{
  "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": 6.29571362806165,
      "learning_rate": 2.0253164556962026e-06,
      "loss": 0.7984,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.20111200213432312,
      "step": 5,
      "valid_targets_mean": 6251.3,
      "valid_targets_min": 4098
    },
    {
      "epoch": 0.064,
      "grad_norm": 4.142747067780766,
      "learning_rate": 4.556962025316456e-06,
      "loss": 0.7726,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1832539439201355,
      "step": 10,
      "valid_targets_mean": 5065.5,
      "valid_targets_min": 3320
    },
    {
      "epoch": 0.096,
      "grad_norm": 2.4969988994122945,
      "learning_rate": 7.08860759493671e-06,
      "loss": 0.7267,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.17620849609375,
      "step": 15,
      "valid_targets_mean": 4723.8,
      "valid_targets_min": 1847
    },
    {
      "epoch": 0.128,
      "grad_norm": 0.9782459900739924,
      "learning_rate": 9.620253164556963e-06,
      "loss": 0.6784,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.16910651326179504,
      "step": 20,
      "valid_targets_mean": 5319.5,
      "valid_targets_min": 3392
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8769838874619504,
      "learning_rate": 1.2151898734177216e-05,
      "loss": 0.6379,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.16586852073669434,
      "step": 25,
      "valid_targets_mean": 5131.4,
      "valid_targets_min": 1477
    },
    {
      "epoch": 0.192,
      "grad_norm": 0.5875668827467091,
      "learning_rate": 1.468354430379747e-05,
      "loss": 0.6028,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.14500227570533752,
      "step": 30,
      "valid_targets_mean": 4813.0,
      "valid_targets_min": 2362
    },
    {
      "epoch": 0.224,
      "grad_norm": 0.5055270161827264,
      "learning_rate": 1.7215189873417723e-05,
      "loss": 0.5818,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.15424823760986328,
      "step": 35,
      "valid_targets_mean": 5623.2,
      "valid_targets_min": 2022
    },
    {
      "epoch": 0.256,
      "grad_norm": 0.4112979236864466,
      "learning_rate": 1.974683544303798e-05,
      "loss": 0.5553,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.14260552823543549,
      "step": 40,
      "valid_targets_mean": 5440.1,
      "valid_targets_min": 2784
    },
    {
      "epoch": 0.288,
      "grad_norm": 0.37381208805120997,
      "learning_rate": 2.2278481012658228e-05,
      "loss": 0.533,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1266845166683197,
      "step": 45,
      "valid_targets_mean": 4593.1,
      "valid_targets_min": 1775
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.31903497688938987,
      "learning_rate": 2.481012658227848e-05,
      "loss": 0.5175,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10548511147499084,
      "step": 50,
      "valid_targets_mean": 4285.8,
      "valid_targets_min": 1231
    },
    {
      "epoch": 0.352,
      "grad_norm": 0.3013548874512016,
      "learning_rate": 2.7341772151898737e-05,
      "loss": 0.4966,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.13925188779830933,
      "step": 55,
      "valid_targets_mean": 5948.7,
      "valid_targets_min": 4283
    },
    {
      "epoch": 0.384,
      "grad_norm": 0.3216710577735999,
      "learning_rate": 2.987341772151899e-05,
      "loss": 0.4959,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11888957023620605,
      "step": 60,
      "valid_targets_mean": 5152.1,
      "valid_targets_min": 3420
    },
    {
      "epoch": 0.416,
      "grad_norm": 0.2642700688606322,
      "learning_rate": 3.240506329113924e-05,
      "loss": 0.4813,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11175169795751572,
      "step": 65,
      "valid_targets_mean": 4866.6,
      "valid_targets_min": 1999
    },
    {
      "epoch": 0.448,
      "grad_norm": 0.24266021950026248,
      "learning_rate": 3.49367088607595e-05,
      "loss": 0.4741,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1105281412601471,
      "step": 70,
      "valid_targets_mean": 5549.7,
      "valid_targets_min": 3180
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.28056322403698214,
      "learning_rate": 3.746835443037975e-05,
      "loss": 0.4724,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12893617153167725,
      "step": 75,
      "valid_targets_mean": 5813.1,
      "valid_targets_min": 2239
    },
    {
      "epoch": 0.512,
      "grad_norm": 0.2946600514035525,
      "learning_rate": 4e-05,
      "loss": 0.4544,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.12273722887039185,
      "step": 80,
      "valid_targets_mean": 5693.5,
      "valid_targets_min": 1794
    },
    {
      "epoch": 0.544,
      "grad_norm": 0.29942414728251926,
      "learning_rate": 3.999504991751045e-05,
      "loss": 0.4435,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0959555059671402,
      "step": 85,
      "valid_targets_mean": 4361.2,
      "valid_targets_min": 2180
    },
    {
      "epoch": 0.576,
      "grad_norm": 0.25712919180660754,
      "learning_rate": 3.9980202120373464e-05,
      "loss": 0.4466,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10809490084648132,
      "step": 90,
      "valid_targets_mean": 4999.4,
      "valid_targets_min": 1858
    },
    {
      "epoch": 0.608,
      "grad_norm": 0.27418518750411786,
      "learning_rate": 3.995546395837111e-05,
      "loss": 0.4435,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11591640114784241,
      "step": 95,
      "valid_targets_mean": 5033.2,
      "valid_targets_min": 3420
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.27722526478163395,
      "learning_rate": 3.992084767709763e-05,
      "loss": 0.4372,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.13024362921714783,
      "step": 100,
      "valid_targets_mean": 5340.6,
      "valid_targets_min": 3747
    },
    {
