341463414634,
      "grad_norm": 0.49544790387153625,
      "learning_rate": 0.002828,
      "loss": 1.1031,
      "step": 527488
    },
    {
      "epoch": 39.59114446529081,
      "grad_norm": 0.6462889909744263,
      "learning_rate": 0.002828,
      "loss": 1.099,
      "step": 527552
    },
    {
      "epoch": 39.59594746716698,
      "grad_norm": 0.7103201150894165,
      "learning_rate": 0.002828,
      "loss": 1.1006,
      "step": 527616
    },
    {
      "epoch": 39.60075046904315,
      "grad_norm": 0.5468045473098755,
      "learning_rate": 0.002828,
      "loss": 1.1072,
      "step": 527680
    },
    {
      "epoch": 39.605553470919325,
      "grad_norm": 0.5340071320533752,
      "learning_rate": 0.002828,
      "loss": 1.1011,
      "step": 527744
    },
    {
      "epoch": 39.610356472795495,
      "grad_norm": 0.6542819142341614,
      "learning_rate": 0.002828,
      "loss": 1.1011,
      "step": 527808
    },
    {
      "epoch": 39.61515947467167,
      "grad_norm": 0.6813800930976868,
      "learning_rate": 0.002828,
      "loss": 1.1031,
      "step": 527872
    },
    {
      "epoch": 39.61996247654784,
      "grad_norm": 0.7469828724861145,
      "learning_rate": 0.002828,
      "loss": 1.0996,
      "step": 527936
    },
    {
      "epoch": 39.62476547842402,
      "grad_norm": 0.7847934365272522,
      "learning_rate": 0.002828,
      "loss": 1.0987,
      "step": 528000
    },
    {
      "epoch": 39.629568480300186,
      "grad_norm": 0.5818659067153931,
      "learning_rate": 0.002828,
      "loss": 1.099,
      "step": 528064
    },
    {
      "epoch": 39.63437148217636,
      "grad_norm": 0.6419061422348022,
      "learning_rate": 0.002828,
      "loss": 1.1005,
      "step": 528128
    },
    {
      "epoch": 39.63917448405253,
      "grad_norm": 0.6728861927986145,
      "learning_rate": 0.002828,
      "loss": 1.1004,
      "step": 528192
    },
    {
      "epoch": 39.64397748592871,
      "grad_norm": 0.645656943321228,
      "learning_rate": 0.002828,
      "loss": 1.101,
      "step": 528256
    },
    {
      "epoch": 39.64878048780488,
      "grad_norm": 0.5696425437927246,
      "learning_rate": 0.002828,
      "loss": 1.1031,
      "step": 528320
    },
    {
      "epoch": 39.653583489681054,
      "grad_norm": 1.0262229442596436,
      "learning_rate": 0.002828,
      "loss": 1.1,
      "step": 528384
    },
    {
      "epoch": 39.65838649155722,
      "grad_norm": 0.6700289249420166,
      "learning_rate": 0.002828,
      "loss": 1.1002,
      "step": 528448
    },
    {
      "epoch": 39.66318949343339,
      "grad_norm": 0.5933250188827515,
      "learning_rate": 0.002828,
      "loss": 1.0997,
      "step": 528512
    },
    {
      "epoch": 39.66799249530957,
      "grad_norm": 0.5513843297958374,
      "learning_rate": 0.002828,
      "loss": 1.1087,
      "step": 528576
    },
    {
      "epoch": 39.67279549718574,
      "grad_norm": 0.632848858833313,
      "learning_rate": 0.002828,
      "loss": 1.1055,
      "step": 528640
    },
    {
      "epoch": 39.677598499061915,
      "grad_norm": 0.6517565846443176,
      "learning_rate": 0.002828,
      "loss": 1.0998,
      "step": 528704
    },
    {
      "epoch": 39.682401500938084,
      "grad_norm": 0.5785037875175476,
      "learning_rate": 0.002828,
      "loss": 1.0996,
      "step": 528768
    },
    {
      "epoch": 39.68720450281426,
      "grad_norm": 0.6588200926780701,
      "learning_rate": 0.002828,
      "loss": 1.1026,
      "step": 528832
    },
    {
      "epoch": 39.69200750469043,
      "grad_norm": 0.5650362968444824,
      "learning_rate": 0.002828,
      "loss": 1.106,
      "step": 528896
    },
    {
      "epoch": 39.69681050656661,
      "grad_norm": 0.6296866536140442,
      "learning_rate": 0.002828,
      "loss": 1.1033,
      "step": 528960
    },
    {
      "epoch": 39.701613508442776,
      "grad_norm": 0.7417605519294739,
      "learning_rate": 0.002828,
      "loss": 1.109,
      "step": 529024
    },
    {
      "epoch": 39.70641651031895,
      "grad_norm": 0.5952714681625366,
      "learning_rate": 0.002828,
      "loss": 1.1063,
      "step": 529088
    },
    {
      "epoch": 39.71121951219512,
      "grad_norm": 0.5962236523628235,
      "learning_rate": 0.002828,
      "loss": 1.102,
      "step": 529152
    },
    {
      "epoch": 39.7160225140713,
      "grad_norm": 0.6925379633903503,
      "learning_rate": 0.002828,
      "loss": 1.1073,
      "step": 529216
    },
    {
      "epoch": 39.72082551594747,
      "grad_norm": 0.5629957914352417,
      "learning_rate": 0.002828,
      "loss": 1.1037,
      "step": 529280
    },
    {
      "epoch": 39.72562851782364,
      "grad_norm": 0.6818074584007263,
      "learning_rate": 0.002828,
      "loss": 1.1017,
      "step": 529344
    },
    {
      "epoch": 39.73043151969981,
      "grad_norm": 0.8665837049484253,
      "learning_rate": 0.002828,
      "loss": 1.1023,
      "step": 529408
    },
    {
      "epoch": 39.73523452157598,
      "grad_norm": 0.608192503452301,
      "learning_rate": 0.002828,
      "loss": 1.1088,
      "step": 529472
    },
    {
      "epoch": 39.74003752345216,
      "grad_norm": 0.6119167804718018,
      "learning_rate": 0.002828,
      "loss": 1.1112,
      "step": 529536
    },
    {
      "epoch": 39.74484052532833,
      "grad_norm": 0.5534200072288513,
      "learning_rate": 0.002828,
      "loss": 1.1018,
      "step": 529600
    },
    {
      "epoch": 39.749643527204505,
      "grad_norm": 0.5940262675285339,
      "learning_rate": 0.002828,
      "loss": 1.1045,
      "step": 529664
    },
    {
      "epoch": 39.754446529080674,
      "grad_norm": 0.611477792263031,
      "learning_rate": 0.002828,
      "loss": 1.1031,
      "step": 529728
    },
    {
      "epoch": 39.75924953095685,
      "grad_norm": 0.7395573854446411,
      "learning_rate": 0.002828,
      "loss": 1.0994,
      "step": 529792
    },
    {
      "epoch": 39.76405253283302,
      "grad_norm": 0.6626569628715515,
      "learning_rate": 0.002828,
      "loss": 1.1132,
      "step": 529856
    },
    {
      "epoch": 39.768855534709196,
      "grad_norm": 0.596015989780426,
      "learning_rate": 0.002828,
      "loss": 1.1109,
      "step": 529920
    },
    {
      "epoch": 39.773658536585366,
      "grad_norm": 0.6616296768188477,
      "learning_rate": 0.002828,
      "loss": 1.1065,
      "step": 529984
    },
    {
      "epoch": 39.778461538461535,
      "grad_norm": 0.6416288614273071,
      "learning_rate": 0.002828,
      "loss": 1.1039,
      "step": 530048
    },
    {
      "epoch": 39.78326454033771,
      "grad_norm": 0.6856739521026611,
      "learning_rate": 0.002828,
      "loss": 1.1038,
      "step": 530112
    },
    {
      "epoch": 39.78806754221388,
      "grad_norm": 0.5775629878044128,
      "learning_rate": 0.002828,
      "loss": 1.1103,
      "step": 530176
    },
    {
      "epoch": 39.79287054409006,
      "grad_norm": 0.8429564237594604,
      "learning_rate": 0.002828,
      "loss": 1.1096,
      "step": 530240
    },
    {
      "epoch": 39.79767354596623,
      "grad_norm": 0.571973443031311,
      "learning_rate": 0.002828,
      "loss": 1.1081,
      "step": 530304
    },
    {
      "epoch": 39.8024765478424,
      "grad_norm": 0.9216853380203247,
      "learning_rate": 0.002828,
      "loss": 1.1028,
      "step": 530368
    },
    {
      "epoch": 39.80727954971857,
      "grad_norm": 0.6850279569625854,
      "learning_rate": 0.002828,
      "loss": 1.1081,
      "step": 530432
    },
    {
      "epoch": 39.81208255159475,
      "grad_norm": 0.5708699226379395,
      "learning_rate": 0.002828,
      "loss": 1.1084,
      "step": 530496
    },
    {
      "epoch": 39.81688555347092,
      "grad_norm": 0.655859112739563,
      "learning_rate": 0.002828,
      "loss": 1.1068,
      "step": 530560
    },
    {
      "epoch": 39.821688555347095,
      "grad_norm": 0.5955530405044556,
      "learning_rate": 0.002828,
      "loss": 1.1093,
      "step": 530624
    },
    {
      "epoch": 39.826491557223264,
      "grad_norm": 0.587763249874115,
      "learning_rate": 0.002828,
      "loss": 1.1067,
      "step": 530688
    },
    {
      "epoch": 39.83129455909944,
      "grad_norm": 0.6162307858467102,
      "learning_rate": 0.002828,
      "loss": 1.1025,
      "step": 530752
    },
    {
      "epoch": 39.83609756097561,
      "grad_norm": 0.6205756664276123,
      "learning_rate": 0.002828,
      "loss": 1.1049,
      "step": 530816
    },
    {
      "epoch": 39.84090056285178,
      "grad_norm": 0.5409461855888367,
      "learning_rate": 0.002828,
      "loss": 1.1084,
      "step": 530880
    },
    {
      "epoch": 39.845703564727955,
      "grad_norm": 1.0385624170303345,
      "learning_rate": 0.002828,
      "loss": 1.1074,
      "step": 530944
    },
    {
      "epoch": 39.850506566604125,
      "grad_norm": 0.6885086297988892,
      "learning_rate": 0.002828,
      "loss": 1.113,
      "step": 531008
    },
    {
      "epoch": 39.8553095684803,
      "grad_norm": 0.694943904876709,
      "learning_rate": 0.002828,
      "loss": 1.11,
      "step": 531072
    },
    {
      "epoch": 39.86011257035647,
      "grad_norm": 0.7824503779411316,
      "learning_rate": 0.002828,
      "loss": 1.1064,
      "step": 531136
    },
    {
      "epoch": 39.86491557223265,
      "grad_norm": 0.562761127948761,
      "learning_rate": 0.002828,
      "loss": 1.1019,
      "step": 531200
    },
    {
      "epoch": 39.869718574108816,
      "grad_norm": 0.5979791283607483,
      "learning_rate": 0.002828,
      "loss": 1.1135,
      "step": 531264
    },
    {
      "epoch": 39.87452157598499,
      "grad_norm": 0.5886710286140442,
      "learning_rate": 0.002828,
      "loss": 1.1056,
      "step": 531328
    },
    {
      "epoch": 39.87932457786116,
      "grad_norm": 0.6354160904884338,
      "learning_rate": 0.002828,
      "loss": 1.1073,
      "step": 531392
    },
    {
      "epoch": 39.88412757973734,
      "grad_norm": 0.7226580381393433,
      "learning_rate": 0.002828,
      "loss": 1.1098,
      "step": 531456
    },
    {
      "epoch": 39.88893058161351,
      "grad_norm": 0.6297502517700195,
      "learning_rate": 0.002828,
      "loss": 1.1076,
      "step": 531520
    },
    {
      "epoch": 39.893733583489684,
      "grad_norm": 0.5878957509994507,
      "learning_rate": 0.002828,
      "loss": 1.1136,
      "step": 531584
    },
    {
      "epoch": 39.898536585365854,
      "grad_norm": 0.7568423748016357,
      "learning_rate": 0.002828,
      "loss": 1.1085,
      "step": 531648
    },
    {
      "epoch": 39.90333958724202,
      "grad_norm": 0.5480137467384338,
      "learning_rate": 0.002828,
      "loss": 1.1044,
      "step": 531712
    },
    {
      "epoch": 39.9081425891182,
      "grad_norm": 0.5894506573677063,
      "learning_rate": 0.002828,
      "loss": 1.1104,
      "step": 531776
    },
    {
      "epoch": 39.91294559099437,
      "grad_norm": 0.6330196261405945,
      "learning_rate": 0.002828,
      "loss": 1.1087,
      "step": 531840
    },
    {
      "epoch": 39.917748592870545,
      "grad_norm": 0.7971309423446655,
      "learning_rate": 0.002828,
      "loss": 1.106,
      "step": 531904
    },
    {
      "epoch": 39.922551594746714,
      "grad_norm": 0.6329578757286072,
      "learning_rate": 0.002828,
      "loss": 1.1129,
      "step": 531968
    },
    {
      "epoch": 39.92735459662289,
      "grad_norm": 0.5038153529167175,
      "learning_rate": 0.002828,
      "loss": 1.1095,
      "step": 532032
    },
    {
      "epoch": 39.93215759849906,
      "grad_norm": 0.6094999313354492,
      "learning_rate": 0.002828,
      "loss": 1.1125,
      "step": 532096
    },
    {
      "epoch": 39.93696060037524,
      "grad_norm": 0.5079559087753296,
      "learning_rate": 0.002828,
      "loss": 1.1029,
      "step": 532160
    },
    {
      "epoch": 39.941763602251406,
      "grad_norm": 0.55616694688797,
      "learning_rate": 0.002828,
      "loss": 1.1044,
      "step": 532224
    },
    {
      "epoch": 39.94656660412758,
      "grad_norm": 0.5412519574165344,
      "learning_rate": 0.002828,
      "loss": 1.1061,
      "step": 532288
    },
    {
      "epoch": 39.95136960600375,
      "grad_norm": 0.6757495403289795,
      "learning_rate": 0.002828,
      "loss": 1.1099,
      "step": 532352
    },
    {
      "epoch": 39.95617260787993,
      "grad_norm": 0.6678786873817444,
      "learning_rate": 0.002828,
      "loss": 1.1036,
      "step": 532416
    },
    {
      "epoch": 39.9609756097561,
      "grad_norm": 0.6607096791267395,
      "learning_rate": 0.002828,
      "loss": 1.1037,
      "step": 532480
    },
    {
      "epoch": 39.96577861163227,
      "grad_norm": 0.5916298031806946,
      "learning_rate": 0.002828,
      "loss": 1.1094,
      "step": 532544
    },
    {
      "epoch": 39.97058161350844,
      "grad_norm": 0.6358587145805359,
      "learning_rate": 0.002828,
      "loss": 1.1145,
      "step": 532608
    },
    {
      "epoch": 39.97538461538461,
      "grad_norm": 0.6469662189483643,
      "learning_rate": 0.002828,
      "loss": 1.1119,
      "step": 532672
    },
    {
      "epoch": 39.98018761726079,
      "grad_norm": 0.6851943731307983,
      "learning_rate": 0.002828,
      "loss": 1.1107,
      "step": 532736
    },
    {
      "epoch": 39.98499061913696,
      "grad_norm": 0.6589726209640503,
      "learning_rate": 0.002828,
      "loss": 1.1127,
      "step": 532800
    },
    {
      "epoch": 39.989793621013135,
      "grad_norm": 0.5190383791923523,
      "learning_rate": 0.002828,
      "loss": 1.1059,
      "step": 532864
    },
    {
      "epoch": 39.994596622889304,
      "grad_norm": 0.5913875699043274,
      "learning_rate": 0.002828,
      "loss": 1.107,
      "step": 532928
    },
    {
      "epoch": 39.99939962476548,
      "grad_norm": 0.6283708810806274,
      "learning_rate": 0.002828,
      "loss": 1.1123,
      "step": 532992
    },
    {
      "epoch": 40.00420262664165,
      "grad_norm": 0.5936568379402161,
      "learning_rate": 0.002828,
      "loss": 1.0796,
      "step": 533056
    },
    {
      "epoch": 40.009005628517826,
      "grad_norm": 0.6228441596031189,
      "learning_rate": 0.002828,
      "loss": 1.0738,
      "step": 533120
    },
    {
      "epoch": 40.013808630393996,
      "grad_norm": 0.5124711394309998,
      "learning_rate": 0.002828,
      "loss": 1.0717,
      "step": 533184
    },
    {
      "epoch": 40.01861163227017,
      "grad_norm": 0.7940961718559265,
      "learning_rate": 0.002828,
      "loss": 1.0734,
      "step": 533248
    },
    {
      "epoch": 40.02341463414634,
      "grad_norm": 0.5992420315742493,
      "learning_rate": 0.002828,
      "loss": 1.0736,
      "step": 533312
    },
    {
      "epoch": 40.02821763602251,
      "grad_norm": 0.5343099236488342,
      "learning_rate": 0.002828,
      "loss": 1.071,
      "step": 533376
    },
    {
      "epoch": 40.03302063789869,
      "grad_norm": 0.5241178274154663,
      "learning_rate": 0.002828,
      "loss": 1.074,
      "step": 533440
    },
    {
      "epoch": 40.03782363977486,
      "grad_norm": 0.7192960977554321,
      "learning_rate": 0.002828,
      "loss": 1.0745,
      "step": 533504
    },
    {
      "epoch": 40.04262664165103,
      "grad_norm": 0.7904562950134277,
      "learning_rate": 0.002828,
      "loss": 1.0774,
      "step": 533568
    },
    {
      "epoch": 40.0474296435272,
      "grad_norm": 0.760181725025177,
      "learning_rate": 0.002828,
      "loss": 1.0721,
      "step": 533632
    },
    {
      "epoch": 40.05223264540338,
      "grad_norm": 0.5834547877311707,
      "learning_rate": 0.002828,
      "loss": 1.074,
      "step": 533696
    },
    {
      "epoch": 40.05703564727955,
      "grad_norm": 0.5778083205223083,
      "learning_rate": 0.002828,
      "loss": 1.0751,
      "step": 533760
    },
    {
      "epoch": 40.061838649155725,
      "grad_norm": 0.6584703326225281,
      "learning_rate": 0.002828,
      "loss": 1.0745,
      "step": 533824
    },
    {
      "epoch": 40.066641651031894,
      "grad_norm": 0.5253800749778748,
      "learning_rate": 0.002828,
      "loss": 1.0711,
      "step": 533888
    },
    {
      "epoch": 40.07144465290807,
      "grad_norm": 0.49245938658714294,
      "learning_rate": 0.002828,
      "loss": 1.0791,
      "step": 533952
    },
    {
      "epoch": 40.07624765478424,
      "grad_norm": 0.6337112188339233,
      "learning_rate": 0.002828,
      "loss": 1.0807,
      "step": 534016
    },
    {
      "epoch": 40.081050656660416,
      "grad_norm": 0.7335975170135498,
      "learning_rate": 0.002828,
      "loss": 1.0733,
      "step": 534080
    },
    {
      "epoch": 40.085853658536585,
      "grad_norm": 0.5771627426147461,
      "learning_rate": 0.002828,
      "loss": 1.0753,
      "step": 534144
    },
    {
      "epoch": 40.090656660412755,
      "grad_norm": 0.6392300724983215,
      "learning_rate": 0.002828,
      "loss": 1.081,
      "step": 534208
    },
    {
      "epoch": 40.09545966228893,
      "grad_norm": 0.5753986239433289,
      "learning_rate": 0.002828,
      "loss": 1.0789,
      "step": 534272
    },
    {
      "epoch": 40.1002626641651,
      "grad_norm": 0.5396976470947266,
      "learning_rate": 0.002828,
      "loss": 1.0792,
      "step": 534336
    },
    {
      "epoch": 40.10506566604128,
      "grad_norm": 0.621372401714325,
      "learning_rate": 0.002828,
      "loss": 1.079,
      "step": 534400
    },
    {
      "epoch": 40.109868667917446,
      "grad_norm": 0.6752663850784302,
      "learning_rate": 0.002828,
      "loss": 1.081,
      "step": 534464
    },
    {
      "epoch": 40.11467166979362,
      "grad_norm": 0.6094245910644531,
      "learning_rate": 0.002828,
      "loss": 1.0864,
      "step": 534528
    },
    {
      "epoch": 40.11947467166979,
      "grad_norm": 0.6214587092399597,
      "learning_rate": 0.002828,
      "loss": 1.0774,
      "step": 534592
    },
    {
      "epoch": 40.12427767354597,
      "grad_norm": 0.6940475106239319,
      "learning_rate": 0.002828,
      "loss": 1.0802,
      "step": 534656
    },
    {
      "epoch": 40.12908067542214,
      "grad_norm": 0.5746902823448181,
      "learning_rate": 0.002828,
      "loss": 1.079,
      "step": 534720
    },
    {
      "epoch": 40.133883677298314,
      "grad_norm": 0.7482483983039856,
      "learning_rate": 0.002828,
      "loss": 1.0785,
      "step": 534784
    },
    {
      "epoch": 40.138686679174484,
      "grad_norm": 0.6373165845870972,
      "learning_rate": 0.002828,
      "loss": 1.0808,
      "step": 534848
    },
    {
      "epoch": 40.14348968105066,
      "grad_norm": 0.6416199207305908,
      "learning_rate": 0.002828,
      "loss": 1.08,
      "step": 534912
    },
    {
      "epoch": 40.14829268292683,
      "grad_norm": 0.7983708381652832,
      "learning_rate": 0.002828,
      "loss": 1.0809,
      "step": 534976
    },
    {
      "epoch": 40.153095684803,
      "grad_norm": 0.6383090615272522,
      "learning_rate": 0.002828,
      "loss": 1.0832,
      "step": 535040
    },
    {
      "epoch": 40.157898686679175,
      "grad_norm": 0.5542482733726501,
      "learning_rate": 0.002828,
      "loss": 1.0791,
      "step": 535104
    },
    {
      "epoch": 40.162701688555345,
      "grad_norm": 0.5444024801254272,
      "learning_rate": 0.002828,
      "loss": 1.0781,
      "step": 535168
    },
    {
      "epoch": 40.16750469043152,
      "grad_norm": 0.5228294134140015,
      "learning_rate": 0.002828,
      "loss": 1.0819,
      "step": 535232
    },
    {
      "epoch": 40.17230769230769,
      "grad_norm": 0.5700907707214355,
      "learning_rate": 0.002828,
      "loss": 1.084,
      "step": 535296
    },
    {
      "epoch": 40.17711069418387,
      "grad_norm": 0.7073768377304077,
      "learning_rate": 0.002828,
      "loss": 1.0811,
      "step": 535360
    },
    {
      "epoch": 40.181913696060036,
      "grad_norm": 0.5951655507087708,
      "learning_rate": 0.002828,
      "loss": 1.0865,
      "step": 535424
    },
    {
      "epoch": 40.18671669793621,
      "grad_norm": 0.5553143620491028,
      "learning_rate": 0.002828,
      "loss": 1.0865,
      "step": 535488
    },
    {
      "epoch": 40.19151969981238,
      "grad_norm": 0.6570543646812439,
      "learning_rate": 0.002828,
      "loss": 1.084,
      "step": 535552
    },
    {
      "epoch": 40.19632270168856,
      "grad_norm": 0.555578887462616,
      "learning_rate": 0.002828,
      "loss": 1.0806,
      "step": 535616
    },
    {
      "epoch": 40.20112570356473,
      "grad_norm": 0.5957068800926208,
      "learning_rate": 0.002828,
      "loss": 1.0834,
      "step": 535680
    },
    {
      "epoch": 40.205928705440904,
      "grad_norm": 0.660434365272522,
      "learning_rate": 0.002828,
      "loss": 1.0789,
      "step": 535744
    },
    {
      "epoch": 40.21073170731707,
      "grad_norm": 0.616719126701355,
      "learning_rate": 0.002828,
      "loss": 1.0808,
      "step": 535808
    },
    {
      "epoch": 40.21553470919324,
      "grad_norm": 0.7140281200408936,
      "learning_rate": 0.002828,
      "loss": 1.0857,
      "step": 535872
    },
    {
      "epoch": 40.22033771106942,
      "grad_norm": 0.7330237627029419,
      "learning_rate": 0.002828,
      "loss": 1.0897,
      "step": 535936
    },
    {
      "epoch": 40.22514071294559,
      "grad_norm": 0.6709555387496948,
      "learning_rate": 0.002828,
      "loss": 1.0825,
      "step": 536000
    },
    {
      "epoch": 40.229943714821765,
      "grad_norm": 0.591925859451294,
      "learning_rate": 0.002828,
      "loss": 1.0851,
      "step": 536064
    },
    {
      "epoch": 40.234746716697934,
      "grad_norm": 0.520167350769043,
      "learning_rate": 0.002828,
      "loss": 1.079,
      "step": 536128
    },
    {
      "epoch": 40.23954971857411,
      "grad_norm": 0.5892237424850464,
      "learning_rate": 0.002828,
      "loss": 1.0828,
      "step": 536192
    },
    {
      "epoch": 40.24435272045028,
      "grad_norm": 0.5957344770431519,
      "learning_rate": 0.002828,
      "loss": 1.0843,
      "step": 536256
    },
    {
      "epoch": 40.249155722326456,
      "grad_norm": 0.5429134368896484,
      "learning_rate": 0.002828,
      "loss": 1.0875,
      "step": 536320
    },
    {
      "epoch": 40.253958724202626,
      "grad_norm": 0.6112959384918213,
      "learning_rate": 0.002828,
      "loss": 1.0886,
      "step": 536384
    },
    {
      "epoch": 40.2587617260788,
      "grad_norm": 0.5770933032035828,
      "learning_rate": 0.002828,
      "loss": 1.086,
      "step": 536448
    },
    {
      "epoch": 40.26356472795497,
      "grad_norm": 0.6509702801704407,
      "learning_rate": 0.002828,
      "loss": 1.0912,
      "step": 536512
    },
    {
      "epoch": 40.26836772983115,
      "grad_norm": 0.5557456016540527,
      "learning_rate": 0.002828,
      "loss": 1.0845,
      "step": 536576
    },
    {
      "epoch": 40.27317073170732,
      "grad_norm": 0.5438624024391174,
      "learning_rate": 0.002828,
      "loss": 1.0823,
      "step": 536640
    },
    {
      "epoch": 40.27797373358349,
      "grad_norm": 0.6417461037635803,
      "learning_rate": 0.002828,
      "loss": 1.0856,
      "step": 536704
    },
    {
      "epoch": 40.28277673545966,
      "grad_norm": 0.5874984860420227,
      "learning_rate": 0.002828,
      "loss": 1.0885,
      "step": 536768
    },
    {
      "epoch": 40.28757973733583,
      "grad_norm": 0.6954288482666016,
      "learning_rate": 0.002828,
      "loss": 1.0822,
      "step": 536832
    },
    {
      "epoch": 40.29238273921201,
      "grad_norm": 0.5061184167861938,
      "learning_rate": 0.002828,
      "loss": 1.078,
      "step": 536896
    },
    {
      "epoch": 40.29718574108818,
      "grad_norm": 0.7347505688667297,
      "learning_rate": 0.002828,
      "loss": 1.0901,
      "step": 536960
    },
    {
      "epoch": 40.301988742964355,
      "grad_norm": 0.6875437498092651,
      "learning_rate": 0.002828,
      "loss": 1.09,
      "step": 537024
    },
    {
      "epoch": 40.306791744840524,
      "grad_norm": 0.7001727819442749,
      "learning_rate": 0.002828,
      "loss": 1.087,
      "step": 537088
    },
    {
      "epoch": 40.3115947467167,
      "grad_norm": 0.7775195240974426,
      "learning_rate": 0.002828,
      "loss": 1.0962,
      "step": 537152
    },
    {
      "epoch": 40.31639774859287,
      "grad_norm": 0.6886974573135376,
      "learning_rate": 0.002828,
      "loss": 1.0867,
      "step": 537216
    },
    {
      "epoch": 40.321200750469046,
      "grad_norm": 0.6425182223320007,
      "learning_rate": 0.002828,
      "loss": 1.0902,
      "step": 537280
    },
    {
      "epoch": 40.326003752345216,
      "grad_norm": 0.6771668195724487,
      "learning_rate": 0.002828,
      "loss": 1.0861,
      "step": 537344
    },
    {
      "epoch": 40.330806754221385,
      "grad_norm": 0.6500647664070129,
      "learning_rate": 0.002828,
      "loss": 1.0899,
      "step": 537408
    },
    {
      "epoch": 40.33560975609756,
      "grad_norm": 0.6931434273719788,
      "learning_rate": 0.002828,
      "loss": 1.088,
      "step": 537472
    },
    {
      "epoch": 40.34041275797373,
      "grad_norm": 0.7521113753318787,
      "learning_rate": 0.002828,
      "loss": 1.0854,
      "step": 537536
    },
    {
      "epoch": 40.34521575984991,
      "grad_norm": 0.7151188850402832,
      "learning_rate": 0.002828,
      "loss": 1.0876,
      "step": 537600
    },
    {
      "epoch": 40.350018761726076,
      "grad_norm": 0.5576344132423401,
      "learning_rate": 0.002828,
      "loss": 1.0917,
      "step": 537664
    },
    {
      "epoch": 40.35482176360225,
      "grad_norm": 0.6216950416564941,
      "learning_rate": 0.002828,
      "loss": 1.0861,
      "step": 537728
    },
    {
      "epoch": 40.35962476547842,
      "grad_norm": 0.6059309840202332,
      "learning_rate": 0.002828,
      "loss": 1.0857,
      "step": 537792
    },
    {
      "epoch": 40.3644277673546,
      "grad_norm": 0.5524047613143921,
      "learning_rate": 0.002828,
      "loss": 1.0887,
      "step": 537856
    },
    {
      "epoch": 40.36923076923077,
      "grad_norm": 0.6283742189407349,
      "learning_rate": 0.002828,
      "loss": 1.0864,
      "step": 537920
    },
    {
      "epoch": 40.374033771106944,
      "grad_norm": 0.7623336911201477,
      "learning_rate": 0.002828,
      "loss": 1.087,
      "step": 537984
    },
    {
      "epoch": 40.378836772983114,
      "grad_norm": 0.5618724226951599,
      "learning_rate": 0.002828,
      "loss": 1.0896,
      "step": 538048
    },
    {
      "epoch": 40.38363977485929,
      "grad_norm": 0.5732715129852295,
      "learning_rate": 0.002828,
      "loss": 1.0899,
      "step": 538112
    },
    {
      "epoch": 40.38844277673546,
      "grad_norm": 0.607079267501831,
      "learning_rate": 0.002828,
      "loss": 1.0862,
      "step": 538176
    },
    {
      "epoch": 40.39324577861163,
      "grad_norm": 0.6852363348007202,
      "learning_rate": 0.002828,
      "loss": 1.0849,
      "step": 538240
    },
    {
      "epoch": 40.398048780487805,
      "grad_norm": 0.7007236480712891,
      "learning_rate": 0.002828,
      "loss": 1.0916,
      "step": 538304
    },
    {
      "epoch": 40.402851782363975,
      "grad_norm": 0.5718969106674194,
      "learning_rate": 0.002828,
      "loss": 1.0869,
      "step": 538368
    },
    {
      "epoch": 40.40765478424015,
      "grad_norm": 0.6075987219810486,
      "learning_rate": 0.002828,
      "loss": 1.0912,
      "step": 538432
    },
    {
      "epoch": 40.41245778611632,
      "grad_norm": 0.6271284222602844,
      "learning_rate": 0.002828,
      "loss": 1.0908,
      "step": 538496
    },
    {
      "epoch": 40.4172607879925,
      "grad_norm": 0.7049885988235474,
      "learning_rate": 0.002828,
      "loss": 1.0912,
      "step": 538560
    },
    {
      "epoch": 40.422063789868666,
      "grad_norm": 0.6199770569801331,
      "learning_rate": 0.002828,
      "loss": 1.0889,
      "step": 538624
    },
    {
      "epoch": 40.42686679174484,
      "grad_norm": 0.6718481183052063,
      "learning_rate": 0.002828,
      "loss": 1.0894,
      "step": 538688
    },
    {
      "epoch": 40.43166979362101,
      "grad_norm": 0.8460226655006409,
      "learning_rate": 0.002828,
      "loss": 1.0894,
      "step": 538752
    },
    {
      "epoch": 40.43647279549719,
      "grad_norm": 0.5630996227264404,
      "learning_rate": 0.002828,
      "loss": 1.097,
      "step": 538816
    },
    {
      "epoch": 40.44127579737336,
      "grad_norm": 0.580254077911377,
      "learning_rate": 0.002828,
      "loss": 1.0858,
      "step": 538880
    },
    {
      "epoch": 40.446078799249534,
      "grad_norm": 0.5695968270301819,
      "learning_rate": 0.002828,
      "loss": 1.0947,
      "step": 538944
    },
    {
      "epoch": 40.4508818011257,
      "grad_norm": 0.6187861561775208,
      "learning_rate": 0.002828,
      "loss": 1.0981,
      "step": 539008
    },
    {
      "epoch": 40.45568480300187,
      "grad_norm": 0.5952305793762207,
      "learning_rate": 0.002828,
      "loss": 1.0931,
      "step": 539072
    },
    {
      "epoch": 40.46048780487805,
      "grad_norm": 0.5588722229003906,
      "learning_rate": 0.002828,
      "loss": 1.0858,
      "step": 539136
    },
    {
      "epoch": 40.46529080675422,
      "grad_norm": 0.5876528024673462,
      "learning_rate": 0.002828,
      "loss": 1.0847,
      "step": 539200
    },
    {
      "epoch": 40.470093808630395,
      "grad_norm": 0.9118528962135315,
      "learning_rate": 0.002828,
      "loss": 1.0904,
      "step": 539264
    },
    {
      "epoch": 40.474896810506564,
      "grad_norm": 0.5637339949607849,
      "learning_rate": 0.002828,
      "loss": 1.0855,
      "step": 539328
    },
    {
      "epoch": 40.47969981238274,
      "grad_norm": 0.6483489274978638,
      "learning_rate": 0.002828,
      "loss": 1.0969,
      "step": 539392
    },
    {
      "epoch": 40.48450281425891,
      "grad_norm": 0.8003821969032288,
      "learning_rate": 0.002828,
      "loss": 1.0917,
      "step": 539456
    },
    {
      "epoch": 40.48930581613509,
      "grad_norm": 0.6287664175033569,
      "learning_rate": 0.002828,
      "loss": 1.0893,
      "step": 539520
    },
    {
      "epoch": 40.494108818011256,
      "grad_norm": 0.6048157215118408,
      "learning_rate": 0.002828,
      "loss": 1.0967,
      "step": 539584
    },
    {
      "epoch": 40.49891181988743,
      "grad_norm": 0.6489182710647583,
      "learning_rate": 0.002828,
      "loss": 1.0941,
      "step": 539648
    },
    {
      "epoch": 40.5037148217636,
      "grad_norm": 0.6671035885810852,
      "learning_rate": 0.002828,
      "loss": 1.0936,
      "step": 539712
    },
    {
      "epoch": 40.50851782363978,
      "grad_norm": 0.5299566984176636,
      "learning_rate": 0.002828,
      "loss": 1.0953,
      "step": 539776
    },
    {
      "epoch": 40.51332082551595,
      "grad_norm": 0.653409481048584,
      "learning_rate": 0.002828,
      "loss": 1.0909,
      "step": 539840
    },
    {
      "epoch": 40.51812382739212,
      "grad_norm": 0.6193459033966064,
      "learning_rate": 0.002828,
      "loss": 1.0925,
      "step": 539904
    },
    {
      "epoch": 40.52292682926829,
      "grad_norm": 0.5770994424819946,
      "learning_rate": 0.002828,
      "loss": 1.0909,
      "step": 539968
    },
    {
      "epoch": 40.52772983114446,
      "grad_norm": 0.5876451730728149,
      "learning_rate": 0.002828,
      "loss": 1.0905,
      "step": 540032
    },
    {
      "epoch": 40.53253283302064,
      "grad_norm": 0.6242850422859192,
      "learning_rate": 0.002828,
      "loss": 1.0878,
      "step": 540096
    },
    {
      "epoch": 40.53733583489681,
      "grad_norm": 0.6436592936515808,
      "learning_rate": 0.002828,
      "loss": 1.0934,
      "step": 540160
    },
    {
      "epoch": 40.542138836772985,
      "grad_norm": 0.6318769454956055,
      "learning_rate": 0.002828,
      "loss": 1.0946,
      "step": 540224
    },
    {
      "epoch": 40.546941838649154,
      "grad_norm": 0.6540458798408508,
      "learning_rate": 0.002828,
      "loss": 1.0924,
      "step": 540288
    },
    {
      "epoch": 40.55174484052533,
      "grad_norm": 0.6689574718475342,
      "learning_rate": 0.002828,
      "loss": 1.0954,
      "step": 540352
    },
    {
      "epoch": 40.5565478424015,
      "grad_norm": 0.5450055003166199,
      "learning_rate": 0.002828,
      "loss": 1.092,
      "step": 540416
    },
    {
      "epoch": 40.561350844277676,
      "grad_norm": 0.5304721593856812,
      "learning_rate": 0.002828,
      "loss": 1.0933,
      "step": 540480
    },
    {
      "epoch": 40.566153846153846,
      "grad_norm": 0.676292896270752,
      "learning_rate": 0.002828,
      "loss": 1.0944,
      "step": 540544
    },
    {
      "epoch": 40.57095684803002,
      "grad_norm": 0.877575159072876,
      "learning_rate": 0.002828,
      "loss": 1.0927,
      "step": 540608
    },
    {
      "epoch": 40.57575984990619,
      "grad_norm": 0.5048443675041199,
      "learning_rate": 0.002828,
      "loss": 1.0995,
      "step": 540672
    },
    {
      "epoch": 40.58056285178236,
      "grad_norm": 0.6128528118133545,
      "learning_rate": 0.002828,
      "loss": 1.0891,
      "step": 540736
    },
    {
      "epoch": 40.58536585365854,
      "grad_norm": 0.606031596660614,
      "learning_rate": 0.002828,
      "loss": 1.0974,
      "step": 540800
    },
    {
      "epoch": 40.590168855534706,
      "grad_norm": 0.8159749507904053,
      "learning_rate": 0.002828,
      "loss": 1.0903,
      "step": 540864
    },
    {
      "epoch": 40.59497185741088,
      "grad_norm": 0.5860930681228638,
      "learning_rate": 0.002828,
      "loss": 1.0918,
      "step": 540928
    },
    {
      "epoch": 40.59977485928705,
      "grad_norm": 0.5595285892486572,
      "learning_rate": 0.002828,
      "loss": 1.0936,
      "step": 540992
    },
    {
      "epoch": 40.60457786116323,
      "grad_norm": 0.7063301205635071,
      "learning_rate": 0.002828,
      "loss": 1.0919,
      "step": 541056
    },
    {
      "epoch": 40.6093808630394,
      "grad_norm": 0.5304481983184814,
      "learning_rate": 0.002828,
      "loss": 1.0892,
      "step": 541120
    },
    {
      "epoch": 40.614183864915574,
      "grad_norm": 0.7145516872406006,
      "learning_rate": 0.002828,
      "loss": 1.096,
      "step": 541184
    },
    {
      "epoch": 40.618986866791744,
      "grad_norm": 0.5920206904411316,
      "learning_rate": 0.002828,
      "loss": 1.0949,
      "step": 541248
    },
    {
      "epoch": 40.62378986866792,
      "grad_norm": 0.6007336378097534,
      "learning_rate": 0.002828,
      "loss": 1.0978,
      "step": 541312
    },
    {
      "epoch": 40.62859287054409,
      "grad_norm": 0.6843711137771606,
      "learning_rate": 0.002828,
      "loss": 1.1003,
      "step": 541376
    },
    {
      "epoch": 40.633395872420266,
      "grad_norm": 0.566557765007019,
      "learning_rate": 0.002828,
      "loss": 1.0945,
      "step": 541440
    },
    {
      "epoch": 40.638198874296435,
      "grad_norm": 0.5469267964363098,
      "learning_rate": 0.002828,
      "loss": 1.0921,
      "step": 541504
    },
    {
      "epoch": 40.643001876172605,
      "grad_norm": 0.6030517816543579,
      "learning_rate": 0.002828,
      "loss": 1.0944,
      "step": 541568
    },
    {
      "epoch": 40.64780487804878,
      "grad_norm": 0.740609347820282,
      "learning_rate": 0.002828,
      "loss": 1.0956,
      "step": 541632
    },
    {
      "epoch": 40.65260787992495,
      "grad_norm": 0.6031094789505005,
      "learning_rate": 0.002828,
      "loss": 1.1013,
      "step": 541696
    },
    {
      "epoch": 40.65741088180113,
      "grad_norm": 0.5991085767745972,
      "learning_rate": 0.002828,
      "loss": 1.095,
      "step": 541760
    },
    {
      "epoch": 40.662213883677296,
      "grad_norm": 0.6071657538414001,
      "learning_rate": 0.002828,
      "loss": 1.0972,
      "step": 541824
    },
    {
      "epoch": 40.66701688555347,
      "grad_norm": 0.6647424101829529,
      "learning_rate": 0.002828,
      "loss": 1.0956,
      "step": 541888
    },
    {
      "epoch": 40.67181988742964,
      "grad_norm": 0.6323827505111694,
      "learning_rate": 0.002828,
      "loss": 1.1023,
      "step": 541952
    },
    {
      "epoch": 40.67662288930582,
      "grad_norm": 0.6064316034317017,
      "learning_rate": 0.002828,
      "loss": 1.0975,
      "step": 542016
    },
    {
      "epoch": 40.68142589118199,
      "grad_norm": 0.515994131565094,
      "learning_rate": 0.002828,
      "loss": 1.1038,
      "step": 542080
    },
    {
      "epoch": 40.686228893058164,
      "grad_norm": 0.5427857637405396,
      "learning_rate": 0.002828,
      "loss": 1.0952,
      "step": 542144
    },
    {
      "epoch": 40.69103189493433,
      "grad_norm": 0.6527212858200073,
      "learning_rate": 0.002828,
      "loss": 1.0949,
      "step": 542208
    },
    {
      "epoch": 40.69583489681051,
      "grad_norm": 0.6460388898849487,
      "learning_rate": 0.002828,
      "loss": 1.0974,
      "step": 542272
    },
    {
      "epoch": 40.70063789868668,
      "grad_norm": 0.6885424852371216,
      "learning_rate": 0.002828,
      "loss": 1.0957,
      "step": 542336
    },
    {
      "epoch": 40.70544090056285,
      "grad_norm": 0.6290729641914368,
      "learning_rate": 0.002828,
      "loss": 1.0961,
      "step": 542400
    },
    {
      "epoch": 40.710243902439025,
      "grad_norm": 0.5251750349998474,
      "learning_rate": 0.002828,
      "loss": 1.0976,
      "step": 542464
    },
    {
      "epoch": 40.715046904315194,
      "grad_norm": 0.5548868775367737,
      "learning_rate": 0.002828,
      "loss": 1.0948,
      "step": 542528
    },
    {
      "epoch": 40.71984990619137,
      "grad_norm": 0.6257069110870361,
      "learning_rate": 0.002828,
      "loss": 1.099,
      "step": 542592
    },
    {
      "epoch": 40.72465290806754,
      "grad_norm": 0.5366240739822388,
      "learning_rate": 0.002828,
      "loss": 1.091,
      "step": 542656
    },
    {
      "epoch": 40.72945590994372,
      "grad_norm": 0.7957614660263062,
      "learning_rate": 0.002828,
      "loss": 1.1006,
      "step": 542720
    },
    {
      "epoch": 40.734258911819886,
      "grad_norm": 0.5318620204925537,
      "learning_rate": 0.002828,
      "loss": 1.1019,
      "step": 542784
    },
    {
      "epoch": 40.73906191369606,
      "grad_norm": 0.6072466969490051,
      "learning_rate": 0.002828,
      "loss": 1.1016,
      "step": 542848
    },
    {
      "epoch": 40.74386491557223,
      "grad_norm": 0.6516512632369995,
      "learning_rate": 0.002828,
      "loss": 1.1004,
      "step": 542912
    },
    {
      "epoch": 40.74866791744841,
      "grad_norm": 0.6152677536010742,
      "learning_rate": 0.002828,
      "loss": 1.0987,
      "step": 542976
    },
    {
      "epoch": 40.75347091932458,
      "grad_norm": 0.7000126242637634,
      "learning_rate": 0.002828,
      "loss": 1.0917,
      "step": 543040
    },
    {
      "epoch": 40.758273921200754,
      "grad_norm": 0.7065082788467407,
      "learning_rate": 0.002828,
      "loss": 1.0984,
      "step": 543104
    },
    {
      "epoch": 40.76307692307692,
      "grad_norm": 0.5523930191993713,
      "learning_rate": 0.002828,
      "loss": 1.0957,
      "step": 543168
    },
    {
      "epoch": 40.76787992495309,
      "grad_norm": 0.4815172255039215,
      "learning_rate": 0.002828,
      "loss": 1.0991,
      "step": 543232
    },
    {
      "epoch": 40.77268292682927,
      "grad_norm": 0.768750011920929,
      "learning_rate": 0.002828,
      "loss": 1.0962,
      "step": 543296
    },
    {
      "epoch": 40.77748592870544,
      "grad_norm": 0.5711806416511536,
      "learning_rate": 0.002828,
      "loss": 1.1019,
      "step": 543360
    },
    {
      "epoch": 40.782288930581615,
      "grad_norm": 0.7046821713447571,
      "learning_rate": 0.002828,
      "loss": 1.0976,
      "step": 543424
    },
    {
      "epoch": 40.787091932457784,
      "grad_norm": 0.7322063446044922,
      "learning_rate": 0.002828,
      "loss": 1.0948,
      "step": 543488
    },
    {
      "epoch": 40.79189493433396,
      "grad_norm": 0.5857354402542114,
      "learning_rate": 0.002828,
      "loss": 1.106,
      "step": 543552
    },
    {
      "epoch": 40.79669793621013,
      "grad_norm": 0.6323885917663574,
      "learning_rate": 0.002828,
      "loss": 1.0987,
      "step": 543616
    },
    {
      "epoch": 40.801500938086306,
      "grad_norm": 0.6436946392059326,
      "learning_rate": 0.002828,
      "loss": 1.104,
      "step": 543680
    },
    {
      "epoch": 40.806303939962476,
      "grad_norm": 0.7924228310585022,
      "learning_rate": 0.002828,
      "loss": 1.0973,
      "step": 543744
    },
    {
      "epoch": 40.81110694183865,
      "grad_norm": 0.5841827988624573,
      "learning_rate": 0.002828,
      "loss": 1.0949,
      "step": 543808
    },
    {
      "epoch": 40.81590994371482,
      "grad_norm": 0.5505257248878479,
      "learning_rate": 0.002828,
      "loss": 1.1047,
      "step": 543872
    },
    {
      "epoch": 40.820712945591,
      "grad_norm": 0.7143750190734863,
      "learning_rate": 0.002828,
      "loss": 1.0979,
      "step": 543936
    },
    {
      "epoch": 40.82551594746717,
      "grad_norm": 0.7817658185958862,
      "learning_rate": 0.002828,
      "loss": 1.1134,
      "step": 544000
    },
    {
      "epoch": 40.83031894934334,
      "grad_norm": 0.7498055100440979,
      "learning_rate": 0.002828,
      "loss": 1.1031,
      "step": 544064
    },
    {
      "epoch": 40.83512195121951,
      "grad_norm": 0.6902478933334351,
      "learning_rate": 0.002828,
      "loss": 1.0984,
      "step": 544128
    },
    {
      "epoch": 40.83992495309568,
      "grad_norm": 0.7068681120872498,
      "learning_rate": 0.002828,
      "loss": 1.1016,
      "step": 544192
    },
    {
      "epoch": 40.84472795497186,
      "grad_norm": 0.6595316529273987,
      "learning_rate": 0.002828,
      "loss": 1.1013,
      "step": 544256
    },
    {
      "epoch": 40.84953095684803,
      "grad_norm": 0.5899179577827454,
      "learning_rate": 0.002828,
      "loss": 1.0953,
      "step": 544320
    },
    {
      "epoch": 40.854333958724204,
      "grad_norm": 0.5795630216598511,
      "learning_rate": 0.002828,
      "loss": 1.0997,
      "step": 544384
    },
    {
      "epoch": 40.859136960600374,
      "grad_norm": 0.5591762661933899,
      "learning_rate": 0.002828,
      "loss": 1.0958,
      "step": 544448
    },
    {
      "epoch": 40.86393996247655,
      "grad_norm": 0.6017462611198425,
      "learning_rate": 0.002828,
      "loss": 1.1009,
      "step": 544512
    },
    {
      "epoch": 40.86874296435272,
      "grad_norm": 0.4877956509590149,
      "learning_rate": 0.002828,
      "loss": 1.0972,
      "step": 544576
    },
    {
      "epoch": 40.873545966228896,
      "grad_norm": 0.6430007219314575,
      "learning_rate": 0.002828,
      "loss": 1.098,
      "step": 544640
    },
    {
      "epoch": 40.878348968105065,
      "grad_norm": 0.6140983700752258,
      "learning_rate": 0.002828,
      "loss": 1.1026,
      "step": 544704
    },
    {
      "epoch": 40.88315196998124,
      "grad_norm": 0.5751563310623169,
      "learning_rate": 0.002828,
      "loss": 1.0949,
      "step": 544768
    },
    {
      "epoch": 40.88795497185741,
      "grad_norm": 1.0069924592971802,
