11099 lines
308 KiB
JSON
11099 lines
308 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 7.0,
|
|
"eval_steps": 500,
|
|
"global_step": 5026,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.006963788300835654,
|
|
"grad_norm": 16.70025919180339,
|
|
"learning_rate": 3.180914512922465e-07,
|
|
"loss": 0.6604,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.31003373861312866,
|
|
"step": 5,
|
|
"valid_targets_mean": 7389.0,
|
|
"valid_targets_min": 5627
|
|
},
|
|
{
|
|
"epoch": 0.013927576601671309,
|
|
"grad_norm": 15.950547440699722,
|
|
"learning_rate": 7.157057654075547e-07,
|
|
"loss": 0.6983,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23548319935798645,
|
|
"step": 10,
|
|
"valid_targets_mean": 1294.8,
|
|
"valid_targets_min": 422
|
|
},
|
|
{
|
|
"epoch": 0.020891364902506964,
|
|
"grad_norm": 13.622952098366161,
|
|
"learning_rate": 1.113320079522863e-06,
|
|
"loss": 0.6693,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.31765997409820557,
|
|
"step": 15,
|
|
"valid_targets_mean": 8539.9,
|
|
"valid_targets_min": 6761
|
|
},
|
|
{
|
|
"epoch": 0.027855153203342618,
|
|
"grad_norm": 9.836072650243178,
|
|
"learning_rate": 1.5109343936381709e-06,
|
|
"loss": 0.5881,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27282822132110596,
|
|
"step": 20,
|
|
"valid_targets_mean": 7579.0,
|
|
"valid_targets_min": 6174
|
|
},
|
|
{
|
|
"epoch": 0.034818941504178275,
|
|
"grad_norm": 6.354639801298961,
|
|
"learning_rate": 1.9085487077534792e-06,
|
|
"loss": 0.5734,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2713155746459961,
|
|
"step": 25,
|
|
"valid_targets_mean": 7328.2,
|
|
"valid_targets_min": 4839
|
|
},
|
|
{
|
|
"epoch": 0.04178272980501393,
|
|
"grad_norm": 5.430407653710302,
|
|
"learning_rate": 2.3061630218687873e-06,
|
|
"loss": 0.559,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29282844066619873,
|
|
"step": 30,
|
|
"valid_targets_mean": 6838.2,
|
|
"valid_targets_min": 5034
|
|
},
|
|
{
|
|
"epoch": 0.04874651810584958,
|
|
"grad_norm": 4.877388127505224,
|
|
"learning_rate": 2.703777335984096e-06,
|
|
"loss": 0.5283,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25860846042633057,
|
|
"step": 35,
|
|
"valid_targets_mean": 7248.2,
|
|
"valid_targets_min": 5567
|
|
},
|
|
{
|
|
"epoch": 0.055710306406685235,
|
|
"grad_norm": 2.846484730228359,
|
|
"learning_rate": 3.101391650099404e-06,
|
|
"loss": 0.4665,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21636009216308594,
|
|
"step": 40,
|
|
"valid_targets_mean": 6234.0,
|
|
"valid_targets_min": 5603
|
|
},
|
|
{
|
|
"epoch": 0.06267409470752089,
|
|
"grad_norm": 2.963648252530947,
|
|
"learning_rate": 3.499005964214712e-06,
|
|
"loss": 0.4547,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10212111473083496,
|
|
"step": 45,
|
|
"valid_targets_mean": 602.0,
|
|
"valid_targets_min": 152
|
|
},
|
|
{
|
|
"epoch": 0.06963788300835655,
|
|
"grad_norm": 1.0973687974221902,
|
|
"learning_rate": 3.89662027833002e-06,
|
|
"loss": 0.4081,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20101504027843475,
|
|
"step": 50,
|
|
"valid_targets_mean": 7299.9,
|
|
"valid_targets_min": 6158
|
|
},
|
|
{
|
|
"epoch": 0.0766016713091922,
|
|
"grad_norm": 0.8984488229396836,
|
|
"learning_rate": 4.294234592445328e-06,
|
|
"loss": 0.4052,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1902889907360077,
|
|
"step": 55,
|
|
"valid_targets_mean": 6140.0,
|
|
"valid_targets_min": 4648
|
|
},
|
|
{
|
|
"epoch": 0.08356545961002786,
|
|
"grad_norm": 0.78646128561226,
|
|
"learning_rate": 4.6918489065606366e-06,
|
|
"loss": 0.3897,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1944657266139984,
|
|
"step": 60,
|
|
"valid_targets_mean": 6603.0,
|
|
"valid_targets_min": 4882
|
|
},
|
|
{
|
|
"epoch": 0.0905292479108635,
|
|
"grad_norm": 0.6308887930432894,
|
|
"learning_rate": 5.089463220675944e-06,
|
|
"loss": 0.371,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1896740198135376,
|
|
"step": 65,
|
|
"valid_targets_mean": 8015.0,
|
|
"valid_targets_min": 6344
|
|
},
|
|
{
|
|
"epoch": 0.09749303621169916,
|
|
"grad_norm": 0.7528753110445529,
|
|
"learning_rate": 5.487077534791253e-06,
|
|
"loss": 0.3587,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17312809824943542,
|
|
"step": 70,
|
|
"valid_targets_mean": 6305.2,
|
|
"valid_targets_min": 4622
|
|
},
|
|
{
|
|
"epoch": 0.10445682451253482,
|
|
"grad_norm": 0.5487849390115586,
|
|
"learning_rate": 5.8846918489065606e-06,
|
|
"loss": 0.3612,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18349185585975647,
|
|
"step": 75,
|
|
"valid_targets_mean": 7890.8,
|
|
"valid_targets_min": 5485
|
|
},
|
|
{
|
|
"epoch": 0.11142061281337047,
|
|
"grad_norm": 0.5189028061387841,
|
|
"learning_rate": 6.28230616302187e-06,
|
|
"loss": 0.3552,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1606183648109436,
|
|
"step": 80,
|
|
"valid_targets_mean": 7016.6,
|
|
"valid_targets_min": 4960
|
|
},
|
|
{
|
|
"epoch": 0.11838440111420613,
|
|
"grad_norm": 0.5214627334126402,
|
|
"learning_rate": 6.679920477137178e-06,
|
|
"loss": 0.3415,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.160720095038414,
|
|
"step": 85,
|
|
"valid_targets_mean": 6527.1,
|
|
"valid_targets_min": 5399
|
|
},
|
|
{
|
|
"epoch": 0.12534818941504178,
|
|
"grad_norm": 0.6006447673375296,
|
|
"learning_rate": 7.077534791252486e-06,
|
|
"loss": 0.3419,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17021866142749786,
|
|
"step": 90,
|
|
"valid_targets_mean": 4995.0,
|
|
"valid_targets_min": 618
|
|
},
|
|
{
|
|
"epoch": 0.13231197771587744,
|
|
"grad_norm": 0.49819624129860873,
|
|
"learning_rate": 7.475149105367794e-06,
|
|
"loss": 0.3511,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1750391721725464,
|
|
"step": 95,
|
|
"valid_targets_mean": 7219.4,
|
|
"valid_targets_min": 5764
|
|
},
|
|
{
|
|
"epoch": 0.1392757660167131,
|
|
"grad_norm": 0.49092772422498776,
|
|
"learning_rate": 7.872763419483102e-06,
|
|
"loss": 0.3305,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17389681935310364,
|
|
"step": 100,
|
|
"valid_targets_mean": 7998.5,
|
|
"valid_targets_min": 5902
|
|
},
|
|
{
|
|
"epoch": 0.14623955431754876,
|
|
"grad_norm": 0.5105786705340714,
|
|
"learning_rate": 8.27037773359841e-06,
|
|
"loss": 0.3031,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13476799428462982,
|
|
"step": 105,
|
|
"valid_targets_mean": 7377.4,
|
|
"valid_targets_min": 5015
|
|
},
|
|
{
|
|
"epoch": 0.1532033426183844,
|
|
"grad_norm": 0.4395069780745515,
|
|
"learning_rate": 8.667992047713719e-06,
|
|
"loss": 0.3071,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.134280264377594,
|
|
"step": 110,
|
|
"valid_targets_mean": 6794.0,
|
|
"valid_targets_min": 5283
|
|
},
|
|
{
|
|
"epoch": 0.16016713091922005,
|
|
"grad_norm": 0.5344731312163281,
|
|
"learning_rate": 9.065606361829027e-06,
|
|
"loss": 0.3196,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17999088764190674,
|
|
"step": 115,
|
|
"valid_targets_mean": 6656.4,
|
|
"valid_targets_min": 4364
|
|
},
|
|
{
|
|
"epoch": 0.1671309192200557,
|
|
"grad_norm": 0.46349547442614725,
|
|
"learning_rate": 9.463220675944334e-06,
|
|
"loss": 0.3083,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1460513472557068,
|
|
"step": 120,
|
|
"valid_targets_mean": 7178.4,
|
|
"valid_targets_min": 4448
|
|
},
|
|
{
|
|
"epoch": 0.17409470752089137,
|
|
"grad_norm": 0.49695390233410575,
|
|
"learning_rate": 9.860834990059643e-06,
|
|
"loss": 0.304,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17316694557666779,
|
|
"step": 125,
|
|
"valid_targets_mean": 7619.8,
|
|
"valid_targets_min": 5737
|
|
},
|
|
{
|
|
"epoch": 0.181058495821727,
|
|
"grad_norm": 0.4131284400571314,
|
|
"learning_rate": 1.0258449304174951e-05,
|
|
"loss": 0.2685,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13033336400985718,
|
|
"step": 130,
|
|
"valid_targets_mean": 8754.4,
|
|
"valid_targets_min": 5752
|
|
},
|
|
{
|
|
"epoch": 0.18802228412256267,
|
|
"grad_norm": 0.46954213145258955,
|
|
"learning_rate": 1.065606361829026e-05,
|
|
"loss": 0.267,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14443400502204895,
|
|
"step": 135,
|
|
"valid_targets_mean": 6341.0,
|
|
"valid_targets_min": 4543
|
|
},
|
|
{
|
|
"epoch": 0.19498607242339833,
|
|
"grad_norm": 0.5305150388984193,
|
|
"learning_rate": 1.1053677932405567e-05,
|
|
"loss": 0.2569,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14701451361179352,
|
|
"step": 140,
|
|
"valid_targets_mean": 7181.9,
|
|
"valid_targets_min": 4563
|
|
},
|
|
{
|
|
"epoch": 0.201949860724234,
|
|
"grad_norm": 1.1100672340753313,
|
|
"learning_rate": 1.1451292246520875e-05,
|
|
"loss": 0.3149,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11581391841173172,
|
|
"step": 145,
|
|
"valid_targets_mean": 1104.4,
|
|
"valid_targets_min": 161
|
|
},
|
|
{
|
|
"epoch": 0.20891364902506965,
|
|
"grad_norm": 0.42654296534182173,
|
|
"learning_rate": 1.1848906560636184e-05,
|
|
"loss": 0.2978,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13008113205432892,
|
|
"step": 150,
|
|
"valid_targets_mean": 7018.2,
|
|
"valid_targets_min": 5424
|
|
},
|
|
{
|
|
"epoch": 0.21587743732590528,
|
|
"grad_norm": 0.4847320156022274,
|
|
"learning_rate": 1.2246520874751492e-05,
|
|
"loss": 0.3005,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15399551391601562,
|
|
"step": 155,
|
|
"valid_targets_mean": 7738.4,
|
|
"valid_targets_min": 5329
|
|
},
|
|
{
|
|
"epoch": 0.22284122562674094,
|
|
"grad_norm": 0.5316398632314259,
|
|
"learning_rate": 1.26441351888668e-05,
|
|
"loss": 0.3071,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14879417419433594,
|
|
"step": 160,
|
|
"valid_targets_mean": 6426.6,
|
|
"valid_targets_min": 5560
|
|
},
|
|
{
|
|
"epoch": 0.2298050139275766,
|
|
"grad_norm": 0.64130569981037,
|
|
"learning_rate": 1.3041749502982108e-05,
|
|
"loss": 0.2979,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15116575360298157,
|
|
"step": 165,
|
|
"valid_targets_mean": 6169.9,
|
|
"valid_targets_min": 3446
|
|
},
|
|
{
|
|
"epoch": 0.23676880222841226,
|
|
"grad_norm": 0.4742255633480007,
|
|
"learning_rate": 1.3439363817097418e-05,
|
|
"loss": 0.284,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14647752046585083,
|
|
"step": 170,
|
|
"valid_targets_mean": 7204.6,
|
|
"valid_targets_min": 5381
|
|
},
|
|
{
|
|
"epoch": 0.24373259052924792,
|
|
"grad_norm": 0.47419429273153524,
|
|
"learning_rate": 1.3836978131212723e-05,
|
|
"loss": 0.2885,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14805693924427032,
|
|
"step": 175,
|
|
"valid_targets_mean": 7600.4,
|
|
"valid_targets_min": 5241
|
|
},
|
|
{
|
|
"epoch": 0.25069637883008355,
|
|
"grad_norm": 0.4888837565186023,
|
|
"learning_rate": 1.4234592445328033e-05,
|
|
"loss": 0.2797,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15302909910678864,
|
|
"step": 180,
|
|
"valid_targets_mean": 7098.5,
|
|
"valid_targets_min": 5607
|
|
},
|
|
{
|
|
"epoch": 0.2576601671309192,
|
|
"grad_norm": 0.5085230199199996,
|
|
"learning_rate": 1.4632206759443342e-05,
|
|
"loss": 0.2851,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14871110022068024,
|
|
"step": 185,
|
|
"valid_targets_mean": 6955.1,
|
|
"valid_targets_min": 5530
|
|
},
|
|
{
|
|
"epoch": 0.2646239554317549,
|
|
"grad_norm": 0.47969863426103976,
|
|
"learning_rate": 1.502982107355865e-05,
|
|
"loss": 0.2799,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14307832717895508,
|
|
"step": 190,
|
|
"valid_targets_mean": 6748.1,
|
|
"valid_targets_min": 5364
|
|
},
|
|
{
|
|
"epoch": 0.27158774373259054,
|
|
"grad_norm": 0.5300381897608597,
|
|
"learning_rate": 1.5427435387673956e-05,
|
|
"loss": 0.2836,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13810153305530548,
|
|
"step": 195,
|
|
"valid_targets_mean": 6273.9,
|
|
"valid_targets_min": 5438
|
|
},
|
|
{
|
|
"epoch": 0.2785515320334262,
|
|
"grad_norm": 0.4680707912423161,
|
|
"learning_rate": 1.5825049701789266e-05,
|
|
"loss": 0.2836,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13786497712135315,
|
|
"step": 200,
|
|
"valid_targets_mean": 6419.5,
|
|
"valid_targets_min": 4879
|
|
},
|
|
{
|
|
"epoch": 0.28551532033426186,
|
|
"grad_norm": 0.4720452212866539,
|
|
"learning_rate": 1.6222664015904573e-05,
|
|
"loss": 0.2793,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13714374601840973,
|
|
"step": 205,
|
|
"valid_targets_mean": 6483.4,
|
|
"valid_targets_min": 4309
|
|
},
|
|
{
|
|
"epoch": 0.2924791086350975,
|
|
"grad_norm": 0.5255652733594837,
|
|
"learning_rate": 1.662027833001988e-05,
|
|
"loss": 0.2755,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1564159393310547,
|
|
"step": 210,
|
|
"valid_targets_mean": 6825.0,
|
|
"valid_targets_min": 5205
|
|
},
|
|
{
|
|
"epoch": 0.2994428969359331,
|
|
"grad_norm": 0.49603459688272655,
|
|
"learning_rate": 1.701789264413519e-05,
|
|
"loss": 0.2724,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12593704462051392,
|
|
"step": 215,
|
|
"valid_targets_mean": 6729.4,
|
|
"valid_targets_min": 5200
|
|
},
|
|
{
|
|
"epoch": 0.3064066852367688,
|
|
"grad_norm": 0.509332149215568,
|
|
"learning_rate": 1.74155069582505e-05,
|
|
"loss": 0.2696,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13264459371566772,
|
|
"step": 220,
|
|
"valid_targets_mean": 6094.0,
|
|
"valid_targets_min": 5302
|
|
},
|
|
{
|
|
"epoch": 0.31337047353760444,
|
|
"grad_norm": 0.5608083132142618,
|
|
"learning_rate": 1.7813121272365807e-05,
|
|
"loss": 0.2732,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14229202270507812,
|
|
"step": 225,
|
|
"valid_targets_mean": 7140.0,
|
|
"valid_targets_min": 5270
|
|
},
|
|
{
|
|
"epoch": 0.3203342618384401,
|
|
"grad_norm": 0.4518894667306929,
|
|
"learning_rate": 1.8210735586481114e-05,
|
|
"loss": 0.2633,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12612324953079224,
|
|
"step": 230,
|
|
"valid_targets_mean": 7425.0,
|
|
"valid_targets_min": 5281
|
|
},
|
|
{
|
|
"epoch": 0.32729805013927576,
|
|
"grad_norm": 0.5269894067130169,
|
|
"learning_rate": 1.8608349900596424e-05,
|
|
"loss": 0.2705,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16548216342926025,
|
|
"step": 235,
|
|
"valid_targets_mean": 7063.4,
|
|
"valid_targets_min": 5386
|
|
},
|
|
{
|
|
"epoch": 0.3342618384401114,
|
|
"grad_norm": 0.5088318301963545,
|
|
"learning_rate": 1.900596421471173e-05,
|
|
"loss": 0.2701,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1421387791633606,
|
|
"step": 240,
|
|
"valid_targets_mean": 6827.0,
|
|
"valid_targets_min": 4547
|
|
},
|
|
{
|
|
"epoch": 0.3412256267409471,
|
|
"grad_norm": 0.8073008646600404,
|
|
"learning_rate": 1.9403578528827038e-05,
|
|
"loss": 0.2496,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08542703092098236,
|
|
"step": 245,
|
|
"valid_targets_mean": 2047.0,
|
|
"valid_targets_min": 165
|
|
},
|
|
{
|
|
"epoch": 0.34818941504178275,
|
|
"grad_norm": 0.4927784128489709,
|
|
"learning_rate": 1.9801192842942348e-05,
|
|
"loss": 0.2596,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13067278265953064,
|
|
"step": 250,
|
|
"valid_targets_mean": 6754.1,
|
|
"valid_targets_min": 5196
|
|
},
|
|
{
|
|
"epoch": 0.3551532033426184,
|
|
"grad_norm": 0.5689943591952936,
|
|
"learning_rate": 2.0198807157057655e-05,
|
|
"loss": 0.2606,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13194392621517181,
|
|
"step": 255,
|
|
"valid_targets_mean": 5985.6,
|
|
"valid_targets_min": 5028
|
|
},
|
|
{
|
|
"epoch": 0.362116991643454,
|
|
"grad_norm": 0.6584472361090068,
|
|
"learning_rate": 2.0596421471172965e-05,
|
|
"loss": 0.2658,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12018608301877975,
|
|
"step": 260,
|
|
"valid_targets_mean": 5963.9,
|
|
"valid_targets_min": 5237
|
|
},
|
|
{
|
|
"epoch": 0.36908077994428967,
|
|
"grad_norm": 0.507426793822691,
|
|
"learning_rate": 2.0994035785288272e-05,
|
|
"loss": 0.2602,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1359572857618332,
|
|
"step": 265,
|
|
"valid_targets_mean": 6319.0,
|
|
"valid_targets_min": 5412
|
|
},
|
|
{
|
|
"epoch": 0.37604456824512533,
|
|
"grad_norm": 0.4533447276132855,
|
|
"learning_rate": 2.139165009940358e-05,
|
|
"loss": 0.2469,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10632900148630142,
|
|
"step": 270,
|
|
"valid_targets_mean": 7125.8,
|
|
"valid_targets_min": 5080
|
|
},
|
|
{
|
|
"epoch": 0.383008356545961,
|
|
"grad_norm": 0.5072627844159278,
|
|
"learning_rate": 2.1789264413518886e-05,
|
|
"loss": 0.2607,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12942767143249512,
|
|
"step": 275,
|
|
"valid_targets_mean": 6123.5,
|
|
"valid_targets_min": 4521
|
|
},
|
|
{
|
|
"epoch": 0.38997214484679665,
|
|
"grad_norm": 2.3716958270807,
|
|
"learning_rate": 2.2186878727634196e-05,
|
|
"loss": 0.3214,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23617935180664062,
|
|
"step": 280,
|
|
"valid_targets_mean": 1886.0,
|
|
"valid_targets_min": 1111
|
|
},
|
|
{
|
|
"epoch": 0.3969359331476323,
|
|
"grad_norm": 1.5308735634592507,
|
|
"learning_rate": 2.2584493041749503e-05,
|
|
"loss": 0.3968,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19828379154205322,
|
|
"step": 285,
|
|
"valid_targets_mean": 1497.6,
|
|
"valid_targets_min": 893
|
|
},
|
|
{
|
|
"epoch": 0.403899721448468,
|
|
"grad_norm": 1.1543013502299828,
|
|
"learning_rate": 2.2982107355864813e-05,
|
|
"loss": 0.3449,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1195836067199707,
|
|
"step": 290,
|
|
"valid_targets_mean": 1356.0,
|
|
"valid_targets_min": 914
|
|
},
|
|
{
|
|
"epoch": 0.41086350974930363,
|
|
"grad_norm": 1.094348041750749,
|
|
"learning_rate": 2.3379721669980124e-05,
|
|
"loss": 0.3274,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13568729162216187,
|
|
"step": 295,
|
|
"valid_targets_mean": 1324.5,
|
|
"valid_targets_min": 680
|
|
},
|
|
{
|
|
"epoch": 0.4178272980501393,
|
|
"grad_norm": 0.9321428551455399,
|
|
"learning_rate": 2.377733598409543e-05,
|
|
"loss": 0.3174,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14562851190567017,
|
|
"step": 300,
|
|
"valid_targets_mean": 1931.8,
|
|
"valid_targets_min": 876
|
|
},
|
|
{
|
|
"epoch": 0.42479108635097496,
|
|
"grad_norm": 1.2058603525575806,
|
|
"learning_rate": 2.4174950298210734e-05,
|
|
"loss": 0.3023,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1342366337776184,
|
|
"step": 305,
|
|
"valid_targets_mean": 1292.2,
|
|
"valid_targets_min": 734
|
|
},
|
|
{
|
|
"epoch": 0.43175487465181056,
|
|
"grad_norm": 1.2425037495378959,
|
|
"learning_rate": 2.4572564612326044e-05,
|
|
"loss": 0.3084,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1206846684217453,
|
|
"step": 310,
|
|
"valid_targets_mean": 1469.6,
|
|
"valid_targets_min": 655
|
|
},
|
|
{
|
|
"epoch": 0.4387186629526462,
|
|
"grad_norm": 0.970667265555978,
|
|
"learning_rate": 2.4970178926441354e-05,
|
|
"loss": 0.2771,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1465504765510559,
|
|
"step": 315,
|
|
"valid_targets_mean": 1548.4,
|
|
"valid_targets_min": 737
|
|
},
|
|
{
|
|
"epoch": 0.4456824512534819,
|
|
"grad_norm": 1.0165307573128308,
|
|
"learning_rate": 2.536779324055666e-05,
|
|
"loss": 0.2969,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1261579394340515,
|
|
"step": 320,
|
|
"valid_targets_mean": 1316.5,
|
|
"valid_targets_min": 759
|
|
},
|
|
{
|
|
"epoch": 0.45264623955431754,
|
|
"grad_norm": 0.9803325403945289,
|
|
"learning_rate": 2.576540755467197e-05,
|
|
"loss": 0.2766,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10474079847335815,
|
|
"step": 325,
|
|
"valid_targets_mean": 1249.6,
|
|
"valid_targets_min": 753
|
|
},
|
|
{
|
|
"epoch": 0.4596100278551532,
|
|
"grad_norm": 0.9660896867872805,
|
|
"learning_rate": 2.616302186878728e-05,
|
|
"loss": 0.2763,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17885830998420715,
|
|
"step": 330,
|
|
"valid_targets_mean": 1833.0,
|
|
"valid_targets_min": 927
|
|
},
|
|
{
|
|
"epoch": 0.46657381615598886,
|
|
"grad_norm": 0.9889163176342366,
|
|
"learning_rate": 2.656063618290259e-05,
|
|
"loss": 0.2858,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1363910287618637,
|
|
"step": 335,
|
|
"valid_targets_mean": 1338.4,
|
|
"valid_targets_min": 679
|
|
},
|
|
{
|
|
"epoch": 0.4735376044568245,
|
|
"grad_norm": 1.061457665973345,
|
|
"learning_rate": 2.6958250497017892e-05,
|
|
"loss": 0.2733,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12158133089542389,
|
|
"step": 340,
|
|
"valid_targets_mean": 1244.6,
|
|
"valid_targets_min": 771
|
|
},
|
|
{
|
|
"epoch": 0.4805013927576602,
|
|
"grad_norm": 1.0563598410355024,
|
|
"learning_rate": 2.7355864811133202e-05,
|
|
"loss": 0.2559,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12138071656227112,
|
|
"step": 345,
|
|
"valid_targets_mean": 1732.5,
|
|
"valid_targets_min": 986
|
|
},
|
|
{
|
|
"epoch": 0.48746518105849584,
|
|
"grad_norm": 1.1235998467468378,
|
|
"learning_rate": 2.775347912524851e-05,
|
|
"loss": 0.2646,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11856169253587723,
|
|
"step": 350,
|
|
"valid_targets_mean": 1389.9,
|
|
"valid_targets_min": 1103
|
|
},
|
|
{
|
|
"epoch": 0.49442896935933145,
|
|
"grad_norm": 0.942754195687271,
|
|
"learning_rate": 2.815109343936382e-05,
|
|
"loss": 0.2673,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14676061272621155,
|
|
"step": 355,
|
|
"valid_targets_mean": 1593.2,
|
|
"valid_targets_min": 675
|
|
},
|
|
{
|
|
"epoch": 0.5013927576601671,
|
|
"grad_norm": 0.927084600615998,
|
|
"learning_rate": 2.8548707753479126e-05,
|
|
"loss": 0.2662,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16312474012374878,
|
|
"step": 360,
|
|
"valid_targets_mean": 1926.6,
|
|
"valid_targets_min": 974
|
|
},
|
|
{
|
|
"epoch": 0.5083565459610028,
|
|
"grad_norm": 1.0089719119822702,
|
|
"learning_rate": 2.8946322067594437e-05,
|
|
"loss": 0.2719,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16004151105880737,
|
|
"step": 365,
|
|
"valid_targets_mean": 1953.1,
|
|
"valid_targets_min": 869
|
|
},
|
|
{
|
|
"epoch": 0.5153203342618384,
|
|
"grad_norm": 1.0547290812174692,
|
|
"learning_rate": 2.9343936381709747e-05,
|
|
"loss": 0.2683,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11815130710601807,
|
|
"step": 370,
|
|
"valid_targets_mean": 1327.2,
|
|
"valid_targets_min": 1023
|
|
},
|
|
{
|
|
"epoch": 0.5222841225626741,
|
|
"grad_norm": 1.091197129252169,
|
|
"learning_rate": 2.974155069582505e-05,
|
|
"loss": 0.271,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14482921361923218,
|
|
"step": 375,
|
|
"valid_targets_mean": 1717.9,
|
|
"valid_targets_min": 709
|
|
},
|
|
{
|
|
"epoch": 0.5292479108635098,
|
|
"grad_norm": 1.041832550945947,
|
|
"learning_rate": 3.0139165009940357e-05,
|
|
"loss": 0.2561,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.111148402094841,
|
|
"step": 380,
|
|
"valid_targets_mean": 1139.9,
|
|
"valid_targets_min": 658
|
|
},
|
|
{
|
|
"epoch": 0.5362116991643454,
|
|
"grad_norm": 1.0373743215959594,
|
|
"learning_rate": 3.053677932405567e-05,
|
|
"loss": 0.2671,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12807971239089966,
|
|
"step": 385,
|
|
"valid_targets_mean": 1444.8,
|
|
"valid_targets_min": 662
|
|
},
|
|
{
|
|
"epoch": 0.5431754874651811,
|
|
"grad_norm": 0.9362737502243705,
|
|
"learning_rate": 3.0934393638170974e-05,
|
|
"loss": 0.2708,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0938352644443512,
|
|
"step": 390,
|
|
"valid_targets_mean": 1242.2,
|
|
"valid_targets_min": 859
|
|
},
|
|
{
|
|
"epoch": 0.5501392757660167,
|
|
"grad_norm": 0.9658781825606017,
|
|
"learning_rate": 3.133200795228629e-05,
|
|
"loss": 0.2447,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.149136483669281,
|
|
"step": 395,
|
|
"valid_targets_mean": 1569.2,
|
|
"valid_targets_min": 950
|
|
},
|
|
{
|
|
"epoch": 0.5571030640668524,
|
|
"grad_norm": 1.1094086785951527,
|
|
"learning_rate": 3.1729622266401595e-05,
|
|
"loss": 0.2579,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12185294181108475,
|
|
"step": 400,
|
|
"valid_targets_mean": 1539.8,
|
|
"valid_targets_min": 686
|
|
},
|
|
{
|
|
"epoch": 0.564066852367688,
|
|
"grad_norm": 0.8987556090799284,
|
|
"learning_rate": 3.21272365805169e-05,
|
|
"loss": 0.2507,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13875404000282288,
|
|
"step": 405,
|
|
"valid_targets_mean": 1610.6,
|
|
"valid_targets_min": 702
|
|
},
|
|
{
|
|
"epoch": 0.5710306406685237,
|
|
"grad_norm": 0.9570507306238789,
|
|
"learning_rate": 3.252485089463221e-05,
|
|
"loss": 0.2456,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09257258474826813,
|
|
"step": 410,
|
|
"valid_targets_mean": 1190.4,
|
|
"valid_targets_min": 779
|
|
},
|
|
{
|
|
"epoch": 0.5779944289693594,
|
|
"grad_norm": 1.041172830465715,
|
|
"learning_rate": 3.2922465208747515e-05,
|
|
"loss": 0.2626,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13315311074256897,
|
|
"step": 415,
|
|
"valid_targets_mean": 1467.1,
|
|
"valid_targets_min": 822
|
|
},
|
|
{
|
|
"epoch": 0.584958217270195,
|
|
"grad_norm": 0.986140542035632,
|
|
"learning_rate": 3.332007952286282e-05,
|
|
"loss": 0.2411,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10449384897947311,
|
|
"step": 420,
|
|
"valid_targets_mean": 1350.5,
|
|
"valid_targets_min": 868
|
|
},
|
|
{
|
|
"epoch": 0.5919220055710307,
|
|
"grad_norm": 1.065472844147368,
|
|
"learning_rate": 3.3717693836978136e-05,
|
|
"loss": 0.2507,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10323739796876907,
|
|
"step": 425,
|
|
"valid_targets_mean": 1283.5,
|
|
"valid_targets_min": 781
|
|
},
|
|
{
|
|
"epoch": 0.5988857938718662,
|
|
"grad_norm": 0.9665463285582397,
|
|
"learning_rate": 3.411530815109344e-05,
|
|
"loss": 0.251,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10586751997470856,
|
|
"step": 430,
|
|
"valid_targets_mean": 1303.2,
|
|
"valid_targets_min": 793
|
|
},
|
|
{
|
|
"epoch": 0.6058495821727019,
|
|
"grad_norm": 1.2852362210237431,
|
|
"learning_rate": 3.451292246520875e-05,
|
|
"loss": 0.2494,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10657232999801636,
|
|
"step": 435,
|
|
"valid_targets_mean": 1309.1,
|
|
"valid_targets_min": 806
|
|
},
|
|
{
|
|
"epoch": 0.6128133704735376,
|
|
"grad_norm": 0.8736026107192932,
|
|
"learning_rate": 3.491053677932406e-05,
|
|
"loss": 0.2474,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1021880954504013,
|
|
"step": 440,
|
|
"valid_targets_mean": 1397.6,
|
|
"valid_targets_min": 803
|
|
},
|
|
{
|
|
"epoch": 0.6197771587743732,
|
|
"grad_norm": 1.0411104476153328,
|
|
"learning_rate": 3.530815109343936e-05,
|
|
"loss": 0.2446,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0923558846116066,
|
|
"step": 445,
|
|
"valid_targets_mean": 1187.8,
|
|
"valid_targets_min": 660
|
|
},
|
|
{
|
|
"epoch": 0.6267409470752089,
|
|
"grad_norm": 0.9460837866342159,
|
|
"learning_rate": 3.570576540755467e-05,
|
|
"loss": 0.229,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10013886541128159,
|
|
"step": 450,
|
|
"valid_targets_mean": 1547.0,
|
|
"valid_targets_min": 853
|
|
},
|
|
{
|
|
"epoch": 0.6337047353760445,
|
|
"grad_norm": 0.9098298102835145,
|
|
"learning_rate": 3.6103379721669984e-05,
|
|
"loss": 0.2482,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1290132701396942,
|
|
"step": 455,
|
|
"valid_targets_mean": 1704.2,
|
|
"valid_targets_min": 719
|
|
},
|
|
{
|
|
"epoch": 0.6406685236768802,
|
|
"grad_norm": 0.9535602596956675,
|
|
"learning_rate": 3.650099403578529e-05,
|
|
"loss": 0.2486,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08892850577831268,
|
|
"step": 460,
|
|
"valid_targets_mean": 1255.0,
|
|
"valid_targets_min": 765
|
|
},
|
|
{
|
|
"epoch": 0.6476323119777159,
|
|
"grad_norm": 0.9327456495578426,
|
|
"learning_rate": 3.68986083499006e-05,
|
|
"loss": 0.2428,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11986593157052994,
|
|
"step": 465,
|
|
"valid_targets_mean": 1550.8,
|
|
"valid_targets_min": 875
|
|
},
|
|
{
|
|
"epoch": 0.6545961002785515,
|
|
"grad_norm": 1.0928873361476414,
|
|
"learning_rate": 3.729622266401591e-05,
|
|
"loss": 0.2413,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12141738831996918,
|
|
"step": 470,
|
|
"valid_targets_mean": 1362.0,
|
|
"valid_targets_min": 739
|
|
},
|
|
{
|
|
"epoch": 0.6615598885793872,
|
|
"grad_norm": 0.9405266158146244,
|
|
"learning_rate": 3.769383697813122e-05,
|
|
"loss": 0.2504,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13972696661949158,
|
|
"step": 475,
|
|
"valid_targets_mean": 1666.8,
|
|
"valid_targets_min": 760
|
|
},
|
|
{
|
|
"epoch": 0.6685236768802229,
|
|
"grad_norm": 0.9803157445663641,
|
|
"learning_rate": 3.8091451292246525e-05,
|
|
"loss": 0.2348,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10272186994552612,
|
|
"step": 480,
|
|
"valid_targets_mean": 1319.6,
|
|
"valid_targets_min": 1046
|
|
},
|
|
{
|
|
"epoch": 0.6754874651810585,
|
|
"grad_norm": 0.8300161576779523,
|
|
"learning_rate": 3.848906560636183e-05,
|
|
"loss": 0.254,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13779720664024353,
|
|
"step": 485,
|
|
"valid_targets_mean": 1727.9,
|
|
"valid_targets_min": 851
|
|
},
|
|
{
|
|
"epoch": 0.6824512534818942,
|
|
"grad_norm": 0.958305960647266,
|
|
"learning_rate": 3.888667992047714e-05,
|
|
"loss": 0.2521,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.129207044839859,
|
|
"step": 490,
|
|
"valid_targets_mean": 1643.2,
|
|
"valid_targets_min": 907
|
|
},
|
|
{
|
|
"epoch": 0.6894150417827298,
|
|
"grad_norm": 0.9358380856267636,
|
|
"learning_rate": 3.9284294234592445e-05,
|
|
"loss": 0.2581,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11121398210525513,
|
|
"step": 495,
|
|
"valid_targets_mean": 1478.5,
|
|
"valid_targets_min": 764
|
|
},
|
|
{
|
|
"epoch": 0.6963788300835655,
|
|
"grad_norm": 0.8280865710705787,
|
|
"learning_rate": 3.968190854870776e-05,
|
|
"loss": 0.2378,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1098678931593895,
|
|
"step": 500,
|
|
"valid_targets_mean": 1522.6,
|
|
"valid_targets_min": 822
|
|
},
|
|
{
|
|
"epoch": 0.7033426183844012,
|
|
"grad_norm": 0.8854144055119295,
|
|
"learning_rate": 3.999999517556397e-05,
|
|
"loss": 0.2363,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11844191700220108,
|
|
"step": 505,
|
|
"valid_targets_mean": 1589.0,
|
|
"valid_targets_min": 982
|
|
},
|
|
{
|
|
"epoch": 0.7103064066852368,
|
|
"grad_norm": 0.889147230178005,
|
|
"learning_rate": 3.9999826320547226e-05,
|
|
"loss": 0.2489,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12125623971223831,
|
|
"step": 510,
|
|
"valid_targets_mean": 1574.8,
|
|
"valid_targets_min": 816
|
|
},
|
|
{
|
|
"epoch": 0.7172701949860725,
|
|
"grad_norm": 1.0297104353022954,
|
|
"learning_rate": 3.999941624605637e-05,
|
|
"loss": 0.238,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12160146236419678,
|
|
"step": 515,
|
|
"valid_targets_mean": 1422.0,
|
|
"valid_targets_min": 721
|
|
},
|
|
{
|
|
"epoch": 0.724233983286908,
|
|
"grad_norm": 0.9715916674508144,
|
|
"learning_rate": 3.999876495703734e-05,
|
|
"loss": 0.2319,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08852733671665192,
|
|
"step": 520,
|
|
"valid_targets_mean": 1150.6,
|
|
"valid_targets_min": 741
|
|
},
|
|
{
|
|
"epoch": 0.7311977715877437,
|
|
"grad_norm": 0.9348675328466968,
|
|
"learning_rate": 3.999787246134539e-05,
|
|
"loss": 0.24,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1302892565727234,
|
|
"step": 525,
|
|
"valid_targets_mean": 1684.2,
|
|
"valid_targets_min": 734
|
|
},
|
|
{
|
|
"epoch": 0.7381615598885793,
|
|
"grad_norm": 0.9829460553497974,
|
|
"learning_rate": 3.999673876974498e-05,
|
|
"loss": 0.2424,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14294780790805817,
|
|
"step": 530,
|
|
"valid_targets_mean": 1937.1,
|
|
"valid_targets_min": 744
|
|
},
|
|
{
|
|
"epoch": 0.745125348189415,
|
|
"grad_norm": 1.00886311547871,
|
|
"learning_rate": 3.9995363895909654e-05,
|
|
"loss": 0.2399,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1358570158481598,
|
|
"step": 535,
|
|
"valid_targets_mean": 1842.2,
|
|
"valid_targets_min": 620
|
|
},
|
|
{
|
|
"epoch": 0.7520891364902507,
|
|
"grad_norm": 0.9029995225438505,
|
|
"learning_rate": 3.9993747856421865e-05,
|
|
"loss": 0.2335,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13714322447776794,
|
|
"step": 540,
|
|
"valid_targets_mean": 1637.6,
|
|
"valid_targets_min": 785
|
|
},
|
|
{
|
|
"epoch": 0.7590529247910863,
|
|
"grad_norm": 0.8367188043813011,
|
|
"learning_rate": 3.99918906707728e-05,
|
|
"loss": 0.2376,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11780621111392975,
|
|
"step": 545,
|
|
"valid_targets_mean": 1969.9,
|
|
"valid_targets_min": 740
|
|
},
|
|
{
|
|
"epoch": 0.766016713091922,
|
|
"grad_norm": 0.9884389058338642,
|
|
"learning_rate": 3.9989792361362124e-05,
|
|
"loss": 0.2344,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1071939691901207,
|
|
"step": 550,
|
|
"valid_targets_mean": 1432.1,
|
|
"valid_targets_min": 711
|
|
},
|
|
{
|
|
"epoch": 0.7729805013927576,
|
|
"grad_norm": 0.9502178485105263,
|
|
"learning_rate": 3.99874529534977e-05,
|
|
"loss": 0.2346,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12912839651107788,
|
|
"step": 555,
|
|
"valid_targets_mean": 1695.6,
|
|
"valid_targets_min": 1122
|
|
},
|
|
{
|
|
"epoch": 0.7799442896935933,
|
|
"grad_norm": 0.9668592085821732,
|
|
"learning_rate": 3.9984872475395325e-05,
|
|
"loss": 0.2326,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12528224289417267,
|
|
"step": 560,
|
|
"valid_targets_mean": 1524.5,
|
|
"valid_targets_min": 722
|
|
},
|
|
{
|
|
"epoch": 0.786908077994429,
|
|
"grad_norm": 0.8799325389323717,
|
|
"learning_rate": 3.998205095817833e-05,
|
|
"loss": 0.2207,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10128602385520935,
|
|
"step": 565,
|
|
"valid_targets_mean": 1289.2,
|
|
"valid_targets_min": 661
|
|
},
|
|
{
|
|
"epoch": 0.7938718662952646,
|
|
"grad_norm": 0.9446705425289179,
|
|
"learning_rate": 3.997898843587727e-05,
|
|
"loss": 0.225,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1078827977180481,
|
|
"step": 570,
|
|
"valid_targets_mean": 1519.5,
|
|
"valid_targets_min": 1011
|
|
},
|
|
{
|
|
"epoch": 0.8008356545961003,
|
|
"grad_norm": 0.8325950401643598,
|
|
"learning_rate": 3.997568494542946e-05,
|
|
"loss": 0.2356,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12279678881168365,
|
|
"step": 575,
|
|
"valid_targets_mean": 1731.4,
|
|
"valid_targets_min": 1277
|
|
},
|
|
{
|
|
"epoch": 0.807799442896936,
|
|
"grad_norm": 0.898136637442764,
|
|
"learning_rate": 3.997214052667856e-05,
|
|
"loss": 0.2232,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11773974448442459,
|
|
"step": 580,
|
|
"valid_targets_mean": 1486.1,
|
|
"valid_targets_min": 807
|
|
},
|
|
{
|
|
"epoch": 0.8147632311977716,
|
|
"grad_norm": 0.8718990576135462,
|
|
"learning_rate": 3.996835522237408e-05,
|
|
"loss": 0.237,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13190338015556335,
|
|
"step": 585,
|
|
"valid_targets_mean": 1549.9,
|
|
"valid_targets_min": 679
|
|
},
|
|
{
|
|
"epoch": 0.8217270194986073,
|
|
"grad_norm": 0.9293519703318798,
|
|
"learning_rate": 3.9964329078170874e-05,
|
|
"loss": 0.2326,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11136806011199951,
|
|
"step": 590,
|
|
"valid_targets_mean": 1656.0,
|
|
"valid_targets_min": 1294
|
|
},
|
|
{
|
|
"epoch": 0.8286908077994429,
|
|
"grad_norm": 0.9027962521886033,
|
|
"learning_rate": 3.996006214262858e-05,
|
|
"loss": 0.2327,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10627582669258118,
|
|
"step": 595,
|
|
"valid_targets_mean": 1368.8,
|
|
"valid_targets_min": 790
|
|
},
|
|
{
|
|
"epoch": 0.8356545961002786,
|
|
"grad_norm": 0.9202444501195384,
|
|
"learning_rate": 3.995555446721104e-05,
|
|
"loss": 0.2293,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12929676473140717,
|
|
"step": 600,
|
|
"valid_targets_mean": 1765.8,
|
|
"valid_targets_min": 592
|
|
},
|
|
{
|
|
"epoch": 0.8426183844011143,
|
|
"grad_norm": 0.8667866984685644,
|
|
"learning_rate": 3.9950806106285694e-05,
|
|
"loss": 0.2347,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12393373250961304,
|
|
"step": 605,
|
|
"valid_targets_mean": 1586.8,
|
|
"valid_targets_min": 662
|
|
},
|
|
{
|
|
"epoch": 0.8495821727019499,
|
|
"grad_norm": 0.9721183405017587,
|
|
"learning_rate": 3.9945817117122884e-05,
|
|
"loss": 0.2379,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1137949600815773,
|
|
"step": 610,
|
|
"valid_targets_mean": 1336.9,
|
|
"valid_targets_min": 622
|
|
},
|
|
{
|
|
"epoch": 0.8565459610027855,
|
|
"grad_norm": 0.9071022828100963,
|
|
"learning_rate": 3.994058755989521e-05,
|
|
"loss": 0.234,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07271943986415863,
|
|
"step": 615,
|
|
"valid_targets_mean": 1097.1,
|
|
"valid_targets_min": 625
|
|
},
|
|
{
|
|
"epoch": 0.8635097493036211,
|
|
"grad_norm": 0.9327235573302172,
|
|
"learning_rate": 3.993511749767676e-05,
|
|
"loss": 0.2465,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09269745647907257,
|
|
"step": 620,
|
|
"valid_targets_mean": 1484.9,
|
|
"valid_targets_min": 888
|
|
},
|
|
{
|
|
"epoch": 0.8704735376044568,
|
|
"grad_norm": 0.8523136084427271,
|
|
"learning_rate": 3.992940699644238e-05,
|
|
"loss": 0.2387,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12552563846111298,
|
|
"step": 625,
|
|
"valid_targets_mean": 1757.0,
|
|
"valid_targets_min": 635
|
|
},
|
|
{
|
|
"epoch": 0.8774373259052924,
|
|
"grad_norm": 0.9675785553764196,
|
|
"learning_rate": 3.9923456125066886e-05,
|
|
