{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 785,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.032,
      "grad_norm": 5.453780786232042,
      "learning_rate": 2.0253164556962026e-06,
      "loss": 0.8032,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.20371782779693604,
      "step": 5,
      "valid_targets_mean": 6761.0,
      "valid_targets_min": 1767
    },
    {
      "epoch": 0.064,
      "grad_norm": 3.792389792983448,
      "learning_rate": 4.556962025316456e-06,
      "loss": 0.7643,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1712636798620224,
      "step": 10,
      "valid_targets_mean": 4999.5,
      "valid_targets_min": 1304
    },
    {
      "epoch": 0.096,
      "grad_norm": 2.063032028483624,
      "learning_rate": 7.08860759493671e-06,
      "loss": 0.7318,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.17536824941635132,
      "step": 15,
      "valid_targets_mean": 5833.2,
      "valid_targets_min": 1211
    },
    {
      "epoch": 0.128,
      "grad_norm": 0.8871678160368911,
      "learning_rate": 9.620253164556963e-06,
      "loss": 0.6933,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.18351516127586365,
      "step": 20,
      "valid_targets_mean": 6014.7,
      "valid_targets_min": 3717
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7382530806327504,
      "learning_rate": 1.2151898734177216e-05,
      "loss": 0.6776,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1541610211133957,
      "step": 25,
      "valid_targets_mean": 5068.2,
      "valid_targets_min": 1347
    },
    {
      "epoch": 0.192,
      "grad_norm": 0.514129978379908,
      "learning_rate": 1.468354430379747e-05,
      "loss": 0.6547,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.15168946981430054,
      "step": 30,
      "valid_targets_mean": 5006.1,
      "valid_targets_min": 881
    },
    {
      "epoch": 0.224,
      "grad_norm": 0.4336197541753502,
      "learning_rate": 1.7215189873417723e-05,
      "loss": 0.6208,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.14679864048957825,
      "step": 35,
      "valid_targets_mean": 5579.1,
      "valid_targets_min": 1357
    },
    {
      "epoch": 0.256,
      "grad_norm": 0.36433161326142466,
      "learning_rate": 1.974683544303798e-05,
      "loss": 0.6069,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.15517079830169678,
      "step": 40,
      "valid_targets_mean": 5445.0,
      "valid_targets_min": 1474
    },
    {
      "epoch": 0.288,
      "grad_norm": 0.32965171941859056,
      "learning_rate": 2.2278481012658228e-05,
      "loss": 0.5728,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.14498282968997955,
      "step": 45,
      "valid_targets_mean": 6170.8,
      "valid_targets_min": 3922
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2872624196401524,
      "learning_rate": 2.481012658227848e-05,
      "loss": 0.5765,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1473766267299652,
      "step": 50,
      "valid_targets_mean": 6663.2,
      "valid_targets_min": 4765
    },
    {
      "epoch": 0.352,
      "grad_norm": 0.27100975287589185,
      "learning_rate": 2.7341772151898737e-05,
      "loss": 0.5616,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.14131703972816467,
      "step": 55,
      "valid_targets_mean": 5866.8,
      "valid_targets_min": 1425
    },
    {
      "epoch": 0.384,
      "grad_norm": 0.2637945073634897,
      "learning_rate": 2.987341772151899e-05,
      "loss": 0.5483,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.13468310236930847,
      "step": 60,
      "valid_targets_mean": 5623.7,
      "valid_targets_min": 1606
    },
    {
      "epoch": 0.416,
      "grad_norm": 0.2609484255153037,
      "learning_rate": 3.240506329113924e-05,
      "loss": 0.5464,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10717941075563431,
      "step": 65,
      "valid_targets_mean": 5325.4,
      "valid_targets_min": 1780
    },
    {
      "epoch": 0.448,
      "grad_norm": 0.2475619693697882,
      "learning_rate": 3.49367088607595e-05,
      "loss": 0.5473,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.14007271826267242,
      "step": 70,
      "valid_targets_mean": 5930.2,
      "valid_targets_min": 1506
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.24464093539865395,
      "learning_rate": 3.746835443037975e-05,
      "loss": 0.5233,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12224975228309631,
      "step": 75,
      "valid_targets_mean": 5512.3,
      "valid_targets_min": 2940
    },
    {
      "epoch": 0.512,
      "grad_norm": 0.22999667928124048,
      "learning_rate": 4e-05,
      "loss": 0.5253,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12176463007926941,
      "step": 80,
      "valid_targets_mean": 5304.3,
      "valid_targets_min": 3470
    },
    {
      "epoch": 0.544,
      "grad_norm": 0.23934636533674483,
      "learning_rate": 3.999504991751045e-05,
      "loss": 0.5338,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.13291212916374207,
      "step": 85,
      "valid_targets_mean": 5400.6,
      "valid_targets_min": 1624
    },
    {
      "epoch": 0.576,
      "grad_norm": 0.22974570019649057,
      "learning_rate": 3.9980202120373464e-05,
      "loss": 0.5139,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12326923757791519,
      "step": 90,
      "valid_targets_mean": 4998.4,
      "valid_targets_min": 1247
    },
    {
      "epoch": 0.608,
      "grad_norm": 0.2926220288463877,
      "learning_rate": 3.995546395837111e-05,
      "loss": 0.5038,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1356104016304016,
      "step": 95,
      "valid_targets_mean": 5303.9,
      "valid_targets_min": 1326
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.23427791767385714,
      "learning_rate": 3.992084767709763e-05,
      "loss": 0.5066,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1405172497034073,
      "step": 100,
      "valid_targets_mean": 5991.0,