      "learning_rate": 0.002828,
      "loss": 1.0938,
      "step": 544832
    },
    {
      "epoch": 40.89275797373358,
      "grad_norm": 0.5524895787239075,
      "learning_rate": 0.002828,
      "loss": 1.0963,
      "step": 544896
    },
    {
      "epoch": 40.89756097560976,
      "grad_norm": 0.550719678401947,
      "learning_rate": 0.002828,
      "loss": 1.1027,
      "step": 544960
    },
    {
      "epoch": 40.902363977485926,
      "grad_norm": 0.5717636346817017,
      "learning_rate": 0.002828,
      "loss": 1.1007,
      "step": 545024
    },
    {
      "epoch": 40.9071669793621,
      "grad_norm": 1.6790406703948975,
      "learning_rate": 0.002828,
      "loss": 1.0998,
      "step": 545088
    },
    {
      "epoch": 40.91196998123827,
      "grad_norm": 0.6707130670547485,
      "learning_rate": 0.002828,
      "loss": 1.0965,
      "step": 545152
    },
    {
      "epoch": 40.91677298311445,
      "grad_norm": 0.6442863941192627,
      "learning_rate": 0.002828,
      "loss": 1.1022,
      "step": 545216
    },
    {
      "epoch": 40.92157598499062,
      "grad_norm": 0.6229715347290039,
      "learning_rate": 0.002828,
      "loss": 1.1009,
      "step": 545280
    },
    {
      "epoch": 40.926378986866794,
      "grad_norm": 0.649745523929596,
      "learning_rate": 0.002828,
      "loss": 1.1047,
      "step": 545344
    },
    {
      "epoch": 40.93118198874296,
      "grad_norm": 0.5942361354827881,
      "learning_rate": 0.002828,
      "loss": 1.099,
      "step": 545408
    },
    {
      "epoch": 40.93598499061914,
      "grad_norm": 0.6187747120857239,
      "learning_rate": 0.002828,
      "loss": 1.1,
      "step": 545472
    },
    {
      "epoch": 40.94078799249531,
      "grad_norm": 0.7702499628067017,
      "learning_rate": 0.002828,
      "loss": 1.1031,
      "step": 545536
    },
    {
      "epoch": 40.945590994371486,
      "grad_norm": 0.6279484033584595,
      "learning_rate": 0.002828,
      "loss": 1.1026,
      "step": 545600
    },
    {
      "epoch": 40.950393996247655,
      "grad_norm": 0.5844237208366394,
      "learning_rate": 0.002828,
      "loss": 1.103,
      "step": 545664
    },
    {
      "epoch": 40.955196998123824,
      "grad_norm": 0.5983730554580688,
      "learning_rate": 0.002828,
      "loss": 1.0975,
      "step": 545728
    },
    {
      "epoch": 40.96,
      "grad_norm": 0.5255396366119385,
      "learning_rate": 0.002828,
      "loss": 1.1032,
      "step": 545792
    },
    {
      "epoch": 40.96480300187617,
      "grad_norm": 0.651672899723053,
      "learning_rate": 0.002828,
      "loss": 1.0965,
      "step": 545856
    },
    {
      "epoch": 40.96960600375235,
      "grad_norm": 0.6900015473365784,
      "learning_rate": 0.002828,
      "loss": 1.0953,
      "step": 545920
    },
    {
      "epoch": 40.974409005628516,
      "grad_norm": 0.6562929153442383,
      "learning_rate": 0.002828,
      "loss": 1.1003,
      "step": 545984
    },
    {
      "epoch": 40.97921200750469,
      "grad_norm": 0.5922935605049133,
      "learning_rate": 0.002828,
      "loss": 1.1015,
      "step": 546048
    },
    {
      "epoch": 40.98401500938086,
      "grad_norm": 0.6217307448387146,
      "learning_rate": 0.002828,
      "loss": 1.1096,
      "step": 546112
    },
    {
      "epoch": 40.98881801125704,
      "grad_norm": 0.6182146072387695,
      "learning_rate": 0.002828,
      "loss": 1.0996,
      "step": 546176
    },
    {
      "epoch": 40.99362101313321,
      "grad_norm": 0.5867896676063538,
      "learning_rate": 0.002828,
      "loss": 1.102,
      "step": 546240
    },
    {
      "epoch": 40.998424015009384,
      "grad_norm": 0.6607248187065125,
      "learning_rate": 0.002828,
      "loss": 1.1067,
      "step": 546304
    },
    {
      "epoch": 41.00322701688555,
      "grad_norm": 0.5025272965431213,
      "learning_rate": 0.002828,
      "loss": 1.0785,
      "step": 546368
    },
    {
      "epoch": 41.00803001876172,
      "grad_norm": 0.5673875212669373,
      "learning_rate": 0.002828,
      "loss": 1.0593,
      "step": 546432
    },
    {
      "epoch": 41.0128330206379,
      "grad_norm": 0.6235575079917908,
      "learning_rate": 0.002828,
      "loss": 1.0606,
      "step": 546496
    },
    {
      "epoch": 41.01763602251407,
      "grad_norm": 0.6856295466423035,
      "learning_rate": 0.002828,
      "loss": 1.0666,
      "step": 546560
    },
    {
      "epoch": 41.022439024390245,
      "grad_norm": 0.591202437877655,
      "learning_rate": 0.002828,
      "loss": 1.0644,
      "step": 546624
    },
    {
      "epoch": 41.027242026266414,
      "grad_norm": 0.6456746459007263,
      "learning_rate": 0.002828,
      "loss": 1.068,
      "step": 546688
    },
    {
      "epoch": 41.03204502814259,
      "grad_norm": 0.8349180221557617,
      "learning_rate": 0.002828,
      "loss": 1.0669,
      "step": 546752
    },
    {
      "epoch": 41.03684803001876,
      "grad_norm": 0.6748512387275696,
      "learning_rate": 0.002828,
      "loss": 1.0691,
      "step": 546816
    },
    {
      "epoch": 41.041651031894936,
      "grad_norm": 0.6579387187957764,
      "learning_rate": 0.002828,
      "loss": 1.0695,
      "step": 546880
    },
    {
      "epoch": 41.046454033771106,
      "grad_norm": 0.6043445467948914,
      "learning_rate": 0.002828,
      "loss": 1.0627,
      "step": 546944
    },
    {
      "epoch": 41.05125703564728,
      "grad_norm": 0.853763997554779,
      "learning_rate": 0.002828,
      "loss": 1.0677,
      "step": 547008
    },
    {
      "epoch": 41.05606003752345,
      "grad_norm": 0.703504741191864,
      "learning_rate": 0.002828,
      "loss": 1.067,
      "step": 547072
    },
    {
      "epoch": 41.06086303939963,
      "grad_norm": 0.5170365571975708,
      "learning_rate": 0.002828,
      "loss": 1.0711,
      "step": 547136
    },
    {
      "epoch": 41.0656660412758,
      "grad_norm": 0.7496975660324097,
      "learning_rate": 0.002828,
      "loss": 1.0742,
      "step": 547200
    },
    {
      "epoch": 41.07046904315197,
      "grad_norm": 0.5645685195922852,
      "learning_rate": 0.002828,
      "loss": 1.0664,
      "step": 547264
    },
    {
      "epoch": 41.07527204502814,
      "grad_norm": 0.6629078388214111,
      "learning_rate": 0.002828,
      "loss": 1.0731,
      "step": 547328
    },
    {
      "epoch": 41.08007504690431,
      "grad_norm": 0.6933609247207642,
      "learning_rate": 0.002828,
      "loss": 1.0642,
      "step": 547392
    },
    {
      "epoch": 41.08487804878049,
      "grad_norm": 0.5706798434257507,
      "learning_rate": 0.002828,
      "loss": 1.0701,
      "step": 547456
    },
    {
      "epoch": 41.08968105065666,
      "grad_norm": 0.7231749892234802,
      "learning_rate": 0.002828,
      "loss": 1.0729,
      "step": 547520
    },
    {
      "epoch": 41.094484052532835,
      "grad_norm": 0.6114044189453125,
      "learning_rate": 0.002828,
      "loss": 1.074,
      "step": 547584
    },
    {
      "epoch": 41.099287054409004,
      "grad_norm": 0.5615615248680115,
      "learning_rate": 0.002828,
      "loss": 1.08,
      "step": 547648
    },
    {
      "epoch": 41.10409005628518,
      "grad_norm": 0.5425843000411987,
      "learning_rate": 0.002828,
      "loss": 1.071,
      "step": 547712
    },
    {
      "epoch": 41.10889305816135,
      "grad_norm": 0.72618168592453,
      "learning_rate": 0.002828,
      "loss": 1.0716,
      "step": 547776
    },
    {
      "epoch": 41.113696060037526,
      "grad_norm": 0.7147377133369446,
      "learning_rate": 0.002828,
      "loss": 1.0764,
      "step": 547840
    },
    {
      "epoch": 41.118499061913695,
      "grad_norm": 0.6635823249816895,
      "learning_rate": 0.002828,
      "loss": 1.0703,
      "step": 547904
    },
    {
      "epoch": 41.12330206378987,
      "grad_norm": 0.724740743637085,
      "learning_rate": 0.002828,
      "loss": 1.0793,
      "step": 547968
    },
    {
      "epoch": 41.12810506566604,
      "grad_norm": 0.6134108901023865,
      "learning_rate": 0.002828,
      "loss": 1.0714,
      "step": 548032
    },
    {
      "epoch": 41.13290806754221,
      "grad_norm": 0.6556316018104553,
      "learning_rate": 0.002828,
      "loss": 1.0735,
      "step": 548096
    },
    {
      "epoch": 41.13771106941839,
      "grad_norm": 0.567468523979187,
      "learning_rate": 0.002828,
      "loss": 1.0748,
      "step": 548160
    },
    {
      "epoch": 41.142514071294556,
      "grad_norm": 0.4976537227630615,
      "learning_rate": 0.002828,
      "loss": 1.0669,
      "step": 548224
    },
    {
      "epoch": 41.14731707317073,
      "grad_norm": 0.5294303297996521,
      "learning_rate": 0.002828,
      "loss": 1.0752,
      "step": 548288
    },
    {
      "epoch": 41.1521200750469,
      "grad_norm": 0.7013516426086426,
      "learning_rate": 0.002828,
      "loss": 1.0757,
      "step": 548352
    },
    {
      "epoch": 41.15692307692308,
      "grad_norm": 0.803352415561676,
      "learning_rate": 0.002828,
      "loss": 1.0707,
      "step": 548416
    },
    {
      "epoch": 41.16172607879925,
      "grad_norm": 0.711155891418457,
      "learning_rate": 0.002828,
      "loss": 1.0781,
      "step": 548480
    },
    {
      "epoch": 41.166529080675424,
      "grad_norm": 0.6899096965789795,
      "learning_rate": 0.002828,
      "loss": 1.0741,
      "step": 548544
    },
    {
      "epoch": 41.171332082551594,
      "grad_norm": 0.713313639163971,
      "learning_rate": 0.002828,
      "loss": 1.074,
      "step": 548608
    },
    {
      "epoch": 41.17613508442777,
      "grad_norm": 0.6300371289253235,
      "learning_rate": 0.002828,
      "loss": 1.0737,
      "step": 548672
    },
    {
      "epoch": 41.18093808630394,
      "grad_norm": 0.5804073810577393,
      "learning_rate": 0.002828,
      "loss": 1.0755,
      "step": 548736
    },
    {
      "epoch": 41.185741088180116,
      "grad_norm": 0.5882118940353394,
      "learning_rate": 0.002828,
      "loss": 1.076,
      "step": 548800
    },
    {
      "epoch": 41.190544090056285,
      "grad_norm": 0.6168975830078125,
      "learning_rate": 0.002828,
      "loss": 1.0765,
      "step": 548864
    },
    {
      "epoch": 41.195347091932454,
      "grad_norm": 0.7686378955841064,
      "learning_rate": 0.002828,
      "loss": 1.08,
      "step": 548928
    },
    {
      "epoch": 41.20015009380863,
      "grad_norm": 0.6872912049293518,
      "learning_rate": 0.002828,
      "loss": 1.0768,
      "step": 548992
    },
    {
      "epoch": 41.2049530956848,
      "grad_norm": 0.6079062819480896,
      "learning_rate": 0.002828,
      "loss": 1.076,
      "step": 549056
    },
    {
      "epoch": 41.20975609756098,
      "grad_norm": 0.6001817584037781,
      "learning_rate": 0.002828,
      "loss": 1.0793,
      "step": 549120
    },
    {
      "epoch": 41.214559099437146,
      "grad_norm": 0.8148412108421326,
      "learning_rate": 0.002828,
      "loss": 1.0822,
      "step": 549184
    },
    {
      "epoch": 41.21936210131332,
      "grad_norm": 0.6967695951461792,
      "learning_rate": 0.002828,
      "loss": 1.078,
      "step": 549248
    },
    {
      "epoch": 41.22416510318949,
      "grad_norm": 0.5821467638015747,
      "learning_rate": 0.002828,
      "loss": 1.0762,
      "step": 549312
    },
    {
      "epoch": 41.22896810506567,
      "grad_norm": 0.592621386051178,
      "learning_rate": 0.002828,
      "loss": 1.0737,
      "step": 549376
    },
    {
      "epoch": 41.23377110694184,
      "grad_norm": 0.5450983643531799,
      "learning_rate": 0.002828,
      "loss": 1.078,
      "step": 549440
    },
    {
      "epoch": 41.238574108818014,
      "grad_norm": 0.6808087229728699,
      "learning_rate": 0.002828,
      "loss": 1.0754,
      "step": 549504
    },
    {
      "epoch": 41.24337711069418,
      "grad_norm": 0.8498914837837219,
      "learning_rate": 0.002828,
      "loss": 1.0768,
      "step": 549568
    },
    {
      "epoch": 41.24818011257036,
      "grad_norm": 0.6844555139541626,
      "learning_rate": 0.002828,
      "loss": 1.0779,
      "step": 549632
    },
    {
      "epoch": 41.25298311444653,
      "grad_norm": 0.5792409181594849,
      "learning_rate": 0.002828,
      "loss": 1.0793,
      "step": 549696
    },
    {
      "epoch": 41.2577861163227,
      "grad_norm": 0.7109765410423279,
      "learning_rate": 0.002828,
      "loss": 1.0782,
      "step": 549760
    },
    {
      "epoch": 41.262589118198875,
      "grad_norm": 0.5593913793563843,
      "learning_rate": 0.002828,
      "loss": 1.0805,
      "step": 549824
    },
    {
      "epoch": 41.267392120075044,
      "grad_norm": 0.6583130955696106,
      "learning_rate": 0.002828,
      "loss": 1.0802,
      "step": 549888
    },
    {
      "epoch": 41.27219512195122,
      "grad_norm": 0.6874505877494812,
      "learning_rate": 0.002828,
      "loss": 1.0821,
      "step": 549952
    },
    {
      "epoch": 41.27699812382739,
      "grad_norm": 0.6103010177612305,
      "learning_rate": 0.002828,
      "loss": 1.0768,
      "step": 550016
    },
    {
      "epoch": 41.281801125703566,
      "grad_norm": 0.507485568523407,
      "learning_rate": 0.002828,
      "loss": 1.0839,
      "step": 550080
    },
    {
      "epoch": 41.286604127579736,
      "grad_norm": 0.5909212231636047,
      "learning_rate": 0.002828,
      "loss": 1.0764,
      "step": 550144
    },
    {
      "epoch": 41.29140712945591,
      "grad_norm": 0.6002563834190369,
      "learning_rate": 0.002828,
      "loss": 1.0735,
      "step": 550208
    },
    {
      "epoch": 41.29621013133208,
      "grad_norm": 0.6409716010093689,
      "learning_rate": 0.002828,
      "loss": 1.077,
      "step": 550272
    },
    {
      "epoch": 41.30101313320826,
      "grad_norm": 0.534680187702179,
      "learning_rate": 0.002828,
      "loss": 1.0798,
      "step": 550336
    },
    {
      "epoch": 41.30581613508443,
      "grad_norm": 0.7306864261627197,
      "learning_rate": 0.002828,
      "loss": 1.0778,
      "step": 550400
    },
    {
      "epoch": 41.310619136960604,
      "grad_norm": 0.7213438153266907,
      "learning_rate": 0.002828,
      "loss": 1.0782,
      "step": 550464
    },
    {
      "epoch": 41.31542213883677,
      "grad_norm": 0.6151289343833923,
      "learning_rate": 0.002828,
      "loss": 1.0804,
      "step": 550528
    },
    {
      "epoch": 41.32022514071294,
      "grad_norm": 0.6030688881874084,
      "learning_rate": 0.002828,
      "loss": 1.0813,
      "step": 550592
    },
    {
      "epoch": 41.32502814258912,
      "grad_norm": 0.6703411340713501,
      "learning_rate": 0.002828,
      "loss": 1.0803,
      "step": 550656
    },
    {
      "epoch": 41.32983114446529,
      "grad_norm": 0.6406719088554382,
      "learning_rate": 0.002828,
      "loss": 1.0799,
      "step": 550720
    },
    {
      "epoch": 41.334634146341465,
      "grad_norm": 0.6375758647918701,
      "learning_rate": 0.002828,
      "loss": 1.0795,
      "step": 550784
    },
    {
      "epoch": 41.339437148217634,
      "grad_norm": 0.6271138191223145,
      "learning_rate": 0.002828,
      "loss": 1.0772,
      "step": 550848
    },
    {
      "epoch": 41.34424015009381,
      "grad_norm": 0.5871312618255615,
      "learning_rate": 0.002828,
      "loss": 1.0793,
      "step": 550912
    },
    {
      "epoch": 41.34904315196998,
      "grad_norm": 0.7420594692230225,
      "learning_rate": 0.002828,
      "loss": 1.0838,
      "step": 550976
    },
    {
      "epoch": 41.353846153846156,
      "grad_norm": 0.6966577172279358,
      "learning_rate": 0.002828,
      "loss": 1.0814,
      "step": 551040
    },
    {
      "epoch": 41.358649155722325,
      "grad_norm": 0.6194618940353394,
      "learning_rate": 0.002828,
      "loss": 1.0822,
      "step": 551104
    },
    {
      "epoch": 41.3634521575985,
      "grad_norm": 0.6598330140113831,
      "learning_rate": 0.002828,
      "loss": 1.0817,
      "step": 551168
    },
    {
      "epoch": 41.36825515947467,
      "grad_norm": 0.6324847340583801,
      "learning_rate": 0.002828,
      "loss": 1.0772,
      "step": 551232
    },
    {
      "epoch": 41.37305816135085,
      "grad_norm": 0.5739226937294006,
      "learning_rate": 0.002828,
      "loss": 1.0782,
      "step": 551296
    },
    {
      "epoch": 41.37786116322702,
      "grad_norm": 0.5579782724380493,
      "learning_rate": 0.002828,
      "loss": 1.0834,
      "step": 551360
    },
    {
      "epoch": 41.382664165103186,
      "grad_norm": 0.6816611886024475,
      "learning_rate": 0.002828,
      "loss": 1.0829,
      "step": 551424
    },
    {
      "epoch": 41.38746716697936,
      "grad_norm": 0.5585365891456604,
      "learning_rate": 0.002828,
      "loss": 1.0804,
      "step": 551488
    },
    {
      "epoch": 41.39227016885553,
      "grad_norm": 0.6159017086029053,
      "learning_rate": 0.002828,
      "loss": 1.0819,
      "step": 551552
    },
    {
      "epoch": 41.39707317073171,
      "grad_norm": 0.5967724323272705,
      "learning_rate": 0.002828,
      "loss": 1.0869,
      "step": 551616
    },
    {
      "epoch": 41.40187617260788,
      "grad_norm": 0.6096137166023254,
      "learning_rate": 0.002828,
      "loss": 1.0821,
      "step": 551680
    },
    {
      "epoch": 41.406679174484054,
      "grad_norm": 0.7553973197937012,
      "learning_rate": 0.002828,
      "loss": 1.0882,
      "step": 551744
    },
    {
      "epoch": 41.411482176360224,
      "grad_norm": 0.7201663255691528,
      "learning_rate": 0.002828,
      "loss": 1.0743,
      "step": 551808
    },
    {
      "epoch": 41.4162851782364,
      "grad_norm": 0.7154415845870972,
      "learning_rate": 0.002828,
      "loss": 1.0782,
      "step": 551872
    },
    {
      "epoch": 41.42108818011257,
      "grad_norm": 0.5563473105430603,
      "learning_rate": 0.002828,
      "loss": 1.082,
      "step": 551936
    },
    {
      "epoch": 41.425891181988746,
      "grad_norm": 0.7097707986831665,
      "learning_rate": 0.002828,
      "loss": 1.0813,
      "step": 552000
    },
    {
      "epoch": 41.430694183864915,
      "grad_norm": 0.6488261818885803,
      "learning_rate": 0.002828,
      "loss": 1.0801,
      "step": 552064
    },
    {
      "epoch": 41.43549718574109,
      "grad_norm": 0.5918857455253601,
      "learning_rate": 0.002828,
      "loss": 1.0844,
      "step": 552128
    },
    {
      "epoch": 41.44030018761726,
      "grad_norm": 0.6212255954742432,
      "learning_rate": 0.002828,
      "loss": 1.0822,
      "step": 552192
    },
    {
      "epoch": 41.44510318949343,
      "grad_norm": 0.696937620639801,
      "learning_rate": 0.002828,
      "loss": 1.0863,
      "step": 552256
    },
    {
      "epoch": 41.44990619136961,
      "grad_norm": 0.5609291195869446,
      "learning_rate": 0.002828,
      "loss": 1.0809,
      "step": 552320
    },
    {
      "epoch": 41.454709193245776,
      "grad_norm": 0.5481732487678528,
      "learning_rate": 0.002828,
      "loss": 1.0829,
      "step": 552384
    },
    {
      "epoch": 41.45951219512195,
      "grad_norm": 0.676578164100647,
      "learning_rate": 0.002828,
      "loss": 1.0877,
      "step": 552448
    },
    {
      "epoch": 41.46431519699812,
      "grad_norm": 0.5765472650527954,
      "learning_rate": 0.002828,
      "loss": 1.08,
      "step": 552512
    },
    {
      "epoch": 41.4691181988743,
      "grad_norm": 0.508776843547821,
      "learning_rate": 0.002828,
      "loss": 1.0824,
      "step": 552576
    },
    {
      "epoch": 41.47392120075047,
      "grad_norm": 0.6089415550231934,
      "learning_rate": 0.002828,
      "loss": 1.0807,
      "step": 552640
    },
    {
      "epoch": 41.478724202626644,
      "grad_norm": 0.7184882164001465,
      "learning_rate": 0.002828,
      "loss": 1.0858,
      "step": 552704
    },
    {
      "epoch": 41.48352720450281,
      "grad_norm": 0.5389273762702942,
      "learning_rate": 0.002828,
      "loss": 1.0862,
      "step": 552768
    },
    {
      "epoch": 41.48833020637899,
      "grad_norm": 0.5319380164146423,
      "learning_rate": 0.002828,
      "loss": 1.0869,
      "step": 552832
    },
    {
      "epoch": 41.49313320825516,
      "grad_norm": 0.6539042592048645,
      "learning_rate": 0.002828,
      "loss": 1.0829,
      "step": 552896
    },
    {
      "epoch": 41.497936210131336,
      "grad_norm": 0.7643285989761353,
      "learning_rate": 0.002828,
      "loss": 1.0806,
      "step": 552960
    },
    {
      "epoch": 41.502739212007505,
      "grad_norm": 0.6605548858642578,
      "learning_rate": 0.002828,
      "loss": 1.0857,
      "step": 553024
    },
    {
      "epoch": 41.507542213883674,
      "grad_norm": 0.5355969071388245,
      "learning_rate": 0.002828,
      "loss": 1.0899,
      "step": 553088
    },
    {
      "epoch": 41.51234521575985,
      "grad_norm": 0.6309624910354614,
      "learning_rate": 0.002828,
      "loss": 1.0881,
      "step": 553152
    },
    {
      "epoch": 41.51714821763602,
      "grad_norm": 0.7098450064659119,
      "learning_rate": 0.002828,
      "loss": 1.0886,
      "step": 553216
    },
    {
      "epoch": 41.5219512195122,
      "grad_norm": 0.6140933632850647,
      "learning_rate": 0.002828,
      "loss": 1.0859,
      "step": 553280
    },
    {
      "epoch": 41.526754221388366,
      "grad_norm": 0.6573895812034607,
      "learning_rate": 0.002828,
      "loss": 1.0836,
      "step": 553344
    },
    {
      "epoch": 41.53155722326454,
      "grad_norm": 0.642605721950531,
      "learning_rate": 0.002828,
      "loss": 1.0935,
      "step": 553408
    },
    {
      "epoch": 41.53636022514071,
      "grad_norm": 0.5573644638061523,
      "learning_rate": 0.002828,
      "loss": 1.0861,
      "step": 553472
    },
    {
      "epoch": 41.54116322701689,
      "grad_norm": 0.48514696955680847,
      "learning_rate": 0.002828,
      "loss": 1.0862,
      "step": 553536
    },
    {
      "epoch": 41.54596622889306,
      "grad_norm": 0.6432372331619263,
      "learning_rate": 0.002828,
      "loss": 1.0882,
      "step": 553600
    },
    {
      "epoch": 41.550769230769234,
      "grad_norm": 0.564486026763916,
      "learning_rate": 0.002828,
      "loss": 1.093,
      "step": 553664
    },
    {
      "epoch": 41.5555722326454,
      "grad_norm": 0.6009403467178345,
      "learning_rate": 0.002828,
      "loss": 1.0884,
      "step": 553728
    },
    {
      "epoch": 41.56037523452157,
      "grad_norm": 0.6792386770248413,
      "learning_rate": 0.002828,
      "loss": 1.0835,
      "step": 553792
    },
    {
      "epoch": 41.56517823639775,
      "grad_norm": 0.49415624141693115,
      "learning_rate": 0.002828,
      "loss": 1.0933,
      "step": 553856
    },
    {
      "epoch": 41.56998123827392,
      "grad_norm": 0.7057024836540222,
      "learning_rate": 0.002828,
      "loss": 1.0842,
      "step": 553920
    },
    {
      "epoch": 41.574784240150095,
      "grad_norm": 0.576983630657196,
      "learning_rate": 0.002828,
      "loss": 1.0857,
      "step": 553984
    },
    {
      "epoch": 41.579587242026264,
      "grad_norm": 0.6637498736381531,
      "learning_rate": 0.002828,
      "loss": 1.0828,
      "step": 554048
    },
    {
      "epoch": 41.58439024390244,
      "grad_norm": 0.7337366938591003,
      "learning_rate": 0.002828,
      "loss": 1.0884,
      "step": 554112
    },
    {
      "epoch": 41.58919324577861,
      "grad_norm": 0.6840522289276123,
      "learning_rate": 0.002828,
      "loss": 1.0846,
      "step": 554176
    },
    {
      "epoch": 41.593996247654786,
      "grad_norm": 0.5753034949302673,
      "learning_rate": 0.002828,
      "loss": 1.081,
      "step": 554240
    },
    {
      "epoch": 41.598799249530956,
      "grad_norm": 0.6625789999961853,
      "learning_rate": 0.002828,
      "loss": 1.0843,
      "step": 554304
    },
    {
      "epoch": 41.60360225140713,
      "grad_norm": 0.6237214207649231,
      "learning_rate": 0.002828,
      "loss": 1.088,
      "step": 554368
    },
    {
      "epoch": 41.6084052532833,
      "grad_norm": 0.6372871994972229,
      "learning_rate": 0.002828,
      "loss": 1.0848,
      "step": 554432
    },
    {
      "epoch": 41.61320825515948,
      "grad_norm": 0.6904455423355103,
      "learning_rate": 0.002828,
      "loss": 1.0899,
      "step": 554496
    },
    {
      "epoch": 41.61801125703565,
      "grad_norm": 0.7011419534683228,
      "learning_rate": 0.002828,
      "loss": 1.0889,
      "step": 554560
    },
    {
      "epoch": 41.622814258911816,
      "grad_norm": 0.620915412902832,
      "learning_rate": 0.002828,
      "loss": 1.0887,
      "step": 554624
    },
    {
      "epoch": 41.62761726078799,
      "grad_norm": 0.5954155921936035,
      "learning_rate": 0.002828,
      "loss": 1.0886,
      "step": 554688
    },
    {
      "epoch": 41.63242026266416,
      "grad_norm": 0.615436851978302,
      "learning_rate": 0.002828,
      "loss": 1.0932,
      "step": 554752
    },
    {
      "epoch": 41.63722326454034,
      "grad_norm": 0.6013243198394775,
      "learning_rate": 0.002828,
      "loss": 1.0903,
      "step": 554816
    },
    {
      "epoch": 41.64202626641651,
      "grad_norm": 0.5366150736808777,
      "learning_rate": 0.002828,
      "loss": 1.0909,
      "step": 554880
    },
    {
      "epoch": 41.646829268292684,
      "grad_norm": 0.5783130526542664,
      "learning_rate": 0.002828,
      "loss": 1.0903,
      "step": 554944
    },
    {
      "epoch": 41.651632270168854,
      "grad_norm": 0.7552841901779175,
      "learning_rate": 0.002828,
      "loss": 1.0856,
      "step": 555008
    },
    {
      "epoch": 41.65643527204503,
      "grad_norm": 0.6122512817382812,
      "learning_rate": 0.002828,
      "loss": 1.0872,
      "step": 555072
    },
    {
      "epoch": 41.6612382739212,
      "grad_norm": 0.5579005479812622,
      "learning_rate": 0.002828,
      "loss": 1.0885,
      "step": 555136
    },
    {
      "epoch": 41.666041275797376,
      "grad_norm": 0.6808589696884155,
      "learning_rate": 0.002828,
      "loss": 1.0896,
      "step": 555200
    },
    {
      "epoch": 41.670844277673545,
      "grad_norm": 0.6607319712638855,
      "learning_rate": 0.002828,
      "loss": 1.092,
      "step": 555264
    },
    {
      "epoch": 41.67564727954972,
      "grad_norm": 0.6016877889633179,
      "learning_rate": 0.002828,
      "loss": 1.0902,
      "step": 555328
    },
    {
      "epoch": 41.68045028142589,
      "grad_norm": 0.546603262424469,
      "learning_rate": 0.002828,
      "loss": 1.092,
      "step": 555392
    },
    {
      "epoch": 41.68525328330206,
      "grad_norm": 0.7585775852203369,
      "learning_rate": 0.002828,
      "loss": 1.0935,
      "step": 555456
    },
    {
      "epoch": 41.69005628517824,
      "grad_norm": 0.5713100433349609,
      "learning_rate": 0.002828,
      "loss": 1.0885,
      "step": 555520
    },
    {
      "epoch": 41.694859287054406,
      "grad_norm": 0.5373910069465637,
      "learning_rate": 0.002828,
      "loss": 1.0914,
      "step": 555584
    },
    {
      "epoch": 41.69966228893058,
      "grad_norm": 0.5792741775512695,
      "learning_rate": 0.002828,
      "loss": 1.0889,
      "step": 555648
    },
    {
      "epoch": 41.70446529080675,
      "grad_norm": 0.5292046070098877,
      "learning_rate": 0.002828,
      "loss": 1.0892,
      "step": 555712
    },
    {
      "epoch": 41.70926829268293,
      "grad_norm": 0.5693057179450989,
      "learning_rate": 0.002828,
      "loss": 1.085,
      "step": 555776
    },
    {
      "epoch": 41.7140712945591,
      "grad_norm": 0.7015483379364014,
      "learning_rate": 0.002828,
      "loss": 1.0875,
      "step": 555840
    },
    {
      "epoch": 41.718874296435274,
      "grad_norm": 0.508471667766571,
      "learning_rate": 0.002828,
      "loss": 1.0934,
      "step": 555904
    },
    {
      "epoch": 41.72367729831144,
      "grad_norm": 0.7722852826118469,
      "learning_rate": 0.002828,
      "loss": 1.0912,
      "step": 555968
    },
    {
      "epoch": 41.72848030018762,
      "grad_norm": 0.8116500973701477,
      "learning_rate": 0.002828,
      "loss": 1.0876,
      "step": 556032
    },
    {
      "epoch": 41.73328330206379,
      "grad_norm": 0.5903668999671936,
      "learning_rate": 0.002828,
      "loss": 1.0884,
      "step": 556096
    },
    {
      "epoch": 41.738086303939966,
      "grad_norm": 0.6233097910881042,
      "learning_rate": 0.002828,
      "loss": 1.0913,
      "step": 556160
    },
    {
      "epoch": 41.742889305816135,
      "grad_norm": 0.5414838194847107,
      "learning_rate": 0.002828,
      "loss": 1.0879,
      "step": 556224
    },
    {
      "epoch": 41.747692307692304,
      "grad_norm": 0.704927384853363,
      "learning_rate": 0.002828,
      "loss": 1.0885,
      "step": 556288
    },
    {
      "epoch": 41.75249530956848,
      "grad_norm": 0.6467387676239014,
      "learning_rate": 0.002828,
      "loss": 1.0921,
      "step": 556352
    },
    {
      "epoch": 41.75729831144465,
      "grad_norm": 0.5254771113395691,
      "learning_rate": 0.002828,
      "loss": 1.0931,
      "step": 556416
    },
    {
      "epoch": 41.76210131332083,
      "grad_norm": 0.5494102835655212,
      "learning_rate": 0.002828,
      "loss": 1.0928,
      "step": 556480
    },
    {
      "epoch": 41.766904315196996,
      "grad_norm": 0.7095513343811035,
      "learning_rate": 0.002828,
      "loss": 1.0889,
      "step": 556544
    },
    {
      "epoch": 41.77170731707317,
      "grad_norm": 0.5700022578239441,
      "learning_rate": 0.002828,
      "loss": 1.0934,
      "step": 556608
    },
    {
      "epoch": 41.77651031894934,
      "grad_norm": 0.7591579556465149,
      "learning_rate": 0.002828,
      "loss": 1.0907,
      "step": 556672
    },
    {
      "epoch": 41.78131332082552,
      "grad_norm": 0.5941113829612732,
      "learning_rate": 0.002828,
      "loss": 1.0973,
      "step": 556736
    },
    {
      "epoch": 41.78611632270169,
      "grad_norm": 0.5992494821548462,
      "learning_rate": 0.002828,
      "loss": 1.0896,
      "step": 556800
    },
    {
      "epoch": 41.790919324577864,
      "grad_norm": 0.585478663444519,
      "learning_rate": 0.002828,
      "loss": 1.0916,
      "step": 556864
    },
    {
      "epoch": 41.79572232645403,
      "grad_norm": 0.5099025964736938,
      "learning_rate": 0.002828,
      "loss": 1.0938,
      "step": 556928
    },
    {
      "epoch": 41.80052532833021,
      "grad_norm": 0.5065568685531616,
      "learning_rate": 0.002828,
      "loss": 1.0908,
      "step": 556992
    },
    {
      "epoch": 41.80532833020638,
      "grad_norm": 0.6000770926475525,
      "learning_rate": 0.002828,
      "loss": 1.0907,
      "step": 557056
    },
    {
      "epoch": 41.81013133208255,
      "grad_norm": 0.5981700420379639,
      "learning_rate": 0.002828,
      "loss": 1.0897,
      "step": 557120
    },
    {
      "epoch": 41.814934333958725,
      "grad_norm": 0.7779768705368042,
      "learning_rate": 0.002828,
      "loss": 1.0911,
      "step": 557184
    },
    {
      "epoch": 41.819737335834894,
      "grad_norm": 0.5827203989028931,
      "learning_rate": 0.002828,
      "loss": 1.0905,
      "step": 557248
    },
    {
      "epoch": 41.82454033771107,
      "grad_norm": 0.5996590256690979,
      "learning_rate": 0.002828,
      "loss": 1.0911,
      "step": 557312
    },
    {
      "epoch": 41.82934333958724,
      "grad_norm": 0.6872954964637756,
      "learning_rate": 0.002828,
      "loss": 1.0918,
      "step": 557376
    },
    {
      "epoch": 41.834146341463416,
      "grad_norm": 0.6303712725639343,
      "learning_rate": 0.002828,
      "loss": 1.0872,
      "step": 557440
    },
    {
      "epoch": 41.838949343339586,
      "grad_norm": 0.5935240387916565,
      "learning_rate": 0.002828,
      "loss": 1.0914,
      "step": 557504
    },
    {
      "epoch": 41.84375234521576,
      "grad_norm": 0.6666385531425476,
      "learning_rate": 0.002828,
      "loss": 1.0858,
      "step": 557568
    },
    {
      "epoch": 41.84855534709193,
      "grad_norm": 0.6222832202911377,
      "learning_rate": 0.002828,
      "loss": 1.0932,
      "step": 557632
    },
    {
      "epoch": 41.85335834896811,
      "grad_norm": 0.5753621459007263,
      "learning_rate": 0.002828,
      "loss": 1.0935,
      "step": 557696
    },
    {
      "epoch": 41.85816135084428,
      "grad_norm": 0.6276858448982239,
      "learning_rate": 0.002828,
      "loss": 1.0982,
      "step": 557760
    },
    {
      "epoch": 41.862964352720454,
      "grad_norm": 0.7073177695274353,
      "learning_rate": 0.002828,
      "loss": 1.0942,
      "step": 557824
    },
    {
      "epoch": 41.86776735459662,
      "grad_norm": 0.5554404854774475,
      "learning_rate": 0.002828,
      "loss": 1.0938,
      "step": 557888
    },
    {
      "epoch": 41.87257035647279,
      "grad_norm": 0.5417864918708801,
      "learning_rate": 0.002828,
      "loss": 1.0922,
      "step": 557952
    },
    {
      "epoch": 41.87737335834897,
      "grad_norm": 0.5264939665794373,
      "learning_rate": 0.002828,
      "loss": 1.0935,
      "step": 558016
    },
    {
      "epoch": 41.88217636022514,
      "grad_norm": 0.5799403190612793,
      "learning_rate": 0.002828,
      "loss": 1.0956,
      "step": 558080
    },
    {
      "epoch": 41.886979362101314,
      "grad_norm": 0.6113492250442505,
      "learning_rate": 0.002828,
      "loss": 1.0918,
      "step": 558144
    },
    {
      "epoch": 41.891782363977484,
      "grad_norm": 0.619025707244873,
      "learning_rate": 0.002828,
      "loss": 1.0907,
      "step": 558208
    },
    {
      "epoch": 41.89658536585366,
      "grad_norm": 0.6421979665756226,
      "learning_rate": 0.002828,
      "loss": 1.0933,
      "step": 558272
    },
    {
      "epoch": 41.90138836772983,
      "grad_norm": 0.5564460158348083,
      "learning_rate": 0.002828,
      "loss": 1.088,
      "step": 558336
    },
    {
      "epoch": 41.906191369606006,
      "grad_norm": 0.7520004510879517,
      "learning_rate": 0.002828,
      "loss": 1.0876,
      "step": 558400
    },
    {
      "epoch": 41.910994371482175,
      "grad_norm": 0.5554314851760864,
      "learning_rate": 0.002828,
      "loss": 1.0947,
      "step": 558464
    },
    {
      "epoch": 41.91579737335835,
      "grad_norm": 0.562497079372406,
      "learning_rate": 0.002828,
      "loss": 1.0915,
      "step": 558528
    },
    {
      "epoch": 41.92060037523452,
      "grad_norm": 0.5724356174468994,
      "learning_rate": 0.002828,
      "loss": 1.0886,
      "step": 558592
    },
    {
      "epoch": 41.9254033771107,
      "grad_norm": 0.5435642004013062,
      "learning_rate": 0.002828,
      "loss": 1.1026,
      "step": 558656
    },
    {
      "epoch": 41.93020637898687,
      "grad_norm": 0.7154808640480042,
      "learning_rate": 0.002828,
      "loss": 1.0907,
      "step": 558720
    },
    {
      "epoch": 41.935009380863036,
      "grad_norm": 0.6489702463150024,
      "learning_rate": 0.002828,
      "loss": 1.0937,
      "step": 558784
    },
    {
      "epoch": 41.93981238273921,
      "grad_norm": 0.5514934659004211,
      "learning_rate": 0.002828,
      "loss": 1.0975,
      "step": 558848
    },
    {
      "epoch": 41.94461538461538,
      "grad_norm": 0.838455855846405,
      "learning_rate": 0.002828,
      "loss": 1.096,
      "step": 558912
    },
    {
      "epoch": 41.94941838649156,
      "grad_norm": 0.5009222030639648,
      "learning_rate": 0.002828,
      "loss": 1.0892,
      "step": 558976
    },
    {
      "epoch": 41.95422138836773,
      "grad_norm": 0.5667208433151245,
      "learning_rate": 0.002828,
      "loss": 1.0963,
      "step": 559040
    },
    {
      "epoch": 41.959024390243904,
      "grad_norm": 0.6530715823173523,
      "learning_rate": 0.002828,
      "loss": 1.0912,
      "step": 559104
    },
    {
      "epoch": 41.96382739212007,
      "grad_norm": 0.621535062789917,
      "learning_rate": 0.002828,
      "loss": 1.0917,
      "step": 559168
    },
    {
      "epoch": 41.96863039399625,
      "grad_norm": 0.5420042872428894,
      "learning_rate": 0.002828,
      "loss": 1.0946,
      "step": 559232
    },
    {
      "epoch": 41.97343339587242,
      "grad_norm": 0.5837153792381287,
      "learning_rate": 0.002828,
      "loss": 1.0953,
      "step": 559296
    },
    {
      "epoch": 41.978236397748596,
      "grad_norm": 0.6675498485565186,
      "learning_rate": 0.002828,
      "loss": 1.0909,
      "step": 559360
    },
    {
      "epoch": 41.983039399624765,
      "grad_norm": 0.6540782451629639,
      "learning_rate": 0.002828,
      "loss": 1.0977,
      "step": 559424
    },
    {
      "epoch": 41.98784240150094,
      "grad_norm": 0.5713112354278564,
      "learning_rate": 0.002828,
      "loss": 1.0951,
      "step": 559488
    },
    {
      "epoch": 41.99264540337711,
      "grad_norm": 0.5297588109970093,
      "learning_rate": 0.002828,
      "loss": 1.0938,
      "step": 559552
    },
    {
      "epoch": 41.99744840525328,
      "grad_norm": 0.6510352492332458,
      "learning_rate": 0.002828,
      "loss": 1.0962,
      "step": 559616
    },
    {
      "epoch": 42.00225140712946,
      "grad_norm": 0.5886886715888977,
      "learning_rate": 0.002828,
      "loss": 1.0745,
      "step": 559680
    },
    {
      "epoch": 42.007054409005626,
      "grad_norm": 0.6688248515129089,
      "learning_rate": 0.002828,
      "loss": 1.0537,
      "step": 559744
    },
    {
      "epoch": 42.0118574108818,
      "grad_norm": 0.6012933850288391,
      "learning_rate": 0.002828,
      "loss": 1.0622,
      "step": 559808
    },
    {
      "epoch": 42.01666041275797,
      "grad_norm": 0.6174977421760559,
      "learning_rate": 0.002828,
      "loss": 1.0622,
      "step": 559872
    },
    {
      "epoch": 42.02146341463415,
      "grad_norm": 0.5445956587791443,
      "learning_rate": 0.002828,
      "loss": 1.057,
      "step": 559936
    },
    {
      "epoch": 42.02626641651032,
      "grad_norm": 0.6119956374168396,
      "learning_rate": 0.002828,
      "loss": 1.0605,
      "step": 560000
    },
    {
      "epoch": 42.031069418386494,
      "grad_norm": 0.6327849626541138,
      "learning_rate": 0.002828,
      "loss": 1.0595,
      "step": 560064
    },
    {
      "epoch": 42.03587242026266,
      "grad_norm": 0.6399878859519958,
      "learning_rate": 0.002828,
      "loss": 1.0573,
      "step": 560128
    },
    {
      "epoch": 42.04067542213884,
      "grad_norm": 0.6213284730911255,
      "learning_rate": 0.002828,
      "loss": 1.062,
      "step": 560192
    },
    {
      "epoch": 42.04547842401501,
      "grad_norm": 0.5296103358268738,
      "learning_rate": 0.002828,
      "loss": 1.0645,
      "step": 560256
    },
    {
      "epoch": 42.050281425891185,
      "grad_norm": 0.698832094669342,
      "learning_rate": 0.002828,
      "loss": 1.0657,
      "step": 560320
    },
    {
      "epoch": 42.055084427767355,
      "grad_norm": 0.6121209859848022,
      "learning_rate": 0.002828,
      "loss": 1.0576,
      "step": 560384
    },
    {
      "epoch": 42.059887429643524,
      "grad_norm": 0.6628225445747375,
      "learning_rate": 0.002828,
      "loss": 1.0587,
      "step": 560448
    },
    {
      "epoch": 42.0646904315197,
      "grad_norm": 0.522656261920929,
      "learning_rate": 0.002828,
      "loss": 1.0611,
      "step": 560512
    },
    {
      "epoch": 42.06949343339587,
      "grad_norm": 0.5090030431747437,
      "learning_rate": 0.002828,
      "loss": 1.0588,
      "step": 560576
    },
    {
      "epoch": 42.074296435272046,
      "grad_norm": 0.7143341898918152,
      "learning_rate": 0.002828,
      "loss": 1.0633,
      "step": 560640
    },
    {
      "epoch": 42.079099437148216,
      "grad_norm": 0.7170779705047607,
      "learning_rate": 0.002828,
      "loss": 1.0654,
      "step": 560704
    },
    {
      "epoch": 42.08390243902439,
      "grad_norm": 0.5964574217796326,
      "learning_rate": 0.002828,
      "loss": 1.0649,
      "step": 560768
    },
    {
      "epoch": 42.08870544090056,
      "grad_norm": 0.6003646850585938,
      "learning_rate": 0.002828,
      "loss": 1.0667,
      "step": 560832
    },
    {
      "epoch": 42.09350844277674,
      "grad_norm": 0.6014326214790344,
      "learning_rate": 0.002828,
      "loss": 1.0605,
      "step": 560896
    },
    {
      "epoch": 42.09831144465291,
      "grad_norm": 0.6906946301460266,
      "learning_rate": 0.002828,
      "loss": 1.0681,
      "step": 560960
    },
    {
      "epoch": 42.103114446529084,
      "grad_norm": 0.5337896347045898,
      "learning_rate": 0.002828,
      "loss": 1.0667,
      "step": 561024
    },
    {
      "epoch": 42.10791744840525,
      "grad_norm": 0.5389870405197144,
      "learning_rate": 0.002828,
      "loss": 1.0629,
      "step": 561088
    },
    {
      "epoch": 42.11272045028143,
      "grad_norm": 0.6632516980171204,
      "learning_rate": 0.002828,
      "loss": 1.0687,
      "step": 561152
    },
    {
      "epoch": 42.1175234521576,
      "grad_norm": 0.6831262707710266,
      "learning_rate": 0.002828,
      "loss": 1.0695,
      "step": 561216
    },
    {
      "epoch": 42.12232645403377,
      "grad_norm": 0.657807469367981,
      "learning_rate": 0.002828,
      "loss": 1.0654,
      "step": 561280
    },
    {
      "epoch": 42.127129455909945,
      "grad_norm": 0.5806097388267517,
      "learning_rate": 0.002828,
      "loss": 1.064,
      "step": 561344
    },
    {
      "epoch": 42.131932457786114,
      "grad_norm": 0.6003397107124329,
      "learning_rate": 0.002828,
      "loss": 1.0703,
      "step": 561408
    },
    {
      "epoch": 42.13673545966229,
      "grad_norm": 0.6153790950775146,
      "learning_rate": 0.002828,
      "loss": 1.0617,
      "step": 561472
    },
    {
      "epoch": 42.14153846153846,
      "grad_norm": 0.6227737069129944,
      "learning_rate": 0.002828,
      "loss": 1.0715,
      "step": 561536
    },
    {
      "epoch": 42.146341463414636,
      "grad_norm": 0.6207556128501892,
      "learning_rate": 0.002828,
      "loss": 1.0633,
      "step": 561600
    },
    {
      "epoch": 42.151144465290805,
      "grad_norm": 0.6176641583442688,
      "learning_rate": 0.002828,
      "loss": 1.0667,
      "step": 561664
    },
    {
      "epoch": 42.15594746716698,
      "grad_norm": 0.7300615310668945,
      "learning_rate": 0.002828,
      "loss": 1.0666,
      "step": 561728
    },
    {
      "epoch": 42.16075046904315,
      "grad_norm": 0.6501210331916809,
      "learning_rate": 0.002828,
      "loss": 1.0702,
      "step": 561792
    },
    {
      "epoch": 42.16555347091933,
      "grad_norm": 0.6231697201728821,
      "learning_rate": 0.002828,
      "loss": 1.0685,
      "step": 561856
    },
    {
      "epoch": 42.1703564727955,
      "grad_norm": 0.5661913156509399,
      "learning_rate": 0.002828,
      "loss": 1.069,
      "step": 561920
    },
    {
      "epoch": 42.175159474671666,
      "grad_norm": 0.6047375798225403,
      "learning_rate": 0.002828,
      "loss": 1.0628,
      "step": 561984
    },
    {
      "epoch": 42.17996247654784,
      "grad_norm": 0.6263827085494995,
      "learning_rate": 0.002828,
      "loss": 1.0689,
      "step": 562048
    },
    {
      "epoch": 42.18476547842401,
      "grad_norm": 0.5567244291305542,
      "learning_rate": 0.002828,
      "loss": 1.0702,
      "step": 562112
    },
    {
      "epoch": 42.18956848030019,
      "grad_norm": 0.5582324266433716,
      "learning_rate": 0.002828,
      "loss": 1.0701,
