8309 lines
230 KiB
JSON
8309 lines
230 KiB
JSON
|
{
|
||
|
|
"best_global_step": null,
|
||
|
|
"best_metric": null,
|
||
|
|
"best_model_checkpoint": null,
|
||
|
|
"epoch": 7.0,
|
||
|
|
"eval_steps": 500,
|
||
|
|
"global_step": 3759,
|
||
|
|
"is_hyper_param_search": false,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 0.009319664492078284,
|
||
|
|
"grad_norm": 17.328271616488834,
|
||
|
|
"learning_rate": 4.2553191489361704e-07,
|
||
|
|
"loss": 0.7935,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5236566066741943,
|
||
|
|
"step": 5,
|
||
|
|
"valid_targets_mean": 5452.5,
|
||
|
|
"valid_targets_min": 684
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.01863932898415657,
|
||
|
|
"grad_norm": 17.97721725967326,
|
||
|
|
"learning_rate": 9.574468085106384e-07,
|
||
|
|
"loss": 0.7802,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3532390594482422,
|
||
|
|
"step": 10,
|
||
|
|
"valid_targets_mean": 6383.5,
|
||
|
|
"valid_targets_min": 5193
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.027958993476234855,
|
||
|
|
"grad_norm": 15.141679426572683,
|
||
|
|
"learning_rate": 1.4893617021276596e-06,
|
||
|
|
"loss": 0.7546,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4239236116409302,
|
||
|
|
"step": 15,
|
||
|
|
"valid_targets_mean": 4557.8,
|
||
|
|
"valid_targets_min": 771
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03727865796831314,
|
||
|
|
"grad_norm": 9.378955061393178,
|
||
|
|
"learning_rate": 2.021276595744681e-06,
|
||
|
|
"loss": 0.7139,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3642452359199524,
|
||
|
|
"step": 20,
|
||
|
|
"valid_targets_mean": 5100.2,
|
||
|
|
"valid_targets_min": 889
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.046598322460391424,
|
||
|
|
"grad_norm": 5.02241216579534,
|
||
|
|
"learning_rate": 2.553191489361702e-06,
|
||
|
|
"loss": 0.6459,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.34934231638908386,
|
||
|
|
"step": 25,
|
||
|
|
"valid_targets_mean": 5152.5,
|
||
|
|
"valid_targets_min": 903
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05591798695246971,
|
||
|
|
"grad_norm": 2.9123745407592954,
|
||
|
|
"learning_rate": 3.0851063829787237e-06,
|
||
|
|
"loss": 0.632,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.32005417346954346,
|
||
|
|
"step": 30,
|
||
|
|
"valid_targets_mean": 4549.9,
|
||
|
|
"valid_targets_min": 537
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06523765144454799,
|
||
|
|
"grad_norm": 1.6920482389107077,
|
||
|
|
"learning_rate": 3.6170212765957453e-06,
|
||
|
|
"loss": 0.5667,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3148297667503357,
|
||
|
|
"step": 35,
|
||
|
|
"valid_targets_mean": 6886.1,
|
||
|
|
"valid_targets_min": 709
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07455731593662628,
|
||
|
|
"grad_norm": 1.3709573242376476,
|
||
|
|
"learning_rate": 4.148936170212766e-06,
|
||
|
|
"loss": 0.5821,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.343390554189682,
|
||
|
|
"step": 40,
|
||
|
|
"valid_targets_mean": 6529.5,
|
||
|
|
"valid_targets_min": 647
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08387698042870456,
|
||
|
|
"grad_norm": 1.0741937881242667,
|
||
|
|
"learning_rate": 4.680851063829788e-06,
|
||
|
|
"loss": 0.5299,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2307893931865692,
|
||
|
|
"step": 45,
|
||
|
|
"valid_targets_mean": 3443.8,
|
||
|
|
"valid_targets_min": 569
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09319664492078285,
|
||
|
|
"grad_norm": 0.8435997685943285,
|
||
|
|
"learning_rate": 5.212765957446809e-06,
|
||
|
|
"loss": 0.4991,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21582701802253723,
|
||
|
|
"step": 50,
|
||
|
|
"valid_targets_mean": 5229.2,
|
||
|
|
"valid_targets_min": 575
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.10251630941286113,
|
||
|
|
"grad_norm": 0.8342469594094312,
|
||
|
|
"learning_rate": 5.744680851063831e-06,
|
||
|
|
"loss": 0.4901,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.281514048576355,
|
||
|
|
"step": 55,
|
||
|
|
"valid_targets_mean": 6156.5,
|
||
|
|
"valid_targets_min": 4180
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11183597390493942,
|
||
|
|
"grad_norm": 0.7027250602014875,
|
||
|
|
"learning_rate": 6.276595744680851e-06,
|
||
|
|
"loss": 0.5416,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2606807351112366,
|
||
|
|
"step": 60,
|
||
|
|
"valid_targets_mean": 4894.9,
|
||
|
|
"valid_targets_min": 608
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1211556383970177,
|
||
|
|
"grad_norm": 0.633758993458475,
|
||
|
|
"learning_rate": 6.808510638297873e-06,
|
||
|
|
"loss": 0.5136,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2218017876148224,
|
||
|
|
"step": 65,
|
||
|
|
"valid_targets_mean": 4408.1,
|
||
|
|
"valid_targets_min": 867
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13047530288909598,
|
||
|
|
"grad_norm": 0.6377164677945605,
|
||
|
|
"learning_rate": 7.340425531914894e-06,
|
||
|
|
"loss": 0.4918,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22672636806964874,
|
||
|
|
"step": 70,
|
||
|
|
"valid_targets_mean": 4031.2,
|
||
|
|
"valid_targets_min": 635
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13979496738117428,
|
||
|
|
"grad_norm": 0.579018125720568,
|
||
|
|
"learning_rate": 7.872340425531916e-06,
|
||
|
|
"loss": 0.5172,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24489057064056396,
|
||
|
|
"step": 75,
|
||
|
|
"valid_targets_mean": 5357.6,
|
||
|
|
"valid_targets_min": 889
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14911463187325255,
|
||
|
|
"grad_norm": 0.6298913374435575,
|
||
|
|
"learning_rate": 8.404255319148937e-06,
|
||
|
|
"loss": 0.4768,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2623286247253418,
|
||
|
|
"step": 80,
|
||
|
|
"valid_targets_mean": 5387.4,
|
||
|
|
"valid_targets_min": 704
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15843429636533085,
|
||
|
|
"grad_norm": 0.614752881453651,
|
||
|
|
"learning_rate": 8.936170212765958e-06,
|
||
|
|
"loss": 0.4658,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24994345009326935,
|
||
|
|
"step": 85,
|
||
|
|
"valid_targets_mean": 4533.9,
|
||
|
|
"valid_targets_min": 717
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16775396085740912,
|
||
|
|
"grad_norm": 0.49529478262605703,
|
||
|
|
"learning_rate": 9.46808510638298e-06,
|
||
|
|
"loss": 0.4688,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22537221014499664,
|
||
|
|
"step": 90,
|
||
|
|
"valid_targets_mean": 5335.8,
|
||
|
|
"valid_targets_min": 762
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17707362534948742,
|
||
|
|
"grad_norm": 0.6586467493515626,
|
||
|
|
"learning_rate": 1e-05,
|
||
|
|
"loss": 0.4931,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2837511897087097,
|
||
|
|
"step": 95,
|
||
|
|
"valid_targets_mean": 4185.6,
|
||
|
|
"valid_targets_min": 581
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1863932898415657,
|
||
|
|
"grad_norm": 0.551249121756281,
|
||
|
|
"learning_rate": 1.0531914893617022e-05,
|
||
|
|
"loss": 0.4654,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16866718232631683,
|
||
|
|
"step": 100,
|
||
|
|
"valid_targets_mean": 5124.4,
|
||
|
|
"valid_targets_min": 562
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.195712954333644,
|
||
|
|
"grad_norm": 0.6072938611230441,
|
||
|
|
"learning_rate": 1.1063829787234044e-05,
|
||
|
|
"loss": 0.4451,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3024511933326721,
|
||
|
|
"step": 105,
|
||
|
|
"valid_targets_mean": 5655.0,
|
||
|
|
"valid_targets_min": 735
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.20503261882572227,
|
||
|
|
"grad_norm": 0.7581102723676115,
|
||
|
|
"learning_rate": 1.1595744680851065e-05,
|
||
|
|
"loss": 0.4687,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20314525067806244,
|
||
|
|
"step": 110,
|
||
|
|
"valid_targets_mean": 5226.1,
|
||
|
|
"valid_targets_min": 1133
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21435228331780057,
|
||
|
|
"grad_norm": 0.6381361641267049,
|
||
|
|
"learning_rate": 1.2127659574468087e-05,
|
||
|
|
"loss": 0.494,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23310333490371704,
|
||
|
|
"step": 115,
|
||
|
|
"valid_targets_mean": 4492.2,
|
||
|
|
"valid_targets_min": 493
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22367194780987884,
|
||
|
|
"grad_norm": 0.690122938737561,
|
||
|
|
"learning_rate": 1.2659574468085108e-05,
|
||
|
|
"loss": 0.4451,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19993986189365387,
|
||
|
|
"step": 120,
|
||
|
|
"valid_targets_mean": 3240.0,
|
||
|
|
"valid_targets_min": 400
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23299161230195714,
|
||
|
|
"grad_norm": 0.5600285807596176,
|
||
|
|
"learning_rate": 1.3191489361702127e-05,
|
||
|
|
"loss": 0.4553,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1823575496673584,
|
||
|
|
"step": 125,
|
||
|
|
"valid_targets_mean": 5417.1,
|
||
|
|
"valid_targets_min": 672
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2423112767940354,
|
||
|
|
"grad_norm": 0.5940461097812059,
|
||
|
|
"learning_rate": 1.372340425531915e-05,
|
||
|
|
"loss": 0.4269,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18409493565559387,
|
||
|
|
"step": 130,
|
||
|
|
"valid_targets_mean": 4215.6,
|
||
|
|
"valid_targets_min": 593
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2516309412861137,
|
||
|
|
"grad_norm": 0.764630577972403,
|
||
|
|
"learning_rate": 1.425531914893617e-05,
|
||
|
|
"loss": 0.4467,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2847345769405365,
|
||
|
|
"step": 135,
|
||
|
|
"valid_targets_mean": 5891.0,
|
||
|
|
"valid_targets_min": 793
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.26095060577819196,
|
||
|
|
"grad_norm": 0.5170267975660888,
|
||
|
|
"learning_rate": 1.4787234042553193e-05,
|
||
|
|
"loss": 0.4079,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2328973263502121,
|
||
|
|
"step": 140,
|
||
|
|
"valid_targets_mean": 6454.4,
|
||
|
|
"valid_targets_min": 989
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2702702702702703,
|
||
|
|
"grad_norm": 0.5680319002477419,
|
||
|
|
"learning_rate": 1.5319148936170214e-05,
|
||
|
|
"loss": 0.4036,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1624707132577896,
|
||
|
|
"step": 145,
|
||
|
|
"valid_targets_mean": 4467.4,
|
||
|
|
"valid_targets_min": 683
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.27958993476234856,
|
||
|
|
"grad_norm": 0.6788939566033827,
|
||
|
|
"learning_rate": 1.5851063829787235e-05,
|
||
|
|
"loss": 0.423,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23229336738586426,
|
||
|
|
"step": 150,
|
||
|
|
"valid_targets_mean": 4603.8,
|
||
|
|
"valid_targets_min": 851
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.28890959925442683,
|
||
|
|
"grad_norm": 0.6768339579668715,
|
||
|
|
"learning_rate": 1.6382978723404255e-05,
|
||
|
|
"loss": 0.418,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2668764591217041,
|
||
|
|
"step": 155,
|
||
|
|
"valid_targets_mean": 4593.2,
|
||
|
|
"valid_targets_min": 938
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2982292637465051,
|
||
|
|
"grad_norm": 0.6251276508466813,
|
||
|
|
"learning_rate": 1.6914893617021276e-05,
|
||
|
|
"loss": 0.4218,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21148009598255157,
|
||
|
|
"step": 160,
|
||
|
|
"valid_targets_mean": 4730.5,
|
||
|
|
"valid_targets_min": 630
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.30754892823858343,
|
||
|
|
"grad_norm": 0.531231603690842,
|
||
|
|
"learning_rate": 1.74468085106383e-05,
|
||
|
|
"loss": 0.4324,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2506332993507385,
|
||
|
|
"step": 165,
|
||
|
|
"valid_targets_mean": 6544.5,
|
||
|
|
"valid_targets_min": 675
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3168685927306617,
|
||
|
|
"grad_norm": 0.6812995447979688,
|
||
|
|
"learning_rate": 1.797872340425532e-05,
|
||
|
|
"loss": 0.4169,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19858166575431824,
|
||
|
|
"step": 170,
|
||
|
|
"valid_targets_mean": 3202.2,
|
||
|
|
"valid_targets_min": 565
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.32618825722274,
|
||
|
|
"grad_norm": 0.6062524956578069,
|
||
|
|
"learning_rate": 1.8510638297872342e-05,
|
||
|
|
"loss": 0.3888,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1954929232597351,
|
||
|
|
"step": 175,
|
||
|
|
"valid_targets_mean": 3711.8,
|
||
|
|
"valid_targets_min": 503
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.33550792171481825,
|
||
|
|
"grad_norm": 0.590552963720586,
|
||
|
|
"learning_rate": 1.9042553191489363e-05,
|
||
|
|
"loss": 0.4036,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18431639671325684,
|
||
|
|
"step": 180,
|
||
|
|
"valid_targets_mean": 4595.4,
|
||
|
|
"valid_targets_min": 664
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3448275862068966,
|
||
|
|
"grad_norm": 0.5348090745564571,
|
||
|
|
"learning_rate": 1.9574468085106384e-05,
|
||
|
|
"loss": 0.3999,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24849465489387512,
|
||
|
|
"step": 185,
|
||
|
|
"valid_targets_mean": 6448.2,
|
||
|
|
"valid_targets_min": 1013
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.35414725069897485,
|
||
|
|
"grad_norm": 0.6030708995461287,
|
||
|
|
"learning_rate": 2.0106382978723404e-05,
|
||
|
|
"loss": 0.4144,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20483532547950745,
|
||
|
|
"step": 190,
|
||
|
|
"valid_targets_mean": 4653.0,
|
||
|
|
"valid_targets_min": 452
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3634669151910531,
|
||
|
|
"grad_norm": 0.8358685234002906,
|
||
|
|
"learning_rate": 2.063829787234043e-05,
|
||
|
|
"loss": 0.4034,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23766618967056274,
