9684 lines
269 KiB
JSON
9684 lines
269 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 7.0,
|
|
"eval_steps": 500,
|
|
"global_step": 4382,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.007987220447284345,
|
|
"grad_norm": 16.214843166507823,
|
|
"learning_rate": 3.644646924829157e-07,
|
|
"loss": 0.6667,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.6293683052062988,
|
|
"step": 5,
|
|
"valid_targets_mean": 5840.6,
|
|
"valid_targets_min": 1683
|
|
},
|
|
{
|
|
"epoch": 0.01597444089456869,
|
|
"grad_norm": 16.536060280334816,
|
|
"learning_rate": 8.200455580865605e-07,
|
|
"loss": 0.7237,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.7005735635757446,
|
|
"step": 10,
|
|
"valid_targets_mean": 4628.3,
|
|
"valid_targets_min": 851
|
|
},
|
|
{
|
|
"epoch": 0.023961661341853034,
|
|
"grad_norm": 15.852663272157008,
|
|
"learning_rate": 1.2756264236902052e-06,
|
|
"loss": 0.711,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.665267825126648,
|
|
"step": 15,
|
|
"valid_targets_mean": 5435.8,
|
|
"valid_targets_min": 3190
|
|
},
|
|
{
|
|
"epoch": 0.03194888178913738,
|
|
"grad_norm": 10.711815863929866,
|
|
"learning_rate": 1.7312072892938498e-06,
|
|
"loss": 0.6221,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.5724351406097412,
|
|
"step": 20,
|
|
"valid_targets_mean": 4856.0,
|
|
"valid_targets_min": 924
|
|
},
|
|
{
|
|
"epoch": 0.039936102236421724,
|
|
"grad_norm": 7.345110084184065,
|
|
"learning_rate": 2.1867881548974945e-06,
|
|
"loss": 0.5904,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.49307093024253845,
|
|
"step": 25,
|
|
"valid_targets_mean": 4609.2,
|
|
"valid_targets_min": 566
|
|
},
|
|
{
|
|
"epoch": 0.04792332268370607,
|
|
"grad_norm": 4.330174769661457,
|
|
"learning_rate": 2.642369020501139e-06,
|
|
"loss": 0.5453,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.47903314232826233,
|
|
"step": 30,
|
|
"valid_targets_mean": 5747.5,
|
|
"valid_targets_min": 1355
|
|
},
|
|
{
|
|
"epoch": 0.05591054313099041,
|
|
"grad_norm": 2.4085938196558376,
|
|
"learning_rate": 3.0979498861047843e-06,
|
|
"loss": 0.4956,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4651956558227539,
|
|
"step": 35,
|
|
"valid_targets_mean": 5118.4,
|
|
"valid_targets_min": 800
|
|
},
|
|
{
|
|
"epoch": 0.06389776357827476,
|
|
"grad_norm": 1.516310546910285,
|
|
"learning_rate": 3.5535307517084285e-06,
|
|
"loss": 0.4872,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4683207869529724,
|
|
"step": 40,
|
|
"valid_targets_mean": 4522.1,
|
|
"valid_targets_min": 838
|
|
},
|
|
{
|
|
"epoch": 0.07188498402555911,
|
|
"grad_norm": 1.0901568008604194,
|
|
"learning_rate": 4.009111617312073e-06,
|
|
"loss": 0.434,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4305180311203003,
|
|
"step": 45,
|
|
"valid_targets_mean": 5100.6,
|
|
"valid_targets_min": 829
|
|
},
|
|
{
|
|
"epoch": 0.07987220447284345,
|
|
"grad_norm": 1.0694363953729904,
|
|
"learning_rate": 4.464692482915718e-06,
|
|
"loss": 0.4373,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4268118143081665,
|
|
"step": 50,
|
|
"valid_targets_mean": 3879.9,
|
|
"valid_targets_min": 590
|
|
},
|
|
{
|
|
"epoch": 0.0878594249201278,
|
|
"grad_norm": 0.9467308768467405,
|
|
"learning_rate": 4.920273348519363e-06,
|
|
"loss": 0.4271,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.475871741771698,
|
|
"step": 55,
|
|
"valid_targets_mean": 3926.7,
|
|
"valid_targets_min": 2067
|
|
},
|
|
{
|
|
"epoch": 0.09584664536741214,
|
|
"grad_norm": 0.7979303257083482,
|
|
"learning_rate": 5.375854214123008e-06,
|
|
"loss": 0.4191,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4331532418727875,
|
|
"step": 60,
|
|
"valid_targets_mean": 5110.4,
|
|
"valid_targets_min": 886
|
|
},
|
|
{
|
|
"epoch": 0.10383386581469649,
|
|
"grad_norm": 0.5833315958168446,
|
|
"learning_rate": 5.831435079726651e-06,
|
|
"loss": 0.3782,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.33264434337615967,
|
|
"step": 65,
|
|
"valid_targets_mean": 5157.1,
|
|
"valid_targets_min": 920
|
|
},
|
|
{
|
|
"epoch": 0.11182108626198083,
|
|
"grad_norm": 0.534341175313362,
|
|
"learning_rate": 6.287015945330297e-06,
|
|
"loss": 0.3964,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.36213618516921997,
|
|
"step": 70,
|
|
"valid_targets_mean": 5796.7,
|
|
"valid_targets_min": 1392
|
|
},
|
|
{
|
|
"epoch": 0.11980830670926518,
|
|
"grad_norm": 0.6474280808218216,
|
|
"learning_rate": 6.742596810933942e-06,
|
|
"loss": 0.4077,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4014986455440521,
|
|
"step": 75,
|
|
"valid_targets_mean": 4715.4,
|
|
"valid_targets_min": 1523
|
|
},
|
|
{
|
|
"epoch": 0.12779552715654952,
|
|
"grad_norm": 0.6245790708527251,
|
|
"learning_rate": 7.1981776765375854e-06,
|
|
"loss": 0.371,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.36042672395706177,
|
|
"step": 80,
|
|
"valid_targets_mean": 4445.1,
|
|
"valid_targets_min": 1210
|
|
},
|
|
{
|
|
"epoch": 0.13578274760383385,
|
|
"grad_norm": 0.5352596369045343,
|
|
"learning_rate": 7.65375854214123e-06,
|
|
"loss": 0.3556,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.36412060260772705,
|
|
"step": 85,
|
|
"valid_targets_mean": 5408.6,
|
|
"valid_targets_min": 1124
|
|
},
|
|
{
|
|
"epoch": 0.14376996805111822,
|
|
"grad_norm": 0.522019132641173,
|
|
"learning_rate": 8.109339407744875e-06,
|
|
"loss": 0.3712,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3510782718658447,
|
|
"step": 90,
|
|
"valid_targets_mean": 5306.1,
|
|
"valid_targets_min": 1987
|
|
},
|
|
{
|
|
"epoch": 0.15175718849840256,
|
|
"grad_norm": 0.5834541923968949,
|
|
"learning_rate": 8.564920273348521e-06,
|
|
"loss": 0.3543,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.37941974401474,
|
|
"step": 95,
|
|
"valid_targets_mean": 4713.2,
|
|
"valid_targets_min": 857
|
|
},
|
|
{
|
|
"epoch": 0.1597444089456869,
|
|
"grad_norm": 0.5977305337467345,
|
|
"learning_rate": 9.020501138952164e-06,
|
|
"loss": 0.3535,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.34058240056037903,
|
|
"step": 100,
|
|
"valid_targets_mean": 4175.4,
|
|
"valid_targets_min": 901
|
|
},
|
|
{
|
|
"epoch": 0.16773162939297126,
|
|
"grad_norm": 0.5777396640406296,
|
|
"learning_rate": 9.47608200455581e-06,
|
|
"loss": 0.3411,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.37314867973327637,
|
|
"step": 105,
|
|
"valid_targets_mean": 4329.9,
|
|
"valid_targets_min": 320
|
|
},
|
|
{
|
|
"epoch": 0.1757188498402556,
|
|
"grad_norm": 0.5383539023704225,
|
|
"learning_rate": 9.931662870159453e-06,
|
|
"loss": 0.357,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3683736324310303,
|
|
"step": 110,
|
|
"valid_targets_mean": 5286.6,
|
|
"valid_targets_min": 1453
|
|
},
|
|
{
|
|
"epoch": 0.18370607028753994,
|
|
"grad_norm": 0.7872582146315561,
|
|
"learning_rate": 1.03872437357631e-05,
|
|
"loss": 0.326,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.30175650119781494,
|
|
"step": 115,
|
|
"valid_targets_mean": 4889.8,
|
|
"valid_targets_min": 1692
|
|
},
|
|
{
|
|
"epoch": 0.19169329073482427,
|
|
"grad_norm": 0.598028210832286,
|
|
"learning_rate": 1.0842824601366744e-05,
|
|
"loss": 0.3108,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29703736305236816,
|
|
"step": 120,
|
|
"valid_targets_mean": 4932.6,
|
|
"valid_targets_min": 657
|
|
},
|
|
{
|
|
"epoch": 0.19968051118210864,
|
|
"grad_norm": 0.5924147320645817,
|
|
"learning_rate": 1.1298405466970387e-05,
|
|
"loss": 0.338,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.30937889218330383,
|
|
"step": 125,
|
|
"valid_targets_mean": 5179.6,
|
|
"valid_targets_min": 1355
|
|
},
|
|
{
|
|
"epoch": 0.20766773162939298,
|
|
"grad_norm": 0.5681641346639285,
|
|
"learning_rate": 1.1753986332574032e-05,
|
|
"loss": 0.3357,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.34611114859580994,
|
|
"step": 130,
|
|
"valid_targets_mean": 5334.6,
|
|
"valid_targets_min": 1313
|
|
},
|
|
{
|
|
"epoch": 0.21565495207667731,
|
|
"grad_norm": 0.5082080515535616,
|
|
"learning_rate": 1.2209567198177677e-05,
|
|
"loss": 0.3366,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3541598618030548,
|
|
"step": 135,
|
|
"valid_targets_mean": 5861.4,
|
|
"valid_targets_min": 2080
|
|
},
|
|
{
|
|
"epoch": 0.22364217252396165,
|
|
"grad_norm": 0.5598608784555669,
|
|
"learning_rate": 1.2665148063781323e-05,
|
|
"loss": 0.2984,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.31298428773880005,
|
|
"step": 140,
|
|
"valid_targets_mean": 4757.7,
|
|
"valid_targets_min": 2006
|
|
},
|
|
{
|
|
"epoch": 0.23162939297124602,
|
|
"grad_norm": 0.5517176609096509,
|
|
"learning_rate": 1.3120728929384968e-05,
|
|
"loss": 0.3104,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28146520256996155,
|
|
"step": 145,
|
|
"valid_targets_mean": 4815.2,
|
|
"valid_targets_min": 714
|
|
},
|
|
{
|
|
"epoch": 0.23961661341853036,
|
|
"grad_norm": 0.6508519385951486,
|
|
"learning_rate": 1.357630979498861e-05,
|
|
"loss": 0.3298,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.34707149863243103,
|
|
"step": 150,
|
|
"valid_targets_mean": 3598.0,
|
|
"valid_targets_min": 735
|
|
},
|
|
{
|
|
"epoch": 0.2476038338658147,
|
|
"grad_norm": 0.5496388523379975,
|
|
"learning_rate": 1.4031890660592255e-05,
|
|
"loss": 0.305,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29744917154312134,
|
|
"step": 155,
|
|
"valid_targets_mean": 4697.5,
|
|
"valid_targets_min": 933
|
|
},
|
|
{
|
|
"epoch": 0.25559105431309903,
|
|
"grad_norm": 0.4789728949464768,
|
|
"learning_rate": 1.4487471526195902e-05,
|
|
"loss": 0.3128,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.30035269260406494,
|
|
"step": 160,
|
|
"valid_targets_mean": 5251.2,
|
|
"valid_targets_min": 1848
|
|
},
|
|
{
|
|
"epoch": 0.26357827476038337,
|
|
"grad_norm": 0.6371209181942088,
|
|
"learning_rate": 1.4943052391799546e-05,
|
|
"loss": 0.3018,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3188396692276001,
|
|
"step": 165,
|
|
"valid_targets_mean": 4774.2,
|
|
"valid_targets_min": 363
|
|
},
|
|
{
|
|
"epoch": 0.2715654952076677,
|
|
"grad_norm": 0.5513751212341264,
|
|
"learning_rate": 1.539863325740319e-05,
|
|
"loss": 0.2944,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27828067541122437,
|
|
"step": 170,
|
|
"valid_targets_mean": 4899.8,
|
|
"valid_targets_min": 1382
|
|
},
|
|
{
|
|
"epoch": 0.2795527156549521,
|
|
"grad_norm": 0.47809556821894617,
|
|
"learning_rate": 1.5854214123006836e-05,
|
|
"loss": 0.3152,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.33486631512641907,
|
|
"step": 175,
|
|
"valid_targets_mean": 5406.3,
|
|
"valid_targets_min": 546
|
|
},
|
|
{
|
|
"epoch": 0.28753993610223644,
|
|
"grad_norm": 0.5552245835273288,
|
|
"learning_rate": 1.630979498861048e-05,
|
|
"loss": 0.29,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2863185405731201,
|
|
"step": 180,
|
|
"valid_targets_mean": 3716.8,
|
|
"valid_targets_min": 813
|
|
},
|
|
{
|
|
"epoch": 0.2955271565495208,
|
|
"grad_norm": 0.7370634203226984,
|
|
"learning_rate": 1.6765375854214125e-05,
|
|
"loss": 0.2889,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.30801016092300415,
|
|
"step": 185,
|
|
"valid_targets_mean": 4590.7,
|
|
"valid_targets_min": 596
|
|
},
|
|
{
|
|
"epoch": 0.3035143769968051,
|
|
"grad_norm": 0.5009946980417955,
|
|
"learning_rate": 1.722095671981777e-05,
|
|
"loss": 0.3077,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2993805706501007,
|
|
"step": 190,
|
|
"valid_targets_mean": 5422.0,
|
|
"valid_targets_min": 288
|
|
},
|
|
{
|
|
"epoch": 0.31150159744408945,
|
|
"grad_norm": 0.5358598801386117,
|
|
"learning_rate": 1.7676537585421415e-05,
|
|
"loss": 0.2763,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29835355281829834,
|
|
"step": 195,
|
|
"valid_targets_mean": 5053.3,
|
|
"valid_targets_min": 924
|
|
},
|
|
{
|
|
"epoch": 0.3194888178913738,
|
|
"grad_norm": 1.1064965236867108,
|
|
"learning_rate": 1.813211845102506e-05,
|
|
"loss": 0.2933,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28204622864723206,
|
|
"step": 200,
|
|
"valid_targets_mean": 4445.6,
|
|
"valid_targets_min": 329
|
|
},
|
|
{
|
|
"epoch": 0.3274760383386581,
|
|
"grad_norm": 0.6238934475229897,
|
|
"learning_rate": 1.8587699316628704e-05,
|
|
"loss": 0.3052,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28925684094429016,
|
|
"step": 205,
|
|
"valid_targets_mean": 4103.9,
|
|
"valid_targets_min": 737
|
|
},
|
|
{
|
|
"epoch": 0.3354632587859425,
|
|
"grad_norm": 0.5595733482793672,
|
|
"learning_rate": 1.904328018223235e-05,
|
|
"loss": 0.2935,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29793277382850647,
|
|
"step": 210,
|
|
"valid_targets_mean": 4934.9,
|
|
"valid_targets_min": 1633
|
|
},
|
|
{
|
|
"epoch": 0.34345047923322686,
|
|
"grad_norm": 0.5589513052143709,
|
|
"learning_rate": 1.9498861047835993e-05,
|
|
"loss": 0.3022,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3029059171676636,
|
|
"step": 215,
|
|
"valid_targets_mean": 4674.2,
|
|
"valid_targets_min": 2145
|
|
},
|
|
{
|
|
"epoch": 0.3514376996805112,
|
|
"grad_norm": 0.5602768175734714,
|
|
"learning_rate": 1.9954441913439638e-05,
|
|
"loss": 0.2844,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2719416916370392,
|
|
"step": 220,
|
|
"valid_targets_mean": 5259.6,
|
|
"valid_targets_min": 2728
|
|
},
|
|
{
|
|
"epoch": 0.35942492012779553,
|
|
"grad_norm": 0.5840002344805434,
|
|
"learning_rate": 2.0410022779043283e-05,
|
|
"loss": 0.2622,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.252376914024353,
|
|
"step": 225,
|
|
"valid_targets_mean": 3708.0,
|
|
"valid_targets_min": 1563
|
|
},
|
|
{
|
|
"epoch": 0.36741214057507987,
|
|
"grad_norm": 0.4988357514360904,
|
|
"learning_rate": 2.0865603644646927e-05,
|
|
"loss": 0.2785,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2672957479953766,
|
|
"step": 230,
|
|
"valid_targets_mean": 5539.1,
|
|
"valid_targets_min": 1948
|
|
},
|
|
{
|
|
"epoch": 0.3753993610223642,
|
|
"grad_norm": 0.5265055363647089,
|
|
"learning_rate": 2.1321184510250572e-05,
|
|
"loss": 0.2832,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2662750482559204,
|
|
"step": 235,
|
|
"valid_targets_mean": 5089.4,
|
|
"valid_targets_min": 2441
|
|
},
|
|
{
|
|
"epoch": 0.38338658146964855,
|
|
"grad_norm": 0.6319243139561913,
|
|
"learning_rate": 2.1776765375854217e-05,
|
|
"loss": 0.2694,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2903909683227539,
|
|
"step": 240,
|
|
"valid_targets_mean": 4345.3,
|
|
"valid_targets_min": 1426
|
|
},
|
|
{
|
|
"epoch": 0.3913738019169329,
|
|
"grad_norm": 0.5332922577584055,
|
|
"learning_rate": 2.223234624145786e-05,
|
|
"loss": 0.2692,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2847251892089844,
|
|
"step": 245,
|
|
"valid_targets_mean": 5051.9,
|
|
"valid_targets_min": 601
|
|
},
|
|
{
|
|
"epoch": 0.3993610223642173,
|
|
"grad_norm": 0.530308481283379,
|
|
"learning_rate": 2.2687927107061506e-05,
|
|
"loss": 0.2896,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2778535485267639,
|
|
"step": 250,
|
|
"valid_targets_mean": 4761.2,
|
|
"valid_targets_min": 1097
|
|
},
|
|
{
|
|
"epoch": 0.4073482428115016,
|
|
"grad_norm": 0.6098521445637076,
|
|
"learning_rate": 2.314350797266515e-05,
|
|
"loss": 0.297,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27581244707107544,
|
|
"step": 255,
|
|
"valid_targets_mean": 3908.1,
|
|
"valid_targets_min": 1588
|
|
},
|
|
{
|
|
"epoch": 0.41533546325878595,
|
|
"grad_norm": 0.5921423858911565,
|
|
"learning_rate": 2.3599088838268792e-05,
|
|
"loss": 0.2734,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2547699809074402,
|
|
"step": 260,
|
|
"valid_targets_mean": 5054.8,
|
|
"valid_targets_min": 358
|
|
},
|
|
{
|
|
"epoch": 0.4233226837060703,
|
|
"grad_norm": 0.5289923217272506,
|
|
"learning_rate": 2.4054669703872436e-05,
|
|
"loss": 0.2595,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2866251468658447,
|
|
"step": 265,
|
|
"valid_targets_mean": 5726.3,
|
|
"valid_targets_min": 2125
|
|
},
|
|
{
|
|
"epoch": 0.43130990415335463,
|
|
"grad_norm": 0.5537781627686953,
|
|
"learning_rate": 2.4510250569476085e-05,
|
|
"loss": 0.3091,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.32708555459976196,
|
|
"step": 270,
|
|
"valid_targets_mean": 5520.9,
|
|
"valid_targets_min": 3173
|
|
},
|
|
{
|
|
"epoch": 0.43929712460063897,
|
|
"grad_norm": 0.5104465395111398,
|
|
"learning_rate": 2.496583143507973e-05,
|
|
"loss": 0.2773,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3148839771747589,
|
|
"step": 275,
|
|
"valid_targets_mean": 5590.6,
|
|
"valid_targets_min": 2062
|
|
},
|
|
{
|
|
"epoch": 0.4472843450479233,
|
|
"grad_norm": 0.5551638885961803,
|
|
"learning_rate": 2.5421412300683374e-05,
|
|
"loss": 0.2862,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28861215710639954,
|
|
"step": 280,
|
|
"valid_targets_mean": 4605.2,
|
|
"valid_targets_min": 2405
|
|
},
|
|
{
|
|
"epoch": 0.45527156549520764,
|
|
"grad_norm": 0.5774373008716863,
|
|
"learning_rate": 2.587699316628702e-05,
|
|
"loss": 0.2948,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2806679904460907,
|
|
"step": 285,
|
|
"valid_targets_mean": 4729.7,
|
|
"valid_targets_min": 1175
|
|
},
|
|
{
|
|
"epoch": 0.46325878594249204,
|
|
"grad_norm": 0.5428102603769059,
|
|
"learning_rate": 2.6332574031890663e-05,
|
|
"loss": 0.2946,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.32112547755241394,
|
|
"step": 290,
|
|
"valid_targets_mean": 5787.0,
|
|
"valid_targets_min": 1645
|
|
},
|
|
{
|
|
"epoch": 0.4712460063897764,
|
|
"grad_norm": 0.521358822134473,
|
|
"learning_rate": 2.6788154897494308e-05,
|
|
"loss": 0.2796,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28433695435523987,
|
|
"step": 295,
|
|
"valid_targets_mean": 5907.2,
|
|
"valid_targets_min": 1100
|
|
},
|
|
{
|
|
"epoch": 0.4792332268370607,
|
|
"grad_norm": 0.5337346140514683,
|
|
"learning_rate": 2.7243735763097953e-05,
|
|
"loss": 0.2778,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2725232243537903,
|
|
"step": 300,
|
|
"valid_targets_mean": 5427.4,
|
|
"valid_targets_min": 2111
|
|
},
|
|
{
|
|
"epoch": 0.48722044728434505,
|
|
"grad_norm": 0.5470089886286064,
|
|
"learning_rate": 2.7699316628701597e-05,
|
|
"loss": 0.2721,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26878929138183594,
|
|
"step": 305,
|
|
"valid_targets_mean": 4654.0,
|
|
"valid_targets_min": 728
|
|
},
|
|
{
|
|
"epoch": 0.4952076677316294,
|
|
"grad_norm": 0.5190354642636552,
|
|
"learning_rate": 2.815489749430524e-05,
|
|
"loss": 0.2708,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2548917233943939,
|
|
"step": 310,
|
|
"valid_targets_mean": 5067.2,
|
|
"valid_targets_min": 1822
|
|
},
|
|
{
|
|
"epoch": 0.5031948881789138,
|
|
"grad_norm": 0.5687012066316108,
|
|
"learning_rate": 2.8610478359908883e-05,
|
|
"loss": 0.2798,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2866740822792053,
|
|
"step": 315,
|
|
"valid_targets_mean": 5067.9,
|
|
"valid_targets_min": 1552
|
|
},
|
|
{
|
|
"epoch": 0.5111821086261981,
|
|
"grad_norm": 0.5373160925653192,
|
|
"learning_rate": 2.906605922551253e-05,
|
|
"loss": 0.2718,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27760303020477295,
|
|
"step": 320,
|
|
"valid_targets_mean": 5024.9,
|
|
"valid_targets_min": 1192
|
|
},
|
|
{
|
|
"epoch": 0.5191693290734825,
|
|
"grad_norm": 0.5362070010772059,
|
|
"learning_rate": 2.9521640091116176e-05,
|
|
"loss": 0.2727,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28479108214378357,
|
|
"step": 325,
|
|
"valid_targets_mean": 5303.2,
|
|
"valid_targets_min": 2418
|
|
},
|
|
{
|
|
"epoch": 0.5271565495207667,
|
|
"grad_norm": 0.5848089779265616,
|
|
"learning_rate": 2.997722095671982e-05,
|
|
"loss": 0.2645,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26311156153678894,
|
|
"step": 330,
|
|
"valid_targets_mean": 4052.6,
|
|
"valid_targets_min": 337
|
|
},
|
|
{
|
|
"epoch": 0.5351437699680511,
|
|
"grad_norm": 0.5493496138236031,
|
|
"learning_rate": 3.0432801822323465e-05,
|
|
"loss": 0.267,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2734183967113495,
|
|
"step": 335,
|
|
"valid_targets_mean": 5034.6,
|
|
"valid_targets_min": 2061
|
|
},
|
|
{
|
|
"epoch": 0.5431309904153354,
|
|
"grad_norm": 0.5635667614887097,
|
|
"learning_rate": 3.088838268792711e-05,
|
|
"loss": 0.2864,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3333064615726471,
|
|
"step": 340,
|
|
"valid_targets_mean": 5823.9,
|
|
"valid_targets_min": 333
|
|
},
|
|
{
|
|
"epoch": 0.5511182108626198,
|
|
"grad_norm": 0.5019872954405129,
|
|
"learning_rate": 3.1343963553530755e-05,
|
|
"loss": 0.2706,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29065948724746704,
|
|
"step": 345,
|
|
"valid_targets_mean": 5628.3,
|
|
"valid_targets_min": 2412
|
|
},
|
|
{
|
|
"epoch": 0.5591054313099042,
|
|
"grad_norm": 0.5374160693839819,
|
|
"learning_rate": 3.17995444191344e-05,
|
|
"loss": 0.2604,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25003743171691895,
|
|
"step": 350,
|
|
"valid_targets_mean": 4885.5,
|
|
"valid_targets_min": 1883
|
|
},
|
|
{
|
|
"epoch": 0.5670926517571885,
|
|
"grad_norm": 0.5230300161449793,
|
|
"learning_rate": 3.2255125284738044e-05,
|
|
"loss": 0.2593,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3082168996334076,
|
|
"step": 355,
|
|
"valid_targets_mean": 5003.2,
|
|
"valid_targets_min": 919
|
|
},
|
|
{
|
|
"epoch": 0.5750798722044729,
|
|
"grad_norm": 0.5801770052473694,
|
|
"learning_rate": 3.271070615034169e-05,
|
|
"loss": 0.2514,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29632386565208435,
|
|
"step": 360,
|
|
"valid_targets_mean": 4708.2,
|
|
"valid_targets_min": 948
|
|
},
|
|
{
|
|
"epoch": 0.5830670926517572,
|
|
"grad_norm": 0.5472241673959384,
|
|
"learning_rate": 3.316628701594533e-05,
|
|
"loss": 0.286,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25985270738601685,
|
|
"step": 365,
|
|
"valid_targets_mean": 4563.3,
|
|
"valid_targets_min": 997
|
|
},
|
|
{
|
|
"epoch": 0.5910543130990416,
|
|
"grad_norm": 0.5149939473439978,
|
|
"learning_rate": 3.362186788154898e-05,
|
|
"loss": 0.2661,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27894413471221924,
|
|
"step": 370,
|
|
"valid_targets_mean": 5198.9,
|
|
"valid_targets_min": 720
|
|
},
|
|
{
|
|
"epoch": 0.5990415335463258,
|
|
"grad_norm": 0.580767045784304,
|
|
"learning_rate": 3.407744874715262e-05,
|
|
"loss": 0.2671,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2662809491157532,
|
|
"step": 375,
|
|
"valid_targets_mean": 4736.9,
|
|
"valid_targets_min": 521
|
|
},
|
|
{
|
|
"epoch": 0.6070287539936102,
|
|
"grad_norm": 0.5837236073120515,
|
|
"learning_rate": 3.453302961275627e-05,
|
|
"loss": 0.2658,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29299938678741455,
|
|
"step": 380,
|
|
"valid_targets_mean": 4359.5,
|
|
"valid_targets_min": 626
|
|
},
|
|
{
|
|
"epoch": 0.6150159744408946,
|
|
"grad_norm": 0.5683559146502843,
|
|
"learning_rate": 3.498861047835991e-05,
|
|
"loss": 0.2605,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2716815173625946,
|
|
"step": 385,
|
|
"valid_targets_mean": 4500.7,
|
|
"valid_targets_min": 1009
|
|
},
|
|
{
|
|
"epoch": 0.6230031948881789,
|
|
"grad_norm": 0.5521731757690844,
|
|
"learning_rate": 3.5444191343963557e-05,
|
|
"loss": 0.2642,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2861851453781128,
|
|
"step": 390,
|
|
"valid_targets_mean": 4939.9,
|
|
"valid_targets_min": 1717
|
|
},
|
|
{
|
|
"epoch": 0.6309904153354633,
|
|
"grad_norm": 0.5454459023619236,
|
|
"learning_rate": 3.58997722095672e-05,
|
|
"loss": 0.271,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2945636212825775,
|
|
"step": 395,
|
|
"valid_targets_mean": 5303.6,
|
|
"valid_targets_min": 1049
|
|
},
|
|
{
|
|
"epoch": 0.6389776357827476,
|
|
"grad_norm": 0.537113572597406,
|
|
"learning_rate": 3.6355353075170846e-05,
|
|
"loss": 0.2666,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2924715578556061,
|
|
"step": 400,
|
|
"valid_targets_mean": 5877.7,
|
|
"valid_targets_min": 1820
|
|
},
|
|
{
|
|
"epoch": 0.646964856230032,
|
|
"grad_norm": 0.5741616450872754,
|
|
"learning_rate": 3.681093394077449e-05,
|
|
"loss": 0.2591,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3024194836616516,
|
|
"step": 405,
|
|
"valid_targets_mean": 4992.8,
|
|
"valid_targets_min": 807
|
|
},
|
|
{
|
|
"epoch": 0.6549520766773163,
|
|
"grad_norm": 0.5581797703618503,
|
|
"learning_rate": 3.7266514806378135e-05,
|
|
"loss": 0.2613,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23057574033737183,
|
|
"step": 410,
|
|
"valid_targets_mean": 4829.6,
|
|
"valid_targets_min": 995
|
|
},
|
|
{
|
|
"epoch": 0.6629392971246006,
|
|
"grad_norm": 0.5027904434256412,
|
|
"learning_rate": 3.772209567198178e-05,
|
|
"loss": 0.2555,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22512824833393097,
|
|
"step": 415,
|
|
"valid_targets_mean": 5073.9,
|
|
"valid_targets_min": 1198
|
|
},
|
|
{
|
|
"epoch": 0.670926517571885,
|
|
"grad_norm": 0.4590703964041238,
|
|
"learning_rate": 3.8177676537585425e-05,
|
|
"loss": 0.2651,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28053027391433716,
|
|
"step": 420,
|
|
"valid_targets_mean": 5108.2,
|
|
"valid_targets_min": 974
|
|
},
|
|
{
|
|
"epoch": 0.6789137380191693,
|
|
"grad_norm": 0.5634136170884916,
|
|
"learning_rate": 3.863325740318907e-05,
|
|
"loss": 0.2611,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2774794101715088,
|
|
"step": 425,
|
|
"valid_targets_mean": 5321.6,
|
|
"valid_targets_min": 1760
|
|
},
|
|
{
|
|
"epoch": 0.6869009584664537,
|
|
"grad_norm": 0.49462936802536345,
|
|
"learning_rate": 3.9088838268792714e-05,
|
|
"loss": 0.2594,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23789596557617188,
|
|
"step": 430,
|
|
"valid_targets_mean": 4688.5,
|
|
"valid_targets_min": 1566
|
|
},
|
|
{
|
|
"epoch": 0.694888178913738,
|
|
"grad_norm": 0.5767236738768104,
|
|
"learning_rate": 3.954441913439636e-05,
|
|
"loss": 0.2549,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23438569903373718,
|
|
"step": 435,
|
|
"valid_targets_mean": 5462.2,
|
|
"valid_targets_min": 988
|
|
},
|
|
{
|
|
"epoch": 0.7028753993610224,
|
|
"grad_norm": 0.5721728971305822,
|
|
"learning_rate": 4e-05,
|
|
"loss": 0.2583,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2589830160140991,
|
|
"step": 440,
|
|
"valid_targets_mean": 4021.7,
|
|
"valid_targets_min": 680
|
|
},
|
|
{
|
|
"epoch": 0.7108626198083067,
|
|
"grad_norm": 0.640470205466098,
|
|
"learning_rate": 3.999984129682125e-05,
|
|
"loss": 0.2541,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.263258695602417,
|
|
"step": 445,
|
|
"valid_targets_mean": 4029.4,
|
|
"valid_targets_min": 1150
|
|
},
|
|
{
|
|
"epoch": 0.7188498402555911,
|
|
"grad_norm": 0.56011564240242,
|
|
"learning_rate": 3.9999365189803684e-05,
|
|
"loss": 0.2561,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23699337244033813,
|
|
"step": 450,
|
|
"valid_targets_mean": 4542.3,
|
|
"valid_targets_min": 614
|
|
},
|
|
{
|
|
"epoch": 0.7268370607028753,
|
|
"grad_norm": 0.5005609148594315,
|
|
"learning_rate": 3.9998571686503264e-05,
|
|
"loss": 0.2502,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2354847490787506,
|
|
"step": 455,
|
|
"valid_targets_mean": 5020.0,
|
|
"valid_targets_min": 395
|
|
},
|
|
{
|
|
"epoch": 0.7348242811501597,
|
|
"grad_norm": 0.46714059891659654,
|
|
"learning_rate": 3.9997460799513134e-05,
|
|
"loss": 0.2479,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2476460188627243,
|
|
"step": 460,
|
|
"valid_targets_mean": 6124.8,
|
|
"valid_targets_min": 1554
|
|
},
|
|
{
|
|
"epoch": 0.7428115015974441,
|
|
"grad_norm": 0.4917282383555329,
|
|
"learning_rate": 3.999603254646343e-05,
|
|
"loss": 0.2575,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23978930711746216,
|
|
"step": 465,
|
|
"valid_targets_mean": 3983.5,
|
|
"valid_targets_min": 1464
|
|
},
|
|
{
|
|
"epoch": 0.7507987220447284,
|
|
"grad_norm": 0.5176375158379338,
|
|
"learning_rate": 3.9994286950020986e-05,
|
|
"loss": 0.272,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29831236600875854,
|
|
"step": 470,
|
|
"valid_targets_mean": 5806.6,
|
|
"valid_targets_min": 2786
|
|
},
|
|
{
|
|
"epoch": 0.7587859424920128,
|
|
"grad_norm": 0.5702966288748866,
|
|
"learning_rate": 3.999222403788896e-05,
|
|
"loss": 0.2716,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2796366214752197,
|
|
"step": 475,
|
|
"valid_targets_mean": 3448.6,
|
|
"valid_targets_min": 817
|
|
},
|
|
{
|
|
"epoch": 0.7667731629392971,
|
|
"grad_norm": 0.4757010714725201,
|
|
"learning_rate": 3.9989843842806435e-05,
|
|
"loss": 0.2547,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28887617588043213,
|
|
"step": 480,
|
|
"valid_targets_mean": 5631.5,
|
|
"valid_targets_min": 968
|
|
},
|
|
{
|
|
"epoch": 0.7747603833865815,
|
|
"grad_norm": 0.4685350745552442,
|
|
"learning_rate": 3.998714640254786e-05,
|
|
"loss": 0.266,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2430667281150818,
|
|
"step": 485,
|
|
"valid_targets_mean": 5380.9,
|
|
"valid_targets_min": 2037
|
|
},
|
|
{
|
|
"epoch": 0.7827476038338658,
|
|
"grad_norm": 0.48364161750691215,
|
|
"learning_rate": 3.998413175992247e-05,
|
|
"loss": 0.2625,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2936822175979614,
|
|
"step": 490,
|
|
"valid_targets_mean": 5195.9,
|
|
"valid_targets_min": 1019
|
|
},
|
|
{
|
|
"epoch": 0.7907348242811502,
|
|
"grad_norm": 0.5358890980558945,
|
|
"learning_rate": 3.99807999627736e-05,
|
|
"loss": 0.2714,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2451843023300171,
|
|
"step": 495,
|
|
"valid_targets_mean": 4305.3,
|
|
"valid_targets_min": 1344
|
|
},
|
|
{
|
|
"epoch": 0.7987220447284346,
|
|
"grad_norm": 0.49266507882701177,
|
|
"learning_rate": 3.997715106397794e-05,
|
|
"loss": 0.2602,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21446114778518677,
|
|
"step": 500,
|
|
"valid_targets_mean": 4101.4,
|
|
"valid_targets_min": 553
|
|
},
|
|
{
|
|
"epoch": 0.8067092651757188,
|
|
"grad_norm": 0.5283191167485831,
|
|
"learning_rate": 3.997318512144465e-05,
|
|
"loss": 0.261,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27231770753860474,
|
|
"step": 505,
|
|
"valid_targets_mean": 4392.6,
|
|
"valid_targets_min": 617
|
|
},
|
|
{
|
|
"epoch": 0.8146964856230032,
|
|
"grad_norm": 0.4637854327346454,
|
|
"learning_rate": 3.9968902198114516e-05,
|
|
"loss": 0.2513,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23250697553157806,
|
|
"step": 510,
|
|
"valid_targets_mean": 5592.4,
|
|
"valid_targets_min": 1339
|
|
},
|
|
{
|
|
"epoch": 0.8226837060702875,
|
|
"grad_norm": 0.512928763182033,
|
|
"learning_rate": 3.996430236195889e-05,
|
|
"loss": 0.2575,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2848711311817169,
|
|
"step": 515,
|
|
"valid_targets_mean": 5160.4,
|
|
"valid_targets_min": 1245
|
|
},
|
|
{
|
|
"epoch": 0.8306709265175719,
|
|
"grad_norm": 0.6688205624295005,
|
|
"learning_rate": 3.995938568597864e-05,
|
|
"loss": 0.2594,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28169065713882446,
|
|
"step": 520,
|
|
"valid_targets_mean": 4845.2,
|
|
"valid_targets_min": 1200
|
|
},
|
|
{
|
|
"epoch": 0.8386581469648562,
|
|
"grad_norm": 0.6530982098422367,
|
|
"learning_rate": 3.995415224820297e-05,
|
|
"loss": 0.252,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2486608326435089,
|
|
"step": 525,
|
|
"valid_targets_mean": 4801.1,
|
|
"valid_targets_min": 1232
|
|
},
|
|
{
|
|
"epoch": 0.8466453674121406,
|
|
"grad_norm": 0.654809754081559,
|
|
"learning_rate": 3.994860213168819e-05,
|
|
"loss": 0.2574,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23041307926177979,
|
|
"step": 530,
|
|
"valid_targets_mean": 4810.8,
|
|
"valid_targets_min": 1533
|
|
},
|
|
{
|
|
"epoch": 0.854632587859425,
|
|
"grad_norm": 0.7831567102686312,
|
|
"learning_rate": 3.9942735424516435e-05,
|
|
"loss": 0.2829,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21508295834064484,
|
|
"step": 535,
|
|
"valid_targets_mean": 3856.6,
|
|
"valid_targets_min": 697
|
|
},
|
|
{
|
|
"epoch": 0.8626198083067093,
|
|
"grad_norm": 0.4712439919508373,
|
|
"learning_rate": 3.9936552219794196e-05,
|
|
"loss": 0.2523,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2780511677265167,
|
|
"step": 540,
|
|
"valid_targets_mean": 5232.4,
|
|
"valid_targets_min": 1528
|
|
},
|
|
{
|
|
"epoch": 0.8706070287539937,
|
|
"grad_norm": 0.4492945585686436,
|
|
"learning_rate": 3.993005261565091e-05,
|
|
"loss": 0.2701,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25370901823043823,
|
|
"step": 545,
|
|
"valid_targets_mean": 5525.1,
|
|
"valid_targets_min": 785
|
|
},
|
|
{
|
|
"epoch": 0.8785942492012779,
|
|
"grad_norm": 0.5040607967651063,
|
|
"learning_rate": 3.992323671523735e-05,
|
|
