4e220f32fa
Model: laion/exp-syh-r2egym-askllm-constrained_glm_4-7_traces_jupiter_cleaned-exp-gfi-swesmith-random-fi Source: Original Platform
3025 lines
84 KiB
JSON
3025 lines
84 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 7.0,
|
|
"eval_steps": 500,
|
|
"global_step": 1358,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.02577319587628866,
|
|
"grad_norm": 19.361754624645346,
|
|
"learning_rate": 1.1764705882352942e-06,
|
|
"loss": 0.7739,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2674316167831421,
|
|
"step": 5,
|
|
"valid_targets_mean": 6342.4,
|
|
"valid_targets_min": 796
|
|
},
|
|
{
|
|
"epoch": 0.05154639175257732,
|
|
"grad_norm": 14.776293018924607,
|
|
"learning_rate": 2.647058823529412e-06,
|
|
"loss": 0.747,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26964008808135986,
|
|
"step": 10,
|
|
"valid_targets_mean": 6470.9,
|
|
"valid_targets_min": 589
|
|
},
|
|
{
|
|
"epoch": 0.07731958762886598,
|
|
"grad_norm": 6.299200585171311,
|
|
"learning_rate": 4.11764705882353e-06,
|
|
"loss": 0.6522,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19556403160095215,
|
|
"step": 15,
|
|
"valid_targets_mean": 5640.0,
|
|
"valid_targets_min": 568
|
|
},
|
|
{
|
|
"epoch": 0.10309278350515463,
|
|
"grad_norm": 2.1706585789929167,
|
|
"learning_rate": 5.588235294117647e-06,
|
|
"loss": 0.5807,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20123359560966492,
|
|
"step": 20,
|
|
"valid_targets_mean": 5792.0,
|
|
"valid_targets_min": 421
|
|
},
|
|
{
|
|
"epoch": 0.12886597938144329,
|
|
"grad_norm": 1.281667358551557,
|
|
"learning_rate": 7.058823529411766e-06,
|
|
"loss": 0.5389,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18983393907546997,
|
|
"step": 25,
|
|
"valid_targets_mean": 6190.2,
|
|
"valid_targets_min": 637
|
|
},
|
|
{
|
|
"epoch": 0.15463917525773196,
|
|
"grad_norm": 0.9543622306940317,
|
|
"learning_rate": 8.529411764705883e-06,
|
|
"loss": 0.5272,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1809738576412201,
|
|
"step": 30,
|
|
"valid_targets_mean": 5972.4,
|
|
"valid_targets_min": 657
|
|
},
|
|
{
|
|
"epoch": 0.18041237113402062,
|
|
"grad_norm": 0.6809974107949935,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.496,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17243939638137817,
|
|
"step": 35,
|
|
"valid_targets_mean": 6295.2,
|
|
"valid_targets_min": 687
|
|
},
|
|
{
|
|
"epoch": 0.20618556701030927,
|
|
"grad_norm": 0.5174417340859547,
|
|
"learning_rate": 1.1470588235294118e-05,
|
|
"loss": 0.487,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12923963367938995,
|
|
"step": 40,
|
|
"valid_targets_mean": 5336.0,
|
|
"valid_targets_min": 587
|
|
},
|
|
{
|
|
"epoch": 0.23195876288659795,
|
|
"grad_norm": 0.4256636161095827,
|
|
"learning_rate": 1.2941176470588238e-05,
|
|
"loss": 0.4661,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1391327828168869,
|
|
"step": 45,
|
|
"valid_targets_mean": 6423.3,
|
|
"valid_targets_min": 567
|
|
},
|
|
{
|
|
"epoch": 0.25773195876288657,
|
|
"grad_norm": 0.3673686393591969,
|
|
"learning_rate": 1.4411764705882353e-05,
|
|
"loss": 0.4478,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13855600357055664,
|
|
"step": 50,
|
|
"valid_targets_mean": 5495.0,
|
|
"valid_targets_min": 540
|
|
},
|
|
{
|
|
"epoch": 0.28350515463917525,
|
|
"grad_norm": 0.28758629083178733,
|
|
"learning_rate": 1.5882352941176473e-05,
|
|
"loss": 0.4307,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12716147303581238,
|
|
"step": 55,
|
|
"valid_targets_mean": 5351.4,
|
|
"valid_targets_min": 455
|
|
},
|
|
{
|
|
"epoch": 0.30927835051546393,
|
|
"grad_norm": 0.2803915310544906,
|
|
"learning_rate": 1.735294117647059e-05,
|
|
"loss": 0.4226,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13830243051052094,
|
|
"step": 60,
|
|
"valid_targets_mean": 5775.8,
|
|
"valid_targets_min": 526
|
|
},
|
|
{
|
|
"epoch": 0.33505154639175255,
|
|
"grad_norm": 0.2975359105410568,
|
|
"learning_rate": 1.8823529411764708e-05,
|
|
"loss": 0.4034,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13560298085212708,
|
|
"step": 65,
|
|
"valid_targets_mean": 6934.9,
|
|
"valid_targets_min": 619
|
|
},
|
|
{
|
|
"epoch": 0.36082474226804123,
|
|
"grad_norm": 0.25312341649003284,
|
|
"learning_rate": 2.0294117647058825e-05,
|
|
"loss": 0.4074,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13240735232830048,
|
|
"step": 70,
|
|
"valid_targets_mean": 6072.6,
|
|
"valid_targets_min": 575
|
|
},
|
|
{
|
|
"epoch": 0.3865979381443299,
|
|
"grad_norm": 0.2327782906957274,
|
|
"learning_rate": 2.1764705882352943e-05,
|
|
"loss": 0.3876,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12153078615665436,
|
|
"step": 75,
|
|
"valid_targets_mean": 5931.8,
|
|
"valid_targets_min": 770
|
|
},
|
|
{
|
|
"epoch": 0.41237113402061853,
|
|
"grad_norm": 0.23209410691794588,
|
|
"learning_rate": 2.3235294117647064e-05,
|
|
"loss": 0.376,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12323503196239471,
|
|
"step": 80,
|
|
"valid_targets_mean": 5867.5,
|
|
"valid_targets_min": 627
|
|
},
|
|
{
|
|
"epoch": 0.4381443298969072,
|
|
"grad_norm": 0.22608571307773745,
|
|
"learning_rate": 2.470588235294118e-05,
|
|
"loss": 0.3845,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1206689402461052,
|
|
"step": 85,
|
|
"valid_targets_mean": 5408.6,
|
|
"valid_targets_min": 674
|
|
},
|
|
{
|
|
"epoch": 0.4639175257731959,
|
|
"grad_norm": 0.6310853681606622,
|
|
"learning_rate": 2.6176470588235295e-05,
|
|
"loss": 0.3754,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11591410636901855,
|
|
"step": 90,
|
|
"valid_targets_mean": 5860.6,
|
|
"valid_targets_min": 629
|
|
},
|
|
{
|
|
"epoch": 0.4896907216494845,
|
|
"grad_norm": 0.2303573287176781,
|
|
"learning_rate": 2.7647058823529413e-05,
|
|
"loss": 0.3617,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10666414350271225,
|
|
"step": 95,
|
|
"valid_targets_mean": 5568.6,
|
|
"valid_targets_min": 484
|
|
},
|
|
{
|
|
"epoch": 0.5154639175257731,
|
|
"grad_norm": 0.2302775696607361,
|
|
"learning_rate": 2.911764705882353e-05,
|
|
"loss": 0.3555,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13573025166988373,
|
|
"step": 100,
|
|
"valid_targets_mean": 6281.9,
|
|
"valid_targets_min": 614
|
|
},
|
|
{
|
|
"epoch": 0.5412371134020618,
|
|
"grad_norm": 0.21637067325838502,
|
|
"learning_rate": 3.0588235294117644e-05,
|
|
"loss": 0.3549,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0937468409538269,
|
|
"step": 105,
|
|
"valid_targets_mean": 5264.0,
|
|
"valid_targets_min": 542
|
|
},
|
|
{
|
|
"epoch": 0.5670103092783505,
|
|
"grad_norm": 0.23809642219447041,
|
|
"learning_rate": 3.205882352941177e-05,
|
|
"loss": 0.344,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10166369378566742,
|
|
"step": 110,
|
|
"valid_targets_mean": 6304.4,
|
|
"valid_targets_min": 524
|
|
},
|
|
{
|
|
"epoch": 0.5927835051546392,
|
|
"grad_norm": 0.23288667622755851,
|
|
"learning_rate": 3.3529411764705886e-05,
|
|
"loss": 0.3511,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10661657154560089,
|
|
"step": 115,
|
|
"valid_targets_mean": 5671.1,
|
|
"valid_targets_min": 632
|
|
},
|
|
{
|
|
"epoch": 0.6185567010309279,
|
|
"grad_norm": 0.2484689669811016,
|
|
"learning_rate": 3.5000000000000004e-05,
|
|
"loss": 0.341,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11113717406988144,
|
|
"step": 120,
|
|
"valid_targets_mean": 5165.4,
|
|
"valid_targets_min": 647
|
|
},
|
|
{
|
|
"epoch": 0.6443298969072165,
|
|
"grad_norm": 0.2482938299277476,
|
|
"learning_rate": 3.647058823529412e-05,
|
|
"loss": 0.3343,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10907314717769623,
|
|
"step": 125,
|
|
"valid_targets_mean": 5740.9,
|
|
"valid_targets_min": 541
|
|
},
|
|
{
|
|
"epoch": 0.6701030927835051,
|
|
"grad_norm": 0.25658063294793965,
|
|
"learning_rate": 3.794117647058824e-05,
|
|
"loss": 0.3471,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1199379414319992,
|
|
"step": 130,
|
|
"valid_targets_mean": 5379.9,
|
|
"valid_targets_min": 538
|
|
},
|
|
{
|
|
"epoch": 0.6958762886597938,
|
|
"grad_norm": 0.280087729238946,
|
|
"learning_rate": 3.9411764705882356e-05,
|
|
"loss": 0.3413,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11225313693284988,
|
|
"step": 135,
|
|
"valid_targets_mean": 5651.6,
|
|
"valid_targets_min": 607
|
|
},
|
|
{
|
|
"epoch": 0.7216494845360825,
|
|
"grad_norm": 0.24721279712124644,
|
|
"learning_rate": 3.999940516338956e-05,
|
|
"loss": 0.3306,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1001933217048645,
|
|
"step": 140,
|
|
"valid_targets_mean": 5565.4,
|
|
"valid_targets_min": 624
|
|
},
|
|
{
|
|
"epoch": 0.7474226804123711,
|
|
"grad_norm": 0.23337875442078515,
|
|
"learning_rate": 3.99957701789056e-05,
|
|
"loss": 0.3175,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10798051208257675,
|
|
"step": 145,
|
|
"valid_targets_mean": 5987.2,
|
|
"valid_targets_min": 506
|
|
},
|
|
{
|
|
"epoch": 0.7731958762886598,
|
|
"grad_norm": 0.2493552441513084,
|
|
"learning_rate": 3.998883127460509e-05,
|
|
"loss": 0.3232,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.101970374584198,
|
|
"step": 150,
|
|
"valid_targets_mean": 5937.7,
|
|
"valid_targets_min": 605
|
|
},
|
|
{
|
|
"epoch": 0.7989690721649485,
|
|
"grad_norm": 0.24775631537939216,
|
|
"learning_rate": 3.997858959700969e-05,
|
|
"loss": 0.3359,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10961902141571045,
|
|
"step": 155,
|
|
"valid_targets_mean": 6309.8,
|
|
"valid_targets_min": 594
|
|
},
|
|
{
|
|
"epoch": 0.8247422680412371,
|
|
"grad_norm": 0.25978597983421364,
|
|
"learning_rate": 3.996504683836136e-05,
|
|
"loss": 0.324,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11169438064098358,
|
|
"step": 160,
|
|
"valid_targets_mean": 5201.8,
|
|
"valid_targets_min": 420
|
|
},
|
|
{
|
|
"epoch": 0.8505154639175257,
|
|
"grad_norm": 0.23898990849913665,
|
|
"learning_rate": 3.994820523634279e-05,
|
|
"loss": 0.3245,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09977433830499649,
|
|
"step": 165,
|
|
"valid_targets_mean": 5932.4,
|
|
"valid_targets_min": 290
|
|
},
|
|
{
|
|
"epoch": 0.8762886597938144,
|
|
"grad_norm": 0.24238386639193504,