|
||
|
|
"step": 195,
|
||
|
|
"valid_targets_mean": 2602.1,
|
||
|
|
"valid_targets_min": 290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3727865796831314,
|
||
|
|
"grad_norm": 0.5980242075654355,
|
||
|
|
"learning_rate": 2.117021276595745e-05,
|
||
|
|
"loss": 0.4168,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16135120391845703,
|
||
|
|
"step": 200,
|
||
|
|
"valid_targets_mean": 5545.6,
|
||
|
|
"valid_targets_min": 539
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3821062441752097,
|
||
|
|
"grad_norm": 0.6377861142528783,
|
||
|
|
"learning_rate": 2.1702127659574467e-05,
|
||
|
|
"loss": 0.4021,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29473748803138733,
|
||
|
|
"step": 205,
|
||
|
|
"valid_targets_mean": 6053.9,
|
||
|
|
"valid_targets_min": 1006
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.391425908667288,
|
||
|
|
"grad_norm": 0.5496840455487942,
|
||
|
|
"learning_rate": 2.223404255319149e-05,
|
||
|
|
"loss": 0.383,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14974994957447052,
|
||
|
|
"step": 210,
|
||
|
|
"valid_targets_mean": 5467.9,
|
||
|
|
"valid_targets_min": 882
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.40074557315936626,
|
||
|
|
"grad_norm": 0.7239852828923016,
|
||
|
|
"learning_rate": 2.2765957446808512e-05,
|
||
|
|
"loss": 0.3942,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1629229038953781,
|
||
|
|
"step": 215,
|
||
|
|
"valid_targets_mean": 3374.0,
|
||
|
|
"valid_targets_min": 571
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.41006523765144454,
|
||
|
|
"grad_norm": 0.6111171496246975,
|
||
|
|
"learning_rate": 2.3297872340425536e-05,
|
||
|
|
"loss": 0.3723,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23944884538650513,
|
||
|
|
"step": 220,
|
||
|
|
"valid_targets_mean": 5255.5,
|
||
|
|
"valid_targets_min": 567
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4193849021435228,
|
||
|
|
"grad_norm": 0.6272225885248738,
|
||
|
|
"learning_rate": 2.3829787234042553e-05,
|
||
|
|
"loss": 0.4125,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23209282755851746,
|
||
|
|
"step": 225,
|
||
|
|
"valid_targets_mean": 5272.5,
|
||
|
|
"valid_targets_min": 4005
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.42870456663560114,
|
||
|
|
"grad_norm": 0.6599237422148458,
|
||
|
|
"learning_rate": 2.4361702127659578e-05,
|
||
|
|
"loss": 0.3748,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17184603214263916,
|
||
|
|
"step": 230,
|
||
|
|
"valid_targets_mean": 8307.5,
|
||
|
|
"valid_targets_min": 4016
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4380242311276794,
|
||
|
|
"grad_norm": 0.6601349556228789,
|
||
|
|
"learning_rate": 2.48936170212766e-05,
|
||
|
|
"loss": 0.4284,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24236704409122467,
|
||
|
|
"step": 235,
|
||
|
|
"valid_targets_mean": 4578.5,
|
||
|
|
"valid_targets_min": 574
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4473438956197577,
|
||
|
|
"grad_norm": 0.6044661624200424,
|
||
|
|
"learning_rate": 2.5425531914893616e-05,
|
||
|
|
"loss": 0.3934,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16210556030273438,
|
||
|
|
"step": 240,
|
||
|
|
"valid_targets_mean": 3993.9,
|
||
|
|
"valid_targets_min": 473
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45666356011183595,
|
||
|
|
"grad_norm": 0.5907553537635412,
|
||
|
|
"learning_rate": 2.595744680851064e-05,
|
||
|
|
"loss": 0.3815,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22304849326610565,
|
||
|
|
"step": 245,
|
||
|
|
"valid_targets_mean": 4994.6,
|
||
|
|
"valid_targets_min": 928
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4659832246039143,
|
||
|
|
"grad_norm": 0.6997264151099214,
|
||
|
|
"learning_rate": 2.6489361702127664e-05,
|
||
|
|
"loss": 0.3944,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24181726574897766,
|
||
|
|
"step": 250,
|
||
|
|
"valid_targets_mean": 4197.9,
|
||
|
|
"valid_targets_min": 579
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47530288909599255,
|
||
|
|
"grad_norm": 0.6833409561960894,
|
||
|
|
"learning_rate": 2.702127659574468e-05,
|
||
|
|
"loss": 0.3955,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18640238046646118,
|
||
|
|
"step": 255,
|
||
|
|
"valid_targets_mean": 4685.2,
|
||
|
|
"valid_targets_min": 709
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4846225535880708,
|
||
|
|
"grad_norm": 0.5995343818958372,
|
||
|
|
"learning_rate": 2.7553191489361702e-05,
|
||
|
|
"loss": 0.3747,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17144665122032166,
|
||
|
|
"step": 260,
|
||
|
|
"valid_targets_mean": 4672.5,
|
||
|
|
"valid_targets_min": 705
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4939422180801491,
|
||
|
|
"grad_norm": 0.6285974997091741,
|
||
|
|
"learning_rate": 2.8085106382978727e-05,
|
||
|
|
"loss": 0.4138,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18355786800384521,
|
||
|
|
"step": 265,
|
||
|
|
"valid_targets_mean": 3932.1,
|
||
|
|
"valid_targets_min": 505
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5032618825722274,
|
||
|
|
"grad_norm": 0.7188128645756522,
|
||
|
|
"learning_rate": 2.8617021276595747e-05,
|
||
|
|
"loss": 0.3856,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23198002576828003,
|
||
|
|
"step": 270,
|
||
|
|
"valid_targets_mean": 4684.9,
|
||
|
|
"valid_targets_min": 627
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5125815470643057,
|
||
|
|
"grad_norm": 0.6119958008395558,
|
||
|
|
"learning_rate": 2.9148936170212768e-05,
|
||
|
|
"loss": 0.3765,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21383114159107208,
|
||
|
|
"step": 275,
|
||
|
|
"valid_targets_mean": 5689.2,
|
||
|
|
"valid_targets_min": 909
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5219012115563839,
|
||
|
|
"grad_norm": 0.7708939421703783,
|
||
|
|
"learning_rate": 2.968085106382979e-05,
|
||
|
|
"loss": 0.398,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19482740759849548,
|
||
|
|
"step": 280,
|
||
|
|
"valid_targets_mean": 5765.0,
|
||
|
|
"valid_targets_min": 926
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5312208760484622,
|
||
|
|
"grad_norm": 0.5290994978248169,
|
||
|
|
"learning_rate": 3.0212765957446813e-05,
|
||
|
|
"loss": 0.4143,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24696442484855652,
|
||
|
|
"step": 285,
|
||
|
|
"valid_targets_mean": 7307.4,
|
||
|
|
"valid_targets_min": 1160
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5405405405405406,
|
||
|
|
"grad_norm": 0.7319541893357391,
|
||
|
|
"learning_rate": 3.074468085106383e-05,
|
||
|
|
"loss": 0.3472,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20133322477340698,
|
||
|
|
"step": 290,
|
||
|
|
"valid_targets_mean": 6405.5,
|
||
|
|
"valid_targets_min": 574
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5498602050326188,
|
||
|
|
"grad_norm": 0.5495310464427745,
|
||
|
|
"learning_rate": 3.127659574468085e-05,
|
||
|
|
"loss": 0.3881,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1541924625635147,
|
||
|
|
"step": 295,
|
||
|
|
"valid_targets_mean": 4248.1,
|
||
|
|
"valid_targets_min": 716
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5591798695246971,
|
||
|
|
"grad_norm": 0.6806940235072374,
|
||
|
|
"learning_rate": 3.180851063829788e-05,
|
||
|
|
"loss": 0.36,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2041209191083908,
|
||
|
|
"step": 300,
|
||
|
|
"valid_targets_mean": 6825.1,
|
||
|
|
"valid_targets_min": 2173
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5684995340167754,
|
||
|
|
"grad_norm": 0.5516200900154095,
|
||
|
|
"learning_rate": 3.234042553191489e-05,
|
||
|
|
"loss": 0.3589,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19386395812034607,
|
||
|
|
"step": 305,
|
||
|
|
"valid_targets_mean": 6026.0,
|
||
|
|
"valid_targets_min": 4582
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5778191985088537,
|
||
|
|
"grad_norm": 0.5300983545390556,
|
||
|
|
"learning_rate": 3.2872340425531914e-05,
|
||
|
|
"loss": 0.3584,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13933435082435608,
|
||
|
|
"step": 310,
|
||
|
|
"valid_targets_mean": 4965.0,
|
||
|
|
"valid_targets_min": 714
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.587138863000932,
|
||
|
|
"grad_norm": 0.5308180966419191,
|
||
|
|
"learning_rate": 3.340425531914894e-05,
|
||
|
|
"loss": 0.3649,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18362927436828613,
|
||
|
|
"step": 315,
|
||
|
|
"valid_targets_mean": 4132.4,
|
||
|
|
"valid_targets_min": 705
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5964585274930102,
|
||
|
|
"grad_norm": 0.5598534024330523,
|
||
|
|
"learning_rate": 3.393617021276596e-05,
|
||
|
|
"loss": 0.3869,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1434950977563858,
|
||
|
|
"step": 320,
|
||
|
|
"valid_targets_mean": 3850.2,
|
||
|
|
"valid_targets_min": 855
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6057781919850885,
|
||
|
|
"grad_norm": 0.5778599262306042,
|
||
|
|
"learning_rate": 3.446808510638298e-05,
|
||
|
|
"loss": 0.3848,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15269169211387634,
|
||
|
|
"step": 325,
|
||
|
|
"valid_targets_mean": 4394.8,
|
||
|
|
"valid_targets_min": 704
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6150978564771669,
|
||
|
|
"grad_norm": 0.5323473969018429,
|
||
|
|
"learning_rate": 3.5000000000000004e-05,
|
||
|
|
"loss": 0.3574,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17514492571353912,
|
||
|
|
"step": 330,
|
||
|
|
"valid_targets_mean": 5784.4,
|
||
|
|
"valid_targets_min": 813
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6244175209692451,
|
||
|
|
"grad_norm": 0.491129255950744,
|
||
|
|
"learning_rate": 3.5531914893617025e-05,
|
||
|
|
"loss": 0.3639,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14187946915626526,
|
||
|
|
"step": 335,
|
||
|
|
"valid_targets_mean": 4756.6,
|
||
|
|
"valid_targets_min": 682
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6337371854613234,
|
||
|
|
"grad_norm": 0.6677220904917631,
|
||
|
|
"learning_rate": 3.6063829787234045e-05,
|
||
|
|
"loss": 0.3648,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20674963295459747,
|
||
|
|
"step": 340,
|
||
|
|
"valid_targets_mean": 4066.4,
|
||
|
|
"valid_targets_min": 707
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6430568499534017,
|
||
|
|
"grad_norm": 0.37915333829057685,
|
||
|
|
"learning_rate": 3.6595744680851066e-05,
|
||
|
|
"loss": 0.356,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12468498945236206,
|
||
|
|
"step": 345,
|
||
|
|
"valid_targets_mean": 7457.2,
|
||
|
|
"valid_targets_min": 830
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.65237651444548,
|
||
|
|
"grad_norm": 0.6620147416847397,
|
||
|
|
"learning_rate": 3.712765957446809e-05,
|
||
|
|
"loss": 0.3613,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18722066283226013,
|
||
|
|
"step": 350,
|
||
|
|
"valid_targets_mean": 5494.4,
|
||
|
|
"valid_targets_min": 796
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6616961789375583,
|
||
|
|
"grad_norm": 0.565905110432582,
|
||
|
|
"learning_rate": 3.7659574468085114e-05,
|
||
|
|
"loss": 0.3689,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16835922002792358,
|
||
|
|
"step": 355,
|
||
|
|
"valid_targets_mean": 3932.1,
|
||
|
|
"valid_targets_min": 589
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6710158434296365,
|
||
|
|
"grad_norm": 0.583286134059242,
|
||
|
|
"learning_rate": 3.819148936170213e-05,
|
||
|
|
"loss": 0.3447,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18441668152809143,
|
||
|
|
"step": 360,
|
||
|
|
"valid_targets_mean": 5088.5,
|
||
|
|
"valid_targets_min": 984
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6803355079217148,
|
||
|
|
"grad_norm": 0.562758062776971,
|
||
|
|
"learning_rate": 3.872340425531915e-05,
|
||
|
|
"loss": 0.356,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22111114859580994,
|
||
|
|
"step": 365,
|
||
|
|
"valid_targets_mean": 5984.8,
|
||
|
|
"valid_targets_min": 543
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6896551724137931,
|
||
|
|
"grad_norm": 0.5833542318109176,
|
||
|
|
"learning_rate": 3.925531914893618e-05,
|
||
|
|
"loss": 0.3405,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17007574439048767,
|
||
|
|
"step": 370,
|
||
|
|
"valid_targets_mean": 6250.1,
|
||
|
|
"valid_targets_min": 728
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6989748369058714,
|
||
|
|
"grad_norm": 0.6855690036683716,
|
||
|
|
"learning_rate": 3.978723404255319e-05,
|
||
|
|
"loss": 0.3688,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22733967006206512,
|
||
|
|
"step": 375,
|
||
|
|
"valid_targets_mean": 4538.1,
|
||
|
|
"valid_targets_min": 658
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7082945013979497,
|
||
|
|
"grad_norm": 0.5965384244069067,
|
||
|
|
"learning_rate": 3.999992238637315e-05,
|
||
|
|
"loss": 0.3749,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2017524391412735,
|
||
|
|
"step": 380,
|
||
|
|
"valid_targets_mean": 5670.5,
|
||
|
|
"valid_targets_min": 764
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7176141658900279,
|
||
|
|
"grad_norm": 0.5910899359400373,
|
||
|
|
"learning_rate": 3.9999448083057144e-05,
|
||
|
|
"loss": 0.3559,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16736526787281036,
|
||
|
|
"step": 385,
|
||
|
|
"valid_targets_mean": 5757.4,
|
||
|
|
"valid_targets_min": 489
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7269338303821062,
|
||
|
|
"grad_norm": 0.5312459686890377,
|
||
|
|
"learning_rate": 3.999854260531999e-05,
|
||
|
|
"loss": 0.3816,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19784556329250336,
|
||
|
|
"step": 390,
|
||
|
|
"valid_targets_mean": 6786.0,
|
||
|
|
"valid_targets_min": 695