"loss": 0.2537,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24419763684272766,
|
|
"step": 550,
|
|
"valid_targets_mean": 4753.2,
|
|
"valid_targets_min": 2476
|
|
},
|
|
{
|
|
"epoch": 0.8865814696485623,
|
|
"grad_norm": 0.4571433610607522,
|
|
"learning_rate": 3.991610462672403e-05,
|
|
"loss": 0.2576,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22853198647499084,
|
|
"step": 555,
|
|
"valid_targets_mean": 5481.8,
|
|
"valid_targets_min": 1745
|
|
},
|
|
{
|
|
"epoch": 0.8945686900958466,
|
|
"grad_norm": 0.4811065332797884,
|
|
"learning_rate": 3.9908656463299456e-05,
|
|
"loss": 0.2573,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2230932116508484,
|
|
"step": 560,
|
|
"valid_targets_mean": 4151.3,
|
|
"valid_targets_min": 2275
|
|
},
|
|
{
|
|
"epoch": 0.902555910543131,
|
|
"grad_norm": 0.4478004354255293,
|
|
"learning_rate": 3.990089234316835e-05,
|
|
"loss": 0.2281,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2097976952791214,
|
|
"step": 565,
|
|
"valid_targets_mean": 4595.0,
|
|
"valid_targets_min": 846
|
|
},
|
|
{
|
|
"epoch": 0.9105431309904153,
|
|
"grad_norm": 0.4939456900920226,
|
|
"learning_rate": 3.989281238954978e-05,
|
|
"loss": 0.2513,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23206022381782532,
|
|
"step": 570,
|
|
"valid_targets_mean": 4879.6,
|
|
"valid_targets_min": 441
|
|
},
|
|
{
|
|
"epoch": 0.9185303514376997,
|
|
"grad_norm": 0.5749964628221267,
|
|
"learning_rate": 3.9884416730675155e-05,
|
|
"loss": 0.253,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23972266912460327,
|
|
"step": 575,
|
|
"valid_targets_mean": 4225.3,
|
|
"valid_targets_min": 1024
|
|
},
|
|
{
|
|
"epoch": 0.9265175718849841,
|
|
"grad_norm": 0.5589879109135186,
|
|
"learning_rate": 3.987570549978626e-05,
|
|
"loss": 0.2361,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2693712115287781,
|
|
"step": 580,
|
|
"valid_targets_mean": 4750.5,
|
|
"valid_targets_min": 1402
|
|
},
|
|
{
|
|
"epoch": 0.9345047923322684,
|
|
"grad_norm": 0.495003839937131,
|
|
"learning_rate": 3.986667883513311e-05,
|
|
"loss": 0.2489,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23675037920475006,
|
|
"step": 585,
|
|
"valid_targets_mean": 4722.5,
|
|
"valid_targets_min": 1095
|
|
},
|
|
{
|
|
"epoch": 0.9424920127795527,
|
|
"grad_norm": 0.4958347061039251,
|
|
"learning_rate": 3.985733687997173e-05,
|
|
"loss": 0.2548,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27246877551078796,
|
|
"step": 590,
|
|
"valid_targets_mean": 4859.9,
|
|
"valid_targets_min": 1067
|
|
},
|
|
{
|
|
"epoch": 0.950479233226837,
|
|
"grad_norm": 0.5214337152337335,
|
|
"learning_rate": 3.984767978256192e-05,
|
|
"loss": 0.2522,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2561061680316925,
|
|
"step": 595,
|
|
"valid_targets_mean": 4810.3,
|
|
"valid_targets_min": 962
|
|
},
|
|
{
|
|
"epoch": 0.9584664536741214,
|
|
"grad_norm": 0.599666677230963,
|
|
"learning_rate": 3.983770769616488e-05,
|
|
"loss": 0.2532,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24689209461212158,
|
|
"step": 600,
|
|
"valid_targets_mean": 4667.0,
|
|
"valid_targets_min": 611
|
|
},
|
|
{
|
|
"epoch": 0.9664536741214057,
|
|
"grad_norm": 0.42495129537443144,
|
|
"learning_rate": 3.9827420779040805e-05,
|
|
"loss": 0.2632,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22384385764598846,
|
|
"step": 605,
|
|
"valid_targets_mean": 5402.8,
|
|
"valid_targets_min": 1674
|
|
},
|
|
{
|
|
"epoch": 0.9744408945686901,
|
|
"grad_norm": 0.5919379724029944,
|
|
"learning_rate": 3.981681919444633e-05,
|
|
"loss": 0.2492,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21028448641300201,
|
|
"step": 610,
|
|
"valid_targets_mean": 4113.6,
|
|
"valid_targets_min": 1029
|
|
},
|
|
{
|
|
"epoch": 0.9824281150159745,
|
|
"grad_norm": 0.5057594048145762,
|
|
"learning_rate": 3.980590311063197e-05,
|
|
"loss": 0.2482,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.263153612613678,
|
|
"step": 615,
|
|
"valid_targets_mean": 4787.4,
|
|
"valid_targets_min": 332
|
|
},
|
|
{
|
|
"epoch": 0.9904153354632588,
|
|
"grad_norm": 0.5473428636517159,
|
|
"learning_rate": 3.9794672700839455e-05,
|
|
"loss": 0.2485,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25911781191825867,
|
|
"step": 620,
|
|
"valid_targets_mean": 5511.7,
|
|
"valid_targets_min": 1610
|
|
},
|
|
{
|
|
"epoch": 0.9984025559105432,
|
|
"grad_norm": 0.6548522189800827,
|
|
"learning_rate": 3.9783128143298945e-05,
|
|
"loss": 0.2445,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23794244229793549,
|
|
"step": 625,
|
|
"valid_targets_mean": 3679.9,
|
|
"valid_targets_min": 1378
|
|
},
|
|
{
|
|
"epoch": 1.0063897763578276,
|
|
"grad_norm": 0.4637817317582565,
|
|
"learning_rate": 3.977126962122625e-05,
|
|
"loss": 0.2547,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24145828187465668,
|
|
"step": 630,
|
|
"valid_targets_mean": 4904.1,
|
|
"valid_targets_min": 1583
|
|
},
|
|
{
|
|
"epoch": 1.0143769968051117,
|
|
"grad_norm": 0.5360031330976764,
|
|
"learning_rate": 3.975909732281988e-05,
|
|
"loss": 0.2381,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2447453737258911,
|
|
"step": 635,
|
|
"valid_targets_mean": 4470.6,
|
|
"valid_targets_min": 1193
|
|
},
|
|
{
|
|
"epoch": 1.0223642172523961,
|
|
"grad_norm": 0.4599528851816257,
|
|
"learning_rate": 3.974661144125808e-05,
|
|
"loss": 0.2356,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23513412475585938,
|
|
"step": 640,
|
|
"valid_targets_mean": 5180.8,
|
|
"valid_targets_min": 1535
|
|
},
|
|
{
|
|
"epoch": 1.0303514376996805,
|
|
"grad_norm": 0.5028441117758059,
|
|
"learning_rate": 3.973381217469576e-05,
|
|
"loss": 0.2334,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23187929391860962,
|
|
"step": 645,
|
|
"valid_targets_mean": 5562.6,
|
|
"valid_targets_min": 835
|
|
},
|
|
{
|
|
"epoch": 1.038338658146965,
|
|
"grad_norm": 0.46649378993081314,
|
|
"learning_rate": 3.972069972626135e-05,
|
|
"loss": 0.2307,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2249964475631714,
|
|
"step": 650,
|
|
"valid_targets_mean": 5799.1,
|
|
"valid_targets_min": 1617
|
|
},
|
|
{
|
|
"epoch": 1.0463258785942493,
|
|
"grad_norm": 0.4890492009118198,
|
|
"learning_rate": 3.970727430405357e-05,
|
|
"loss": 0.2245,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2165513038635254,
|
|
"step": 655,
|
|
"valid_targets_mean": 4514.2,
|
|
"valid_targets_min": 777
|
|
},
|
|
{
|
|
"epoch": 1.0543130990415335,
|
|
"grad_norm": 0.5326777464844199,
|
|
"learning_rate": 3.969353612113815e-05,
|
|
"loss": 0.2363,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23171201348304749,
|
|
"step": 660,
|
|
"valid_targets_mean": 4665.6,
|
|
"valid_targets_min": 835
|
|
},
|
|
{
|
|
"epoch": 1.0623003194888179,
|
|
"grad_norm": 0.4418803882160922,
|
|
"learning_rate": 3.96794853955444e-05,
|
|
"loss": 0.2177,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19331520795822144,
|
|
"step": 665,
|
|
"valid_targets_mean": 5251.9,
|
|
"valid_targets_min": 1005
|
|
},
|
|
{
|
|
"epoch": 1.0702875399361023,
|
|
"grad_norm": 0.45451533824288753,
|
|
"learning_rate": 3.966512235026182e-05,
|
|
"loss": 0.2309,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2011176496744156,
|
|
"step": 670,
|
|
"valid_targets_mean": 5191.7,
|
|
"valid_targets_min": 723
|
|
},
|
|
{
|
|
"epoch": 1.0782747603833867,
|
|
"grad_norm": 0.5484392391440085,
|
|
"learning_rate": 3.96504472132365e-05,
|
|
"loss": 0.2251,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2403126060962677,
|
|
"step": 675,
|
|
"valid_targets_mean": 4600.9,
|
|
"valid_targets_min": 940
|
|
},
|
|
{
|
|
"epoch": 1.0862619808306708,
|
|
"grad_norm": 0.4836244631836265,
|
|
"learning_rate": 3.9635460217367513e-05,
|
|
"loss": 0.2347,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2545502781867981,
|
|
"step": 680,
|
|
"valid_targets_mean": 5117.4,
|
|
"valid_targets_min": 1430
|
|
},
|
|
{
|
|
"epoch": 1.0942492012779552,
|
|
"grad_norm": 0.5347142043189543,
|
|
"learning_rate": 3.962016160050327e-05,
|
|
"loss": 0.2411,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24589413404464722,
|
|
"step": 685,
|
|
"valid_targets_mean": 4828.7,
|
|
"valid_targets_min": 1024
|
|
},
|
|
{
|
|
"epoch": 1.1022364217252396,
|
|
"grad_norm": 0.4826007302397816,
|
|
"learning_rate": 3.960455160543767e-05,
|
|
"loss": 0.2264,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2529512345790863,
|
|
"step": 690,
|
|
"valid_targets_mean": 5452.4,
|
|
"valid_targets_min": 406
|
|
},
|
|
{
|
|
"epoch": 1.110223642172524,
|
|
"grad_norm": 0.6187571170123001,
|
|
"learning_rate": 3.958863047990631e-05,
|
|
"loss": 0.2408,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27497977018356323,
|
|
"step": 695,
|
|
"valid_targets_mean": 4824.9,
|
|
"valid_targets_min": 590
|
|
},
|
|
{
|
|
"epoch": 1.1182108626198084,
|
|
"grad_norm": 0.5218472166818443,
|
|
"learning_rate": 3.95723984765825e-05,
|
|
"loss": 0.2384,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2636116147041321,
|
|
"step": 700,
|
|
"valid_targets_mean": 5057.1,
|
|
"valid_targets_min": 625
|
|
},
|
|
{
|
|
"epoch": 1.1261980830670926,
|
|
"grad_norm": 0.5566646689747374,
|
|
"learning_rate": 3.955585585307329e-05,
|
|
"loss": 0.2246,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23090273141860962,
|
|
"step": 705,
|
|
"valid_targets_mean": 5316.9,
|
|
"valid_targets_min": 1154
|
|
},
|
|
{
|
|
"epoch": 1.134185303514377,
|
|
"grad_norm": 0.5316587953258529,
|
|
"learning_rate": 3.9539002871915395e-05,
|
|
"loss": 0.2286,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2314445823431015,
|
|
"step": 710,
|
|
"valid_targets_mean": 4499.5,
|
|
"valid_targets_min": 850
|
|
},
|
|
{
|
|
"epoch": 1.1421725239616614,
|
|
"grad_norm": 0.47010947677754694,
|
|
"learning_rate": 3.952183980057096e-05,
|
|
"loss": 0.2133,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18813036382198334,
|
|
"step": 715,
|
|
"valid_targets_mean": 4771.4,
|
|
"valid_targets_min": 2418
|
|
},
|
|
{
|
|
"epoch": 1.1501597444089458,
|
|
"grad_norm": 0.5355744283119177,
|
|
"learning_rate": 3.950436691142339e-05,
|
|
"loss": 0.2316,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23436841368675232,
|
|
"step": 720,
|
|
"valid_targets_mean": 5125.1,
|
|
"valid_targets_min": 1756
|
|
},
|
|
{
|
|
"epoch": 1.15814696485623,
|
|
"grad_norm": 0.45835728873246123,
|
|
"learning_rate": 3.948658448177299e-05,
|
|
"loss": 0.2291,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24731190502643585,
|
|
"step": 725,
|
|
"valid_targets_mean": 5959.6,
|
|
"valid_targets_min": 566
|
|
},
|
|
{
|
|
"epoch": 1.1661341853035143,
|
|
"grad_norm": 0.453683250662003,
|
|
"learning_rate": 3.946849279383258e-05,
|
|
"loss": 0.2519,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23882758617401123,
|
|
"step": 730,
|
|
"valid_targets_mean": 5065.3,
|
|
"valid_targets_min": 1942
|
|
},
|
|
{
|
|
"epoch": 1.1741214057507987,
|
|
"grad_norm": 0.5202455361914546,
|
|
"learning_rate": 3.9450092134722984e-05,
|
|
"loss": 0.2393,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24885712563991547,
|
|
"step": 735,
|
|
"valid_targets_mean": 5193.4,
|
|
"valid_targets_min": 1693
|
|
},
|
|
{
|
|
"epoch": 1.182108626198083,
|
|
"grad_norm": 0.5590629911303716,
|
|
"learning_rate": 3.943138279646853e-05,
|
|
"loss": 0.2392,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2640933394432068,
|
|
"step": 740,
|
|
"valid_targets_mean": 4765.4,
|
|
"valid_targets_min": 2236
|
|
},
|
|
{
|
|
"epoch": 1.1900958466453675,
|
|
"grad_norm": 0.4568215659004477,
|
|
"learning_rate": 3.941236507599234e-05,
|
|
"loss": 0.2538,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23051276803016663,
|
|
"step": 745,
|
|
"valid_targets_mean": 5046.9,
|
|
"valid_targets_min": 1931
|
|
},
|
|
{
|
|
"epoch": 1.1980830670926517,
|
|
"grad_norm": 0.44712112441589846,
|
|
"learning_rate": 3.93930392751117e-05,
|
|
"loss": 0.2225,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21604947745800018,
|
|
"step": 750,
|
|
"valid_targets_mean": 5026.4,
|
|
"valid_targets_min": 1153
|
|
},
|
|
{
|
|
"epoch": 1.206070287539936,
|
|
"grad_norm": 0.5591497129271406,
|
|
"learning_rate": 3.9373405700533204e-05,
|
|
"loss": 0.221,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21242642402648926,
|
|
"step": 755,
|
|
"valid_targets_mean": 4334.6,
|
|
"valid_targets_min": 1623
|
|
},
|
|
{
|
|
"epoch": 1.2140575079872205,
|
|
"grad_norm": 0.5943967241943648,
|
|
"learning_rate": 3.935346466384793e-05,
|
|
"loss": 0.2416,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29953110218048096,
|
|
"step": 760,
|
|
"valid_targets_mean": 5227.2,
|
|
"valid_targets_min": 920
|
|
},
|
|
{
|
|
"epoch": 1.2220447284345048,
|
|
"grad_norm": 0.4042075117948494,
|
|
"learning_rate": 3.933321648152646e-05,
|
|
"loss": 0.246,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22984498739242554,
|
|
"step": 765,
|
|
"valid_targets_mean": 6071.8,
|
|
"valid_targets_min": 2067
|
|
},
|
|
{
|
|
"epoch": 1.230031948881789,
|
|
"grad_norm": 0.48485906424603825,
|
|
"learning_rate": 3.931266147491389e-05,
|
|
"loss": 0.2363,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23701247572898865,
|
|
"step": 770,
|
|
"valid_targets_mean": 4539.6,
|
|
"valid_targets_min": 912
|
|
},
|
|
{
|
|
"epoch": 1.2380191693290734,
|
|
"grad_norm": 0.484609973987031,
|
|
"learning_rate": 3.929179997022471e-05,
|
|
"loss": 0.2355,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22221249341964722,
|
|
"step": 775,
|
|
"valid_targets_mean": 4264.4,
|
|
"valid_targets_min": 2231
|
|
},
|
|
{
|
|
"epoch": 1.2460063897763578,
|
|
"grad_norm": 0.4634913749349423,
|
|
"learning_rate": 3.927063229853763e-05,
|
|
"loss": 0.2505,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2315971553325653,
|
|
"step": 780,
|
|
"valid_targets_mean": 4269.0,
|
|
"valid_targets_min": 1298
|
|
},
|
|
{
|
|
"epoch": 1.2539936102236422,
|
|
"grad_norm": 0.46003637169332934,
|
|
"learning_rate": 3.9249158795790316e-05,
|
|
"loss": 0.233,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21694815158843994,
|
|
"step": 785,
|
|
"valid_targets_mean": 4534.9,
|
|
"valid_targets_min": 1367
|
|
},
|
|
{
|
|
"epoch": 1.2619808306709266,
|
|
"grad_norm": 0.5159552397005117,
|
|
"learning_rate": 3.9227379802774106e-05,
|
|
"loss": 0.2468,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22513447701931,
|
|
"step": 790,
|
|
"valid_targets_mean": 4702.6,
|
|
"valid_targets_min": 933
|
|
},
|
|
{
|
|
"epoch": 1.269968051118211,
|
|
"grad_norm": 1.1512499859539236,
|
|
"learning_rate": 3.920529566512852e-05,
|
|
"loss": 0.2435,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23088482022285461,
|
|
"step": 795,
|
|
"valid_targets_mean": 5759.7,
|
|
"valid_targets_min": 2373
|
|
},
|
|
{
|
|
"epoch": 1.2779552715654952,
|
|
"grad_norm": 0.5434796049479874,
|
|
"learning_rate": 3.918290673333585e-05,
|
|
"loss": 0.2382,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2833731472492218,
|
|
"step": 800,
|
|
"valid_targets_mean": 4076.2,
|
|
"valid_targets_min": 568
|
|
},
|
|
{
|
|
"epoch": 1.2859424920127795,
|
|
"grad_norm": 0.5180856784834859,
|
|
"learning_rate": 3.916021336271556e-05,
|
|
"loss": 0.2491,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23785400390625,
|
|
"step": 805,
|
|
"valid_targets_mean": 4955.7,
|
|
"valid_targets_min": 924
|
|
},
|
|
{
|
|
"epoch": 1.293929712460064,
|
|
"grad_norm": 0.5149958361755231,
|
|
"learning_rate": 3.913721591341867e-05,
|
|
"loss": 0.2219,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25627684593200684,
|
|
"step": 810,
|
|
"valid_targets_mean": 5241.1,
|
|
"valid_targets_min": 1588
|
|
},
|
|
{
|
|
"epoch": 1.3019169329073481,
|
|
"grad_norm": 0.4502540860682754,
|
|
"learning_rate": 3.9113914750421985e-05,
|
|
"loss": 0.219,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25584468245506287,
|
|
"step": 815,
|
|
"valid_targets_mean": 6717.8,
|
|
"valid_targets_min": 1683
|
|
},
|
|
{
|
|
"epoch": 1.3099041533546325,
|
|
"grad_norm": 0.5006857836953049,
|
|
"learning_rate": 3.9090310243522394e-05,
|
|
"loss": 0.2337,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2447993904352188,
|
|
"step": 820,
|
|
"valid_targets_mean": 4522.4,
|
|
"valid_targets_min": 557
|
|
},
|
|
{
|
|
"epoch": 1.317891373801917,
|
|
"grad_norm": 0.530082191187644,
|
|
"learning_rate": 3.90664027673309e-05,
|
|
"loss": 0.2421,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28234124183654785,
|
|
"step": 825,
|
|
"valid_targets_mean": 5410.7,
|
|
"valid_targets_min": 2137
|
|
},
|
|
{
|
|
"epoch": 1.3258785942492013,
|
|
"grad_norm": 0.4414104140990789,
|
|
"learning_rate": 3.904219270126677e-05,
|
|
"loss": 0.2199,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20832520723342896,
|
|
"step": 830,
|
|
"valid_targets_mean": 4990.1,
|
|
"valid_targets_min": 1554
|
|
},
|
|
{
|
|
"epoch": 1.3338658146964857,
|
|
"grad_norm": 1.2054942942284992,
|
|
"learning_rate": 3.901768042955144e-05,
|
|
"loss": 0.2331,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2427096664905548,
|
|
"step": 835,
|
|
"valid_targets_mean": 5274.6,
|
|
"valid_targets_min": 2181
|
|
},
|
|
{
|
|
"epoch": 1.34185303514377,
|
|
"grad_norm": 0.44040986559266476,
|
|
"learning_rate": 3.8992866341202446e-05,
|
|
"loss": 0.2206,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23897823691368103,
|
|
"step": 840,
|
|
"valid_targets_mean": 6298.9,
|
|
"valid_targets_min": 320
|
|
},
|
|
{
|
|
"epoch": 1.3498402555910542,
|
|
"grad_norm": 0.42791841484545096,
|
|
"learning_rate": 3.8967750830027277e-05,
|
|
"loss": 0.2271,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21991094946861267,
|
|
"step": 845,
|
|
"valid_targets_mean": 4994.8,
|
|
"valid_targets_min": 1678
|
|
},
|
|
{
|
|
"epoch": 1.3578274760383386,
|
|
"grad_norm": 0.4765755408468607,
|
|
"learning_rate": 3.894233429461706e-05,
|
|
"loss": 0.2293,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24188624322414398,
|
|
"step": 850,
|
|
"valid_targets_mean": 4151.7,
|
|
"valid_targets_min": 640
|
|
},
|
|
{
|
|
"epoch": 1.365814696485623,
|
|
"grad_norm": 0.4702467492894455,
|
|
"learning_rate": 3.89166171383403e-05,
|
|
"loss": 0.2607,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2250172346830368,
|
|
"step": 855,
|
|
"valid_targets_mean": 5215.4,
|
|
"valid_targets_min": 1781
|
|
},
|
|
{
|
|
"epoch": 1.3738019169329074,
|
|
"grad_norm": 0.5280309031579677,
|
|
"learning_rate": 3.889059976933644e-05,
|
|
"loss": 0.2176,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22099585831165314,
|
|
"step": 860,
|
|
"valid_targets_mean": 5249.2,
|
|
"valid_targets_min": 1828
|
|
},
|
|
{
|
|
"epoch": 1.3817891373801916,
|
|
"grad_norm": 0.5191741974294681,
|
|
"learning_rate": 3.88642826005094e-05,
|
|
"loss": 0.2209,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23172377049922943,
|
|
"step": 865,
|
|
"valid_targets_mean": 4564.9,
|
|
"valid_targets_min": 299
|
|
},
|
|
{
|
|
"epoch": 1.389776357827476,
|
|
"grad_norm": 0.4931770156824684,
|
|
"learning_rate": 3.883766604952102e-05,
|
|
"loss": 0.2405,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22133614122867584,
|
|
"step": 870,
|
|
"valid_targets_mean": 4580.8,
|
|
"valid_targets_min": 1788
|
|
},
|
|
{
|
|
"epoch": 1.3977635782747604,
|
|
"grad_norm": 0.43973258214989486,
|
|
"learning_rate": 3.8810750538784404e-05,
|
|
"loss": 0.2397,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2410700023174286,
|
|
"step": 875,
|
|
"valid_targets_mean": 5454.4,
|
|
"valid_targets_min": 268
|
|
},
|
|
{
|
|
"epoch": 1.4057507987220448,
|
|
"grad_norm": 0.46352392220990385,
|
|
"learning_rate": 3.878353649545728e-05,
|
|
"loss": 0.2305,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2691129446029663,
|
|
"step": 880,
|
|
"valid_targets_mean": 4870.0,
|
|
"valid_targets_min": 1357
|
|
},
|
|
{
|
|
"epoch": 1.4137380191693292,
|
|
"grad_norm": 0.45638607377842,
|
|
"learning_rate": 3.875602435143517e-05,
|
|
"loss": 0.2413,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2274504452943802,
|
|
"step": 885,
|
|
"valid_targets_mean": 5388.2,
|
|
"valid_targets_min": 1962
|
|
},
|
|
{
|
|
"epoch": 1.4217252396166133,
|
|
"grad_norm": 0.4790466979102037,
|
|
"learning_rate": 3.872821454334453e-05,
|
|
"loss": 0.2224,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2431117296218872,
|
|
"step": 890,
|
|
"valid_targets_mean": 5412.2,
|
|
"valid_targets_min": 1351
|
|
},
|
|
{
|
|
"epoch": 1.4297124600638977,
|
|
"grad_norm": 0.4117592654298577,
|
|
"learning_rate": 3.870010751253587e-05,
|
|
"loss": 0.2244,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2094022035598755,
|
|
"step": 895,
|
|
"valid_targets_mean": 5338.6,
|
|
"valid_targets_min": 337
|
|
},
|
|
{
|
|
"epoch": 1.4376996805111821,
|
|
"grad_norm": 0.5463838416850257,
|
|
"learning_rate": 3.867170370507668e-05,
|
|
"loss": 0.2105,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21328377723693848,
|
|
"step": 900,
|
|
"valid_targets_mean": 3389.1,
|
|
"valid_targets_min": 846
|
|
},
|
|
{
|
|
"epoch": 1.4456869009584665,
|
|
"grad_norm": 0.4394986594361576,
|
|
"learning_rate": 3.8643003571744445e-05,
|
|
"loss": 0.2202,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21873891353607178,
|
|
"step": 905,
|
|
"valid_targets_mean": 5016.2,
|
|
"valid_targets_min": 1875
|
|
},
|
|
{
|
|
"epoch": 1.4536741214057507,
|
|
"grad_norm": 0.4092707030943055,
|
|
"learning_rate": 3.861400756801938e-05,
|
|
"loss": 0.2216,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21018297970294952,
|
|
"step": 910,
|
|
"valid_targets_mean": 5767.4,
|
|
"valid_targets_min": 2816
|
|
},
|
|
{
|
|
"epoch": 1.461661341853035,
|
|
"grad_norm": 0.465636175172399,
|
|
"learning_rate": 3.8584716154077286e-05,
|
|
"loss": 0.2195,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22354570031166077,
|
|
"step": 915,
|
|
"valid_targets_mean": 6138.7,
|
|
"valid_targets_min": 2160
|
|
},
|
|
{
|
|
"epoch": 1.4696485623003195,
|
|
"grad_norm": 0.434468625549387,
|
|
"learning_rate": 3.855512979478222e-05,
|
|
"loss": 0.2457,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22415438294410706,
|
|
"step": 920,
|
|
"valid_targets_mean": 6146.2,
|
|
"valid_targets_min": 609
|
|
},
|
|
{
|
|
"epoch": 1.4776357827476039,
|
|
"grad_norm": 0.44757829016572626,
|
|
"learning_rate": 3.852524895967911e-05,
|
|
"loss": 0.2387,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23147371411323547,
|
|
"step": 925,
|
|
"valid_targets_mean": 5348.4,
|
|
"valid_targets_min": 2054
|
|
},
|
|
{
|
|
"epoch": 1.4856230031948883,
|
|
"grad_norm": 0.5097635877620077,
|
|
"learning_rate": 3.8495074122986296e-05,
|
|
"loss": 0.2309,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27335506677627563,
|
|
"step": 930,
|
|
"valid_targets_mean": 5118.7,
|
|
"valid_targets_min": 1367
|
|
},
|
|
{
|
|
"epoch": 1.4936102236421724,
|
|
"grad_norm": 0.4776053358725652,
|
|
"learning_rate": 3.846460576358804e-05,
|
|
"loss": 0.224,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22878849506378174,
|
|
"step": 935,
|
|
"valid_targets_mean": 4780.4,
|
|
"valid_targets_min": 1610
|
|
},
|
|
{
|
|
"epoch": 1.5015974440894568,
|
|
"grad_norm": 0.5356334827731245,
|
|
"learning_rate": 3.843384436502688e-05,
|
|
"loss": 0.2424,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22441717982292175,
|
|
"step": 940,
|
|
"valid_targets_mean": 4326.4,
|
|
"valid_targets_min": 476
|
|
},
|
|
{
|
|
"epoch": 1.5095846645367412,
|
|
"grad_norm": 0.4512491618502969,
|
|
"learning_rate": 3.8402790415496e-05,
|
|
"loss": 0.2285,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21335360407829285,
|
|
"step": 945,
|
|
"valid_targets_mean": 5230.7,
|
|
"valid_targets_min": 617
|
|
},
|
|
{
|
|
"epoch": 1.5175718849840254,
|
|
"grad_norm": 0.45906002413677505,
|
|
"learning_rate": 3.837144440783144e-05,
|
|
"loss": 0.2332,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24478338658809662,
|
|
"step": 950,
|
|
"valid_targets_mean": 5632.8,
|
|
"valid_targets_min": 1674
|
|
},
|
|
{
|
|
"epoch": 1.5255591054313098,
|
|
"grad_norm": 0.49019467193719773,
|
|
"learning_rate": 3.833980683950431e-05,
|
|
"loss": 0.2411,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2519046664237976,
|
|
"step": 955,
|
|
"valid_targets_mean": 5137.8,
|
|
"valid_targets_min": 932
|
|
},
|
|
{
|
|
"epoch": 1.5335463258785942,
|
|
"grad_norm": 0.42945027027443916,
|
|
"learning_rate": 3.8307878212612886e-05,
|
|
"loss": 0.2311,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1915363073348999,
|
|
"step": 960,
|
|
"valid_targets_mean": 4416.3,
|
|
"valid_targets_min": 761
|
|
},
|
|
{
|
|
"epoch": 1.5415335463258786,
|
|
"grad_norm": 0.5579055331895597,
|
|
"learning_rate": 3.827565903387461e-05,
|
|
"loss": 0.2176,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21044248342514038,
|
|
"step": 965,
|
|
"valid_targets_mean": 5284.6,
|
|
"valid_targets_min": 2506
|
|
},
|
|
{
|
|
"epoch": 1.549520766773163,
|
|
"grad_norm": 0.5356246975818258,
|
|
"learning_rate": 3.82431498146181e-05,
|
|
"loss": 0.2525,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2801210284233093,
|
|
"step": 970,
|
|
"valid_targets_mean": 5111.0,
|
|
"valid_targets_min": 648
|
|
},
|
|
{
|
|
"epoch": 1.5575079872204474,
|
|
"grad_norm": 0.4676912946078301,
|
|
"learning_rate": 3.821035107077499e-05,
|
|
"loss": 0.2415,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2880561053752899,
|
|
"step": 975,
|
|
"valid_targets_mean": 5908.9,
|
|
"valid_targets_min": 729
|
|
},
|
|
{
|
|
"epoch": 1.5654952076677318,
|
|
"grad_norm": 0.418525383451211,
|
|
"learning_rate": 3.817726332287179e-05,
|
|
"loss": 0.2233,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21174396574497223,
|
|
"step": 980,
|
|
"valid_targets_mean": 5006.8,
|
|
"valid_targets_min": 817
|
|
},
|
|
{
|
|
"epoch": 1.573482428115016,
|
|
"grad_norm": 0.5929561302080186,
|
|
"learning_rate": 3.814388709602155e-05,
|
|
"loss": 0.2348,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2387070208787918,
|
|
"step": 985,
|
|
"valid_targets_mean": 4222.9,
|
|
"valid_targets_min": 1167
|
|
},
|
|
{
|
|
"epoch": 1.5814696485623003,
|
|
"grad_norm": 0.4595638312331319,
|
|
"learning_rate": 3.8110222919915626e-05,
|
|
"loss": 0.2233,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23978567123413086,
|
|
"step": 990,
|
|
"valid_targets_mean": 5538.3,
|
|
"valid_targets_min": 1108
|
|
},
|
|
{
|
|
"epoch": 1.5894568690095847,
|
|
"grad_norm": 0.5512689817880697,
|
|
"learning_rate": 3.8076271328815175e-05,
|
|
"loss": 0.2346,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23280058801174164,
|
|
"step": 995,
|
|
"valid_targets_mean": 4920.5,
|
|
"valid_targets_min": 724
|
|
},
|
|
{
|
|
"epoch": 1.5974440894568689,
|
|
"grad_norm": 0.42805027733164774,
|
|
"learning_rate": 3.804203286154275e-05,
|
|
"loss": 0.2214,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21907193958759308,
|
|
"step": 1000,
|
|
"valid_targets_mean": 5069.7,
|
|
"valid_targets_min": 1920
|
|
},
|
|
{
|
|
"epoch": 1.6054313099041533,
|
|
"grad_norm": 0.4858809130152022,
|
|
"learning_rate": 3.800750806147371e-05,
|
|
"loss": 0.2374,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22444787621498108,
|
|
"step": 1005,
|
|
"valid_targets_mean": 4515.1,
|
|
"valid_targets_min": 714
|
|
},
|
|
{
|
|
"epoch": 1.6134185303514377,
|
|
"grad_norm": 0.5302128778041387,
|
|
"learning_rate": 3.79726974765276e-05,
|
|
"loss": 0.2356,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25729191303253174,
|
|
"step": 1010,
|
|
"valid_targets_mean": 4565.3,
|
|
"valid_targets_min": 1548
|
|
},
|
|
{
|
|
"epoch": 1.621405750798722,
|
|
"grad_norm": 0.4829977405730259,
|
|
"learning_rate": 3.793760165915947e-05,
|
|
"loss": 0.2341,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23642590641975403,
|
|
"step": 1015,
|
|
"valid_targets_mean": 4304.2,
|
|
"valid_targets_min": 478
|
|
},
|
|
{
|
|
"epoch": 1.6293929712460065,
|
|
"grad_norm": 0.4594198791814113,
|
|
"learning_rate": 3.7902221166351106e-05,
|
|
"loss": 0.2234,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20123255252838135,
|
|
"step": 1020,
|
|
"valid_targets_mean": 5075.9,
|
|
"valid_targets_min": 1258
|
|
},
|
|
{
|
|
"epoch": 1.6373801916932909,
|
|
"grad_norm": 0.4426802622554693,
|
|
"learning_rate": 3.786655655960216e-05,
|
|
"loss": 0.2255,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23082798719406128,
|
|
"step": 1025,
|
|
"valid_targets_mean": 5212.6,
|
|
"valid_targets_min": 1279
|
|
},
|
|
{
|
|
"epoch": 1.645367412140575,
|
|
"grad_norm": 0.4870481188557817,
|
|
"learning_rate": 3.7830608404921294e-05,
|
|
"loss": 0.2074,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21210849285125732,
|
|
"step": 1030,
|
|
"valid_targets_mean": 4509.0,
|
|
"valid_targets_min": 1346
|
|
},
|
|
{
|
|
"epoch": 1.6533546325878594,
|
|
"grad_norm": 0.4698781770865886,
|
|
"learning_rate": 3.7794377272817144e-05,
|
|
"loss": 0.2173,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2248748391866684,
|
|
"step": 1035,
|
|
"valid_targets_mean": 5013.8,
|
|
"valid_targets_min": 829
|
|
},
|
|
{
|
|
"epoch": 1.6613418530351438,
|
|
"grad_norm": 0.4361307082713528,
|
|
"learning_rate": 3.775786373828929e-05,
|
|
"loss": 0.2286,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.222589910030365,
|
|
"step": 1040,
|
|
"valid_targets_mean": 5144.6,
|
|
"valid_targets_min": 770
|
|
},
|
|
{
|
|
"epoch": 1.669329073482428,
|
|
"grad_norm": 0.49522338344118094,
|
|
"learning_rate": 3.7721068380819135e-05,
|
|
"loss": 0.2193,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2181740403175354,
|
|
"step": 1045,
|
|
"valid_targets_mean": 4062.8,
|
|
"valid_targets_min": 860
|
|
},
|
|
{
|
|
"epoch": 1.6773162939297124,
|
|
"grad_norm": 0.49702593993884003,
|
|
"learning_rate": 3.768399178436069e-05,
|
|
"loss": 0.2349,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23673710227012634,
|
|
"step": 1050,
|
|
"valid_targets_mean": 4153.0,
|
|
"valid_targets_min": 1755
|
|
},
|
|
{
|
|
"epoch": 1.6853035143769968,
|
|
"grad_norm": 0.455513257026603,
|
|
"learning_rate": 3.764663453733135e-05,
|
|
"loss": 0.2378,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2705381512641907,
|
|
"step": 1055,
|
|
"valid_targets_mean": 5011.4,
|
|
"valid_targets_min": 1194
|
|
},
|
|
{
|
|
"epoch": 1.6932907348242812,
|
|
"grad_norm": 0.48342636718296034,
|
|
"learning_rate": 3.7608997232602475e-05,
|
|
"loss": 0.2342,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21209147572517395,
|
|
"step": 1060,
|
|
"valid_targets_mean": 4200.3,
|
|
"valid_targets_min": 1032
|
|
},
|
|
{
|
|
"epoch": 1.7012779552715656,
|
|
"grad_norm": 0.4680188702163582,
|
|
"learning_rate": 3.757108046749006e-05,
|
|
"loss": 0.2299,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26718616485595703,
|
|
"step": 1065,
|
|
"valid_targets_mean": 4864.6,
|
|
"valid_targets_min": 710
|
|
},
|
|
{
|
|
"epoch": 1.70926517571885,
|
|
"grad_norm": 0.46173075527590035,
|
|
"learning_rate": 3.753288484374524e-05,
|
|
"loss": 0.2297,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2084069550037384,
|
|
"step": 1070,
|
|
"valid_targets_mean": 4525.0,
|
|
"valid_targets_min": 1873
|
|
},
|
|
{
|
|
"epoch": 1.7172523961661343,
|
|
"grad_norm": 0.45688747124270174,
|
|
"learning_rate": 3.7494410967544674e-05,
|
|
"loss": 0.2318,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22160950303077698,
|
|
"step": 1075,
|
|
"valid_targets_mean": 5390.1,
|
|
"valid_targets_min": 3406
|
|
},
|
|
{
|
|
"epoch": 1.7252396166134185,
|
|
"grad_norm": 0.5994722237990151,
|
|
"learning_rate": 3.745565944948103e-05,
|
|
"loss": 0.2156,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21359854936599731,
|
|
"step": 1080,
|
|
"valid_targets_mean": 4494.8,
|
|
"valid_targets_min": 1172
|
|
},
|
|
{
|
|
"epoch": 1.733226837060703,
|
|
"grad_norm": 0.42158057178019015,
|
|
"learning_rate": 3.7416630904553205e-05,
|
|
"loss": 0.2287,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2346889078617096,
|
|
"step": 1085,
|
|
"valid_targets_mean": 5971.8,
|
|
"valid_targets_min": 2006
|
|
},
|
|
{
|
|
"epoch": 1.741214057507987,
|
|
"grad_norm": 0.41370122622856514,
|
|
"learning_rate": 3.737732595215663e-05,
|
|
"loss": 0.2244,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2176918089389801,
|
|
"step": 1090,
|
|
"valid_targets_mean": 5168.2,
|
|
"valid_targets_min": 1622
|
|
},
|
|
{
|
|
"epoch": 1.7492012779552715,
|
|
"grad_norm": 0.42953107682395736,
|
|
"learning_rate": 3.733774521607338e-05,
|
|
"loss": 0.2282,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23724493384361267,
|
|
"step": 1095,
|
|
"valid_targets_mean": 5557.6,
|
|
"valid_targets_min": 1198
|
|
},
|
|
{
|
|
"epoch": 1.7571884984025559,
|
|
"grad_norm": 0.5657788201117734,
|
|
"learning_rate": 3.729788932446231e-05,
|
|