|
|
"learning_rate": 3.992806757370764e-05,
|
|
"loss": 0.3248,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1041790246963501,
|
|
"step": 170,
|
|
"valid_targets_mean": 5472.7,
|
|
"valid_targets_min": 603
|
|
},
|
|
{
|
|
"epoch": 0.9020618556701031,
|
|
"grad_norm": 0.25746086870152124,
|
|
"learning_rate": 3.990463717782074e-05,
|
|
"loss": 0.3137,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1092878133058548,
|
|
"step": 175,
|
|
"valid_targets_mean": 5549.4,
|
|
"valid_targets_min": 470
|
|
},
|
|
{
|
|
"epoch": 0.9278350515463918,
|
|
"grad_norm": 0.2592996660818364,
|
|
"learning_rate": 3.987791792010834e-05,
|
|
"loss": 0.323,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10904739797115326,
|
|
"step": 180,
|
|
"valid_targets_mean": 6642.5,
|
|
"valid_targets_min": 678
|
|
},
|
|
{
|
|
"epoch": 0.9536082474226805,
|
|
"grad_norm": 0.2550877179756528,
|
|
"learning_rate": 3.984791421541839e-05,
|
|
"loss": 0.323,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11759485304355621,
|
|
"step": 185,
|
|
"valid_targets_mean": 5979.9,
|
|
"valid_targets_min": 678
|
|
},
|
|
{
|
|
"epoch": 0.979381443298969,
|
|
"grad_norm": 0.23574276810911696,
|
|
"learning_rate": 3.981463102129109e-05,
|
|
"loss": 0.3266,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0982227697968483,
|
|
"step": 190,
|
|
"valid_targets_mean": 5994.5,
|
|
"valid_targets_min": 583
|
|
},
|
|
{
|
|
"epoch": 1.0051546391752577,
|
|
"grad_norm": 0.2409637458488068,
|
|
"learning_rate": 3.9778073837139735e-05,
|
|
"loss": 0.3167,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09220079332590103,
|
|
"step": 195,
|
|
"valid_targets_mean": 5070.1,
|
|
"valid_targets_min": 620
|
|
},
|
|
{
|
|
"epoch": 1.0309278350515463,
|
|
"grad_norm": 0.2380204035855985,
|
|
"learning_rate": 3.9738248703342085e-05,
|
|
"loss": 0.3124,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09644968062639236,
|
|
"step": 200,
|
|
"valid_targets_mean": 6010.5,
|
|
"valid_targets_min": 604
|
|
},
|
|
{
|
|
"epoch": 1.056701030927835,
|
|
"grad_norm": 0.2549648587514221,
|
|
"learning_rate": 3.9695162200242256e-05,
|
|
"loss": 0.3132,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11107391119003296,
|
|
"step": 205,
|
|
"valid_targets_mean": 6042.8,
|
|
"valid_targets_min": 777
|
|
},
|
|
{
|
|
"epoch": 1.0824742268041236,
|
|
"grad_norm": 0.2534182080407131,
|
|
"learning_rate": 3.964882144706347e-05,
|
|
"loss": 0.3077,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10496579855680466,
|
|
"step": 210,
|
|
"valid_targets_mean": 5895.7,
|
|
"valid_targets_min": 574
|
|
},
|
|
{
|
|
"epoch": 1.1082474226804124,
|
|
"grad_norm": 0.25768862239255363,
|
|
"learning_rate": 3.959923410073174e-05,
|
|
"loss": 0.3141,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10322555154561996,
|
|
"step": 215,
|
|
"valid_targets_mean": 5244.7,
|
|
"valid_targets_min": 593
|
|
},
|
|
{
|
|
"epoch": 1.134020618556701,
|
|
"grad_norm": 0.2426564982364801,
|
|
"learning_rate": 3.954640835461069e-05,
|
|
"loss": 0.3158,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10308908671140671,
|
|
"step": 220,
|
|
"valid_targets_mean": 6443.2,
|
|
"valid_targets_min": 590
|
|
},
|
|
{
|
|
"epoch": 1.1597938144329896,
|
|
"grad_norm": 0.23655770946779622,
|
|
"learning_rate": 3.9490352937147786e-05,
|
|
"loss": 0.3133,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11381332576274872,
|
|
"step": 225,
|
|
"valid_targets_mean": 6552.3,
|
|
"valid_targets_min": 575
|
|
},
|
|
{
|
|
"epoch": 1.1855670103092784,
|
|
"grad_norm": 0.24301765299422307,
|
|
"learning_rate": 3.943107711043211e-05,
|
|
"loss": 0.3031,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10668733716011047,
|
|
"step": 230,
|
|
"valid_targets_mean": 5882.4,
|
|
"valid_targets_min": 601
|
|
},
|
|
{
|
|
"epoch": 1.211340206185567,
|
|
"grad_norm": 0.2514857854558027,
|
|
"learning_rate": 3.936859066866397e-05,
|
|
"loss": 0.3109,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08552701771259308,
|
|
"step": 235,
|
|
"valid_targets_mean": 5791.9,
|
|
"valid_targets_min": 618
|
|
},
|
|
{
|
|
"epoch": 1.2371134020618557,
|
|
"grad_norm": 0.2877896952104453,
|
|
"learning_rate": 3.93029039365366e-05,
|
|
"loss": 0.3147,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09683221578598022,
|
|
"step": 240,
|
|
"valid_targets_mean": 5507.1,
|
|
"valid_targets_min": 635
|
|
},
|
|
{
|
|
"epoch": 1.2628865979381443,
|
|
"grad_norm": 0.23920924217001596,
|
|
"learning_rate": 3.923402776753022e-05,
|
|
"loss": 0.3011,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11555798351764679,
|
|
"step": 245,
|
|
"valid_targets_mean": 6155.2,
|
|
"valid_targets_min": 806
|
|
},
|
|
{
|
|
"epoch": 1.2886597938144329,
|
|
"grad_norm": 0.25899916480361196,
|
|
"learning_rate": 3.9161973542118686e-05,
|
|
"loss": 0.3044,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10108867287635803,
|
|
"step": 250,
|
|
"valid_targets_mean": 6389.3,
|
|
"valid_targets_min": 689
|
|
},
|
|
{
|
|
"epoch": 1.3144329896907216,
|
|
"grad_norm": 0.23959656238848795,
|
|
"learning_rate": 3.9086753165889076e-05,
|
|
"loss": 0.2985,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0946207121014595,
|
|
"step": 255,
|
|
"valid_targets_mean": 5665.8,
|
|
"valid_targets_min": 596
|
|
},
|
|
{
|
|
"epoch": 1.3402061855670104,
|
|
"grad_norm": 0.23205059436417375,
|
|
"learning_rate": 3.900837906757454e-05,
|
|
"loss": 0.3096,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1049540713429451,
|
|
"step": 260,
|
|
"valid_targets_mean": 6088.0,
|
|
"valid_targets_min": 549
|
|
},
|
|
{
|
|
"epoch": 1.365979381443299,
|
|
"grad_norm": 0.23826424065683088,
|
|
"learning_rate": 3.8926864197000676e-05,
|
|
"loss": 0.3124,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10210095345973969,
|
|
"step": 265,
|
|
"valid_targets_mean": 5778.2,
|
|
"valid_targets_min": 455
|
|
},
|
|
{
|
|
"epoch": 1.3917525773195876,
|
|
"grad_norm": 0.23824515889271916,
|
|
"learning_rate": 3.8842222022945823e-05,
|
|
"loss": 0.3049,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10111156105995178,
|
|
"step": 270,
|
|
"valid_targets_mean": 6070.1,
|
|
"valid_targets_min": 605
|
|
},
|
|
{
|
|
"epoch": 1.4175257731958764,
|
|
"grad_norm": 0.23845633943933653,
|
|
"learning_rate": 3.875446653091559e-05,
|
|
"loss": 0.2924,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11315014958381653,
|
|
"step": 275,
|
|
"valid_targets_mean": 6917.2,
|
|
"valid_targets_min": 983
|
|
},
|
|
{
|
|
"epoch": 1.443298969072165,
|
|
"grad_norm": 0.2310439199773325,
|
|
"learning_rate": 3.866361222083205e-05,
|
|
"loss": 0.3095,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08991458266973495,
|
|
"step": 280,
|
|
"valid_targets_mean": 5686.5,
|
|
"valid_targets_min": 627
|
|
},
|
|
{
|
|
"epoch": 1.4690721649484537,
|
|
"grad_norm": 0.22655104228409462,
|
|
"learning_rate": 3.856967410463787e-05,
|
|
"loss": 0.2981,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09699489921331406,
|
|
"step": 285,
|
|
"valid_targets_mean": 5802.7,
|
|
"valid_targets_min": 458
|
|
},
|
|
{
|
|
"epoch": 1.4948453608247423,
|
|
"grad_norm": 0.22292542695729753,
|
|
"learning_rate": 3.8472667703815876e-05,
|
|
"loss": 0.3017,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09988431632518768,
|
|
"step": 290,
|
|
"valid_targets_mean": 6589.3,
|
|
"valid_targets_min": 559
|
|
},
|
|
{
|
|
"epoch": 1.5206185567010309,
|
|
"grad_norm": 0.2408453057328623,
|
|
"learning_rate": 3.837260904682445e-05,
|
|
"loss": 0.2977,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09748954325914383,
|
|
"step": 295,
|
|
"valid_targets_mean": 6321.9,
|
|
"valid_targets_min": 586
|
|
},
|
|
{
|
|
"epoch": 1.5463917525773194,
|
|
"grad_norm": 0.23558386336961148,
|
|
"learning_rate": 3.8269514666449116e-05,
|
|
"loss": 0.3025,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10613813996315002,
|
|
"step": 300,
|
|
"valid_targets_mean": 6418.5,
|
|
"valid_targets_min": 806
|
|
},
|
|
{
|
|
"epoch": 1.5721649484536082,
|
|
"grad_norm": 0.24812086882951945,
|
|
"learning_rate": 3.816340159707079e-05,
|
|
"loss": 0.298,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10738293081521988,
|
|
"step": 305,
|
|
"valid_targets_mean": 5983.7,
|
|
"valid_targets_min": 601
|
|
},
|
|
{
|
|
"epoch": 1.597938144329897,
|
|
"grad_norm": 0.21029547243360874,
|
|
"learning_rate": 3.805428737185124e-05,
|
|
"loss": 0.2903,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08348864316940308,
|
|
"step": 310,
|
|
"valid_targets_mean": 5838.5,
|
|
"valid_targets_min": 657
|
|
},
|
|
{
|
|
"epoch": 1.6237113402061856,
|
|
"grad_norm": 0.23325282068417247,
|
|
"learning_rate": 3.794219001983597e-05,
|
|
"loss": 0.2854,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08546201884746552,
|
|
"step": 315,
|
|
"valid_targets_mean": 5487.7,
|
|
"valid_targets_min": 503
|
|
},
|
|
{
|
|
"epoch": 1.6494845360824741,
|
|
"grad_norm": 0.23941733339783547,
|
|
"learning_rate": 3.782712806297536e-05,
|
|
"loss": 0.298,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09802323579788208,
|
|
"step": 320,
|
|
"valid_targets_mean": 5105.8,
|
|
"valid_targets_min": 687
|
|
},
|
|
{
|
|
"epoch": 1.675257731958763,
|
|
"grad_norm": 0.2429264263419137,
|
|
"learning_rate": 3.7709120513064196e-05,
|
|
"loss": 0.2957,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10877049714326859,
|
|
"step": 325,
|
|
"valid_targets_mean": 6018.4,
|
|
"valid_targets_min": 524
|
|
},
|
|
{
|
|
"epoch": 1.7010309278350515,
|
|
"grad_norm": 0.2286086286790759,
|
|
"learning_rate": 3.758818686860038e-05,
|
|
"loss": 0.2941,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09833472967147827,
|
|
"step": 330,
|
|
"valid_targets_mean": 6012.3,
|
|
"valid_targets_min": 683
|
|
},
|
|
{
|
|
"epoch": 1.7268041237113403,
|
|
"grad_norm": 0.23910119281288197,
|
|
"learning_rate": 3.746434711156314e-05,
|
|
"loss": 0.2939,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11784946173429489,
|
|
"step": 335,
|
|
"valid_targets_mean": 6342.0,
|
|
"valid_targets_min": 659
|
|
},
|
|
{
|
|
"epoch": 1.7525773195876289,
|
|
"grad_norm": 0.24850545007143537,
|
|
"learning_rate": 3.733762170411138e-05,
|
|