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7362534948741846,
|
||
|
|
"grad_norm": 0.6708304860475924,
|
||
|
|
"learning_rate": 3.9997205972683174e-05,
|
||
|
|
"loss": 0.3408,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19118860363960266,
|
||
|
|
"step": 395,
|
||
|
|
"valid_targets_mean": 4075.1,
|
||
|
|
"valid_targets_min": 541
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7455731593662628,
|
||
|
|
"grad_norm": 0.6595097551730269,
|
||
|
|
"learning_rate": 3.999543821396357e-05,
|
||
|
|
"loss": 0.3582,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21778111159801483,
|
||
|
|
"step": 400,
|
||
|
|
"valid_targets_mean": 4638.6,
|
||
|
|
"valid_targets_min": 574
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7548928238583411,
|
||
|
|
"grad_norm": 0.5161475817985995,
|
||
|
|
"learning_rate": 3.999323936727285e-05,
|
||
|
|
"loss": 0.3539,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16146428883075714,
|
||
|
|
"step": 405,
|
||
|
|
"valid_targets_mean": 5282.5,
|
||
|
|
"valid_targets_min": 1077
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7642124883504194,
|
||
|
|
"grad_norm": 0.5966760561989775,
|
||
|
|
"learning_rate": 3.999060948001665e-05,
|
||
|
|
"loss": 0.3441,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1795726716518402,
|
||
|
|
"step": 410,
|
||
|
|
"valid_targets_mean": 4953.9,
|
||
|
|
"valid_targets_min": 663
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7735321528424977,
|
||
|
|
"grad_norm": 0.5509093266432215,
|
||
|
|
"learning_rate": 3.998754860889353e-05,
|
||
|
|
"loss": 0.3844,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19792440533638,
|
||
|
|
"step": 415,
|
||
|
|
"valid_targets_mean": 5953.9,
|
||
|
|
"valid_targets_min": 737
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.782851817334576,
|
||
|
|
"grad_norm": 0.6390030178994652,
|
||
|
|
"learning_rate": 3.998405681989378e-05,
|
||
|
|
"loss": 0.3773,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20613934099674225,
|
||
|
|
"step": 420,
|
||
|
|
"valid_targets_mean": 4224.4,
|
||
|
|
"valid_targets_min": 674
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7921714818266542,
|
||
|
|
"grad_norm": 0.7057303943337726,
|
||
|
|
"learning_rate": 3.998013418829799e-05,
|
||
|
|
"loss": 0.3477,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.136399507522583,
|
||
|
|
"step": 425,
|
||
|
|
"valid_targets_mean": 3159.8,
|
||
|
|
"valid_targets_min": 647
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8014911463187325,
|
||
|
|
"grad_norm": 0.6704146022933026,
|
||
|
|
"learning_rate": 3.997578079867542e-05,
|
||
|
|
"loss": 0.3399,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18061327934265137,
|
||
|
|
"step": 430,
|
||
|
|
"valid_targets_mean": 3761.4,
|
||
|
|
"valid_targets_min": 591
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8108108108108109,
|
||
|
|
"grad_norm": 0.5723969687493538,
|
||
|
|
"learning_rate": 3.997099674488215e-05,
|
||
|
|
"loss": 0.3236,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13792851567268372,
|
||
|
|
"step": 435,
|
||
|
|
"valid_targets_mean": 4989.6,
|
||
|
|
"valid_targets_min": 728
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8201304753028891,
|
||
|
|
"grad_norm": 0.6008175377452926,
|
||
|
|
"learning_rate": 3.9965782130059124e-05,
|
||
|
|
"loss": 0.3517,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2434007078409195,
|
||
|
|
"step": 440,
|
||
|
|
"valid_targets_mean": 6817.4,
|
||
|
|
"valid_targets_min": 5962
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8294501397949674,
|
||
|
|
"grad_norm": 0.5894199117275819,
|
||
|
|
"learning_rate": 3.996013706662987e-05,
|
||
|
|
"loss": 0.3546,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1499970257282257,
|
||
|
|
"step": 445,
|
||
|
|
"valid_targets_mean": 4534.5,
|
||
|
|
"valid_targets_min": 848
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8387698042870456,
|
||
|
|
"grad_norm": 0.559422167684271,
|
||
|
|
"learning_rate": 3.995406167629809e-05,
|
||
|
|
"loss": 0.3576,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16677437722682953,
|
||
|
|
"step": 450,
|
||
|
|
"valid_targets_mean": 5276.2,
|
||
|
|
"valid_targets_min": 564
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.848089468779124,
|
||
|
|
"grad_norm": 0.6130383655536883,
|
||
|
|
"learning_rate": 3.994755609004505e-05,
|
||
|
|
"loss": 0.3652,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15164849162101746,
|
||
|
|
"step": 455,
|
||
|
|
"valid_targets_mean": 3934.4,
|
||
|
|
"valid_targets_min": 581
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8574091332712023,
|
||
|
|
"grad_norm": 0.5487568808003801,
|
||
|
|
"learning_rate": 3.994062044812673e-05,
|
||
|
|
"loss": 0.3383,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19470691680908203,
|
||
|
|
"step": 460,
|
||
|
|
"valid_targets_mean": 5974.9,
|
||
|
|
"valid_targets_min": 893
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8667287977632805,
|
||
|
|
"grad_norm": 0.5610070145742456,
|
||
|
|
"learning_rate": 3.993325490007083e-05,
|
||
|
|
"loss": 0.338,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12349021434783936,
|
||
|
|
"step": 465,
|
||
|
|
"valid_targets_mean": 4138.9,
|
||
|
|
"valid_targets_min": 619
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8760484622553588,
|
||
|
|
"grad_norm": 0.6844738694880523,
|
||
|
|
"learning_rate": 3.992545960467353e-05,
|
||
|
|
"loss": 0.3738,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2368139624595642,
|
||
|
|
"step": 470,
|
||
|
|
"valid_targets_mean": 4311.0,
|
||
|
|
"valid_targets_min": 831
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8853681267474371,
|
||
|
|
"grad_norm": 0.5751791323614228,
|
||
|
|
"learning_rate": 3.9917234729996065e-05,
|
||
|
|
"loss": 0.3569,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1378014236688614,
|
||
|
|
"step": 475,
|
||
|
|
"valid_targets_mean": 3293.0,
|
||
|
|
"valid_targets_min": 589
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8946877912395154,
|
||
|
|
"grad_norm": 0.583182959498461,
|
||
|
|
"learning_rate": 3.990858045336111e-05,
|
||
|
|
"loss": 0.3609,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16992974281311035,
|
||
|
|
"step": 480,
|
||
|
|
"valid_targets_mean": 4164.1,
|
||
|
|
"valid_targets_min": 613
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9040074557315937,
|
||
|
|
"grad_norm": 0.6403424979255861,
|
||
|
|
"learning_rate": 3.989949696134894e-05,
|
||
|
|
"loss": 0.3205,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16153624653816223,
|
||
|
|
"step": 485,
|
||
|
|
"valid_targets_mean": 3746.4,
|
||
|
|
"valid_targets_min": 549
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9133271202236719,
|
||
|
|
"grad_norm": 0.6948822709888964,
|
||
|
|
"learning_rate": 3.988998444979345e-05,
|
||
|
|
"loss": 0.3496,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20749089121818542,
|
||
|
|
"step": 490,
|
||
|
|
"valid_targets_mean": 6376.2,
|
||
|
|
"valid_targets_min": 679
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9226467847157502,
|
||
|
|
"grad_norm": 0.5834516993934473,
|
||
|
|
"learning_rate": 3.988004312377786e-05,
|
||
|
|
"loss": 0.3546,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2187417447566986,
|
||
|
|
"step": 495,
|
||
|
|
"valid_targets_mean": 6500.9,
|
||
|
|
"valid_targets_min": 960
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9319664492078286,
|
||
|
|
"grad_norm": 0.535301297890541,
|
||
|
|
"learning_rate": 3.986967319763038e-05,
|
||
|
|
"loss": 0.3717,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11342230439186096,
|
||
|
|
"step": 500,
|
||
|
|
"valid_targets_mean": 3759.2,
|
||
|
|
"valid_targets_min": 492
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9412861136999068,
|
||
|
|
"grad_norm": 0.5983702475310047,
|
||
|
|
"learning_rate": 3.9858874894919516e-05,
|
||
|
|
"loss": 0.3474,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2573143243789673,
|
||
|
|
"step": 505,
|
||
|
|
"valid_targets_mean": 7248.2,
|
||
|
|
"valid_targets_min": 2153
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9506057781919851,
|
||
|
|
"grad_norm": 0.4709475167469776,
|
||
|
|
"learning_rate": 3.9847648448449274e-05,
|
||
|
|
"loss": 0.3287,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.151879221200943,
|
||
|
|
"step": 510,
|
||
|
|
"valid_targets_mean": 4898.1,
|
||
|
|
"valid_targets_min": 625
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9599254426840633,
|
||
|
|
"grad_norm": 0.570872080651508,
|
||
|
|
"learning_rate": 3.983599410025418e-05,
|
||
|
|
"loss": 0.3696,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2124394029378891,
|
||
|
|
"step": 515,
|
||
|
|
"valid_targets_mean": 6515.0,
|
||
|
|
"valid_targets_min": 669
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9692451071761417,
|
||
|
|
"grad_norm": 0.4694994835651028,
|
||
|
|
"learning_rate": 3.9823912101594e-05,
|
||
|
|
"loss": 0.3603,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12680339813232422,
|
||
|
|
"step": 520,
|
||
|
|
"valid_targets_mean": 5629.4,
|
||
|
|
"valid_targets_min": 924
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.97856477166822,
|
||
|
|
"grad_norm": 0.48332393826324077,
|
||
|
|
"learning_rate": 3.981140271294837e-05,
|
||
|
|
"loss": 0.3475,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17299839854240417,
|
||
|
|
"step": 525,
|
||
|
|
"valid_targets_mean": 7334.0,
|
||
|
|
"valid_targets_min": 891
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9878844361602982,
|
||
|
|
"grad_norm": 0.6619447063947715,
|
||
|
|
"learning_rate": 3.979846620401115e-05,
|
||
|
|
"loss": 0.359,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14597058296203613,
|
||
|
|
"step": 530,
|
||
|
|
"valid_targets_mean": 3610.5,
|
||
|
|
"valid_targets_min": 540
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9972041006523765,
|
||
|
|
"grad_norm": 0.5849320007817558,
|
||
|
|
"learning_rate": 3.9785102853684614e-05,
|
||
|
|
"loss": 0.3552,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18137356638908386,
|
||
|
|
"step": 535,
|
||
|
|
"valid_targets_mean": 4398.8,
|
||
|
|
"valid_targets_min": 588
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.005591798695247,
|
||
|
|
"grad_norm": 0.43363028927312053,
|
||
|
|
"learning_rate": 3.9771312950073464e-05,
|
||
|
|
"loss": 0.35,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1500912606716156,
|
||
|
|
"step": 540,
|
||
|
|
"valid_targets_mean": 8273.5,
|
||
|
|
"valid_targets_min": 475
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0149114631873253,
|
||
|
|
"grad_norm": 0.6985209161624261,
|
||
|
|
"learning_rate": 3.9757096790478585e-05,
|
||
|
|
"loss": 0.3341,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28995752334594727,
|
||
|
|
"step": 545,
|
||
|
|
"valid_targets_mean": 3989.1,
|
||
|
|
"valid_targets_min": 593
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0242311276794036,
|
||
|
|
"grad_norm": 0.6642090322238782,
|
||
|
|
"learning_rate": 3.974245468139066e-05,
|
||
|
|
"loss": 0.3544,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1888938993215561,
|
||
|
|
"step": 550,
|
||
|
|
"valid_targets_mean": 3940.4,
|
||
|
|
"valid_targets_min": 502
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0335507921714817,
|
||
|
|
"grad_norm": 0.5675769125928385,
|
||
|
|
"learning_rate": 3.972738693848354e-05,
|
||
|
|
"loss": 0.3362,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12539042532444,
|
||
|
|
"step": 555,
|
||
|
|
"valid_targets_mean": 3347.2,
|
||
|
|
"valid_targets_min": 775
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.04287045666356,
|
||
|
|
"grad_norm": 1.3235961608283708,
|
||
|
|
"learning_rate": 3.971189388660747e-05,
|
||
|
|
"loss": 0.3217,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17603641748428345,
|
||
|
|
"step": 560,
|
||
|
|
"valid_targets_mean": 7054.6,
|
||
|
|
"valid_targets_min": 4693
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0521901211556384,
|
||
|
|
"grad_norm": 0.5747441747449616,
|
||
|
|
"learning_rate": 3.9695975859782025e-05,
|
||
|
|
"loss": 0.3578,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15910890698432922,
|
||
|
|
"step": 565,
|
||
|
|
"valid_targets_mean": 5128.5,
|
||
|
|
"valid_targets_min": 1011
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0615097856477167,
|
||
|
|
"grad_norm": 0.48426730871725376,
|
||
|
|
"learning_rate": 3.9679633201188996e-05,
|
||
|
|
"loss": 0.3299,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1317993402481079,
|
||
|
|
"step": 570,
|
||
|
|
"valid_targets_mean": 4944.9,
|
||
|
|
"valid_targets_min": 599
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.070829450139795,
|
||
|
|
"grad_norm": 0.6036726732399292,
|
||
|
|
"learning_rate": 3.966286626316491e-05,
|
||
|
|
"loss": 0.3598,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22306659817695618,
|
||
|
|
"step": 575,
|
||
|
|
"valid_targets_mean": 4907.4,
|
||
|
|
"valid_targets_min": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0801491146318734,
|
||
|
|
"grad_norm": 0.549982571052518,
|
||
|
|
"learning_rate": 3.96456754071935e-05,
|
||
|
|
"loss": 0.3385,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12971454858779907,
|
||
|
|
"step": 580,
|
||
|
|
"valid_targets_mean": 4649.6,
|
||
|
|
"valid_targets_min": 647
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0894687791239515,
|
||
|
|
"grad_norm": 0.54490925245613,
|
||
|
|
"learning_rate": 3.962806100389785e-05,
|
||
|
|
"loss": 0.3365,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12257365882396698,
|
||
|
|
"step": 585,
|
||
|
|
"valid_targets_mean": 4001.2,
|
||
|
|
"valid_targets_min": 543
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0987884436160298,
|
||
|
|
"grad_norm": 0.6134397027587086,
|
||
|
|
"learning_rate": 3.961002343303245e-05,
|
