{
  "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": 7.992597780897032,
      "learning_rate": 2.0253164556962026e-06,
      "loss": 0.7346,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1965068280696869,
      "step": 5,
      "valid_targets_mean": 5113.5,
      "valid_targets_min": 3169
    },
    {
      "epoch": 0.064,
      "grad_norm": 4.905164971122138,
      "learning_rate": 4.556962025316456e-06,
      "loss": 0.7062,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.16786593198776245,
      "step": 10,
      "valid_targets_mean": 5742.6,
      "valid_targets_min": 4001
    },
    {
      "epoch": 0.096,
      "grad_norm": 2.4265542623454994,
      "learning_rate": 7.08860759493671e-06,
      "loss": 0.6469,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1609123945236206,
      "step": 15,
      "valid_targets_mean": 5070.8,
      "valid_targets_min": 1187
    },
    {
      "epoch": 0.128,
      "grad_norm": 0.9695481834207988,
      "learning_rate": 9.620253164556963e-06,
      "loss": 0.5955,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.152688130736351,
      "step": 20,
      "valid_targets_mean": 5043.4,
      "valid_targets_min": 3778
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8556904894549953,
      "learning_rate": 1.2151898734177216e-05,
      "loss": 0.5631,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1266036331653595,
      "step": 25,
      "valid_targets_mean": 5028.5,
      "valid_targets_min": 3598
    },
    {
      "epoch": 0.192,
      "grad_norm": 0.5406877764182748,
      "learning_rate": 1.468354430379747e-05,
      "loss": 0.5366,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12929826974868774,
      "step": 30,
      "valid_targets_mean": 5412.2,
      "valid_targets_min": 3452
    },
    {
      "epoch": 0.224,
      "grad_norm": 0.4491232079388601,
      "learning_rate": 1.7215189873417723e-05,
      "loss": 0.5279,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12588968873023987,
      "step": 35,
      "valid_targets_mean": 5830.6,
      "valid_targets_min": 1236
    },
    {
      "epoch": 0.256,
      "grad_norm": 0.3734935489185054,
      "learning_rate": 1.974683544303798e-05,
      "loss": 0.4878,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1219080314040184,
      "step": 40,
      "valid_targets_mean": 5462.8,
      "valid_targets_min": 3876
    },
    {
      "epoch": 0.288,
      "grad_norm": 0.3243375480472286,
      "learning_rate": 2.2278481012658228e-05,
      "loss": 0.4516,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11703348159790039,
      "step": 45,
      "valid_targets_mean": 5427.9,
      "valid_targets_min": 3616
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2882643420638927,
      "learning_rate": 2.481012658227848e-05,
      "loss": 0.4482,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10204213112592697,
      "step": 50,
      "valid_targets_mean": 5454.8,
      "valid_targets_min": 1708
    },
    {
      "epoch": 0.352,
      "grad_norm": 0.2769504438340826,
      "learning_rate": 2.7341772151898737e-05,
      "loss": 0.4342,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11272697150707245,
      "step": 55,
      "valid_targets_mean": 5620.4,
      "valid_targets_min": 1322
    },
    {
      "epoch": 0.384,
      "grad_norm": 0.2620427633098365,
      "learning_rate": 2.987341772151899e-05,
      "loss": 0.4211,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10937047749757767,
      "step": 60,
      "valid_targets_mean": 5090.8,
      "valid_targets_min": 3842
    },
    {
      "epoch": 0.416,
      "grad_norm": 0.2463435677646114,
      "learning_rate": 3.240506329113924e-05,
      "loss": 0.4135,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10371287912130356,
      "step": 65,
      "valid_targets_mean": 5944.2,
      "valid_targets_min": 3980
    },
    {
      "epoch": 0.448,
      "grad_norm": 0.2439380748245345,
      "learning_rate": 3.49367088607595e-05,
      "loss": 0.4112,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11104129254817963,
      "step": 70,
      "valid_targets_mean": 6091.2,
      "valid_targets_min": 4103
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2592892650393887,
      "learning_rate": 3.746835443037975e-05,
      "loss": 0.392,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09229408204555511,
      "step": 75,
      "valid_targets_mean": 5581.8,
      "valid_targets_min": 3931
    },
    {
      "epoch": 0.512,
      "grad_norm": 0.27965263535503615,
      "learning_rate": 4e-05,
      "loss": 0.3945,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09553131461143494,
      "step": 80,
      "valid_targets_mean": 5852.2,
      "valid_targets_min": 3736
    },
    {
      "epoch": 0.544,
      "grad_norm": 0.23703955142453972,
      "learning_rate": 3.999504991751045e-05,
      "loss": 0.3795,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09883347153663635,
      "step": 85,
      "valid_targets_mean": 6116.4,
      "valid_targets_min": 1188
    },
    {
      "epoch": 0.576,
      "grad_norm": 0.27006174007491235,
      "learning_rate": 3.9980202120373464e-05,
      "loss": 0.392,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09063629806041718,
      "step": 90,
      "valid_targets_mean": 4680.5,
      "valid_targets_min": 638
    },
    {
      "epoch": 0.608,
      "grad_norm": 0.26947357129102495,
      "learning_rate": 3.995546395837111e-05,
      "loss": 0.3846,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1091674342751503,
      "step": 95,
      "valid_targets_mean": 6072.8,
      "valid_targets_min": 3603
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2927613874979733,
      "learning_rate": 3.992084767709763e-05,
      "loss": 0.3735,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09993791580200195,
      "step": 100,
      "valid_targets_mean": 5355.8,
      "valid_targets_min": 826
    },
    {
      "epoch": 0.672,