      "valid_targets_min": 3481
    },
    {
      "epoch": 0.672,
      "grad_norm": 0.28437672036074757,
      "learning_rate": 3.987637041189781e-05,
      "loss": 0.512,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1347738653421402,
      "step": 105,
      "valid_targets_mean": 6278.3,
      "valid_targets_min": 2344
    },
    {
      "epoch": 0.704,
      "grad_norm": 0.24924033321191602,
      "learning_rate": 3.982205417938482e-05,
      "loss": 0.4975,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11735053360462189,
      "step": 110,
      "valid_targets_mean": 5851.7,
      "valid_targets_min": 1129
    },
    {
      "epoch": 0.736,
      "grad_norm": 0.2507030670655199,
      "learning_rate": 3.975792586654179e-05,
      "loss": 0.5084,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12411265820264816,
      "step": 115,
      "valid_targets_mean": 5674.4,
      "valid_targets_min": 3051
    },
    {
      "epoch": 0.768,
      "grad_norm": 0.2500377635060955,
      "learning_rate": 3.968401721741259e-05,
      "loss": 0.5022,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12196642905473709,
      "step": 120,
      "valid_targets_mean": 6003.1,
      "valid_targets_min": 1515
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.24135329786073206,
      "learning_rate": 3.960036481738819e-05,
      "loss": 0.5008,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12246815860271454,
      "step": 125,
      "valid_targets_mean": 5971.2,
      "valid_targets_min": 2866
    },
    {
      "epoch": 0.832,
      "grad_norm": 0.22519547761877387,
      "learning_rate": 3.950701007509667e-05,
      "loss": 0.4981,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09827093780040741,
      "step": 130,
      "valid_targets_mean": 5276.2,
      "valid_targets_min": 1312
    },
    {
      "epoch": 0.864,
      "grad_norm": 0.24777405687491888,
      "learning_rate": 3.940399920190552e-05,
      "loss": 0.4924,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1192011833190918,
      "step": 135,
      "valid_targets_mean": 5829.4,
      "valid_targets_min": 1973
    },
    {
      "epoch": 0.896,
      "grad_norm": 0.2461001388229048,
      "learning_rate": 3.92913831890467e-05,
      "loss": 0.5026,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.127201110124588,
      "step": 140,
      "valid_targets_mean": 6177.5,
      "valid_targets_min": 3491
    },
    {
      "epoch": 0.928,
      "grad_norm": 0.26205526754490166,
      "learning_rate": 3.916921778237556e-05,
      "loss": 0.4921,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12378434836864471,
      "step": 145,
      "valid_targets_mean": 6577.8,
      "valid_targets_min": 2300
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2527022155127535,
      "learning_rate": 3.903756345477612e-05,
      "loss": 0.4917,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11821117997169495,
      "step": 150,
      "valid_targets_mean": 5569.9,
      "valid_targets_min": 1507
    },
    {
      "epoch": 0.992,
      "grad_norm": 0.24869143664799934,
      "learning_rate": 3.889648537622657e-05,
      "loss": 0.482,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11581720411777496,
      "step": 155,
      "valid_targets_mean": 5312.9,
      "valid_targets_min": 2593
    },
    {
      "epoch": 1.0192,
      "grad_norm": 0.23417657797639116,
      "learning_rate": 3.874605338153952e-05,
      "loss": 0.4875,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1262468695640564,
      "step": 160,
      "valid_targets_mean": 6192.8,
      "valid_targets_min": 4127
    },
    {
      "epoch": 1.0512,
      "grad_norm": 0.27974191809541077,
      "learning_rate": 3.8586341935793265e-05,
      "loss": 0.4818,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12143966555595398,
      "step": 165,
      "valid_targets_mean": 5954.1,
      "valid_targets_min": 1743
    },
    {
      "epoch": 1.0832,
      "grad_norm": 0.25435642517002544,
      "learning_rate": 3.841743009747089e-05,
      "loss": 0.4815,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.13246093690395355,
      "step": 170,
      "valid_targets_mean": 6993.4,
      "valid_targets_min": 3801
    },
    {
      "epoch": 1.1152,
      "grad_norm": 0.25881160063054937,
      "learning_rate": 3.8239401479325714e-05,
      "loss": 0.4693,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11434349417686462,
      "step": 175,
      "valid_targets_mean": 4978.4,
      "valid_targets_min": 1260
    },
    {
      "epoch": 1.1472,
      "grad_norm": 0.2278502886084013,
      "learning_rate": 3.8052344206992276e-05,
      "loss": 0.4787,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12233040481805801,
      "step": 180,
      "valid_targets_mean": 7108.7,
      "valid_targets_min": 2155
    },
    {
      "epoch": 1.1792,
      "grad_norm": 0.25565502704610965,
      "learning_rate": 3.7856350875363396e-05,
      "loss": 0.4801,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11371868848800659,
      "step": 185,
      "valid_targets_mean": 6057.6,
      "valid_targets_min": 3039
    },
    {
      "epoch": 1.2112,
      "grad_norm": 0.24238202719508706,
      "learning_rate": 3.765151850275497e-05,
      "loss": 0.475,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10759443044662476,
      "step": 190,
      "valid_targets_mean": 5236.0,
      "valid_targets_min": 1548
    },
    {
      "epoch": 1.2432,
      "grad_norm": 0.2707586320724312,
      "learning_rate": 3.7437948482881104e-05,
      "loss": 0.4838,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10973644256591797,
      "step": 195,
      "valid_targets_mean": 4837.2,
      "valid_targets_min": 1166
    },
    {
      "epoch": 1.2752,
      "grad_norm": 0.26143437324759183,
      "learning_rate": 3.721574653466336e-05,
      "loss": 0.4737,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.13594186305999756,
      "step": 200,