      "step": 562176
    },
    {
      "epoch": 42.19437148217636,
      "grad_norm": 0.6602678894996643,
      "learning_rate": 0.002828,
      "loss": 1.067,
      "step": 562240
    },
    {
      "epoch": 42.199174484052534,
      "grad_norm": 0.6724734902381897,
      "learning_rate": 0.002828,
      "loss": 1.0668,
      "step": 562304
    },
    {
      "epoch": 42.203977485928704,
      "grad_norm": 0.6203201413154602,
      "learning_rate": 0.002828,
      "loss": 1.0708,
      "step": 562368
    },
    {
      "epoch": 42.20878048780488,
      "grad_norm": 0.5524185299873352,
      "learning_rate": 0.002828,
      "loss": 1.066,
      "step": 562432
    },
    {
      "epoch": 42.21358348968105,
      "grad_norm": 0.7851992845535278,
      "learning_rate": 0.002828,
      "loss": 1.0666,
      "step": 562496
    },
    {
      "epoch": 42.218386491557226,
      "grad_norm": 0.5382575988769531,
      "learning_rate": 0.002828,
      "loss": 1.0748,
      "step": 562560
    },
    {
      "epoch": 42.223189493433395,
      "grad_norm": 0.631264865398407,
      "learning_rate": 0.002828,
      "loss": 1.069,
      "step": 562624
    },
    {
      "epoch": 42.22799249530957,
      "grad_norm": 0.6270348429679871,
      "learning_rate": 0.002828,
      "loss": 1.0713,
      "step": 562688
    },
    {
      "epoch": 42.23279549718574,
      "grad_norm": 0.7309645414352417,
      "learning_rate": 0.002828,
      "loss": 1.0723,
      "step": 562752
    },
    {
      "epoch": 42.23759849906191,
      "grad_norm": 0.6442151069641113,
      "learning_rate": 0.002828,
      "loss": 1.0686,
      "step": 562816
    },
    {
      "epoch": 42.24240150093809,
      "grad_norm": 0.6158406734466553,
      "learning_rate": 0.002828,
      "loss": 1.0712,
      "step": 562880
    },
    {
      "epoch": 42.247204502814256,
      "grad_norm": 0.5718555450439453,
      "learning_rate": 0.002828,
      "loss": 1.0751,
      "step": 562944
    },
    {
      "epoch": 42.25200750469043,
      "grad_norm": 0.6009352207183838,
      "learning_rate": 0.002828,
      "loss": 1.0723,
      "step": 563008
    },
    {
      "epoch": 42.2568105065666,
      "grad_norm": 0.648350715637207,
      "learning_rate": 0.002828,
      "loss": 1.0691,
      "step": 563072
    },
    {
      "epoch": 42.26161350844278,
      "grad_norm": 0.6446778178215027,
      "learning_rate": 0.002828,
      "loss": 1.0757,
      "step": 563136
    },
    {
      "epoch": 42.26641651031895,
      "grad_norm": 0.6527370810508728,
      "learning_rate": 0.002828,
      "loss": 1.0731,
      "step": 563200
    },
    {
      "epoch": 42.271219512195124,
      "grad_norm": 0.7441429495811462,
      "learning_rate": 0.002828,
      "loss": 1.0691,
      "step": 563264
    },
    {
      "epoch": 42.27602251407129,
      "grad_norm": 0.6982424855232239,
      "learning_rate": 0.002828,
      "loss": 1.0702,
      "step": 563328
    },
    {
      "epoch": 42.28082551594747,
      "grad_norm": 0.5313721299171448,
      "learning_rate": 0.002828,
      "loss": 1.0748,
      "step": 563392
    },
    {
      "epoch": 42.28562851782364,
      "grad_norm": 0.7577272057533264,
      "learning_rate": 0.002828,
      "loss": 1.0728,
      "step": 563456
    },
    {
      "epoch": 42.290431519699816,
      "grad_norm": 0.6524193286895752,
      "learning_rate": 0.002828,
      "loss": 1.0744,
      "step": 563520
    },
    {
      "epoch": 42.295234521575985,
      "grad_norm": 0.681130051612854,
      "learning_rate": 0.002828,
      "loss": 1.0771,
      "step": 563584
    },
    {
      "epoch": 42.300037523452154,
      "grad_norm": 0.5935671329498291,
      "learning_rate": 0.002828,
      "loss": 1.0698,
      "step": 563648
    },
    {
      "epoch": 42.30484052532833,
      "grad_norm": 0.7082101106643677,
      "learning_rate": 0.002828,
      "loss": 1.0705,
      "step": 563712
    },
    {
      "epoch": 42.3096435272045,
      "grad_norm": 0.7285563945770264,
      "learning_rate": 0.002828,
      "loss": 1.0719,
      "step": 563776
    },
    {
      "epoch": 42.314446529080676,
      "grad_norm": 0.6174675226211548,
      "learning_rate": 0.002828,
      "loss": 1.0679,
      "step": 563840
    },
    {
      "epoch": 42.319249530956846,
      "grad_norm": 0.7859613299369812,
      "learning_rate": 0.002828,
      "loss": 1.0709,
      "step": 563904
    },
    {
      "epoch": 42.32405253283302,
      "grad_norm": 0.6008791923522949,
      "learning_rate": 0.002828,
      "loss": 1.0693,
      "step": 563968
    },
    {
      "epoch": 42.32885553470919,
      "grad_norm": 0.7026347517967224,
      "learning_rate": 0.002828,
      "loss": 1.0731,
      "step": 564032
    },
    {
      "epoch": 42.33365853658537,
      "grad_norm": 0.569503128528595,
      "learning_rate": 0.002828,
      "loss": 1.0694,
      "step": 564096
    },
    {
      "epoch": 42.33846153846154,
      "grad_norm": 0.5854909420013428,
      "learning_rate": 0.002828,
      "loss": 1.0727,
      "step": 564160
    },
    {
      "epoch": 42.343264540337714,
      "grad_norm": 0.5679054260253906,
      "learning_rate": 0.002828,
      "loss": 1.0707,
      "step": 564224
    },
    {
      "epoch": 42.34806754221388,
      "grad_norm": 0.6394909620285034,
      "learning_rate": 0.002828,
      "loss": 1.0756,
      "step": 564288
    },
    {
      "epoch": 42.35287054409006,
      "grad_norm": 0.6406596899032593,
      "learning_rate": 0.002828,
      "loss": 1.0762,
      "step": 564352
    },
    {
      "epoch": 42.35767354596623,
      "grad_norm": 0.6130920648574829,
      "learning_rate": 0.002828,
      "loss": 1.0683,
      "step": 564416
    },
    {
      "epoch": 42.3624765478424,
      "grad_norm": 0.5922021865844727,
      "learning_rate": 0.002828,
      "loss": 1.0765,
      "step": 564480
    },
    {
      "epoch": 42.367279549718575,
      "grad_norm": 0.7147934436798096,
      "learning_rate": 0.002828,
      "loss": 1.0741,
      "step": 564544
    },
    {
      "epoch": 42.372082551594744,
      "grad_norm": 0.6006954908370972,
      "learning_rate": 0.002828,
      "loss": 1.0758,
      "step": 564608
    },
    {
      "epoch": 42.37688555347092,
      "grad_norm": 0.7138851881027222,
      "learning_rate": 0.002828,
      "loss": 1.0811,
      "step": 564672
    },
    {
      "epoch": 42.38168855534709,
      "grad_norm": 0.6090694069862366,
      "learning_rate": 0.002828,
      "loss": 1.0751,
      "step": 564736
    },
    {
      "epoch": 42.386491557223266,
      "grad_norm": 0.5797130465507507,
      "learning_rate": 0.002828,
      "loss": 1.0714,
      "step": 564800
    },
    {
      "epoch": 42.391294559099435,
      "grad_norm": 0.718805193901062,
      "learning_rate": 0.002828,
      "loss": 1.0741,
      "step": 564864
    },
    {
      "epoch": 42.39609756097561,
      "grad_norm": 0.6358516216278076,
      "learning_rate": 0.002828,
      "loss": 1.0725,
      "step": 564928
    },
    {
      "epoch": 42.40090056285178,
      "grad_norm": 0.5884746313095093,
      "learning_rate": 0.002828,
      "loss": 1.0737,
      "step": 564992
    },
    {
      "epoch": 42.40570356472796,
      "grad_norm": 0.5733670592308044,
      "learning_rate": 0.002828,
      "loss": 1.0815,
      "step": 565056
    },
    {
      "epoch": 42.41050656660413,
      "grad_norm": 0.650057852268219,
      "learning_rate": 0.002828,
      "loss": 1.0763,
      "step": 565120
    },
    {
      "epoch": 42.4153095684803,
      "grad_norm": 0.7314160466194153,
      "learning_rate": 0.002828,
      "loss": 1.0779,
      "step": 565184
    },
    {
      "epoch": 42.42011257035647,
      "grad_norm": 0.5718480348587036,
      "learning_rate": 0.002828,
      "loss": 1.0746,
      "step": 565248
    },
    {
      "epoch": 42.42491557223264,
      "grad_norm": 0.5505502820014954,
      "learning_rate": 0.002828,
      "loss": 1.0741,
      "step": 565312
    },
    {
      "epoch": 42.42971857410882,
      "grad_norm": 0.5616229772567749,
      "learning_rate": 0.002828,
      "loss": 1.0761,
      "step": 565376
    },
    {
      "epoch": 42.43452157598499,
      "grad_norm": 0.6704441905021667,
      "learning_rate": 0.002828,
      "loss": 1.0763,
      "step": 565440
    },
    {
      "epoch": 42.439324577861164,
      "grad_norm": 0.5727601051330566,
      "learning_rate": 0.002828,
      "loss": 1.0808,
      "step": 565504
    },
    {
      "epoch": 42.444127579737334,
      "grad_norm": 0.5947756767272949,
      "learning_rate": 0.002828,
      "loss": 1.0725,
      "step": 565568
    },
    {
      "epoch": 42.44893058161351,
      "grad_norm": 0.6589171290397644,
      "learning_rate": 0.002828,
      "loss": 1.0739,
      "step": 565632
    },
    {
      "epoch": 42.45373358348968,
      "grad_norm": 0.829900324344635,
      "learning_rate": 0.002828,
      "loss": 1.0781,
      "step": 565696
    },
    {
      "epoch": 42.458536585365856,
      "grad_norm": 0.7435891628265381,
      "learning_rate": 0.002828,
      "loss": 1.0774,
      "step": 565760
    },
    {
      "epoch": 42.463339587242025,
      "grad_norm": 0.633306086063385,
      "learning_rate": 0.002828,
      "loss": 1.0755,
      "step": 565824
    },
    {
      "epoch": 42.4681425891182,
      "grad_norm": 0.7354848980903625,
      "learning_rate": 0.002828,
      "loss": 1.0779,
      "step": 565888
    },
    {
      "epoch": 42.47294559099437,
      "grad_norm": 0.7187036871910095,
      "learning_rate": 0.002828,
      "loss": 1.079,
      "step": 565952
    },
    {
      "epoch": 42.47774859287055,
      "grad_norm": 0.6875394582748413,
      "learning_rate": 0.002828,
      "loss": 1.0839,
      "step": 566016
    },
    {
      "epoch": 42.48255159474672,
      "grad_norm": 0.5287431478500366,
      "learning_rate": 0.002828,
      "loss": 1.0786,
      "step": 566080
    },
    {
      "epoch": 42.487354596622886,
      "grad_norm": 0.6273989677429199,
      "learning_rate": 0.002828,
      "loss": 1.0789,
      "step": 566144
    },
    {
      "epoch": 42.49215759849906,
      "grad_norm": 0.7115692496299744,
      "learning_rate": 0.002828,
      "loss": 1.0782,
      "step": 566208
    },
    {
      "epoch": 42.49696060037523,
      "grad_norm": 0.5426573157310486,
      "learning_rate": 0.002828,
      "loss": 1.0771,
      "step": 566272
    },
    {
      "epoch": 42.50176360225141,
      "grad_norm": 0.6621776223182678,
      "learning_rate": 0.002828,
      "loss": 1.0761,
      "step": 566336
    },
    {
      "epoch": 42.50656660412758,
      "grad_norm": 0.5490167140960693,
      "learning_rate": 0.002828,
      "loss": 1.0758,
      "step": 566400
    },
    {
      "epoch": 42.511369606003754,
      "grad_norm": 0.5877586603164673,
      "learning_rate": 0.002828,
      "loss": 1.0807,
      "step": 566464
    },
    {
      "epoch": 42.51617260787992,
      "grad_norm": 0.8068798780441284,
      "learning_rate": 0.002828,
      "loss": 1.0761,
      "step": 566528
    },
    {
      "epoch": 42.5209756097561,
      "grad_norm": 0.6804494261741638,
      "learning_rate": 0.002828,
      "loss": 1.0793,
      "step": 566592
    },
    {
      "epoch": 42.52577861163227,
      "grad_norm": 0.605664074420929,
      "learning_rate": 0.002828,
      "loss": 1.0742,
      "step": 566656
    },
    {
      "epoch": 42.530581613508446,
      "grad_norm": 0.5913964509963989,
      "learning_rate": 0.002828,
      "loss": 1.0801,
      "step": 566720
    },
    {
      "epoch": 42.535384615384615,
      "grad_norm": 0.5786881446838379,
      "learning_rate": 0.002828,
      "loss": 1.0806,
      "step": 566784
    },
    {
      "epoch": 42.54018761726079,
      "grad_norm": 0.5988232493400574,
      "learning_rate": 0.002828,
      "loss": 1.0774,
      "step": 566848
    },
    {
      "epoch": 42.54499061913696,
      "grad_norm": 0.529296875,
      "learning_rate": 0.002828,
      "loss": 1.0777,
      "step": 566912
    },
    {
      "epoch": 42.54979362101313,
      "grad_norm": 0.5336984395980835,
      "learning_rate": 0.002828,
      "loss": 1.0807,
      "step": 566976
    },
    {
      "epoch": 42.554596622889306,
      "grad_norm": 0.5777416229248047,
      "learning_rate": 0.002828,
      "loss": 1.0805,
      "step": 567040
    },
    {
      "epoch": 42.559399624765476,
      "grad_norm": 0.5394503474235535,
      "learning_rate": 0.002828,
      "loss": 1.0783,
      "step": 567104
    },
    {
      "epoch": 42.56420262664165,
      "grad_norm": 0.5473426580429077,
      "learning_rate": 0.002828,
      "loss": 1.0749,
      "step": 567168
    },
    {
      "epoch": 42.56900562851782,
      "grad_norm": 0.6526885032653809,
      "learning_rate": 0.002828,
      "loss": 1.0792,
      "step": 567232
    },
    {
      "epoch": 42.573808630394,
      "grad_norm": 0.5946779251098633,
      "learning_rate": 0.002828,
      "loss": 1.0803,
      "step": 567296
    },
    {
      "epoch": 42.57861163227017,
      "grad_norm": 0.5849866271018982,
      "learning_rate": 0.002828,
      "loss": 1.0775,
      "step": 567360
    },
    {
      "epoch": 42.583414634146344,
      "grad_norm": 0.7357564568519592,
      "learning_rate": 0.002828,
      "loss": 1.0806,
      "step": 567424
    },
    {
      "epoch": 42.58821763602251,
      "grad_norm": 0.6068862080574036,
      "learning_rate": 0.002828,
      "loss": 1.0816,
      "step": 567488
    },
    {
      "epoch": 42.59302063789869,
      "grad_norm": 0.5769990086555481,
      "learning_rate": 0.002828,
      "loss": 1.0853,
      "step": 567552
    },
    {
      "epoch": 42.59782363977486,
      "grad_norm": 0.8670127391815186,
      "learning_rate": 0.002828,
      "loss": 1.0748,
      "step": 567616
    },
    {
      "epoch": 42.602626641651035,
      "grad_norm": 0.5923072695732117,
      "learning_rate": 0.002828,
      "loss": 1.0848,
      "step": 567680
    },
    {
      "epoch": 42.607429643527205,
      "grad_norm": 0.6279094815254211,
      "learning_rate": 0.002828,
      "loss": 1.0821,
      "step": 567744
    },
    {
      "epoch": 42.612232645403374,
      "grad_norm": 0.6318846344947815,
      "learning_rate": 0.002828,
      "loss": 1.0801,
      "step": 567808
    },
    {
      "epoch": 42.61703564727955,
      "grad_norm": 0.5974023342132568,
      "learning_rate": 0.002828,
      "loss": 1.0829,
      "step": 567872
    },
    {
      "epoch": 42.62183864915572,
      "grad_norm": 0.5550240278244019,
      "learning_rate": 0.002828,
      "loss": 1.0812,
      "step": 567936
    },
    {
      "epoch": 42.626641651031896,
      "grad_norm": 0.6231825351715088,
      "learning_rate": 0.002828,
      "loss": 1.0753,
      "step": 568000
    },
    {
      "epoch": 42.631444652908066,
      "grad_norm": 0.6439521908760071,
      "learning_rate": 0.002828,
      "loss": 1.0791,
      "step": 568064
    },
    {
      "epoch": 42.63624765478424,
      "grad_norm": 0.6360310912132263,
      "learning_rate": 0.002828,
      "loss": 1.0796,
      "step": 568128
    },
    {
      "epoch": 42.64105065666041,
      "grad_norm": 0.5486919283866882,
      "learning_rate": 0.002828,
      "loss": 1.083,
      "step": 568192
    },
    {
      "epoch": 42.64585365853659,
      "grad_norm": 0.6617389917373657,
      "learning_rate": 0.002828,
      "loss": 1.0778,
      "step": 568256
    },
    {
      "epoch": 42.65065666041276,
      "grad_norm": 0.6045265197753906,
      "learning_rate": 0.002828,
      "loss": 1.0819,
      "step": 568320
    },
    {
      "epoch": 42.65545966228893,
      "grad_norm": 0.6709670424461365,
      "learning_rate": 0.002828,
      "loss": 1.0834,
      "step": 568384
    },
    {
      "epoch": 42.6602626641651,
      "grad_norm": 0.6926343441009521,
      "learning_rate": 0.002828,
      "loss": 1.0829,
      "step": 568448
    },
    {
      "epoch": 42.66506566604128,
      "grad_norm": 0.6111130714416504,
      "learning_rate": 0.002828,
      "loss": 1.0837,
      "step": 568512
    },
    {
      "epoch": 42.66986866791745,
      "grad_norm": 0.5871363878250122,
      "learning_rate": 0.002828,
      "loss": 1.0829,
      "step": 568576
    },
    {
      "epoch": 42.67467166979362,
      "grad_norm": 0.567190408706665,
      "learning_rate": 0.002828,
      "loss": 1.0809,
      "step": 568640
    },
    {
      "epoch": 42.679474671669794,
      "grad_norm": 0.6697739362716675,
      "learning_rate": 0.002828,
      "loss": 1.0859,
      "step": 568704
    },
    {
      "epoch": 42.684277673545964,
      "grad_norm": 0.6716319918632507,
      "learning_rate": 0.002828,
      "loss": 1.0769,
      "step": 568768
    },
    {
      "epoch": 42.68908067542214,
      "grad_norm": 0.611808180809021,
      "learning_rate": 0.002828,
      "loss": 1.0888,
      "step": 568832
    },
    {
      "epoch": 42.69388367729831,
      "grad_norm": 0.531807005405426,
      "learning_rate": 0.002828,
      "loss": 1.0809,
      "step": 568896
    },
    {
      "epoch": 42.698686679174486,
      "grad_norm": 0.5492876768112183,
      "learning_rate": 0.002828,
      "loss": 1.08,
      "step": 568960
    },
    {
      "epoch": 42.703489681050655,
      "grad_norm": 0.5824979543685913,
      "learning_rate": 0.002828,
      "loss": 1.0874,
      "step": 569024
    },
    {
      "epoch": 42.70829268292683,
      "grad_norm": 0.617067813873291,
      "learning_rate": 0.002828,
      "loss": 1.0808,
      "step": 569088
    },
    {
      "epoch": 42.713095684803,
      "grad_norm": 0.6360995769500732,
      "learning_rate": 0.002828,
      "loss": 1.0853,
      "step": 569152
    },
    {
      "epoch": 42.71789868667918,
      "grad_norm": 0.6540601253509521,
      "learning_rate": 0.002828,
      "loss": 1.0849,
      "step": 569216
    },
    {
      "epoch": 42.72270168855535,
      "grad_norm": 0.7369455099105835,
      "learning_rate": 0.002828,
      "loss": 1.0827,
      "step": 569280
    },
    {
      "epoch": 42.727504690431516,
      "grad_norm": 0.690152108669281,
      "learning_rate": 0.002828,
      "loss": 1.0824,
      "step": 569344
    },
    {
      "epoch": 42.73230769230769,
      "grad_norm": 0.613617479801178,
      "learning_rate": 0.002828,
      "loss": 1.0839,
      "step": 569408
    },
    {
      "epoch": 42.73711069418386,
      "grad_norm": 0.6699286103248596,
      "learning_rate": 0.002828,
      "loss": 1.0859,
      "step": 569472
    },
    {
      "epoch": 42.74191369606004,
      "grad_norm": 0.6793148517608643,
      "learning_rate": 0.002828,
      "loss": 1.0866,
      "step": 569536
    },
    {
      "epoch": 42.74671669793621,
      "grad_norm": 0.6059393286705017,
      "learning_rate": 0.002828,
      "loss": 1.0796,
      "step": 569600
    },
    {
      "epoch": 42.751519699812384,
      "grad_norm": 0.58489590883255,
      "learning_rate": 0.002828,
      "loss": 1.0818,
      "step": 569664
    },
    {
      "epoch": 42.75632270168855,
      "grad_norm": 0.6535030007362366,
      "learning_rate": 0.002828,
      "loss": 1.089,
      "step": 569728
    },
    {
      "epoch": 42.76112570356473,
      "grad_norm": 0.7264353036880493,
      "learning_rate": 0.002828,
      "loss": 1.0853,
      "step": 569792
    },
    {
      "epoch": 42.7659287054409,
      "grad_norm": 0.7156565189361572,
      "learning_rate": 0.002828,
      "loss": 1.0861,
      "step": 569856
    },
    {
      "epoch": 42.770731707317076,
      "grad_norm": 0.56122887134552,
      "learning_rate": 0.002828,
      "loss": 1.0835,
      "step": 569920
    },
    {
      "epoch": 42.775534709193245,
      "grad_norm": 0.5848171710968018,
      "learning_rate": 0.002828,
      "loss": 1.0839,
      "step": 569984
    },
    {
      "epoch": 42.78033771106942,
      "grad_norm": 0.6172915697097778,
      "learning_rate": 0.002828,
      "loss": 1.0779,
      "step": 570048
    },
    {
      "epoch": 42.78514071294559,
      "grad_norm": 0.5863094329833984,
      "learning_rate": 0.002828,
      "loss": 1.0871,
      "step": 570112
    },
    {
      "epoch": 42.78994371482176,
      "grad_norm": 0.5800777673721313,
      "learning_rate": 0.002828,
      "loss": 1.0828,
      "step": 570176
    },
    {
      "epoch": 42.79474671669794,
      "grad_norm": 0.6153944134712219,
      "learning_rate": 0.002828,
      "loss": 1.0826,
      "step": 570240
    },
    {
      "epoch": 42.799549718574106,
      "grad_norm": 0.6796879768371582,
      "learning_rate": 0.002828,
      "loss": 1.0891,
      "step": 570304
    },
    {
      "epoch": 42.80435272045028,
      "grad_norm": 0.688765287399292,
      "learning_rate": 0.002828,
      "loss": 1.0849,
      "step": 570368
    },
    {
      "epoch": 42.80915572232645,
      "grad_norm": 0.6226803660392761,
      "learning_rate": 0.002828,
      "loss": 1.0852,
      "step": 570432
    },
    {
      "epoch": 42.81395872420263,
      "grad_norm": 0.5279849171638489,
      "learning_rate": 0.002828,
      "loss": 1.0821,
      "step": 570496
    },
    {
      "epoch": 42.8187617260788,
      "grad_norm": 0.7052308320999146,
      "learning_rate": 0.002828,
      "loss": 1.0825,
      "step": 570560
    },
    {
      "epoch": 42.823564727954974,
      "grad_norm": 0.5396923422813416,
      "learning_rate": 0.002828,
      "loss": 1.0828,
      "step": 570624
    },
    {
      "epoch": 42.82836772983114,
      "grad_norm": 0.6663107872009277,
      "learning_rate": 0.002828,
      "loss": 1.087,
      "step": 570688
    },
    {
      "epoch": 42.83317073170732,
      "grad_norm": 0.4901922941207886,
      "learning_rate": 0.002828,
      "loss": 1.0942,
      "step": 570752
    },
    {
      "epoch": 42.83797373358349,
      "grad_norm": 0.5605421662330627,
      "learning_rate": 0.002828,
      "loss": 1.0865,
      "step": 570816
    },
    {
      "epoch": 42.842776735459665,
      "grad_norm": 0.7057970762252808,
      "learning_rate": 0.002828,
      "loss": 1.0844,
      "step": 570880
    },
    {
      "epoch": 42.847579737335835,
      "grad_norm": 0.5384121537208557,
      "learning_rate": 0.002828,
      "loss": 1.0857,
      "step": 570944
    },
    {
      "epoch": 42.852382739212004,
      "grad_norm": 0.6306588649749756,
      "learning_rate": 0.002828,
      "loss": 1.0856,
      "step": 571008
    },
    {
      "epoch": 42.85718574108818,
      "grad_norm": 0.5484234094619751,
      "learning_rate": 0.002828,
      "loss": 1.0833,
      "step": 571072
    },
    {
      "epoch": 42.86198874296435,
      "grad_norm": 0.7848870754241943,
      "learning_rate": 0.002828,
      "loss": 1.0875,
      "step": 571136
    },
    {
      "epoch": 42.866791744840526,
      "grad_norm": 0.6511049270629883,
      "learning_rate": 0.002828,
      "loss": 1.091,
      "step": 571200
    },
    {
      "epoch": 42.871594746716696,
      "grad_norm": 0.6501409411430359,
      "learning_rate": 0.002828,
      "loss": 1.0861,
      "step": 571264
    },
    {
      "epoch": 42.87639774859287,
      "grad_norm": 0.6331605315208435,
      "learning_rate": 0.002828,
      "loss": 1.0844,
      "step": 571328
    },
    {
      "epoch": 42.88120075046904,
      "grad_norm": 0.7007723450660706,
      "learning_rate": 0.002828,
      "loss": 1.0881,
      "step": 571392
    },
    {
      "epoch": 42.88600375234522,
      "grad_norm": 0.6939073801040649,
      "learning_rate": 0.002828,
      "loss": 1.082,
      "step": 571456
    },
    {
      "epoch": 42.89080675422139,
      "grad_norm": 0.6270707249641418,
      "learning_rate": 0.002828,
      "loss": 1.0808,
      "step": 571520
    },
    {
      "epoch": 42.89560975609756,
      "grad_norm": 0.7300127148628235,
      "learning_rate": 0.002828,
      "loss": 1.0849,
      "step": 571584
    },
    {
      "epoch": 42.90041275797373,
      "grad_norm": 0.5953959822654724,
      "learning_rate": 0.002828,
      "loss": 1.0864,
      "step": 571648
    },
    {
      "epoch": 42.90521575984991,
      "grad_norm": 0.6020718216896057,
      "learning_rate": 0.002828,
      "loss": 1.0895,
      "step": 571712
    },
    {
      "epoch": 42.91001876172608,
      "grad_norm": 0.6176451444625854,
      "learning_rate": 0.002828,
      "loss": 1.0872,
      "step": 571776
    },
    {
      "epoch": 42.91482176360225,
      "grad_norm": 0.7385719418525696,
      "learning_rate": 0.002828,
      "loss": 1.0854,
      "step": 571840
    },
    {
      "epoch": 42.919624765478424,
      "grad_norm": 0.591037929058075,
      "learning_rate": 0.002828,
      "loss": 1.083,
      "step": 571904
    },
    {
      "epoch": 42.924427767354594,
      "grad_norm": 0.750694751739502,
      "learning_rate": 0.002828,
      "loss": 1.0846,
      "step": 571968
    },
    {
      "epoch": 42.92923076923077,
      "grad_norm": 0.7448388934135437,
      "learning_rate": 0.002828,
      "loss": 1.0909,
      "step": 572032
    },
    {
      "epoch": 42.93403377110694,
      "grad_norm": 0.5910031795501709,
      "learning_rate": 0.002828,
      "loss": 1.0899,
      "step": 572096
    },
    {
      "epoch": 42.938836772983116,
      "grad_norm": 0.652076244354248,
      "learning_rate": 0.002828,
      "loss": 1.0898,
      "step": 572160
    },
    {
      "epoch": 42.943639774859285,
      "grad_norm": 0.5189456939697266,
      "learning_rate": 0.002828,
      "loss": 1.0874,
      "step": 572224
    },
    {
      "epoch": 42.94844277673546,
      "grad_norm": 0.8578477501869202,
      "learning_rate": 0.002828,
      "loss": 1.0873,
      "step": 572288
    },
    {
      "epoch": 42.95324577861163,
      "grad_norm": 0.5538474321365356,
      "learning_rate": 0.002828,
      "loss": 1.0907,
      "step": 572352
    },
    {
      "epoch": 42.95804878048781,
      "grad_norm": 0.6879174709320068,
      "learning_rate": 0.002828,
      "loss": 1.0847,
      "step": 572416
    },
    {
      "epoch": 42.96285178236398,
      "grad_norm": 0.5544927716255188,
      "learning_rate": 0.002828,
      "loss": 1.086,
      "step": 572480
    },
    {
      "epoch": 42.96765478424015,
      "grad_norm": 0.57198166847229,
      "learning_rate": 0.002828,
      "loss": 1.0837,
      "step": 572544
    },
    {
      "epoch": 42.97245778611632,
      "grad_norm": 0.6882445216178894,
      "learning_rate": 0.002828,
      "loss": 1.0904,
      "step": 572608
    },
    {
      "epoch": 42.97726078799249,
      "grad_norm": 0.5579478740692139,
      "learning_rate": 0.002828,
      "loss": 1.0857,
      "step": 572672
    },
    {
      "epoch": 42.98206378986867,
      "grad_norm": 0.7191713452339172,
      "learning_rate": 0.002828,
      "loss": 1.0855,
      "step": 572736
    },
    {
      "epoch": 42.98686679174484,
      "grad_norm": 0.6459416151046753,
      "learning_rate": 0.002828,
      "loss": 1.0915,
      "step": 572800
    },
    {
      "epoch": 42.991669793621014,
      "grad_norm": 0.5977568030357361,
      "learning_rate": 0.002828,
      "loss": 1.0839,
      "step": 572864
    },
    {
      "epoch": 42.99647279549718,
      "grad_norm": 0.6233012676239014,
      "learning_rate": 0.002828,
      "loss": 1.0823,
      "step": 572928
    },
    {
      "epoch": 43.00127579737336,
      "grad_norm": 0.5843372344970703,
      "learning_rate": 0.002828,
      "loss": 1.0782,
      "step": 572992
    },
    {
      "epoch": 43.00607879924953,
      "grad_norm": 0.5951461791992188,
      "learning_rate": 0.002828,
      "loss": 1.0502,
      "step": 573056
    },
    {
      "epoch": 43.010881801125706,
      "grad_norm": 0.7060664296150208,
      "learning_rate": 0.002828,
      "loss": 1.0491,
      "step": 573120
    },
    {
      "epoch": 43.015684803001875,
      "grad_norm": 0.6400389075279236,
      "learning_rate": 0.002828,
      "loss": 1.0487,
      "step": 573184
    },
    {
      "epoch": 43.02048780487805,
      "grad_norm": 0.7333444952964783,
      "learning_rate": 0.002828,
      "loss": 1.051,
      "step": 573248
    },
    {
      "epoch": 43.02529080675422,
      "grad_norm": 0.7291139960289001,
      "learning_rate": 0.002828,
      "loss": 1.0521,
      "step": 573312
    },
    {
      "epoch": 43.0300938086304,
      "grad_norm": 0.5190329551696777,
      "learning_rate": 0.002828,
      "loss": 1.0572,
      "step": 573376
    },
    {
      "epoch": 43.03489681050657,
      "grad_norm": 0.7910537719726562,
      "learning_rate": 0.002828,
      "loss": 1.0555,
      "step": 573440
    },
    {
      "epoch": 43.039699812382736,
      "grad_norm": 0.7930744886398315,
      "learning_rate": 0.002828,
      "loss": 1.0526,
      "step": 573504
    },
    {
      "epoch": 43.04450281425891,
      "grad_norm": 0.5120797157287598,
      "learning_rate": 0.002828,
      "loss": 1.0537,
      "step": 573568
    },
    {
      "epoch": 43.04930581613508,
      "grad_norm": 0.6616923809051514,
      "learning_rate": 0.002828,
      "loss": 1.0574,
      "step": 573632
    },
    {
      "epoch": 43.05410881801126,
      "grad_norm": 0.6355065703392029,
      "learning_rate": 0.002828,
      "loss": 1.0548,
      "step": 573696
    },
    {
      "epoch": 43.05891181988743,
      "grad_norm": 0.5768610835075378,
      "learning_rate": 0.002828,
      "loss": 1.0571,
      "step": 573760
    },
    {
      "epoch": 43.063714821763604,
      "grad_norm": 0.782618522644043,
      "learning_rate": 0.002828,
      "loss": 1.0526,
      "step": 573824
    },
    {
      "epoch": 43.06851782363977,
      "grad_norm": 0.5912759304046631,
      "learning_rate": 0.002828,
      "loss": 1.0564,
      "step": 573888
    },
    {
      "epoch": 43.07332082551595,
      "grad_norm": 0.6178007125854492,
      "learning_rate": 0.002828,
      "loss": 1.0585,
      "step": 573952
    },
    {
      "epoch": 43.07812382739212,
      "grad_norm": 0.597594678401947,
      "learning_rate": 0.002828,
      "loss": 1.0582,
      "step": 574016
    },
    {
      "epoch": 43.082926829268295,
      "grad_norm": 0.5913254618644714,
      "learning_rate": 0.002828,
      "loss": 1.0552,
      "step": 574080
    },
    {
      "epoch": 43.087729831144465,
      "grad_norm": 0.691646933555603,
      "learning_rate": 0.002828,
      "loss": 1.055,
      "step": 574144
    },
    {
      "epoch": 43.09253283302064,
      "grad_norm": 0.5682713389396667,
      "learning_rate": 0.002828,
      "loss": 1.0642,
      "step": 574208
    },
    {
      "epoch": 43.09733583489681,
      "grad_norm": 0.6842798590660095,
      "learning_rate": 0.002828,
      "loss": 1.0599,
      "step": 574272
    },
    {
      "epoch": 43.10213883677298,
      "grad_norm": 0.584943950176239,
      "learning_rate": 0.002828,
      "loss": 1.0569,
      "step": 574336
    },
    {
      "epoch": 43.106941838649156,
      "grad_norm": 0.4885464608669281,
      "learning_rate": 0.002828,
      "loss": 1.0596,
      "step": 574400
    },
    {
      "epoch": 43.111744840525326,
      "grad_norm": 0.5184704065322876,
      "learning_rate": 0.002828,
      "loss": 1.0612,
      "step": 574464
    },
    {
      "epoch": 43.1165478424015,
      "grad_norm": 0.7813940644264221,
      "learning_rate": 0.002828,
      "loss": 1.0629,
      "step": 574528
    },
    {
      "epoch": 43.12135084427767,
      "grad_norm": 0.6903042793273926,
      "learning_rate": 0.002828,
      "loss": 1.0614,
      "step": 574592
    },
    {
      "epoch": 43.12615384615385,
      "grad_norm": 0.5398491621017456,
      "learning_rate": 0.002828,
      "loss": 1.058,
      "step": 574656
    },
    {
      "epoch": 43.13095684803002,
      "grad_norm": 0.6805057525634766,
      "learning_rate": 0.002828,
      "loss": 1.0587,
      "step": 574720
    },
    {
      "epoch": 43.135759849906194,
      "grad_norm": 0.7092175483703613,
      "learning_rate": 0.002828,
      "loss": 1.0644,
      "step": 574784
    },
    {
      "epoch": 43.14056285178236,
      "grad_norm": 0.6952199935913086,
      "learning_rate": 0.002828,
      "loss": 1.0601,
      "step": 574848
    },
    {
      "epoch": 43.14536585365854,
      "grad_norm": 0.6295602321624756,
      "learning_rate": 0.002828,
      "loss": 1.0563,
      "step": 574912
    },
    {
      "epoch": 43.15016885553471,
      "grad_norm": 0.4714643657207489,
      "learning_rate": 0.002828,
      "loss": 1.0581,
      "step": 574976
    },
    {
      "epoch": 43.154971857410885,
      "grad_norm": 0.5860217809677124,
      "learning_rate": 0.002828,
      "loss": 1.0595,
      "step": 575040
    },
    {
      "epoch": 43.159774859287054,
      "grad_norm": 0.5973192453384399,
      "learning_rate": 0.002828,
      "loss": 1.0611,
      "step": 575104
    },
    {
      "epoch": 43.164577861163224,
      "grad_norm": 0.692756712436676,
      "learning_rate": 0.002828,
      "loss": 1.0639,
      "step": 575168
    },
    {
      "epoch": 43.1693808630394,
      "grad_norm": 0.6680248379707336,
      "learning_rate": 0.002828,
      "loss": 1.0607,
      "step": 575232
    },
    {
      "epoch": 43.17418386491557,
      "grad_norm": 0.8391013741493225,
      "learning_rate": 0.002828,
      "loss": 1.0617,
      "step": 575296
    },
    {
      "epoch": 43.178986866791746,
      "grad_norm": 0.7862709164619446,
      "learning_rate": 0.002828,
      "loss": 1.0632,
      "step": 575360
    },
    {
      "epoch": 43.183789868667915,
      "grad_norm": 0.6694145202636719,
      "learning_rate": 0.002828,
      "loss": 1.0604,
      "step": 575424
    },
    {
      "epoch": 43.18859287054409,
      "grad_norm": 0.5881288051605225,
      "learning_rate": 0.002828,
      "loss": 1.0608,
      "step": 575488
    },
    {
      "epoch": 43.19339587242026,
      "grad_norm": 0.6086690425872803,
      "learning_rate": 0.002828,
      "loss": 1.0658,
      "step": 575552
    },
    {
      "epoch": 43.19819887429644,
      "grad_norm": 0.5388079285621643,
      "learning_rate": 0.002828,
      "loss": 1.0613,
      "step": 575616
    },
    {
      "epoch": 43.20300187617261,
      "grad_norm": 0.5871914625167847,
      "learning_rate": 0.002828,
      "loss": 1.0581,
      "step": 575680
    },
    {
      "epoch": 43.20780487804878,
      "grad_norm": 1.0138866901397705,
      "learning_rate": 0.002828,
      "loss": 1.0594,
      "step": 575744
    },
    {
      "epoch": 43.21260787992495,
      "grad_norm": 0.8155320286750793,
      "learning_rate": 0.002828,
      "loss": 1.0638,
      "step": 575808
    },
    {
      "epoch": 43.21741088180113,
      "grad_norm": 0.5958095788955688,
      "learning_rate": 0.002828,
      "loss": 1.0669,
      "step": 575872
    },
    {
      "epoch": 43.2222138836773,
      "grad_norm": 0.7571079730987549,
      "learning_rate": 0.002828,
      "loss": 1.0629,
      "step": 575936
    },
    {
      "epoch": 43.22701688555347,
      "grad_norm": 0.5348535776138306,
      "learning_rate": 0.002828,
      "loss": 1.0587,
      "step": 576000
    },
    {
      "epoch": 43.231819887429644,
      "grad_norm": 0.5872513651847839,
      "learning_rate": 0.002828,
      "loss": 1.0663,
      "step": 576064
    },
    {
      "epoch": 43.23662288930581,
      "grad_norm": 0.5647656917572021,
      "learning_rate": 0.002828,
      "loss": 1.0622,
      "step": 576128
    },
    {
      "epoch": 43.24142589118199,
      "grad_norm": 0.5898438692092896,
      "learning_rate": 0.002828,
      "loss": 1.066,
      "step": 576192
    },
    {
      "epoch": 43.24622889305816,
      "grad_norm": 0.590237557888031,
      "learning_rate": 0.002828,
      "loss": 1.0615,
      "step": 576256
    },
    {
      "epoch": 43.251031894934336,
      "grad_norm": 0.7051697373390198,
      "learning_rate": 0.002828,
      "loss": 1.0634,
      "step": 576320
    },
    {
      "epoch": 43.255834896810505,
      "grad_norm": 0.6399511098861694,
      "learning_rate": 0.002828,
      "loss": 1.0649,
      "step": 576384
    },
    {
      "epoch": 43.26063789868668,
      "grad_norm": 0.6372509598731995,
      "learning_rate": 0.002828,
      "loss": 1.0607,
      "step": 576448
    },
    {
      "epoch": 43.26544090056285,
      "grad_norm": 0.5469287037849426,
      "learning_rate": 0.002828,
      "loss": 1.0645,
      "step": 576512
    },
    {
      "epoch": 43.27024390243903,
      "grad_norm": 0.5622618198394775,
      "learning_rate": 0.002828,
      "loss": 1.0631,
      "step": 576576
    },
    {
      "epoch": 43.2750469043152,
      "grad_norm": 0.6582178473472595,
      "learning_rate": 0.002828,
      "loss": 1.0637,
      "step": 576640
    },
    {
      "epoch": 43.27984990619137,
      "grad_norm": 0.5956276655197144,
      "learning_rate": 0.002828,
      "loss": 1.0637,
      "step": 576704
    },
    {
      "epoch": 43.28465290806754,
      "grad_norm": 0.7692244052886963,
      "learning_rate": 0.002828,
      "loss": 1.0628,
      "step": 576768
    },
    {
      "epoch": 43.28945590994371,
      "grad_norm": 0.6546646356582642,
      "learning_rate": 0.002828,
      "loss": 1.061,
      "step": 576832
    },
    {
      "epoch": 43.29425891181989,
      "grad_norm": 0.6064482927322388,
      "learning_rate": 0.002828,
      "loss": 1.0622,
      "step": 576896
    },
    {
      "epoch": 43.29906191369606,
      "grad_norm": 0.7595977783203125,
      "learning_rate": 0.002828,
      "loss": 1.0584,
      "step": 576960
    },
    {
      "epoch": 43.303864915572234,
      "grad_norm": 0.6731221675872803,
      "learning_rate": 0.002828,
      "loss": 1.0669,
      "step": 577024
    },
    {
      "epoch": 43.3086679174484,
      "grad_norm": 0.6657878756523132,
      "learning_rate": 0.002828,
      "loss": 1.066,
      "step": 577088
    },
    {
      "epoch": 43.31347091932458,
      "grad_norm": 0.5610875487327576,
      "learning_rate": 0.002828,
      "loss": 1.0667,
      "step": 577152
    },
    {
      "epoch": 43.31827392120075,
      "grad_norm": 0.6538919806480408,
      "learning_rate": 0.002828,
      "loss": 1.0663,
      "step": 577216
    },
    {
      "epoch": 43.323076923076925,
      "grad_norm": 0.763543426990509,
      "learning_rate": 0.002828,
      "loss": 1.0694,
      "step": 577280
    },
    {
      "epoch": 43.327879924953095,
      "grad_norm": 0.6737272143363953,
      "learning_rate": 0.002828,
      "loss": 1.065,
      "step": 577344
    },
    {
      "epoch": 43.33268292682927,
      "grad_norm": 0.6752422451972961,
      "learning_rate": 0.002828,
      "loss": 1.0697,
      "step": 577408
    },
    {
      "epoch": 43.33748592870544,
      "grad_norm": 0.7812511920928955,
      "learning_rate": 0.002828,
      "loss": 1.0639,
      "step": 577472
    },
    {
      "epoch": 43.34228893058162,
      "grad_norm": 0.6621730327606201,
      "learning_rate": 0.002828,
      "loss": 1.0697,
      "step": 577536
    },
    {
      "epoch": 43.347091932457786,
      "grad_norm": 0.697150707244873,
      "learning_rate": 0.002828,
      "loss": 1.0702,
      "step": 577600
    },
    {
      "epoch": 43.351894934333956,
      "grad_norm": 0.5495653748512268,
      "learning_rate": 0.002828,
      "loss": 1.0665,
      "step": 577664
    },
    {
      "epoch": 43.35669793621013,
      "grad_norm": 0.6246775984764099,
      "learning_rate": 0.002828,
      "loss": 1.0644,
      "step": 577728
    },
    {
      "epoch": 43.3615009380863,
      "grad_norm": 0.6434612274169922,
      "learning_rate": 0.002828,
      "loss": 1.069,
      "step": 577792
    },
    {
      "epoch": 43.36630393996248,
      "grad_norm": 0.8342929482460022,
      "learning_rate": 0.002828,
      "loss": 1.0649,
      "step": 577856
    },
    {
      "epoch": 43.37110694183865,
      "grad_norm": 0.7098569273948669,
      "learning_rate": 0.002828,
      "loss": 1.0673,
      "step": 577920
    },
    {
      "epoch": 43.375909943714824,
      "grad_norm": 0.6844133138656616,
      "learning_rate": 0.002828,
      "loss": 1.0622,
      "step": 577984
    },
    {
      "epoch": 43.38071294559099,
      "grad_norm": 0.6284855008125305,
      "learning_rate": 0.002828,
      "loss": 1.0678,
      "step": 578048
    },
    {
      "epoch": 43.38551594746717,
      "grad_norm": 0.5995723605155945,
      "learning_rate": 0.002828,
      "loss": 1.0661,
      "step": 578112
    },
    {
      "epoch": 43.39031894934334,
      "grad_norm": 0.5386275053024292,
      "learning_rate": 0.002828,
      "loss": 1.0683,
      "step": 578176
    },
    {
      "epoch": 43.395121951219515,
      "grad_norm": 0.8105739951133728,
      "learning_rate": 0.002828,
      "loss": 1.0637,
      "step": 578240
    },
    {
      "epoch": 43.399924953095685,
      "grad_norm": 0.5679133534431458,
      "learning_rate": 0.002828,
      "loss": 1.0674,
      "step": 578304
    },
    {
      "epoch": 43.40472795497186,
      "grad_norm": 0.6953145861625671,
      "learning_rate": 0.002828,
      "loss": 1.0727,
      "step": 578368
    },
    {
      "epoch": 43.40953095684803,
      "grad_norm": 0.5408474802970886,
      "learning_rate": 0.002828,
      "loss": 1.073,
      "step": 578432
    },
    {
      "epoch": 43.4143339587242,
      "grad_norm": 0.5147654414176941,
      "learning_rate": 0.002828,
      "loss": 1.0679,
      "step": 578496
    },
    {
      "epoch": 43.419136960600376,
      "grad_norm": 0.5925143957138062,
      "learning_rate": 0.002828,
      "loss": 1.0722,
      "step": 578560
    },
    {
      "epoch": 43.423939962476545,
      "grad_norm": 0.5820292234420776,
      "learning_rate": 0.002828,
      "loss": 1.0686,
      "step": 578624
    },
    {
      "epoch": 43.42874296435272,
      "grad_norm": 0.6305953860282898,
      "learning_rate": 0.002828,
      "loss": 1.0691,
      "step": 578688
    },
    {
      "epoch": 43.43354596622889,
      "grad_norm": 0.6557329893112183,
      "learning_rate": 0.002828,
      "loss": 1.0656,
      "step": 578752
    },
    {
      "epoch": 43.43834896810507,
      "grad_norm": 0.6649267673492432,
      "learning_rate": 0.002828,
      "loss": 1.077,
      "step": 578816
    },
    {
      "epoch": 43.44315196998124,
      "grad_norm": 0.7626684904098511,
      "learning_rate": 0.002828,
      "loss": 1.0657,
      "step": 578880
    },
    {
      "epoch": 43.44795497185741,
      "grad_norm": 0.6433974504470825,
      "learning_rate": 0.002828,
      "loss": 1.0776,
      "step": 578944
    },
    {
      "epoch": 43.45275797373358,
      "grad_norm": 0.7001895904541016,
      "learning_rate": 0.002828,
      "loss": 1.0702,
      "step": 579008
    },
    {
      "epoch": 43.45756097560976,
      "grad_norm": 0.7059563994407654,
      "learning_rate": 0.002828,
      "loss": 1.0674,
      "step": 579072
    },
    {
      "epoch": 43.46236397748593,
      "grad_norm": 0.6160516738891602,
      "learning_rate": 0.002828,
      "loss": 1.0698,
      "step": 579136
    },
    {
      "epoch": 43.4671669793621,
      "grad_norm": 0.6591466665267944,
      "learning_rate": 0.002828,
      "loss": 1.0658,
      "step": 579200
    },
    {
      "epoch": 43.471969981238274,
      "grad_norm": 0.6705019474029541,
      "learning_rate": 0.002828,
      "loss": 1.0735,
      "step": 579264
    },
    {
      "epoch": 43.476772983114444,
      "grad_norm": 0.6403300762176514,
      "learning_rate": 0.002828,
      "loss": 1.0678,
      "step": 579328
    },
    {
      "epoch": 43.48157598499062,
      "grad_norm": 0.613906979560852,
      "learning_rate": 0.002828,
      "loss": 1.0704,
      "step": 579392
    },
    {
      "epoch": 43.48637898686679,
      "grad_norm": 0.6602962017059326,
      "learning_rate": 0.002828,
      "loss": 1.0747,
      "step": 579456
    },
    {
      "epoch": 43.491181988742966,
      "grad_norm": 0.6226909160614014,
      "learning_rate": 0.002828,
      "loss": 1.0704,
      "step": 579520
    },
    {