"loss": 0.2239,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25702691078186035,
|
|
"step": 1100,
|
|
"valid_targets_mean": 4413.9,
|
|
"valid_targets_min": 794
|
|
},
|
|
{
|
|
"epoch": 1.7651757188498403,
|
|
"grad_norm": 0.417477496167328,
|
|
"learning_rate": 3.7257758909849107e-05,
|
|
"loss": 0.2238,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1908511519432068,
|
|
"step": 1105,
|
|
"valid_targets_mean": 4390.8,
|
|
"valid_targets_min": 765
|
|
},
|
|
{
|
|
"epoch": 1.7731629392971247,
|
|
"grad_norm": 0.4384968778126525,
|
|
"learning_rate": 3.72173546091162e-05,
|
|
"loss": 0.2114,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20780757069587708,
|
|
"step": 1110,
|
|
"valid_targets_mean": 5119.9,
|
|
"valid_targets_min": 1113
|
|
},
|
|
{
|
|
"epoch": 1.781150159744409,
|
|
"grad_norm": 0.469616288326366,
|
|
"learning_rate": 3.7176677063492685e-05,
|
|
"loss": 0.242,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2339574694633484,
|
|
"step": 1115,
|
|
"valid_targets_mean": 4781.5,
|
|
"valid_targets_min": 1392
|
|
},
|
|
{
|
|
"epoch": 1.7891373801916934,
|
|
"grad_norm": 0.40034591514089823,
|
|
"learning_rate": 3.713572691854414e-05,
|
|
"loss": 0.2339,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23980014026165009,
|
|
"step": 1120,
|
|
"valid_targets_mean": 6481.1,
|
|
"valid_targets_min": 1932
|
|
},
|
|
{
|
|
"epoch": 1.7971246006389776,
|
|
"grad_norm": 0.4320316231238413,
|
|
"learning_rate": 3.709450482416239e-05,
|
|
"loss": 0.2419,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25469911098480225,
|
|
"step": 1125,
|
|
"valid_targets_mean": 6072.1,
|
|
"valid_targets_min": 2053
|
|
},
|
|
{
|
|
"epoch": 1.805111821086262,
|
|
"grad_norm": 0.4330826211238907,
|
|
"learning_rate": 3.7053011434555165e-05,
|
|
"loss": 0.2195,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2246796190738678,
|
|
"step": 1130,
|
|
"valid_targets_mean": 4668.4,
|
|
"valid_targets_min": 2346
|
|
},
|
|
{
|
|
"epoch": 1.8130990415335462,
|
|
"grad_norm": 0.4364883586042583,
|
|
"learning_rate": 3.701124740823575e-05,
|
|
"loss": 0.2313,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22576244175434113,
|
|
"step": 1135,
|
|
"valid_targets_mean": 4683.9,
|
|
"valid_targets_min": 936
|
|
},
|
|
{
|
|
"epoch": 1.8210862619808306,
|
|
"grad_norm": 0.48768121572116485,
|
|
"learning_rate": 3.696921340801253e-05,
|
|
"loss": 0.23,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23971682786941528,
|
|
"step": 1140,
|
|
"valid_targets_mean": 4992.1,
|
|
"valid_targets_min": 1319
|
|
},
|
|
{
|
|
"epoch": 1.829073482428115,
|
|
"grad_norm": 0.46689741570673715,
|
|
"learning_rate": 3.6926910100978444e-05,
|
|
"loss": 0.2209,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20149345695972443,
|
|
"step": 1145,
|
|
"valid_targets_mean": 4508.4,
|
|
"valid_targets_min": 1663
|
|
},
|
|
{
|
|
"epoch": 1.8370607028753994,
|
|
"grad_norm": 0.4650335934315107,
|
|
"learning_rate": 3.688433815850041e-05,
|
|
"loss": 0.2246,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23571673035621643,
|
|
"step": 1150,
|
|
"valid_targets_mean": 4445.2,
|
|
"valid_targets_min": 942
|
|
},
|
|
{
|
|
"epoch": 1.8450479233226837,
|
|
"grad_norm": 0.49587787282658763,
|
|
"learning_rate": 3.68414982562087e-05,
|
|
"loss": 0.2345,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25807803869247437,
|
|
"step": 1155,
|
|
"valid_targets_mean": 4298.6,
|
|
"valid_targets_min": 1095
|
|
},
|
|
{
|
|
"epoch": 1.8530351437699681,
|
|
"grad_norm": 0.4711400556625992,
|
|
"learning_rate": 3.679839107398618e-05,
|
|
"loss": 0.2073,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22391392290592194,
|
|
"step": 1160,
|
|
"valid_targets_mean": 4605.8,
|
|
"valid_targets_min": 1079
|
|
},
|
|
{
|
|
"epoch": 1.8610223642172525,
|
|
"grad_norm": 0.4952899600117989,
|
|
"learning_rate": 3.6755017295957536e-05,
|
|
"loss": 0.2273,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2246803343296051,
|
|
"step": 1165,
|
|
"valid_targets_mean": 4714.9,
|
|
"valid_targets_min": 1269
|
|
},
|
|
{
|
|
"epoch": 1.8690095846645367,
|
|
"grad_norm": 0.4505280090095968,
|
|
"learning_rate": 3.67113776104784e-05,
|
|
"loss": 0.2324,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25178074836730957,
|
|
"step": 1170,
|
|
"valid_targets_mean": 5578.4,
|
|
"valid_targets_min": 1360
|
|
},
|
|
{
|
|
"epoch": 1.876996805111821,
|
|
"grad_norm": 0.45568302444103076,
|
|
"learning_rate": 3.6667472710124475e-05,
|
|
"loss": 0.2292,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22000402212142944,
|
|
"step": 1175,
|
|
"valid_targets_mean": 6081.1,
|
|
"valid_targets_min": 1049
|
|
},
|
|
{
|
|
"epoch": 1.8849840255591053,
|
|
"grad_norm": 0.521807718623954,
|
|
"learning_rate": 3.6623303291680465e-05,
|
|
"loss": 0.2214,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2112407088279724,
|
|
"step": 1180,
|
|
"valid_targets_mean": 3739.1,
|
|
"valid_targets_min": 483
|
|
},
|
|
{
|
|
"epoch": 1.8929712460063897,
|
|
"grad_norm": 0.4625676303413823,
|
|
"learning_rate": 3.6578870056129086e-05,
|
|
"loss": 0.2522,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22916141152381897,
|
|
"step": 1185,
|
|
"valid_targets_mean": 4772.2,
|
|
"valid_targets_min": 2216
|
|
},
|
|
{
|
|
"epoch": 1.900958466453674,
|
|
"grad_norm": 0.4563568093746638,
|
|
"learning_rate": 3.653417370863992e-05,
|
|
"loss": 0.2389,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23207005858421326,
|
|
"step": 1190,
|
|
"valid_targets_mean": 4272.0,
|
|
"valid_targets_min": 1679
|
|
},
|
|
{
|
|
"epoch": 1.9089456869009584,
|
|
"grad_norm": 0.4874312100681554,
|
|
"learning_rate": 3.6489214958558206e-05,
|
|
"loss": 0.2092,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21660256385803223,
|
|
"step": 1195,
|
|
"valid_targets_mean": 4120.2,
|
|
"valid_targets_min": 1090
|
|
},
|
|
{
|
|
"epoch": 1.9169329073482428,
|
|
"grad_norm": 0.42785031525017525,
|
|
"learning_rate": 3.644399451939358e-05,
|
|
"loss": 0.2256,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21264520287513733,
|
|
"step": 1200,
|
|
"valid_targets_mean": 4690.1,
|
|
"valid_targets_min": 1768
|
|
},
|
|
{
|
|
"epoch": 1.9249201277955272,
|
|
"grad_norm": 0.43725812400863223,
|
|
"learning_rate": 3.639851310880881e-05,
|
|
"loss": 0.2355,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24090486764907837,
|
|
"step": 1205,
|
|
"valid_targets_mean": 4846.9,
|
|
"valid_targets_min": 2437
|
|
},
|
|
{
|
|
"epoch": 1.9329073482428116,
|
|
"grad_norm": 0.4716273434408524,
|
|
"learning_rate": 3.635277144860834e-05,
|
|
"loss": 0.2362,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24312490224838257,
|
|
"step": 1210,
|
|
"valid_targets_mean": 6083.0,
|
|
"valid_targets_min": 2349
|
|
},
|
|
{
|
|
"epoch": 1.9408945686900958,
|
|
"grad_norm": 0.40108176875471574,
|
|
"learning_rate": 3.630677026472684e-05,
|
|
"loss": 0.2285,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21307267248630524,
|
|
"step": 1215,
|
|
"valid_targets_mean": 5714.7,
|
|
"valid_targets_min": 2174
|
|
},
|
|
{
|
|
"epoch": 1.9488817891373802,
|
|
"grad_norm": 0.49671707842355,
|
|
"learning_rate": 3.626051028721773e-05,
|
|
"loss": 0.2248,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2293505221605301,
|
|
"step": 1220,
|
|
"valid_targets_mean": 3788.6,
|
|
"valid_targets_min": 543
|
|
},
|
|
{
|
|
"epoch": 1.9568690095846646,
|
|
"grad_norm": 0.44107663935637476,
|
|
"learning_rate": 3.621399225024156e-05,
|
|
"loss": 0.2232,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20223724842071533,
|
|
"step": 1225,
|
|
"valid_targets_mean": 4862.7,
|
|
"valid_targets_min": 943
|
|
},
|
|
{
|
|
"epoch": 1.9648562300319488,
|
|
"grad_norm": 0.576803228028713,
|
|
"learning_rate": 3.616721689205436e-05,
|
|
"loss": 0.22,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21243348717689514,
|
|
"step": 1230,
|
|
"valid_targets_mean": 4010.9,
|
|
"valid_targets_min": 850
|
|
},
|
|
{
|
|
"epoch": 1.9728434504792332,
|
|
"grad_norm": 0.4387186194383008,
|
|
"learning_rate": 3.612018495499594e-05,
|
|
"loss": 0.2303,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21914973855018616,
|
|
"step": 1235,
|
|
"valid_targets_mean": 5108.9,
|
|
"valid_targets_min": 2566
|
|
},
|
|
{
|
|
"epoch": 1.9808306709265175,
|
|
"grad_norm": 0.45509359256484455,
|
|
"learning_rate": 3.6072897185478074e-05,
|
|
"loss": 0.2057,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19633805751800537,
|
|
"step": 1240,
|
|
"valid_targets_mean": 5676.6,
|
|
"valid_targets_min": 2015
|
|
},
|
|
{
|
|
"epoch": 1.988817891373802,
|
|
"grad_norm": 0.4285410701862073,
|
|
"learning_rate": 3.6025354333972714e-05,
|
|
"loss": 0.24,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23625974357128143,
|
|
"step": 1245,
|
|
"valid_targets_mean": 5396.7,
|
|
"valid_targets_min": 2667
|
|
},
|
|
{
|
|
"epoch": 1.9968051118210863,
|
|
"grad_norm": 0.5719947285802028,
|
|
"learning_rate": 3.597755715500002e-05,
|
|
"loss": 0.2423,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25348833203315735,
|
|
"step": 1250,
|
|
"valid_targets_mean": 3761.4,
|
|
"valid_targets_min": 1730
|
|
},
|
|
{
|
|
"epoch": 2.0047923322683707,
|
|
"grad_norm": 0.4535662103492767,
|
|
"learning_rate": 3.592950640711642e-05,
|
|
"loss": 0.2229,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2162315994501114,
|
|
"step": 1255,
|
|
"valid_targets_mean": 3989.9,
|
|
"valid_targets_min": 907
|
|
},
|
|
{
|
|
"epoch": 2.012779552715655,
|
|
"grad_norm": 0.4410351357350173,
|
|
"learning_rate": 3.5881202852902543e-05,
|
|
"loss": 0.2057,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20752325654029846,
|
|
"step": 1260,
|
|
"valid_targets_mean": 5998.4,
|
|
"valid_targets_min": 2837
|
|
},
|
|
{
|
|
"epoch": 2.0207667731629395,
|
|
"grad_norm": 0.47898384165843794,
|
|
"learning_rate": 3.583264725895117e-05,
|
|
"loss": 0.2162,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20195890963077545,
|
|
"step": 1265,
|
|
"valid_targets_mean": 4670.1,
|
|
"valid_targets_min": 1350
|
|
},
|
|
{
|
|
"epoch": 2.0287539936102235,
|
|
"grad_norm": 0.5515945007484138,
|
|
"learning_rate": 3.5783840395854984e-05,
|
|
"loss": 0.205,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19788508117198944,
|
|
"step": 1270,
|
|
"valid_targets_mean": 5088.9,
|
|
"valid_targets_min": 1475
|
|
},
|
|
{
|
|
"epoch": 2.036741214057508,
|
|
"grad_norm": 0.5431215978248436,
|
|
"learning_rate": 3.5734783038194446e-05,
|
|
"loss": 0.2078,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24707621335983276,
|
|
"step": 1275,
|
|
"valid_targets_mean": 4470.7,
|
|
"valid_targets_min": 572
|
|
},
|
|
{
|
|
"epoch": 2.0447284345047922,
|
|
"grad_norm": 0.6459598360479897,
|
|
"learning_rate": 3.5685475964525396e-05,
|
|
"loss": 0.2232,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2134454846382141,
|
|
"step": 1280,
|
|
"valid_targets_mean": 4424.7,
|
|
"valid_targets_min": 1468
|
|
},
|
|
{
|
|
"epoch": 2.0527156549520766,
|
|
"grad_norm": 0.47235874120995064,
|
|
"learning_rate": 3.5635919957366775e-05,
|
|
"loss": 0.2059,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17809467017650604,
|
|
"step": 1285,
|
|
"valid_targets_mean": 4389.7,
|
|
"valid_targets_min": 634
|
|
},
|
|
{
|
|
"epoch": 2.060702875399361,
|
|
"grad_norm": 0.43600370251565135,
|
|
"learning_rate": 3.558611580318817e-05,
|
|
"loss": 0.2093,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19783927500247955,
|
|
"step": 1290,
|
|
"valid_targets_mean": 5044.0,
|
|
"valid_targets_min": 2058
|
|
},
|
|
{
|
|
"epoch": 2.0686900958466454,
|
|
"grad_norm": 0.4309467855687483,
|
|
"learning_rate": 3.5536064292397334e-05,
|
|
"loss": 0.2084,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1924149990081787,
|
|
"step": 1295,
|
|
"valid_targets_mean": 5567.2,
|
|
"valid_targets_min": 2060
|
|
},
|
|
{
|
|
"epoch": 2.07667731629393,
|
|
"grad_norm": 0.4579013316960597,
|
|
"learning_rate": 3.548576621932766e-05,
|
|
"loss": 0.2062,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20921730995178223,
|
|
"step": 1300,
|
|
"valid_targets_mean": 5577.4,
|
|
"valid_targets_min": 889
|
|
},
|
|
{
|
|
"epoch": 2.084664536741214,
|
|
"grad_norm": 0.4435730045136153,
|
|
"learning_rate": 3.543522238222555e-05,
|
|
"loss": 0.2119,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18302898108959198,
|
|
"step": 1305,
|
|
"valid_targets_mean": 5364.7,
|
|
"valid_targets_min": 2309
|
|
},
|
|
{
|
|
"epoch": 2.0926517571884986,
|
|
"grad_norm": 0.4845583718038434,
|
|
"learning_rate": 3.538443358323777e-05,
|
|
"loss": 0.2076,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2209482342004776,
|
|
"step": 1310,
|
|
"valid_targets_mean": 4514.1,
|
|
"valid_targets_min": 1366
|
|
},
|
|
{
|
|
"epoch": 2.1006389776357826,
|
|
"grad_norm": 0.596915693931467,
|
|
"learning_rate": 3.5333400628398696e-05,
|
|
"loss": 0.199,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1920449137687683,
|
|
"step": 1315,
|
|
"valid_targets_mean": 5001.5,
|
|
"valid_targets_min": 941
|
|
},
|
|
{
|
|
"epoch": 2.108626198083067,
|
|
"grad_norm": 0.4537412759793582,
|
|
"learning_rate": 3.5282124327617556e-05,
|
|
"loss": 0.2168,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23330473899841309,
|
|
"step": 1320,
|
|
"valid_targets_mean": 4856.5,
|
|
"valid_targets_min": 851
|
|
},
|
|
{
|
|
"epoch": 2.1166134185303513,
|
|
"grad_norm": 0.5170397096645349,
|
|
"learning_rate": 3.5230605494665535e-05,
|
|
"loss": 0.2095,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24425125122070312,
|
|
"step": 1325,
|
|
"valid_targets_mean": 5382.1,
|
|
"valid_targets_min": 1194
|
|
},
|
|
{
|
|
"epoch": 2.1246006389776357,
|
|
"grad_norm": 0.4546000330875769,
|
|
"learning_rate": 3.517884494716289e-05,
|
|
"loss": 0.2061,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19844010472297668,
|
|
"step": 1330,
|
|
"valid_targets_mean": 4694.7,
|
|
"valid_targets_min": 2177
|
|
},
|
|
{
|
|
"epoch": 2.13258785942492,
|
|
"grad_norm": 0.4666137402009551,
|
|
"learning_rate": 3.512684350656596e-05,
|
|
"loss": 0.207,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22991979122161865,
|
|
"step": 1335,
|
|
"valid_targets_mean": 4580.0,
|
|
"valid_targets_min": 1281
|
|
},
|
|
{
|
|
"epoch": 2.1405750798722045,
|
|
"grad_norm": 0.43841659587073956,
|
|
"learning_rate": 3.507460199815414e-05,
|
|
"loss": 0.2329,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23546649515628815,
|
|
"step": 1340,
|
|
"valid_targets_mean": 5230.1,
|
|
"valid_targets_min": 2442
|
|
},
|
|
{
|
|
"epoch": 2.148562300319489,
|
|
"grad_norm": 0.43409838130564055,
|
|
"learning_rate": 3.5022121251016786e-05,
|
|
"loss": 0.2084,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20348697900772095,
|
|
"step": 1345,
|
|
"valid_targets_mean": 5317.4,
|
|
"valid_targets_min": 1682
|
|
},
|
|
{
|
|
"epoch": 2.1565495207667733,
|
|
"grad_norm": 0.44719714225413865,
|
|
"learning_rate": 3.4969402098040015e-05,
|
|
"loss": 0.2087,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17266574501991272,
|
|
"step": 1350,
|
|
"valid_targets_mean": 4226.5,
|
|
"valid_targets_min": 659
|
|
},
|
|
{
|
|
"epoch": 2.1645367412140577,
|
|
"grad_norm": 0.4185009785089201,
|
|
"learning_rate": 3.491644537589356e-05,
|
|
"loss": 0.2182,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20402270555496216,
|
|
"step": 1355,
|
|
"valid_targets_mean": 5367.0,
|
|
"valid_targets_min": 1574
|
|
},
|
|
{
|
|
"epoch": 2.1725239616613417,
|
|
"grad_norm": 0.4300283341003427,
|
|
"learning_rate": 3.486325192501743e-05,
|
|
"loss": 0.2023,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19802656769752502,
|
|
"step": 1360,
|
|
"valid_targets_mean": 4707.4,
|
|
"valid_targets_min": 1393
|
|
},
|
|
{
|
|
"epoch": 2.180511182108626,
|
|
"grad_norm": 0.4244014902369462,
|
|
"learning_rate": 3.480982258960859e-05,
|
|
"loss": 0.2025,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18263283371925354,
|
|
"step": 1365,
|
|
"valid_targets_mean": 4805.6,
|
|
"valid_targets_min": 1855
|
|
},
|
|
{
|
|
"epoch": 2.1884984025559104,
|
|
"grad_norm": 0.4259899136783193,
|
|
"learning_rate": 3.47561582176076e-05,
|
|
"loss": 0.2063,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20305916666984558,
|
|
"step": 1370,
|
|
"valid_targets_mean": 5255.1,
|
|
"valid_targets_min": 2441
|
|
},
|
|
{
|
|
"epoch": 2.196485623003195,
|
|
"grad_norm": 0.4774516932266148,
|
|
"learning_rate": 3.470225966068507e-05,
|
|
"loss": 0.2082,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2115797996520996,
|
|
"step": 1375,
|
|
"valid_targets_mean": 5119.3,
|
|
"valid_targets_min": 835
|
|
},
|
|
{
|
|
"epoch": 2.2044728434504792,
|
|
"grad_norm": 0.4548863057712348,
|
|
"learning_rate": 3.464812777422826e-05,
|
|
"loss": 0.2188,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1970842033624649,
|
|
"step": 1380,
|
|
"valid_targets_mean": 4329.1,
|
|
"valid_targets_min": 1090
|
|
},
|
|
{
|
|
"epoch": 2.2124600638977636,
|
|
"grad_norm": 0.4892500528484616,
|
|
"learning_rate": 3.459376341732741e-05,
|
|
"loss": 0.1901,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.220895454287529,
|
|
"step": 1385,
|
|
"valid_targets_mean": 5527.6,
|
|
"valid_targets_min": 1416
|
|
},
|
|
{
|
|
"epoch": 2.220447284345048,
|
|
"grad_norm": 0.489398730051602,
|
|
"learning_rate": 3.453916745276213e-05,
|
|
"loss": 0.2029,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20596617460250854,
|
|
"step": 1390,
|
|
"valid_targets_mean": 4585.8,
|
|
"valid_targets_min": 587
|
|
},
|
|
{
|
|
"epoch": 2.2284345047923324,
|
|
"grad_norm": 0.47959886490092324,
|
|
"learning_rate": 3.4484340746987745e-05,
|
|
"loss": 0.2066,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18780018389225006,
|
|
"step": 1395,
|
|
"valid_targets_mean": 4551.4,
|
|
"valid_targets_min": 2053
|
|
},
|
|
{
|
|
"epoch": 2.236421725239617,
|
|
"grad_norm": 0.4620077078997459,
|
|
"learning_rate": 3.4429284170121494e-05,
|
|
"loss": 0.2117,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23137012124061584,
|
|
"step": 1400,
|
|
"valid_targets_mean": 5430.0,
|
|
"valid_targets_min": 1432
|
|
},
|
|
{
|
|
"epoch": 2.244408945686901,
|
|
"grad_norm": 0.49180601508096844,
|
|
"learning_rate": 3.4373998595928764e-05,
|
|
"loss": 0.2076,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20962712168693542,
|
|
"step": 1405,
|
|
"valid_targets_mean": 4032.1,
|
|
"valid_targets_min": 1012
|
|
},
|
|
{
|
|
"epoch": 2.252396166134185,
|
|
"grad_norm": 0.48861738455509557,
|
|
"learning_rate": 3.431848490180919e-05,
|
|
"loss": 0.2037,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21679922938346863,
|
|
"step": 1410,
|
|
"valid_targets_mean": 4853.1,
|
|
"valid_targets_min": 916
|
|
},
|
|
{
|
|
"epoch": 2.2603833865814695,
|
|
"grad_norm": 0.408604716929785,
|
|
"learning_rate": 3.4262743968782736e-05,
|
|
"loss": 0.201,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16209134459495544,
|
|
"step": 1415,
|
|
"valid_targets_mean": 4628.4,
|
|
"valid_targets_min": 2578
|
|
},
|
|
{
|
|
"epoch": 2.268370607028754,
|
|
"grad_norm": 0.43752774127215555,
|
|
"learning_rate": 3.4206776681475737e-05,
|
|
"loss": 0.2193,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24136966466903687,
|
|
"step": 1420,
|
|
"valid_targets_mean": 5033.4,
|
|
"valid_targets_min": 909
|
|
},
|
|
{
|
|
"epoch": 2.2763578274760383,
|
|
"grad_norm": 0.43560432902877066,
|
|
"learning_rate": 3.415058392810682e-05,
|
|
"loss": 0.2027,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.198669895529747,
|
|
"step": 1425,
|
|
"valid_targets_mean": 4891.9,
|
|
"valid_targets_min": 784
|
|
},
|
|
{
|
|
"epoch": 2.2843450479233227,
|
|
"grad_norm": 0.580378603983258,
|
|
"learning_rate": 3.409416660047286e-05,
|
|
"loss": 0.2174,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21435877680778503,
|
|
"step": 1430,
|
|
"valid_targets_mean": 4078.2,
|
|
"valid_targets_min": 1318
|
|
},
|
|
{
|
|
"epoch": 2.292332268370607,
|
|
"grad_norm": 0.40700558817808163,
|
|
"learning_rate": 3.403752559393477e-05,
|
|
"loss": 0.2027,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18513426184654236,
|
|
"step": 1435,
|
|
"valid_targets_mean": 5148.9,
|
|
"valid_targets_min": 1067
|
|
},
|
|
{
|
|
"epoch": 2.3003194888178915,
|
|
"grad_norm": 0.4799030454563342,
|
|
"learning_rate": 3.398066180740332e-05,
|
|
"loss": 0.2173,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22550812363624573,
|
|
"step": 1440,
|
|
"valid_targets_mean": 5266.4,
|
|
"valid_targets_min": 1313
|
|
},
|
|
{
|
|
"epoch": 2.308306709265176,
|
|
"grad_norm": 0.4974331710461038,
|
|
"learning_rate": 3.39235761433249e-05,
|
|
"loss": 0.2133,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2103099226951599,
|
|
"step": 1445,
|
|
"valid_targets_mean": 4404.1,
|
|
"valid_targets_min": 2121
|
|
},
|
|
{
|
|
"epoch": 2.31629392971246,
|
|
"grad_norm": 0.4340003337295617,
|
|
"learning_rate": 3.386626950766713e-05,
|
|
"loss": 0.2111,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20411810278892517,
|
|
"step": 1450,
|
|
"valid_targets_mean": 4893.8,
|
|
"valid_targets_min": 1976
|
|
},
|
|
{
|
|
"epoch": 2.3242811501597442,
|
|
"grad_norm": 0.4577279718543424,
|
|
"learning_rate": 3.380874280990454e-05,
|
|
"loss": 0.1954,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1925685852766037,
|
|
"step": 1455,
|
|
"valid_targets_mean": 4346.2,
|
|
"valid_targets_min": 682
|
|
},
|
|
{
|
|
"epoch": 2.3322683706070286,
|
|
"grad_norm": 0.43413317097279497,
|
|
"learning_rate": 3.37509969630041e-05,
|
|
"loss": 0.207,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21768368780612946,
|
|
"step": 1460,
|
|
"valid_targets_mean": 5565.8,
|
|
"valid_targets_min": 2098
|
|
},
|
|
{
|
|
"epoch": 2.340255591054313,
|
|
"grad_norm": 0.4536871654290314,
|
|
"learning_rate": 3.369303288341078e-05,
|
|
"loss": 0.2123,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1936158984899521,
|
|
"step": 1465,
|
|
"valid_targets_mean": 4365.2,
|
|
"valid_targets_min": 1036
|
|
},
|
|
{
|
|
"epoch": 2.3482428115015974,
|
|
"grad_norm": 0.4567310654580342,
|
|
"learning_rate": 3.363485149103292e-05,
|
|
"loss": 0.2183,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25850218534469604,
|
|
"step": 1470,
|
|
"valid_targets_mean": 5031.1,
|
|
"valid_targets_min": 2001
|
|
},
|
|
{
|
|
"epoch": 2.356230031948882,
|
|
"grad_norm": 0.5977141302333794,
|
|
"learning_rate": 3.357645370922772e-05,
|
|
"loss": 0.2017,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23247145116329193,
|
|
"step": 1475,
|
|
"valid_targets_mean": 3937.1,
|
|
"valid_targets_min": 625
|
|
},
|
|
{
|
|
"epoch": 2.364217252396166,
|
|
"grad_norm": 0.44895842484039244,
|
|
"learning_rate": 3.351784046478655e-05,
|
|
"loss": 0.2212,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21234402060508728,
|
|
"step": 1480,
|
|
"valid_targets_mean": 5188.9,
|
|
"valid_targets_min": 819
|
|
},
|
|
{
|
|
"epoch": 2.3722044728434506,
|
|
"grad_norm": 0.47003250547900366,
|
|
"learning_rate": 3.345901268792022e-05,
|
|
"loss": 0.2032,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19264158606529236,
|
|
"step": 1485,
|
|
"valid_targets_mean": 4191.7,
|
|
"valid_targets_min": 568
|
|
},
|
|
{
|
|
"epoch": 2.380191693290735,
|
|
"grad_norm": 0.44506494499803045,
|
|
"learning_rate": 3.3399971312244255e-05,
|
|
"loss": 0.2104,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22497370839118958,
|
|
"step": 1490,
|
|
"valid_targets_mean": 5244.2,
|
|
"valid_targets_min": 1405
|
|
},
|
|
{
|
|
"epoch": 2.3881789137380194,
|
|
"grad_norm": 0.46933729486270825,
|
|
"learning_rate": 3.334071727476406e-05,
|
|
"loss": 0.2138,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22563546895980835,
|
|
"step": 1495,
|
|
"valid_targets_mean": 4995.9,
|
|
"valid_targets_min": 1358
|
|
},
|
|
{
|
|
"epoch": 2.3961661341853033,
|
|
"grad_norm": 0.5019361637475704,
|
|
"learning_rate": 3.3281251515860035e-05,
|
|
"loss": 0.2143,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23603831231594086,
|
|
"step": 1500,
|
|
"valid_targets_mean": 5936.2,
|
|
"valid_targets_min": 288
|
|
},
|
|
{
|
|
"epoch": 2.4041533546325877,
|
|
"grad_norm": 0.40308147209479217,
|
|
"learning_rate": 3.3221574979272676e-05,
|
|
"loss": 0.2141,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20887164771556854,
|
|
"step": 1505,
|
|
"valid_targets_mean": 6166.1,
|
|
"valid_targets_min": 836
|
|
},
|
|
{
|
|
"epoch": 2.412140575079872,
|
|
"grad_norm": 0.4398473748615588,
|
|
"learning_rate": 3.316168861208759e-05,
|
|
"loss": 0.2071,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20377591252326965,
|
|
"step": 1510,
|
|
"valid_targets_mean": 4693.8,
|
|
"valid_targets_min": 590
|
|
},
|
|
{
|
|
"epoch": 2.4201277955271565,
|
|
"grad_norm": 0.5309344597523309,
|
|
"learning_rate": 3.310159336472047e-05,
|
|
"loss": 0.2061,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21097645163536072,
|
|
"step": 1515,
|
|
"valid_targets_mean": 4200.9,
|
|
"valid_targets_min": 761
|
|
},
|
|
{
|
|
"epoch": 2.428115015974441,
|
|
"grad_norm": 0.5532178599586695,
|
|
"learning_rate": 3.304129019090198e-05,
|
|
"loss": 0.2098,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2257629632949829,
|
|
"step": 1520,
|
|
"valid_targets_mean": 4738.1,
|
|
"valid_targets_min": 968
|
|
},
|
|
{
|
|
"epoch": 2.4361022364217253,
|
|
"grad_norm": 0.4607360256928386,
|
|
"learning_rate": 3.298078004766267e-05,
|
|
"loss": 0.1935,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17515632510185242,
|
|
"step": 1525,
|
|
"valid_targets_mean": 3853.8,
|
|
"valid_targets_min": 1100
|
|
},
|
|
{
|
|
"epoch": 2.4440894568690097,
|
|
"grad_norm": 0.4370375317963132,
|
|
"learning_rate": 3.292006389531774e-05,
|
|
"loss": 0.1946,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16992877423763275,
|
|
"step": 1530,
|
|
"valid_targets_mean": 4876.9,
|
|
"valid_targets_min": 1554
|
|
},
|
|
{
|
|
"epoch": 2.452076677316294,
|
|
"grad_norm": 0.41401464203003635,
|
|
"learning_rate": 3.2859142697451835e-05,
|
|
"loss": 0.218,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21531742811203003,
|
|
"step": 1535,
|
|
"valid_targets_mean": 6467.6,
|
|
"valid_targets_min": 2188
|
|
},
|
|
{
|
|
"epoch": 2.460063897763578,
|
|
"grad_norm": 0.6483636089392388,
|
|
"learning_rate": 3.279801742090372e-05,
|
|
"loss": 0.2113,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26298800110816956,
|
|
"step": 1540,
|
|
"valid_targets_mean": 5260.5,
|
|
"valid_targets_min": 724
|
|
},
|
|
{
|
|
"epoch": 2.4680511182108624,
|
|
"grad_norm": 0.4802414273434107,
|
|
"learning_rate": 3.2736689035750975e-05,
|
|
"loss": 0.2031,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20041510462760925,
|
|
"step": 1545,
|
|
"valid_targets_mean": 4248.7,
|
|
"valid_targets_min": 943
|
|
},
|
|
{
|
|
"epoch": 2.476038338658147,
|
|
"grad_norm": 0.4443471057730885,
|
|
"learning_rate": 3.2675158515294554e-05,
|
|
"loss": 0.2026,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1807590126991272,
|
|
"step": 1550,
|
|
"valid_targets_mean": 4097.3,
|
|
"valid_targets_min": 1009
|
|
},
|
|
{
|
|
"epoch": 2.484025559105431,
|
|
"grad_norm": 0.4902750001832227,
|
|
"learning_rate": 3.2613426836043386e-05,
|
|
"loss": 0.1964,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2025105357170105,
|
|
"step": 1555,
|
|
"valid_targets_mean": 4429.0,
|
|
"valid_targets_min": 862
|
|
},
|
|
{
|
|
"epoch": 2.4920127795527156,
|
|
"grad_norm": 0.4582351957066707,
|
|
"learning_rate": 3.2551494977698844e-05,
|
|
"loss": 0.1926,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18224765360355377,
|
|
"step": 1560,
|
|
"valid_targets_mean": 4185.1,
|
|
"valid_targets_min": 519
|
|
},
|
|
{
|
|
"epoch": 2.5,
|
|
"grad_norm": 0.4254457086333195,
|
|
"learning_rate": 3.248936392313921e-05,
|
|
"loss": 0.2195,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1733468770980835,
|
|
"step": 1565,
|
|
"valid_targets_mean": 4750.7,
|
|
"valid_targets_min": 1263
|
|
},
|
|
{
|
|
"epoch": 2.5079872204472844,
|
|
"grad_norm": 0.45224970751667104,
|
|
"learning_rate": 3.2427034658404056e-05,
|
|
"loss": 0.2269,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21232850849628448,
|
|
"step": 1570,
|
|
"valid_targets_mean": 4944.6,
|
|
"valid_targets_min": 1634
|
|
},
|
|
{
|
|
"epoch": 2.515974440894569,
|
|
"grad_norm": 0.43711126622124813,
|
|
"learning_rate": 3.236450817267863e-05,
|
|
"loss": 0.2055,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18365254998207092,
|
|
"step": 1575,
|
|
"valid_targets_mean": 4473.3,
|
|
"valid_targets_min": 721
|
|
},
|
|
{
|
|
"epoch": 2.523961661341853,
|
|
"grad_norm": 0.48918317077505047,
|
|
"learning_rate": 3.230178545827814e-05,
|
|
"loss": 0.2165,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2108568251132965,
|
|
"step": 1580,
|
|
"valid_targets_mean": 4876.8,
|
|
"valid_targets_min": 592
|
|
},
|
|
{
|
|
"epoch": 2.5319488817891376,
|
|
"grad_norm": 0.3925030886140674,
|
|
"learning_rate": 3.223886751063201e-05,
|
|
"loss": 0.2179,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20117145776748657,
|
|
"step": 1585,
|
|
"valid_targets_mean": 6112.6,
|
|
"valid_targets_min": 2397
|
|
},
|
|
{
|
|
"epoch": 2.539936102236422,
|
|
"grad_norm": 0.4505753806395508,
|
|
"learning_rate": 3.2175755328268054e-05,
|
|
"loss": 0.216,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2128993570804596,
|
|
"step": 1590,
|
|
"valid_targets_mean": 5152.8,
|
|
"valid_targets_min": 1554
|
|
},
|
|
{
|
|
"epoch": 2.547923322683706,
|
|
"grad_norm": 0.40790132537857193,
|
|
"learning_rate": 3.2112449912796675e-05,
|
|
"loss": 0.2115,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1775086373090744,
|
|
"step": 1595,
|
|
"valid_targets_mean": 5171.2,
|
|
"valid_targets_min": 1857
|
|
},
|
|
{
|
|
"epoch": 2.5559105431309903,
|
|
"grad_norm": 0.4318086807503156,
|
|
"learning_rate": 3.204895226889494e-05,
|
|
"loss": 0.1933,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19410687685012817,
|
|
"step": 1600,
|
|
"valid_targets_mean": 4738.9,
|
|
"valid_targets_min": 1355
|
|
},
|
|
{
|
|
"epoch": 2.5638977635782747,
|
|
"grad_norm": 0.4636942859708295,
|
|
"learning_rate": 3.1985263404290634e-05,
|
|
"loss": 0.2164,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2244848757982254,
|
|
"step": 1605,
|
|
"valid_targets_mean": 5147.4,
|
|
"valid_targets_min": 2813
|
|
},
|
|
{
|
|
"epoch": 2.571884984025559,
|
|
"grad_norm": 0.47397849207507037,
|
|
"learning_rate": 3.19213843297463e-05,
|
|
"loss": 0.2069,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21858221292495728,
|
|
"step": 1610,
|
|
"valid_targets_mean": 4920.2,
|
|
"valid_targets_min": 1932
|
|
},
|
|
{
|
|
"epoch": 2.5798722044728435,
|
|
"grad_norm": 0.45197351575402667,
|
|
"learning_rate": 3.1857316059043144e-05,
|
|
"loss": 0.1974,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17527048289775848,
|
|
"step": 1615,
|
|
"valid_targets_mean": 4120.8,
|
|
"valid_targets_min": 526
|
|
},
|
|
{
|
|
"epoch": 2.587859424920128,
|
|
"grad_norm": 0.5005051161387774,
|
|
"learning_rate": 3.1793059608964986e-05,
|
|
"loss": 0.1925,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21094384789466858,
|
|
"step": 1620,
|
|
"valid_targets_mean": 5580.8,
|
|
"valid_targets_min": 1367
|
|
},
|
|
{
|
|
"epoch": 2.5958466453674123,
|
|
"grad_norm": 0.47871718041244976,
|
|
"learning_rate": 3.172861599928212e-05,
|
|
"loss": 0.202,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17966912686824799,
|
|
"step": 1625,
|
|
"valid_targets_mean": 3921.2,
|
|
"valid_targets_min": 939
|
|
},
|
|
{
|
|
"epoch": 2.6038338658146962,
|
|
"grad_norm": 0.4375260791787283,
|
|
"learning_rate": 3.166398625273512e-05,
|
|
"loss": 0.2078,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18762406706809998,
|
|
"step": 1630,
|
|
"valid_targets_mean": 4905.2,
|
|
"valid_targets_min": 1067
|
|
},
|
|
{
|
|
"epoch": 2.6118210862619806,
|
|
"grad_norm": 0.3983808594765356,
|
|
"learning_rate": 3.15991713950186e-05,
|
|
"loss": 0.2009,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21603745222091675,
|
|
"step": 1635,
|
|
"valid_targets_mean": 5733.9,
|
|
"valid_targets_min": 2853
|
|
},
|
|
{
|
|
"epoch": 2.619808306709265,
|
|
"grad_norm": 0.45201676841219657,
|
|
"learning_rate": 3.153417245476495e-05,
|
|
"loss": 0.2186,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23388880491256714,
|
|
"step": 1640,
|
|
"valid_targets_mean": 6155.7,
|
|
"valid_targets_min": 2887
|
|
},
|
|
{
|
|
"epoch": 2.6277955271565494,
|
|
"grad_norm": 0.46574710638041417,
|
|
"learning_rate": 3.146899046352804e-05,
|
|
"loss": 0.1929,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1933613419532776,
|
|
"step": 1645,
|
|
"valid_targets_mean": 4483.0,
|
|
"valid_targets_min": 1141
|
|
},
|
|
{
|
|
"epoch": 2.635782747603834,
|
|
"grad_norm": 0.49115410149229843,