"loss": 0.2412,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1370333880186081,
|
|
"step": 630,
|
|
"valid_targets_mean": 1476.1,
|
|
"valid_targets_min": 522
|
|
},
|
|
{
|
|
"epoch": 0.8844011142061281,
|
|
"grad_norm": 0.9240372392580922,
|
|
"learning_rate": 3.9917264955324205e-05,
|
|
"loss": 0.2281,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10322962701320648,
|
|
"step": 635,
|
|
"valid_targets_mean": 1230.9,
|
|
"valid_targets_min": 697
|
|
},
|
|
{
|
|
"epoch": 0.8913649025069638,
|
|
"grad_norm": 0.8039407980641069,
|
|
"learning_rate": 3.991083356188651e-05,
|
|
"loss": 0.2261,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12441124022006989,
|
|
"step": 640,
|
|
"valid_targets_mean": 1671.0,
|
|
"valid_targets_min": 1010
|
|
},
|
|
{
|
|
"epoch": 0.8983286908077994,
|
|
"grad_norm": 0.901298627076697,
|
|
"learning_rate": 3.9904162022323356e-05,
|
|
"loss": 0.231,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11574780941009521,
|
|
"step": 645,
|
|
"valid_targets_mean": 1411.0,
|
|
"valid_targets_min": 769
|
|
},
|
|
{
|
|
"epoch": 0.9052924791086351,
|
|
"grad_norm": 0.8076274204627811,
|
|
"learning_rate": 3.9897250417100685e-05,
|
|
"loss": 0.2334,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.097521111369133,
|
|
"step": 650,
|
|
"valid_targets_mean": 1298.6,
|
|
"valid_targets_min": 745
|
|
},
|
|
{
|
|
"epoch": 0.9122562674094707,
|
|
"grad_norm": 0.8790050634268944,
|
|
"learning_rate": 3.9890098829579936e-05,
|
|
"loss": 0.2284,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09707553684711456,
|
|
"step": 655,
|
|
"valid_targets_mean": 1118.9,
|
|
"valid_targets_min": 680
|
|
},
|
|
{
|
|
"epoch": 0.9192200557103064,
|
|
"grad_norm": 0.8222830712195175,
|
|
"learning_rate": 3.988270734601694e-05,
|
|
"loss": 0.222,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13638588786125183,
|
|
"step": 660,
|
|
"valid_targets_mean": 1905.9,
|
|
"valid_targets_min": 1347
|
|
},
|
|
{
|
|
"epoch": 0.9261838440111421,
|
|
"grad_norm": 0.9010068300752175,
|
|
"learning_rate": 3.987507605556098e-05,
|
|
"loss": 0.2376,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16486553847789764,
|
|
"step": 665,
|
|
"valid_targets_mean": 1636.0,
|
|
"valid_targets_min": 851
|
|
},
|
|
{
|
|
"epoch": 0.9331476323119777,
|
|
"grad_norm": 0.8103801455626374,
|
|
"learning_rate": 3.9867205050253635e-05,
|
|
"loss": 0.224,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09920711815357208,
|
|
"step": 670,
|
|
"valid_targets_mean": 1457.2,
|
|
"valid_targets_min": 903
|
|
},
|
|
{
|
|
"epoch": 0.9401114206128134,
|
|
"grad_norm": 0.8043472181124449,
|
|
"learning_rate": 3.9859094425027726e-05,
|
|
"loss": 0.2217,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12532712519168854,
|
|
"step": 675,
|
|
"valid_targets_mean": 1674.4,
|
|
"valid_targets_min": 1240
|
|
},
|
|
{
|
|
"epoch": 0.947075208913649,
|
|
"grad_norm": 0.8118897965994559,
|
|
"learning_rate": 3.985074427770614e-05,
|
|
"loss": 0.2201,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09223470091819763,
|
|
"step": 680,
|
|
"valid_targets_mean": 1697.1,
|
|
"valid_targets_min": 917
|
|
},
|
|
{
|
|
"epoch": 0.9540389972144847,
|
|
"grad_norm": 0.9472041237654983,
|
|
"learning_rate": 3.9842154709000665e-05,
|
|
"loss": 0.2071,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09532526135444641,
|
|
"step": 685,
|
|
"valid_targets_mean": 1580.9,
|
|
"valid_targets_min": 930
|
|
},
|
|
{
|
|
"epoch": 0.9610027855153204,
|
|
"grad_norm": 0.8083428002058612,
|
|
"learning_rate": 3.983332582251074e-05,
|
|
"loss": 0.2289,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11902359873056412,
|
|
"step": 690,
|
|
"valid_targets_mean": 1655.1,
|
|
"valid_targets_min": 1227
|
|
},
|
|
{
|
|
"epoch": 0.967966573816156,
|
|
"grad_norm": 0.8148921698054701,
|
|
"learning_rate": 3.982425772472228e-05,
|
|
"loss": 0.2243,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11095002293586731,
|
|
"step": 695,
|
|
"valid_targets_mean": 1565.8,
|
|
"valid_targets_min": 870
|
|
},
|
|
{
|
|
"epoch": 0.9749303621169917,
|
|
"grad_norm": 2.1396789291672396,
|
|
"learning_rate": 3.9814950525006315e-05,
|
|
"loss": 0.2271,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10898259282112122,
|
|
"step": 700,
|
|
"valid_targets_mean": 1564.6,
|
|
"valid_targets_min": 822
|
|
},
|
|
{
|
|
"epoch": 0.9818941504178273,
|
|
"grad_norm": 0.8780065601127116,
|
|
"learning_rate": 3.980540433561771e-05,
|
|
"loss": 0.2318,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10233960300683975,
|
|
"step": 705,
|
|
"valid_targets_mean": 1540.5,
|
|
"valid_targets_min": 1214
|
|
},
|
|
{
|
|
"epoch": 0.9888579387186629,
|
|
"grad_norm": 0.8908774625256577,
|
|
"learning_rate": 3.979561927169381e-05,
|
|
"loss": 0.2206,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10410676151514053,
|
|
"step": 710,
|
|
"valid_targets_mean": 1268.6,
|
|
"valid_targets_min": 591
|
|
},
|
|
{
|
|
"epoch": 0.9958217270194986,
|
|
"grad_norm": 0.8377782810822102,
|
|
"learning_rate": 3.978559545125303e-05,
|
|
"loss": 0.2351,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08468751609325409,
|
|
"step": 715,
|
|
"valid_targets_mean": 1232.5,
|
|
"valid_targets_min": 646
|
|
},
|
|
{
|
|
"epoch": 1.0027855153203342,
|
|
"grad_norm": 0.5045448459523831,
|
|
"learning_rate": 3.977533299519347e-05,
|
|
"loss": 0.2503,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12714359164237976,
|
|
"step": 720,
|
|
"valid_targets_mean": 8514.0,
|
|
"valid_targets_min": 6788
|
|
},
|
|
{
|
|
"epoch": 1.00974930362117,
|
|
"grad_norm": 0.5387200781137093,
|
|
"learning_rate": 3.97648320272914e-05,
|
|
"loss": 0.2851,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13830497860908508,
|
|
"step": 725,
|
|
"valid_targets_mean": 6668.0,
|
|
"valid_targets_min": 5933
|
|
},
|
|
{
|
|
"epoch": 1.0167130919220055,
|
|
"grad_norm": 0.6200314371011584,
|
|
"learning_rate": 3.9754092674199816e-05,
|
|
"loss": 0.2725,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12560366094112396,
|
|
"step": 730,
|
|
"valid_targets_mean": 4107.0,
|
|
"valid_targets_min": 161
|
|
},
|
|
{
|
|
"epoch": 1.0236768802228413,
|
|
"grad_norm": 0.41903186731709363,
|
|
"learning_rate": 3.974311506544692e-05,
|
|
"loss": 0.2614,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11801932752132416,
|
|
"step": 735,
|
|
"valid_targets_mean": 7946.0,
|
|
"valid_targets_min": 5596
|
|
},
|
|
{
|
|
"epoch": 1.0306406685236769,
|
|
"grad_norm": 0.43595924603479874,
|
|
"learning_rate": 3.973189933343449e-05,
|
|
"loss": 0.248,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13217604160308838,
|
|
"step": 740,
|
|
"valid_targets_mean": 7392.8,
|
|
"valid_targets_min": 5629
|
|
},
|
|
{
|
|
"epoch": 1.0376044568245126,
|
|
"grad_norm": 0.5245267930858062,
|
|
"learning_rate": 3.972044561343636e-05,
|
|
"loss": 0.2511,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12668457627296448,
|
|
"step": 745,
|
|
"valid_targets_mean": 6879.6,
|
|
"valid_targets_min": 5113
|
|
},
|
|
{
|
|
"epoch": 1.0445682451253482,
|
|
"grad_norm": 0.4794835613989842,
|
|
"learning_rate": 3.970875404359675e-05,
|
|
"loss": 0.27,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1310746967792511,
|
|
"step": 750,
|
|
"valid_targets_mean": 6919.5,
|
|
"valid_targets_min": 5080
|
|
},
|
|
{
|
|
"epoch": 1.051532033426184,
|
|
"grad_norm": 0.47039907099593237,
|
|
"learning_rate": 3.969682476492858e-05,
|
|
"loss": 0.2624,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1175151839852333,
|
|
"step": 755,
|
|
"valid_targets_mean": 6377.0,
|
|
"valid_targets_min": 3939
|
|
},
|
|
{
|
|
"epoch": 1.0584958217270195,
|
|
"grad_norm": 0.4472868406904229,
|
|
"learning_rate": 3.968465792131184e-05,
|
|
"loss": 0.2482,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12472754716873169,
|
|
"step": 760,
|
|
"valid_targets_mean": 6392.2,
|
|
"valid_targets_min": 4355
|
|
},
|
|
{
|
|
"epoch": 1.065459610027855,
|
|
"grad_norm": 0.5109948443641175,
|
|
"learning_rate": 3.9672253659491776e-05,
|
|
"loss": 0.2449,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14166294038295746,
|
|
"step": 765,
|
|
"valid_targets_mean": 5196.5,
|
|
"valid_targets_min": 137
|
|
},
|
|
{
|
|
"epoch": 1.0724233983286908,
|
|
"grad_norm": 0.3950884119973961,
|
|
"learning_rate": 3.965961212907716e-05,
|
|
"loss": 0.2446,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10841463506221771,
|
|
"step": 770,
|
|
"valid_targets_mean": 7344.4,
|
|
"valid_targets_min": 5244
|
|
},
|
|
{
|
|
"epoch": 1.0793871866295264,
|
|
"grad_norm": 0.39325931847169565,
|
|
"learning_rate": 3.9646733482538486e-05,
|
|
"loss": 0.2358,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10372970998287201,
|
|
"step": 775,
|
|
"valid_targets_mean": 7921.0,
|
|
"valid_targets_min": 5311
|
|
},
|
|
{
|
|
"epoch": 1.0863509749303621,
|
|
"grad_norm": 0.39685130314701833,
|
|
"learning_rate": 3.963361787520612e-05,
|
|
"loss": 0.2462,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12329264730215073,
|
|
"step": 780,
|
|
"valid_targets_mean": 7931.2,
|
|
"valid_targets_min": 6015
|
|
},
|
|
{
|
|
"epoch": 1.0933147632311977,
|
|
"grad_norm": 0.36954967114170856,
|
|
"learning_rate": 3.9620265465268423e-05,
|
|
"loss": 0.2338,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11054964363574982,
|
|
"step": 785,
|
|
"valid_targets_mean": 7640.8,
|
|
"valid_targets_min": 5248
|
|
},
|
|
{
|
|
"epoch": 1.1002785515320335,
|
|
"grad_norm": 0.39689900322648536,
|
|
"learning_rate": 3.960667641376987e-05,
|
|
"loss": 0.232,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11496052145957947,
|
|
"step": 790,
|
|
"valid_targets_mean": 7249.6,
|
|
"valid_targets_min": 5217
|
|
},
|
|
{
|
|
"epoch": 1.107242339832869,
|
|
"grad_norm": 0.433531433289986,
|
|
"learning_rate": 3.959285088460908e-05,
|
|
"loss": 0.2396,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13216888904571533,
|
|
"step": 795,
|
|
"valid_targets_mean": 7619.9,
|
|
"valid_targets_min": 5094
|
|
},
|
|
{
|
|
"epoch": 1.1142061281337048,
|
|
"grad_norm": 0.39505484410680386,
|
|
"learning_rate": 3.957878904453683e-05,
|
|
"loss": 0.2324,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11113603413105011,
|
|
"step": 800,
|
|
"valid_targets_mean": 7732.8,
|
|
"valid_targets_min": 5842
|
|
},
|
|
{
|
|
"epoch": 1.1211699164345403,
|
|
"grad_norm": 0.4052047229981209,
|
|
"learning_rate": 3.9564491063154084e-05,
|
|
"loss": 0.2346,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12098728865385056,
|
|
"step": 805,
|
|
"valid_targets_mean": 7952.9,
|
|
"valid_targets_min": 5494
|
|
},
|
|
{
|
|
"epoch": 1.128133704735376,
|
|
"grad_norm": 0.417279220495771,
|
|
"learning_rate": 3.9549957112909914e-05,
|
|
"loss": 0.255,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12448830902576447,
|
|
"step": 810,
|
|
"valid_targets_mean": 8001.0,
|
|
"valid_targets_min": 5454
|
|
},
|
|
{
|
|
"epoch": 1.1350974930362117,
|
|
"grad_norm": 0.5319248244035725,
|
|
"learning_rate": 3.9535187369099437e-05,
|
|
"loss": 0.2584,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09340626001358032,
|
|
"step": 815,
|
|
"valid_targets_mean": 2985.1,
|
|
"valid_targets_min": 1896
|
|
},
|
|
{
|
|
"epoch": 1.1420612813370474,
|
|
"grad_norm": 0.39891658368515515,
|
|
"learning_rate": 3.952018200986169e-05,
|
|
"loss": 0.2183,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10278548300266266,
|
|
"step": 820,
|
|
"valid_targets_mean": 6714.6,
|
|
"valid_targets_min": 4917
|
|
},
|
|
{
|
|
"epoch": 1.149025069637883,
|
|
"grad_norm": 0.4249036845709375,
|
|
"learning_rate": 3.950494121617749e-05,
|
|
"loss": 0.2296,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1260872483253479,
|
|
"step": 825,
|
|
"valid_targets_mean": 7595.2,
|
|
"valid_targets_min": 5488
|
|
},
|
|
{
|
|
"epoch": 1.1559888579387188,
|
|
"grad_norm": 0.45070868901900607,
|
|
"learning_rate": 3.9489465171867254e-05,
|
|
"loss": 0.2326,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1257646679878235,
|
|
"step": 830,
|
|
"valid_targets_mean": 7046.1,
|
|
"valid_targets_min": 5579
|
|
},
|
|
{
|
|
"epoch": 1.1629526462395543,
|
|
"grad_norm": 0.4439435807963078,
|
|
"learning_rate": 3.947375406358874e-05,
|
|
"loss": 0.2402,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1169155091047287,
|
|
"step": 835,
|
|
"valid_targets_mean": 7108.0,
|
|
"valid_targets_min": 4963
|
|
},
|
|
{
|
|
"epoch": 1.16991643454039,
|
|
"grad_norm": 0.44221583815751264,
|
|
"learning_rate": 3.9457808080834884e-05,
|
|
"loss": 0.2344,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11854035407304764,
|
|
"step": 840,
|
|
"valid_targets_mean": 6810.1,
|
|
"valid_targets_min": 5666
|
|
},
|
|
{
|
|
"epoch": 1.1768802228412256,
|
|
"grad_norm": 0.4316908401901056,
|
|
"learning_rate": 3.944162741593142e-05,
|
|
"loss": 0.225,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11046448349952698,
|
|
"step": 845,
|
|
"valid_targets_mean": 8068.5,
|
|
"valid_targets_min": 5524
|
|
},
|
|
{
|
|
"epoch": 1.1838440111420612,
|
|
"grad_norm": 0.4180604360052473,
|
|
"learning_rate": 3.9425212264034626e-05,
|
|
"loss": 0.2035,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09698522835969925,
|
|
"step": 850,
|
|
"valid_targets_mean": 6553.8,
|
|
"valid_targets_min": 3621
|
|
},
|
|
{
|
|
"epoch": 1.190807799442897,
|
|
"grad_norm": 0.4189841877717661,
|
|
"learning_rate": 3.940856282312893e-05,
|
|
"loss": 0.2085,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10995703935623169,
|
|
"step": 855,
|
|
"valid_targets_mean": 6762.5,
|
|
"valid_targets_min": 4968
|
|
},
|
|
{
|
|
"epoch": 1.1977715877437327,
|
|
"grad_norm": 0.45595104424848715,
|
|
"learning_rate": 3.9391679294024543e-05,
|
|
"loss": 0.2224,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11162762343883514,
|
|
"step": 860,
|
|
"valid_targets_mean": 6143.9,
|
|
"valid_targets_min": 4917
|
|
},
|
|
{
|
|
"epoch": 1.2047353760445683,
|
|
"grad_norm": 0.5851346783403598,
|
|
"learning_rate": 3.937456188035504e-05,
|
|
"loss": 0.2377,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16354359686374664,
|
|
"step": 865,
|
|
"valid_targets_mean": 4583.8,
|
|
"valid_targets_min": 215
|
|
},
|
|
{
|
|
"epoch": 1.2116991643454038,
|
|
"grad_norm": 0.4055676263805544,
|
|
"learning_rate": 3.935721078857489e-05,
|
|
"loss": 0.2371,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11995077133178711,
|
|
"step": 870,
|
|
"valid_targets_mean": 7317.1,
|
|
"valid_targets_min": 4948
|
|
},
|
|
{
|
|
"epoch": 1.2186629526462396,
|
|
"grad_norm": 0.42172517542326765,
|
|
"learning_rate": 3.9339626227956966e-05,
|
|
"loss": 0.2521,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12088988721370697,
|
|
"step": 875,
|
|
"valid_targets_mean": 6437.6,
|
|
"valid_targets_min": 4510
|
|
},
|
|
{
|
|
"epoch": 1.2256267409470751,
|
|
"grad_norm": 0.4535203146647295,
|
|
"learning_rate": 3.932180841059004e-05,
|
|
"loss": 0.2433,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13094569742679596,
|
|
"step": 880,
|
|
"valid_targets_mean": 7707.9,
|
|
"valid_targets_min": 5796
|
|
},
|
|
{
|
|
"epoch": 1.232590529247911,
|
|
"grad_norm": 0.418577272093497,
|
|
"learning_rate": 3.93037575513762e-05,
|
|
"loss": 0.2305,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11241194605827332,
|
|
"step": 885,
|
|
"valid_targets_mean": 7900.9,
|
|
"valid_targets_min": 5643
|
|
},
|
|
{
|
|
"epoch": 1.2395543175487465,
|
|
"grad_norm": 0.4344807498847784,
|
|
"learning_rate": 3.9285473868028287e-05,
|
|
"loss": 0.2308,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12634864449501038,
|
|
"step": 890,
|
|
"valid_targets_mean": 7627.2,
|
|
"valid_targets_min": 5742
|
|
},
|
|
{
|
|
"epoch": 1.2465181058495822,
|
|
"grad_norm": 0.40801176738421346,
|
|
"learning_rate": 3.9266957581067215e-05,
|
|
"loss": 0.2346,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13307298719882965,
|
|
"step": 895,
|
|
"valid_targets_mean": 8725.6,
|
|
"valid_targets_min": 5141
|
|
},
|
|
{
|
|
"epoch": 1.2534818941504178,
|
|
"grad_norm": 0.4451588968606175,
|
|
"learning_rate": 3.924820891381939e-05,
|
|
"loss": 0.2309,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12867535650730133,
|
|
"step": 900,
|
|
"valid_targets_mean": 6967.8,
|
|
"valid_targets_min": 6088
|
|
},
|
|
{
|
|
"epoch": 1.2604456824512535,
|
|
"grad_norm": 0.4207887725907819,
|
|
"learning_rate": 3.922922809241397e-05,
|
|
"loss": 0.2333,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10731957852840424,
|
|
"step": 905,
|
|
"valid_targets_mean": 7140.1,
|
|
"valid_targets_min": 4834
|
|
},
|
|
{
|
|
"epoch": 1.267409470752089,
|
|
"grad_norm": 0.43935252315939416,
|
|
"learning_rate": 3.92100153457801e-05,
|
|
"loss": 0.2313,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10727177560329437,
|
|
"step": 910,
|
|
"valid_targets_mean": 6706.9,
|
|
"valid_targets_min": 5108
|
|
},
|
|
{
|
|
"epoch": 1.2743732590529249,
|
|
"grad_norm": 0.4987156443279082,
|
|
"learning_rate": 3.9190570905644254e-05,
|
|
"loss": 0.236,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11678919196128845,
|
|
"step": 915,
|
|
"valid_targets_mean": 5931.4,
|
|
"valid_targets_min": 4708
|
|
},
|
|
{
|
|
"epoch": 1.2813370473537604,
|
|
"grad_norm": 0.4407354572740225,
|
|
"learning_rate": 3.917089500652733e-05,
|
|
"loss": 0.2273,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10812367498874664,
|
|
"step": 920,
|
|
"valid_targets_mean": 7145.6,
|
|
"valid_targets_min": 5280
|
|
},
|
|
{
|
|
"epoch": 1.2883008356545962,
|
|
"grad_norm": 0.7644739613411758,
|
|
"learning_rate": 3.915098788574191e-05,
|
|
"loss": 0.2231,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0803007185459137,
|
|
"step": 925,
|
|
"valid_targets_mean": 1554.9,
|
|
"valid_targets_min": 592
|
|
},
|
|
{
|
|
"epoch": 1.2952646239554317,
|
|
"grad_norm": 0.45530435313643286,
|
|
"learning_rate": 3.913084978338931e-05,
|
|
"loss": 0.2281,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12958484888076782,
|
|
"step": 930,
|
|
"valid_targets_mean": 7069.5,
|
|
"valid_targets_min": 4765
|
|
},
|
|
{
|
|
"epoch": 1.3022284122562673,
|
|
"grad_norm": 0.4187104102266161,
|
|
"learning_rate": 3.911048094235679e-05,
|
|
"loss": 0.2262,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11318781226873398,
|
|
"step": 935,
|
|
"valid_targets_mean": 6193.4,
|
|
"valid_targets_min": 5587
|
|
},
|
|
{
|
|
"epoch": 1.309192200557103,
|
|
"grad_norm": 0.4642694802664071,
|
|
"learning_rate": 3.9089881608314534e-05,
|
|
"loss": 0.2268,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11810522526502609,
|
|
"step": 940,
|
|
"valid_targets_mean": 6612.2,
|
|
"valid_targets_min": 4894
|
|
},
|
|
{
|
|
"epoch": 1.3161559888579388,
|
|
"grad_norm": 0.40454438050769786,
|
|
"learning_rate": 3.9069052029712713e-05,
|
|
"loss": 0.2201,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10395780950784683,
|
|
"step": 945,
|
|
"valid_targets_mean": 6605.9,
|
|
"valid_targets_min": 3795
|
|
},
|
|
{
|
|
"epoch": 1.3231197771587744,
|
|
"grad_norm": 0.4215605431440107,
|
|
"learning_rate": 3.9047992457778524e-05,
|
|
"loss": 0.2247,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11420217156410217,
|
|
"step": 950,
|
|
"valid_targets_mean": 7186.8,
|
|
"valid_targets_min": 4890
|
|
},
|
|
{
|
|
"epoch": 1.33008356545961,
|
|
"grad_norm": 0.620429539040831,
|
|
"learning_rate": 3.90267031465131e-05,
|
|
"loss": 0.2287,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1147976964712143,
|
|
"step": 955,
|
|
"valid_targets_mean": 6544.4,
|
|
"valid_targets_min": 4615
|
|
},
|
|
{
|
|
"epoch": 1.3370473537604457,
|
|
"grad_norm": 0.47863602052631893,
|
|
"learning_rate": 3.9005184352688514e-05,
|
|
"loss": 0.2318,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1220189705491066,
|
|
"step": 960,
|
|
"valid_targets_mean": 6339.2,
|
|
"valid_targets_min": 4798
|
|
},
|
|
{
|
|
"epoch": 1.3440111420612815,
|
|
"grad_norm": 0.6133989682966983,
|
|
"learning_rate": 3.89834363358446e-05,
|
|
"loss": 0.1994,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1313786804676056,
|
|
"step": 965,
|
|
"valid_targets_mean": 4558.8,
|
|
"valid_targets_min": 148
|
|
},
|
|
{
|
|
"epoch": 1.350974930362117,
|
|
"grad_norm": 0.45702297334659797,
|
|
"learning_rate": 3.896145935828592e-05,
|
|
"loss": 0.2258,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.124158576130867,
|
|
"step": 970,
|
|
"valid_targets_mean": 6807.9,
|
|
"valid_targets_min": 5164
|
|
},
|
|
{
|
|
"epoch": 1.3579387186629526,
|
|
"grad_norm": 0.4313724233091901,
|
|
"learning_rate": 3.893925368507852e-05,
|
|
"loss": 0.2178,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11191289126873016,
|
|
"step": 975,
|
|
"valid_targets_mean": 7011.8,
|
|
"valid_targets_min": 4974
|
|
},
|
|
{
|
|
"epoch": 1.3649025069637883,
|
|
"grad_norm": 0.46720364315569657,
|
|
"learning_rate": 3.891681958404675e-05,
|
|
"loss": 0.2258,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10171587765216827,
|
|
"step": 980,
|
|
"valid_targets_mean": 5794.2,
|
|
"valid_targets_min": 4809
|
|
},
|
|
{
|
|
"epoch": 1.371866295264624,
|
|
"grad_norm": 0.43110533517562993,
|
|
"learning_rate": 3.8894157325770085e-05,
|
|
"loss": 0.2137,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10188580304384232,
|
|
"step": 985,
|
|
"valid_targets_mean": 6401.8,
|
|
"valid_targets_min": 4842
|
|
},
|
|
{
|
|
"epoch": 1.3788300835654597,
|
|
"grad_norm": 0.44932522936305014,
|
|
"learning_rate": 3.887126718357978e-05,
|
|
"loss": 0.2193,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1125284880399704,
|
|
"step": 990,
|
|
"valid_targets_mean": 6182.1,
|
|
"valid_targets_min": 5116
|
|
},
|
|
{
|
|
"epoch": 1.3857938718662952,
|
|
"grad_norm": 0.42140650517718253,
|
|
"learning_rate": 3.884814943355565e-05,
|
|
"loss": 0.2192,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10108678042888641,
|
|
"step": 995,
|
|
"valid_targets_mean": 6165.1,
|
|
"valid_targets_min": 5764
|
|
},
|
|
{
|
|
"epoch": 1.392757660167131,
|
|
"grad_norm": 0.9460709949341402,
|
|
"learning_rate": 3.882480435452268e-05,
|
|
"loss": 0.2509,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09195755422115326,
|
|
"step": 1000,
|
|
"valid_targets_mean": 1428.2,
|
|
"valid_targets_min": 1062
|
|
},
|
|
{
|
|
"epoch": 1.3997214484679665,
|
|
"grad_norm": 1.0910007125977519,
|
|
"learning_rate": 3.88012322280477e-05,
|
|
"loss": 0.2267,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10902673006057739,
|
|
"step": 1005,
|
|
"valid_targets_mean": 1663.0,
|
|
"valid_targets_min": 960
|
|
},
|
|
{
|
|
"epoch": 1.4066852367688023,
|
|
"grad_norm": 0.8694966611159818,
|
|
"learning_rate": 3.877743333843599e-05,
|
|
"loss": 0.2226,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10752002894878387,
|
|
"step": 1010,
|
|
"valid_targets_mean": 1476.5,
|
|
"valid_targets_min": 933
|
|
},
|
|
{
|
|
"epoch": 1.4136490250696379,
|
|
"grad_norm": 0.8363762394596167,
|
|
"learning_rate": 3.875340797272781e-05,
|
|
"loss": 0.2181,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10708215087652206,
|
|
"step": 1015,
|
|
"valid_targets_mean": 1478.1,
|
|
"valid_targets_min": 881
|
|
},
|
|
{
|
|
"epoch": 1.4206128133704734,
|
|
"grad_norm": 0.9058452941123465,
|
|
"learning_rate": 3.872915642069499e-05,
|
|
"loss": 0.2195,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1283268928527832,
|
|
"step": 1020,
|
|
"valid_targets_mean": 1640.6,
|
|
"valid_targets_min": 869
|
|
},
|
|
{
|
|
"epoch": 1.4275766016713092,
|
|
"grad_norm": 1.1890706352537899,
|
|
"learning_rate": 3.87046789748374e-05,
|
|
"loss": 0.2227,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11645996570587158,
|
|
"step": 1025,
|
|
"valid_targets_mean": 1709.0,
|
|
"valid_targets_min": 1090
|
|
},
|
|
{
|
|
"epoch": 1.434540389972145,
|
|
"grad_norm": 0.8343148724615661,
|
|
"learning_rate": 3.867997593037943e-05,
|
|
"loss": 0.2117,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08550204336643219,
|
|
"step": 1030,
|
|
"valid_targets_mean": 1157.9,
|
|
"valid_targets_min": 761
|
|
},
|
|
{
|
|
"epoch": 1.4415041782729805,
|
|
"grad_norm": 0.7596563407523057,
|
|
"learning_rate": 3.865504758526644e-05,
|
|
"loss": 0.2283,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1036420613527298,
|
|
"step": 1035,
|
|
"valid_targets_mean": 1553.6,
|
|
"valid_targets_min": 1241
|
|
},
|
|
{
|
|
"epoch": 1.448467966573816,
|
|
"grad_norm": 0.8589106034441814,
|
|
"learning_rate": 3.8629894240161153e-05,
|
|
"loss": 0.2103,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09890624135732651,
|
|
"step": 1040,
|
|
"valid_targets_mean": 1538.1,
|
|
"valid_targets_min": 733
|
|
},
|
|
{
|
|
"epoch": 1.4554317548746518,
|
|
"grad_norm": 0.7911151876930106,
|
|
"learning_rate": 3.8604516198440036e-05,
|
|
"loss": 0.217,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06528353691101074,
|
|
"step": 1045,
|
|
"valid_targets_mean": 1250.6,
|
|
"valid_targets_min": 835
|
|
},
|
|
{
|
|
"epoch": 1.4623955431754876,
|
|
"grad_norm": 0.787767064205962,
|
|
"learning_rate": 3.857891376618965e-05,
|
|
"loss": 0.2079,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12352828681468964,
|
|
"step": 1050,
|
|
"valid_targets_mean": 1596.2,
|
|
"valid_targets_min": 1046
|
|
},
|
|
{
|
|
"epoch": 1.4693593314763231,
|
|
"grad_norm": 0.8571652833546958,
|
|
"learning_rate": 3.855308725220293e-05,
|
|
"loss": 0.2211,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10267695039510727,
|
|
"step": 1055,
|
|
"valid_targets_mean": 1646.5,
|
|
"valid_targets_min": 797
|
|
},
|
|
{
|
|
"epoch": 1.4763231197771587,
|
|
"grad_norm": 0.9386572516522071,
|
|
"learning_rate": 3.85270369679755e-05,
|
|
"loss": 0.2091,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1088155061006546,
|
|
"step": 1060,
|
|
"valid_targets_mean": 1448.9,
|
|
"valid_targets_min": 570
|
|
},
|
|
{
|
|
"epoch": 1.4832869080779945,
|
|
"grad_norm": 0.8090527333348528,
|
|
"learning_rate": 3.850076322770186e-05,
|
|
"loss": 0.2018,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07706055045127869,
|
|
"step": 1065,
|
|
"valid_targets_mean": 1164.1,
|
|
"valid_targets_min": 548
|
|
},
|
|
{
|
|
"epoch": 1.49025069637883,
|
|
"grad_norm": 0.888098593113638,
|
|
"learning_rate": 3.847426634827167e-05,
|
|
"loss": 0.2002,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1056225374341011,
|
|
"step": 1070,
|
|
"valid_targets_mean": 1338.0,
|
|
"valid_targets_min": 682
|
|
},
|
|
{
|
|
"epoch": 1.4972144846796658,
|
|
"grad_norm": 0.7475892854829238,
|
|
"learning_rate": 3.844754664926586e-05,
|
|
"loss": 0.2168,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10795626044273376,
|
|
"step": 1075,
|
|
"valid_targets_mean": 1632.8,
|
|
"valid_targets_min": 896
|
|
},
|
|
{
|
|
"epoch": 1.5041782729805013,
|
|
"grad_norm": 0.7767225037367806,
|
|
"learning_rate": 3.842060445295283e-05,
|
|
"loss": 0.2178,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13912582397460938,
|
|
"step": 1080,
|
|
"valid_targets_mean": 1806.5,
|
|
"valid_targets_min": 597
|
|
},
|
|
{
|
|
"epoch": 1.511142061281337,
|
|
"grad_norm": 0.8482626365324997,
|
|
"learning_rate": 3.839344008428451e-05,
|
|
"loss": 0.2188,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09307943284511566,
|
|
"step": 1085,
|
|
"valid_targets_mean": 1289.9,
|
|
"valid_targets_min": 706
|
|
},
|
|
{
|
|
"epoch": 1.5181058495821727,
|
|
"grad_norm": 0.7497328533096577,
|
|
"learning_rate": 3.836605387089248e-05,
|
|
"loss": 0.2125,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06710005551576614,
|
|
"step": 1090,
|
|
"valid_targets_mean": 1354.1,
|
|
"valid_targets_min": 754
|
|
},
|
|
{
|
|
"epoch": 1.5250696378830084,
|
|
"grad_norm": 0.7897926297952617,
|
|
"learning_rate": 3.8338446143084015e-05,
|
|
"loss": 0.2073,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0963955968618393,
|
|
"step": 1095,
|
|
"valid_targets_mean": 1461.9,
|
|
"valid_targets_min": 819
|
|
},
|
|
{
|
|
"epoch": 1.532033426183844,
|
|
"grad_norm": 0.8054353338785877,
|
|
"learning_rate": 3.8310617233838094e-05,
|
|
"loss": 0.2164,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08029267191886902,
|
|
"step": 1100,
|
|
"valid_targets_mean": 1225.5,
|
|
"valid_targets_min": 684
|
|
},
|
|
{
|
|
"epoch": 1.5389972144846795,
|
|
"grad_norm": 0.9118859088683984,
|
|
"learning_rate": 3.828256747880137e-05,
|
|
"loss": 0.2219,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10978655517101288,
|
|
"step": 1105,
|
|
"valid_targets_mean": 1282.2,
|
|
"valid_targets_min": 711
|
|
},
|
|
{
|
|
"epoch": 1.5459610027855153,
|
|
"grad_norm": 0.7673580373354333,
|
|
"learning_rate": 3.825429721628411e-05,
|
|
"loss": 0.2065,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11380324512720108,
|
|
"step": 1110,
|
|
"valid_targets_mean": 1634.8,
|
|
"valid_targets_min": 1190
|
|
},
|
|
{
|
|
"epoch": 1.552924791086351,
|
|
"grad_norm": 0.8522675201936903,
|
|
"learning_rate": 3.822580678725621e-05,
|
|
"loss": 0.2015,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10675722360610962,
|
|
"step": 1115,
|
|
"valid_targets_mean": 1615.9,
|
|
"valid_targets_min": 774
|
|
},
|
|
{
|
|
"epoch": 1.5598885793871866,
|
|
"grad_norm": 0.7180000643578018,
|
|
"learning_rate": 3.819709653534295e-05,
|
|
"loss": 0.2055,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10761082917451859,
|
|
"step": 1120,
|
|
"valid_targets_mean": 1848.4,
|
|
"valid_targets_min": 734
|
|
},
|
|
{
|
|
"epoch": 1.5668523676880222,
|
|
"grad_norm": 0.8256061679097826,
|
|
"learning_rate": 3.8168166806820924e-05,
|
|
"loss": 0.2035,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10491684079170227,
|
|
"step": 1125,
|
|
"valid_targets_mean": 1570.6,
|
|
"valid_targets_min": 1011
|
|
},
|
|
{
|
|
"epoch": 1.573816155988858,
|
|
"grad_norm": 0.800219846734056,
|
|
"learning_rate": 3.813901795061387e-05,
|
|
"loss": 0.2072,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1188449040055275,
|
|
"step": 1130,
|
|
"valid_targets_mean": 1697.0,
|
|
"valid_targets_min": 803
|
|
},
|
|
{
|
|
"epoch": 1.5807799442896937,
|
|
"grad_norm": 0.7849874656744029,
|
|
"learning_rate": 3.810965031828843e-05,
|
|
"loss": 0.2079,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08988340198993683,
|
|
"step": 1135,
|
|
"valid_targets_mean": 1307.1,
|
|
"valid_targets_min": 782
|
|
},
|
|
{
|
|
"epoch": 1.5877437325905293,
|
|
"grad_norm": 0.7519337668408274,
|
|
"learning_rate": 3.808006426404991e-05,
|
|
"loss": 0.2009,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10832203924655914,
|
|
"step": 1140,
|
|
"valid_targets_mean": 1596.6,
|
|
"valid_targets_min": 646
|
|
},
|
|
{
|
|
"epoch": 1.5947075208913648,
|
|
"grad_norm": 0.7913090801912336,
|
|
"learning_rate": 3.805026014473803e-05,
|
|
"loss": 0.2113,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14530369639396667,
|
|
"step": 1145,
|
|
"valid_targets_mean": 1817.5,
|
|
"valid_targets_min": 922
|
|
},
|
|
{
|
|
"epoch": 1.6016713091922006,
|
|
"grad_norm": 0.8006210281974334,
|
|
"learning_rate": 3.802023831982262e-05,
|
|
"loss": 0.2089,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11583166569471359,
|
|
"step": 1150,
|
|
"valid_targets_mean": 1822.2,
|
|
"valid_targets_min": 796
|
|
},
|
|
{
|
|
"epoch": 1.6086350974930363,
|
|
"grad_norm": 0.8429365694331502,
|
|
"learning_rate": 3.798999915139926e-05,
|
|
"loss": 0.2013,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10505574196577072,
|
|
"step": 1155,
|
|
"valid_targets_mean": 1542.4,
|
|
"valid_targets_min": 903
|
|
},
|
|
{
|
|
"epoch": 1.615598885793872,
|
|
"grad_norm": 0.9071692515157804,
|
|
"learning_rate": 3.795954300418494e-05,
|
|
"loss": 0.1985,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11454838514328003,
|
|
"step": 1160,
|
|
"valid_targets_mean": 1459.5,
|
|
"valid_targets_min": 780
|
|
},
|
|
{
|
|
"epoch": 1.6225626740947074,
|
|
"grad_norm": 0.7700813936036295,
|
|
"learning_rate": 3.792887024551363e-05,
|
|
"loss": 0.1972,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10751248896121979,
|
|
"step": 1165,
|
|
"valid_targets_mean": 1581.1,
|
|
"valid_targets_min": 783
|
|
},
|
|
{
|
|
"epoch": 1.6295264623955432,
|
|
"grad_norm": 0.7587182482455052,
|
|
"learning_rate": 3.7897981245331895e-05,
|
|
"loss": 0.1996,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07905155420303345,
|
|
"step": 1170,
|
|
"valid_targets_mean": 1342.1,
|
|
"valid_targets_min": 662
|
|
},
|
|
{
|
|
"epoch": 1.636490250696379,
|
|
"grad_norm": 0.8016804160384552,
|
|
"learning_rate": 3.786687637619437e-05,
|
|
"loss": 0.2016,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08902774751186371,
|
|
"step": 1175,
|
|
"valid_targets_mean": 1269.1,
|
|
"valid_targets_min": 724
|
|
},
|
|
{
|
|
"epoch": 1.6434540389972145,
|
|
"grad_norm": 0.8762625648078411,
|
|
"learning_rate": 3.783555601325933e-05,
|
|
"loss": 0.1977,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11289331316947937,
|
|
"step": 1180,
|
|
"valid_targets_mean": 1510.1,
|
|
"valid_targets_min": 966
|
|
},
|
|
{
|
|
"epoch": 1.65041782729805,
|
|
"grad_norm": 0.8362696841731547,
|
|
"learning_rate": 3.7804020534284125e-05,
|
|
"loss": 0.2035,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08795975148677826,
|
|
"step": 1185,
|
|
"valid_targets_mean": 1327.8,
|
|
"valid_targets_min": 773
|
|
},
|
|
{
|
|
"epoch": 1.6573816155988856,
|
|
"grad_norm": 0.8217589431615828,
|
|
"learning_rate": 3.7772270319620655e-05,
|
|
"loss": 0.2014,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08339729905128479,
|
|
"step": 1190,
|
|
"valid_targets_mean": 1353.2,
|
|
"valid_targets_min": 807
|
|
},
|
|
{
|
|
"epoch": 1.6643454038997214,
|
|
"grad_norm": 0.8481129752613321,
|
|
"learning_rate": 3.774030575221073e-05,
|
|
"loss": 0.2051,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07550103962421417,
|
|
"step": 1195,
|
|
"valid_targets_mean": 1142.5,
|
|
"valid_targets_min": 782
|
|
},
|
|
{
|
|
"epoch": 1.6713091922005572,
|
|
"grad_norm": 0.7883067686921773,
|
|
"learning_rate": 3.770812721758151e-05,
|
|
"loss": 0.1964,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.112126424908638,
|
|
"step": 1200,
|
|
"valid_targets_mean": 1849.9,
|
|
"valid_targets_min": 634
|
|
},
|
|
{
|
|
"epoch": 1.6782729805013927,
|
|
"grad_norm": 0.7411457071163579,
|
|
"learning_rate": 3.7675735103840844e-05,
|
|
"loss": 0.2122,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10241081565618515,
|
|
"step": 1205,
|
|
"valid_targets_mean": 1599.6,
|
|
"valid_targets_min": 573
|
|
},
|
|
{
|
|
"epoch": 1.6852367688022283,
|
|
"grad_norm": 0.7086572852054175,
|
|
"learning_rate": 3.7643129801672535e-05,
|
|
"loss": 0.2029,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11208580434322357,
|
|
"step": 1210,
|
|
"valid_targets_mean": 1994.9,
|
|
"valid_targets_min": 1102
|
|
},
|
|
{
|
|
"epoch": 1.692200557103064,
|
|
"grad_norm": 0.8779677269614039,
|
|
"learning_rate": 3.76103117043317e-05,
|
|
"loss": 0.2095,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07884174585342407,
|
|
"step": 1215,
|
|
"valid_targets_mean": 1239.2,
|
|
"valid_targets_min": 841
|
|
},
|
|
{
|
|
"epoch": 1.6991643454038998,
|
|
"grad_norm": 0.7715841944120366,
|
|
"learning_rate": 3.757728120763998e-05,
|
|
"loss": 0.1967,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09706147015094757,
|
|
"step": 1220,
|
|
"valid_targets_mean": 1467.5,
|
|
"valid_targets_min": 956
|
|
},
|
|
{
|
|
"epoch": 1.7061281337047354,
|
|
"grad_norm": 0.8840496408494232,
|
|
"learning_rate": 3.754403870998079e-05,
|
|
"loss": 0.2035,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10566884279251099,
|
|
"step": 1225,
|
|
"valid_targets_mean": 1300.9,
|
|
"valid_targets_min": 857
|
|
},
|
|
{
|
|
"epoch": 1.713091922005571,
|
|
"grad_norm": 0.9178621719926586,
|
|
"learning_rate": 3.7510584612294506e-05,
|
|
"loss": 0.2045,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10823386162519455,
|
|
"step": 1230,
|
|
"valid_targets_mean": 1389.8,
|
|
"valid_targets_min": 808
|
|
},
|
|
{
|
|
"epoch": 1.7200557103064067,
|
|
"grad_norm": 0.938154160967897,
|
|
"learning_rate": 3.747691931807362e-05,
|
|
"loss": 0.1897,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08545945584774017,
|
|
"step": 1235,
|
|
"valid_targets_mean": 1303.8,
|
|
"valid_targets_min": 677
|
|
},
|
|
{
|
|
"epoch": 1.7270194986072425,
|
|
"grad_norm": 0.8217723401413547,
|
|
"learning_rate": 3.744304323335787e-05,
|
|
"loss": 0.1962,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0866972804069519,
|
|
"step": 1240,
|
|
"valid_targets_mean": 1220.1,
|
|
"valid_targets_min": 727
|
|
},
|
|
{
|
|
"epoch": 1.733983286908078,
|
|
"grad_norm": 0.7965558518328687,
|
|
"learning_rate": 3.7408956766729406e-05,
|
|
"loss": 0.1997,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0839056596159935,
|
|
"step": 1245,
|
|
"valid_targets_mean": 1262.0,
|
|
"valid_targets_min": 726
|
|
},
|
|
{
|
|
"epoch": 1.7409470752089136,
|
|
"grad_norm": 0.7985361430147639,
|
|
"learning_rate": 3.737466032930775e-05,
|
|
"loss": 0.2069,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11801943182945251,
|
|
"step": 1250,
|
|
"valid_targets_mean": 1615.0,
|
|
"valid_targets_min": 669
|
|
},
|
|
{
|
|
"epoch": 1.7479108635097493,
|
|
"grad_norm": 0.7163362819024491,
|
|
"learning_rate": 3.7340154334744933e-05,
|
|
"loss": 0.1998,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11403729021549225,
|
|
"step": 1255,
|
|
"valid_targets_mean": 1806.5,
|
|
"valid_targets_min": 792
|
|
},
|
|
{
|
|
"epoch": 1.754874651810585,
|
|
"grad_norm": 0.76114767590763,
|
|
"learning_rate": 3.7305439199220466e-05,
|
|
"loss": 0.197,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09969079494476318,