      "epoch": 43.495984990619135,
      "grad_norm": 0.6018249988555908,
      "learning_rate": 0.002828,
      "loss": 1.0696,
      "step": 579584
    },
    {
      "epoch": 43.50078799249531,
      "grad_norm": 0.5919440388679504,
      "learning_rate": 0.002828,
      "loss": 1.0725,
      "step": 579648
    },
    {
      "epoch": 43.50559099437148,
      "grad_norm": 10.90912914276123,
      "learning_rate": 0.002828,
      "loss": 1.0727,
      "step": 579712
    },
    {
      "epoch": 43.51039399624766,
      "grad_norm": 0.6816866397857666,
      "learning_rate": 0.002828,
      "loss": 1.0742,
      "step": 579776
    },
    {
      "epoch": 43.51519699812383,
      "grad_norm": 0.5978453159332275,
      "learning_rate": 0.002828,
      "loss": 1.0799,
      "step": 579840
    },
    {
      "epoch": 43.52,
      "grad_norm": 0.6959922909736633,
      "learning_rate": 0.002828,
      "loss": 1.0714,
      "step": 579904
    },
    {
      "epoch": 43.52480300187617,
      "grad_norm": 0.5744678378105164,
      "learning_rate": 0.002828,
      "loss": 1.0667,
      "step": 579968
    },
    {
      "epoch": 43.52960600375234,
      "grad_norm": 0.6247631907463074,
      "learning_rate": 0.002828,
      "loss": 1.0722,
      "step": 580032
    },
    {
      "epoch": 43.53440900562852,
      "grad_norm": 0.7486253976821899,
      "learning_rate": 0.002828,
      "loss": 1.0726,
      "step": 580096
    },
    {
      "epoch": 43.53921200750469,
      "grad_norm": 0.6821879744529724,
      "learning_rate": 0.002828,
      "loss": 1.071,
      "step": 580160
    },
    {
      "epoch": 43.544015009380864,
      "grad_norm": 0.5762830972671509,
      "learning_rate": 0.002828,
      "loss": 1.0728,
      "step": 580224
    },
    {
      "epoch": 43.54881801125703,
      "grad_norm": 0.5583580732345581,
      "learning_rate": 0.002828,
      "loss": 1.0672,
      "step": 580288
    },
    {
      "epoch": 43.55362101313321,
      "grad_norm": 0.5891516804695129,
      "learning_rate": 0.002828,
      "loss": 1.0694,
      "step": 580352
    },
    {
      "epoch": 43.55842401500938,
      "grad_norm": 0.7095462679862976,
      "learning_rate": 0.002828,
      "loss": 1.0733,
      "step": 580416
    },
    {
      "epoch": 43.563227016885556,
      "grad_norm": 0.5964438915252686,
      "learning_rate": 0.002828,
      "loss": 1.0733,
      "step": 580480
    },
    {
      "epoch": 43.568030018761725,
      "grad_norm": 0.6678001284599304,
      "learning_rate": 0.002828,
      "loss": 1.0711,
      "step": 580544
    },
    {
      "epoch": 43.5728330206379,
      "grad_norm": 0.5241270065307617,
      "learning_rate": 0.002828,
      "loss": 1.0736,
      "step": 580608
    },
    {
      "epoch": 43.57763602251407,
      "grad_norm": 0.5979251265525818,
      "learning_rate": 0.002828,
      "loss": 1.0711,
      "step": 580672
    },
    {
      "epoch": 43.58243902439025,
      "grad_norm": 0.5383949279785156,
      "learning_rate": 0.002828,
      "loss": 1.0789,
      "step": 580736
    },
    {
      "epoch": 43.587242026266416,
      "grad_norm": 0.5068698525428772,
      "learning_rate": 0.002828,
      "loss": 1.0756,
      "step": 580800
    },
    {
      "epoch": 43.592045028142586,
      "grad_norm": 0.6026465892791748,
      "learning_rate": 0.002828,
      "loss": 1.0743,
      "step": 580864
    },
    {
      "epoch": 43.59684803001876,
      "grad_norm": 0.7454937696456909,
      "learning_rate": 0.002828,
      "loss": 1.08,
      "step": 580928
    },
    {
      "epoch": 43.60165103189493,
      "grad_norm": 0.7301528453826904,
      "learning_rate": 0.002828,
      "loss": 1.0736,
      "step": 580992
    },
    {
      "epoch": 43.60645403377111,
      "grad_norm": 0.5982566475868225,
      "learning_rate": 0.002828,
      "loss": 1.0758,
      "step": 581056
    },
    {
      "epoch": 43.61125703564728,
      "grad_norm": 0.6397075653076172,
      "learning_rate": 0.002828,
      "loss": 1.0818,
      "step": 581120
    },
    {
      "epoch": 43.616060037523454,
      "grad_norm": 0.6900129914283752,
      "learning_rate": 0.002828,
      "loss": 1.079,
      "step": 581184
    },
    {
      "epoch": 43.62086303939962,
      "grad_norm": 0.6174712181091309,
      "learning_rate": 0.002828,
      "loss": 1.0774,
      "step": 581248
    },
    {
      "epoch": 43.6256660412758,
      "grad_norm": 0.6319519281387329,
      "learning_rate": 0.002828,
      "loss": 1.0751,
      "step": 581312
    },
    {
      "epoch": 43.63046904315197,
      "grad_norm": 0.5982273817062378,
      "learning_rate": 0.002828,
      "loss": 1.0732,
      "step": 581376
    },
    {
      "epoch": 43.635272045028145,
      "grad_norm": 0.6787490248680115,
      "learning_rate": 0.002828,
      "loss": 1.077,
      "step": 581440
    },
    {
      "epoch": 43.640075046904315,
      "grad_norm": 0.6518996357917786,
      "learning_rate": 0.002828,
      "loss": 1.0745,
      "step": 581504
    },
    {
      "epoch": 43.64487804878049,
      "grad_norm": 0.6295955777168274,
      "learning_rate": 0.002828,
      "loss": 1.0759,
      "step": 581568
    },
    {
      "epoch": 43.64968105065666,
      "grad_norm": 0.5694141983985901,
      "learning_rate": 0.002828,
      "loss": 1.076,
      "step": 581632
    },
    {
      "epoch": 43.65448405253283,
      "grad_norm": 0.6541298627853394,
      "learning_rate": 0.002828,
      "loss": 1.0776,
      "step": 581696
    },
    {
      "epoch": 43.659287054409006,
      "grad_norm": 0.5805025696754456,
      "learning_rate": 0.002828,
      "loss": 1.0752,
      "step": 581760
    },
    {
      "epoch": 43.664090056285175,
      "grad_norm": 0.5690178275108337,
      "learning_rate": 0.002828,
      "loss": 1.0786,
      "step": 581824
    },
    {
      "epoch": 43.66889305816135,
      "grad_norm": 0.8229773044586182,
      "learning_rate": 0.002828,
      "loss": 1.0703,
      "step": 581888
    },
    {
      "epoch": 43.67369606003752,
      "grad_norm": 0.7628452181816101,
      "learning_rate": 0.002828,
      "loss": 1.0792,
      "step": 581952
    },
    {
      "epoch": 43.6784990619137,
      "grad_norm": 0.5986396074295044,
      "learning_rate": 0.002828,
      "loss": 1.0772,
      "step": 582016
    },
    {
      "epoch": 43.68330206378987,
      "grad_norm": 0.632666826248169,
      "learning_rate": 0.002828,
      "loss": 1.0767,
      "step": 582080
    },
    {
      "epoch": 43.68810506566604,
      "grad_norm": 0.6015210151672363,
      "learning_rate": 0.002828,
      "loss": 1.071,
      "step": 582144
    },
    {
      "epoch": 43.69290806754221,
      "grad_norm": 0.6710034608840942,
      "learning_rate": 0.002828,
      "loss": 1.0756,
      "step": 582208
    },
    {
      "epoch": 43.69771106941839,
      "grad_norm": 0.5373445153236389,
      "learning_rate": 0.002828,
      "loss": 1.0787,
      "step": 582272
    },
    {
      "epoch": 43.70251407129456,
      "grad_norm": 0.7191179990768433,
      "learning_rate": 0.002828,
      "loss": 1.0803,
      "step": 582336
    },
    {
      "epoch": 43.707317073170735,
      "grad_norm": 0.8347383141517639,
      "learning_rate": 0.002828,
      "loss": 1.0688,
      "step": 582400
    },
    {
      "epoch": 43.712120075046904,
      "grad_norm": 0.5719173550605774,
      "learning_rate": 0.002828,
      "loss": 1.0793,
      "step": 582464
    },
    {
      "epoch": 43.716923076923074,
      "grad_norm": 0.6953842639923096,
      "learning_rate": 0.002828,
      "loss": 1.0763,
      "step": 582528
    },
    {
      "epoch": 43.72172607879925,
      "grad_norm": 0.6387263536453247,
      "learning_rate": 0.002828,
      "loss": 1.0747,
      "step": 582592
    },
    {
      "epoch": 43.72652908067542,
      "grad_norm": 0.6053597331047058,
      "learning_rate": 0.002828,
      "loss": 1.079,
      "step": 582656
    },
    {
      "epoch": 43.731332082551596,
      "grad_norm": 0.7519330978393555,
      "learning_rate": 0.002828,
      "loss": 1.076,
      "step": 582720
    },
    {
      "epoch": 43.736135084427765,
      "grad_norm": 0.5185713171958923,
      "learning_rate": 0.002828,
      "loss": 1.0813,
      "step": 582784
    },
    {
      "epoch": 43.74093808630394,
      "grad_norm": 0.4951317310333252,
      "learning_rate": 0.002828,
      "loss": 1.0802,
      "step": 582848
    },
    {
      "epoch": 43.74574108818011,
      "grad_norm": 0.5165672302246094,
      "learning_rate": 0.002828,
      "loss": 1.0713,
      "step": 582912
    },
    {
      "epoch": 43.75054409005629,
      "grad_norm": 0.6204631924629211,
      "learning_rate": 0.002828,
      "loss": 1.0694,
      "step": 582976
    },
    {
      "epoch": 43.75534709193246,
      "grad_norm": 0.5312516689300537,
      "learning_rate": 0.002828,
      "loss": 1.0771,
      "step": 583040
    },
    {
      "epoch": 43.76015009380863,
      "grad_norm": 0.6661979556083679,
      "learning_rate": 0.002828,
      "loss": 1.0804,
      "step": 583104
    },
    {
      "epoch": 43.7649530956848,
      "grad_norm": 0.5828707218170166,
      "learning_rate": 0.002828,
      "loss": 1.0827,
      "step": 583168
    },
    {
      "epoch": 43.76975609756098,
      "grad_norm": 0.5881485939025879,
      "learning_rate": 0.002828,
      "loss": 1.0765,
      "step": 583232
    },
    {
      "epoch": 43.77455909943715,
      "grad_norm": 0.6520361304283142,
      "learning_rate": 0.002828,
      "loss": 1.0808,
      "step": 583296
    },
    {
      "epoch": 43.77936210131332,
      "grad_norm": 0.7659959197044373,
      "learning_rate": 0.002828,
      "loss": 1.0765,
      "step": 583360
    },
    {
      "epoch": 43.784165103189494,
      "grad_norm": 0.4846315085887909,
      "learning_rate": 0.002828,
      "loss": 1.0789,
      "step": 583424
    },
    {
      "epoch": 43.78896810506566,
      "grad_norm": 0.5142455101013184,
      "learning_rate": 0.002828,
      "loss": 1.0785,
      "step": 583488
    },
    {
      "epoch": 43.79377110694184,
      "grad_norm": 0.6952672600746155,
      "learning_rate": 0.002828,
      "loss": 1.0783,
      "step": 583552
    },
    {
      "epoch": 43.79857410881801,
      "grad_norm": 0.7225334048271179,
      "learning_rate": 0.002828,
      "loss": 1.0759,
      "step": 583616
    },
    {
      "epoch": 43.803377110694186,
      "grad_norm": 1.8566055297851562,
      "learning_rate": 0.002828,
      "loss": 1.0816,
      "step": 583680
    },
    {
      "epoch": 43.808180112570355,
      "grad_norm": 0.5677652359008789,
      "learning_rate": 0.002828,
      "loss": 1.0816,
      "step": 583744
    },
    {
      "epoch": 43.81298311444653,
      "grad_norm": 0.7882415056228638,
      "learning_rate": 0.002828,
      "loss": 1.0761,
      "step": 583808
    },
    {
      "epoch": 43.8177861163227,
      "grad_norm": 0.7066271901130676,
      "learning_rate": 0.002828,
      "loss": 1.0751,
      "step": 583872
    },
    {
      "epoch": 43.82258911819888,
      "grad_norm": 0.6859885454177856,
      "learning_rate": 0.002828,
      "loss": 1.0749,
      "step": 583936
    },
    {
      "epoch": 43.82739212007505,
      "grad_norm": 0.5205126404762268,
      "learning_rate": 0.002828,
      "loss": 1.0768,
      "step": 584000
    },
    {
      "epoch": 43.83219512195122,
      "grad_norm": 0.5213777422904968,
      "learning_rate": 0.002828,
      "loss": 1.0791,
      "step": 584064
    },
    {
      "epoch": 43.83699812382739,
      "grad_norm": 0.6113591194152832,
      "learning_rate": 0.002828,
      "loss": 1.0808,
      "step": 584128
    },
    {
      "epoch": 43.84180112570356,
      "grad_norm": 0.7442314624786377,
      "learning_rate": 0.002828,
      "loss": 1.0756,
      "step": 584192
    },
    {
      "epoch": 43.84660412757974,
      "grad_norm": 0.7046990990638733,
      "learning_rate": 0.002828,
      "loss": 1.0815,
      "step": 584256
    },
    {
      "epoch": 43.85140712945591,
      "grad_norm": 0.5735657215118408,
      "learning_rate": 0.002828,
      "loss": 1.0778,
      "step": 584320
    },
    {
      "epoch": 43.856210131332084,
      "grad_norm": 0.5720263719558716,
      "learning_rate": 0.002828,
      "loss": 1.0824,
      "step": 584384
    },
    {
      "epoch": 43.86101313320825,
      "grad_norm": 0.7149404883384705,
      "learning_rate": 0.002828,
      "loss": 1.0763,
      "step": 584448
    },
    {
      "epoch": 43.86581613508443,
      "grad_norm": 0.6320396065711975,
      "learning_rate": 0.002828,
      "loss": 1.0862,
      "step": 584512
    },
    {
      "epoch": 43.8706191369606,
      "grad_norm": 0.6746490001678467,
      "learning_rate": 0.002828,
      "loss": 1.0848,
      "step": 584576
    },
    {
      "epoch": 43.875422138836775,
      "grad_norm": 0.5930066108703613,
      "learning_rate": 0.002828,
      "loss": 1.0818,
      "step": 584640
    },
    {
      "epoch": 43.880225140712945,
      "grad_norm": 0.6815347671508789,
      "learning_rate": 0.002828,
      "loss": 1.0772,
      "step": 584704
    },
    {
      "epoch": 43.88502814258912,
      "grad_norm": 0.5609988570213318,
      "learning_rate": 0.002828,
      "loss": 1.0778,
      "step": 584768
    },
    {
      "epoch": 43.88983114446529,
      "grad_norm": 0.6229777336120605,
      "learning_rate": 0.002828,
      "loss": 1.0802,
      "step": 584832
    },
    {
      "epoch": 43.89463414634147,
      "grad_norm": 0.5853741765022278,
      "learning_rate": 0.002828,
      "loss": 1.0754,
      "step": 584896
    },
    {
      "epoch": 43.899437148217636,
      "grad_norm": 0.8227654695510864,
      "learning_rate": 0.002828,
      "loss": 1.0778,
      "step": 584960
    },
    {
      "epoch": 43.904240150093806,
      "grad_norm": 0.5244843363761902,
      "learning_rate": 0.002828,
      "loss": 1.078,
      "step": 585024
    },
    {
      "epoch": 43.90904315196998,
      "grad_norm": 0.6417165398597717,
      "learning_rate": 0.002828,
      "loss": 1.0807,
      "step": 585088
    },
    {
      "epoch": 43.91384615384615,
      "grad_norm": 0.7093722820281982,
      "learning_rate": 0.002828,
      "loss": 1.0742,
      "step": 585152
    },
    {
      "epoch": 43.91864915572233,
      "grad_norm": 0.6699938774108887,
      "learning_rate": 0.002828,
      "loss": 1.0833,
      "step": 585216
    },
    {
      "epoch": 43.9234521575985,
      "grad_norm": 0.6436761021614075,
      "learning_rate": 0.002828,
      "loss": 1.0748,
      "step": 585280
    },
    {
      "epoch": 43.92825515947467,
      "grad_norm": 0.7439795136451721,
      "learning_rate": 0.002828,
      "loss": 1.08,
      "step": 585344
    },
    {
      "epoch": 43.93305816135084,
      "grad_norm": 0.5588270425796509,
      "learning_rate": 0.002828,
      "loss": 1.0853,
      "step": 585408
    },
    {
      "epoch": 43.93786116322702,
      "grad_norm": 0.7395601868629456,
      "learning_rate": 0.002828,
      "loss": 1.0813,
      "step": 585472
    },
    {
      "epoch": 43.94266416510319,
      "grad_norm": 0.6798150539398193,
      "learning_rate": 0.002828,
      "loss": 1.082,
      "step": 585536
    },
    {
      "epoch": 43.947467166979365,
      "grad_norm": 0.6237505078315735,
      "learning_rate": 0.002828,
      "loss": 1.0837,
      "step": 585600
    },
    {
      "epoch": 43.952270168855534,
      "grad_norm": 0.6804483532905579,
      "learning_rate": 0.002828,
      "loss": 1.0781,
      "step": 585664
    },
    {
      "epoch": 43.957073170731704,
      "grad_norm": 0.6526710987091064,
      "learning_rate": 0.002828,
      "loss": 1.0844,
      "step": 585728
    },
    {
      "epoch": 43.96187617260788,
      "grad_norm": 0.7000369429588318,
      "learning_rate": 0.002828,
      "loss": 1.0795,
      "step": 585792
    },
    {
      "epoch": 43.96667917448405,
      "grad_norm": 0.6432533264160156,
      "learning_rate": 0.002828,
      "loss": 1.0808,
      "step": 585856
    },
    {
      "epoch": 43.971482176360226,
      "grad_norm": 0.4752485156059265,
      "learning_rate": 0.002828,
      "loss": 1.0787,
      "step": 585920
    },
    {
      "epoch": 43.976285178236395,
      "grad_norm": 0.5716511607170105,
      "learning_rate": 0.002828,
      "loss": 1.081,
      "step": 585984
    },
    {
      "epoch": 43.98108818011257,
      "grad_norm": 0.5871129631996155,
      "learning_rate": 0.002828,
      "loss": 1.0887,
      "step": 586048
    },
    {
      "epoch": 43.98589118198874,
      "grad_norm": 0.5385404229164124,
      "learning_rate": 0.002828,
      "loss": 1.0806,
      "step": 586112
    },
    {
      "epoch": 43.99069418386492,
      "grad_norm": 0.6556265354156494,
      "learning_rate": 0.002828,
      "loss": 1.0862,
      "step": 586176
    },
    {
      "epoch": 43.99549718574109,
      "grad_norm": 0.547139585018158,
      "learning_rate": 0.002828,
      "loss": 1.0825,
      "step": 586240
    },
    {
      "epoch": 44.00030018761726,
      "grad_norm": 0.6162726283073425,
      "learning_rate": 0.002828,
      "loss": 1.0855,
      "step": 586304
    },
    {
      "epoch": 44.00510318949343,
      "grad_norm": 0.6631591320037842,
      "learning_rate": 0.002828,
      "loss": 1.0443,
      "step": 586368
    },
    {
      "epoch": 44.00990619136961,
      "grad_norm": 0.6567460894584656,
      "learning_rate": 0.002828,
      "loss": 1.0485,
      "step": 586432
    },
    {
      "epoch": 44.01470919324578,
      "grad_norm": 0.6258338093757629,
      "learning_rate": 0.002828,
      "loss": 1.0489,
      "step": 586496
    },
    {
      "epoch": 44.01951219512195,
      "grad_norm": 0.6194806694984436,
      "learning_rate": 0.002828,
      "loss": 1.0441,
      "step": 586560
    },
    {
      "epoch": 44.024315196998124,
      "grad_norm": 0.6310887932777405,
      "learning_rate": 0.002828,
      "loss": 1.0494,
      "step": 586624
    },
    {
      "epoch": 44.02911819887429,
      "grad_norm": 0.590854287147522,
      "learning_rate": 0.002828,
      "loss": 1.048,
      "step": 586688
    },
    {
      "epoch": 44.03392120075047,
      "grad_norm": 0.5728686451911926,
      "learning_rate": 0.002828,
      "loss": 1.0507,
      "step": 586752
    },
    {
      "epoch": 44.03872420262664,
      "grad_norm": 0.6099998354911804,
      "learning_rate": 0.002828,
      "loss": 1.047,
      "step": 586816
    },
    {
      "epoch": 44.043527204502816,
      "grad_norm": 0.5928614139556885,
      "learning_rate": 0.002828,
      "loss": 1.0467,
      "step": 586880
    },
    {
      "epoch": 44.048330206378985,
      "grad_norm": 0.4924786686897278,
      "learning_rate": 0.002828,
      "loss": 1.046,
      "step": 586944
    },
    {
      "epoch": 44.05313320825516,
      "grad_norm": 0.6266118884086609,
      "learning_rate": 0.002828,
      "loss": 1.0441,
      "step": 587008
    },
    {
      "epoch": 44.05793621013133,
      "grad_norm": 0.6437507271766663,
      "learning_rate": 0.002828,
      "loss": 1.0523,
      "step": 587072
    },
    {
      "epoch": 44.06273921200751,
      "grad_norm": 0.6502929329872131,
      "learning_rate": 0.002828,
      "loss": 1.0534,
      "step": 587136
    },
    {
      "epoch": 44.06754221388368,
      "grad_norm": 0.5448206067085266,
      "learning_rate": 0.002828,
      "loss": 1.0495,
      "step": 587200
    },
    {
      "epoch": 44.07234521575985,
      "grad_norm": 0.6379390954971313,
      "learning_rate": 0.002828,
      "loss": 1.0465,
      "step": 587264
    },
    {
      "epoch": 44.07714821763602,
      "grad_norm": 0.6880658268928528,
      "learning_rate": 0.002828,
      "loss": 1.0496,
      "step": 587328
    },
    {
      "epoch": 44.08195121951219,
      "grad_norm": 0.6843789219856262,
      "learning_rate": 0.002828,
      "loss": 1.0494,
      "step": 587392
    },
    {
      "epoch": 44.08675422138837,
      "grad_norm": 0.6257761716842651,
      "learning_rate": 0.002828,
      "loss": 1.0552,
      "step": 587456
    },
    {
      "epoch": 44.09155722326454,
      "grad_norm": 0.5892012119293213,
      "learning_rate": 0.002828,
      "loss": 1.0548,
      "step": 587520
    },
    {
      "epoch": 44.096360225140714,
      "grad_norm": 0.5341951251029968,
      "learning_rate": 0.002828,
      "loss": 1.0527,
      "step": 587584
    },
    {
      "epoch": 44.10116322701688,
      "grad_norm": 0.5494747161865234,
      "learning_rate": 0.002828,
      "loss": 1.0548,
      "step": 587648
    },
    {
      "epoch": 44.10596622889306,
      "grad_norm": 0.6075941324234009,
      "learning_rate": 0.002828,
      "loss": 1.0527,
      "step": 587712
    },
    {
      "epoch": 44.11076923076923,
      "grad_norm": 0.5817174315452576,
      "learning_rate": 0.002828,
      "loss": 1.0448,
      "step": 587776
    },
    {
      "epoch": 44.115572232645405,
      "grad_norm": 0.593499481678009,
      "learning_rate": 0.002828,
      "loss": 1.0499,
      "step": 587840
    },
    {
      "epoch": 44.120375234521575,
      "grad_norm": 0.7568128108978271,
      "learning_rate": 0.002828,
      "loss": 1.0579,
      "step": 587904
    },
    {
      "epoch": 44.12517823639775,
      "grad_norm": 0.6690353751182556,
      "learning_rate": 0.002828,
      "loss": 1.0514,
      "step": 587968
    },
    {
      "epoch": 44.12998123827392,
      "grad_norm": 0.6533127427101135,
      "learning_rate": 0.002828,
      "loss": 1.0552,
      "step": 588032
    },
    {
      "epoch": 44.1347842401501,
      "grad_norm": 0.6924563646316528,
      "learning_rate": 0.002828,
      "loss": 1.0515,
      "step": 588096
    },
    {
      "epoch": 44.139587242026266,
      "grad_norm": 0.5739848613739014,
      "learning_rate": 0.002828,
      "loss": 1.0506,
      "step": 588160
    },
    {
      "epoch": 44.144390243902436,
      "grad_norm": 0.5882086157798767,
      "learning_rate": 0.002828,
      "loss": 1.0541,
      "step": 588224
    },
    {
      "epoch": 44.14919324577861,
      "grad_norm": 0.5814594626426697,
      "learning_rate": 0.002828,
      "loss": 1.0537,
      "step": 588288
    },
    {
      "epoch": 44.15399624765478,
      "grad_norm": 0.5676882266998291,
      "learning_rate": 0.002828,
      "loss": 1.0546,
      "step": 588352
    },
    {
      "epoch": 44.15879924953096,
      "grad_norm": 0.5608579516410828,
      "learning_rate": 0.002828,
      "loss": 1.0525,
      "step": 588416
    },
    {
      "epoch": 44.16360225140713,
      "grad_norm": 0.6067665219306946,
      "learning_rate": 0.002828,
      "loss": 1.0546,
      "step": 588480
    },
    {
      "epoch": 44.168405253283304,
      "grad_norm": 0.6528283953666687,
      "learning_rate": 0.002828,
      "loss": 1.0599,
      "step": 588544
    },
    {
      "epoch": 44.17320825515947,
      "grad_norm": 0.6116089820861816,
      "learning_rate": 0.002828,
      "loss": 1.0542,
      "step": 588608
    },
    {
      "epoch": 44.17801125703565,
      "grad_norm": 0.5617401599884033,
      "learning_rate": 0.002828,
      "loss": 1.0509,
      "step": 588672
    },
    {
      "epoch": 44.18281425891182,
      "grad_norm": 0.624115526676178,
      "learning_rate": 0.002828,
      "loss": 1.0551,
      "step": 588736
    },
    {
      "epoch": 44.187617260787995,
      "grad_norm": 0.6719458699226379,
      "learning_rate": 0.002828,
      "loss": 1.057,
      "step": 588800
    },
    {
      "epoch": 44.192420262664164,
      "grad_norm": 0.6795137524604797,
      "learning_rate": 0.002828,
      "loss": 1.0526,
      "step": 588864
    },
    {
      "epoch": 44.19722326454034,
      "grad_norm": 0.8369677066802979,
      "learning_rate": 0.002828,
      "loss": 1.058,
      "step": 588928
    },
    {
      "epoch": 44.20202626641651,
      "grad_norm": 0.6024190187454224,
      "learning_rate": 0.002828,
      "loss": 1.0584,
      "step": 588992
    },
    {
      "epoch": 44.20682926829268,
      "grad_norm": 0.6553812623023987,
      "learning_rate": 0.002828,
      "loss": 1.0631,
      "step": 589056
    },
    {
      "epoch": 44.211632270168856,
      "grad_norm": 0.7060786485671997,
      "learning_rate": 0.002828,
      "loss": 1.0599,
      "step": 589120
    },
    {
      "epoch": 44.216435272045025,
      "grad_norm": 0.7535731792449951,
      "learning_rate": 0.002828,
      "loss": 1.0558,
      "step": 589184
    },
    {
      "epoch": 44.2212382739212,
      "grad_norm": 0.6029979586601257,
      "learning_rate": 0.002828,
      "loss": 1.0588,
      "step": 589248
    },
    {
      "epoch": 44.22604127579737,
      "grad_norm": 0.6775466799736023,
      "learning_rate": 0.002828,
      "loss": 1.0511,
      "step": 589312
    },
    {
      "epoch": 44.23084427767355,
      "grad_norm": 0.5644216537475586,
      "learning_rate": 0.002828,
      "loss": 1.0551,
      "step": 589376
    },
    {
      "epoch": 44.23564727954972,
      "grad_norm": 0.55975741147995,
      "learning_rate": 0.002828,
      "loss": 1.0575,
      "step": 589440
    },
    {
      "epoch": 44.24045028142589,
      "grad_norm": 0.4965690076351166,
      "learning_rate": 0.002828,
      "loss": 1.0561,
      "step": 589504
    },
    {
      "epoch": 44.24525328330206,
      "grad_norm": 0.6054225564002991,
      "learning_rate": 0.002828,
      "loss": 1.0546,
      "step": 589568
    },
    {
      "epoch": 44.25005628517824,
      "grad_norm": 0.591658890247345,
      "learning_rate": 0.002828,
      "loss": 1.0581,
      "step": 589632
    },
    {
      "epoch": 44.25485928705441,
      "grad_norm": 0.7469866871833801,
      "learning_rate": 0.002828,
      "loss": 1.0621,
      "step": 589696
    },
    {
      "epoch": 44.259662288930585,
      "grad_norm": 0.5662860870361328,
      "learning_rate": 0.002828,
      "loss": 1.0586,
      "step": 589760
    },
    {
      "epoch": 44.264465290806754,
      "grad_norm": 0.6618179082870483,
      "learning_rate": 0.002828,
      "loss": 1.0568,
      "step": 589824
    },
    {
      "epoch": 44.26926829268292,
      "grad_norm": 0.6655414700508118,
      "learning_rate": 0.002828,
      "loss": 1.0598,
      "step": 589888
    },
    {
      "epoch": 44.2740712945591,
      "grad_norm": 0.5782995223999023,
      "learning_rate": 0.002828,
      "loss": 1.0534,
      "step": 589952
    },
    {
      "epoch": 44.27887429643527,
      "grad_norm": 0.5280641317367554,
      "learning_rate": 0.002828,
      "loss": 1.0596,
      "step": 590016
    },
    {
      "epoch": 44.283677298311446,
      "grad_norm": 0.7310107946395874,
      "learning_rate": 0.002828,
      "loss": 1.058,
      "step": 590080
    },
    {
      "epoch": 44.288480300187615,
      "grad_norm": 0.6584124565124512,
      "learning_rate": 0.002828,
      "loss": 1.0576,
      "step": 590144
    },
    {
      "epoch": 44.29328330206379,
      "grad_norm": 0.5985362529754639,
      "learning_rate": 0.002828,
      "loss": 1.0604,
      "step": 590208
    },
    {
      "epoch": 44.29808630393996,
      "grad_norm": 0.6973876357078552,
      "learning_rate": 0.002828,
      "loss": 1.0635,
      "step": 590272
    },
    {
      "epoch": 44.30288930581614,
      "grad_norm": 0.542786717414856,
      "learning_rate": 0.002828,
      "loss": 1.0587,
      "step": 590336
    },
    {
      "epoch": 44.30769230769231,
      "grad_norm": 0.5430316925048828,
      "learning_rate": 0.002828,
      "loss": 1.0616,
      "step": 590400
    },
    {
      "epoch": 44.31249530956848,
      "grad_norm": 0.6455052495002747,
      "learning_rate": 0.002828,
      "loss": 1.0537,
      "step": 590464
    },
    {
      "epoch": 44.31729831144465,
      "grad_norm": 0.6836392283439636,
      "learning_rate": 0.002828,
      "loss": 1.0507,
      "step": 590528
    },
    {
      "epoch": 44.32210131332083,
      "grad_norm": 0.6295143365859985,
      "learning_rate": 0.002828,
      "loss": 1.0554,
      "step": 590592
    },
    {
      "epoch": 44.326904315197,
      "grad_norm": 0.742729127407074,
      "learning_rate": 0.002828,
      "loss": 1.0575,
      "step": 590656
    },
    {
      "epoch": 44.33170731707317,
      "grad_norm": 0.6139900088310242,
      "learning_rate": 0.002828,
      "loss": 1.0603,
      "step": 590720
    },
    {
      "epoch": 44.336510318949344,
      "grad_norm": 0.5415821075439453,
      "learning_rate": 0.002828,
      "loss": 1.0553,
      "step": 590784
    },
    {
      "epoch": 44.34131332082551,
      "grad_norm": 0.5520257949829102,
      "learning_rate": 0.002828,
      "loss": 1.0552,
      "step": 590848
    },
    {
      "epoch": 44.34611632270169,
      "grad_norm": 0.5664125084877014,
      "learning_rate": 0.002828,
      "loss": 1.0595,
      "step": 590912
    },
    {
      "epoch": 44.35091932457786,
      "grad_norm": 0.5885429382324219,
      "learning_rate": 0.002828,
      "loss": 1.0568,
      "step": 590976
    },
    {
      "epoch": 44.355722326454035,
      "grad_norm": 0.7497702836990356,
      "learning_rate": 0.002828,
      "loss": 1.056,
      "step": 591040
    },
    {
      "epoch": 44.360525328330205,
      "grad_norm": 0.6161512732505798,
      "learning_rate": 0.002828,
      "loss": 1.0547,
      "step": 591104
    },
    {
      "epoch": 44.36532833020638,
      "grad_norm": 0.6655434370040894,
      "learning_rate": 0.002828,
      "loss": 1.0586,
      "step": 591168
    },
    {
      "epoch": 44.37013133208255,
      "grad_norm": 0.5859283208847046,
      "learning_rate": 0.002828,
      "loss": 1.06,
      "step": 591232
    },
    {
      "epoch": 44.37493433395873,
      "grad_norm": 0.7172648906707764,
      "learning_rate": 0.002828,
      "loss": 1.0526,
      "step": 591296
    },
    {
      "epoch": 44.379737335834896,
      "grad_norm": 0.6022630333900452,
      "learning_rate": 0.002828,
      "loss": 1.055,
      "step": 591360
    },
    {
      "epoch": 44.38454033771107,
      "grad_norm": 0.6173017024993896,
      "learning_rate": 0.002828,
      "loss": 1.0574,
      "step": 591424
    },
    {
      "epoch": 44.38934333958724,
      "grad_norm": 0.7868699431419373,
      "learning_rate": 0.002828,
      "loss": 1.0516,
      "step": 591488
    },
    {
      "epoch": 44.39414634146341,
      "grad_norm": 0.559699535369873,
      "learning_rate": 0.002828,
      "loss": 1.0522,
      "step": 591552
    },
    {
      "epoch": 44.39894934333959,
      "grad_norm": 0.7050228714942932,
      "learning_rate": 0.002828,
      "loss": 1.0564,
      "step": 591616
    },
    {
      "epoch": 44.40375234521576,
      "grad_norm": 0.6284798383712769,
      "learning_rate": 0.002828,
      "loss": 1.0459,
      "step": 591680
    },
    {
      "epoch": 44.408555347091934,
      "grad_norm": 0.5507165789604187,
      "learning_rate": 0.002828,
      "loss": 1.0461,
      "step": 591744
    },
    {
      "epoch": 44.4133583489681,
      "grad_norm": 0.6204633712768555,
      "learning_rate": 0.002828,
      "loss": 1.0416,
      "step": 591808
    },
    {
      "epoch": 44.41816135084428,
      "grad_norm": 0.5714849829673767,
      "learning_rate": 0.002828,
      "loss": 1.0496,
      "step": 591872
    },
    {
      "epoch": 44.42296435272045,
      "grad_norm": 0.6399840712547302,
      "learning_rate": 0.002828,
      "loss": 1.0458,
      "step": 591936
    },
    {
      "epoch": 44.427767354596625,
      "grad_norm": 0.5837627053260803,
      "learning_rate": 0.002828,
      "loss": 1.0512,
      "step": 592000
    },
    {
      "epoch": 44.432570356472795,
      "grad_norm": 0.6572296023368835,
      "learning_rate": 0.002828,
      "loss": 1.0509,
      "step": 592064
    },
    {
      "epoch": 44.43737335834897,
      "grad_norm": 0.6775990724563599,
      "learning_rate": 0.002828,
      "loss": 1.0559,
      "step": 592128
    },
    {
      "epoch": 44.44217636022514,
      "grad_norm": 0.6193767189979553,
      "learning_rate": 0.002828,
      "loss": 1.0516,
      "step": 592192
    },
    {
      "epoch": 44.44697936210132,
      "grad_norm": 0.6084091663360596,
      "learning_rate": 0.002828,
      "loss": 1.0495,
      "step": 592256
    },
    {
      "epoch": 44.451782363977486,
      "grad_norm": 0.6361521482467651,
      "learning_rate": 0.002828,
      "loss": 1.0467,
      "step": 592320
    },
    {
      "epoch": 44.456585365853655,
      "grad_norm": 0.6244654655456543,
      "learning_rate": 0.002828,
      "loss": 1.0485,
      "step": 592384
    },
    {
      "epoch": 44.46138836772983,
      "grad_norm": 0.6915056109428406,
      "learning_rate": 0.002828,
      "loss": 1.049,
      "step": 592448
    },
    {
      "epoch": 44.466191369606,
      "grad_norm": 0.5671323537826538,
      "learning_rate": 0.002828,
      "loss": 1.0476,
      "step": 592512
    },
    {
      "epoch": 44.47099437148218,
      "grad_norm": 0.6313081979751587,
      "learning_rate": 0.002828,
      "loss": 1.054,
      "step": 592576
    },
    {
      "epoch": 44.47579737335835,
      "grad_norm": 0.5822582840919495,
      "learning_rate": 0.002828,
      "loss": 1.0507,
      "step": 592640
    },
    {
      "epoch": 44.48060037523452,
      "grad_norm": 0.6180717945098877,
      "learning_rate": 0.002828,
      "loss": 1.0482,
      "step": 592704
    },
    {
      "epoch": 44.48540337711069,
      "grad_norm": 0.7753960490226746,
      "learning_rate": 0.002828,
      "loss": 1.0501,
      "step": 592768
    },
    {
      "epoch": 44.49020637898687,
      "grad_norm": 0.887162983417511,
      "learning_rate": 0.002828,
      "loss": 1.0539,
      "step": 592832
    },
    {
      "epoch": 44.49500938086304,
      "grad_norm": 0.6842669248580933,
      "learning_rate": 0.002828,
      "loss": 1.0504,
      "step": 592896
    },
    {
      "epoch": 44.499812382739215,
      "grad_norm": 0.5897248983383179,
      "learning_rate": 0.002828,
      "loss": 1.0542,
      "step": 592960
    },
    {
      "epoch": 44.504615384615384,
      "grad_norm": 0.6817046999931335,
      "learning_rate": 0.002828,
      "loss": 1.0452,
      "step": 593024
    },
    {
      "epoch": 44.50941838649156,
      "grad_norm": 0.6712552905082703,
      "learning_rate": 0.002828,
      "loss": 1.0497,
      "step": 593088
    },
    {
      "epoch": 44.51422138836773,
      "grad_norm": 0.7757641673088074,
      "learning_rate": 0.002828,
      "loss": 1.0498,
      "step": 593152
    },
    {
      "epoch": 44.5190243902439,
      "grad_norm": 0.6338831186294556,
      "learning_rate": 0.002828,
      "loss": 1.0529,
      "step": 593216
    },
    {
      "epoch": 44.523827392120076,
      "grad_norm": 0.5562050342559814,
      "learning_rate": 0.002828,
      "loss": 1.0446,
      "step": 593280
    },
    {
      "epoch": 44.528630393996245,
      "grad_norm": 0.6536921262741089,
      "learning_rate": 0.002828,
      "loss": 1.0569,
      "step": 593344
    },
    {
      "epoch": 44.53343339587242,
      "grad_norm": 0.5887206792831421,
      "learning_rate": 0.002828,
      "loss": 1.0521,
      "step": 593408
    },
    {
      "epoch": 44.53823639774859,
      "grad_norm": 0.7711880803108215,
      "learning_rate": 0.002828,
      "loss": 1.0535,
      "step": 593472
    },
    {
      "epoch": 44.54303939962477,
      "grad_norm": 0.604606032371521,
      "learning_rate": 0.002828,
      "loss": 1.0512,
      "step": 593536
    },
    {
      "epoch": 44.54784240150094,
      "grad_norm": 0.5851953029632568,
      "learning_rate": 0.002828,
      "loss": 1.049,
      "step": 593600
    },
    {
      "epoch": 44.55264540337711,
      "grad_norm": 0.8617910146713257,
      "learning_rate": 0.002828,
      "loss": 1.0514,
      "step": 593664
    },
    {
      "epoch": 44.55744840525328,
      "grad_norm": 0.6344445943832397,
      "learning_rate": 0.002828,
      "loss": 1.0521,
      "step": 593728
    },
    {
      "epoch": 44.56225140712946,
      "grad_norm": 0.6127328276634216,
      "learning_rate": 0.002828,
      "loss": 1.0516,
      "step": 593792
    },
    {
      "epoch": 44.56705440900563,
      "grad_norm": 0.6657578945159912,
      "learning_rate": 0.002828,
      "loss": 1.0553,
      "step": 593856
    },
    {
      "epoch": 44.571857410881805,
      "grad_norm": 0.7788225412368774,
      "learning_rate": 0.002828,
      "loss": 1.0466,
      "step": 593920
    },
    {
      "epoch": 44.576660412757974,
      "grad_norm": 0.5609893798828125,
      "learning_rate": 0.002828,
      "loss": 1.0561,
      "step": 593984
    },
    {
      "epoch": 44.58146341463414,
      "grad_norm": 0.6387379765510559,
      "learning_rate": 0.002828,
      "loss": 1.0482,
      "step": 594048
    },
    {
      "epoch": 44.58626641651032,
      "grad_norm": 0.6570412516593933,
      "learning_rate": 0.002828,
      "loss": 1.0516,
      "step": 594112
    },
    {
      "epoch": 44.59106941838649,
      "grad_norm": 0.494945228099823,
      "learning_rate": 0.002828,
      "loss": 1.0541,
      "step": 594176
    },
    {
      "epoch": 44.595872420262666,
      "grad_norm": 0.6139020323753357,
      "learning_rate": 0.002828,
      "loss": 1.0536,
      "step": 594240
    },
    {
      "epoch": 44.600675422138835,
      "grad_norm": 0.7307867407798767,
      "learning_rate": 0.002828,
      "loss": 1.0527,
      "step": 594304
    },
    {
      "epoch": 44.60547842401501,
      "grad_norm": 0.6578233242034912,
      "learning_rate": 0.002828,
      "loss": 1.0518,
      "step": 594368
    },
    {
      "epoch": 44.61028142589118,
      "grad_norm": 0.6268816590309143,
      "learning_rate": 0.002828,
      "loss": 1.0562,
      "step": 594432
    },
    {
      "epoch": 44.61508442776736,
      "grad_norm": 0.6375176310539246,
      "learning_rate": 0.002828,
      "loss": 1.0543,
      "step": 594496
    },
    {
      "epoch": 44.619887429643526,
      "grad_norm": 0.5896298885345459,
      "learning_rate": 0.002828,
      "loss": 1.0482,
      "step": 594560
    },
    {
      "epoch": 44.6246904315197,
      "grad_norm": 0.6761197447776794,
      "learning_rate": 0.002828,
      "loss": 1.0599,
      "step": 594624
    },
    {
      "epoch": 44.62949343339587,
      "grad_norm": 0.5964671969413757,
      "learning_rate": 0.002828,
      "loss": 1.0609,
      "step": 594688
    },
    {
      "epoch": 44.63429643527205,
      "grad_norm": 0.6479743123054504,
      "learning_rate": 0.002828,
      "loss": 1.0531,
      "step": 594752
    },
    {
      "epoch": 44.63909943714822,
      "grad_norm": 0.6290108561515808,
      "learning_rate": 0.002828,
      "loss": 1.0567,
      "step": 594816
    },
    {
      "epoch": 44.64390243902439,
      "grad_norm": 0.7283443808555603,
      "learning_rate": 0.002828,
      "loss": 1.0541,
      "step": 594880
    },
    {
      "epoch": 44.648705440900564,
      "grad_norm": 0.6581125855445862,
      "learning_rate": 0.002828,
      "loss": 1.0503,
      "step": 594944
    },
    {
      "epoch": 44.65350844277673,
      "grad_norm": 0.6930689811706543,
      "learning_rate": 0.002828,
      "loss": 1.0572,
      "step": 595008
    },
    {
      "epoch": 44.65831144465291,
      "grad_norm": 0.5703624486923218,
      "learning_rate": 0.002828,
      "loss": 1.0597,
      "step": 595072
    },
    {
      "epoch": 44.66311444652908,
      "grad_norm": 0.7384238243103027,
      "learning_rate": 0.002828,
      "loss": 1.0528,
      "step": 595136
    },
    {
      "epoch": 44.667917448405255,
      "grad_norm": 0.6679426431655884,
      "learning_rate": 0.002828,
      "loss": 1.0638,
      "step": 595200
    },
    {
      "epoch": 44.672720450281425,
      "grad_norm": 0.5475023984909058,
      "learning_rate": 0.002828,
      "loss": 1.0582,
      "step": 595264
    },
    {
      "epoch": 44.6775234521576,
      "grad_norm": 0.5508772730827332,
      "learning_rate": 0.002828,
      "loss": 1.0594,
      "step": 595328
    },
    {
      "epoch": 44.68232645403377,
      "grad_norm": 0.6432110071182251,
      "learning_rate": 0.002828,
      "loss": 1.0586,
      "step": 595392
    },
    {
      "epoch": 44.68712945590995,
      "grad_norm": 0.5725296139717102,
      "learning_rate": 0.002828,
      "loss": 1.0557,
      "step": 595456
    },
    {
      "epoch": 44.691932457786116,
      "grad_norm": 0.679014265537262,
      "learning_rate": 0.002828,
      "loss": 1.0536,
      "step": 595520
    },
    {
      "epoch": 44.696735459662285,
      "grad_norm": 0.7023448944091797,
      "learning_rate": 0.002828,
      "loss": 1.0614,
      "step": 595584
    },
    {
      "epoch": 44.70153846153846,
      "grad_norm": 0.563921332359314,
      "learning_rate": 0.002828,
      "loss": 1.0599,
      "step": 595648
    },
    {
      "epoch": 44.70634146341463,
      "grad_norm": 0.5842373967170715,
      "learning_rate": 0.002828,
      "loss": 1.0508,
      "step": 595712
    },
    {
      "epoch": 44.71114446529081,
      "grad_norm": 0.5589472651481628,
      "learning_rate": 0.002828,
      "loss": 1.0557,
      "step": 595776
    },
    {
      "epoch": 44.71594746716698,
      "grad_norm": 0.6475051045417786,
      "learning_rate": 0.002828,
      "loss": 1.0557,
      "step": 595840
    },
    {
      "epoch": 44.72075046904315,
      "grad_norm": 0.5432165861129761,
      "learning_rate": 0.002828,
      "loss": 1.057,
      "step": 595904
    },
    {
      "epoch": 44.72555347091932,
      "grad_norm": 0.5137358903884888,
      "learning_rate": 0.002828,
      "loss": 1.062,
      "step": 595968
    },
    {
      "epoch": 44.7303564727955,
      "grad_norm": 0.6988866925239563,
      "learning_rate": 0.002828,
      "loss": 1.0518,
      "step": 596032
    },
    {
      "epoch": 44.73515947467167,
      "grad_norm": 0.7478548884391785,
      "learning_rate": 0.002828,
      "loss": 1.0593,
      "step": 596096
    },
    {
      "epoch": 44.739962476547845,
      "grad_norm": 0.4760184586048126,
      "learning_rate": 0.002828,
      "loss": 1.0603,
      "step": 596160
    },
    {
      "epoch": 44.744765478424014,
      "grad_norm": 0.7488265633583069,
      "learning_rate": 0.002828,
      "loss": 1.0586,
      "step": 596224
    },
    {
      "epoch": 44.74956848030019,
      "grad_norm": 0.6295368671417236,
      "learning_rate": 0.002828,
      "loss": 1.0557,
      "step": 596288
    },
    {
      "epoch": 44.75437148217636,
      "grad_norm": 0.4992520809173584,
      "learning_rate": 0.002828,
      "loss": 1.0539,
      "step": 596352
    },
    {
      "epoch": 44.75917448405253,
      "grad_norm": 0.6316255927085876,
      "learning_rate": 0.002828,
      "loss": 1.0576,
      "step": 596416
    },
    {
      "epoch": 44.763977485928706,
      "grad_norm": 0.6836703419685364,
      "learning_rate": 0.002828,
      "loss": 1.0535,
      "step": 596480
    },
    {
      "epoch": 44.768780487804875,
      "grad_norm": 0.5981588959693909,
      "learning_rate": 0.002828,
      "loss": 1.0597,
      "step": 596544
    },
    {
      "epoch": 44.77358348968105,
      "grad_norm": 0.8249488472938538,
      "learning_rate": 0.002828,
      "loss": 1.0532,
      "step": 596608
    },
    {
      "epoch": 44.77838649155722,
      "grad_norm": 0.6853210926055908,
      "learning_rate": 0.002828,
      "loss": 1.0572,
      "step": 596672
    },
    {
      "epoch": 44.7831894934334,
      "grad_norm": 0.5512163043022156,
      "learning_rate": 0.002828,
      "loss": 1.0565,
      "step": 596736
    },
    {
      "epoch": 44.78799249530957,
      "grad_norm": 0.6343268156051636,
      "learning_rate": 0.002828,
      "loss": 1.0494,
      "step": 596800
    },
    {
      "epoch": 44.79279549718574,
      "grad_norm": 0.6317381262779236,
      "learning_rate": 0.002828,
      "loss": 1.0577,
      "step": 596864
    },
    {
      "epoch": 44.79759849906191,
      "grad_norm": 0.6074554324150085,