"loss": 0.3005,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09112188220024109,
|
|
"step": 340,
|
|
"valid_targets_mean": 5050.5,
|
|
"valid_targets_min": 506
|
|
},
|
|
{
|
|
"epoch": 1.7783505154639174,
|
|
"grad_norm": 0.24896810002197928,
|
|
"learning_rate": 3.720803158520279e-05,
|
|
"loss": 0.2991,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11645789444446564,
|
|
"step": 345,
|
|
"valid_targets_mean": 6646.2,
|
|
"valid_targets_min": 992
|
|
},
|
|
{
|
|
"epoch": 1.8041237113402062,
|
|
"grad_norm": 0.24171025750101124,
|
|
"learning_rate": 3.707559816713395e-05,
|
|
"loss": 0.2936,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08031776547431946,
|
|
"step": 350,
|
|
"valid_targets_mean": 5258.9,
|
|
"valid_targets_min": 503
|
|
},
|
|
{
|
|
"epoch": 1.829896907216495,
|
|
"grad_norm": 0.22904770581010406,
|
|
"learning_rate": 3.694034333200245e-05,
|
|
"loss": 0.301,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0882028341293335,
|
|
"step": 355,
|
|
"valid_targets_mean": 5434.7,
|
|
"valid_targets_min": 676
|
|
},
|
|
{
|
|
"epoch": 1.8556701030927836,
|
|
"grad_norm": 0.2582656548833247,
|
|
"learning_rate": 3.680228942809126e-05,
|
|
"loss": 0.292,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09634621441364288,
|
|
"step": 360,
|
|
"valid_targets_mean": 5531.5,
|
|
"valid_targets_min": 612
|
|
},
|
|
{
|
|
"epoch": 1.8814432989690721,
|
|
"grad_norm": 0.21524059520826663,
|
|
"learning_rate": 3.666145926617609e-05,
|
|
"loss": 0.2888,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.096525639295578,
|
|
"step": 365,
|
|
"valid_targets_mean": 5507.4,
|
|
"valid_targets_min": 682
|
|
},
|
|
{
|
|
"epoch": 1.9072164948453607,
|
|
"grad_norm": 0.24255084476459574,
|
|
"learning_rate": 3.6517876115756376e-05,
|
|
"loss": 0.2922,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09641002118587494,
|
|
"step": 370,
|
|
"valid_targets_mean": 5803.1,
|
|
"valid_targets_min": 811
|
|
},
|
|
{
|
|
"epoch": 1.9329896907216495,
|
|
"grad_norm": 0.23858816600196303,
|
|
"learning_rate": 3.6371563701210406e-05,
|
|
"loss": 0.3038,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10171850025653839,
|
|
"step": 375,
|
|
"valid_targets_mean": 6099.8,
|
|
"valid_targets_min": 801
|
|
},
|
|
{
|
|
"epoch": 1.9587628865979383,
|
|
"grad_norm": 0.23332041195703118,
|
|
"learning_rate": 3.622254619787533e-05,
|
|
"loss": 0.2941,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10116893798112869,
|
|
"step": 380,
|
|
"valid_targets_mean": 6078.7,
|
|
"valid_targets_min": 733
|
|
},
|
|
{
|
|
"epoch": 1.9845360824742269,
|
|
"grad_norm": 0.2420312100632208,
|
|
"learning_rate": 3.607084822805268e-05,
|
|
"loss": 0.2951,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08722677826881409,
|
|
"step": 385,
|
|
"valid_targets_mean": 5265.4,
|
|
"valid_targets_min": 591
|
|
},
|
|
{
|
|
"epoch": 2.0103092783505154,
|
|
"grad_norm": 0.2343248517342809,
|
|
"learning_rate": 3.5916494856939944e-05,
|
|
"loss": 0.2949,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08683840185403824,
|
|
"step": 390,
|
|
"valid_targets_mean": 5935.9,
|
|
"valid_targets_min": 645
|
|
},
|
|
{
|
|
"epoch": 2.036082474226804,
|
|
"grad_norm": 0.24099732747355063,
|
|
"learning_rate": 3.575951158848905e-05,
|
|
"loss": 0.2882,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10633479058742523,
|
|
"step": 395,
|
|
"valid_targets_mean": 6427.1,
|
|
"valid_targets_min": 576
|
|
},
|
|
{
|
|
"epoch": 2.0618556701030926,
|
|
"grad_norm": 0.2658298537622557,
|
|
"learning_rate": 3.5599924361192335e-05,
|
|
"loss": 0.2835,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09219761192798615,
|
|
"step": 400,
|
|
"valid_targets_mean": 6387.1,
|
|
"valid_targets_min": 674
|
|
},
|
|
{
|
|
"epoch": 2.0876288659793816,
|
|
"grad_norm": 0.24887435726906582,
|
|
"learning_rate": 3.54377595437967e-05,
|
|
"loss": 0.2872,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09384647756814957,
|
|
"step": 405,
|
|
"valid_targets_mean": 5790.4,
|
|
"valid_targets_min": 618
|
|
},
|
|
{
|
|
"epoch": 2.11340206185567,
|
|
"grad_norm": 0.2342275822421684,
|
|
"learning_rate": 3.527304393094665e-05,
|
|
"loss": 0.2822,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09381407499313354,
|
|
"step": 410,
|
|
"valid_targets_mean": 5896.6,
|
|
"valid_targets_min": 618
|
|
},
|
|
{
|
|
"epoch": 2.1391752577319587,
|
|
"grad_norm": 0.2534942268345584,
|
|
"learning_rate": 3.510580473875701e-05,
|
|
"loss": 0.2917,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1218668520450592,
|
|
"step": 415,
|
|
"valid_targets_mean": 6681.8,
|
|
"valid_targets_min": 518
|
|
},
|
|
{
|
|
"epoch": 2.1649484536082473,
|
|
"grad_norm": 0.23284944470966143,
|
|
"learning_rate": 3.493606960031601e-05,
|
|
"loss": 0.2757,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09857223927974701,
|
|
"step": 420,
|
|
"valid_targets_mean": 6568.3,
|
|
"valid_targets_min": 527
|
|
},
|
|
{
|
|
"epoch": 2.1907216494845363,
|
|
"grad_norm": 0.2421756941998698,
|
|
"learning_rate": 3.476386656111938e-05,
|
|
"loss": 0.2851,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09131501615047455,
|
|
"step": 425,
|
|
"valid_targets_mean": 5952.6,
|
|
"valid_targets_min": 750
|
|
},
|
|
{
|
|
"epoch": 2.216494845360825,
|
|
"grad_norm": 0.25771606949400594,
|
|
"learning_rate": 3.458922407443643e-05,
|
|
"loss": 0.2844,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09061141312122345,
|
|
"step": 430,
|
|
"valid_targets_mean": 5639.3,
|
|
"valid_targets_min": 734
|
|
},
|
|
{
|
|
"epoch": 2.2422680412371134,
|
|
"grad_norm": 0.24545909429015159,
|
|
"learning_rate": 3.441217099660864e-05,
|
|
"loss": 0.2853,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11367497593164444,
|
|
"step": 435,
|
|
"valid_targets_mean": 6035.3,
|
|
"valid_targets_min": 689
|
|
},
|
|
{
|
|
"epoch": 2.268041237113402,
|
|
"grad_norm": 0.28375620819640673,
|
|
"learning_rate": 3.423273658228171e-05,
|
|
"loss": 0.2887,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09130064398050308,
|
|
"step": 440,
|
|
"valid_targets_mean": 5681.8,
|
|
"valid_targets_min": 594
|
|
},
|
|
{
|
|
"epoch": 2.2938144329896906,
|
|
"grad_norm": 0.251365825220021,
|
|
"learning_rate": 3.405095047957183e-05,
|
|
"loss": 0.2923,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09749715775251389,
|
|
"step": 445,
|
|
"valid_targets_mean": 5888.2,
|
|
"valid_targets_min": 960
|
|
},
|
|
{
|
|
"epoch": 2.319587628865979,
|
|
"grad_norm": 0.2402309094467951,
|
|
"learning_rate": 3.3866842725166855e-05,
|
|
"loss": 0.2828,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0872880220413208,
|
|
"step": 450,
|
|
"valid_targets_mean": 5899.3,
|
|
"valid_targets_min": 769
|
|
},
|
|
{
|
|
"epoch": 2.345360824742268,
|
|
"grad_norm": 0.25176884070476524,
|
|
"learning_rate": 3.3680443739363304e-05,
|
|
"loss": 0.2829,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1044222041964531,
|
|
"step": 455,
|
|
"valid_targets_mean": 5661.7,
|
|
"valid_targets_min": 503
|
|
},
|
|
{
|
|
"epoch": 2.3711340206185567,
|
|
"grad_norm": 0.23866003231483893,
|
|
"learning_rate": 3.349178432104001e-05,
|
|
"loss": 0.2866,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09630560874938965,
|
|
"step": 460,
|
|
"valid_targets_mean": 6392.2,
|
|
"valid_targets_min": 612
|
|
},
|
|
{
|
|
"epoch": 2.3969072164948453,
|
|
"grad_norm": 0.23396899390869666,
|
|
"learning_rate": 3.330089564256919e-05,
|
|
"loss": 0.2875,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10524716228246689,
|
|
"step": 465,
|
|
"valid_targets_mean": 6265.0,
|
|
"valid_targets_min": 524
|
|
},
|
|
{
|
|
"epoch": 2.422680412371134,
|
|
"grad_norm": 0.25436406743642886,
|
|
"learning_rate": 3.3107809244665805e-05,
|
|
"loss": 0.2908,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09861588478088379,
|
|
"step": 470,
|
|
"valid_targets_mean": 5925.4,
|
|
"valid_targets_min": 548
|
|
},
|
|
{
|
|
"epoch": 2.448453608247423,
|
|
"grad_norm": 0.2315961680878081,
|
|
"learning_rate": 3.291255703117605e-05,
|
|
"loss": 0.2847,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07914268970489502,
|
|
"step": 475,
|
|
"valid_targets_mean": 5032.7,
|
|
"valid_targets_min": 549
|
|
},
|
|
{
|
|
"epoch": 2.4742268041237114,
|
|
"grad_norm": 0.24059524256760667,
|
|
"learning_rate": 3.271517126380586e-05,
|
|
"loss": 0.2827,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09832991659641266,
|
|
"step": 480,
|
|
"valid_targets_mean": 5955.6,
|
|
"valid_targets_min": 877
|
|
},
|
|
{
|
|
"epoch": 2.5,
|
|
"grad_norm": 0.23560310629184877,
|
|
"learning_rate": 3.251568455679025e-05,
|
|
"loss": 0.2891,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11293981969356537,
|
|
"step": 485,
|
|
"valid_targets_mean": 7011.7,
|
|
"valid_targets_min": 761
|
|
},
|
|
{
|
|
"epoch": 2.5257731958762886,
|
|
"grad_norm": 0.24957734753553512,
|
|
"learning_rate": 3.2314129871504504e-05,
|
|
"loss": 0.2853,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1026647537946701,
|
|
"step": 490,
|
|
"valid_targets_mean": 5802.8,
|
|
"valid_targets_min": 582
|
|
},
|
|
{
|
|
"epoch": 2.551546391752577,
|
|
"grad_norm": 0.22947364439758133,
|
|
"learning_rate": 3.211054051101785e-05,
|
|
"loss": 0.2766,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0943123921751976,
|
|
"step": 495,
|
|
"valid_targets_mean": 5864.0,
|
|
"valid_targets_min": 610
|
|
},
|
|
{
|
|
"epoch": 2.5773195876288657,
|
|
"grad_norm": 0.23770497464938728,
|
|
"learning_rate": 3.190495011459082e-05,
|
|
"loss": 0.2869,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08464314043521881,
|
|
"step": 500,
|
|
"valid_targets_mean": 5466.0,
|
|
"valid_targets_min": 659
|
|
},
|
|
{
|
|
"epoch": 2.6030927835051547,
|
|
"grad_norm": 0.2529741357172592,
|
|
"learning_rate": 3.1697392652116996e-05,
|
|
"loss": 0.2766,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.107028529047966,
|
|
"step": 505,
|
|
"valid_targets_mean": 6372.2,
|
|
"valid_targets_min": 661
|
|
},
|
|
{
|
|
"epoch": 2.6288659793814433,
|
|
"grad_norm": 0.21932740574304702,
|
|
"learning_rate": 3.1487902418510095e-05,
|
|
"loss": 0.2857,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09716858714818954,
|
|