||
|
|
"loss": 0.3303,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14924025535583496,
|
||
|
|
"step": 590,
|
||
|
|
"valid_targets_mean": 5355.5,
|
||
|
|
"valid_targets_min": 717
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1081081081081081,
|
||
|
|
"grad_norm": 0.5728026741448146,
|
||
|
|
"learning_rate": 3.9591563083474996e-05,
|
||
|
|
"loss": 0.349,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19922178983688354,
|
||
|
|
"step": 595,
|
||
|
|
"valid_targets_mean": 5504.0,
|
||
|
|
"valid_targets_min": 1088
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1174277726001864,
|
||
|
|
"grad_norm": 0.586450994670148,
|
||
|
|
"learning_rate": 3.9572680353217984e-05,
|
||
|
|
"loss": 0.3334,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22876465320587158,
|
||
|
|
"step": 600,
|
||
|
|
"valid_targets_mean": 6451.4,
|
||
|
|
"valid_targets_min": 867
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1267474370922648,
|
||
|
|
"grad_norm": 0.6436714669738806,
|
||
|
|
"learning_rate": 3.9553375649360175e-05,
|
||
|
|
"loss": 0.3116,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16142940521240234,
|
||
|
|
"step": 605,
|
||
|
|
"valid_targets_mean": 2945.1,
|
||
|
|
"valid_targets_min": 677
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1360671015843429,
|
||
|
|
"grad_norm": 0.6247728446337487,
|
||
|
|
"learning_rate": 3.953364938809777e-05,
|
||
|
|
"loss": 0.342,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11677239835262299,
|
||
|
|
"step": 610,
|
||
|
|
"valid_targets_mean": 3510.6,
|
||
|
|
"valid_targets_min": 550
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1453867660764212,
|
||
|
|
"grad_norm": 0.7684760065183371,
|
||
|
|
"learning_rate": 3.9513501994715476e-05,
|
||
|
|
"loss": 0.3426,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16630516946315765,
|
||
|
|
"step": 615,
|
||
|
|
"valid_targets_mean": 4120.8,
|
||
|
|
"valid_targets_min": 804
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1547064305684995,
|
||
|
|
"grad_norm": 0.6608260666678305,
|
||
|
|
"learning_rate": 3.9492933903577316e-05,
|
||
|
|
"loss": 0.3552,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14233127236366272,
|
||
|
|
"step": 620,
|
||
|
|
"valid_targets_mean": 2758.2,
|
||
|
|
"valid_targets_min": 536
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1640260950605779,
|
||
|
|
"grad_norm": 0.5571453067524983,
|
||
|
|
"learning_rate": 3.947194555811725e-05,
|
||
|
|
"loss": 0.3278,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16653169691562653,
|
||
|
|
"step": 625,
|
||
|
|
"valid_targets_mean": 4923.9,
|
||
|
|
"valid_targets_min": 752
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1733457595526562,
|
||
|
|
"grad_norm": 0.5820864700068743,
|
||
|
|
"learning_rate": 3.9450537410829674e-05,
|
||
|
|
"loss": 0.341,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20728662610054016,
|
||
|
|
"step": 630,
|
||
|
|
"valid_targets_mean": 4607.9,
|
||
|
|
"valid_targets_min": 432
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1826654240447343,
|
||
|
|
"grad_norm": 0.624432632114962,
|
||
|
|
"learning_rate": 3.9428709923259586e-05,
|
||
|
|
"loss": 0.3258,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20635253190994263,
|
||
|
|
"step": 635,
|
||
|
|
"valid_targets_mean": 4581.9,
|
||
|
|
"valid_targets_min": 674
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1919850885368126,
|
||
|
|
"grad_norm": 0.45123216340902716,
|
||
|
|
"learning_rate": 3.940646356599269e-05,
|
||
|
|
"loss": 0.3123,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15497656166553497,
|
||
|
|
"step": 640,
|
||
|
|
"valid_targets_mean": 7466.4,
|
||
|
|
"valid_targets_min": 2655
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.201304753028891,
|
||
|
|
"grad_norm": 0.5942472892259401,
|
||
|
|
"learning_rate": 3.9383798818645236e-05,
|
||
|
|
"loss": 0.329,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1420000046491623,
|
||
|
|
"step": 645,
|
||
|
|
"valid_targets_mean": 4790.5,
|
||
|
|
"valid_targets_min": 659
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2106244175209693,
|
||
|
|
"grad_norm": 0.5058169692207616,
|
||
|
|
"learning_rate": 3.936071616985366e-05,
|
||
|
|
"loss": 0.3217,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17437744140625,
|
||
|
|
"step": 650,
|
||
|
|
"valid_targets_mean": 5501.8,
|
||
|
|
"valid_targets_min": 526
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2199440820130476,
|
||
|
|
"grad_norm": 0.5113071884757048,
|
||
|
|
"learning_rate": 3.9337216117264106e-05,
|
||
|
|
"loss": 0.3266,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15600627660751343,
|
||
|
|
"step": 655,
|
||
|
|
"valid_targets_mean": 5536.9,
|
||
|
|
"valid_targets_min": 873
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2292637465051257,
|
||
|
|
"grad_norm": 0.789996401422049,
|
||
|
|
"learning_rate": 3.931329916752162e-05,
|
||
|
|
"loss": 0.3412,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11787241697311401,
|
||
|
|
"step": 660,
|
||
|
|
"valid_targets_mean": 2247.5,
|
||
|
|
"valid_targets_min": 778
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.238583410997204,
|
||
|
|
"grad_norm": 0.5009140820050977,
|
||
|
|
"learning_rate": 3.928896583625927e-05,
|
||
|
|
"loss": 0.368,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17678555846214294,
|
||
|
|
"step": 665,
|
||
|
|
"valid_targets_mean": 6132.1,
|
||
|
|
"valid_targets_min": 914
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2479030754892824,
|
||
|
|
"grad_norm": 0.6747826571373701,
|
||
|
|
"learning_rate": 3.926421664808706e-05,
|
||
|
|
"loss": 0.3366,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1372922658920288,
|
||
|
|
"step": 670,
|
||
|
|
"valid_targets_mean": 3936.2,
|
||
|
|
"valid_targets_min": 606
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2572227399813607,
|
||
|
|
"grad_norm": 0.44949575108064793,
|
||
|
|
"learning_rate": 3.9239052136580536e-05,
|
||
|
|
"loss": 0.3086,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15726375579833984,
|
||
|
|
"step": 675,
|
||
|
|
"valid_targets_mean": 6250.9,
|
||
|
|
"valid_targets_min": 583
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.266542404473439,
|
||
|
|
"grad_norm": 0.613789873141341,
|
||
|
|
"learning_rate": 3.921347284426935e-05,
|
||
|
|
"loss": 0.3344,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25219327211380005,
|
||
|
|
"step": 680,
|
||
|
|
"valid_targets_mean": 5632.4,
|
||
|
|
"valid_targets_min": 588
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2758620689655173,
|
||
|
|
"grad_norm": 0.5506226900699908,
|
||
|
|
"learning_rate": 3.918747932262558e-05,
|
||
|
|
"loss": 0.3129,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18796980381011963,
|
||
|
|
"step": 685,
|
||
|
|
"valid_targets_mean": 5110.0,
|
||
|
|
"valid_targets_min": 738
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2851817334575955,
|
||
|
|
"grad_norm": 0.5210386177006626,
|
||
|
|
"learning_rate": 3.916107213205174e-05,
|
||
|
|
"loss": 0.3299,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10750293731689453,
|
||
|
|
"step": 690,
|
||
|
|
"valid_targets_mean": 3022.9,
|
||
|
|
"valid_targets_min": 787
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2945013979496738,
|
||
|
|
"grad_norm": 0.5523726388732116,
|
||
|
|
"learning_rate": 3.9134251841868806e-05,
|
||
|
|
"loss": 0.3344,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.213568776845932,
|
||
|
|
"step": 695,
|
||
|
|
"valid_targets_mean": 6226.5,
|
||
|
|
"valid_targets_min": 4641
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.303821062441752,
|
||
|
|
"grad_norm": 0.5242573387661804,
|
||
|
|
"learning_rate": 3.91070190303039e-05,
|
||
|
|
"loss": 0.3484,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19567181169986725,
|
||
|
|
"step": 700,
|
||
|
|
"valid_targets_mean": 5893.1,
|
||
|
|
"valid_targets_min": 722
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3131407269338304,
|
||
|
|
"grad_norm": 0.5454687072478831,
|
||
|
|
"learning_rate": 3.9079374284477804e-05,
|
||
|
|
"loss": 0.3391,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1507263481616974,
|
||
|
|
"step": 705,
|
||
|
|
"valid_targets_mean": 3655.0,
|
||
|
|
"valid_targets_min": 549
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3224603914259085,
|
||
|
|
"grad_norm": 0.6044588461136069,
|
||
|
|
"learning_rate": 3.905131820039232e-05,
|
||
|
|
"loss": 0.3169,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12113272398710251,
|
||
|
|
"step": 710,
|
||
|
|
"valid_targets_mean": 3249.0,
|
||
|
|
"valid_targets_min": 624
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3317800559179869,
|
||
|
|
"grad_norm": 0.4905516634460778,
|
||
|
|
"learning_rate": 3.902285138291745e-05,
|
||
|
|
"loss": 0.3129,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14931780099868774,
|
||
|
|
"step": 715,
|
||
|
|
"valid_targets_mean": 5316.0,
|
||
|
|
"valid_targets_min": 573
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3410997204100652,
|
||
|
|
"grad_norm": 0.5730998632069458,
|
||
|
|
"learning_rate": 3.899397444577829e-05,
|
||
|
|
"loss": 0.3598,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20939984917640686,
|
||
|
|
"step": 720,
|
||
|
|
"valid_targets_mean": 5578.0,
|
||
|
|
"valid_targets_min": 644
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3504193849021435,
|
||
|
|
"grad_norm": 0.47210705512524664,
|
||
|
|
"learning_rate": 3.8964688011541864e-05,
|
||
|
|
"loss": 0.3305,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14459006488323212,
|
||
|
|
"step": 725,
|
||
|
|
"valid_targets_mean": 6343.6,
|
||
|
|
"valid_targets_min": 546
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3597390493942219,
|
||
|
|
"grad_norm": 0.5542098724577234,
|
||
|
|
"learning_rate": 3.893499271160366e-05,
|
||
|
|
"loss": 0.3256,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1409851461648941,
|
||
|
|
"step": 730,
|
||
|
|
"valid_targets_mean": 4484.9,
|
||
|
|
"valid_targets_min": 693
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3690587138863002,
|
||
|
|
"grad_norm": 0.5737211052493484,
|
||
|
|
"learning_rate": 3.890488918617403e-05,
|
||
|
|
"loss": 0.3168,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19839328527450562,
|
||
|
|
"step": 735,
|
||
|
|
"valid_targets_mean": 4829.2,
|
||
|
|
"valid_targets_min": 734
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3783783783783785,
|
||
|
|
"grad_norm": 0.5591401690331682,
|
||
|
|
"learning_rate": 3.887437808426439e-05,
|
||
|
|
"loss": 0.3105,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12328825891017914,
|
||
|
|
"step": 740,
|
||
|
|
"valid_targets_mean": 3544.5,
|
||
|
|
"valid_targets_min": 570
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3876980428704566,
|
||
|
|
"grad_norm": 0.49805559087260903,
|
||
|
|
"learning_rate": 3.884346006367324e-05,
|
||
|
|
"loss": 0.3162,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17507076263427734,
|
||
|
|
"step": 745,
|
||
|
|
"valid_targets_mean": 6099.1,
|
||
|
|
"valid_targets_min": 509
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.397017707362535,
|
||
|
|
"grad_norm": 0.5583471608773602,
|
||
|
|
"learning_rate": 3.8812135790971946e-05,
|
||
|
|
"loss": 0.3621,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17341488599777222,
|
||
|
|
"step": 750,
|
||
|
|
"valid_targets_mean": 5978.5,
|
||
|
|
"valid_targets_min": 815
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4063373718546133,
|
||
|
|
"grad_norm": 0.5147837121321843,
|
||
|
|
"learning_rate": 3.87804059414904e-05,
|
||
|
|
"loss": 0.3101,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14275220036506653,
|
||
|
|
"step": 755,
|
||
|
|
"valid_targets_mean": 4095.9,
|
||
|
|
"valid_targets_min": 794
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4156570363466916,
|
||
|
|
"grad_norm": 0.5185769725079494,
|
||
|
|
"learning_rate": 3.8748271199302465e-05,
|
||
|
|
"loss": 0.2968,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1769077628850937,
|
||
|
|
"step": 760,
|
||
|
|
"valid_targets_mean": 6859.1,
|
||
|
|
"valid_targets_min": 1558
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4249767008387697,
|
||
|
|
"grad_norm": 0.5428709156022019,
|
||
|
|
"learning_rate": 3.871573225721119e-05,
|
||
|
|
"loss": 0.3459,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18135955929756165,
|
||
|
|
"step": 765,
|
||
|
|
"valid_targets_mean": 6159.9,
|
||
|
|
"valid_targets_min": 584
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.434296365330848,
|
||
|
|
"grad_norm": 0.6720508008702379,
|
||
|
|
"learning_rate": 3.868278981673391e-05,
|
||
|
|
"loss": 0.3376,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09273894131183624,
|
||
|
|
"step": 770,
|
||
|
|
"valid_targets_mean": 1395.1,
|
||
|
|
"valid_targets_min": 596
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4436160298229264,
|
||
|
|
"grad_norm": 0.6699374087323592,
|
||
|
|
"learning_rate": 3.864944458808712e-05,
|
||
|
|
"loss": 0.3243,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10325013101100922,
|
||
|
|
"step": 775,
|
||
|
|
"valid_targets_mean": 2982.6,
|
||
|
|
"valid_targets_min": 592
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4529356943150047,
|
||
|
|
"grad_norm": 0.5171516817749161,
|
||
|
|
"learning_rate": 3.8615697290171135e-05,
|
||
|
|
"loss": 0.3124,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18036475777626038,
|
||
|
|
"step": 780,
|
||
|
|
"valid_targets_mean": 5915.0,
|
||
|
|
"valid_targets_min": 784
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.462255358807083,
|
||
|
|
"grad_norm": 0.6309432638911677,
|
||
|
|
"learning_rate": 3.858154865055461e-05,
|
||
|
|
"loss": 0.3126,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1574413776397705,
|
||
|
|
"step": 785,
|
||
|
|
"valid_targets_mean": 4905.1,