      "learning_rate": 0.002828,
      "loss": 1.0542,
      "step": 596928
    },
    {
      "epoch": 44.80240150093809,
      "grad_norm": 0.5829271078109741,
      "learning_rate": 0.002828,
      "loss": 1.0538,
      "step": 596992
    },
    {
      "epoch": 44.80720450281426,
      "grad_norm": 0.5427216291427612,
      "learning_rate": 0.002828,
      "loss": 1.0613,
      "step": 597056
    },
    {
      "epoch": 44.812007504690435,
      "grad_norm": 0.7172414660453796,
      "learning_rate": 0.002828,
      "loss": 1.0536,
      "step": 597120
    },
    {
      "epoch": 44.816810506566604,
      "grad_norm": 0.5971596837043762,
      "learning_rate": 0.002828,
      "loss": 1.0618,
      "step": 597184
    },
    {
      "epoch": 44.82161350844277,
      "grad_norm": 0.5736137628555298,
      "learning_rate": 0.002828,
      "loss": 1.0551,
      "step": 597248
    },
    {
      "epoch": 44.82641651031895,
      "grad_norm": 0.7229673266410828,
      "learning_rate": 0.002828,
      "loss": 1.0606,
      "step": 597312
    },
    {
      "epoch": 44.83121951219512,
      "grad_norm": 0.79429030418396,
      "learning_rate": 0.002828,
      "loss": 1.0533,
      "step": 597376
    },
    {
      "epoch": 44.836022514071296,
      "grad_norm": 0.5500268340110779,
      "learning_rate": 0.002828,
      "loss": 1.0581,
      "step": 597440
    },
    {
      "epoch": 44.840825515947465,
      "grad_norm": 0.6283418536186218,
      "learning_rate": 0.002828,
      "loss": 1.0618,
      "step": 597504
    },
    {
      "epoch": 44.84562851782364,
      "grad_norm": 0.5786862969398499,
      "learning_rate": 0.002828,
      "loss": 1.0523,
      "step": 597568
    },
    {
      "epoch": 44.85043151969981,
      "grad_norm": 0.6707375645637512,
      "learning_rate": 0.002828,
      "loss": 1.0596,
      "step": 597632
    },
    {
      "epoch": 44.85523452157599,
      "grad_norm": 0.6113284826278687,
      "learning_rate": 0.002828,
      "loss": 1.0573,
      "step": 597696
    },
    {
      "epoch": 44.86003752345216,
      "grad_norm": 0.7945263385772705,
      "learning_rate": 0.002828,
      "loss": 1.0632,
      "step": 597760
    },
    {
      "epoch": 44.86484052532833,
      "grad_norm": 0.8358811736106873,
      "learning_rate": 0.002828,
      "loss": 1.0602,
      "step": 597824
    },
    {
      "epoch": 44.8696435272045,
      "grad_norm": 0.5964508652687073,
      "learning_rate": 0.002828,
      "loss": 1.057,
      "step": 597888
    },
    {
      "epoch": 44.87444652908068,
      "grad_norm": 0.6710835099220276,
      "learning_rate": 0.002828,
      "loss": 1.0595,
      "step": 597952
    },
    {
      "epoch": 44.87924953095685,
      "grad_norm": 0.6213023066520691,
      "learning_rate": 0.002828,
      "loss": 1.057,
      "step": 598016
    },
    {
      "epoch": 44.88405253283302,
      "grad_norm": 0.6238622665405273,
      "learning_rate": 0.002828,
      "loss": 1.0514,
      "step": 598080
    },
    {
      "epoch": 44.888855534709194,
      "grad_norm": 0.7188548445701599,
      "learning_rate": 0.002828,
      "loss": 1.0622,
      "step": 598144
    },
    {
      "epoch": 44.89365853658536,
      "grad_norm": 0.5461704730987549,
      "learning_rate": 0.002828,
      "loss": 1.0625,
      "step": 598208
    },
    {
      "epoch": 44.89846153846154,
      "grad_norm": 0.5917800068855286,
      "learning_rate": 0.002828,
      "loss": 1.0594,
      "step": 598272
    },
    {
      "epoch": 44.90326454033771,
      "grad_norm": 0.5917359590530396,
      "learning_rate": 0.002828,
      "loss": 1.0653,
      "step": 598336
    },
    {
      "epoch": 44.908067542213885,
      "grad_norm": 0.7558372020721436,
      "learning_rate": 0.002828,
      "loss": 1.0642,
      "step": 598400
    },
    {
      "epoch": 44.912870544090055,
      "grad_norm": 0.5604377388954163,
      "learning_rate": 0.002828,
      "loss": 1.0558,
      "step": 598464
    },
    {
      "epoch": 44.91767354596623,
      "grad_norm": 0.6405630111694336,
      "learning_rate": 0.002828,
      "loss": 1.0626,
      "step": 598528
    },
    {
      "epoch": 44.9224765478424,
      "grad_norm": 0.6783610582351685,
      "learning_rate": 0.002828,
      "loss": 1.0597,
      "step": 598592
    },
    {
      "epoch": 44.92727954971858,
      "grad_norm": 0.6869999766349792,
      "learning_rate": 0.002828,
      "loss": 1.0595,
      "step": 598656
    },
    {
      "epoch": 44.932082551594746,
      "grad_norm": 0.5955403447151184,
      "learning_rate": 0.002828,
      "loss": 1.058,
      "step": 598720
    },
    {
      "epoch": 44.93688555347092,
      "grad_norm": 0.5020884275436401,
      "learning_rate": 0.002828,
      "loss": 1.0592,
      "step": 598784
    },
    {
      "epoch": 44.94168855534709,
      "grad_norm": 0.6526339054107666,
      "learning_rate": 0.002828,
      "loss": 1.0546,
      "step": 598848
    },
    {
      "epoch": 44.94649155722326,
      "grad_norm": 0.7076884508132935,
      "learning_rate": 0.002828,
      "loss": 1.0607,
      "step": 598912
    },
    {
      "epoch": 44.95129455909944,
      "grad_norm": 0.7215730547904968,
      "learning_rate": 0.002828,
      "loss": 1.0632,
      "step": 598976
    },
    {
      "epoch": 44.95609756097561,
      "grad_norm": 0.6171411871910095,
      "learning_rate": 0.002828,
      "loss": 1.0619,
      "step": 599040
    },
    {
      "epoch": 44.96090056285178,
      "grad_norm": 0.5962241291999817,
      "learning_rate": 0.002828,
      "loss": 1.0596,
      "step": 599104
    },
    {
      "epoch": 44.96570356472795,
      "grad_norm": 0.5820527672767639,
      "learning_rate": 0.002828,
      "loss": 1.0633,
      "step": 599168
    },
    {
      "epoch": 44.97050656660413,
      "grad_norm": 0.5785463452339172,
      "learning_rate": 0.002828,
      "loss": 1.0584,
      "step": 599232
    },
    {
      "epoch": 44.9753095684803,
      "grad_norm": 0.6721450090408325,
      "learning_rate": 0.002828,
      "loss": 1.0603,
      "step": 599296
    },
    {
      "epoch": 44.980112570356475,
      "grad_norm": 0.5723146200180054,
      "learning_rate": 0.002828,
      "loss": 1.0643,
      "step": 599360
    },
    {
      "epoch": 44.984915572232644,
      "grad_norm": 0.5571084022521973,
      "learning_rate": 0.002828,
      "loss": 1.0574,
      "step": 599424
    },
    {
      "epoch": 44.98971857410882,
      "grad_norm": 0.6663368344306946,
      "learning_rate": 0.002828,
      "loss": 1.0591,
      "step": 599488
    },
    {
      "epoch": 44.99452157598499,
      "grad_norm": 0.6615773439407349,
      "learning_rate": 0.002828,
      "loss": 1.0612,
      "step": 599552
    },
    {
      "epoch": 44.99932457786117,
      "grad_norm": 0.6352889537811279,
      "learning_rate": 0.002828,
      "loss": 1.0617,
      "step": 599616
    },
    {
      "epoch": 45.004127579737336,
      "grad_norm": 0.6930376291275024,
      "learning_rate": 0.002828,
      "loss": 1.0447,
      "step": 599680
    },
    {
      "epoch": 45.008930581613505,
      "grad_norm": 0.6235330104827881,
      "learning_rate": 0.002828,
      "loss": 1.0442,
      "step": 599744
    },
    {
      "epoch": 45.01373358348968,
      "grad_norm": 0.6568100452423096,
      "learning_rate": 0.002828,
      "loss": 1.0478,
      "step": 599808
    },
    {
      "epoch": 45.01853658536585,
      "grad_norm": 0.6526299118995667,
      "learning_rate": 0.002828,
      "loss": 1.0404,
      "step": 599872
    },
    {
      "epoch": 45.02333958724203,
      "grad_norm": 0.7410742044448853,
      "learning_rate": 0.002828,
      "loss": 1.0454,
      "step": 599936
    },
    {
      "epoch": 45.0281425891182,
      "grad_norm": 0.6709450483322144,
      "learning_rate": 0.002828,
      "loss": 1.0456,
      "step": 600000
    },
    {
      "epoch": 45.03294559099437,
      "grad_norm": 0.6473307013511658,
      "learning_rate": 0.002828,
      "loss": 1.05,
      "step": 600064
    },
    {
      "epoch": 45.03774859287054,
      "grad_norm": 0.6768342852592468,
      "learning_rate": 0.002828,
      "loss": 1.0417,
      "step": 600128
    },
    {
      "epoch": 45.04255159474672,
      "grad_norm": 0.5911619067192078,
      "learning_rate": 0.002828,
      "loss": 1.0458,
      "step": 600192
    },
    {
      "epoch": 45.04735459662289,
      "grad_norm": 0.5454323291778564,
      "learning_rate": 0.002828,
      "loss": 1.0384,
      "step": 600256
    },
    {
      "epoch": 45.052157598499065,
      "grad_norm": 0.7044894099235535,
      "learning_rate": 0.002828,
      "loss": 1.0402,
      "step": 600320
    },
    {
      "epoch": 45.056960600375234,
      "grad_norm": 0.6176587343215942,
      "learning_rate": 0.002828,
      "loss": 1.0492,
      "step": 600384
    },
    {
      "epoch": 45.06176360225141,
      "grad_norm": 0.6300610303878784,
      "learning_rate": 0.002828,
      "loss": 1.044,
      "step": 600448
    },
    {
      "epoch": 45.06656660412758,
      "grad_norm": 0.7107115387916565,
      "learning_rate": 0.002828,
      "loss": 1.0383,
      "step": 600512
    },
    {
      "epoch": 45.07136960600375,
      "grad_norm": 0.6533573865890503,
      "learning_rate": 0.002828,
      "loss": 1.0458,
      "step": 600576
    },
    {
      "epoch": 45.076172607879926,
      "grad_norm": 0.6106192469596863,
      "learning_rate": 0.002828,
      "loss": 1.0454,
      "step": 600640
    },
    {
      "epoch": 45.080975609756095,
      "grad_norm": 0.7222683429718018,
      "learning_rate": 0.002828,
      "loss": 1.0459,
      "step": 600704
    },
    {
      "epoch": 45.08577861163227,
      "grad_norm": 0.6331077814102173,
      "learning_rate": 0.002828,
      "loss": 1.0471,
      "step": 600768
    },
    {
      "epoch": 45.09058161350844,
      "grad_norm": 0.5737351179122925,
      "learning_rate": 0.002828,
      "loss": 1.0424,
      "step": 600832
    },
    {
      "epoch": 45.09538461538462,
      "grad_norm": 0.6387905478477478,
      "learning_rate": 0.002828,
      "loss": 1.0538,
      "step": 600896
    },
    {
      "epoch": 45.10018761726079,
      "grad_norm": 0.558758556842804,
      "learning_rate": 0.002828,
      "loss": 1.05,
      "step": 600960
    },
    {
      "epoch": 45.10499061913696,
      "grad_norm": 0.6962189078330994,
      "learning_rate": 0.002828,
      "loss": 1.0503,
      "step": 601024
    },
    {
      "epoch": 45.10979362101313,
      "grad_norm": 0.6257447600364685,
      "learning_rate": 0.002828,
      "loss": 1.0464,
      "step": 601088
    },
    {
      "epoch": 45.11459662288931,
      "grad_norm": 0.6099736094474792,
      "learning_rate": 0.002828,
      "loss": 1.0466,
      "step": 601152
    },
    {
      "epoch": 45.11939962476548,
      "grad_norm": 0.6403970718383789,
      "learning_rate": 0.002828,
      "loss": 1.0487,
      "step": 601216
    },
    {
      "epoch": 45.124202626641654,
      "grad_norm": 0.8435704708099365,
      "learning_rate": 0.002828,
      "loss": 1.0434,
      "step": 601280
    },
    {
      "epoch": 45.129005628517824,
      "grad_norm": 0.6793568134307861,
      "learning_rate": 0.002828,
      "loss": 1.0476,
      "step": 601344
    },
    {
      "epoch": 45.13380863039399,
      "grad_norm": 0.5509681701660156,
      "learning_rate": 0.002828,
      "loss": 1.0538,
      "step": 601408
    },
    {
      "epoch": 45.13861163227017,
      "grad_norm": 0.5556575655937195,
      "learning_rate": 0.002828,
      "loss": 1.0492,
      "step": 601472
    },
    {
      "epoch": 45.14341463414634,
      "grad_norm": 0.6454841494560242,
      "learning_rate": 0.002828,
      "loss": 1.0442,
      "step": 601536
    },
    {
      "epoch": 45.148217636022515,
      "grad_norm": 0.5697574019432068,
      "learning_rate": 0.002828,
      "loss": 1.0418,
      "step": 601600
    },
    {
      "epoch": 45.153020637898685,
      "grad_norm": 0.8331282734870911,
      "learning_rate": 0.002828,
      "loss": 1.0482,
      "step": 601664
    },
    {
      "epoch": 45.15782363977486,
      "grad_norm": 0.6625939011573792,
      "learning_rate": 0.002828,
      "loss": 1.0433,
      "step": 601728
    },
    {
      "epoch": 45.16262664165103,
      "grad_norm": 0.7174499034881592,
      "learning_rate": 0.002828,
      "loss": 1.05,
      "step": 601792
    },
    {
      "epoch": 45.16742964352721,
      "grad_norm": 0.7669702768325806,
      "learning_rate": 0.002828,
      "loss": 1.0489,
      "step": 601856
    },
    {
      "epoch": 45.172232645403376,
      "grad_norm": 0.6364301443099976,
      "learning_rate": 0.002828,
      "loss": 1.0477,
      "step": 601920
    },
    {
      "epoch": 45.17703564727955,
      "grad_norm": 0.5847406387329102,
      "learning_rate": 0.002828,
      "loss": 1.057,
      "step": 601984
    },
    {
      "epoch": 45.18183864915572,
      "grad_norm": 0.6891112923622131,
      "learning_rate": 0.002828,
      "loss": 1.0542,
      "step": 602048
    },
    {
      "epoch": 45.1866416510319,
      "grad_norm": 0.559404194355011,
      "learning_rate": 0.002828,
      "loss": 1.051,
      "step": 602112
    },
    {
      "epoch": 45.19144465290807,
      "grad_norm": 0.6273938417434692,
      "learning_rate": 0.002828,
      "loss": 1.0541,
      "step": 602176
    },
    {
      "epoch": 45.19624765478424,
      "grad_norm": 0.6550655364990234,
      "learning_rate": 0.002828,
      "loss": 1.0464,
      "step": 602240
    },
    {
      "epoch": 45.20105065666041,
      "grad_norm": 0.6907812356948853,
      "learning_rate": 0.002828,
      "loss": 1.0524,
      "step": 602304
    },
    {
      "epoch": 45.20585365853658,
      "grad_norm": 0.5884630084037781,
      "learning_rate": 0.002828,
      "loss": 1.0524,
      "step": 602368
    },
    {
      "epoch": 45.21065666041276,
      "grad_norm": 0.5913063287734985,
      "learning_rate": 0.002828,
      "loss": 1.053,
      "step": 602432
    },
    {
      "epoch": 45.21545966228893,
      "grad_norm": 0.6427793502807617,
      "learning_rate": 0.002828,
      "loss": 1.0524,
      "step": 602496
    },
    {
      "epoch": 45.220262664165105,
      "grad_norm": 0.5619731545448303,
      "learning_rate": 0.002828,
      "loss": 1.0514,
      "step": 602560
    },
    {
      "epoch": 45.225065666041274,
      "grad_norm": 0.5781331062316895,
      "learning_rate": 0.002828,
      "loss": 1.0546,
      "step": 602624
    },
    {
      "epoch": 45.22986866791745,
      "grad_norm": 0.5728752613067627,
      "learning_rate": 0.002828,
      "loss": 1.0498,
      "step": 602688
    },
    {
      "epoch": 45.23467166979362,
      "grad_norm": 0.5659903883934021,
      "learning_rate": 0.002828,
      "loss": 1.0552,
      "step": 602752
    },
    {
      "epoch": 45.2394746716698,
      "grad_norm": 0.6563093066215515,
      "learning_rate": 0.002828,
      "loss": 1.0511,
      "step": 602816
    },
    {
      "epoch": 45.244277673545966,
      "grad_norm": 0.5482872724533081,
      "learning_rate": 0.002828,
      "loss": 1.0545,
      "step": 602880
    },
    {
      "epoch": 45.249080675422135,
      "grad_norm": 0.5136545300483704,
      "learning_rate": 0.002828,
      "loss": 1.0573,
      "step": 602944
    },
    {
      "epoch": 45.25388367729831,
      "grad_norm": 0.5944323539733887,
      "learning_rate": 0.002828,
      "loss": 1.0615,
      "step": 603008
    },
    {
      "epoch": 45.25868667917448,
      "grad_norm": 0.8258227109909058,
      "learning_rate": 0.002828,
      "loss": 1.0524,
      "step": 603072
    },
    {
      "epoch": 45.26348968105066,
      "grad_norm": 0.68766850233078,
      "learning_rate": 0.002828,
      "loss": 1.0561,
      "step": 603136
    },
    {
      "epoch": 45.26829268292683,
      "grad_norm": 0.5555733442306519,
      "learning_rate": 0.002828,
      "loss": 1.0525,
      "step": 603200
    },
    {
      "epoch": 45.273095684803,
      "grad_norm": 0.7993525266647339,
      "learning_rate": 0.002828,
      "loss": 1.0562,
      "step": 603264
    },
    {
      "epoch": 45.27789868667917,
      "grad_norm": 0.6797704696655273,
      "learning_rate": 0.002828,
      "loss": 1.053,
      "step": 603328
    },
    {
      "epoch": 45.28270168855535,
      "grad_norm": 0.8095536231994629,
      "learning_rate": 0.002828,
      "loss": 1.0536,
      "step": 603392
    },
    {
      "epoch": 45.28750469043152,
      "grad_norm": 0.6387770175933838,
      "learning_rate": 0.002828,
      "loss": 1.0532,
      "step": 603456
    },
    {
      "epoch": 45.292307692307695,
      "grad_norm": 0.7661316990852356,
      "learning_rate": 0.002828,
      "loss": 1.0567,
      "step": 603520
    },
    {
      "epoch": 45.297110694183864,
      "grad_norm": 0.6616008877754211,
      "learning_rate": 0.002828,
      "loss": 1.0502,
      "step": 603584
    },
    {
      "epoch": 45.30191369606004,
      "grad_norm": 0.6984626054763794,
      "learning_rate": 0.002828,
      "loss": 1.049,
      "step": 603648
    },
    {
      "epoch": 45.30671669793621,
      "grad_norm": 0.6384255886077881,
      "learning_rate": 0.002828,
      "loss": 1.058,
      "step": 603712
    },
    {
      "epoch": 45.31151969981238,
      "grad_norm": 0.6473172307014465,
      "learning_rate": 0.002828,
      "loss": 1.0551,
      "step": 603776
    },
    {
      "epoch": 45.316322701688556,
      "grad_norm": 0.7297527194023132,
      "learning_rate": 0.002828,
      "loss": 1.0615,
      "step": 603840
    },
    {
      "epoch": 45.321125703564725,
      "grad_norm": 0.6076101064682007,
      "learning_rate": 0.002828,
      "loss": 1.0537,
      "step": 603904
    },
    {
      "epoch": 45.3259287054409,
      "grad_norm": 0.6504645347595215,
      "learning_rate": 0.002828,
      "loss": 1.054,
      "step": 603968
    },
    {
      "epoch": 45.33073170731707,
      "grad_norm": 0.7188733220100403,
      "learning_rate": 0.002828,
      "loss": 1.0501,
      "step": 604032
    },
    {
      "epoch": 45.33553470919325,
      "grad_norm": 0.6785242557525635,
      "learning_rate": 0.002828,
      "loss": 1.0502,
      "step": 604096
    },
    {
      "epoch": 45.34033771106942,
      "grad_norm": 0.7085686326026917,
      "learning_rate": 0.002828,
      "loss": 1.0577,
      "step": 604160
    },
    {
      "epoch": 45.34514071294559,
      "grad_norm": 0.5494343638420105,
      "learning_rate": 0.002828,
      "loss": 1.0569,
      "step": 604224
    },
    {
      "epoch": 45.34994371482176,
      "grad_norm": 0.6185113787651062,
      "learning_rate": 0.002828,
      "loss": 1.0554,
      "step": 604288
    },
    {
      "epoch": 45.35474671669794,
      "grad_norm": 0.6287076473236084,
      "learning_rate": 0.002828,
      "loss": 1.059,
      "step": 604352
    },
    {
      "epoch": 45.35954971857411,
      "grad_norm": 0.6744497418403625,
      "learning_rate": 0.002828,
      "loss": 1.0573,
      "step": 604416
    },
    {
      "epoch": 45.364352720450285,
      "grad_norm": 0.6540379524230957,
      "learning_rate": 0.002828,
      "loss": 1.0519,
      "step": 604480
    },
    {
      "epoch": 45.369155722326454,
      "grad_norm": 0.648766040802002,
      "learning_rate": 0.002828,
      "loss": 1.0582,
      "step": 604544
    },
    {
      "epoch": 45.37395872420262,
      "grad_norm": 0.7529312968254089,
      "learning_rate": 0.002828,
      "loss": 1.0562,
      "step": 604608
    },
    {
      "epoch": 45.3787617260788,
      "grad_norm": 0.6116892695426941,
      "learning_rate": 0.002828,
      "loss": 1.0572,
      "step": 604672
    },
    {
      "epoch": 45.38356472795497,
      "grad_norm": 0.6726174354553223,
      "learning_rate": 0.002828,
      "loss": 1.0555,
      "step": 604736
    },
    {
      "epoch": 45.388367729831145,
      "grad_norm": 0.5719484090805054,
      "learning_rate": 0.002828,
      "loss": 1.0567,
      "step": 604800
    },
    {
      "epoch": 45.393170731707315,
      "grad_norm": 0.7066259980201721,
      "learning_rate": 0.002828,
      "loss": 1.0599,
      "step": 604864
    },
    {
      "epoch": 45.39797373358349,
      "grad_norm": 0.6202705502510071,
      "learning_rate": 0.002828,
      "loss": 1.0587,
      "step": 604928
    },
    {
      "epoch": 45.40277673545966,
      "grad_norm": 0.8698914051055908,
      "learning_rate": 0.002828,
      "loss": 1.0569,
      "step": 604992
    },
    {
      "epoch": 45.40757973733584,
      "grad_norm": 0.6540534496307373,
      "learning_rate": 0.002828,
      "loss": 1.0578,
      "step": 605056
    },
    {
      "epoch": 45.412382739212006,
      "grad_norm": 0.5570470094680786,
      "learning_rate": 0.002828,
      "loss": 1.0609,
      "step": 605120
    },
    {
      "epoch": 45.41718574108818,
      "grad_norm": 0.7164893746376038,
      "learning_rate": 0.002828,
      "loss": 1.0548,
      "step": 605184
    },
    {
      "epoch": 45.42198874296435,
      "grad_norm": 0.5541973114013672,
      "learning_rate": 0.002828,
      "loss": 1.0545,
      "step": 605248
    },
    {
      "epoch": 45.42679174484053,
      "grad_norm": 0.7188869118690491,
      "learning_rate": 0.002828,
      "loss": 1.0525,
      "step": 605312
    },
    {
      "epoch": 45.4315947467167,
      "grad_norm": 0.6505350470542908,
      "learning_rate": 0.002828,
      "loss": 1.0601,
      "step": 605376
    },
    {
      "epoch": 45.43639774859287,
      "grad_norm": 0.5822009444236755,
      "learning_rate": 0.002828,
      "loss": 1.0615,
      "step": 605440
    },
    {
      "epoch": 45.441200750469044,
      "grad_norm": 0.7058424353599548,
      "learning_rate": 0.002828,
      "loss": 1.0595,
      "step": 605504
    },
    {
      "epoch": 45.44600375234521,
      "grad_norm": 0.5275965929031372,
      "learning_rate": 0.002828,
      "loss": 1.0624,
      "step": 605568
    },
    {
      "epoch": 45.45080675422139,
      "grad_norm": 0.5741568207740784,
      "learning_rate": 0.002828,
      "loss": 1.0561,
      "step": 605632
    },
    {
      "epoch": 45.45560975609756,
      "grad_norm": 0.6492908596992493,
      "learning_rate": 0.002828,
      "loss": 1.063,
      "step": 605696
    },
    {
      "epoch": 45.460412757973735,
      "grad_norm": 0.5441504716873169,
      "learning_rate": 0.002828,
      "loss": 1.0581,
      "step": 605760
    },
    {
      "epoch": 45.465215759849904,
      "grad_norm": 0.6401174664497375,
      "learning_rate": 0.002828,
      "loss": 1.0585,
      "step": 605824
    },
    {
      "epoch": 45.47001876172608,
      "grad_norm": 0.6390235424041748,
      "learning_rate": 0.002828,
      "loss": 1.0655,
      "step": 605888
    },
    {
      "epoch": 45.47482176360225,
      "grad_norm": 0.6424334049224854,
      "learning_rate": 0.002828,
      "loss": 1.0606,
      "step": 605952
    },
    {
      "epoch": 45.47962476547843,
      "grad_norm": 0.5485963821411133,
      "learning_rate": 0.002828,
      "loss": 1.0646,
      "step": 606016
    },
    {
      "epoch": 45.484427767354596,
      "grad_norm": 0.8131200075149536,
      "learning_rate": 0.002828,
      "loss": 1.058,
      "step": 606080
    },
    {
      "epoch": 45.48923076923077,
      "grad_norm": 0.7325833439826965,
      "learning_rate": 0.002828,
      "loss": 1.0633,
      "step": 606144
    },
    {
      "epoch": 45.49403377110694,
      "grad_norm": 0.6808172464370728,
      "learning_rate": 0.002828,
      "loss": 1.0572,
      "step": 606208
    },
    {
      "epoch": 45.49883677298311,
      "grad_norm": 0.7559350728988647,
      "learning_rate": 0.002828,
      "loss": 1.0633,
      "step": 606272
    },
    {
      "epoch": 45.50363977485929,
      "grad_norm": 0.723712146282196,
      "learning_rate": 0.002828,
      "loss": 1.0573,
      "step": 606336
    },
    {
      "epoch": 45.50844277673546,
      "grad_norm": 0.7043648958206177,
      "learning_rate": 0.002828,
      "loss": 1.059,
      "step": 606400
    },
    {
      "epoch": 45.51324577861163,
      "grad_norm": 0.630072832107544,
      "learning_rate": 0.002828,
      "loss": 1.0598,
      "step": 606464
    },
    {
      "epoch": 45.5180487804878,
      "grad_norm": 0.6712070107460022,
      "learning_rate": 0.002828,
      "loss": 1.0591,
      "step": 606528
    },
    {
      "epoch": 45.52285178236398,
      "grad_norm": 0.6040600538253784,
      "learning_rate": 0.002828,
      "loss": 1.0606,
      "step": 606592
    },
    {
      "epoch": 45.52765478424015,
      "grad_norm": 0.6480143666267395,
      "learning_rate": 0.002828,
      "loss": 1.0629,
      "step": 606656
    },
    {
      "epoch": 45.532457786116325,
      "grad_norm": 0.5866736769676208,
      "learning_rate": 0.002828,
      "loss": 1.0608,
      "step": 606720
    },
    {
      "epoch": 45.537260787992494,
      "grad_norm": 0.5952351093292236,
      "learning_rate": 0.002828,
      "loss": 1.0573,
      "step": 606784
    },
    {
      "epoch": 45.54206378986867,
      "grad_norm": 0.6766398549079895,
      "learning_rate": 0.002828,
      "loss": 1.0595,
      "step": 606848
    },
    {
      "epoch": 45.54686679174484,
      "grad_norm": 0.71608966588974,
      "learning_rate": 0.002828,
      "loss": 1.0592,
      "step": 606912
    },
    {
      "epoch": 45.551669793621016,
      "grad_norm": 0.7299851179122925,
      "learning_rate": 0.002828,
      "loss": 1.0594,
      "step": 606976
    },
    {
      "epoch": 45.556472795497186,
      "grad_norm": 0.6860955953598022,
      "learning_rate": 0.002828,
      "loss": 1.0664,
      "step": 607040
    },
    {
      "epoch": 45.561275797373355,
      "grad_norm": 0.5690156817436218,
      "learning_rate": 0.002828,
      "loss": 1.0674,
      "step": 607104
    },
    {
      "epoch": 45.56607879924953,
      "grad_norm": 0.7102258801460266,
      "learning_rate": 0.002828,
      "loss": 1.0603,
      "step": 607168
    },
    {
      "epoch": 45.5708818011257,
      "grad_norm": 0.5700670480728149,
      "learning_rate": 0.002828,
      "loss": 1.0582,
      "step": 607232
    },
    {
      "epoch": 45.57568480300188,
      "grad_norm": 0.628460168838501,
      "learning_rate": 0.002828,
      "loss": 1.0626,
      "step": 607296
    },
    {
      "epoch": 45.58048780487805,
      "grad_norm": 0.5459248423576355,
      "learning_rate": 0.002828,
      "loss": 1.0644,
      "step": 607360
    },
    {
      "epoch": 45.58529080675422,
      "grad_norm": 0.7937043905258179,
      "learning_rate": 0.002828,
      "loss": 1.0617,
      "step": 607424
    },
    {
      "epoch": 45.59009380863039,
      "grad_norm": 0.8268114924430847,
      "learning_rate": 0.002828,
      "loss": 1.0699,
      "step": 607488
    },
    {
      "epoch": 45.59489681050657,
      "grad_norm": 0.5842559337615967,
      "learning_rate": 0.002828,
      "loss": 1.0617,
      "step": 607552
    },
    {
      "epoch": 45.59969981238274,
      "grad_norm": 0.5671669840812683,
      "learning_rate": 0.002828,
      "loss": 1.0591,
      "step": 607616
    },
    {
      "epoch": 45.604502814258915,
      "grad_norm": 0.7660974264144897,
      "learning_rate": 0.002828,
      "loss": 1.0595,
      "step": 607680
    },
    {
      "epoch": 45.609305816135084,
      "grad_norm": 0.6504864692687988,
      "learning_rate": 0.002828,
      "loss": 1.0621,
      "step": 607744
    },
    {
      "epoch": 45.61410881801126,
      "grad_norm": 0.590907633304596,
      "learning_rate": 0.002828,
      "loss": 1.0661,
      "step": 607808
    },
    {
      "epoch": 45.61891181988743,
      "grad_norm": 0.7128926515579224,
      "learning_rate": 0.002828,
      "loss": 1.0601,
      "step": 607872
    },
    {
      "epoch": 45.6237148217636,
      "grad_norm": 0.643826961517334,
      "learning_rate": 0.002828,
      "loss": 1.0661,
      "step": 607936
    },
    {
      "epoch": 45.628517823639775,
      "grad_norm": 0.6559561491012573,
      "learning_rate": 0.002828,
      "loss": 1.0649,
      "step": 608000
    },
    {
      "epoch": 45.633320825515945,
      "grad_norm": 0.6678355932235718,
      "learning_rate": 0.002828,
      "loss": 1.0637,
      "step": 608064
    },
    {
      "epoch": 45.63812382739212,
      "grad_norm": 0.5866308808326721,
      "learning_rate": 0.002828,
      "loss": 1.0636,
      "step": 608128
    },
    {
      "epoch": 45.64292682926829,
      "grad_norm": 0.4977904260158539,
      "learning_rate": 0.002828,
      "loss": 1.064,
      "step": 608192
    },
    {
      "epoch": 45.64772983114447,
      "grad_norm": 0.6667592525482178,
      "learning_rate": 0.002828,
      "loss": 1.0693,
      "step": 608256
    },
    {
      "epoch": 45.652532833020636,
      "grad_norm": 0.571887195110321,
      "learning_rate": 0.002828,
      "loss": 1.0597,
      "step": 608320
    },
    {
      "epoch": 45.65733583489681,
      "grad_norm": 0.7032132744789124,
      "learning_rate": 0.002828,
      "loss": 1.0719,
      "step": 608384
    },
    {
      "epoch": 45.66213883677298,
      "grad_norm": 0.5857048630714417,
      "learning_rate": 0.002828,
      "loss": 1.0604,
      "step": 608448
    },
    {
      "epoch": 45.66694183864916,
      "grad_norm": 0.5976815223693848,
      "learning_rate": 0.002828,
      "loss": 1.0618,
      "step": 608512
    },
    {
      "epoch": 45.67174484052533,
      "grad_norm": 0.59645015001297,
      "learning_rate": 0.002828,
      "loss": 1.0643,
      "step": 608576
    },
    {
      "epoch": 45.676547842401504,
      "grad_norm": 0.6468033790588379,
      "learning_rate": 0.002828,
      "loss": 1.0637,
      "step": 608640
    },
    {
      "epoch": 45.681350844277674,
      "grad_norm": 0.5100387334823608,
      "learning_rate": 0.002828,
      "loss": 1.0612,
      "step": 608704
    },
    {
      "epoch": 45.68615384615384,
      "grad_norm": 0.6013880968093872,
      "learning_rate": 0.002828,
      "loss": 1.0661,
      "step": 608768
    },
    {
      "epoch": 45.69095684803002,
      "grad_norm": 0.7553115487098694,
      "learning_rate": 0.002828,
      "loss": 1.0637,
      "step": 608832
    },
    {
      "epoch": 45.69575984990619,
      "grad_norm": 0.7213432788848877,
      "learning_rate": 0.002828,
      "loss": 1.0638,
      "step": 608896
    },
    {
      "epoch": 45.700562851782365,
      "grad_norm": 0.7093005776405334,
      "learning_rate": 0.002828,
      "loss": 1.0617,
      "step": 608960
    },
    {
      "epoch": 45.705365853658535,
      "grad_norm": 0.5940427780151367,
      "learning_rate": 0.002828,
      "loss": 1.0667,
      "step": 609024
    },
    {
      "epoch": 45.71016885553471,
      "grad_norm": 0.5454390048980713,
      "learning_rate": 0.002828,
      "loss": 1.0621,
      "step": 609088
    },
    {
      "epoch": 45.71497185741088,
      "grad_norm": 0.6252509951591492,
      "learning_rate": 0.002828,
      "loss": 1.0669,
      "step": 609152
    },
    {
      "epoch": 45.71977485928706,
      "grad_norm": 0.7278851270675659,
      "learning_rate": 0.002828,
      "loss": 1.0644,
      "step": 609216
    },
    {
      "epoch": 45.724577861163226,
      "grad_norm": 0.6815802454948425,
      "learning_rate": 0.002828,
      "loss": 1.0657,
      "step": 609280
    },
    {
      "epoch": 45.7293808630394,
      "grad_norm": 0.6935381889343262,
      "learning_rate": 0.002828,
      "loss": 1.0701,
      "step": 609344
    },
    {
      "epoch": 45.73418386491557,
      "grad_norm": 0.5729061365127563,
      "learning_rate": 0.002828,
      "loss": 1.0662,
      "step": 609408
    },
    {
      "epoch": 45.73898686679175,
      "grad_norm": 0.6517598628997803,
      "learning_rate": 0.002828,
      "loss": 1.0643,
      "step": 609472
    },
    {
      "epoch": 45.74378986866792,
      "grad_norm": 0.6475855112075806,
      "learning_rate": 0.002828,
      "loss": 1.0675,
      "step": 609536
    },
    {
      "epoch": 45.74859287054409,
      "grad_norm": 0.7667732834815979,
      "learning_rate": 0.002828,
      "loss": 1.0653,
      "step": 609600
    },
    {
      "epoch": 45.75339587242026,
      "grad_norm": 0.782156229019165,
      "learning_rate": 0.002828,
      "loss": 1.062,
      "step": 609664
    },
    {
      "epoch": 45.75819887429643,
      "grad_norm": 0.6261060237884521,
      "learning_rate": 0.002828,
      "loss": 1.0675,
      "step": 609728
    },
    {
      "epoch": 45.76300187617261,
      "grad_norm": 0.5714720487594604,
      "learning_rate": 0.002828,
      "loss": 1.0648,
      "step": 609792
    },
    {
      "epoch": 45.76780487804878,
      "grad_norm": 0.7018872499465942,
      "learning_rate": 0.002828,
      "loss": 1.0621,
      "step": 609856
    },
    {
      "epoch": 45.772607879924955,
      "grad_norm": 0.5345858931541443,
      "learning_rate": 0.002828,
      "loss": 1.0628,
      "step": 609920
    },
    {
      "epoch": 45.777410881801124,
      "grad_norm": 0.728009819984436,
      "learning_rate": 0.002828,
      "loss": 1.0657,
      "step": 609984
    },
    {
      "epoch": 45.7822138836773,
      "grad_norm": 0.5718938112258911,
      "learning_rate": 0.002828,
      "loss": 1.0606,
      "step": 610048
    },
    {
      "epoch": 45.78701688555347,
      "grad_norm": 0.533854603767395,
      "learning_rate": 0.002828,
      "loss": 1.0623,
      "step": 610112
    },
    {
      "epoch": 45.79181988742965,
      "grad_norm": 0.5917041301727295,
      "learning_rate": 0.002828,
      "loss": 1.0697,
      "step": 610176
    },
    {
      "epoch": 45.796622889305816,
      "grad_norm": 0.7023472785949707,
      "learning_rate": 0.002828,
      "loss": 1.0664,
      "step": 610240
    },
    {
      "epoch": 45.80142589118199,
      "grad_norm": 0.5526425838470459,
      "learning_rate": 0.002828,
      "loss": 1.0667,
      "step": 610304
    },
    {
      "epoch": 45.80622889305816,
      "grad_norm": 0.7164841294288635,
      "learning_rate": 0.002828,
      "loss": 1.0693,
      "step": 610368
    },
    {
      "epoch": 45.81103189493433,
      "grad_norm": 0.6742995381355286,
      "learning_rate": 0.002828,
      "loss": 1.0676,
      "step": 610432
    },
    {
      "epoch": 45.81583489681051,
      "grad_norm": 0.5527735948562622,
      "learning_rate": 0.002828,
      "loss": 1.0698,
      "step": 610496
    },
    {
      "epoch": 45.82063789868668,
      "grad_norm": 0.6233502626419067,
      "learning_rate": 0.002828,
      "loss": 1.065,
      "step": 610560
    },
    {
      "epoch": 45.82544090056285,
      "grad_norm": 0.5897585153579712,
      "learning_rate": 0.002828,
      "loss": 1.0648,
      "step": 610624
    },
    {
      "epoch": 45.83024390243902,
      "grad_norm": 0.6187131404876709,
      "learning_rate": 0.002828,
      "loss": 1.0725,
      "step": 610688
    },
    {
      "epoch": 45.8350469043152,
      "grad_norm": 0.7255659699440002,
      "learning_rate": 0.002828,
      "loss": 1.0625,
      "step": 610752
    },
    {
      "epoch": 45.83984990619137,
      "grad_norm": 0.9766494631767273,
      "learning_rate": 0.002828,
      "loss": 1.0674,
      "step": 610816
    },
    {
      "epoch": 45.844652908067545,
      "grad_norm": 0.6275309324264526,
      "learning_rate": 0.002828,
      "loss": 1.0621,
      "step": 610880
    },
    {
      "epoch": 45.849455909943714,
      "grad_norm": 0.6769102811813354,
      "learning_rate": 0.002828,
      "loss": 1.0727,
      "step": 610944
    },
    {
      "epoch": 45.85425891181989,
      "grad_norm": 0.5380213260650635,
      "learning_rate": 0.002828,
      "loss": 1.0707,
      "step": 611008
    },
    {
      "epoch": 45.85906191369606,
      "grad_norm": 0.7997820377349854,
      "learning_rate": 0.002828,
      "loss": 1.0693,
      "step": 611072
    },
    {
      "epoch": 45.863864915572236,
      "grad_norm": 0.5203222632408142,
      "learning_rate": 0.002828,
      "loss": 1.0704,
      "step": 611136
    },
    {
      "epoch": 45.868667917448406,
      "grad_norm": 0.6101234555244446,
      "learning_rate": 0.002828,
      "loss": 1.0677,
      "step": 611200
    },
    {
      "epoch": 45.873470919324575,
      "grad_norm": 0.5174065232276917,
      "learning_rate": 0.002828,
      "loss": 1.0717,
      "step": 611264
    },
    {
      "epoch": 45.87827392120075,
      "grad_norm": 0.6431335210800171,
      "learning_rate": 0.002828,
      "loss": 1.069,
      "step": 611328
    },
    {
      "epoch": 45.88307692307692,
      "grad_norm": 0.5777249932289124,
      "learning_rate": 0.002828,
      "loss": 1.0649,
      "step": 611392
    },
    {
      "epoch": 45.8878799249531,
      "grad_norm": 0.5198315978050232,
      "learning_rate": 0.002828,
      "loss": 1.0688,
      "step": 611456
    },
    {
      "epoch": 45.892682926829266,
      "grad_norm": 0.7460737824440002,
      "learning_rate": 0.002828,
      "loss": 1.0706,
      "step": 611520
    },
    {
      "epoch": 45.89748592870544,
      "grad_norm": 0.6131591796875,
      "learning_rate": 0.002828,
      "loss": 1.0666,
      "step": 611584
    },
    {
      "epoch": 45.90228893058161,
      "grad_norm": 0.7207826972007751,
      "learning_rate": 0.002828,
      "loss": 1.0662,
      "step": 611648
    },
    {
      "epoch": 45.90709193245779,
      "grad_norm": 0.5071738362312317,
      "learning_rate": 0.002828,
      "loss": 1.0707,
      "step": 611712
    },
    {
      "epoch": 45.91189493433396,
      "grad_norm": 0.638817310333252,
      "learning_rate": 0.002828,
      "loss": 1.0696,
      "step": 611776
    },
    {
      "epoch": 45.916697936210134,
      "grad_norm": 0.5947349071502686,
      "learning_rate": 0.002828,
      "loss": 1.0707,
      "step": 611840
    },
    {
      "epoch": 45.921500938086304,
      "grad_norm": 0.7027119994163513,
      "learning_rate": 0.002828,
      "loss": 1.0681,
      "step": 611904
    },
    {
      "epoch": 45.92630393996247,
      "grad_norm": 0.667210042476654,
      "learning_rate": 0.002828,
      "loss": 1.0658,
      "step": 611968
    },
    {
      "epoch": 45.93110694183865,
      "grad_norm": 0.7699005603790283,
      "learning_rate": 0.002828,
      "loss": 1.0635,
      "step": 612032
    },
    {
      "epoch": 45.93590994371482,
      "grad_norm": 0.7679214477539062,
      "learning_rate": 0.002828,
      "loss": 1.0624,
      "step": 612096
    },
    {
      "epoch": 45.940712945590995,
      "grad_norm": 0.5503223538398743,
      "learning_rate": 0.002828,
      "loss": 1.0688,
      "step": 612160
    },
    {
      "epoch": 45.945515947467165,
      "grad_norm": 0.7197269201278687,
      "learning_rate": 0.002828,
      "loss": 1.0723,
      "step": 612224
    },
    {
      "epoch": 45.95031894934334,
      "grad_norm": 0.8216593861579895,
      "learning_rate": 0.002828,
      "loss": 1.07,
      "step": 612288
    },
    {
      "epoch": 45.95512195121951,
      "grad_norm": 0.7016892433166504,
      "learning_rate": 0.002828,
      "loss": 1.0717,
      "step": 612352
    },
    {
      "epoch": 45.95992495309569,
      "grad_norm": 0.568098783493042,
      "learning_rate": 0.002828,
      "loss": 1.0734,
      "step": 612416
    },
    {
      "epoch": 45.964727954971856,
      "grad_norm": 0.7623087167739868,
      "learning_rate": 0.002828,
      "loss": 1.07,
      "step": 612480
    },
    {
      "epoch": 45.96953095684803,
      "grad_norm": 0.6430860161781311,
      "learning_rate": 0.002828,
      "loss": 1.0715,
      "step": 612544
    },
    {
      "epoch": 45.9743339587242,
      "grad_norm": 0.7953311800956726,
      "learning_rate": 0.002828,
      "loss": 1.0693,
      "step": 612608
    },
    {
      "epoch": 45.97913696060038,
      "grad_norm": 0.555494487285614,
      "learning_rate": 0.002828,
      "loss": 1.0678,
      "step": 612672
    },
    {
      "epoch": 45.98393996247655,
      "grad_norm": 0.5385116934776306,
      "learning_rate": 0.002828,
      "loss": 1.0696,
      "step": 612736
    },
    {
      "epoch": 45.98874296435272,
      "grad_norm": 0.5468719601631165,
      "learning_rate": 0.002828,
      "loss": 1.073,
      "step": 612800
    },
    {
      "epoch": 45.99354596622889,
      "grad_norm": 0.5610455870628357,
      "learning_rate": 0.002828,
      "loss": 1.0707,
      "step": 612864
    },
    {
      "epoch": 45.99834896810506,
      "grad_norm": 0.7394833564758301,
      "learning_rate": 0.002828,
      "loss": 1.0752,
      "step": 612928
    },
    {
      "epoch": 46.00315196998124,
      "grad_norm": 0.529012143611908,
      "learning_rate": 0.002828,
      "loss": 1.037,
      "step": 612992
    },
    {
      "epoch": 46.00795497185741,
      "grad_norm": 0.7306380271911621,
      "learning_rate": 0.002828,
      "loss": 1.042,
      "step": 613056
    },
    {
      "epoch": 46.012757973733585,
      "grad_norm": 0.5341888070106506,
      "learning_rate": 0.002828,
      "loss": 1.036,
      "step": 613120
    },
    {
      "epoch": 46.017560975609754,
      "grad_norm": 0.6116080284118652,
      "learning_rate": 0.002828,
      "loss": 1.0346,
      "step": 613184
    },
    {
      "epoch": 46.02236397748593,
      "grad_norm": 0.6336046457290649,
      "learning_rate": 0.002828,
      "loss": 1.0421,
      "step": 613248
    },
    {
      "epoch": 46.0271669793621,
      "grad_norm": 0.8078818917274475,
      "learning_rate": 0.002828,
      "loss": 1.0385,
      "step": 613312
    },
    {
      "epoch": 46.03196998123828,
      "grad_norm": 0.6536026000976562,
      "learning_rate": 0.002828,
      "loss": 1.0373,
      "step": 613376
    },
    {
      "epoch": 46.036772983114446,
      "grad_norm": 0.6699345111846924,
      "learning_rate": 0.002828,
      "loss": 1.0352,
      "step": 613440
    },
    {
      "epoch": 46.04157598499062,
      "grad_norm": 0.7356686592102051,
      "learning_rate": 0.002828,
      "loss": 1.0307,
      "step": 613504
    },
    {
      "epoch": 46.04637898686679,
      "grad_norm": 0.5745231509208679,
      "learning_rate": 0.002828,
      "loss": 1.0375,
      "step": 613568
    },
    {
      "epoch": 46.05118198874296,
      "grad_norm": 0.5645381212234497,
      "learning_rate": 0.002828,
      "loss": 1.0389,
      "step": 613632
    },
    {
      "epoch": 46.05598499061914,
      "grad_norm": 0.6182840466499329,
      "learning_rate": 0.002828,
      "loss": 1.0426,
      "step": 613696
    },
    {
      "epoch": 46.06078799249531,
      "grad_norm": 0.5951291918754578,
      "learning_rate": 0.002828,
      "loss": 1.0327,
      "step": 613760
    },
    {
      "epoch": 46.06559099437148,
      "grad_norm": 0.7547361850738525,
      "learning_rate": 0.002828,
      "loss": 1.035,
      "step": 613824
    },
    {
      "epoch": 46.07039399624765,
      "grad_norm": 0.6684466004371643,
      "learning_rate": 0.002828,
      "loss": 1.0391,
      "step": 613888
    },
    {
      "epoch": 46.07519699812383,
      "grad_norm": 0.8261184692382812,
      "learning_rate": 0.002828,
      "loss": 1.0406,
      "step": 613952
    },
    {
      "epoch": 46.08,
      "grad_norm": 0.6962255239486694,
      "learning_rate": 0.002828,
      "loss": 1.0452,
      "step": 614016
    },
    {
      "epoch": 46.084803001876175,
      "grad_norm": 0.594768762588501,
      "learning_rate": 0.002828,
      "loss": 1.0394,
      "step": 614080
    },
    {
      "epoch": 46.089606003752344,
      "grad_norm": 0.6519895792007446,
      "learning_rate": 0.002828,
      "loss": 1.0406,
      "step": 614144
    },
    {
      "epoch": 46.09440900562852,
      "grad_norm": 0.605659008026123,
      "learning_rate": 0.002828,
      "loss": 1.0396,
      "step": 614208
    },
    {
      "epoch": 46.09921200750469,
      "grad_norm": 0.798794150352478,
      "learning_rate": 0.002828,
      "loss": 1.0375,