|
|
"step": 1260,
|
|
"valid_targets_mean": 1627.6,
|
|
"valid_targets_min": 736
|
|
},
|
|
{
|
|
"epoch": 1.7618384401114207,
|
|
"grad_norm": 0.7967512334914618,
|
|
"learning_rate": 3.727051534143631e-05,
|
|
"loss": 0.2039,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14327603578567505,
|
|
"step": 1265,
|
|
"valid_targets_mean": 1822.9,
|
|
"valid_targets_min": 694
|
|
},
|
|
{
|
|
"epoch": 1.7688022284122562,
|
|
"grad_norm": 0.7047699568272487,
|
|
"learning_rate": 3.723538318261186e-05,
|
|
"loss": 0.1954,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1259545087814331,
|
|
"step": 1270,
|
|
"valid_targets_mean": 1848.9,
|
|
"valid_targets_min": 1437
|
|
},
|
|
{
|
|
"epoch": 1.775766016713092,
|
|
"grad_norm": 0.7684434862741877,
|
|
"learning_rate": 3.720004314647884e-05,
|
|
"loss": 0.1958,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12424354255199432,
|
|
"step": 1275,
|
|
"valid_targets_mean": 1852.4,
|
|
"valid_targets_min": 809
|
|
},
|
|
{
|
|
"epoch": 1.7827298050139275,
|
|
"grad_norm": 0.7865791196195834,
|
|
"learning_rate": 3.7164495659276195e-05,
|
|
"loss": 0.1946,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09065093100070953,
|
|
"step": 1280,
|
|
"valid_targets_mean": 1375.5,
|
|
"valid_targets_min": 790
|
|
},
|
|
{
|
|
"epoch": 1.7896935933147633,
|
|
"grad_norm": 0.8290020685682563,
|
|
"learning_rate": 3.712874114974496e-05,
|
|
"loss": 0.1854,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0918719470500946,
|
|
"step": 1285,
|
|
"valid_targets_mean": 1338.8,
|
|
"valid_targets_min": 750
|
|
},
|
|
{
|
|
"epoch": 1.7966573816155988,
|
|
"grad_norm": 0.7994498344708911,
|
|
"learning_rate": 3.7092780049123076e-05,
|
|
"loss": 0.1984,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12238575518131256,
|
|
"step": 1290,
|
|
"valid_targets_mean": 1778.8,
|
|
"valid_targets_min": 826
|
|
},
|
|
{
|
|
"epoch": 1.8036211699164344,
|
|
"grad_norm": 0.7645358833299734,
|
|
"learning_rate": 3.7056612791140204e-05,
|
|
"loss": 0.1938,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08983242511749268,
|
|
"step": 1295,
|
|
"valid_targets_mean": 1600.8,
|
|
"valid_targets_min": 907
|
|
},
|
|
{
|
|
"epoch": 1.8105849582172702,
|
|
"grad_norm": 0.7839077663753583,
|
|
"learning_rate": 3.7020239812012473e-05,
|
|
"loss": 0.1917,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1056416928768158,
|
|
"step": 1300,
|
|
"valid_targets_mean": 1474.5,
|
|
"valid_targets_min": 1093
|
|
},
|
|
{
|
|
"epoch": 1.817548746518106,
|
|
"grad_norm": 0.76163948748665,
|
|
"learning_rate": 3.698366155043724e-05,
|
|
"loss": 0.1981,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10014639794826508,
|
|
"step": 1305,
|
|
"valid_targets_mean": 1577.2,
|
|
"valid_targets_min": 1119
|
|
},
|
|
{
|
|
"epoch": 1.8245125348189415,
|
|
"grad_norm": 1.7707818510491133,
|
|
"learning_rate": 3.694687844758779e-05,
|
|
"loss": 0.1962,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11413945257663727,
|
|
"step": 1310,
|
|
"valid_targets_mean": 1512.1,
|
|
"valid_targets_min": 700
|
|
},
|
|
{
|
|
"epoch": 1.831476323119777,
|
|
"grad_norm": 0.7668240309562283,
|
|
"learning_rate": 3.6909890947108016e-05,
|
|
"loss": 0.196,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07934726029634476,
|
|
"step": 1315,
|
|
"valid_targets_mean": 1288.2,
|
|
"valid_targets_min": 810
|
|
},
|
|
{
|
|
"epoch": 1.8384401114206128,
|
|
"grad_norm": 0.8996028117422565,
|
|
"learning_rate": 3.687269949510705e-05,
|
|
"loss": 0.2007,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09808281809091568,
|
|
"step": 1320,
|
|
"valid_targets_mean": 1328.9,
|
|
"valid_targets_min": 635
|
|
},
|
|
{
|
|
"epoch": 1.8454038997214486,
|
|
"grad_norm": 0.8176602096459092,
|
|
"learning_rate": 3.6835304540153914e-05,
|
|
"loss": 0.1931,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11697147786617279,
|
|
"step": 1325,
|
|
"valid_targets_mean": 1542.5,
|
|
"valid_targets_min": 673
|
|
},
|
|
{
|
|
"epoch": 1.8523676880222841,
|
|
"grad_norm": 1.1979420435140948,
|
|
"learning_rate": 3.67977065332721e-05,
|
|
"loss": 0.2077,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1085534319281578,
|
|
"step": 1330,
|
|
"valid_targets_mean": 1215.6,
|
|
"valid_targets_min": 641
|
|
},
|
|
{
|
|
"epoch": 1.8593314763231197,
|
|
"grad_norm": 0.8466439137236119,
|
|
"learning_rate": 3.6759905927934115e-05,
|
|
"loss": 0.1943,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08859957754611969,
|
|
"step": 1335,
|
|
"valid_targets_mean": 1222.6,
|
|
"valid_targets_min": 701
|
|
},
|
|
{
|
|
"epoch": 1.8662952646239555,
|
|
"grad_norm": 0.7520419595750739,
|
|
"learning_rate": 3.672190318005603e-05,
|
|
"loss": 0.211,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10533758252859116,
|
|
"step": 1340,
|
|
"valid_targets_mean": 1850.6,
|
|
"valid_targets_min": 1202
|
|
},
|
|
{
|
|
"epoch": 1.8732590529247912,
|
|
"grad_norm": 0.7954441568911064,
|
|
"learning_rate": 3.6683698747991963e-05,
|
|
"loss": 0.1974,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07728315144777298,
|
|
"step": 1345,
|
|
"valid_targets_mean": 1265.5,
|
|
"valid_targets_min": 657
|
|
},
|
|
{
|
|
"epoch": 1.8802228412256268,
|
|
"grad_norm": 0.8167699792397287,
|
|
"learning_rate": 3.6645293092528565e-05,
|
|
"loss": 0.2028,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09353001415729523,
|
|
"step": 1350,
|
|
"valid_targets_mean": 1388.0,
|
|
"valid_targets_min": 672
|
|
},
|
|
{
|
|
"epoch": 1.8871866295264623,
|
|
"grad_norm": 0.7414620593948343,
|
|
"learning_rate": 3.660668667687946e-05,
|
|
"loss": 0.1954,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.119552381336689,
|
|
"step": 1355,
|
|
"valid_targets_mean": 1813.6,
|
|
"valid_targets_min": 1019
|
|
},
|
|
{
|
|
"epoch": 1.894150417827298,
|
|
"grad_norm": 0.8151102750049719,
|
|
"learning_rate": 3.6567879966679667e-05,
|
|
"loss": 0.1945,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07849185168743134,
|
|
"step": 1360,
|
|
"valid_targets_mean": 1114.6,
|
|
"valid_targets_min": 732
|
|
},
|
|
{
|
|
"epoch": 1.9011142061281339,
|
|
"grad_norm": 0.8226820998476463,
|
|
"learning_rate": 3.6528873429979946e-05,
|
|
"loss": 0.2021,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07085934281349182,
|
|
"step": 1365,
|
|
"valid_targets_mean": 1194.5,
|
|
"valid_targets_min": 634
|
|
},
|
|
{
|
|
"epoch": 1.9080779944289694,
|
|
"grad_norm": 0.6954859825201096,
|
|
"learning_rate": 3.64896675372412e-05,
|
|
"loss": 0.1917,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10201038420200348,
|
|
"step": 1370,
|
|
"valid_targets_mean": 1756.5,
|
|
"valid_targets_min": 905
|
|
},
|
|
{
|
|
"epoch": 1.915041782729805,
|
|
"grad_norm": 0.7713340565719015,
|
|
"learning_rate": 3.645026276132877e-05,
|
|
"loss": 0.1925,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1128055602312088,
|
|
"step": 1375,
|
|
"valid_targets_mean": 1554.9,
|
|
"valid_targets_min": 596
|
|
},
|
|
{
|
|
"epoch": 1.9220055710306405,
|
|
"grad_norm": 0.7927271064032854,
|
|
"learning_rate": 3.6410659577506773e-05,
|
|
"loss": 0.1873,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07899089902639389,
|
|
"step": 1380,
|
|
"valid_targets_mean": 1214.6,
|
|
"valid_targets_min": 687
|
|
},
|
|
{
|
|
"epoch": 1.9289693593314763,
|
|
"grad_norm": 0.7718601693739504,
|
|
"learning_rate": 3.63708584634323e-05,
|
|
"loss": 0.2054,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11290781199932098,
|
|
"step": 1385,
|
|
"valid_targets_mean": 1955.1,
|
|
"valid_targets_min": 1199
|
|
},
|
|
{
|
|
"epoch": 1.935933147632312,
|
|
"grad_norm": 0.7193801549469071,
|
|
"learning_rate": 3.6330859899149694e-05,
|
|
"loss": 0.193,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10013289004564285,
|
|
"step": 1390,
|
|
"valid_targets_mean": 1758.6,
|
|
"valid_targets_min": 1214
|
|
},
|
|
{
|
|
"epoch": 1.9428969359331476,
|
|
"grad_norm": 0.726016091373521,
|
|
"learning_rate": 3.629066436708481e-05,
|
|
"loss": 0.1894,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09914419054985046,
|
|
"step": 1395,
|
|
"valid_targets_mean": 1711.5,
|
|
"valid_targets_min": 1122
|
|
},
|
|
{
|
|
"epoch": 1.9498607242339832,
|
|
"grad_norm": 0.8210382354675453,
|
|
"learning_rate": 3.625027235203909e-05,
|
|
"loss": 0.1888,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09899426996707916,
|
|
"step": 1400,
|
|
"valid_targets_mean": 1489.6,
|
|
"valid_targets_min": 850
|
|
},
|
|
{
|
|
"epoch": 1.956824512534819,
|
|
"grad_norm": 0.7429817246292151,
|
|
"learning_rate": 3.6209684341183803e-05,
|
|
"loss": 0.1786,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08339495956897736,
|
|
"step": 1405,
|
|
"valid_targets_mean": 1572.4,
|
|
"valid_targets_min": 1318
|
|
},
|
|
{
|
|
"epoch": 1.9637883008356547,
|
|
"grad_norm": 0.7739403851396445,
|
|
"learning_rate": 3.6168900824054135e-05,
|
|
"loss": 0.197,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1083778664469719,
|
|
"step": 1410,
|
|
"valid_targets_mean": 1500.0,
|
|
"valid_targets_min": 1043
|
|
},
|
|
{
|
|
"epoch": 1.9707520891364902,
|
|
"grad_norm": 0.7568148924154536,
|
|
"learning_rate": 3.612792229254327e-05,
|
|
"loss": 0.1966,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11948154121637344,
|
|
"step": 1415,
|
|
"valid_targets_mean": 1774.9,
|
|
"valid_targets_min": 920
|
|
},
|
|
{
|
|
"epoch": 1.9777158774373258,
|
|
"grad_norm": 0.6992261616231107,
|
|
"learning_rate": 3.608674924089651e-05,
|
|
"loss": 0.1958,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10174979269504547,
|
|
"step": 1420,
|
|
"valid_targets_mean": 1547.1,
|
|
"valid_targets_min": 847
|
|
},
|
|
{
|
|
"epoch": 1.9846796657381616,
|
|
"grad_norm": 0.7035138286444061,
|
|
"learning_rate": 3.6045382165705255e-05,
|
|
"loss": 0.1942,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10589592158794403,
|
|
"step": 1425,
|
|
"valid_targets_mean": 1769.4,
|
|
"valid_targets_min": 738
|
|
},
|
|
{
|
|
"epoch": 1.9916434540389973,
|
|
"grad_norm": 0.8155181230292006,
|
|
"learning_rate": 3.600382156590104e-05,
|
|
"loss": 0.2003,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10955852270126343,
|
|
"step": 1430,
|
|
"valid_targets_mean": 1540.1,
|
|
"valid_targets_min": 594
|
|
},
|
|
{
|
|
"epoch": 1.998607242339833,
|
|
"grad_norm": 0.6347666367578908,
|
|
"learning_rate": 3.5962067942749516e-05,
|
|
"loss": 0.1994,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10341192036867142,
|
|
"step": 1435,
|
|
"valid_targets_mean": 2228.6,
|
|
"valid_targets_min": 1393
|
|
},
|
|
{
|
|
"epoch": 2.0055710306406684,
|
|
"grad_norm": 0.5577761958239622,
|
|
"learning_rate": 3.592012179984442e-05,
|
|
"loss": 0.226,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18662209808826447,
|
|
"step": 1440,
|
|
"valid_targets_mean": 8029.9,
|
|
"valid_targets_min": 6142
|
|
},
|
|
{
|
|
"epoch": 2.012534818941504,
|
|
"grad_norm": 0.483618877635035,
|
|
"learning_rate": 3.587798364310146e-05,
|
|
"loss": 0.2322,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11201878637075424,
|
|
"step": 1445,
|
|
"valid_targets_mean": 6664.0,
|
|
"valid_targets_min": 5470
|
|
},
|
|
{
|
|
"epoch": 2.01949860724234,
|
|
"grad_norm": 0.5185691459202105,
|
|
"learning_rate": 3.583565398075226e-05,
|
|
"loss": 0.2276,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11615484952926636,
|
|
"step": 1450,
|
|
"valid_targets_mean": 7126.9,
|
|
"valid_targets_min": 5368
|
|
},
|
|
{
|
|
"epoch": 2.0264623955431755,
|
|
"grad_norm": 0.438228030552986,
|
|
"learning_rate": 3.579313332333819e-05,
|
|
"loss": 0.2123,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10276132822036743,
|
|
"step": 1455,
|
|
"valid_targets_mean": 7634.9,
|
|
"valid_targets_min": 5614
|
|
},
|
|
{
|
|
"epoch": 2.033426183844011,
|
|
"grad_norm": 0.3948344228513168,
|
|
"learning_rate": 3.575042218370423e-05,
|
|
"loss": 0.2174,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1019166111946106,
|
|
"step": 1460,
|
|
"valid_targets_mean": 8630.8,
|
|
"valid_targets_min": 6362
|
|
},
|
|
{
|
|
"epoch": 2.0403899721448466,
|
|
"grad_norm": 0.4535316383686147,
|
|
"learning_rate": 3.570752107699281e-05,
|
|
"loss": 0.2212,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10765604674816132,
|
|
"step": 1465,
|
|
"valid_targets_mean": 7383.6,
|
|
"valid_targets_min": 5844
|
|
},
|
|
{
|
|
"epoch": 2.0473537604456826,
|
|
"grad_norm": 0.4216201437635594,
|
|
"learning_rate": 3.566443052063751e-05,
|
|
"loss": 0.2323,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12464028596878052,
|
|
"step": 1470,
|
|
"valid_targets_mean": 8017.0,
|
|
"valid_targets_min": 5283
|
|
},
|
|
{
|
|
"epoch": 2.054317548746518,
|
|
"grad_norm": 0.3816491538208156,
|
|
"learning_rate": 3.5621151034356927e-05,
|
|
"loss": 0.2176,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09279941767454147,
|
|
"step": 1475,
|
|
"valid_targets_mean": 7051.1,
|
|
"valid_targets_min": 4135
|
|
},
|
|
{
|
|
"epoch": 2.0612813370473537,
|
|
"grad_norm": 0.41992932447893627,
|
|
"learning_rate": 3.557768314014834e-05,
|
|
"loss": 0.2121,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09113673865795135,
|
|
"step": 1480,
|
|
"valid_targets_mean": 5370.4,
|
|
"valid_targets_min": 3919
|
|
},
|
|
{
|
|
"epoch": 2.0682451253481893,
|
|
"grad_norm": 0.3912975995428693,
|
|
"learning_rate": 3.5534027362281436e-05,
|
|
"loss": 0.2023,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0919366329908371,
|
|
"step": 1485,
|
|
"valid_targets_mean": 6063.6,
|
|
"valid_targets_min": 4863
|
|
},
|
|
{
|
|
"epoch": 2.0752089136490253,
|
|
"grad_norm": 0.4107622241608645,
|
|
"learning_rate": 3.549018422729197e-05,
|
|
"loss": 0.2167,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10401890426874161,
|
|
"step": 1490,
|
|
"valid_targets_mean": 7001.2,
|
|
"valid_targets_min": 6020
|
|
},
|
|
{
|
|
"epoch": 2.082172701949861,
|
|
"grad_norm": 0.39061276611048656,
|
|
"learning_rate": 3.544615426397543e-05,
|
|
"loss": 0.2031,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.101812943816185,
|
|
"step": 1495,
|
|
"valid_targets_mean": 6181.9,
|
|
"valid_targets_min": 4669
|
|
},
|
|
{
|
|
"epoch": 2.0891364902506964,
|
|
"grad_norm": 0.38035206361099594,
|
|
"learning_rate": 3.5401938003380666e-05,
|
|
"loss": 0.2093,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09549552947282791,
|
|
"step": 1500,
|
|
"valid_targets_mean": 6250.2,
|
|
"valid_targets_min": 4900
|
|
},
|
|
{
|
|
"epoch": 2.096100278551532,
|
|
"grad_norm": 0.43005142535025387,
|
|
"learning_rate": 3.5357535978803455e-05,
|
|
"loss": 0.2014,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09210428595542908,
|
|
"step": 1505,
|
|
"valid_targets_mean": 7156.4,
|
|
"valid_targets_min": 5323
|
|
},
|
|
{
|
|
"epoch": 2.103064066852368,
|
|
"grad_norm": 0.40901081252647176,
|
|
"learning_rate": 3.53129487257801e-05,
|
|
"loss": 0.2052,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10530374944210052,
|
|
"step": 1510,
|
|
"valid_targets_mean": 7215.8,
|
|
"valid_targets_min": 4720
|
|
},
|
|
{
|
|
"epoch": 2.1100278551532035,
|
|
"grad_norm": 0.412691959227041,
|
|
"learning_rate": 3.5268176782080967e-05,
|
|
"loss": 0.2108,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10234518349170685,
|
|
"step": 1515,
|
|
"valid_targets_mean": 6550.8,
|
|
"valid_targets_min": 5739
|
|
},
|
|
{
|
|
"epoch": 2.116991643454039,
|
|
"grad_norm": 0.3904193639606634,
|
|
"learning_rate": 3.522322068770397e-05,
|
|
"loss": 0.2026,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10454997420310974,
|
|
"step": 1520,
|
|
"valid_targets_mean": 6726.8,
|
|
"valid_targets_min": 4540
|
|
},
|
|
{
|
|
"epoch": 2.1239554317548746,
|
|
"grad_norm": 0.5086857986092044,
|
|
"learning_rate": 3.51780809848681e-05,
|
|
"loss": 0.212,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11639285087585449,
|
|
"step": 1525,
|
|
"valid_targets_mean": 4573.4,
|
|
"valid_targets_min": 1039
|
|
},
|
|
{
|
|
"epoch": 2.13091922005571,
|
|
"grad_norm": 0.4916720484004185,
|
|
"learning_rate": 3.513275821800684e-05,
|
|
"loss": 0.2247,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10517747700214386,
|
|
"step": 1530,
|
|
"valid_targets_mean": 6430.0,
|
|
"valid_targets_min": 5412
|
|
},
|
|
{
|
|
"epoch": 2.137883008356546,
|
|
"grad_norm": 0.7279504673107768,
|
|
"learning_rate": 3.5087252933761654e-05,
|
|
"loss": 0.2099,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.03320852294564247,
|
|
"step": 1535,
|
|
"valid_targets_mean": 583.1,
|
|
"valid_targets_min": 158
|
|
},
|
|
{
|
|
"epoch": 2.1448467966573816,
|
|
"grad_norm": 0.38537163974446353,
|
|
"learning_rate": 3.504156568097533e-05,
|
|
"loss": 0.1918,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10623390972614288,
|
|
"step": 1540,
|
|
"valid_targets_mean": 8230.9,
|
|
"valid_targets_min": 5061
|
|
},
|
|
{
|
|
"epoch": 2.151810584958217,
|
|
"grad_norm": 0.41759228803004284,
|
|
"learning_rate": 3.4995697010685415e-05,
|
|
"loss": 0.205,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11333325505256653,
|
|
"step": 1545,
|
|
"valid_targets_mean": 7238.9,
|
|
"valid_targets_min": 5112
|
|
},
|
|
{
|
|
"epoch": 2.1587743732590527,
|
|
"grad_norm": 0.4142412953268547,
|
|
"learning_rate": 3.4949647476117535e-05,
|
|
"loss": 0.204,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09155681729316711,
|
|
"step": 1550,
|
|
"valid_targets_mean": 6322.8,
|
|
"valid_targets_min": 4867
|
|
},
|
|
{
|
|
"epoch": 2.1657381615598887,
|
|
"grad_norm": 0.38370921033703687,
|
|
"learning_rate": 3.4903417632678746e-05,
|
|
"loss": 0.212,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10269880294799805,
|
|
"step": 1555,
|
|
"valid_targets_mean": 7887.0,
|
|
"valid_targets_min": 6004
|
|
},
|
|
{
|
|
"epoch": 2.1727019498607243,
|
|
"grad_norm": 0.3711170733226334,
|
|
"learning_rate": 3.4857008037950814e-05,
|
|
"loss": 0.2024,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09217008948326111,
|
|
"step": 1560,
|
|
"valid_targets_mean": 7090.2,
|
|
"valid_targets_min": 4917
|
|
},
|
|
{
|
|
"epoch": 2.17966573816156,
|
|
"grad_norm": 0.3940923954443781,
|
|
"learning_rate": 3.481041925168349e-05,
|
|
"loss": 0.1942,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09533364325761795,
|
|
"step": 1565,
|
|
"valid_targets_mean": 7441.9,
|
|
"valid_targets_min": 4202
|
|
},
|
|
{
|
|
"epoch": 2.1866295264623954,
|
|
"grad_norm": 0.38589644774474163,
|
|
"learning_rate": 3.476365183578779e-05,
|
|
"loss": 0.1794,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09489143639802933,
|
|
"step": 1570,
|
|
"valid_targets_mean": 7901.8,
|
|
"valid_targets_min": 5060
|
|
},
|
|
{
|
|
"epoch": 2.1935933147632314,
|
|
"grad_norm": 0.32887624393579473,
|
|
"learning_rate": 3.471670635432919e-05,
|
|
"loss": 0.1815,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08562236279249191,
|
|
"step": 1575,
|
|
"valid_targets_mean": 8557.6,
|
|
"valid_targets_min": 6331
|
|
},
|
|
{
|
|
"epoch": 2.200557103064067,
|
|
"grad_norm": 0.45867578912244944,
|
|
"learning_rate": 3.466958337352082e-05,
|
|
"loss": 0.2123,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09923288971185684,
|
|
"step": 1580,
|
|
"valid_targets_mean": 4976.4,
|
|
"valid_targets_min": 2847
|
|
},
|
|
{
|
|
"epoch": 2.2075208913649025,
|
|
"grad_norm": 0.4238519335072428,
|
|
"learning_rate": 3.462228346171664e-05,
|
|
"loss": 0.2029,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11293122917413712,
|
|
"step": 1585,
|
|
"valid_targets_mean": 7129.1,
|
|
"valid_targets_min": 5331
|
|
},
|
|
{
|
|
"epoch": 2.214484679665738,
|
|
"grad_norm": 0.4164534419411376,
|
|
"learning_rate": 3.457480718940461e-05,
|
|
"loss": 0.2147,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1091599240899086,
|
|
"step": 1590,
|
|
"valid_targets_mean": 6864.0,
|
|
"valid_targets_min": 4697
|
|
},
|
|
{
|
|
"epoch": 2.2214484679665736,
|
|
"grad_norm": 0.42010359781278844,
|
|
"learning_rate": 3.4527155129199766e-05,
|
|
"loss": 0.2187,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10069143772125244,
|
|
"step": 1595,
|
|
"valid_targets_mean": 6631.9,
|
|
"valid_targets_min": 4623
|
|
},
|
|
{
|
|
"epoch": 2.2284122562674096,
|
|
"grad_norm": 0.43617878030396867,
|
|
"learning_rate": 3.447932785583735e-05,
|
|
"loss": 0.212,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09616230428218842,
|
|
"step": 1600,
|
|
"valid_targets_mean": 6330.4,
|
|
"valid_targets_min": 5424
|
|
},
|
|
{
|
|
"epoch": 2.235376044568245,
|
|
"grad_norm": 0.41414881589071445,
|
|
"learning_rate": 3.4431325946165847e-05,
|
|
"loss": 0.2036,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09712429344654083,
|
|
"step": 1605,
|
|
"valid_targets_mean": 6814.8,
|
|
"valid_targets_min": 4964
|
|
},
|
|
{
|
|
"epoch": 2.2423398328690807,
|
|
"grad_norm": 0.4444215191301066,
|
|
"learning_rate": 3.438314997914006e-05,
|
|
"loss": 0.2125,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11258673667907715,
|
|
"step": 1610,
|
|
"valid_targets_mean": 6915.8,
|
|
"valid_targets_min": 5321
|
|
},
|
|
{
|
|
"epoch": 2.2493036211699167,
|
|
"grad_norm": 0.41522878354353365,
|
|
"learning_rate": 3.433480053581412e-05,
|
|
"loss": 0.2035,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10380226373672485,
|
|
"step": 1615,
|
|
"valid_targets_mean": 6464.5,
|
|
"valid_targets_min": 5185
|
|
},
|
|
{
|
|
"epoch": 2.256267409470752,
|
|
"grad_norm": 0.4467050398625796,
|
|
"learning_rate": 3.4286278199334436e-05,
|
|
"loss": 0.2077,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10517797619104385,
|
|
"step": 1620,
|
|
"valid_targets_mean": 7048.6,
|
|
"valid_targets_min": 4913
|
|
},
|
|
{
|
|
"epoch": 2.2632311977715878,
|
|
"grad_norm": 0.4232921124371884,
|
|
"learning_rate": 3.4237583554932726e-05,
|
|
"loss": 0.2093,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11401194334030151,
|
|
"step": 1625,
|
|
"valid_targets_mean": 7469.2,
|
|
"valid_targets_min": 5426
|
|
},
|
|
{
|
|
"epoch": 2.2701949860724233,
|
|
"grad_norm": 0.4259278689375421,
|
|
"learning_rate": 3.4188717189918915e-05,
|
|
"loss": 0.2091,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10883718729019165,
|
|
"step": 1630,
|
|
"valid_targets_mean": 6819.4,
|
|
"valid_targets_min": 5286
|
|
},
|
|
{
|
|
"epoch": 2.277158774373259,
|
|
"grad_norm": 0.412251669431946,
|
|
"learning_rate": 3.413967969367406e-05,
|
|
"loss": 0.2085,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10999740660190582,
|
|
"step": 1635,
|
|
"valid_targets_mean": 7107.5,
|
|
"valid_targets_min": 5201
|
|
},
|
|
{
|
|
"epoch": 2.284122562674095,
|
|
"grad_norm": 0.42374140207835653,
|
|
"learning_rate": 3.4090471657643254e-05,
|
|
"loss": 0.2064,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09911487251520157,
|
|
"step": 1640,
|
|
"valid_targets_mean": 5642.6,
|
|
"valid_targets_min": 4838
|
|
},
|
|
{
|
|
"epoch": 2.2910863509749304,
|
|
"grad_norm": 0.9807420220774704,
|
|
"learning_rate": 3.404109367532847e-05,
|
|
"loss": 0.182,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19164110720157623,
|
|
"step": 1645,
|
|
"valid_targets_mean": 2054.1,
|
|
"valid_targets_min": 134
|
|
},
|
|
{
|
|
"epoch": 2.298050139275766,
|
|
"grad_norm": 0.5566049180562685,
|
|
"learning_rate": 3.399154634228144e-05,
|
|
"loss": 0.2081,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0960545465350151,
|
|
"step": 1650,
|
|
"valid_targets_mean": 6686.4,
|
|
"valid_targets_min": 5256
|
|
},
|
|
{
|
|
"epoch": 2.3050139275766015,
|
|
"grad_norm": 0.4253696878618078,
|
|
"learning_rate": 3.394183025609644e-05,
|
|
"loss": 0.1998,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1074969694018364,
|
|
"step": 1655,
|
|
"valid_targets_mean": 6216.6,
|
|
"valid_targets_min": 5542
|
|
},
|
|
{
|
|
"epoch": 2.3119777158774375,
|
|
"grad_norm": 0.4131104336589319,
|
|
"learning_rate": 3.3891946016403056e-05,
|
|
"loss": 0.2038,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10091003775596619,
|
|
"step": 1660,
|
|
"valid_targets_mean": 6486.1,
|
|
"valid_targets_min": 5531
|
|
},
|
|
{
|
|
"epoch": 2.318941504178273,
|
|
"grad_norm": 0.45874934809459283,
|
|
"learning_rate": 3.384189422485904e-05,
|
|
"loss": 0.1992,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09876036643981934,
|
|
"step": 1665,
|
|
"valid_targets_mean": 5797.4,
|
|
"valid_targets_min": 3911
|
|
},
|
|
{
|
|
"epoch": 2.3259052924791086,
|
|
"grad_norm": 0.4246615076495944,
|
|
"learning_rate": 3.3791675485142974e-05,
|
|
"loss": 0.1991,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10160474479198456,
|
|
"step": 1670,
|
|
"valid_targets_mean": 6357.5,
|
|
"valid_targets_min": 5195
|
|
},
|
|
{
|
|
"epoch": 2.332869080779944,
|
|
"grad_norm": 0.43609990408164573,
|
|
"learning_rate": 3.374129040294701e-05,
|
|
"loss": 0.2091,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10332215577363968,
|
|
"step": 1675,
|
|
"valid_targets_mean": 5808.9,
|
|
"valid_targets_min": 4959
|
|
},
|
|
{
|
|
"epoch": 2.33983286908078,
|
|
"grad_norm": 0.5400090427254614,
|
|
"learning_rate": 3.369073958596958e-05,
|
|
"loss": 0.2043,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06865261495113373,
|
|
"step": 1680,
|
|
"valid_targets_mean": 2591.5,
|
|
"valid_targets_min": 842
|
|
},
|
|
{
|
|
"epoch": 2.3467966573816157,
|
|
"grad_norm": 0.457933957702517,
|
|
"learning_rate": 3.364002364390805e-05,
|
|
"loss": 0.1793,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09762382507324219,
|
|
"step": 1685,
|
|
"valid_targets_mean": 6006.9,
|
|
"valid_targets_min": 4769
|
|
},
|
|
{
|
|
"epoch": 2.3537604456824512,
|
|
"grad_norm": 0.4088785055658562,
|
|
"learning_rate": 3.358914318845138e-05,
|
|
"loss": 0.1974,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09737879037857056,
|
|
"step": 1690,
|
|
"valid_targets_mean": 6764.2,
|
|
"valid_targets_min": 5046
|
|
},
|
|
{
|
|
"epoch": 2.360724233983287,
|
|
"grad_norm": 0.47400359842514794,
|
|
"learning_rate": 3.353809883327273e-05,
|
|
"loss": 0.2029,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10614047944545746,
|
|
"step": 1695,
|
|
"valid_targets_mean": 6055.8,
|
|
"valid_targets_min": 5182
|
|
},
|
|
{
|
|
"epoch": 2.3676880222841223,
|
|
"grad_norm": 0.44156250020466503,
|
|
"learning_rate": 3.348689119402209e-05,
|
|
"loss": 0.1989,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10033164173364639,
|
|
"step": 1700,
|
|
"valid_targets_mean": 7101.6,
|
|
"valid_targets_min": 5366
|
|
},
|
|
{
|
|
"epoch": 2.3746518105849583,
|
|
"grad_norm": 0.41213498850675423,
|
|
"learning_rate": 3.343552088831878e-05,
|
|
"loss": 0.195,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1027466282248497,
|
|
"step": 1705,
|
|
"valid_targets_mean": 7111.9,
|
|
"valid_targets_min": 5192
|
|
},
|
|
{
|
|
"epoch": 2.381615598885794,
|
|
"grad_norm": 0.5040432860111047,
|
|
"learning_rate": 3.338398853574411e-05,
|
|
"loss": 0.1985,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09895579516887665,
|
|
"step": 1710,
|
|
"valid_targets_mean": 5509.9,
|
|
"valid_targets_min": 4790
|
|
},
|
|
{
|
|
"epoch": 2.3885793871866294,
|
|
"grad_norm": 1.1363379165814331,
|
|
"learning_rate": 3.333229475783383e-05,
|
|
"loss": 0.2019,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11968301236629486,
|
|
"step": 1715,
|
|
"valid_targets_mean": 1581.9,
|
|
"valid_targets_min": 864
|
|
},
|
|
{
|
|
"epoch": 2.3955431754874654,
|
|
"grad_norm": 0.8698074566188687,
|
|
"learning_rate": 3.3280440178070635e-05,
|
|
"loss": 0.2104,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09715799242258072,
|
|
"step": 1720,
|
|
"valid_targets_mean": 1333.2,
|
|
"valid_targets_min": 632
|
|
},
|
|
{
|
|
"epoch": 2.402506963788301,
|
|
"grad_norm": 0.8444689746099655,
|
|
"learning_rate": 3.322842542187669e-05,
|
|
"loss": 0.194,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10113231092691422,
|
|
"step": 1725,
|
|
"valid_targets_mean": 1602.9,
|
|
"valid_targets_min": 875
|
|
},
|
|
{
|
|
"epoch": 2.4094707520891365,
|
|
"grad_norm": 1.1270431114328114,
|
|
"learning_rate": 3.317625111660605e-05,
|
|
"loss": 0.1921,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1020176112651825,
|
|
"step": 1730,
|
|
"valid_targets_mean": 1340.4,
|
|
"valid_targets_min": 701
|
|
},
|
|
{
|
|
"epoch": 2.416434540389972,
|
|
"grad_norm": 0.7839393431669954,
|
|
"learning_rate": 3.3123917891537085e-05,
|
|
"loss": 0.1823,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11173247545957565,
|
|
"step": 1735,
|
|
"valid_targets_mean": 1792.9,
|
|
"valid_targets_min": 898
|
|
},
|
|
{
|
|
"epoch": 2.4233983286908076,
|
|
"grad_norm": 0.7608101315047425,
|
|
"learning_rate": 3.307142637786496e-05,
|
|
"loss": 0.187,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09726438671350479,
|
|
"step": 1740,
|
|
"valid_targets_mean": 1433.1,
|
|
"valid_targets_min": 1070
|
|
},
|
|
{
|
|
"epoch": 2.4303621169916436,
|
|
"grad_norm": 0.851244310222369,
|
|
"learning_rate": 3.3018777208693916e-05,
|
|
"loss": 0.1976,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10597909986972809,
|
|
"step": 1745,
|
|
"valid_targets_mean": 1589.5,
|
|
"valid_targets_min": 970
|
|
},
|
|
{
|
|
"epoch": 2.437325905292479,
|
|
"grad_norm": 0.7562891748345674,
|
|
"learning_rate": 3.2965971019029716e-05,
|
|
"loss": 0.1741,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08589346706867218,
|
|
"step": 1750,
|
|
"valid_targets_mean": 1270.5,
|
|
"valid_targets_min": 648
|
|
},
|
|
{
|
|
"epoch": 2.4442896935933147,
|
|
"grad_norm": 0.8072271025740217,
|
|
"learning_rate": 3.291300844577196e-05,
|
|
"loss": 0.2016,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1143551915884018,
|
|
"step": 1755,
|
|
"valid_targets_mean": 1674.5,
|
|
"valid_targets_min": 727
|
|
},
|
|
{
|
|
"epoch": 2.4512534818941503,
|
|
"grad_norm": 0.8040766542030429,
|
|
"learning_rate": 3.28598901277064e-05,
|
|
"loss": 0.1776,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07331328094005585,
|
|
"step": 1760,
|
|
"valid_targets_mean": 1406.0,
|
|
"valid_targets_min": 770
|
|
},
|
|
{
|
|
"epoch": 2.4582172701949863,
|
|
"grad_norm": 0.7600004146401925,
|
|
"learning_rate": 3.280661670549723e-05,
|
|
"loss": 0.1849,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08032012730836868,
|
|
"step": 1765,
|
|
"valid_targets_mean": 1381.2,
|
|
"valid_targets_min": 572
|
|
},
|
|
{
|
|
"epoch": 2.465181058495822,
|
|
"grad_norm": 0.8186128164361567,
|
|
"learning_rate": 3.275318882167938e-05,
|
|
"loss": 0.1889,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11326928436756134,
|
|
"step": 1770,
|
|
"valid_targets_mean": 1580.4,
|
|
"valid_targets_min": 633
|
|
},
|
|
{
|
|
"epoch": 2.4721448467966574,
|
|
"grad_norm": 0.7878473077583293,
|
|
"learning_rate": 3.269960712065074e-05,
|
|
"loss": 0.1846,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09199025481939316,
|
|
"step": 1775,
|
|
"valid_targets_mean": 1478.8,
|
|
"valid_targets_min": 923
|
|
},
|
|
{
|
|
"epoch": 2.479108635097493,
|
|
"grad_norm": 0.7408103486539382,
|
|
"learning_rate": 3.264587224866442e-05,
|
|
"loss": 0.1749,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08334987610578537,
|
|
"step": 1780,
|
|
"valid_targets_mean": 1401.0,
|
|
"valid_targets_min": 740
|
|
},
|
|
{
|
|
"epoch": 2.486072423398329,
|
|
"grad_norm": 0.7785323575154897,
|
|
"learning_rate": 3.2591984853820926e-05,
|
|
"loss": 0.1769,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08059869706630707,
|
|
"step": 1785,
|
|
"valid_targets_mean": 1588.9,
|
|
"valid_targets_min": 870
|
|
},
|
|
{
|
|
"epoch": 2.4930362116991645,
|
|
"grad_norm": 0.7501213652690097,
|
|
"learning_rate": 3.2537945586060336e-05,
|
|
"loss": 0.1766,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12979084253311157,
|
|
"step": 1790,
|
|
"valid_targets_mean": 2008.4,
|
|
"valid_targets_min": 1042
|
|
},
|
|
{
|
|
"epoch": 2.5,
|
|
"grad_norm": 0.769300550573875,
|
|
"learning_rate": 3.248375509715452e-05,
|
|
"loss": 0.1832,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07505079358816147,
|
|
"step": 1795,
|
|
"valid_targets_mean": 1209.0,
|
|
"valid_targets_min": 870
|
|
},
|
|
{
|
|
"epoch": 2.5069637883008355,
|
|
"grad_norm": 0.8297708977009772,
|
|
"learning_rate": 3.242941404069921e-05,
|
|
"loss": 0.1878,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09335260093212128,
|
|
"step": 1800,
|
|
"valid_targets_mean": 1535.5,
|
|
"valid_targets_min": 740
|
|
},
|
|
{
|
|
"epoch": 2.513927576601671,
|
|
"grad_norm": 0.731844058605371,
|
|
"learning_rate": 3.237492307210614e-05,
|
|
"loss": 0.1882,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09308585524559021,
|
|
"step": 1805,
|
|
"valid_targets_mean": 1642.9,
|
|
"valid_targets_min": 906
|
|
},
|
|
{
|
|
"epoch": 2.520891364902507,
|
|
"grad_norm": 0.8193149089522122,
|
|
"learning_rate": 3.2320282848595174e-05,
|
|
"loss": 0.1826,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09537345170974731,
|
|
"step": 1810,
|
|
"valid_targets_mean": 1570.5,
|
|
"valid_targets_min": 814
|
|
},
|
|
{
|
|
"epoch": 2.5278551532033426,
|
|
"grad_norm": 0.8371651941131215,
|
|
"learning_rate": 3.226549402918631e-05,
|
|
"loss": 0.1826,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08008214086294174,
|
|
"step": 1815,
|
|
"valid_targets_mean": 1722.1,
|
|
"valid_targets_min": 832
|
|
},
|
|
{
|
|
"epoch": 2.534818941504178,
|
|
"grad_norm": 0.7018441413273226,
|
|
"learning_rate": 3.221055727469182e-05,
|
|
"loss": 0.1831,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06970568746328354,
|
|
"step": 1820,
|
|
"valid_targets_mean": 1453.5,
|
|
"valid_targets_min": 891
|
|
},
|
|
{
|
|
"epoch": 2.541782729805014,
|
|
"grad_norm": 0.7963146448093934,
|
|
"learning_rate": 3.21554732477082e-05,
|
|
"loss": 0.1832,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11181323230266571,
|
|
"step": 1825,
|
|
"valid_targets_mean": 1584.4,
|
|
"valid_targets_min": 734
|
|
},
|
|
{
|
|
"epoch": 2.5487465181058497,
|
|
"grad_norm": 0.7592311759031534,
|
|
"learning_rate": 3.2100242612608213e-05,
|
|
"loss": 0.1705,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08187158405780792,
|
|
"step": 1830,
|
|
"valid_targets_mean": 1495.2,
|
|
"valid_targets_min": 737
|
|
},
|
|
{
|
|
"epoch": 2.5557103064066853,
|
|
"grad_norm": 0.773359400074154,
|
|
"learning_rate": 3.204486603553289e-05,
|
|
"loss": 0.1799,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0989374965429306,
|
|
"step": 1835,
|
|
"valid_targets_mean": 1737.0,
|
|
"valid_targets_min": 872
|
|
},
|
|
{
|
|
"epoch": 2.562674094707521,
|
|
"grad_norm": 0.7743236506768809,
|
|
"learning_rate": 3.1989344184383476e-05,
|
|
"loss": 0.173,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09391649067401886,
|
|
"step": 1840,
|
|
"valid_targets_mean": 1431.6,
|
|
"valid_targets_min": 805
|
|
},
|
|
{
|
|
"epoch": 2.5696378830083564,
|
|
"grad_norm": 0.7358590006225955,
|
|
"learning_rate": 3.1933677728813364e-05,
|
|
"loss": 0.1816,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10584774613380432,
|
|
"step": 1845,
|
|
"valid_targets_mean": 2060.2,
|
|
"valid_targets_min": 1253
|
|
},
|
|
{
|
|
"epoch": 2.5766016713091924,
|
|
"grad_norm": 0.7193105312572781,
|
|
"learning_rate": 3.1877867340220045e-05,
|
|
"loss": 0.1825,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10305032879114151,
|
|
"step": 1850,
|
|
"valid_targets_mean": 2063.9,
|
|
"valid_targets_min": 1264
|
|
},
|
|
{
|
|
"epoch": 2.583565459610028,
|
|
"grad_norm": 0.7695266994780475,
|
|
"learning_rate": 3.1821913691736995e-05,
|
|
"loss": 0.1744,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0683041363954544,
|
|
"step": 1855,
|
|
"valid_targets_mean": 1036.5,
|
|
"valid_targets_min": 613
|
|
},
|
|
{
|
|
"epoch": 2.5905292479108635,
|
|
"grad_norm": 0.7101389867132969,
|
|
"learning_rate": 3.1765817458225546e-05,
|
|
"loss": 0.1747,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08477064967155457,
|
|
"step": 1860,
|
|
"valid_targets_mean": 1674.4,
|
|
"valid_targets_min": 768
|
|
},
|
|
{
|
|
"epoch": 2.597493036211699,
|
|
"grad_norm": 0.8599774854634431,
|
|
"learning_rate": 3.170957931626679e-05,
|
|
"loss": 0.1752,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.080399289727211,
|
|
"step": 1865,
|
|
"valid_targets_mean": 1249.8,
|
|
"valid_targets_min": 714
|
|
},
|
|
{
|
|
"epoch": 2.6044568245125346,
|
|
"grad_norm": 0.7993028212492834,
|
|
"learning_rate": 3.165319994415336e-05,
|
|
"loss": 0.1842,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06606724113225937,
|
|
"step": 1870,
|
|
"valid_targets_mean": 1114.8,
|
|
"valid_targets_min": 800
|
|
},
|
|
{
|
|
"epoch": 2.6114206128133706,
|
|
"grad_norm": 0.7781447333017487,
|
|
"learning_rate": 3.159668002188128e-05,
|
|
"loss": 0.175,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10036128759384155,
|
|
"step": 1875,
|
|
"valid_targets_mean": 1780.9,
|
|
"valid_targets_min": 933
|
|
},
|
|
{
|
|
"epoch": 2.618384401114206,
|
|
"grad_norm": 0.8858832068872401,
|
|
"learning_rate": 3.1540020231141774e-05,
|
|
"loss": 0.1717,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07642105966806412,
|
|
"step": 1880,
|
|
"valid_targets_mean": 1113.6,
|
|
"valid_targets_min": 746
|
|
},
|
|
{
|
|
"epoch": 2.6253481894150417,
|
|
"grad_norm": 0.8236948964912306,
|
|
"learning_rate": 3.148322125531302e-05,
|
|
"loss": 0.1637,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09135651588439941,
|
|
"step": 1885,
|
|
"valid_targets_mean": 1265.0,
|
|
"valid_targets_min": 814
|
|
},
|
|
{
|
|
"epoch": 2.6323119777158777,
|
|
"grad_norm": 0.7402380099739323,
|
|
"learning_rate": 3.142628377945191e-05,
|
|