|
||
|
|
"valid_targets_min": 838
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4715750232991613,
|
||
|
|
"grad_norm": 0.7129786250801856,
|
||
|
|
"learning_rate": 3.8546999405458876e-05,
|
||
|
|
"loss": 0.3396,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29485881328582764,
|
||
|
|
"step": 790,
|
||
|
|
"valid_targets_mean": 4582.6,
|
||
|
|
"valid_targets_min": 567
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4808946877912395,
|
||
|
|
"grad_norm": 0.5159905861145383,
|
||
|
|
"learning_rate": 3.8512050299742005e-05,
|
||
|
|
"loss": 0.3187,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18507157266139984,
|
||
|
|
"step": 795,
|
||
|
|
"valid_targets_mean": 6410.5,
|
||
|
|
"valid_targets_min": 645
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4902143522833178,
|
||
|
|
"grad_norm": 0.6760436399385918,
|
||
|
|
"learning_rate": 3.847670208688282e-05,
|
||
|
|
"loss": 0.334,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17075665295124054,
|
||
|
|
"step": 800,
|
||
|
|
"valid_targets_mean": 4103.9,
|
||
|
|
"valid_targets_min": 365
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.499534016775396,
|
||
|
|
"grad_norm": 0.7543870220742888,
|
||
|
|
"learning_rate": 3.844095552896461e-05,
|
||
|
|
"loss": 0.329,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20135506987571716,
|
||
|
|
"step": 805,
|
||
|
|
"valid_targets_mean": 5593.0,
|
||
|
|
"valid_targets_min": 613
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5088536812674742,
|
||
|
|
"grad_norm": 0.5303278442838655,
|
||
|
|
"learning_rate": 3.840481139665872e-05,
|
||
|
|
"loss": 0.3102,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14774736762046814,
|
||
|
|
"step": 810,
|
||
|
|
"valid_targets_mean": 4865.1,
|
||
|
|
"valid_targets_min": 601
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5181733457595525,
|
||
|
|
"grad_norm": 0.5994077480263036,
|
||
|
|
"learning_rate": 3.836827046920791e-05,
|
||
|
|
"loss": 0.3186,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1644090712070465,
|
||
|
|
"step": 815,
|
||
|
|
"valid_targets_mean": 5103.2,
|
||
|
|
"valid_targets_min": 703
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5274930102516309,
|
||
|
|
"grad_norm": 0.527753970095113,
|
||
|
|
"learning_rate": 3.8331333534409594e-05,
|
||
|
|
"loss": 0.3126,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17244011163711548,
|
||
|
|
"step": 820,
|
||
|
|
"valid_targets_mean": 6099.6,
|
||
|
|
"valid_targets_min": 975
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5368126747437092,
|
||
|
|
"grad_norm": 0.5118693430880751,
|
||
|
|
"learning_rate": 3.8294001388598807e-05,
|
||
|
|
"loss": 0.3239,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16559144854545593,
|
||
|
|
"step": 825,
|
||
|
|
"valid_targets_mean": 6211.9,
|
||
|
|
"valid_targets_min": 490
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5461323392357875,
|
||
|
|
"grad_norm": 0.46846632404221605,
|
||
|
|
"learning_rate": 3.825627483663109e-05,
|
||
|
|
"loss": 0.3407,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13697263598442078,
|
||
|
|
"step": 830,
|
||
|
|
"valid_targets_mean": 7358.0,
|
||
|
|
"valid_targets_min": 856
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5554520037278659,
|
||
|
|
"grad_norm": 0.5232076902798352,
|
||
|
|
"learning_rate": 3.82181546918651e-05,
|
||
|
|
"loss": 0.3287,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1596425473690033,
|
||
|
|
"step": 835,
|
||
|
|
"valid_targets_mean": 4988.4,
|
||
|
|
"valid_targets_min": 650
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5647716682199442,
|
||
|
|
"grad_norm": 0.4198035118832552,
|
||
|
|
"learning_rate": 3.8179641776145067e-05,
|
||
|
|
"loss": 0.342,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17914628982543945,
|
||
|
|
"step": 840,
|
||
|
|
"valid_targets_mean": 7219.0,
|
||
|
|
"valid_targets_min": 613
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5740913327120225,
|
||
|
|
"grad_norm": 0.4942413218230838,
|
||
|
|
"learning_rate": 3.814073691978313e-05,
|
||
|
|
"loss": 0.3306,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2236957848072052,
|
||
|
|
"step": 845,
|
||
|
|
"valid_targets_mean": 6277.0,
|
||
|
|
"valid_targets_min": 588
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5834109972041006,
|
||
|
|
"grad_norm": 0.6638976147489284,
|
||
|
|
"learning_rate": 3.810144096154137e-05,
|
||
|
|
"loss": 0.3259,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1160690039396286,
|
||
|
|
"step": 850,
|
||
|
|
"valid_targets_mean": 1440.6,
|
||
|
|
"valid_targets_min": 563
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.592730661696179,
|
||
|
|
"grad_norm": 0.6021182296813001,
|
||
|
|
"learning_rate": 3.8061754748613766e-05,
|
||
|
|
"loss": 0.3474,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18292996287345886,
|
||
|
|
"step": 855,
|
||
|
|
"valid_targets_mean": 4387.2,
|
||
|
|
"valid_targets_min": 784
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6020503261882573,
|
||
|
|
"grad_norm": 0.5041550974070712,
|
||
|
|
"learning_rate": 3.802167913660794e-05,
|
||
|
|
"loss": 0.3112,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17557063698768616,
|
||
|
|
"step": 860,
|
||
|
|
"valid_targets_mean": 5462.0,
|
||
|
|
"valid_targets_min": 664
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6113699906803354,
|
||
|
|
"grad_norm": 0.5793518572042334,
|
||
|
|
"learning_rate": 3.7981214989526664e-05,
|
||
|
|
"loss": 0.3432,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1741921752691269,
|
||
|
|
"step": 865,
|
||
|
|
"valid_targets_mean": 4235.1,
|
||
|
|
"valid_targets_min": 817
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6206896551724137,
|
||
|
|
"grad_norm": 0.716125307895724,
|
||
|
|
"learning_rate": 3.7940363179749274e-05,
|
||
|
|
"loss": 0.3061,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15217386186122894,
|
||
|
|
"step": 870,
|
||
|
|
"valid_targets_mean": 4950.9,
|
||
|
|
"valid_targets_min": 286
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.630009319664492,
|
||
|
|
"grad_norm": 0.5310873940998737,
|
||
|
|
"learning_rate": 3.7899124588012844e-05,
|
||
|
|
"loss": 0.3146,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14660096168518066,
|
||
|
|
"step": 875,
|
||
|
|
"valid_targets_mean": 5473.9,
|
||
|
|
"valid_targets_min": 962
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6393289841565704,
|
||
|
|
"grad_norm": 0.5609298405623432,
|
||
|
|
"learning_rate": 3.785750010339321e-05,
|
||
|
|
"loss": 0.3,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1352173089981079,
|
||
|
|
"step": 880,
|
||
|
|
"valid_targets_mean": 4093.6,
|
||
|
|
"valid_targets_min": 841
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6486486486486487,
|
||
|
|
"grad_norm": 0.47131669523310027,
|
||
|
|
"learning_rate": 3.7815490623285776e-05,
|
||
|
|
"loss": 0.3089,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11407679319381714,
|
||
|
|
"step": 885,
|
||
|
|
"valid_targets_mean": 4251.2,
|
||
|
|
"valid_targets_min": 684
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.657968313140727,
|
||
|
|
"grad_norm": 0.5599375956672767,
|
||
|
|
"learning_rate": 3.7773097053386214e-05,
|
||
|
|
"loss": 0.3359,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1903909593820572,
|
||
|
|
"step": 890,
|
||
|
|
"valid_targets_mean": 4750.5,
|
||
|
|
"valid_targets_min": 565
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6672879776328053,
|
||
|
|
"grad_norm": 0.624655323750245,
|
||
|
|
"learning_rate": 3.7730320307670896e-05,
|
||
|
|
"loss": 0.324,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10537172853946686,
|
||
|
|
"step": 895,
|
||
|
|
"valid_targets_mean": 2529.1,
|
||
|
|
"valid_targets_min": 694
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6766076421248837,
|
||
|
|
"grad_norm": 0.4452862062274481,
|
||
|
|
"learning_rate": 3.7687161308377185e-05,
|
||
|
|
"loss": 0.3195,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15276660025119781,
|
||
|
|
"step": 900,
|
||
|
|
"valid_targets_mean": 5813.2,
|
||
|
|
"valid_targets_min": 622
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6859273066169618,
|
||
|
|
"grad_norm": 0.5523927476075841,
|
||
|
|
"learning_rate": 3.7643620985983594e-05,
|
||
|
|
"loss": 0.3523,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24211472272872925,
|
||
|
|
"step": 905,
|
||
|
|
"valid_targets_mean": 6134.5,
|
||
|
|
"valid_targets_min": 708
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.69524697110904,
|
||
|
|
"grad_norm": 0.4996953557298608,
|
||
|
|
"learning_rate": 3.759970027918969e-05,
|
||
|
|
"loss": 0.3252,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12529046833515167,
|
||
|
|
"step": 910,
|
||
|
|
"valid_targets_mean": 4631.6,
|
||
|
|
"valid_targets_min": 867
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7045666356011182,
|
||
|
|
"grad_norm": 0.49041059118253344,
|
||
|
|
"learning_rate": 3.7555400134895875e-05,
|
||
|
|
"loss": 0.3085,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1190144270658493,
|
||
|
|
"step": 915,
|
||
|
|
"valid_targets_mean": 4393.9,
|
||
|
|
"valid_targets_min": 438
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7138863000931965,
|
||
|
|
"grad_norm": 0.4916082642448217,
|
||
|
|
"learning_rate": 3.751072150818296e-05,
|
||
|
|
"loss": 0.3164,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1375523805618286,
|
||
|
|
"step": 920,
|
||
|
|
"valid_targets_mean": 4066.9,
|
||
|
|
"valid_targets_min": 781
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7232059645852749,
|
||
|
|
"grad_norm": 0.6000469861407015,
|
||
|
|
"learning_rate": 3.7465665362291564e-05,
|
||
|
|
"loss": 0.3391,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26054567098617554,
|
||
|
|
"step": 925,
|
||
|
|
"valid_targets_mean": 5831.6,
|
||
|
|
"valid_targets_min": 657
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7325256290773532,
|
||
|
|
"grad_norm": 0.6122867693208235,
|
||
|
|
"learning_rate": 3.742023266860139e-05,
|
||
|
|
"loss": 0.3246,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21604666113853455,
|
||
|
|
"step": 930,
|
||
|
|
"valid_targets_mean": 6167.9,
|
||
|
|
"valid_targets_min": 503
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7418452935694315,
|
||
|
|
"grad_norm": 0.6672921767879697,
|
||
|
|
"learning_rate": 3.737442440661023e-05,
|
||
|
|
"loss": 0.3208,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1308538168668747,
|
||
|
|
"step": 935,
|
||
|
|
"valid_targets_mean": 3656.5,
|
||
|
|
"valid_targets_min": 638
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7511649580615098,
|
||
|
|
"grad_norm": 0.4686503831015379,
|
||
|
|
"learning_rate": 3.7328241563912874e-05,
|
||
|
|
"loss": 0.3022,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15487657487392426,
|
||
|
|
"step": 940,
|
||
|
|
"valid_targets_mean": 6589.2,
|
||
|
|
"valid_targets_min": 889
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7604846225535882,
|
||
|
|
"grad_norm": 0.5088845871904579,
|
||
|
|
"learning_rate": 3.728168513617984e-05,
|
||
|
|
"loss": 0.2863,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13339266180992126,
|
||
|
|
"step": 945,
|
||
|
|
"valid_targets_mean": 4995.2,
|
||
|
|
"valid_targets_min": 508
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7698042870456665,
|
||
|
|
"grad_norm": 0.5275636860381216,
|
||
|
|
"learning_rate": 3.723475612713585e-05,
|
||
|
|
"loss": 0.3402,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19692593812942505,
|
||
|
|
"step": 950,
|
||
|
|
"valid_targets_mean": 5999.4,
|
||
|
|
"valid_targets_min": 395
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7791239515377446,
|
||
|
|
"grad_norm": 0.4730066021133068,
|
||
|
|
"learning_rate": 3.7187455548538245e-05,
|
||
|
|
"loss": 0.3213,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15709280967712402,
|
||
|
|
"step": 955,
|
||
|
|
"valid_targets_mean": 5757.8,
|
||
|
|
"valid_targets_min": 689
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.788443616029823,
|
||
|
|
"grad_norm": 0.6522715161923418,
|
||
|
|
"learning_rate": 3.7139784420155145e-05,
|
||
|
|
"loss": 0.3213,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15685850381851196,
|
||
|
|
"step": 960,
|
||
|
|
"valid_targets_mean": 5051.6,
|
||
|
|
"valid_targets_min": 696
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7977632805219013,
|
||
|
|
"grad_norm": 0.5097486836276304,
|
||
|
|
"learning_rate": 3.709174376974347e-05,
|
||
|
|
"loss": 0.3061,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1090770810842514,
|
||
|
|
"step": 965,
|
||
|
|
"valid_targets_mean": 3604.4,
|
||
|
|
"valid_targets_min": 588
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8070829450139794,
|
||
|
|
"grad_norm": 0.5739396130125296,
|
||
|
|
"learning_rate": 3.7043334633026785e-05,
|
||
|
|
"loss": 0.3261,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1537359654903412,
|
||
|
|
"step": 970,
|
||
|
|
"valid_targets_mean": 3502.8,
|
||
|
|
"valid_targets_min": 765
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8164026095060577,
|
||
|
|
"grad_norm": 0.7029508584970757,
|
||
|
|
"learning_rate": 3.699455805367296e-05,
|
||
|
|
"loss": 0.3118,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14871211349964142,
|
||
|
|
"step": 975,
|
||
|
|
"valid_targets_mean": 4196.9,
|
||
|
|
"valid_targets_min": 714
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.825722273998136,
|
||
|
|
"grad_norm": 0.4645454504105169,
|
||
|
|
"learning_rate": 3.69454150832717e-05,
|
||
|
|
"loss": 0.3084,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13736394047737122,
|
||
|
|
"step": 980,
|
||
|
|
"valid_targets_mean": 5685.4,
|
||
|
|
"valid_targets_min": 862
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8350419384902144,
|
||
|
|
"grad_norm": 0.5214199973559711,