      "epoch": 0.672,
      "grad_norm": 0.2651607714751492,
      "learning_rate": 3.987637041189781e-05,
      "loss": 0.4374,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09705384075641632,
      "step": 105,
      "valid_targets_mean": 5213.8,
      "valid_targets_min": 2677
    },
    {
      "epoch": 0.704,
      "grad_norm": 0.26864768650565807,
      "learning_rate": 3.982205417938482e-05,
      "loss": 0.4294,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1038064956665039,
      "step": 110,
      "valid_targets_mean": 4970.3,
      "valid_targets_min": 2452
    },
    {
      "epoch": 0.736,
      "grad_norm": 0.25491986794110943,
      "learning_rate": 3.975792586654179e-05,
      "loss": 0.4277,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10245868563652039,
      "step": 115,
      "valid_targets_mean": 5669.7,
      "valid_targets_min": 2359
    },
    {
      "epoch": 0.768,
      "grad_norm": 0.2657109287601836,
      "learning_rate": 3.968401721741259e-05,
      "loss": 0.4299,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11396533250808716,
      "step": 120,
      "valid_targets_mean": 5281.9,
      "valid_targets_min": 3595
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3887533903373934,
      "learning_rate": 3.960036481738819e-05,
      "loss": 0.4271,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10856138169765472,
      "step": 125,
      "valid_targets_mean": 5841.7,
      "valid_targets_min": 3554
    },
    {
      "epoch": 0.832,
      "grad_norm": 0.29903074041842825,
      "learning_rate": 3.950701007509667e-05,
      "loss": 0.443,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10608379542827606,
      "step": 130,
      "valid_targets_mean": 4909.1,
      "valid_targets_min": 3166
    },
    {
      "epoch": 0.864,
      "grad_norm": 0.27614957787333827,
      "learning_rate": 3.940399920190552e-05,
      "loss": 0.4267,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09728549420833588,
      "step": 135,
      "valid_targets_mean": 5028.8,
      "valid_targets_min": 2367
    },
    {
      "epoch": 0.896,
      "grad_norm": 0.28545155080578316,
      "learning_rate": 3.92913831890467e-05,
      "loss": 0.4208,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09202072769403458,
      "step": 140,
      "valid_targets_mean": 4792.6,
      "valid_targets_min": 2658
    },
    {
      "epoch": 0.928,
      "grad_norm": 0.2586185803194004,
      "learning_rate": 3.916921778237556e-05,
      "loss": 0.4191,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11039420962333679,
      "step": 145,
      "valid_targets_mean": 5334.1,
      "valid_targets_min": 2420
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.26060174254291524,
      "learning_rate": 3.903756345477612e-05,
      "loss": 0.432,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11281171441078186,
      "step": 150,
      "valid_targets_mean": 4962.4,
      "valid_targets_min": 3008
    },
    {
      "epoch": 0.992,
      "grad_norm": 0.2622634552688161,
      "learning_rate": 3.889648537622657e-05,
      "loss": 0.4282,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11850865185260773,
      "step": 155,
      "valid_targets_mean": 5644.2,
      "valid_targets_min": 2831
    },
    {
      "epoch": 1.0192,
      "grad_norm": 0.28400802556266697,
      "learning_rate": 3.874605338153952e-05,
      "loss": 0.424,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10352396965026855,
      "step": 160,
      "valid_targets_mean": 5019.7,
      "valid_targets_min": 1802
    },
    {
      "epoch": 1.0512,
      "grad_norm": 0.27543230154003945,
      "learning_rate": 3.8586341935793265e-05,
      "loss": 0.4193,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09966562688350677,
      "step": 165,
      "valid_targets_mean": 5189.9,
      "valid_targets_min": 2784
    },
    {
      "epoch": 1.0832,
      "grad_norm": 0.27407233775279605,
      "learning_rate": 3.841743009747089e-05,
      "loss": 0.4142,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09901313483715057,
      "step": 170,
      "valid_targets_mean": 5483.7,
      "valid_targets_min": 2255
    },
    {
      "epoch": 1.1152,
      "grad_norm": 0.2921685541962091,
      "learning_rate": 3.8239401479325714e-05,
      "loss": 0.412,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09235119074583054,
      "step": 175,
      "valid_targets_mean": 4900.9,
      "valid_targets_min": 2069
    },
    {
      "epoch": 1.1472,
      "grad_norm": 0.2698012514337229,
      "learning_rate": 3.8052344206992276e-05,
      "loss": 0.4113,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09564970433712006,
      "step": 180,
      "valid_targets_mean": 4774.1,
      "valid_targets_min": 2037
    },
    {
      "epoch": 1.1792,
      "grad_norm": 0.28601965577687405,
      "learning_rate": 3.7856350875363396e-05,
      "loss": 0.4153,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09468363970518112,
      "step": 185,
      "valid_targets_mean": 4779.0,
      "valid_targets_min": 1838
    },
    {
      "epoch": 1.2112,
      "grad_norm": 0.2654708997786682,
      "learning_rate": 3.765151850275497e-05,
      "loss": 0.4137,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10876516997814178,
      "step": 190,
      "valid_targets_mean": 5857.3,
      "valid_targets_min": 1887
    },
    {
      "epoch": 1.2432,
      "grad_norm": 0.27179047105633725,
      "learning_rate": 3.7437948482881104e-05,
      "loss": 0.4116,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09206192195415497,
      "step": 195,
      "valid_targets_mean": 4488.4,
      "valid_targets_min": 1445
    },
    {
      "epoch": 1.2752,
      "grad_norm": 0.2901671425980728,
      "learning_rate": 3.721574653466336e-05,
      "loss": 0.4124,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10212432593107224,
      "step": 200,
      "valid_targets_mean": 5330.1,
      "valid_targets_min": 2786