      "valid_targets_mean": 5692.9,
      "valid_targets_min": 4103
    },
    {
      "epoch": 1.3072,
      "grad_norm": 0.26530637558666487,
      "learning_rate": 3.698502264989903e-05,
      "loss": 0.474,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0969756618142128,
      "step": 205,
      "valid_targets_mean": 4845.5,
      "valid_targets_min": 2870
    },
    {
      "epoch": 1.3392,
      "grad_norm": 0.24008064698409048,
      "learning_rate": 3.674589103881432e-05,
      "loss": 0.4667,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12318138033151627,
      "step": 210,
      "valid_targets_mean": 5510.6,
      "valid_targets_min": 1680
    },
    {
      "epoch": 1.3712,
      "grad_norm": 0.25529145100152656,
      "learning_rate": 3.64984700735293e-05,
      "loss": 0.4761,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12145581841468811,
      "step": 215,
      "valid_targets_mean": 5900.8,
      "valid_targets_min": 2659
    },
    {
      "epoch": 1.4032,
      "grad_norm": 0.25794769045722116,
      "learning_rate": 3.624288222946273e-05,
      "loss": 0.484,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1101190522313118,
      "step": 220,
      "valid_targets_mean": 5558.0,
      "valid_targets_min": 1262
    },
    {
      "epoch": 1.4352,
      "grad_norm": 0.2782121058872405,
      "learning_rate": 3.597925402470578e-05,
      "loss": 0.4809,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.13891330361366272,
      "step": 225,
      "valid_targets_mean": 6163.1,
      "valid_targets_min": 2175
    },
    {
      "epoch": 1.4672,
      "grad_norm": 0.2511379752274597,
      "learning_rate": 3.570771595739445e-05,
      "loss": 0.4669,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1152762770652771,
      "step": 230,
      "valid_targets_mean": 6515.6,
      "valid_targets_min": 1578
    },
    {
      "epoch": 1.4992,
      "grad_norm": 0.27590481971133046,
      "learning_rate": 3.5428402441111964e-05,
      "loss": 0.4806,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12730714678764343,
      "step": 235,
      "valid_targets_mean": 5110.8,
      "valid_targets_min": 1657
    },
    {
      "epoch": 1.5312000000000001,
      "grad_norm": 0.254323226627503,
      "learning_rate": 3.5141451738352936e-05,
      "loss": 0.4659,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10027951002120972,
      "step": 240,
      "valid_targets_mean": 5534.1,
      "valid_targets_min": 891
    },
    {
      "epoch": 1.5632000000000001,
      "grad_norm": 0.2580174702832105,
      "learning_rate": 3.4847005892082266e-05,
      "loss": 0.4675,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11096903681755066,
      "step": 245,
      "valid_targets_mean": 5662.7,
      "valid_targets_min": 1644
    },
    {
      "epoch": 1.5952,
      "grad_norm": 0.30698403507646943,
      "learning_rate": 3.454521065542273e-05,
      "loss": 0.4859,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1288014054298401,
      "step": 250,
      "valid_targets_mean": 5168.7,
      "valid_targets_min": 1162
    },
    {
      "epoch": 1.6272,
      "grad_norm": 0.25805156829059184,
      "learning_rate": 3.423621541950597e-05,
      "loss": 0.464,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.129526287317276,
      "step": 255,
      "valid_targets_mean": 7147.4,
      "valid_targets_min": 3884
    },
    {
      "epoch": 1.6592,
      "grad_norm": 0.23616819648599033,
      "learning_rate": 3.3920173139522664e-05,
      "loss": 0.4673,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11364764720201492,
      "step": 260,
      "valid_targets_mean": 5991.1,
      "valid_targets_min": 1280
    },
    {
      "epoch": 1.6912,
      "grad_norm": 0.2810806381450421,
      "learning_rate": 3.35972402590084e-05,
      "loss": 0.471,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1247754693031311,
      "step": 265,
      "valid_targets_mean": 5614.1,
      "valid_targets_min": 1199
    },
    {
      "epoch": 1.7231999999999998,
      "grad_norm": 0.2529117867416003,
      "learning_rate": 3.326757663240291e-05,
      "loss": 0.4793,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1115766316652298,
      "step": 270,
      "valid_targets_mean": 5206.1,
      "valid_targets_min": 1946
    },
    {
      "epoch": 1.7551999999999999,
      "grad_norm": 0.24073937834464354,
      "learning_rate": 3.293134544592073e-05,
      "loss": 0.463,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12958145141601562,
      "step": 275,
      "valid_targets_mean": 6616.4,
      "valid_targets_min": 3579
    },
    {
      "epoch": 1.7872,
      "grad_norm": 0.23555835370029385,
      "learning_rate": 3.258871313677274e-05,
      "loss": 0.4677,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.13404905796051025,
      "step": 280,
      "valid_targets_mean": 7205.4,
      "valid_targets_min": 1356
    },
    {
      "epoch": 1.8192,
      "grad_norm": 0.2513641254982323,
      "learning_rate": 3.2239849310778316e-05,
      "loss": 0.4636,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1133008599281311,
      "step": 285,
      "valid_targets_mean": 5637.1,
      "valid_targets_min": 3779
    },
    {
      "epoch": 1.8512,
      "grad_norm": 0.22918433139423086,
      "learning_rate": 3.188492665840909e-05,
      "loss": 0.4649,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10168616473674774,
      "step": 290,
      "valid_targets_mean": 5500.5,
      "valid_targets_min": 1523
    },
    {
      "epoch": 1.8832,
      "grad_norm": 0.25315031376778907,
      "learning_rate": 3.1524120869305726e-05,
      "loss": 0.4538,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11145422607660294,
      "step": 295,
      "valid_targets_mean": 4931.9,
      "valid_targets_min": 1168
    },
    {
      "epoch": 1.9152,
      "grad_norm": 0.22105397485658976,
      "learning_rate": 3.11576105453101e-05,
      "loss": 0.4589,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11858774721622467,