"step": 510,
|
|
"valid_targets_mean": 6433.1,
|
|
"valid_targets_min": 681
|
|
},
|
|
{
|
|
"epoch": 2.654639175257732,
|
|
"grad_norm": 0.21132373673187144,
|
|
"learning_rate": 3.1276514028037444e-05,
|
|
"loss": 0.2809,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09688623994588852,
|
|
"step": 515,
|
|
"valid_targets_mean": 6873.9,
|
|
"valid_targets_min": 655
|
|
},
|
|
{
|
|
"epoch": 2.680412371134021,
|
|
"grad_norm": 0.22391906122745997,
|
|
"learning_rate": 3.10632624086006e-05,
|
|
"loss": 0.2746,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09009232372045517,
|
|
"step": 520,
|
|
"valid_targets_mean": 5922.0,
|
|
"valid_targets_min": 626
|
|
},
|
|
{
|
|
"epoch": 2.7061855670103094,
|
|
"grad_norm": 0.2416840076675489,
|
|
"learning_rate": 3.084818279596417e-05,
|
|
"loss": 0.2819,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09669722616672516,
|
|
"step": 525,
|
|
"valid_targets_mean": 5536.5,
|
|
"valid_targets_min": 463
|
|
},
|
|
{
|
|
"epoch": 2.731958762886598,
|
|
"grad_norm": 0.24005876397100998,
|
|
"learning_rate": 3.063131072793379e-05,
|
|
"loss": 0.283,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09082949161529541,
|
|
"step": 530,
|
|
"valid_targets_mean": 5989.2,
|
|
"valid_targets_min": 581
|
|
},
|
|
{
|
|
"epoch": 2.7577319587628866,
|
|
"grad_norm": 0.22663868595846726,
|
|
"learning_rate": 3.0412682038484215e-05,
|
|
"loss": 0.2741,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09188930690288544,
|
|
"step": 535,
|
|
"valid_targets_mean": 6203.5,
|
|
"valid_targets_min": 707
|
|
},
|
|
{
|
|
"epoch": 2.783505154639175,
|
|
"grad_norm": 0.21908930480458605,
|
|
"learning_rate": 3.019233285183835e-05,
|
|
"loss": 0.2753,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10089852660894394,
|
|
"step": 540,
|
|
"valid_targets_mean": 6096.4,
|
|
"valid_targets_min": 749
|
|
},
|
|
{
|
|
"epoch": 2.8092783505154637,
|
|
"grad_norm": 0.2515063915320478,
|
|
"learning_rate": 2.9970299576498495e-05,
|
|
"loss": 0.2858,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09591610729694366,
|
|
"step": 545,
|
|
"valid_targets_mean": 5674.9,
|
|
"valid_targets_min": 592
|
|
},
|
|
{
|
|
"epoch": 2.8350515463917527,
|
|
"grad_norm": 0.2328364488943388,
|
|
"learning_rate": 2.974661889923048e-05,
|
|
"loss": 0.2906,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12141162902116776,
|
|
"step": 550,
|
|
"valid_targets_mean": 6436.8,
|
|
"valid_targets_min": 651
|
|
},
|
|
{
|
|
"epoch": 2.8608247422680413,
|
|
"grad_norm": 0.244877435814816,
|
|
"learning_rate": 2.9521327779001906e-05,
|
|
"loss": 0.2789,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10483044385910034,
|
|
"step": 555,
|
|
"valid_targets_mean": 6250.4,
|
|
"valid_targets_min": 795
|
|
},
|
|
{
|
|
"epoch": 2.88659793814433,
|
|
"grad_norm": 0.24326293126897247,
|
|
"learning_rate": 2.9294463440875375e-05,
|
|
"loss": 0.2846,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08272717893123627,
|
|
"step": 560,
|
|
"valid_targets_mean": 5923.4,
|
|
"valid_targets_min": 503
|
|
},
|
|
{
|
|
"epoch": 2.9123711340206184,
|
|
"grad_norm": 0.24352900559746452,
|
|
"learning_rate": 2.906606336985775e-05,
|
|
"loss": 0.2779,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08788428455591202,
|
|
"step": 565,
|
|
"valid_targets_mean": 4883.2,
|
|
"valid_targets_min": 627
|
|
},
|
|
{
|
|
"epoch": 2.9381443298969074,
|
|
"grad_norm": 0.24432741772867364,
|
|
"learning_rate": 2.883616530470648e-05,
|
|
"loss": 0.2772,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10530249029397964,
|
|
"step": 570,
|
|
"valid_targets_mean": 5787.4,
|
|
"valid_targets_min": 599
|
|
},
|
|
{
|
|
"epoch": 2.963917525773196,
|
|
"grad_norm": 0.2652726308860743,
|
|
"learning_rate": 2.8604807231693987e-05,
|
|
"loss": 0.2854,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08077148348093033,
|
|
"step": 575,
|
|
"valid_targets_mean": 5646.8,
|
|
"valid_targets_min": 603
|
|
},
|
|
{
|
|
"epoch": 2.9896907216494846,
|
|
"grad_norm": 0.23411552434676475,
|
|
"learning_rate": 2.8372027378331148e-05,
|
|
"loss": 0.2783,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08464553207159042,
|
|
"step": 580,
|
|
"valid_targets_mean": 5979.0,
|
|
"valid_targets_min": 613
|
|
},
|
|
{
|
|
"epoch": 3.015463917525773,
|
|
"grad_norm": 0.20260770993636365,
|
|
"learning_rate": 2.8137864207050945e-05,
|
|
"loss": 0.2746,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09096153825521469,
|
|
"step": 585,
|
|
"valid_targets_mean": 6076.0,
|
|
"valid_targets_min": 682
|
|
},
|
|
{
|
|
"epoch": 3.0412371134020617,
|
|
"grad_norm": 0.28486265482999723,
|
|
"learning_rate": 2.790235640885327e-05,
|
|
"loss": 0.2675,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10470966249704361,
|
|
"step": 590,
|
|
"valid_targets_mean": 6595.6,
|
|
"valid_targets_min": 546
|
|
},
|
|
{
|
|
"epoch": 3.0670103092783507,
|
|
"grad_norm": 0.24942742541858107,
|
|
"learning_rate": 2.7665542896911993e-05,
|
|
"loss": 0.2764,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09387102723121643,
|
|
"step": 595,
|
|
"valid_targets_mean": 6898.8,
|
|
"valid_targets_min": 2944
|
|
},
|
|
{
|
|
"epoch": 3.0927835051546393,
|
|
"grad_norm": 0.22996526245052193,
|
|
"learning_rate": 2.742746280014527e-05,
|
|
"loss": 0.2661,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08765529841184616,
|
|
"step": 600,
|
|
"valid_targets_mean": 5356.6,
|
|
"valid_targets_min": 683
|
|
},
|
|
{
|
|
"epoch": 3.118556701030928,
|
|
"grad_norm": 0.26407748188095326,
|
|
"learning_rate": 2.7188155456750257e-05,
|
|
"loss": 0.2814,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0783463567495346,
|
|
"step": 605,
|
|
"valid_targets_mean": 4844.9,
|
|
"valid_targets_min": 733
|
|
},
|
|
{
|
|
"epoch": 3.1443298969072164,
|
|
"grad_norm": 0.22348367760182972,
|
|
"learning_rate": 2.694766040770323e-05,
|
|
"loss": 0.2728,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09305690228939056,
|
|
"step": 610,
|
|
"valid_targets_mean": 6214.3,
|
|
"valid_targets_min": 558
|
|
},
|
|
{
|
|
"epoch": 3.170103092783505,
|
|
"grad_norm": 0.2361655369361659,
|
|
"learning_rate": 2.6706017390226162e-05,
|
|
"loss": 0.2754,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09483394771814346,
|
|
"step": 615,
|
|
"valid_targets_mean": 6181.8,
|
|
"valid_targets_min": 817
|
|
},
|
|
{
|
|
"epoch": 3.195876288659794,
|
|
"grad_norm": 0.22483897147525417,
|
|
"learning_rate": 2.646326633122094e-05,
|
|
"loss": 0.2702,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08822141587734222,
|
|
"step": 620,
|
|
"valid_targets_mean": 5600.7,
|
|
"valid_targets_min": 568
|
|
},
|
|
{
|
|
"epoch": 3.2216494845360826,
|
|
"grad_norm": 0.23594366918535295,
|
|
"learning_rate": 2.621944734067218e-05,
|
|
"loss": 0.28,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10289697349071503,
|
|
"step": 625,
|
|
"valid_targets_mean": 6439.7,
|
|
"valid_targets_min": 829
|
|
},
|
|
{
|
|
"epoch": 3.247422680412371,
|
|
"grad_norm": 0.23205688764514465,
|
|
"learning_rate": 2.597460070501983e-05,
|
|
"loss": 0.2775,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08696689456701279,
|
|
"step": 630,
|
|
"valid_targets_mean": 5283.2,
|
|
"valid_targets_min": 634
|
|
},
|
|
{
|
|
"epoch": 3.2731958762886597,
|
|
"grad_norm": 0.22585622138198921,
|
|
"learning_rate": 2.5728766880502604e-05,
|
|
"loss": 0.2752,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08871736377477646,
|
|
"step": 635,
|
|
"valid_targets_mean": 5726.8,
|
|
"valid_targets_min": 610
|
|
},
|
|
{
|
|
"epoch": 3.2989690721649483,
|
|
"grad_norm": 0.22220502134440065,
|
|
"learning_rate": 2.5481986486473352e-05,
|
|
"loss": 0.2757,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07732512056827545,
|
|
"step": 640,
|
|
"valid_targets_mean": 5585.0,
|
|
"valid_targets_min": 701
|
|
},
|
|
{
|
|
"epoch": 3.3247422680412373,
|
|
"grad_norm": 0.23991535815705303,
|
|
"learning_rate": 2.523430029868749e-05,
|
|
"loss": 0.27,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08161801099777222,
|
|
"step": 645,
|
|
"valid_targets_mean": 5979.8,
|
|
"valid_targets_min": 524
|
|
},
|
|
{
|
|
"epoch": 3.350515463917526,
|
|
"grad_norm": 0.2200888804921485,
|
|
"learning_rate": 2.498574924256558e-05,
|
|
"loss": 0.268,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08938449621200562,
|
|
"step": 650,
|
|
"valid_targets_mean": 5659.7,
|
|
"valid_targets_min": 581
|
|
},
|
|
{
|
|
"epoch": 3.3762886597938144,
|
|
"grad_norm": 0.22680091617135775,
|
|
"learning_rate": 2.4736374386431194e-05,
|
|
"loss": 0.2701,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08103431016206741,
|
|
"step": 655,
|
|
"valid_targets_mean": 6188.1,
|
|
"valid_targets_min": 509
|
|
},
|
|
{
|
|
"epoch": 3.402061855670103,
|
|
"grad_norm": 0.2304373114730706,
|
|
"learning_rate": 2.4486216934725143e-05,
|
|
"loss": 0.2815,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10041002184152603,
|
|
"step": 660,
|
|
"valid_targets_mean": 6363.6,
|
|
"valid_targets_min": 996
|
|
},
|
|
{
|
|
"epoch": 3.4278350515463916,
|
|
"grad_norm": 0.23661270764251344,
|
|
"learning_rate": 2.4235318221197258e-05,
|
|
"loss": 0.2749,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0988386869430542,
|
|
"step": 665,
|
|
"valid_targets_mean": 6608.8,
|
|
"valid_targets_min": 958
|
|
},
|
|
{
|
|
"epoch": 3.4536082474226806,
|
|
"grad_norm": 0.2515269393218624,
|
|
"learning_rate": 2.3983719702076724e-05,
|
|
"loss": 0.2785,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09459570050239563,
|
|
"step": 670,
|
|
"valid_targets_mean": 5628.4,
|
|
"valid_targets_min": 492
|
|
},
|
|
{
|
|
"epoch": 3.479381443298969,
|
|
"grad_norm": 0.22183184343635845,
|
|
"learning_rate": 2.3731462949222282e-05,
|
|
"loss": 0.2685,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10383935272693634,
|
|
"step": 675,
|
|
"valid_targets_mean": 6286.9,
|
|
"valid_targets_min": 610
|
|
},
|
|
{
|
|
"epoch": 3.5051546391752577,
|
|
"grad_norm": 0.218173519793643,
|
|
"learning_rate": 2.347858964325324e-05,
|
|
"loss": 0.2697,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08168500661849976,
|
|
"step": 680,
|
|
"valid_targets_mean": 5534.5,
|
|
"valid_targets_min": 569