      "grad_norm": 0.24121791608110918,
      "learning_rate": 3.987637041189781e-05,
      "loss": 0.3845,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09499676525592804,
      "step": 105,
      "valid_targets_mean": 6090.1,
      "valid_targets_min": 3190
    },
    {
      "epoch": 0.704,
      "grad_norm": 0.24568965637675186,
      "learning_rate": 3.982205417938482e-05,
      "loss": 0.3783,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09854920208454132,
      "step": 110,
      "valid_targets_mean": 5208.7,
      "valid_targets_min": 3716
    },
    {
      "epoch": 0.736,
      "grad_norm": 0.2440314012575419,
      "learning_rate": 3.975792586654179e-05,
      "loss": 0.3661,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08172737061977386,
      "step": 115,
      "valid_targets_mean": 5737.1,
      "valid_targets_min": 2032
    },
    {
      "epoch": 0.768,
      "grad_norm": 0.25094578582914895,
      "learning_rate": 3.968401721741259e-05,
      "loss": 0.3729,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08649879693984985,
      "step": 120,
      "valid_targets_mean": 4845.7,
      "valid_targets_min": 2490
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2454746912712756,
      "learning_rate": 3.960036481738819e-05,
      "loss": 0.3772,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08654075860977173,
      "step": 125,
      "valid_targets_mean": 5857.2,
      "valid_targets_min": 4130
    },
    {
      "epoch": 0.832,
      "grad_norm": 0.27286819412305224,
      "learning_rate": 3.950701007509667e-05,
      "loss": 0.3698,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0876830592751503,
      "step": 130,
      "valid_targets_mean": 6018.9,
      "valid_targets_min": 3598
    },
    {
      "epoch": 0.864,
      "grad_norm": 0.24040936227390755,
      "learning_rate": 3.940399920190552e-05,
      "loss": 0.3615,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09337051212787628,
      "step": 135,
      "valid_targets_mean": 5848.9,
      "valid_targets_min": 4036
    },
    {
      "epoch": 0.896,
      "grad_norm": 0.26092439770692516,
      "learning_rate": 3.92913831890467e-05,
      "loss": 0.3687,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0913911908864975,
      "step": 140,
      "valid_targets_mean": 5912.9,
      "valid_targets_min": 4229
    },
    {
      "epoch": 0.928,
      "grad_norm": 0.24371078285635853,
      "learning_rate": 3.916921778237556e-05,
      "loss": 0.3664,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08543594181537628,
      "step": 145,
      "valid_targets_mean": 5582.9,
      "valid_targets_min": 3696
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.217048732240177,
      "learning_rate": 3.903756345477612e-05,
      "loss": 0.3586,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08604587614536285,
      "step": 150,
      "valid_targets_mean": 5474.2,
      "valid_targets_min": 3167
    },
    {
      "epoch": 0.992,
      "grad_norm": 0.26711655142264756,
      "learning_rate": 3.889648537622657e-05,
      "loss": 0.3612,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08616804331541061,
      "step": 155,
      "valid_targets_mean": 5076.6,
      "valid_targets_min": 709
    },
    {
      "epoch": 1.0192,
      "grad_norm": 0.24511649097845134,
      "learning_rate": 3.874605338153952e-05,
      "loss": 0.3586,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.090129055082798,
      "step": 160,
      "valid_targets_mean": 5228.8,
      "valid_targets_min": 3914
    },
    {
      "epoch": 1.0512,
      "grad_norm": 0.25785229272339794,
      "learning_rate": 3.8586341935793265e-05,
      "loss": 0.3559,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07839885354042053,
      "step": 165,
      "valid_targets_mean": 4965.1,
      "valid_targets_min": 1130
    },
    {
      "epoch": 1.0832,
      "grad_norm": 0.26965616757629585,
      "learning_rate": 3.841743009747089e-05,
      "loss": 0.3532,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08936138451099396,
      "step": 170,
      "valid_targets_mean": 6028.9,
      "valid_targets_min": 3924
    },
    {
      "epoch": 1.1152,
      "grad_norm": 0.2253315208645885,
      "learning_rate": 3.8239401479325714e-05,
      "loss": 0.3526,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08879570662975311,
      "step": 175,
      "valid_targets_mean": 5262.8,
      "valid_targets_min": 3437
    },
    {
      "epoch": 1.1472,
      "grad_norm": 0.2225876874756375,
      "learning_rate": 3.8052344206992276e-05,
      "loss": 0.3571,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08561760932207108,
      "step": 180,
      "valid_targets_mean": 5392.8,
      "valid_targets_min": 3094
    },
    {
      "epoch": 1.1792,
      "grad_norm": 0.25198165027120606,
      "learning_rate": 3.7856350875363396e-05,
      "loss": 0.3588,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08976033329963684,
      "step": 185,
      "valid_targets_mean": 5986.1,
      "valid_targets_min": 671
    },
    {
      "epoch": 1.2112,
      "grad_norm": 0.2340713016794323,
      "learning_rate": 3.765151850275497e-05,
      "loss": 0.3493,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08710001409053802,
      "step": 190,
      "valid_targets_mean": 5964.0,
      "valid_targets_min": 3471
    },
    {
      "epoch": 1.2432,
      "grad_norm": 0.2342881184173051,
      "learning_rate": 3.7437948482881104e-05,
      "loss": 0.3452,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0860249400138855,
      "step": 195,
      "valid_targets_mean": 6094.3,
      "valid_targets_min": 3818
    },
    {
      "epoch": 1.2752,
      "grad_norm": 0.2525665937119514,
      "learning_rate": 3.721574653466336e-05,
      "loss": 0.3504,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08649658411741257,
      "step": 200,
      "valid_targets_mean": 6088.2,
      "valid_targets_min": 3539
    },