|
||
|
|
"learning_rate": 3.689590678131181e-05,
|
||
|
|
"loss": 0.2974,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1477317214012146,
|
||
|
|
"step": 985,
|
||
|
|
"valid_targets_mean": 4724.9,
|
||
|
|
"valid_targets_min": 838
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8443616029822927,
|
||
|
|
"grad_norm": 0.5239880760179282,
|
||
|
|
"learning_rate": 3.684603421515844e-05,
|
||
|
|
"loss": 0.3295,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19311471283435822,
|
||
|
|
"step": 990,
|
||
|
|
"valid_targets_mean": 5576.5,
|
||
|
|
"valid_targets_min": 527
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.853681267474371,
|
||
|
|
"grad_norm": 0.46647588839942317,
|
||
|
|
"learning_rate": 3.6795798460029996e-05,
|
||
|
|
"loss": 0.322,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11346167325973511,
|
||
|
|
"step": 995,
|
||
|
|
"valid_targets_mean": 4905.8,
|
||
|
|
"valid_targets_min": 578
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8630009319664493,
|
||
|
|
"grad_norm": 0.7049309109318336,
|
||
|
|
"learning_rate": 3.6745200598974987e-05,
|
||
|
|
"loss": 0.3101,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17578113079071045,
|
||
|
|
"step": 1000,
|
||
|
|
"valid_targets_mean": 4574.0,
|
||
|
|
"valid_targets_min": 762
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8723205964585274,
|
||
|
|
"grad_norm": 0.6466046684134269,
|
||
|
|
"learning_rate": 3.66942417228487e-05,
|
||
|
|
"loss": 0.3386,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13234451413154602,
|
||
|
|
"step": 1005,
|
||
|
|
"valid_targets_mean": 3809.0,
|
||
|
|
"valid_targets_min": 945
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8816402609506058,
|
||
|
|
"grad_norm": 0.6010100369352087,
|
||
|
|
"learning_rate": 3.664292293028964e-05,
|
||
|
|
"loss": 0.331,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16403965651988983,
|
||
|
|
"step": 1010,
|
||
|
|
"valid_targets_mean": 5981.1,
|
||
|
|
"valid_targets_min": 743
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.890959925442684,
|
||
|
|
"grad_norm": 0.5196718950815074,
|
||
|
|
"learning_rate": 3.659124532769587e-05,
|
||
|
|
"loss": 0.2988,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15700331330299377,
|
||
|
|
"step": 1015,
|
||
|
|
"valid_targets_mean": 5260.5,
|
||
|
|
"valid_targets_min": 690
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9002795899347622,
|
||
|
|
"grad_norm": 0.6406857500820201,
|
||
|
|
"learning_rate": 3.6539210029201174e-05,
|
||
|
|
"loss": 0.3201,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19720548391342163,
|
||
|
|
"step": 1020,
|
||
|
|
"valid_targets_mean": 4249.5,
|
||
|
|
"valid_targets_min": 710
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9095992544268405,
|
||
|
|
"grad_norm": 0.6546161187486346,
|
||
|
|
"learning_rate": 3.6486818156650985e-05,
|
||
|
|
"loss": 0.3339,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30151045322418213,
|
||
|
|
"step": 1025,
|
||
|
|
"valid_targets_mean": 5563.1,
|
||
|
|
"valid_targets_min": 541
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9189189189189189,
|
||
|
|
"grad_norm": 0.7047046673369397,
|
||
|
|
"learning_rate": 3.643407083957823e-05,
|
||
|
|
"loss": 0.3084,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17458941042423248,
|
||
|
|
"step": 1030,
|
||
|
|
"valid_targets_mean": 4984.8,
|
||
|
|
"valid_targets_min": 1056
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9282385834109972,
|
||
|
|
"grad_norm": 0.5287139427665362,
|
||
|
|
"learning_rate": 3.6380969215178994e-05,
|
||
|
|
"loss": 0.3144,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15954937040805817,
|
||
|
|
"step": 1035,
|
||
|
|
"valid_targets_mean": 5800.4,
|
||
|
|
"valid_targets_min": 802
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9375582479030755,
|
||
|
|
"grad_norm": 0.8600973668352561,
|
||
|
|
"learning_rate": 3.6327514428287985e-05,
|
||
|
|
"loss": 0.339,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1817929446697235,
|
||
|
|
"step": 1040,
|
||
|
|
"valid_targets_mean": 4888.4,
|
||
|
|
"valid_targets_min": 463
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9468779123951538,
|
||
|
|
"grad_norm": 0.4815461383636214,
|
||
|
|
"learning_rate": 3.627370763135384e-05,
|
||
|
|
"loss": 0.2934,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.149854838848114,
|
||
|
|
"step": 1045,
|
||
|
|
"valid_targets_mean": 6049.5,
|
||
|
|
"valid_targets_min": 959
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9561975768872322,
|
||
|
|
"grad_norm": 0.5458847480325585,
|
||
|
|
"learning_rate": 3.6219549984414293e-05,
|
||
|
|
"loss": 0.296,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12161937355995178,
|
||
|
|
"step": 1050,
|
||
|
|
"valid_targets_mean": 3814.2,
|
||
|
|
"valid_targets_min": 657
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9655172413793105,
|
||
|
|
"grad_norm": 0.5383421343286,
|
||
|
|
"learning_rate": 3.616504265507116e-05,
|
||
|
|
"loss": 0.3271,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14529967308044434,
|
||
|
|
"step": 1055,
|
||
|
|
"valid_targets_mean": 6147.9,
|
||
|
|
"valid_targets_min": 5629
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9748369058713886,
|
||
|
|
"grad_norm": 0.530079646398552,
|
||
|
|
"learning_rate": 3.61101868184652e-05,
|
||
|
|
"loss": 0.3104,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16821002960205078,
|
||
|
|
"step": 1060,
|
||
|
|
"valid_targets_mean": 5541.6,
|
||
|
|
"valid_targets_min": 784
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.984156570363467,
|
||
|
|
"grad_norm": 0.5746914310113385,
|
||
|
|
"learning_rate": 3.605498365725073e-05,
|
||
|
|
"loss": 0.2978,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11010649055242538,
|
||
|
|
"step": 1065,
|
||
|
|
"valid_targets_mean": 3736.4,
|
||
|
|
"valid_targets_min": 628
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.993476234855545,
|
||
|
|
"grad_norm": 0.5567626388134503,
|
||
|
|
"learning_rate": 3.599943436157012e-05,
|
||
|
|
"loss": 0.3273,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17329010367393494,
|
||
|
|
"step": 1070,
|
||
|
|
"valid_targets_mean": 4686.0,
|
||
|
|
"valid_targets_min": 682
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0018639328984156,
|
||
|
|
"grad_norm": 0.7689294545168546,
|
||
|
|
"learning_rate": 3.594354012902821e-05,
|
||
|
|
"loss": 0.3123,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09438365697860718,
|
||
|
|
"step": 1075,
|
||
|
|
"valid_targets_mean": 2043.1,
|
||
|
|
"valid_targets_min": 626
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.011183597390494,
|
||
|
|
"grad_norm": 0.540501706753981,
|
||
|
|
"learning_rate": 3.588730216466642e-05,
|
||
|
|
"loss": 0.2968,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14411693811416626,
|
||
|
|
"step": 1080,
|
||
|
|
"valid_targets_mean": 4466.2,
|
||
|
|
"valid_targets_min": 1030
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0205032618825722,
|
||
|
|
"grad_norm": 0.782787831432509,
|
||
|
|
"learning_rate": 3.58307216809368e-05,
|
||
|
|
"loss": 0.2839,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1342252939939499,
|
||
|
|
"step": 1085,
|
||
|
|
"valid_targets_mean": 3723.0,
|
||
|
|
"valid_targets_min": 690
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0298229263746506,
|
||
|
|
"grad_norm": 0.5775233131027144,
|
||
|
|
"learning_rate": 3.5773799897675865e-05,
|
||
|
|
"loss": 0.2974,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1644313782453537,
|
||
|
|
"step": 1090,
|
||
|
|
"valid_targets_mean": 5271.2,
|
||
|
|
"valid_targets_min": 636
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.039142590866729,
|
||
|
|
"grad_norm": 0.6389493169234576,
|
||
|
|
"learning_rate": 3.5716538042078333e-05,
|
||
|
|
"loss": 0.3025,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10115222632884979,
|
||
|
|
"step": 1095,
|
||
|
|
"valid_targets_mean": 2417.4,
|
||
|
|
"valid_targets_min": 644
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.048462255358807,
|
||
|
|
"grad_norm": 0.61845216492274,
|
||
|
|
"learning_rate": 3.565893734867065e-05,
|
||
|
|
"loss": 0.3091,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1815635859966278,
|
||
|
|
"step": 1100,
|
||
|
|
"valid_targets_mean": 4211.2,
|
||
|
|
"valid_targets_min": 395
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0577819198508855,
|
||
|
|
"grad_norm": 0.5938421819404118,
|
||
|
|
"learning_rate": 3.560099905928437e-05,
|
||
|
|
"loss": 0.327,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14733126759529114,
|
||
|
|
"step": 1105,
|
||
|
|
"valid_targets_mean": 5618.6,
|
||
|
|
"valid_targets_min": 857
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0671015843429634,
|
||
|
|
"grad_norm": 0.5634283495637461,
|
||
|
|
"learning_rate": 3.554272442302936e-05,
|
||
|
|
"loss": 0.2901,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13202795386314392,
|
||
|
|
"step": 1110,
|
||
|
|
"valid_targets_mean": 8129.0,
|
||
|
|
"valid_targets_min": 814
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0764212488350418,
|
||
|
|
"grad_norm": 0.5802031178829163,
|
||
|
|
"learning_rate": 3.548411469626694e-05,
|
||
|
|
"loss": 0.3413,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1378626525402069,
|
||
|
|
"step": 1115,
|
||
|
|
"valid_targets_mean": 4162.9,
|
||
|
|
"valid_targets_min": 596
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.08574091332712,
|
||
|
|
"grad_norm": 0.5670261263769482,
|
||
|
|
"learning_rate": 3.5425171142582725e-05,
|
||
|
|
"loss": 0.2922,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15551522374153137,
|
||
|
|
"step": 1120,
|
||
|
|
"valid_targets_mean": 4608.6,
|
||
|
|
"valid_targets_min": 630
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0950605778191984,
|
||
|
|
"grad_norm": 0.6419211267762933,
|
||
|
|
"learning_rate": 3.536589503275941e-05,
|
||
|
|
"loss": 0.3171,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13013221323490143,
|
||
|
|
"step": 1125,
|
||
|
|
"valid_targets_mean": 3345.0,
|
||
|
|
"valid_targets_min": 613
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1043802423112767,
|
||
|
|
"grad_norm": 0.49733716646449716,
|
||
|
|
"learning_rate": 3.53062876447494e-05,
|
||
|
|
"loss": 0.3013,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15403276681900024,
|
||
|
|
"step": 1130,
|
||
|
|
"valid_targets_mean": 5789.8,
|
||
|
|
"valid_targets_min": 459
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.113699906803355,
|
||
|
|
"grad_norm": 0.4662100180832762,
|
||
|
|
"learning_rate": 3.5246350263647175e-05,
|
||
|
|
"loss": 0.2959,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13433662056922913,
|
||
|
|
"step": 1135,
|
||
|
|
"valid_targets_mean": 5245.5,
|
||
|
|
"valid_targets_min": 975
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1230195712954334,
|
||
|
|
"grad_norm": 0.5525036948705264,
|
||
|
|
"learning_rate": 3.518608418166169e-05,
|
||
|
|
"loss": 0.3043,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1270814836025238,
|
||
|
|
"step": 1140,
|
||
|
|
"valid_targets_mean": 4083.5,
|
||
|
|
"valid_targets_min": 804
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1323392357875117,
|
||
|
|
"grad_norm": 0.5924601063350385,
|
||
|
|
"learning_rate": 3.512549069808846e-05,
|
||
|
|
"loss": 0.2928,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10555174946784973,
|
||
|
|
"step": 1145,
|
||
|
|
"valid_targets_mean": 3731.4,
|
||
|
|
"valid_targets_min": 506
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.14165890027959,
|
||
|
|
"grad_norm": 0.5186984108313448,
|
||
|
|
"learning_rate": 3.5064571119281535e-05,
|
||
|
|
"loss": 0.3021,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1472579538822174,
|
||
|
|
"step": 1150,
|
||
|
|
"valid_targets_mean": 5399.6,
|
||
|
|
"valid_targets_min": 596
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1509785647716684,
|
||
|
|
"grad_norm": 0.5556634400122727,
|
||
|
|
"learning_rate": 3.500332675862537e-05,
|
||
|
|
"loss": 0.3008,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14503410458564758,
|
||
|
|
"step": 1155,
|
||
|
|
"valid_targets_mean": 4572.8,
|
||
|
|
"valid_targets_min": 564
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1602982292637467,
|
||
|
|
"grad_norm": 0.5408178980685083,
|
||
|
|
"learning_rate": 3.4941758936506484e-05,
|
||
|
|
"loss": 0.3069,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07900405675172806,
|
||
|
|
"step": 1160,
|
||
|
|
"valid_targets_mean": 2506.4,
|
||
|
|
"valid_targets_min": 540
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1696178937558246,
|
||
|
|
"grad_norm": 0.5803311841565066,
|
||
|
|
"learning_rate": 3.4879868980285017e-05,
|
||
|
|
"loss": 0.2695,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15611647069454193,
|
||
|
|
"step": 1165,
|
||
|
|
"valid_targets_mean": 4209.6,
|
||
|
|
"valid_targets_min": 595
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.178937558247903,
|
||
|
|
"grad_norm": 0.5670362113872505,
|
||
|
|
"learning_rate": 3.481765822426609e-05,
|
||
|
|
"loss": 0.3043,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18629573285579681,
|
||
|
|
"step": 1170,
|
||
|
|
"valid_targets_mean": 5123.9,
|
||
|
|
"valid_targets_min": 595
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1882572227399812,
|
||
|
|
"grad_norm": 0.5997845522286626,
|
||
|
|
"learning_rate": 3.4755128009671055e-05,
|
||
|
|
"loss": 0.3066,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12389770150184631,
|
||
|
|
"step": 1175,
|
||
|
|
"valid_targets_mean": 3379.6,
|
||
|
|
"valid_targets_min": 635
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1975768872320596,
|
||
|
|
"grad_norm": 0.9724274005581592,
|
||
|
|
"learning_rate": 3.4692279684608554e-05,
|
||
|
|
"loss": 0.3004,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11610507220029831,