"loss": 0.1751,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10815448313951492,
|
|
"step": 1890,
|
|
"valid_targets_mean": 1652.1,
|
|
"valid_targets_min": 635
|
|
},
|
|
{
|
|
"epoch": 2.639275766016713,
|
|
"grad_norm": 0.7766341389183821,
|
|
"learning_rate": 3.1369208490285824e-05,
|
|
"loss": 0.173,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06988343596458435,
|
|
"step": 1895,
|
|
"valid_targets_mean": 1231.2,
|
|
"valid_targets_min": 739
|
|
},
|
|
{
|
|
"epoch": 2.6462395543175488,
|
|
"grad_norm": 0.790490819348294,
|
|
"learning_rate": 3.13119960762043e-05,
|
|
"loss": 0.1739,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10846837610006332,
|
|
"step": 1900,
|
|
"valid_targets_mean": 1682.4,
|
|
"valid_targets_min": 831
|
|
},
|
|
{
|
|
"epoch": 2.6532033426183843,
|
|
"grad_norm": 0.8701861923326255,
|
|
"learning_rate": 3.125464722725074e-05,
|
|
"loss": 0.1763,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08927632868289948,
|
|
"step": 1905,
|
|
"valid_targets_mean": 1571.6,
|
|
"valid_targets_min": 1015
|
|
},
|
|
{
|
|
"epoch": 2.66016713091922,
|
|
"grad_norm": 0.7672559496303158,
|
|
"learning_rate": 3.119716263511412e-05,
|
|
"loss": 0.1731,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07645729184150696,
|
|
"step": 1910,
|
|
"valid_targets_mean": 1276.8,
|
|
"valid_targets_min": 711
|
|
},
|
|
{
|
|
"epoch": 2.667130919220056,
|
|
"grad_norm": 0.7633251851517129,
|
|
"learning_rate": 3.1139542993120615e-05,
|
|
"loss": 0.1689,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05591302365064621,
|
|
"step": 1915,
|
|
"valid_targets_mean": 1147.5,
|
|
"valid_targets_min": 733
|
|
},
|
|
{
|
|
"epoch": 2.6740947075208914,
|
|
"grad_norm": 0.8548710751284283,
|
|
"learning_rate": 3.108178899622524e-05,
|
|
"loss": 0.1789,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11267301440238953,
|
|
"step": 1920,
|
|
"valid_targets_mean": 1604.0,
|
|
"valid_targets_min": 953
|
|
},
|
|
{
|
|
"epoch": 2.681058495821727,
|
|
"grad_norm": 0.7154752716285275,
|
|
"learning_rate": 3.102390134100349e-05,
|
|
"loss": 0.183,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08892469853162766,
|
|
"step": 1925,
|
|
"valid_targets_mean": 1784.0,
|
|
"valid_targets_min": 1011
|
|
},
|
|
{
|
|
"epoch": 2.688022284122563,
|
|
"grad_norm": 0.7886303529310777,
|
|
"learning_rate": 3.096588072564291e-05,
|
|
"loss": 0.1856,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10861388593912125,
|
|
"step": 1930,
|
|
"valid_targets_mean": 1729.0,
|
|
"valid_targets_min": 1143
|
|
},
|
|
{
|
|
"epoch": 2.6949860724233985,
|
|
"grad_norm": 0.7782781896814367,
|
|
"learning_rate": 3.09077278499347e-05,
|
|
"loss": 0.1723,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0805855244398117,
|
|
"step": 1935,
|
|
"valid_targets_mean": 1483.9,
|
|
"valid_targets_min": 699
|
|
},
|
|
{
|
|
"epoch": 2.701949860724234,
|
|
"grad_norm": 0.7716708466878998,
|
|
"learning_rate": 3.084944341526524e-05,
|
|
"loss": 0.1655,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07467707991600037,
|
|
"step": 1940,
|
|
"valid_targets_mean": 1198.4,
|
|
"valid_targets_min": 680
|
|
},
|
|
{
|
|
"epoch": 2.7089136490250696,
|
|
"grad_norm": 0.7956657891036718,
|
|
"learning_rate": 3.079102812460769e-05,
|
|
"loss": 0.1809,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08717083930969238,
|
|
"step": 1945,
|
|
"valid_targets_mean": 1365.2,
|
|
"valid_targets_min": 687
|
|
},
|
|
{
|
|
"epoch": 2.715877437325905,
|
|
"grad_norm": 0.8040752534275205,
|
|
"learning_rate": 3.0732482682513435e-05,
|
|
"loss": 0.1742,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06229158490896225,
|
|
"step": 1950,
|
|
"valid_targets_mean": 1142.9,
|
|
"valid_targets_min": 835
|
|
},
|
|
{
|
|
"epoch": 2.722841225626741,
|
|
"grad_norm": 0.7747436174550746,
|
|
"learning_rate": 3.067380779510365e-05,
|
|
"loss": 0.1672,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07470732927322388,
|
|
"step": 1955,
|
|
"valid_targets_mean": 1316.9,
|
|
"valid_targets_min": 829
|
|
},
|
|
{
|
|
"epoch": 2.7298050139275767,
|
|
"grad_norm": 0.7825972178164834,
|
|
"learning_rate": 3.061500417006076e-05,
|
|
"loss": 0.1665,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09210141003131866,
|
|
"step": 1960,
|
|
"valid_targets_mean": 1739.2,
|
|
"valid_targets_min": 1138
|
|
},
|
|
{
|
|
"epoch": 2.7367688022284122,
|
|
"grad_norm": 0.696392311199642,
|
|
"learning_rate": 3.0556072516619896e-05,
|
|
"loss": 0.181,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07831001281738281,
|
|
"step": 1965,
|
|
"valid_targets_mean": 1592.2,
|
|
"valid_targets_min": 1174
|
|
},
|
|
{
|
|
"epoch": 2.743732590529248,
|
|
"grad_norm": 0.7770652602858239,
|
|
"learning_rate": 3.0497013545560354e-05,
|
|
"loss": 0.1763,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08863939344882965,
|
|
"step": 1970,
|
|
"valid_targets_mean": 1543.1,
|
|
"valid_targets_min": 554
|
|
},
|
|
{
|
|
"epoch": 2.7506963788300833,
|
|
"grad_norm": 0.7372103268972405,
|
|
"learning_rate": 3.043782796919701e-05,
|
|
"loss": 0.1711,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08466736972332001,
|
|
"step": 1975,
|
|
"valid_targets_mean": 1430.0,
|
|
"valid_targets_min": 843
|
|
},
|
|
{
|
|
"epoch": 2.7576601671309193,
|
|
"grad_norm": 0.7288721894678064,
|
|
"learning_rate": 3.0378516501371752e-05,
|
|
"loss": 0.1694,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08267127722501755,
|
|
"step": 1980,
|
|
"valid_targets_mean": 1622.9,
|
|
"valid_targets_min": 849
|
|
},
|
|
{
|
|
"epoch": 2.764623955431755,
|
|
"grad_norm": 0.819855661582502,
|
|
"learning_rate": 3.0319079857444844e-05,
|
|
"loss": 0.1776,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08106249570846558,
|
|
"step": 1985,
|
|
"valid_targets_mean": 1429.0,
|
|
"valid_targets_min": 612
|
|
},
|
|
{
|
|
"epoch": 2.7715877437325904,
|
|
"grad_norm": 0.7552847804482032,
|
|
"learning_rate": 3.0259518754286302e-05,
|
|
"loss": 0.171,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07017084956169128,
|
|
"step": 1990,
|
|
"valid_targets_mean": 1283.0,
|
|
"valid_targets_min": 782
|
|
},
|
|
{
|
|
"epoch": 2.7785515320334264,
|
|
"grad_norm": 0.8464332743776155,
|
|
"learning_rate": 3.0199833910267286e-05,
|
|
"loss": 0.1702,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09966123104095459,
|
|
"step": 1995,
|
|
"valid_targets_mean": 1457.0,
|
|
"valid_targets_min": 926
|
|
},
|
|
{
|
|
"epoch": 2.785515320334262,
|
|
"grad_norm": 0.7123507198714567,
|
|
"learning_rate": 3.0140026045251357e-05,
|
|
"loss": 0.1646,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0700862854719162,
|
|
"step": 2000,
|
|
"valid_targets_mean": 1399.6,
|
|
"valid_targets_min": 822
|
|
},
|
|
{
|
|
"epoch": 2.7924791086350975,
|
|
"grad_norm": 0.7070416875399016,
|
|
"learning_rate": 3.0080095880585884e-05,
|
|
"loss": 0.159,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07886691391468048,
|
|
"step": 2005,
|
|
"valid_targets_mean": 1464.6,
|
|
"valid_targets_min": 670
|
|
},
|
|
{
|
|
"epoch": 2.799442896935933,
|
|
"grad_norm": 0.7839739267460378,
|
|
"learning_rate": 3.002004413909326e-05,
|
|
"loss": 0.1743,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08037430047988892,
|
|
"step": 2010,
|
|
"valid_targets_mean": 1504.8,
|
|
"valid_targets_min": 647
|
|
},
|
|
{
|
|
"epoch": 2.8064066852367686,
|
|
"grad_norm": 0.7325386392599472,
|
|
"learning_rate": 2.995987154506228e-05,
|
|
"loss": 0.1702,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06354961544275284,
|
|
"step": 2015,
|
|
"valid_targets_mean": 1276.9,
|
|
"valid_targets_min": 791
|
|
},
|
|
{
|
|
"epoch": 2.8133704735376046,
|
|
"grad_norm": 0.7963527637303378,
|
|
"learning_rate": 2.98995788242393e-05,
|
|
"loss": 0.1682,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07502847909927368,
|
|
"step": 2020,
|
|
"valid_targets_mean": 1291.4,
|
|
"valid_targets_min": 746
|
|
},
|
|
{
|
|
"epoch": 2.82033426183844,
|
|
"grad_norm": 0.7325341255976993,
|
|
"learning_rate": 2.9839166703819572e-05,
|
|
"loss": 0.1698,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06790463626384735,
|
|
"step": 2025,
|
|
"valid_targets_mean": 1409.1,
|
|
"valid_targets_min": 892
|
|
},
|
|
{
|
|
"epoch": 2.8272980501392757,
|
|
"grad_norm": 1.0267772645703437,
|
|
"learning_rate": 2.97786359124384e-05,
|
|
"loss": 0.1806,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10208472609519958,
|
|
"step": 2030,
|
|
"valid_targets_mean": 1581.9,
|
|
"valid_targets_min": 914
|
|
},
|
|
{
|
|
"epoch": 2.8342618384401113,
|
|
"grad_norm": 0.7487266217807981,
|
|
"learning_rate": 2.971798718016242e-05,
|
|
"loss": 0.169,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07919836044311523,
|
|
"step": 2035,
|
|
"valid_targets_mean": 1647.6,
|
|
"valid_targets_min": 636
|
|
},
|
|
{
|
|
"epoch": 2.841225626740947,
|
|
"grad_norm": 0.9158258240874332,
|
|
"learning_rate": 2.9657221238480746e-05,
|
|
"loss": 0.1687,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0867113322019577,
|
|
"step": 2040,
|
|
"valid_targets_mean": 1523.5,
|
|
"valid_targets_min": 641
|
|
},
|
|
{
|
|
"epoch": 2.848189415041783,
|
|
"grad_norm": 0.8495953544791401,
|
|
"learning_rate": 2.9596338820296163e-05,
|
|
"loss": 0.1733,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09921315312385559,
|
|
"step": 2045,
|
|
"valid_targets_mean": 1629.4,
|
|
"valid_targets_min": 1020
|
|
},
|
|
{
|
|
"epoch": 2.8551532033426184,
|
|
"grad_norm": 1.1190071551544958,
|
|
"learning_rate": 2.953534065991631e-05,
|
|
"loss": 0.1744,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08030430972576141,
|
|
"step": 2050,
|
|
"valid_targets_mean": 1163.4,
|
|
"valid_targets_min": 729
|
|
},
|
|
{
|
|
"epoch": 2.862116991643454,
|
|
"grad_norm": 0.7607775978376603,
|
|
"learning_rate": 2.947422749304476e-05,
|
|
"loss": 0.177,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10393047332763672,
|
|
"step": 2055,
|
|
"valid_targets_mean": 1613.2,
|
|
"valid_targets_min": 1001
|
|
},
|
|
{
|
|
"epoch": 2.86908077994429,
|
|
"grad_norm": 0.783991662666305,
|
|
"learning_rate": 2.9413000056772225e-05,
|
|
"loss": 0.1773,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0888073593378067,
|
|
"step": 2060,
|
|
"valid_targets_mean": 1567.0,
|
|
"valid_targets_min": 1175
|
|
},
|
|
{
|
|
"epoch": 2.8760445682451254,
|
|
"grad_norm": 0.7826353186492123,
|
|
"learning_rate": 2.9351659089567622e-05,
|
|
"loss": 0.1756,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10893246531486511,
|
|
"step": 2065,
|
|
"valid_targets_mean": 1609.0,
|
|
"valid_targets_min": 892
|
|
},
|
|
{
|
|
"epoch": 2.883008356545961,
|
|
"grad_norm": 0.7934696501334912,
|
|
"learning_rate": 2.9290205331269158e-05,
|
|
"loss": 0.1759,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08275732398033142,
|
|
"step": 2070,
|
|
"valid_targets_mean": 1425.2,
|
|
"valid_targets_min": 801
|
|
},
|
|
{
|
|
"epoch": 2.8899721448467965,
|
|
"grad_norm": 0.7172842896059596,
|
|
"learning_rate": 2.9228639523075432e-05,
|
|
"loss": 0.1621,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08390860259532928,
|
|
"step": 2075,
|
|
"valid_targets_mean": 1462.4,
|
|
"valid_targets_min": 564
|
|
},
|
|
{
|
|
"epoch": 2.896935933147632,
|
|
"grad_norm": 0.8356533188243083,
|
|
"learning_rate": 2.9166962407536496e-05,
|
|
"loss": 0.1664,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08387964218854904,
|
|
"step": 2080,
|
|
"valid_targets_mean": 1749.1,
|
|
"valid_targets_min": 651
|
|
},
|
|
{
|
|
"epoch": 2.903899721448468,
|
|
"grad_norm": 0.780853509813084,
|
|
"learning_rate": 2.910517472854487e-05,
|
|
"loss": 0.1763,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07162684202194214,
|
|
"step": 2085,
|
|
"valid_targets_mean": 1161.0,
|
|
"valid_targets_min": 716
|
|
},
|
|
{
|
|
"epoch": 2.9108635097493036,
|
|
"grad_norm": 0.7403317269239412,
|
|
"learning_rate": 2.9043277231326606e-05,
|
|
"loss": 0.1673,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08211035281419754,
|
|
"step": 2090,
|
|
"valid_targets_mean": 1425.6,
|
|
"valid_targets_min": 744
|
|
},
|
|
{
|
|
"epoch": 2.917827298050139,
|
|
"grad_norm": 0.7512823712045082,
|
|
"learning_rate": 2.8981270662432258e-05,
|
|
"loss": 0.1617,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0736178457736969,
|
|
"step": 2095,
|
|
"valid_targets_mean": 1391.1,
|
|
"valid_targets_min": 1218
|
|
},
|
|
{
|
|
"epoch": 2.924791086350975,
|
|
"grad_norm": 0.8217643276597543,
|
|
"learning_rate": 2.8919155769727932e-05,
|
|
"loss": 0.1685,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09656500816345215,
|
|
"step": 2100,
|
|
"valid_targets_mean": 1724.5,
|
|
"valid_targets_min": 793
|
|
},
|
|
{
|
|
"epoch": 2.9317548746518107,
|
|
"grad_norm": 0.7855411813191986,
|
|
"learning_rate": 2.8856933302386224e-05,
|
|
"loss": 0.1823,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07395291328430176,
|
|
"step": 2105,
|
|
"valid_targets_mean": 1366.8,
|
|
"valid_targets_min": 846
|
|
},
|
|
{
|
|
"epoch": 2.9387186629526463,
|
|
"grad_norm": 0.9080574970183085,
|
|
"learning_rate": 2.879460401087718e-05,
|
|
"loss": 0.1652,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09661687910556793,
|
|
"step": 2110,
|
|
"valid_targets_mean": 1772.1,
|
|
"valid_targets_min": 979
|
|
},
|
|
{
|
|
"epoch": 2.945682451253482,
|
|
"grad_norm": 0.7185745540261005,
|
|
"learning_rate": 2.873216864695928e-05,
|
|
"loss": 0.1714,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09203220903873444,
|
|
"step": 2115,
|
|
"valid_targets_mean": 1778.9,
|
|
"valid_targets_min": 1093
|
|
},
|
|
{
|
|
"epoch": 2.9526462395543174,
|
|
"grad_norm": 0.8058469860251909,
|
|
"learning_rate": 2.8669627963670348e-05,
|
|
"loss": 0.1517,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0805043950676918,
|
|
"step": 2120,
|
|
"valid_targets_mean": 1510.8,
|
|
"valid_targets_min": 595
|
|
},
|
|
{
|
|
"epoch": 2.9596100278551534,
|
|
"grad_norm": 0.8161844468728507,
|
|
"learning_rate": 2.8606982715318467e-05,
|
|
"loss": 0.1689,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08108193427324295,
|
|
"step": 2125,
|
|
"valid_targets_mean": 1330.4,
|
|
"valid_targets_min": 868
|
|
},
|
|
{
|
|
"epoch": 2.966573816155989,
|
|
"grad_norm": 0.8048088395167377,
|
|
"learning_rate": 2.8544233657472893e-05,
|
|
"loss": 0.168,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10407143086194992,
|
|
"step": 2130,
|
|
"valid_targets_mean": 1504.2,
|
|
"valid_targets_min": 914
|
|
},
|
|
{
|
|
"epoch": 2.9735376044568245,
|
|
"grad_norm": 0.7696287813690902,
|
|
"learning_rate": 2.8481381546954944e-05,
|
|
"loss": 0.1713,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10349928587675095,
|
|
"step": 2135,
|
|
"valid_targets_mean": 1972.9,
|
|
"valid_targets_min": 1080
|
|
},
|
|
{
|
|
"epoch": 2.98050139275766,
|
|
"grad_norm": 0.7248553330646451,
|
|
"learning_rate": 2.8418427141828836e-05,
|
|
"loss": 0.1721,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08698803186416626,
|
|
"step": 2140,
|
|
"valid_targets_mean": 1519.6,
|
|
"valid_targets_min": 658
|
|
},
|
|
{
|
|
"epoch": 2.9874651810584956,
|
|
"grad_norm": 0.8081002064785113,
|
|
"learning_rate": 2.835537120139261e-05,
|
|
"loss": 0.1733,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07442046701908112,
|
|
"step": 2145,
|
|
"valid_targets_mean": 1362.0,
|
|
"valid_targets_min": 1018
|
|
},
|
|
{
|
|
"epoch": 2.9944289693593316,
|
|
"grad_norm": 0.7441884200422766,
|
|
"learning_rate": 2.8292214486168887e-05,
|
|
"loss": 0.1812,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1013420969247818,
|
|
"step": 2150,
|
|
"valid_targets_mean": 1623.0,
|
|
"valid_targets_min": 1041
|
|
},
|
|
{
|
|
"epoch": 3.001392757660167,
|
|
"grad_norm": 0.582670213892772,
|
|
"learning_rate": 2.8228957757895787e-05,
|
|
"loss": 0.1756,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10588406026363373,
|
|
"step": 2155,
|
|
"valid_targets_mean": 8215.2,
|
|
"valid_targets_min": 6163
|
|
},
|
|
{
|
|
"epoch": 3.0083565459610027,
|
|
"grad_norm": 0.45807589528981335,
|
|
"learning_rate": 2.816560177951765e-05,
|
|
"loss": 0.214,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10979750752449036,
|
|
"step": 2160,
|
|
"valid_targets_mean": 8594.4,
|
|
"valid_targets_min": 6006
|
|
},
|
|
{
|
|
"epoch": 3.0153203342618387,
|
|
"grad_norm": 0.5425898684830011,
|
|
"learning_rate": 2.810214731517592e-05,
|
|
"loss": 0.2073,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0921827107667923,
|
|
"step": 2165,
|
|
"valid_targets_mean": 3578.4,
|
|
"valid_targets_min": 289
|
|
},
|
|
{
|
|
"epoch": 3.022284122562674,
|
|
"grad_norm": 0.41010535238599494,
|
|
"learning_rate": 2.8038595130199846e-05,
|
|
"loss": 0.2047,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10162421315908432,
|
|
"step": 2170,
|
|
"valid_targets_mean": 7019.9,
|
|
"valid_targets_min": 4985
|
|
},
|
|
{
|
|
"epoch": 3.0292479108635098,
|
|
"grad_norm": 0.408252787870625,
|
|
"learning_rate": 2.7974945991097343e-05,
|
|
"loss": 0.1902,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09866312146186829,
|
|
"step": 2175,
|
|
"valid_targets_mean": 6956.6,
|
|
"valid_targets_min": 1457
|
|
},
|
|
{
|
|
"epoch": 3.0362116991643453,
|
|
"grad_norm": 0.376422224239574,
|
|
"learning_rate": 2.791120066554565e-05,
|
|
"loss": 0.1963,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09282627701759338,
|
|
"step": 2180,
|
|
"valid_targets_mean": 7361.6,
|
|
"valid_targets_min": 4961
|
|
},
|
|
{
|
|
"epoch": 3.043175487465181,
|
|
"grad_norm": 0.4174310411579024,
|
|
"learning_rate": 2.7847359922382154e-05,
|
|
"loss": 0.2118,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12435085326433182,
|
|
"step": 2185,
|
|
"valid_targets_mean": 7058.5,
|
|
"valid_targets_min": 5985
|
|
},
|
|
{
|
|
"epoch": 3.050139275766017,
|
|
"grad_norm": 0.40649864588422974,
|
|
"learning_rate": 2.7783424531595057e-05,
|
|
"loss": 0.2059,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11022967100143433,
|
|
"step": 2190,
|
|
"valid_targets_mean": 7664.2,
|
|
"valid_targets_min": 5169
|
|
},
|
|
{
|
|
"epoch": 3.0571030640668524,
|
|
"grad_norm": 0.3894067905652123,
|
|
"learning_rate": 2.7719395264314125e-05,
|
|
"loss": 0.1931,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09706021845340729,
|
|
"step": 2195,
|
|
"valid_targets_mean": 7111.8,
|
|
"valid_targets_min": 5348
|
|
},
|
|
{
|
|
"epoch": 3.064066852367688,
|
|
"grad_norm": 0.4571915199208844,
|
|
"learning_rate": 2.7655272892801376e-05,
|
|
"loss": 0.1844,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0859590619802475,
|
|
"step": 2200,
|
|
"valid_targets_mean": 4238.9,
|
|
"valid_targets_min": 152
|
|
},
|
|
{
|
|
"epoch": 3.0710306406685235,
|
|
"grad_norm": 0.442146850622646,
|
|
"learning_rate": 2.7591058190441754e-05,
|
|
"loss": 0.1868,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0988345593214035,
|
|
"step": 2205,
|
|
"valid_targets_mean": 7199.4,
|
|
"valid_targets_min": 4258
|
|
},
|
|
{
|
|
"epoch": 3.0779944289693595,
|
|
"grad_norm": 0.3888970877948061,
|
|
"learning_rate": 2.7526751931733824e-05,
|
|
"loss": 0.1931,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09356151521205902,
|
|
"step": 2210,
|
|
"valid_targets_mean": 7243.4,
|
|
"valid_targets_min": 5831
|
|
},
|
|
{
|
|
"epoch": 3.084958217270195,
|
|
"grad_norm": 0.3880326891417797,
|
|
"learning_rate": 2.7462354892280428e-05,
|
|
"loss": 0.1894,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09559017419815063,
|
|
"step": 2215,
|
|
"valid_targets_mean": 6950.6,
|
|
"valid_targets_min": 3862
|
|
},
|
|
{
|
|
"epoch": 3.0919220055710306,
|
|
"grad_norm": 0.38716280782428913,
|
|
"learning_rate": 2.7397867848779296e-05,
|
|
"loss": 0.1883,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10002546012401581,
|
|
"step": 2220,
|
|
"valid_targets_mean": 6991.9,
|
|
"valid_targets_min": 5696
|
|
},
|
|
{
|
|
"epoch": 3.098885793871866,
|
|
"grad_norm": 0.3892772405020075,
|
|
"learning_rate": 2.733329157901373e-05,
|
|
"loss": 0.1829,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1018202155828476,
|
|
"step": 2225,
|
|
"valid_targets_mean": 6963.5,
|
|
"valid_targets_min": 4947
|
|
},
|
|
{
|
|
"epoch": 3.105849582172702,
|
|
"grad_norm": 0.3954300382500997,
|
|
"learning_rate": 2.7268626861843176e-05,
|
|
"loss": 0.1871,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08901640772819519,
|
|
"step": 2230,
|
|
"valid_targets_mean": 6436.9,
|
|
"valid_targets_min": 4988
|
|
},
|
|
{
|
|
"epoch": 3.1128133704735377,
|
|
"grad_norm": 0.38253440455307475,
|
|
"learning_rate": 2.7203874477193867e-05,
|
|
"loss": 0.1905,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09357216954231262,
|
|
"step": 2235,
|
|
"valid_targets_mean": 7518.0,
|
|
"valid_targets_min": 5476
|
|
},
|
|
{
|
|
"epoch": 3.1197771587743732,
|
|
"grad_norm": 0.37760495023063456,
|
|
"learning_rate": 2.7139035206049392e-05,
|
|
"loss": 0.1848,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0883115828037262,
|
|
"step": 2240,
|
|
"valid_targets_mean": 7422.6,
|
|
"valid_targets_min": 5226
|
|
},
|
|
{
|
|
"epoch": 3.1267409470752088,
|
|
"grad_norm": 0.4492359216101069,
|
|
"learning_rate": 2.707410983044128e-05,
|
|
"loss": 0.1991,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12339232116937637,
|
|
"step": 2245,
|
|
"valid_targets_mean": 7080.0,
|
|
"valid_targets_min": 3753
|
|
},
|
|
{
|
|
"epoch": 3.1337047353760448,
|
|
"grad_norm": 0.478752949833473,
|
|
"learning_rate": 2.7009099133439588e-05,
|
|
"loss": 0.2132,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10847386717796326,
|
|
"step": 2250,
|
|
"valid_targets_mean": 6064.6,
|
|
"valid_targets_min": 5161
|
|
},
|
|
{
|
|
"epoch": 3.1406685236768803,
|
|
"grad_norm": 0.41876052780288764,
|
|
"learning_rate": 2.6944003899143424e-05,
|
|
"loss": 0.1624,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09066791832447052,
|
|
"step": 2255,
|
|
"valid_targets_mean": 7388.1,
|
|
"valid_targets_min": 5795
|
|
},
|
|
{
|
|
"epoch": 3.147632311977716,
|
|
"grad_norm": 0.4035596406534088,
|
|
"learning_rate": 2.6878824912671526e-05,
|
|
"loss": 0.1802,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0864686593413353,
|
|
"step": 2260,
|
|
"valid_targets_mean": 7696.0,
|
|
"valid_targets_min": 4658
|
|
},
|
|
{
|
|
"epoch": 3.1545961002785514,
|
|
"grad_norm": 0.3924938647790918,
|
|
"learning_rate": 2.6813562960152754e-05,
|
|
"loss": 0.1848,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10020910203456879,
|
|
"step": 2265,
|
|
"valid_targets_mean": 7799.1,
|
|
"valid_targets_min": 5508
|
|
},
|
|
{
|
|
"epoch": 3.1615598885793874,
|
|
"grad_norm": 0.39783779066844704,
|
|
"learning_rate": 2.6748218828716637e-05,
|
|
"loss": 0.1947,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0940442681312561,
|
|
"step": 2270,
|
|
"valid_targets_mean": 7341.6,
|
|
"valid_targets_min": 6454
|
|
},
|
|
{
|
|
"epoch": 3.168523676880223,
|
|
"grad_norm": 0.40685251066765943,
|
|
"learning_rate": 2.6682793306483875e-05,
|
|
"loss": 0.189,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08199045807123184,
|
|
"step": 2275,
|
|
"valid_targets_mean": 6818.0,
|
|
"valid_targets_min": 4392
|
|
},
|
|
{
|
|
"epoch": 3.1754874651810585,
|
|
"grad_norm": 0.3774943376487801,
|
|
"learning_rate": 2.6617287182556818e-05,
|
|
"loss": 0.1872,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08261384814977646,
|
|
"step": 2280,
|
|
"valid_targets_mean": 7857.6,
|
|
"valid_targets_min": 5394
|
|
},
|
|
{
|
|
"epoch": 3.182451253481894,
|
|
"grad_norm": 0.3962509851654937,
|
|
"learning_rate": 2.6551701247009974e-05,
|
|
"loss": 0.1647,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07782858610153198,
|
|
"step": 2285,
|
|
"valid_targets_mean": 8314.1,
|
|
"valid_targets_min": 4837
|
|
},
|
|
{
|
|
"epoch": 3.1894150417827296,
|
|
"grad_norm": 0.3805935986013425,
|
|
"learning_rate": 2.6486036290880447e-05,
|
|
"loss": 0.1662,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07052335143089294,
|
|
"step": 2290,
|
|
"valid_targets_mean": 7368.2,
|
|
"valid_targets_min": 6027
|
|
},
|
|
{
|
|
"epoch": 3.1963788300835656,
|
|
"grad_norm": 0.4494256622684818,
|
|
"learning_rate": 2.6420293106158425e-05,
|
|
"loss": 0.1742,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09394653886556625,
|
|
"step": 2295,
|
|
"valid_targets_mean": 6435.0,
|
|
"valid_targets_min": 4238
|
|
},
|
|
{
|
|
"epoch": 3.203342618384401,
|
|
"grad_norm": 1.3891357549119348,
|
|
"learning_rate": 2.6354472485777627e-05,
|
|
"loss": 0.1767,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05725764483213425,
|
|
"step": 2300,
|
|
"valid_targets_mean": 247.8,
|
|
"valid_targets_min": 137
|
|
},
|
|
{
|
|
"epoch": 3.2103064066852367,
|
|
"grad_norm": 0.44440730067879897,
|
|
"learning_rate": 2.6288575223605703e-05,
|
|
"loss": 0.1942,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09099040925502777,
|
|
"step": 2305,
|
|
"valid_targets_mean": 6807.0,
|
|
"valid_targets_min": 5165
|
|
},
|
|
{
|
|
"epoch": 3.2172701949860723,
|
|
"grad_norm": 0.47854206372685043,
|
|
"learning_rate": 2.622260211443471e-05,
|
|
"loss": 0.2009,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10358612984418869,
|
|
"step": 2310,
|
|
"valid_targets_mean": 6989.9,
|
|
"valid_targets_min": 5104
|
|
},
|
|
{
|
|
"epoch": 3.2242339832869082,
|
|
"grad_norm": 0.40218643285685773,
|
|
"learning_rate": 2.615655395397149e-05,
|
|
"loss": 0.1964,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10578115284442902,
|
|
"step": 2315,
|
|
"valid_targets_mean": 6692.2,
|
|
"valid_targets_min": 5068
|
|
},
|
|
{
|
|
"epoch": 3.231197771587744,
|
|
"grad_norm": 0.4823984099669414,
|
|
"learning_rate": 2.6090431538828095e-05,
|
|
"loss": 0.1926,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11674507707357407,
|
|
"step": 2320,
|
|
"valid_targets_mean": 7419.6,
|
|
"valid_targets_min": 5578
|
|
},
|
|
{
|
|
"epoch": 3.2381615598885793,
|
|
"grad_norm": 0.3808006367102266,
|
|
"learning_rate": 2.6024235666512154e-05,
|
|
"loss": 0.1855,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09444583207368851,
|
|
"step": 2325,
|
|
"valid_targets_mean": 7536.5,
|
|
"valid_targets_min": 5075
|
|
},
|
|
{
|
|
"epoch": 3.245125348189415,
|
|
"grad_norm": 0.3735886845835376,
|
|
"learning_rate": 2.5957967135417272e-05,
|
|
"loss": 0.1931,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08822908252477646,
|
|
"step": 2330,
|
|
"valid_targets_mean": 7221.5,
|
|
"valid_targets_min": 5841
|
|
},
|
|
{
|
|
"epoch": 3.252089136490251,
|
|
"grad_norm": 0.38530492214279993,
|
|
"learning_rate": 2.5891626744813413e-05,
|
|
"loss": 0.1871,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0928116887807846,
|
|
"step": 2335,
|
|
"valid_targets_mean": 7287.5,
|
|
"valid_targets_min": 5151
|
|
},
|
|
{
|
|
"epoch": 3.2590529247910864,
|
|
"grad_norm": 0.431169270507617,
|
|
"learning_rate": 2.5825215294837213e-05,
|
|
"loss": 0.1971,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09144345670938492,
|
|
"step": 2340,
|
|
"valid_targets_mean": 6738.4,
|
|
"valid_targets_min": 4042
|
|
},
|
|
{
|
|
"epoch": 3.266016713091922,
|
|
"grad_norm": 0.42057933866822494,
|
|
"learning_rate": 2.5758733586482402e-05,
|
|
"loss": 0.1888,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10509088635444641,
|
|
"step": 2345,
|
|
"valid_targets_mean": 6331.0,
|
|
"valid_targets_min": 5496
|
|
},
|
|
{
|
|
"epoch": 3.2729805013927575,
|
|
"grad_norm": 0.43887987652105004,
|
|
"learning_rate": 2.5692182421590062e-05,
|
|
"loss": 0.1905,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0931042730808258,
|
|
"step": 2350,
|
|
"valid_targets_mean": 6271.1,
|
|
"valid_targets_min": 4695
|
|
},
|
|
{
|
|
"epoch": 3.279944289693593,
|
|
"grad_norm": 0.6506631337130584,
|
|
"learning_rate": 2.5625562602839024e-05,
|
|
"loss": 0.1902,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09925953298807144,
|
|
"step": 2355,
|
|
"valid_targets_mean": 5790.8,
|
|
"valid_targets_min": 5021
|
|
},
|
|
{
|
|
"epoch": 3.286908077994429,
|
|
"grad_norm": 0.5048077739195413,
|
|
"learning_rate": 2.555887493373614e-05,
|
|
"loss": 0.1813,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07177004218101501,
|
|
"step": 2360,
|
|
"valid_targets_mean": 3386.8,
|
|
"valid_targets_min": 2279
|
|
},
|
|
{
|
|
"epoch": 3.2938718662952646,
|
|
"grad_norm": 0.6120098262207729,
|
|
"learning_rate": 2.549212021860662e-05,
|
|
"loss": 0.1669,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09277001023292542,
|
|
"step": 2365,
|
|
"valid_targets_mean": 6240.5,
|
|
"valid_targets_min": 5093
|
|
},
|
|
{
|
|
"epoch": 3.3008356545961,
|
|
"grad_norm": 0.7108808669473043,
|
|
"learning_rate": 2.542529926258433e-05,
|
|
"loss": 0.1864,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09597060084342957,
|
|
"step": 2370,
|
|
"valid_targets_mean": 6329.5,
|
|
"valid_targets_min": 4589
|
|
},
|
|
{
|
|
"epoch": 3.307799442896936,
|
|
"grad_norm": 0.3884309764014742,
|
|
"learning_rate": 2.5358412871602046e-05,
|
|
"loss": 0.1842,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0994221642613411,
|
|
"step": 2375,
|
|
"valid_targets_mean": 7589.2,
|
|
"valid_targets_min": 4963
|
|
},
|
|
{
|
|
"epoch": 3.3147632311977717,
|
|
"grad_norm": 0.43936327155895843,
|
|
"learning_rate": 2.5291461852381788e-05,
|
|
"loss": 0.185,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07934075593948364,
|
|
"step": 2380,
|
|
"valid_targets_mean": 5577.9,
|
|
"valid_targets_min": 4458
|
|
},
|
|
{
|
|
"epoch": 3.3217270194986073,
|
|
"grad_norm": 0.4459328442427666,
|
|
"learning_rate": 2.522444701242505e-05,
|
|
"loss": 0.184,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0892050564289093,
|
|
"step": 2385,
|
|
"valid_targets_mean": 5432.5,
|
|
"valid_targets_min": 4433
|
|
},
|
|
{
|
|
"epoch": 3.328690807799443,
|
|
"grad_norm": 0.5161665476844713,
|
|
"learning_rate": 2.5157369160003066e-05,
|
|
"loss": 0.1883,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09683055430650711,
|
|
"step": 2390,
|
|
"valid_targets_mean": 6361.8,
|
|
"valid_targets_min": 5201
|
|
},
|
|
{
|
|
"epoch": 3.3356545961002784,
|
|
"grad_norm": 0.45408075599291947,
|
|
"learning_rate": 2.509022910414708e-05,
|
|
"loss": 0.1913,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08833605796098709,
|
|
"step": 2395,
|
|
"valid_targets_mean": 5824.4,
|
|
"valid_targets_min": 5129
|
|
},
|
|
{
|
|
"epoch": 3.3426183844011144,
|
|
"grad_norm": 0.7702050101973215,
|
|
"learning_rate": 2.5023027654638576e-05,
|
|
"loss": 0.1539,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10458929091691971,
|
|
"step": 2400,
|
|
"valid_targets_mean": 1637.6,
|
|
"valid_targets_min": 137
|
|
},
|
|
{
|
|
"epoch": 3.34958217270195,
|
|
"grad_norm": 0.4785675018453149,
|
|
"learning_rate": 2.4955765621999505e-05,
|
|
"loss": 0.1838,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09321224689483643,
|
|
"step": 2405,
|
|
"valid_targets_mean": 6474.6,
|
|
"valid_targets_min": 4751
|
|
},
|
|
{
|
|
"epoch": 3.3565459610027855,
|
|
"grad_norm": 0.413066098207994,
|
|
"learning_rate": 2.4888443817482517e-05,
|
|
"loss": 0.1821,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10013020038604736,
|
|
"step": 2410,
|
|
"valid_targets_mean": 7186.2,
|
|
"valid_targets_min": 5358
|
|
},
|
|
{
|
|
"epoch": 3.363509749303621,
|
|
"grad_norm": 0.3906129731653948,
|
|
"learning_rate": 2.4821063053061184e-05,
|
|
"loss": 0.1889,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09480290114879608,
|
|
"step": 2415,
|
|
"valid_targets_mean": 7126.8,
|
|
"valid_targets_min": 4078
|
|
},
|
|
{
|
|
"epoch": 3.370473537604457,
|
|
"grad_norm": 0.45841696400076865,
|
|
"learning_rate": 2.4753624141420182e-05,
|
|
"loss": 0.1811,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09096138924360275,
|
|
"step": 2420,
|
|
"valid_targets_mean": 7022.1,
|
|
"valid_targets_min": 4940
|
|
},
|
|
{
|
|
"epoch": 3.3774373259052926,
|
|
"grad_norm": 0.4351904530245512,
|
|
"learning_rate": 2.4686127895945517e-05,
|
|
"loss": 0.1776,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09502632170915604,
|
|
"step": 2425,
|
|
"valid_targets_mean": 5900.6,
|
|
"valid_targets_min": 4779
|
|
},
|
|
{
|
|
"epoch": 3.384401114206128,
|
|
"grad_norm": 0.47258906281645247,
|
|
"learning_rate": 2.4618575130714707e-05,
|
|
"loss": 0.1856,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09541469812393188,
|
|
"step": 2430,
|
|
"valid_targets_mean": 6290.1,
|
|
"valid_targets_min": 4620
|
|
},
|
|
{
|
|
"epoch": 3.3913649025069637,
|
|
"grad_norm": 0.886229632768184,
|
|
"learning_rate": 2.4550966660486944e-05,
|
|
"loss": 0.1915,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07395320385694504,
|
|
"step": 2435,
|
|
"valid_targets_mean": 1162.2,
|
|
"valid_targets_min": 788
|
|
},
|
|
{
|
|
"epoch": 3.3983286908077996,
|
|
"grad_norm": 0.7835854833673302,
|
|
"learning_rate": 2.4483303300693296e-05,
|
|
"loss": 0.1761,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08432234823703766,
|
|
"step": 2440,
|
|
"valid_targets_mean": 1500.6,
|
|
"valid_targets_min": 957
|
|
},
|
|
{
|
|
"epoch": 3.405292479108635,
|
|
"grad_norm": 0.7364784582289423,
|
|
"learning_rate": 2.4415585867426854e-05,
|
|
"loss": 0.1651,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07511076331138611,
|
|
"step": 2445,
|
|
"valid_targets_mean": 1371.2,
|
|
"valid_targets_min": 796
|
|
},
|
|
{
|
|
"epoch": 3.4122562674094707,
|
|
"grad_norm": 0.9103744401835016,
|
|
"learning_rate": 2.434781517743289e-05,
|
|
"loss": 0.1689,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09593386203050613,
|
|
"step": 2450,
|
|
"valid_targets_mean": 1433.6,
|
|
"valid_targets_min": 697
|
|
},
|
|
{
|
|
"epoch": 3.4192200557103063,
|
|
"grad_norm": 0.7736223123121384,
|
|
"learning_rate": 2.4279992048099016e-05,
|
|
"loss": 0.1636,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07917076349258423,
|
|
"step": 2455,
|
|
"valid_targets_mean": 1578.9,
|
|
"valid_targets_min": 638
|
|
},
|
|
{
|
|
"epoch": 3.426183844011142,
|
|
"grad_norm": 0.7034601262769054,
|
|
"learning_rate": 2.4212117297445307e-05,
|
|
"loss": 0.1623,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07970526069402695,
|
|
"step": 2460,
|
|
"valid_targets_mean": 1746.4,
|
|
"valid_targets_min": 727
|
|
},
|
|
{
|
|
"epoch": 3.433147632311978,
|
|
"grad_norm": 0.7382178076531261,
|
|
"learning_rate": 2.4144191744114457e-05,
|
|
"loss": 0.1698,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06873296201229095,
|
|
"step": 2465,
|
|
"valid_targets_mean": 1309.9,
|
|
"valid_targets_min": 761
|
|
},
|
|
{
|
|
"epoch": 3.4401114206128134,
|
|
"grad_norm": 0.8425832615711547,
|
|
"learning_rate": 2.4076216207361904e-05,
|
|
"loss": 0.1529,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09206151962280273,
|
|
"step": 2470,
|
|
"valid_targets_mean": 1368.6,
|
|
"valid_targets_min": 714
|
|
},
|
|
{
|
|
"epoch": 3.447075208913649,
|
|
"grad_norm": 0.8395407250078996,
|
|
"learning_rate": 2.4008191507045913e-05,
|
|
"loss": 0.1737,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08047881722450256,
|
|
"step": 2475,
|
|
"valid_targets_mean": 1491.8,
|
|
"valid_targets_min": 867
|
|
},
|
|
{
|
|
"epoch": 3.4540389972144845,
|
|
"grad_norm": 0.7755881495928278,
|
|
"learning_rate": 2.3940118463617745e-05,
|
|
"loss": 0.1638,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09353938698768616,
|
|
"step": 2480,
|
|
"valid_targets_mean": 1994.0,
|
|
"valid_targets_min": 790
|
|
},
|
|
{
|
|
"epoch": 3.4610027855153205,
|
|
"grad_norm": 0.7814533436270437,
|
|
"learning_rate": 2.3871997898111707e-05,
|
|
"loss": 0.1559,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08098438382148743,
|
|
"step": 2485,
|
|
"valid_targets_mean": 1523.4,
|
|
"valid_targets_min": 896
|
|
},
|
|
{
|
|
"epoch": 3.467966573816156,
|
|
"grad_norm": 0.8288270044083298,
|
|
"learning_rate": 2.3803830632135284e-05,
|
|
"loss": 0.1641,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08329111337661743,
|
|
"step": 2490,
|
|
"valid_targets_mean": 1790.1,
|
|
"valid_targets_min": 1009
|
|
},
|
|
{
|
|
"epoch": 3.4749303621169916,
|
|
"grad_norm": 0.7061032285936508,
|
|
"learning_rate": 2.3735617487859224e-05,
|
|
"loss": 0.1568,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06442536413669586,
|
|
"step": 2495,
|
|
"valid_targets_mean": 1394.0,
|
|
"valid_targets_min": 717
|
|
},
|
|
{
|
|
"epoch": 3.481894150417827,
|
|
"grad_norm": 0.9020364718977384,
|
|
"learning_rate": 2.3667359288007612e-05,
|
|
"loss": 0.1481,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06079908832907677,
|
|
"step": 2500,
|
|
"valid_targets_mean": 1129.0,
|
|
"valid_targets_min": 659
|
|
},
|
|
{
|
|
"epoch": 3.488857938718663,
|
|
"grad_norm": 0.7566586323370277,
|
|
"learning_rate": 2.3599056855847937e-05,
|
|
"loss": 0.1501,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07075653970241547,
|
|
"step": 2505,
|
|
"valid_targets_mean": 1466.9,
|
|
"valid_targets_min": 642
|
|
},
|
|
{
|
|
"epoch": 3.4958217270194987,
|
|
"grad_norm": 0.7734326889825535,
|
|
"learning_rate": 2.353071101518121e-05,
|
|
"loss": 0.1537,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.054345741868019104,
|
|
"step": 2510,
|
|
"valid_targets_mean": 1129.4,
|
|
"valid_targets_min": 765
|
|
},
|
|
{
|
|
"epoch": 3.502785515320334,
|
|
"grad_norm": 0.804981371612974,
|
|
"learning_rate": 2.346232259033196e-05,
|
|
"loss": 0.1637,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07415035367012024,
|
|
"step": 2515,
|
|
"valid_targets_mean": 1322.8,
|
|
"valid_targets_min": 631
|
|
},
|
|
{
|
|
"epoch": 3.5097493036211698,
|
|