    {
      "epoch": 1.3072,
      "grad_norm": 0.2274695837603368,
      "learning_rate": 3.698502264989903e-05,
      "loss": 0.3452,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0824115127325058,
      "step": 205,
      "valid_targets_mean": 5638.9,
      "valid_targets_min": 4279
    },
    {
      "epoch": 1.3392,
      "grad_norm": 0.26782770813054557,
      "learning_rate": 3.674589103881432e-05,
      "loss": 0.3476,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09814299643039703,
      "step": 210,
      "valid_targets_mean": 5398.6,
      "valid_targets_min": 3777
    },
    {
      "epoch": 1.3712,
      "grad_norm": 0.25592916799225457,
      "learning_rate": 3.64984700735293e-05,
      "loss": 0.3524,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08779767900705338,
      "step": 215,
      "valid_targets_mean": 5396.3,
      "valid_targets_min": 4022
    },
    {
      "epoch": 1.4032,
      "grad_norm": 0.26084030673703285,
      "learning_rate": 3.624288222946273e-05,
      "loss": 0.3556,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08763320744037628,
      "step": 220,
      "valid_targets_mean": 5151.4,
      "valid_targets_min": 1167
    },
    {
      "epoch": 1.4352,
      "grad_norm": 0.2297953014020372,
      "learning_rate": 3.597925402470578e-05,
      "loss": 0.3466,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08031567931175232,
      "step": 225,
      "valid_targets_mean": 5586.9,
      "valid_targets_min": 3479
    },
    {
      "epoch": 1.4672,
      "grad_norm": 0.25603642117773706,
      "learning_rate": 3.570771595739445e-05,
      "loss": 0.3455,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09005199372768402,
      "step": 230,
      "valid_targets_mean": 5418.8,
      "valid_targets_min": 3343
    },
    {
      "epoch": 1.4992,
      "grad_norm": 0.2259912054719558,
      "learning_rate": 3.5428402441111964e-05,
      "loss": 0.3501,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08260194212198257,
      "step": 235,
      "valid_targets_mean": 5300.5,
      "valid_targets_min": 2971
    },
    {
      "epoch": 1.5312000000000001,
      "grad_norm": 0.29512010642162345,
      "learning_rate": 3.5141451738352936e-05,
      "loss": 0.3516,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08057820051908493,
      "step": 240,
      "valid_targets_mean": 4825.9,
      "valid_targets_min": 757
    },
    {
      "epoch": 1.5632000000000001,
      "grad_norm": 0.2357451414355105,
      "learning_rate": 3.4847005892082266e-05,
      "loss": 0.3447,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08529014140367508,
      "step": 245,
      "valid_targets_mean": 5788.1,
      "valid_targets_min": 292
    },
    {
      "epoch": 1.5952,
      "grad_norm": 0.243031110151129,
      "learning_rate": 3.454521065542273e-05,
      "loss": 0.3403,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08439987897872925,
      "step": 250,
      "valid_targets_mean": 5117.9,
      "valid_targets_min": 2601
    },
    {
      "epoch": 1.6272,
      "grad_norm": 0.2631373773545141,
      "learning_rate": 3.423621541950597e-05,
      "loss": 0.3406,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08441022783517838,
      "step": 255,
      "valid_targets_mean": 5009.3,
      "valid_targets_min": 1060
    },
    {
      "epoch": 1.6592,
      "grad_norm": 0.25653559181125485,
      "learning_rate": 3.3920173139522664e-05,
      "loss": 0.3475,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09525300562381744,
      "step": 260,
      "valid_targets_mean": 6077.0,
      "valid_targets_min": 4176
    },
    {
      "epoch": 1.6912,
      "grad_norm": 0.21963523463947673,
      "learning_rate": 3.35972402590084e-05,
      "loss": 0.3334,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08650698512792587,
      "step": 265,
      "valid_targets_mean": 6181.9,
      "valid_targets_min": 3567
    },
    {
      "epoch": 1.7231999999999998,
      "grad_norm": 0.2654405120685201,
      "learning_rate": 3.326757663240291e-05,
      "loss": 0.3428,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08159274607896805,
      "step": 270,
      "valid_targets_mean": 5128.4,
      "valid_targets_min": 1748
    },
    {
      "epoch": 1.7551999999999999,
      "grad_norm": 0.24111341689755342,
      "learning_rate": 3.293134544592073e-05,
      "loss": 0.3408,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08390471339225769,
      "step": 275,
      "valid_targets_mean": 5030.9,
      "valid_targets_min": 4017
    },
    {
      "epoch": 1.7872,
      "grad_norm": 0.2168712885276149,
      "learning_rate": 3.258871313677274e-05,
      "loss": 0.3408,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08097031712532043,
      "step": 280,
      "valid_targets_mean": 5817.1,
      "valid_targets_min": 1078
    },
    {
      "epoch": 1.8192,
      "grad_norm": 0.24477381681141186,
      "learning_rate": 3.2239849310778316e-05,
      "loss": 0.3442,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08444704860448837,
      "step": 285,
      "valid_targets_mean": 5124.5,
      "valid_targets_min": 2765
    },
    {
      "epoch": 1.8512,
      "grad_norm": 0.23685277446824146,
      "learning_rate": 3.188492665840909e-05,
      "loss": 0.3392,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0879988819360733,
      "step": 290,
      "valid_targets_mean": 5757.4,
      "valid_targets_min": 4257
    },
    {
      "epoch": 1.8832,
      "grad_norm": 0.24431534481839284,
      "learning_rate": 3.1524120869305726e-05,
      "loss": 0.3477,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0799805298447609,
      "step": 295,
      "valid_targets_mean": 5013.1,
      "valid_targets_min": 2408
    },
    {
      "epoch": 1.9152,
      "grad_norm": 0.23717088002345657,
      "learning_rate": 3.11576105453101e-05,
      "loss": 0.336,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08305468410253525,