|
||
|
|
"step": 1180,
|
||
|
|
"valid_targets_mean": 6035.0,
|
||
|
|
"valid_targets_min": 615
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.206896551724138,
|
||
|
|
"grad_norm": 0.376935091728397,
|
||
|
|
"learning_rate": 3.46291146040455e-05,
|
||
|
|
"loss": 0.2645,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10374706983566284,
|
||
|
|
"step": 1185,
|
||
|
|
"valid_targets_mean": 7277.0,
|
||
|
|
"valid_targets_min": 282
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2162162162162162,
|
||
|
|
"grad_norm": 0.6606426186708854,
|
||
|
|
"learning_rate": 3.456563412977783e-05,
|
||
|
|
"loss": 0.3096,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15126419067382812,
|
||
|
|
"step": 1190,
|
||
|
|
"valid_targets_mean": 3607.6,
|
||
|
|
"valid_targets_min": 753
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2255358807082946,
|
||
|
|
"grad_norm": 0.5080746156718853,
|
||
|
|
"learning_rate": 3.4501839630401136e-05,
|
||
|
|
"loss": 0.294,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16678392887115479,
|
||
|
|
"step": 1195,
|
||
|
|
"valid_targets_mean": 5061.5,
|
||
|
|
"valid_targets_min": 644
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.234855545200373,
|
||
|
|
"grad_norm": 0.537667797345181,
|
||
|
|
"learning_rate": 3.443773248128119e-05,
|
||
|
|
"loss": 0.3025,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17888835072517395,
|
||
|
|
"step": 1200,
|
||
|
|
"valid_targets_mean": 5744.6,
|
||
|
|
"valid_targets_min": 669
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.244175209692451,
|
||
|
|
"grad_norm": 0.5858445968190366,
|
||
|
|
"learning_rate": 3.437331406452429e-05,
|
||
|
|
"loss": 0.2742,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14848342537879944,
|
||
|
|
"step": 1205,
|
||
|
|
"valid_targets_mean": 4501.0,
|
||
|
|
"valid_targets_min": 620
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2534948741845295,
|
||
|
|
"grad_norm": 0.48481597269700144,
|
||
|
|
"learning_rate": 3.4308585768947424e-05,
|
||
|
|
"loss": 0.2853,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10843102633953094,
|
||
|
|
"step": 1210,
|
||
|
|
"valid_targets_mean": 4104.1,
|
||
|
|
"valid_targets_min": 843
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.262814538676608,
|
||
|
|
"grad_norm": 0.48700378537851235,
|
||
|
|
"learning_rate": 3.424354899004839e-05,
|
||
|
|
"loss": 0.2975,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13847170770168304,
|
||
|
|
"step": 1215,
|
||
|
|
"valid_targets_mean": 5362.1,
|
||
|
|
"valid_targets_min": 576
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2721342031686858,
|
||
|
|
"grad_norm": 0.5442025877600544,
|
||
|
|
"learning_rate": 3.417820512997564e-05,
|
||
|
|
"loss": 0.3044,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14482352137565613,
|
||
|
|
"step": 1220,
|
||
|
|
"valid_targets_mean": 5346.2,
|
||
|
|
"valid_targets_min": 687
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.281453867660764,
|
||
|
|
"grad_norm": 0.5338323794605577,
|
||
|
|
"learning_rate": 3.411255559749811e-05,
|
||
|
|
"loss": 0.3067,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17702502012252808,
|
||
|
|
"step": 1225,
|
||
|
|
"valid_targets_mean": 6127.4,
|
||
|
|
"valid_targets_min": 838
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2907735321528424,
|
||
|
|
"grad_norm": 0.5603226526718844,
|
||
|
|
"learning_rate": 3.404660180797481e-05,
|
||
|
|
"loss": 0.2841,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13995474576950073,
|
||
|
|
"step": 1230,
|
||
|
|
"valid_targets_mean": 4215.9,
|
||
|
|
"valid_targets_min": 718
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3000931966449207,
|
||
|
|
"grad_norm": 0.5696120166424857,
|
||
|
|
"learning_rate": 3.3980345183324344e-05,
|
||
|
|
"loss": 0.3128,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19453905522823334,
|
||
|
|
"step": 1235,
|
||
|
|
"valid_targets_mean": 6069.8,
|
||
|
|
"valid_targets_min": 867
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.309412861136999,
|
||
|
|
"grad_norm": 0.6499865364279552,
|
||
|
|
"learning_rate": 3.391378715199419e-05,
|
||
|
|
"loss": 0.291,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1384582817554474,
|
||
|
|
"step": 1240,
|
||
|
|
"valid_targets_mean": 4951.4,
|
||
|
|
"valid_targets_min": 889
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3187325256290774,
|
||
|
|
"grad_norm": 0.5826705757395046,
|
||
|
|
"learning_rate": 3.384692914893002e-05,
|
||
|
|
"loss": 0.2975,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12952333688735962,
|
||
|
|
"step": 1245,
|
||
|
|
"valid_targets_mean": 2905.0,
|
||
|
|
"valid_targets_min": 590
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3280521901211557,
|
||
|
|
"grad_norm": 0.4822889385112366,
|
||
|
|
"learning_rate": 3.377977261554462e-05,
|
||
|
|
"loss": 0.2775,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10987751930952072,
|
||
|
|
"step": 1250,
|
||
|
|
"valid_targets_mean": 4208.8,
|
||
|
|
"valid_targets_min": 576
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.337371854613234,
|
||
|
|
"grad_norm": 0.6303045948453817,
|
||
|
|
"learning_rate": 3.3712318999686914e-05,
|
||
|
|
"loss": 0.3093,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16873258352279663,
|
||
|
|
"step": 1255,
|
||
|
|
"valid_targets_mean": 4039.8,
|
||
|
|
"valid_targets_min": 497
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3466915191053124,
|
||
|
|
"grad_norm": 0.5257802333158621,
|
||
|
|
"learning_rate": 3.3644569755610745e-05,
|
||
|
|
"loss": 0.3017,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18518994748592377,
|
||
|
|
"step": 1260,
|
||
|
|
"valid_targets_mean": 6295.6,
|
||
|
|
"valid_targets_min": 548
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3560111835973903,
|
||
|
|
"grad_norm": 0.4795720032885019,
|
||
|
|
"learning_rate": 3.357652634394345e-05,
|
||
|
|
"loss": 0.2958,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12140154838562012,
|
||
|
|
"step": 1265,
|
||
|
|
"valid_targets_mean": 4020.0,
|
||
|
|
"valid_targets_min": 786
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3653308480894686,
|
||
|
|
"grad_norm": 0.477717719334079,
|
||
|
|
"learning_rate": 3.350819023165446e-05,
|
||
|
|
"loss": 0.2858,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16924908757209778,
|
||
|
|
"step": 1270,
|
||
|
|
"valid_targets_mean": 6727.4,
|
||
|
|
"valid_targets_min": 822
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.374650512581547,
|
||
|
|
"grad_norm": 0.44839860708291224,
|
||
|
|
"learning_rate": 3.343956289202361e-05,
|
||
|
|
"loss": 0.3062,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20942535996437073,
|
||
|
|
"step": 1275,
|
||
|
|
"valid_targets_mean": 8636.0,
|
||
|
|
"valid_targets_min": 2001
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3839701770736252,
|
||
|
|
"grad_norm": 0.48647022036067333,
|
||
|
|
"learning_rate": 3.33706458046094e-05,
|
||
|
|
"loss": 0.3088,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10708896815776825,
|
||
|
|
"step": 1280,
|
||
|
|
"valid_targets_mean": 4129.6,
|
||
|
|
"valid_targets_min": 659
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3932898415657036,
|
||
|
|
"grad_norm": 0.49526987594606975,
|
||
|
|
"learning_rate": 3.33014404552171e-05,
|
||
|
|
"loss": 0.3023,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12043149024248123,
|
||
|
|
"step": 1285,
|
||
|
|
"valid_targets_mean": 5282.8,
|
||
|
|
"valid_targets_min": 838
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.402609506057782,
|
||
|
|
"grad_norm": 0.5017831716790211,
|
||
|
|
"learning_rate": 3.323194833586669e-05,
|
||
|
|
"loss": 0.3071,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20298132300376892,
|
||
|
|
"step": 1290,
|
||
|
|
"valid_targets_mean": 6172.1,
|
||
|
|
"valid_targets_min": 761
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4119291705498602,
|
||
|
|
"grad_norm": 0.4165506370836097,
|
||
|
|
"learning_rate": 3.316217094476076e-05,
|
||
|
|
"loss": 0.2873,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11466651409864426,
|
||
|
|
"step": 1295,
|
||
|
|
"valid_targets_mean": 6480.1,
|
||
|
|
"valid_targets_min": 642
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4212488350419386,
|
||
|
|
"grad_norm": 0.6732478616819767,
|
||
|
|
"learning_rate": 3.3092109786252105e-05,
|
||
|
|
"loss": 0.3056,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16846883296966553,
|
||
|
|
"step": 1300,
|
||
|
|
"valid_targets_mean": 3246.2,
|
||
|
|
"valid_targets_min": 606
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.430568499534017,
|
||
|
|
"grad_norm": 0.5188585258168104,
|
||
|
|
"learning_rate": 3.3021766370811403e-05,
|
||
|
|
"loss": 0.3195,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15453115105628967,
|
||
|
|
"step": 1305,
|
||
|
|
"valid_targets_mean": 5965.8,
|
||
|
|
"valid_targets_min": 606
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.439888164026095,
|
||
|
|
"grad_norm": 0.4624042832371253,
|
||
|
|
"learning_rate": 3.2951142214994565e-05,
|
||
|
|
"loss": 0.2842,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15097807347774506,
|
||
|
|
"step": 1310,
|
||
|
|
"valid_targets_mean": 6282.1,
|
||
|
|
"valid_targets_min": 963
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4492078285181735,
|
||
|
|
"grad_norm": 0.512672294103711,
|
||
|
|
"learning_rate": 3.2880238841410086e-05,
|
||
|
|
"loss": 0.2778,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13124805688858032,
|
||
|
|
"step": 1315,
|
||
|
|
"valid_targets_mean": 4953.6,
|
||
|
|
"valid_targets_min": 923
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4585274930102514,
|
||
|
|
"grad_norm": 0.5049249460110912,
|
||
|
|
"learning_rate": 3.280905777868621e-05,
|
||
|
|
"loss": 0.3032,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13191038370132446,
|
||
|
|
"step": 1320,
|
||
|
|
"valid_targets_mean": 5471.6,
|
||
|
|
"valid_targets_min": 565
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4678471575023297,
|
||
|
|
"grad_norm": 0.5474248447764726,
|
||
|
|
"learning_rate": 3.273760056143795e-05,
|
||
|
|
"loss": 0.2962,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13751721382141113,
|
||
|
|
"step": 1325,
|
||
|
|
"valid_targets_mean": 4946.1,
|
||
|
|
"valid_targets_min": 998
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.477166821994408,
|
||
|
|
"grad_norm": 0.5025585506115298,
|
||
|
|
"learning_rate": 3.266586873023404e-05,
|
||
|
|
"loss": 0.3039,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1211470291018486,
|
||
|
|
"step": 1330,
|
||
|
|
"valid_targets_mean": 4520.5,
|
||
|
|
"valid_targets_min": 757
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4864864864864864,
|
||
|
|
"grad_norm": 0.5368634371439516,
|
||
|
|
"learning_rate": 3.259386383156369e-05,
|
||
|
|
"loss": 0.2919,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11575411260128021,
|
||
|
|
"step": 1335,
|
||
|
|
"valid_targets_mean": 5470.5,
|
||
|
|
"valid_targets_min": 650
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4958061509785647,
|
||
|
|
"grad_norm": 0.595034223658956,
|
||
|
|
"learning_rate": 3.252158741780328e-05,
|
||
|
|
"loss": 0.284,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1273195743560791,
|
||
|
|
"step": 1340,
|
||
|
|
"valid_targets_mean": 3258.5,
|
||
|
|
"valid_targets_min": 728
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.505125815470643,
|
||
|
|
"grad_norm": 0.5277178557013645,
|
||
|
|
"learning_rate": 3.244904104718284e-05,
|
||
|
|
"loss": 0.281,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14186760783195496,
|
||
|
|
"step": 1345,
|
||
|
|
"valid_targets_mean": 5108.2,
|
||
|
|
"valid_targets_min": 845
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5144454799627214,
|
||
|
|
"grad_norm": 0.5142855931923779,
|
||
|
|
"learning_rate": 3.23762262837525e-05,
|
||
|
|
"loss": 0.2813,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13558357954025269,
|
||
|
|
"step": 1350,
|
||
|
|
"valid_targets_mean": 5724.4,
|
||
|
|
"valid_targets_min": 716
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5237651444547997,
|
||
|
|
"grad_norm": 0.5643724377012218,
|
||
|
|
"learning_rate": 3.230314469734877e-05,
|
||
|
|
"loss": 0.3035,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12268234044313431,
|
||
|
|
"step": 1355,
|
||
|
|
"valid_targets_mean": 3152.4,
|
||
|
|
"valid_targets_min": 694
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.533084808946878,
|
||
|
|
"grad_norm": 0.5445224694099998,
|
||
|
|
"learning_rate": 3.222979786356064e-05,
|
||
|
|
"loss": 0.3292,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14697018265724182,
|
||
|
|
"step": 1360,
|
||
|
|
"valid_targets_mean": 4632.9,
|
||
|
|
"valid_targets_min": 503
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.542404473438956,
|
||
|
|
"grad_norm": 0.495361487650719,
|
||
|
|
"learning_rate": 3.21561873636957e-05,
|
||
|
|
"loss": 0.2865,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12937918305397034,
|
||
|
|
"step": 1365,
|
||
|
|
"valid_targets_mean": 4836.1,
|
||
|
|
"valid_targets_min": 1079
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5517241379310347,
|
||
|
|
"grad_norm": 0.5423292142943388,
|
||
|
|
"learning_rate": 3.208231478474596e-05,
|
||
|
|
"loss": 0.3074,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1531848907470703,
|
||
|
|
"step": 1370,
|
||
|
|
"valid_targets_mean": 5012.9,
|
||
|
|
"valid_targets_min": 568
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5610438024231126,
|
||
|
|
"grad_norm": 0.575417686245814,
|
||
|
|
"learning_rate": 3.200818171935371e-05,
|
||
|
|
"loss": 0.2939,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.185256227850914,
|
||
|
|
"step": 1375,
|
||
|
|
"valid_targets_mean": 5524.9,
|
||
|