    },
    {
      "epoch": 1.3072,
      "grad_norm": 0.2464328564671245,
      "learning_rate": 3.698502264989903e-05,
      "loss": 0.4072,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1031215563416481,
      "step": 205,
      "valid_targets_mean": 5465.9,
      "valid_targets_min": 3181
    },
    {
      "epoch": 1.3392,
      "grad_norm": 0.2827725046484686,
      "learning_rate": 3.674589103881432e-05,
      "loss": 0.4073,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1113610491156578,
      "step": 210,
      "valid_targets_mean": 4905.2,
      "valid_targets_min": 3405
    },
    {
      "epoch": 1.3712,
      "grad_norm": 0.30317708460592974,
      "learning_rate": 3.64984700735293e-05,
      "loss": 0.4012,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10195107758045197,
      "step": 215,
      "valid_targets_mean": 4707.1,
      "valid_targets_min": 2806
    },
    {
      "epoch": 1.4032,
      "grad_norm": 0.26819515805856425,
      "learning_rate": 3.624288222946273e-05,
      "loss": 0.4043,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1035236120223999,
      "step": 220,
      "valid_targets_mean": 5102.1,
      "valid_targets_min": 2239
    },
    {
      "epoch": 1.4352,
      "grad_norm": 0.2856944378477626,
      "learning_rate": 3.597925402470578e-05,
      "loss": 0.4149,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09435896575450897,
      "step": 225,
      "valid_targets_mean": 4920.2,
      "valid_targets_min": 1738
    },
    {
      "epoch": 1.4672,
      "grad_norm": 0.24727875964553028,
      "learning_rate": 3.570771595739445e-05,
      "loss": 0.406,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.101529560983181,
      "step": 230,
      "valid_targets_mean": 5505.2,
      "valid_targets_min": 2919
    },
    {
      "epoch": 1.4992,
      "grad_norm": 0.26885540380619355,
      "learning_rate": 3.5428402441111964e-05,
      "loss": 0.4034,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0941242128610611,
      "step": 235,
      "valid_targets_mean": 5094.9,
      "valid_targets_min": 3434
    },
    {
      "epoch": 1.5312000000000001,
      "grad_norm": 0.2715774828654331,
      "learning_rate": 3.5141451738352936e-05,
      "loss": 0.397,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09768666326999664,
      "step": 240,
      "valid_targets_mean": 4866.4,
      "valid_targets_min": 1674
    },
    {
      "epoch": 1.5632000000000001,
      "grad_norm": 0.25270229060393723,
      "learning_rate": 3.4847005892082266e-05,
      "loss": 0.4025,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10096217691898346,
      "step": 245,
      "valid_targets_mean": 5329.2,
      "valid_targets_min": 3194
    },
    {
      "epoch": 1.5952,
      "grad_norm": 0.2854130456171178,
      "learning_rate": 3.454521065542273e-05,
      "loss": 0.401,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09778477251529694,
      "step": 250,
      "valid_targets_mean": 4605.8,
      "valid_targets_min": 2859
    },
    {
      "epoch": 1.6272,
      "grad_norm": 0.3104764200129568,
      "learning_rate": 3.423621541950597e-05,
      "loss": 0.404,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10908493399620056,
      "step": 255,
      "valid_targets_mean": 5343.1,
      "valid_targets_min": 2933
    },
    {
      "epoch": 1.6592,
      "grad_norm": 0.27642655790611487,
      "learning_rate": 3.3920173139522664e-05,
      "loss": 0.4038,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10530358552932739,
      "step": 260,
      "valid_targets_mean": 4896.8,
      "valid_targets_min": 2167
    },
    {
      "epoch": 1.6912,
      "grad_norm": 0.28119496367041413,
      "learning_rate": 3.35972402590084e-05,
      "loss": 0.3983,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11041218042373657,
      "step": 265,
      "valid_targets_mean": 5682.6,
      "valid_targets_min": 2949
    },
    {
      "epoch": 1.7231999999999998,
      "grad_norm": 0.26282396375631545,
      "learning_rate": 3.326757663240291e-05,
      "loss": 0.4023,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0962829738855362,
      "step": 270,
      "valid_targets_mean": 5131.3,
      "valid_targets_min": 1477
    },
    {
      "epoch": 1.7551999999999999,
      "grad_norm": 0.2842178924969545,
      "learning_rate": 3.293134544592073e-05,
      "loss": 0.4038,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0953085720539093,
      "step": 275,
      "valid_targets_mean": 4846.5,
      "valid_targets_min": 2602
    },
    {
      "epoch": 1.7872,
      "grad_norm": 0.2869789137432663,
      "learning_rate": 3.258871313677274e-05,
      "loss": 0.4006,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11130845546722412,
      "step": 280,
      "valid_targets_mean": 5805.8,
      "valid_targets_min": 3280
    },
    {
      "epoch": 1.8192,
      "grad_norm": 0.3200883285438623,
      "learning_rate": 3.2239849310778316e-05,
      "loss": 0.3986,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09632055461406708,
      "step": 285,
      "valid_targets_mean": 5025.1,
      "valid_targets_min": 2567
    },
    {
      "epoch": 1.8512,
      "grad_norm": 0.2691895947294203,
      "learning_rate": 3.188492665840909e-05,
      "loss": 0.4008,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09088140726089478,
      "step": 290,
      "valid_targets_mean": 5353.7,
      "valid_targets_min": 2179
    },
    {
      "epoch": 1.8832,
      "grad_norm": 0.2537469004297996,
      "learning_rate": 3.1524120869305726e-05,
      "loss": 0.3956,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11161912977695465,
      "step": 295,
      "valid_targets_mean": 6161.1,
      "valid_targets_min": 3659
    },
    {
      "epoch": 1.9152,
      "grad_norm": 0.2674684777522305,
      "learning_rate": 3.11576105453101e-05,
      "loss": 0.4008,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09953461587429047,