      "step": 300,
      "valid_targets_mean": 5636.8,
      "valid_targets_min": 3806
    },
    {
      "epoch": 1.9472,
      "grad_norm": 0.23843928568213113,
      "learning_rate": 3.0785577112055916e-05,
      "loss": 0.3325,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08401763439178467,
      "step": 305,
      "valid_targets_mean": 5015.1,
      "valid_targets_min": 4040
    },
    {
      "epoch": 1.9792,
      "grad_norm": 0.22230671454556905,
      "learning_rate": 3.040820472916153e-05,
      "loss": 0.338,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08706588298082352,
      "step": 310,
      "valid_targets_mean": 5607.3,
      "valid_targets_min": 2762
    },
    {
      "epoch": 2.0064,
      "grad_norm": 0.24335412699803988,
      "learning_rate": 3.002568019906939e-05,
      "loss": 0.3354,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08025571703910828,
      "step": 315,
      "valid_targets_mean": 5478.8,
      "valid_targets_min": 3702
    },
    {
      "epoch": 2.0384,
      "grad_norm": 0.24330841804288073,
      "learning_rate": 2.963819287457733e-05,
      "loss": 0.3357,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07753533124923706,
      "step": 320,
      "valid_targets_mean": 6092.9,
      "valid_targets_min": 3663
    },
    {
      "epoch": 2.0704,
      "grad_norm": 0.25064283317993247,
      "learning_rate": 2.924593456510733e-05,
      "loss": 0.3356,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08604463934898376,
      "step": 325,
      "valid_targets_mean": 5441.4,
      "valid_targets_min": 3454
    },
    {
      "epoch": 2.1024,
      "grad_norm": 0.24013319984882203,
      "learning_rate": 2.8849099441758306e-05,
      "loss": 0.3402,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0855434238910675,
      "step": 330,
      "valid_targets_mean": 6197.1,
      "valid_targets_min": 4166
    },
    {
      "epoch": 2.1344,
      "grad_norm": 0.23527928586707625,
      "learning_rate": 2.844788394118979e-05,
      "loss": 0.3414,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07986783981323242,
      "step": 335,
      "valid_targets_mean": 4266.2,
      "valid_targets_min": 1054
    },
    {
      "epoch": 2.1664,
      "grad_norm": 0.2476030088095561,
      "learning_rate": 2.8042486668384164e-05,
      "loss": 0.3335,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08800746500492096,
      "step": 340,
      "valid_targets_mean": 5671.7,
      "valid_targets_min": 2845
    },
    {
      "epoch": 2.1984,
      "grad_norm": 0.2212675652031375,
      "learning_rate": 2.7633108298335582e-05,
      "loss": 0.3239,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0804094448685646,
      "step": 345,
      "valid_targets_mean": 6075.1,
      "valid_targets_min": 2695
    },
    {
      "epoch": 2.2304,
      "grad_norm": 0.25952169809602754,
      "learning_rate": 2.721995147671416e-05,
      "loss": 0.323,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09103801101446152,
      "step": 350,
      "valid_targets_mean": 5498.5,
      "valid_targets_min": 2659
    },
    {
      "epoch": 2.2624,
      "grad_norm": 0.2383291871019339,
      "learning_rate": 2.68032207195547e-05,
      "loss": 0.3403,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09590868651866913,
      "step": 355,
      "valid_targets_mean": 5983.8,
      "valid_targets_min": 3808
    },
    {
      "epoch": 2.2944,
      "grad_norm": 0.2442383332052773,
      "learning_rate": 2.6383122312019604e-05,
      "loss": 0.3331,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09253934025764465,
      "step": 360,
      "valid_targets_mean": 5522.2,
      "valid_targets_min": 4197
    },
    {
      "epoch": 2.3264,
      "grad_norm": 0.26697574295179705,
      "learning_rate": 2.595986420628597e-05,
      "loss": 0.3431,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08339850604534149,
      "step": 365,
      "valid_targets_mean": 5238.8,
      "valid_targets_min": 4166
    },
    {
      "epoch": 2.3584,
      "grad_norm": 0.22410653282363405,
      "learning_rate": 2.5533655918607573e-05,
      "loss": 0.3386,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07287449389696121,
      "step": 370,
      "valid_targets_mean": 4816.2,
      "valid_targets_min": 1188
    },
    {
      "epoch": 2.3904,
      "grad_norm": 0.24337444824543214,
      "learning_rate": 2.510470842560259e-05,
      "loss": 0.3355,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09005071222782135,
      "step": 375,
      "valid_targets_mean": 5576.5,
      "valid_targets_min": 1602
    },
    {
      "epoch": 2.4224,
      "grad_norm": 0.23066395747891116,
      "learning_rate": 2.467323405981841e-05,
      "loss": 0.3242,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08278749883174896,
      "step": 380,
      "valid_targets_mean": 5321.2,
      "valid_targets_min": 2324
    },
    {
      "epoch": 2.4544,
      "grad_norm": 0.2193267803280765,
      "learning_rate": 2.423944640462533e-05,
      "loss": 0.3271,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08832499384880066,
      "step": 385,
      "valid_targets_mean": 5073.8,
      "valid_targets_min": 1054
    },
    {
      "epoch": 2.4864,
      "grad_norm": 0.21806083908641838,
      "learning_rate": 2.3803560188490968e-05,
      "loss": 0.3303,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08088940382003784,
      "step": 390,
      "valid_targets_mean": 6070.9,
      "valid_targets_min": 3083
    },
    {
      "epoch": 2.5183999999999997,
      "grad_norm": 0.22164532444220172,
      "learning_rate": 2.336579117868789e-05,
      "loss": 0.335,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09069826453924179,
      "step": 395,
      "valid_targets_mean": 5716.8,
      "valid_targets_min": 1389
    },
    {
      "epoch": 2.5504,
      "grad_norm": 0.23682560385891782,
      "learning_rate": 2.292635607448711e-05,