|
|
},
|
|
{
|
|
"epoch": 3.5309278350515463,
|
|
"grad_norm": 0.24391334613816987,
|
|
"learning_rate": 2.3225141566662556e-05,
|
|
"loss": 0.2817,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1026201918721199,
|
|
"step": 685,
|
|
"valid_targets_mean": 5892.1,
|
|
"valid_targets_min": 587
|
|
},
|
|
{
|
|
"epoch": 3.556701030927835,
|
|
"grad_norm": 0.22283037278700582,
|
|
"learning_rate": 2.297116059691307e-05,
|
|
"loss": 0.2717,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09592479467391968,
|
|
"step": 690,
|
|
"valid_targets_mean": 5870.2,
|
|
"valid_targets_min": 692
|
|
},
|
|
{
|
|
"epoch": 3.582474226804124,
|
|
"grad_norm": 0.2327362706557194,
|
|
"learning_rate": 2.2716688699518063e-05,
|
|
"loss": 0.2703,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08582796156406403,
|
|
"step": 695,
|
|
"valid_targets_mean": 5697.4,
|
|
"valid_targets_min": 617
|
|
},
|
|
{
|
|
"epoch": 3.6082474226804124,
|
|
"grad_norm": 0.22590384878782008,
|
|
"learning_rate": 2.2461767921107255e-05,
|
|
"loss": 0.2741,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0797477662563324,
|
|
"step": 700,
|
|
"valid_targets_mean": 5632.4,
|
|
"valid_targets_min": 574
|
|
},
|
|
{
|
|
"epoch": 3.634020618556701,
|
|
"grad_norm": 0.29798031320705237,
|
|
"learning_rate": 2.2206440382479378e-05,
|
|
"loss": 0.2709,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08003167808055878,
|
|
"step": 705,
|
|
"valid_targets_mean": 5537.1,
|
|
"valid_targets_min": 515
|
|
},
|
|
{
|
|
"epoch": 3.6597938144329896,
|
|
"grad_norm": 0.21131751862752976,
|
|
"learning_rate": 2.195074827164256e-05,
|
|
"loss": 0.2674,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1092803105711937,
|
|
"step": 710,
|
|
"valid_targets_mean": 6254.5,
|
|
"valid_targets_min": 710
|
|
},
|
|
{
|
|
"epoch": 3.6855670103092786,
|
|
"grad_norm": 0.2154469780366876,
|
|
"learning_rate": 2.169473383684352e-05,
|
|
"loss": 0.2679,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08612493425607681,
|
|
"step": 715,
|
|
"valid_targets_mean": 6330.3,
|
|
"valid_targets_min": 466
|
|
},
|
|
{
|
|
"epoch": 3.711340206185567,
|
|
"grad_norm": 0.23087260590653957,
|
|
"learning_rate": 2.143843937958687e-05,
|
|
"loss": 0.2736,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08392924070358276,
|
|
"step": 720,
|
|
"valid_targets_mean": 5510.4,
|
|
"valid_targets_min": 806
|
|
},
|
|
{
|
|
"epoch": 3.7371134020618557,
|
|
"grad_norm": 0.23084663151007864,
|
|
"learning_rate": 2.118190724764559e-05,
|
|
"loss": 0.2741,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08487777411937714,
|
|
"step": 725,
|
|
"valid_targets_mean": 5964.2,
|
|
"valid_targets_min": 618
|
|
},
|
|
{
|
|
"epoch": 3.7628865979381443,
|
|
"grad_norm": 0.22025011833513305,
|
|
"learning_rate": 2.0925179828063875e-05,
|
|
"loss": 0.2812,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10198937356472015,
|
|
"step": 730,
|
|
"valid_targets_mean": 5891.6,
|
|
"valid_targets_min": 709
|
|
},
|
|
{
|
|
"epoch": 3.788659793814433,
|
|
"grad_norm": 0.22121842128951696,
|
|
"learning_rate": 2.0668299540153494e-05,
|
|
"loss": 0.2803,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09764805436134338,
|
|
"step": 735,
|
|
"valid_targets_mean": 6472.3,
|
|
"valid_targets_min": 807
|
|
},
|
|
{
|
|
"epoch": 3.8144329896907214,
|
|
"grad_norm": 0.22565018419259,
|
|
"learning_rate": 2.0411308828484766e-05,
|
|
"loss": 0.2728,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09720548987388611,
|
|
"step": 740,
|
|
"valid_targets_mean": 6513.4,
|
|
"valid_targets_min": 796
|
|
},
|
|
{
|
|
"epoch": 3.8402061855670104,
|
|
"grad_norm": 0.21971367853705612,
|
|
"learning_rate": 2.0154250155873454e-05,
|
|
"loss": 0.2708,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08757640421390533,
|
|
"step": 745,
|
|
"valid_targets_mean": 5806.6,
|
|
"valid_targets_min": 531
|
|
},
|
|
{
|
|
"epoch": 3.865979381443299,
|
|
"grad_norm": 0.23684124646299223,
|
|
"learning_rate": 1.989716599636456e-05,
|
|
"loss": 0.2733,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09203958511352539,
|
|
"step": 750,
|
|
"valid_targets_mean": 6173.7,
|
|
"valid_targets_min": 635
|
|
},
|
|
{
|
|
"epoch": 3.8917525773195876,
|
|
"grad_norm": 0.2208836110906776,
|
|
"learning_rate": 1.964009882821429e-05,
|
|
"loss": 0.2699,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08275872468948364,
|
|
"step": 755,
|
|
"valid_targets_mean": 5502.0,
|
|
"valid_targets_min": 527
|
|
},
|
|
{
|
|
"epoch": 3.917525773195876,
|
|
"grad_norm": 0.21857410574502745,
|
|
"learning_rate": 1.9383091126871406e-05,
|
|
"loss": 0.2695,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09316932410001755,
|
|
"step": 760,
|
|
"valid_targets_mean": 5920.6,
|
|
"valid_targets_min": 585
|
|
},
|
|
{
|
|
"epoch": 3.943298969072165,
|
|
"grad_norm": 0.21398463578295063,
|
|
"learning_rate": 1.9126185357958863e-05,
|
|
"loss": 0.2678,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09043984115123749,
|
|
"step": 765,
|
|
"valid_targets_mean": 6321.4,
|
|
"valid_targets_min": 887
|
|
},
|
|
{
|
|
"epoch": 3.9690721649484537,
|
|
"grad_norm": 0.21415830494203367,
|
|
"learning_rate": 1.8869423970257243e-05,
|
|
"loss": 0.2687,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09687180817127228,
|
|
"step": 770,
|
|
"valid_targets_mean": 6493.0,
|
|
"valid_targets_min": 593
|
|
},
|
|
{
|
|
"epoch": 3.9948453608247423,
|
|
"grad_norm": 0.23164224876228207,
|
|
"learning_rate": 1.8612849388690893e-05,
|
|
"loss": 0.2704,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0891723483800888,
|
|
"step": 775,
|
|
"valid_targets_mean": 5776.0,
|
|
"valid_targets_min": 907
|
|
},
|
|
{
|
|
"epoch": 4.020618556701031,
|
|
"grad_norm": 0.21756533495233116,
|
|
"learning_rate": 1.835650400731799e-05,
|
|
"loss": 0.2642,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08937852084636688,
|
|
"step": 780,
|
|
"valid_targets_mean": 6833.9,
|
|
"valid_targets_min": 675
|
|
},
|
|
{
|
|
"epoch": 4.046391752577319,
|
|
"grad_norm": 0.21433750105865554,
|
|
"learning_rate": 1.810043018232577e-05,
|
|
"loss": 0.2687,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08626782894134521,
|
|
"step": 785,
|
|
"valid_targets_mean": 6847.7,
|
|
"valid_targets_min": 1132
|
|
},
|
|
{
|
|
"epoch": 4.072164948453608,
|
|
"grad_norm": 0.2288990231618256,
|
|
"learning_rate": 1.7844670225031934e-05,
|
|
"loss": 0.269,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07571521401405334,
|
|
"step": 790,
|
|
"valid_targets_mean": 5339.3,
|
|
"valid_targets_min": 524
|
|
},
|
|
{
|
|
"epoch": 4.097938144329897,
|
|
"grad_norm": 0.23192269065818466,
|
|
"learning_rate": 1.758926639489354e-05,
|
|
"loss": 0.265,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07901392877101898,
|
|
"step": 795,
|
|
"valid_targets_mean": 4847.0,
|
|
"valid_targets_min": 503
|
|
},
|
|
{
|
|
"epoch": 4.123711340206185,
|
|
"grad_norm": 0.23478575143774874,
|
|
"learning_rate": 1.733426089252443e-05,
|
|
"loss": 0.2697,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09161382913589478,
|
|
"step": 800,
|
|
"valid_targets_mean": 5251.7,
|
|
"valid_targets_min": 548
|
|
},
|
|
{
|
|
"epoch": 4.149484536082475,
|
|
"grad_norm": 0.22000507442418243,
|
|
"learning_rate": 1.7079695852722347e-05,
|
|
"loss": 0.2635,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09962445497512817,
|
|
"step": 805,
|
|
"valid_targets_mean": 5863.6,
|
|
"valid_targets_min": 685
|
|
},
|
|
{
|
|
"epoch": 4.175257731958763,
|
|
"grad_norm": 0.22765250973668155,
|
|
"learning_rate": 1.682561333750703e-05,
|
|
"loss": 0.2581,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0884147435426712,
|
|
"step": 810,
|
|
"valid_targets_mean": 5735.3,
|
|
"valid_targets_min": 653
|
|
},
|
|
{
|
|
"epoch": 4.201030927835052,
|
|
"grad_norm": 0.2095776550984572,
|
|
"learning_rate": 1.6572055329170214e-05,
|
|
"loss": 0.2652,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07988305389881134,
|
|
"step": 815,
|
|
"valid_targets_mean": 6059.3,
|
|
"valid_targets_min": 421
|
|
},
|
|
{
|
|
"epoch": 4.22680412371134,
|
|
"grad_norm": 0.20422248570855148,
|
|
"learning_rate": 1.6319063723338855e-05,
|
|
"loss": 0.2688,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0876869261264801,
|
|
"step": 820,
|
|
"valid_targets_mean": 6198.9,
|
|
"valid_targets_min": 726
|
|
},
|
|
{
|
|
"epoch": 4.252577319587629,
|
|
"grad_norm": 0.2270878274423797,
|
|
"learning_rate": 1.6066680322052703e-05,
|
|
"loss": 0.26,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08421394973993301,
|
|
"step": 825,
|
|
"valid_targets_mean": 6588.2,
|
|
"valid_targets_min": 563
|
|
},
|
|
{
|
|
"epoch": 4.278350515463917,
|
|
"grad_norm": 0.2178736819595442,
|
|
"learning_rate": 1.5814946826857323e-05,
|
|
"loss": 0.2669,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0885683000087738,
|
|
"step": 830,
|
|
"valid_targets_mean": 5630.2,
|
|
"valid_targets_min": 475
|
|
},
|
|
{
|
|
"epoch": 4.304123711340206,
|
|
"grad_norm": 0.21081740441324276,
|
|
"learning_rate": 1.5563904831913663e-05,
|
|
"loss": 0.2722,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09665192663669586,
|
|
"step": 835,
|
|
"valid_targets_mean": 6812.8,
|
|
"valid_targets_min": 665
|
|
},
|
|
{
|
|
"epoch": 4.329896907216495,
|
|
"grad_norm": 0.22526468296223778,
|
|
"learning_rate": 1.5313595817125445e-05,
|
|
"loss": 0.2647,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08404377102851868,
|
|
"step": 840,
|
|
"valid_targets_mean": 6131.6,
|
|
"valid_targets_min": 528
|
|
},
|
|
{
|
|
"epoch": 4.355670103092783,
|
|
"grad_norm": 0.2146685650480542,
|
|
"learning_rate": 1.5064061141285397e-05,
|
|
"loss": 0.2711,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08425067365169525,
|
|
"step": 845,
|
|
"valid_targets_mean": 6123.2,
|
|
"valid_targets_min": 587
|
|
},
|
|
{
|
|
"epoch": 4.381443298969073,
|
|
"grad_norm": 0.2127817486043924,
|
|
"learning_rate": 1.481534203524149e-05,
|
|
"loss": 0.2645,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07339270412921906,
|
|
"step": 850,
|
|
"valid_targets_mean": 5728.7,
|
|
"valid_targets_min": 580
|
|
},
|
|
{
|
|
"epoch": 4.407216494845361,
|
|
"grad_norm": 0.22478007486889792,