      "step": 300,
      "valid_targets_mean": 4538.9,
      "valid_targets_min": 1923
    },
    {
      "epoch": 1.9472,
      "grad_norm": 0.27521471996294616,
      "learning_rate": 3.0785577112055916e-05,
      "loss": 0.4029,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09671811759471893,
      "step": 305,
      "valid_targets_mean": 4993.0,
      "valid_targets_min": 1793
    },
    {
      "epoch": 1.9792,
      "grad_norm": 0.2861590648650816,
      "learning_rate": 3.040820472916153e-05,
      "loss": 0.4006,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10373485088348389,
      "step": 310,
      "valid_targets_mean": 5229.4,
      "valid_targets_min": 1468
    },
    {
      "epoch": 2.0064,
      "grad_norm": 0.2917886264361851,
      "learning_rate": 3.002568019906939e-05,
      "loss": 0.3894,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09146748483181,
      "step": 315,
      "valid_targets_mean": 4844.1,
      "valid_targets_min": 2742
    },
    {
      "epoch": 2.0384,
      "grad_norm": 0.2686711069342525,
      "learning_rate": 2.963819287457733e-05,
      "loss": 0.394,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09616793692111969,
      "step": 320,
      "valid_targets_mean": 4960.5,
      "valid_targets_min": 1487
    },
    {
      "epoch": 2.0704,
      "grad_norm": 0.27640522156112673,
      "learning_rate": 2.924593456510733e-05,
      "loss": 0.3915,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09300155192613602,
      "step": 325,
      "valid_targets_mean": 4701.9,
      "valid_targets_min": 2016
    },
    {
      "epoch": 2.1024,
      "grad_norm": 0.27372794504488607,
      "learning_rate": 2.8849099441758306e-05,
      "loss": 0.3913,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.11577819287776947,
      "step": 330,
      "valid_targets_mean": 5657.6,
      "valid_targets_min": 2743
    },
    {
      "epoch": 2.1344,
      "grad_norm": 0.2706996358221534,
      "learning_rate": 2.844788394118979e-05,
      "loss": 0.3893,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09498875588178635,
      "step": 335,
      "valid_targets_mean": 4847.1,
      "valid_targets_min": 2247
    },
    {
      "epoch": 2.1664,
      "grad_norm": 0.2541610009797207,
      "learning_rate": 2.8042486668384164e-05,
      "loss": 0.3887,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10963079333305359,
      "step": 340,
      "valid_targets_mean": 5559.4,
      "valid_targets_min": 3098
    },
    {
      "epoch": 2.1984,
      "grad_norm": 0.27395615254158723,
      "learning_rate": 2.7633108298335582e-05,
      "loss": 0.3946,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09737113118171692,
      "step": 345,
      "valid_targets_mean": 4693.8,
      "valid_targets_min": 2866
    },
    {
      "epoch": 2.2304,
      "grad_norm": 0.2607202712442059,
      "learning_rate": 2.721995147671416e-05,
      "loss": 0.388,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07866384834051132,
      "step": 350,
      "valid_targets_mean": 4349.5,
      "valid_targets_min": 1280
    },
    {
      "epoch": 2.2624,
      "grad_norm": 0.2659711633615929,
      "learning_rate": 2.68032207195547e-05,
      "loss": 0.387,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10025221109390259,
      "step": 355,
      "valid_targets_mean": 5812.4,
      "valid_targets_min": 3468
    },
    {
      "epoch": 2.2944,
      "grad_norm": 0.2609745674927213,
      "learning_rate": 2.6383122312019604e-05,
      "loss": 0.3935,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08885324746370316,
      "step": 360,
      "valid_targets_mean": 4752.0,
      "valid_targets_min": 1854
    },
    {
      "epoch": 2.3264,
      "grad_norm": 0.26181806015994075,
      "learning_rate": 2.595986420628597e-05,
      "loss": 0.3912,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10182733833789825,
      "step": 365,
      "valid_targets_mean": 4989.9,
      "valid_targets_min": 2267
    },
    {
      "epoch": 2.3584,
      "grad_norm": 0.25967767358969746,
      "learning_rate": 2.5533655918607573e-05,
      "loss": 0.3911,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0928272157907486,
      "step": 370,
      "valid_targets_mean": 5317.7,
      "valid_targets_min": 3002
    },
    {
      "epoch": 2.3904,
      "grad_norm": 0.2845917477486873,
      "learning_rate": 2.510470842560259e-05,
      "loss": 0.3927,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09599171578884125,
      "step": 375,
      "valid_targets_mean": 4927.9,
      "valid_targets_min": 2191
    },
    {
      "epoch": 2.4224,
      "grad_norm": 0.25308081819402717,
      "learning_rate": 2.467323405981841e-05,
      "loss": 0.3884,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10373442620038986,
      "step": 380,
      "valid_targets_mean": 5235.2,
      "valid_targets_min": 3074
    },
    {
      "epoch": 2.4544,
      "grad_norm": 0.2451835858290873,
      "learning_rate": 2.423944640462533e-05,
      "loss": 0.3875,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09769438952207565,
      "step": 385,
      "valid_targets_mean": 5193.5,
      "valid_targets_min": 2405
    },
    {
      "epoch": 2.4864,
      "grad_norm": 0.26963311715651334,
      "learning_rate": 2.3803560188490968e-05,
      "loss": 0.3934,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08894319832324982,
      "step": 390,
      "valid_targets_mean": 4663.1,
      "valid_targets_min": 3580
    },
    {
      "epoch": 2.5183999999999997,
      "grad_norm": 0.25513095898821697,
      "learning_rate": 2.336579117868789e-05,
      "loss": 0.3875,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10441608726978302,
      "step": 395,
      "valid_targets_mean": 5480.6,
      "valid_targets_min": 3288
    },
    {
      "epoch": 2.5504,
      "grad_norm": 0.26536634409119253,
      "learning_rate": 2.292635607448711e-05,