      "loss": 0.325,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08261828124523163,
      "step": 400,
      "valid_targets_mean": 5917.8,
      "valid_targets_min": 1918
    },
    {
      "epoch": 2.5824,
      "grad_norm": 0.24346477170249317,
      "learning_rate": 2.248547239989008e-05,
      "loss": 0.3371,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08386968076229095,
      "step": 405,
      "valid_targets_mean": 5211.4,
      "valid_targets_min": 3124
    },
    {
      "epoch": 2.6144,
      "grad_norm": 0.19472948812532634,
      "learning_rate": 2.204335839595255e-05,
      "loss": 0.3188,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07493388652801514,
      "step": 410,
      "valid_targets_mean": 5276.1,
      "valid_targets_min": 2549
    },
    {
      "epoch": 2.6464,
      "grad_norm": 0.22442481140650106,
      "learning_rate": 2.1600232912753452e-05,
      "loss": 0.3321,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07857003808021545,
      "step": 415,
      "valid_targets_mean": 5059.2,
      "valid_targets_min": 1478
    },
    {
      "epoch": 2.6784,
      "grad_norm": 0.21770822481824514,
      "learning_rate": 2.1156315301062293e-05,
      "loss": 0.3325,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08654189109802246,
      "step": 420,
      "valid_targets_mean": 6176.6,
      "valid_targets_min": 3999
    },
    {
      "epoch": 2.7104,
      "grad_norm": 0.2303772265208128,
      "learning_rate": 2.0711825303758712e-05,
      "loss": 0.3315,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07916657626628876,
      "step": 425,
      "valid_targets_mean": 4857.8,
      "valid_targets_min": 1871
    },
    {
      "epoch": 2.7424,
      "grad_norm": 0.2362702636474169,
      "learning_rate": 2.0266982947057962e-05,
      "loss": 0.3331,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08266505599021912,
      "step": 430,
      "valid_targets_mean": 5273.8,
      "valid_targets_min": 571
    },
    {
      "epoch": 2.7744,
      "grad_norm": 0.23238610336716467,
      "learning_rate": 1.9822008431596083e-05,
      "loss": 0.3318,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0778215304017067,
      "step": 435,
      "valid_targets_mean": 5079.6,
      "valid_targets_min": 1018
    },
    {
      "epoch": 2.8064,
      "grad_norm": 0.22745238988383099,
      "learning_rate": 1.937712202342881e-05,
      "loss": 0.3324,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08198985457420349,
      "step": 440,
      "valid_targets_mean": 5034.9,
      "valid_targets_min": 2190
    },
    {
      "epoch": 2.8384,
      "grad_norm": 0.22031663400522064,
      "learning_rate": 1.8932543944998037e-05,
      "loss": 0.3362,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09263623505830765,
      "step": 445,
      "valid_targets_mean": 5411.8,
      "valid_targets_min": 4103
    },
    {
      "epoch": 2.8704,
      "grad_norm": 0.23446135635447007,
      "learning_rate": 1.8488494266119877e-05,
      "loss": 0.3245,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08320371061563492,
      "step": 450,
      "valid_targets_mean": 5275.1,
      "valid_targets_min": 3131
    },
    {
      "epoch": 2.9024,
      "grad_norm": 0.2234904380138594,
      "learning_rate": 1.804519279504834e-05,
      "loss": 0.3251,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07954467833042145,
      "step": 455,
      "valid_targets_mean": 4903.2,
      "valid_targets_min": 1309
    },
    {
      "epoch": 2.9344,
      "grad_norm": 0.2274787643306054,
      "learning_rate": 1.7602858969668365e-05,
      "loss": 0.3294,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08588364720344543,
      "step": 460,
      "valid_targets_mean": 6065.0,
      "valid_targets_min": 4174
    },
    {
      "epoch": 2.9664,
      "grad_norm": 0.21541040235652797,
      "learning_rate": 1.716171174887231e-05,
      "loss": 0.3223,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08300478011369705,
      "step": 465,
      "valid_targets_mean": 5221.2,
      "valid_targets_min": 2466
    },
    {
      "epoch": 2.9984,
      "grad_norm": 0.20577577268402722,
      "learning_rate": 1.6721969504173484e-05,
      "loss": 0.328,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08499614894390106,
      "step": 470,
      "valid_targets_mean": 6225.9,
      "valid_targets_min": 3937
    },
    {
      "epoch": 3.0256,
      "grad_norm": 0.21751044527982824,
      "learning_rate": 1.628384991161041e-05,
      "loss": 0.3194,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08121082186698914,
      "step": 475,
      "valid_targets_mean": 6044.2,
      "valid_targets_min": 4658
    },
    {
      "epoch": 3.0576,
      "grad_norm": 0.21018229320526427,
      "learning_rate": 1.5847569843995452e-05,
      "loss": 0.3234,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07966670393943787,
      "step": 480,
      "valid_targets_mean": 5516.9,
      "valid_targets_min": 3439
    },
    {
      "epoch": 3.0896,
      "grad_norm": 0.21179087254975096,
      "learning_rate": 1.5413345263560922e-05,
      "loss": 0.3269,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08021973818540573,
      "step": 485,
      "valid_targets_mean": 5511.8,
      "valid_targets_min": 625
    },
    {
      "epoch": 3.1216,
      "grad_norm": 0.21533786371733335,
      "learning_rate": 1.4981391115056032e-05,
      "loss": 0.3151,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08071337640285492,
      "step": 490,
      "valid_targets_mean": 6069.5,
      "valid_targets_min": 4143
    },
    {
      "epoch": 3.1536,
      "grad_norm": 0.22340718433208287,
      "learning_rate": 1.455192121934748e-05,
      "loss": 0.3138,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07231708616018295,
      "step": 495,
      "valid_targets_mean": 5085.9,
      "valid_targets_min": 813
    },
    {
      "epoch": 3.1856,
      "grad_norm": 0.23470120256913535,
      "learning_rate": 1.4125148167576303e-05,