      "step": 300,
      "valid_targets_mean": 6575.4,
      "valid_targets_min": 4043
    },
    {
      "epoch": 1.9472,
      "grad_norm": 0.2668714334524866,
      "learning_rate": 3.0785577112055916e-05,
      "loss": 0.467,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1110314130783081,
      "step": 305,
      "valid_targets_mean": 4943.4,
      "valid_targets_min": 3259
    },
    {
      "epoch": 1.9792,
      "grad_norm": 0.27142228891732906,
      "learning_rate": 3.040820472916153e-05,
      "loss": 0.474,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10838167369365692,
      "step": 310,
      "valid_targets_mean": 5198.4,
      "valid_targets_min": 1470
    },
    {
      "epoch": 2.0064,
      "grad_norm": 0.2567890914244406,
      "learning_rate": 3.002568019906939e-05,
      "loss": 0.463,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1107785701751709,
      "step": 315,
      "valid_targets_mean": 5847.9,
      "valid_targets_min": 2018
    },
    {
      "epoch": 2.0384,
      "grad_norm": 0.2523608373300584,
      "learning_rate": 2.963819287457733e-05,
      "loss": 0.4488,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10997898876667023,
      "step": 320,
      "valid_targets_mean": 5916.2,
      "valid_targets_min": 2413
    },
    {
      "epoch": 2.0704,
      "grad_norm": 0.23463415840591811,
      "learning_rate": 2.924593456510733e-05,
      "loss": 0.4674,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10450360178947449,
      "step": 325,
      "valid_targets_mean": 5486.4,
      "valid_targets_min": 1264
    },
    {
      "epoch": 2.1024,
      "grad_norm": 0.24843951325249042,
      "learning_rate": 2.8849099441758306e-05,
      "loss": 0.4558,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11403744667768478,
      "step": 330,
      "valid_targets_mean": 5439.9,
      "valid_targets_min": 1138
    },
    {
      "epoch": 2.1344,
      "grad_norm": 0.24610362358361257,
      "learning_rate": 2.844788394118979e-05,
      "loss": 0.453,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11807750910520554,
      "step": 335,
      "valid_targets_mean": 5861.1,
      "valid_targets_min": 1565
    },
    {
      "epoch": 2.1664,
      "grad_norm": 0.33040794254123873,
      "learning_rate": 2.8042486668384164e-05,
      "loss": 0.4545,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11304843425750732,
      "step": 340,
      "valid_targets_mean": 6610.1,
      "valid_targets_min": 2124
    },
    {
      "epoch": 2.1984,
      "grad_norm": 0.25162468452460407,
      "learning_rate": 2.7633108298335582e-05,
      "loss": 0.4554,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11378766596317291,
      "step": 345,
      "valid_targets_mean": 5402.4,
      "valid_targets_min": 1247
    },
    {
      "epoch": 2.2304,
      "grad_norm": 0.25754716013199486,
      "learning_rate": 2.721995147671416e-05,
      "loss": 0.4466,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10926654934883118,
      "step": 350,
      "valid_targets_mean": 6230.7,
      "valid_targets_min": 1159
    },
    {
      "epoch": 2.2624,
      "grad_norm": 0.25363357284595456,
      "learning_rate": 2.68032207195547e-05,
      "loss": 0.4655,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12160837650299072,
      "step": 355,
      "valid_targets_mean": 5901.7,
      "valid_targets_min": 1400
    },
    {
      "epoch": 2.2944,
      "grad_norm": 0.2371953068408656,
      "learning_rate": 2.6383122312019604e-05,
      "loss": 0.4593,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10440811514854431,
      "step": 360,
      "valid_targets_mean": 5618.4,
      "valid_targets_min": 2764
    },
    {
      "epoch": 2.3264,
      "grad_norm": 0.24670634376756886,
      "learning_rate": 2.595986420628597e-05,
      "loss": 0.4554,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11458228528499603,
      "step": 365,
      "valid_targets_mean": 5704.5,
      "valid_targets_min": 1938
    },
    {
      "epoch": 2.3584,
      "grad_norm": 0.23895194020127508,
      "learning_rate": 2.5533655918607573e-05,
      "loss": 0.4585,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10346005856990814,
      "step": 370,
      "valid_targets_mean": 5528.0,
      "valid_targets_min": 1275
    },
    {
      "epoch": 2.3904,
      "grad_norm": 0.2408178690827922,
      "learning_rate": 2.510470842560259e-05,
      "loss": 0.4519,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11147673428058624,
      "step": 375,
      "valid_targets_mean": 5869.6,
      "valid_targets_min": 1637
    },
    {
      "epoch": 2.4224,
      "grad_norm": 0.23255787756846094,
      "learning_rate": 2.467323405981841e-05,
      "loss": 0.4607,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10699373483657837,
      "step": 380,
      "valid_targets_mean": 5419.7,
      "valid_targets_min": 2956
    },
    {
      "epoch": 2.4544,
      "grad_norm": 0.24230598627352132,
      "learning_rate": 2.423944640462533e-05,
      "loss": 0.4633,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10441893339157104,
      "step": 385,
      "valid_targets_mean": 4641.1,
      "valid_targets_min": 1544
    },
    {
      "epoch": 2.4864,
      "grad_norm": 0.24609359447675086,
      "learning_rate": 2.3803560188490968e-05,
      "loss": 0.4507,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.13390594720840454,
      "step": 390,
      "valid_targets_mean": 6129.4,
      "valid_targets_min": 1506
    },
    {
      "epoch": 2.5183999999999997,
      "grad_norm": 0.2450309981216017,
      "learning_rate": 2.336579117868789e-05,
      "loss": 0.4505,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09820785373449326,
      "step": 395,
      "valid_targets_mean": 4850.6,
      "valid_targets_min": 1443
    },
    {
      "epoch": 2.5504,
      "grad_norm": 0.26353410163682456,
      "learning_rate": 2.292635607448711e-05,