      "loss": 0.3895,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09016616642475128,
      "step": 400,
      "valid_targets_mean": 4590.4,
      "valid_targets_min": 2208
    },
    {
      "epoch": 2.5824,
      "grad_norm": 0.39471222196553674,
      "learning_rate": 2.248547239989008e-05,
      "loss": 0.3827,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09323616325855255,
      "step": 405,
      "valid_targets_mean": 5016.6,
      "valid_targets_min": 1556
    },
    {
      "epoch": 2.6144,
      "grad_norm": 0.2606139190879748,
      "learning_rate": 2.204335839595255e-05,
      "loss": 0.3808,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09083975106477737,
      "step": 410,
      "valid_targets_mean": 5140.6,
      "valid_targets_min": 2865
    },
    {
      "epoch": 2.6464,
      "grad_norm": 0.2877965557679335,
      "learning_rate": 2.1600232912753452e-05,
      "loss": 0.3871,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09244799613952637,
      "step": 415,
      "valid_targets_mean": 5237.1,
      "valid_targets_min": 2819
    },
    {
      "epoch": 2.6784,
      "grad_norm": 0.24868387551811333,
      "learning_rate": 2.1156315301062293e-05,
      "loss": 0.388,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09515254944562912,
      "step": 420,
      "valid_targets_mean": 5689.1,
      "valid_targets_min": 3369
    },
    {
      "epoch": 2.7104,
      "grad_norm": 0.2906948619356932,
      "learning_rate": 2.0711825303758712e-05,
      "loss": 0.3892,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10640858113765717,
      "step": 425,
      "valid_targets_mean": 5483.8,
      "valid_targets_min": 3029
    },
    {
      "epoch": 2.7424,
      "grad_norm": 0.2524550152768104,
      "learning_rate": 2.0266982947057962e-05,
      "loss": 0.3903,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0917065218091011,
      "step": 430,
      "valid_targets_mean": 4898.9,
      "valid_targets_min": 1945
    },
    {
      "epoch": 2.7744,
      "grad_norm": 0.26647428909618837,
      "learning_rate": 1.9822008431596083e-05,
      "loss": 0.3911,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09158390760421753,
      "step": 435,
      "valid_targets_mean": 5036.7,
      "valid_targets_min": 2309
    },
    {
      "epoch": 2.8064,
      "grad_norm": 0.24939838895355831,
      "learning_rate": 1.937712202342881e-05,
      "loss": 0.3891,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08817063271999359,
      "step": 440,
      "valid_targets_mean": 4600.1,
      "valid_targets_min": 2310
    },
    {
      "epoch": 2.8384,
      "grad_norm": 0.25153383248016903,
      "learning_rate": 1.8932543944998037e-05,
      "loss": 0.3841,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09152311086654663,
      "step": 445,
      "valid_targets_mean": 5435.9,
      "valid_targets_min": 2855
    },
    {
      "epoch": 2.8704,
      "grad_norm": 0.2309646725764094,
      "learning_rate": 1.8488494266119877e-05,
      "loss": 0.3857,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09221877157688141,
      "step": 450,
      "valid_targets_mean": 5408.8,
      "valid_targets_min": 1571
    },
    {
      "epoch": 2.9024,
      "grad_norm": 0.2395570853377421,
      "learning_rate": 1.804519279504834e-05,
      "loss": 0.3914,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10076689720153809,
      "step": 455,
      "valid_targets_mean": 5394.1,
      "valid_targets_min": 2743
    },
    {
      "epoch": 2.9344,
      "grad_norm": 0.269070593097617,
      "learning_rate": 1.7602858969668365e-05,
      "loss": 0.3852,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.07979879528284073,
      "step": 460,
      "valid_targets_mean": 4112.2,
      "valid_targets_min": 1739
    },
    {
      "epoch": 2.9664,
      "grad_norm": 0.2837163704255437,
      "learning_rate": 1.716171174887231e-05,
      "loss": 0.389,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09907828271389008,
      "step": 465,
      "valid_targets_mean": 5093.3,
      "valid_targets_min": 1944
    },
    {
      "epoch": 2.9984,
      "grad_norm": 0.2709088957473118,
      "learning_rate": 1.6721969504173484e-05,
      "loss": 0.3844,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09028175473213196,
      "step": 470,
      "valid_targets_mean": 4735.1,
      "valid_targets_min": 2318
    },
    {
      "epoch": 3.0256,
      "grad_norm": 0.2827116313476474,
      "learning_rate": 1.628384991161041e-05,
      "loss": 0.3812,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08898613601922989,
      "step": 475,
      "valid_targets_mean": 4397.9,
      "valid_targets_min": 2325
    },
    {
      "epoch": 3.0576,
      "grad_norm": 0.2700317966307477,
      "learning_rate": 1.5847569843995452e-05,
      "loss": 0.3798,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10498139262199402,
      "step": 480,
      "valid_targets_mean": 5153.1,
      "valid_targets_min": 3170
    },
    {
      "epoch": 3.0896,
      "grad_norm": 0.27160569678940877,
      "learning_rate": 1.5413345263560922e-05,
      "loss": 0.3797,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09167727082967758,
      "step": 485,
      "valid_targets_mean": 4856.2,
      "valid_targets_min": 1445
    },
    {
      "epoch": 3.1216,
      "grad_norm": 0.2457536697129015,
      "learning_rate": 1.4981391115056032e-05,
      "loss": 0.3766,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1018785610795021,
      "step": 490,
      "valid_targets_mean": 5481.6,
      "valid_targets_min": 3322
    },
    {
      "epoch": 3.1536,
      "grad_norm": 0.2631344841371176,
      "learning_rate": 1.455192121934748e-05,
      "loss": 0.3764,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10155577957630157,
      "step": 495,
      "valid_targets_mean": 5058.2,
      "valid_targets_min": 2192
    },
    {
      "epoch": 3.1856,
      "grad_norm": 0.2566124726202203,
      "learning_rate": 1.4125148167576303e-05,