      "loss": 0.3316,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08944922685623169,
      "step": 500,
      "valid_targets_mean": 5925.7,
      "valid_targets_min": 4258
    },
    {
      "epoch": 3.2176,
      "grad_norm": 0.2225970343777138,
      "learning_rate": 1.3701283215923563e-05,
      "loss": 0.3198,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07638940215110779,
      "step": 505,
      "valid_targets_mean": 5141.3,
      "valid_targets_min": 1120
    },
    {
      "epoch": 3.2496,
      "grad_norm": 0.22812978410814985,
      "learning_rate": 1.328053618103677e-05,
      "loss": 0.3283,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08584798872470856,
      "step": 510,
      "valid_targets_mean": 5091.2,
      "valid_targets_min": 1999
    },
    {
      "epoch": 3.2816,
      "grad_norm": 0.2108055801479295,
      "learning_rate": 1.2863115336168916e-05,
      "loss": 0.3257,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07783376425504684,
      "step": 515,
      "valid_targets_mean": 5384.1,
      "valid_targets_min": 935
    },
    {
      "epoch": 3.3136,
      "grad_norm": 0.27065537006118323,
      "learning_rate": 1.2449227308081509e-05,
      "loss": 0.3207,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07168179750442505,
      "step": 520,
      "valid_targets_mean": 5510.8,
      "valid_targets_min": 686
    },
    {
      "epoch": 3.3456,
      "grad_norm": 0.22197248150979024,
      "learning_rate": 1.2039076974762587e-05,
      "loss": 0.3254,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08753179013729095,
      "step": 525,
      "valid_targets_mean": 5997.3,
      "valid_targets_min": 4451
    },
    {
      "epoch": 3.3776,
      "grad_norm": 0.21204994803562507,
      "learning_rate": 1.163286736401044e-05,
      "loss": 0.3295,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07973788678646088,
      "step": 530,
      "valid_targets_mean": 5424.2,
      "valid_targets_min": 2094
    },
    {
      "epoch": 3.4096,
      "grad_norm": 0.20565663141003207,
      "learning_rate": 1.123079955293322e-05,
      "loss": 0.3268,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07733616232872009,
      "step": 535,
      "valid_targets_mean": 5321.2,
      "valid_targets_min": 1101
    },
    {
      "epoch": 3.4416,
      "grad_norm": 0.21601342130688855,
      "learning_rate": 1.0833072568414037e-05,
      "loss": 0.3238,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08604863286018372,
      "step": 540,
      "valid_targets_mean": 5389.1,
      "valid_targets_min": 2707
    },
    {
      "epoch": 3.4736000000000002,
      "grad_norm": 0.2266250484856398,
      "learning_rate": 1.0439883288591057e-05,
      "loss": 0.3234,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08474144339561462,
      "step": 545,
      "valid_targets_mean": 5556.5,
      "valid_targets_min": 3476
    },
    {
      "epoch": 3.5056000000000003,
      "grad_norm": 0.21000231515568052,
      "learning_rate": 1.0051426345401202e-05,
      "loss": 0.3226,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07983237504959106,
      "step": 550,
      "valid_targets_mean": 5647.8,
      "valid_targets_min": 1239
    },
    {
      "epoch": 3.5376,
      "grad_norm": 0.21908337779577544,
      "learning_rate": 9.667894028235704e-06,
      "loss": 0.328,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07918648421764374,
      "step": 555,
      "valid_targets_mean": 5250.9,
      "valid_targets_min": 2989
    },
    {
      "epoch": 3.5696,
      "grad_norm": 0.20775201883443636,
      "learning_rate": 9.289476188755315e-06,
      "loss": 0.3207,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08295097947120667,
      "step": 560,
      "valid_targets_mean": 6207.6,
      "valid_targets_min": 4217
    },
    {
      "epoch": 3.6016,
      "grad_norm": 0.21294278678462036,
      "learning_rate": 8.916360146912122e-06,
      "loss": 0.3194,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07971015572547913,
      "step": 565,
      "valid_targets_mean": 5787.9,
      "valid_targets_min": 3808
    },
    {
      "epoch": 3.6336,
      "grad_norm": 0.22408542647313262,
      "learning_rate": 8.548730598224646e-06,
      "loss": 0.3242,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08619304746389389,
      "step": 570,
      "valid_targets_mean": 5396.3,
      "valid_targets_min": 3498
    },
    {
      "epoch": 3.6656,
      "grad_norm": 0.2289853339804503,
      "learning_rate": 8.186769522352053e-06,
      "loss": 0.3204,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08720386028289795,
      "step": 575,
      "valid_targets_mean": 5984.4,
      "valid_targets_min": 4048
    },
    {
      "epoch": 3.6976,
      "grad_norm": 0.21073530692466028,
      "learning_rate": 7.830656093012714e-06,
      "loss": 0.3286,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08480887115001678,
      "step": 580,
      "valid_targets_mean": 5643.3,
      "valid_targets_min": 2574
    },
    {
      "epoch": 3.7296,
      "grad_norm": 0.22372831454903153,
      "learning_rate": 7.480566589291696e-06,
      "loss": 0.3271,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0835227221250534,
      "step": 585,
      "valid_targets_mean": 5632.9,
      "valid_targets_min": 3630
    },
    {
      "epoch": 3.7616,
      "grad_norm": 0.19869385568005432,
      "learning_rate": 7.1366743083812285e-06,
      "loss": 0.3139,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07951802760362625,
      "step": 590,
      "valid_targets_mean": 5470.8,
      "valid_targets_min": 3152
    },
    {
      "epoch": 3.7936,
      "grad_norm": 0.20574173001103932,
      "learning_rate": 6.799149479797101e-06,
      "loss": 0.3233,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0819997638463974,
      "step": 595,
      "valid_targets_mean": 5568.1,
      "valid_targets_min": 4110
    },
    {
      "epoch": 3.8256,