      "loss": 0.4558,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12506261467933655,
      "step": 400,
      "valid_targets_mean": 6348.2,
      "valid_targets_min": 3938
    },
    {
      "epoch": 2.5824,
      "grad_norm": 0.2591867666837128,
      "learning_rate": 2.248547239989008e-05,
      "loss": 0.4644,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1142532080411911,
      "step": 405,
      "valid_targets_mean": 5473.3,
      "valid_targets_min": 3717
    },
    {
      "epoch": 2.6144,
      "grad_norm": 0.23939041830365473,
      "learning_rate": 2.204335839595255e-05,
      "loss": 0.4579,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.126010924577713,
      "step": 410,
      "valid_targets_mean": 7091.4,
      "valid_targets_min": 2590
    },
    {
      "epoch": 2.6464,
      "grad_norm": 0.27523073180915464,
      "learning_rate": 2.1600232912753452e-05,
      "loss": 0.4524,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12722241878509521,
      "step": 415,
      "valid_targets_mean": 5683.2,
      "valid_targets_min": 1399
    },
    {
      "epoch": 2.6784,
      "grad_norm": 0.2555993394486925,
      "learning_rate": 2.1156315301062293e-05,
      "loss": 0.4419,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10284954309463501,
      "step": 420,
      "valid_targets_mean": 5005.5,
      "valid_targets_min": 934
    },
    {
      "epoch": 2.7104,
      "grad_norm": 0.26731870473795194,
      "learning_rate": 2.0711825303758712e-05,
      "loss": 0.4433,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11506682634353638,
      "step": 425,
      "valid_targets_mean": 6046.0,
      "valid_targets_min": 1647
    },
    {
      "epoch": 2.7424,
      "grad_norm": 0.2407561884924848,
      "learning_rate": 2.0266982947057962e-05,
      "loss": 0.4559,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10325722396373749,
      "step": 430,
      "valid_targets_mean": 5949.6,
      "valid_targets_min": 2688
    },
    {
      "epoch": 2.7744,
      "grad_norm": 0.2336404714487898,
      "learning_rate": 1.9822008431596083e-05,
      "loss": 0.4522,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10586457699537277,
      "step": 435,
      "valid_targets_mean": 6859.5,
      "valid_targets_min": 4222
    },
    {
      "epoch": 2.8064,
      "grad_norm": 0.22137813612500373,
      "learning_rate": 1.937712202342881e-05,
      "loss": 0.4408,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12141789495944977,
      "step": 440,
      "valid_targets_mean": 6645.0,
      "valid_targets_min": 3372
    },
    {
      "epoch": 2.8384,
      "grad_norm": 0.2361975809490191,
      "learning_rate": 1.8932543944998037e-05,
      "loss": 0.4621,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11469966918230057,
      "step": 445,
      "valid_targets_mean": 6750.8,
      "valid_targets_min": 3870
    },
    {
      "epoch": 2.8704,
      "grad_norm": 0.2569963385739716,
      "learning_rate": 1.8488494266119877e-05,
      "loss": 0.4558,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10432308912277222,
      "step": 450,
      "valid_targets_mean": 5330.1,
      "valid_targets_min": 1132
    },
    {
      "epoch": 2.9024,
      "grad_norm": 0.23032242260685118,
      "learning_rate": 1.804519279504834e-05,
      "loss": 0.4526,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11718428134918213,
      "step": 455,
      "valid_targets_mean": 6316.1,
      "valid_targets_min": 881
    },
    {
      "epoch": 2.9344,
      "grad_norm": 0.23444798682474463,
      "learning_rate": 1.7602858969668365e-05,
      "loss": 0.4619,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10966004431247711,
      "step": 460,
      "valid_targets_mean": 5196.4,
      "valid_targets_min": 1548
    },
    {
      "epoch": 2.9664,
      "grad_norm": 0.24127707453224004,
      "learning_rate": 1.716171174887231e-05,
      "loss": 0.4417,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10286953300237656,
      "step": 465,
      "valid_targets_mean": 5809.4,
      "valid_targets_min": 1703
    },
    {
      "epoch": 2.9984,
      "grad_norm": 0.24085134696567514,
      "learning_rate": 1.6721969504173484e-05,
      "loss": 0.443,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11106819659471512,
      "step": 470,
      "valid_targets_mean": 5545.3,
      "valid_targets_min": 1533
    },
    {
      "epoch": 3.0256,
      "grad_norm": 0.2541813174463172,
      "learning_rate": 1.628384991161041e-05,
      "loss": 0.4528,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11498390138149261,
      "step": 475,
      "valid_targets_mean": 5463.5,
      "valid_targets_min": 1925
    },
    {
      "epoch": 3.0576,
      "grad_norm": 0.2353622062322344,
      "learning_rate": 1.5847569843995452e-05,
      "loss": 0.4449,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1087237298488617,
      "step": 480,
      "valid_targets_mean": 5840.9,
      "valid_targets_min": 1507
    },
    {
      "epoch": 3.0896,
      "grad_norm": 0.22580596321042834,
      "learning_rate": 1.5413345263560922e-05,
      "loss": 0.4421,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11730214208364487,
      "step": 485,
      "valid_targets_mean": 7615.9,
      "valid_targets_min": 4895
    },
    {
      "epoch": 3.1216,
      "grad_norm": 0.2346986705939085,
      "learning_rate": 1.4981391115056032e-05,
      "loss": 0.4484,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08560355007648468,
      "step": 490,
      "valid_targets_mean": 5496.8,
      "valid_targets_min": 2426
    },
    {
      "epoch": 3.1536,
      "grad_norm": 0.2474046202773989,
      "learning_rate": 1.455192121934748e-05,
      "loss": 0.4392,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10706610977649689,
      "step": 495,
      "valid_targets_mean": 6306.4,
      "valid_targets_min": 1468
    },
    {
      "epoch": 3.1856,
      "grad_norm": 0.24598088386643885,
      "learning_rate": 1.4125148167576303e-05,