      "step": 614272
    },
    {
      "epoch": 46.104015009380866,
      "grad_norm": 0.6516727209091187,
      "learning_rate": 0.002828,
      "loss": 1.0366,
      "step": 614336
    },
    {
      "epoch": 46.108818011257036,
      "grad_norm": 0.5554437637329102,
      "learning_rate": 0.002828,
      "loss": 1.0429,
      "step": 614400
    },
    {
      "epoch": 46.113621013133205,
      "grad_norm": 0.7370342016220093,
      "learning_rate": 0.002828,
      "loss": 1.044,
      "step": 614464
    },
    {
      "epoch": 46.11842401500938,
      "grad_norm": 0.5281388163566589,
      "learning_rate": 0.002828,
      "loss": 1.038,
      "step": 614528
    },
    {
      "epoch": 46.12322701688555,
      "grad_norm": 0.6068486571311951,
      "learning_rate": 0.002828,
      "loss": 1.043,
      "step": 614592
    },
    {
      "epoch": 46.12803001876173,
      "grad_norm": 0.5642166137695312,
      "learning_rate": 0.002828,
      "loss": 1.0401,
      "step": 614656
    },
    {
      "epoch": 46.1328330206379,
      "grad_norm": 0.7300256490707397,
      "learning_rate": 0.002828,
      "loss": 1.0428,
      "step": 614720
    },
    {
      "epoch": 46.13763602251407,
      "grad_norm": 0.5526905655860901,
      "learning_rate": 0.002828,
      "loss": 1.0444,
      "step": 614784
    },
    {
      "epoch": 46.14243902439024,
      "grad_norm": 0.5648698806762695,
      "learning_rate": 0.002828,
      "loss": 1.0403,
      "step": 614848
    },
    {
      "epoch": 46.14724202626642,
      "grad_norm": 0.5436872839927673,
      "learning_rate": 0.002828,
      "loss": 1.0417,
      "step": 614912
    },
    {
      "epoch": 46.15204502814259,
      "grad_norm": 0.7099076509475708,
      "learning_rate": 0.002828,
      "loss": 1.0422,
      "step": 614976
    },
    {
      "epoch": 46.156848030018764,
      "grad_norm": 0.668596088886261,
      "learning_rate": 0.002828,
      "loss": 1.0413,
      "step": 615040
    },
    {
      "epoch": 46.161651031894934,
      "grad_norm": 0.6142210960388184,
      "learning_rate": 0.002828,
      "loss": 1.0408,
      "step": 615104
    },
    {
      "epoch": 46.16645403377111,
      "grad_norm": 0.5688113570213318,
      "learning_rate": 0.002828,
      "loss": 1.04,
      "step": 615168
    },
    {
      "epoch": 46.17125703564728,
      "grad_norm": 0.5963445901870728,
      "learning_rate": 0.002828,
      "loss": 1.0417,
      "step": 615232
    },
    {
      "epoch": 46.17606003752345,
      "grad_norm": 0.6413314938545227,
      "learning_rate": 0.002828,
      "loss": 1.0467,
      "step": 615296
    },
    {
      "epoch": 46.180863039399625,
      "grad_norm": 0.729892909526825,
      "learning_rate": 0.002828,
      "loss": 1.0436,
      "step": 615360
    },
    {
      "epoch": 46.185666041275795,
      "grad_norm": 0.5900475382804871,
      "learning_rate": 0.002828,
      "loss": 1.0513,
      "step": 615424
    },
    {
      "epoch": 46.19046904315197,
      "grad_norm": 0.5937836766242981,
      "learning_rate": 0.002828,
      "loss": 1.0463,
      "step": 615488
    },
    {
      "epoch": 46.19527204502814,
      "grad_norm": 0.5667184591293335,
      "learning_rate": 0.002828,
      "loss": 1.044,
      "step": 615552
    },
    {
      "epoch": 46.20007504690432,
      "grad_norm": 0.7023417949676514,
      "learning_rate": 0.002828,
      "loss": 1.0459,
      "step": 615616
    },
    {
      "epoch": 46.204878048780486,
      "grad_norm": 0.6091719269752502,
      "learning_rate": 0.002828,
      "loss": 1.0431,
      "step": 615680
    },
    {
      "epoch": 46.20968105065666,
      "grad_norm": 0.6026754379272461,
      "learning_rate": 0.002828,
      "loss": 1.0404,
      "step": 615744
    },
    {
      "epoch": 46.21448405253283,
      "grad_norm": 0.7550497651100159,
      "learning_rate": 0.002828,
      "loss": 1.0404,
      "step": 615808
    },
    {
      "epoch": 46.21928705440901,
      "grad_norm": 0.6541439294815063,
      "learning_rate": 0.002828,
      "loss": 1.0424,
      "step": 615872
    },
    {
      "epoch": 46.22409005628518,
      "grad_norm": 0.678283154964447,
      "learning_rate": 0.002828,
      "loss": 1.0498,
      "step": 615936
    },
    {
      "epoch": 46.228893058161354,
      "grad_norm": 0.8214104771614075,
      "learning_rate": 0.002828,
      "loss": 1.0465,
      "step": 616000
    },
    {
      "epoch": 46.23369606003752,
      "grad_norm": 0.629660427570343,
      "learning_rate": 0.002828,
      "loss": 1.0469,
      "step": 616064
    },
    {
      "epoch": 46.23849906191369,
      "grad_norm": 0.5089216232299805,
      "learning_rate": 0.002828,
      "loss": 1.0481,
      "step": 616128
    },
    {
      "epoch": 46.24330206378987,
      "grad_norm": 0.6166875958442688,
      "learning_rate": 0.002828,
      "loss": 1.044,
      "step": 616192
    },
    {
      "epoch": 46.24810506566604,
      "grad_norm": 0.6610133051872253,
      "learning_rate": 0.002828,
      "loss": 1.0462,
      "step": 616256
    },
    {
      "epoch": 46.252908067542215,
      "grad_norm": 0.6949422359466553,
      "learning_rate": 0.002828,
      "loss": 1.0411,
      "step": 616320
    },
    {
      "epoch": 46.257711069418384,
      "grad_norm": 0.6649113297462463,
      "learning_rate": 0.002828,
      "loss": 1.0463,
      "step": 616384
    },
    {
      "epoch": 46.26251407129456,
      "grad_norm": 0.7349320650100708,
      "learning_rate": 0.002828,
      "loss": 1.0439,
      "step": 616448
    },
    {
      "epoch": 46.26731707317073,
      "grad_norm": 0.6438102722167969,
      "learning_rate": 0.002828,
      "loss": 1.0438,
      "step": 616512
    },
    {
      "epoch": 46.27212007504691,
      "grad_norm": 0.688700258731842,
      "learning_rate": 0.002828,
      "loss": 1.0475,
      "step": 616576
    },
    {
      "epoch": 46.276923076923076,
      "grad_norm": 0.49075180292129517,
      "learning_rate": 0.002828,
      "loss": 1.0411,
      "step": 616640
    },
    {
      "epoch": 46.28172607879925,
      "grad_norm": 0.4977070987224579,
      "learning_rate": 0.002828,
      "loss": 1.0468,
      "step": 616704
    },
    {
      "epoch": 46.28652908067542,
      "grad_norm": 0.6536993980407715,
      "learning_rate": 0.002828,
      "loss": 1.0524,
      "step": 616768
    },
    {
      "epoch": 46.2913320825516,
      "grad_norm": 0.5656940340995789,
      "learning_rate": 0.002828,
      "loss": 1.0464,
      "step": 616832
    },
    {
      "epoch": 46.29613508442777,
      "grad_norm": 0.7046701908111572,
      "learning_rate": 0.002828,
      "loss": 1.0495,
      "step": 616896
    },
    {
      "epoch": 46.30093808630394,
      "grad_norm": 0.7004103660583496,
      "learning_rate": 0.002828,
      "loss": 1.0455,
      "step": 616960
    },
    {
      "epoch": 46.30574108818011,
      "grad_norm": 0.6272410154342651,
      "learning_rate": 0.002828,
      "loss": 1.0444,
      "step": 617024
    },
    {
      "epoch": 46.31054409005628,
      "grad_norm": 0.6549649834632874,
      "learning_rate": 0.002828,
      "loss": 1.0468,
      "step": 617088
    },
    {
      "epoch": 46.31534709193246,
      "grad_norm": 0.6230428218841553,
      "learning_rate": 0.002828,
      "loss": 1.0448,
      "step": 617152
    },
    {
      "epoch": 46.32015009380863,
      "grad_norm": 0.576143741607666,
      "learning_rate": 0.002828,
      "loss": 1.0445,
      "step": 617216
    },
    {
      "epoch": 46.324953095684805,
      "grad_norm": 0.6095119118690491,
      "learning_rate": 0.002828,
      "loss": 1.0525,
      "step": 617280
    },
    {
      "epoch": 46.329756097560974,
      "grad_norm": 0.7142927050590515,
      "learning_rate": 0.002828,
      "loss": 1.0497,
      "step": 617344
    },
    {
      "epoch": 46.33455909943715,
      "grad_norm": 0.6191550493240356,
      "learning_rate": 0.002828,
      "loss": 1.0512,
      "step": 617408
    },
    {
      "epoch": 46.33936210131332,
      "grad_norm": 0.681013286113739,
      "learning_rate": 0.002828,
      "loss": 1.0467,
      "step": 617472
    },
    {
      "epoch": 46.344165103189496,
      "grad_norm": 0.5898663401603699,
      "learning_rate": 0.002828,
      "loss": 1.0524,
      "step": 617536
    },
    {
      "epoch": 46.348968105065666,
      "grad_norm": 0.7214446067810059,
      "learning_rate": 0.002828,
      "loss": 1.0506,
      "step": 617600
    },
    {
      "epoch": 46.35377110694184,
      "grad_norm": 0.6058822274208069,
      "learning_rate": 0.002828,
      "loss": 1.0548,
      "step": 617664
    },
    {
      "epoch": 46.35857410881801,
      "grad_norm": 0.6141964197158813,
      "learning_rate": 0.002828,
      "loss": 1.0531,
      "step": 617728
    },
    {
      "epoch": 46.36337711069418,
      "grad_norm": 0.5327640175819397,
      "learning_rate": 0.002828,
      "loss": 1.0462,
      "step": 617792
    },
    {
      "epoch": 46.36818011257036,
      "grad_norm": 0.7058151364326477,
      "learning_rate": 0.002828,
      "loss": 1.0535,
      "step": 617856
    },
    {
      "epoch": 46.37298311444653,
      "grad_norm": 0.46748828887939453,
      "learning_rate": 0.002828,
      "loss": 1.0495,
      "step": 617920
    },
    {
      "epoch": 46.3777861163227,
      "grad_norm": 0.5134317874908447,
      "learning_rate": 0.002828,
      "loss": 1.0479,
      "step": 617984
    },
    {
      "epoch": 46.38258911819887,
      "grad_norm": 0.799745500087738,
      "learning_rate": 0.002828,
      "loss": 1.0535,
      "step": 618048
    },
    {
      "epoch": 46.38739212007505,
      "grad_norm": 0.5732191801071167,
      "learning_rate": 0.002828,
      "loss": 1.0489,
      "step": 618112
    },
    {
      "epoch": 46.39219512195122,
      "grad_norm": 0.5947842597961426,
      "learning_rate": 0.002828,
      "loss": 1.0524,
      "step": 618176
    },
    {
      "epoch": 46.396998123827395,
      "grad_norm": 0.5764126181602478,
      "learning_rate": 0.002828,
      "loss": 1.0514,
      "step": 618240
    },
    {
      "epoch": 46.401801125703564,
      "grad_norm": 0.4855901300907135,
      "learning_rate": 0.002828,
      "loss": 1.0534,
      "step": 618304
    },
    {
      "epoch": 46.40660412757974,
      "grad_norm": 0.6275175213813782,
      "learning_rate": 0.002828,
      "loss": 1.0553,
      "step": 618368
    },
    {
      "epoch": 46.41140712945591,
      "grad_norm": 0.6019368171691895,
      "learning_rate": 0.002828,
      "loss": 1.0534,
      "step": 618432
    },
    {
      "epoch": 46.416210131332086,
      "grad_norm": 0.531459391117096,
      "learning_rate": 0.002828,
      "loss": 1.0519,
      "step": 618496
    },
    {
      "epoch": 46.421013133208255,
      "grad_norm": 0.5653847455978394,
      "learning_rate": 0.002828,
      "loss": 1.055,
      "step": 618560
    },
    {
      "epoch": 46.425816135084425,
      "grad_norm": 0.6705509424209595,
      "learning_rate": 0.002828,
      "loss": 1.0548,
      "step": 618624
    },
    {
      "epoch": 46.4306191369606,
      "grad_norm": 0.6440606117248535,
      "learning_rate": 0.002828,
      "loss": 1.0522,
      "step": 618688
    },
    {
      "epoch": 46.43542213883677,
      "grad_norm": 0.6682266592979431,
      "learning_rate": 0.002828,
      "loss": 1.0521,
      "step": 618752
    },
    {
      "epoch": 46.44022514071295,
      "grad_norm": 0.6298613548278809,
      "learning_rate": 0.002828,
      "loss": 1.0512,
      "step": 618816
    },
    {
      "epoch": 46.445028142589116,
      "grad_norm": 0.776021420955658,
      "learning_rate": 0.002828,
      "loss": 1.0534,
      "step": 618880
    },
    {
      "epoch": 46.44983114446529,
      "grad_norm": 0.5949306488037109,
      "learning_rate": 0.002828,
      "loss": 1.0521,
      "step": 618944
    },
    {
      "epoch": 46.45463414634146,
      "grad_norm": 0.6268242597579956,
      "learning_rate": 0.002828,
      "loss": 1.0556,
      "step": 619008
    },
    {
      "epoch": 46.45943714821764,
      "grad_norm": 0.5667731761932373,
      "learning_rate": 0.002828,
      "loss": 1.0518,
      "step": 619072
    },
    {
      "epoch": 46.46424015009381,
      "grad_norm": 0.5637649297714233,
      "learning_rate": 0.002828,
      "loss": 1.052,
      "step": 619136
    },
    {
      "epoch": 46.469043151969984,
      "grad_norm": 0.6626964211463928,
      "learning_rate": 0.002828,
      "loss": 1.0514,
      "step": 619200
    },
    {
      "epoch": 46.473846153846154,
      "grad_norm": 0.8213684558868408,
      "learning_rate": 0.002828,
      "loss": 1.0564,
      "step": 619264
    },
    {
      "epoch": 46.47864915572232,
      "grad_norm": 0.7015941739082336,
      "learning_rate": 0.002828,
      "loss": 1.0542,
      "step": 619328
    },
    {
      "epoch": 46.4834521575985,
      "grad_norm": 0.6902109980583191,
      "learning_rate": 0.002828,
      "loss": 1.0554,
      "step": 619392
    },
    {
      "epoch": 46.48825515947467,
      "grad_norm": 0.6375514268875122,
      "learning_rate": 0.002828,
      "loss": 1.0525,
      "step": 619456
    },
    {
      "epoch": 46.493058161350845,
      "grad_norm": 0.6744531989097595,
      "learning_rate": 0.002828,
      "loss": 1.0577,
      "step": 619520
    },
    {
      "epoch": 46.497861163227014,
      "grad_norm": 0.6031131148338318,
      "learning_rate": 0.002828,
      "loss": 1.0535,
      "step": 619584
    },
    {
      "epoch": 46.50266416510319,
      "grad_norm": 0.5269550681114197,
      "learning_rate": 0.002828,
      "loss": 1.0565,
      "step": 619648
    },
    {
      "epoch": 46.50746716697936,
      "grad_norm": 0.5548364520072937,
      "learning_rate": 0.002828,
      "loss": 1.0579,
      "step": 619712
    },
    {
      "epoch": 46.51227016885554,
      "grad_norm": 0.6689755320549011,
      "learning_rate": 0.002828,
      "loss": 1.0569,
      "step": 619776
    },
    {
      "epoch": 46.517073170731706,
      "grad_norm": 0.5496863722801208,
      "learning_rate": 0.002828,
      "loss": 1.0548,
      "step": 619840
    },
    {
      "epoch": 46.52187617260788,
      "grad_norm": 0.5853809714317322,
      "learning_rate": 0.002828,
      "loss": 1.051,
      "step": 619904
    },
    {
      "epoch": 46.52667917448405,
      "grad_norm": 0.6034921407699585,
      "learning_rate": 0.002828,
      "loss": 1.0507,
      "step": 619968
    },
    {
      "epoch": 46.53148217636023,
      "grad_norm": 0.6705649495124817,
      "learning_rate": 0.002828,
      "loss": 1.0521,
      "step": 620032
    },
    {
      "epoch": 46.5362851782364,
      "grad_norm": 0.5398948192596436,
      "learning_rate": 0.002828,
      "loss": 1.0468,
      "step": 620096
    },
    {
      "epoch": 46.54108818011257,
      "grad_norm": 0.5841556787490845,
      "learning_rate": 0.002828,
      "loss": 1.0577,
      "step": 620160
    },
    {
      "epoch": 46.54589118198874,
      "grad_norm": 0.6120521426200867,
      "learning_rate": 0.002828,
      "loss": 1.0484,
      "step": 620224
    },
    {
      "epoch": 46.55069418386491,
      "grad_norm": 0.6358661651611328,
      "learning_rate": 0.002828,
      "loss": 1.0567,
      "step": 620288
    },
    {
      "epoch": 46.55549718574109,
      "grad_norm": 0.7871175408363342,
      "learning_rate": 0.002828,
      "loss": 1.054,
      "step": 620352
    },
    {
      "epoch": 46.56030018761726,
      "grad_norm": 0.6011291742324829,
      "learning_rate": 0.002828,
      "loss": 1.0511,
      "step": 620416
    },
    {
      "epoch": 46.565103189493435,
      "grad_norm": 0.7039676308631897,
      "learning_rate": 0.002828,
      "loss": 1.0543,
      "step": 620480
    },
    {
      "epoch": 46.569906191369604,
      "grad_norm": 0.7038636207580566,
      "learning_rate": 0.002828,
      "loss": 1.0553,
      "step": 620544
    },
    {
      "epoch": 46.57470919324578,
      "grad_norm": 0.6768410205841064,
      "learning_rate": 0.002828,
      "loss": 1.0512,
      "step": 620608
    },
    {
      "epoch": 46.57951219512195,
      "grad_norm": 0.6707714796066284,
      "learning_rate": 0.002828,
      "loss": 1.0508,
      "step": 620672
    },
    {
      "epoch": 46.584315196998126,
      "grad_norm": 0.678458034992218,
      "learning_rate": 0.002828,
      "loss": 1.0575,
      "step": 620736
    },
    {
      "epoch": 46.589118198874296,
      "grad_norm": 0.6657276749610901,
      "learning_rate": 0.002828,
      "loss": 1.057,
      "step": 620800
    },
    {
      "epoch": 46.59392120075047,
      "grad_norm": 0.6960893273353577,
      "learning_rate": 0.002828,
      "loss": 1.0588,
      "step": 620864
    },
    {
      "epoch": 46.59872420262664,
      "grad_norm": 0.6260180473327637,
      "learning_rate": 0.002828,
      "loss": 1.0576,
      "step": 620928
    },
    {
      "epoch": 46.60352720450281,
      "grad_norm": 0.5803308486938477,
      "learning_rate": 0.002828,
      "loss": 1.0544,
      "step": 620992
    },
    {
      "epoch": 46.60833020637899,
      "grad_norm": 0.563851535320282,
      "learning_rate": 0.002828,
      "loss": 1.0559,
      "step": 621056
    },
    {
      "epoch": 46.61313320825516,
      "grad_norm": 0.6698217988014221,
      "learning_rate": 0.002828,
      "loss": 1.0547,
      "step": 621120
    },
    {
      "epoch": 46.61793621013133,
      "grad_norm": 0.670448899269104,
      "learning_rate": 0.002828,
      "loss": 1.0593,
      "step": 621184
    },
    {
      "epoch": 46.6227392120075,
      "grad_norm": 0.5835123658180237,
      "learning_rate": 0.002828,
      "loss": 1.0567,
      "step": 621248
    },
    {
      "epoch": 46.62754221388368,
      "grad_norm": 0.6441203951835632,
      "learning_rate": 0.002828,
      "loss": 1.0545,
      "step": 621312
    },
    {
      "epoch": 46.63234521575985,
      "grad_norm": 0.666936457157135,
      "learning_rate": 0.002828,
      "loss": 1.0577,
      "step": 621376
    },
    {
      "epoch": 46.637148217636025,
      "grad_norm": 0.6317396759986877,
      "learning_rate": 0.002828,
      "loss": 1.0607,
      "step": 621440
    },
    {
      "epoch": 46.641951219512194,
      "grad_norm": 0.5824635028839111,
      "learning_rate": 0.002828,
      "loss": 1.0611,
      "step": 621504
    },
    {
      "epoch": 46.64675422138837,
      "grad_norm": 0.6023385524749756,
      "learning_rate": 0.002828,
      "loss": 1.0541,
      "step": 621568
    },
    {
      "epoch": 46.65155722326454,
      "grad_norm": 0.6010758876800537,
      "learning_rate": 0.002828,
      "loss": 1.0565,
      "step": 621632
    },
    {
      "epoch": 46.656360225140716,
      "grad_norm": 0.6052330732345581,
      "learning_rate": 0.002828,
      "loss": 1.0539,
      "step": 621696
    },
    {
      "epoch": 46.661163227016885,
      "grad_norm": 0.6076634526252747,
      "learning_rate": 0.002828,
      "loss": 1.0563,
      "step": 621760
    },
    {
      "epoch": 46.665966228893055,
      "grad_norm": 0.6403579711914062,
      "learning_rate": 0.002828,
      "loss": 1.0607,
      "step": 621824
    },
    {
      "epoch": 46.67076923076923,
      "grad_norm": 0.6364386081695557,
      "learning_rate": 0.002828,
      "loss": 1.0626,
      "step": 621888
    },
    {
      "epoch": 46.6755722326454,
      "grad_norm": 0.6403948068618774,
      "learning_rate": 0.002828,
      "loss": 1.0538,
      "step": 621952
    },
    {
      "epoch": 46.68037523452158,
      "grad_norm": 0.48538199067115784,
      "learning_rate": 0.002828,
      "loss": 1.0576,
      "step": 622016
    },
    {
      "epoch": 46.685178236397746,
      "grad_norm": 0.6486366391181946,
      "learning_rate": 0.002828,
      "loss": 1.0551,
      "step": 622080
    },
    {
      "epoch": 46.68998123827392,
      "grad_norm": 0.650784969329834,
      "learning_rate": 0.002828,
      "loss": 1.0575,
      "step": 622144
    },
    {
      "epoch": 46.69478424015009,
      "grad_norm": 0.7481862902641296,
      "learning_rate": 0.002828,
      "loss": 1.0558,
      "step": 622208
    },
    {
      "epoch": 46.69958724202627,
      "grad_norm": 0.6436601281166077,
      "learning_rate": 0.002828,
      "loss": 1.0596,
      "step": 622272
    },
    {
      "epoch": 46.70439024390244,
      "grad_norm": 0.6628262996673584,
      "learning_rate": 0.002828,
      "loss": 1.0623,
      "step": 622336
    },
    {
      "epoch": 46.709193245778614,
      "grad_norm": 0.680276095867157,
      "learning_rate": 0.002828,
      "loss": 1.0563,
      "step": 622400
    },
    {
      "epoch": 46.713996247654784,
      "grad_norm": 0.6068404912948608,
      "learning_rate": 0.002828,
      "loss": 1.0491,
      "step": 622464
    },
    {
      "epoch": 46.71879924953096,
      "grad_norm": 0.5882270932197571,
      "learning_rate": 0.002828,
      "loss": 1.06,
      "step": 622528
    },
    {
      "epoch": 46.72360225140713,
      "grad_norm": 0.6210886240005493,
      "learning_rate": 0.002828,
      "loss": 1.0512,
      "step": 622592
    },
    {
      "epoch": 46.7284052532833,
      "grad_norm": 0.5923880934715271,
      "learning_rate": 0.002828,
      "loss": 1.0589,
      "step": 622656
    },
    {
      "epoch": 46.733208255159475,
      "grad_norm": 0.6892109513282776,
      "learning_rate": 0.002828,
      "loss": 1.0578,
      "step": 622720
    },
    {
      "epoch": 46.738011257035645,
      "grad_norm": 0.6397147178649902,
      "learning_rate": 0.002828,
      "loss": 1.0595,
      "step": 622784
    },
    {
      "epoch": 46.74281425891182,
      "grad_norm": 0.6659414172172546,
      "learning_rate": 0.002828,
      "loss": 1.0609,
      "step": 622848
    },
    {
      "epoch": 46.74761726078799,
      "grad_norm": 0.7534034252166748,
      "learning_rate": 0.002828,
      "loss": 1.0582,
      "step": 622912
    },
    {
      "epoch": 46.75242026266417,
      "grad_norm": 0.6626340746879578,
      "learning_rate": 0.002828,
      "loss": 1.0608,
      "step": 622976
    },
    {
      "epoch": 46.757223264540336,
      "grad_norm": 0.6399165391921997,
      "learning_rate": 0.002828,
      "loss": 1.0617,
      "step": 623040
    },
    {
      "epoch": 46.76202626641651,
      "grad_norm": 0.5546966195106506,
      "learning_rate": 0.002828,
      "loss": 1.0594,
      "step": 623104
    },
    {
      "epoch": 46.76682926829268,
      "grad_norm": 0.6276941299438477,
      "learning_rate": 0.002828,
      "loss": 1.0606,
      "step": 623168
    },
    {
      "epoch": 46.77163227016886,
      "grad_norm": 0.6073575615882874,
      "learning_rate": 0.002828,
      "loss": 1.0576,
      "step": 623232
    },
    {
      "epoch": 46.77643527204503,
      "grad_norm": 0.688251793384552,
      "learning_rate": 0.002828,
      "loss": 1.0573,
      "step": 623296
    },
    {
      "epoch": 46.781238273921204,
      "grad_norm": 0.6676405668258667,
      "learning_rate": 0.002828,
      "loss": 1.0561,
      "step": 623360
    },
    {
      "epoch": 46.78604127579737,
      "grad_norm": 0.578522264957428,
      "learning_rate": 0.002828,
      "loss": 1.0615,
      "step": 623424
    },
    {
      "epoch": 46.79084427767354,
      "grad_norm": 0.6574742197990417,
      "learning_rate": 0.002828,
      "loss": 1.0592,
      "step": 623488
    },
    {
      "epoch": 46.79564727954972,
      "grad_norm": 0.7181845307350159,
      "learning_rate": 0.002828,
      "loss": 1.066,
      "step": 623552
    },
    {
      "epoch": 46.80045028142589,
      "grad_norm": 0.5383839011192322,
      "learning_rate": 0.002828,
      "loss": 1.056,
      "step": 623616
    },
    {
      "epoch": 46.805253283302065,
      "grad_norm": 0.5833193063735962,
      "learning_rate": 0.002828,
      "loss": 1.0584,
      "step": 623680
    },
    {
      "epoch": 46.810056285178234,
      "grad_norm": 0.6272100210189819,
      "learning_rate": 0.002828,
      "loss": 1.0568,
      "step": 623744
    },
    {
      "epoch": 46.81485928705441,
      "grad_norm": 0.5736045241355896,
      "learning_rate": 0.002828,
      "loss": 1.0598,
      "step": 623808
    },
    {
      "epoch": 46.81966228893058,
      "grad_norm": 0.8262968063354492,
      "learning_rate": 0.002828,
      "loss": 1.0644,
      "step": 623872
    },
    {
      "epoch": 46.82446529080676,
      "grad_norm": 0.6368528604507446,
      "learning_rate": 0.002828,
      "loss": 1.0644,
      "step": 623936
    },
    {
      "epoch": 46.829268292682926,
      "grad_norm": 0.6689131259918213,
      "learning_rate": 0.002828,
      "loss": 1.06,
      "step": 624000
    },
    {
      "epoch": 46.8340712945591,
      "grad_norm": 0.5683980584144592,
      "learning_rate": 0.002828,
      "loss": 1.0604,
      "step": 624064
    },
    {
      "epoch": 46.83887429643527,
      "grad_norm": 0.5765275955200195,
      "learning_rate": 0.002828,
      "loss": 1.0654,
      "step": 624128
    },
    {
      "epoch": 46.84367729831145,
      "grad_norm": 0.5761871933937073,
      "learning_rate": 0.002828,
      "loss": 1.0644,
      "step": 624192
    },
    {
      "epoch": 46.84848030018762,
      "grad_norm": 0.5934112071990967,
      "learning_rate": 0.002828,
      "loss": 1.0633,
      "step": 624256
    },
    {
      "epoch": 46.85328330206379,
      "grad_norm": 0.521324634552002,
      "learning_rate": 0.002828,
      "loss": 1.0687,
      "step": 624320
    },
    {
      "epoch": 46.85808630393996,
      "grad_norm": 0.6639213562011719,
      "learning_rate": 0.002828,
      "loss": 1.0542,
      "step": 624384
    },
    {
      "epoch": 46.86288930581613,
      "grad_norm": 0.6850979328155518,
      "learning_rate": 0.002828,
      "loss": 1.0618,
      "step": 624448
    },
    {
      "epoch": 46.86769230769231,
      "grad_norm": 0.6221501231193542,
      "learning_rate": 0.002828,
      "loss": 1.0626,
      "step": 624512
    },
    {
      "epoch": 46.87249530956848,
      "grad_norm": 0.6977725028991699,
      "learning_rate": 0.002828,
      "loss": 1.0613,
      "step": 624576
    },
    {
      "epoch": 46.877298311444655,
      "grad_norm": 0.6750850081443787,
      "learning_rate": 0.002828,
      "loss": 1.0663,
      "step": 624640
    },
    {
      "epoch": 46.882101313320824,
      "grad_norm": 0.6050744652748108,
      "learning_rate": 0.002828,
      "loss": 1.0634,
      "step": 624704
    },
    {
      "epoch": 46.886904315197,
      "grad_norm": 0.7393004298210144,
      "learning_rate": 0.002828,
      "loss": 1.0625,
      "step": 624768
    },
    {
      "epoch": 46.89170731707317,
      "grad_norm": 0.669156551361084,
      "learning_rate": 0.002828,
      "loss": 1.0646,
      "step": 624832
    },
    {
      "epoch": 46.896510318949346,
      "grad_norm": 0.5985300540924072,
      "learning_rate": 0.002828,
      "loss": 1.0641,
      "step": 624896
    },
    {
      "epoch": 46.901313320825516,
      "grad_norm": 0.5303430557250977,
      "learning_rate": 0.002828,
      "loss": 1.0651,
      "step": 624960
    },
    {
      "epoch": 46.90611632270169,
      "grad_norm": 0.5656881332397461,
      "learning_rate": 0.002828,
      "loss": 1.0672,
      "step": 625024
    },
    {
      "epoch": 46.91091932457786,
      "grad_norm": 0.6489809155464172,
      "learning_rate": 0.002828,
      "loss": 1.0628,
      "step": 625088
    },
    {
      "epoch": 46.91572232645403,
      "grad_norm": 0.5730379223823547,
      "learning_rate": 0.002828,
      "loss": 1.0618,
      "step": 625152
    },
    {
      "epoch": 46.92052532833021,
      "grad_norm": 0.6608423590660095,
      "learning_rate": 0.002828,
      "loss": 1.0624,
      "step": 625216
    },
    {
      "epoch": 46.925328330206376,
      "grad_norm": 0.4727746248245239,
      "learning_rate": 0.002828,
      "loss": 1.0652,
      "step": 625280
    },
    {
      "epoch": 46.93013133208255,
      "grad_norm": 0.6244372129440308,
      "learning_rate": 0.002828,
      "loss": 1.0659,
      "step": 625344
    },
    {
      "epoch": 46.93493433395872,
      "grad_norm": 0.7188080549240112,
      "learning_rate": 0.002828,
      "loss": 1.0652,
      "step": 625408
    },
    {
      "epoch": 46.9397373358349,
      "grad_norm": 0.62159663438797,
      "learning_rate": 0.002828,
      "loss": 1.0667,
      "step": 625472
    },
    {
      "epoch": 46.94454033771107,
      "grad_norm": 0.6020349264144897,
      "learning_rate": 0.002828,
      "loss": 1.0609,
      "step": 625536
    },
    {
      "epoch": 46.949343339587244,
      "grad_norm": 0.7522795796394348,
      "learning_rate": 0.002828,
      "loss": 1.0665,
      "step": 625600
    },
    {
      "epoch": 46.954146341463414,
      "grad_norm": 0.6020083427429199,
      "learning_rate": 0.002828,
      "loss": 1.0613,
      "step": 625664
    },
    {
      "epoch": 46.95894934333959,
      "grad_norm": 0.6429307460784912,
      "learning_rate": 0.002828,
      "loss": 1.0667,
      "step": 625728
    },
    {
      "epoch": 46.96375234521576,
      "grad_norm": 0.5400229692459106,
      "learning_rate": 0.002828,
      "loss": 1.0607,
      "step": 625792
    },
    {
      "epoch": 46.968555347091936,
      "grad_norm": 0.5039812326431274,
      "learning_rate": 0.002828,
      "loss": 1.0663,
      "step": 625856
    },
    {
      "epoch": 46.973358348968105,
      "grad_norm": 0.5142475366592407,
      "learning_rate": 0.002828,
      "loss": 1.0598,
      "step": 625920
    },
    {
      "epoch": 46.978161350844275,
      "grad_norm": 0.6738988161087036,
      "learning_rate": 0.002828,
      "loss": 1.0629,
      "step": 625984
    },
    {
      "epoch": 46.98296435272045,
      "grad_norm": 0.5502144694328308,
      "learning_rate": 0.002828,
      "loss": 1.0555,
      "step": 626048
    },
    {
      "epoch": 46.98776735459662,
      "grad_norm": 0.5779480934143066,
      "learning_rate": 0.002828,
      "loss": 1.0604,
      "step": 626112
    },
    {
      "epoch": 46.9925703564728,
      "grad_norm": 0.809315025806427,
      "learning_rate": 0.002828,
      "loss": 1.0589,
      "step": 626176
    },
    {
      "epoch": 46.997373358348966,
      "grad_norm": 0.7586283087730408,
      "learning_rate": 0.002828,
      "loss": 1.0606,
      "step": 626240
    },
    {
      "epoch": 47.00217636022514,
      "grad_norm": 0.5281000733375549,
      "learning_rate": 0.002828,
      "loss": 1.0499,
      "step": 626304
    },
    {
      "epoch": 47.00697936210131,
      "grad_norm": 0.6618357300758362,
      "learning_rate": 0.002828,
      "loss": 1.026,
      "step": 626368
    },
    {
      "epoch": 47.01178236397749,
      "grad_norm": 0.6292099356651306,
      "learning_rate": 0.002828,
      "loss": 1.0272,
      "step": 626432
    },
    {
      "epoch": 47.01658536585366,
      "grad_norm": 0.6515204906463623,
      "learning_rate": 0.002828,
      "loss": 1.0305,
      "step": 626496
    },
    {
      "epoch": 47.021388367729834,
      "grad_norm": 0.7430369257926941,
      "learning_rate": 0.002828,
      "loss": 1.0339,
      "step": 626560
    },
    {
      "epoch": 47.026191369606,
      "grad_norm": 0.6072025895118713,
      "learning_rate": 0.002828,
      "loss": 1.0312,
      "step": 626624
    },
    {
      "epoch": 47.03099437148218,
      "grad_norm": 0.5867000222206116,
      "learning_rate": 0.002828,
      "loss": 1.0357,
      "step": 626688
    },
    {
      "epoch": 47.03579737335835,
      "grad_norm": 0.6820483207702637,
      "learning_rate": 0.002828,
      "loss": 1.03,
      "step": 626752
    },
    {
      "epoch": 47.04060037523452,
      "grad_norm": 0.528099536895752,
      "learning_rate": 0.002828,
      "loss": 1.0369,
      "step": 626816
    },
    {
      "epoch": 47.045403377110695,
      "grad_norm": 0.7703579664230347,
      "learning_rate": 0.002828,
      "loss": 1.0248,
      "step": 626880
    },
    {
      "epoch": 47.050206378986864,
      "grad_norm": 0.6891136765480042,
      "learning_rate": 0.002828,
      "loss": 1.0322,
      "step": 626944
    },
    {
      "epoch": 47.05500938086304,
      "grad_norm": 0.6598283648490906,
      "learning_rate": 0.002828,
      "loss": 1.0344,
      "step": 627008
    },
    {
      "epoch": 47.05981238273921,
      "grad_norm": 0.7029429078102112,
      "learning_rate": 0.002828,
      "loss": 1.0324,
      "step": 627072
    },
    {
      "epoch": 47.06461538461539,
      "grad_norm": 0.6200289130210876,
      "learning_rate": 0.002828,
      "loss": 1.0355,
      "step": 627136
    },
    {
      "epoch": 47.069418386491556,
      "grad_norm": 0.7782018780708313,
      "learning_rate": 0.002828,
      "loss": 1.0301,
      "step": 627200
    },
    {
      "epoch": 47.07422138836773,
      "grad_norm": 0.5827069282531738,
      "learning_rate": 0.002828,
      "loss": 1.0322,
      "step": 627264
    },
    {
      "epoch": 47.0790243902439,
      "grad_norm": 0.6420133113861084,
      "learning_rate": 0.002828,
      "loss": 1.0348,
      "step": 627328
    },
    {
      "epoch": 47.08382739212008,
      "grad_norm": 0.6323050856590271,
      "learning_rate": 0.002828,
      "loss": 1.0299,
      "step": 627392
    },
    {
      "epoch": 47.08863039399625,
      "grad_norm": 0.6225181221961975,
      "learning_rate": 0.002828,
      "loss": 1.0318,
      "step": 627456
    },
    {
      "epoch": 47.09343339587242,
      "grad_norm": 0.5773589611053467,
      "learning_rate": 0.002828,
      "loss": 1.0342,
      "step": 627520
    },
    {
      "epoch": 47.09823639774859,
      "grad_norm": 0.6008696556091309,
      "learning_rate": 0.002828,
      "loss": 1.0334,
      "step": 627584
    },
    {
      "epoch": 47.10303939962476,
      "grad_norm": 0.5864893198013306,
      "learning_rate": 0.002828,
      "loss": 1.0331,
      "step": 627648
    },
    {
      "epoch": 47.10784240150094,
      "grad_norm": 0.6275175213813782,
      "learning_rate": 0.002828,
      "loss": 1.0303,
      "step": 627712
    },
    {
      "epoch": 47.11264540337711,
      "grad_norm": 0.5716657638549805,
      "learning_rate": 0.002828,
      "loss": 1.033,
      "step": 627776
    },
    {
      "epoch": 47.117448405253285,
      "grad_norm": 0.5904713273048401,
      "learning_rate": 0.002828,
      "loss": 1.0367,
      "step": 627840
    },
    {
      "epoch": 47.122251407129454,
      "grad_norm": 0.6342942714691162,
      "learning_rate": 0.002828,
      "loss": 1.0411,
      "step": 627904
    },
    {
      "epoch": 47.12705440900563,
      "grad_norm": 0.5893776416778564,
      "learning_rate": 0.002828,
      "loss": 1.0379,
      "step": 627968
    },
    {
      "epoch": 47.1318574108818,
      "grad_norm": 0.6401630640029907,
      "learning_rate": 0.002828,
      "loss": 1.0359,
      "step": 628032
    },
    {
      "epoch": 47.136660412757976,
      "grad_norm": 0.7856540083885193,
      "learning_rate": 0.002828,
      "loss": 1.0388,
      "step": 628096
    },
    {
      "epoch": 47.141463414634146,
      "grad_norm": 0.5844777822494507,
      "learning_rate": 0.002828,
      "loss": 1.0367,
      "step": 628160
    },
    {
      "epoch": 47.14626641651032,
      "grad_norm": 0.7515383958816528,
      "learning_rate": 0.002828,
      "loss": 1.0368,
      "step": 628224
    },
    {
      "epoch": 47.15106941838649,
      "grad_norm": 0.59804368019104,
      "learning_rate": 0.002828,
      "loss": 1.0379,
      "step": 628288
    },
    {
      "epoch": 47.15587242026266,
      "grad_norm": 0.5939375162124634,
      "learning_rate": 0.002828,
      "loss": 1.0351,
      "step": 628352
    },
    {
      "epoch": 47.16067542213884,
      "grad_norm": 0.5455921292304993,
      "learning_rate": 0.002828,
      "loss": 1.0341,
      "step": 628416
    },
    {
      "epoch": 47.16547842401501,
      "grad_norm": 0.6681742668151855,
      "learning_rate": 0.002828,
      "loss": 1.0352,
      "step": 628480
    },
    {
      "epoch": 47.17028142589118,
      "grad_norm": 0.6716008186340332,
      "learning_rate": 0.002828,
      "loss": 1.0383,
      "step": 628544
    },
    {
      "epoch": 47.17508442776735,
      "grad_norm": 0.6943315863609314,
      "learning_rate": 0.002828,
      "loss": 1.0372,
      "step": 628608
    },
    {
      "epoch": 47.17988742964353,
      "grad_norm": 0.6753778457641602,
      "learning_rate": 0.002828,
      "loss": 1.0434,
      "step": 628672
    },
    {
      "epoch": 47.1846904315197,
      "grad_norm": 0.7257936000823975,
      "learning_rate": 0.002828,
      "loss": 1.0382,
      "step": 628736
    },
    {
      "epoch": 47.189493433395874,
      "grad_norm": 0.6734530329704285,
      "learning_rate": 0.002828,
      "loss": 1.0364,
      "step": 628800
    },
    {
      "epoch": 47.194296435272044,
      "grad_norm": 0.7898691892623901,
      "learning_rate": 0.002828,
      "loss": 1.0407,
      "step": 628864
    },
    {
      "epoch": 47.19909943714822,
      "grad_norm": 0.6546264886856079,
      "learning_rate": 0.002828,
      "loss": 1.0377,
      "step": 628928
    },
    {
      "epoch": 47.20390243902439,
      "grad_norm": 0.5156117677688599,
      "learning_rate": 0.002828,
      "loss": 1.041,
      "step": 628992
    },
    {
      "epoch": 47.208705440900566,
      "grad_norm": 0.6237413883209229,
      "learning_rate": 0.002828,
      "loss": 1.0378,
      "step": 629056
    },
    {
      "epoch": 47.213508442776735,
      "grad_norm": 0.7476790547370911,
      "learning_rate": 0.002828,
      "loss": 1.0435,
      "step": 629120
    },
    {
      "epoch": 47.218311444652905,
      "grad_norm": 0.643621027469635,
      "learning_rate": 0.002828,
      "loss": 1.0387,
      "step": 629184
    },
    {
      "epoch": 47.22311444652908,
      "grad_norm": 0.7706505060195923,
      "learning_rate": 0.002828,
      "loss": 1.0363,
      "step": 629248
    },
    {
      "epoch": 47.22791744840525,
      "grad_norm": 0.582042932510376,
      "learning_rate": 0.002828,
      "loss": 1.0398,
      "step": 629312
    },
    {
      "epoch": 47.23272045028143,
      "grad_norm": 0.6429329514503479,
      "learning_rate": 0.002828,
      "loss": 1.0377,
      "step": 629376
    },
    {
      "epoch": 47.237523452157596,
      "grad_norm": 0.5040108561515808,
      "learning_rate": 0.002828,
      "loss": 1.0448,
      "step": 629440
    },
    {
      "epoch": 47.24232645403377,
      "grad_norm": 0.7274630069732666,
      "learning_rate": 0.002828,
      "loss": 1.0394,
      "step": 629504
    },
    {
      "epoch": 47.24712945590994,
      "grad_norm": 0.6705653071403503,
      "learning_rate": 0.002828,
      "loss": 1.0376,
      "step": 629568
    },
    {
      "epoch": 47.25193245778612,
      "grad_norm": 0.6059752106666565,
      "learning_rate": 0.002828,
      "loss": 1.0384,
      "step": 629632
    },
    {
      "epoch": 47.25673545966229,
      "grad_norm": 0.5783113241195679,
      "learning_rate": 0.002828,
      "loss": 1.0454,
      "step": 629696
    },
    {
      "epoch": 47.261538461538464,
      "grad_norm": 0.505687415599823,
      "learning_rate": 0.002828,
      "loss": 1.0363,
      "step": 629760
    },
    {
      "epoch": 47.26634146341463,
      "grad_norm": 0.5358524918556213,
      "learning_rate": 0.002828,
      "loss": 1.0411,
      "step": 629824
    },
    {
      "epoch": 47.27114446529081,
      "grad_norm": 0.8537362813949585,
      "learning_rate": 0.002828,
      "loss": 1.0409,
      "step": 629888
    },
    {
      "epoch": 47.27594746716698,
      "grad_norm": 0.5460085868835449,
      "learning_rate": 0.002828,
      "loss": 1.0437,
      "step": 629952
    },
    {
      "epoch": 47.28075046904315,
      "grad_norm": 0.788359522819519,
      "learning_rate": 0.002828,
      "loss": 1.0475,
      "step": 630016
    },
    {
      "epoch": 47.285553470919325,
      "grad_norm": 0.5909889340400696,
      "learning_rate": 0.002828,
      "loss": 1.0402,
      "step": 630080
    },
    {
      "epoch": 47.290356472795494,
      "grad_norm": 0.6027221083641052,
      "learning_rate": 0.002828,
      "loss": 1.0388,
      "step": 630144
    },
    {
      "epoch": 47.29515947467167,
      "grad_norm": 0.6677124500274658,
      "learning_rate": 0.002828,
      "loss": 1.0417,
      "step": 630208
    },
    {
      "epoch": 47.29996247654784,
      "grad_norm": 0.6280375719070435,
      "learning_rate": 0.002828,
      "loss": 1.0443,
      "step": 630272
    },
    {
      "epoch": 47.30476547842402,
      "grad_norm": 0.5601394176483154,
      "learning_rate": 0.002828,
      "loss": 1.0407,
      "step": 630336
    },
    {
      "epoch": 47.309568480300186,
      "grad_norm": 0.7195338606834412,
      "learning_rate": 0.002828,
      "loss": 1.0456,
      "step": 630400
    },
    {
      "epoch": 47.31437148217636,
      "grad_norm": 0.6633152961730957,
      "learning_rate": 0.002828,
      "loss": 1.047,
      "step": 630464
    },
    {
      "epoch": 47.31917448405253,
      "grad_norm": 0.595317542552948,
      "learning_rate": 0.002828,
      "loss": 1.0443,
      "step": 630528
    },
    {
      "epoch": 47.32397748592871,
      "grad_norm": 0.6574602723121643,
      "learning_rate": 0.002828,
      "loss": 1.0383,
      "step": 630592
    },
    {
      "epoch": 47.32878048780488,
      "grad_norm": 0.5694827437400818,
      "learning_rate": 0.002828,
      "loss": 1.0408,
      "step": 630656
    },
    {
      "epoch": 47.333583489681054,
      "grad_norm": 0.5001234412193298,
      "learning_rate": 0.002828,
      "loss": 1.041,
      "step": 630720
    },
    {
      "epoch": 47.33838649155722,
      "grad_norm": 0.5457995533943176,
      "learning_rate": 0.002828,
      "loss": 1.0394,
      "step": 630784
    },
    {
      "epoch": 47.34318949343339,
      "grad_norm": 0.6352452039718628,
      "learning_rate": 0.002828,
      "loss": 1.042,
      "step": 630848
    },
    {
      "epoch": 47.34799249530957,
      "grad_norm": 0.565528154373169,
      "learning_rate": 0.002828,
      "loss": 1.0451,
      "step": 630912
    },
    {
      "epoch": 47.35279549718574,
      "grad_norm": 0.5789456963539124,
      "learning_rate": 0.002828,
      "loss": 1.0471,
      "step": 630976
    },
    {
      "epoch": 47.357598499061915,
      "grad_norm": 0.7093526124954224,
      "learning_rate": 0.002828,
      "loss": 1.0432,
      "step": 631040
    },
    {
      "epoch": 47.362401500938084,
      "grad_norm": 0.5827301144599915,
      "learning_rate": 0.002828,
      "loss": 1.0422,
      "step": 631104
    },
    {
      "epoch": 47.36720450281426,
      "grad_norm": 0.5498403906822205,
      "learning_rate": 0.002828,
      "loss": 1.0393,
      "step": 631168
    },
    {
      "epoch": 47.37200750469043,
      "grad_norm": 0.7173840403556824,
      "learning_rate": 0.002828,
      "loss": 1.046,
      "step": 631232
    },
    {
      "epoch": 47.376810506566606,
      "grad_norm": 0.6041868329048157,
      "learning_rate": 0.002828,
      "loss": 1.042,
      "step": 631296
    },
    {
      "epoch": 47.381613508442776,
      "grad_norm": 0.6391956806182861,
      "learning_rate": 0.002828,
      "loss": 1.0457,
      "step": 631360
    },
    {
      "epoch": 47.38641651031895,
      "grad_norm": 0.5060358047485352,
      "learning_rate": 0.002828,
      "loss": 1.0422,
      "step": 631424
    },
    {
      "epoch": 47.39121951219512,
      "grad_norm": 0.6481066346168518,
      "learning_rate": 0.002828,
      "loss": 1.0487,
      "step": 631488
    },
    {
      "epoch": 47.3960225140713,
      "grad_norm": 0.6985580325126648,
      "learning_rate": 0.002828,
      "loss": 1.0431,
      "step": 631552
    },
    {
      "epoch": 47.40082551594747,
      "grad_norm": 0.7035173773765564,
      "learning_rate": 0.002828,
      "loss": 1.0418,
      "step": 631616
    },
    {