      "grad_norm": 0.2152111168894627,
      "learning_rate": 6.4681591811137e-06,
      "loss": 0.3264,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09062743186950684,
      "step": 600,
      "valid_targets_mean": 5654.9,
      "valid_targets_min": 2936
    },
    {
      "epoch": 3.8576,
      "grad_norm": 0.2053653642091841,
      "learning_rate": 6.143867255259197e-06,
      "loss": 0.3286,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08145453035831451,
      "step": 605,
      "valid_targets_mean": 5277.7,
      "valid_targets_min": 958
    },
    {
      "epoch": 3.8895999999999997,
      "grad_norm": 0.19618687651688638,
      "learning_rate": 5.8264342294119504e-06,
      "loss": 0.3307,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07444492727518082,
      "step": 610,
      "valid_targets_mean": 5692.4,
      "valid_targets_min": 1054
    },
    {
      "epoch": 3.9215999999999998,
      "grad_norm": 0.4633944919440031,
      "learning_rate": 5.516017235538258e-06,
      "loss": 0.3218,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07183433324098587,
      "step": 615,
      "valid_targets_mean": 6619.4,
      "valid_targets_min": 4473
    },
    {
      "epoch": 3.9536,
      "grad_norm": 0.19886084345698227,
      "learning_rate": 5.212769932610695e-06,
      "loss": 0.3221,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0788281261920929,
      "step": 620,
      "valid_targets_mean": 5528.3,
      "valid_targets_min": 3187
    },
    {
      "epoch": 3.9856,
      "grad_norm": 0.208092162317957,
      "learning_rate": 4.916842430545681e-06,
      "loss": 0.3307,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08409751206636429,
      "step": 625,
      "valid_targets_mean": 6215.4,
      "valid_targets_min": 3757
    },
    {
      "epoch": 4.0128,
      "grad_norm": 0.21479215026797172,
      "learning_rate": 4.628381215897837e-06,
      "loss": 0.3245,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08276808261871338,
      "step": 630,
      "valid_targets_mean": 4926.9,
      "valid_targets_min": 2898
    },
    {
      "epoch": 4.0448,
      "grad_norm": 0.20550296016432856,
      "learning_rate": 4.347529079347914e-06,
      "loss": 0.3226,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07196902483701706,
      "step": 635,
      "valid_targets_mean": 5715.8,
      "valid_targets_min": 710
    },
    {
      "epoch": 4.0768,
      "grad_norm": 0.19841931672206817,
      "learning_rate": 4.074425045020247e-06,
      "loss": 0.3055,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07841131091117859,
      "step": 640,
      "valid_targets_mean": 5529.1,
      "valid_targets_min": 881
    },
    {
      "epoch": 4.1088,
      "grad_norm": 0.20632895716592087,
      "learning_rate": 3.8092043016646487e-06,
      "loss": 0.3238,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08663403987884521,
      "step": 645,
      "valid_targets_mean": 5540.6,
      "valid_targets_min": 646
    },
    {
      "epoch": 4.1408,
      "grad_norm": 0.19479668863200172,
      "learning_rate": 3.551998135736867e-06,
      "loss": 0.3247,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08184544742107391,
      "step": 650,
      "valid_targets_mean": 5659.1,
      "valid_targets_min": 3048
    },
    {
      "epoch": 4.1728,
      "grad_norm": 0.21722025161657196,
      "learning_rate": 3.3029338664107267e-06,
      "loss": 0.3205,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08144575357437134,
      "step": 655,
      "valid_targets_mean": 5178.1,
      "valid_targets_min": 571
    },
    {
      "epoch": 4.2048,
      "grad_norm": 0.20576725177526284,
      "learning_rate": 3.0621347825540625e-06,
      "loss": 0.3249,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07952291518449783,
      "step": 660,
      "valid_targets_mean": 5043.1,
      "valid_targets_min": 1090
    },
    {
      "epoch": 4.2368,
      "grad_norm": 0.2137647103914308,
      "learning_rate": 2.8297200816997183e-06,
      "loss": 0.3243,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08988995850086212,
      "step": 665,
      "valid_targets_mean": 6148.6,
      "valid_targets_min": 4401
    },
    {
      "epoch": 4.2688,
      "grad_norm": 0.20096638447003157,
      "learning_rate": 2.605804811041803e-06,
      "loss": 0.3129,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0822472795844078,
      "step": 670,
      "valid_targets_mean": 5288.3,
      "valid_targets_min": 3220
    },
    {
      "epoch": 4.3008,
      "grad_norm": 0.20183040579550365,
      "learning_rate": 2.390499810486351e-06,
      "loss": 0.3338,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08840819448232651,
      "step": 675,
      "valid_targets_mean": 5098.6,
      "valid_targets_min": 3391
    },
    {
      "epoch": 4.3328,
      "grad_norm": 0.19426925166522613,
      "learning_rate": 2.183911657784685e-06,
      "loss": 0.3189,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08456306159496307,
      "step": 680,
      "valid_targets_mean": 5405.1,
      "valid_targets_min": 3206
    },
    {
      "epoch": 4.3648,
      "grad_norm": 0.20116698442114167,
      "learning_rate": 1.986142615776532e-06,
      "loss": 0.3304,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07993205636739731,
      "step": 685,
      "valid_targets_mean": 5235.6,
      "valid_targets_min": 1167
    },
    {
      "epoch": 4.3968,
      "grad_norm": 0.20291008340159364,
      "learning_rate": 1.7972905817690644e-06,
      "loss": 0.3178,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08226853609085083,
      "step": 690,
      "valid_targets_mean": 5901.6,
      "valid_targets_min": 4097
    },
    {
      "epoch": 4.4288,
      "grad_norm": 0.1970309784297005,
      "learning_rate": 1.617449039076955e-06,
      "loss": 0.3205,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0889642983675003,
      "step": 695,
      "valid_targets_mean": 6463.2,