|
|
"learning_rate": 3.140362645576676e-05,
|
|
"loss": 0.2122,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2165064811706543,
|
|
"step": 1650,
|
|
"valid_targets_mean": 4406.9,
|
|
"valid_targets_min": 632
|
|
},
|
|
{
|
|
"epoch": 2.643769968051118,
|
|
"grad_norm": 0.4422886145501157,
|
|
"learning_rate": 3.133808146882871e-05,
|
|
"loss": 0.196,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1859450340270996,
|
|
"step": 1655,
|
|
"valid_targets_mean": 4923.4,
|
|
"valid_targets_min": 729
|
|
},
|
|
{
|
|
"epoch": 2.6517571884984026,
|
|
"grad_norm": 0.5105266767260843,
|
|
"learning_rate": 3.1272356542933654e-05,
|
|
"loss": 0.1918,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21490055322647095,
|
|
"step": 1660,
|
|
"valid_targets_mean": 5973.7,
|
|
"valid_targets_min": 3336
|
|
},
|
|
{
|
|
"epoch": 2.659744408945687,
|
|
"grad_norm": 0.6953455819856134,
|
|
"learning_rate": 3.120645272115707e-05,
|
|
"loss": 0.2086,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18827220797538757,
|
|
"step": 1665,
|
|
"valid_targets_mean": 4309.4,
|
|
"valid_targets_min": 1276
|
|
},
|
|
{
|
|
"epoch": 2.6677316293929714,
|
|
"grad_norm": 0.5154053185300087,
|
|
"learning_rate": 3.114037104941355e-05,
|
|
"loss": 0.1944,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19753989577293396,
|
|
"step": 1670,
|
|
"valid_targets_mean": 4340.2,
|
|
"valid_targets_min": 1424
|
|
},
|
|
{
|
|
"epoch": 2.6757188498402558,
|
|
"grad_norm": 0.4782161700404796,
|
|
"learning_rate": 3.1074112576440236e-05,
|
|
"loss": 0.2027,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.175480455160141,
|
|
"step": 1675,
|
|
"valid_targets_mean": 3959.3,
|
|
"valid_targets_min": 680
|
|
},
|
|
{
|
|
"epoch": 2.68370607028754,
|
|
"grad_norm": 0.4453801116013753,
|
|
"learning_rate": 3.1007678353780154e-05,
|
|
"loss": 0.2154,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20716232061386108,
|
|
"step": 1680,
|
|
"valid_targets_mean": 5072.2,
|
|
"valid_targets_min": 1962
|
|
},
|
|
{
|
|
"epoch": 2.6916932907348246,
|
|
"grad_norm": 0.42387415100705655,
|
|
"learning_rate": 3.094106943576553e-05,
|
|
"loss": 0.2021,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20682209730148315,
|
|
"step": 1685,
|
|
"valid_targets_mean": 5182.8,
|
|
"valid_targets_min": 2198
|
|
},
|
|
{
|
|
"epoch": 2.6996805111821085,
|
|
"grad_norm": 0.8505431396696655,
|
|
"learning_rate": 3.087428687950108e-05,
|
|
"loss": 0.1886,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1699068248271942,
|
|
"step": 1690,
|
|
"valid_targets_mean": 5186.1,
|
|
"valid_targets_min": 1578
|
|
},
|
|
{
|
|
"epoch": 2.707667731629393,
|
|
"grad_norm": 0.42454000480579385,
|
|
"learning_rate": 3.080733174484719e-05,
|
|
"loss": 0.202,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22431977093219757,
|
|
"step": 1695,
|
|
"valid_targets_mean": 5460.5,
|
|
"valid_targets_min": 1251
|
|
},
|
|
{
|
|
"epoch": 2.7156549520766773,
|
|
"grad_norm": 0.515113513985079,
|
|
"learning_rate": 3.074020509440313e-05,
|
|
"loss": 0.2024,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18955373764038086,
|
|
"step": 1700,
|
|
"valid_targets_mean": 4786.7,
|
|
"valid_targets_min": 1748
|
|
},
|
|
{
|
|
"epoch": 2.7236421725239617,
|
|
"grad_norm": 0.4441426398491621,
|
|
"learning_rate": 3.0672907993490183e-05,
|
|
"loss": 0.2096,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19987380504608154,
|
|
"step": 1705,
|
|
"valid_targets_mean": 5302.4,
|
|
"valid_targets_min": 1298
|
|
},
|
|
{
|
|
"epoch": 2.731629392971246,
|
|
"grad_norm": 0.5688162028968159,
|
|
"learning_rate": 3.060544151013473e-05,
|
|
"loss": 0.2023,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20857849717140198,
|
|
"step": 1710,
|
|
"valid_targets_mean": 4687.8,
|
|
"valid_targets_min": 1559
|
|
},
|
|
{
|
|
"epoch": 2.7396166134185305,
|
|
"grad_norm": 0.4557968966348482,
|
|
"learning_rate": 3.053780671505132e-05,
|
|
"loss": 0.2105,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19488662481307983,
|
|
"step": 1715,
|
|
"valid_targets_mean": 4964.9,
|
|
"valid_targets_min": 2075
|
|
},
|
|
{
|
|
"epoch": 2.747603833865815,
|
|
"grad_norm": 0.4645476414446028,
|
|
"learning_rate": 3.047000468162563e-05,
|
|
"loss": 0.2087,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19949454069137573,
|
|
"step": 1720,
|
|
"valid_targets_mean": 4732.9,
|
|
"valid_targets_min": 543
|
|
},
|
|
{
|
|
"epoch": 2.755591054313099,
|
|
"grad_norm": 0.3966169427070985,
|
|
"learning_rate": 3.0402036485897496e-05,
|
|
"loss": 0.1972,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1724996268749237,
|
|
"step": 1725,
|
|
"valid_targets_mean": 5957.0,
|
|
"valid_targets_min": 2236
|
|
},
|
|
{
|
|
"epoch": 2.763578274760383,
|
|
"grad_norm": 0.46798109501484725,
|
|
"learning_rate": 3.0333903206543786e-05,
|
|
"loss": 0.2059,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2168780416250229,
|
|
"step": 1730,
|
|
"valid_targets_mean": 4231.0,
|
|
"valid_targets_min": 1070
|
|
},
|
|
{
|
|
"epoch": 2.7715654952076676,
|
|
"grad_norm": 0.4214469481286826,
|
|
"learning_rate": 3.02656059248613e-05,
|
|
"loss": 0.2054,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19142383337020874,
|
|
"step": 1735,
|
|
"valid_targets_mean": 5115.2,
|
|
"valid_targets_min": 665
|
|
},
|
|
{
|
|
"epoch": 2.779552715654952,
|
|
"grad_norm": 0.4303775537672929,
|
|
"learning_rate": 3.0197145724749615e-05,
|
|
"loss": 0.2029,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16952644288539886,
|
|
"step": 1740,
|
|
"valid_targets_mean": 4943.2,
|
|
"valid_targets_min": 983
|
|
},
|
|
{
|
|
"epoch": 2.7875399361022364,
|
|
"grad_norm": 0.529210583397824,
|
|
"learning_rate": 3.0128523692693865e-05,
|
|
"loss": 0.2129,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21570533514022827,
|
|
"step": 1745,
|
|
"valid_targets_mean": 5128.9,
|
|
"valid_targets_min": 1272
|
|
},
|
|
{
|
|
"epoch": 2.7955271565495208,
|
|
"grad_norm": 0.4875526018314679,
|
|
"learning_rate": 3.005974091774751e-05,
|
|
"loss": 0.2108,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20090654492378235,
|
|
"step": 1750,
|
|
"valid_targets_mean": 5847.1,
|
|
"valid_targets_min": 3140
|
|
},
|
|
{
|
|
"epoch": 2.803514376996805,
|
|
"grad_norm": 0.49510928662774695,
|
|
"learning_rate": 2.999079849151505e-05,
|
|
"loss": 0.2039,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1958474963903427,
|
|
"step": 1755,
|
|
"valid_targets_mean": 4237.8,
|
|
"valid_targets_min": 1373
|
|
},
|
|
{
|
|
"epoch": 2.8115015974440896,
|
|
"grad_norm": 0.44073636843613645,
|
|
"learning_rate": 2.992169750813471e-05,
|
|
"loss": 0.2125,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22938530147075653,
|
|
"step": 1760,
|
|
"valid_targets_mean": 5790.8,
|
|
"valid_targets_min": 592
|
|
},
|
|
{
|
|
"epoch": 2.819488817891374,
|
|
"grad_norm": 0.48287316764874594,
|
|
"learning_rate": 2.985243906426106e-05,
|
|
"loss": 0.1985,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18632280826568604,
|
|
"step": 1765,
|
|
"valid_targets_mean": 4863.8,
|
|
"valid_targets_min": 1510
|
|
},
|
|
{
|
|
"epoch": 2.8274760383386583,
|
|
"grad_norm": 0.4345674430560167,
|
|
"learning_rate": 2.9783024259047617e-05,
|
|
"loss": 0.2004,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19696888327598572,
|
|
"step": 1770,
|
|
"valid_targets_mean": 5276.3,
|
|
"valid_targets_min": 1217
|
|
},
|
|
{
|
|
"epoch": 2.8354632587859427,
|
|
"grad_norm": 0.4842821725079989,
|
|
"learning_rate": 2.971345419412941e-05,
|
|
"loss": 0.1918,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18247222900390625,
|
|
"step": 1775,
|
|
"valid_targets_mean": 4082.5,
|
|
"valid_targets_min": 1215
|
|
},
|
|
{
|
|
"epoch": 2.8434504792332267,
|
|
"grad_norm": 0.42005980553544303,
|
|
"learning_rate": 2.964372997360548e-05,
|
|
"loss": 0.189,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17339880764484406,
|
|
"step": 1780,
|
|
"valid_targets_mean": 5439.2,
|
|
"valid_targets_min": 1617
|
|
},
|
|
{
|
|
"epoch": 2.851437699680511,
|
|
"grad_norm": 0.47263819048565253,
|
|
"learning_rate": 2.957385270402137e-05,
|
|
"loss": 0.1952,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1739785075187683,
|
|
"step": 1785,
|
|
"valid_targets_mean": 3770.3,
|
|
"valid_targets_min": 735
|
|
},
|
|
{
|
|
"epoch": 2.8594249201277955,
|
|
"grad_norm": 0.47326009235291333,
|
|
"learning_rate": 2.9503823494351565e-05,
|
|
"loss": 0.2067,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2062799632549286,
|
|
"step": 1790,
|
|
"valid_targets_mean": 4597.7,
|
|
"valid_targets_min": 1477
|
|
},
|
|
{
|
|
"epoch": 2.86741214057508,
|
|
"grad_norm": 0.45549075497929276,
|
|
"learning_rate": 2.9433643455981874e-05,
|
|
"loss": 0.2079,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19905966520309448,
|
|
"step": 1795,
|
|
"valid_targets_mean": 4486.9,
|
|
"valid_targets_min": 1078
|
|
},
|
|
{
|
|
"epoch": 2.8753993610223643,
|
|
"grad_norm": 0.41464722420988726,
|
|
"learning_rate": 2.9363313702691827e-05,
|
|
"loss": 0.2041,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19828450679779053,
|
|
"step": 1800,
|
|
"valid_targets_mean": 5414.1,
|
|
"valid_targets_min": 372
|
|
},
|
|
{
|
|
"epoch": 2.8833865814696487,
|
|
"grad_norm": 0.4753585149002599,
|
|
"learning_rate": 2.9292835350636957e-05,
|
|
"loss": 0.214,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2153431475162506,
|
|
"step": 1805,
|
|
"valid_targets_mean": 5618.3,
|
|
"valid_targets_min": 299
|
|
},
|
|
{
|
|
"epoch": 2.891373801916933,
|
|
"grad_norm": 0.5840186243183315,
|
|
"learning_rate": 2.922220951833111e-05,
|
|
"loss": 0.2008,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23015804588794708,
|
|
"step": 1810,
|
|
"valid_targets_mean": 4671.9,
|
|
"valid_targets_min": 867
|
|
},
|
|
{
|
|
"epoch": 2.899361022364217,
|
|
"grad_norm": 0.4781950760792564,
|
|
"learning_rate": 2.9151437326628706e-05,
|
|
"loss": 0.2012,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19905801117420197,
|
|
"step": 1815,
|
|
"valid_targets_mean": 4259.0,
|
|
"valid_targets_min": 1696
|
|
},
|
|
{
|
|
"epoch": 2.9073482428115014,
|
|
"grad_norm": 0.5309793198861545,
|
|
"learning_rate": 2.908051989870692e-05,
|
|
"loss": 0.2024,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20656386017799377,
|
|
"step": 1820,
|
|
"valid_targets_mean": 3589.2,
|
|
"valid_targets_min": 657
|
|
},
|
|
{
|
|
"epoch": 2.915335463258786,
|
|
"grad_norm": 0.45103136219819573,
|
|
"learning_rate": 2.9009458360047867e-05,
|
|
"loss": 0.1986,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1953534185886383,
|
|
"step": 1825,
|
|
"valid_targets_mean": 4900.6,
|
|
"valid_targets_min": 727
|
|
},
|
|
{
|
|
"epoch": 2.92332268370607,
|
|
"grad_norm": 0.43991916307467194,
|
|
"learning_rate": 2.8938253838420762e-05,
|
|
"loss": 0.2008,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20834563672542572,
|
|
"step": 1830,
|
|
"valid_targets_mean": 4385.7,
|
|
"valid_targets_min": 1359
|
|
},
|
|
{
|
|
"epoch": 2.9313099041533546,
|
|
"grad_norm": 0.5114257063223862,
|
|
"learning_rate": 2.8866907463864006e-05,
|
|
"loss": 0.2157,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23517729341983795,
|
|
"step": 1835,
|
|
"valid_targets_mean": 4265.8,
|
|
"valid_targets_min": 1344
|
|
},
|
|
{
|
|
"epoch": 2.939297124600639,
|
|
"grad_norm": 0.6007582058907062,
|
|
"learning_rate": 2.8795420368667225e-05,
|
|
"loss": 0.2237,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17349465191364288,
|
|
"step": 1840,
|
|
"valid_targets_mean": 4591.4,
|
|
"valid_targets_min": 1352
|
|
},
|
|
{
|
|
"epoch": 2.9472843450479234,
|
|
"grad_norm": 0.47686092522453105,
|
|
"learning_rate": 2.8723793687353354e-05,
|
|
"loss": 0.2054,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21949972212314606,
|
|
"step": 1845,
|
|
"valid_targets_mean": 4742.1,
|
|
"valid_targets_min": 737
|
|
},
|
|
{
|
|
"epoch": 2.9552715654952078,
|
|
"grad_norm": 0.4541073469272048,
|
|
"learning_rate": 2.8652028556660593e-05,
|
|
"loss": 0.2149,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18832948803901672,
|
|
"step": 1850,
|
|
"valid_targets_mean": 5189.2,
|
|
"valid_targets_min": 644
|
|
},
|
|
{
|
|
"epoch": 2.963258785942492,
|
|
"grad_norm": 0.4881400662727766,
|
|
"learning_rate": 2.8580126115524373e-05,
|
|
"loss": 0.2125,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1738438755273819,
|
|
"step": 1855,
|
|
"valid_targets_mean": 4233.8,
|
|
"valid_targets_min": 1245
|
|
},
|
|
{
|
|
"epoch": 2.9712460063897765,
|
|
"grad_norm": 0.4820316182908142,
|
|
"learning_rate": 2.8508087505059302e-05,
|
|
"loss": 0.2161,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1900295913219452,
|
|
"step": 1860,
|
|
"valid_targets_mean": 4443.1,
|
|
"valid_targets_min": 870
|
|
},
|
|
{
|
|
"epoch": 2.979233226837061,
|
|
"grad_norm": 0.4534930639820006,
|
|
"learning_rate": 2.843591386854102e-05,
|
|
"loss": 0.2046,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22404572367668152,
|
|
"step": 1865,
|
|
"valid_targets_mean": 4954.1,
|
|
"valid_targets_min": 1848
|
|
},
|
|
{
|
|
"epoch": 2.987220447284345,
|
|
"grad_norm": 0.5923245030683633,
|
|
"learning_rate": 2.8363606351388068e-05,
|
|
"loss": 0.2011,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20152445137500763,
|
|
"step": 1870,
|
|
"valid_targets_mean": 5321.8,
|
|
"valid_targets_min": 358
|
|
},
|
|
{
|
|
"epoch": 2.9952076677316293,
|
|
"grad_norm": 0.46168562051066675,
|
|
"learning_rate": 2.829116610114375e-05,
|
|
"loss": 0.2094,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2535361349582672,
|
|
"step": 1875,
|
|
"valid_targets_mean": 5529.3,
|
|
"valid_targets_min": 2116
|
|
},
|
|
{
|
|
"epoch": 3.0031948881789137,
|
|
"grad_norm": 0.4203505928859863,
|
|
"learning_rate": 2.821859426745786e-05,
|
|
"loss": 0.1864,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16350862383842468,
|
|
"step": 1880,
|
|
"valid_targets_mean": 4606.1,
|
|
"valid_targets_min": 1318
|
|
},
|
|
{
|
|
"epoch": 3.011182108626198,
|
|
"grad_norm": 0.4596351934581623,
|
|
"learning_rate": 2.8145892002068454e-05,
|
|
"loss": 0.193,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20174235105514526,
|
|
"step": 1885,
|
|
"valid_targets_mean": 5120.6,
|
|
"valid_targets_min": 984
|
|
},
|
|
{
|
|
"epoch": 3.0191693290734825,
|
|
"grad_norm": 0.5067879256603599,
|
|
"learning_rate": 2.8073060458783606e-05,
|
|
"loss": 0.1917,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18824288249015808,
|
|
"step": 1890,
|
|
"valid_targets_mean": 4709.0,
|
|
"valid_targets_min": 1568
|
|
},
|
|
{
|
|
"epoch": 3.027156549520767,
|
|
"grad_norm": 0.46646953173336714,
|
|
"learning_rate": 2.8000100793463056e-05,
|
|
"loss": 0.1786,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16396231949329376,
|
|
"step": 1895,
|
|
"valid_targets_mean": 4063.9,
|
|
"valid_targets_min": 1475
|
|
},
|
|
{
|
|
"epoch": 3.0351437699680512,
|
|
"grad_norm": 0.5062153381404296,
|
|
"learning_rate": 2.792701416399988e-05,
|
|
"loss": 0.1844,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17092962563037872,
|
|
"step": 1900,
|
|
"valid_targets_mean": 4888.4,
|
|
"valid_targets_min": 1583
|
|
},
|
|
{
|
|
"epoch": 3.0431309904153356,
|
|
"grad_norm": 0.4849825982152225,
|
|
"learning_rate": 2.7853801730302134e-05,
|
|
"loss": 0.179,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18273237347602844,
|
|
"step": 1905,
|
|
"valid_targets_mean": 4441.9,
|
|
"valid_targets_min": 1202
|
|
},
|
|
{
|
|
"epoch": 3.0511182108626196,
|
|
"grad_norm": 0.4427472193966588,
|
|
"learning_rate": 2.7780464654274396e-05,
|
|
"loss": 0.1979,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1674235314130783,
|
|
"step": 1910,
|
|
"valid_targets_mean": 6172.0,
|
|
"valid_targets_min": 1178
|
|
},
|
|
{
|
|
"epoch": 3.059105431309904,
|
|
"grad_norm": 0.47957103937687784,
|
|
"learning_rate": 2.770700409979938e-05,
|
|
"loss": 0.1863,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17426815629005432,
|
|
"step": 1915,
|
|
"valid_targets_mean": 5122.2,
|
|
"valid_targets_min": 1210
|
|
},
|
|
{
|
|
"epoch": 3.0670926517571884,
|
|
"grad_norm": 0.4464980807866025,
|
|
"learning_rate": 2.7633421232719442e-05,
|
|
"loss": 0.1863,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1803584098815918,
|
|
"step": 1920,
|
|
"valid_targets_mean": 5715.9,
|
|
"valid_targets_min": 916
|
|
},
|
|
{
|
|
"epoch": 3.0750798722044728,
|
|
"grad_norm": 0.447283516112869,
|
|
"learning_rate": 2.7559717220818067e-05,
|
|
"loss": 0.1863,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17208212614059448,
|
|
"step": 1925,
|
|
"valid_targets_mean": 5718.5,
|
|
"valid_targets_min": 1113
|
|
},
|
|
{
|
|
"epoch": 3.083067092651757,
|
|
"grad_norm": 0.7231060695564152,
|
|
"learning_rate": 2.7485893233801366e-05,
|
|
"loss": 0.1932,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2085496187210083,
|
|
"step": 1930,
|
|
"valid_targets_mean": 5697.9,
|
|
"valid_targets_min": 1392
|
|
},
|
|
{
|
|
"epoch": 3.0910543130990416,
|
|
"grad_norm": 0.43218420898922144,
|
|
"learning_rate": 2.7411950443279458e-05,
|
|
"loss": 0.1844,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17368361353874207,
|
|
"step": 1935,
|
|
"valid_targets_mean": 5389.1,
|
|
"valid_targets_min": 2315
|
|
},
|
|
{
|
|
"epoch": 3.099041533546326,
|
|
"grad_norm": 0.48232381925589013,
|
|
"learning_rate": 2.7337890022747945e-05,
|
|
"loss": 0.1967,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18132419884204865,
|
|
"step": 1940,
|
|
"valid_targets_mean": 5897.9,
|
|
"valid_targets_min": 1980
|
|
},
|
|
{
|
|
"epoch": 3.1070287539936103,
|
|
"grad_norm": 0.44350201563584263,
|
|
"learning_rate": 2.7263713147569243e-05,
|
|
"loss": 0.182,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17033743858337402,
|
|
"step": 1945,
|
|
"valid_targets_mean": 4619.8,
|
|
"valid_targets_min": 1470
|
|
},
|
|
{
|
|
"epoch": 3.1150159744408947,
|
|
"grad_norm": 0.48205388274643085,
|
|
"learning_rate": 2.7189420994953945e-05,
|
|
"loss": 0.1836,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19110047817230225,
|
|
"step": 1950,
|
|
"valid_targets_mean": 4604.9,
|
|
"valid_targets_min": 952
|
|
},
|
|
{
|
|
"epoch": 3.123003194888179,
|
|
"grad_norm": 0.4367361113266311,
|
|
"learning_rate": 2.7115014743942114e-05,
|
|
"loss": 0.186,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1712811291217804,
|
|
"step": 1955,
|
|
"valid_targets_mean": 4904.7,
|
|
"valid_targets_min": 2258
|
|
},
|
|
{
|
|
"epoch": 3.130990415335463,
|
|
"grad_norm": 0.438479271064866,
|
|
"learning_rate": 2.704049557538461e-05,
|
|
"loss": 0.1991,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17538374662399292,
|
|
"step": 1960,
|
|
"valid_targets_mean": 5102.3,
|
|
"valid_targets_min": 1947
|
|
},
|
|
{
|
|
"epoch": 3.1389776357827475,
|
|
"grad_norm": 0.51415649262304,
|
|
"learning_rate": 2.696586467192434e-05,
|
|
"loss": 0.1802,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1985991895198822,
|
|
"step": 1965,
|
|
"valid_targets_mean": 3896.9,
|
|
"valid_targets_min": 917
|
|
},
|
|
{
|
|
"epoch": 3.146964856230032,
|
|
"grad_norm": 0.43403612119727447,
|
|
"learning_rate": 2.6891123217977448e-05,
|
|
"loss": 0.184,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1899278461933136,
|
|
"step": 1970,
|
|
"valid_targets_mean": 5828.6,
|
|
"valid_targets_min": 596
|
|
},
|
|
{
|
|
"epoch": 3.1549520766773163,
|
|
"grad_norm": 0.4985226013639757,
|
|
"learning_rate": 2.681627239971458e-05,
|
|
"loss": 0.1963,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20802980661392212,
|
|
"step": 1975,
|
|
"valid_targets_mean": 5021.2,
|
|
"valid_targets_min": 543
|
|
},
|
|
{
|
|
"epoch": 3.1629392971246006,
|
|
"grad_norm": 0.4771929236008774,
|
|
"learning_rate": 2.6741313405041997e-05,
|
|
"loss": 0.1926,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2114463746547699,
|
|
"step": 1980,
|
|
"valid_targets_mean": 5528.2,
|
|
"valid_targets_min": 1181
|
|
},
|
|
{
|
|
"epoch": 3.170926517571885,
|
|
"grad_norm": 0.4423988902881132,
|
|
"learning_rate": 2.666624742358279e-05,
|
|
"loss": 0.2001,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18834856152534485,
|
|
"step": 1985,
|
|
"valid_targets_mean": 5797.9,
|
|
"valid_targets_min": 2396
|
|
},
|
|
{
|
|
"epoch": 3.1789137380191694,
|
|
"grad_norm": 0.4743308128905031,
|
|
"learning_rate": 2.659107564665794e-05,
|
|
"loss": 0.2013,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19523614645004272,
|
|
"step": 1990,
|
|
"valid_targets_mean": 5063.4,
|
|
"valid_targets_min": 2050
|
|
},
|
|
{
|
|
"epoch": 3.186900958466454,
|
|
"grad_norm": 0.49163208262052766,
|
|
"learning_rate": 2.6515799267267445e-05,
|
|
"loss": 0.1928,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22430217266082764,
|
|
"step": 1995,
|
|
"valid_targets_mean": 4761.1,
|
|
"valid_targets_min": 1387
|
|
},
|
|
{
|
|
"epoch": 3.194888178913738,
|
|
"grad_norm": 0.45501689903001435,
|
|
"learning_rate": 2.6440419480071366e-05,
|
|
"loss": 0.1832,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17785659432411194,
|
|
"step": 2000,
|
|
"valid_targets_mean": 4346.3,
|
|
"valid_targets_min": 1566
|
|
},
|
|
{
|
|
"epoch": 3.202875399361022,
|
|
"grad_norm": 0.5011874844026918,
|
|
"learning_rate": 2.6364937481370887e-05,
|
|
"loss": 0.1982,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20688596367835999,
|
|
"step": 2005,
|
|
"valid_targets_mean": 4911.2,
|
|
"valid_targets_min": 634
|
|
},
|
|
{
|
|
"epoch": 3.2108626198083066,
|
|
"grad_norm": 0.49758161473324886,
|
|
"learning_rate": 2.628935446908933e-05,
|
|
"loss": 0.1846,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19779793918132782,
|
|
"step": 2010,
|
|
"valid_targets_mean": 5070.7,
|
|
"valid_targets_min": 1535
|
|
},
|
|
{
|
|
"epoch": 3.218849840255591,
|
|
"grad_norm": 0.4693482728109941,
|
|
"learning_rate": 2.6213671642753124e-05,
|
|
"loss": 0.1884,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17302462458610535,
|
|
"step": 2015,
|
|
"valid_targets_mean": 5099.9,
|
|
"valid_targets_min": 1001
|
|
},
|
|
{
|
|
"epoch": 3.2268370607028753,
|
|
"grad_norm": 1.2388840168096895,
|
|
"learning_rate": 2.6137890203472786e-05,
|
|
"loss": 0.2081,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20180654525756836,
|
|
"step": 2020,
|
|
"valid_targets_mean": 4193.2,
|
|
"valid_targets_min": 713
|
|
},
|
|
{
|
|
"epoch": 3.2348242811501597,
|
|
"grad_norm": 0.451425368532569,
|
|
"learning_rate": 2.606201135392383e-05,
|
|
"loss": 0.1914,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17804279923439026,
|
|
"step": 2025,
|
|
"valid_targets_mean": 4523.9,
|
|
"valid_targets_min": 1052
|
|
},
|
|
{
|
|
"epoch": 3.242811501597444,
|
|
"grad_norm": 0.47890074437923297,
|
|
"learning_rate": 2.5986036298327725e-05,
|
|
"loss": 0.1906,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1896950602531433,
|
|
"step": 2030,
|
|
"valid_targets_mean": 5309.2,
|
|
"valid_targets_min": 570
|
|
},
|
|
{
|
|
"epoch": 3.2507987220447285,
|
|
"grad_norm": 0.47760863077509297,
|
|
"learning_rate": 2.590996624243276e-05,
|
|
"loss": 0.1991,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17177632451057434,
|
|
"step": 2035,
|
|
"valid_targets_mean": 4826.1,
|
|
"valid_targets_min": 1044
|
|
},
|
|
{
|
|
"epoch": 3.258785942492013,
|
|
"grad_norm": 0.44892367909412995,
|
|
"learning_rate": 2.5833802393494897e-05,
|
|
"loss": 0.1884,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19597885012626648,
|
|
"step": 2040,
|
|
"valid_targets_mean": 5032.6,
|
|
"valid_targets_min": 1418
|
|
},
|
|
{
|
|
"epoch": 3.2667731629392973,
|
|
"grad_norm": 0.5222618703273273,
|
|
"learning_rate": 2.5757545960258638e-05,
|
|
"loss": 0.1808,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16722160577774048,
|
|
"step": 2045,
|
|
"valid_targets_mean": 3837.4,
|
|
"valid_targets_min": 935
|
|
},
|
|
{
|
|
"epoch": 3.2747603833865817,
|
|
"grad_norm": 0.43989273642822957,
|
|
"learning_rate": 2.5681198152937804e-05,
|
|
"loss": 0.1954,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1844216287136078,
|
|
"step": 2050,
|
|
"valid_targets_mean": 4744.3,
|
|
"valid_targets_min": 1090
|
|
},
|
|
{
|
|
"epoch": 3.2827476038338657,
|
|
"grad_norm": 0.4442392514686647,
|
|
"learning_rate": 2.5604760183196377e-05,
|
|
"loss": 0.2047,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22049251198768616,
|
|
"step": 2055,
|
|
"valid_targets_mean": 5748.8,
|
|
"valid_targets_min": 1707
|
|
},
|
|
{
|
|
"epoch": 3.29073482428115,
|
|
"grad_norm": 0.478703439267164,
|
|
"learning_rate": 2.5528233264129228e-05,
|
|
"loss": 0.1941,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22547464072704315,
|
|
"step": 2060,
|
|
"valid_targets_mean": 5527.6,
|
|
"valid_targets_min": 827
|
|
},
|
|
{
|
|
"epoch": 3.2987220447284344,
|
|
"grad_norm": 0.5588579373056553,
|
|
"learning_rate": 2.54516186102429e-05,
|
|
"loss": 0.1846,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15393394231796265,
|
|
"step": 2065,
|
|
"valid_targets_mean": 3522.1,
|
|
"valid_targets_min": 1510
|
|
},
|
|
{
|
|
"epoch": 3.306709265175719,
|
|
"grad_norm": 0.41920245325374705,
|
|
"learning_rate": 2.5374917437436294e-05,
|
|
"loss": 0.185,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19683489203453064,
|
|
"step": 2070,
|
|
"valid_targets_mean": 5266.0,
|
|
"valid_targets_min": 465
|
|
},
|
|
{
|
|
"epoch": 3.3146964856230032,
|
|
"grad_norm": 0.48083992951050875,
|
|
"learning_rate": 2.5298130962981403e-05,
|
|
"loss": 0.1867,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24016697704792023,
|
|
"step": 2075,
|
|
"valid_targets_mean": 5231.2,
|
|
"valid_targets_min": 1409
|
|
},
|
|
{
|
|
"epoch": 3.3226837060702876,
|
|
"grad_norm": 0.6082496138674709,
|
|
"learning_rate": 2.5221260405503985e-05,
|
|
"loss": 0.1966,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23262649774551392,
|
|
"step": 2080,
|
|
"valid_targets_mean": 3419.9,
|
|
"valid_targets_min": 1292
|
|
},
|
|
{
|
|
"epoch": 3.330670926517572,
|
|
"grad_norm": 0.4860773084588283,
|
|
"learning_rate": 2.5144306984964225e-05,
|
|
"loss": 0.1822,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1687614917755127,
|
|
"step": 2085,
|
|
"valid_targets_mean": 3893.1,
|
|
"valid_targets_min": 776
|
|
},
|
|
{
|
|
"epoch": 3.3386581469648564,
|
|
"grad_norm": 0.42498160473633834,
|
|
"learning_rate": 2.5067271922637377e-05,
|
|
"loss": 0.1981,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1992039680480957,
|
|
"step": 2090,
|
|
"valid_targets_mean": 4862.2,
|
|
"valid_targets_min": 2834
|
|
},
|
|
{
|
|
"epoch": 3.3466453674121404,
|
|
"grad_norm": 0.4258825619968381,
|
|
"learning_rate": 2.499015644109435e-05,
|
|
"loss": 0.1923,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17836719751358032,
|
|
"step": 2095,
|
|
"valid_targets_mean": 4995.8,
|
|
"valid_targets_min": 1723
|
|
},
|
|
{
|
|
"epoch": 3.3546325878594248,
|
|
"grad_norm": 0.46930617839559285,
|
|
"learning_rate": 2.491296176418236e-05,
|
|
"loss": 0.1812,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1799635887145996,
|
|
"step": 2100,
|
|
"valid_targets_mean": 4621.9,
|
|
"valid_targets_min": 1217
|
|
},
|
|
{
|
|
"epoch": 3.362619808306709,
|
|
"grad_norm": 0.6584943289712364,
|
|
"learning_rate": 2.4835689117005472e-05,
|
|
"loss": 0.1755,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20586544275283813,
|
|
"step": 2105,
|
|
"valid_targets_mean": 5239.2,
|
|
"valid_targets_min": 1339
|
|
},
|
|
{
|
|
"epoch": 3.3706070287539935,
|
|
"grad_norm": 0.45731375171363,
|
|
"learning_rate": 2.4758339725905152e-05,
|
|
"loss": 0.1975,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.199666365981102,
|
|
"step": 2110,
|
|
"valid_targets_mean": 5336.4,
|
|
"valid_targets_min": 2376
|
|
},
|
|
{
|
|
"epoch": 3.378594249201278,
|
|
"grad_norm": 0.43058193211917495,
|
|
"learning_rate": 2.4680914818440823e-05,
|
|
"loss": 0.1818,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16980770230293274,
|
|
"step": 2115,
|
|
"valid_targets_mean": 5046.4,
|
|
"valid_targets_min": 1190
|
|
},
|
|
{
|
|
"epoch": 3.3865814696485623,
|
|
"grad_norm": 0.4266548230980135,
|
|
"learning_rate": 2.4603415623370387e-05,
|
|
"loss": 0.1931,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1820763200521469,
|
|
"step": 2120,
|
|
"valid_targets_mean": 5194.9,
|
|
"valid_targets_min": 896
|
|
},
|
|
{
|
|
"epoch": 3.3945686900958467,
|
|
"grad_norm": 0.45809211730759314,
|
|
"learning_rate": 2.4525843370630697e-05,
|
|
"loss": 0.1843,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17680346965789795,
|
|
"step": 2125,
|
|
"valid_targets_mean": 4570.6,
|
|
"valid_targets_min": 792
|
|
},
|
|
{
|
|
"epoch": 3.402555910543131,
|
|
"grad_norm": 0.6772631764618228,
|
|
"learning_rate": 2.4448199291318058e-05,
|
|
"loss": 0.1922,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1788870245218277,
|
|
"step": 2130,
|
|
"valid_targets_mean": 5122.0,
|
|
"valid_targets_min": 2157
|
|
},
|
|
{
|
|
"epoch": 3.4105431309904155,
|
|
"grad_norm": 0.39859839161760713,
|
|
"learning_rate": 2.4370484617668707e-05,
|
|
"loss": 0.1759,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1690652221441269,
|
|
"step": 2135,
|
|
"valid_targets_mean": 5479.8,
|
|
"valid_targets_min": 1940
|
|
},
|
|
{
|
|
"epoch": 3.4185303514377,
|
|
"grad_norm": 0.6639191506771926,
|
|
"learning_rate": 2.4292700583039194e-05,
|
|
"loss": 0.1962,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1719520092010498,
|
|
"step": 2140,
|
|
"valid_targets_mean": 4791.0,
|
|
"valid_targets_min": 490
|
|
},
|
|
{
|
|
"epoch": 3.426517571884984,
|
|
"grad_norm": 0.42668237867151115,
|
|
"learning_rate": 2.4214848421886893e-05,
|
|
"loss": 0.1814,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18518546223640442,
|
|
"step": 2145,
|
|
"valid_targets_mean": 5195.2,
|
|
"valid_targets_min": 1095
|
|
},
|
|
{
|
|
"epoch": 3.4345047923322682,
|
|
"grad_norm": 0.6674059346938613,
|
|
"learning_rate": 2.4136929369750344e-05,
|
|
"loss": 0.2049,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19148018956184387,
|
|
"step": 2150,
|
|
"valid_targets_mean": 4607.1,
|
|
"valid_targets_min": 1097
|
|
},
|
|
{
|
|
"epoch": 3.4424920127795526,
|
|
"grad_norm": 0.44213355246762426,
|
|
"learning_rate": 2.4058944663229672e-05,
|
|
"loss": 0.1841,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19993336498737335,
|
|
"step": 2155,
|
|
"valid_targets_mean": 4826.8,
|
|
"valid_targets_min": 1777
|
|
},
|
|
{
|
|
"epoch": 3.450479233226837,
|
|
"grad_norm": 0.45019819902087,
|
|
"learning_rate": 2.398089553996696e-05,
|
|
"loss": 0.1899,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18793454766273499,
|
|
"step": 2160,
|
|
"valid_targets_mean": 5297.8,
|
|
"valid_targets_min": 2098
|
|
},
|
|
{
|
|
"epoch": 3.4584664536741214,
|
|
"grad_norm": 0.4694883248667954,
|
|
"learning_rate": 2.3902783238626604e-05,
|
|
"loss": 0.1952,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18550071120262146,
|
|
"step": 2165,
|
|
"valid_targets_mean": 4982.4,
|
|
"valid_targets_min": 2397
|
|
},
|
|
{
|
|
"epoch": 3.466453674121406,
|
|
"grad_norm": 0.5087812268690848,
|
|
"learning_rate": 2.3824608998875652e-05,
|
|
"loss": 0.1935,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16924448311328888,
|
|
"step": 2170,
|
|
"valid_targets_mean": 4626.5,
|
|
"valid_targets_min": 813
|
|
},
|
|
{
|
|
"epoch": 3.47444089456869,
|
|
"grad_norm": 0.4032588568241463,
|
|
"learning_rate": 2.3746374061364138e-05,
|
|
"loss": 0.1874,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15398786962032318,
|
|
"step": 2175,
|
|
"valid_targets_mean": 4960.7,
|
|
"valid_targets_min": 1709
|
|
},
|
|
{
|
|
"epoch": 3.4824281150159746,
|
|
"grad_norm": 0.4604507803407548,
|
|
"learning_rate": 2.3668079667705404e-05,
|
|
"loss": 0.1996,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19481347501277924,
|
|
"step": 2180,
|
|
"valid_targets_mean": 4630.8,
|
|
"valid_targets_min": 1025
|
|
},
|
|
{
|
|
"epoch": 3.4904153354632586,
|
|
"grad_norm": 0.46373238726464927,
|
|
"learning_rate": 2.3589727060456342e-05,
|
|
"loss": 0.1847,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2064315527677536,
|
|
"step": 2185,
|
|
"valid_targets_mean": 4385.4,
|
|
"valid_targets_min": 367
|
|
},
|
|
{
|
|
"epoch": 3.498402555910543,
|
|
"grad_norm": 0.5185491121751089,
|
|
"learning_rate": 2.3511317483097753e-05,
|
|
"loss": 0.1854,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17522519826889038,
|
|
"step": 2190,
|
|
"valid_targets_mean": 4921.1,
|
|
"valid_targets_min": 543
|
|
},
|
|
{
|
|
"epoch": 3.5063897763578273,
|
|
"grad_norm": 0.47737053459541356,
|
|
"learning_rate": 2.3432852180014548e-05,
|
|
"loss": 0.2012,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19549396634101868,
|
|
"step": 2195,
|
|
"valid_targets_mean": 5227.9,