      "loss": 0.4517,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0977165475487709,
      "step": 500,
      "valid_targets_mean": 5140.1,
      "valid_targets_min": 2124
    },
    {
      "epoch": 3.2176,
      "grad_norm": 0.23772157152624504,
      "learning_rate": 1.3701283215923563e-05,
      "loss": 0.4515,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12374231219291687,
      "step": 505,
      "valid_targets_mean": 6650.8,
      "valid_targets_min": 2146
    },
    {
      "epoch": 3.2496,
      "grad_norm": 0.2381861982363929,
      "learning_rate": 1.328053618103677e-05,
      "loss": 0.4524,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08905310928821564,
      "step": 510,
      "valid_targets_mean": 4238.1,
      "valid_targets_min": 1246
    },
    {
      "epoch": 3.2816,
      "grad_norm": 0.24129917399571515,
      "learning_rate": 1.2863115336168916e-05,
      "loss": 0.4424,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10858754813671112,
      "step": 515,
      "valid_targets_mean": 5508.2,
      "valid_targets_min": 1990
    },
    {
      "epoch": 3.3136,
      "grad_norm": 0.2299768195746735,
      "learning_rate": 1.2449227308081509e-05,
      "loss": 0.4305,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10494785010814667,
      "step": 520,
      "valid_targets_mean": 5587.0,
      "valid_targets_min": 1414
    },
    {
      "epoch": 3.3456,
      "grad_norm": 0.22756287907781986,
      "learning_rate": 1.2039076974762587e-05,
      "loss": 0.4412,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09129714220762253,
      "step": 525,
      "valid_targets_mean": 5011.6,
      "valid_targets_min": 1275
    },
    {
      "epoch": 3.3776,
      "grad_norm": 0.2490236621582409,
      "learning_rate": 1.163286736401044e-05,
      "loss": 0.4353,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10486893355846405,
      "step": 530,
      "valid_targets_mean": 5423.1,
      "valid_targets_min": 1137
    },
    {
      "epoch": 3.4096,
      "grad_norm": 0.22017430759615417,
      "learning_rate": 1.123079955293322e-05,
      "loss": 0.4427,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1027197316288948,
      "step": 535,
      "valid_targets_mean": 5919.5,
      "valid_targets_min": 1816
    },
    {
      "epoch": 3.4416,
      "grad_norm": 0.2140066307199504,
      "learning_rate": 1.0833072568414037e-05,
      "loss": 0.4454,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11336848139762878,
      "step": 540,
      "valid_targets_mean": 7503.7,
      "valid_targets_min": 4812
    },
    {
      "epoch": 3.4736000000000002,
      "grad_norm": 0.23150362740195682,
      "learning_rate": 1.0439883288591057e-05,
      "loss": 0.4561,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11124958097934723,
      "step": 545,
      "valid_targets_mean": 6104.5,
      "valid_targets_min": 3593
    },
    {
      "epoch": 3.5056000000000003,
      "grad_norm": 0.2239460927463848,
      "learning_rate": 1.0051426345401202e-05,
      "loss": 0.4383,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08680635690689087,
      "step": 550,
      "valid_targets_mean": 4237.5,
      "valid_targets_min": 1464
    },
    {
      "epoch": 3.5376,
      "grad_norm": 0.2276675294214827,
      "learning_rate": 9.667894028235704e-06,
      "loss": 0.4512,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12256583571434021,
      "step": 555,
      "valid_targets_mean": 6583.3,
      "valid_targets_min": 3849
    },
    {
      "epoch": 3.5696,
      "grad_norm": 0.23874452325722306,
      "learning_rate": 9.289476188755315e-06,
      "loss": 0.4354,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10612036287784576,
      "step": 560,
      "valid_targets_mean": 6009.8,
      "valid_targets_min": 2688
    },
    {
      "epoch": 3.6016,
      "grad_norm": 0.21514329189036926,
      "learning_rate": 8.916360146912122e-06,
      "loss": 0.4469,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11689209938049316,
      "step": 565,
      "valid_targets_mean": 6188.1,
      "valid_targets_min": 3993
    },
    {
      "epoch": 3.6336,
      "grad_norm": 0.2362020094484568,
      "learning_rate": 8.548730598224646e-06,
      "loss": 0.4484,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11842970550060272,
      "step": 570,
      "valid_targets_mean": 5449.2,
      "valid_targets_min": 1275
    },
    {
      "epoch": 3.6656,
      "grad_norm": 0.2401716944941585,
      "learning_rate": 8.186769522352053e-06,
      "loss": 0.4512,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11459657549858093,
      "step": 575,
      "valid_targets_mean": 5901.6,
      "valid_targets_min": 3112
    },
    {
      "epoch": 3.6976,
      "grad_norm": 0.2260946660210682,
      "learning_rate": 7.830656093012714e-06,
      "loss": 0.45,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11391419172286987,
      "step": 580,
      "valid_targets_mean": 6314.2,
      "valid_targets_min": 2536
    },
    {
      "epoch": 3.7296,
      "grad_norm": 0.23490628110150955,
      "learning_rate": 7.480566589291696e-06,
      "loss": 0.4402,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1118554174900055,
      "step": 585,
      "valid_targets_mean": 6000.4,
      "valid_targets_min": 2420
    },
    {
      "epoch": 3.7616,
      "grad_norm": 0.22394540910322736,
      "learning_rate": 7.1366743083812285e-06,
      "loss": 0.4336,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11416637897491455,
      "step": 590,
      "valid_targets_mean": 5651.9,
      "valid_targets_min": 1312
    },
    {
      "epoch": 3.7936,
      "grad_norm": 0.23434924832059395,
      "learning_rate": 6.799149479797101e-06,
      "loss": 0.4524,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1163535937666893,
      "step": 595,
      "valid_targets_mean": 6361.0,
      "valid_targets_min": 3963
    },
    {
      "epoch": 3.8256,