|
|
"learning_rate": 1.4567479595084349e-05,
|
|
"loss": 0.2629,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07707114517688751,
|
|
"step": 855,
|
|
"valid_targets_mean": 5482.8,
|
|
"valid_targets_min": 760
|
|
},
|
|
{
|
|
"epoch": 4.43298969072165,
|
|
"grad_norm": 0.24636487762866363,
|
|
"learning_rate": 1.4320514775356878e-05,
|
|
"loss": 0.2661,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10012729465961456,
|
|
"step": 860,
|
|
"valid_targets_mean": 6463.2,
|
|
"valid_targets_min": 665
|
|
},
|
|
{
|
|
"epoch": 4.458762886597938,
|
|
"grad_norm": 0.20615408947011518,
|
|
"learning_rate": 1.4074488382287324e-05,
|
|
"loss": 0.267,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08085936307907104,
|
|
"step": 865,
|
|
"valid_targets_mean": 5675.8,
|
|
"valid_targets_min": 705
|
|
},
|
|
{
|
|
"epoch": 4.484536082474227,
|
|
"grad_norm": 0.22354565382405714,
|
|
"learning_rate": 1.3829441067046822e-05,
|
|
"loss": 0.2708,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0899420902132988,
|
|
"step": 870,
|
|
"valid_targets_mean": 5963.5,
|
|
"valid_targets_min": 579
|
|
},
|
|
{
|
|
"epoch": 4.510309278350515,
|
|
"grad_norm": 0.2136805096947886,
|
|
"learning_rate": 1.358541331903255e-05,
|
|
"loss": 0.2626,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09212556481361389,
|
|
"step": 875,
|
|
"valid_targets_mean": 5723.1,
|
|
"valid_targets_min": 630
|
|
},
|
|
{
|
|
"epoch": 4.536082474226804,
|
|
"grad_norm": 0.21156293091841508,
|
|
"learning_rate": 1.3342445459177658e-05,
|
|
"loss": 0.2674,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09288851916790009,
|
|
"step": 880,
|
|
"valid_targets_mean": 6316.2,
|
|
"valid_targets_min": 761
|
|
},
|
|
{
|
|
"epoch": 4.561855670103093,
|
|
"grad_norm": 0.21227368812862335,
|
|
"learning_rate": 1.3100577633288967e-05,
|
|
"loss": 0.262,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08863004297018051,
|
|
"step": 885,
|
|
"valid_targets_mean": 6136.3,
|
|
"valid_targets_min": 700
|
|
},
|
|
{
|
|
"epoch": 4.587628865979381,
|
|
"grad_norm": 0.21992367316398848,
|
|
"learning_rate": 1.2859849805413664e-05,
|
|
"loss": 0.271,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09031090140342712,
|
|
"step": 890,
|
|
"valid_targets_mean": 5882.1,
|
|
"valid_targets_min": 624
|
|
},
|
|
{
|
|
"epoch": 4.61340206185567,
|
|
"grad_norm": 0.22712167753510576,
|
|
"learning_rate": 1.262030175123599e-05,
|
|
"loss": 0.2638,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09249001741409302,
|
|
"step": 895,
|
|
"valid_targets_mean": 6656.8,
|
|
"valid_targets_min": 877
|
|
},
|
|
{
|
|
"epoch": 4.639175257731958,
|
|
"grad_norm": 0.2181886728130532,
|
|
"learning_rate": 1.2381973051505089e-05,
|
|
"loss": 0.2613,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09235621988773346,
|
|
"step": 900,
|
|
"valid_targets_mean": 6152.6,
|
|
"valid_targets_min": 519
|
|
},
|
|
{
|
|
"epoch": 4.664948453608248,
|
|
"grad_norm": 0.21173138896730861,
|
|
"learning_rate": 1.2144903085495021e-05,
|
|
"loss": 0.266,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09007187932729721,
|
|
"step": 905,
|
|
"valid_targets_mean": 6109.0,
|
|
"valid_targets_min": 571
|
|
},
|
|
{
|
|
"epoch": 4.690721649484536,
|
|
"grad_norm": 0.21384173047360272,
|
|
"learning_rate": 1.190913102449809e-05,
|
|
"loss": 0.2693,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09088248014450073,
|
|
"step": 910,
|
|
"valid_targets_mean": 6272.6,
|
|
"valid_targets_min": 516
|
|
},
|
|
{
|
|
"epoch": 4.716494845360825,
|
|
"grad_norm": 0.2471444990761087,
|
|
"learning_rate": 1.1674695825352544e-05,
|
|
"loss": 0.267,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08068041503429413,
|
|
"step": 915,
|
|
"valid_targets_mean": 5634.2,
|
|
"valid_targets_min": 537
|
|
},
|
|
{
|
|
"epoch": 4.742268041237113,
|
|
"grad_norm": 0.20417738889614037,
|
|
"learning_rate": 1.1441636224005684e-05,
|
|
"loss": 0.2692,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09102384746074677,
|
|
"step": 920,
|
|
"valid_targets_mean": 5985.6,
|
|
"valid_targets_min": 549
|
|
},
|
|
{
|
|
"epoch": 4.768041237113402,
|
|
"grad_norm": 0.20933575423306253,
|
|
"learning_rate": 1.120999072911354e-05,
|
|
"loss": 0.2632,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09504446387290955,
|
|
"step": 925,
|
|
"valid_targets_mean": 6498.6,
|
|
"valid_targets_min": 826
|
|
},
|
|
{
|
|
"epoch": 4.793814432989691,
|
|
"grad_norm": 0.21495610205009033,
|
|
"learning_rate": 1.097979761567796e-05,
|
|
"loss": 0.2624,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08593370765447617,
|
|
"step": 930,
|
|
"valid_targets_mean": 6293.3,
|
|
"valid_targets_min": 836
|
|
},
|
|
{
|
|
"epoch": 4.819587628865979,
|
|
"grad_norm": 0.21628152707138062,
|
|
"learning_rate": 1.0751094918722487e-05,
|
|
"loss": 0.2714,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09788992255926132,
|
|
"step": 935,
|
|
"valid_targets_mean": 5808.1,
|
|
"valid_targets_min": 543
|
|
},
|
|
{
|
|
"epoch": 4.845360824742268,
|
|
"grad_norm": 0.20876818864253993,
|
|
"learning_rate": 1.0523920427007717e-05,
|
|
"loss": 0.2641,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10079794377088547,
|
|
"step": 940,
|
|
"valid_targets_mean": 6872.3,
|
|
"valid_targets_min": 844
|
|
},
|
|
{
|
|
"epoch": 4.871134020618557,
|
|
"grad_norm": 0.21111742947633577,
|
|
"learning_rate": 1.0298311676787474e-05,
|
|
"loss": 0.2715,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08679303526878357,
|
|
"step": 945,
|
|
"valid_targets_mean": 5815.4,
|
|
"valid_targets_min": 518
|
|
},
|
|
{
|
|
"epoch": 4.896907216494846,
|
|
"grad_norm": 0.21632908151924193,
|
|
"learning_rate": 1.0074305945606684e-05,
|
|
"loss": 0.2656,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08715079724788666,
|
|
"step": 950,
|
|
"valid_targets_mean": 5100.9,
|
|
"valid_targets_min": 574
|
|
},
|
|
{
|
|
"epoch": 4.922680412371134,
|
|
"grad_norm": 0.20770269931366217,
|
|
"learning_rate": 9.851940246141888e-06,
|
|
"loss": 0.2653,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10157003253698349,
|
|
"step": 955,
|
|
"valid_targets_mean": 6592.1,
|
|
"valid_targets_min": 669
|
|
},
|
|
{
|
|
"epoch": 4.948453608247423,
|
|
"grad_norm": 0.21189368814109807,
|
|
"learning_rate": 9.63125132008569e-06,
|
|
"loss": 0.2655,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07998209446668625,
|
|
"step": 960,
|
|
"valid_targets_mean": 5278.8,
|
|
"valid_targets_min": 640
|
|
},
|
|
{
|
|
"epoch": 4.974226804123711,
|
|
"grad_norm": 0.22322550354193166,
|
|
"learning_rate": 9.412275632075827e-06,
|
|
"loss": 0.2633,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09199348092079163,
|
|
"step": 965,
|
|
"valid_targets_mean": 5675.4,
|
|
"valid_targets_min": 705
|
|
},
|
|
{
|
|
"epoch": 5.0,
|
|
"grad_norm": 0.20908447032677452,
|
|
"learning_rate": 9.195049363670125e-06,
|
|
"loss": 0.266,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08515501767396927,
|
|
"step": 970,
|
|
"valid_targets_mean": 5548.7,
|
|
"valid_targets_min": 542
|
|
},
|
|
{
|
|
"epoch": 5.025773195876289,
|
|
"grad_norm": 0.20769139189806285,
|
|
"learning_rate": 8.979608407368182e-06,
|
|
"loss": 0.2674,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0799143984913826,
|
|
"step": 975,
|
|
"valid_targets_mean": 5533.4,
|
|
"valid_targets_min": 549
|
|
},
|
|
{
|
|
"epoch": 5.051546391752577,
|
|
"grad_norm": 0.20853254730464704,
|
|
"learning_rate": 8.765988360680772e-06,
|
|
"loss": 0.2621,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07116574048995972,
|
|
"step": 980,
|
|
"valid_targets_mean": 5815.3,
|
|
"valid_targets_min": 663
|
|
},
|
|
{
|
|
"epoch": 5.077319587628866,
|
|
"grad_norm": 0.2121997853645642,
|
|
"learning_rate": 8.554224520248115e-06,
|
|
"loss": 0.2641,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08537183701992035,
|
|
"step": 985,
|
|
"valid_targets_mean": 5817.0,
|
|
"valid_targets_min": 317
|
|
},
|
|
{
|
|
"epoch": 5.103092783505154,
|
|
"grad_norm": 0.21085290904277704,
|
|
"learning_rate": 8.344351876007686e-06,
|
|
"loss": 0.2644,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.097097247838974,
|
|
"step": 990,
|
|
"valid_targets_mean": 6354.1,
|
|
"valid_targets_min": 625
|
|
},
|
|
{
|
|
"epoch": 5.128865979381443,
|
|
"grad_norm": 0.21286339815700664,
|
|
"learning_rate": 8.136405105412897e-06,
|
|
"loss": 0.266,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07448627799749374,
|
|
"step": 995,
|
|
"valid_targets_mean": 5040.7,
|
|
"valid_targets_min": 710
|
|
},
|
|
{
|
|
"epoch": 5.154639175257732,
|
|
"grad_norm": 0.21468210382569625,
|
|
"learning_rate": 7.930418567703215e-06,
|
|
"loss": 0.2541,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0872785896062851,
|
|
"step": 1000,
|
|
"valid_targets_mean": 6107.7,
|
|
"valid_targets_min": 601
|
|
},
|
|
{
|
|
"epoch": 5.180412371134021,
|
|
"grad_norm": 0.1997157686733035,
|
|
"learning_rate": 7.726426298226996e-06,
|
|
"loss": 0.2656,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08862429857254028,
|
|
"step": 1005,
|
|
"valid_targets_mean": 6700.6,
|
|
"valid_targets_min": 687
|
|
},
|
|
{
|
|
"epoch": 5.206185567010309,
|
|
"grad_norm": 0.20206102721595298,
|
|
"learning_rate": 7.524462002817809e-06,
|
|
"loss": 0.2592,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08948960900306702,
|
|
"step": 1010,
|
|
"valid_targets_mean": 5725.0,
|
|
"valid_targets_min": 574
|
|
},
|
|
{
|
|
"epoch": 5.231958762886598,
|
|
"grad_norm": 0.2440073772913018,
|
|
"learning_rate": 7.324559052225162e-06,
|
|
"loss": 0.2559,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09645431488752365,
|
|
"step": 1015,
|
|
"valid_targets_mean": 6422.7,
|
|
"valid_targets_min": 915
|
|
},
|
|
{
|
|
"epoch": 5.257731958762887,
|
|
"grad_norm": 0.21606760726516505,
|
|
"learning_rate": 7.126750476600635e-06,
|
|
"loss": 0.2595,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08492911607027054,
|
|
"step": 1020,
|
|
"valid_targets_mean": 6023.5,
|
|
"valid_targets_min": 560
|
|
},
|
|
{
|
|
"epoch": 5.283505154639175,
|
|
"grad_norm": 0.20849113781355197,
|
|
"learning_rate": 6.931068960040248e-06,
|
|