|
|
"valid_targets_min": 1785
|
|
},
|
|
{
|
|
"epoch": 3.5143769968051117,
|
|
"grad_norm": 0.9355774311363018,
|
|
"learning_rate": 2.3354332396476026e-05,
|
|
"loss": 0.1885,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17945951223373413,
|
|
"step": 2200,
|
|
"valid_targets_mean": 4325.2,
|
|
"valid_targets_min": 625
|
|
},
|
|
{
|
|
"epoch": 3.522364217252396,
|
|
"grad_norm": 0.40783334210073574,
|
|
"learning_rate": 2.327575937861612e-05,
|
|
"loss": 0.1773,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16562104225158691,
|
|
"step": 2205,
|
|
"valid_targets_mean": 5723.5,
|
|
"valid_targets_min": 1929
|
|
},
|
|
{
|
|
"epoch": 3.5303514376996805,
|
|
"grad_norm": 0.4527972615076226,
|
|
"learning_rate": 2.3197134373413595e-05,
|
|
"loss": 0.1892,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22815603017807007,
|
|
"step": 2210,
|
|
"valid_targets_mean": 5091.8,
|
|
"valid_targets_min": 838
|
|
},
|
|
{
|
|
"epoch": 3.538338658146965,
|
|
"grad_norm": 0.47388983182379485,
|
|
"learning_rate": 2.3118458628672272e-05,
|
|
"loss": 0.1917,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18909385800361633,
|
|
"step": 2215,
|
|
"valid_targets_mean": 4816.7,
|
|
"valid_targets_min": 1347
|
|
},
|
|
{
|
|
"epoch": 3.5463258785942493,
|
|
"grad_norm": 0.4250838131885362,
|
|
"learning_rate": 2.3039733393001233e-05,
|
|
"loss": 0.1904,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2078056037425995,
|
|
"step": 2220,
|
|
"valid_targets_mean": 6035.9,
|
|
"valid_targets_min": 1642
|
|
},
|
|
{
|
|
"epoch": 3.5543130990415337,
|
|
"grad_norm": 0.4913873201029617,
|
|
"learning_rate": 2.2960959915794998e-05,
|
|
"loss": 0.1842,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19340744614601135,
|
|
"step": 2225,
|
|
"valid_targets_mean": 4638.5,
|
|
"valid_targets_min": 643
|
|
},
|
|
{
|
|
"epoch": 3.562300319488818,
|
|
"grad_norm": 0.40839037061724476,
|
|
"learning_rate": 2.2882139447213673e-05,
|
|
"loss": 0.1782,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16877365112304688,
|
|
"step": 2230,
|
|
"valid_targets_mean": 5349.2,
|
|
"valid_targets_min": 1593
|
|
},
|
|
{
|
|
"epoch": 3.5702875399361025,
|
|
"grad_norm": 0.4675729076472275,
|
|
"learning_rate": 2.280327323816317e-05,
|
|
"loss": 0.2037,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.223651722073555,
|
|
"step": 2235,
|
|
"valid_targets_mean": 4993.3,
|
|
"valid_targets_min": 1785
|
|
},
|
|
{
|
|
"epoch": 3.5782747603833864,
|
|
"grad_norm": 0.5048609910600323,
|
|
"learning_rate": 2.2724362540275288e-05,
|
|
"loss": 0.1899,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19978858530521393,
|
|
"step": 2240,
|
|
"valid_targets_mean": 3995.2,
|
|
"valid_targets_min": 1343
|
|
},
|
|
{
|
|
"epoch": 3.586261980830671,
|
|
"grad_norm": 1.0705437953436956,
|
|
"learning_rate": 2.2645408605887882e-05,
|
|
"loss": 0.2024,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2068878561258316,
|
|
"step": 2245,
|
|
"valid_targets_mean": 4488.3,
|
|
"valid_targets_min": 860
|
|
},
|
|
{
|
|
"epoch": 3.594249201277955,
|
|
"grad_norm": 0.46081975658418306,
|
|
"learning_rate": 2.2566412688025003e-05,
|
|
"loss": 0.1976,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17880478501319885,
|
|
"step": 2250,
|
|
"valid_targets_mean": 4766.0,
|
|
"valid_targets_min": 1981
|
|
},
|
|
{
|
|
"epoch": 3.6022364217252396,
|
|
"grad_norm": 0.49359472491199236,
|
|
"learning_rate": 2.2487376040376958e-05,
|
|
"loss": 0.1949,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1768798828125,
|
|
"step": 2255,
|
|
"valid_targets_mean": 3792.6,
|
|
"valid_targets_min": 1838
|
|
},
|
|
{
|
|
"epoch": 3.610223642172524,
|
|
"grad_norm": 0.4522379907468209,
|
|
"learning_rate": 2.240829991728048e-05,
|
|
"loss": 0.1927,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18313640356063843,
|
|
"step": 2260,
|
|
"valid_targets_mean": 5166.9,
|
|
"valid_targets_min": 1360
|
|
},
|
|
{
|
|
"epoch": 3.6182108626198084,
|
|
"grad_norm": 0.4840769217866764,
|
|
"learning_rate": 2.2329185573698778e-05,
|
|
"loss": 0.196,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18533888459205627,
|
|
"step": 2265,
|
|
"valid_targets_mean": 4982.4,
|
|
"valid_targets_min": 358
|
|
},
|
|
{
|
|
"epoch": 3.626198083067093,
|
|
"grad_norm": 0.4336567815547825,
|
|
"learning_rate": 2.225003426520163e-05,
|
|
"loss": 0.1953,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18483304977416992,
|
|
"step": 2270,
|
|
"valid_targets_mean": 6042.6,
|
|
"valid_targets_min": 2980
|
|
},
|
|
{
|
|
"epoch": 3.6341853035143767,
|
|
"grad_norm": 0.4814668528965902,
|
|
"learning_rate": 2.2170847247945466e-05,
|
|
"loss": 0.1779,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18856918811798096,
|
|
"step": 2275,
|
|
"valid_targets_mean": 4287.3,
|
|
"valid_targets_min": 2095
|
|
},
|
|
{
|
|
"epoch": 3.642172523961661,
|
|
"grad_norm": 0.43225765042897346,
|
|
"learning_rate": 2.209162577865342e-05,
|
|
"loss": 0.1866,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21974483132362366,
|
|
"step": 2280,
|
|
"valid_targets_mean": 5710.0,
|
|
"valid_targets_min": 772
|
|
},
|
|
{
|
|
"epoch": 3.6501597444089455,
|
|
"grad_norm": 0.5391791704251936,
|
|
"learning_rate": 2.201237111459539e-05,
|
|
"loss": 0.1989,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19698813557624817,
|
|
"step": 2285,
|
|
"valid_targets_mean": 4624.1,
|
|
"valid_targets_min": 707
|
|
},
|
|
{
|
|
"epoch": 3.65814696485623,
|
|
"grad_norm": 0.46099188882052544,
|
|
"learning_rate": 2.193308451356809e-05,
|
|
"loss": 0.1795,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18069559335708618,
|
|
"step": 2290,
|
|
"valid_targets_mean": 4457.1,
|
|
"valid_targets_min": 829
|
|
},
|
|
{
|
|
"epoch": 3.6661341853035143,
|
|
"grad_norm": 0.4316480786621061,
|
|
"learning_rate": 2.1853767233875084e-05,
|
|
"loss": 0.1834,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2013440579175949,
|
|
"step": 2295,
|
|
"valid_targets_mean": 4825.4,
|
|
"valid_targets_min": 931
|
|
},
|
|
{
|
|
"epoch": 3.6741214057507987,
|
|
"grad_norm": 0.4305778870477568,
|
|
"learning_rate": 2.1774420534306812e-05,
|
|
"loss": 0.1941,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17159786820411682,
|
|
"step": 2300,
|
|
"valid_targets_mean": 5305.0,
|
|
"valid_targets_min": 2340
|
|
},
|
|
{
|
|
"epoch": 3.682108626198083,
|
|
"grad_norm": 0.4675706015705595,
|
|
"learning_rate": 2.1695045674120615e-05,
|
|
"loss": 0.1953,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1822780966758728,
|
|
"step": 2305,
|
|
"valid_targets_mean": 4895.9,
|
|
"valid_targets_min": 1010
|
|
},
|
|
{
|
|
"epoch": 3.6900958466453675,
|
|
"grad_norm": 0.49409976963921004,
|
|
"learning_rate": 2.1615643913020755e-05,
|
|
"loss": 0.1981,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18701043725013733,
|
|
"step": 2310,
|
|
"valid_targets_mean": 4383.5,
|
|
"valid_targets_min": 601
|
|
},
|
|
{
|
|
"epoch": 3.698083067092652,
|
|
"grad_norm": 0.4247423680969255,
|
|
"learning_rate": 2.1536216511138423e-05,
|
|
"loss": 0.1701,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15091648697853088,
|
|
"step": 2315,
|
|
"valid_targets_mean": 4364.2,
|
|
"valid_targets_min": 663
|
|
},
|
|
{
|
|
"epoch": 3.7060702875399363,
|
|
"grad_norm": 0.4152343558887996,
|
|
"learning_rate": 2.145676472901174e-05,
|
|
"loss": 0.185,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18406620621681213,
|
|
"step": 2320,
|
|
"valid_targets_mean": 5993.6,
|
|
"valid_targets_min": 936
|
|
},
|
|
{
|
|
"epoch": 3.7140575079872207,
|
|
"grad_norm": 0.4548151227454303,
|
|
"learning_rate": 2.1377289827565734e-05,
|
|
"loss": 0.1875,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19774073362350464,
|
|
"step": 2325,
|
|
"valid_targets_mean": 4848.2,
|
|
"valid_targets_min": 1688
|
|
},
|
|
{
|
|
"epoch": 3.722044728434505,
|
|
"grad_norm": 0.5024804371842532,
|
|
"learning_rate": 2.129779306809236e-05,
|
|
"loss": 0.1845,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.183456152677536,
|
|
"step": 2330,
|
|
"valid_targets_mean": 3705.8,
|
|
"valid_targets_min": 806
|
|
},
|
|
{
|
|
"epoch": 3.730031948881789,
|
|
"grad_norm": 0.4418380559472342,
|
|
"learning_rate": 2.121827571223046e-05,
|
|
"loss": 0.1948,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18899336457252502,
|
|
"step": 2335,
|
|
"valid_targets_mean": 5399.1,
|
|
"valid_targets_min": 899
|
|
},
|
|
{
|
|
"epoch": 3.7380191693290734,
|
|
"grad_norm": 0.5754123014682926,
|
|
"learning_rate": 2.113873902194575e-05,
|
|
"loss": 0.1858,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1759827435016632,
|
|
"step": 2340,
|
|
"valid_targets_mean": 4815.8,
|
|
"valid_targets_min": 1535
|
|
},
|
|
{
|
|
"epoch": 3.746006389776358,
|
|
"grad_norm": 0.5185080250250155,
|
|
"learning_rate": 2.1059184259510786e-05,
|
|
"loss": 0.1813,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17565274238586426,
|
|
"step": 2345,
|
|
"valid_targets_mean": 4331.1,
|
|
"valid_targets_min": 620
|
|
},
|
|
{
|
|
"epoch": 3.753993610223642,
|
|
"grad_norm": 0.4877335595382726,
|
|
"learning_rate": 2.0979612687484935e-05,
|
|
"loss": 0.1786,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18762779235839844,
|
|
"step": 2350,
|
|
"valid_targets_mean": 3845.9,
|
|
"valid_targets_min": 1194
|
|
},
|
|
{
|
|
"epoch": 3.7619808306709266,
|
|
"grad_norm": 0.43410975598882556,
|
|
"learning_rate": 2.0900025568694334e-05,
|
|
"loss": 0.1869,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15621888637542725,
|
|
"step": 2355,
|
|
"valid_targets_mean": 4890.7,
|
|
"valid_targets_min": 1382
|
|
},
|
|
{
|
|
"epoch": 3.769968051118211,
|
|
"grad_norm": 0.48305637781442534,
|
|
"learning_rate": 2.082042416621187e-05,
|
|
"loss": 0.2007,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.210737407207489,
|
|
"step": 2360,
|
|
"valid_targets_mean": 4516.4,
|
|
"valid_targets_min": 862
|
|
},
|
|
{
|
|
"epoch": 3.777955271565495,
|
|
"grad_norm": 0.5022806770180145,
|
|
"learning_rate": 2.0740809743337095e-05,
|
|
"loss": 0.191,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18494747579097748,
|
|
"step": 2365,
|
|
"valid_targets_mean": 4611.1,
|
|
"valid_targets_min": 680
|
|
},
|
|
{
|
|
"epoch": 3.7859424920127793,
|
|
"grad_norm": 0.5581835227268139,
|
|
"learning_rate": 2.0661183563576202e-05,
|
|
"loss": 0.192,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2176152467727661,
|
|
"step": 2370,
|
|
"valid_targets_mean": 4084.2,
|
|
"valid_targets_min": 320
|
|
},
|
|
{
|
|
"epoch": 3.7939297124600637,
|
|
"grad_norm": 0.5241288006020112,
|
|
"learning_rate": 2.0581546890621986e-05,
|
|
"loss": 0.1854,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18710482120513916,
|
|
"step": 2375,
|
|
"valid_targets_mean": 4911.0,
|
|
"valid_targets_min": 732
|
|
},
|
|
{
|
|
"epoch": 3.801916932907348,
|
|
"grad_norm": 0.43506359784608306,
|
|
"learning_rate": 2.0501900988333755e-05,
|
|
"loss": 0.1759,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17661073803901672,
|
|
"step": 2380,
|
|
"valid_targets_mean": 5269.9,
|
|
"valid_targets_min": 1674
|
|
},
|
|
{
|
|
"epoch": 3.8099041533546325,
|
|
"grad_norm": 0.47473570226755557,
|
|
"learning_rate": 2.0422247120717304e-05,
|
|
"loss": 0.1925,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19318652153015137,
|
|
"step": 2385,
|
|
"valid_targets_mean": 4951.1,
|
|
"valid_targets_min": 1510
|
|
},
|
|
{
|
|
"epoch": 3.817891373801917,
|
|
"grad_norm": 0.50739056377129,
|
|
"learning_rate": 2.0342586551904824e-05,
|
|
"loss": 0.2081,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20048654079437256,
|
|
"step": 2390,
|
|
"valid_targets_mean": 4932.8,
|
|
"valid_targets_min": 1456
|
|
},
|
|
{
|
|
"epoch": 3.8258785942492013,
|
|
"grad_norm": 0.5076070725831601,
|
|
"learning_rate": 2.026292054613486e-05,
|
|
"loss": 0.1941,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21501141786575317,
|
|
"step": 2395,
|
|
"valid_targets_mean": 4507.7,
|
|
"valid_targets_min": 617
|
|
},
|
|
{
|
|
"epoch": 3.8338658146964857,
|
|
"grad_norm": 0.5532415716131805,
|
|
"learning_rate": 2.0183250367732253e-05,
|
|
"loss": 0.1949,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20856130123138428,
|
|
"step": 2400,
|
|
"valid_targets_mean": 4624.4,
|
|
"valid_targets_min": 682
|
|
},
|
|
{
|
|
"epoch": 3.84185303514377,
|
|
"grad_norm": 0.45671458342233145,
|
|
"learning_rate": 2.0103577281088065e-05,
|
|
"loss": 0.184,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18179446458816528,
|
|
"step": 2405,
|
|
"valid_targets_mean": 4898.2,
|
|
"valid_targets_min": 1537
|
|
},
|
|
{
|
|
"epoch": 3.8498402555910545,
|
|
"grad_norm": 0.4759919873634848,
|
|
"learning_rate": 2.00239025506395e-05,
|
|
"loss": 0.1992,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21135717630386353,
|
|
"step": 2410,
|
|
"valid_targets_mean": 5055.8,
|
|
"valid_targets_min": 1394
|
|
},
|
|
{
|
|
"epoch": 3.857827476038339,
|
|
"grad_norm": 0.46319188180737886,
|
|
"learning_rate": 1.9944227440849856e-05,
|
|
"loss": 0.1801,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16708174347877502,
|
|
"step": 2415,
|
|
"valid_targets_mean": 4287.0,
|
|
"valid_targets_min": 1070
|
|
},
|
|
{
|
|
"epoch": 3.8658146964856233,
|
|
"grad_norm": 0.5031704543649632,
|
|
"learning_rate": 1.9864553216188454e-05,
|
|
"loss": 0.1878,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21082769334316254,
|
|
"step": 2420,
|
|
"valid_targets_mean": 4561.1,
|
|
"valid_targets_min": 996
|
|
},
|
|
{
|
|
"epoch": 3.873801916932907,
|
|
"grad_norm": 0.5225932563691561,
|
|
"learning_rate": 1.978488114111057e-05,
|
|
"loss": 0.1711,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18447518348693848,
|
|
"step": 2425,
|
|
"valid_targets_mean": 4778.2,
|
|
"valid_targets_min": 1629
|
|
},
|
|
{
|
|
"epoch": 3.8817891373801916,
|
|
"grad_norm": 0.4478785968060911,
|
|
"learning_rate": 1.970521248003735e-05,
|
|
"loss": 0.1719,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18033427000045776,
|
|
"step": 2430,
|
|
"valid_targets_mean": 5120.3,
|
|
"valid_targets_min": 721
|
|
},
|
|
{
|
|
"epoch": 3.889776357827476,
|
|
"grad_norm": 0.44839878262804944,
|
|
"learning_rate": 1.9625548497335783e-05,
|
|
"loss": 0.1894,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2015233039855957,
|
|
"step": 2435,
|
|
"valid_targets_mean": 5338.6,
|
|
"valid_targets_min": 2090
|
|
},
|
|
{
|
|
"epoch": 3.8977635782747604,
|
|
"grad_norm": 0.5033175049463399,
|
|
"learning_rate": 1.9545890457298592e-05,
|
|
"loss": 0.1763,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17985785007476807,
|
|
"step": 2440,
|
|
"valid_targets_mean": 4605.1,
|
|
"valid_targets_min": 854
|
|
},
|
|
{
|
|
"epoch": 3.905750798722045,
|
|
"grad_norm": 0.6939870310820266,
|
|
"learning_rate": 1.9466239624124203e-05,
|
|
"loss": 0.1811,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19669967889785767,
|
|
"step": 2445,
|
|
"valid_targets_mean": 4070.8,
|
|
"valid_targets_min": 933
|
|
},
|
|
{
|
|
"epoch": 3.913738019169329,
|
|
"grad_norm": 0.43504461484987295,
|
|
"learning_rate": 1.938659726189663e-05,
|
|
"loss": 0.1897,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20107465982437134,
|
|
"step": 2450,
|
|
"valid_targets_mean": 5109.3,
|
|
"valid_targets_min": 1514
|
|
},
|
|
{
|
|
"epoch": 3.9217252396166136,
|
|
"grad_norm": 0.49934803104249825,
|
|
"learning_rate": 1.9306964634565505e-05,
|
|
"loss": 0.1857,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19940102100372314,
|
|
"step": 2455,
|
|
"valid_targets_mean": 5095.8,
|
|
"valid_targets_min": 1756
|
|
},
|
|
{
|
|
"epoch": 3.9297124600638975,
|
|
"grad_norm": 0.3953376883568536,
|
|
"learning_rate": 1.9227343005925926e-05,
|
|
"loss": 0.1871,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17035652697086334,
|
|
"step": 2460,
|
|
"valid_targets_mean": 5663.9,
|
|
"valid_targets_min": 1635
|
|
},
|
|
{
|
|
"epoch": 3.937699680511182,
|
|
"grad_norm": 0.4369568213838696,
|
|
"learning_rate": 1.914773363959845e-05,
|
|
"loss": 0.1808,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19834986329078674,
|
|
"step": 2465,
|
|
"valid_targets_mean": 5365.0,
|
|
"valid_targets_min": 1859
|
|
},
|
|
{
|
|
"epoch": 3.9456869009584663,
|
|
"grad_norm": 0.5030102742032438,
|
|
"learning_rate": 1.906813779900903e-05,
|
|
"loss": 0.1863,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1819533109664917,
|
|
"step": 2470,
|
|
"valid_targets_mean": 4127.9,
|
|
"valid_targets_min": 1521
|
|
},
|
|
{
|
|
"epoch": 3.9536741214057507,
|
|
"grad_norm": 0.4731427364779601,
|
|
"learning_rate": 1.8988556747368953e-05,
|
|
"loss": 0.1894,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17888499796390533,
|
|
"step": 2475,
|
|
"valid_targets_mean": 4367.8,
|
|
"valid_targets_min": 785
|
|
},
|
|
{
|
|
"epoch": 3.961661341853035,
|
|
"grad_norm": 0.5557284048366862,
|
|
"learning_rate": 1.890899174765481e-05,
|
|
"loss": 0.1837,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18574580550193787,
|
|
"step": 2480,
|
|
"valid_targets_mean": 4279.9,
|
|
"valid_targets_min": 1718
|
|
},
|
|
{
|
|
"epoch": 3.9696485623003195,
|
|
"grad_norm": 0.47991964027385436,
|
|
"learning_rate": 1.8829444062588433e-05,
|
|
"loss": 0.1917,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19404660165309906,
|
|
"step": 2485,
|
|
"valid_targets_mean": 4322.4,
|
|
"valid_targets_min": 857
|
|
},
|
|
{
|
|
"epoch": 3.977635782747604,
|
|
"grad_norm": 0.42202690145985944,
|
|
"learning_rate": 1.8749914954616882e-05,
|
|
"loss": 0.1762,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14783018827438354,
|
|
"step": 2490,
|
|
"valid_targets_mean": 5029.9,
|
|
"valid_targets_min": 1786
|
|
},
|
|
{
|
|
"epoch": 3.9856230031948883,
|
|
"grad_norm": 0.4410364424689328,
|
|
"learning_rate": 1.867040568589236e-05,
|
|
"loss": 0.1844,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15357394516468048,
|
|
"step": 2495,
|
|
"valid_targets_mean": 4057.6,
|
|
"valid_targets_min": 1562
|
|
},
|
|
{
|
|
"epoch": 3.9936102236421727,
|
|
"grad_norm": 0.5197355081404011,
|
|
"learning_rate": 1.859091751825225e-05,
|
|
"loss": 0.1952,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20624461770057678,
|
|
"step": 2500,
|
|
"valid_targets_mean": 4384.0,
|
|
"valid_targets_min": 611
|
|
},
|
|
{
|
|
"epoch": 4.001597444089457,
|
|
"grad_norm": 0.4699014479471355,
|
|
"learning_rate": 1.8511451713199038e-05,
|
|
"loss": 0.1917,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20383623242378235,
|
|
"step": 2505,
|
|
"valid_targets_mean": 4910.9,
|
|
"valid_targets_min": 1243
|
|
},
|
|
{
|
|
"epoch": 4.0095846645367414,
|
|
"grad_norm": 0.44610459030943744,
|
|
"learning_rate": 1.843200953188031e-05,
|
|
"loss": 0.1767,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1724563091993332,
|
|
"step": 2510,
|
|
"valid_targets_mean": 4851.7,
|
|
"valid_targets_min": 1271
|
|
},
|
|
{
|
|
"epoch": 4.017571884984026,
|
|
"grad_norm": 0.5342425830230646,
|
|
"learning_rate": 1.8352592235068735e-05,
|
|
"loss": 0.1632,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17269504070281982,
|
|
"step": 2515,
|
|
"valid_targets_mean": 4625.4,
|
|
"valid_targets_min": 1260
|
|
},
|
|
{
|
|
"epoch": 4.02555910543131,
|
|
"grad_norm": 0.4544261008139438,
|
|
"learning_rate": 1.8273201083142056e-05,
|
|
"loss": 0.18,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15527205169200897,
|
|
"step": 2520,
|
|
"valid_targets_mean": 4469.7,
|
|
"valid_targets_min": 770
|
|
},
|
|
{
|
|
"epoch": 4.033546325878595,
|
|
"grad_norm": 0.45777968965714255,
|
|
"learning_rate": 1.8193837336063096e-05,
|
|
"loss": 0.1659,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18351805210113525,
|
|
"step": 2525,
|
|
"valid_targets_mean": 5344.1,
|
|
"valid_targets_min": 1393
|
|
},
|
|
{
|
|
"epoch": 4.041533546325879,
|
|
"grad_norm": 0.48743720071330254,
|
|
"learning_rate": 1.8114502253359745e-05,
|
|
"loss": 0.1678,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15608066320419312,
|
|
"step": 2530,
|
|
"valid_targets_mean": 4386.8,
|
|
"valid_targets_min": 806
|
|
},
|
|
{
|
|
"epoch": 4.0495207667731625,
|
|
"grad_norm": 0.5102889380748086,
|
|
"learning_rate": 1.8035197094104986e-05,
|
|
"loss": 0.1627,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16816166043281555,
|
|
"step": 2535,
|
|
"valid_targets_mean": 4606.9,
|
|
"valid_targets_min": 1176
|
|
},
|
|
{
|
|
"epoch": 4.057507987220447,
|
|
"grad_norm": 0.49384781849437476,
|
|
"learning_rate": 1.7955923116896908e-05,
|
|
"loss": 0.183,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1960548460483551,
|
|
"step": 2540,
|
|
"valid_targets_mean": 4988.2,
|
|
"valid_targets_min": 1467
|
|
},
|
|
{
|
|
"epoch": 4.065495207667731,
|
|
"grad_norm": 0.47524965354500476,
|
|
"learning_rate": 1.787668157983872e-05,
|
|
"loss": 0.1804,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16807019710540771,
|
|
"step": 2545,
|
|
"valid_targets_mean": 5086.1,
|
|
"valid_targets_min": 667
|
|
},
|
|
{
|
|
"epoch": 4.073482428115016,
|
|
"grad_norm": 0.4840983111087558,
|
|
"learning_rate": 1.779747374051881e-05,
|
|
"loss": 0.1824,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.159114807844162,
|
|
"step": 2550,
|
|
"valid_targets_mean": 4995.7,
|
|
"valid_targets_min": 1213
|
|
},
|
|
{
|
|
"epoch": 4.0814696485623,
|
|
"grad_norm": 0.45450548714319483,
|
|
"learning_rate": 1.7718300855990767e-05,
|
|
"loss": 0.1808,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19937041401863098,
|
|
"step": 2555,
|
|
"valid_targets_mean": 5728.3,
|
|
"valid_targets_min": 1841
|
|
},
|
|
{
|
|
"epoch": 4.0894568690095845,
|
|
"grad_norm": 0.45562843993445606,
|
|
"learning_rate": 1.7639164182753434e-05,
|
|
"loss": 0.1732,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1393558382987976,
|
|
"step": 2560,
|
|
"valid_targets_mean": 4458.7,
|
|
"valid_targets_min": 1285
|
|
},
|
|
{
|
|
"epoch": 4.097444089456869,
|
|
"grad_norm": 0.542728331932225,
|
|
"learning_rate": 1.7560064976730967e-05,
|
|
"loss": 0.1805,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16445447504520416,
|
|
"step": 2565,
|
|
"valid_targets_mean": 4540.4,
|
|
"valid_targets_min": 1036
|
|
},
|
|
{
|
|
"epoch": 4.105431309904153,
|
|
"grad_norm": 0.44164521814861474,
|
|
"learning_rate": 1.7481004493252915e-05,
|
|
"loss": 0.183,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17348837852478027,
|
|
"step": 2570,
|
|
"valid_targets_mean": 5027.4,
|
|
"valid_targets_min": 917
|
|
},
|
|
{
|
|
"epoch": 4.113418530351438,
|
|
"grad_norm": 0.4759337103330901,
|
|
"learning_rate": 1.740198398703428e-05,
|
|
"loss": 0.1878,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19265320897102356,
|
|
"step": 2575,
|
|
"valid_targets_mean": 5355.0,
|
|
"valid_targets_min": 1435
|
|
},
|
|
{
|
|
"epoch": 4.121405750798722,
|
|
"grad_norm": 0.505547502504362,
|
|
"learning_rate": 1.7323004712155613e-05,
|
|
"loss": 0.1632,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.152913436293602,
|
|
"step": 2580,
|
|
"valid_targets_mean": 5178.0,
|
|
"valid_targets_min": 996
|
|
},
|
|
{
|
|
"epoch": 4.1293929712460065,
|
|
"grad_norm": 0.48447911256072,
|
|
"learning_rate": 1.724406792204312e-05,
|
|
"loss": 0.1673,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18921968340873718,
|
|
"step": 2585,
|
|
"valid_targets_mean": 5014.1,
|
|
"valid_targets_min": 1543
|
|
},
|
|
{
|
|
"epoch": 4.137380191693291,
|
|
"grad_norm": 0.5964352002018138,
|
|
"learning_rate": 1.7165174869448738e-05,
|
|
"loss": 0.1837,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19785529375076294,
|
|
"step": 2590,
|
|
"valid_targets_mean": 4668.9,
|
|
"valid_targets_min": 2454
|
|
},
|
|
{
|
|
"epoch": 4.145367412140575,
|
|
"grad_norm": 0.5362905258520115,
|
|
"learning_rate": 1.7086326806430298e-05,
|
|
"loss": 0.1779,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19170859456062317,
|
|
"step": 2595,
|
|
"valid_targets_mean": 4329.1,
|
|
"valid_targets_min": 1001
|
|
},
|
|
{
|
|
"epoch": 4.15335463258786,
|
|
"grad_norm": 0.5063383812064016,
|
|
"learning_rate": 1.7007524984331623e-05,
|
|
"loss": 0.1617,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14627587795257568,
|
|
"step": 2600,
|
|
"valid_targets_mean": 4513.1,
|
|
"valid_targets_min": 1287
|
|
},
|
|
{
|
|
"epoch": 4.161341853035144,
|
|
"grad_norm": 0.4571185442372499,
|
|
"learning_rate": 1.692877065376268e-05,
|
|
"loss": 0.1794,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20067718625068665,
|
|
"step": 2605,
|
|
"valid_targets_mean": 5342.1,
|
|
"valid_targets_min": 1884
|
|
},
|
|
{
|
|
"epoch": 4.169329073482428,
|
|
"grad_norm": 0.5531440170736265,
|
|
"learning_rate": 1.6850065064579724e-05,
|
|
"loss": 0.1653,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1775854527950287,
|
|
"step": 2610,
|
|
"valid_targets_mean": 3792.1,
|
|
"valid_targets_min": 2240
|
|
},
|
|
{
|
|
"epoch": 4.177316293929713,
|
|
"grad_norm": 0.4707463377254241,
|
|
"learning_rate": 1.677140946586548e-05,
|
|
"loss": 0.1789,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1646420657634735,
|
|
"step": 2615,
|
|
"valid_targets_mean": 4516.5,
|
|
"valid_targets_min": 721
|
|
},
|
|
{
|
|
"epoch": 4.185303514376997,
|
|
"grad_norm": 0.5068778613138127,
|
|
"learning_rate": 1.66928051059093e-05,
|
|
"loss": 0.1793,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19891464710235596,
|
|
"step": 2620,
|
|
"valid_targets_mean": 4863.6,
|
|
"valid_targets_min": 2881
|
|
},
|
|
{
|
|
"epoch": 4.193290734824281,
|
|
"grad_norm": 0.48265152952562435,
|
|
"learning_rate": 1.6614253232187363e-05,
|
|
"loss": 0.1719,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21110501885414124,
|
|
"step": 2625,
|
|
"valid_targets_mean": 5051.4,
|
|
"valid_targets_min": 912
|
|
},
|
|
{
|
|
"epoch": 4.201277955271565,
|
|
"grad_norm": 0.4701304397256498,
|
|
"learning_rate": 1.6535755091342876e-05,
|
|
"loss": 0.1743,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16952627897262573,
|
|
"step": 2630,
|
|
"valid_targets_mean": 4840.1,
|
|
"valid_targets_min": 1343
|
|
},
|
|
{
|
|
"epoch": 4.2092651757188495,
|
|
"grad_norm": 0.47877634257198126,
|
|
"learning_rate": 1.6457311929166286e-05,
|
|
"loss": 0.1622,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1714383214712143,
|
|
"step": 2635,
|
|
"valid_targets_mean": 4488.4,
|
|
"valid_targets_min": 657
|
|
},
|
|
{
|
|
"epoch": 4.217252396166134,
|
|
"grad_norm": 0.4932893039114935,
|
|
"learning_rate": 1.637892499057551e-05,
|
|
"loss": 0.1878,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17138069868087769,
|
|
"step": 2640,
|
|
"valid_targets_mean": 4592.1,
|
|
"valid_targets_min": 819
|
|
},
|
|
{
|
|
"epoch": 4.225239616613418,
|
|
"grad_norm": 0.43679655489643965,
|
|
"learning_rate": 1.6300595519596177e-05,
|
|
"loss": 0.1628,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15296795964241028,
|
|
"step": 2645,
|
|
"valid_targets_mean": 4857.1,
|
|
"valid_targets_min": 966
|
|
},
|
|
{
|
|
"epoch": 4.233226837060703,
|
|
"grad_norm": 0.6163573944435584,
|
|
"learning_rate": 1.6222324759341897e-05,
|
|
"loss": 0.1723,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17561312019824982,
|
|
"step": 2650,
|
|
"valid_targets_mean": 4857.5,
|
|
"valid_targets_min": 813
|
|
},
|
|
{
|
|
"epoch": 4.241214057507987,
|
|
"grad_norm": 0.521141354455912,
|
|
"learning_rate": 1.6144113951994515e-05,
|
|
"loss": 0.1751,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.145559623837471,
|
|
"step": 2655,
|
|
"valid_targets_mean": 3869.6,
|
|
"valid_targets_min": 561
|
|
},
|
|
{
|
|
"epoch": 4.2492012779552715,
|
|
"grad_norm": 0.4466498466808147,
|
|
"learning_rate": 1.60659643387844e-05,
|
|
"loss": 0.162,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15559345483779907,
|
|
"step": 2660,
|
|
"valid_targets_mean": 5260.7,
|
|
"valid_targets_min": 643
|
|
},
|
|
{
|
|
"epoch": 4.257188498402556,
|
|
"grad_norm": 0.5191430163552699,
|
|
"learning_rate": 1.5987877159970765e-05,
|
|
"loss": 0.1965,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23058605194091797,
|
|
"step": 2665,
|
|
"valid_targets_mean": 4721.8,
|
|
"valid_targets_min": 2121
|
|
},
|
|
{
|
|
"epoch": 4.26517571884984,
|
|
"grad_norm": 0.49558878513171556,
|
|
"learning_rate": 1.5909853654821947e-05,
|
|
"loss": 0.1851,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23383599519729614,
|
|
"step": 2670,
|
|
"valid_targets_mean": 4648.8,
|
|
"valid_targets_min": 1276
|
|
},
|
|
{
|
|
"epoch": 4.273162939297125,
|
|
"grad_norm": 0.41929939973228997,
|
|
"learning_rate": 1.583189506159579e-05,
|
|
"loss": 0.1818,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1661728322505951,
|
|
"step": 2675,
|
|
"valid_targets_mean": 6055.9,
|
|
"valid_targets_min": 1690
|
|
},
|
|
{
|
|
"epoch": 4.281150159744409,
|
|
"grad_norm": 0.4335583132231686,
|
|
"learning_rate": 1.575400261751993e-05,
|
|
"loss": 0.1765,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16649028658866882,
|
|
"step": 2680,
|
|
"valid_targets_mean": 5270.9,
|
|
"valid_targets_min": 1863
|
|
},
|
|
{
|
|
"epoch": 4.289137380191693,
|
|
"grad_norm": 0.5033624044429718,
|
|
"learning_rate": 1.567617755877223e-05,
|
|
"loss": 0.1796,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21152611076831818,
|
|
"step": 2685,
|
|
"valid_targets_mean": 4367.6,
|
|
"valid_targets_min": 596
|
|
},
|
|
{
|
|
"epoch": 4.297124600638978,
|
|
"grad_norm": 0.4637755962720511,
|
|
"learning_rate": 1.5598421120461106e-05,
|
|
"loss": 0.1768,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2031022310256958,
|
|
"step": 2690,
|
|
"valid_targets_mean": 5368.4,
|
|
"valid_targets_min": 1109
|
|
},
|
|
{
|
|
"epoch": 4.305111821086262,
|
|
"grad_norm": 0.4756304172705233,
|
|
"learning_rate": 1.552073453660595e-05,
|
|
"loss": 0.1763,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19398033618927002,
|
|
"step": 2695,
|
|
"valid_targets_mean": 5216.4,
|
|
"valid_targets_min": 2283
|
|
},
|
|
{
|
|
"epoch": 4.313099041533547,
|
|
"grad_norm": 0.5287810944260913,
|
|
"learning_rate": 1.5443119040117544e-05,
|
|
"loss": 0.1681,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18938207626342773,
|
|
"step": 2700,
|
|
"valid_targets_mean": 3709.1,
|
|
"valid_targets_min": 1009
|
|
},
|
|
{
|
|
"epoch": 4.321086261980831,
|
|
"grad_norm": 0.5088421190380102,
|
|
"learning_rate": 1.536557586277849e-05,
|
|
"loss": 0.1789,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.209833025932312,
|
|
"step": 2705,
|
|
"valid_targets_mean": 5181.1,
|
|
"valid_targets_min": 1443
|
|
},
|
|
{
|
|
"epoch": 4.329073482428115,
|
|
"grad_norm": 0.42223055697102546,
|
|
"learning_rate": 1.5288106235223655e-05,
|
|
"loss": 0.1696,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13359196484088898,
|
|
"step": 2710,
|
|
"valid_targets_mean": 4941.1,
|
|
"valid_targets_min": 1419
|
|
},
|
|
{
|
|
"epoch": 4.3370607028754,
|
|
"grad_norm": 0.444464250701453,
|
|
"learning_rate": 1.521071138692066e-05,
|
|
"loss": 0.1606,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14987829327583313,
|
|
"step": 2715,
|
|
"valid_targets_mean": 4453.1,
|
|
"valid_targets_min": 405
|
|
},
|
|
{
|
|
"epoch": 4.345047923322683,
|
|
"grad_norm": 0.4876366454348983,
|
|
"learning_rate": 1.5133392546150347e-05,
|
|
"loss": 0.1897,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21039533615112305,
|
|
"step": 2720,
|
|
"valid_targets_mean": 4893.1,
|
|
"valid_targets_min": 817
|
|
},
|
|
{
|
|
"epoch": 4.353035143769968,
|
|
"grad_norm": 0.49731859469202605,
|
|
"learning_rate": 1.5056150939987303e-05,
|
|
"loss": 0.1799,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19343283772468567,
|
|
"step": 2725,
|
|
"valid_targets_mean": 4593.2,
|
|
"valid_targets_min": 762
|
|
},
|
|
{
|
|
"epoch": 4.361022364217252,
|
|
"grad_norm": 0.4727820962488383,
|
|
"learning_rate": 1.4978987794280363e-05,
|
|
"loss": 0.1745,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16634801030158997,
|
|
"step": 2730,
|
|
"valid_targets_mean": 4974.9,
|
|
"valid_targets_min": 1897
|
|
},
|
|
{
|
|
"epoch": 4.3690095846645365,
|
|
"grad_norm": 0.4837015825511962,
|
|
"learning_rate": 1.4901904333633179e-05,
|
|
"loss": 0.1625,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1642087996006012,
|
|
"step": 2735,
|
|
"valid_targets_mean": 4487.1,
|
|
"valid_targets_min": 765
|
|
},
|
|
{
|
|
"epoch": 4.376996805111821,
|
|
"grad_norm": 0.4568084735081673,
|
|
"learning_rate": 1.4824901781384777e-05,
|
|
"loss": 0.1695,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16266146302223206,
|
|
"step": 2740,
|
|
"valid_targets_mean": 4704.2,
|
|
"valid_targets_min": 590
|
|
},
|
|
{
|
|
"epoch": 4.384984025559105,
|
|
"grad_norm": 0.5317320056579059,
|
|
"learning_rate": 1.4747981359590138e-05,
|
|