"grad_norm": 0.9012347000535628,
|
|
"learning_rate": 2.3393892406138354e-05,
|
|
"loss": 0.1646,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09443515539169312,
|
|
"step": 2520,
|
|
"valid_targets_mean": 1612.0,
|
|
"valid_targets_min": 623
|
|
},
|
|
{
|
|
"epoch": 3.5167130919220053,
|
|
"grad_norm": 0.8529475358392916,
|
|
"learning_rate": 2.332542128794221e-05,
|
|
"loss": 0.1568,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.060877248644828796,
|
|
"step": 2525,
|
|
"valid_targets_mean": 1402.1,
|
|
"valid_targets_min": 940
|
|
},
|
|
{
|
|
"epoch": 3.5236768802228413,
|
|
"grad_norm": 0.7346114867964298,
|
|
"learning_rate": 2.325691006157905e-05,
|
|
"loss": 0.1598,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06927518546581268,
|
|
"step": 2530,
|
|
"valid_targets_mean": 1337.8,
|
|
"valid_targets_min": 776
|
|
},
|
|
{
|
|
"epoch": 3.530640668523677,
|
|
"grad_norm": 0.7848390780850379,
|
|
"learning_rate": 2.3188359553368157e-05,
|
|
"loss": 0.1593,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08660085499286652,
|
|
"step": 2535,
|
|
"valid_targets_mean": 1525.5,
|
|
"valid_targets_min": 623
|
|
},
|
|
{
|
|
"epoch": 3.5376044568245124,
|
|
"grad_norm": 0.7918791101050114,
|
|
"learning_rate": 2.3119770590102585e-05,
|
|
"loss": 0.1569,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08223545551300049,
|
|
"step": 2540,
|
|
"valid_targets_mean": 1441.0,
|
|
"valid_targets_min": 693
|
|
},
|
|
{
|
|
"epoch": 3.5445682451253484,
|
|
"grad_norm": 0.9253216126700171,
|
|
"learning_rate": 2.3051143999039198e-05,
|
|
"loss": 0.1548,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.042962588369846344,
|
|
"step": 2545,
|
|
"valid_targets_mean": 847.9,
|
|
"valid_targets_min": 611
|
|
},
|
|
{
|
|
"epoch": 3.551532033426184,
|
|
"grad_norm": 0.8194995897407035,
|
|
"learning_rate": 2.2982480607888693e-05,
|
|
"loss": 0.1492,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07983898371458054,
|
|
"step": 2550,
|
|
"valid_targets_mean": 1524.9,
|
|
"valid_targets_min": 793
|
|
},
|
|
{
|
|
"epoch": 3.5584958217270195,
|
|
"grad_norm": 0.7123126085956808,
|
|
"learning_rate": 2.291378124480563e-05,
|
|
"loss": 0.1519,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08361306041479111,
|
|
"step": 2555,
|
|
"valid_targets_mean": 1779.4,
|
|
"valid_targets_min": 563
|
|
},
|
|
{
|
|
"epoch": 3.565459610027855,
|
|
"grad_norm": 0.7375463773516223,
|
|
"learning_rate": 2.284504673837839e-05,
|
|
"loss": 0.1529,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08410098403692245,
|
|
"step": 2560,
|
|
"valid_targets_mean": 1741.2,
|
|
"valid_targets_min": 1341
|
|
},
|
|
{
|
|
"epoch": 3.5724233983286906,
|
|
"grad_norm": 0.813717402129795,
|
|
"learning_rate": 2.277627791761927e-05,
|
|
"loss": 0.157,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0976366251707077,
|
|
"step": 2565,
|
|
"valid_targets_mean": 1875.6,
|
|
"valid_targets_min": 1107
|
|
},
|
|
{
|
|
"epoch": 3.5793871866295266,
|
|
"grad_norm": 0.7638558548512067,
|
|
"learning_rate": 2.270747561195441e-05,
|
|
"loss": 0.1608,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07316169142723083,
|
|
"step": 2570,
|
|
"valid_targets_mean": 1419.5,
|
|
"valid_targets_min": 680
|
|
},
|
|
{
|
|
"epoch": 3.586350974930362,
|
|
"grad_norm": 0.7546648374433426,
|
|
"learning_rate": 2.2638640651213804e-05,
|
|
"loss": 0.1496,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10009348392486572,
|
|
"step": 2575,
|
|
"valid_targets_mean": 1968.2,
|
|
"valid_targets_min": 981
|
|
},
|
|
{
|
|
"epoch": 3.5933147632311977,
|
|
"grad_norm": 0.8517766302015852,
|
|
"learning_rate": 2.2569773865621324e-05,
|
|
"loss": 0.1528,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07636626064777374,
|
|
"step": 2580,
|
|
"valid_targets_mean": 1568.9,
|
|
"valid_targets_min": 937
|
|
},
|
|
{
|
|
"epoch": 3.6002785515320337,
|
|
"grad_norm": 0.8067654097110404,
|
|
"learning_rate": 2.2500876085784665e-05,
|
|
"loss": 0.1577,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08284150809049606,
|
|
"step": 2585,
|
|
"valid_targets_mean": 1645.2,
|
|
"valid_targets_min": 804
|
|
},
|
|
{
|
|
"epoch": 3.6072423398328692,
|
|
"grad_norm": 0.7691207000641539,
|
|
"learning_rate": 2.2431948142685352e-05,
|
|
"loss": 0.1583,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07658159732818604,
|
|
"step": 2590,
|
|
"valid_targets_mean": 1574.9,
|
|
"valid_targets_min": 1019
|
|
},
|
|
{
|
|
"epoch": 3.614206128133705,
|
|
"grad_norm": 0.7783596723806455,
|
|
"learning_rate": 2.236299086766873e-05,
|
|
"loss": 0.1492,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06561608612537384,
|
|
"step": 2595,
|
|
"valid_targets_mean": 1121.6,
|
|
"valid_targets_min": 882
|
|
},
|
|
{
|
|
"epoch": 3.6211699164345403,
|
|
"grad_norm": 0.7775958409848589,
|
|
"learning_rate": 2.2294005092433887e-05,
|
|
"loss": 0.1461,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08176523447036743,
|
|
"step": 2600,
|
|
"valid_targets_mean": 1818.1,
|
|
"valid_targets_min": 1168
|
|
},
|
|
{
|
|
"epoch": 3.628133704735376,
|
|
"grad_norm": 0.828811804727648,
|
|
"learning_rate": 2.2224991649023675e-05,
|
|
"loss": 0.141,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07202639430761337,
|
|
"step": 2605,
|
|
"valid_targets_mean": 1417.8,
|
|
"valid_targets_min": 678
|
|
},
|
|
{
|
|
"epoch": 3.635097493036212,
|
|
"grad_norm": 0.8282689257518919,
|
|
"learning_rate": 2.2155951369814644e-05,
|
|
"loss": 0.1523,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.058761902153491974,
|
|
"step": 2610,
|
|
"valid_targets_mean": 1051.0,
|
|
"valid_targets_min": 676
|
|
},
|
|
{
|
|
"epoch": 3.6420612813370474,
|
|
"grad_norm": 1.7079878973692406,
|
|
"learning_rate": 2.2086885087507027e-05,
|
|
"loss": 0.1484,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06798002123832703,
|
|
"step": 2615,
|
|
"valid_targets_mean": 1431.1,
|
|
"valid_targets_min": 794
|
|
},
|
|
{
|
|
"epoch": 3.649025069637883,
|
|
"grad_norm": 0.7717379455457729,
|
|
"learning_rate": 2.2017793635114667e-05,
|
|
"loss": 0.1514,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07599664479494095,
|
|
"step": 2620,
|
|
"valid_targets_mean": 1486.9,
|
|
"valid_targets_min": 742
|
|
},
|
|
{
|
|
"epoch": 3.6559888579387185,
|
|
"grad_norm": 0.7698310794801155,
|
|
"learning_rate": 2.1948677845954985e-05,
|
|
"loss": 0.1498,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08982954919338226,
|
|
"step": 2625,
|
|
"valid_targets_mean": 1734.5,
|
|
"valid_targets_min": 741
|
|
},
|
|
{
|
|
"epoch": 3.662952646239554,
|
|
"grad_norm": 0.8944739797043322,
|
|
"learning_rate": 2.1879538553638942e-05,
|
|
"loss": 0.1557,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09024101495742798,
|
|
"step": 2630,
|
|
"valid_targets_mean": 1712.5,
|
|
"valid_targets_min": 1157
|
|
},
|
|
{
|
|
"epoch": 3.66991643454039,
|
|
"grad_norm": 0.7894287756846752,
|
|
"learning_rate": 2.1810376592060968e-05,
|
|
"loss": 0.1405,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09558722376823425,
|
|
"step": 2635,
|
|
"valid_targets_mean": 1723.8,
|
|
"valid_targets_min": 862
|
|
},
|
|
{
|
|
"epoch": 3.6768802228412256,
|
|
"grad_norm": 0.8796292419764679,
|
|
"learning_rate": 2.174119279538891e-05,
|
|
"loss": 0.1612,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09259238839149475,
|
|
"step": 2640,
|
|
"valid_targets_mean": 1633.1,
|
|
"valid_targets_min": 1019
|
|
},
|
|
{
|
|
"epoch": 3.683844011142061,
|
|
"grad_norm": 0.7361323056285416,
|
|
"learning_rate": 2.1671987998053954e-05,
|
|
"loss": 0.152,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0679299384355545,
|
|
"step": 2645,
|
|
"valid_targets_mean": 1376.1,
|
|
"valid_targets_min": 1058
|
|
},
|
|
{
|
|
"epoch": 3.690807799442897,
|
|
"grad_norm": 0.7916992537280872,
|
|
"learning_rate": 2.16027630347406e-05,
|
|
"loss": 0.1623,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08028727769851685,
|
|
"step": 2650,
|
|
"valid_targets_mean": 1576.5,
|
|
"valid_targets_min": 618
|
|
},
|
|
{
|
|
"epoch": 3.6977715877437327,
|
|
"grad_norm": 0.8165492508178849,
|
|
"learning_rate": 2.1533518740376557e-05,
|
|
"loss": 0.1451,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0871560126543045,
|
|
"step": 2655,
|
|
"valid_targets_mean": 1615.8,
|
|
"valid_targets_min": 629
|
|
},
|
|
{
|
|
"epoch": 3.7047353760445683,
|
|
"grad_norm": 0.8981802581015453,
|
|
"learning_rate": 2.146425595012269e-05,
|
|
"loss": 0.1494,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08319033682346344,
|
|
"step": 2660,
|
|
"valid_targets_mean": 1881.2,
|
|
"valid_targets_min": 1167
|
|
},
|
|
{
|
|
"epoch": 3.711699164345404,
|
|
"grad_norm": 0.7955838043549589,
|
|
"learning_rate": 2.1394975499362947e-05,
|
|
"loss": 0.1521,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07869289815425873,
|
|
"step": 2665,
|
|
"valid_targets_mean": 1367.2,
|
|
"valid_targets_min": 699
|
|
},
|
|
{
|
|
"epoch": 3.7186629526462394,
|
|
"grad_norm": 0.7732510809048744,
|
|
"learning_rate": 2.132567822369428e-05,
|
|
"loss": 0.1429,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.058514297008514404,
|
|
"step": 2670,
|
|
"valid_targets_mean": 1174.9,
|
|
"valid_targets_min": 679
|
|
},
|
|
{
|
|
"epoch": 3.7256267409470754,
|
|
"grad_norm": 0.7978233555615979,
|
|
"learning_rate": 2.1256364958916564e-05,
|
|
"loss": 0.1469,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07325758785009384,
|
|
"step": 2675,
|
|
"valid_targets_mean": 1556.2,
|
|
"valid_targets_min": 504
|
|
},
|
|
{
|
|
"epoch": 3.732590529247911,
|
|
"grad_norm": 0.8402148455143513,
|
|
"learning_rate": 2.1187036541022533e-05,
|
|
"loss": 0.1496,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08283115178346634,
|
|
"step": 2680,
|
|
"valid_targets_mean": 1885.0,
|
|
"valid_targets_min": 1331
|
|
},
|
|
{
|
|
"epoch": 3.7395543175487465,
|
|
"grad_norm": 0.764780642210825,
|
|
"learning_rate": 2.1117693806187657e-05,
|
|
"loss": 0.1545,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.055060356855392456,
|
|
"step": 2685,
|
|
"valid_targets_mean": 1306.1,
|
|
"valid_targets_min": 661
|
|
},
|
|
{
|
|
"epoch": 3.7465181058495824,
|
|
"grad_norm": 0.7707860162726192,
|
|
"learning_rate": 2.104833759076011e-05,
|
|
"loss": 0.1562,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05953007936477661,
|
|
"step": 2690,
|
|
"valid_targets_mean": 1330.1,
|
|
"valid_targets_min": 958
|
|
},
|
|
{
|
|
"epoch": 3.7534818941504176,
|
|
"grad_norm": 0.7132388417704622,
|
|
"learning_rate": 2.0978968731250648e-05,
|
|
"loss": 0.143,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.057646650820970535,
|
|
"step": 2695,
|
|
"valid_targets_mean": 1465.0,
|
|
"valid_targets_min": 963
|
|
},
|
|
{
|
|
"epoch": 3.7604456824512535,
|
|
"grad_norm": 0.7777931157465899,
|
|
"learning_rate": 2.0909588064322524e-05,
|
|
"loss": 0.1523,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07116229832172394,
|
|
"step": 2700,
|
|
"valid_targets_mean": 1428.0,
|
|
"valid_targets_min": 793
|
|
},
|
|
{
|
|
"epoch": 3.767409470752089,
|
|
"grad_norm": 0.7925330730772479,
|
|
"learning_rate": 2.08401964267814e-05,
|
|
"loss": 0.1481,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.091750867664814,
|
|
"step": 2705,
|
|
"valid_targets_mean": 1633.1,
|
|
"valid_targets_min": 917
|
|
},
|
|
{
|
|
"epoch": 3.7743732590529246,
|
|
"grad_norm": 0.8377874283930322,
|
|
"learning_rate": 2.0770794655565278e-05,
|
|
"loss": 0.1489,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06515245139598846,
|
|
"step": 2710,
|
|
"valid_targets_mean": 1246.5,
|
|
"valid_targets_min": 872
|
|
},
|
|
{
|
|
"epoch": 3.7813370473537606,
|
|
"grad_norm": 0.7500409300068362,
|
|
"learning_rate": 2.0701383587734334e-05,
|
|
"loss": 0.1496,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07608701288700104,
|
|
"step": 2715,
|
|
"valid_targets_mean": 1903.6,
|
|
"valid_targets_min": 1220
|
|
},
|
|
{
|
|
"epoch": 3.788300835654596,
|
|
"grad_norm": 0.8435711995319485,
|
|
"learning_rate": 2.0631964060460934e-05,
|
|
"loss": 0.1357,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06794526427984238,
|
|
"step": 2720,
|
|
"valid_targets_mean": 1391.6,
|
|
"valid_targets_min": 712
|
|
},
|
|
{
|
|
"epoch": 3.7952646239554317,
|
|
"grad_norm": 0.8724016823017468,
|
|
"learning_rate": 2.0562536911019415e-05,
|
|
"loss": 0.1453,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0759831964969635,
|
|
"step": 2725,
|
|
"valid_targets_mean": 1295.0,
|
|
"valid_targets_min": 605
|
|
},
|
|
{
|
|
"epoch": 3.8022284122562673,
|
|
"grad_norm": 0.7990588269247201,
|
|
"learning_rate": 2.049310297677609e-05,
|
|
"loss": 0.1533,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05404563248157501,
|
|
"step": 2730,
|
|
"valid_targets_mean": 1414.8,
|
|
"valid_targets_min": 746
|
|
},
|
|
{
|
|
"epoch": 3.809192200557103,
|
|
"grad_norm": 0.9100646789026491,
|
|
"learning_rate": 2.0423663095179084e-05,
|
|
"loss": 0.1403,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06111488863825798,
|
|
"step": 2735,
|
|
"valid_targets_mean": 1059.4,
|
|
"valid_targets_min": 723
|
|
},
|
|
{
|
|
"epoch": 3.816155988857939,
|
|
"grad_norm": 0.7835072806883097,
|
|
"learning_rate": 2.0354218103748253e-05,
|
|
"loss": 0.1491,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08211217820644379,
|
|
"step": 2740,
|
|
"valid_targets_mean": 1634.2,
|
|
"valid_targets_min": 1327
|
|
},
|
|
{
|
|
"epoch": 3.8231197771587744,
|
|
"grad_norm": 0.8355043521576563,
|
|
"learning_rate": 2.028476884006508e-05,
|
|
"loss": 0.1507,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07008926570415497,
|
|
"step": 2745,
|
|
"valid_targets_mean": 1412.5,
|
|
"valid_targets_min": 902
|
|
},
|
|
{
|
|
"epoch": 3.83008356545961,
|
|
"grad_norm": 0.8578826264946082,
|
|
"learning_rate": 2.02153161417626e-05,
|
|
"loss": 0.153,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07012935727834702,
|
|
"step": 2750,
|
|
"valid_targets_mean": 1558.8,
|
|
"valid_targets_min": 803
|
|
},
|
|
{
|
|
"epoch": 3.837047353760446,
|
|
"grad_norm": 0.8206701722875304,
|
|
"learning_rate": 2.0145860846515234e-05,
|
|
"loss": 0.1512,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07594504952430725,
|
|
"step": 2755,
|
|
"valid_targets_mean": 1441.2,
|
|
"valid_targets_min": 1079
|
|
},
|
|
{
|
|
"epoch": 3.8440111420612815,
|
|
"grad_norm": 0.7481367084277267,
|
|
"learning_rate": 2.0076403792028757e-05,
|
|
"loss": 0.1447,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06753340363502502,
|
|
"step": 2760,
|
|
"valid_targets_mean": 1449.8,
|
|
"valid_targets_min": 1263
|
|
},
|
|
{
|
|
"epoch": 3.850974930362117,
|
|
"grad_norm": 0.7968317810485137,
|
|
"learning_rate": 2.000694581603016e-05,
|
|
"loss": 0.1524,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07419832050800323,
|
|
"step": 2765,
|
|
"valid_targets_mean": 1487.4,
|
|
"valid_targets_min": 844
|
|
},
|
|
{
|
|
"epoch": 3.8579387186629526,
|
|
"grad_norm": 0.8690761427120194,
|
|
"learning_rate": 1.993748775625752e-05,
|
|
"loss": 0.1499,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06871486455202103,
|
|
"step": 2770,
|
|
"valid_targets_mean": 1421.4,
|
|
"valid_targets_min": 667
|
|
},
|
|
{
|
|
"epoch": 3.864902506963788,
|
|
"grad_norm": 0.8975013022838169,
|
|
"learning_rate": 1.986803045044996e-05,
|
|
"loss": 0.1543,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08886584639549255,
|
|
"step": 2775,
|
|
"valid_targets_mean": 1822.6,
|
|
"valid_targets_min": 1041
|
|
},
|
|
{
|
|
"epoch": 3.871866295264624,
|
|
"grad_norm": 0.7977117949143756,
|
|
"learning_rate": 1.9798574736337487e-05,
|
|
"loss": 0.1519,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06609257310628891,
|
|
"step": 2780,
|
|
"valid_targets_mean": 1327.4,
|
|
"valid_targets_min": 777
|
|
},
|
|
{
|
|
"epoch": 3.8788300835654597,
|
|
"grad_norm": 0.7438548739680065,
|
|
"learning_rate": 1.972912145163093e-05,
|
|
"loss": 0.1504,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07537013292312622,
|
|
"step": 2785,
|
|
"valid_targets_mean": 1823.8,
|
|
"valid_targets_min": 1283
|
|
},
|
|
{
|
|
"epoch": 3.885793871866295,
|
|
"grad_norm": 0.928372055697338,
|
|
"learning_rate": 1.965967143401178e-05,
|
|
"loss": 0.1439,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05227617174386978,
|
|
"step": 2790,
|
|
"valid_targets_mean": 924.0,
|
|
"valid_targets_min": 620
|
|
},
|
|
{
|
|
"epoch": 3.8927576601671308,
|
|
"grad_norm": 0.8016846902818193,
|
|
"learning_rate": 1.9590225521122167e-05,
|
|
"loss": 0.1455,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06370696425437927,
|
|
"step": 2795,
|
|
"valid_targets_mean": 1480.2,
|
|
"valid_targets_min": 656
|
|
},
|
|
{
|
|
"epoch": 3.8997214484679663,
|
|
"grad_norm": 0.8062692815757603,
|
|
"learning_rate": 1.9520784550554683e-05,
|
|
"loss": 0.1481,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.059848278760910034,
|
|
"step": 2800,
|
|
"valid_targets_mean": 1415.5,
|
|
"valid_targets_min": 697
|
|
},
|
|
{
|
|
"epoch": 3.9066852367688023,
|
|
"grad_norm": 0.8353157154283097,
|
|
"learning_rate": 1.9451349359842332e-05,
|
|
"loss": 0.1442,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07082884013652802,
|
|
"step": 2805,
|
|
"valid_targets_mean": 1655.8,
|
|
"valid_targets_min": 816
|
|
},
|
|
{
|
|
"epoch": 3.913649025069638,
|
|
"grad_norm": 0.7597621427921378,
|
|
"learning_rate": 1.938192078644839e-05,
|
|
"loss": 0.1417,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06765986979007721,
|
|
"step": 2810,
|
|
"valid_targets_mean": 1553.0,
|
|
"valid_targets_min": 850
|
|
},
|
|
{
|
|
"epoch": 3.9206128133704734,
|
|
"grad_norm": 0.7816154228464142,
|
|
"learning_rate": 1.931249966775633e-05,
|
|
"loss": 0.1411,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06977061927318573,
|
|
"step": 2815,
|
|
"valid_targets_mean": 1353.9,
|
|
"valid_targets_min": 747
|
|
},
|
|
{
|
|
"epoch": 3.9275766016713094,
|
|
"grad_norm": 0.8523628518802676,
|
|
"learning_rate": 1.92430868410597e-05,
|
|
"loss": 0.1548,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0653572827577591,
|
|
"step": 2820,
|
|
"valid_targets_mean": 1420.9,
|
|
"valid_targets_min": 933
|
|
},
|
|
{
|
|
"epoch": 3.934540389972145,
|
|
"grad_norm": 0.801382105679613,
|
|
"learning_rate": 1.9173683143552057e-05,
|
|
"loss": 0.147,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07180759310722351,
|
|
"step": 2825,
|
|
"valid_targets_mean": 1394.9,
|
|
"valid_targets_min": 827
|
|
},
|
|
{
|
|
"epoch": 3.9415041782729805,
|
|
"grad_norm": 0.7523245478800936,
|
|
"learning_rate": 1.910428941231684e-05,
|
|
"loss": 0.1421,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06203988194465637,
|
|
"step": 2830,
|
|
"valid_targets_mean": 1383.2,
|
|
"valid_targets_min": 909
|
|
},
|
|
{
|
|
"epoch": 3.948467966573816,
|
|
"grad_norm": 0.7403715943582305,
|
|
"learning_rate": 1.903490648431728e-05,
|
|
"loss": 0.1493,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08447936922311783,
|
|
"step": 2835,
|
|
"valid_targets_mean": 1682.4,
|
|
"valid_targets_min": 1094
|
|
},
|
|
{
|
|
"epoch": 3.9554317548746516,
|
|
"grad_norm": 0.8253011305912571,
|
|
"learning_rate": 1.8965535196386326e-05,
|
|
"loss": 0.1358,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06693306565284729,
|
|
"step": 2840,
|
|
"valid_targets_mean": 1118.4,
|
|
"valid_targets_min": 720
|
|
},
|
|
{
|
|
"epoch": 3.9623955431754876,
|
|
"grad_norm": 0.7530842490472827,
|
|
"learning_rate": 1.889617638521651e-05,
|
|
"loss": 0.1487,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07407623529434204,
|
|
"step": 2845,
|
|
"valid_targets_mean": 1538.2,
|
|
"valid_targets_min": 610
|
|
},
|
|
{
|
|
"epoch": 3.969359331476323,
|
|
"grad_norm": 0.7546263356082901,
|
|
"learning_rate": 1.8826830887349902e-05,
|
|
"loss": 0.1437,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07467672973871231,
|
|
"step": 2850,
|
|
"valid_targets_mean": 1570.1,
|
|
"valid_targets_min": 713
|
|
},
|
|
{
|
|
"epoch": 3.9763231197771587,
|
|
"grad_norm": 0.833106804892126,
|
|
"learning_rate": 1.8757499539167986e-05,
|
|
"loss": 0.1587,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07438725978136063,
|
|
"step": 2855,
|
|
"valid_targets_mean": 1537.8,
|
|
"valid_targets_min": 874
|
|
},
|
|
{
|
|
"epoch": 3.9832869080779947,
|
|
"grad_norm": 0.7504076181098892,
|
|
"learning_rate": 1.8688183176881596e-05,
|
|
"loss": 0.1472,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07164924591779709,
|
|
"step": 2860,
|
|
"valid_targets_mean": 1471.2,
|
|
"valid_targets_min": 765
|
|
},
|
|
{
|
|
"epoch": 3.9902506963788302,
|
|
"grad_norm": 0.7785848799373203,
|
|
"learning_rate": 1.861888263652081e-05,
|
|
"loss": 0.1485,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06348396092653275,
|
|
"step": 2865,
|
|
"valid_targets_mean": 1254.4,
|
|
"valid_targets_min": 816
|
|
},
|
|
{
|
|
"epoch": 3.997214484679666,
|
|
"grad_norm": 0.78598015466159,
|
|
"learning_rate": 1.8549598753924894e-05,
|
|
"loss": 0.1527,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0708591490983963,
|
|
"step": 2870,
|
|
"valid_targets_mean": 1482.9,
|
|
"valid_targets_min": 1182
|
|
},
|
|
{
|
|
"epoch": 4.004178272980502,
|
|
"grad_norm": 0.7745411560000336,
|
|
"learning_rate": 1.8480332364732178e-05,
|
|
"loss": 0.1787,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09853234887123108,
|
|
"step": 2875,
|
|
"valid_targets_mean": 5435.1,
|
|
"valid_targets_min": 3516
|
|
},
|
|
{
|
|
"epoch": 4.011142061281337,
|
|
"grad_norm": 0.45693392884456113,
|
|
"learning_rate": 1.8411084304370014e-05,
|
|
"loss": 0.1987,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09516393393278122,
|
|
"step": 2880,
|
|
"valid_targets_mean": 7654.2,
|
|
"valid_targets_min": 6354
|
|
},
|
|
{
|
|
"epoch": 4.018105849582173,
|
|
"grad_norm": 0.4319211248040884,
|
|
"learning_rate": 1.8341855408044697e-05,
|
|
"loss": 0.1906,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10908873379230499,
|
|
"step": 2885,
|
|
"valid_targets_mean": 7341.6,
|
|
"valid_targets_min": 5754
|
|
},
|
|
{
|
|
"epoch": 4.025069637883008,
|
|
"grad_norm": 0.35087638698700635,
|
|
"learning_rate": 1.8272646510731375e-05,
|
|
"loss": 0.1834,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08147092163562775,
|
|
"step": 2890,
|
|
"valid_targets_mean": 7666.2,
|
|
"valid_targets_min": 5666
|
|
},
|
|
{
|
|
"epoch": 4.032033426183844,
|
|
"grad_norm": 0.4079352838694494,
|
|
"learning_rate": 1.820345844716399e-05,
|
|
"loss": 0.1835,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09540951251983643,
|
|
"step": 2895,
|
|
"valid_targets_mean": 7469.0,
|
|
"valid_targets_min": 5646
|
|
},
|
|
{
|
|
"epoch": 4.03899721448468,
|
|
"grad_norm": 0.453891412990921,
|
|
"learning_rate": 1.81342920518252e-05,
|
|
"loss": 0.1844,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1076955646276474,
|
|
"step": 2900,
|
|
"valid_targets_mean": 7046.9,
|
|
"valid_targets_min": 5571
|
|
},
|
|
{
|
|
"epoch": 4.045961002785515,
|
|
"grad_norm": 0.44912248872165283,
|
|
"learning_rate": 1.8065148158936323e-05,
|
|
"loss": 0.1985,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10332390666007996,
|
|
"step": 2905,
|
|
"valid_targets_mean": 7827.6,
|
|
"valid_targets_min": 5405
|
|
},
|
|
{
|
|
"epoch": 4.052924791086351,
|
|
"grad_norm": 0.3855281872703947,
|
|
"learning_rate": 1.799602760244728e-05,
|
|
"loss": 0.1839,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09584302455186844,
|
|
"step": 2910,
|
|
"valid_targets_mean": 7023.5,
|
|
"valid_targets_min": 5664
|
|
},
|
|
{
|
|
"epoch": 4.059888579387187,
|
|
"grad_norm": 0.39674294021757817,
|
|
"learning_rate": 1.7926931216026518e-05,
|
|
"loss": 0.178,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08769885450601578,
|
|
"step": 2915,
|
|
"valid_targets_mean": 7069.8,
|
|
"valid_targets_min": 5103
|
|
},
|
|
{
|
|
"epoch": 4.066852367688022,
|
|
"grad_norm": 0.4247588326582294,
|
|
"learning_rate": 1.7857859833050983e-05,
|
|
"loss": 0.1628,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08552022278308868,
|
|
"step": 2920,
|
|
"valid_targets_mean": 6712.4,
|
|
"valid_targets_min": 5469
|
|
},
|
|
{
|
|
"epoch": 4.073816155988858,
|
|
"grad_norm": 0.4334738323877378,
|
|
"learning_rate": 1.7788814286596044e-05,
|
|
"loss": 0.1798,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0862928107380867,
|
|
"step": 2925,
|
|
"valid_targets_mean": 7289.2,
|
|
"valid_targets_min": 5513
|
|
},
|
|
{
|
|
"epoch": 4.080779944289693,
|
|
"grad_norm": 0.5671431858648565,
|
|
"learning_rate": 1.7719795409425443e-05,
|
|
"loss": 0.1718,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08433561772108078,
|
|
"step": 2930,
|
|
"valid_targets_mean": 7022.1,
|
|
"valid_targets_min": 5691
|
|
},
|
|
{
|
|
"epoch": 4.087743732590529,
|
|
"grad_norm": 0.5417102538090298,
|
|
"learning_rate": 1.765080403398127e-05,
|
|
"loss": 0.1801,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0926387831568718,
|
|
"step": 2935,
|
|
"valid_targets_mean": 6736.0,
|
|
"valid_targets_min": 5550
|
|
},
|
|
{
|
|
"epoch": 4.094707520891365,
|
|
"grad_norm": 0.3961734843289232,
|
|
"learning_rate": 1.758184099237393e-05,
|
|
"loss": 0.1671,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08144709467887878,
|
|
"step": 2940,
|
|
"valid_targets_mean": 6346.9,
|
|
"valid_targets_min": 5067
|
|
},
|
|
{
|
|
"epoch": 4.1016713091922,
|
|
"grad_norm": 0.3937587901125671,
|
|
"learning_rate": 1.7512907116372066e-05,
|
|
"loss": 0.1718,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08808314055204391,
|
|
"step": 2945,
|
|
"valid_targets_mean": 6784.0,
|
|
"valid_targets_min": 5618
|
|
},
|
|
{
|
|
"epoch": 4.108635097493036,
|
|
"grad_norm": 0.43597320037991466,
|
|
"learning_rate": 1.7444003237392567e-05,
|
|
"loss": 0.1784,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10079539567232132,
|
|
"step": 2950,
|
|
"valid_targets_mean": 7568.4,
|
|
"valid_targets_min": 5856
|
|
},
|
|
{
|
|
"epoch": 4.1155988857938715,
|
|
"grad_norm": 0.3994536656356681,
|
|
"learning_rate": 1.737513018649053e-05,
|
|
"loss": 0.1707,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08414926379919052,
|
|
"step": 2955,
|
|
"valid_targets_mean": 7198.5,
|
|
"valid_targets_min": 4793
|
|
},
|
|
{
|
|
"epoch": 4.1225626740947074,
|
|
"grad_norm": 0.4184797204828292,
|
|
"learning_rate": 1.7306288794349213e-05,
|
|
"loss": 0.1761,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08882571756839752,
|
|
"step": 2960,
|
|
"valid_targets_mean": 6680.5,
|
|
"valid_targets_min": 5350
|
|
},
|
|
{
|
|
"epoch": 4.129526462395543,
|
|
"grad_norm": 0.4148372539820588,
|
|
"learning_rate": 1.7237479891270054e-05,
|
|
"loss": 0.1876,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0935942679643631,
|
|
"step": 2965,
|
|
"valid_targets_mean": 6770.6,
|
|
"valid_targets_min": 5753
|
|
},
|
|
{
|
|
"epoch": 4.1364902506963785,
|
|
"grad_norm": 0.7626972756641216,
|
|
"learning_rate": 1.7168704307162625e-05,
|
|
"loss": 0.1835,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06254291534423828,
|
|
"step": 2970,
|
|
"valid_targets_mean": 1419.8,
|
|
"valid_targets_min": 289
|
|
},
|
|
{
|
|
"epoch": 4.1434540389972145,
|
|
"grad_norm": 0.4371572099046906,
|
|
"learning_rate": 1.7099962871534644e-05,
|
|
"loss": 0.1518,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09101256728172302,
|
|
"step": 2975,
|
|
"valid_targets_mean": 6784.2,
|
|
"valid_targets_min": 5042
|
|
},
|
|
{
|
|
"epoch": 4.1504178272980505,
|
|
"grad_norm": 0.4078695342570583,
|
|
"learning_rate": 1.703125641348197e-05,
|
|
"loss": 0.169,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08425340801477432,
|
|
"step": 2980,
|
|
"valid_targets_mean": 6774.9,
|
|
"valid_targets_min": 3875
|
|
},
|
|
{
|
|
"epoch": 4.157381615598886,
|
|
"grad_norm": 0.43908909204234714,
|
|
"learning_rate": 1.696258576167856e-05,
|
|
"loss": 0.1744,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09084954857826233,
|
|
"step": 2985,
|
|
"valid_targets_mean": 7047.6,
|
|
"valid_targets_min": 6134
|
|
},
|
|
{
|
|
"epoch": 4.164345403899722,
|
|
"grad_norm": 0.3923313188578705,
|
|
"learning_rate": 1.6893951744366547e-05,
|
|
"loss": 0.1783,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09309709817171097,
|
|
"step": 2990,
|
|
"valid_targets_mean": 6886.6,
|
|
"valid_targets_min": 5869
|
|
},
|
|
{
|
|
"epoch": 4.171309192200557,
|
|
"grad_norm": 0.377006462886014,
|
|
"learning_rate": 1.6825355189346192e-05,
|
|
"loss": 0.1745,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08231081813573837,
|
|
"step": 2995,
|
|
"valid_targets_mean": 7559.8,
|
|
"valid_targets_min": 6145
|
|
},
|
|
{
|
|
"epoch": 4.178272980501393,
|
|
"grad_norm": 0.3411126289914851,
|
|
"learning_rate": 1.675679692396593e-05,
|
|
"loss": 0.1631,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07891160249710083,
|
|
"step": 3000,
|
|
"valid_targets_mean": 8920.5,
|
|
"valid_targets_min": 4780
|
|
},
|
|
{
|
|
"epoch": 4.185236768802229,
|
|
"grad_norm": 0.3592395577774327,
|
|
"learning_rate": 1.6688277775112377e-05,
|
|
"loss": 0.1517,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06996404379606247,
|
|
"step": 3005,
|
|
"valid_targets_mean": 7393.4,
|
|
"valid_targets_min": 5116
|
|
},
|
|
{
|
|
"epoch": 4.192200557103064,
|
|
"grad_norm": 0.34174396201498236,
|
|
"learning_rate": 1.6619798569200368e-05,
|
|
"loss": 0.1552,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07682168483734131,
|
|
"step": 3010,
|
|
"valid_targets_mean": 8477.5,
|
|
"valid_targets_min": 4033
|
|
},
|
|
{
|
|
"epoch": 4.1991643454039,
|
|
"grad_norm": 0.40130345703058956,
|
|
"learning_rate": 1.6551360132162967e-05,
|
|
"loss": 0.173,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08580783009529114,
|
|
"step": 3015,
|
|
"valid_targets_mean": 6736.6,
|
|
"valid_targets_min": 4646
|
|
},
|
|
{
|
|
"epoch": 4.206128133704736,
|
|
"grad_norm": 0.43606325450974787,
|
|
"learning_rate": 1.6482963289441533e-05,
|
|
"loss": 0.1575,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09041762351989746,
|
|
"step": 3020,
|
|
"valid_targets_mean": 7551.5,
|
|
"valid_targets_min": 5461
|
|
},
|
|
{
|
|
"epoch": 4.213091922005571,
|
|
"grad_norm": 0.45829099636483556,
|
|
"learning_rate": 1.6414608865975757e-05,
|
|
"loss": 0.1824,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10397613048553467,
|
|
"step": 3025,
|
|
"valid_targets_mean": 7091.0,
|
|
"valid_targets_min": 4879
|
|
},
|
|
{
|
|
"epoch": 4.220055710306407,
|
|
"grad_norm": 0.42080713802214453,
|
|
"learning_rate": 1.6346297686193698e-05,
|
|
"loss": 0.1869,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08283068239688873,
|
|
"step": 3030,
|
|
"valid_targets_mean": 6269.8,
|
|
"valid_targets_min": 4606
|
|
},
|
|
{
|
|
"epoch": 4.227019498607242,
|
|
"grad_norm": 0.4001310643885332,
|
|
"learning_rate": 1.6278030574001863e-05,
|
|
"loss": 0.1834,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08124437183141708,
|
|
"step": 3035,
|
|
"valid_targets_mean": 6818.4,
|
|
"valid_targets_min": 5525
|
|
},
|
|
{
|
|
"epoch": 4.233983286908078,
|
|
"grad_norm": 0.3928643150423594,
|
|
"learning_rate": 1.6209808352775227e-05,
|
|
"loss": 0.1725,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08757513761520386,
|
|
"step": 3040,
|
|
"valid_targets_mean": 7365.4,
|
|
"valid_targets_min": 5000
|
|
},
|
|
{
|
|
"epoch": 4.240947075208914,
|
|
"grad_norm": 0.42455615486180276,
|
|
"learning_rate": 1.6141631845347368e-05,
|
|
"loss": 0.1761,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07783573120832443,
|
|
"step": 3045,
|
|
"valid_targets_mean": 6212.1,
|
|
"valid_targets_min": 3988
|
|
},
|
|
{
|
|
"epoch": 4.247910863509749,
|
|
"grad_norm": 0.42241913812780074,
|
|
"learning_rate": 1.607350187400049e-05,
|
|
"loss": 0.1778,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08539704978466034,
|
|
"step": 3050,
|
|
"valid_targets_mean": 8396.9,
|
|
"valid_targets_min": 6131
|
|
},
|
|
{
|
|
"epoch": 4.254874651810585,
|
|
"grad_norm": 0.4237454815884405,
|
|
"learning_rate": 1.6005419260455512e-05,
|
|
"loss": 0.1759,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0880412757396698,
|
|
"step": 3055,
|
|
"valid_targets_mean": 7651.0,
|
|
"valid_targets_min": 4061
|
|
},
|
|
{
|
|
"epoch": 4.26183844011142,
|
|
"grad_norm": 0.37280072499256484,
|
|
"learning_rate": 1.593738482586219e-05,
|
|
"loss": 0.1785,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0859137699007988,
|
|
"step": 3060,
|
|
"valid_targets_mean": 7090.1,
|
|
"valid_targets_min": 5203
|
|
},
|
|
{
|
|
"epoch": 4.268802228412256,
|
|
"grad_norm": 0.41477357509750673,
|
|
"learning_rate": 1.5869399390789172e-05,
|
|
"loss": 0.1778,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08433390408754349,
|
|
"step": 3065,
|
|
"valid_targets_mean": 7001.0,
|
|
"valid_targets_min": 5362
|
|
},
|
|
{
|
|
"epoch": 4.275766016713092,
|
|
"grad_norm": 0.40793877263257894,
|
|
"learning_rate": 1.5801463775214118e-05,
|
|
"loss": 0.1781,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08703209459781647,
|
|
"step": 3070,
|
|
"valid_targets_mean": 6359.2,
|
|
"valid_targets_min": 5236
|
|
},
|
|
{
|
|
"epoch": 4.282729805013927,
|
|
"grad_norm": 0.44350806675129867,
|
|
"learning_rate": 1.573357879851381e-05,
|
|
"loss": 0.177,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09586170315742493,
|
|
"step": 3075,
|
|
"valid_targets_mean": 5939.8,
|
|
"valid_targets_min": 4693
|
|
},
|
|
{
|
|
"epoch": 4.289693593314763,
|
|
"grad_norm": 1.146676173490332,
|
|
"learning_rate": 1.5665745279454294e-05,
|
|
"loss": 0.1441,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.030542651191353798,
|
|
"step": 3080,
|
|
"valid_targets_mean": 208.6,
|
|
"valid_targets_min": 134
|
|
},
|
|
{
|
|
"epoch": 4.296657381615599,
|
|
"grad_norm": 0.48379943817559756,
|
|
"learning_rate": 1.5597964036180955e-05,
|
|
"loss": 0.1751,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09201765060424805,
|
|
"step": 3085,
|
|
"valid_targets_mean": 6008.5,
|
|
"valid_targets_min": 4956
|
|
},
|
|
{
|
|
"epoch": 4.303621169916434,
|
|
"grad_norm": 0.4202491008226685,
|
|
"learning_rate": 1.553023588620869e-05,
|
|
"loss": 0.171,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07570789754390717,
|
|
"step": 3090,
|
|
"valid_targets_mean": 6042.0,
|
|
"valid_targets_min": 4817
|
|
},
|
|
{
|
|
"epoch": 4.31058495821727,
|
|
"grad_norm": 0.42477738905964485,
|
|
"learning_rate": 1.5462561646412028e-05,
|
|
"loss": 0.1765,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08630920946598053,
|
|
"step": 3095,
|
|
"valid_targets_mean": 6608.6,
|
|
"valid_targets_min": 5102
|
|
},
|
|
{
|
|
"epoch": 4.3175487465181055,
|
|
"grad_norm": 0.42688040529738014,
|
|
"learning_rate": 1.5394942133015278e-05,
|
|
"loss": 0.1679,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08623873442411423,
|
|
"step": 3100,
|
|
"valid_targets_mean": 6670.1,
|
|
"valid_targets_min": 4869
|
|
},
|
|
{
|
|
"epoch": 4.3245125348189415,
|
|
"grad_norm": 0.42003992455965294,
|
|
"learning_rate": 1.5327378161582697e-05,
|
|
"loss": 0.1685,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07713023573160172,
|
|
"step": 3105,
|
|
"valid_targets_mean": 6088.4,
|
|
"valid_targets_min": 4946
|
|
},
|
|
{
|
|
"epoch": 4.3314763231197775,
|
|
"grad_norm": 0.4216278012962236,
|
|
"learning_rate": 1.5259870547008647e-05,
|
|
"loss": 0.1797,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08279541879892349,
|
|
"step": 3110,
|
|
"valid_targets_mean": 6228.4,
|
|
"valid_targets_min": 4856
|
|
},
|
|
{
|
|
"epoch": 4.338440111420613,
|
|
"grad_norm": 0.44787796967160803,
|
|
"learning_rate": 1.5192420103507765e-05,
|
|
"loss": 0.1769,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08231884241104126,
|
|
"step": 3115,
|
|
"valid_targets_mean": 5376.6,
|
|
"valid_targets_min": 4145
|
|
},
|
|
{
|
|
"epoch": 4.345403899721449,
|
|
"grad_norm": 0.4676976765001585,
|
|
"learning_rate": 1.5125027644605143e-05,
|
|
"loss": 0.1369,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08625025302171707,
|
|
"step": 3120,
|
|
"valid_targets_mean": 6309.2,
|
|
"valid_targets_min": 5305
|
|
},
|
|
{
|
|
"epoch": 4.352367688022284,
|
|
"grad_norm": 0.4889513820698066,
|
|
"learning_rate": 1.50576939831265e-05,
|
|
"loss": 0.1704,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08322583138942719,
|
|
"step": 3125,
|
|
"valid_targets_mean": 6221.5,
|
|
"valid_targets_min": 4636
|
|
},
|
|
{
|
|
"epoch": 4.35933147632312,
|
|
"grad_norm": 0.4369115623225211,
|
|
"learning_rate": 1.4990419931188425e-05,
|
|
"loss": 0.1698,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09070675075054169,
|
|
"step": 3130,
|
|
"valid_targets_mean": 6831.5,
|
|
"valid_targets_min": 5668
|
|
},
|
|
{
|
|
"epoch": 4.366295264623956,
|
|
"grad_norm": 0.42542268683362244,
|
|
"learning_rate": 1.4923206300188522e-05,
|
|
"loss": 0.1722,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09038935601711273,
|
|
"step": 3135,
|
|
"valid_targets_mean": 6437.1,
|
|
"valid_targets_min": 5004
|
|
},
|
|
{
|
|
"epoch": 4.373259052924791,
|
|
"grad_norm": 0.5378332721568458,
|
|
"learning_rate": 1.4856053900795682e-05,
|
|
"loss": 0.1655,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08066794276237488,
|
|
"step": 3140,
|
|
"valid_targets_mean": 6205.8,
|
|
"valid_targets_min": 5341
|
|
},
|
|
{
|
|
"epoch": 4.380222841225627,
|
|
"grad_norm": 0.4295836677794995,
|
|
"learning_rate": 1.4788963542940251e-05,
|
|
"loss": 0.1698,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09073355048894882,
|
|
"step": 3145,
|
|
"valid_targets_mean": 6748.0,
|
|