      "loss": 0.384,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.1064523309469223,
      "step": 500,
      "valid_targets_mean": 5357.4,
      "valid_targets_min": 2590
    },
    {
      "epoch": 3.2176,
      "grad_norm": 0.2531334902826478,
      "learning_rate": 1.3701283215923563e-05,
      "loss": 0.3866,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09294597059488297,
      "step": 505,
      "valid_targets_mean": 5169.3,
      "valid_targets_min": 2382
    },
    {
      "epoch": 3.2496,
      "grad_norm": 0.23525196551056587,
      "learning_rate": 1.328053618103677e-05,
      "loss": 0.376,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08874399214982986,
      "step": 510,
      "valid_targets_mean": 4844.4,
      "valid_targets_min": 1842
    },
    {
      "epoch": 3.2816,
      "grad_norm": 0.2737109957875444,
      "learning_rate": 1.2863115336168916e-05,
      "loss": 0.3849,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09585750848054886,
      "step": 515,
      "valid_targets_mean": 4731.6,
      "valid_targets_min": 1612
    },
    {
      "epoch": 3.3136,
      "grad_norm": 0.23954548298620934,
      "learning_rate": 1.2449227308081509e-05,
      "loss": 0.3805,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0933152511715889,
      "step": 520,
      "valid_targets_mean": 4974.9,
      "valid_targets_min": 2857
    },
    {
      "epoch": 3.3456,
      "grad_norm": 0.24950456760336664,
      "learning_rate": 1.2039076974762587e-05,
      "loss": 0.3846,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10132192820310593,
      "step": 525,
      "valid_targets_mean": 5588.5,
      "valid_targets_min": 2903
    },
    {
      "epoch": 3.3776,
      "grad_norm": 0.24698701379700322,
      "learning_rate": 1.163286736401044e-05,
      "loss": 0.3812,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08419165760278702,
      "step": 530,
      "valid_targets_mean": 4778.9,
      "valid_targets_min": 1750
    },
    {
      "epoch": 3.4096,
      "grad_norm": 0.2594736038626073,
      "learning_rate": 1.123079955293322e-05,
      "loss": 0.3942,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10506877303123474,
      "step": 535,
      "valid_targets_mean": 4694.1,
      "valid_targets_min": 1293
    },
    {
      "epoch": 3.4416,
      "grad_norm": 0.24024356133504324,
      "learning_rate": 1.0833072568414037e-05,
      "loss": 0.3825,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09017220139503479,
      "step": 540,
      "valid_targets_mean": 4607.1,
      "valid_targets_min": 2087
    },
    {
      "epoch": 3.4736000000000002,
      "grad_norm": 0.25582028963503706,
      "learning_rate": 1.0439883288591057e-05,
      "loss": 0.3822,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09501463174819946,
      "step": 545,
      "valid_targets_mean": 4746.0,
      "valid_targets_min": 2419
    },
    {
      "epoch": 3.5056000000000003,
      "grad_norm": 0.25629545313149577,
      "learning_rate": 1.0051426345401202e-05,
      "loss": 0.3782,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.090950608253479,
      "step": 550,
      "valid_targets_mean": 4597.5,
      "valid_targets_min": 1687
    },
    {
      "epoch": 3.5376,
      "grad_norm": 0.2504937990170045,
      "learning_rate": 9.667894028235704e-06,
      "loss": 0.3784,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0809982493519783,
      "step": 555,
      "valid_targets_mean": 4798.9,
      "valid_targets_min": 2752
    },
    {
      "epoch": 3.5696,
      "grad_norm": 0.25603058098352444,
      "learning_rate": 9.289476188755315e-06,
      "loss": 0.3814,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0900259017944336,
      "step": 560,
      "valid_targets_mean": 4771.6,
      "valid_targets_min": 2304
    },
    {
      "epoch": 3.6016,
      "grad_norm": 0.2748497981490769,
      "learning_rate": 8.916360146912122e-06,
      "loss": 0.3787,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.101971834897995,
      "step": 565,
      "valid_targets_mean": 5472.9,
      "valid_targets_min": 3183
    },
    {
      "epoch": 3.6336,
      "grad_norm": 0.2960903656486721,
      "learning_rate": 8.548730598224646e-06,
      "loss": 0.385,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10061405599117279,
      "step": 570,
      "valid_targets_mean": 5165.8,
      "valid_targets_min": 1862
    },
    {
      "epoch": 3.6656,
      "grad_norm": 0.2439653717975467,
      "learning_rate": 8.186769522352053e-06,
      "loss": 0.3842,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0856606662273407,
      "step": 575,
      "valid_targets_mean": 4714.6,
      "valid_targets_min": 1932
    },
    {
      "epoch": 3.6976,
      "grad_norm": 0.23035154697789606,
      "learning_rate": 7.830656093012714e-06,
      "loss": 0.3839,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09490450471639633,
      "step": 580,
      "valid_targets_mean": 5547.4,
      "valid_targets_min": 3290
    },
    {
      "epoch": 3.7296,
      "grad_norm": 0.24783968662120118,
      "learning_rate": 7.480566589291696e-06,
      "loss": 0.379,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10084158927202225,
      "step": 585,
      "valid_targets_mean": 6062.2,
      "valid_targets_min": 3361
    },
    {
      "epoch": 3.7616,
      "grad_norm": 0.23597209171825934,
      "learning_rate": 7.1366743083812285e-06,
      "loss": 0.3801,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08867701888084412,
      "step": 590,
      "valid_targets_mean": 5557.6,
      "valid_targets_min": 2364
    },
    {
      "epoch": 3.7936,
      "grad_norm": 0.2533568999438392,
      "learning_rate": 6.799149479797101e-06,
      "loss": 0.3751,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09318187832832336,
      "step": 595,
      "valid_targets_mean": 4899.3,
      "valid_targets_min": 2059
    },
    {
      "epoch": 3.8256,