"loss": 0.1768,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17885902523994446,
|
|
"step": 2745,
|
|
"valid_targets_mean": 5585.6,
|
|
"valid_targets_min": 2346
|
|
},
|
|
{
|
|
"epoch": 4.39297124600639,
|
|
"grad_norm": 0.47355838191742555,
|
|
"learning_rate": 1.4671144289000806e-05,
|
|
"loss": 0.1836,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1866658627986908,
|
|
"step": 2750,
|
|
"valid_targets_mean": 4586.2,
|
|
"valid_targets_min": 860
|
|
},
|
|
{
|
|
"epoch": 4.400958466453674,
|
|
"grad_norm": 0.4596060420334498,
|
|
"learning_rate": 1.4594391789045513e-05,
|
|
"loss": 0.181,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14949265122413635,
|
|
"step": 2755,
|
|
"valid_targets_mean": 4949.5,
|
|
"valid_targets_min": 808
|
|
},
|
|
{
|
|
"epoch": 4.4089456869009584,
|
|
"grad_norm": 0.4902124697286356,
|
|
"learning_rate": 1.4517725077810835e-05,
|
|
"loss": 0.1803,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18611708283424377,
|
|
"step": 2760,
|
|
"valid_targets_mean": 4757.1,
|
|
"valid_targets_min": 2281
|
|
},
|
|
{
|
|
"epoch": 4.416932907348243,
|
|
"grad_norm": 0.4397051481953006,
|
|
"learning_rate": 1.4441145372021847e-05,
|
|
"loss": 0.1683,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16182464361190796,
|
|
"step": 2765,
|
|
"valid_targets_mean": 5520.4,
|
|
"valid_targets_min": 1653
|
|
},
|
|
{
|
|
"epoch": 4.424920127795527,
|
|
"grad_norm": 0.49132049788954635,
|
|
"learning_rate": 1.4364653887022831e-05,
|
|
"loss": 0.1669,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.183456152677536,
|
|
"step": 2770,
|
|
"valid_targets_mean": 4734.7,
|
|
"valid_targets_min": 1523
|
|
},
|
|
{
|
|
"epoch": 4.432907348242812,
|
|
"grad_norm": 0.5022396485100658,
|
|
"learning_rate": 1.4288251836757956e-05,
|
|
"loss": 0.179,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16363565623760223,
|
|
"step": 2775,
|
|
"valid_targets_mean": 4242.3,
|
|
"valid_targets_min": 1927
|
|
},
|
|
{
|
|
"epoch": 4.440894568690096,
|
|
"grad_norm": 0.46423648756327707,
|
|
"learning_rate": 1.4211940433752052e-05,
|
|
"loss": 0.1708,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1824829876422882,
|
|
"step": 2780,
|
|
"valid_targets_mean": 5115.5,
|
|
"valid_targets_min": 747
|
|
},
|
|
{
|
|
"epoch": 4.44888178913738,
|
|
"grad_norm": 0.5437782791421605,
|
|
"learning_rate": 1.4135720889091342e-05,
|
|
"loss": 0.1893,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24194525182247162,
|
|
"step": 2785,
|
|
"valid_targets_mean": 3960.6,
|
|
"valid_targets_min": 665
|
|
},
|
|
{
|
|
"epoch": 4.456869009584665,
|
|
"grad_norm": 0.47207100051740214,
|
|
"learning_rate": 1.4059594412404227e-05,
|
|
"loss": 0.1695,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15635071694850922,
|
|
"step": 2790,
|
|
"valid_targets_mean": 4844.9,
|
|
"valid_targets_min": 770
|
|
},
|
|
{
|
|
"epoch": 4.464856230031949,
|
|
"grad_norm": 0.7725352514241253,
|
|
"learning_rate": 1.3983562211842094e-05,
|
|
"loss": 0.1754,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16280369460582733,
|
|
"step": 2795,
|
|
"valid_targets_mean": 4732.6,
|
|
"valid_targets_min": 2517
|
|
},
|
|
{
|
|
"epoch": 4.472843450479234,
|
|
"grad_norm": 0.556527290971323,
|
|
"learning_rate": 1.390762549406013e-05,
|
|
"loss": 0.1817,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19617465138435364,
|
|
"step": 2800,
|
|
"valid_targets_mean": 4603.3,
|
|
"valid_targets_min": 896
|
|
},
|
|
{
|
|
"epoch": 4.480830670926517,
|
|
"grad_norm": 0.4522080581527669,
|
|
"learning_rate": 1.3831785464198188e-05,
|
|
"loss": 0.182,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17426979541778564,
|
|
"step": 2805,
|
|
"valid_targets_mean": 5406.4,
|
|
"valid_targets_min": 1533
|
|
},
|
|
{
|
|
"epoch": 4.488817891373802,
|
|
"grad_norm": 0.4491497927742097,
|
|
"learning_rate": 1.3756043325861648e-05,
|
|
"loss": 0.1685,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1735353171825409,
|
|
"step": 2810,
|
|
"valid_targets_mean": 5382.9,
|
|
"valid_targets_min": 1207
|
|
},
|
|
{
|
|
"epoch": 4.496805111821086,
|
|
"grad_norm": 0.5070465660484962,
|
|
"learning_rate": 1.3680400281102326e-05,
|
|
"loss": 0.1786,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19638991355895996,
|
|
"step": 2815,
|
|
"valid_targets_mean": 4763.1,
|
|
"valid_targets_min": 1971
|
|
},
|
|
{
|
|
"epoch": 4.50479233226837,
|
|
"grad_norm": 0.4717610993795897,
|
|
"learning_rate": 1.3604857530399378e-05,
|
|
"loss": 0.1688,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1694454550743103,
|
|
"step": 2820,
|
|
"valid_targets_mean": 4820.9,
|
|
"valid_targets_min": 720
|
|
},
|
|
{
|
|
"epoch": 4.512779552715655,
|
|
"grad_norm": 0.4228838522176023,
|
|
"learning_rate": 1.3529416272640278e-05,
|
|
"loss": 0.1693,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14357447624206543,
|
|
"step": 2825,
|
|
"valid_targets_mean": 5209.4,
|
|
"valid_targets_min": 1897
|
|
},
|
|
{
|
|
"epoch": 4.520766773162939,
|
|
"grad_norm": 0.46720980296374576,
|
|
"learning_rate": 1.3454077705101766e-05,
|
|
"loss": 0.1739,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1450195610523224,
|
|
"step": 2830,
|
|
"valid_targets_mean": 5413.9,
|
|
"valid_targets_min": 2133
|
|
},
|
|
{
|
|
"epoch": 4.5287539936102235,
|
|
"grad_norm": 0.4595243786600848,
|
|
"learning_rate": 1.3378843023430856e-05,
|
|
"loss": 0.1804,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19309571385383606,
|
|
"step": 2835,
|
|
"valid_targets_mean": 5553.4,
|
|
"valid_targets_min": 1046
|
|
},
|
|
{
|
|
"epoch": 4.536741214057508,
|
|
"grad_norm": 0.4859604764765627,
|
|
"learning_rate": 1.3303713421625864e-05,
|
|
"loss": 0.1676,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1785299926996231,
|
|
"step": 2840,
|
|
"valid_targets_mean": 5517.8,
|
|
"valid_targets_min": 1231
|
|
},
|
|
{
|
|
"epoch": 4.544728434504792,
|
|
"grad_norm": 0.513803510247055,
|
|
"learning_rate": 1.3228690092017452e-05,
|
|
"loss": 0.1747,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16955965757369995,
|
|
"step": 2845,
|
|
"valid_targets_mean": 5190.0,
|
|
"valid_targets_min": 1084
|
|
},
|
|
{
|
|
"epoch": 4.552715654952077,
|
|
"grad_norm": 0.45254755090215193,
|
|
"learning_rate": 1.3153774225249705e-05,
|
|
"loss": 0.1803,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15954682230949402,
|
|
"step": 2850,
|
|
"valid_targets_mean": 4924.6,
|
|
"valid_targets_min": 1679
|
|
},
|
|
{
|
|
"epoch": 4.560702875399361,
|
|
"grad_norm": 0.4588250749939126,
|
|
"learning_rate": 1.3078967010261243e-05,
|
|
"loss": 0.1828,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16771024465560913,
|
|
"step": 2855,
|
|
"valid_targets_mean": 5032.6,
|
|
"valid_targets_min": 1681
|
|
},
|
|
{
|
|
"epoch": 4.568690095846645,
|
|
"grad_norm": 0.45471209980664273,
|
|
"learning_rate": 1.3004269634266357e-05,
|
|
"loss": 0.1841,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18457409739494324,
|
|
"step": 2860,
|
|
"valid_targets_mean": 5483.9,
|
|
"valid_targets_min": 1857
|
|
},
|
|
{
|
|
"epoch": 4.57667731629393,
|
|
"grad_norm": 0.5381067225547658,
|
|
"learning_rate": 1.2929683282736135e-05,
|
|
"loss": 0.1756,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19067108631134033,
|
|
"step": 2865,
|
|
"valid_targets_mean": 4096.4,
|
|
"valid_targets_min": 806
|
|
},
|
|
{
|
|
"epoch": 4.584664536741214,
|
|
"grad_norm": 0.4900463699350829,
|
|
"learning_rate": 1.285520913937969e-05,
|
|
"loss": 0.1733,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1881490796804428,
|
|
"step": 2870,
|
|
"valid_targets_mean": 4364.2,
|
|
"valid_targets_min": 855
|
|
},
|
|
{
|
|
"epoch": 4.592651757188499,
|
|
"grad_norm": 0.4331601483409343,
|
|
"learning_rate": 1.2780848386125354e-05,
|
|
"loss": 0.1717,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15843982994556427,
|
|
"step": 2875,
|
|
"valid_targets_mean": 4837.6,
|
|
"valid_targets_min": 625
|
|
},
|
|
{
|
|
"epoch": 4.600638977635783,
|
|
"grad_norm": 0.42410902438561765,
|
|
"learning_rate": 1.2706602203101913e-05,
|
|
"loss": 0.1728,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1685314178466797,
|
|
"step": 2880,
|
|
"valid_targets_mean": 5168.2,
|
|
"valid_targets_min": 2244
|
|
},
|
|
{
|
|
"epoch": 4.608626198083067,
|
|
"grad_norm": 0.44351881797251413,
|
|
"learning_rate": 1.2632471768619894e-05,
|
|
"loss": 0.1756,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14001521468162537,
|
|
"step": 2885,
|
|
"valid_targets_mean": 4807.2,
|
|
"valid_targets_min": 753
|
|
},
|
|
{
|
|
"epoch": 4.616613418530352,
|
|
"grad_norm": 0.5210509360425715,
|
|
"learning_rate": 1.255845825915286e-05,
|
|
"loss": 0.1663,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.163040429353714,
|
|
"step": 2890,
|
|
"valid_targets_mean": 3828.9,
|
|
"valid_targets_min": 1267
|
|
},
|
|
{
|
|
"epoch": 4.624600638977636,
|
|
"grad_norm": 0.4940486964325761,
|
|
"learning_rate": 1.248456284931873e-05,
|
|
"loss": 0.1656,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17665278911590576,
|
|
"step": 2895,
|
|
"valid_targets_mean": 4694.3,
|
|
"valid_targets_min": 682
|
|
},
|
|
{
|
|
"epoch": 4.63258785942492,
|
|
"grad_norm": 0.5312612354322576,
|
|
"learning_rate": 1.241078671186115e-05,
|
|
"loss": 0.1679,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1842440664768219,
|
|
"step": 2900,
|
|
"valid_targets_mean": 4051.1,
|
|
"valid_targets_min": 553
|
|
},
|
|
{
|
|
"epoch": 4.640575079872205,
|
|
"grad_norm": 0.5199309221773137,
|
|
"learning_rate": 1.2337131017630873e-05,
|
|
"loss": 0.1716,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1977030634880066,
|
|
"step": 2905,
|
|
"valid_targets_mean": 3997.2,
|
|
"valid_targets_min": 803
|
|
},
|
|
{
|
|
"epoch": 4.6485623003194885,
|
|
"grad_norm": 0.49556792854851944,
|
|
"learning_rate": 1.2263596935567175e-05,
|
|
"loss": 0.1694,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17178234457969666,
|
|
"step": 2910,
|
|
"valid_targets_mean": 4453.0,
|
|
"valid_targets_min": 1163
|
|
},
|
|
{
|
|
"epoch": 4.656549520766773,
|
|
"grad_norm": 0.4918010969834742,
|
|
"learning_rate": 1.2190185632679316e-05,
|
|
"loss": 0.1806,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2153959572315216,
|
|
"step": 2915,
|
|
"valid_targets_mean": 5196.2,
|
|
"valid_targets_min": 2456
|
|
},
|
|
{
|
|
"epoch": 4.664536741214057,
|
|
"grad_norm": 0.5504834449476904,
|
|
"learning_rate": 1.2116898274028008e-05,
|
|
"loss": 0.1681,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17798376083374023,
|
|
"step": 2920,
|
|
"valid_targets_mean": 4049.2,
|
|
"valid_targets_min": 1668
|
|
},
|
|
{
|
|
"epoch": 4.672523961661342,
|
|
"grad_norm": 0.441802781277938,
|
|
"learning_rate": 1.2043736022706934e-05,
|
|
"loss": 0.1796,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15885400772094727,
|
|
"step": 2925,
|
|
"valid_targets_mean": 4954.7,
|
|
"valid_targets_min": 1132
|
|
},
|
|
{
|
|
"epoch": 4.680511182108626,
|
|
"grad_norm": 0.5473584807237512,
|
|
"learning_rate": 1.1970700039824271e-05,
|
|
"loss": 0.1816,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21114695072174072,
|
|
"step": 2930,
|
|
"valid_targets_mean": 4388.6,
|
|
"valid_targets_min": 792
|
|
},
|
|
{
|
|
"epoch": 4.68849840255591,
|
|
"grad_norm": 0.4338073131012635,
|
|
"learning_rate": 1.1897791484484284e-05,
|
|
"loss": 0.1699,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15443475544452667,
|
|
"step": 2935,
|
|
"valid_targets_mean": 4838.2,
|
|
"valid_targets_min": 874
|
|
},
|
|
{
|
|
"epoch": 4.696485623003195,
|
|
"grad_norm": 0.4625102868784867,
|
|
"learning_rate": 1.182501151376893e-05,
|
|
"loss": 0.1811,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19839084148406982,
|
|
"step": 2940,
|
|
"valid_targets_mean": 5425.4,
|
|
"valid_targets_min": 3203
|
|
},
|
|
{
|
|
"epoch": 4.704472843450479,
|
|
"grad_norm": 0.466180176987445,
|
|
"learning_rate": 1.1752361282719472e-05,
|
|
"loss": 0.1751,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1526743769645691,
|
|
"step": 2945,
|
|
"valid_targets_mean": 4683.6,
|
|
"valid_targets_min": 1552
|
|
},
|
|
{
|
|
"epoch": 4.712460063897764,
|
|
"grad_norm": 0.47551837484440984,
|
|
"learning_rate": 1.1679841944318171e-05,
|
|
"loss": 0.1763,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17545554041862488,
|
|
"step": 2950,
|
|
"valid_targets_mean": 4613.1,
|
|
"valid_targets_min": 420
|
|
},
|
|
{
|
|
"epoch": 4.720447284345048,
|
|
"grad_norm": 0.39710037175579455,
|
|
"learning_rate": 1.1607454649469991e-05,
|
|
"loss": 0.1697,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.167159304022789,
|
|
"step": 2955,
|
|
"valid_targets_mean": 6828.9,
|
|
"valid_targets_min": 1557
|
|
},
|
|
{
|
|
"epoch": 4.728434504792332,
|
|
"grad_norm": 0.45927767227442196,
|
|
"learning_rate": 1.1535200546984291e-05,
|
|
"loss": 0.1731,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13732823729515076,
|
|
"step": 2960,
|
|
"valid_targets_mean": 4008.6,
|
|
"valid_targets_min": 336
|
|
},
|
|
{
|
|
"epoch": 4.736421725239617,
|
|
"grad_norm": 0.4457328214905737,
|
|
"learning_rate": 1.1463080783556663e-05,
|
|
"loss": 0.1742,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1593000888824463,
|
|
"step": 2965,
|
|
"valid_targets_mean": 4668.5,
|
|
"valid_targets_min": 2312
|
|
},
|
|
{
|
|
"epoch": 4.744408945686901,
|
|
"grad_norm": 0.45706099623559,
|
|
"learning_rate": 1.139109650375066e-05,
|
|
"loss": 0.1663,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15481191873550415,
|
|
"step": 2970,
|
|
"valid_targets_mean": 4386.1,
|
|
"valid_targets_min": 557
|
|
},
|
|
{
|
|
"epoch": 4.752396166134186,
|
|
"grad_norm": 0.4384834971282695,
|
|
"learning_rate": 1.1319248849979705e-05,
|
|
"loss": 0.1805,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17784811556339264,
|
|
"step": 2975,
|
|
"valid_targets_mean": 5961.6,
|
|
"valid_targets_min": 924
|
|
},
|
|
{
|
|
"epoch": 4.76038338658147,
|
|
"grad_norm": 0.4580626059711276,
|
|
"learning_rate": 1.1247538962488882e-05,
|
|
"loss": 0.1784,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1837904453277588,
|
|
"step": 2980,
|
|
"valid_targets_mean": 5080.2,
|
|
"valid_targets_min": 1940
|
|
},
|
|
{
|
|
"epoch": 4.768370607028754,
|
|
"grad_norm": 0.4769567025754341,
|
|
"learning_rate": 1.1175967979336913e-05,
|
|
"loss": 0.1793,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15448135137557983,
|
|
"step": 2985,
|
|
"valid_targets_mean": 3807.9,
|
|
"valid_targets_min": 566
|
|
},
|
|
{
|
|
"epoch": 4.776357827476039,
|
|
"grad_norm": 0.4453754802734512,
|
|
"learning_rate": 1.1104537036378054e-05,
|
|
"loss": 0.1654,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13982991874217987,
|
|
"step": 2990,
|
|
"valid_targets_mean": 5076.4,
|
|
"valid_targets_min": 1683
|
|
},
|
|
{
|
|
"epoch": 4.784345047923322,
|
|
"grad_norm": 1.2063322370643315,
|
|
"learning_rate": 1.1033247267244063e-05,
|
|
"loss": 0.1715,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14099296927452087,
|
|
"step": 2995,
|
|
"valid_targets_mean": 5085.6,
|
|
"valid_targets_min": 2205
|
|
},
|
|
{
|
|
"epoch": 4.792332268370607,
|
|
"grad_norm": 0.5668162128968549,
|
|
"learning_rate": 1.0962099803326257e-05,
|
|
"loss": 0.1744,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16944657266139984,
|
|
"step": 3000,
|
|
"valid_targets_mean": 4677.1,
|
|
"valid_targets_min": 2127
|
|
},
|
|
{
|
|
"epoch": 4.800319488817891,
|
|
"grad_norm": 0.5039649642727122,
|
|
"learning_rate": 1.089109577375748e-05,
|
|
"loss": 0.1821,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20508737862110138,
|
|
"step": 3005,
|
|
"valid_targets_mean": 5167.5,
|
|
"valid_targets_min": 2174
|
|
},
|
|
{
|
|
"epoch": 4.8083067092651754,
|
|
"grad_norm": 0.46177989402877045,
|
|
"learning_rate": 1.082023630539427e-05,
|
|
"loss": 0.1807,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19288091361522675,
|
|
"step": 3010,
|
|
"valid_targets_mean": 5167.2,
|
|
"valid_targets_min": 614
|
|
},
|
|
{
|
|
"epoch": 4.81629392971246,
|
|
"grad_norm": 0.5194814075761426,
|
|
"learning_rate": 1.07495225227989e-05,
|
|
"loss": 0.1787,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20017312467098236,
|
|
"step": 3015,
|
|
"valid_targets_mean": 4127.4,
|
|
"valid_targets_min": 1210
|
|
},
|
|
{
|
|
"epoch": 4.824281150159744,
|
|
"grad_norm": 0.5111828873940237,
|
|
"learning_rate": 1.0678955548221595e-05,
|
|
"loss": 0.1748,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16314226388931274,
|
|
"step": 3020,
|
|
"valid_targets_mean": 4261.6,
|
|
"valid_targets_min": 337
|
|
},
|
|
{
|
|
"epoch": 4.832268370607029,
|
|
"grad_norm": 0.48761163101608956,
|
|
"learning_rate": 1.0608536501582654e-05,
|
|
"loss": 0.1681,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20200759172439575,
|
|
"step": 3025,
|
|
"valid_targets_mean": 4762.9,
|
|
"valid_targets_min": 1243
|
|
},
|
|
{
|
|
"epoch": 4.840255591054313,
|
|
"grad_norm": 0.4630695814574282,
|
|
"learning_rate": 1.0538266500454739e-05,
|
|
"loss": 0.1761,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17577683925628662,
|
|
"step": 3030,
|
|
"valid_targets_mean": 5338.1,
|
|
"valid_targets_min": 961
|
|
},
|
|
{
|
|
"epoch": 4.848242811501597,
|
|
"grad_norm": 0.42697756027113803,
|
|
"learning_rate": 1.0468146660045118e-05,
|
|
"loss": 0.1736,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17181923985481262,
|
|
"step": 3035,
|
|
"valid_targets_mean": 6130.8,
|
|
"valid_targets_min": 2510
|
|
},
|
|
{
|
|
"epoch": 4.856230031948882,
|
|
"grad_norm": 0.4518844364317005,
|
|
"learning_rate": 1.0398178093177928e-05,
|
|
"loss": 0.167,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15323194861412048,
|
|
"step": 3040,
|
|
"valid_targets_mean": 4782.7,
|
|
"valid_targets_min": 940
|
|
},
|
|
{
|
|
"epoch": 4.864217252396166,
|
|
"grad_norm": 0.5664237237310001,
|
|
"learning_rate": 1.0328361910276592e-05,
|
|
"loss": 0.1705,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16498196125030518,
|
|
"step": 3045,
|
|
"valid_targets_mean": 4808.9,
|
|
"valid_targets_min": 1623
|
|
},
|
|
{
|
|
"epoch": 4.872204472843451,
|
|
"grad_norm": 0.47173286112075913,
|
|
"learning_rate": 1.0258699219346091e-05,
|
|
"loss": 0.1754,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17309358716011047,
|
|
"step": 3050,
|
|
"valid_targets_mean": 4583.5,
|
|
"valid_targets_min": 2051
|
|
},
|
|
{
|
|
"epoch": 4.880191693290735,
|
|
"grad_norm": 0.5718052290052063,
|
|
"learning_rate": 1.0189191125955503e-05,
|
|
"loss": 0.1674,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17422272264957428,
|
|
"step": 3055,
|
|
"valid_targets_mean": 3853.5,
|
|
"valid_targets_min": 805
|
|
},
|
|
{
|
|
"epoch": 4.888178913738019,
|
|
"grad_norm": 0.4620315195605763,
|
|
"learning_rate": 1.0119838733220342e-05,
|
|
"loss": 0.1689,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1558941751718521,
|
|
"step": 3060,
|
|
"valid_targets_mean": 4822.6,
|
|
"valid_targets_min": 695
|
|
},
|
|
{
|
|
"epoch": 4.896166134185304,
|
|
"grad_norm": 0.5131232008912642,
|
|
"learning_rate": 1.0050643141785148e-05,
|
|
"loss": 0.1805,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16869670152664185,
|
|
"step": 3065,
|
|
"valid_targets_mean": 4437.9,
|
|
"valid_targets_min": 656
|
|
},
|
|
{
|
|
"epoch": 4.904153354632588,
|
|
"grad_norm": 0.4410859299182515,
|
|
"learning_rate": 9.981605449805933e-06,
|
|
"loss": 0.1736,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18083491921424866,
|
|
"step": 3070,
|
|
"valid_targets_mean": 5510.3,
|
|
"valid_targets_min": 1818
|
|
},
|
|
{
|
|
"epoch": 4.912140575079873,
|
|
"grad_norm": 0.4514321375646226,
|
|
"learning_rate": 9.912726752932823e-06,
|
|
"loss": 0.1635,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16155874729156494,
|
|
"step": 3075,
|
|
"valid_targets_mean": 5034.1,
|
|
"valid_targets_min": 1344
|
|
},
|
|
{
|
|
"epoch": 4.920127795527156,
|
|
"grad_norm": 0.4980585873144301,
|
|
"learning_rate": 9.844008144292643e-06,
|
|
"loss": 0.1929,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24201665818691254,
|
|
"step": 3080,
|
|
"valid_targets_mean": 4619.4,
|
|
"valid_targets_min": 571
|
|
},
|
|
{
|
|
"epoch": 4.928115015974441,
|
|
"grad_norm": 0.559552867245746,
|
|
"learning_rate": 9.775450714471537e-06,
|
|
"loss": 0.181,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1549963653087616,
|
|
"step": 3085,
|
|
"valid_targets_mean": 4439.6,
|
|
"valid_targets_min": 1307
|
|
},
|
|
{
|
|
"epoch": 4.936102236421725,
|
|
"grad_norm": 0.44659416714775546,
|
|
"learning_rate": 9.707055551497734e-06,
|
|
"loss": 0.1723,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1799061894416809,
|
|
"step": 3090,
|
|
"valid_targets_mean": 5328.0,
|
|
"valid_targets_min": 1392
|
|
},
|
|
{
|
|
"epoch": 4.944089456869009,
|
|
"grad_norm": 0.4828672393131135,
|
|
"learning_rate": 9.638823740824177e-06,
|
|
"loss": 0.1768,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19210804998874664,
|
|
"step": 3095,
|
|
"valid_targets_mean": 5063.3,
|
|
"valid_targets_min": 916
|
|
},
|
|
{
|
|
"epoch": 4.952076677316294,
|
|
"grad_norm": 0.6302157500037942,
|
|
"learning_rate": 9.570756365311418e-06,
|
|
"loss": 0.1751,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19101989269256592,
|
|
"step": 3100,
|
|
"valid_targets_mean": 3066.4,
|
|
"valid_targets_min": 626
|
|
},
|
|
{
|
|
"epoch": 4.960063897763578,
|
|
"grad_norm": 0.4651924692921601,
|
|
"learning_rate": 9.502854505210326e-06,
|
|
"loss": 0.1764,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1738903820514679,
|
|
"step": 3105,
|
|
"valid_targets_mean": 5021.1,
|
|
"valid_targets_min": 1392
|
|
},
|
|
{
|
|
"epoch": 4.968051118210862,
|
|
"grad_norm": 0.5080118101677551,
|
|
"learning_rate": 9.435119238145018e-06,
|
|
"loss": 0.1791,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1990775465965271,
|
|
"step": 3110,
|
|
"valid_targets_mean": 5799.6,
|
|
"valid_targets_min": 1404
|
|
},
|
|
{
|
|
"epoch": 4.976038338658147,
|
|
"grad_norm": 0.4644747451905339,
|
|
"learning_rate": 9.367551639095704e-06,
|
|
"loss": 0.1674,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14686700701713562,
|
|
"step": 3115,
|
|
"valid_targets_mean": 4626.5,
|
|
"valid_targets_min": 409
|
|
},
|
|
{
|
|
"epoch": 4.984025559105431,
|
|
"grad_norm": 0.42687008353899464,
|
|
"learning_rate": 9.300152780381664e-06,
|
|
"loss": 0.1719,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16838842630386353,
|
|
"step": 3120,
|
|
"valid_targets_mean": 5399.3,
|
|
"valid_targets_min": 1914
|
|
},
|
|
{
|
|
"epoch": 4.992012779552716,
|
|
"grad_norm": 0.38161676379593085,
|
|
"learning_rate": 9.232923731644216e-06,
|
|
"loss": 0.1738,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14308978617191315,
|
|
"step": 3125,
|
|
"valid_targets_mean": 5909.3,
|
|
"valid_targets_min": 2080
|
|
},
|
|
{
|
|
"epoch": 5.0,
|
|
"grad_norm": 0.47394287634536164,
|
|
"learning_rate": 9.16586555982972e-06,
|
|
"loss": 0.171,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1663876175880432,
|
|
"step": 3130,
|
|
"valid_targets_mean": 4612.8,
|
|
"valid_targets_min": 1419
|
|
},
|
|
{
|
|
"epoch": 5.007987220447284,
|
|
"grad_norm": 0.45724309774968813,
|
|
"learning_rate": 9.098979329172702e-06,
|
|
"loss": 0.1589,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15543916821479797,
|
|
"step": 3135,
|
|
"valid_targets_mean": 5156.6,
|
|
"valid_targets_min": 1269
|
|
},
|
|
{
|
|
"epoch": 5.015974440894569,
|
|
"grad_norm": 0.5375515015395808,
|
|
"learning_rate": 9.032266101178872e-06,
|
|
"loss": 0.1714,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1786627471446991,
|
|
"step": 3140,
|
|
"valid_targets_mean": 4052.9,
|
|
"valid_targets_min": 409
|
|
},
|
|
{
|
|
"epoch": 5.023961661341853,
|
|
"grad_norm": 0.5138573644283863,
|
|
"learning_rate": 8.965726934608392e-06,
|
|
"loss": 0.1594,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17196139693260193,
|
|
"step": 3145,
|
|
"valid_targets_mean": 6069.1,
|
|
"valid_targets_min": 2498
|
|
},
|
|
{
|
|
"epoch": 5.031948881789138,
|
|
"grad_norm": 0.43963520310638937,
|
|
"learning_rate": 8.899362885458964e-06,
|
|
"loss": 0.1724,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15448343753814697,
|
|
"step": 3150,
|
|
"valid_targets_mean": 4924.6,
|
|
"valid_targets_min": 1058
|
|
},
|
|
{
|
|
"epoch": 5.039936102236422,
|
|
"grad_norm": 0.5289418769845741,
|
|
"learning_rate": 8.833175006949165e-06,
|
|
"loss": 0.1554,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15173891186714172,
|
|
"step": 3155,
|
|
"valid_targets_mean": 3554.7,
|
|
"valid_targets_min": 714
|
|
},
|
|
{
|
|
"epoch": 5.047923322683706,
|
|
"grad_norm": 0.5972361696404577,
|
|
"learning_rate": 8.767164349501648e-06,
|
|
"loss": 0.1563,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18872520327568054,
|
|
"step": 3160,
|
|
"valid_targets_mean": 5194.0,
|
|
"valid_targets_min": 611
|
|
},
|
|
{
|
|
"epoch": 5.055910543130991,
|
|
"grad_norm": 0.4522349154786805,
|
|
"learning_rate": 8.70133196072654e-06,
|
|
"loss": 0.1654,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1551792174577713,
|
|
"step": 3165,
|
|
"valid_targets_mean": 5324.4,
|
|
"valid_targets_min": 1254
|
|
},
|
|
{
|
|
"epoch": 5.063897763578275,
|
|
"grad_norm": 0.5166382639587666,
|
|
"learning_rate": 8.63567888540478e-06,
|
|
"loss": 0.1637,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1354677975177765,
|
|
"step": 3170,
|
|
"valid_targets_mean": 3944.1,
|
|
"valid_targets_min": 1318
|
|
},
|
|
{
|
|
"epoch": 5.0718849840255595,
|
|
"grad_norm": 0.5458370995208548,
|
|
"learning_rate": 8.570206165471535e-06,
|
|
"loss": 0.1603,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21429777145385742,
|
|
"step": 3175,
|
|
"valid_targets_mean": 4476.5,
|
|
"valid_targets_min": 1940
|
|
},
|
|
{
|
|
"epoch": 5.079872204472843,
|
|
"grad_norm": 0.5005297076281998,
|
|
"learning_rate": 8.504914839999691e-06,
|
|
"loss": 0.1615,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14907950162887573,
|
|
"step": 3180,
|
|
"valid_targets_mean": 4650.1,
|
|
"valid_targets_min": 1352
|
|
},
|
|
{
|
|
"epoch": 5.087859424920127,
|
|
"grad_norm": 0.4579906641875055,
|
|
"learning_rate": 8.439805945183333e-06,
|
|
"loss": 0.161,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20461249351501465,
|
|
"step": 3185,
|
|
"valid_targets_mean": 6407.1,
|
|
"valid_targets_min": 1645
|
|
},
|
|
{
|
|
"epoch": 5.095846645367412,
|
|
"grad_norm": 0.5348372856234672,
|
|
"learning_rate": 8.374880514321318e-06,
|
|
"loss": 0.1626,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15267446637153625,
|
|
"step": 3190,
|
|
"valid_targets_mean": 3976.1,
|
|
"valid_targets_min": 957
|
|
},
|
|
{
|
|
"epoch": 5.103833865814696,
|
|
"grad_norm": 0.42872203869233483,
|
|
"learning_rate": 8.310139577800864e-06,
|
|
"loss": 0.1645,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15673330426216125,
|
|
"step": 3195,
|
|
"valid_targets_mean": 5642.8,
|
|
"valid_targets_min": 2409
|
|
},
|
|
{
|
|
"epoch": 5.111821086261981,
|
|
"grad_norm": 0.4520202286446449,
|
|
"learning_rate": 8.245584163081228e-06,
|
|
"loss": 0.1535,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15110281109809875,
|
|
"step": 3200,
|
|
"valid_targets_mean": 5858.1,
|
|
"valid_targets_min": 3241
|
|
},
|
|
{
|
|
"epoch": 5.119808306709265,
|
|
"grad_norm": 0.4759288491846095,
|
|
"learning_rate": 8.18121529467735e-06,
|
|
"loss": 0.1579,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.135267436504364,
|
|
"step": 3205,
|
|
"valid_targets_mean": 4472.4,
|
|
"valid_targets_min": 1280
|
|
},
|
|
{
|
|
"epoch": 5.127795527156549,
|
|
"grad_norm": 0.5449524467506791,
|
|
"learning_rate": 8.117033994143635e-06,
|
|
"loss": 0.1642,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1479802131652832,
|
|
"step": 3210,
|
|
"valid_targets_mean": 4995.2,
|
|
"valid_targets_min": 739
|
|
},
|
|
{
|
|
"epoch": 5.135782747603834,
|
|
"grad_norm": 0.5234094996626493,
|
|
"learning_rate": 8.053041280057733e-06,
|
|
"loss": 0.1732,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1646098494529724,
|
|
"step": 3215,
|
|
"valid_targets_mean": 4460.2,
|
|
"valid_targets_min": 823
|
|
},
|
|
{
|
|
"epoch": 5.143769968051118,
|
|
"grad_norm": 0.4668031288928073,
|
|
"learning_rate": 7.989238168004347e-06,
|
|
"loss": 0.1598,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1528545469045639,
|
|
"step": 3220,
|
|
"valid_targets_mean": 4892.1,
|
|
"valid_targets_min": 1459
|
|
},
|
|
{
|
|
"epoch": 5.151757188498403,
|
|
"grad_norm": 0.46150987530655635,
|
|
"learning_rate": 7.925625670559158e-06,
|
|
"loss": 0.1627,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16160641610622406,
|
|
"step": 3225,
|
|
"valid_targets_mean": 5333.8,
|
|
"valid_targets_min": 1736
|
|
},
|
|
{
|
|
"epoch": 5.159744408945687,
|
|
"grad_norm": 0.5603329278900041,
|
|
"learning_rate": 7.862204797272716e-06,
|
|
"loss": 0.1607,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15708664059638977,
|
|
"step": 3230,
|
|
"valid_targets_mean": 4700.1,
|
|
"valid_targets_min": 791
|
|
},
|
|
{
|
|
"epoch": 5.167731629392971,
|
|
"grad_norm": 0.45539642954548165,
|
|
"learning_rate": 7.798976554654438e-06,
|
|
"loss": 0.1618,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1746443808078766,
|
|
"step": 3235,
|
|
"valid_targets_mean": 5515.9,
|
|
"valid_targets_min": 1947
|
|
},
|
|
{
|
|
"epoch": 5.175718849840256,
|
|
"grad_norm": 0.4994398855561148,
|
|
"learning_rate": 7.73594194615663e-06,
|
|
"loss": 0.1672,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1738542914390564,
|
|
"step": 3240,
|
|
"valid_targets_mean": 4582.4,
|
|
"valid_targets_min": 572
|
|
},
|
|
{
|
|
"epoch": 5.18370607028754,
|
|
"grad_norm": 0.5240890324025992,
|
|
"learning_rate": 7.673101972158572e-06,
|
|
"loss": 0.1552,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17766988277435303,
|
|
"step": 3245,
|
|
"valid_targets_mean": 4482.9,
|
|
"valid_targets_min": 617
|
|
},
|
|
{
|
|
"epoch": 5.1916932907348246,
|
|
"grad_norm": 0.45091164589846017,
|
|
"learning_rate": 7.610457629950621e-06,
|
|
"loss": 0.1589,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15332496166229248,
|
|
"step": 3250,
|
|
"valid_targets_mean": 5425.2,
|
|
"valid_targets_min": 936
|
|
},
|
|
{
|
|
"epoch": 5.199680511182109,
|
|
"grad_norm": 0.4892456016566229,
|
|
"learning_rate": 7.548009913718402e-06,
|
|
"loss": 0.165,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1631796956062317,
|
|
"step": 3255,
|
|
"valid_targets_mean": 4840.7,
|
|
"valid_targets_min": 901
|
|
},
|
|
{
|
|
"epoch": 5.207667731629393,
|
|
"grad_norm": 0.49628523854824014,
|
|
"learning_rate": 7.485759814527034e-06,
|
|
"loss": 0.1687,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15594470500946045,
|
|
"step": 3260,
|
|
"valid_targets_mean": 4835.2,
|
|
"valid_targets_min": 2353
|
|
},
|
|
{
|
|
"epoch": 5.215654952076678,
|
|
"grad_norm": 0.4767934545124,
|
|
"learning_rate": 7.423708320305361e-06,
|
|
"loss": 0.1548,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15322643518447876,
|
|
"step": 3265,
|
|
"valid_targets_mean": 4935.9,
|
|
"valid_targets_min": 1314
|
|
},
|
|
{
|
|
"epoch": 5.223642172523961,
|
|
"grad_norm": 0.46585801665699067,
|
|
"learning_rate": 7.361856415830335e-06,
|
|
"loss": 0.1734,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17025049030780792,
|
|
"step": 3270,
|
|
"valid_targets_mean": 5506.8,
|
|
"valid_targets_min": 942
|
|
},
|
|
{
|
|
"epoch": 5.231629392971246,
|
|
"grad_norm": 0.49838933432010535,
|
|
"learning_rate": 7.300205082711336e-06,
|
|
"loss": 0.1725,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16799825429916382,
|
|
"step": 3275,
|
|
"valid_targets_mean": 5049.1,
|
|
"valid_targets_min": 1453
|
|
},
|
|
{
|
|
"epoch": 5.23961661341853,
|
|
"grad_norm": 0.4496138993305811,
|
|
"learning_rate": 7.2387552993746204e-06,
|
|
"loss": 0.165,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1527540236711502,
|
|
"step": 3280,
|
|
"valid_targets_mean": 6115.8,
|
|
"valid_targets_min": 786
|
|
},
|
|
{
|
|
"epoch": 5.247603833865814,
|
|
"grad_norm": 0.520151851078282,
|
|
"learning_rate": 7.177508041047769e-06,
|
|
"loss": 0.1563,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16873812675476074,
|
|
"step": 3285,
|
|
"valid_targets_mean": 4611.3,
|
|
"valid_targets_min": 2114
|
|
},
|
|
{
|
|
"epoch": 5.255591054313099,
|
|
"grad_norm": 0.6413493447918366,
|
|
"learning_rate": 7.116464279744262e-06,
|
|
"loss": 0.1806,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1801944375038147,
|
|
"step": 3290,
|
|
"valid_targets_mean": 5371.3,
|
|
"valid_targets_min": 2600
|
|
},
|
|
{
|
|
"epoch": 5.263578274760383,
|
|
"grad_norm": 0.42212429264005713,
|
|
"learning_rate": 7.055624984247977e-06,
|
|