      "valid_targets_min": 3151
    },
    {
      "epoch": 4.4608,
      "grad_norm": 0.21957818241096985,
      "learning_rate": 1.4467070107473413e-06,
      "loss": 0.3102,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08201165497303009,
      "step": 700,
      "valid_targets_mean": 5256.1,
      "valid_targets_min": 1931
    },
    {
      "epoch": 4.4928,
      "grad_norm": 0.20381806326480958,
      "learning_rate": 1.2851490154926816e-06,
      "loss": 0.3217,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08283894509077072,
      "step": 705,
      "valid_targets_mean": 5474.8,
      "valid_targets_min": 3711
    },
    {
      "epoch": 4.5248,
      "grad_norm": 0.20007664300882147,
      "learning_rate": 1.1328550258533211e-06,
      "loss": 0.3195,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08363893628120422,
      "step": 710,
      "valid_targets_mean": 5336.9,
      "valid_targets_min": 4024
    },
    {
      "epoch": 4.5568,
      "grad_norm": 0.2027444945770559,
      "learning_rate": 9.899004286103953e-07,
      "loss": 0.3145,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08135296404361725,
      "step": 715,
      "valid_targets_mean": 5425.2,
      "valid_targets_min": 3646
    },
    {
      "epoch": 4.5888,
      "grad_norm": 0.18936965807900735,
      "learning_rate": 8.5635598746876e-07,
      "loss": 0.3156,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07710923254489899,
      "step": 720,
      "valid_targets_mean": 6018.6,
      "valid_targets_min": 4053
    },
    {
      "epoch": 4.6208,
      "grad_norm": 0.1847915315828253,
      "learning_rate": 7.32287808028389e-07,
      "loss": 0.3265,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07985624670982361,
      "step": 725,
      "valid_targets_mean": 6699.7,
      "valid_targets_min": 2723
    },
    {
      "epoch": 4.6528,
      "grad_norm": 0.2099701962926105,
      "learning_rate": 6.177573050615327e-07,
      "loss": 0.3154,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08430498838424683,
      "step": 730,
      "valid_targets_mean": 5119.5,
      "valid_targets_min": 3867
    },
    {
      "epoch": 4.6848,
      "grad_norm": 0.18915002670306677,
      "learning_rate": 5.128211721119213e-07,
      "loss": 0.3188,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0837610512971878,
      "step": 735,
      "valid_targets_mean": 6211.8,
      "valid_targets_min": 4771
    },
    {
      "epoch": 4.7168,
      "grad_norm": 0.21175210312067896,
      "learning_rate": 4.175313534309755e-07,
      "loss": 0.3217,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08258293569087982,
      "step": 740,
      "valid_targets_mean": 5606.8,
      "valid_targets_min": 4067
    },
    {
      "epoch": 4.7488,
      "grad_norm": 0.19171064939070626,
      "learning_rate": 3.319350182649861e-07,
      "loss": 0.3207,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08525727689266205,
      "step": 745,
      "valid_targets_mean": 6320.4,
      "valid_targets_min": 3489
    },
    {
      "epoch": 4.7808,
      "grad_norm": 0.20140414971721998,
      "learning_rate": 2.560745375059392e-07,
      "loss": 0.3179,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07746146619319916,
      "step": 750,
      "valid_targets_mean": 5373.7,
      "valid_targets_min": 3737
    },
    {
      "epoch": 4.8128,
      "grad_norm": 0.19725191792852176,
      "learning_rate": 1.8998746271758016e-07,
      "loss": 0.3231,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07618357986211777,
      "step": 755,
      "valid_targets_mean": 5033.5,
      "valid_targets_min": 1995
    },
    {
      "epoch": 4.8448,
      "grad_norm": 0.18764613291156892,
      "learning_rate": 1.337065075470778e-07,
      "loss": 0.3189,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0758911669254303,
      "step": 760,
      "valid_targets_mean": 5730.7,
      "valid_targets_min": 3598
    },
    {
      "epoch": 4.8768,
      "grad_norm": 0.18701696026878642,
      "learning_rate": 8.725953153150279e-08,
      "loss": 0.3149,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07627741992473602,
      "step": 765,
      "valid_targets_mean": 5283.1,
      "valid_targets_min": 389
    },
    {
      "epoch": 4.9088,
      "grad_norm": 0.20233391129510425,
      "learning_rate": 5.066952630711886e-08,
      "loss": 0.3263,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08208300173282623,
      "step": 770,
      "valid_targets_mean": 5043.8,
      "valid_targets_min": 645
    },
    {
      "epoch": 4.9408,
      "grad_norm": 0.19082641034730213,
      "learning_rate": 2.3954604228342283e-08,
      "loss": 0.3236,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07420461624860764,
      "step": 775,
      "valid_targets_mean": 5601.7,
      "valid_targets_min": 4300
    },
    {
      "epoch": 4.9728,
      "grad_norm": 0.19359894736289907,
      "learning_rate": 7.12798940197601e-09,
      "loss": 0.3208,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08184170722961426,
      "step": 780,
      "valid_targets_mean": 5441.5,
      "valid_targets_min": 1112
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.4056907192862938,
      "learning_rate": 1.9801114115480802e-10,
      "loss": 0.3226,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.3350988030433655,
      "step": 785,
      "valid_targets_mean": 5269.4,
      "valid_targets_min": 3557
    },
    {
      "epoch": 5.0,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.3350988030433655,
      "step": 785,
      "total_flos": 2.1121566765544899e+18,
      "train_loss": 0.35298769489215437,
      "train_runtime": 22279.9871,
      "train_samples_per_second": 2.242,
      "train_steps_per_second": 0.035,
      "valid_targets_mean": 5269.4,
      "valid_targets_min": 3557
    }
  ],
  "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": 2.1121566765544899e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}