"loss": 0.2688,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1017090231180191,
|
|
"step": 1025,
|
|
"valid_targets_mean": 6501.2,
|
|
"valid_targets_min": 663
|
|
},
|
|
{
|
|
"epoch": 5.309278350515464,
|
|
"grad_norm": 0.2513768070437953,
|
|
"learning_rate": 6.737546835184101e-06,
|
|
"loss": 0.2632,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08266648650169373,
|
|
"step": 1030,
|
|
"valid_targets_mean": 5260.0,
|
|
"valid_targets_min": 828
|
|
},
|
|
{
|
|
"epoch": 5.335051546391752,
|
|
"grad_norm": 0.21151722209757032,
|
|
"learning_rate": 6.546216077873941e-06,
|
|
"loss": 0.2673,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08756110072135925,
|
|
"step": 1035,
|
|
"valid_targets_mean": 6145.3,
|
|
"valid_targets_min": 657
|
|
},
|
|
{
|
|
"epoch": 5.360824742268041,
|
|
"grad_norm": 0.22684065543714102,
|
|
"learning_rate": 6.3571083018698385e-06,
|
|
"loss": 0.2631,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09769861400127411,
|
|
"step": 1040,
|
|
"valid_targets_mean": 6334.3,
|
|
"valid_targets_min": 615
|
|
},
|
|
{
|
|
"epoch": 5.38659793814433,
|
|
"grad_norm": 0.20785089313976238,
|
|
"learning_rate": 6.17025475362657e-06,
|
|
"loss": 0.2594,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07463595271110535,
|
|
"step": 1045,
|
|
"valid_targets_mean": 5333.0,
|
|
"valid_targets_min": 345
|
|
},
|
|
{
|
|
"epoch": 5.412371134020619,
|
|
"grad_norm": 0.2213657046336635,
|
|
"learning_rate": 5.985686307130738e-06,
|
|
"loss": 0.2646,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09124470502138138,
|
|
"step": 1050,
|
|
"valid_targets_mean": 5560.1,
|
|
"valid_targets_min": 749
|
|
},
|
|
{
|
|
"epoch": 5.438144329896907,
|
|
"grad_norm": 0.21717198912442856,
|
|
"learning_rate": 5.80343345879947e-06,
|
|
"loss": 0.2628,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09935463219881058,
|
|
"step": 1055,
|
|
"valid_targets_mean": 5976.7,
|
|
"valid_targets_min": 748
|
|
},
|
|
{
|
|
"epoch": 5.463917525773196,
|
|
"grad_norm": 0.22034970299362733,
|
|
"learning_rate": 5.623526322441417e-06,
|
|
"loss": 0.2652,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09560218453407288,
|
|
"step": 1060,
|
|
"valid_targets_mean": 5304.5,
|
|
"valid_targets_min": 568
|
|
},
|
|
{
|
|
"epoch": 5.489690721649485,
|
|
"grad_norm": 0.21099542255888487,
|
|
"learning_rate": 5.445994624281059e-06,
|
|
"loss": 0.255,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07948188483715057,
|
|
"step": 1065,
|
|
"valid_targets_mean": 4990.7,
|
|
"valid_targets_min": 613
|
|
},
|
|
{
|
|
"epoch": 5.515463917525773,
|
|
"grad_norm": 0.24965216915028757,
|
|
"learning_rate": 5.270867698047009e-06,
|
|
"loss": 0.2588,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09611229598522186,
|
|
"step": 1070,
|
|
"valid_targets_mean": 6180.8,
|
|
"valid_targets_min": 839
|
|
},
|
|
{
|
|
"epoch": 5.541237113402062,
|
|
"grad_norm": 0.21537926365810592,
|
|
"learning_rate": 5.098174480125138e-06,
|
|
"loss": 0.266,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07472441345453262,
|
|
"step": 1075,
|
|
"valid_targets_mean": 5670.4,
|
|
"valid_targets_min": 685
|
|
},
|
|
{
|
|
"epoch": 5.56701030927835,
|
|
"grad_norm": 0.2062264532200198,
|
|
"learning_rate": 4.927943504777439e-06,
|
|
"loss": 0.2625,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08477666974067688,
|
|
"step": 1080,
|
|
"valid_targets_mean": 6193.4,
|
|
"valid_targets_min": 787
|
|
},
|
|
{
|
|
"epoch": 5.592783505154639,
|
|
"grad_norm": 0.22807479360755165,
|
|
"learning_rate": 4.760202899427251e-06,
|
|
"loss": 0.2662,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10754773020744324,
|
|
"step": 1085,
|
|
"valid_targets_mean": 6542.5,
|
|
"valid_targets_min": 716
|
|
},
|
|
{
|
|
"epoch": 5.618556701030927,
|
|
"grad_norm": 0.25063151638562114,
|
|
"learning_rate": 4.594980380011756e-06,
|
|
"loss": 0.2648,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08312106132507324,
|
|
"step": 1090,
|
|
"valid_targets_mean": 5348.1,
|
|
"valid_targets_min": 661
|
|
},
|
|
{
|
|
"epoch": 5.644329896907217,
|
|
"grad_norm": 0.2123995849638101,
|
|
"learning_rate": 4.432303246402423e-06,
|
|
"loss": 0.2609,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08281916379928589,
|
|
"step": 1095,
|
|
"valid_targets_mean": 5451.6,
|
|
"valid_targets_min": 682
|
|
},
|
|
{
|
|
"epoch": 5.670103092783505,
|
|
"grad_norm": 0.21442466144358932,
|
|
"learning_rate": 4.272198377894266e-06,
|
|
"loss": 0.2682,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08195334672927856,
|
|
"step": 1100,
|
|
"valid_targets_mean": 5530.9,
|
|
"valid_targets_min": 432
|
|
},
|
|
{
|
|
"epoch": 5.695876288659794,
|
|
"grad_norm": 0.2205772428631778,
|
|
"learning_rate": 4.114692228764506e-06,
|
|
"loss": 0.2615,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08610519766807556,
|
|
"step": 1105,
|
|
"valid_targets_mean": 5663.6,
|
|
"valid_targets_min": 659
|
|
},
|
|
{
|
|
"epoch": 5.721649484536083,
|
|
"grad_norm": 0.21322779902205768,
|
|
"learning_rate": 3.959810823901544e-06,
|
|
"loss": 0.2613,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08498895913362503,
|
|
"step": 1110,
|
|
"valid_targets_mean": 5395.2,
|
|
"valid_targets_min": 634
|
|
},
|
|
{
|
|
"epoch": 5.747422680412371,
|
|
"grad_norm": 0.20512049497797466,
|
|
"learning_rate": 3.8075797545048352e-06,
|
|
"loss": 0.2606,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10353204607963562,
|
|
"step": 1115,
|
|
"valid_targets_mean": 6570.0,
|
|
"valid_targets_min": 715
|
|
},
|
|
{
|
|
"epoch": 5.77319587628866,
|
|
"grad_norm": 0.20013648862760422,
|
|
"learning_rate": 3.658024173856407e-06,
|
|
"loss": 0.2607,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08529584854841232,
|
|
"step": 1120,
|
|
"valid_targets_mean": 5956.6,
|
|
"valid_targets_min": 654
|
|
},
|
|
{
|
|
"epoch": 5.798969072164948,
|
|
"grad_norm": 0.21094661785666766,
|
|
"learning_rate": 3.5111687931647984e-06,
|
|
"loss": 0.2605,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08977287262678146,
|
|
"step": 1125,
|
|
"valid_targets_mean": 6174.0,
|
|
"valid_targets_min": 775
|
|
},
|
|
{
|
|
"epoch": 5.824742268041237,
|
|
"grad_norm": 0.22215911229802232,
|
|
"learning_rate": 3.3670378774819602e-06,
|
|
"loss": 0.2577,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08660276234149933,
|
|
"step": 1130,
|
|
"valid_targets_mean": 5474.1,
|
|
"valid_targets_min": 804
|
|
},
|
|
{
|
|
"epoch": 5.850515463917525,
|
|
"grad_norm": 0.2133539586562487,
|
|
"learning_rate": 3.225655241693959e-06,
|
|
"loss": 0.2595,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0850963145494461,
|
|
"step": 1135,
|
|
"valid_targets_mean": 5754.8,
|
|
"valid_targets_min": 760
|
|
},
|
|
{
|
|
"epoch": 5.876288659793815,
|
|
"grad_norm": 0.20575597260563536,
|
|
"learning_rate": 3.0870442465859995e-06,
|
|
"loss": 0.2667,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08850501477718353,
|
|
"step": 1140,
|
|
"valid_targets_mean": 6029.8,
|
|
"valid_targets_min": 566
|
|
},
|
|
{
|
|
"epoch": 5.902061855670103,
|
|
"grad_norm": 0.2343081260192309,
|
|
"learning_rate": 2.9512277949824964e-06,
|
|
"loss": 0.2575,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08381161838769913,
|
|
"step": 1145,
|
|
"valid_targets_mean": 5860.2,
|
|
"valid_targets_min": 664
|
|
},
|
|
{
|
|
"epoch": 5.927835051546392,
|
|
"grad_norm": 0.20310701719507557,
|
|
"learning_rate": 2.8182283279628286e-06,
|
|
"loss": 0.2589,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0917319804430008,
|
|
"step": 1150,
|
|
"valid_targets_mean": 5665.4,
|
|
"valid_targets_min": 492
|
|
},
|
|
{
|
|
"epoch": 5.953608247422681,
|
|
"grad_norm": 0.21157422016009103,
|
|
"learning_rate": 2.688067821153384e-06,
|
|
"loss": 0.2542,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08067169785499573,
|
|
"step": 1155,
|
|
"valid_targets_mean": 5880.8,
|
|
"valid_targets_min": 831
|
|
},
|
|
{
|
|
"epoch": 5.979381443298969,
|
|
"grad_norm": 0.2526644899017494,
|
|
"learning_rate": 2.5607677810964936e-06,
|
|
"loss": 0.2565,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06718919426202774,
|
|
"step": 1160,
|
|
"valid_targets_mean": 5461.6,
|
|
"valid_targets_min": 481
|
|
},
|
|
{
|
|
"epoch": 6.005154639175258,
|
|
"grad_norm": 0.20806354652395836,
|
|
"learning_rate": 2.436349241696885e-06,
|
|
"loss": 0.2605,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07857988774776459,
|
|
"step": 1165,
|
|
"valid_targets_mean": 5099.2,
|
|
"valid_targets_min": 536
|
|
},
|
|
{
|
|
"epoch": 6.030927835051546,
|
|
"grad_norm": 0.19114186705459127,
|
|
"learning_rate": 2.314832760746244e-06,
|
|
"loss": 0.2577,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07678833603858948,
|
|
"step": 1170,
|
|
"valid_targets_mean": 5982.6,
|
|
"valid_targets_min": 521
|
|
},
|
|
{
|
|
"epoch": 6.056701030927835,
|
|
"grad_norm": 0.21278822715210383,
|
|
"learning_rate": 2.196238416526415e-06,
|
|
"loss": 0.2567,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08630746603012085,
|
|
"step": 1175,
|
|
"valid_targets_mean": 5385.6,
|
|
"valid_targets_min": 801
|
|
},
|
|
{
|
|
"epoch": 6.082474226804123,
|
|
"grad_norm": 0.19982606934036368,
|
|
"learning_rate": 2.0805858044918638e-06,
|
|
"loss": 0.2603,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07132503390312195,
|
|
"step": 1180,
|
|
"valid_targets_mean": 5036.2,
|
|
"valid_targets_min": 703
|
|
},
|
|
{
|
|
"epoch": 6.108247422680412,
|
|
"grad_norm": 0.2025276145294527,
|
|
"learning_rate": 1.9678940340318984e-06,
|
|
"loss": 0.2543,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08077646791934967,
|
|
"step": 1185,
|
|
"valid_targets_mean": 6119.1,
|
|
"valid_targets_min": 593
|
|
},
|
|
{
|
|
"epoch": 6.134020618556701,
|
|
"grad_norm": 0.2146826565190349,
|
|
"learning_rate": 1.8581817253131861e-06,
|
|
"loss": 0.2585,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08421634137630463,
|
|
"step": 1190,
|
|
"valid_targets_mean": 5804.8,
|
|
"valid_targets_min": 721
|
|
},
|
|
{
|
|
"epoch": 6.15979381443299,
|
|
"grad_norm": 0.21850253076866163,
|
|
"learning_rate": 1.751467006203158e-06,
|
|
"loss": 0.2571,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08505451679229736,
|
|