"loss": 0.1579,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15957115590572357,
|
|
"step": 3295,
|
|
"valid_targets_mean": 5400.9,
|
|
"valid_targets_min": 896
|
|
},
|
|
{
|
|
"epoch": 5.271565495207668,
|
|
"grad_norm": 0.46092174693212845,
|
|
"learning_rate": 6.9949911200978805e-06,
|
|
"loss": 0.1643,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1509469449520111,
|
|
"step": 3300,
|
|
"valid_targets_mean": 4963.5,
|
|
"valid_targets_min": 1635
|
|
},
|
|
{
|
|
"epoch": 5.279552715654952,
|
|
"grad_norm": 0.4572057037635674,
|
|
"learning_rate": 6.934563649572678e-06,
|
|
"loss": 0.1589,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14917771518230438,
|
|
"step": 3305,
|
|
"valid_targets_mean": 4919.5,
|
|
"valid_targets_min": 984
|
|
},
|
|
{
|
|
"epoch": 5.287539936102236,
|
|
"grad_norm": 0.4623875098548041,
|
|
"learning_rate": 6.874343531675521e-06,
|
|
"loss": 0.1593,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15402501821517944,
|
|
"step": 3310,
|
|
"valid_targets_mean": 4710.2,
|
|
"valid_targets_min": 1263
|
|
},
|
|
{
|
|
"epoch": 5.295527156549521,
|
|
"grad_norm": 0.4523223248875542,
|
|
"learning_rate": 6.814331722118837e-06,
|
|
"loss": 0.1707,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18308000266551971,
|
|
"step": 3315,
|
|
"valid_targets_mean": 5763.9,
|
|
"valid_targets_min": 2312
|
|
},
|
|
{
|
|
"epoch": 5.303514376996805,
|
|
"grad_norm": 0.5024336853774208,
|
|
"learning_rate": 6.754529173309112e-06,
|
|
"loss": 0.159,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14715887606143951,
|
|
"step": 3320,
|
|
"valid_targets_mean": 4335.7,
|
|
"valid_targets_min": 1574
|
|
},
|
|
{
|
|
"epoch": 5.31150159744409,
|
|
"grad_norm": 0.5485512556247358,
|
|
"learning_rate": 6.694936834331809e-06,
|
|
"loss": 0.1668,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17354059219360352,
|
|
"step": 3325,
|
|
"valid_targets_mean": 4979.9,
|
|
"valid_targets_min": 1732
|
|
},
|
|
{
|
|
"epoch": 5.319488817891374,
|
|
"grad_norm": 0.47928325678399936,
|
|
"learning_rate": 6.635555650936278e-06,
|
|
"loss": 0.1604,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17333224415779114,
|
|
"step": 3330,
|
|
"valid_targets_mean": 5068.4,
|
|
"valid_targets_min": 1320
|
|
},
|
|
{
|
|
"epoch": 5.327476038338658,
|
|
"grad_norm": 0.5934479215992917,
|
|
"learning_rate": 6.576386565520794e-06,
|
|
"loss": 0.1696,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15551719069480896,
|
|
"step": 3335,
|
|
"valid_targets_mean": 6251.5,
|
|
"valid_targets_min": 2807
|
|
},
|
|
{
|
|
"epoch": 5.335463258785943,
|
|
"grad_norm": 0.48745271437862536,
|
|
"learning_rate": 6.5174305171175336e-06,
|
|
"loss": 0.1544,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17440325021743774,
|
|
"step": 3340,
|
|
"valid_targets_mean": 5215.0,
|
|
"valid_targets_min": 1470
|
|
},
|
|
{
|
|
"epoch": 5.343450479233227,
|
|
"grad_norm": 0.4802249791337759,
|
|
"learning_rate": 6.458688441377734e-06,
|
|
"loss": 0.1698,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18174077570438385,
|
|
"step": 3345,
|
|
"valid_targets_mean": 5208.0,
|
|
"valid_targets_min": 1072
|
|
},
|
|
{
|
|
"epoch": 5.3514376996805115,
|
|
"grad_norm": 0.4664532449254744,
|
|
"learning_rate": 6.400161270556817e-06,
|
|
"loss": 0.1556,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14379502832889557,
|
|
"step": 3350,
|
|
"valid_targets_mean": 4037.0,
|
|
"valid_targets_min": 634
|
|
},
|
|
{
|
|
"epoch": 5.359424920127796,
|
|
"grad_norm": 0.43779060372472667,
|
|
"learning_rate": 6.341849933499573e-06,
|
|
"loss": 0.1675,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1506979912519455,
|
|
"step": 3355,
|
|
"valid_targets_mean": 6327.8,
|
|
"valid_targets_min": 857
|
|
},
|
|
{
|
|
"epoch": 5.36741214057508,
|
|
"grad_norm": 0.49630264813010194,
|
|
"learning_rate": 6.283755355625472e-06,
|
|
"loss": 0.1664,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18854346871376038,
|
|
"step": 3360,
|
|
"valid_targets_mean": 5617.9,
|
|
"valid_targets_min": 1552
|
|
},
|
|
{
|
|
"epoch": 5.375399361022364,
|
|
"grad_norm": 0.5457885435607882,
|
|
"learning_rate": 6.225878458913917e-06,
|
|
"loss": 0.1578,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17461910843849182,
|
|
"step": 3365,
|
|
"valid_targets_mean": 5613.8,
|
|
"valid_targets_min": 1821
|
|
},
|
|
{
|
|
"epoch": 5.383386581469648,
|
|
"grad_norm": 0.48035325618255165,
|
|
"learning_rate": 6.168220161889678e-06,
|
|
"loss": 0.1853,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1818755567073822,
|
|
"step": 3370,
|
|
"valid_targets_mean": 5177.7,
|
|
"valid_targets_min": 2053
|
|
},
|
|
{
|
|
"epoch": 5.391373801916933,
|
|
"grad_norm": 0.5405350831620226,
|
|
"learning_rate": 6.110781379608226e-06,
|
|
"loss": 0.1608,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1981622874736786,
|
|
"step": 3375,
|
|
"valid_targets_mean": 4290.1,
|
|
"valid_targets_min": 433
|
|
},
|
|
{
|
|
"epoch": 5.399361022364217,
|
|
"grad_norm": 1.301410002327568,
|
|
"learning_rate": 6.053563023641318e-06,
|
|
"loss": 0.1679,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17580270767211914,
|
|
"step": 3380,
|
|
"valid_targets_mean": 5276.1,
|
|
"valid_targets_min": 1344
|
|
},
|
|
{
|
|
"epoch": 5.407348242811501,
|
|
"grad_norm": 0.501406306040324,
|
|
"learning_rate": 5.996566002062439e-06,
|
|
"loss": 0.1649,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14683012664318085,
|
|
"step": 3385,
|
|
"valid_targets_mean": 4519.6,
|
|
"valid_targets_min": 926
|
|
},
|
|
{
|
|
"epoch": 5.415335463258786,
|
|
"grad_norm": 0.5051757518619181,
|
|
"learning_rate": 5.939791219432442e-06,
|
|
"loss": 0.1816,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18291693925857544,
|
|
"step": 3390,
|
|
"valid_targets_mean": 6008.5,
|
|
"valid_targets_min": 586
|
|
},
|
|
{
|
|
"epoch": 5.42332268370607,
|
|
"grad_norm": 0.43841993125264633,
|
|
"learning_rate": 5.8832395767851846e-06,
|
|
"loss": 0.1565,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1574784219264984,
|
|
"step": 3395,
|
|
"valid_targets_mean": 6095.5,
|
|
"valid_targets_min": 1844
|
|
},
|
|
{
|
|
"epoch": 5.431309904153355,
|
|
"grad_norm": 0.5745704992069267,
|
|
"learning_rate": 5.826911971613198e-06,
|
|
"loss": 0.1728,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1651586890220642,
|
|
"step": 3400,
|
|
"valid_targets_mean": 4589.2,
|
|
"valid_targets_min": 483
|
|
},
|
|
{
|
|
"epoch": 5.439297124600639,
|
|
"grad_norm": 0.4989786194389621,
|
|
"learning_rate": 5.770809297853495e-06,
|
|
"loss": 0.1512,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14755919575691223,
|
|
"step": 3405,
|
|
"valid_targets_mean": 5938.4,
|
|
"valid_targets_min": 2031
|
|
},
|
|
{
|
|
"epoch": 5.447284345047923,
|
|
"grad_norm": 0.49443831260201854,
|
|
"learning_rate": 5.714932445873325e-06,
|
|
"loss": 0.1616,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1464129090309143,
|
|
"step": 3410,
|
|
"valid_targets_mean": 4288.4,
|
|
"valid_targets_min": 1100
|
|
},
|
|
{
|
|
"epoch": 5.455271565495208,
|
|
"grad_norm": 0.44461458578301255,
|
|
"learning_rate": 5.659282302456115e-06,
|
|
"loss": 0.1568,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14050893485546112,
|
|
"step": 3415,
|
|
"valid_targets_mean": 4480.2,
|
|
"valid_targets_min": 1900
|
|
},
|
|
{
|
|
"epoch": 5.463258785942492,
|
|
"grad_norm": 0.46880506564818625,
|
|
"learning_rate": 5.603859750787302e-06,
|
|
"loss": 0.1691,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16493158042430878,
|
|
"step": 3420,
|
|
"valid_targets_mean": 4905.8,
|
|
"valid_targets_min": 729
|
|
},
|
|
{
|
|
"epoch": 5.4712460063897765,
|
|
"grad_norm": 0.5510903509258099,
|
|
"learning_rate": 5.548665670440418e-06,
|
|
"loss": 0.1739,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1883580982685089,
|
|
"step": 3425,
|
|
"valid_targets_mean": 5244.1,
|
|
"valid_targets_min": 1231
|
|
},
|
|
{
|
|
"epoch": 5.479233226837061,
|
|
"grad_norm": 0.5762253046887472,
|
|
"learning_rate": 5.4937009373630535e-06,
|
|
"loss": 0.1668,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2006421685218811,
|
|
"step": 3430,
|
|
"valid_targets_mean": 4248.2,
|
|
"valid_targets_min": 391
|
|
},
|
|
{
|
|
"epoch": 5.487220447284345,
|
|
"grad_norm": 0.4097504486851553,
|
|
"learning_rate": 5.438966423862997e-06,
|
|
"loss": 0.167,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15338310599327087,
|
|
"step": 3435,
|
|
"valid_targets_mean": 5802.8,
|
|
"valid_targets_min": 3432
|
|
},
|
|
{
|
|
"epoch": 5.49520766773163,
|
|
"grad_norm": 0.4823098748493422,
|
|
"learning_rate": 5.384462998594384e-06,
|
|
"loss": 0.1726,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19534525275230408,
|
|
"step": 3440,
|
|
"valid_targets_mean": 4855.6,
|
|
"valid_targets_min": 1535
|
|
},
|
|
{
|
|
"epoch": 5.503194888178914,
|
|
"grad_norm": 0.5244003410758836,
|
|
"learning_rate": 5.330191526543884e-06,
|
|
"loss": 0.1617,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15437617897987366,
|
|
"step": 3445,
|
|
"valid_targets_mean": 4552.6,
|
|
"valid_targets_min": 1260
|
|
},
|
|
{
|
|
"epoch": 5.511182108626198,
|
|
"grad_norm": 0.7176789774329526,
|
|
"learning_rate": 5.276152869017028e-06,
|
|
"loss": 0.1751,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17744609713554382,
|
|
"step": 3450,
|
|
"valid_targets_mean": 3896.1,
|
|
"valid_targets_min": 747
|
|
},
|
|
{
|
|
"epoch": 5.519169329073483,
|
|
"grad_norm": 0.5937799295103032,
|
|
"learning_rate": 5.22234788362447e-06,
|
|
"loss": 0.1523,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1442209631204605,
|
|
"step": 3455,
|
|
"valid_targets_mean": 4732.0,
|
|
"valid_targets_min": 2590
|
|
},
|
|
{
|
|
"epoch": 5.527156549520766,
|
|
"grad_norm": 0.45301538134789104,
|
|
"learning_rate": 5.168777424268454e-06,
|
|
"loss": 0.1707,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15300558507442474,
|
|
"step": 3460,
|
|
"valid_targets_mean": 5012.1,
|
|
"valid_targets_min": 1063
|
|
},
|
|
{
|
|
"epoch": 5.535143769968051,
|
|
"grad_norm": 0.5152594689501937,
|
|
"learning_rate": 5.115442341129171e-06,
|
|
"loss": 0.1646,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18040233850479126,
|
|
"step": 3465,
|
|
"valid_targets_mean": 4254.9,
|
|
"valid_targets_min": 2080
|
|
},
|
|
{
|
|
"epoch": 5.543130990415335,
|
|
"grad_norm": 0.49154076018681375,
|
|
"learning_rate": 5.062343480651364e-06,
|
|
"loss": 0.1674,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.161627858877182,
|
|
"step": 3470,
|
|
"valid_targets_mean": 4199.4,
|
|
"valid_targets_min": 641
|
|
},
|
|
{
|
|
"epoch": 5.55111821086262,
|
|
"grad_norm": 0.48941128766239755,
|
|
"learning_rate": 5.009481685530817e-06,
|
|
"loss": 0.148,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15404511988162994,
|
|
"step": 3475,
|
|
"valid_targets_mean": 4560.2,
|
|
"valid_targets_min": 1653
|
|
},
|
|
{
|
|
"epoch": 5.559105431309904,
|
|
"grad_norm": 0.525948500348585,
|
|
"learning_rate": 4.956857794701026e-06,
|
|
"loss": 0.1695,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1696540117263794,
|
|
"step": 3480,
|
|
"valid_targets_mean": 3846.1,
|
|
"valid_targets_min": 625
|
|
},
|
|
{
|
|
"epoch": 5.567092651757188,
|
|
"grad_norm": 0.49936612277457104,
|
|
"learning_rate": 4.904472643319873e-06,
|
|
"loss": 0.169,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15820062160491943,
|
|
"step": 3485,
|
|
"valid_targets_mean": 4445.6,
|
|
"valid_targets_min": 1523
|
|
},
|
|
{
|
|
"epoch": 5.575079872204473,
|
|
"grad_norm": 0.4729205139876077,
|
|
"learning_rate": 4.852327062756352e-06,
|
|
"loss": 0.1664,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16825735569000244,
|
|
"step": 3490,
|
|
"valid_targets_mean": 4923.2,
|
|
"valid_targets_min": 1543
|
|
},
|
|
{
|
|
"epoch": 5.583067092651757,
|
|
"grad_norm": 0.4560925106569269,
|
|
"learning_rate": 4.800421880577411e-06,
|
|
"loss": 0.1719,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16373009979724884,
|
|
"step": 3495,
|
|
"valid_targets_mean": 5574.2,
|
|
"valid_targets_min": 1357
|
|
},
|
|
{
|
|
"epoch": 5.5910543130990416,
|
|
"grad_norm": 0.466756243293937,
|
|
"learning_rate": 4.748757920534779e-06,
|
|
"loss": 0.1776,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19758476316928864,
|
|
"step": 3500,
|
|
"valid_targets_mean": 5720.0,
|
|
"valid_targets_min": 1298
|
|
},
|
|
{
|
|
"epoch": 5.599041533546326,
|
|
"grad_norm": 0.5396498066660705,
|
|
"learning_rate": 4.697336002551947e-06,
|
|
"loss": 0.1684,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19277673959732056,
|
|
"step": 3505,
|
|
"valid_targets_mean": 6450.8,
|
|
"valid_targets_min": 2080
|
|
},
|
|
{
|
|
"epoch": 5.60702875399361,
|
|
"grad_norm": 0.4782218522459289,
|
|
"learning_rate": 4.6461569427110684e-06,
|
|
"loss": 0.1601,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16561949253082275,
|
|
"step": 3510,
|
|
"valid_targets_mean": 5124.3,
|
|
"valid_targets_min": 1392
|
|
},
|
|
{
|
|
"epoch": 5.615015974440895,
|
|
"grad_norm": 0.5304339963977932,
|
|
"learning_rate": 4.5952215532401146e-06,
|
|
"loss": 0.1653,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16678208112716675,
|
|
"step": 3515,
|
|
"valid_targets_mean": 4472.6,
|
|
"valid_targets_min": 1156
|
|
},
|
|
{
|
|
"epoch": 5.623003194888179,
|
|
"grad_norm": 0.47151339401839326,
|
|
"learning_rate": 4.544530642499894e-06,
|
|
"loss": 0.1683,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16763369739055634,
|
|
"step": 3520,
|
|
"valid_targets_mean": 5364.7,
|
|
"valid_targets_min": 614
|
|
},
|
|
{
|
|
"epoch": 5.6309904153354635,
|
|
"grad_norm": 0.5501456164083334,
|
|
"learning_rate": 4.4940850149712765e-06,
|
|
"loss": 0.161,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1649385541677475,
|
|
"step": 3525,
|
|
"valid_targets_mean": 4224.4,
|
|
"valid_targets_min": 1217
|
|
},
|
|
{
|
|
"epoch": 5.638977635782748,
|
|
"grad_norm": 0.5241537472573441,
|
|
"learning_rate": 4.443885471242418e-06,
|
|
"loss": 0.1884,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24142885208129883,
|
|
"step": 3530,
|
|
"valid_targets_mean": 4711.8,
|
|
"valid_targets_min": 852
|
|
},
|
|
{
|
|
"epoch": 5.646964856230032,
|
|
"grad_norm": 0.5267524806152389,
|
|
"learning_rate": 4.393932807996017e-06,
|
|
"loss": 0.1605,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16665439307689667,
|
|
"step": 3535,
|
|
"valid_targets_mean": 5355.3,
|
|
"valid_targets_min": 2521
|
|
},
|
|
{
|
|
"epoch": 5.654952076677317,
|
|
"grad_norm": 0.5167579479066717,
|
|
"learning_rate": 4.344227817996735e-06,
|
|
"loss": 0.1614,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22691744565963745,
|
|
"step": 3540,
|
|
"valid_targets_mean": 5392.4,
|
|
"valid_targets_min": 1781
|
|
},
|
|
{
|
|
"epoch": 5.6629392971246,
|
|
"grad_norm": 0.5392095499369923,
|
|
"learning_rate": 4.294771290078548e-06,
|
|
"loss": 0.1574,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1580587923526764,
|
|
"step": 3545,
|
|
"valid_targets_mean": 4370.6,
|
|
"valid_targets_min": 1450
|
|
},
|
|
{
|
|
"epoch": 5.6709265175718855,
|
|
"grad_norm": 0.4500098175323538,
|
|
"learning_rate": 4.245564009132293e-06,
|
|
"loss": 0.1577,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13657471537590027,
|
|
"step": 3550,
|
|
"valid_targets_mean": 5291.3,
|
|
"valid_targets_min": 974
|
|
},
|
|
{
|
|
"epoch": 5.678913738019169,
|
|
"grad_norm": 0.4882435750022714,
|
|
"learning_rate": 4.196606756093138e-06,
|
|
"loss": 0.1635,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14333054423332214,
|
|
"step": 3555,
|
|
"valid_targets_mean": 5113.7,
|
|
"valid_targets_min": 1822
|
|
},
|
|
{
|
|
"epoch": 5.686900958466453,
|
|
"grad_norm": 0.539438359838628,
|
|
"learning_rate": 4.147900307928268e-06,
|
|
"loss": 0.1567,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15099304914474487,
|
|
"step": 3560,
|
|
"valid_targets_mean": 4687.0,
|
|
"valid_targets_min": 1194
|
|
},
|
|
{
|
|
"epoch": 5.694888178913738,
|
|
"grad_norm": 0.5099354746112605,
|
|
"learning_rate": 4.099445437624487e-06,
|
|
"loss": 0.1629,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17994922399520874,
|
|
"step": 3565,
|
|
"valid_targets_mean": 4840.9,
|
|
"valid_targets_min": 508
|
|
},
|
|
{
|
|
"epoch": 5.702875399361022,
|
|
"grad_norm": 0.4910940595601601,
|
|
"learning_rate": 4.051242914175995e-06,
|
|
"loss": 0.1649,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16600370407104492,
|
|
"step": 3570,
|
|
"valid_targets_mean": 5274.4,
|
|
"valid_targets_min": 568
|
|
},
|
|
{
|
|
"epoch": 5.710862619808307,
|
|
"grad_norm": 0.4937393027275975,
|
|
"learning_rate": 4.003293502572163e-06,
|
|
"loss": 0.1638,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18773019313812256,
|
|
"step": 3575,
|
|
"valid_targets_mean": 4499.5,
|
|
"valid_targets_min": 420
|
|
},
|
|
{
|
|
"epoch": 5.718849840255591,
|
|
"grad_norm": 0.44621833560642143,
|
|
"learning_rate": 3.955597963785391e-06,
|
|
"loss": 0.148,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15206709504127502,
|
|
"step": 3580,
|
|
"valid_targets_mean": 5360.9,
|
|
"valid_targets_min": 773
|
|
},
|
|
{
|
|
"epoch": 5.726837060702875,
|
|
"grad_norm": 0.46879550301828304,
|
|
"learning_rate": 3.908157054759048e-06,
|
|
"loss": 0.169,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15934661030769348,
|
|
"step": 3585,
|
|
"valid_targets_mean": 5322.0,
|
|
"valid_targets_min": 909
|
|
},
|
|
{
|
|
"epoch": 5.73482428115016,
|
|
"grad_norm": 0.4680808523084378,
|
|
"learning_rate": 3.860971528395427e-06,
|
|
"loss": 0.1832,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17729443311691284,
|
|
"step": 3590,
|
|
"valid_targets_mean": 5461.9,
|
|
"valid_targets_min": 2250
|
|
},
|
|
{
|
|
"epoch": 5.742811501597444,
|
|
"grad_norm": 0.5267451694944348,
|
|
"learning_rate": 3.814042133543847e-06,
|
|
"loss": 0.1626,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16615736484527588,
|
|
"step": 3595,
|
|
"valid_targets_mean": 3859.7,
|
|
"valid_targets_min": 646
|
|
},
|
|
{
|
|
"epoch": 5.7507987220447285,
|
|
"grad_norm": 0.514060632798016,
|
|
"learning_rate": 3.7673696149887117e-06,
|
|
"loss": 0.1663,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16031639277935028,
|
|
"step": 3600,
|
|
"valid_targets_mean": 5673.9,
|
|
"valid_targets_min": 3182
|
|
},
|
|
{
|
|
"epoch": 5.758785942492013,
|
|
"grad_norm": 0.49974165896330014,
|
|
"learning_rate": 3.72095471343773e-06,
|
|
"loss": 0.1657,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18452566862106323,
|
|
"step": 3605,
|
|
"valid_targets_mean": 4920.8,
|
|
"valid_targets_min": 1172
|
|
},
|
|
{
|
|
"epoch": 5.766773162939297,
|
|
"grad_norm": 0.6833706381129956,
|
|
"learning_rate": 3.674798165510136e-06,
|
|
"loss": 0.156,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17326435446739197,
|
|
"step": 3610,
|
|
"valid_targets_mean": 3820.6,
|
|
"valid_targets_min": 336
|
|
},
|
|
{
|
|
"epoch": 5.774760383386582,
|
|
"grad_norm": 0.48946150619688195,
|
|
"learning_rate": 3.6289007037250244e-06,
|
|
"loss": 0.1663,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1674729883670807,
|
|
"step": 3615,
|
|
"valid_targets_mean": 4756.6,
|
|
"valid_targets_min": 546
|
|
},
|
|
{
|
|
"epoch": 5.782747603833866,
|
|
"grad_norm": 0.6981098068418168,
|
|
"learning_rate": 3.5832630564897073e-06,
|
|
"loss": 0.1674,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14979082345962524,
|
|
"step": 3620,
|
|
"valid_targets_mean": 6019.2,
|
|
"valid_targets_min": 1232
|
|
},
|
|
{
|
|
"epoch": 5.7907348242811505,
|
|
"grad_norm": 0.4415917414218253,
|
|
"learning_rate": 3.5378859480881443e-06,
|
|
"loss": 0.1726,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1633869707584381,
|
|
"step": 3625,
|
|
"valid_targets_mean": 5033.4,
|
|
"valid_targets_min": 1642
|
|
},
|
|
{
|
|
"epoch": 5.798722044728435,
|
|
"grad_norm": 0.5082100658653667,
|
|
"learning_rate": 3.492770098669478e-06,
|
|
"loss": 0.1623,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17087781429290771,
|
|
"step": 3630,
|
|
"valid_targets_mean": 4868.6,
|
|
"valid_targets_min": 2039
|
|
},
|
|
{
|
|
"epoch": 5.806709265175719,
|
|
"grad_norm": 0.4705356790307018,
|
|
"learning_rate": 3.4479162242365717e-06,
|
|
"loss": 0.1571,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1566528081893921,
|
|
"step": 3635,
|
|
"valid_targets_mean": 5087.8,
|
|
"valid_targets_min": 428
|
|
},
|
|
{
|
|
"epoch": 5.814696485623003,
|
|
"grad_norm": 0.5289220901032725,
|
|
"learning_rate": 3.403325036634679e-06,
|
|
"loss": 0.1689,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16809199750423431,
|
|
"step": 3640,
|
|
"valid_targets_mean": 4230.8,
|
|
"valid_targets_min": 840
|
|
},
|
|
{
|
|
"epoch": 5.822683706070287,
|
|
"grad_norm": 0.4958668832343072,
|
|
"learning_rate": 3.3589972435401184e-06,
|
|
"loss": 0.1516,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16835638880729675,
|
|
"step": 3645,
|
|
"valid_targets_mean": 4882.5,
|
|
"valid_targets_min": 2241
|
|
},
|
|
{
|
|
"epoch": 5.830670926517572,
|
|
"grad_norm": 0.48811905886136925,
|
|
"learning_rate": 3.3149335484490553e-06,
|
|
"loss": 0.174,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16630032658576965,
|
|
"step": 3650,
|
|
"valid_targets_mean": 5152.8,
|
|
"valid_targets_min": 562
|
|
},
|
|
{
|
|
"epoch": 5.838658146964856,
|
|
"grad_norm": 0.512069651873867,
|
|
"learning_rate": 3.2711346506663346e-06,
|
|
"loss": 0.1692,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15492770075798035,
|
|
"step": 3655,
|
|
"valid_targets_mean": 3918.2,
|
|
"valid_targets_min": 1267
|
|
},
|
|
{
|
|
"epoch": 5.84664536741214,
|
|
"grad_norm": 0.4683577080630245,
|
|
"learning_rate": 3.2276012452943893e-06,
|
|
"loss": 0.1515,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13782991468906403,
|
|
"step": 3660,
|
|
"valid_targets_mean": 5204.9,
|
|
"valid_targets_min": 1990
|
|
},
|
|
{
|
|
"epoch": 5.854632587859425,
|
|
"grad_norm": 0.4545248711358778,
|
|
"learning_rate": 3.1843340232222064e-06,
|
|
"loss": 0.1676,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14945927262306213,
|
|
"step": 3665,
|
|
"valid_targets_mean": 4706.4,
|
|
"valid_targets_min": 1169
|
|
},
|
|
{
|
|
"epoch": 5.862619808306709,
|
|
"grad_norm": 0.45995293124692316,
|
|
"learning_rate": 3.1413336711143437e-06,
|
|
"loss": 0.1617,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14870786666870117,
|
|
"step": 3670,
|
|
"valid_targets_mean": 4540.5,
|
|
"valid_targets_min": 1963
|
|
},
|
|
{
|
|
"epoch": 5.8706070287539935,
|
|
"grad_norm": 0.47027428301454915,
|
|
"learning_rate": 3.0986008714000703e-06,
|
|
"loss": 0.1646,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19357925653457642,
|
|
"step": 3675,
|
|
"valid_targets_mean": 5395.3,
|
|
"valid_targets_min": 661
|
|
},
|
|
{
|
|
"epoch": 5.878594249201278,
|
|
"grad_norm": 0.47417060520818427,
|
|
"learning_rate": 3.056136302262489e-06,
|
|
"loss": 0.1584,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17412704229354858,
|
|
"step": 3680,
|
|
"valid_targets_mean": 5844.1,
|
|
"valid_targets_min": 2300
|
|
},
|
|
{
|
|
"epoch": 5.886581469648562,
|
|
"grad_norm": 0.4974733331740804,
|
|
"learning_rate": 3.0139406376278212e-06,
|
|
"loss": 0.1649,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16733218729496002,
|
|
"step": 3685,
|
|
"valid_targets_mean": 4846.6,
|
|
"valid_targets_min": 936
|
|
},
|
|
{
|
|
"epoch": 5.894568690095847,
|
|
"grad_norm": 0.5231984609118597,
|
|
"learning_rate": 2.972014547154671e-06,
|
|
"loss": 0.1609,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1489640772342682,
|
|
"step": 3690,
|
|
"valid_targets_mean": 4862.9,
|
|
"valid_targets_min": 651
|
|
},
|
|
{
|
|
"epoch": 5.902555910543131,
|
|
"grad_norm": 0.4859221837793371,
|
|
"learning_rate": 2.930358696223423e-06,
|
|
"loss": 0.1682,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1759262979030609,
|
|
"step": 3695,
|
|
"valid_targets_mean": 5368.2,
|
|
"valid_targets_min": 899
|
|
},
|
|
{
|
|
"epoch": 5.9105431309904155,
|
|
"grad_norm": 0.47077473092171224,
|
|
"learning_rate": 2.8889737459256695e-06,
|
|
"loss": 0.1664,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16548126935958862,
|
|
"step": 3700,
|
|
"valid_targets_mean": 5086.6,
|
|
"valid_targets_min": 784
|
|
},
|
|
{
|
|
"epoch": 5.9185303514377,
|
|
"grad_norm": 0.49432424854437196,
|
|
"learning_rate": 2.8478603530537285e-06,
|
|
"loss": 0.1659,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15692386031150818,
|
|
"step": 3705,
|
|
"valid_targets_mean": 4616.9,
|
|
"valid_targets_min": 1698
|
|
},
|
|
{
|
|
"epoch": 5.926517571884984,
|
|
"grad_norm": 0.49861584831517203,
|
|
"learning_rate": 2.8070191700902194e-06,
|
|
"loss": 0.1692,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14717496931552887,
|
|
"step": 3710,
|
|
"valid_targets_mean": 4043.9,
|
|
"valid_targets_min": 358
|
|
},
|
|
{
|
|
"epoch": 5.934504792332269,
|
|
"grad_norm": 0.5227096968437029,
|
|
"learning_rate": 2.7664508451976903e-06,
|
|
"loss": 0.1601,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14311034977436066,
|
|
"step": 3715,
|
|
"valid_targets_mean": 5310.3,
|
|
"valid_targets_min": 822
|
|
},
|
|
{
|
|
"epoch": 5.942492012779553,
|
|
"grad_norm": 0.5401942371530758,
|
|
"learning_rate": 2.726156022208362e-06,
|
|
"loss": 0.1681,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19872379302978516,
|
|
"step": 3720,
|
|
"valid_targets_mean": 4996.3,
|
|
"valid_targets_min": 1360
|
|
},
|
|
{
|
|
"epoch": 5.950479233226837,
|
|
"grad_norm": 0.5036722120094557,
|
|
"learning_rate": 2.6861353406138713e-06,
|
|
"loss": 0.1653,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18039855360984802,
|
|
"step": 3725,
|
|
"valid_targets_mean": 5059.4,
|
|
"valid_targets_min": 835
|
|
},
|
|
{
|
|
"epoch": 5.958466453674122,
|
|
"grad_norm": 0.4768709414023673,
|
|
"learning_rate": 2.646389435555172e-06,
|
|
"loss": 0.1705,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16467642784118652,
|
|
"step": 3730,
|
|
"valid_targets_mean": 4775.9,
|
|
"valid_targets_min": 1113
|
|
},
|
|
{
|
|
"epoch": 5.966453674121405,
|
|
"grad_norm": 0.4505152766422281,
|
|
"learning_rate": 2.6069189378124015e-06,
|
|
"loss": 0.1686,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1375581920146942,
|
|
"step": 3735,
|
|
"valid_targets_mean": 5153.4,
|
|
"valid_targets_min": 1586
|
|
},
|
|
{
|
|
"epoch": 5.97444089456869,
|
|
"grad_norm": 0.6295109583708369,
|
|
"learning_rate": 2.567724473794908e-06,
|
|
"loss": 0.177,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18594983220100403,
|
|
"step": 3740,
|
|
"valid_targets_mean": 3786.9,
|
|
"valid_targets_min": 683
|
|
},
|
|
{
|
|
"epoch": 5.982428115015974,
|
|
"grad_norm": 0.4200857765813741,
|
|
"learning_rate": 2.5288066655312914e-06,
|
|
"loss": 0.1649,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.163049578666687,
|
|
"step": 3745,
|
|
"valid_targets_mean": 6365.8,
|
|
"valid_targets_min": 2095
|
|
},
|
|
{
|
|
"epoch": 5.9904153354632586,
|
|
"grad_norm": 0.6254325903030274,
|
|
"learning_rate": 2.4901661306595414e-06,
|
|
"loss": 0.1744,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17417556047439575,
|
|
"step": 3750,
|
|
"valid_targets_mean": 4794.1,
|
|
"valid_targets_min": 501
|
|
},
|
|
{
|
|
"epoch": 5.998402555910543,
|
|
"grad_norm": 0.5264492916089583,
|
|
"learning_rate": 2.451803482417234e-06,
|
|
"loss": 0.1722,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14926278591156006,
|
|
"step": 3755,
|
|
"valid_targets_mean": 4241.8,
|
|
"valid_targets_min": 1562
|
|
},
|
|
{
|
|
"epoch": 6.006389776357827,
|
|
"grad_norm": 0.49203753066718664,
|
|
"learning_rate": 2.413719329631785e-06,
|
|
"loss": 0.1527,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15311649441719055,
|
|
"step": 3760,
|
|
"valid_targets_mean": 4905.8,
|
|
"valid_targets_min": 1987
|
|
},
|
|
{
|
|
"epoch": 6.014376996805112,
|
|
"grad_norm": 0.5951326896601777,
|
|
"learning_rate": 2.375914276710811e-06,
|
|
"loss": 0.1635,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16578024625778198,
|
|
"step": 3765,
|
|
"valid_targets_mean": 4779.8,
|
|
"valid_targets_min": 1100
|
|
},
|
|
{
|
|
"epoch": 6.022364217252396,
|
|
"grad_norm": 0.48183368867035825,
|
|
"learning_rate": 2.338388923632513e-06,
|
|
"loss": 0.1723,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16086134314537048,
|
|
"step": 3770,
|
|
"valid_targets_mean": 4555.1,
|
|
"valid_targets_min": 680
|
|
},
|
|
{
|
|
"epoch": 6.0303514376996805,
|
|
"grad_norm": 0.4991514794168638,
|
|
"learning_rate": 2.3011438659361794e-06,
|
|
"loss": 0.1461,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15352702140808105,
|
|
"step": 3775,
|
|
"valid_targets_mean": 5390.8,
|
|
"valid_targets_min": 2574
|
|
},
|
|
{
|
|
"epoch": 6.038338658146965,
|
|
"grad_norm": 0.5004302788098899,
|
|
"learning_rate": 2.2641796947127114e-06,
|
|
"loss": 0.1715,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16071294248104095,
|
|
"step": 3780,
|
|
"valid_targets_mean": 5227.7,
|
|
"valid_targets_min": 1983
|
|
},
|
|
{
|
|
"epoch": 6.046325878594249,
|
|
"grad_norm": 0.47919983329397353,
|
|
"learning_rate": 2.2274969965952553e-06,
|
|
"loss": 0.1402,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14981648325920105,
|
|
"step": 3785,
|
|
"valid_targets_mean": 4976.9,
|
|
"valid_targets_min": 2021
|
|
},
|
|
{
|
|
"epoch": 6.054313099041534,
|
|
"grad_norm": 0.47542744449520763,
|
|
"learning_rate": 2.1910963537498887e-06,
|
|
"loss": 0.1541,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14725175499916077,
|
|
"step": 3790,
|
|
"valid_targets_mean": 4814.4,
|
|
"valid_targets_min": 690
|
|
},
|
|
{
|
|
"epoch": 6.062300319488818,
|
|
"grad_norm": 0.4842460038834621,
|
|
"learning_rate": 2.1549783438663872e-06,
|
|
"loss": 0.1615,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1420721560716629,
|
|
"step": 3795,
|
|
"valid_targets_mean": 5436.4,
|
|
"valid_targets_min": 1793
|
|
},
|
|
{
|
|
"epoch": 6.0702875399361025,
|
|
"grad_norm": 0.47140984563889077,
|
|
"learning_rate": 2.1191435401490534e-06,
|
|
"loss": 0.1655,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1739579290151596,
|
|
"step": 3800,
|
|
"valid_targets_mean": 5437.5,
|
|
"valid_targets_min": 919
|
|
},
|
|
{
|
|
"epoch": 6.078274760383387,
|
|
"grad_norm": 0.7446969509152094,
|
|
"learning_rate": 2.0835925113076062e-06,
|
|
"loss": 0.1566,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16405190527439117,
|
|
"step": 3805,
|
|
"valid_targets_mean": 5858.0,
|
|
"valid_targets_min": 1543
|
|
},
|
|
{
|
|
"epoch": 6.086261980830671,
|
|
"grad_norm": 0.4456085804468523,
|
|
"learning_rate": 2.0483258215481784e-06,
|
|
"loss": 0.1574,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14001686871051788,
|
|
"step": 3810,
|
|
"valid_targets_mean": 5576.1,
|
|
"valid_targets_min": 998
|
|
},
|
|
{
|
|
"epoch": 6.094249201277956,
|
|
"grad_norm": 0.4581135228921414,
|
|
"learning_rate": 2.0133440305643413e-06,
|
|
"loss": 0.1482,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1370004415512085,
|
|
"step": 3815,
|
|
"valid_targets_mean": 4656.1,
|
|
"valid_targets_min": 1325
|
|
},
|
|
{
|
|
"epoch": 6.102236421725239,
|
|
"grad_norm": 0.4478316277948835,
|
|
"learning_rate": 1.9786476935282463e-06,
|
|
"loss": 0.1559,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14399506151676178,
|
|
"step": 3820,
|
|
"valid_targets_mean": 5733.0,
|
|
"valid_targets_min": 1816
|
|
},
|
|
{
|
|
"epoch": 6.110223642172524,
|
|
"grad_norm": 0.445274051117867,
|
|
"learning_rate": 1.944237361081782e-06,
|
|
"loss": 0.1586,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14635354280471802,
|
|
"step": 3825,
|
|
"valid_targets_mean": 5607.6,
|
|
"valid_targets_min": 428
|
|
},
|
|
{
|
|
"epoch": 6.118210862619808,
|
|
"grad_norm": 0.6415233850020196,
|
|
"learning_rate": 1.9101135793278746e-06,
|
|
"loss": 0.1646,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1825276017189026,
|
|
"step": 3830,
|
|
"valid_targets_mean": 4352.9,
|
|
"valid_targets_min": 329
|
|
},
|
|
{
|
|
"epoch": 6.126198083067092,
|
|
"grad_norm": 0.5276389299287585,
|
|
"learning_rate": 1.8762768898217732e-06,
|
|
"loss": 0.1634,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1704082190990448,
|
|
"step": 3835,
|
|
"valid_targets_mean": 4400.9,
|
|
"valid_targets_min": 942
|
|
},
|
|
{
|
|
"epoch": 6.134185303514377,
|
|
"grad_norm": 0.55009561912505,
|
|
"learning_rate": 1.8427278295625006e-06,
|
|
"loss": 0.1743,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19751335680484772,
|
|
"step": 3840,
|
|
"valid_targets_mean": 4488.4,
|
|
"valid_targets_min": 896
|
|
},
|
|
{
|
|
"epoch": 6.142172523961661,
|
|
"grad_norm": 0.48859746009960264,
|
|
"learning_rate": 1.8094669309843161e-06,
|
|