      "epoch": 47.40562851782364,
      "grad_norm": 0.6084043383598328,
      "learning_rate": 0.002828,
      "loss": 1.0437,
      "step": 631680
    },
    {
      "epoch": 47.41043151969981,
      "grad_norm": 0.6706401705741882,
      "learning_rate": 0.002828,
      "loss": 1.0426,
      "step": 631744
    },
    {
      "epoch": 47.41523452157598,
      "grad_norm": 0.7065619230270386,
      "learning_rate": 0.002828,
      "loss": 1.0459,
      "step": 631808
    },
    {
      "epoch": 47.42003752345216,
      "grad_norm": 0.6161830425262451,
      "learning_rate": 0.002828,
      "loss": 1.0483,
      "step": 631872
    },
    {
      "epoch": 47.42484052532833,
      "grad_norm": 0.6157974600791931,
      "learning_rate": 0.002828,
      "loss": 1.0434,
      "step": 631936
    },
    {
      "epoch": 47.429643527204504,
      "grad_norm": 0.782438337802887,
      "learning_rate": 0.002828,
      "loss": 1.0485,
      "step": 632000
    },
    {
      "epoch": 47.434446529080674,
      "grad_norm": 0.6360306739807129,
      "learning_rate": 0.002828,
      "loss": 1.0463,
      "step": 632064
    },
    {
      "epoch": 47.43924953095685,
      "grad_norm": 0.6451125741004944,
      "learning_rate": 0.002828,
      "loss": 1.0484,
      "step": 632128
    },
    {
      "epoch": 47.44405253283302,
      "grad_norm": 0.7209149599075317,
      "learning_rate": 0.002828,
      "loss": 1.0443,
      "step": 632192
    },
    {
      "epoch": 47.448855534709196,
      "grad_norm": 0.5689561367034912,
      "learning_rate": 0.002828,
      "loss": 1.0479,
      "step": 632256
    },
    {
      "epoch": 47.453658536585365,
      "grad_norm": 0.632634699344635,
      "learning_rate": 0.002828,
      "loss": 1.0456,
      "step": 632320
    },
    {
      "epoch": 47.45846153846154,
      "grad_norm": 0.6760092377662659,
      "learning_rate": 0.002828,
      "loss": 1.0422,
      "step": 632384
    },
    {
      "epoch": 47.46326454033771,
      "grad_norm": 0.563585638999939,
      "learning_rate": 0.002828,
      "loss": 1.0476,
      "step": 632448
    },
    {
      "epoch": 47.46806754221388,
      "grad_norm": 0.6286882162094116,
      "learning_rate": 0.002828,
      "loss": 1.0466,
      "step": 632512
    },
    {
      "epoch": 47.47287054409006,
      "grad_norm": 0.7374217510223389,
      "learning_rate": 0.002828,
      "loss": 1.0463,
      "step": 632576
    },
    {
      "epoch": 47.477673545966226,
      "grad_norm": 0.5990324020385742,
      "learning_rate": 0.002828,
      "loss": 1.0424,
      "step": 632640
    },
    {
      "epoch": 47.4824765478424,
      "grad_norm": 0.5541861653327942,
      "learning_rate": 0.002828,
      "loss": 1.0487,
      "step": 632704
    },
    {
      "epoch": 47.48727954971857,
      "grad_norm": 0.49984487891197205,
      "learning_rate": 0.002828,
      "loss": 1.0469,
      "step": 632768
    },
    {
      "epoch": 47.49208255159475,
      "grad_norm": 0.7705683708190918,
      "learning_rate": 0.002828,
      "loss": 1.0461,
      "step": 632832
    },
    {
      "epoch": 47.49688555347092,
      "grad_norm": 0.5586681365966797,
      "learning_rate": 0.002828,
      "loss": 1.0466,
      "step": 632896
    },
    {
      "epoch": 47.501688555347094,
      "grad_norm": 0.6764875054359436,
      "learning_rate": 0.002828,
      "loss": 1.0536,
      "step": 632960
    },
    {
      "epoch": 47.506491557223264,
      "grad_norm": 0.6699135303497314,
      "learning_rate": 0.002828,
      "loss": 1.0484,
      "step": 633024
    },
    {
      "epoch": 47.51129455909944,
      "grad_norm": 0.7383409738540649,
      "learning_rate": 0.002828,
      "loss": 1.05,
      "step": 633088
    },
    {
      "epoch": 47.51609756097561,
      "grad_norm": 0.6897132396697998,
      "learning_rate": 0.002828,
      "loss": 1.0514,
      "step": 633152
    },
    {
      "epoch": 47.520900562851786,
      "grad_norm": 0.5044294595718384,
      "learning_rate": 0.002828,
      "loss": 1.0491,
      "step": 633216
    },
    {
      "epoch": 47.525703564727955,
      "grad_norm": 0.520473301410675,
      "learning_rate": 0.002828,
      "loss": 1.044,
      "step": 633280
    },
    {
      "epoch": 47.530506566604124,
      "grad_norm": 0.5634440183639526,
      "learning_rate": 0.002828,
      "loss": 1.046,
      "step": 633344
    },
    {
      "epoch": 47.5353095684803,
      "grad_norm": 0.6947305798530579,
      "learning_rate": 0.002828,
      "loss": 1.0506,
      "step": 633408
    },
    {
      "epoch": 47.54011257035647,
      "grad_norm": 0.6053649187088013,
      "learning_rate": 0.002828,
      "loss": 1.0543,
      "step": 633472
    },
    {
      "epoch": 47.54491557223265,
      "grad_norm": 0.644389271736145,
      "learning_rate": 0.002828,
      "loss": 1.0462,
      "step": 633536
    },
    {
      "epoch": 47.549718574108816,
      "grad_norm": 0.5872748494148254,
      "learning_rate": 0.002828,
      "loss": 1.0452,
      "step": 633600
    },
    {
      "epoch": 47.55452157598499,
      "grad_norm": 0.7785335779190063,
      "learning_rate": 0.002828,
      "loss": 1.0488,
      "step": 633664
    },
    {
      "epoch": 47.55932457786116,
      "grad_norm": 0.6512903571128845,
      "learning_rate": 0.002828,
      "loss": 1.0492,
      "step": 633728
    },
    {
      "epoch": 47.56412757973734,
      "grad_norm": 0.6293616890907288,
      "learning_rate": 0.002828,
      "loss": 1.0518,
      "step": 633792
    },
    {
      "epoch": 47.56893058161351,
      "grad_norm": 0.6513379216194153,
      "learning_rate": 0.002828,
      "loss": 1.0479,
      "step": 633856
    },
    {
      "epoch": 47.573733583489684,
      "grad_norm": 0.6688045859336853,
      "learning_rate": 0.002828,
      "loss": 1.0514,
      "step": 633920
    },
    {
      "epoch": 47.57853658536585,
      "grad_norm": 0.8207405209541321,
      "learning_rate": 0.002828,
      "loss": 1.0461,
      "step": 633984
    },
    {
      "epoch": 47.58333958724203,
      "grad_norm": 0.7107495069503784,
      "learning_rate": 0.002828,
      "loss": 1.0482,
      "step": 634048
    },
    {
      "epoch": 47.5881425891182,
      "grad_norm": 0.6240464448928833,
      "learning_rate": 0.002828,
      "loss": 1.0461,
      "step": 634112
    },
    {
      "epoch": 47.59294559099437,
      "grad_norm": 0.7034191489219666,
      "learning_rate": 0.002828,
      "loss": 1.0495,
      "step": 634176
    },
    {
      "epoch": 47.597748592870545,
      "grad_norm": 0.606263279914856,
      "learning_rate": 0.002828,
      "loss": 1.0536,
      "step": 634240
    },
    {
      "epoch": 47.602551594746714,
      "grad_norm": 0.6022868752479553,
      "learning_rate": 0.002828,
      "loss": 1.0513,
      "step": 634304
    },
    {
      "epoch": 47.60735459662289,
      "grad_norm": 0.9316113591194153,
      "learning_rate": 0.002828,
      "loss": 1.0488,
      "step": 634368
    },
    {
      "epoch": 47.61215759849906,
      "grad_norm": 0.5415453910827637,
      "learning_rate": 0.002828,
      "loss": 1.0522,
      "step": 634432
    },
    {
      "epoch": 47.616960600375236,
      "grad_norm": 0.6964948177337646,
      "learning_rate": 0.002828,
      "loss": 1.0508,
      "step": 634496
    },
    {
      "epoch": 47.621763602251406,
      "grad_norm": 0.6361346244812012,
      "learning_rate": 0.002828,
      "loss": 1.0522,
      "step": 634560
    },
    {
      "epoch": 47.62656660412758,
      "grad_norm": 0.778515100479126,
      "learning_rate": 0.002828,
      "loss": 1.053,
      "step": 634624
    },
    {
      "epoch": 47.63136960600375,
      "grad_norm": 0.6617556214332581,
      "learning_rate": 0.002828,
      "loss": 1.0443,
      "step": 634688
    },
    {
      "epoch": 47.63617260787993,
      "grad_norm": 0.642494261264801,
      "learning_rate": 0.002828,
      "loss": 1.0483,
      "step": 634752
    },
    {
      "epoch": 47.6409756097561,
      "grad_norm": 0.6483986973762512,
      "learning_rate": 0.002828,
      "loss": 1.0541,
      "step": 634816
    },
    {
      "epoch": 47.64577861163227,
      "grad_norm": 0.5259507894515991,
      "learning_rate": 0.002828,
      "loss": 1.0493,
      "step": 634880
    },
    {
      "epoch": 47.65058161350844,
      "grad_norm": 0.5775465369224548,
      "learning_rate": 0.002828,
      "loss": 1.0492,
      "step": 634944
    },
    {
      "epoch": 47.65538461538461,
      "grad_norm": 0.6609646677970886,
      "learning_rate": 0.002828,
      "loss": 1.0489,
      "step": 635008
    },
    {
      "epoch": 47.66018761726079,
      "grad_norm": 0.6297633647918701,
      "learning_rate": 0.002828,
      "loss": 1.0525,
      "step": 635072
    },
    {
      "epoch": 47.66499061913696,
      "grad_norm": 0.6732721924781799,
      "learning_rate": 0.002828,
      "loss": 1.0498,
      "step": 635136
    },
    {
      "epoch": 47.669793621013135,
      "grad_norm": 0.5661013722419739,
      "learning_rate": 0.002828,
      "loss": 1.0517,
      "step": 635200
    },
    {
      "epoch": 47.674596622889304,
      "grad_norm": 0.5473097562789917,
      "learning_rate": 0.002828,
      "loss": 1.0491,
      "step": 635264
    },
    {
      "epoch": 47.67939962476548,
      "grad_norm": 0.6529828906059265,
      "learning_rate": 0.002828,
      "loss": 1.0518,
      "step": 635328
    },
    {
      "epoch": 47.68420262664165,
      "grad_norm": 0.7554152607917786,
      "learning_rate": 0.002828,
      "loss": 1.0479,
      "step": 635392
    },
    {
      "epoch": 47.689005628517826,
      "grad_norm": 0.7058707475662231,
      "learning_rate": 0.002828,
      "loss": 1.0535,
      "step": 635456
    },
    {
      "epoch": 47.693808630393995,
      "grad_norm": 0.5582723021507263,
      "learning_rate": 0.002828,
      "loss": 1.0546,
      "step": 635520
    },
    {
      "epoch": 47.69861163227017,
      "grad_norm": 0.6613869667053223,
      "learning_rate": 0.002828,
      "loss": 1.0532,
      "step": 635584
    },
    {
      "epoch": 47.70341463414634,
      "grad_norm": 0.5572169423103333,
      "learning_rate": 0.002828,
      "loss": 1.0467,
      "step": 635648
    },
    {
      "epoch": 47.70821763602251,
      "grad_norm": 0.6240736246109009,
      "learning_rate": 0.002828,
      "loss": 1.0477,
      "step": 635712
    },
    {
      "epoch": 47.71302063789869,
      "grad_norm": 0.7397733926773071,
      "learning_rate": 0.002828,
      "loss": 1.0547,
      "step": 635776
    },
    {
      "epoch": 47.717823639774856,
      "grad_norm": 0.5110957026481628,
      "learning_rate": 0.002828,
      "loss": 1.0523,
      "step": 635840
    },
    {
      "epoch": 47.72262664165103,
      "grad_norm": 0.8373028039932251,
      "learning_rate": 0.002828,
      "loss": 1.0462,
      "step": 635904
    },
    {
      "epoch": 47.7274296435272,
      "grad_norm": 0.7218450903892517,
      "learning_rate": 0.002828,
      "loss": 1.057,
      "step": 635968
    },
    {
      "epoch": 47.73223264540338,
      "grad_norm": 0.6247406601905823,
      "learning_rate": 0.002828,
      "loss": 1.0528,
      "step": 636032
    },
    {
      "epoch": 47.73703564727955,
      "grad_norm": 0.720674455165863,
      "learning_rate": 0.002828,
      "loss": 1.0491,
      "step": 636096
    },
    {
      "epoch": 47.741838649155724,
      "grad_norm": 0.7045348286628723,
      "learning_rate": 0.002828,
      "loss": 1.0559,
      "step": 636160
    },
    {
      "epoch": 47.746641651031894,
      "grad_norm": 0.5386797189712524,
      "learning_rate": 0.002828,
      "loss": 1.0483,
      "step": 636224
    },
    {
      "epoch": 47.75144465290807,
      "grad_norm": 0.5467872619628906,
      "learning_rate": 0.002828,
      "loss": 1.0584,
      "step": 636288
    },
    {
      "epoch": 47.75624765478424,
      "grad_norm": 0.5429150462150574,
      "learning_rate": 0.002828,
      "loss": 1.0543,
      "step": 636352
    },
    {
      "epoch": 47.761050656660416,
      "grad_norm": 0.6549360752105713,
      "learning_rate": 0.002828,
      "loss": 1.0504,
      "step": 636416
    },
    {
      "epoch": 47.765853658536585,
      "grad_norm": 0.6696134209632874,
      "learning_rate": 0.002828,
      "loss": 1.0576,
      "step": 636480
    },
    {
      "epoch": 47.770656660412754,
      "grad_norm": 0.6573948264122009,
      "learning_rate": 0.002828,
      "loss": 1.0522,
      "step": 636544
    },
    {
      "epoch": 47.77545966228893,
      "grad_norm": 0.6131066083908081,
      "learning_rate": 0.002828,
      "loss": 1.0602,
      "step": 636608
    },
    {
      "epoch": 47.7802626641651,
      "grad_norm": 0.6040283441543579,
      "learning_rate": 0.002828,
      "loss": 1.0636,
      "step": 636672
    },
    {
      "epoch": 47.78506566604128,
      "grad_norm": 0.6244000196456909,
      "learning_rate": 0.002828,
      "loss": 1.0531,
      "step": 636736
    },
    {
      "epoch": 47.789868667917446,
      "grad_norm": 0.7662160992622375,
      "learning_rate": 0.002828,
      "loss": 1.0522,
      "step": 636800
    },
    {
      "epoch": 47.79467166979362,
      "grad_norm": 0.5642756223678589,
      "learning_rate": 0.002828,
      "loss": 1.0548,
      "step": 636864
    },
    {
      "epoch": 47.79947467166979,
      "grad_norm": 0.7008233070373535,
      "learning_rate": 0.002828,
      "loss": 1.0548,
      "step": 636928
    },
    {
      "epoch": 47.80427767354597,
      "grad_norm": 0.6942220330238342,
      "learning_rate": 0.002828,
      "loss": 1.0566,
      "step": 636992
    },
    {
      "epoch": 47.80908067542214,
      "grad_norm": 0.7647560238838196,
      "learning_rate": 0.002828,
      "loss": 1.0545,
      "step": 637056
    },
    {
      "epoch": 47.813883677298314,
      "grad_norm": 0.6823753714561462,
      "learning_rate": 0.002828,
      "loss": 1.0528,
      "step": 637120
    },
    {
      "epoch": 47.81868667917448,
      "grad_norm": 0.5762351155281067,
      "learning_rate": 0.002828,
      "loss": 1.0522,
      "step": 637184
    },
    {
      "epoch": 47.82348968105066,
      "grad_norm": 0.6646673083305359,
      "learning_rate": 0.002828,
      "loss": 1.0487,
      "step": 637248
    },
    {
      "epoch": 47.82829268292683,
      "grad_norm": 0.604059100151062,
      "learning_rate": 0.002828,
      "loss": 1.0624,
      "step": 637312
    },
    {
      "epoch": 47.833095684803,
      "grad_norm": 0.6984003186225891,
      "learning_rate": 0.002828,
      "loss": 1.0504,
      "step": 637376
    },
    {
      "epoch": 47.837898686679175,
      "grad_norm": 0.5845495462417603,
      "learning_rate": 0.002828,
      "loss": 1.0545,
      "step": 637440
    },
    {
      "epoch": 47.842701688555344,
      "grad_norm": 0.6453345417976379,
      "learning_rate": 0.002828,
      "loss": 1.0536,
      "step": 637504
    },
    {
      "epoch": 47.84750469043152,
      "grad_norm": 0.5302745699882507,
      "learning_rate": 0.002828,
      "loss": 1.0527,
      "step": 637568
    },
    {
      "epoch": 47.85230769230769,
      "grad_norm": 0.6094585061073303,
      "learning_rate": 0.002828,
      "loss": 1.055,
      "step": 637632
    },
    {
      "epoch": 47.857110694183866,
      "grad_norm": 0.5993672013282776,
      "learning_rate": 0.002828,
      "loss": 1.0542,
      "step": 637696
    },
    {
      "epoch": 47.861913696060036,
      "grad_norm": 0.6446161866188049,
      "learning_rate": 0.002828,
      "loss": 1.0516,
      "step": 637760
    },
    {
      "epoch": 47.86671669793621,
      "grad_norm": 0.6541599631309509,
      "learning_rate": 0.002828,
      "loss": 1.0557,
      "step": 637824
    },
    {
      "epoch": 47.87151969981238,
      "grad_norm": 0.6368933916091919,
      "learning_rate": 0.002828,
      "loss": 1.0582,
      "step": 637888
    },
    {
      "epoch": 47.87632270168856,
      "grad_norm": 0.7711507678031921,
      "learning_rate": 0.002828,
      "loss": 1.0542,
      "step": 637952
    },
    {
      "epoch": 47.88112570356473,
      "grad_norm": 0.7595430016517639,
      "learning_rate": 0.002828,
      "loss": 1.0588,
      "step": 638016
    },
    {
      "epoch": 47.885928705440904,
      "grad_norm": 0.5265303254127502,
      "learning_rate": 0.002828,
      "loss": 1.0572,
      "step": 638080
    },
    {
      "epoch": 47.89073170731707,
      "grad_norm": 0.7377369999885559,
      "learning_rate": 0.002828,
      "loss": 1.0622,
      "step": 638144
    },
    {
      "epoch": 47.89553470919324,
      "grad_norm": 0.6574912667274475,
      "learning_rate": 0.002828,
      "loss": 1.0527,
      "step": 638208
    },
    {
      "epoch": 47.90033771106942,
      "grad_norm": 0.6473036408424377,
      "learning_rate": 0.002828,
      "loss": 1.0571,
      "step": 638272
    },
    {
      "epoch": 47.90514071294559,
      "grad_norm": 0.6601776480674744,
      "learning_rate": 0.002828,
      "loss": 1.0571,
      "step": 638336
    },
    {
      "epoch": 47.909943714821765,
      "grad_norm": 0.6472011208534241,
      "learning_rate": 0.002828,
      "loss": 1.0577,
      "step": 638400
    },
    {
      "epoch": 47.914746716697934,
      "grad_norm": 0.701826274394989,
      "learning_rate": 0.002828,
      "loss": 1.0603,
      "step": 638464
    },
    {
      "epoch": 47.91954971857411,
      "grad_norm": 0.5442734360694885,
      "learning_rate": 0.002828,
      "loss": 1.0587,
      "step": 638528
    },
    {
      "epoch": 47.92435272045028,
      "grad_norm": 0.683586061000824,
      "learning_rate": 0.002828,
      "loss": 1.0598,
      "step": 638592
    },
    {
      "epoch": 47.929155722326456,
      "grad_norm": 0.6531975269317627,
      "learning_rate": 0.002828,
      "loss": 1.054,
      "step": 638656
    },
    {
      "epoch": 47.933958724202625,
      "grad_norm": 0.7838884592056274,
      "learning_rate": 0.002828,
      "loss": 1.0581,
      "step": 638720
    },
    {
      "epoch": 47.9387617260788,
      "grad_norm": 0.6613420248031616,
      "learning_rate": 0.002828,
      "loss": 1.0596,
      "step": 638784
    },
    {
      "epoch": 47.94356472795497,
      "grad_norm": 0.6097777485847473,
      "learning_rate": 0.002828,
      "loss": 1.0533,
      "step": 638848
    },
    {
      "epoch": 47.94836772983115,
      "grad_norm": 0.5991477370262146,
      "learning_rate": 0.002828,
      "loss": 1.0531,
      "step": 638912
    },
    {
      "epoch": 47.95317073170732,
      "grad_norm": 0.741007924079895,
      "learning_rate": 0.002828,
      "loss": 1.0585,
      "step": 638976
    },
    {
      "epoch": 47.957973733583486,
      "grad_norm": 0.6317166090011597,
      "learning_rate": 0.002828,
      "loss": 1.0596,
      "step": 639040
    },
    {
      "epoch": 47.96277673545966,
      "grad_norm": 0.6188191771507263,
      "learning_rate": 0.002828,
      "loss": 1.0558,
      "step": 639104
    },
    {
      "epoch": 47.96757973733583,
      "grad_norm": 0.7189934253692627,
      "learning_rate": 0.002828,
      "loss": 1.0571,
      "step": 639168
    },
    {
      "epoch": 47.97238273921201,
      "grad_norm": 0.6403295993804932,
      "learning_rate": 0.002828,
      "loss": 1.0533,
      "step": 639232
    },
    {
      "epoch": 47.97718574108818,
      "grad_norm": 0.6693562865257263,
      "learning_rate": 0.002828,
      "loss": 1.0531,
      "step": 639296
    },
    {
      "epoch": 47.981988742964354,
      "grad_norm": 0.545728325843811,
      "learning_rate": 0.002828,
      "loss": 1.0589,
      "step": 639360
    },
    {
      "epoch": 47.986791744840524,
      "grad_norm": 0.6077553629875183,
      "learning_rate": 0.002828,
      "loss": 1.0634,
      "step": 639424
    },
    {
      "epoch": 47.9915947467167,
      "grad_norm": 0.5200755000114441,
      "learning_rate": 0.002828,
      "loss": 1.061,
      "step": 639488
    },
    {
      "epoch": 47.99639774859287,
      "grad_norm": 0.589529812335968,
      "learning_rate": 0.002828,
      "loss": 1.0567,
      "step": 639552
    },
    {
      "epoch": 48.001200750469046,
      "grad_norm": 0.7281575202941895,
      "learning_rate": 0.002828,
      "loss": 1.0468,
      "step": 639616
    },
    {
      "epoch": 48.006003752345215,
      "grad_norm": 0.6403776407241821,
      "learning_rate": 0.002828,
      "loss": 1.0214,
      "step": 639680
    },
    {
      "epoch": 48.01080675422139,
      "grad_norm": 0.5161173939704895,
      "learning_rate": 0.002828,
      "loss": 1.0208,
      "step": 639744
    },
    {
      "epoch": 48.01560975609756,
      "grad_norm": 0.5105540752410889,
      "learning_rate": 0.002828,
      "loss": 1.0238,
      "step": 639808
    },
    {
      "epoch": 48.02041275797373,
      "grad_norm": 0.6909409761428833,
      "learning_rate": 0.002828,
      "loss": 1.02,
      "step": 639872
    },
    {
      "epoch": 48.02521575984991,
      "grad_norm": 0.6992323398590088,
      "learning_rate": 0.002828,
      "loss": 1.0245,
      "step": 639936
    },
    {
      "epoch": 48.030018761726076,
      "grad_norm": 0.6982917785644531,
      "learning_rate": 0.002828,
      "loss": 1.0298,
      "step": 640000
    },
    {
      "epoch": 48.03482176360225,
      "grad_norm": 0.7123545408248901,
      "learning_rate": 0.002828,
      "loss": 1.019,
      "step": 640064
    },
    {
      "epoch": 48.03962476547842,
      "grad_norm": 0.6596068143844604,
      "learning_rate": 0.002828,
      "loss": 1.0297,
      "step": 640128
    },
    {
      "epoch": 48.0444277673546,
      "grad_norm": 0.7315285205841064,
      "learning_rate": 0.002828,
      "loss": 1.0268,
      "step": 640192
    },
    {
      "epoch": 48.04923076923077,
      "grad_norm": 0.6878241896629333,
      "learning_rate": 0.002828,
      "loss": 1.0234,
      "step": 640256
    },
    {
      "epoch": 48.054033771106944,
      "grad_norm": 0.5641777515411377,
      "learning_rate": 0.002828,
      "loss": 1.0302,
      "step": 640320
    },
    {
      "epoch": 48.05883677298311,
      "grad_norm": 0.6245089769363403,
      "learning_rate": 0.002828,
      "loss": 1.0281,
      "step": 640384
    },
    {
      "epoch": 48.06363977485929,
      "grad_norm": 0.7929434180259705,
      "learning_rate": 0.002828,
      "loss": 1.0297,
      "step": 640448
    },
    {
      "epoch": 48.06844277673546,
      "grad_norm": 0.5667207837104797,
      "learning_rate": 0.002828,
      "loss": 1.0263,
      "step": 640512
    },
    {
      "epoch": 48.073245778611636,
      "grad_norm": 0.6004430055618286,
      "learning_rate": 0.002828,
      "loss": 1.0275,
      "step": 640576
    },
    {
      "epoch": 48.078048780487805,
      "grad_norm": 0.6360554695129395,
      "learning_rate": 0.002828,
      "loss": 1.0265,
      "step": 640640
    },
    {
      "epoch": 48.082851782363974,
      "grad_norm": 0.6814728379249573,
      "learning_rate": 0.002828,
      "loss": 1.031,
      "step": 640704
    },
    {
      "epoch": 48.08765478424015,
      "grad_norm": 0.657783567905426,
      "learning_rate": 0.002828,
      "loss": 1.0234,
      "step": 640768
    },
    {
      "epoch": 48.09245778611632,
      "grad_norm": 0.5107567310333252,
      "learning_rate": 0.002828,
      "loss": 1.0254,
      "step": 640832
    },
    {
      "epoch": 48.0972607879925,
      "grad_norm": 0.6924241781234741,
      "learning_rate": 0.002828,
      "loss": 1.0326,
      "step": 640896
    },
    {
      "epoch": 48.102063789868666,
      "grad_norm": 0.6154158115386963,
      "learning_rate": 0.002828,
      "loss": 1.0229,
      "step": 640960
    },
    {
      "epoch": 48.10686679174484,
      "grad_norm": 0.685716986656189,
      "learning_rate": 0.002828,
      "loss": 1.0256,
      "step": 641024
    },
    {
      "epoch": 48.11166979362101,
      "grad_norm": 0.6853328347206116,
      "learning_rate": 0.002828,
      "loss": 1.0376,
      "step": 641088
    },
    {
      "epoch": 48.11647279549719,
      "grad_norm": 0.5760686993598938,
      "learning_rate": 0.002828,
      "loss": 1.028,
      "step": 641152
    },
    {
      "epoch": 48.12127579737336,
      "grad_norm": 0.5979732275009155,
      "learning_rate": 0.002828,
      "loss": 1.0299,
      "step": 641216
    },
    {
      "epoch": 48.126078799249534,
      "grad_norm": 0.6231614947319031,
      "learning_rate": 0.002828,
      "loss": 1.0352,
      "step": 641280
    },
    {
      "epoch": 48.1308818011257,
      "grad_norm": 0.6504188179969788,
      "learning_rate": 0.002828,
      "loss": 1.0261,
      "step": 641344
    },
    {
      "epoch": 48.13568480300188,
      "grad_norm": 0.6307108402252197,
      "learning_rate": 0.002828,
      "loss": 1.0322,
      "step": 641408
    },
    {
      "epoch": 48.14048780487805,
      "grad_norm": 0.591966986656189,
      "learning_rate": 0.002828,
      "loss": 1.0313,
      "step": 641472
    },
    {
      "epoch": 48.14529080675422,
      "grad_norm": 0.5324594378471375,
      "learning_rate": 0.002828,
      "loss": 1.0335,
      "step": 641536
    },
    {
      "epoch": 48.150093808630395,
      "grad_norm": 0.7049140930175781,
      "learning_rate": 0.002828,
      "loss": 1.0277,
      "step": 641600
    },
    {
      "epoch": 48.154896810506564,
      "grad_norm": 0.7575385570526123,
      "learning_rate": 0.002828,
      "loss": 1.0301,
      "step": 641664
    },
    {
      "epoch": 48.15969981238274,
      "grad_norm": 0.7463407516479492,
      "learning_rate": 0.002828,
      "loss": 1.0307,
      "step": 641728
    },
    {
      "epoch": 48.16450281425891,
      "grad_norm": 0.6462583541870117,
      "learning_rate": 0.002828,
      "loss": 1.0286,
      "step": 641792
    },
    {
      "epoch": 48.169305816135086,
      "grad_norm": 0.6489346623420715,
      "learning_rate": 0.002828,
      "loss": 1.0273,
      "step": 641856
    },
    {
      "epoch": 48.174108818011256,
      "grad_norm": 0.6180921196937561,
      "learning_rate": 0.002828,
      "loss": 1.0381,
      "step": 641920
    },
    {
      "epoch": 48.17891181988743,
      "grad_norm": 0.7339633703231812,
      "learning_rate": 0.002828,
      "loss": 1.0339,
      "step": 641984
    },
    {
      "epoch": 48.1837148217636,
      "grad_norm": 0.6105630993843079,
      "learning_rate": 0.002828,
      "loss": 1.0335,
      "step": 642048
    },
    {
      "epoch": 48.18851782363978,
      "grad_norm": 0.5919303894042969,
      "learning_rate": 0.002828,
      "loss": 1.0318,
      "step": 642112
    },
    {
      "epoch": 48.19332082551595,
      "grad_norm": 0.6174924969673157,
      "learning_rate": 0.002828,
      "loss": 1.0308,
      "step": 642176
    },
    {
      "epoch": 48.19812382739212,
      "grad_norm": 0.7035306692123413,
      "learning_rate": 0.002828,
      "loss": 1.0316,
      "step": 642240
    },
    {
      "epoch": 48.20292682926829,
      "grad_norm": 0.5673324465751648,
      "learning_rate": 0.002828,
      "loss": 1.0362,
      "step": 642304
    },
    {
      "epoch": 48.20772983114446,
      "grad_norm": 0.6283726692199707,
      "learning_rate": 0.002828,
      "loss": 1.0305,
      "step": 642368
    },
    {
      "epoch": 48.21253283302064,
      "grad_norm": 0.6875553727149963,
      "learning_rate": 0.002828,
      "loss": 1.0311,
      "step": 642432
    },
    {
      "epoch": 48.21733583489681,
      "grad_norm": 0.7149839997291565,
      "learning_rate": 0.002828,
      "loss": 1.0359,
      "step": 642496
    },
    {
      "epoch": 48.222138836772984,
      "grad_norm": 0.5285342931747437,
      "learning_rate": 0.002828,
      "loss": 1.0351,
      "step": 642560
    },
    {
      "epoch": 48.226941838649154,
      "grad_norm": 0.5428668856620789,
      "learning_rate": 0.002828,
      "loss": 1.0336,
      "step": 642624
    },
    {
      "epoch": 48.23174484052533,
      "grad_norm": 0.5632164478302002,
      "learning_rate": 0.002828,
      "loss": 1.037,
      "step": 642688
    },
    {
      "epoch": 48.2365478424015,
      "grad_norm": 0.710423469543457,
      "learning_rate": 0.002828,
      "loss": 1.0398,
      "step": 642752
    },
    {
      "epoch": 48.241350844277676,
      "grad_norm": 0.7015545964241028,
      "learning_rate": 0.002828,
      "loss": 1.034,
      "step": 642816
    },
    {
      "epoch": 48.246153846153845,
      "grad_norm": 0.6140884757041931,
      "learning_rate": 0.002828,
      "loss": 1.0347,
      "step": 642880
    },
    {
      "epoch": 48.25095684803002,
      "grad_norm": 0.6183640360832214,
      "learning_rate": 0.002828,
      "loss": 1.0361,
      "step": 642944
    },
    {
      "epoch": 48.25575984990619,
      "grad_norm": 0.6626155376434326,
      "learning_rate": 0.002828,
      "loss": 1.038,
      "step": 643008
    },
    {
      "epoch": 48.26056285178237,
      "grad_norm": 0.7320151329040527,
      "learning_rate": 0.002828,
      "loss": 1.0372,
      "step": 643072
    },
    {
      "epoch": 48.26536585365854,
      "grad_norm": 0.6230993866920471,
      "learning_rate": 0.002828,
      "loss": 1.0276,
      "step": 643136
    },
    {
      "epoch": 48.270168855534706,
      "grad_norm": 0.658531904220581,
      "learning_rate": 0.002828,
      "loss": 1.0434,
      "step": 643200
    },
    {
      "epoch": 48.27497185741088,
      "grad_norm": 0.5257073044776917,
      "learning_rate": 0.002828,
      "loss": 1.0349,
      "step": 643264
    },
    {
      "epoch": 48.27977485928705,
      "grad_norm": 0.7808026671409607,
      "learning_rate": 0.002828,
      "loss": 1.0312,
      "step": 643328
    },
    {
      "epoch": 48.28457786116323,
      "grad_norm": 0.6385923027992249,
      "learning_rate": 0.002828,
      "loss": 1.0333,
      "step": 643392
    },
    {
      "epoch": 48.2893808630394,
      "grad_norm": 0.693492591381073,
      "learning_rate": 0.002828,
      "loss": 1.0359,
      "step": 643456
    },
    {
      "epoch": 48.294183864915574,
      "grad_norm": 0.5596438050270081,
      "learning_rate": 0.002828,
      "loss": 1.0354,
      "step": 643520
    },
    {
      "epoch": 48.29898686679174,
      "grad_norm": 0.6546857357025146,
      "learning_rate": 0.002828,
      "loss": 1.0341,
      "step": 643584
    },
    {
      "epoch": 48.30378986866792,
      "grad_norm": 0.5163006782531738,
      "learning_rate": 0.002828,
      "loss": 1.0316,
      "step": 643648
    },
    {
      "epoch": 48.30859287054409,
      "grad_norm": 0.553073525428772,
      "learning_rate": 0.002828,
      "loss": 1.0349,
      "step": 643712
    },
    {
      "epoch": 48.313395872420266,
      "grad_norm": 0.7049180865287781,
      "learning_rate": 0.002828,
      "loss": 1.0343,
      "step": 643776
    },
    {
      "epoch": 48.318198874296435,
      "grad_norm": 0.6629697680473328,
      "learning_rate": 0.002828,
      "loss": 1.0348,
      "step": 643840
    },
    {
      "epoch": 48.32300187617261,
      "grad_norm": 0.6142928600311279,
      "learning_rate": 0.002828,
      "loss": 1.0374,
      "step": 643904
    },
    {
      "epoch": 48.32780487804878,
      "grad_norm": 0.8215116262435913,
      "learning_rate": 0.002828,
      "loss": 1.0362,
      "step": 643968
    },
    {
      "epoch": 48.33260787992495,
      "grad_norm": 0.7114266753196716,
      "learning_rate": 0.002828,
      "loss": 1.0352,
      "step": 644032
    },
    {
      "epoch": 48.33741088180113,
      "grad_norm": 0.742658257484436,
      "learning_rate": 0.002828,
      "loss": 1.0356,
      "step": 644096
    },
    {
      "epoch": 48.342213883677296,
      "grad_norm": 0.5757499933242798,
      "learning_rate": 0.002828,
      "loss": 1.0346,
      "step": 644160
    },
    {
      "epoch": 48.34701688555347,
      "grad_norm": 0.6741271615028381,
      "learning_rate": 0.002828,
      "loss": 1.0449,
      "step": 644224
    },
    {
      "epoch": 48.35181988742964,
      "grad_norm": 0.7909224033355713,
      "learning_rate": 0.002828,
      "loss": 1.0351,
      "step": 644288
    },
    {
      "epoch": 48.35662288930582,
      "grad_norm": 0.6528041958808899,
      "learning_rate": 0.002828,
      "loss": 1.0401,
      "step": 644352
    },
    {
      "epoch": 48.36142589118199,
      "grad_norm": 0.6354573965072632,
      "learning_rate": 0.002828,
      "loss": 1.0327,
      "step": 644416
    },
    {
      "epoch": 48.366228893058164,
      "grad_norm": 0.7101730704307556,
      "learning_rate": 0.002828,
      "loss": 1.0344,
      "step": 644480
    },
    {
      "epoch": 48.37103189493433,
      "grad_norm": 0.5995579361915588,
      "learning_rate": 0.002828,
      "loss": 1.0401,
      "step": 644544
    },
    {
      "epoch": 48.37583489681051,
      "grad_norm": 0.6640355587005615,
      "learning_rate": 0.002828,
      "loss": 1.0396,
      "step": 644608
    },
    {
      "epoch": 48.38063789868668,
      "grad_norm": 0.6031660437583923,
      "learning_rate": 0.002828,
      "loss": 1.0399,
      "step": 644672
    },
    {
      "epoch": 48.38544090056285,
      "grad_norm": 0.6285581588745117,
      "learning_rate": 0.002828,
      "loss": 1.0368,
      "step": 644736
    },
    {
      "epoch": 48.390243902439025,
      "grad_norm": 0.6844165325164795,
      "learning_rate": 0.002828,
      "loss": 1.0411,
      "step": 644800
    },
    {
      "epoch": 48.395046904315194,
      "grad_norm": 0.6001917719841003,
      "learning_rate": 0.002828,
      "loss": 1.0373,
      "step": 644864
    },
    {
      "epoch": 48.39984990619137,
      "grad_norm": 0.6581211686134338,
      "learning_rate": 0.002828,
      "loss": 1.0319,
      "step": 644928
    },
    {
      "epoch": 48.40465290806754,
      "grad_norm": 0.6621804237365723,
      "learning_rate": 0.002828,
      "loss": 1.0385,
      "step": 644992
    },
    {
      "epoch": 48.409455909943716,
      "grad_norm": 0.6346020102500916,
      "learning_rate": 0.002828,
      "loss": 1.0392,
      "step": 645056
    },
    {
      "epoch": 48.414258911819886,
      "grad_norm": 0.6103173494338989,
      "learning_rate": 0.002828,
      "loss": 1.0404,
      "step": 645120
    },
    {
      "epoch": 48.41906191369606,
      "grad_norm": 0.7052111029624939,
      "learning_rate": 0.002828,
      "loss": 1.0412,
      "step": 645184
    },
    {
      "epoch": 48.42386491557223,
      "grad_norm": 0.6104574799537659,
      "learning_rate": 0.002828,
      "loss": 1.0333,
      "step": 645248
    },
    {
      "epoch": 48.42866791744841,
      "grad_norm": 0.5670404434204102,
      "learning_rate": 0.002828,
      "loss": 1.0371,
      "step": 645312
    },
    {
      "epoch": 48.43347091932458,
      "grad_norm": 0.6457066535949707,
      "learning_rate": 0.002828,
      "loss": 1.0381,
      "step": 645376
    },
    {
      "epoch": 48.438273921200754,
      "grad_norm": 0.7141438722610474,
      "learning_rate": 0.002828,
      "loss": 1.04,
      "step": 645440
    },
    {
      "epoch": 48.44307692307692,
      "grad_norm": 0.6793937683105469,
      "learning_rate": 0.002828,
      "loss": 1.0392,
      "step": 645504
    },
    {
      "epoch": 48.44787992495309,
      "grad_norm": 0.6794419288635254,
      "learning_rate": 0.002828,
      "loss": 1.0315,
      "step": 645568
    },
    {
      "epoch": 48.45268292682927,
      "grad_norm": 0.5492742657661438,
      "learning_rate": 0.002828,
      "loss": 1.0426,
      "step": 645632
    },
    {
      "epoch": 48.45748592870544,
      "grad_norm": 0.5703292489051819,
      "learning_rate": 0.002828,
      "loss": 1.0382,
      "step": 645696
    },
    {
      "epoch": 48.462288930581614,
      "grad_norm": 0.6498845815658569,
      "learning_rate": 0.002828,
      "loss": 1.0413,
      "step": 645760
    },
    {
      "epoch": 48.467091932457784,
      "grad_norm": 0.6196924448013306,
      "learning_rate": 0.002828,
      "loss": 1.0419,
      "step": 645824
    },
    {
      "epoch": 48.47189493433396,
      "grad_norm": 0.8179911375045776,
      "learning_rate": 0.002828,
      "loss": 1.0418,
      "step": 645888
    },
    {
      "epoch": 48.47669793621013,
      "grad_norm": 0.6045671701431274,
      "learning_rate": 0.002828,
      "loss": 1.046,
      "step": 645952
    },
    {
      "epoch": 48.481500938086306,
      "grad_norm": 0.6320502161979675,
      "learning_rate": 0.002828,
      "loss": 1.0441,
      "step": 646016
    },
    {
      "epoch": 48.486303939962475,
      "grad_norm": 0.6962401866912842,
      "learning_rate": 0.002828,
      "loss": 1.041,
      "step": 646080
    },
    {
      "epoch": 48.49110694183865,
      "grad_norm": 0.6397041082382202,
      "learning_rate": 0.002828,
      "loss": 1.0428,
      "step": 646144
    },
    {
      "epoch": 48.49590994371482,
      "grad_norm": 0.599310040473938,
      "learning_rate": 0.002828,
      "loss": 1.0461,
      "step": 646208
    },
    {
      "epoch": 48.500712945591,
      "grad_norm": 0.5711956024169922,
      "learning_rate": 0.002828,
      "loss": 1.0386,
      "step": 646272
    },
    {
      "epoch": 48.50551594746717,
      "grad_norm": 0.6898519396781921,
      "learning_rate": 0.002828,
      "loss": 1.0441,
      "step": 646336
    },
    {
      "epoch": 48.510318949343336,
      "grad_norm": 0.818206787109375,
      "learning_rate": 0.002828,
      "loss": 1.0452,
      "step": 646400
    },
    {
      "epoch": 48.51512195121951,
      "grad_norm": 0.571017861366272,
      "learning_rate": 0.002828,
      "loss": 1.0436,
      "step": 646464
    },
    {
      "epoch": 48.51992495309568,
      "grad_norm": 0.5874201059341431,
      "learning_rate": 0.002828,
      "loss": 1.0453,
      "step": 646528
    },
    {
      "epoch": 48.52472795497186,
      "grad_norm": 0.5294089317321777,
      "learning_rate": 0.002828,
      "loss": 1.0431,
      "step": 646592
    },
    {
      "epoch": 48.52953095684803,
      "grad_norm": 0.6514833569526672,
      "learning_rate": 0.002828,
      "loss": 1.0457,
      "step": 646656
    },
    {
      "epoch": 48.534333958724204,
      "grad_norm": 0.5631481409072876,
      "learning_rate": 0.002828,
      "loss": 1.0405,
      "step": 646720
    },
    {
      "epoch": 48.53913696060037,
      "grad_norm": 0.7069897055625916,
      "learning_rate": 0.002828,
      "loss": 1.0445,
      "step": 646784
    },
    {
      "epoch": 48.54393996247655,
      "grad_norm": 0.589514434337616,
      "learning_rate": 0.002828,
      "loss": 1.0424,
      "step": 646848
    },
    {
      "epoch": 48.54874296435272,
      "grad_norm": 0.6267862915992737,
      "learning_rate": 0.002828,
      "loss": 1.0413,
      "step": 646912
    },
    {
      "epoch": 48.553545966228896,
      "grad_norm": 0.637648344039917,
      "learning_rate": 0.002828,
      "loss": 1.0412,
      "step": 646976
    },
    {
      "epoch": 48.558348968105065,
      "grad_norm": 0.5455642342567444,
      "learning_rate": 0.002828,
      "loss": 1.0476,
      "step": 647040
    },
    {
      "epoch": 48.56315196998124,
      "grad_norm": 0.7661427855491638,
      "learning_rate": 0.002828,
      "loss": 1.0432,
      "step": 647104
    },
    {
      "epoch": 48.56795497185741,
      "grad_norm": 0.7244403958320618,
      "learning_rate": 0.002828,
      "loss": 1.046,
      "step": 647168
    },
    {
      "epoch": 48.57275797373358,
      "grad_norm": 0.5974761247634888,
      "learning_rate": 0.002828,
      "loss": 1.0377,
      "step": 647232
    },
    {
      "epoch": 48.57756097560976,
      "grad_norm": 0.7058935761451721,
      "learning_rate": 0.002828,
      "loss": 1.0409,
      "step": 647296
    },
    {
      "epoch": 48.582363977485926,
      "grad_norm": 0.5798015594482422,
      "learning_rate": 0.002828,
      "loss": 1.0429,
      "step": 647360
    },
    {
      "epoch": 48.5871669793621,
      "grad_norm": 0.5887682437896729,
      "learning_rate": 0.002828,
      "loss": 1.0444,
      "step": 647424
    },
    {
      "epoch": 48.59196998123827,
      "grad_norm": 0.680789589881897,
      "learning_rate": 0.002828,
      "loss": 1.0446,
      "step": 647488
    },
    {
      "epoch": 48.59677298311445,
      "grad_norm": 0.5927484035491943,
      "learning_rate": 0.002828,
      "loss": 1.0449,
      "step": 647552
    },
    {
      "epoch": 48.60157598499062,
      "grad_norm": 0.5905951857566833,
      "learning_rate": 0.002828,
      "loss": 1.0444,
      "step": 647616
    },
    {
      "epoch": 48.606378986866794,
      "grad_norm": 0.6388547420501709,
      "learning_rate": 0.002828,
      "loss": 1.0458,
      "step": 647680
    },
    {
      "epoch": 48.61118198874296,
      "grad_norm": 0.6019654870033264,
      "learning_rate": 0.002828,
      "loss": 1.0469,
      "step": 647744
    },
    {
      "epoch": 48.61598499061914,
      "grad_norm": 0.6199590563774109,
      "learning_rate": 0.002828,
      "loss": 1.0444,
      "step": 647808
    },
    {
      "epoch": 48.62078799249531,
      "grad_norm": 0.6658002138137817,
      "learning_rate": 0.002828,
      "loss": 1.0449,
      "step": 647872
    },
    {
      "epoch": 48.625590994371485,
      "grad_norm": 0.6782828569412231,
      "learning_rate": 0.002828,
      "loss": 1.0459,
      "step": 647936
    },
    {
      "epoch": 48.630393996247655,
      "grad_norm": 0.8401302695274353,
      "learning_rate": 0.002828,
      "loss": 1.05,
      "step": 648000
    },
    {
      "epoch": 48.635196998123824,
      "grad_norm": 0.4938647150993347,
      "learning_rate": 0.002828,
      "loss": 1.0463,
      "step": 648064
    },
    {
      "epoch": 48.64,
      "grad_norm": 0.8528944253921509,
      "learning_rate": 0.002828,
      "loss": 1.0463,
      "step": 648128
    },
    {
      "epoch": 48.64480300187617,
      "grad_norm": 0.6374745965003967,
      "learning_rate": 0.002828,
      "loss": 1.0427,
      "step": 648192
    },
    {
      "epoch": 48.649606003752346,
      "grad_norm": 0.5849046111106873,
      "learning_rate": 0.002828,
      "loss": 1.0441,
      "step": 648256
    },
    {
      "epoch": 48.654409005628516,
      "grad_norm": 0.576451301574707,
      "learning_rate": 0.002828,
      "loss": 1.0453,
      "step": 648320
    },
    {
      "epoch": 48.65921200750469,
      "grad_norm": 0.5208295583724976,
      "learning_rate": 0.002828,
      "loss": 1.0482,
      "step": 648384
    },
    {
      "epoch": 48.66401500938086,
      "grad_norm": 0.649178147315979,
      "learning_rate": 0.002828,
      "loss": 1.0494,
      "step": 648448
    },
    {
      "epoch": 48.66881801125704,
      "grad_norm": 0.6612587571144104,
      "learning_rate": 0.002828,
      "loss": 1.0453,
      "step": 648512
    },
    {
      "epoch": 48.67362101313321,
      "grad_norm": 0.8569740653038025,
      "learning_rate": 0.002828,
      "loss": 1.0443,
      "step": 648576
    },
    {
      "epoch": 48.678424015009384,
      "grad_norm": 0.6810051798820496,
      "learning_rate": 0.002828,
      "loss": 1.053,
      "step": 648640
    },
    {
      "epoch": 48.68322701688555,
      "grad_norm": 0.6148619055747986,
      "learning_rate": 0.002828,
      "loss": 1.0457,
      "step": 648704
    },
    {
      "epoch": 48.68803001876173,
      "grad_norm": 0.5363982915878296,
      "learning_rate": 0.002828,
      "loss": 1.0462,
      "step": 648768
    },
    {
      "epoch": 48.6928330206379,
      "grad_norm": 0.662300169467926,
      "learning_rate": 0.002828,
      "loss": 1.0437,
      "step": 648832
    },
    {
      "epoch": 48.69763602251407,
      "grad_norm": 0.5750449895858765,
      "learning_rate": 0.002828,
      "loss": 1.0484,
      "step": 648896
    },
    {
      "epoch": 48.702439024390245,
      "grad_norm": 0.5501517057418823,
      "learning_rate": 0.002828,
      "loss": 1.0393,
      "step": 648960
    },
    {
      "epoch": 48.707242026266414,
      "grad_norm": 0.721763014793396,