      "grad_norm": 0.2508770160677488,
      "learning_rate": 6.4681591811137e-06,
      "loss": 0.3769,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09285284578800201,
      "step": 600,
      "valid_targets_mean": 4989.8,
      "valid_targets_min": 1581
    },
    {
      "epoch": 3.8576,
      "grad_norm": 0.23622472978985473,
      "learning_rate": 6.143867255259197e-06,
      "loss": 0.3735,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08706752210855484,
      "step": 605,
      "valid_targets_mean": 4942.9,
      "valid_targets_min": 2892
    },
    {
      "epoch": 3.8895999999999997,
      "grad_norm": 0.23126551300776224,
      "learning_rate": 5.8264342294119504e-06,
      "loss": 0.3821,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08980019390583038,
      "step": 610,
      "valid_targets_mean": 4755.7,
      "valid_targets_min": 2766
    },
    {
      "epoch": 3.9215999999999998,
      "grad_norm": 0.2284966594454513,
      "learning_rate": 5.516017235538258e-06,
      "loss": 0.3769,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09404298663139343,
      "step": 615,
      "valid_targets_mean": 5714.6,
      "valid_targets_min": 3685
    },
    {
      "epoch": 3.9536,
      "grad_norm": 0.24562408527658194,
      "learning_rate": 5.212769932610695e-06,
      "loss": 0.3752,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10075512528419495,
      "step": 620,
      "valid_targets_mean": 5231.1,
      "valid_targets_min": 3311
    },
    {
      "epoch": 3.9856,
      "grad_norm": 0.2292179176251408,
      "learning_rate": 4.916842430545681e-06,
      "loss": 0.3734,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08822321146726608,
      "step": 625,
      "valid_targets_mean": 5240.1,
      "valid_targets_min": 3467
    },
    {
      "epoch": 4.0128,
      "grad_norm": 0.23411898804377318,
      "learning_rate": 4.628381215897837e-06,
      "loss": 0.3764,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08666276931762695,
      "step": 630,
      "valid_targets_mean": 4693.2,
      "valid_targets_min": 1855
    },
    {
      "epoch": 4.0448,
      "grad_norm": 0.23060643397343694,
      "learning_rate": 4.347529079347914e-06,
      "loss": 0.3853,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09197352081537247,
      "step": 635,
      "valid_targets_mean": 5232.1,
      "valid_targets_min": 1794
    },
    {
      "epoch": 4.0768,
      "grad_norm": 0.22819232578093349,
      "learning_rate": 4.074425045020247e-06,
      "loss": 0.3785,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09138046950101852,
      "step": 640,
      "valid_targets_mean": 5464.4,
      "valid_targets_min": 1880
    },
    {
      "epoch": 4.1088,
      "grad_norm": 0.23147490384309233,
      "learning_rate": 3.8092043016646487e-06,
      "loss": 0.3707,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08915231376886368,
      "step": 645,
      "valid_targets_mean": 5191.8,
      "valid_targets_min": 3102
    },
    {
      "epoch": 4.1408,
      "grad_norm": 0.25385017338015203,
      "learning_rate": 3.551998135736867e-06,
      "loss": 0.3785,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09559452533721924,
      "step": 650,
      "valid_targets_mean": 5393.1,
      "valid_targets_min": 2150
    },
    {
      "epoch": 4.1728,
      "grad_norm": 0.2320722979585512,
      "learning_rate": 3.3029338664107267e-06,
      "loss": 0.3677,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0835668295621872,
      "step": 655,
      "valid_targets_mean": 4711.2,
      "valid_targets_min": 2111
    },
    {
      "epoch": 4.2048,
      "grad_norm": 0.23664960938578167,
      "learning_rate": 3.0621347825540625e-06,
      "loss": 0.373,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10337899625301361,
      "step": 660,
      "valid_targets_mean": 5375.4,
      "valid_targets_min": 3218
    },
    {
      "epoch": 4.2368,
      "grad_norm": 0.24417105041207457,
      "learning_rate": 2.8297200816997183e-06,
      "loss": 0.3715,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09414561837911606,
      "step": 665,
      "valid_targets_mean": 4996.1,
      "valid_targets_min": 3554
    },
    {
      "epoch": 4.2688,
      "grad_norm": 0.23450850680611232,
      "learning_rate": 2.605804811041803e-06,
      "loss": 0.3787,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.107195645570755,
      "step": 670,
      "valid_targets_mean": 5131.9,
      "valid_targets_min": 2741
    },
    {
      "epoch": 4.3008,
      "grad_norm": 0.2430621302240459,
      "learning_rate": 2.390499810486351e-06,
      "loss": 0.3752,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10133292526006699,
      "step": 675,
      "valid_targets_mean": 5571.7,
      "valid_targets_min": 2870
    },
    {
      "epoch": 4.3328,
      "grad_norm": 0.23138694256795886,
      "learning_rate": 2.183911657784685e-06,
      "loss": 0.3785,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10721897333860397,
      "step": 680,
      "valid_targets_mean": 6100.7,
      "valid_targets_min": 2130
    },
    {
      "epoch": 4.3648,
      "grad_norm": 0.23313740461049393,
      "learning_rate": 1.986142615776532e-06,
      "loss": 0.3774,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08411256223917007,
      "step": 685,
      "valid_targets_mean": 4805.2,
      "valid_targets_min": 3090
    },
    {
      "epoch": 4.3968,
      "grad_norm": 0.24400690620390336,
      "learning_rate": 1.7972905817690644e-06,
      "loss": 0.3792,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08684651553630829,
      "step": 690,
      "valid_targets_mean": 4691.9,
      "valid_targets_min": 2029
    },
    {
      "epoch": 4.4288,
      "grad_norm": 0.21672504221878636,
      "learning_rate": 1.617449039076955e-06,
      "loss": 0.3781,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08824565261602402,
      "step": 695,
      "valid_targets_mean": 5133.9,