"step": 1195,
|
|
"valid_targets_mean": 5829.6,
|
|
"valid_targets_min": 438
|
|
},
|
|
{
|
|
"epoch": 6.185567010309279,
|
|
"grad_norm": 0.3345352586895775,
|
|
"learning_rate": 1.6477675092747113e-06,
|
|
"loss": 0.256,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08121395111083984,
|
|
"step": 1200,
|
|
"valid_targets_mean": 5465.2,
|
|
"valid_targets_min": 828
|
|
},
|
|
{
|
|
"epoch": 6.211340206185567,
|
|
"grad_norm": 0.20015585293909394,
|
|
"learning_rate": 1.5471003688927423e-06,
|
|
"loss": 0.2606,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09846120327711105,
|
|
"step": 1205,
|
|
"valid_targets_mean": 6461.5,
|
|
"valid_targets_min": 714
|
|
},
|
|
{
|
|
"epoch": 6.237113402061856,
|
|
"grad_norm": 0.22613313561787032,
|
|
"learning_rate": 1.4494822183830604e-06,
|
|
"loss": 0.2615,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.083452008664608,
|
|
"step": 1210,
|
|
"valid_targets_mean": 5220.8,
|
|
"valid_targets_min": 799
|
|
},
|
|
{
|
|
"epoch": 6.262886597938144,
|
|
"grad_norm": 0.21798846465090554,
|
|
"learning_rate": 1.3549291872840087e-06,
|
|
"loss": 0.2617,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08671265840530396,
|
|
"step": 1215,
|
|
"valid_targets_mean": 5615.6,
|
|
"valid_targets_min": 775
|
|
},
|
|
{
|
|
"epoch": 6.288659793814433,
|
|
"grad_norm": 0.206486585157575,
|
|
"learning_rate": 1.2634568986813878e-06,
|
|
"loss": 0.2652,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08534083515405655,
|
|
"step": 1220,
|
|
"valid_targets_mean": 5792.2,
|
|
"valid_targets_min": 616
|
|
},
|
|
{
|
|
"epoch": 6.314432989690721,
|
|
"grad_norm": 0.2625585223869907,
|
|
"learning_rate": 1.1750804666270365e-06,
|
|
"loss": 0.2576,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10572131723165512,
|
|
"step": 1225,
|
|
"valid_targets_mean": 6684.8,
|
|
"valid_targets_min": 811
|
|
},
|
|
{
|
|
"epoch": 6.34020618556701,
|
|
"grad_norm": 0.20979953889098482,
|
|
"learning_rate": 1.0898144936415168e-06,
|
|
"loss": 0.2672,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09444086253643036,
|
|
"step": 1230,
|
|
"valid_targets_mean": 5211.2,
|
|
"valid_targets_min": 562
|
|
},
|
|
{
|
|
"epoch": 6.365979381443299,
|
|
"grad_norm": 0.2039944878938327,
|
|
"learning_rate": 1.007673068301327e-06,
|
|
"loss": 0.2564,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08865610510110855,
|
|
"step": 1235,
|
|
"valid_targets_mean": 6050.2,
|
|
"valid_targets_min": 756
|
|
},
|
|
{
|
|
"epoch": 6.391752577319588,
|
|
"grad_norm": 0.20895313957264638,
|
|
"learning_rate": 9.286697629110319e-07,
|
|
"loss": 0.2603,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10159533470869064,
|
|
"step": 1240,
|
|
"valid_targets_mean": 6348.9,
|
|
"valid_targets_min": 749
|
|
},
|
|
{
|
|
"epoch": 6.417525773195877,
|
|
"grad_norm": 0.20966345980821396,
|
|
"learning_rate": 8.528176312607072e-07,
|
|
"loss": 0.2599,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08785787224769592,
|
|
"step": 1245,
|
|
"valid_targets_mean": 6008.3,
|
|
"valid_targets_min": 701
|
|
},
|
|
{
|
|
"epoch": 6.443298969072165,
|
|
"grad_norm": 0.20376284624443033,
|
|
"learning_rate": 7.80129206469038e-07,
|
|
"loss": 0.2599,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07996776700019836,
|
|
"step": 1250,
|
|
"valid_targets_mean": 5809.2,
|
|
"valid_targets_min": 763
|
|
},
|
|
{
|
|
"epoch": 6.469072164948454,
|
|
"grad_norm": 0.20952438750802332,
|
|
"learning_rate": 7.106164989124708e-07,
|
|
"loss": 0.2577,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10448366403579712,
|
|
"step": 1255,
|
|
"valid_targets_mean": 6406.7,
|
|
"valid_targets_min": 829
|
|
},
|
|
{
|
|
"epoch": 6.494845360824742,
|
|
"grad_norm": 0.20402910434899638,
|
|
"learning_rate": 6.442909942407105e-07,
|
|
"loss": 0.2589,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08333670347929001,
|
|
"step": 1260,
|
|
"valid_targets_mean": 6146.5,
|
|
"valid_targets_min": 721
|
|
},
|
|
{
|
|
"epoch": 6.520618556701031,
|
|
"grad_norm": 0.2039557306296135,
|
|
"learning_rate": 5.811636514789598e-07,
|
|
"loss": 0.275,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0835859477519989,
|
|
"step": 1265,
|
|
"valid_targets_mean": 6323.6,
|
|
"valid_targets_min": 732
|
|
},
|
|
{
|
|
"epoch": 6.546391752577319,
|
|
"grad_norm": 0.20320012215508462,
|
|
"learning_rate": 5.212449012171306e-07,
|
|
"loss": 0.2588,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09487879276275635,
|
|
"step": 1270,
|
|
"valid_targets_mean": 6336.9,
|
|
"valid_targets_min": 634
|
|
},
|
|
{
|
|
"epoch": 6.572164948453608,
|
|
"grad_norm": 0.2025812510858642,
|
|
"learning_rate": 4.64544643886391e-07,
|
|
"loss": 0.2622,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0825139656662941,
|
|
"step": 1275,
|
|
"valid_targets_mean": 5189.3,
|
|
"valid_targets_min": 644
|
|
},
|
|
{
|
|
"epoch": 6.597938144329897,
|
|
"grad_norm": 0.1982534662923328,
|
|
"learning_rate": 4.1107224812331867e-07,
|
|
"loss": 0.2633,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08331938087940216,
|
|
"step": 1280,
|
|
"valid_targets_mean": 5979.3,
|
|
"valid_targets_min": 549
|
|
},
|
|
{
|
|
"epoch": 6.623711340206185,
|
|
"grad_norm": 0.2056936358029195,
|
|
"learning_rate": 3.6083654922189237e-07,
|
|
"loss": 0.2535,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08464473485946655,
|
|
"step": 1285,
|
|
"valid_targets_mean": 5499.7,
|
|
"valid_targets_min": 594
|
|
},
|
|
{
|
|
"epoch": 6.649484536082475,
|
|
"grad_norm": 0.19965629057607698,
|
|
"learning_rate": 3.1384584767365276e-07,
|
|
"loss": 0.2625,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08106464892625809,
|
|
"step": 1290,
|
|
"valid_targets_mean": 6474.9,
|
|
"valid_targets_min": 475
|
|
},
|
|
{
|
|
"epoch": 6.675257731958763,
|
|
"grad_norm": 0.21892884828822426,
|
|
"learning_rate": 2.701079077961821e-07,
|
|
"loss": 0.2582,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08687181770801544,
|
|
"step": 1295,
|
|
"valid_targets_mean": 5231.6,
|
|
"valid_targets_min": 666
|
|
},
|
|
{
|
|
"epoch": 6.701030927835052,
|
|
"grad_norm": 0.20896179005298743,
|
|
"learning_rate": 2.2962995645021469e-07,
|
|
"loss": 0.263,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08677627146244049,
|
|
"step": 1300,
|
|
"valid_targets_mean": 5321.3,
|
|
"valid_targets_min": 497
|
|
},
|
|
{
|
|
"epoch": 6.72680412371134,
|
|
"grad_norm": 0.20409776458009016,
|
|
"learning_rate": 1.9241868184552537e-07,
|
|
"loss": 0.2718,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08831295371055603,
|
|
"step": 1305,
|
|
"valid_targets_mean": 6504.9,
|
|
"valid_targets_min": 820
|
|
},
|
|
{
|
|
"epoch": 6.752577319587629,
|
|
"grad_norm": 0.20230584513721114,
|
|
"learning_rate": 1.5848023243583143e-07,
|
|
"loss": 0.261,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09413200616836548,
|
|
"step": 1310,
|
|
"valid_targets_mean": 6212.2,
|
|
"valid_targets_min": 689
|
|
},
|
|
{
|
|
"epoch": 6.778350515463917,
|
|
"grad_norm": 0.21423837116119837,
|
|
"learning_rate": 1.278202159028896e-07,
|
|
"loss": 0.2663,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07308657467365265,
|
|
"step": 1315,
|
|
"valid_targets_mean": 4973.8,
|
|
"valid_targets_min": 675
|
|
},
|
|
{
|
|
"epoch": 6.804123711340206,
|
|
"grad_norm": 0.1912335151833021,
|
|
"learning_rate": 1.0044369822991729e-07,
|
|
"loss": 0.2565,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0846131220459938,
|
|
"step": 1320,
|
|
"valid_targets_mean": 6746.4,
|
|
"valid_targets_min": 707
|
|
},
|
|
{
|
|
"epoch": 6.829896907216495,
|
|
"grad_norm": 0.2214596123527528,
|
|
"learning_rate": 7.635520286454868e-08,
|
|
"loss": 0.2505,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09434468299150467,
|
|
"step": 1325,
|
|
"valid_targets_mean": 5639.7,
|
|
"valid_targets_min": 661
|
|
},
|
|
{
|
|
"epoch": 6.855670103092783,
|
|
"grad_norm": 0.20921188363271367,
|
|
"learning_rate": 5.555870997140833e-08,
|
|
"loss": 0.2571,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07504785060882568,
|
|
"step": 1330,
|
|
"valid_targets_mean": 5425.8,
|
|
"valid_targets_min": 759
|
|
},
|
|
{
|
|
"epoch": 6.881443298969073,
|
|
"grad_norm": 0.20876849858807778,
|
|
"learning_rate": 3.8057655774479395e-08,
|
|
"loss": 0.2577,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09874755144119263,
|
|
"step": 1335,
|
|
"valid_targets_mean": 6840.6,
|
|
"valid_targets_min": 694
|
|
},
|
|
{
|
|
"epoch": 6.907216494845361,
|
|
"grad_norm": 0.20718159609812295,
|
|
"learning_rate": 2.3854931989319986e-08,
|
|
"loss": 0.2604,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08516525477170944,
|
|
"step": 1340,
|
|
"valid_targets_mean": 6252.4,
|
|
"valid_targets_min": 804
|
|
},
|
|
{
|
|
"epoch": 6.93298969072165,
|
|
"grad_norm": 0.2050064910368948,
|
|
"learning_rate": 1.295288534527206e-08,
|
|
"loss": 0.26,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09851793199777603,
|
|
"step": 1345,
|
|
"valid_targets_mean": 6319.2,
|
|
"valid_targets_min": 734
|
|
},
|
|
{
|
|
"epoch": 6.958762886597938,
|
|
"grad_norm": 0.2216176200645684,
|
|
"learning_rate": 5.353317197702712e-09,
|
|
"loss": 0.2617,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10352624952793121,
|
|
"step": 1350,
|
|
"valid_targets_mean": 6360.4,
|
|
"valid_targets_min": 669
|
|
},
|
|
{
|
|
"epoch": 6.984536082474227,
|
|
"grad_norm": 0.20373965786689494,
|
|
"learning_rate": 1.0574832303689164e-09,
|
|
"loss": 0.2569,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08267886936664581,
|
|
"step": 1355,
|
|
"valid_targets_mean": 6156.2,
|
|
"valid_targets_min": 664
|
|
},
|
|
{
|
|
"epoch": 7.0,
|
|
"step": 1358,
|
|
"total_flos": 6.436468142603502e+18,
|
|
"train_loss": 0.0,
|
|
"train_runtime": 0.8245,
|
|
"train_samples_per_second": 157900.906,
|
|
"train_steps_per_second": 1647.012
|
|
}
|
|
],
|
|
"logging_steps": 5,
|
|
"max_steps": 1358,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 7,
|
|
"save_steps": 300,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": true
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 6.436468142603502e+18,
|
|
"train_batch_size": 1,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|