      "learning_rate": 0.002828,
      "loss": 1.0528,
      "step": 649024
    },
    {
      "epoch": 48.71204502814259,
      "grad_norm": 0.6454340219497681,
      "learning_rate": 0.002828,
      "loss": 1.0426,
      "step": 649088
    },
    {
      "epoch": 48.71684803001876,
      "grad_norm": 0.6845133304595947,
      "learning_rate": 0.002828,
      "loss": 1.0474,
      "step": 649152
    },
    {
      "epoch": 48.721651031894936,
      "grad_norm": 0.762546718120575,
      "learning_rate": 0.002828,
      "loss": 1.0476,
      "step": 649216
    },
    {
      "epoch": 48.726454033771105,
      "grad_norm": 0.6332879662513733,
      "learning_rate": 0.002828,
      "loss": 1.0498,
      "step": 649280
    },
    {
      "epoch": 48.73125703564728,
      "grad_norm": 0.6962564587593079,
      "learning_rate": 0.002828,
      "loss": 1.0439,
      "step": 649344
    },
    {
      "epoch": 48.73606003752345,
      "grad_norm": 0.6551742553710938,
      "learning_rate": 0.002828,
      "loss": 1.0438,
      "step": 649408
    },
    {
      "epoch": 48.74086303939963,
      "grad_norm": 0.6957951784133911,
      "learning_rate": 0.002828,
      "loss": 1.0486,
      "step": 649472
    },
    {
      "epoch": 48.7456660412758,
      "grad_norm": 0.5979375839233398,
      "learning_rate": 0.002828,
      "loss": 1.0501,
      "step": 649536
    },
    {
      "epoch": 48.75046904315197,
      "grad_norm": 0.6078873872756958,
      "learning_rate": 0.002828,
      "loss": 1.0492,
      "step": 649600
    },
    {
      "epoch": 48.75527204502814,
      "grad_norm": 0.5456064343452454,
      "learning_rate": 0.002828,
      "loss": 1.0495,
      "step": 649664
    },
    {
      "epoch": 48.76007504690431,
      "grad_norm": 0.6340984106063843,
      "learning_rate": 0.002828,
      "loss": 1.0475,
      "step": 649728
    },
    {
      "epoch": 48.76487804878049,
      "grad_norm": 0.6775336265563965,
      "learning_rate": 0.002828,
      "loss": 1.0435,
      "step": 649792
    },
    {
      "epoch": 48.76968105065666,
      "grad_norm": 0.6024465560913086,
      "learning_rate": 0.002828,
      "loss": 1.0489,
      "step": 649856
    },
    {
      "epoch": 48.774484052532834,
      "grad_norm": 0.6876002550125122,
      "learning_rate": 0.002828,
      "loss": 1.0454,
      "step": 649920
    },
    {
      "epoch": 48.779287054409004,
      "grad_norm": 0.7455543875694275,
      "learning_rate": 0.002828,
      "loss": 1.0491,
      "step": 649984
    },
    {
      "epoch": 48.78409005628518,
      "grad_norm": 0.7086321711540222,
      "learning_rate": 0.002828,
      "loss": 1.0489,
      "step": 650048
    },
    {
      "epoch": 48.78889305816135,
      "grad_norm": 0.614976167678833,
      "learning_rate": 0.002828,
      "loss": 1.0462,
      "step": 650112
    },
    {
      "epoch": 48.793696060037526,
      "grad_norm": 0.6302147507667542,
      "learning_rate": 0.002828,
      "loss": 1.0449,
      "step": 650176
    },
    {
      "epoch": 48.798499061913695,
      "grad_norm": 0.6637048125267029,
      "learning_rate": 0.002828,
      "loss": 1.0494,
      "step": 650240
    },
    {
      "epoch": 48.80330206378987,
      "grad_norm": 0.6396692991256714,
      "learning_rate": 0.002828,
      "loss": 1.0497,
      "step": 650304
    },
    {
      "epoch": 48.80810506566604,
      "grad_norm": 0.5633668303489685,
      "learning_rate": 0.002828,
      "loss": 1.0413,
      "step": 650368
    },
    {
      "epoch": 48.81290806754222,
      "grad_norm": 0.5798957347869873,
      "learning_rate": 0.002828,
      "loss": 1.0498,
      "step": 650432
    },
    {
      "epoch": 48.81771106941839,
      "grad_norm": 0.6401728391647339,
      "learning_rate": 0.002828,
      "loss": 1.0499,
      "step": 650496
    },
    {
      "epoch": 48.822514071294556,
      "grad_norm": 0.6117183566093445,
      "learning_rate": 0.002828,
      "loss": 1.0484,
      "step": 650560
    },
    {
      "epoch": 48.82731707317073,
      "grad_norm": 0.5531995296478271,
      "learning_rate": 0.002828,
      "loss": 1.0527,
      "step": 650624
    },
    {
      "epoch": 48.8321200750469,
      "grad_norm": 0.4812498390674591,
      "learning_rate": 0.002828,
      "loss": 1.049,
      "step": 650688
    },
    {
      "epoch": 48.83692307692308,
      "grad_norm": 0.5269946455955505,
      "learning_rate": 0.002828,
      "loss": 1.0485,
      "step": 650752
    },
    {
      "epoch": 48.84172607879925,
      "grad_norm": 0.5963234305381775,
      "learning_rate": 0.002828,
      "loss": 1.0452,
      "step": 650816
    },
    {
      "epoch": 48.846529080675424,
      "grad_norm": 0.6707912087440491,
      "learning_rate": 0.002828,
      "loss": 1.0521,
      "step": 650880
    },
    {
      "epoch": 48.85133208255159,
      "grad_norm": 0.754862368106842,
      "learning_rate": 0.002828,
      "loss": 1.0533,
      "step": 650944
    },
    {
      "epoch": 48.85613508442777,
      "grad_norm": 0.7602565884590149,
      "learning_rate": 0.002828,
      "loss": 1.0457,
      "step": 651008
    },
    {
      "epoch": 48.86093808630394,
      "grad_norm": 0.5208454728126526,
      "learning_rate": 0.002828,
      "loss": 1.0479,
      "step": 651072
    },
    {
      "epoch": 48.865741088180116,
      "grad_norm": 0.706495463848114,
      "learning_rate": 0.002828,
      "loss": 1.0501,
      "step": 651136
    },
    {
      "epoch": 48.870544090056285,
      "grad_norm": 0.7677738070487976,
      "learning_rate": 0.002828,
      "loss": 1.0482,
      "step": 651200
    },
    {
      "epoch": 48.875347091932454,
      "grad_norm": 0.7193800806999207,
      "learning_rate": 0.002828,
      "loss": 1.0475,
      "step": 651264
    },
    {
      "epoch": 48.88015009380863,
      "grad_norm": 0.5489779114723206,
      "learning_rate": 0.002828,
      "loss": 1.0489,
      "step": 651328
    },
    {
      "epoch": 48.8849530956848,
      "grad_norm": 0.5512523055076599,
      "learning_rate": 0.002828,
      "loss": 1.0491,
      "step": 651392
    },
    {
      "epoch": 48.889756097560976,
      "grad_norm": 0.7432524561882019,
      "learning_rate": 0.002828,
      "loss": 1.0496,
      "step": 651456
    },
    {
      "epoch": 48.894559099437146,
      "grad_norm": 0.6426169872283936,
      "learning_rate": 0.002828,
      "loss": 1.0565,
      "step": 651520
    },
    {
      "epoch": 48.89936210131332,
      "grad_norm": 0.5901021361351013,
      "learning_rate": 0.002828,
      "loss": 1.0486,
      "step": 651584
    },
    {
      "epoch": 48.90416510318949,
      "grad_norm": 0.7789266109466553,
      "learning_rate": 0.002828,
      "loss": 1.0471,
      "step": 651648
    },
    {
      "epoch": 48.90896810506567,
      "grad_norm": 0.7148862481117249,
      "learning_rate": 0.002828,
      "loss": 1.0497,
      "step": 651712
    },
    {
      "epoch": 48.91377110694184,
      "grad_norm": 0.6462157964706421,
      "learning_rate": 0.002828,
      "loss": 1.0537,
      "step": 651776
    },
    {
      "epoch": 48.918574108818014,
      "grad_norm": 0.6770671606063843,
      "learning_rate": 0.002828,
      "loss": 1.054,
      "step": 651840
    },
    {
      "epoch": 48.92337711069418,
      "grad_norm": 0.6750553250312805,
      "learning_rate": 0.002828,
      "loss": 1.0534,
      "step": 651904
    },
    {
      "epoch": 48.92818011257036,
      "grad_norm": 0.57192462682724,
      "learning_rate": 0.002828,
      "loss": 1.049,
      "step": 651968
    },
    {
      "epoch": 48.93298311444653,
      "grad_norm": 0.7276706099510193,
      "learning_rate": 0.002828,
      "loss": 1.0529,
      "step": 652032
    },
    {
      "epoch": 48.9377861163227,
      "grad_norm": 0.648541271686554,
      "learning_rate": 0.002828,
      "loss": 1.0511,
      "step": 652096
    },
    {
      "epoch": 48.942589118198875,
      "grad_norm": 0.6642624139785767,
      "learning_rate": 0.002828,
      "loss": 1.0544,
      "step": 652160
    },
    {
      "epoch": 48.947392120075044,
      "grad_norm": 0.755593478679657,
      "learning_rate": 0.002828,
      "loss": 1.0518,
      "step": 652224
    },
    {
      "epoch": 48.95219512195122,
      "grad_norm": 0.5870972871780396,
      "learning_rate": 0.002828,
      "loss": 1.0566,
      "step": 652288
    },
    {
      "epoch": 48.95699812382739,
      "grad_norm": 0.8194005489349365,
      "learning_rate": 0.002828,
      "loss": 1.0498,
      "step": 652352
    },
    {
      "epoch": 48.961801125703566,
      "grad_norm": 0.6303133964538574,
      "learning_rate": 0.002828,
      "loss": 1.054,
      "step": 652416
    },
    {
      "epoch": 48.966604127579735,
      "grad_norm": 0.5689576268196106,
      "learning_rate": 0.002828,
      "loss": 1.0464,
      "step": 652480
    },
    {
      "epoch": 48.97140712945591,
      "grad_norm": 0.7027059197425842,
      "learning_rate": 0.002828,
      "loss": 1.048,
      "step": 652544
    },
    {
      "epoch": 48.97621013133208,
      "grad_norm": 0.6509045362472534,
      "learning_rate": 0.002828,
      "loss": 1.0536,
      "step": 652608
    },
    {
      "epoch": 48.98101313320826,
      "grad_norm": 0.6776425242424011,
      "learning_rate": 0.002828,
      "loss": 1.0528,
      "step": 652672
    },
    {
      "epoch": 48.98581613508443,
      "grad_norm": 0.8358839750289917,
      "learning_rate": 0.002828,
      "loss": 1.0489,
      "step": 652736
    },
    {
      "epoch": 48.9906191369606,
      "grad_norm": 0.5223692655563354,
      "learning_rate": 0.002828,
      "loss": 1.0551,
      "step": 652800
    },
    {
      "epoch": 48.99542213883677,
      "grad_norm": 0.5597832798957825,
      "learning_rate": 0.002828,
      "loss": 1.0536,
      "step": 652864
    },
    {
      "epoch": 49.00022514071294,
      "grad_norm": 0.5572521686553955,
      "learning_rate": 0.002828,
      "loss": 1.0484,
      "step": 652928
    },
    {
      "epoch": 49.00502814258912,
      "grad_norm": 0.6252709031105042,
      "learning_rate": 0.002828,
      "loss": 1.0213,
      "step": 652992
    },
    {
      "epoch": 49.00983114446529,
      "grad_norm": 0.7065907716751099,
      "learning_rate": 0.002828,
      "loss": 1.0139,
      "step": 653056
    },
    {
      "epoch": 49.014634146341464,
      "grad_norm": 0.6138619184494019,
      "learning_rate": 0.002828,
      "loss": 1.0155,
      "step": 653120
    },
    {
      "epoch": 49.019437148217634,
      "grad_norm": 0.640450656414032,
      "learning_rate": 0.002828,
      "loss": 1.0183,
      "step": 653184
    },
    {
      "epoch": 49.02424015009381,
      "grad_norm": 0.7467250227928162,
      "learning_rate": 0.002828,
      "loss": 1.0183,
      "step": 653248
    },
    {
      "epoch": 49.02904315196998,
      "grad_norm": 0.5681378841400146,
      "learning_rate": 0.002828,
      "loss": 1.0222,
      "step": 653312
    },
    {
      "epoch": 49.033846153846156,
      "grad_norm": 0.5548416972160339,
      "learning_rate": 0.002828,
      "loss": 1.0175,
      "step": 653376
    },
    {
      "epoch": 49.038649155722325,
      "grad_norm": 0.5206427574157715,
      "learning_rate": 0.002828,
      "loss": 1.0199,
      "step": 653440
    },
    {
      "epoch": 49.0434521575985,
      "grad_norm": 0.5543241500854492,
      "learning_rate": 0.002828,
      "loss": 1.0229,
      "step": 653504
    },
    {
      "epoch": 49.04825515947467,
      "grad_norm": 0.701314389705658,
      "learning_rate": 0.002828,
      "loss": 1.0198,
      "step": 653568
    },
    {
      "epoch": 49.05305816135085,
      "grad_norm": 0.6811202764511108,
      "learning_rate": 0.002828,
      "loss": 1.0195,
      "step": 653632
    },
    {
      "epoch": 49.05786116322702,
      "grad_norm": 0.7191393375396729,
      "learning_rate": 0.002828,
      "loss": 1.0192,
      "step": 653696
    },
    {
      "epoch": 49.062664165103186,
      "grad_norm": 0.625001072883606,
      "learning_rate": 0.002828,
      "loss": 1.0283,
      "step": 653760
    },
    {
      "epoch": 49.06746716697936,
      "grad_norm": 0.6109411120414734,
      "learning_rate": 0.002828,
      "loss": 1.0214,
      "step": 653824
    },
    {
      "epoch": 49.07227016885553,
      "grad_norm": 0.562427282333374,
      "learning_rate": 0.002828,
      "loss": 1.0236,
      "step": 653888
    },
    {
      "epoch": 49.07707317073171,
      "grad_norm": 0.6617029905319214,
      "learning_rate": 0.002828,
      "loss": 1.0186,
      "step": 653952
    },
    {
      "epoch": 49.08187617260788,
      "grad_norm": 0.5700415968894958,
      "learning_rate": 0.002828,
      "loss": 1.0222,
      "step": 654016
    },
    {
      "epoch": 49.086679174484054,
      "grad_norm": 0.5865738391876221,
      "learning_rate": 0.002828,
      "loss": 1.0226,
      "step": 654080
    },
    {
      "epoch": 49.09148217636022,
      "grad_norm": 0.6235159635543823,
      "learning_rate": 0.002828,
      "loss": 1.0189,
      "step": 654144
    },
    {
      "epoch": 49.0962851782364,
      "grad_norm": 0.5553084015846252,
      "learning_rate": 0.002828,
      "loss": 1.0207,
      "step": 654208
    },
    {
      "epoch": 49.10108818011257,
      "grad_norm": 0.582048237323761,
      "learning_rate": 0.002828,
      "loss": 1.0174,
      "step": 654272
    },
    {
      "epoch": 49.105891181988746,
      "grad_norm": 0.6419191956520081,
      "learning_rate": 0.002828,
      "loss": 1.0255,
      "step": 654336
    },
    {
      "epoch": 49.110694183864915,
      "grad_norm": 0.512915313243866,
      "learning_rate": 0.002828,
      "loss": 1.0275,
      "step": 654400
    },
    {
      "epoch": 49.11549718574109,
      "grad_norm": 0.6591917872428894,
      "learning_rate": 0.002828,
      "loss": 1.0198,
      "step": 654464
    },
    {
      "epoch": 49.12030018761726,
      "grad_norm": 0.6231454014778137,
      "learning_rate": 0.002828,
      "loss": 1.0227,
      "step": 654528
    },
    {
      "epoch": 49.12510318949343,
      "grad_norm": 0.6315057873725891,
      "learning_rate": 0.002828,
      "loss": 1.0215,
      "step": 654592
    },
    {
      "epoch": 49.12990619136961,
      "grad_norm": 0.6685563325881958,
      "learning_rate": 0.002828,
      "loss": 1.0217,
      "step": 654656
    },
    {
      "epoch": 49.134709193245776,
      "grad_norm": 0.5746342539787292,
      "learning_rate": 0.002828,
      "loss": 1.0264,
      "step": 654720
    },
    {
      "epoch": 49.13951219512195,
      "grad_norm": 0.6998280882835388,
      "learning_rate": 0.002828,
      "loss": 1.0225,
      "step": 654784
    },
    {
      "epoch": 49.14431519699812,
      "grad_norm": 0.73170006275177,
      "learning_rate": 0.002828,
      "loss": 1.0283,
      "step": 654848
    },
    {
      "epoch": 49.1491181988743,
      "grad_norm": 0.7164236903190613,
      "learning_rate": 0.002828,
      "loss": 1.0225,
      "step": 654912
    },
    {
      "epoch": 49.15392120075047,
      "grad_norm": 0.7144173979759216,
      "learning_rate": 0.002828,
      "loss": 1.0273,
      "step": 654976
    },
    {
      "epoch": 49.158724202626644,
      "grad_norm": 0.4856126606464386,
      "learning_rate": 0.002828,
      "loss": 1.0245,
      "step": 655040
    },
    {
      "epoch": 49.16352720450281,
      "grad_norm": 0.6092809438705444,
      "learning_rate": 0.002828,
      "loss": 1.0273,
      "step": 655104
    },
    {
      "epoch": 49.16833020637899,
      "grad_norm": 0.6352282762527466,
      "learning_rate": 0.002828,
      "loss": 1.0259,
      "step": 655168
    },
    {
      "epoch": 49.17313320825516,
      "grad_norm": 0.5976561307907104,
      "learning_rate": 0.002828,
      "loss": 1.025,
      "step": 655232
    },
    {
      "epoch": 49.177936210131335,
      "grad_norm": 0.6315568685531616,
      "learning_rate": 0.002828,
      "loss": 1.0252,
      "step": 655296
    },
    {
      "epoch": 49.182739212007505,
      "grad_norm": 0.6927725076675415,
      "learning_rate": 0.002828,
      "loss": 1.0295,
      "step": 655360
    },
    {
      "epoch": 49.187542213883674,
      "grad_norm": 0.6839077472686768,
      "learning_rate": 0.002828,
      "loss": 1.0266,
      "step": 655424
    },
    {
      "epoch": 49.19234521575985,
      "grad_norm": 0.571959376335144,
      "learning_rate": 0.002828,
      "loss": 1.0282,
      "step": 655488
    },
    {
      "epoch": 49.19714821763602,
      "grad_norm": 0.553979754447937,
      "learning_rate": 0.002828,
      "loss": 1.0267,
      "step": 655552
    },
    {
      "epoch": 49.201951219512196,
      "grad_norm": 0.6836845278739929,
      "learning_rate": 0.002828,
      "loss": 1.0314,
      "step": 655616
    },
    {
      "epoch": 49.206754221388366,
      "grad_norm": 0.647217869758606,
      "learning_rate": 0.002828,
      "loss": 1.03,
      "step": 655680
    },
    {
      "epoch": 49.21155722326454,
      "grad_norm": 0.6496232748031616,
      "learning_rate": 0.002828,
      "loss": 1.0259,
      "step": 655744
    },
    {
      "epoch": 49.21636022514071,
      "grad_norm": 0.6329558491706848,
      "learning_rate": 0.002828,
      "loss": 1.0351,
      "step": 655808
    },
    {
      "epoch": 49.22116322701689,
      "grad_norm": 0.5167147517204285,
      "learning_rate": 0.002828,
      "loss": 1.0259,
      "step": 655872
    },
    {
      "epoch": 49.22596622889306,
      "grad_norm": 0.7184466123580933,
      "learning_rate": 0.002828,
      "loss": 1.029,
      "step": 655936
    },
    {
      "epoch": 49.23076923076923,
      "grad_norm": 0.7324831485748291,
      "learning_rate": 0.002828,
      "loss": 1.0298,
      "step": 656000
    },
    {
      "epoch": 49.2355722326454,
      "grad_norm": 0.6538679003715515,
      "learning_rate": 0.002828,
      "loss": 1.0287,
      "step": 656064
    },
    {
      "epoch": 49.24037523452158,
      "grad_norm": 0.6968247890472412,
      "learning_rate": 0.002828,
      "loss": 1.0251,
      "step": 656128
    },
    {
      "epoch": 49.24517823639775,
      "grad_norm": 0.5685263872146606,
      "learning_rate": 0.002828,
      "loss": 1.0242,
      "step": 656192
    },
    {
      "epoch": 49.24998123827392,
      "grad_norm": 0.6809595227241516,
      "learning_rate": 0.002828,
      "loss": 1.0306,
      "step": 656256
    },
    {
      "epoch": 49.254784240150094,
      "grad_norm": 0.6379225850105286,
      "learning_rate": 0.002828,
      "loss": 1.0288,
      "step": 656320
    },
    {
      "epoch": 49.259587242026264,
      "grad_norm": 0.7669009566307068,
      "learning_rate": 0.002828,
      "loss": 1.033,
      "step": 656384
    },
    {
      "epoch": 49.26439024390244,
      "grad_norm": 0.6783488392829895,
      "learning_rate": 0.002828,
      "loss": 1.0265,
      "step": 656448
    },
    {
      "epoch": 49.26919324577861,
      "grad_norm": 0.6145064830780029,
      "learning_rate": 0.002828,
      "loss": 1.031,
      "step": 656512
    },
    {
      "epoch": 49.273996247654786,
      "grad_norm": 0.6485490798950195,
      "learning_rate": 0.002828,
      "loss": 1.0337,
      "step": 656576
    },
    {
      "epoch": 49.278799249530955,
      "grad_norm": 0.636939525604248,
      "learning_rate": 0.002828,
      "loss": 1.0252,
      "step": 656640
    },
    {
      "epoch": 49.28360225140713,
      "grad_norm": 0.5966800451278687,
      "learning_rate": 0.002828,
      "loss": 1.0321,
      "step": 656704
    },
    {
      "epoch": 49.2884052532833,
      "grad_norm": 0.5692200660705566,
      "learning_rate": 0.002828,
      "loss": 1.0303,
      "step": 656768
    },
    {
      "epoch": 49.29320825515948,
      "grad_norm": 0.724468469619751,
      "learning_rate": 0.002828,
      "loss": 1.029,
      "step": 656832
    },
    {
      "epoch": 49.29801125703565,
      "grad_norm": 0.727816641330719,
      "learning_rate": 0.002828,
      "loss": 1.027,
      "step": 656896
    },
    {
      "epoch": 49.30281425891182,
      "grad_norm": 0.6119939684867859,
      "learning_rate": 0.002828,
      "loss": 1.0332,
      "step": 656960
    },
    {
      "epoch": 49.30761726078799,
      "grad_norm": 0.6606029272079468,
      "learning_rate": 0.002828,
      "loss": 1.0348,
      "step": 657024
    },
    {
      "epoch": 49.31242026266416,
      "grad_norm": 0.6363573670387268,
      "learning_rate": 0.002828,
      "loss": 1.0233,
      "step": 657088
    },
    {
      "epoch": 49.31722326454034,
      "grad_norm": 0.6781613230705261,
      "learning_rate": 0.002828,
      "loss": 1.0332,
      "step": 657152
    },
    {
      "epoch": 49.32202626641651,
      "grad_norm": 0.6983715295791626,
      "learning_rate": 0.002828,
      "loss": 1.0245,
      "step": 657216
    },
    {
      "epoch": 49.326829268292684,
      "grad_norm": 0.7369449138641357,
      "learning_rate": 0.002828,
      "loss": 1.0301,
      "step": 657280
    },
    {
      "epoch": 49.33163227016885,
      "grad_norm": 0.6598891615867615,
      "learning_rate": 0.002828,
      "loss": 1.0271,
      "step": 657344
    },
    {
      "epoch": 49.33643527204503,
      "grad_norm": 0.8296746015548706,
      "learning_rate": 0.002828,
      "loss": 1.0339,
      "step": 657408
    },
    {
      "epoch": 49.3412382739212,
      "grad_norm": 0.7491599917411804,
      "learning_rate": 0.002828,
      "loss": 1.0335,
      "step": 657472
    },
    {
      "epoch": 49.346041275797376,
      "grad_norm": 0.7481504082679749,
      "learning_rate": 0.002828,
      "loss": 1.0337,
      "step": 657536
    },
    {
      "epoch": 49.350844277673545,
      "grad_norm": 0.7235509157180786,
      "learning_rate": 0.002828,
      "loss": 1.0284,
      "step": 657600
    },
    {
      "epoch": 49.35564727954972,
      "grad_norm": 0.6248297095298767,
      "learning_rate": 0.002828,
      "loss": 1.0358,
      "step": 657664
    },
    {
      "epoch": 49.36045028142589,
      "grad_norm": 0.717667818069458,
      "learning_rate": 0.002828,
      "loss": 1.0325,
      "step": 657728
    },
    {
      "epoch": 49.36525328330207,
      "grad_norm": 0.6365737915039062,
      "learning_rate": 0.002828,
      "loss": 1.0325,
      "step": 657792
    },
    {
      "epoch": 49.37005628517824,
      "grad_norm": 0.6971416473388672,
      "learning_rate": 0.002828,
      "loss": 1.0308,
      "step": 657856
    },
    {
      "epoch": 49.374859287054406,
      "grad_norm": 0.6160202622413635,
      "learning_rate": 0.002828,
      "loss": 1.0295,
      "step": 657920
    },
    {
      "epoch": 49.37966228893058,
      "grad_norm": 0.5844454765319824,
      "learning_rate": 0.002828,
      "loss": 1.0349,
      "step": 657984
    },
    {
      "epoch": 49.38446529080675,
      "grad_norm": 0.7336181402206421,
      "learning_rate": 0.002828,
      "loss": 1.035,
      "step": 658048
    },
    {
      "epoch": 49.38926829268293,
      "grad_norm": 0.6621116399765015,
      "learning_rate": 0.002828,
      "loss": 1.0389,
      "step": 658112
    },
    {
      "epoch": 49.3940712945591,
      "grad_norm": 0.7765074372291565,
      "learning_rate": 0.002828,
      "loss": 1.0363,
      "step": 658176
    },
    {
      "epoch": 49.398874296435274,
      "grad_norm": 0.6697800159454346,
      "learning_rate": 0.002828,
      "loss": 1.0296,
      "step": 658240
    },
    {
      "epoch": 49.40367729831144,
      "grad_norm": 0.6838088035583496,
      "learning_rate": 0.002828,
      "loss": 1.0321,
      "step": 658304
    },
    {
      "epoch": 49.40848030018762,
      "grad_norm": 0.6083168387413025,
      "learning_rate": 0.002828,
      "loss": 1.0368,
      "step": 658368
    },
    {
      "epoch": 49.41328330206379,
      "grad_norm": 0.7235950231552124,
      "learning_rate": 0.002828,
      "loss": 1.0326,
      "step": 658432
    },
    {
      "epoch": 49.418086303939965,
      "grad_norm": 0.6048862934112549,
      "learning_rate": 0.002828,
      "loss": 1.036,
      "step": 658496
    },
    {
      "epoch": 49.422889305816135,
      "grad_norm": 0.6545407176017761,
      "learning_rate": 0.002828,
      "loss": 1.0306,
      "step": 658560
    },
    {
      "epoch": 49.42769230769231,
      "grad_norm": 0.6188938617706299,
      "learning_rate": 0.002828,
      "loss": 1.0363,
      "step": 658624
    },
    {
      "epoch": 49.43249530956848,
      "grad_norm": 0.5984054207801819,
      "learning_rate": 0.002828,
      "loss": 1.0326,
      "step": 658688
    },
    {
      "epoch": 49.43729831144465,
      "grad_norm": 0.585507333278656,
      "learning_rate": 0.002828,
      "loss": 1.0357,
      "step": 658752
    },
    {
      "epoch": 49.442101313320826,
      "grad_norm": 0.6248952150344849,
      "learning_rate": 0.002828,
      "loss": 1.0356,
      "step": 658816
    },
    {
      "epoch": 49.446904315196996,
      "grad_norm": 0.5708743333816528,
      "learning_rate": 0.002828,
      "loss": 1.035,
      "step": 658880
    },
    {
      "epoch": 49.45170731707317,
      "grad_norm": 0.7817354202270508,
      "learning_rate": 0.002828,
      "loss": 1.0371,
      "step": 658944
    },
    {
      "epoch": 49.45651031894934,
      "grad_norm": 0.6046018600463867,
      "learning_rate": 0.002828,
      "loss": 1.0395,
      "step": 659008
    },
    {
      "epoch": 49.46131332082552,
      "grad_norm": 0.7188746333122253,
      "learning_rate": 0.002828,
      "loss": 1.0356,
      "step": 659072
    },
    {
      "epoch": 49.46611632270169,
      "grad_norm": 0.5960226058959961,
      "learning_rate": 0.002828,
      "loss": 1.0383,
      "step": 659136
    },
    {
      "epoch": 49.470919324577864,
      "grad_norm": 0.6067889928817749,
      "learning_rate": 0.002828,
      "loss": 1.0337,
      "step": 659200
    },
    {
      "epoch": 49.47572232645403,
      "grad_norm": 0.5953211188316345,
      "learning_rate": 0.002828,
      "loss": 1.0352,
      "step": 659264
    },
    {
      "epoch": 49.48052532833021,
      "grad_norm": 0.6781429052352905,
      "learning_rate": 0.002828,
      "loss": 1.0365,
      "step": 659328
    },
    {
      "epoch": 49.48532833020638,
      "grad_norm": 0.765910804271698,
      "learning_rate": 0.002828,
      "loss": 1.0387,
      "step": 659392
    },
    {
      "epoch": 49.490131332082555,
      "grad_norm": 0.6305378675460815,
      "learning_rate": 0.002828,
      "loss": 1.0318,
      "step": 659456
    },
    {
      "epoch": 49.494934333958724,
      "grad_norm": 0.6369653344154358,
      "learning_rate": 0.002828,
      "loss": 1.0384,
      "step": 659520
    },
    {
      "epoch": 49.499737335834894,
      "grad_norm": 0.7140512466430664,
      "learning_rate": 0.002828,
      "loss": 1.0365,
      "step": 659584
    },
    {
      "epoch": 49.50454033771107,
      "grad_norm": 0.640666127204895,
      "learning_rate": 0.002828,
      "loss": 1.0358,
      "step": 659648
    },
    {
      "epoch": 49.50934333958724,
      "grad_norm": 0.601139485836029,
      "learning_rate": 0.002828,
      "loss": 1.0361,
      "step": 659712
    },
    {
      "epoch": 49.514146341463416,
      "grad_norm": 0.5523284673690796,
      "learning_rate": 0.002828,
      "loss": 1.0383,
      "step": 659776
    },
    {
      "epoch": 49.518949343339585,
      "grad_norm": 0.6139888763427734,
      "learning_rate": 0.002828,
      "loss": 1.0397,
      "step": 659840
    },
    {
      "epoch": 49.52375234521576,
      "grad_norm": 0.5877168774604797,
      "learning_rate": 0.002828,
      "loss": 1.035,
      "step": 659904
    },
    {
      "epoch": 49.52855534709193,
      "grad_norm": 0.7172452807426453,
      "learning_rate": 0.002828,
      "loss": 1.0365,
      "step": 659968
    },
    {
      "epoch": 49.53335834896811,
      "grad_norm": 0.6976636052131653,
      "learning_rate": 0.002828,
      "loss": 1.0322,
      "step": 660032
    },
    {
      "epoch": 49.53816135084428,
      "grad_norm": 0.6075940728187561,
      "learning_rate": 0.002828,
      "loss": 1.0372,
      "step": 660096
    },
    {
      "epoch": 49.54296435272045,
      "grad_norm": 0.7370678782463074,
      "learning_rate": 0.002828,
      "loss": 1.034,
      "step": 660160
    },
    {
      "epoch": 49.54776735459662,
      "grad_norm": 0.6457955241203308,
      "learning_rate": 0.002828,
      "loss": 1.0433,
      "step": 660224
    },
    {
      "epoch": 49.5525703564728,
      "grad_norm": 0.6360769867897034,
      "learning_rate": 0.002828,
      "loss": 1.0416,
      "step": 660288
    },
    {
      "epoch": 49.55737335834897,
      "grad_norm": 0.5721134543418884,
      "learning_rate": 0.002828,
      "loss": 1.0366,
      "step": 660352
    },
    {
      "epoch": 49.56217636022514,
      "grad_norm": 0.5355108976364136,
      "learning_rate": 0.002828,
      "loss": 1.0366,
      "step": 660416
    },
    {
      "epoch": 49.566979362101314,
      "grad_norm": 0.6029777526855469,
      "learning_rate": 0.002828,
      "loss": 1.037,
      "step": 660480
    },
    {
      "epoch": 49.57178236397748,
      "grad_norm": 0.6070830821990967,
      "learning_rate": 0.002828,
      "loss": 1.0351,
      "step": 660544
    },
    {
      "epoch": 49.57658536585366,
      "grad_norm": 0.6995278596878052,
      "learning_rate": 0.002828,
      "loss": 1.0385,
      "step": 660608
    },
    {
      "epoch": 49.58138836772983,
      "grad_norm": 0.6078864932060242,
      "learning_rate": 0.002828,
      "loss": 1.0363,
      "step": 660672
    },
    {
      "epoch": 49.586191369606006,
      "grad_norm": 0.6785514950752258,
      "learning_rate": 0.002828,
      "loss": 1.0382,
      "step": 660736
    },
    {
      "epoch": 49.590994371482175,
      "grad_norm": 0.6332456469535828,
      "learning_rate": 0.002828,
      "loss": 1.0358,
      "step": 660800
    },
    {
      "epoch": 49.59579737335835,
      "grad_norm": 0.638705849647522,
      "learning_rate": 0.002828,
      "loss": 1.0395,
      "step": 660864
    },
    {
      "epoch": 49.60060037523452,
      "grad_norm": 0.6583680510520935,
      "learning_rate": 0.002828,
      "loss": 1.0365,
      "step": 660928
    },
    {
      "epoch": 49.6054033771107,
      "grad_norm": 0.6483926773071289,
      "learning_rate": 0.002828,
      "loss": 1.0354,
      "step": 660992
    },
    {
      "epoch": 49.61020637898687,
      "grad_norm": 0.6064408421516418,
      "learning_rate": 0.002828,
      "loss": 1.0344,
      "step": 661056
    },
    {
      "epoch": 49.615009380863036,
      "grad_norm": 0.7157003283500671,
      "learning_rate": 0.002828,
      "loss": 1.0425,
      "step": 661120
    },
    {
      "epoch": 49.61981238273921,
      "grad_norm": 0.5999187231063843,
      "learning_rate": 0.002828,
      "loss": 1.0358,
      "step": 661184
    },
    {
      "epoch": 49.62461538461538,
      "grad_norm": 0.5730012655258179,
      "learning_rate": 0.002828,
      "loss": 1.037,
      "step": 661248
    },
    {
      "epoch": 49.62941838649156,
      "grad_norm": 0.6842864155769348,
      "learning_rate": 0.002828,
      "loss": 1.0373,
      "step": 661312
    },
    {
      "epoch": 49.63422138836773,
      "grad_norm": 0.6744268536567688,
      "learning_rate": 0.002828,
      "loss": 1.0338,
      "step": 661376
    },
    {
      "epoch": 49.639024390243904,
      "grad_norm": 0.6678750514984131,
      "learning_rate": 0.002828,
      "loss": 1.0405,
      "step": 661440
    },
    {
      "epoch": 49.64382739212007,
      "grad_norm": 0.5240787267684937,
      "learning_rate": 0.002828,
      "loss": 1.0375,
      "step": 661504
    },
    {
      "epoch": 49.64863039399625,
      "grad_norm": 0.6266018152236938,
      "learning_rate": 0.002828,
      "loss": 1.0392,
      "step": 661568
    },
    {
      "epoch": 49.65343339587242,
      "grad_norm": 0.6554743647575378,
      "learning_rate": 0.002828,
      "loss": 1.0417,
      "step": 661632
    },
    {
      "epoch": 49.658236397748595,
      "grad_norm": 0.5542384386062622,
      "learning_rate": 0.002828,
      "loss": 1.0337,
      "step": 661696
    },
    {
      "epoch": 49.663039399624765,
      "grad_norm": 0.6251294612884521,
      "learning_rate": 0.002828,
      "loss": 1.0389,
      "step": 661760
    },
    {
      "epoch": 49.66784240150094,
      "grad_norm": 0.5415288805961609,
      "learning_rate": 0.002828,
      "loss": 1.0464,
      "step": 661824
    },
    {
      "epoch": 49.67264540337711,
      "grad_norm": 0.7915424108505249,
      "learning_rate": 0.002828,
      "loss": 1.0446,
      "step": 661888
    },
    {
      "epoch": 49.67744840525328,
      "grad_norm": 0.5885140299797058,
      "learning_rate": 0.002828,
      "loss": 1.0421,
      "step": 661952
    },
    {
      "epoch": 49.682251407129456,
      "grad_norm": 0.6221455931663513,
      "learning_rate": 0.002828,
      "loss": 1.0386,
      "step": 662016
    },
    {
      "epoch": 49.687054409005626,
      "grad_norm": 0.8090603947639465,
      "learning_rate": 0.002828,
      "loss": 1.0413,
      "step": 662080
    },
    {
      "epoch": 49.6918574108818,
      "grad_norm": 0.6631290912628174,
      "learning_rate": 0.002828,
      "loss": 1.0358,
      "step": 662144
    },
    {
      "epoch": 49.69666041275797,
      "grad_norm": 0.5944517254829407,
      "learning_rate": 0.002828,
      "loss": 1.0397,
      "step": 662208
    },
    {
      "epoch": 49.70146341463415,
      "grad_norm": 0.560357391834259,
      "learning_rate": 0.002828,
      "loss": 1.0444,
      "step": 662272
    },
    {
      "epoch": 49.70626641651032,
      "grad_norm": 0.5998170971870422,
      "learning_rate": 0.002828,
      "loss": 1.0386,
      "step": 662336
    },
    {
      "epoch": 49.711069418386494,
      "grad_norm": 0.5775870084762573,
      "learning_rate": 0.002828,
      "loss": 1.0447,
      "step": 662400
    },
    {
      "epoch": 49.71587242026266,
      "grad_norm": 0.6063507199287415,
      "learning_rate": 0.002828,
      "loss": 1.0406,
      "step": 662464
    },
    {
      "epoch": 49.72067542213884,
      "grad_norm": 0.624538779258728,
      "learning_rate": 0.002828,
      "loss": 1.0397,
      "step": 662528
    },
    {
      "epoch": 49.72547842401501,
      "grad_norm": 0.6854320168495178,
      "learning_rate": 0.002828,
      "loss": 1.0418,
      "step": 662592
    },
    {
      "epoch": 49.730281425891185,
      "grad_norm": 0.7011392116546631,
      "learning_rate": 0.002828,
      "loss": 1.0439,
      "step": 662656
    },
    {
      "epoch": 49.735084427767354,
      "grad_norm": 0.6861682534217834,
      "learning_rate": 0.002828,
      "loss": 1.0411,
      "step": 662720
    },
    {
      "epoch": 49.739887429643524,
      "grad_norm": 0.5691295266151428,
      "learning_rate": 0.002828,
      "loss": 1.0449,
      "step": 662784
    },
    {
      "epoch": 49.7446904315197,
      "grad_norm": 0.6943109631538391,
      "learning_rate": 0.002828,
      "loss": 1.0402,
      "step": 662848
    },
    {
      "epoch": 49.74949343339587,
      "grad_norm": 0.6759951710700989,
      "learning_rate": 0.002828,
      "loss": 1.045,
      "step": 662912
    },
    {
      "epoch": 49.754296435272046,
      "grad_norm": 0.5847229361534119,
      "learning_rate": 0.002828,
      "loss": 1.0439,
      "step": 662976
    },
    {
      "epoch": 49.759099437148215,
      "grad_norm": 0.6311315894126892,
      "learning_rate": 0.002828,
      "loss": 1.0403,
      "step": 663040
    },
    {
      "epoch": 49.76390243902439,
      "grad_norm": 0.5579125285148621,
      "learning_rate": 0.002828,
      "loss": 1.0427,
      "step": 663104
    },
    {
      "epoch": 49.76870544090056,
      "grad_norm": 0.6787163019180298,
      "learning_rate": 0.002828,
      "loss": 1.0425,
      "step": 663168
    },
    {
      "epoch": 49.77350844277674,
      "grad_norm": 0.7048472166061401,
      "learning_rate": 0.002828,
      "loss": 1.0473,
      "step": 663232
    },
    {
      "epoch": 49.77831144465291,
      "grad_norm": 0.5743940472602844,
      "learning_rate": 0.002828,
      "loss": 1.0452,
      "step": 663296
    },
    {
      "epoch": 49.78311444652908,
      "grad_norm": 0.6513806581497192,
      "learning_rate": 0.002828,
      "loss": 1.0351,
      "step": 663360
    },
    {
      "epoch": 49.78791744840525,
      "grad_norm": 0.6481301188468933,
      "learning_rate": 0.002828,
      "loss": 1.0443,
      "step": 663424
    },
    {
      "epoch": 49.79272045028143,
      "grad_norm": 0.6559600234031677,
      "learning_rate": 0.002828,
      "loss": 1.0474,
      "step": 663488
    },
    {
      "epoch": 49.7975234521576,
      "grad_norm": 0.571797251701355,
      "learning_rate": 0.002828,
      "loss": 1.0481,
      "step": 663552
    },
    {
      "epoch": 49.80232645403377,
      "grad_norm": 0.6878950595855713,
      "learning_rate": 0.002828,
      "loss": 1.0432,
      "step": 663616
    },
    {
      "epoch": 49.807129455909944,
      "grad_norm": 0.5944818258285522,
      "learning_rate": 0.002828,
      "loss": 1.0453,
      "step": 663680
    },
    {
      "epoch": 49.811932457786114,
      "grad_norm": 0.7458009123802185,
      "learning_rate": 0.002828,
      "loss": 1.0421,
      "step": 663744
    },
    {
      "epoch": 49.81673545966229,
      "grad_norm": 0.6427059769630432,
      "learning_rate": 0.002828,
      "loss": 1.0408,
      "step": 663808
    },
    {
      "epoch": 49.82153846153846,
      "grad_norm": 0.5951507091522217,
      "learning_rate": 0.002828,
      "loss": 1.0432,
      "step": 663872
    },
    {
      "epoch": 49.826341463414636,
      "grad_norm": 0.6637440919876099,
      "learning_rate": 0.002828,
      "loss": 1.0419,
      "step": 663936
    },
    {
      "epoch": 49.831144465290805,
      "grad_norm": 0.6890386343002319,
      "learning_rate": 0.002828,
      "loss": 1.0393,
      "step": 664000
    },
    {
      "epoch": 49.83594746716698,
      "grad_norm": 0.5513873100280762,
      "learning_rate": 0.002828,
      "loss": 1.0435,
      "step": 664064
    },
    {
      "epoch": 49.84075046904315,
      "grad_norm": 0.5582188963890076,
      "learning_rate": 0.002828,
      "loss": 1.0426,
      "step": 664128
    },
    {
      "epoch": 49.84555347091933,
      "grad_norm": 0.7070587873458862,
      "learning_rate": 0.002828,
      "loss": 1.0469,
      "step": 664192
    },
    {
      "epoch": 49.8503564727955,
      "grad_norm": 0.6251951456069946,
      "learning_rate": 0.002828,
      "loss": 1.0463,
      "step": 664256
    },
    {
      "epoch": 49.85515947467167,
      "grad_norm": 0.7980432510375977,
      "learning_rate": 0.002828,
      "loss": 1.046,
      "step": 664320
    },
    {
      "epoch": 49.85996247654784,
      "grad_norm": 0.5912220478057861,
      "learning_rate": 0.002828,
      "loss": 1.0438,
      "step": 664384
    },
    {
      "epoch": 49.86476547842401,
      "grad_norm": 0.651661217212677,
      "learning_rate": 0.002828,
      "loss": 1.0367,
      "step": 664448
    },
    {
      "epoch": 49.86956848030019,
      "grad_norm": 0.5960105657577515,
      "learning_rate": 0.002828,
      "loss": 1.0428,
      "step": 664512
    },
    {
      "epoch": 49.87437148217636,
      "grad_norm": 0.6765246987342834,
      "learning_rate": 0.002828,
      "loss": 1.043,
      "step": 664576
    },
    {
      "epoch": 49.879174484052534,
      "grad_norm": 0.6367417573928833,
      "learning_rate": 0.002828,
      "loss": 1.0459,
      "step": 664640
    },
    {
      "epoch": 49.8839774859287,
      "grad_norm": 0.6240684986114502,
      "learning_rate": 0.002828,
      "loss": 1.0473,
      "step": 664704
    },
    {
      "epoch": 49.88878048780488,
      "grad_norm": 0.5741950869560242,
      "learning_rate": 0.002828,
      "loss": 1.044,
      "step": 664768
    },
    {
      "epoch": 49.89358348968105,
      "grad_norm": 0.6683483123779297,
      "learning_rate": 0.002828,
      "loss": 1.0451,
      "step": 664832
    },
    {
      "epoch": 49.898386491557225,
      "grad_norm": 0.633254885673523,
      "learning_rate": 0.002828,
      "loss": 1.0447,
      "step": 664896
    },
    {
      "epoch": 49.903189493433395,
      "grad_norm": 0.7444667220115662,
      "learning_rate": 0.002828,
      "loss": 1.0422,
      "step": 664960
    },
    {
      "epoch": 49.90799249530957,
      "grad_norm": 0.5886372923851013,
      "learning_rate": 0.002828,
      "loss": 1.0428,
      "step": 665024
    },
    {
      "epoch": 49.91279549718574,
      "grad_norm": 0.595720112323761,
      "learning_rate": 0.002828,
      "loss": 1.0499,
      "step": 665088
    },
    {
      "epoch": 49.91759849906192,
      "grad_norm": 0.6207606792449951,
      "learning_rate": 0.002828,
      "loss": 1.0417,
      "step": 665152
    },
    {
      "epoch": 49.922401500938086,
      "grad_norm": 0.5864999890327454,
      "learning_rate": 0.002828,
      "loss": 1.0456,
      "step": 665216
    },
    {
      "epoch": 49.927204502814256,
      "grad_norm": 0.7752425670623779,
      "learning_rate": 0.002828,
      "loss": 1.0438,
      "step": 665280
    },
    {
      "epoch": 49.93200750469043,
      "grad_norm": 0.552574634552002,
      "learning_rate": 0.002828,
      "loss": 1.0461,
      "step": 665344
    },
    {
      "epoch": 49.9368105065666,
      "grad_norm": 0.576705813407898,
      "learning_rate": 0.002828,
      "loss": 1.0453,
      "step": 665408
    },
    {
      "epoch": 49.94161350844278,
      "grad_norm": 0.6199607849121094,
      "learning_rate": 0.002828,
      "loss": 1.049,
      "step": 665472
    },
    {
      "epoch": 49.94641651031895,
      "grad_norm": 0.6204496622085571,
      "learning_rate": 0.002828,
      "loss": 1.044,
      "step": 665536
    },
    {
      "epoch": 49.951219512195124,
      "grad_norm": 0.5873743891716003,
      "learning_rate": 0.002828,
      "loss": 1.043,
      "step": 665600
    },
    {
      "epoch": 49.95602251407129,
      "grad_norm": 0.8199127316474915,
      "learning_rate": 0.002828,
      "loss": 1.0496,
      "step": 665664
    },
    {
      "epoch": 49.96082551594747,
      "grad_norm": 0.5267244577407837,
      "learning_rate": 0.002828,
      "loss": 1.0469,
      "step": 665728
    },
    {
      "epoch": 49.96562851782364,
      "grad_norm": 0.6444991827011108,
      "learning_rate": 0.002828,
      "loss": 1.0432,
      "step": 665792
    },
    {
      "epoch": 49.970431519699815,
      "grad_norm": 0.7863524556159973,
      "learning_rate": 0.002828,
      "loss": 1.0399,
      "step": 665856
    },
    {
      "epoch": 49.975234521575985,
      "grad_norm": 0.619107186794281,
      "learning_rate": 0.002828,
      "loss": 1.0424,
      "step": 665920
    },
    {
      "epoch": 49.98003752345216,
      "grad_norm": 0.6243489384651184,
      "learning_rate": 0.002828,
      "loss": 1.0404,
      "step": 665984
    },
    {
      "epoch": 49.98484052532833,
      "grad_norm": 0.6714879870414734,
      "learning_rate": 0.002828,
      "loss": 1.0467,
      "step": 666048
    },
    {
      "epoch": 49.9896435272045,
      "grad_norm": 0.6690127849578857,
      "learning_rate": 0.002828,
      "loss": 1.0477,
      "step": 666112
    },
    {
      "epoch": 49.994446529080676,
      "grad_norm": 0.6523921489715576,
      "learning_rate": 0.002828,
      "loss": 1.0537,
      "step": 666176
    },
    {
      "epoch": 49.999249530956845,
      "grad_norm": 0.7129687070846558,
      "learning_rate": 0.002828,
      "loss": 1.0488,
      "step": 666240
    }
  ],
  "logging_steps": 64,
  "max_steps": 666250,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 50,
  "save_steps": 320,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.313475466887168e+19,
  "train_batch_size": 200,
  "trial_name": null,
  "trial_params": null
}