"valid_targets_min": 5363
|
|
},
|
|
{
|
|
"epoch": 4.387186629526463,
|
|
"grad_norm": 0.4850027247112984,
|
|
"learning_rate": 1.4721936035804307e-05,
|
|
"loss": 0.1686,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07909142971038818,
|
|
"step": 3150,
|
|
"valid_targets_mean": 5210.5,
|
|
"valid_targets_min": 791
|
|
},
|
|
{
|
|
"epoch": 4.394150417827298,
|
|
"grad_norm": 0.7383337146665891,
|
|
"learning_rate": 1.4654972187811858e-05,
|
|
"loss": 0.1717,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06561724841594696,
|
|
"step": 3155,
|
|
"valid_targets_mean": 1279.5,
|
|
"valid_targets_min": 746
|
|
},
|
|
{
|
|
"epoch": 4.401114206128134,
|
|
"grad_norm": 0.8310596897559344,
|
|
"learning_rate": 1.4588072806619135e-05,
|
|
"loss": 0.1534,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08013515174388885,
|
|
"step": 3160,
|
|
"valid_targets_mean": 1635.8,
|
|
"valid_targets_min": 661
|
|
},
|
|
{
|
|
"epoch": 4.408077994428969,
|
|
"grad_norm": 0.8516771248517533,
|
|
"learning_rate": 1.4521238699104828e-05,
|
|
"loss": 0.1453,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07185165584087372,
|
|
"step": 3165,
|
|
"valid_targets_mean": 1420.2,
|
|
"valid_targets_min": 826
|
|
},
|
|
{
|
|
"epoch": 4.415041782729805,
|
|
"grad_norm": 0.825781158854519,
|
|
"learning_rate": 1.4454470671360337e-05,
|
|
"loss": 0.1437,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06541028618812561,
|
|
"step": 3170,
|
|
"valid_targets_mean": 1399.9,
|
|
"valid_targets_min": 742
|
|
},
|
|
{
|
|
"epoch": 4.422005571030641,
|
|
"grad_norm": 0.8548496982709529,
|
|
"learning_rate": 1.4387769528680098e-05,
|
|
"loss": 0.1517,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.060420047491788864,
|
|
"step": 3175,
|
|
"valid_targets_mean": 1309.9,
|
|
"valid_targets_min": 718
|
|
},
|
|
{
|
|
"epoch": 4.428969359331476,
|
|
"grad_norm": 0.8030341275951788,
|
|
"learning_rate": 1.4321136075551816e-05,
|
|
"loss": 0.1479,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08500155806541443,
|
|
"step": 3180,
|
|
"valid_targets_mean": 1729.4,
|
|
"valid_targets_min": 764
|
|
},
|
|
{
|
|
"epoch": 4.435933147632312,
|
|
"grad_norm": 0.7574670919758995,
|
|
"learning_rate": 1.4254571115646795e-05,
|
|
"loss": 0.136,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06925354897975922,
|
|
"step": 3185,
|
|
"valid_targets_mean": 1651.4,
|
|
"valid_targets_min": 814
|
|
},
|
|
{
|
|
"epoch": 4.442896935933147,
|
|
"grad_norm": 0.8399886623759784,
|
|
"learning_rate": 1.4188075451810238e-05,
|
|
"loss": 0.1509,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05408865213394165,
|
|
"step": 3190,
|
|
"valid_targets_mean": 1192.0,
|
|
"valid_targets_min": 610
|
|
},
|
|
{
|
|
"epoch": 4.449860724233983,
|
|
"grad_norm": 0.8256278183381764,
|
|
"learning_rate": 1.4121649886051567e-05,
|
|
"loss": 0.1401,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07870268821716309,
|
|
"step": 3195,
|
|
"valid_targets_mean": 1879.0,
|
|
"valid_targets_min": 817
|
|
},
|
|
{
|
|
"epoch": 4.456824512534819,
|
|
"grad_norm": 0.7713275384072368,
|
|
"learning_rate": 1.4055295219534735e-05,
|
|
"loss": 0.143,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06405435502529144,
|
|
"step": 3200,
|
|
"valid_targets_mean": 1500.4,
|
|
"valid_targets_min": 837
|
|
},
|
|
{
|
|
"epoch": 4.463788300835654,
|
|
"grad_norm": 0.9706438394826296,
|
|
"learning_rate": 1.3989012252568596e-05,
|
|
"loss": 0.1394,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09357143938541412,
|
|
"step": 3205,
|
|
"valid_targets_mean": 1661.8,
|
|
"valid_targets_min": 593
|
|
},
|
|
{
|
|
"epoch": 4.47075208913649,
|
|
"grad_norm": 0.7845778542370729,
|
|
"learning_rate": 1.3922801784597191e-05,
|
|
"loss": 0.1384,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06827139109373093,
|
|
"step": 3210,
|
|
"valid_targets_mean": 1537.8,
|
|
"valid_targets_min": 704
|
|
},
|
|
{
|
|
"epoch": 4.477715877437326,
|
|
"grad_norm": 0.7759659105769868,
|
|
"learning_rate": 1.3856664614190186e-05,
|
|
"loss": 0.132,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0502796433866024,
|
|
"step": 3215,
|
|
"valid_targets_mean": 1227.0,
|
|
"valid_targets_min": 722
|
|
},
|
|
{
|
|
"epoch": 4.484679665738161,
|
|
"grad_norm": 0.8287147960258013,
|
|
"learning_rate": 1.3790601539033168e-05,
|
|
"loss": 0.1308,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06461460888385773,
|
|
"step": 3220,
|
|
"valid_targets_mean": 1364.5,
|
|
"valid_targets_min": 642
|
|
},
|
|
{
|
|
"epoch": 4.491643454038997,
|
|
"grad_norm": 0.8589525229582935,
|
|
"learning_rate": 1.3724613355918077e-05,
|
|
"loss": 0.1304,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06513009965419769,
|
|
"step": 3225,
|
|
"valid_targets_mean": 1358.5,
|
|
"valid_targets_min": 518
|
|
},
|
|
{
|
|
"epoch": 4.498607242339833,
|
|
"grad_norm": 0.8202013278047937,
|
|
"learning_rate": 1.3658700860733566e-05,
|
|
"loss": 0.1419,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07893405854701996,
|
|
"step": 3230,
|
|
"valid_targets_mean": 1590.6,
|
|
"valid_targets_min": 728
|
|
},
|
|
{
|
|
"epoch": 4.505571030640668,
|
|
"grad_norm": 0.7966714906034105,
|
|
"learning_rate": 1.3592864848455415e-05,
|
|
"loss": 0.1378,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0855114758014679,
|
|
"step": 3235,
|
|
"valid_targets_mean": 1791.4,
|
|
"valid_targets_min": 820
|
|
},
|
|
{
|
|
"epoch": 4.512534818941504,
|
|
"grad_norm": 0.8234446947551594,
|
|
"learning_rate": 1.3527106113136915e-05,
|
|
"loss": 0.1422,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07350599765777588,
|
|
"step": 3240,
|
|
"valid_targets_mean": 1470.8,
|
|
"valid_targets_min": 818
|
|
},
|
|
{
|
|
"epoch": 4.5194986072423395,
|
|
"grad_norm": 0.909954505776544,
|
|
"learning_rate": 1.3461425447899345e-05,
|
|
"loss": 0.1328,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05728901922702789,
|
|
"step": 3245,
|
|
"valid_targets_mean": 1278.6,
|
|
"valid_targets_min": 625
|
|
},
|
|
{
|
|
"epoch": 4.5264623955431755,
|
|
"grad_norm": 0.7815284768559326,
|
|
"learning_rate": 1.3395823644922357e-05,
|
|
"loss": 0.1374,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08874043077230453,
|
|
"step": 3250,
|
|
"valid_targets_mean": 2313.5,
|
|
"valid_targets_min": 790
|
|
},
|
|
{
|
|
"epoch": 4.5334261838440115,
|
|
"grad_norm": 0.8603759963719848,
|
|
"learning_rate": 1.3330301495434439e-05,
|
|
"loss": 0.1415,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09591110795736313,
|
|
"step": 3255,
|
|
"valid_targets_mean": 2231.9,
|
|
"valid_targets_min": 1335
|
|
},
|
|
{
|
|
"epoch": 4.540389972144847,
|
|
"grad_norm": 0.758467847298457,
|
|
"learning_rate": 1.3264859789703391e-05,
|
|
"loss": 0.1334,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07577726244926453,
|
|
"step": 3260,
|
|
"valid_targets_mean": 1729.5,
|
|
"valid_targets_min": 719
|
|
},
|
|
{
|
|
"epoch": 4.547353760445683,
|
|
"grad_norm": 0.8154388753592663,
|
|
"learning_rate": 1.3199499317026741e-05,
|
|
"loss": 0.1302,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.052854329347610474,
|
|
"step": 3265,
|
|
"valid_targets_mean": 1222.9,
|
|
"valid_targets_min": 697
|
|
},
|
|
{
|
|
"epoch": 4.554317548746518,
|
|
"grad_norm": 0.8552854445354738,
|
|
"learning_rate": 1.3134220865722282e-05,
|
|
"loss": 0.1289,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07314737141132355,
|
|
"step": 3270,
|
|
"valid_targets_mean": 1521.1,
|
|
"valid_targets_min": 998
|
|
},
|
|
{
|
|
"epoch": 4.561281337047354,
|
|
"grad_norm": 0.7820786112177985,
|
|
"learning_rate": 1.3069025223118539e-05,
|
|
"loss": 0.133,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06754308938980103,
|
|
"step": 3275,
|
|
"valid_targets_mean": 1693.9,
|
|
"valid_targets_min": 629
|
|
},
|
|
{
|
|
"epoch": 4.56824512534819,
|
|
"grad_norm": 0.8755511497986203,
|
|
"learning_rate": 1.3003913175545271e-05,
|
|
"loss": 0.134,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07453906536102295,
|
|
"step": 3280,
|
|
"valid_targets_mean": 1630.2,
|
|
"valid_targets_min": 1078
|
|
},
|
|
{
|
|
"epoch": 4.575208913649025,
|
|
"grad_norm": 0.7758564032420037,
|
|
"learning_rate": 1.2938885508323989e-05,
|
|
"loss": 0.1376,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06837262213230133,
|
|
"step": 3285,
|
|
"valid_targets_mean": 1843.0,
|
|
"valid_targets_min": 1075
|
|
},
|
|
{
|
|
"epoch": 4.582172701949861,
|
|
"grad_norm": 0.7809622207382767,
|
|
"learning_rate": 1.28739430057585e-05,
|
|
"loss": 0.14,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07848972082138062,
|
|
"step": 3290,
|
|
"valid_targets_mean": 1848.2,
|
|
"valid_targets_min": 1317
|
|
},
|
|
{
|
|
"epoch": 4.589136490250697,
|
|
"grad_norm": 0.7816839323875953,
|
|
"learning_rate": 1.2809086451125402e-05,
|
|
"loss": 0.1314,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07772564888000488,
|
|
"step": 3295,
|
|
"valid_targets_mean": 1727.4,
|
|
"valid_targets_min": 957
|
|
},
|
|
{
|
|
"epoch": 4.596100278551532,
|
|
"grad_norm": 0.9203545388755756,
|
|
"learning_rate": 1.2744316626664702e-05,
|
|
"loss": 0.1361,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.060780055820941925,
|
|
"step": 3300,
|
|
"valid_targets_mean": 1115.4,
|
|
"valid_targets_min": 643
|
|
},
|
|
{
|
|
"epoch": 4.603064066852368,
|
|
"grad_norm": 0.896065872644034,
|
|
"learning_rate": 1.2679634313570324e-05,
|
|
"loss": 0.14,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07272806763648987,
|
|
"step": 3305,
|
|
"valid_targets_mean": 1547.2,
|
|
"valid_targets_min": 1111
|
|
},
|
|
{
|
|
"epoch": 4.610027855153203,
|
|
"grad_norm": 0.7733642462901773,
|
|
"learning_rate": 1.2615040291980729e-05,
|
|
"loss": 0.1273,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.061603400856256485,
|
|
"step": 3310,
|
|
"valid_targets_mean": 1450.0,
|
|
"valid_targets_min": 705
|
|
},
|
|
{
|
|
"epoch": 4.616991643454039,
|
|
"grad_norm": 0.8143367409475855,
|
|
"learning_rate": 1.2550535340969483e-05,
|
|
"loss": 0.1305,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07132863998413086,
|
|
"step": 3315,
|
|
"valid_targets_mean": 1539.0,
|
|
"valid_targets_min": 1288
|
|
},
|
|
{
|
|
"epoch": 4.623955431754875,
|
|
"grad_norm": 0.8553022073322677,
|
|
"learning_rate": 1.2486120238535845e-05,
|
|
"loss": 0.1224,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06410254538059235,
|
|
"step": 3320,
|
|
"valid_targets_mean": 1321.2,
|
|
"valid_targets_min": 753
|
|
},
|
|
{
|
|
"epoch": 4.63091922005571,
|
|
"grad_norm": 0.8567062695633868,
|
|
"learning_rate": 1.2421795761595424e-05,
|
|
"loss": 0.1312,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0614665150642395,
|
|
"step": 3325,
|
|
"valid_targets_mean": 1480.0,
|
|
"valid_targets_min": 817
|
|
},
|
|
{
|
|
"epoch": 4.637883008356546,
|
|
"grad_norm": 0.8755257408477221,
|
|
"learning_rate": 1.2357562685970787e-05,
|
|
"loss": 0.1262,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05426248908042908,
|
|
"step": 3330,
|
|
"valid_targets_mean": 1293.9,
|
|
"valid_targets_min": 974
|
|
},
|
|
{
|
|
"epoch": 4.644846796657381,
|
|
"grad_norm": 0.9722123694394161,
|
|
"learning_rate": 1.2293421786382094e-05,
|
|
"loss": 0.1333,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07091125100851059,
|
|
"step": 3335,
|
|
"valid_targets_mean": 1529.2,
|
|
"valid_targets_min": 657
|
|
},
|
|
{
|
|
"epoch": 4.651810584958217,
|
|
"grad_norm": 0.9541829768814589,
|
|
"learning_rate": 1.2229373836437764e-05,
|
|
"loss": 0.1304,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07790433615446091,
|
|
"step": 3340,
|
|
"valid_targets_mean": 1678.0,
|
|
"valid_targets_min": 880
|
|
},
|
|
{
|
|
"epoch": 4.658774373259053,
|
|
"grad_norm": 0.8576485652747621,
|
|
"learning_rate": 1.2165419608625157e-05,
|
|
"loss": 0.1317,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07814347743988037,
|
|
"step": 3345,
|
|
"valid_targets_mean": 1609.4,
|
|
"valid_targets_min": 1050
|
|
},
|
|
{
|
|
"epoch": 4.665738161559888,
|
|
"grad_norm": 0.8917597302101115,
|
|
"learning_rate": 1.2101559874301217e-05,
|
|
"loss": 0.1247,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05816866457462311,
|
|
"step": 3350,
|
|
"valid_targets_mean": 1449.0,
|
|
"valid_targets_min": 682
|
|
},
|
|
{
|
|
"epoch": 4.672701949860724,
|
|
"grad_norm": 0.832501950400249,
|
|
"learning_rate": 1.2037795403683212e-05,
|
|
"loss": 0.1314,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06002386286854744,
|
|
"step": 3355,
|
|
"valid_targets_mean": 1344.0,
|
|
"valid_targets_min": 703
|
|
},
|
|
{
|
|
"epoch": 4.67966573816156,
|
|
"grad_norm": 0.8308585092709525,
|
|
"learning_rate": 1.1974126965839434e-05,
|
|
"loss": 0.1385,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06484422087669373,
|
|
"step": 3360,
|
|
"valid_targets_mean": 1469.4,
|
|
"valid_targets_min": 934
|
|
},
|
|
{
|
|
"epoch": 4.686629526462395,
|
|
"grad_norm": 0.8444644777001103,
|
|
"learning_rate": 1.1910555328679896e-05,
|
|
"loss": 0.1413,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1002117171883583,
|
|
"step": 3365,
|
|
"valid_targets_mean": 1829.1,
|
|
"valid_targets_min": 734
|
|
},
|
|
{
|
|
"epoch": 4.693593314763231,
|
|
"grad_norm": 0.7838276598240631,
|
|
"learning_rate": 1.18470812589471e-05,
|
|
"loss": 0.1276,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08434125781059265,
|
|
"step": 3370,
|
|
"valid_targets_mean": 2010.8,
|
|
"valid_targets_min": 1437
|
|
},
|
|
{
|
|
"epoch": 4.7005571030640665,
|
|
"grad_norm": 0.8121696038208687,
|
|
"learning_rate": 1.1783705522206794e-05,
|
|
"loss": 0.1248,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05047055333852768,
|
|
"step": 3375,
|
|
"valid_targets_mean": 1405.2,
|
|
"valid_targets_min": 551
|
|
},
|
|
{
|
|
"epoch": 4.7075208913649025,
|
|
"grad_norm": 0.9702607731512856,
|
|
"learning_rate": 1.1720428882838694e-05,
|
|
"loss": 0.1335,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08291260153055191,
|
|
"step": 3380,
|
|
"valid_targets_mean": 1817.6,
|
|
"valid_targets_min": 931
|
|
},
|
|
{
|
|
"epoch": 4.7144846796657385,
|
|
"grad_norm": 0.820699694440321,
|
|
"learning_rate": 1.1657252104027314e-05,
|
|
"loss": 0.1277,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06308387219905853,
|
|
"step": 3385,
|
|
"valid_targets_mean": 1599.1,
|
|
"valid_targets_min": 904
|
|
},
|
|
{
|
|
"epoch": 4.721448467966574,
|
|
"grad_norm": 0.829670605394756,
|
|
"learning_rate": 1.1594175947752749e-05,
|
|
"loss": 0.1252,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07266199588775635,
|
|
"step": 3390,
|
|
"valid_targets_mean": 1774.8,
|
|
"valid_targets_min": 1158
|
|
},
|
|
{
|
|
"epoch": 4.72841225626741,
|
|
"grad_norm": 0.8896710009000609,
|
|
"learning_rate": 1.1531201174781459e-05,
|
|
"loss": 0.1211,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05510660260915756,
|
|
"step": 3395,
|
|
"valid_targets_mean": 1229.2,
|
|
"valid_targets_min": 756
|
|
},
|
|
{
|
|
"epoch": 4.735376044568245,
|
|
"grad_norm": 1.1555963653569135,
|
|
"learning_rate": 1.1468328544657132e-05,
|
|
"loss": 0.1392,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08370531350374222,
|
|
"step": 3400,
|
|
"valid_targets_mean": 1536.9,
|
|
"valid_targets_min": 854
|
|
},
|
|
{
|
|
"epoch": 4.742339832869081,
|
|
"grad_norm": 0.8354454053066505,
|
|
"learning_rate": 1.1405558815691466e-05,
|
|
"loss": 0.1324,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07298648357391357,
|
|
"step": 3405,
|
|
"valid_targets_mean": 1739.5,
|
|
"valid_targets_min": 1462
|
|
},
|
|
{
|
|
"epoch": 4.749303621169917,
|
|
"grad_norm": 0.8260820753810716,
|
|
"learning_rate": 1.1342892744955112e-05,
|
|
"loss": 0.1282,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05890219286084175,
|
|
"step": 3410,
|
|
"valid_targets_mean": 1475.9,
|
|
"valid_targets_min": 944
|
|
},
|
|
{
|
|
"epoch": 4.756267409470752,
|
|
"grad_norm": 0.8016656660345047,
|
|
"learning_rate": 1.1280331088268452e-05,
|
|
"loss": 0.1269,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.060654304921627045,
|
|
"step": 3415,
|
|
"valid_targets_mean": 1482.6,
|
|
"valid_targets_min": 1227
|
|
},
|
|
{
|
|
"epoch": 4.763231197771588,
|
|
"grad_norm": 0.7864044513954926,
|
|
"learning_rate": 1.1217874600192542e-05,
|
|
"loss": 0.1309,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.051892273128032684,
|
|
"step": 3420,
|
|
"valid_targets_mean": 1276.5,
|
|
"valid_targets_min": 740
|
|
},
|
|
{
|
|
"epoch": 4.770194986072424,
|
|
"grad_norm": 0.774017966732346,
|
|
"learning_rate": 1.1155524034019981e-05,
|
|
"loss": 0.131,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07892292737960815,
|
|
"step": 3425,
|
|
"valid_targets_mean": 2024.8,
|
|
"valid_targets_min": 881
|
|
},
|
|
{
|
|
"epoch": 4.777158774373259,
|
|
"grad_norm": 0.8962743138721668,
|
|
"learning_rate": 1.1093280141765837e-05,
|
|
"loss": 0.1286,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06519632786512375,
|
|
"step": 3430,
|
|
"valid_targets_mean": 1288.6,
|
|
"valid_targets_min": 636
|
|
},
|
|
{
|
|
"epoch": 4.784122562674095,
|
|
"grad_norm": 0.8008310946938783,
|
|
"learning_rate": 1.103114367415858e-05,
|
|
"loss": 0.1255,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04663780331611633,
|
|
"step": 3435,
|
|
"valid_targets_mean": 1103.9,
|
|
"valid_targets_min": 814
|
|
},
|
|
{
|
|
"epoch": 4.791086350974931,
|
|
"grad_norm": 0.857674350118069,
|
|
"learning_rate": 1.0969115380631015e-05,
|
|
"loss": 0.1212,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.061598315834999084,
|
|
"step": 3440,
|
|
"valid_targets_mean": 1392.9,
|
|
"valid_targets_min": 820
|
|
},
|
|
{
|
|
"epoch": 4.798050139275766,
|
|
"grad_norm": 0.8096519542290342,
|
|
"learning_rate": 1.090719600931126e-05,
|
|
"loss": 0.129,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.057360272854566574,
|
|
"step": 3445,
|
|
"valid_targets_mean": 1392.4,
|
|
"valid_targets_min": 661
|
|
},
|
|
{
|
|
"epoch": 4.805013927576602,
|
|
"grad_norm": 0.8515212322986098,
|
|
"learning_rate": 1.0845386307013703e-05,
|
|
"loss": 0.1317,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06106788292527199,
|
|
"step": 3450,
|
|
"valid_targets_mean": 1441.8,
|
|
"valid_targets_min": 728
|
|
},
|
|
{
|
|
"epoch": 4.811977715877437,
|
|
"grad_norm": 0.8509639251731652,
|
|
"learning_rate": 1.078368701923003e-05,
|
|
"loss": 0.1248,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.060364048928022385,
|
|
"step": 3455,
|
|
"valid_targets_mean": 1413.5,
|
|
"valid_targets_min": 692
|
|
},
|
|
{
|
|
"epoch": 4.818941504178273,
|
|
"grad_norm": 0.8405241655525482,
|
|
"learning_rate": 1.0722098890120176e-05,
|
|
"loss": 0.1271,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06148139387369156,
|
|
"step": 3460,
|
|
"valid_targets_mean": 1323.5,
|
|
"valid_targets_min": 840
|
|
},
|
|
{
|
|
"epoch": 4.825905292479108,
|
|
"grad_norm": 0.9119451778532717,
|
|
"learning_rate": 1.0660622662503389e-05,
|
|
"loss": 0.1347,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06766475737094879,
|
|
"step": 3465,
|
|
"valid_targets_mean": 1488.4,
|
|
"valid_targets_min": 950
|
|
},
|
|
{
|
|
"epoch": 4.832869080779944,
|
|
"grad_norm": 0.7915913493434501,
|
|
"learning_rate": 1.0599259077849294e-05,
|
|
"loss": 0.1269,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07292820513248444,
|
|
"step": 3470,
|
|
"valid_targets_mean": 1935.9,
|
|
"valid_targets_min": 878
|
|
},
|
|
{
|
|
"epoch": 4.83983286908078,
|
|
"grad_norm": 0.8489205471637494,
|
|
"learning_rate": 1.0538008876268887e-05,
|
|
"loss": 0.1312,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06822683662176132,
|
|
"step": 3475,
|
|
"valid_targets_mean": 1502.1,
|
|
"valid_targets_min": 836
|
|
},
|
|
{
|
|
"epoch": 4.846796657381615,
|
|
"grad_norm": 0.8786148894939648,
|
|
"learning_rate": 1.0476872796505658e-05,
|
|
"loss": 0.1264,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07060012221336365,
|
|
"step": 3480,
|
|
"valid_targets_mean": 1535.1,
|
|
"valid_targets_min": 1013
|
|
},
|
|
{
|
|
"epoch": 4.853760445682451,
|
|
"grad_norm": 0.8959033919138696,
|
|
"learning_rate": 1.0415851575926656e-05,
|
|
"loss": 0.1312,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05586308240890503,
|
|
"step": 3485,
|
|
"valid_targets_mean": 1346.4,
|
|
"valid_targets_min": 748
|
|
},
|
|
{
|
|
"epoch": 4.860724233983287,
|
|
"grad_norm": 0.9017231573931604,
|
|
"learning_rate": 1.035494595051361e-05,
|
|
"loss": 0.1294,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06896314024925232,
|
|
"step": 3490,
|
|
"valid_targets_mean": 1543.9,
|
|
"valid_targets_min": 729
|
|
},
|
|
{
|
|
"epoch": 4.867688022284122,
|
|
"grad_norm": 0.8540161387269878,
|
|
"learning_rate": 1.0294156654854053e-05,
|
|
"loss": 0.1372,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.058618832379579544,
|
|
"step": 3495,
|
|
"valid_targets_mean": 1541.9,
|
|
"valid_targets_min": 1219
|
|
},
|
|
{
|
|
"epoch": 4.874651810584958,
|
|
"grad_norm": 0.8673351138138804,
|
|
"learning_rate": 1.023348442213244e-05,
|
|
"loss": 0.1298,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06518058478832245,
|
|
"step": 3500,
|
|
"valid_targets_mean": 1632.8,
|
|
"valid_targets_min": 916
|
|
},
|
|
{
|
|
"epoch": 4.881615598885794,
|
|
"grad_norm": 0.8662755506535637,
|
|
"learning_rate": 1.0172929984121333e-05,
|
|
"loss": 0.1292,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0599341094493866,
|
|
"step": 3505,
|
|
"valid_targets_mean": 1349.9,
|
|
"valid_targets_min": 646
|
|
},
|
|
{
|
|
"epoch": 4.888579387186629,
|
|
"grad_norm": 0.8789665027258394,
|
|
"learning_rate": 1.0112494071172549e-05,
|
|
"loss": 0.1254,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06385256350040436,
|
|
"step": 3510,
|
|
"valid_targets_mean": 1383.0,
|
|
"valid_targets_min": 906
|
|
},
|
|
{
|
|
"epoch": 4.895543175487465,
|
|
"grad_norm": 0.8475866768723133,
|
|
"learning_rate": 1.0052177412208401e-05,
|
|
"loss": 0.1225,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.059585411101579666,
|
|
"step": 3515,
|
|
"valid_targets_mean": 1577.5,
|
|
"valid_targets_min": 848
|
|
},
|
|
{
|
|
"epoch": 4.9025069637883005,
|
|
"grad_norm": 0.9212057570976543,
|
|
"learning_rate": 9.991980734712821e-06,
|
|
"loss": 0.1326,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06198536604642868,
|
|
"step": 3520,
|
|
"valid_targets_mean": 1183.5,
|
|
"valid_targets_min": 692
|
|
},
|
|
{
|
|
"epoch": 4.9094707520891365,
|
|
"grad_norm": 0.8950552623059923,
|
|
"learning_rate": 9.931904764722661e-06,
|
|
"loss": 0.123,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.058858513832092285,
|
|
"step": 3525,
|
|
"valid_targets_mean": 1373.9,
|
|
"valid_targets_min": 821
|
|
},
|
|
{
|
|
"epoch": 4.9164345403899725,
|
|
"grad_norm": 0.7942490619902811,
|
|
"learning_rate": 9.871950226818907e-06,
|
|
"loss": 0.1202,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05587601661682129,
|
|
"step": 3530,
|
|
"valid_targets_mean": 1428.1,
|
|
"valid_targets_min": 1013
|
|
},
|
|
{
|
|
"epoch": 4.923398328690808,
|
|
"grad_norm": 0.7959315882544532,
|
|
"learning_rate": 9.812117844117955e-06,
|
|
"loss": 0.1243,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06044112890958786,
|
|
"step": 3535,
|
|
"valid_targets_mean": 1420.9,
|
|
"valid_targets_min": 781
|
|
},
|
|
{
|
|
"epoch": 4.930362116991644,
|
|
"grad_norm": 0.9261429168716236,
|
|
"learning_rate": 9.752408338262864e-06,
|
|
"loss": 0.1365,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06673343479633331,
|
|
"step": 3540,
|
|
"valid_targets_mean": 1255.1,
|
|
"valid_targets_min": 548
|
|
},
|
|
{
|
|
"epoch": 4.937325905292479,
|
|
"grad_norm": 0.7859825717323234,
|
|
"learning_rate": 9.692822429414657e-06,
|
|
"loss": 0.1246,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0637277364730835,
|
|
"step": 3545,
|
|
"valid_targets_mean": 1411.0,
|
|
"valid_targets_min": 694
|
|
},
|
|
{
|
|
"epoch": 4.944289693593315,
|
|
"grad_norm": 0.8542974067863182,
|
|
"learning_rate": 9.633360836243656e-06,
|
|
"loss": 0.1335,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08179077506065369,
|
|
"step": 3550,
|
|
"valid_targets_mean": 1973.8,
|
|
"valid_targets_min": 1030
|
|
},
|
|
{
|
|
"epoch": 4.951253481894151,
|
|
"grad_norm": 0.7455850946114743,
|
|
"learning_rate": 9.57402427592081e-06,
|
|
"loss": 0.1209,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05281627178192139,
|
|
"step": 3555,
|
|
"valid_targets_mean": 1580.2,
|
|
"valid_targets_min": 682
|
|
},
|
|
{
|
|
"epoch": 4.958217270194986,
|
|
"grad_norm": 0.8918246113696654,
|
|
"learning_rate": 9.514813464109023e-06,
|
|
"loss": 0.1229,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06317047029733658,
|
|
"step": 3560,
|
|
"valid_targets_mean": 1425.4,
|
|
"valid_targets_min": 990
|
|
},
|
|
{
|
|
"epoch": 4.965181058495822,
|
|
"grad_norm": 0.8206588096352696,
|
|
"learning_rate": 9.455729114954542e-06,
|
|
"loss": 0.128,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04826410859823227,
|
|
"step": 3565,
|
|
"valid_targets_mean": 1299.6,
|
|
"valid_targets_min": 639
|
|
},
|
|
{
|
|
"epoch": 4.972144846796658,
|
|
"grad_norm": 0.8215144309267559,
|
|
"learning_rate": 9.396771941078335e-06,
|
|
"loss": 0.1254,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.053259532898664474,
|
|
"step": 3570,
|
|
"valid_targets_mean": 1458.8,
|
|
"valid_targets_min": 721
|
|
},
|
|
{
|
|
"epoch": 4.979108635097493,
|
|
"grad_norm": 0.8423487434531587,
|
|
"learning_rate": 9.337942653567504e-06,
|
|
"loss": 0.1361,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05987967178225517,
|
|
"step": 3575,
|
|
"valid_targets_mean": 1329.8,
|
|
"valid_targets_min": 874
|
|
},
|
|
{
|
|
"epoch": 4.986072423398329,
|
|
"grad_norm": 0.7786328247052688,
|
|
"learning_rate": 9.279241961966698e-06,
|
|
"loss": 0.1307,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0674387663602829,
|
|
"step": 3580,
|
|
"valid_targets_mean": 1559.9,
|
|
"valid_targets_min": 686
|
|
},
|
|
{
|
|
"epoch": 4.993036211699164,
|
|
"grad_norm": 0.8885857987523331,
|
|
"learning_rate": 9.220670574269564e-06,
|
|
"loss": 0.1328,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07278092205524445,
|
|
"step": 3585,
|
|
"valid_targets_mean": 1392.5,
|
|
"valid_targets_min": 713
|
|
},
|
|
{
|
|
"epoch": 5.0,
|
|
"grad_norm": 0.9810812179546335,
|
|
"learning_rate": 9.162229196910201e-06,
|
|
"loss": 0.1311,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09730862826108932,
|
|
"step": 3590,
|
|
"valid_targets_mean": 6130.4,
|
|
"valid_targets_min": 920
|
|
},
|
|
{
|
|
"epoch": 5.006963788300836,
|
|
"grad_norm": 0.7395884420407827,
|
|
"learning_rate": 9.103918534754648e-06,
|
|
"loss": 0.19,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09225118160247803,
|
|
"step": 3595,
|
|
"valid_targets_mean": 7389.0,
|
|
"valid_targets_min": 5627
|
|
},
|
|
{
|
|
"epoch": 5.013927576601671,
|
|
"grad_norm": 0.7448900926657113,
|
|
"learning_rate": 9.045739291092377e-06,
|
|
"loss": 0.1886,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05141618102788925,
|
|
"step": 3600,
|
|
"valid_targets_mean": 1294.8,
|
|
"valid_targets_min": 422
|
|
},
|
|
{
|
|
"epoch": 5.020891364902507,
|
|
"grad_norm": 0.43737029497845903,
|
|
"learning_rate": 8.987692167627807e-06,
|
|
"loss": 0.1775,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09226194024085999,
|
|
"step": 3605,
|
|
"valid_targets_mean": 8539.9,
|
|
"valid_targets_min": 6761
|
|
},
|
|
{
|
|
"epoch": 5.027855153203342,
|
|
"grad_norm": 0.40244324306844387,
|
|
"learning_rate": 8.929777864471847e-06,
|
|
"loss": 0.1705,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0802266001701355,
|
|
"step": 3610,
|
|
"valid_targets_mean": 7579.0,
|
|
"valid_targets_min": 6174
|
|
},
|
|
{
|
|
"epoch": 5.034818941504178,
|
|
"grad_norm": 0.3948157476493201,
|
|
"learning_rate": 8.871997080133446e-06,
|
|
"loss": 0.1733,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07746812701225281,
|
|
"step": 3615,
|
|
"valid_targets_mean": 7328.2,
|
|
"valid_targets_min": 4839
|
|
},
|
|
{
|
|
"epoch": 5.041782729805014,
|
|
"grad_norm": 0.42318640045066125,
|
|
"learning_rate": 8.814350511511187e-06,
|
|
"loss": 0.1823,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09257575869560242,
|
|
"step": 3620,
|
|
"valid_targets_mean": 6838.2,
|
|
"valid_targets_min": 5034
|
|
},
|
|
{
|
|
"epoch": 5.048746518105849,
|
|
"grad_norm": 0.4452897616046545,
|
|
"learning_rate": 8.756838853884862e-06,
|
|
"loss": 0.1838,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09431345760822296,
|
|
"step": 3625,
|
|
"valid_targets_mean": 7248.2,
|
|
"valid_targets_min": 5567
|
|
},
|
|
{
|
|
"epoch": 5.055710306406685,
|
|
"grad_norm": 0.4113841425843344,
|
|
"learning_rate": 8.699462800907063e-06,
|
|
"loss": 0.1687,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07702521979808807,
|
|
"step": 3630,
|
|
"valid_targets_mean": 6234.0,
|
|
"valid_targets_min": 5603
|
|
},
|
|
{
|
|
"epoch": 5.062674094707521,
|
|
"grad_norm": 0.6816435531080544,
|
|
"learning_rate": 8.642223044594873e-06,
|
|
"loss": 0.1632,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.023794259876012802,
|
|
"step": 3635,
|
|
"valid_targets_mean": 602.0,
|
|
"valid_targets_min": 152
|
|
},
|
|
{
|
|
"epoch": 5.069637883008356,
|
|
"grad_norm": 0.40935977827091236,
|
|
"learning_rate": 8.585120275321493e-06,
|
|
"loss": 0.1554,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08000779151916504,
|
|
"step": 3640,
|
|
"valid_targets_mean": 7299.9,
|
|
"valid_targets_min": 6158
|
|
},
|
|
{
|
|
"epoch": 5.076601671309192,
|
|
"grad_norm": 0.41147638157549343,
|
|
"learning_rate": 8.528155181807895e-06,
|
|
"loss": 0.1722,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08062512427568436,
|
|
"step": 3645,
|
|
"valid_targets_mean": 6140.0,
|
|
"valid_targets_min": 4648
|
|
},
|
|
{
|
|
"epoch": 5.0835654596100275,
|
|
"grad_norm": 0.44547329868476815,
|
|
"learning_rate": 8.471328451114545e-06,
|
|
"loss": 0.164,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08527650684118271,
|
|
"step": 3650,
|
|
"valid_targets_mean": 6603.0,
|
|
"valid_targets_min": 4882
|
|
},
|
|
{
|
|
"epoch": 5.0905292479108635,
|
|
"grad_norm": 0.40170578080263875,
|
|
"learning_rate": 8.414640768633096e-06,
|
|
"loss": 0.1656,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0836111456155777,
|
|
"step": 3655,
|
|
"valid_targets_mean": 8015.0,
|
|
"valid_targets_min": 6344
|
|
},
|
|
{
|
|
"epoch": 5.0974930362116995,
|
|
"grad_norm": 0.41169416175196893,
|
|
"learning_rate": 8.358092818078133e-06,
|
|
"loss": 0.1579,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07325682044029236,
|
|
"step": 3660,
|
|
"valid_targets_mean": 6305.2,
|
|
"valid_targets_min": 4622
|
|
},
|
|
{
|
|
"epoch": 5.104456824512535,
|
|
"grad_norm": 0.40731776029002104,
|
|
"learning_rate": 8.301685281478923e-06,
|
|
"loss": 0.1662,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.089007668197155,
|
|
"step": 3665,
|
|
"valid_targets_mean": 7890.8,
|
|
"valid_targets_min": 5485
|
|
},
|
|
{
|
|
"epoch": 5.111420612813371,
|
|
"grad_norm": 0.4776234320230322,
|
|
"learning_rate": 8.24541883917119e-06,
|
|
"loss": 0.1663,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07598386704921722,
|
|
"step": 3670,
|
|
"valid_targets_mean": 7016.6,
|
|
"valid_targets_min": 4960
|
|
},
|
|
{
|
|
"epoch": 5.118384401114207,
|
|
"grad_norm": 0.41863282757191406,
|
|
"learning_rate": 8.189294169788913e-06,
|
|
"loss": 0.1626,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07902969419956207,
|
|
"step": 3675,
|
|
"valid_targets_mean": 6527.1,
|
|
"valid_targets_min": 5399
|
|
},
|
|
{
|
|
"epoch": 5.125348189415042,
|
|
"grad_norm": 0.47406483603656724,
|
|
"learning_rate": 8.133311950256127e-06,
|
|
"loss": 0.1675,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08029402792453766,
|
|
"step": 3680,
|
|
"valid_targets_mean": 4995.0,
|
|
"valid_targets_min": 618
|
|
},
|
|
{
|
|
"epoch": 5.132311977715878,
|
|
"grad_norm": 0.4225431739550544,
|
|
"learning_rate": 8.077472855778774e-06,
|
|
"loss": 0.1814,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09067791700363159,
|
|
"step": 3685,
|
|
"valid_targets_mean": 7219.4,
|
|
"valid_targets_min": 5764
|
|
},
|
|
{
|
|
"epoch": 5.139275766016713,
|
|
"grad_norm": 0.4015096498481741,
|
|
"learning_rate": 8.021777559836554e-06,
|
|
"loss": 0.1461,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08319695293903351,
|
|
"step": 3690,
|
|
"valid_targets_mean": 7998.5,
|
|
"valid_targets_min": 5902
|
|
},
|
|
{
|
|
"epoch": 5.146239554317549,
|
|
"grad_norm": 0.38710682418396525,
|
|
"learning_rate": 7.966226734174799e-06,
|
|
"loss": 0.1538,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06552509218454361,
|
|
"step": 3695,
|
|
"valid_targets_mean": 7377.4,
|
|
"valid_targets_min": 5015
|
|
},
|
|
{
|
|
"epoch": 5.153203342618385,
|
|
"grad_norm": 0.3983605888889487,
|
|
"learning_rate": 7.910821048796368e-06,
|
|
"loss": 0.1619,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06865357607603073,
|
|
"step": 3700,
|
|
"valid_targets_mean": 6794.0,
|
|
"valid_targets_min": 5283
|
|
},
|
|
{
|
|
"epoch": 5.16016713091922,
|
|
"grad_norm": 0.44045948176479105,
|
|
"learning_rate": 7.855561171953594e-06,
|
|
"loss": 0.1685,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09348156303167343,
|
|
"step": 3705,
|
|
"valid_targets_mean": 6656.4,
|
|
"valid_targets_min": 4364
|
|
},
|
|
{
|
|
"epoch": 5.167130919220056,
|
|
"grad_norm": 0.43728173137315074,
|
|
"learning_rate": 7.800447770140185e-06,
|
|
"loss": 0.1677,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07885047048330307,
|
|
"step": 3710,
|
|
"valid_targets_mean": 7178.4,
|
|
"valid_targets_min": 4448
|
|
},
|
|
{
|
|
"epoch": 5.174094707520891,
|
|
"grad_norm": 0.4411940412213923,
|
|
"learning_rate": 7.74548150808319e-06,
|
|
"loss": 0.1653,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09826900064945221,
|
|
"step": 3715,
|
|
"valid_targets_mean": 7619.8,
|
|
"valid_targets_min": 5737
|
|
},
|
|
{
|
|
"epoch": 5.181058495821727,
|
|
"grad_norm": 0.3963098926707128,
|
|
"learning_rate": 7.690663048735009e-06,
|
|
"loss": 0.1456,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06886550784111023,
|
|
"step": 3720,
|
|
"valid_targets_mean": 8754.4,
|
|
"valid_targets_min": 5752
|
|
},
|
|
{
|
|
"epoch": 5.188022284122563,
|
|
"grad_norm": 0.4431165807502115,
|
|
"learning_rate": 7.635993053265391e-06,
|
|
"loss": 0.1476,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07675637304782867,
|
|
"step": 3725,
|
|
"valid_targets_mean": 6341.0,
|
|
"valid_targets_min": 4543
|
|
},
|
|
{
|
|
"epoch": 5.194986072423398,
|
|
"grad_norm": 0.4239858813074235,
|
|
"learning_rate": 7.581472181053435e-06,
|
|
"loss": 0.1432,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08128580451011658,
|
|
"step": 3730,
|
|
"valid_targets_mean": 7181.9,
|
|
"valid_targets_min": 4563
|
|
},
|
|
{
|
|
"epoch": 5.201949860724234,
|
|
"grad_norm": 0.7649649723811223,
|
|
"learning_rate": 7.527101089679658e-06,
|
|
"loss": 0.1665,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.051468294113874435,
|
|
"step": 3735,
|
|
"valid_targets_mean": 1104.4,
|
|
"valid_targets_min": 161
|
|
},
|
|
{
|
|
"epoch": 5.20891364902507,
|
|
"grad_norm": 0.41405913060333405,
|
|
"learning_rate": 7.4728804349180575e-06,
|
|
"loss": 0.1523,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07386703789234161,
|
|
"step": 3740,
|
|
"valid_targets_mean": 7018.2,
|
|
"valid_targets_min": 5424
|
|
},
|
|
{
|
|
"epoch": 5.215877437325905,
|
|
"grad_norm": 0.45338644264188654,
|
|
"learning_rate": 7.418810870728203e-06,
|
|
"loss": 0.172,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08742383122444153,
|
|
"step": 3745,
|
|
"valid_targets_mean": 7738.4,
|
|
"valid_targets_min": 5329
|
|
},
|
|
{
|
|
"epoch": 5.222841225626741,
|
|
"grad_norm": 0.4466036832294684,
|
|
"learning_rate": 7.364893049247351e-06,
|
|
"loss": 0.1758,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0829770565032959,
|
|
"step": 3750,
|
|
"valid_targets_mean": 6426.6,
|
|
"valid_targets_min": 5560
|
|
},
|
|
{
|
|
"epoch": 5.229805013927576,
|
|
"grad_norm": 0.48777116575319945,
|
|
"learning_rate": 7.311127620782575e-06,
|
|
"loss": 0.1702,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08484528213739395,
|
|
"step": 3755,
|
|
"valid_targets_mean": 6169.9,
|
|
"valid_targets_min": 3446
|
|
},
|
|
{
|
|
"epoch": 5.236768802228412,
|
|
"grad_norm": 0.4147145687544805,
|
|
"learning_rate": 7.2575152338029255e-06,
|
|
"loss": 0.1669,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08592289686203003,
|
|
"step": 3760,
|
|
"valid_targets_mean": 7204.6,
|
|
"valid_targets_min": 5381
|
|
},
|
|
{
|
|
"epoch": 5.243732590529248,
|
|
"grad_norm": 0.4338983913748669,
|
|
"learning_rate": 7.204056534931607e-06,
|
|
"loss": 0.1689,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08771762251853943,
|
|
"step": 3765,
|
|
"valid_targets_mean": 7600.4,
|
|
"valid_targets_min": 5241
|
|
},
|
|
{
|
|
"epoch": 5.250696378830083,
|
|
"grad_norm": 0.41445196900515596,
|
|
"learning_rate": 7.150752168938182e-06,
|
|
"loss": 0.1649,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08908963948488235,
|
|
"step": 3770,
|
|
"valid_targets_mean": 7098.5,
|
|
"valid_targets_min": 5607
|
|
},
|
|
{
|
|
"epoch": 5.257660167130919,
|
|
"grad_norm": 0.463052253966403,
|
|
"learning_rate": 7.097602778730788e-06,
|
|
"loss": 0.1687,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08907216787338257,
|
|
"step": 3775,
|
|
"valid_targets_mean": 6955.1,