      "grad_norm": 0.20434011194504567,
      "learning_rate": 6.4681591811137e-06,
      "loss": 0.431,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10242234170436859,
      "step": 600,
      "valid_targets_mean": 6733.9,
      "valid_targets_min": 2050
    },
    {
      "epoch": 3.8576,
      "grad_norm": 0.2205683173173195,
      "learning_rate": 6.143867255259197e-06,
      "loss": 0.4469,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10782508552074432,
      "step": 605,
      "valid_targets_mean": 5887.1,
      "valid_targets_min": 1264
    },
    {
      "epoch": 3.8895999999999997,
      "grad_norm": 0.22196701228922652,
      "learning_rate": 5.8264342294119504e-06,
      "loss": 0.4507,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09781044721603394,
      "step": 610,
      "valid_targets_mean": 6692.9,
      "valid_targets_min": 3335
    },
    {
      "epoch": 3.9215999999999998,
      "grad_norm": 0.23419677990052767,
      "learning_rate": 5.516017235538258e-06,
      "loss": 0.4398,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.088894322514534,
      "step": 615,
      "valid_targets_mean": 5359.7,
      "valid_targets_min": 1567
    },
    {
      "epoch": 3.9536,
      "grad_norm": 0.20991276705753859,
      "learning_rate": 5.212769932610695e-06,
      "loss": 0.4465,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08706802129745483,
      "step": 620,
      "valid_targets_mean": 5569.8,
      "valid_targets_min": 1280
    },
    {
      "epoch": 3.9856,
      "grad_norm": 0.2305681012598391,
      "learning_rate": 4.916842430545681e-06,
      "loss": 0.4464,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10766085982322693,
      "step": 625,
      "valid_targets_mean": 6046.6,
      "valid_targets_min": 2956
    },
    {
      "epoch": 4.0128,
      "grad_norm": 0.23248030070161385,
      "learning_rate": 4.628381215897837e-06,
      "loss": 0.4368,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09869658201932907,
      "step": 630,
      "valid_targets_mean": 5463.2,
      "valid_targets_min": 1626
    },
    {
      "epoch": 4.0448,
      "grad_norm": 0.2320693470818531,
      "learning_rate": 4.347529079347914e-06,
      "loss": 0.4401,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12055571377277374,
      "step": 635,
      "valid_targets_mean": 5899.8,
      "valid_targets_min": 4103
    },
    {
      "epoch": 4.0768,
      "grad_norm": 0.2376324397286371,
      "learning_rate": 4.074425045020247e-06,
      "loss": 0.4506,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10627171397209167,
      "step": 640,
      "valid_targets_mean": 4855.4,
      "valid_targets_min": 1443
    },
    {
      "epoch": 4.1088,
      "grad_norm": 0.23365976592481544,
      "learning_rate": 3.8092043016646487e-06,
      "loss": 0.4392,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11408081650733948,
      "step": 645,
      "valid_targets_mean": 5649.8,
      "valid_targets_min": 1275
    },
    {
      "epoch": 4.1408,
      "grad_norm": 0.21456883184147094,
      "learning_rate": 3.551998135736867e-06,
      "loss": 0.4316,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10515138506889343,
      "step": 650,
      "valid_targets_mean": 5795.8,
      "valid_targets_min": 1994
    },
    {
      "epoch": 4.1728,
      "grad_norm": 0.23159726143523057,
      "learning_rate": 3.3029338664107267e-06,
      "loss": 0.4436,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12827849388122559,
      "step": 655,
      "valid_targets_mean": 6799.2,
      "valid_targets_min": 4330
    },
    {
      "epoch": 4.2048,
      "grad_norm": 0.2267271521046096,
      "learning_rate": 3.0621347825540625e-06,
      "loss": 0.4335,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11244725435972214,
      "step": 660,
      "valid_targets_mean": 5340.2,
      "valid_targets_min": 2584
    },
    {
      "epoch": 4.2368,
      "grad_norm": 0.20149080037204617,
      "learning_rate": 2.8297200816997183e-06,
      "loss": 0.4342,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09943097829818726,
      "step": 665,
      "valid_targets_mean": 5967.9,
      "valid_targets_min": 1349
    },
    {
      "epoch": 4.2688,
      "grad_norm": 0.22316435861473205,
      "learning_rate": 2.605804811041803e-06,
      "loss": 0.4475,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1142878532409668,
      "step": 670,
      "valid_targets_mean": 5788.6,
      "valid_targets_min": 1504
    },
    {
      "epoch": 4.3008,
      "grad_norm": 0.23749043061782324,
      "learning_rate": 2.390499810486351e-06,
      "loss": 0.4398,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09940489381551743,
      "step": 675,
      "valid_targets_mean": 5341.0,
      "valid_targets_min": 1211
    },
    {
      "epoch": 4.3328,
      "grad_norm": 0.21705720222967442,
      "learning_rate": 2.183911657784685e-06,
      "loss": 0.4384,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10698309540748596,
      "step": 680,
      "valid_targets_mean": 6206.0,
      "valid_targets_min": 1104
    },
    {
      "epoch": 4.3648,
      "grad_norm": 0.2359689779618734,
      "learning_rate": 1.986142615776532e-06,
      "loss": 0.4354,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12712863087654114,
      "step": 685,
      "valid_targets_mean": 6414.2,
      "valid_targets_min": 2090
    },
    {
      "epoch": 4.3968,
      "grad_norm": 0.2270181783968223,
      "learning_rate": 1.7972905817690644e-06,
      "loss": 0.4482,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11009643971920013,
      "step": 690,
      "valid_targets_mean": 5408.8,
      "valid_targets_min": 1740
    },
    {
      "epoch": 4.4288,
      "grad_norm": 0.1982686366700752,
      "learning_rate": 1.617449039076955e-06,
      "loss": 0.4467,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11948183923959732,
      "step": 695,
      "valid_targets_mean": 8110.3,
      "valid_targets_min": 1565