      "valid_targets_min": 1579
    },
    {
      "epoch": 4.4608,
      "grad_norm": 0.2267660538797704,
      "learning_rate": 1.4467070107473413e-06,
      "loss": 0.3735,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08960177004337311,
      "step": 700,
      "valid_targets_mean": 5434.2,
      "valid_targets_min": 2051
    },
    {
      "epoch": 4.4928,
      "grad_norm": 0.2385381710568613,
      "learning_rate": 1.2851490154926816e-06,
      "loss": 0.3787,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09955689311027527,
      "step": 705,
      "valid_targets_mean": 4987.2,
      "valid_targets_min": 1847
    },
    {
      "epoch": 4.5248,
      "grad_norm": 0.23981043467013477,
      "learning_rate": 1.1328550258533211e-06,
      "loss": 0.3772,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08953072130680084,
      "step": 710,
      "valid_targets_mean": 4642.4,
      "valid_targets_min": 2694
    },
    {
      "epoch": 4.5568,
      "grad_norm": 0.24641092742931273,
      "learning_rate": 9.899004286103953e-07,
      "loss": 0.3704,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09114815294742584,
      "step": 715,
      "valid_targets_mean": 5159.6,
      "valid_targets_min": 1955
    },
    {
      "epoch": 4.5888,
      "grad_norm": 0.22215771099573642,
      "learning_rate": 8.5635598746876e-07,
      "loss": 0.3742,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10481426864862442,
      "step": 720,
      "valid_targets_mean": 6288.1,
      "valid_targets_min": 2885
    },
    {
      "epoch": 4.6208,
      "grad_norm": 0.242262077962136,
      "learning_rate": 7.32287808028389e-07,
      "loss": 0.3793,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.0912296324968338,
      "step": 725,
      "valid_targets_mean": 4778.2,
      "valid_targets_min": 2527
    },
    {
      "epoch": 4.6528,
      "grad_norm": 0.22046299469725933,
      "learning_rate": 6.177573050615327e-07,
      "loss": 0.3791,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09594270586967468,
      "step": 730,
      "valid_targets_mean": 6026.2,
      "valid_targets_min": 3840
    },
    {
      "epoch": 4.6848,
      "grad_norm": 0.2250335766046042,
      "learning_rate": 5.128211721119213e-07,
      "loss": 0.3718,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10690298676490784,
      "step": 735,
      "valid_targets_mean": 5806.8,
      "valid_targets_min": 3778
    },
    {
      "epoch": 4.7168,
      "grad_norm": 0.25141836037445864,
      "learning_rate": 4.175313534309755e-07,
      "loss": 0.382,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08581297099590302,
      "step": 740,
      "valid_targets_mean": 4431.0,
      "valid_targets_min": 1581
    },
    {
      "epoch": 4.7488,
      "grad_norm": 0.23111767245348003,
      "learning_rate": 3.319350182649861e-07,
      "loss": 0.3774,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09311626851558685,
      "step": 745,
      "valid_targets_mean": 5526.9,
      "valid_targets_min": 4201
    },
    {
      "epoch": 4.7808,
      "grad_norm": 0.22439655656891788,
      "learning_rate": 2.560745375059392e-07,
      "loss": 0.377,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09380144625902176,
      "step": 750,
      "valid_targets_mean": 5235.3,
      "valid_targets_min": 2139
    },
    {
      "epoch": 4.8128,
      "grad_norm": 0.2256282889501494,
      "learning_rate": 1.8998746271758016e-07,
      "loss": 0.3754,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10746807605028152,
      "step": 755,
      "valid_targets_mean": 5267.0,
      "valid_targets_min": 2381
    },
    {
      "epoch": 4.8448,
      "grad_norm": 0.22721329000073195,
      "learning_rate": 1.337065075470778e-07,
      "loss": 0.3809,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.10962164402008057,
      "step": 760,
      "valid_targets_mean": 6532.8,
      "valid_targets_min": 3733
    },
    {
      "epoch": 4.8768,
      "grad_norm": 0.225285168321463,
      "learning_rate": 8.725953153150279e-08,
      "loss": 0.3766,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09537424147129059,
      "step": 765,
      "valid_targets_mean": 5370.6,
      "valid_targets_min": 1862
    },
    {
      "epoch": 4.9088,
      "grad_norm": 0.27094629674938736,
      "learning_rate": 5.066952630711886e-08,
      "loss": 0.3764,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09874732047319412,
      "step": 770,
      "valid_targets_mean": 5485.8,
      "valid_targets_min": 2540
    },
    {
      "epoch": 4.9408,
      "grad_norm": 0.22063049624652012,
      "learning_rate": 2.3954604228342283e-08,
      "loss": 0.3744,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.09011753648519516,
      "step": 775,
      "valid_targets_mean": 5431.5,
      "valid_targets_min": 3302
    },
    {
      "epoch": 4.9728,
      "grad_norm": 0.23300194382666267,
      "learning_rate": 7.12798940197601e-09,
      "loss": 0.3799,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.08871814608573914,
      "step": 780,
      "valid_targets_mean": 4908.0,
      "valid_targets_min": 1687
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.4472911577753402,
      "learning_rate": 1.9801114115480802e-10,
      "loss": 0.3798,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.37919333577156067,
      "step": 785,
      "valid_targets_mean": 5141.6,
      "valid_targets_min": 2704
    },
    {
      "epoch": 5.0,
      "loss_nan_ranks": 0,
      "loss_rank_avg": 0.37919333577156067,
      "step": 785,
      "total_flos": 2.363785022508368e+18,
      "train_loss": 0.4118077915944871,
      "train_runtime": 19452.6911,
      "train_samples_per_second": 2.57,
      "train_steps_per_second": 0.04,
      "valid_targets_mean": 5141.6,
      "valid_targets_min": 2704
    }
  ],
  "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": 2.363785022508368e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}