|
|
"valid_targets_min": 5530
|
|
},
|
|
{
|
|
"epoch": 5.264623955431755,
|
|
"grad_norm": 0.4364902349015762,
|
|
"learning_rate": 7.044609005348389e-06,
|
|
"loss": 0.1681,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08470651507377625,
|
|
"step": 3780,
|
|
"valid_targets_mean": 6748.1,
|
|
"valid_targets_min": 5364
|
|
},
|
|
{
|
|
"epoch": 5.27158774373259,
|
|
"grad_norm": 0.4667194728842975,
|
|
"learning_rate": 6.991771487953036e-06,
|
|
"loss": 0.1681,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08043470978736877,
|
|
"step": 3785,
|
|
"valid_targets_mean": 6273.9,
|
|
"valid_targets_min": 5438
|
|
},
|
|
{
|
|
"epoch": 5.278551532033426,
|
|
"grad_norm": 0.42521567334703014,
|
|
"learning_rate": 6.939090863822188e-06,
|
|
"loss": 0.1654,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07761454582214355,
|
|
"step": 3790,
|
|
"valid_targets_mean": 6419.5,
|
|
"valid_targets_min": 4879
|
|
},
|
|
{
|
|
"epoch": 5.2855153203342615,
|
|
"grad_norm": 0.43932142773608807,
|
|
"learning_rate": 6.8865677683409835e-06,
|
|
"loss": 0.1648,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07779448479413986,
|
|
"step": 3795,
|
|
"valid_targets_mean": 6483.4,
|
|
"valid_targets_min": 4309
|
|
},
|
|
{
|
|
"epoch": 5.2924791086350975,
|
|
"grad_norm": 0.47079517415735234,
|
|
"learning_rate": 6.834202834994581e-06,
|
|
"loss": 0.1334,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09646376967430115,
|
|
"step": 3800,
|
|
"valid_targets_mean": 6825.0,
|
|
"valid_targets_min": 5205
|
|
},
|
|
{
|
|
"epoch": 5.2994428969359335,
|
|
"grad_norm": 0.46289740146196456,
|
|
"learning_rate": 6.781996695360553e-06,
|
|
"loss": 0.1652,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07646790146827698,
|
|
"step": 3805,
|
|
"valid_targets_mean": 6729.4,
|
|
"valid_targets_min": 5200
|
|
},
|
|
{
|
|
"epoch": 5.306406685236769,
|
|
"grad_norm": 0.44246422147197345,
|
|
"learning_rate": 6.729949979101256e-06,
|
|
"loss": 0.1619,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07880687713623047,
|
|
"step": 3810,
|
|
"valid_targets_mean": 6094.0,
|
|
"valid_targets_min": 5302
|
|
},
|
|
{
|
|
"epoch": 5.313370473537605,
|
|
"grad_norm": 0.4386645916488665,
|
|
"learning_rate": 6.678063313956211e-06,
|
|
"loss": 0.1659,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0853356197476387,
|
|
"step": 3815,
|
|
"valid_targets_mean": 7140.0,
|
|
"valid_targets_min": 5270
|
|
},
|
|
{
|
|
"epoch": 5.32033426183844,
|
|
"grad_norm": 0.3896192941415339,
|
|
"learning_rate": 6.626337325734564e-06,
|
|
"loss": 0.159,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07671719789505005,
|
|
"step": 3820,
|
|
"valid_targets_mean": 7425.0,
|
|
"valid_targets_min": 5281
|
|
},
|
|
{
|
|
"epoch": 5.327298050139276,
|
|
"grad_norm": 0.43882210400710403,
|
|
"learning_rate": 6.574772638307496e-06,
|
|
"loss": 0.1657,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10439842939376831,
|
|
"step": 3825,
|
|
"valid_targets_mean": 7063.4,
|
|
"valid_targets_min": 5386
|
|
},
|
|
{
|
|
"epoch": 5.334261838440112,
|
|
"grad_norm": 0.4328294638391199,
|
|
"learning_rate": 6.523369873600767e-06,
|
|
"loss": 0.1672,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08555565774440765,
|
|
"step": 3830,
|
|
"valid_targets_mean": 6827.0,
|
|
"valid_targets_min": 4547
|
|
},
|
|
{
|
|
"epoch": 5.341225626740947,
|
|
"grad_norm": 0.5740969553533752,
|
|
"learning_rate": 6.472129651587147e-06,
|
|
"loss": 0.1421,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.03779056295752525,
|
|
"step": 3835,
|
|
"valid_targets_mean": 2047.0,
|
|
"valid_targets_min": 165
|
|
},
|
|
{
|
|
"epoch": 5.348189415041783,
|
|
"grad_norm": 0.4535247047853864,
|
|
"learning_rate": 6.421052590278976e-06,
|
|
"loss": 0.1466,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07807661592960358,
|
|
"step": 3840,
|
|
"valid_targets_mean": 6754.1,
|
|
"valid_targets_min": 5196
|
|
},
|
|
{
|
|
"epoch": 5.355153203342619,
|
|
"grad_norm": 0.4691124981565333,
|
|
"learning_rate": 6.370139305720693e-06,
|
|
"loss": 0.1595,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07937394082546234,
|
|
"step": 3845,
|
|
"valid_targets_mean": 5985.6,
|
|
"valid_targets_min": 5028
|
|
},
|
|
{
|
|
"epoch": 5.362116991643454,
|
|
"grad_norm": 0.4621292650890354,
|
|
"learning_rate": 6.319390411981421e-06,
|
|
"loss": 0.1648,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07576687633991241,
|
|
"step": 3850,
|
|
"valid_targets_mean": 5963.9,
|
|
"valid_targets_min": 5237
|
|
},
|
|
{
|
|
"epoch": 5.36908077994429,
|
|
"grad_norm": 0.47453529144859713,
|
|
"learning_rate": 6.268806521147546e-06,
|
|
"loss": 0.161,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08445766568183899,
|
|
"step": 3855,
|
|
"valid_targets_mean": 6319.0,
|
|
"valid_targets_min": 5412
|
|
},
|
|
{
|
|
"epoch": 5.376044568245125,
|
|
"grad_norm": 0.41147646717777764,
|
|
"learning_rate": 6.218388243315343e-06,
|
|
"loss": 0.1541,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06716381013393402,
|
|
"step": 3860,
|
|
"valid_targets_mean": 7125.8,
|
|
"valid_targets_min": 5080
|
|
},
|
|
{
|
|
"epoch": 5.383008356545961,
|
|
"grad_norm": 0.43779798115412527,
|
|
"learning_rate": 6.168136186583615e-06,
|
|
"loss": 0.1621,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07958684861660004,
|
|
"step": 3865,
|
|
"valid_targets_mean": 6123.5,
|
|
"valid_targets_min": 4521
|
|
},
|
|
{
|
|
"epoch": 5.389972144846797,
|
|
"grad_norm": 0.966196621555397,
|
|
"learning_rate": 6.118050957046358e-06,
|
|
"loss": 0.1589,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09092744439840317,
|
|
"step": 3870,
|
|
"valid_targets_mean": 1886.0,
|
|
"valid_targets_min": 1111
|
|
},
|
|
{
|
|
"epoch": 5.396935933147632,
|
|
"grad_norm": 0.8077855185739582,
|
|
"learning_rate": 6.068133158785463e-06,
|
|
"loss": 0.1481,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07527165114879608,
|
|
"step": 3875,
|
|
"valid_targets_mean": 1497.6,
|
|
"valid_targets_min": 893
|
|
},
|
|
{
|
|
"epoch": 5.403899721448468,
|
|
"grad_norm": 0.7144983965195878,
|
|
"learning_rate": 6.018383393863412e-06,
|
|
"loss": 0.1349,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.049098506569862366,
|
|
"step": 3880,
|
|
"valid_targets_mean": 1356.0,
|
|
"valid_targets_min": 914
|
|
},
|
|
{
|
|
"epoch": 5.410863509749303,
|
|
"grad_norm": 0.8109690813075966,
|
|
"learning_rate": 5.968802262316009e-06,
|
|
"loss": 0.1312,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.054713502526283264,
|
|
"step": 3885,
|
|
"valid_targets_mean": 1324.5,
|
|
"valid_targets_min": 680
|
|
},
|
|
{
|
|
"epoch": 5.417827298050139,
|
|
"grad_norm": 0.8096672959801836,
|
|
"learning_rate": 5.919390362145172e-06,
|
|
"loss": 0.1336,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06651387363672256,
|
|
"step": 3890,
|
|
"valid_targets_mean": 1931.8,
|
|
"valid_targets_min": 876
|
|
},
|
|
{
|
|
"epoch": 5.424791086350975,
|
|
"grad_norm": 0.9084117209145378,
|
|
"learning_rate": 5.870148289311719e-06,
|
|
"loss": 0.1281,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05635681003332138,
|
|
"step": 3895,
|
|
"valid_targets_mean": 1292.2,
|
|
"valid_targets_min": 734
|
|
},
|
|
{
|
|
"epoch": 5.43175487465181,
|
|
"grad_norm": 0.8234461561242756,
|
|
"learning_rate": 5.8210766377281425e-06,
|
|
"loss": 0.1376,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05349376052618027,
|
|
"step": 3900,
|
|
"valid_targets_mean": 1469.6,
|
|
"valid_targets_min": 655
|
|
},
|
|
{
|
|
"epoch": 5.438718662952646,
|
|
"grad_norm": 0.8349797022317297,
|
|
"learning_rate": 5.7721759992514835e-06,
|
|
"loss": 0.1165,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06947162747383118,
|
|
"step": 3905,
|
|
"valid_targets_mean": 1548.4,
|
|
"valid_targets_min": 737
|
|
},
|
|
{
|
|
"epoch": 5.445682451253482,
|
|
"grad_norm": 0.8357949779321237,
|
|
"learning_rate": 5.7234469636761825e-06,
|
|
"loss": 0.1376,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05619405210018158,
|
|
"step": 3910,
|
|
"valid_targets_mean": 1316.5,
|
|
"valid_targets_min": 759
|
|
},
|
|
{
|
|
"epoch": 5.452646239554317,
|
|
"grad_norm": 0.8320721049373022,
|
|
"learning_rate": 5.674890118726955e-06,
|
|
"loss": 0.125,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.048418521881103516,
|
|
"step": 3915,
|
|
"valid_targets_mean": 1249.6,
|
|
"valid_targets_min": 753
|
|
},
|
|
{
|
|
"epoch": 5.459610027855153,
|
|
"grad_norm": 0.8918073241547342,
|
|
"learning_rate": 5.626506050051716e-06,
|
|
"loss": 0.126,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07567094266414642,
|
|
"step": 3920,
|
|
"valid_targets_mean": 1833.0,
|
|
"valid_targets_min": 927
|
|
},
|
|
{
|
|
"epoch": 5.4665738161559885,
|
|
"grad_norm": 0.864812696220783,
|
|
"learning_rate": 5.578295341214519e-06,
|
|
"loss": 0.1259,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06043354421854019,
|
|
"step": 3925,
|
|
"valid_targets_mean": 1338.4,
|
|
"valid_targets_min": 679
|
|
},
|
|
{
|
|
"epoch": 5.4735376044568245,
|
|
"grad_norm": 0.854189693756131,
|
|
"learning_rate": 5.530258573688496e-06,
|
|
"loss": 0.124,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05904326215386391,
|
|
"step": 3930,
|
|
"valid_targets_mean": 1244.6,
|
|
"valid_targets_min": 771
|
|
},
|
|
{
|
|
"epoch": 5.4805013927576605,
|
|
"grad_norm": 0.7843808631480862,
|
|
"learning_rate": 5.482396326848884e-06,
|
|
"loss": 0.1125,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.058021124452352524,
|
|
"step": 3935,
|
|
"valid_targets_mean": 1732.5,
|
|
"valid_targets_min": 986
|
|
},
|
|
{
|
|
"epoch": 5.487465181058496,
|
|
"grad_norm": 0.9558429957238452,
|
|
"learning_rate": 5.434709177965982e-06,
|
|
"loss": 0.1177,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05305638536810875,
|
|
"step": 3940,
|
|
"valid_targets_mean": 1389.9,
|
|
"valid_targets_min": 1103
|
|
},
|
|
{
|
|
"epoch": 5.494428969359332,
|
|
"grad_norm": 0.8557141687664696,
|
|
"learning_rate": 5.387197702198239e-06,
|
|
"loss": 0.1195,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0624840185046196,
|
|
"step": 3945,
|
|
"valid_targets_mean": 1593.2,
|
|
"valid_targets_min": 675
|
|
},
|
|
{
|
|
"epoch": 5.501392757660167,
|
|
"grad_norm": 0.8442709657389822,
|
|
"learning_rate": 5.339862472585284e-06,
|
|
"loss": 0.1221,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07162931561470032,
|
|
"step": 3950,
|
|
"valid_targets_mean": 1926.6,
|
|
"valid_targets_min": 974
|
|
},
|
|
{
|
|
"epoch": 5.508356545961003,
|
|
"grad_norm": 0.8922738009156526,
|
|
"learning_rate": 5.292704060041047e-06,
|
|
"loss": 0.1228,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07262811809778214,
|
|
"step": 3955,
|
|
"valid_targets_mean": 1953.1,
|
|
"valid_targets_min": 869
|
|
},
|
|
{
|
|
"epoch": 5.515320334261839,
|
|
"grad_norm": 0.8628907763362105,
|
|
"learning_rate": 5.245723033346841e-06,
|
|
"loss": 0.1226,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05248372256755829,
|
|
"step": 3960,
|
|
"valid_targets_mean": 1327.2,
|
|
"valid_targets_min": 1023
|
|
},
|
|
{
|
|
"epoch": 5.522284122562674,
|
|
"grad_norm": 1.0689981634779093,
|
|
"learning_rate": 5.198919959144504e-06,
|
|
"loss": 0.1263,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06385475397109985,
|
|
"step": 3965,
|
|
"valid_targets_mean": 1717.9,
|
|
"valid_targets_min": 709
|
|
},
|
|
{
|
|
"epoch": 5.52924791086351,
|
|
"grad_norm": 0.8507547339269755,
|
|
"learning_rate": 5.152295401929581e-06,
|
|
"loss": 0.1222,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04907778650522232,
|
|
"step": 3970,
|
|
"valid_targets_mean": 1139.9,
|
|
"valid_targets_min": 658
|
|
},
|
|
{
|
|
"epoch": 5.536211699164346,
|
|
"grad_norm": 0.8352935218223749,
|
|
"learning_rate": 5.10584992404453e-06,
|
|
"loss": 0.1204,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05608684569597244,
|
|
"step": 3975,
|
|
"valid_targets_mean": 1444.8,
|
|
"valid_targets_min": 662
|
|
},
|
|
{
|
|
"epoch": 5.543175487465181,
|
|
"grad_norm": 0.8360060925828016,
|
|
"learning_rate": 5.059584085671896e-06,
|
|
"loss": 0.1211,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04403482377529144,
|
|
"step": 3980,
|
|
"valid_targets_mean": 1242.2,
|
|
"valid_targets_min": 859
|
|
},
|
|
{
|
|
"epoch": 5.550139275766017,
|
|
"grad_norm": 0.8557031879654616,
|
|
"learning_rate": 5.0134984448275844e-06,
|
|
"loss": 0.11,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06525018811225891,
|
|
"step": 3985,
|
|
"valid_targets_mean": 1569.2,
|
|
"valid_targets_min": 950
|
|
},
|
|
{
|
|
"epoch": 5.557103064066853,
|
|
"grad_norm": 0.8940987834631933,
|
|
"learning_rate": 4.967593557354125e-06,
|
|
"loss": 0.1175,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.057968899607658386,
|
|
"step": 3990,
|
|
"valid_targets_mean": 1539.8,
|
|
"valid_targets_min": 686
|
|
},
|
|
{
|
|
"epoch": 5.564066852367688,
|
|
"grad_norm": 0.8803812608953893,
|
|
"learning_rate": 4.921869976913964e-06,
|
|
"loss": 0.1187,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06881322711706161,
|
|
"step": 3995,
|
|
"valid_targets_mean": 1610.6,
|
|
"valid_targets_min": 702
|
|
},
|
|
{
|
|
"epoch": 5.571030640668524,
|
|
"grad_norm": 0.8857750625346741,
|
|
"learning_rate": 4.876328254982796e-06,
|
|
"loss": 0.1188,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.043028295040130615,
|
|
"step": 4000,
|
|
"valid_targets_mean": 1190.4,
|
|
"valid_targets_min": 779
|
|
},
|
|
{
|
|
"epoch": 5.577994428969359,
|
|
"grad_norm": 0.8537850227769329,
|
|
"learning_rate": 4.8309689408429015e-06,
|
|
"loss": 0.1287,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06296706199645996,
|
|
"step": 4005,
|
|
"valid_targets_mean": 1467.1,
|
|
"valid_targets_min": 822
|
|
},
|
|
{
|
|
"epoch": 5.584958217270195,
|
|
"grad_norm": 0.8285166830028964,
|
|
"learning_rate": 4.785792581576523e-06,
|
|
"loss": 0.1172,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05445345491170883,
|
|
"step": 4010,
|
|
"valid_targets_mean": 1350.5,
|
|
"valid_targets_min": 868
|
|
},
|
|
{
|
|
"epoch": 5.591922005571031,
|
|
"grad_norm": 0.8667031190956394,
|
|
"learning_rate": 4.740799722059269e-06,
|
|
"loss": 0.1196,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04850904271006584,
|
|
"step": 4015,
|
|
"valid_targets_mean": 1283.5,
|
|
"valid_targets_min": 781
|
|
},
|
|
{
|
|
"epoch": 5.598885793871866,
|
|
"grad_norm": 0.8366685664142771,
|
|
"learning_rate": 4.6959909049535665e-06,
|
|
"loss": 0.1198,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05181458219885826,
|
|
"step": 4020,
|
|
"valid_targets_mean": 1303.2,
|
|
"valid_targets_min": 793
|
|
},
|
|
{
|
|
"epoch": 5.605849582172702,
|
|
"grad_norm": 0.7495096178648898,
|
|
"learning_rate": 4.651366670702054e-06,
|
|
"loss": 0.1211,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.047033414244651794,
|
|
"step": 4025,
|
|
"valid_targets_mean": 1309.1,
|
|
"valid_targets_min": 806
|
|
},
|
|
{
|
|
"epoch": 5.612813370473537,
|
|
"grad_norm": 0.811123052386617,
|
|
"learning_rate": 4.606927557521128e-06,
|
|
"loss": 0.117,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04835967719554901,
|
|
"step": 4030,
|
|
"valid_targets_mean": 1397.6,
|
|
"valid_targets_min": 803
|
|
},
|
|
{
|
|
"epoch": 5.619777158774373,
|
|
"grad_norm": 0.851744557513053,
|
|
"learning_rate": 4.562674101394413e-06,
|
|
"loss": 0.1157,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04473861679434776,
|
|
"step": 4035,
|
|
"valid_targets_mean": 1187.8,
|
|
"valid_targets_min": 660
|
|
},
|
|
{
|
|
"epoch": 5.626740947075209,
|
|
"grad_norm": 0.8083228291949228,
|
|
"learning_rate": 4.518606836066328e-06,
|
|
"loss": 0.107,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05075952038168907,
|
|
"step": 4040,
|
|
"valid_targets_mean": 1547.0,
|
|
"valid_targets_min": 853
|
|
},
|
|
{
|
|
"epoch": 5.633704735376044,
|
|
"grad_norm": 0.924317112722506,
|
|
"learning_rate": 4.474726293035614e-06,
|
|
"loss": 0.1172,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0637693852186203,
|
|
"step": 4045,
|
|
"valid_targets_mean": 1704.2,
|
|
"valid_targets_min": 719
|
|
},
|
|
{
|
|
"epoch": 5.64066852367688,
|
|
"grad_norm": 0.8765664950505369,
|
|
"learning_rate": 4.43103300154893e-06,
|
|
"loss": 0.1161,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.044058702886104584,
|
|
"step": 4050,
|
|
"valid_targets_mean": 1255.0,
|
|
"valid_targets_min": 765
|
|
},
|
|
{
|
|
"epoch": 5.647632311977716,
|
|
"grad_norm": 0.8847967785711801,
|
|
"learning_rate": 4.3875274885944965e-06,
|
|
"loss": 0.1184,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06218354403972626,
|
|
"step": 4055,
|
|
"valid_targets_mean": 1550.8,
|
|
"valid_targets_min": 875
|
|
},
|
|
{
|
|
"epoch": 5.654596100278551,
|
|
"grad_norm": 0.8774671561905153,
|
|
"learning_rate": 4.344210278895729e-06,
|
|
"loss": 0.1121,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0588260255753994,
|
|
"step": 4060,
|
|
"valid_targets_mean": 1362.0,
|
|
"valid_targets_min": 739
|
|
},
|
|
{
|
|
"epoch": 5.661559888579387,
|
|
"grad_norm": 0.8793053210734343,
|
|
"learning_rate": 4.301081894904882e-06,
|
|
"loss": 0.117,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06402565538883209,
|
|
"step": 4065,
|
|
"valid_targets_mean": 1666.8,
|
|
"valid_targets_min": 760
|
|
},
|
|
{
|
|
"epoch": 5.6685236768802225,
|
|
"grad_norm": 0.8736445128719057,
|
|
"learning_rate": 4.258142856796779e-06,
|
|
"loss": 0.111,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04810625687241554,
|
|
"step": 4070,
|
|
"valid_targets_mean": 1319.6,
|
|
"valid_targets_min": 1046
|
|
},
|
|
{
|
|
"epoch": 5.6754874651810585,
|
|
"grad_norm": 0.9146961523812506,
|
|
"learning_rate": 4.21539368246253e-06,
|
|
"loss": 0.1235,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06791988015174866,
|
|
"step": 4075,
|
|
"valid_targets_mean": 1727.9,
|
|
"valid_targets_min": 851
|
|
},
|
|
{
|
|
"epoch": 5.6824512534818945,
|
|
"grad_norm": 0.8950204260314059,
|
|
"learning_rate": 4.172834887503277e-06,
|
|
"loss": 0.1226,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06217281520366669,
|
|
"step": 4080,
|
|
"valid_targets_mean": 1643.2,
|
|
"valid_targets_min": 907
|
|
},
|
|
{
|
|
"epoch": 5.68941504178273,
|
|
"grad_norm": 0.8625659343005182,
|
|
"learning_rate": 4.1304669852239865e-06,
|
|
"loss": 0.1242,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.052721407264471054,
|
|
"step": 4085,
|
|
"valid_targets_mean": 1478.5,
|
|
"valid_targets_min": 764
|
|
},
|
|
{
|
|
"epoch": 5.696378830083566,
|
|
"grad_norm": 0.8423037084637137,
|
|
"learning_rate": 4.088290486627249e-06,
|
|
"loss": 0.1117,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.051760852336883545,
|
|
"step": 4090,
|
|
"valid_targets_mean": 1522.6,
|
|
"valid_targets_min": 822
|
|
},
|
|
{
|
|
"epoch": 5.703342618384401,
|
|
"grad_norm": 0.7986133696863431,
|
|
"learning_rate": 4.046305900407123e-06,
|
|
"loss": 0.1089,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05744687840342522,
|
|
"step": 4095,
|
|
"valid_targets_mean": 1589.0,
|
|
"valid_targets_min": 982
|
|
},
|
|
{
|
|
"epoch": 5.710306406685237,
|
|
"grad_norm": 0.8247019511544061,
|
|
"learning_rate": 4.004513732943001e-06,
|
|
"loss": 0.1217,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05563589930534363,
|
|
"step": 4100,
|
|
"valid_targets_mean": 1574.8,
|
|
"valid_targets_min": 816
|
|
},
|
|
{
|
|
"epoch": 5.717270194986073,
|
|
"grad_norm": 0.8220162581325487,
|
|
"learning_rate": 3.962914488293487e-06,
|
|
"loss": 0.1119,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05360320955514908,
|
|
"step": 4105,
|
|
"valid_targets_mean": 1422.0,
|
|
"valid_targets_min": 721
|
|
},
|
|
{
|
|
"epoch": 5.724233983286908,
|
|
"grad_norm": 0.8640387086170787,
|
|
"learning_rate": 3.921508668190337e-06,
|
|
"loss": 0.1102,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04421108961105347,
|
|
"step": 4110,
|
|
"valid_targets_mean": 1150.6,
|
|
"valid_targets_min": 741
|
|
},
|
|
{
|
|
"epoch": 5.731197771587744,
|
|
"grad_norm": 0.9963080986309333,
|
|
"learning_rate": 3.8802967720323944e-06,
|
|
"loss": 0.116,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0667385682463646,
|
|
"step": 4115,
|
|
"valid_targets_mean": 1684.2,
|
|
"valid_targets_min": 734
|
|
},
|
|
{
|
|
"epoch": 5.73816155988858,
|
|
"grad_norm": 0.9033327920465504,
|
|
"learning_rate": 3.839279296879566e-06,
|
|
"loss": 0.122,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07596944272518158,
|
|
"step": 4120,
|
|
"valid_targets_mean": 1937.1,
|
|
"valid_targets_min": 744
|
|
},
|
|
{
|
|
"epoch": 5.745125348189415,
|
|
"grad_norm": 0.9201337682269749,
|
|
"learning_rate": 3.7984567374468495e-06,
|
|
"loss": 0.1182,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06649009138345718,
|
|
"step": 4125,
|
|
"valid_targets_mean": 1842.2,
|
|
"valid_targets_min": 620
|
|
},
|
|
{
|
|
"epoch": 5.752089136490251,
|
|
"grad_norm": 0.8638321399247245,
|
|
"learning_rate": 3.757829586098334e-06,
|
|
"loss": 0.1111,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06432803720235825,
|
|
"step": 4130,
|
|
"valid_targets_mean": 1637.6,
|
|
"valid_targets_min": 785
|
|
},
|
|
{
|
|
"epoch": 5.759052924791086,
|
|
"grad_norm": 0.8229387884326227,
|
|
"learning_rate": 3.717398332841269e-06,
|
|
"loss": 0.1196,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06425435841083527,
|
|
"step": 4135,
|
|
"valid_targets_mean": 1969.9,
|
|
"valid_targets_min": 740
|
|
},
|
|
{
|
|
"epoch": 5.766016713091922,
|
|
"grad_norm": 0.9669615015698776,
|
|
"learning_rate": 3.6771634653201726e-06,
|
|
"loss": 0.1116,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.054495103657245636,
|
|
"step": 4140,
|
|
"valid_targets_mean": 1432.1,
|
|
"valid_targets_min": 711
|
|
},
|
|
{
|
|
"epoch": 5.772980501392758,
|
|
"grad_norm": 0.7975239579352195,
|
|
"learning_rate": 3.6371254688109493e-06,
|
|
"loss": 0.1163,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06108352169394493,
|
|
"step": 4145,
|
|
"valid_targets_mean": 1695.6,
|
|
"valid_targets_min": 1122
|
|
},
|
|
{
|
|
"epoch": 5.779944289693593,
|
|
"grad_norm": 0.8324320935247468,
|
|
"learning_rate": 3.5972848262150084e-06,
|
|
"loss": 0.1145,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06076567620038986,
|
|
"step": 4150,
|
|
"valid_targets_mean": 1524.5,
|
|
"valid_targets_min": 722
|
|
},
|
|
{
|
|
"epoch": 5.786908077994429,
|
|
"grad_norm": 0.8295223085601157,
|
|
"learning_rate": 3.5576420180534666e-06,
|
|
"loss": 0.1109,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.052539028227329254,
|
|
"step": 4155,
|
|
"valid_targets_mean": 1289.2,
|
|
"valid_targets_min": 661
|
|
},
|
|
{
|
|
"epoch": 5.793871866295264,
|
|
"grad_norm": 0.8688174214233373,
|
|
"learning_rate": 3.518197522461344e-06,
|
|
"loss": 0.1093,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05658646672964096,
|
|
"step": 4160,
|
|
"valid_targets_mean": 1519.5,
|
|
"valid_targets_min": 1011
|
|
},
|
|
{
|
|
"epoch": 5.8008356545961,
|
|
"grad_norm": 0.8560277889635767,
|
|
"learning_rate": 3.4789518151817926e-06,
|
|
"loss": 0.1182,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0627206489443779,
|
|
"step": 4165,
|
|
"valid_targets_mean": 1731.4,
|
|
"valid_targets_min": 1277
|
|
},
|
|
{
|
|
"epoch": 5.807799442896936,
|
|
"grad_norm": 0.8221705464226036,
|
|
"learning_rate": 3.43990536956037e-06,
|
|
"loss": 0.1132,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05920272320508957,
|
|
"step": 4170,
|
|
"valid_targets_mean": 1486.1,
|
|
"valid_targets_min": 807
|
|
},
|
|
{
|
|
"epoch": 5.814763231197771,
|
|
"grad_norm": 0.8332753209312879,
|
|
"learning_rate": 3.4010586565393176e-06,
|
|
"loss": 0.1139,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.057332780212163925,
|
|
"step": 4175,
|
|
"valid_targets_mean": 1549.9,
|
|
"valid_targets_min": 679
|
|
},
|
|
{
|
|
"epoch": 5.821727019498607,
|
|
"grad_norm": 0.8619887271447355,
|
|
"learning_rate": 3.36241214465189e-06,
|
|
"loss": 0.116,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05862189084291458,
|
|
"step": 4180,
|
|
"valid_targets_mean": 1656.0,
|
|
"valid_targets_min": 1294
|
|
},
|
|
{
|
|
"epoch": 5.828690807799443,
|
|
"grad_norm": 0.8103161574226729,
|
|
"learning_rate": 3.323966300016699e-06,
|
|
"loss": 0.1199,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.053418777883052826,
|
|
"step": 4185,
|
|
"valid_targets_mean": 1368.8,
|
|
"valid_targets_min": 790
|
|
},
|
|
{
|
|
"epoch": 5.835654596100278,
|
|
"grad_norm": 0.8465138226795567,
|
|
"learning_rate": 3.285721586332089e-06,
|
|
"loss": 0.1161,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06450681388378143,
|
|
"step": 4190,
|
|
"valid_targets_mean": 1765.8,
|
|
"valid_targets_min": 592
|
|
},
|
|
{
|
|
"epoch": 5.842618384401114,
|
|
"grad_norm": 0.8341389356508091,
|
|
"learning_rate": 3.247678464870556e-06,
|
|
"loss": 0.1161,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0624236986041069,
|
|
"step": 4195,
|
|
"valid_targets_mean": 1586.8,
|
|
"valid_targets_min": 662
|
|
},
|
|
{
|
|
"epoch": 5.84958217270195,
|
|
"grad_norm": 0.9648828608657739,
|
|
"learning_rate": 3.2098373944731696e-06,
|
|
"loss": 0.114,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05213096737861633,
|
|
"step": 4200,
|
|
"valid_targets_mean": 1336.9,
|
|
"valid_targets_min": 622
|
|
},
|
|
{
|
|
"epoch": 5.8565459610027855,
|
|
"grad_norm": 0.8693836846048845,
|
|
"learning_rate": 3.1721988315440467e-06,
|
|
"loss": 0.113,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.03882891684770584,
|
|
"step": 4205,
|
|
"valid_targets_mean": 1097.1,
|
|
"valid_targets_min": 625
|
|
},
|
|
{
|
|
"epoch": 5.8635097493036215,
|
|
"grad_norm": 0.8889935458275575,
|
|
"learning_rate": 3.1347632300448528e-06,
|
|
"loss": 0.1223,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05130942910909653,
|
|
"step": 4210,
|
|
"valid_targets_mean": 1484.9,
|
|
"valid_targets_min": 888
|
|
},
|
|
{
|
|
"epoch": 5.870473537604457,
|
|
"grad_norm": 0.8950418860834436,
|
|
"learning_rate": 3.0975310414893147e-06,
|
|
"loss": 0.1216,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06308726966381073,
|
|
"step": 4215,
|
|
"valid_targets_mean": 1757.0,
|
|
"valid_targets_min": 635
|
|
},
|
|
{
|
|
"epoch": 5.8774373259052926,
|
|
"grad_norm": 0.9448517682773246,
|
|
"learning_rate": 3.060502714937772e-06,
|
|
"loss": 0.116,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06284761428833008,
|
|
"step": 4220,
|
|
"valid_targets_mean": 1476.1,
|
|
"valid_targets_min": 522
|
|
},
|
|
{
|
|
"epoch": 5.884401114206128,
|
|
"grad_norm": 0.8513582848638479,
|
|
"learning_rate": 3.023678696991774e-06,
|
|
"loss": 0.1155,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04917547106742859,
|
|
"step": 4225,
|
|
"valid_targets_mean": 1230.9,
|
|
"valid_targets_min": 697
|
|
},
|
|
{
|
|
"epoch": 5.891364902506964,
|
|
"grad_norm": 0.8380241944756079,
|
|
"learning_rate": 2.9870594317887012e-06,
|
|
"loss": 0.109,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.060824502259492874,
|
|
"step": 4230,
|
|
"valid_targets_mean": 1671.0,
|
|
"valid_targets_min": 1010
|
|
},
|
|
{
|
|
"epoch": 5.8983286908078,
|
|
"grad_norm": 0.9611502718725131,
|
|
"learning_rate": 2.950645360996378e-06,
|
|
"loss": 0.111,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05260036140680313,
|
|
"step": 4235,
|
|
"valid_targets_mean": 1411.0,
|
|
"valid_targets_min": 769
|
|
},
|
|
{
|
|
"epoch": 5.905292479108635,
|
|
"grad_norm": 0.8900339205155612,
|
|
"learning_rate": 2.914436923807773e-06,
|
|
"loss": 0.115,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04677753150463104,
|
|
"step": 4240,
|
|
"valid_targets_mean": 1298.6,
|
|
"valid_targets_min": 745
|
|
},
|
|
{
|
|
"epoch": 5.912256267409471,
|
|
"grad_norm": 0.8870011470624816,
|
|
"learning_rate": 2.878434556935672e-06,
|
|
"loss": 0.1108,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.042783405631780624,
|
|
"step": 4245,
|
|
"valid_targets_mean": 1118.9,
|
|
"valid_targets_min": 680
|
|
},
|
|
{
|
|
"epoch": 5.919220055710307,
|
|
"grad_norm": 0.8118186588287939,
|
|
"learning_rate": 2.8426386946074625e-06,
|
|
"loss": 0.1112,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07259643822908401,
|
|
"step": 4250,
|
|
"valid_targets_mean": 1905.9,
|
|
"valid_targets_min": 1347
|
|
},
|
|
{
|
|
"epoch": 5.926183844011142,
|
|
"grad_norm": 0.9846435636971836,
|
|
"learning_rate": 2.807049768559844e-06,
|
|
"loss": 0.1197,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08297774195671082,
|
|
"step": 4255,
|
|
"valid_targets_mean": 1636.0,
|
|
"valid_targets_min": 851
|
|
},
|
|
{
|
|
"epoch": 5.933147632311978,
|
|
"grad_norm": 0.89306065477576,
|
|
"learning_rate": 2.771668208033642e-06,
|
|
"loss": 0.1139,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05090225487947464,
|
|
"step": 4260,
|
|
"valid_targets_mean": 1457.2,
|
|
"valid_targets_min": 903
|
|
},
|
|
{
|
|
"epoch": 5.940111420612814,
|
|
"grad_norm": 0.8381846762533809,
|
|
"learning_rate": 2.7364944397686356e-06,
|
|
"loss": 0.1146,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05966182425618172,
|
|
"step": 4265,
|
|
"valid_targets_mean": 1674.4,
|
|
"valid_targets_min": 1240
|
|
},
|
|
{
|
|
"epoch": 5.947075208913649,
|
|
"grad_norm": 0.6975382895103197,
|
|
"learning_rate": 2.7015288879984013e-06,
|
|
"loss": 0.1174,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05049895867705345,
|
|
"step": 4270,
|
|
"valid_targets_mean": 1697.1,
|
|
"valid_targets_min": 917
|
|
},
|
|
{
|
|
"epoch": 5.954038997214485,
|
|
"grad_norm": 0.8183978547191294,
|
|
"learning_rate": 2.6667719744452037e-06,
|
|
"loss": 0.1058,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05475775897502899,
|
|
"step": 4275,
|
|
"valid_targets_mean": 1580.9,
|
|
"valid_targets_min": 930
|
|
},
|
|
{
|
|
"epoch": 5.96100278551532,
|
|
"grad_norm": 0.8177442897301394,
|
|
"learning_rate": 2.6322241183149057e-06,
|
|
"loss": 0.1153,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.059341538697481155,
|
|
"step": 4280,
|
|
"valid_targets_mean": 1655.1,
|
|
"valid_targets_min": 1227
|
|
},
|
|
{
|
|
"epoch": 5.967966573816156,
|
|
"grad_norm": 0.8272238546293612,
|
|
"learning_rate": 2.5978857362919054e-06,
|
|
"loss": 0.1135,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05763782560825348,
|
|
"step": 4285,
|
|
"valid_targets_mean": 1565.8,
|
|
"valid_targets_min": 870
|
|
},
|
|
{
|
|
"epoch": 5.974930362116992,
|
|
"grad_norm": 0.9573734999330009,
|
|
"learning_rate": 2.5637572425341174e-06,
|
|
"loss": 0.122,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06584827601909637,
|
|
"step": 4290,
|
|
"valid_targets_mean": 1564.6,
|
|
"valid_targets_min": 822
|
|
},
|
|
{
|
|
"epoch": 5.981894150417827,
|
|
"grad_norm": 1.1210065324101028,
|
|
"learning_rate": 2.529839048667997e-06,
|
|
"loss": 0.1222,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05611727386713028,
|
|
"step": 4295,
|
|
"valid_targets_mean": 1540.5,
|
|
"valid_targets_min": 1214
|
|
},
|
|
{
|
|
"epoch": 5.988857938718663,
|
|
"grad_norm": 0.8877917752913989,
|
|
"learning_rate": 2.4961315637835346e-06,
|
|
"loss": 0.1124,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05314347892999649,
|
|
"step": 4300,
|
|
"valid_targets_mean": 1268.6,
|
|
"valid_targets_min": 591
|
|
},
|
|
{
|
|
"epoch": 5.995821727019498,
|
|
"grad_norm": 0.9062573425634036,
|
|
"learning_rate": 2.4626351944293435e-06,
|
|
"loss": 0.1205,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04725968837738037,
|
|
"step": 4305,
|
|
"valid_targets_mean": 1232.5,
|
|
"valid_targets_min": 646
|
|
},
|
|
{
|
|
"epoch": 6.002785515320334,
|
|
"grad_norm": 0.9480677409493633,
|
|
"learning_rate": 2.4293503446077638e-06,
|
|
"loss": 0.1432,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08825192600488663,
|
|
"step": 4310,
|
|
"valid_targets_mean": 8514.0,
|
|
"valid_targets_min": 6788
|
|
},
|
|
{
|
|
"epoch": 6.00974930362117,
|
|
"grad_norm": 0.8792429695725503,
|
|
"learning_rate": 2.3962774157699853e-06,
|
|
"loss": 0.1906,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09347966313362122,
|
|
"step": 4315,
|
|
"valid_targets_mean": 6668.0,
|
|
"valid_targets_min": 5933
|
|
},
|
|
{
|
|
"epoch": 6.016713091922005,
|
|
"grad_norm": 0.7256186554172626,
|
|
"learning_rate": 2.363416806811196e-06,
|
|
"loss": 0.1731,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07888566702604294,
|
|
"step": 4320,
|
|
"valid_targets_mean": 4107.0,
|
|
"valid_targets_min": 161
|
|
},
|
|
{
|
|
"epoch": 6.023676880222841,
|
|
"grad_norm": 0.5337563605817399,
|
|
"learning_rate": 2.3307689140657796e-06,
|
|
"loss": 0.1768,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07886291295289993,
|
|
"step": 4325,
|
|
"valid_targets_mean": 7946.0,
|
|
"valid_targets_min": 5596
|
|
},
|
|
{
|
|
"epoch": 6.030640668523677,
|
|
"grad_norm": 0.5250064440886613,
|
|
"learning_rate": 2.29833413130252e-06,
|
|
"loss": 0.167,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08924835920333862,
|
|
"step": 4330,
|
|
"valid_targets_mean": 7392.8,
|
|
"valid_targets_min": 5629
|
|
},
|
|
{
|
|
"epoch": 6.037604456824512,
|
|
"grad_norm": 1.706269156371303,
|
|
"learning_rate": 2.2661128497198905e-06,
|
|
"loss": 0.1704,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08858384191989899,
|
|
"step": 4335,
|
|
"valid_targets_mean": 6879.6,
|
|
"valid_targets_min": 5113
|
|
},
|
|
{
|
|
"epoch": 6.044568245125348,
|
|
"grad_norm": 0.4944795196550045,
|
|
"learning_rate": 2.234105457941289e-06,
|
|
"loss": 0.1824,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08616100996732712,
|
|
"step": 4340,
|
|
"valid_targets_mean": 6919.5,
|
|
"valid_targets_min": 5080
|
|
},
|
|
{
|
|
"epoch": 6.0515320334261835,
|
|
"grad_norm": 0.46269857441542933,
|
|
"learning_rate": 2.202312342010382e-06,
|
|
"loss": 0.1734,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0758790597319603,
|
|
"step": 4345,
|
|
"valid_targets_mean": 6377.0,
|
|
"valid_targets_min": 3939
|
|
},
|
|
{
|
|
"epoch": 6.0584958217270195,
|
|
"grad_norm": 0.46714693312176175,
|
|
"learning_rate": 2.170733885386431e-06,
|
|
"loss": 0.1623,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08216223120689392,
|
|
"step": 4350,
|
|
"valid_targets_mean": 6392.2,
|
|
"valid_targets_min": 4355
|
|
},
|
|
{
|
|
"epoch": 6.0654596100278555,
|
|
"grad_norm": 0.6249782828216601,
|
|
"learning_rate": 2.1393704689396812e-06,
|
|
"loss": 0.1477,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08479143679141998,
|
|
"step": 4355,
|
|
"valid_targets_mean": 5196.5,
|
|
"valid_targets_min": 137
|
|
},
|
|
{
|
|
"epoch": 6.072423398328691,
|
|
"grad_norm": 0.43679042233173293,
|
|
"learning_rate": 2.108222470946757e-06,
|
|
"loss": 0.1624,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07390058040618896,
|
|
"step": 4360,
|
|
"valid_targets_mean": 7344.4,
|
|
"valid_targets_min": 5244
|
|
},
|
|
{
|
|
"epoch": 6.079387186629527,
|
|
"grad_norm": 0.43303393100622734,
|
|
"learning_rate": 2.077290267086105e-06,
|
|
"loss": 0.1585,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06864619255065918,
|
|
"step": 4365,
|
|
"valid_targets_mean": 7921.0,
|
|
"valid_targets_min": 5311
|
|
},
|
|
{
|
|
"epoch": 6.086350974930362,
|
|
"grad_norm": 0.4246006529195599,
|
|
"learning_rate": 2.04657423043346e-06,
|
|
"loss": 0.1654,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08542105555534363,
|
|
"step": 4370,
|
|
"valid_targets_mean": 7931.2,
|
|
"valid_targets_min": 6015
|
|
},
|
|
{
|
|
"epoch": 6.093314763231198,
|
|
"grad_norm": 0.4106570417894629,
|
|
"learning_rate": 2.016074731457356e-06,
|
|
"loss": 0.157,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07517217099666595,
|
|
"step": 4375,
|
|
"valid_targets_mean": 7640.8,
|
|
"valid_targets_min": 5248
|
|
},
|
|
{
|
|
"epoch": 6.100278551532034,
|
|
"grad_norm": 0.4148098516880654,
|
|
"learning_rate": 1.9857921380146462e-06,
|
|
"loss": 0.155,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07737202942371368,
|
|
"step": 4380,
|
|
"valid_targets_mean": 7249.6,
|
|
"valid_targets_min": 5217
|
|
},
|
|
{
|
|
"epoch": 6.107242339832869,
|
|
"grad_norm": 0.4436920181563948,
|
|
"learning_rate": 1.955726815346053e-06,
|
|
"loss": 0.1622,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08977077901363373,
|
|
"step": 4385,
|
|
"valid_targets_mean": 7619.9,
|
|
"valid_targets_min": 5094
|
|
},
|
|
{
|
|
"epoch": 6.114206128133705,
|
|
"grad_norm": 0.40942198769595695,
|
|
"learning_rate": 1.9258791260717944e-06,
|
|
"loss": 0.1566,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07487683743238449,
|
|
"step": 4390,
|
|
"valid_targets_mean": 7732.8,
|
|
"valid_targets_min": 5842
|
|
},
|
|
{
|
|
"epoch": 6.121169916434541,
|
|
"grad_norm": 0.4096948039067667,
|
|
"learning_rate": 1.896249430187198e-06,
|
|
"loss": 0.1598,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08216005563735962,
|
|
"step": 4395,
|
|
"valid_targets_mean": 7952.9,
|
|
"valid_targets_min": 5494
|
|
},
|
|
{
|
|
"epoch": 6.128133704735376,
|
|
"grad_norm": 0.5382913217347769,
|
|
"learning_rate": 1.866838085058349e-06,
|
|
"loss": 0.1689,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08534446358680725,
|
|
"step": 4400,
|
|
"valid_targets_mean": 8001.0,
|
|
"valid_targets_min": 5454
|
|
},
|
|
{
|
|
"epoch": 6.135097493036212,
|
|
"grad_norm": 0.4993178623034378,
|
|
"learning_rate": 1.837645445417786e-06,
|
|
"loss": 0.1757,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05737616866827011,
|
|
"step": 4405,
|
|
"valid_targets_mean": 2985.1,
|
|