"loss": 0.1552,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12780989706516266,
|
|
"step": 3845,
|
|
"valid_targets_mean": 3926.5,
|
|
"valid_targets_min": 1012
|
|
},
|
|
{
|
|
"epoch": 6.1501597444089455,
|
|
"grad_norm": 0.5607367725075801,
|
|
"learning_rate": 1.776494721948241e-06,
|
|
"loss": 0.1548,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14702245593070984,
|
|
"step": 3850,
|
|
"valid_targets_mean": 4336.1,
|
|
"valid_targets_min": 1453
|
|
},
|
|
{
|
|
"epoch": 6.15814696485623,
|
|
"grad_norm": 0.43671641575643083,
|
|
"learning_rate": 1.7438117257337239e-06,
|
|
"loss": 0.1485,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16328737139701843,
|
|
"step": 3855,
|
|
"valid_targets_mean": 5713.7,
|
|
"valid_targets_min": 1510
|
|
},
|
|
{
|
|
"epoch": 6.166134185303514,
|
|
"grad_norm": 0.4725994169019507,
|
|
"learning_rate": 1.7114184610302964e-06,
|
|
"loss": 0.1561,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16375653445720673,
|
|
"step": 3860,
|
|
"valid_targets_mean": 5364.8,
|
|
"valid_targets_min": 1867
|
|
},
|
|
{
|
|
"epoch": 6.174121405750799,
|
|
"grad_norm": 0.43920140069675684,
|
|
"learning_rate": 1.67931544192937e-06,
|
|
"loss": 0.1551,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1650637686252594,
|
|
"step": 3865,
|
|
"valid_targets_mean": 5834.2,
|
|
"valid_targets_min": 2987
|
|
},
|
|
{
|
|
"epoch": 6.182108626198083,
|
|
"grad_norm": 0.6998947577393563,
|
|
"learning_rate": 1.6475031779160611e-06,
|
|
"loss": 0.1602,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16513022780418396,
|
|
"step": 3870,
|
|
"valid_targets_mean": 5041.4,
|
|
"valid_targets_min": 862
|
|
},
|
|
{
|
|
"epoch": 6.1900958466453675,
|
|
"grad_norm": 0.5083668808608908,
|
|
"learning_rate": 1.6159821738611192e-06,
|
|
"loss": 0.1778,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1575690358877182,
|
|
"step": 3875,
|
|
"valid_targets_mean": 4826.8,
|
|
"valid_targets_min": 519
|
|
},
|
|
{
|
|
"epoch": 6.198083067092652,
|
|
"grad_norm": 0.4593445751109819,
|
|
"learning_rate": 1.5847529300128827e-06,
|
|
"loss": 0.1686,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16691654920578003,
|
|
"step": 3880,
|
|
"valid_targets_mean": 5452.4,
|
|
"valid_targets_min": 1090
|
|
},
|
|
{
|
|
"epoch": 6.206070287539936,
|
|
"grad_norm": 0.5422017808243945,
|
|
"learning_rate": 1.5538159419893895e-06,
|
|
"loss": 0.1559,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15538769960403442,
|
|
"step": 3885,
|
|
"valid_targets_mean": 4313.0,
|
|
"valid_targets_min": 1022
|
|
},
|
|
{
|
|
"epoch": 6.214057507987221,
|
|
"grad_norm": 0.5347913106039871,
|
|
"learning_rate": 1.5231717007704738e-06,
|
|
"loss": 0.1475,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1320032775402069,
|
|
"step": 3890,
|
|
"valid_targets_mean": 4379.3,
|
|
"valid_targets_min": 2080
|
|
},
|
|
{
|
|
"epoch": 6.222044728434505,
|
|
"grad_norm": 0.4474745812734044,
|
|
"learning_rate": 1.4928206926899801e-06,
|
|
"loss": 0.1687,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14763572812080383,
|
|
"step": 3895,
|
|
"valid_targets_mean": 5836.9,
|
|
"valid_targets_min": 2325
|
|
},
|
|
{
|
|
"epoch": 6.2300319488817895,
|
|
"grad_norm": 0.47433310434755555,
|
|
"learning_rate": 1.4627633994280599e-06,
|
|
"loss": 0.1547,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1461184024810791,
|
|
"step": 3900,
|
|
"valid_targets_mean": 5509.9,
|
|
"valid_targets_min": 865
|
|
},
|
|
{
|
|
"epoch": 6.238019169329074,
|
|
"grad_norm": 0.4529194089949568,
|
|
"learning_rate": 1.433000298003504e-06,
|
|
"loss": 0.1537,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1502389758825302,
|
|
"step": 3905,
|
|
"valid_targets_mean": 6055.1,
|
|
"valid_targets_min": 2880
|
|
},
|
|
{
|
|
"epoch": 6.246006389776358,
|
|
"grad_norm": 0.5221343423400612,
|
|
"learning_rate": 1.4035318607662029e-06,
|
|
"loss": 0.1716,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18835774064064026,
|
|
"step": 3910,
|
|
"valid_targets_mean": 5647.4,
|
|
"valid_targets_min": 557
|
|
},
|
|
{
|
|
"epoch": 6.253993610223642,
|
|
"grad_norm": 0.4696728514223722,
|
|
"learning_rate": 1.3743585553896144e-06,
|
|
"loss": 0.1643,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16331344842910767,
|
|
"step": 3915,
|
|
"valid_targets_mean": 5139.0,
|
|
"valid_targets_min": 802
|
|
},
|
|
{
|
|
"epoch": 6.261980830670926,
|
|
"grad_norm": 0.5698993865428698,
|
|
"learning_rate": 1.345480844863376e-06,
|
|
"loss": 0.1615,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1860525906085968,
|
|
"step": 3920,
|
|
"valid_targets_mean": 4267.4,
|
|
"valid_targets_min": 1880
|
|
},
|
|
{
|
|
"epoch": 6.2699680511182105,
|
|
"grad_norm": 0.5391313982087154,
|
|
"learning_rate": 1.316899187485925e-06,
|
|
"loss": 0.1622,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1673540621995926,
|
|
"step": 3925,
|
|
"valid_targets_mean": 5031.1,
|
|
"valid_targets_min": 1518
|
|
},
|
|
{
|
|
"epoch": 6.277955271565495,
|
|
"grad_norm": 0.5119925703283522,
|
|
"learning_rate": 1.288614036857252e-06,
|
|
"loss": 0.1676,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15239587426185608,
|
|
"step": 3930,
|
|
"valid_targets_mean": 4621.6,
|
|
"valid_targets_min": 2205
|
|
},
|
|
{
|
|
"epoch": 6.285942492012779,
|
|
"grad_norm": 0.5002344474981675,
|
|
"learning_rate": 1.260625841871692e-06,
|
|
"loss": 0.1571,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17312490940093994,
|
|
"step": 3935,
|
|
"valid_targets_mean": 4581.0,
|
|
"valid_targets_min": 952
|
|
},
|
|
{
|
|
"epoch": 6.293929712460064,
|
|
"grad_norm": 0.5471905188175354,
|
|
"learning_rate": 1.2329350467107925e-06,
|
|
"loss": 0.1853,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1913260966539383,
|
|
"step": 3940,
|
|
"valid_targets_mean": 4414.9,
|
|
"valid_targets_min": 1948
|
|
},
|
|
{
|
|
"epoch": 6.301916932907348,
|
|
"grad_norm": 0.5928095486490488,
|
|
"learning_rate": 1.2055420908362781e-06,
|
|
"loss": 0.1697,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15695932507514954,
|
|
"step": 3945,
|
|
"valid_targets_mean": 5036.8,
|
|
"valid_targets_min": 587
|
|
},
|
|
{
|
|
"epoch": 6.3099041533546325,
|
|
"grad_norm": 0.7806745270379135,
|
|
"learning_rate": 1.1784474089830612e-06,
|
|
"loss": 0.1549,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.141230046749115,
|
|
"step": 3950,
|
|
"valid_targets_mean": 3909.8,
|
|
"valid_targets_min": 803
|
|
},
|
|
{
|
|
"epoch": 6.317891373801917,
|
|
"grad_norm": 0.5241502993360989,
|
|
"learning_rate": 1.1516514311523607e-06,
|
|
"loss": 0.1572,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1841631531715393,
|
|
"step": 3955,
|
|
"valid_targets_mean": 4971.7,
|
|
"valid_targets_min": 1007
|
|
},
|
|
{
|
|
"epoch": 6.325878594249201,
|
|
"grad_norm": 0.44293207636192655,
|
|
"learning_rate": 1.1251545826048593e-06,
|
|
"loss": 0.1627,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17101489007472992,
|
|
"step": 3960,
|
|
"valid_targets_mean": 5828.6,
|
|
"valid_targets_min": 2349
|
|
},
|
|
{
|
|
"epoch": 6.333865814696486,
|
|
"grad_norm": 0.5147967978992214,
|
|
"learning_rate": 1.098957283853972e-06,
|
|
"loss": 0.1649,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15055087208747864,
|
|
"step": 3965,
|
|
"valid_targets_mean": 4281.2,
|
|
"valid_targets_min": 813
|
|
},
|
|
{
|
|
"epoch": 6.34185303514377,
|
|
"grad_norm": 0.5053144291349435,
|
|
"learning_rate": 1.0730599506591476e-06,
|
|
"loss": 0.1705,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15987616777420044,
|
|
"step": 3970,
|
|
"valid_targets_mean": 4819.4,
|
|
"valid_targets_min": 656
|
|
},
|
|
{
|
|
"epoch": 6.3498402555910545,
|
|
"grad_norm": 0.49378772224517475,
|
|
"learning_rate": 1.0474629940192994e-06,
|
|
"loss": 0.1517,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17134281992912292,
|
|
"step": 3975,
|
|
"valid_targets_mean": 5614.7,
|
|
"valid_targets_min": 936
|
|
},
|
|
{
|
|
"epoch": 6.357827476038339,
|
|
"grad_norm": 0.4942524157645037,
|
|
"learning_rate": 1.0221668201662726e-06,
|
|
"loss": 0.1573,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1388753205537796,
|
|
"step": 3980,
|
|
"valid_targets_mean": 4453.8,
|
|
"valid_targets_min": 958
|
|
},
|
|
{
|
|
"epoch": 6.365814696485623,
|
|
"grad_norm": 0.47105231810783227,
|
|
"learning_rate": 9.971718305583767e-07,
|
|
"loss": 0.171,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1350238174200058,
|
|
"step": 3985,
|
|
"valid_targets_mean": 4756.4,
|
|
"valid_targets_min": 651
|
|
},
|
|
{
|
|
"epoch": 6.373801916932908,
|
|
"grad_norm": 0.4797868903591038,
|
|
"learning_rate": 9.724784218740524e-07,
|
|
"loss": 0.163,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18055656552314758,
|
|
"step": 3990,
|
|
"valid_targets_mean": 5385.4,
|
|
"valid_targets_min": 2538
|
|
},
|
|
{
|
|
"epoch": 6.381789137380192,
|
|
"grad_norm": 0.546317463292646,
|
|
"learning_rate": 9.480869860055364e-07,
|
|
"loss": 0.1485,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12972435355186462,
|
|
"step": 3995,
|
|
"valid_targets_mean": 5110.4,
|
|
"valid_targets_min": 828
|
|
},
|
|
{
|
|
"epoch": 6.389776357827476,
|
|
"grad_norm": 0.501652799871782,
|
|
"learning_rate": 9.239979100526763e-07,
|
|
"loss": 0.1544,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16283018887043,
|
|
"step": 4000,
|
|
"valid_targets_mean": 5847.4,
|
|
"valid_targets_min": 1265
|
|
},
|
|
{
|
|
"epoch": 6.397763578274761,
|
|
"grad_norm": 0.5123995362461974,
|
|
"learning_rate": 9.002115763167585e-07,
|
|
"loss": 0.1526,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16418376564979553,
|
|
"step": 4005,
|
|
"valid_targets_mean": 4781.9,
|
|
"valid_targets_min": 483
|
|
},
|
|
{
|
|
"epoch": 6.405750798722044,
|
|
"grad_norm": 0.5655064938108871,
|
|
"learning_rate": 8.7672836229447e-07,
|
|
"loss": 0.1577,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16947625577449799,
|
|
"step": 4010,
|
|
"valid_targets_mean": 4005.4,
|
|
"valid_targets_min": 791
|
|
},
|
|
{
|
|
"epoch": 6.413738019169329,
|
|
"grad_norm": 0.6651149588120067,
|
|
"learning_rate": 8.535486406718684e-07,
|
|
"loss": 0.1583,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16407115757465363,
|
|
"step": 4015,
|
|
"valid_targets_mean": 5528.8,
|
|
"valid_targets_min": 1297
|
|
},
|
|
{
|
|
"epoch": 6.421725239616613,
|
|
"grad_norm": 0.5062861693449161,
|
|
"learning_rate": 8.306727793185132e-07,
|
|
"loss": 0.1625,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15367011725902557,
|
|
"step": 4020,
|
|
"valid_targets_mean": 4609.2,
|
|
"valid_targets_min": 1156
|
|
},
|
|
{
|
|
"epoch": 6.4297124600638975,
|
|
"grad_norm": 0.5252283287798611,
|
|
"learning_rate": 8.081011412815965e-07,
|
|
"loss": 0.1662,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18086940050125122,
|
|
"step": 4025,
|
|
"valid_targets_mean": 4623.1,
|
|
"valid_targets_min": 797
|
|
},
|
|
{
|
|
"epoch": 6.437699680511182,
|
|
"grad_norm": 0.5181451559409599,
|
|
"learning_rate": 7.858340847801815e-07,
|
|
"loss": 0.1587,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.160737544298172,
|
|
"step": 4030,
|
|
"valid_targets_mean": 4243.2,
|
|
"valid_targets_min": 1583
|
|
},
|
|
{
|
|
"epoch": 6.445686900958466,
|
|
"grad_norm": 0.5212666018139555,
|
|
"learning_rate": 7.638719631995406e-07,
|
|
"loss": 0.1579,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14286597073078156,
|
|
"step": 4035,
|
|
"valid_targets_mean": 4253.1,
|
|
"valid_targets_min": 2188
|
|
},
|
|
{
|
|
"epoch": 6.453674121405751,
|
|
"grad_norm": 0.5004735693367742,
|
|
"learning_rate": 7.422151250855214e-07,
|
|
"loss": 0.148,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15612217783927917,
|
|
"step": 4040,
|
|
"valid_targets_mean": 4539.6,
|
|
"valid_targets_min": 1364
|
|
},
|
|
{
|
|
"epoch": 6.461661341853035,
|
|
"grad_norm": 0.5228627437968858,
|
|
"learning_rate": 7.208639141390295e-07,
|
|
"loss": 0.1553,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16546684503555298,
|
|
"step": 4045,
|
|
"valid_targets_mean": 4428.9,
|
|
"valid_targets_min": 819
|
|
},
|
|
{
|
|
"epoch": 6.4696485623003195,
|
|
"grad_norm": 0.44743730236732104,
|
|
"learning_rate": 6.998186692105657e-07,
|
|
"loss": 0.1544,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15039759874343872,
|
|
"step": 4050,
|
|
"valid_targets_mean": 5451.6,
|
|
"valid_targets_min": 1905
|
|
},
|
|
{
|
|
"epoch": 6.477635782747604,
|
|
"grad_norm": 0.4502056621596025,
|
|
"learning_rate": 6.790797242948644e-07,
|
|
"loss": 0.1606,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15037524700164795,
|
|
"step": 4055,
|
|
"valid_targets_mean": 5427.0,
|
|
"valid_targets_min": 1897
|
|
},
|
|
{
|
|
"epoch": 6.485623003194888,
|
|
"grad_norm": 0.4293053831010077,
|
|
"learning_rate": 6.5864740852557e-07,
|
|
"loss": 0.155,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16043220460414886,
|
|
"step": 4060,
|
|
"valid_targets_mean": 5719.5,
|
|
"valid_targets_min": 757
|
|
},
|
|
{
|
|
"epoch": 6.493610223642173,
|
|
"grad_norm": 0.47108992924603316,
|
|
"learning_rate": 6.385220461700248e-07,
|
|
"loss": 0.1647,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16741064190864563,
|
|
"step": 4065,
|
|
"valid_targets_mean": 5453.7,
|
|
"valid_targets_min": 1758
|
|
},
|
|
{
|
|
"epoch": 6.501597444089457,
|
|
"grad_norm": 0.47782793699178533,
|
|
"learning_rate": 6.187039566241337e-07,
|
|
"loss": 0.159,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1436358541250229,
|
|
"step": 4070,
|
|
"valid_targets_mean": 5147.2,
|
|
"valid_targets_min": 423
|
|
},
|
|
{
|
|
"epoch": 6.5095846645367414,
|
|
"grad_norm": 0.5097003346397011,
|
|
"learning_rate": 5.99193454407272e-07,
|
|
"loss": 0.1498,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.145539790391922,
|
|
"step": 4075,
|
|
"valid_targets_mean": 4628.7,
|
|
"valid_targets_min": 1347
|
|
},
|
|
{
|
|
"epoch": 6.517571884984026,
|
|
"grad_norm": 0.4885563890134189,
|
|
"learning_rate": 5.799908491573148e-07,
|
|
"loss": 0.1713,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16930031776428223,
|
|
"step": 4080,
|
|
"valid_targets_mean": 4541.6,
|
|
"valid_targets_min": 1679
|
|
},
|
|
{
|
|
"epoch": 6.52555910543131,
|
|
"grad_norm": 0.5156008309314725,
|
|
"learning_rate": 5.610964456257107e-07,
|
|
"loss": 0.1545,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18574655055999756,
|
|
"step": 4085,
|
|
"valid_targets_mean": 4505.2,
|
|
"valid_targets_min": 1407
|
|
},
|
|
{
|
|
"epoch": 6.533546325878595,
|
|
"grad_norm": 0.6193155974120058,
|
|
"learning_rate": 5.425105436726496e-07,
|
|
"loss": 0.1556,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1539543867111206,
|
|
"step": 4090,
|
|
"valid_targets_mean": 4921.3,
|
|
"valid_targets_min": 818
|
|
},
|
|
{
|
|
"epoch": 6.541533546325878,
|
|
"grad_norm": 0.48499335224731416,
|
|
"learning_rate": 5.242334382623004e-07,
|
|
"loss": 0.1637,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13866549730300903,
|
|
"step": 4095,
|
|
"valid_targets_mean": 5624.8,
|
|
"valid_targets_min": 1141
|
|
},
|
|
{
|
|
"epoch": 6.549520766773163,
|
|
"grad_norm": 0.49424624741619433,
|
|
"learning_rate": 5.062654194581429e-07,
|
|
"loss": 0.1553,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1799236238002777,
|
|
"step": 4100,
|
|
"valid_targets_mean": 4593.4,
|
|
"valid_targets_min": 1548
|
|
},
|
|
{
|
|
"epoch": 6.557507987220447,
|
|
"grad_norm": 0.44905955898824995,
|
|
"learning_rate": 4.88606772418343e-07,
|
|
"loss": 0.1509,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15588271617889404,
|
|
"step": 4105,
|
|
"valid_targets_mean": 5042.8,
|
|
"valid_targets_min": 1170
|
|
},
|
|
{
|
|
"epoch": 6.565495207667731,
|
|
"grad_norm": 0.520702776011052,
|
|
"learning_rate": 4.7125777739123857e-07,
|
|
"loss": 0.159,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20354077219963074,
|
|
"step": 4110,
|
|
"valid_targets_mean": 5167.0,
|
|
"valid_targets_min": 1373
|
|
},
|
|
{
|
|
"epoch": 6.573482428115016,
|
|
"grad_norm": 0.4435861783742549,
|
|
"learning_rate": 4.54218709710903e-07,
|
|
"loss": 0.14,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14058135449886322,
|
|
"step": 4115,
|
|
"valid_targets_mean": 5767.5,
|
|
"valid_targets_min": 2807
|
|
},
|
|
{
|
|
"epoch": 6.5814696485623,
|
|
"grad_norm": 0.552138536766462,
|
|
"learning_rate": 4.374898397927507e-07,
|
|
"loss": 0.1635,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17940273880958557,
|
|
"step": 4120,
|
|
"valid_targets_mean": 4240.6,
|
|
"valid_targets_min": 657
|
|
},
|
|
{
|
|
"epoch": 6.5894568690095845,
|
|
"grad_norm": 0.4934760675953949,
|
|
"learning_rate": 4.210714331292698e-07,
|
|
"loss": 0.1621,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14860069751739502,
|
|
"step": 4125,
|
|
"valid_targets_mean": 4598.2,
|
|
"valid_targets_min": 1648
|
|
},
|
|
{
|
|
"epoch": 6.597444089456869,
|
|
"grad_norm": 0.47557123063031764,
|
|
"learning_rate": 4.049637502857895e-07,
|
|
"loss": 0.1598,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16057202219963074,
|
|
"step": 4130,
|
|
"valid_targets_mean": 5109.9,
|
|
"valid_targets_min": 1470
|
|
},
|
|
{
|
|
"epoch": 6.605431309904153,
|
|
"grad_norm": 0.48824188715889366,
|
|
"learning_rate": 3.8916704689635707e-07,
|
|
"loss": 0.1522,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1606670618057251,
|
|
"step": 4135,
|
|
"valid_targets_mean": 4935.2,
|
|
"valid_targets_min": 827
|
|
},
|
|
{
|
|
"epoch": 6.613418530351438,
|
|
"grad_norm": 0.584193072096277,
|
|
"learning_rate": 3.736815736596766e-07,
|
|
"loss": 0.1567,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18314039707183838,
|
|
"step": 4140,
|
|
"valid_targets_mean": 4389.9,
|
|
"valid_targets_min": 455
|
|
},
|
|
{
|
|
"epoch": 6.621405750798722,
|
|
"grad_norm": 0.5239645121482677,
|
|
"learning_rate": 3.5850757633513424e-07,
|
|
"loss": 0.1515,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15261167287826538,
|
|
"step": 4145,
|
|
"valid_targets_mean": 3911.5,
|
|
"valid_targets_min": 723
|
|
},
|
|
{
|
|
"epoch": 6.6293929712460065,
|
|
"grad_norm": 0.49798914942498246,
|
|
"learning_rate": 3.4364529573888803e-07,
|
|
"loss": 0.1538,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1832718402147293,
|
|
"step": 4150,
|
|
"valid_targets_mean": 4998.7,
|
|
"valid_targets_min": 1378
|
|
},
|
|
{
|
|
"epoch": 6.637380191693291,
|
|
"grad_norm": 0.5370351684175654,
|
|
"learning_rate": 3.2909496774005344e-07,
|
|
"loss": 0.1615,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14727014303207397,
|
|
"step": 4155,
|
|
"valid_targets_mean": 4993.6,
|
|
"valid_targets_min": 1859
|
|
},
|
|
{
|
|
"epoch": 6.645367412140575,
|
|
"grad_norm": 0.5376065934233798,
|
|
"learning_rate": 3.14856823256966e-07,
|
|
"loss": 0.1601,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13972601294517517,
|
|
"step": 4160,
|
|
"valid_targets_mean": 4538.9,
|
|
"valid_targets_min": 1760
|
|
},
|
|
{
|
|
"epoch": 6.65335463258786,
|
|
"grad_norm": 0.4879257262202097,
|
|
"learning_rate": 3.009310882534999e-07,
|
|
"loss": 0.1656,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18523406982421875,
|
|
"step": 4165,
|
|
"valid_targets_mean": 5296.8,
|
|
"valid_targets_min": 1050
|
|
},
|
|
{
|
|
"epoch": 6.661341853035144,
|
|
"grad_norm": 0.5804989494397974,
|
|
"learning_rate": 2.8731798373550004e-07,
|
|
"loss": 0.1621,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1688893437385559,
|
|
"step": 4170,
|
|
"valid_targets_mean": 4406.4,
|
|
"valid_targets_min": 762
|
|
},
|
|
{
|
|
"epoch": 6.669329073482428,
|
|
"grad_norm": 0.46347021872106603,
|
|
"learning_rate": 2.740177257472576e-07,
|
|
"loss": 0.1562,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1542181670665741,
|
|
"step": 4175,
|
|
"valid_targets_mean": 5176.9,
|
|
"valid_targets_min": 640
|
|
},
|
|
{
|
|
"epoch": 6.677316293929713,
|
|
"grad_norm": 0.48527977721610366,
|
|
"learning_rate": 2.6103052536810226e-07,
|
|
"loss": 0.1532,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16154858469963074,
|
|
"step": 4180,
|
|
"valid_targets_mean": 5032.2,
|
|
"valid_targets_min": 1075
|
|
},
|
|
{
|
|
"epoch": 6.685303514376997,
|
|
"grad_norm": 0.5207002345996271,
|
|
"learning_rate": 2.4835658870902226e-07,
|
|
"loss": 0.1762,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18678952753543854,
|
|
"step": 4185,
|
|
"valid_targets_mean": 4774.8,
|
|
"valid_targets_min": 623
|
|
},
|
|
{
|
|
"epoch": 6.693290734824281,
|
|
"grad_norm": 0.4900949681899387,
|
|
"learning_rate": 2.3599611690943158e-07,
|
|
"loss": 0.1657,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14106178283691406,
|
|
"step": 4190,
|
|
"valid_targets_mean": 4770.2,
|
|
"valid_targets_min": 1290
|
|
},
|
|
{
|
|
"epoch": 6.701277955271565,
|
|
"grad_norm": 0.5658605830474439,
|
|
"learning_rate": 2.2394930613393927e-07,
|
|
"loss": 0.1595,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19314196705818176,
|
|
"step": 4195,
|
|
"valid_targets_mean": 4287.3,
|
|
"valid_targets_min": 1148
|
|
},
|
|
{
|
|
"epoch": 6.7092651757188495,
|
|
"grad_norm": 0.45153452050782694,
|
|
"learning_rate": 2.122163475692629e-07,
|
|
"loss": 0.1577,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15588468313217163,
|
|
"step": 4200,
|
|
"valid_targets_mean": 4994.2,
|
|
"valid_targets_min": 1568
|
|
},
|
|
{
|
|
"epoch": 6.717252396166134,
|
|
"grad_norm": 0.4790409003168309,
|
|
"learning_rate": 2.0079742742118878e-07,
|
|
"loss": 0.1696,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17813389003276825,
|
|
"step": 4205,
|
|
"valid_targets_mean": 5793.6,
|
|
"valid_targets_min": 2043
|
|
},
|
|
{
|
|
"epoch": 6.725239616613418,
|
|
"grad_norm": 0.48525249473687376,
|
|
"learning_rate": 1.8969272691160334e-07,
|
|
"loss": 0.1438,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1606864333152771,
|
|
"step": 4210,
|
|
"valid_targets_mean": 5174.8,
|
|
"valid_targets_min": 959
|
|
},
|
|
{
|
|
"epoch": 6.733226837060703,
|
|
"grad_norm": 0.5323491003812912,
|
|
"learning_rate": 1.789024222756397e-07,
|
|
"loss": 0.1651,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16105014085769653,
|
|
"step": 4215,
|
|
"valid_targets_mean": 3702.8,
|
|
"valid_targets_min": 1032
|
|
},
|
|
{
|
|
"epoch": 6.741214057507987,
|
|
"grad_norm": 0.4968258466615749,
|
|
"learning_rate": 1.6842668475885782e-07,
|
|
"loss": 0.1462,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14075995981693268,
|
|
"step": 4220,
|
|
"valid_targets_mean": 4199.2,
|
|
"valid_targets_min": 1313
|
|
},
|
|
{
|
|
"epoch": 6.7492012779552715,
|
|
"grad_norm": 0.4925534642983092,
|
|
"learning_rate": 1.582656806145444e-07,
|
|
"loss": 0.1639,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15862902998924255,
|
|
"step": 4225,
|
|
"valid_targets_mean": 4477.2,
|
|
"valid_targets_min": 1815
|
|
},
|
|
{
|
|
"epoch": 6.757188498402556,
|
|
"grad_norm": 0.550218936446372,
|
|
"learning_rate": 1.4841957110106388e-07,
|
|
"loss": 0.1677,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18082785606384277,
|
|
"step": 4230,
|
|
"valid_targets_mean": 5861.3,
|
|
"valid_targets_min": 940
|
|
},
|
|
{
|
|
"epoch": 6.76517571884984,
|
|
"grad_norm": 0.4781570840237632,
|
|
"learning_rate": 1.388885124793049e-07,
|
|
"loss": 0.1613,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15590888261795044,
|
|
"step": 4235,
|
|
"valid_targets_mean": 4668.6,
|
|
"valid_targets_min": 695
|
|
},
|
|
{
|
|
"epoch": 6.773162939297125,
|
|
"grad_norm": 0.5355126036837008,
|
|
"learning_rate": 1.2967265601019573e-07,
|
|
"loss": 0.1448,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14738382399082184,
|
|
"step": 4240,
|
|
"valid_targets_mean": 3881.4,
|
|
"valid_targets_min": 337
|
|
},
|
|
{
|
|
"epoch": 6.781150159744409,
|
|
"grad_norm": 0.4581319359373992,
|
|
"learning_rate": 1.207721479523105e-07,
|
|
"loss": 0.1612,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12903811037540436,
|
|
"step": 4245,
|
|
"valid_targets_mean": 4758.4,
|
|
"valid_targets_min": 2440
|
|
},
|
|
{
|
|
"epoch": 6.789137380191693,
|
|
"grad_norm": 0.5172271808327057,
|
|
"learning_rate": 1.1218712955954003e-07,
|
|
"loss": 0.1591,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14677739143371582,
|
|
"step": 4250,
|
|
"valid_targets_mean": 3776.4,
|
|
"valid_targets_min": 592
|
|
},
|
|
{
|
|
"epoch": 6.797124600638978,
|
|
"grad_norm": 0.530851817590877,
|
|
"learning_rate": 1.0391773707885578e-07,
|
|
"loss": 0.1524,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1520860195159912,
|
|
"step": 4255,
|
|
"valid_targets_mean": 3847.8,
|
|
"valid_targets_min": 521
|
|
},
|
|
{
|
|
"epoch": 6.805111821086262,
|
|
"grad_norm": 0.5796714007260149,
|
|
"learning_rate": 9.596410174814497e-08,
|
|
"loss": 0.1503,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.144109845161438,
|
|
"step": 4260,
|
|
"valid_targets_mean": 4218.2,
|
|
"valid_targets_min": 543
|
|
},
|
|
{
|
|
"epoch": 6.813099041533547,
|
|
"grad_norm": 0.5184630243827649,
|
|
"learning_rate": 8.83263497941278e-08,
|
|
"loss": 0.1807,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16317498683929443,
|
|
"step": 4265,
|
|
"valid_targets_mean": 4504.2,
|
|
"valid_targets_min": 1386
|
|
},
|
|
{
|
|
"epoch": 6.821086261980831,
|
|
"grad_norm": 0.5295419275859305,
|
|
"learning_rate": 8.10046024303568e-08,
|
|
"loss": 0.1599,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1961647868156433,
|
|
"step": 4270,
|
|
"valid_targets_mean": 5160.0,
|
|
"valid_targets_min": 2117
|
|
},
|
|
{
|
|
"epoch": 6.8290734824281145,
|
|
"grad_norm": 0.515399801707447,
|
|
"learning_rate": 7.399897585528726e-08,
|
|
"loss": 0.1586,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17954765260219574,
|
|
"step": 4275,
|
|
"valid_targets_mean": 4397.3,
|
|
"valid_targets_min": 1475
|
|
},
|
|
{
|
|
"epoch": 6.8370607028754,
|
|
"grad_norm": 0.48082799148327066,
|
|
"learning_rate": 6.730958125044319e-08,
|
|
"loss": 0.1534,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15425550937652588,
|
|
"step": 4280,
|
|
"valid_targets_mean": 4867.4,
|
|
"valid_targets_min": 902
|
|
},
|
|
{
|
|
"epoch": 6.845047923322683,
|
|
"grad_norm": 0.5125347064202592,
|
|
"learning_rate": 6.093652477864309e-08,
|
|
"loss": 0.1566,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17881545424461365,
|
|
"step": 4285,
|
|
"valid_targets_mean": 4752.0,
|
|
"valid_targets_min": 1767
|
|
},
|
|
{
|
|
"epoch": 6.853035143769968,
|
|
"grad_norm": 0.44550266769959085,
|
|
"learning_rate": 5.4879907582316986e-08,
|
|
"loss": 0.1553,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1376715898513794,
|
|
"step": 4290,
|
|
"valid_targets_mean": 5008.2,
|
|
"valid_targets_min": 1827
|
|
},
|
|
{
|
|
"epoch": 6.861022364217252,
|
|
"grad_norm": 0.49785912971046536,
|
|
"learning_rate": 4.913982578190535e-08,
|
|
"loss": 0.1676,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18554158508777618,
|
|
"step": 4295,
|
|
"valid_targets_mean": 5076.2,
|
|
"valid_targets_min": 1699
|
|
},
|
|
{
|
|
"epoch": 6.8690095846645365,
|
|
"grad_norm": 0.434012862335176,
|
|
"learning_rate": 4.3716370474331527e-08,
|
|
"loss": 0.1533,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15761414170265198,
|
|
"step": 4300,
|
|
"valid_targets_mean": 5683.8,
|
|
"valid_targets_min": 2173
|
|
},
|
|
{
|
|
"epoch": 6.876996805111821,
|
|
"grad_norm": 0.4600701812971065,
|
|
"learning_rate": 3.8609627731558405e-08,
|
|
"loss": 0.1622,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14310933649539948,
|
|
"step": 4305,
|
|
"valid_targets_mean": 5604.4,
|
|
"valid_targets_min": 1873
|
|
},
|
|
{
|
|
"epoch": 6.884984025559105,
|
|
"grad_norm": 0.7704497916826207,
|
|
"learning_rate": 3.381967859920954e-08,
|
|
"loss": 0.159,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16258737444877625,
|
|
"step": 4310,
|
|
"valid_targets_mean": 5497.0,
|
|
"valid_targets_min": 1359
|
|
},
|
|
{
|
|
"epoch": 6.89297124600639,
|
|
"grad_norm": 0.4657209311348018,
|
|
"learning_rate": 2.9346599095305685e-08,
|
|
"loss": 0.1599,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17536340653896332,
|
|
"step": 4315,
|
|
"valid_targets_mean": 5737.7,
|
|
"valid_targets_min": 784
|
|
},
|
|
{
|
|
"epoch": 6.900958466453674,
|
|
"grad_norm": 0.4473310586041997,
|
|
"learning_rate": 2.5190460209039146e-08,
|
|
"loss": 0.1645,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1598825305700302,
|
|
"step": 4320,
|
|
"valid_targets_mean": 5702.4,
|
|
"valid_targets_min": 835
|
|
},
|
|
{
|
|
"epoch": 6.9089456869009584,
|
|
"grad_norm": 0.6716035094415955,
|
|
"learning_rate": 2.1351327899656883e-08,
|
|
"loss": 0.1679,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20283496379852295,
|
|
"step": 4325,
|
|
"valid_targets_mean": 4717.5,
|
|
"valid_targets_min": 372
|
|
},
|
|
{
|
|
"epoch": 6.916932907348243,
|
|
"grad_norm": 0.5446415374246868,
|
|
"learning_rate": 1.782926309540578e-08,
|
|
"loss": 0.1523,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16346979141235352,
|
|
"step": 4330,
|
|
"valid_targets_mean": 4082.8,
|
|
"valid_targets_min": 1344
|
|
},
|
|
{
|
|
"epoch": 6.924920127795527,
|
|
"grad_norm": 0.5727743626029946,
|
|
"learning_rate": 1.462432169257344e-08,
|
|
"loss": 0.1695,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15288791060447693,
|
|
"step": 4335,
|
|
"valid_targets_mean": 5043.2,
|
|
"valid_targets_min": 333
|
|
},
|
|
{
|
|
"epoch": 6.932907348242812,
|
|
"grad_norm": 0.4857450702797858,
|
|
"learning_rate": 1.1736554554604429e-08,
|
|
"loss": 0.156,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1811220347881317,
|
|
"step": 4340,
|
|
"valid_targets_mean": 5184.7,
|
|
"valid_targets_min": 1178
|
|
},
|
|
{
|
|
"epoch": 6.940894568690096,
|
|
"grad_norm": 0.5405114561331039,
|
|
"learning_rate": 9.166007511274278e-09,
|
|
"loss": 0.1566,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.145079106092453,
|
|
"step": 4345,
|
|
"valid_targets_mean": 5249.0,
|
|
"valid_targets_min": 1114
|
|
},
|
|
{
|
|
"epoch": 6.94888178913738,
|
|
"grad_norm": 0.5207983988424999,
|
|
"learning_rate": 6.912721357985597e-09,
|
|
"loss": 0.171,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15930955111980438,
|
|
"step": 4350,
|
|
"valid_targets_mean": 4517.9,
|
|
"valid_targets_min": 794
|
|
},
|
|
{
|
|
"epoch": 6.956869009584665,
|
|
"grad_norm": 0.5174766841014377,
|
|
"learning_rate": 4.976731855104166e-09,
|
|
"loss": 0.1646,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13806484639644623,
|
|
"step": 4355,
|
|
"valid_targets_mean": 4797.9,
|
|
"valid_targets_min": 2094
|
|
},
|
|
{
|
|
"epoch": 6.964856230031949,
|
|
"grad_norm": 0.5135301894652287,
|
|
"learning_rate": 3.3580697274016028e-09,
|
|
"loss": 0.1723,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17942970991134644,
|
|
"step": 4360,
|
|
"valid_targets_mean": 4651.2,
|
|
"valid_targets_min": 1438
|
|
},
|
|
{
|
|
"epoch": 6.972843450479234,
|
|
"grad_norm": 0.5218654864574037,
|
|
"learning_rate": 2.056760663555757e-09,
|
|
"loss": 0.1524,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1661636233329773,
|
|
"step": 4365,
|
|
"valid_targets_mean": 4953.9,
|
|
"valid_targets_min": 857
|
|
},
|
|
{
|
|
"epoch": 6.980830670926517,
|
|
"grad_norm": 0.5617892482358111,
|
|
"learning_rate": 1.0728253157599178e-09,
|
|
"loss": 0.1566,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16045665740966797,
|
|
"step": 4370,
|
|
"valid_targets_mean": 3920.1,
|
|
"valid_targets_min": 288
|
|
},
|
|
{
|
|
"epoch": 6.988817891373802,
|
|
"grad_norm": 0.6107885657111256,
|
|
"learning_rate": 4.062792993786424e-10,
|
|
"loss": 0.1616,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16254091262817383,
|
|
"step": 4375,
|
|
"valid_targets_mean": 3999.9,
|
|
"valid_targets_min": 614
|
|
},
|
|
{
|
|
"epoch": 6.996805111821086,
|
|
"grad_norm": 0.4316133436358314,
|
|
"learning_rate": 5.7133192707947705e-11,
|
|
"loss": 0.1628,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13435956835746765,
|
|
"step": 4380,
|
|
"valid_targets_mean": 5564.4,
|
|
"valid_targets_min": 568
|
|
},
|
|
{
|
|
"epoch": 7.0,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1903749257326126,
|
|
"step": 4382,
|
|
"total_flos": 1474549610774528.0,
|
|
"train_loss": 0.20471333478722384,
|
|
"train_runtime": 24590.8199,
|
|
"train_samples_per_second": 2.851,
|
|
"train_steps_per_second": 0.178,
|
|
"valid_targets_mean": 5778.1,
|
|
"valid_targets_min": 3072
|
|
}
|
|
],
|
|
"logging_steps": 5,
|
|
"max_steps": 4382,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 7,
|
|
"save_steps": 1500,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": true
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 1474549610774528.0,
|
|
"train_batch_size": 1,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|