    },
    {
      "epoch": 4.4608,
      "grad_norm": 0.22182051002480957,
      "learning_rate": 1.4467070107473413e-06,
      "loss": 0.4407,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09709776937961578,
      "step": 700,
      "valid_targets_mean": 5159.9,
      "valid_targets_min": 3161
    },
    {
      "epoch": 4.4928,
      "grad_norm": 0.21510025780610673,
      "learning_rate": 1.2851490154926816e-06,
      "loss": 0.4475,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11582086980342865,
      "step": 705,
      "valid_targets_mean": 6414.6,
      "valid_targets_min": 1497
    },
    {
      "epoch": 4.5248,
      "grad_norm": 0.20748343105814038,
      "learning_rate": 1.1328550258533211e-06,
      "loss": 0.4444,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12172878533601761,
      "step": 710,
      "valid_targets_mean": 6439.6,
      "valid_targets_min": 3898
    },
    {
      "epoch": 4.5568,
      "grad_norm": 0.21381893999640084,
      "learning_rate": 9.899004286103953e-07,
      "loss": 0.4367,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10774854570627213,
      "step": 715,
      "valid_targets_mean": 6118.2,
      "valid_targets_min": 1434
    },
    {
      "epoch": 4.5888,
      "grad_norm": 0.22561568200277468,
      "learning_rate": 8.5635598746876e-07,
      "loss": 0.4344,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11041155457496643,
      "step": 720,
      "valid_targets_mean": 6051.2,
      "valid_targets_min": 3805
    },
    {
      "epoch": 4.6208,
      "grad_norm": 0.20402603591723495,
      "learning_rate": 7.32287808028389e-07,
      "loss": 0.4338,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11558706313371658,
      "step": 725,
      "valid_targets_mean": 6586.0,
      "valid_targets_min": 4236
    },
    {
      "epoch": 4.6528,
      "grad_norm": 0.24563597589147643,
      "learning_rate": 6.177573050615327e-07,
      "loss": 0.4304,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09163513034582138,
      "step": 730,
      "valid_targets_mean": 4575.3,
      "valid_targets_min": 1257
    },
    {
      "epoch": 4.6848,
      "grad_norm": 0.21645901836986417,
      "learning_rate": 5.128211721119213e-07,
      "loss": 0.4337,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11745560169219971,
      "step": 735,
      "valid_targets_mean": 6024.5,
      "valid_targets_min": 1402
    },
    {
      "epoch": 4.7168,
      "grad_norm": 0.21851740632245759,
      "learning_rate": 4.175313534309755e-07,
      "loss": 0.4366,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10883457958698273,
      "step": 740,
      "valid_targets_mean": 6283.2,
      "valid_targets_min": 1555
    },
    {
      "epoch": 4.7488,
      "grad_norm": 0.21932736290236435,
      "learning_rate": 3.319350182649861e-07,
      "loss": 0.4468,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10903538763523102,
      "step": 745,
      "valid_targets_mean": 6333.0,
      "valid_targets_min": 1504
    },
    {
      "epoch": 4.7808,
      "grad_norm": 0.22616118624010112,
      "learning_rate": 2.560745375059392e-07,
      "loss": 0.4382,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11657419800758362,
      "step": 750,
      "valid_targets_mean": 7166.1,
      "valid_targets_min": 3048
    },
    {
      "epoch": 4.8128,
      "grad_norm": 0.22643827545429354,
      "learning_rate": 1.8998746271758016e-07,
      "loss": 0.4369,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10837658494710922,
      "step": 755,
      "valid_targets_mean": 5800.5,
      "valid_targets_min": 3765
    },
    {
      "epoch": 4.8448,
      "grad_norm": 0.21056524475735883,
      "learning_rate": 1.337065075470778e-07,
      "loss": 0.4389,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10905088484287262,
      "step": 760,
      "valid_targets_mean": 6031.9,
      "valid_targets_min": 3851
    },
    {
      "epoch": 4.8768,
      "grad_norm": 0.22673914171528292,
      "learning_rate": 8.725953153150279e-08,
      "loss": 0.4299,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11067554354667664,
      "step": 765,
      "valid_targets_mean": 5955.0,
      "valid_targets_min": 3870
    },
    {
      "epoch": 4.9088,
      "grad_norm": 0.2234901980490257,
      "learning_rate": 5.066952630711886e-08,
      "loss": 0.459,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.13258133828639984,
      "step": 770,
      "valid_targets_mean": 5992.2,
      "valid_targets_min": 1472
    },
    {
      "epoch": 4.9408,
      "grad_norm": 0.22776223805560375,
      "learning_rate": 2.3954604228342283e-08,
      "loss": 0.4454,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10461519658565521,
      "step": 775,
      "valid_targets_mean": 5661.4,
      "valid_targets_min": 1507
    },
    {
      "epoch": 4.9728,
      "grad_norm": 0.2189412079803582,
      "learning_rate": 7.12798940197601e-09,
      "loss": 0.439,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11273640394210815,
      "step": 780,
      "valid_targets_mean": 5641.1,
      "valid_targets_min": 2808
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.381336408442084,
      "learning_rate": 1.9801114115480802e-10,
      "loss": 0.4311,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.3657087981700897,
      "step": 785,
      "valid_targets_mean": 6505.4,
      "valid_targets_min": 1469
    },
    {
      "epoch": 5.0,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.3657087981700897,
      "step": 785,
      "total_flos": 2.0706726654792172e+18,
      "train_loss": 0.47480044941993277,
      "train_runtime": 19659.7391,
      "train_samples_per_second": 2.543,
      "train_steps_per_second": 0.04,
      "valid_targets_mean": 6505.4,
      "valid_targets_min": 1469
    }
  ],
  "logging_steps": 5,
  "max_steps": 785,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.0706726654792172e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}