"valid_targets_min": 1896
|
|
},
|
|
{
|
|
"epoch": 6.142061281337047,
|
|
"grad_norm": 0.5514824724986366,
|
|
"learning_rate": 1.8086718633602252e-06,
|
|
"loss": 0.1294,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07018405199050903,
|
|
"step": 4410,
|
|
"valid_targets_mean": 6714.6,
|
|
"valid_targets_min": 4917
|
|
},
|
|
{
|
|
"epoch": 6.149025069637883,
|
|
"grad_norm": 0.4252531390576798,
|
|
"learning_rate": 1.7799176883383152e-06,
|
|
"loss": 0.1564,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08560578525066376,
|
|
"step": 4415,
|
|
"valid_targets_mean": 7595.2,
|
|
"valid_targets_min": 5488
|
|
},
|
|
{
|
|
"epoch": 6.155988857938719,
|
|
"grad_norm": 0.5092771217817552,
|
|
"learning_rate": 1.751383267158413e-06,
|
|
"loss": 0.157,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08620335161685944,
|
|
"step": 4420,
|
|
"valid_targets_mean": 7046.1,
|
|
"valid_targets_min": 5579
|
|
},
|
|
{
|
|
"epoch": 6.162952646239554,
|
|
"grad_norm": 0.4716862438880787,
|
|
"learning_rate": 1.7230689439764136e-06,
|
|
"loss": 0.1633,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07925520837306976,
|
|
"step": 4425,
|
|
"valid_targets_mean": 7108.0,
|
|
"valid_targets_min": 4963
|
|
},
|
|
{
|
|
"epoch": 6.16991643454039,
|
|
"grad_norm": 0.45619274881293254,
|
|
"learning_rate": 1.694975060293591e-06,
|
|
"loss": 0.1609,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08042465150356293,
|
|
"step": 4430,
|
|
"valid_targets_mean": 6810.1,
|
|
"valid_targets_min": 5666
|
|
},
|
|
{
|
|
"epoch": 6.176880222841225,
|
|
"grad_norm": 0.3901334892783887,
|
|
"learning_rate": 1.6671019549524724e-06,
|
|
"loss": 0.1532,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07437565922737122,
|
|
"step": 4435,
|
|
"valid_targets_mean": 8068.5,
|
|
"valid_targets_min": 5524
|
|
},
|
|
{
|
|
"epoch": 6.183844011142061,
|
|
"grad_norm": 0.40732855082482716,
|
|
"learning_rate": 1.63944996413278e-06,
|
|
"loss": 0.1391,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06510244309902191,
|
|
"step": 4440,
|
|
"valid_targets_mean": 6553.8,
|
|
"valid_targets_min": 3621
|
|
},
|
|
{
|
|
"epoch": 6.190807799442897,
|
|
"grad_norm": 0.42124118933928567,
|
|
"learning_rate": 1.612019421347335e-06,
|
|
"loss": 0.1427,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07448364794254303,
|
|
"step": 4445,
|
|
"valid_targets_mean": 6762.5,
|
|
"valid_targets_min": 4968
|
|
},
|
|
{
|
|
"epoch": 6.197771587743732,
|
|
"grad_norm": 0.4383555593223217,
|
|
"learning_rate": 1.5848106574380694e-06,
|
|
"loss": 0.1528,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07571583241224289,
|
|
"step": 4450,
|
|
"valid_targets_mean": 6143.9,
|
|
"valid_targets_min": 4917
|
|
},
|
|
{
|
|
"epoch": 6.204735376044568,
|
|
"grad_norm": 0.5466494230157721,
|
|
"learning_rate": 1.5578240005720147e-06,
|
|
"loss": 0.1383,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10035134851932526,
|
|
"step": 4455,
|
|
"valid_targets_mean": 4583.8,
|
|
"valid_targets_min": 215
|
|
},
|
|
{
|
|
"epoch": 6.211699164345404,
|
|
"grad_norm": 0.4040453396697649,
|
|
"learning_rate": 1.5310597762373624e-06,
|
|
"loss": 0.1641,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08089716732501984,
|
|
"step": 4460,
|
|
"valid_targets_mean": 7317.1,
|
|
"valid_targets_min": 4948
|
|
},
|
|
{
|
|
"epoch": 6.218662952646239,
|
|
"grad_norm": 0.4727586903795262,
|
|
"learning_rate": 1.504518307239522e-06,
|
|
"loss": 0.1755,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08399631083011627,
|
|
"step": 4465,
|
|
"valid_targets_mean": 6437.6,
|
|
"valid_targets_min": 4510
|
|
},
|
|
{
|
|
"epoch": 6.225626740947075,
|
|
"grad_norm": 0.47933738314929225,
|
|
"learning_rate": 1.478199913697227e-06,
|
|
"loss": 0.1672,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09280038625001907,
|
|
"step": 4470,
|
|
"valid_targets_mean": 7707.9,
|
|
"valid_targets_min": 5796
|
|
},
|
|
{
|
|
"epoch": 6.2325905292479105,
|
|
"grad_norm": 0.41595542358539167,
|
|
"learning_rate": 1.4521049130386854e-06,
|
|
"loss": 0.1604,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07995796948671341,
|
|
"step": 4475,
|
|
"valid_targets_mean": 7900.9,
|
|
"valid_targets_min": 5643
|
|
},
|
|
{
|
|
"epoch": 6.2395543175487465,
|
|
"grad_norm": 0.4778381922849063,
|
|
"learning_rate": 1.4262336199977545e-06,
|
|
"loss": 0.1614,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08848213404417038,
|
|
"step": 4480,
|
|
"valid_targets_mean": 7627.2,
|
|
"valid_targets_min": 5742
|
|
},
|
|
{
|
|
"epoch": 6.2465181058495824,
|
|
"grad_norm": 0.4136225741751865,
|
|
"learning_rate": 1.4005863466101265e-06,
|
|
"loss": 0.1651,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09422549605369568,
|
|
"step": 4485,
|
|
"valid_targets_mean": 8725.6,
|
|
"valid_targets_min": 5141
|
|
},
|
|
{
|
|
"epoch": 6.2534818941504176,
|
|
"grad_norm": 0.5089693736225027,
|
|
"learning_rate": 1.375163402209576e-06,
|
|
"loss": 0.1607,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09053859859704971,
|
|
"step": 4490,
|
|
"valid_targets_mean": 6967.8,
|
|
"valid_targets_min": 6088
|
|
},
|
|
{
|
|
"epoch": 6.2604456824512535,
|
|
"grad_norm": 0.4130639739472889,
|
|
"learning_rate": 1.3499650934242325e-06,
|
|
"loss": 0.1637,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07545335590839386,
|
|
"step": 4495,
|
|
"valid_targets_mean": 7140.1,
|
|
"valid_targets_min": 4834
|
|
},
|
|
{
|
|
"epoch": 6.2674094707520895,
|
|
"grad_norm": 0.43000993148245376,
|
|
"learning_rate": 1.3249917241728704e-06,
|
|
"loss": 0.162,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07579700648784637,
|
|
"step": 4500,
|
|
"valid_targets_mean": 6706.9,
|
|
"valid_targets_min": 5108
|
|
},
|
|
{
|
|
"epoch": 6.274373259052925,
|
|
"grad_norm": 0.5160601513181945,
|
|
"learning_rate": 1.3002435956612614e-06,
|
|
"loss": 0.1647,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08105462789535522,
|
|
"step": 4505,
|
|
"valid_targets_mean": 5931.4,
|
|
"valid_targets_min": 4708
|
|
},
|
|
{
|
|
"epoch": 6.281337047353761,
|
|
"grad_norm": 0.44759379754062206,
|
|
"learning_rate": 1.2757210063785186e-06,
|
|
"loss": 0.1581,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07396161556243896,
|
|
"step": 4510,
|
|
"valid_targets_mean": 7145.6,
|
|
"valid_targets_min": 5280
|
|
},
|
|
{
|
|
"epoch": 6.288300835654596,
|
|
"grad_norm": 0.7344682406855992,
|
|
"learning_rate": 1.2514242520935182e-06,
|
|
"loss": 0.1468,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04381747171282768,
|
|
"step": 4515,
|
|
"valid_targets_mean": 1554.9,
|
|
"valid_targets_min": 592
|
|
},
|
|
{
|
|
"epoch": 6.295264623955432,
|
|
"grad_norm": 0.46744454281739256,
|
|
"learning_rate": 1.2273536258513174e-06,
|
|
"loss": 0.1423,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09317928552627563,
|
|
"step": 4520,
|
|
"valid_targets_mean": 7069.5,
|
|
"valid_targets_min": 4765
|
|
},
|
|
{
|
|
"epoch": 6.302228412256268,
|
|
"grad_norm": 0.44624650053434495,
|
|
"learning_rate": 1.2035094179696305e-06,
|
|
"loss": 0.1594,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08074669539928436,
|
|
"step": 4525,
|
|
"valid_targets_mean": 6193.4,
|
|
"valid_targets_min": 5587
|
|
},
|
|
{
|
|
"epoch": 6.309192200557103,
|
|
"grad_norm": 0.45433035530886273,
|
|
"learning_rate": 1.1798919160353183e-06,
|
|
"loss": 0.1596,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0842660665512085,
|
|
"step": 4530,
|
|
"valid_targets_mean": 6612.2,
|
|
"valid_targets_min": 4894
|
|
},
|
|
{
|
|
"epoch": 6.316155988857939,
|
|
"grad_norm": 0.4219263161331273,
|
|
"learning_rate": 1.1565014049009227e-06,
|
|
"loss": 0.1549,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07519182562828064,
|
|
"step": 4535,
|
|
"valid_targets_mean": 6605.9,
|
|
"valid_targets_min": 3795
|
|
},
|
|
{
|
|
"epoch": 6.323119777158775,
|
|
"grad_norm": 0.428099298209557,
|
|
"learning_rate": 1.1333381666812326e-06,
|
|
"loss": 0.1577,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08015845715999603,
|
|
"step": 4540,
|
|
"valid_targets_mean": 7186.8,
|
|
"valid_targets_min": 4890
|
|
},
|
|
{
|
|
"epoch": 6.33008356545961,
|
|
"grad_norm": 0.5405733766441226,
|
|
"learning_rate": 1.110402480749886e-06,
|
|
"loss": 0.1625,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08327323198318481,
|
|
"step": 4545,
|
|
"valid_targets_mean": 6544.4,
|
|
"valid_targets_min": 4615
|
|
},
|
|
{
|
|
"epoch": 6.337047353760446,
|
|
"grad_norm": 0.48862459251110824,
|
|
"learning_rate": 1.0876946237359887e-06,
|
|
"loss": 0.1632,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0856969878077507,
|
|
"step": 4550,
|
|
"valid_targets_mean": 6339.2,
|
|
"valid_targets_min": 4798
|
|
},
|
|
{
|
|
"epoch": 6.344011142061281,
|
|
"grad_norm": 0.5591305131293233,
|
|
"learning_rate": 1.0652148695207809e-06,
|
|
"loss": 0.1176,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08373311161994934,
|
|
"step": 4555,
|
|
"valid_targets_mean": 4558.8,
|
|
"valid_targets_min": 148
|
|
},
|
|
{
|
|
"epoch": 6.350974930362117,
|
|
"grad_norm": 0.4694300456582176,
|
|
"learning_rate": 1.042963489234341e-06,
|
|
"loss": 0.159,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08754730969667435,
|
|
"step": 4560,
|
|
"valid_targets_mean": 6807.9,
|
|
"valid_targets_min": 5164
|
|
},
|
|
{
|
|
"epoch": 6.357938718662953,
|
|
"grad_norm": 0.4564231493271595,
|
|
"learning_rate": 1.0209407512523194e-06,
|
|
"loss": 0.1522,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07847484946250916,
|
|
"step": 4565,
|
|
"valid_targets_mean": 7011.8,
|
|
"valid_targets_min": 4974
|
|
},
|
|
{
|
|
"epoch": 6.364902506963788,
|
|
"grad_norm": 0.44585558276766535,
|
|
"learning_rate": 9.99146921192682e-07,
|
|
"loss": 0.1594,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07176651060581207,
|
|
"step": 4570,
|
|
"valid_targets_mean": 5794.2,
|
|
"valid_targets_min": 4809
|
|
},
|
|
{
|
|
"epoch": 6.371866295264624,
|
|
"grad_norm": 0.4586375643278151,
|
|
"learning_rate": 9.775822619125242e-07,
|
|
"loss": 0.1503,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06948361545801163,
|
|
"step": 4575,
|
|
"valid_targets_mean": 6401.8,
|
|
"valid_targets_min": 4842
|
|
},
|
|
{
|
|
"epoch": 6.378830083565459,
|
|
"grad_norm": 0.4501958201734831,
|
|
"learning_rate": 9.562470335048912e-07,
|
|
"loss": 0.1557,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07945942133665085,
|
|
"step": 4580,
|
|
"valid_targets_mean": 6182.1,
|
|
"valid_targets_min": 5116
|
|
},
|
|
{
|
|
"epoch": 6.385793871866295,
|
|
"grad_norm": 0.4327240600935404,
|
|
"learning_rate": 9.351414932956482e-07,
|
|
"loss": 0.1553,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0714079886674881,
|
|
"step": 4585,
|
|
"valid_targets_mean": 6165.1,
|
|
"valid_targets_min": 5764
|
|
},
|
|
{
|
|
"epoch": 6.392757660167131,
|
|
"grad_norm": 0.8047252344536302,
|
|
"learning_rate": 9.142658958403694e-07,
|
|
"loss": 0.1503,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05607593059539795,
|
|
"step": 4590,
|
|
"valid_targets_mean": 1428.2,
|
|
"valid_targets_min": 1062
|
|
},
|
|
{
|
|
"epoch": 6.399721448467966,
|
|
"grad_norm": 0.8195753374003149,
|
|
"learning_rate": 8.936204929212766e-07,
|
|
"loss": 0.1329,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06777313351631165,
|
|
"step": 4595,
|
|
"valid_targets_mean": 1663.0,
|
|
"valid_targets_min": 960
|
|
},
|
|
{
|
|
"epoch": 6.406685236768802,
|
|
"grad_norm": 0.833685966360577,
|
|
"learning_rate": 8.732055335441902e-07,
|
|
"loss": 0.1253,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06374421715736389,
|
|
"step": 4600,
|
|
"valid_targets_mean": 1476.5,
|
|
"valid_targets_min": 933
|
|
},
|
|
{
|
|
"epoch": 6.413649025069638,
|
|
"grad_norm": 0.780358085772891,
|
|
"learning_rate": 8.530212639355384e-07,
|
|
"loss": 0.1247,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06148510053753853,
|
|
"step": 4605,
|
|
"valid_targets_mean": 1478.1,
|
|
"valid_targets_min": 881
|
|
},
|
|
{
|
|
"epoch": 6.420612813370473,
|
|
"grad_norm": 0.890308955107473,
|
|
"learning_rate": 8.330679275393816e-07,
|
|
"loss": 0.1272,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07429173588752747,
|
|
"step": 4610,
|
|
"valid_targets_mean": 1640.6,
|
|
"valid_targets_min": 869
|
|
},
|
|
{
|
|
"epoch": 6.427576601671309,
|
|
"grad_norm": 0.9489952322841751,
|
|
"learning_rate": 8.133457650144705e-07,
|
|
"loss": 0.1261,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06858045607805252,
|
|
"step": 4615,
|
|
"valid_targets_mean": 1709.0,
|
|
"valid_targets_min": 1090
|
|
},
|
|
{
|
|
"epoch": 6.4345403899721445,
|
|
"grad_norm": 0.8409646954850704,
|
|
"learning_rate": 7.938550142313594e-07,
|
|
"loss": 0.1177,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.046157389879226685,
|
|
"step": 4620,
|
|
"valid_targets_mean": 1157.9,
|
|
"valid_targets_min": 761
|
|
},
|
|
{
|
|
"epoch": 6.4415041782729805,
|
|
"grad_norm": 0.7487722194005665,
|
|
"learning_rate": 7.7459591026952e-07,
|
|
"loss": 0.1286,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05890098214149475,
|
|
"step": 4625,
|
|
"valid_targets_mean": 1553.6,
|
|
"valid_targets_min": 1241
|
|
},
|
|
{
|
|
"epoch": 6.4484679665738165,
|
|
"grad_norm": 0.7904874191930988,
|
|
"learning_rate": 7.555686854145184e-07,
|
|
"loss": 0.117,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05734875053167343,
|
|
"step": 4630,
|
|
"valid_targets_mean": 1538.1,
|
|
"valid_targets_min": 733
|
|
},
|
|
{
|
|
"epoch": 6.455431754874652,
|
|
"grad_norm": 0.792386346804611,
|
|
"learning_rate": 7.367735691552047e-07,
|
|
"loss": 0.1231,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.03880469501018524,
|
|
"step": 4635,
|
|
"valid_targets_mean": 1250.6,
|
|
"valid_targets_min": 835
|
|
},
|
|
{
|
|
"epoch": 6.462395543175488,
|
|
"grad_norm": 0.860965341679149,
|
|
"learning_rate": 7.182107881809486e-07,
|
|
"loss": 0.1115,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06356509774923325,
|
|
"step": 4640,
|
|
"valid_targets_mean": 1596.2,
|
|
"valid_targets_min": 1046
|
|
},
|
|
{
|
|
"epoch": 6.469359331476323,
|
|
"grad_norm": 0.8718675926426537,
|
|
"learning_rate": 6.998805663789077e-07,
|
|
"loss": 0.1212,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05807361379265785,
|
|
"step": 4645,
|
|
"valid_targets_mean": 1646.5,
|
|
"valid_targets_min": 797
|
|
},
|
|
{
|
|
"epoch": 6.476323119777159,
|
|
"grad_norm": 0.8164546555462905,
|
|
"learning_rate": 6.817831248313279e-07,
|
|
"loss": 0.1123,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05345907807350159,
|
|
"step": 4650,
|
|
"valid_targets_mean": 1448.9,
|
|
"valid_targets_min": 570
|
|
},
|
|
{
|
|
"epoch": 6.483286908077995,
|
|
"grad_norm": 0.7688143714918261,
|
|
"learning_rate": 6.639186818128696e-07,
|
|
"loss": 0.1069,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04237380996346474,
|
|
"step": 4655,
|
|
"valid_targets_mean": 1164.1,
|
|
"valid_targets_min": 548
|
|
},
|
|
{
|
|
"epoch": 6.49025069637883,
|
|
"grad_norm": 0.8494248149460677,
|
|
"learning_rate": 6.462874527879859e-07,
|
|
"loss": 0.108,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05192333832383156,
|
|
"step": 4660,
|
|
"valid_targets_mean": 1338.0,
|
|
"valid_targets_min": 682
|
|
},
|
|
{
|
|
"epoch": 6.497214484679666,
|
|
"grad_norm": 0.9725150847641078,
|
|
"learning_rate": 6.288896504083019e-07,
|
|
"loss": 0.1176,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06402169167995453,
|
|
"step": 4665,
|
|
"valid_targets_mean": 1632.8,
|
|
"valid_targets_min": 896
|
|
},
|
|
{
|
|
"epoch": 6.504178272980502,
|
|
"grad_norm": 0.8615786174263845,
|
|
"learning_rate": 6.117254845100884e-07,
|
|
"loss": 0.115,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0654836893081665,
|
|
"step": 4670,
|
|
"valid_targets_mean": 1806.5,
|
|
"valid_targets_min": 597
|
|
},
|
|
{
|
|
"epoch": 6.511142061281337,
|
|
"grad_norm": 0.8339197139282923,
|
|
"learning_rate": 5.947951621116899e-07,
|
|
"loss": 0.1197,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.049058057367801666,
|
|
"step": 4675,
|
|
"valid_targets_mean": 1289.9,
|
|
"valid_targets_min": 706
|
|
},
|
|
{
|
|
"epoch": 6.518105849582173,
|
|
"grad_norm": 0.7968481465627608,
|
|
"learning_rate": 5.78098887411056e-07,
|
|
"loss": 0.1137,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04030627757310867,
|
|
"step": 4680,
|
|
"valid_targets_mean": 1354.1,
|
|
"valid_targets_min": 754
|
|
},
|
|
{
|
|
"epoch": 6.525069637883008,
|
|
"grad_norm": 0.7476070454913744,
|
|
"learning_rate": 5.616368617832657e-07,
|
|
"loss": 0.1139,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.049573637545108795,
|
|
"step": 4685,
|
|
"valid_targets_mean": 1461.9,
|
|
"valid_targets_min": 819
|
|
},
|
|
{
|
|
"epoch": 6.532033426183844,
|
|
"grad_norm": 1.0053730974789308,
|
|
"learning_rate": 5.454092837781e-07,
|
|
"loss": 0.1164,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.041574910283088684,
|
|
"step": 4690,
|
|
"valid_targets_mean": 1225.5,
|
|
"valid_targets_min": 684
|
|
},
|
|
{
|
|
"epoch": 6.53899721448468,
|
|
"grad_norm": 0.9520563364369184,
|
|
"learning_rate": 5.294163491176507e-07,
|
|
"loss": 0.1177,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05617387592792511,
|
|
"step": 4695,
|
|
"valid_targets_mean": 1282.2,
|
|
"valid_targets_min": 711
|
|
},
|
|
{
|
|
"epoch": 6.545961002785515,
|
|
"grad_norm": 1.2484349046028755,
|
|
"learning_rate": 5.136582506939558e-07,
|
|
"loss": 0.1108,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06170322746038437,
|
|
"step": 4700,
|
|
"valid_targets_mean": 1634.8,
|
|
"valid_targets_min": 1190
|
|
},
|
|
{
|
|
"epoch": 6.552924791086351,
|
|
"grad_norm": 0.8277215108085088,
|
|
"learning_rate": 4.981351785666788e-07,
|
|
"loss": 0.1073,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05962364748120308,
|
|
"step": 4705,
|
|
"valid_targets_mean": 1615.9,
|
|
"valid_targets_min": 774
|
|
},
|
|
{
|
|
"epoch": 6.559888579387186,
|
|
"grad_norm": 1.256443333024305,
|
|
"learning_rate": 4.828473199608086e-07,
|
|
"loss": 0.1114,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.062144938856363297,
|
|
"step": 4710,
|
|
"valid_targets_mean": 1848.4,
|
|
"valid_targets_min": 734
|
|
},
|
|
{
|
|
"epoch": 6.566852367688022,
|
|
"grad_norm": 0.8519945991274293,
|
|
"learning_rate": 4.6779485926440994e-07,
|
|
"loss": 0.1102,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06112103536725044,
|
|
"step": 4715,
|
|
"valid_targets_mean": 1570.6,
|
|
"valid_targets_min": 1011
|
|
},
|
|
{
|
|
"epoch": 6.573816155988858,
|
|
"grad_norm": 0.8867015908107557,
|
|
"learning_rate": 4.529779780263899e-07,
|
|
"loss": 0.1188,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06696411967277527,
|
|
"step": 4720,
|
|
"valid_targets_mean": 1697.0,
|
|
"valid_targets_min": 803
|
|
},
|
|
{
|
|
"epoch": 6.580779944289693,
|
|
"grad_norm": 0.8971331539026275,
|
|
"learning_rate": 4.3839685495431494e-07,
|
|
"loss": 0.1156,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05167017877101898,
|
|
"step": 4725,
|
|
"valid_targets_mean": 1307.1,
|
|
"valid_targets_min": 782
|
|
},
|
|
{
|
|
"epoch": 6.587743732590529,
|
|
"grad_norm": 0.8240677573070857,
|
|
"learning_rate": 4.240516659122529e-07,
|
|
"loss": 0.1112,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05892687290906906,
|
|
"step": 4730,
|
|
"valid_targets_mean": 1596.6,
|
|
"valid_targets_min": 646
|
|
},
|
|
{
|
|
"epoch": 6.594707520891365,
|
|
"grad_norm": 0.8777230504903879,
|
|
"learning_rate": 4.099425839186588e-07,
|
|
"loss": 0.1169,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07540211081504822,
|
|
"step": 4735,
|
|
"valid_targets_mean": 1817.5,
|
|
"valid_targets_min": 922
|
|
},
|
|
{
|
|
"epoch": 6.6016713091922,
|
|
"grad_norm": 0.9233129928932025,
|
|
"learning_rate": 3.960697791442725e-07,
|
|
"loss": 0.1185,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06853315234184265,
|
|
"step": 4740,
|
|
"valid_targets_mean": 1822.2,
|
|
"valid_targets_min": 796
|
|
},
|
|
{
|
|
"epoch": 6.608635097493036,
|
|
"grad_norm": 0.8365359905989758,
|
|
"learning_rate": 3.824334189100864e-07,
|
|
"loss": 0.1067,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05597367510199547,
|
|
"step": 4745,
|
|
"valid_targets_mean": 1542.4,
|
|
"valid_targets_min": 903
|
|
},
|
|
{
|
|
"epoch": 6.615598885793872,
|
|
"grad_norm": 0.8653881395676635,
|
|
"learning_rate": 3.6903366768530347e-07,
|
|
"loss": 0.1093,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.059465356171131134,
|
|
"step": 4750,
|
|
"valid_targets_mean": 1459.5,
|
|
"valid_targets_min": 780
|
|
},
|
|
{
|
|
"epoch": 6.6225626740947074,
|
|
"grad_norm": 1.0220209875020088,
|
|
"learning_rate": 3.558706870853801e-07,
|
|
"loss": 0.1059,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05482107773423195,
|
|
"step": 4755,
|
|
"valid_targets_mean": 1581.1,
|
|
"valid_targets_min": 783
|
|
},
|
|
{
|
|
"epoch": 6.629526462395543,
|
|
"grad_norm": 0.8895541017805985,
|
|
"learning_rate": 3.42944635870055e-07,
|
|
"loss": 0.1085,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04831491783261299,
|
|
"step": 4760,
|
|
"valid_targets_mean": 1342.1,
|
|
"valid_targets_min": 662
|
|
},
|
|
{
|
|
"epoch": 6.6364902506963785,
|
|
"grad_norm": 0.85022678615267,
|
|
"learning_rate": 3.302556699414461e-07,
|
|
"loss": 0.1065,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04755166172981262,
|
|
"step": 4765,
|
|
"valid_targets_mean": 1269.1,
|
|
"valid_targets_min": 724
|
|
},
|
|
{
|
|
"epoch": 6.6434540389972145,
|
|
"grad_norm": 0.9159333691892194,
|
|
"learning_rate": 3.178039423421675e-07,
|
|
"loss": 0.1092,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05955637991428375,
|
|
"step": 4770,
|
|
"valid_targets_mean": 1510.1,
|
|
"valid_targets_min": 966
|
|
},
|
|
{
|
|
"epoch": 6.65041782729805,
|
|
"grad_norm": 0.8794943394432526,
|
|
"learning_rate": 3.055896032534844e-07,
|
|
"loss": 0.1116,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04700024425983429,
|
|
"step": 4775,
|
|
"valid_targets_mean": 1327.8,
|
|
"valid_targets_min": 773
|
|
},
|
|
{
|
|
"epoch": 6.657381615598886,
|
|
"grad_norm": 1.0648559373125812,
|
|
"learning_rate": 2.9361279999349456e-07,
|
|
"loss": 0.1072,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0473455972969532,
|
|
"step": 4780,
|
|
"valid_targets_mean": 1353.2,
|
|
"valid_targets_min": 807
|
|
},
|
|
{
|
|
"epoch": 6.664345403899722,
|
|
"grad_norm": 0.8462867733368669,
|
|
"learning_rate": 2.8187367701536514e-07,
|
|
"loss": 0.1106,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0411439947783947,
|
|
"step": 4785,
|
|
"valid_targets_mean": 1142.5,
|
|
"valid_targets_min": 782
|
|
},
|
|
{
|
|
"epoch": 6.671309192200557,
|
|
"grad_norm": 0.8759881208699287,
|
|
"learning_rate": 2.7037237590557874e-07,
|
|
"loss": 0.1082,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06818887591362,
|
|
"step": 4790,
|
|
"valid_targets_mean": 1849.9,
|
|
"valid_targets_min": 634
|
|
},
|
|
{
|
|
"epoch": 6.678272980501393,
|
|
"grad_norm": 0.8652916414354441,
|
|
"learning_rate": 2.591090353822279e-07,
|
|
"loss": 0.1177,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05813930183649063,
|
|
"step": 4795,
|
|
"valid_targets_mean": 1599.6,
|
|
"valid_targets_min": 573
|
|
},
|
|
{
|
|
"epoch": 6.685236768802229,
|
|
"grad_norm": 0.8289668185741821,
|
|
"learning_rate": 2.480837912933476e-07,
|
|
"loss": 0.1141,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06298094987869263,
|
|
"step": 4800,
|
|
"valid_targets_mean": 1994.9,
|
|
"valid_targets_min": 1102
|
|
},
|
|
{
|
|
"epoch": 6.692200557103064,
|
|
"grad_norm": 0.8437774991054501,
|
|
"learning_rate": 2.372967766152745e-07,
|
|
"loss": 0.1149,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04422305151820183,
|
|
"step": 4805,
|
|
"valid_targets_mean": 1239.2,
|
|
"valid_targets_min": 841
|
|
},
|
|
{
|
|
"epoch": 6.6991643454039,
|
|
"grad_norm": 0.8335016224230695,
|
|
"learning_rate": 2.2674812145103253e-07,
|
|
"loss": 0.1059,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04819972813129425,
|
|
"step": 4810,
|
|
"valid_targets_mean": 1467.5,
|
|
"valid_targets_min": 956
|
|
},
|
|
{
|
|
"epoch": 6.706128133704736,
|
|
"grad_norm": 0.9023177270948384,
|
|
"learning_rate": 2.16437953028783e-07,
|
|
"loss": 0.1115,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05732536315917969,
|
|
"step": 4815,
|
|
"valid_targets_mean": 1300.9,
|
|
"valid_targets_min": 857
|
|
},
|
|
{
|
|
"epoch": 6.713091922005571,
|
|
"grad_norm": 0.8464182537507942,
|
|
"learning_rate": 2.0636639570027706e-07,
|
|
"loss": 0.1107,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.054763518273830414,
|
|
"step": 4820,
|
|
"valid_targets_mean": 1389.8,
|
|
"valid_targets_min": 808
|
|
},
|
|
{
|
|
"epoch": 6.720055710306407,
|
|
"grad_norm": 0.8724602751899221,
|
|
"learning_rate": 1.9653357093935677e-07,
|
|
"loss": 0.1042,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05148616060614586,
|
|
"step": 4825,
|
|
"valid_targets_mean": 1303.8,
|
|
"valid_targets_min": 677
|
|
},
|
|
{
|
|
"epoch": 6.727019498607242,
|
|
"grad_norm": 0.892862606177971,
|
|
"learning_rate": 1.8693959734049194e-07,
|
|
"loss": 0.1064,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04439748823642731,
|
|
"step": 4830,
|
|
"valid_targets_mean": 1220.1,
|
|
"valid_targets_min": 727
|
|
},
|
|
{
|
|
"epoch": 6.733983286908078,
|
|
"grad_norm": 0.8006054734140239,
|
|
"learning_rate": 1.7758459061735235e-07,
|
|
"loss": 0.1099,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04145853966474533,
|
|
"step": 4835,
|
|
"valid_targets_mean": 1262.0,
|
|
"valid_targets_min": 726
|
|
},
|
|
{
|
|
"epoch": 6.740947075208914,
|
|
"grad_norm": 0.9012716233194072,
|
|
"learning_rate": 1.6846866360140434e-07,
|
|
"loss": 0.118,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06492114067077637,
|
|
"step": 4840,
|
|
"valid_targets_mean": 1615.0,
|
|
"valid_targets_min": 669
|
|
},
|
|
{
|
|
"epoch": 6.747910863509749,
|
|
"grad_norm": 0.8349181470475844,
|
|
"learning_rate": 1.595919262405632e-07,
|
|
"loss": 0.1111,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05895448848605156,
|
|
"step": 4845,
|
|
"valid_targets_mean": 1806.5,
|
|
"valid_targets_min": 792
|
|
},
|
|
{
|
|
"epoch": 6.754874651810585,
|
|
"grad_norm": 0.8116954418998069,
|
|
"learning_rate": 1.5095448559785176e-07,
|
|
"loss": 0.1073,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05427628010511398,
|
|
"step": 4850,
|
|
"valid_targets_mean": 1627.6,
|
|
"valid_targets_min": 736
|
|
},
|
|
{
|
|
"epoch": 6.76183844011142,
|
|
"grad_norm": 0.8820001141515799,
|
|
"learning_rate": 1.425564458501194e-07,
|
|
"loss": 0.1152,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.07305572926998138,
|
|
"step": 4855,
|
|
"valid_targets_mean": 1822.9,
|
|
"valid_targets_min": 694
|
|
},
|
|
{
|
|
"epoch": 6.768802228412256,
|
|
"grad_norm": 0.8785588698821919,
|
|
"learning_rate": 1.3439790828678523e-07,
|
|
"loss": 0.1075,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06912118941545486,
|
|
"step": 4860,
|
|
"valid_targets_mean": 1848.9,
|
|
"valid_targets_min": 1437
|
|
},
|
|
{
|
|
"epoch": 6.775766016713092,
|
|
"grad_norm": 0.8363864610673993,
|
|
"learning_rate": 1.2647897130860566e-07,
|
|
"loss": 0.1113,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06591206789016724,
|
|
"step": 4865,
|
|
"valid_targets_mean": 1852.4,
|
|
"valid_targets_min": 809
|
|
},
|
|
{
|
|
"epoch": 6.782729805013927,
|
|
"grad_norm": 0.8474827648363192,
|
|
"learning_rate": 1.187997304265065e-07,
|
|
"loss": 0.108,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05124801769852638,
|
|
"step": 4870,
|
|
"valid_targets_mean": 1375.5,
|
|
"valid_targets_min": 790
|
|
},
|
|
{
|
|
"epoch": 6.789693593314763,
|
|
"grad_norm": 0.8696793467523222,
|
|
"learning_rate": 1.1136027826040841e-07,
|
|
"loss": 0.102,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.048445772379636765,
|
|
"step": 4875,
|
|
"valid_targets_mean": 1338.8,
|
|
"valid_targets_min": 750
|
|
},
|
|
{
|
|
"epoch": 6.796657381615599,
|
|
"grad_norm": 0.9202757557686742,
|
|
"learning_rate": 1.041607045381321e-07,
|
|
"loss": 0.1109,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06679531186819077,
|
|
"step": 4880,
|
|
"valid_targets_mean": 1778.8,
|
|
"valid_targets_min": 826
|
|
},
|
|
{
|
|
"epoch": 6.803621169916434,
|
|
"grad_norm": 0.8516595011162112,
|
|
"learning_rate": 9.720109609429928e-08,
|
|
"loss": 0.1121,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05206713080406189,
|
|
"step": 4885,
|
|
"valid_targets_mean": 1600.8,
|
|
"valid_targets_min": 907
|
|
},
|
|
{
|
|
"epoch": 6.81058495821727,
|
|
"grad_norm": 0.8400966183501765,
|
|
"learning_rate": 9.04815368692935e-08,
|
|
"loss": 0.1044,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05995471403002739,
|
|
"step": 4890,
|
|
"valid_targets_mean": 1474.5,
|
|
"valid_targets_min": 1093
|
|
},
|
|
{
|
|
"epoch": 6.8175487465181055,
|
|
"grad_norm": 0.8562520722201993,
|
|
"learning_rate": 8.400210790824536e-08,
|
|
"loss": 0.109,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06154263764619827,
|
|
"step": 4895,
|
|
"valid_targets_mean": 1577.2,
|
|
"valid_targets_min": 1119
|
|
},
|
|
{
|
|
"epoch": 6.8245125348189415,
|
|
"grad_norm": 0.8814865307617711,
|
|
"learning_rate": 7.776288736005776e-08,
|
|
"loss": 0.1132,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06116368994116783,
|
|
"step": 4900,
|
|
"valid_targets_mean": 1512.1,
|
|
"valid_targets_min": 700
|
|
},
|
|
{
|
|
"epoch": 6.8314763231197775,
|
|
"grad_norm": 0.816015444670367,
|
|
"learning_rate": 7.176395047645557e-08,
|
|
"loss": 0.11,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0455644354224205,
|
|
"step": 4905,
|
|
"valid_targets_mean": 1288.2,
|
|
"valid_targets_min": 810
|
|
},
|
|
{
|
|
"epoch": 6.838440111420613,
|
|
"grad_norm": 0.9432317625659447,
|
|
"learning_rate": 6.600536961109072e-08,
|
|
"loss": 0.1136,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.052437908947467804,
|
|
"step": 4910,
|
|
"valid_targets_mean": 1328.9,
|
|
"valid_targets_min": 635
|
|
},
|
|
{
|
|
"epoch": 6.845403899721449,
|
|
"grad_norm": 0.8794044449829441,
|
|
"learning_rate": 6.048721421865633e-08,
|
|
"loss": 0.1057,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.059289056807756424,
|
|
"step": 4915,
|
|
"valid_targets_mean": 1542.5,
|
|
"valid_targets_min": 673
|
|
},
|
|
{
|
|
"epoch": 6.852367688022284,
|
|
"grad_norm": 0.9326004661735458,
|
|
"learning_rate": 5.5209550854058435e-08,
|
|
"loss": 0.1152,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06115001440048218,
|
|
"step": 4920,
|
|
"valid_targets_mean": 1215.6,
|
|
"valid_targets_min": 641
|
|
},
|
|
{
|
|
"epoch": 6.85933147632312,
|
|
"grad_norm": 0.8767540104775717,
|
|
"learning_rate": 5.017244317160997e-08,
|
|
"loss": 0.1056,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.04882102459669113,
|
|
"step": 4925,
|
|
"valid_targets_mean": 1222.6,
|
|
"valid_targets_min": 701
|
|
},
|
|
{
|
|
"epoch": 6.866295264623956,
|
|
"grad_norm": 0.8971771361885165,
|
|
"learning_rate": 4.537595192425803e-08,
|
|
"loss": 0.1211,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0595490038394928,
|
|
"step": 4930,
|
|
"valid_targets_mean": 1850.6,
|
|
"valid_targets_min": 1202
|
|
},
|
|
{
|
|
"epoch": 6.873259052924791,
|
|
"grad_norm": 0.8299434208974764,
|
|
"learning_rate": 4.0820134962864525e-08,
|
|
"loss": 0.1095,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.044841837137937546,
|
|
"step": 4935,
|
|
"valid_targets_mean": 1265.5,
|
|
"valid_targets_min": 657
|
|
},
|
|
{
|
|
"epoch": 6.880222841225627,
|
|
"grad_norm": 0.901095866607732,
|
|
"learning_rate": 3.650504723549109e-08,
|
|
"loss": 0.1125,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05367933213710785,
|
|
"step": 4940,
|
|
"valid_targets_mean": 1388.0,
|
|
"valid_targets_min": 672
|
|
},
|
|
{
|
|
"epoch": 6.887186629526463,
|
|
"grad_norm": 0.8315063396628217,
|
|
"learning_rate": 3.243074078675301e-08,
|
|
"loss": 0.1088,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06695376336574554,
|
|
"step": 4945,
|
|
"valid_targets_mean": 1813.6,
|
|
"valid_targets_min": 1019
|
|
},
|
|
{
|
|
"epoch": 6.894150417827298,
|
|
"grad_norm": 0.8625526799833706,
|
|
"learning_rate": 2.85972647571775e-08,
|
|
"loss": 0.1029,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.044332314282655716,
|
|
"step": 4950,
|
|
"valid_targets_mean": 1114.6,
|
|
"valid_targets_min": 732
|
|
},
|
|
{
|
|
"epoch": 6.901114206128134,
|
|
"grad_norm": 1.0040487494315289,
|
|
"learning_rate": 2.5004665382619698e-08,
|
|
"loss": 0.1126,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0398283377289772,
|
|
"step": 4955,
|
|
"valid_targets_mean": 1194.5,
|
|
"valid_targets_min": 634
|
|
},
|
|
{
|
|
"epoch": 6.908077994428969,
|
|
"grad_norm": 0.8569836647676721,
|
|
"learning_rate": 2.1652985993705355e-08,
|
|
"loss": 0.1055,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0556265227496624,
|
|
"step": 4960,
|
|
"valid_targets_mean": 1756.5,
|
|
"valid_targets_min": 905
|
|
},
|
|
{
|
|
"epoch": 6.915041782729805,
|
|
"grad_norm": 0.8794612149843942,
|
|
"learning_rate": 1.8542267015297932e-08,
|
|
"loss": 0.1044,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06186116486787796,
|
|
"step": 4965,
|
|
"valid_targets_mean": 1554.9,
|
|
"valid_targets_min": 596
|
|
},
|
|
{
|
|
"epoch": 6.922005571030641,
|
|
"grad_norm": 0.8682797357172517,
|
|
"learning_rate": 1.5672545966025633e-08,
|
|
"loss": 0.105,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0438067801296711,
|
|
"step": 4970,
|
|
"valid_targets_mean": 1214.6,
|
|
"valid_targets_min": 687
|
|
},
|
|
{
|
|
"epoch": 6.928969359331476,
|
|
"grad_norm": 0.8529938938679981,
|
|
"learning_rate": 1.3043857457819553e-08,
|
|
"loss": 0.1169,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06511755287647247,
|
|
"step": 4975,
|
|
"valid_targets_mean": 1955.1,
|
|
"valid_targets_min": 1199
|
|
},
|
|
{
|
|
"epoch": 6.935933147632312,
|
|
"grad_norm": 0.8325142431206025,
|
|
"learning_rate": 1.0656233195498466e-08,
|
|
"loss": 0.1088,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.059841353446245193,
|
|
"step": 4980,
|
|
"valid_targets_mean": 1758.6,
|
|
"valid_targets_min": 1214
|
|
},
|
|
{
|
|
"epoch": 6.942896935933147,
|
|
"grad_norm": 0.8532395148660459,
|
|
"learning_rate": 8.509701976382456e-09,
|
|
"loss": 0.1079,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06080895662307739,
|
|
"step": 4985,
|
|
"valid_targets_mean": 1711.5,
|
|
"valid_targets_min": 1122
|
|
},
|
|
{
|
|
"epoch": 6.949860724233983,
|
|
"grad_norm": 0.8576769303500783,
|
|
"learning_rate": 6.6042896899576324e-09,
|
|
"loss": 0.1106,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.051803942769765854,
|
|
"step": 4990,
|
|
"valid_targets_mean": 1489.6,
|
|
"valid_targets_min": 850
|
|
},
|
|
{
|
|
"epoch": 6.956824512534819,
|
|
"grad_norm": 0.7839478511204715,
|
|
"learning_rate": 4.940019317547506e-09,
|
|
"loss": 0.0999,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.049742192029953,
|
|
"step": 4995,
|
|
"valid_targets_mean": 1572.4,
|
|
"valid_targets_min": 1318
|
|
},
|
|
{
|
|
"epoch": 6.963788300835654,
|
|
"grad_norm": 0.8785812149835993,
|
|
"learning_rate": 3.516910932048756e-09,
|
|
"loss": 0.1116,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.060333505272865295,
|
|
"step": 5000,
|
|
"valid_targets_mean": 1500.0,
|
|
"valid_targets_min": 1043
|
|
},
|
|
{
|
|
"epoch": 6.97075208913649,
|
|
"grad_norm": 0.8822603333113546,
|
|
"learning_rate": 2.3349816976825367e-09,
|
|
"loss": 0.1113,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06858158856630325,
|
|
"step": 5005,
|
|
"valid_targets_mean": 1774.9,
|
|
"valid_targets_min": 920
|
|
},
|
|
{
|
|
"epoch": 6.977715877437326,
|
|
"grad_norm": 0.8367535436815119,
|
|
"learning_rate": 1.3942458697924211e-09,
|
|
"loss": 0.118,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05870440602302551,
|
|
"step": 5010,
|
|
"valid_targets_mean": 1547.1,
|
|
"valid_targets_min": 847
|
|
},
|
|
{
|
|
"epoch": 6.984679665738161,
|
|
"grad_norm": 0.7917458458759404,
|
|
"learning_rate": 6.947147946645416e-10,
|
|
"loss": 0.1116,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.05920833349227905,
|
|
"step": 5015,
|
|
"valid_targets_mean": 1769.4,
|
|
"valid_targets_min": 738
|
|
},
|
|
{
|
|
"epoch": 6.991643454038997,
|
|
"grad_norm": 0.9106170116332665,
|
|
"learning_rate": 2.3639690939880613e-10,
|
|
"loss": 0.1124,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06107974797487259,
|
|
"step": 5020,
|
|
"valid_targets_mean": 1540.1,
|
|
"valid_targets_min": 594
|
|
},
|
|
{
|
|
"epoch": 6.998607242339833,
|
|
"grad_norm": 1.1797464944725595,
|
|
"learning_rate": 1.929774180231547e-11,
|
|
"loss": 0.1135,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.06046333163976669,
|
|
"step": 5025,
|
|
"valid_targets_mean": 2228.6,
|
|
"valid_targets_min": 1393
|
|
},
|
|
{
|
|
"epoch": 7.0,
|
|
"step": 5026,
|
|
"total_flos": 1.6519984675532308e+18,
|
|
"train_loss": 0.0,
|
|
"train_runtime": 1.7845,
|
|
"train_samples_per_second": 45043.632,
|
|
"train_steps_per_second": 2816.453
|
|
}
|
|
],
|
|
"logging_steps": 5,
|
|
"max_steps": 5026,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 7,
|
|
"save_steps": 200,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": true
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 1.6519984675532308e+18,
|
|
"train_batch_size": 1,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|