|
"valid_targets_min": 703
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.570363466915191,
|
||
|
|
"grad_norm": 0.5631354985458626,
|
||
|
|
"learning_rate": 3.193378976577712e-05,
|
||
|
|
"loss": 0.3006,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12874957919120789,
|
||
|
|
"step": 1380,
|
||
|
|
"valid_targets_mean": 3601.6,
|
||
|
|
"valid_targets_min": 581
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5796831314072692,
|
||
|
|
"grad_norm": 0.5520066159358215,
|
||
|
|
"learning_rate": 3.185914052785584e-05,
|
||
|
|
"loss": 0.2782,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14450564980506897,
|
||
|
|
"step": 1385,
|
||
|
|
"valid_targets_mean": 3653.8,
|
||
|
|
"valid_targets_min": 536
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5890027958993476,
|
||
|
|
"grad_norm": 0.5525942245587327,
|
||
|
|
"learning_rate": 3.178423561497636e-05,
|
||
|
|
"loss": 0.2917,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.158339262008667,
|
||
|
|
"step": 1390,
|
||
|
|
"valid_targets_mean": 4360.8,
|
||
|
|
"valid_targets_min": 565
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.598322460391426,
|
||
|
|
"grad_norm": 0.518651221698429,
|
||
|
|
"learning_rate": 3.170907664203739e-05,
|
||
|
|
"loss": 0.2818,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17771437764167786,
|
||
|
|
"step": 1395,
|
||
|
|
"valid_targets_mean": 5481.9,
|
||
|
|
"valid_targets_min": 549
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.607642124883504,
|
||
|
|
"grad_norm": 0.41378717704038076,
|
||
|
|
"learning_rate": 3.1633665229414956e-05,
|
||
|
|
"loss": 0.288,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1330062448978424,
|
||
|
|
"step": 1400,
|
||
|
|
"valid_targets_mean": 6588.8,
|
||
|
|
"valid_targets_min": 1178
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6169617893755825,
|
||
|
|
"grad_norm": 0.5075307281593862,
|
||
|
|
"learning_rate": 3.155800300292755e-05,
|
||
|
|
"loss": 0.2821,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13558664917945862,
|
||
|
|
"step": 1405,
|
||
|
|
"valid_targets_mean": 6072.0,
|
||
|
|
"valid_targets_min": 4966
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.626281453867661,
|
||
|
|
"grad_norm": 0.7261192633084247,
|
||
|
|
"learning_rate": 3.148209159380101e-05,
|
||
|
|
"loss": 0.3054,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13959947228431702,
|
||
|
|
"step": 1410,
|
||
|
|
"valid_targets_mean": 2199.6,
|
||
|
|
"valid_targets_min": 527
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.635601118359739,
|
||
|
|
"grad_norm": 0.6161058564710279,
|
||
|
|
"learning_rate": 3.1405932638633404e-05,
|
||
|
|
"loss": 0.2816,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13546423614025116,
|
||
|
|
"step": 1415,
|
||
|
|
"valid_targets_mean": 3726.9,
|
||
|
|
"valid_targets_min": 561
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.644920782851817,
|
||
|
|
"grad_norm": 0.5128138024637409,
|
||
|
|
"learning_rate": 3.13295277793597e-05,
|
||
|
|
"loss": 0.2609,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1413899064064026,
|
||
|
|
"step": 1420,
|
||
|
|
"valid_targets_mean": 5723.5,
|
||
|
|
"valid_targets_min": 921
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.654240447343896,
|
||
|
|
"grad_norm": 0.6424120946284729,
|
||
|
|
"learning_rate": 3.125287866321643e-05,
|
||
|
|
"loss": 0.3011,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19588042795658112,
|
||
|
|
"step": 1425,
|
||
|
|
"valid_targets_mean": 4232.0,
|
||
|
|
"valid_targets_min": 739
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6635601118359737,
|
||
|
|
"grad_norm": 0.4419489221565109,
|
||
|
|
"learning_rate": 3.117598694270609e-05,
|
||
|
|
"loss": 0.2662,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13978776335716248,
|
||
|
|
"step": 1430,
|
||
|
|
"valid_targets_mean": 6799.5,
|
||
|
|
"valid_targets_min": 829
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.672879776328052,
|
||
|
|
"grad_norm": 0.5746546487521226,
|
||
|
|
"learning_rate": 3.1098854275561565e-05,
|
||
|
|
"loss": 0.2837,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16849973797798157,
|
||
|
|
"step": 1435,
|
||
|
|
"valid_targets_mean": 4461.9,
|
||
|
|
"valid_targets_min": 638
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6821994408201304,
|
||
|
|
"grad_norm": 0.657282411229176,
|
||
|
|
"learning_rate": 3.102148232471043e-05,
|
||
|
|
"loss": 0.2825,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16825127601623535,
|
||
|
|
"step": 1440,
|
||
|
|
"valid_targets_mean": 3673.8,
|
||
|
|
"valid_targets_min": 883
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6915191053122087,
|
||
|
|
"grad_norm": 0.504305062962119,
|
||
|
|
"learning_rate": 3.0943872758239016e-05,
|
||
|
|
"loss": 0.2835,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10637699067592621,
|
||
|
|
"step": 1445,
|
||
|
|
"valid_targets_mean": 4512.4,
|
||
|
|
"valid_targets_min": 763
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.700838769804287,
|
||
|
|
"grad_norm": 0.5533176866574699,
|
||
|
|
"learning_rate": 3.0866027249356484e-05,
|
||
|
|
"loss": 0.305,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16336052119731903,
|
||
|
|
"step": 1450,
|
||
|
|
"valid_targets_mean": 3702.1,
|
||
|
|
"valid_targets_min": 477
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7101584342963654,
|
||
|
|
"grad_norm": 0.6509325664471923,
|
||
|
|
"learning_rate": 3.0787947476358765e-05,
|
||
|
|
"loss": 0.307,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21394076943397522,
|
||
|
|
"step": 1455,
|
||
|
|
"valid_targets_mean": 4565.9,
|
||
|
|
"valid_targets_min": 581
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7194780987884437,
|
||
|
|
"grad_norm": 0.545691157091797,
|
||
|
|
"learning_rate": 3.070963512259235e-05,
|
||
|
|
"loss": 0.3057,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15207439661026,
|
||
|
|
"step": 1460,
|
||
|
|
"valid_targets_mean": 4667.1,
|
||
|
|
"valid_targets_min": 574
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.728797763280522,
|
||
|
|
"grad_norm": 0.46020680468047603,
|
||
|
|
"learning_rate": 3.063109187641803e-05,
|
||
|
|
"loss": 0.2765,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1598256528377533,
|
||
|
|
"step": 1465,
|
||
|
|
"valid_targets_mean": 5772.6,
|
||
|
|
"valid_targets_min": 760
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7381174277726004,
|
||
|
|
"grad_norm": 0.6370971941140389,
|
||
|
|
"learning_rate": 3.055231943117447e-05,
|
||
|
|
"loss": 0.2834,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21424660086631775,
|
||
|
|
"step": 1470,
|
||
|
|
"valid_targets_mean": 5103.8,
|
||
|
|
"valid_targets_min": 726
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7474370922646782,
|
||
|
|
"grad_norm": 0.5074411529933411,
|
||
|
|
"learning_rate": 3.0473319485141702e-05,
|
||
|
|
"loss": 0.2936,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18712572753429413,
|
||
|
|
"step": 1475,
|
||
|
|
"valid_targets_mean": 5550.9,
|
||
|
|
"valid_targets_min": 973
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.756756756756757,
|
||
|
|
"grad_norm": 0.5132177076965325,
|
||
|
|
"learning_rate": 3.039409374150454e-05,
|
||
|
|
"loss": 0.3129,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1562758982181549,
|
||
|
|
"step": 1480,
|
||
|
|
"valid_targets_mean": 4753.8,
|
||
|
|
"valid_targets_min": 567
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.766076421248835,
|
||
|
|
"grad_norm": 0.5256235764243411,
|
||
|
|
"learning_rate": 3.0314643908315812e-05,
|
||
|
|
"loss": 0.3159,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18409791588783264,
|
||
|
|
"step": 1485,
|
||
|
|
"valid_targets_mean": 5339.2,
|
||
|
|
"valid_targets_min": 896
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7753960857409132,
|
||
|
|
"grad_norm": 0.5894249291463771,
|
||
|
|
"learning_rate": 3.0234971698459582e-05,
|
||
|
|
"loss": 0.3123,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1759987771511078,
|
||
|
|
"step": 1490,
|
||
|
|
"valid_targets_mean": 5201.5,
|
||
|
|
"valid_targets_min": 547
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7847157502329916,
|
||
|
|
"grad_norm": 0.4480979155265044,
|
||
|
|
"learning_rate": 3.015507882961421e-05,
|
||
|
|
"loss": 0.3134,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1333552896976471,
|
||
|
|
"step": 1495,
|
||
|
|
"valid_targets_mean": 5493.6,
|
||
|
|
"valid_targets_min": 900
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.79403541472507,
|
||
|
|
"grad_norm": 0.6649380939115621,
|
||
|
|
"learning_rate": 3.007496702421529e-05,
|
||
|
|
"loss": 0.3207,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17558157444000244,
|
||
|
|
"step": 1500,
|
||
|
|
"valid_targets_mean": 3708.4,
|
||
|
|
"valid_targets_min": 631
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.803355079217148,
|
||
|
|
"grad_norm": 0.5686075626204706,
|
||
|
|
"learning_rate": 2.9994638009418552e-05,
|
||
|
|
"loss": 0.2721,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19613277912139893,
|
||
|
|
"step": 1505,
|
||
|
|
"valid_targets_mean": 6333.5,
|
||
|
|
"valid_targets_min": 737
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8126747437092265,
|
||
|
|
"grad_norm": 0.5472980295631082,
|
||
|
|
"learning_rate": 2.9914093517062612e-05,
|
||
|
|
"loss": 0.3059,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16347160935401917,
|
||
|
|
"step": 1510,
|
||
|
|
"valid_targets_mean": 4730.6,
|
||
|
|
"valid_targets_min": 868
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.821994408201305,
|
||
|
|
"grad_norm": 0.6265915578676964,
|
||
|
|
"learning_rate": 2.9833335283631642e-05,
|
||
|
|
"loss": 0.2804,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1574643850326538,
|
||
|
|
"step": 1515,
|
||
|
|
"valid_targets_mean": 4709.0,
|
||
|
|
"valid_targets_min": 757
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.831314072693383,
|
||
|
|
"grad_norm": 0.4356908408200581,
|
||
|
|
"learning_rate": 2.9752365050217898e-05,
|
||
|
|
"loss": 0.2836,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14582863450050354,
|
||
|
|
"step": 1520,
|
||
|
|
"valid_targets_mean": 7536.1,
|
||
|
|
"valid_targets_min": 5604
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8406337371854615,
|
||
|
|
"grad_norm": 0.5037999587904803,
|
||
|
|
"learning_rate": 2.9671184562484244e-05,
|
||
|
|
"loss": 0.2826,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13556590676307678,
|
||
|
|
"step": 1525,
|
||
|
|
"valid_targets_mean": 5021.2,
|
||
|
|
"valid_targets_min": 489
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8499534016775394,
|
||
|
|
"grad_norm": 0.5323584844080301,
|
||
|
|
"learning_rate": 2.958979557062646e-05,
|
||
|
|
"loss": 0.2925,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1633891761302948,
|
||
|
|
"step": 1530,
|
||
|
|
"valid_targets_mean": 5298.1,
|
||
|
|
"valid_targets_min": 873
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8592730661696177,
|
||
|
|
"grad_norm": 0.5344819744890954,
|
||
|
|
"learning_rate": 2.9508199829335543e-05,
|
||
|
|
"loss": 0.2966,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10134637355804443,
|
||
|
|
"step": 1535,
|
||
|
|
"valid_targets_mean": 3073.6,
|
||
|
|
"valid_targets_min": 760
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.868592730661696,
|
||
|
|
"grad_norm": 0.5159025112400663,
|
||
|
|
"learning_rate": 2.942639909775987e-05,
|
||
|
|
"loss": 0.3072,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15362700819969177,
|
||
|
|
"step": 1540,
|
||
|
|
"valid_targets_mean": 6463.4,
|
||
|
|
"valid_targets_min": 1210
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8779123951537744,
|
||
|
|
"grad_norm": 0.5979265240689953,
|
||
|
|
"learning_rate": 2.934439513946726e-05,
|
||
|
|
"loss": 0.2909,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11578511446714401,
|
||
|
|
"step": 1545,
|
||
|
|
"valid_targets_mean": 2746.6,
|
||
|
|
"valid_targets_min": 536
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8872320596458527,
|
||
|
|
"grad_norm": 0.5858529657587352,
|
||
|
|
"learning_rate": 2.9262189722406956e-05,
|
||
|
|
"loss": 0.3027,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15718010067939758,
|
||
|
|
"step": 1550,
|
||
|
|
"valid_targets_mean": 5905.1,
|
||
|
|
"valid_targets_min": 1146
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.896551724137931,
|
||
|
|
"grad_norm": 0.47126713100646117,
|
||
|
|
"learning_rate": 2.917978461887154e-05,
|
||
|
|
"loss": 0.3018,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17965883016586304,
|
||
|
|
"step": 1555,
|
||
|
|
"valid_targets_mean": 6422.1,
|
||
|
|
"valid_targets_min": 691
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9058713886300094,
|
||
|
|
"grad_norm": 0.6593836247084388,
|
||
|
|
"learning_rate": 2.909718160545865e-05,
|
||
|
|
"loss": 0.3334,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1793258786201477,
|
||
|
|
"step": 1560,
|
||
|
|
"valid_targets_mean": 4425.2,
|
||
|
|
"valid_targets_min": 838
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9151910531220877,
|
||
|
|
"grad_norm": 0.5204676005818271,
|
||
|
|
"learning_rate": 2.9014382463032782e-05,
|
||
|
|
"loss": 0.3037,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17872509360313416,
|
||
|
|
"step": 1565,
|
||
|
|
"valid_targets_mean": 6217.2,
|
||
|
|
"valid_targets_min": 887
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.924510717614166,
|
||
|
|
"grad_norm": 0.5307464612031242,
|
||
|
|
"learning_rate": 2.89313889766868e-05,
|
||
|
|
"loss": 0.2947,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16371187567710876,
|
||
|
|
"step": 1570,
|
||
|
|
"valid_targets_mean": 5345.2,
|
||
|
|
"valid_targets_min": 616
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.933830382106244,
|
||
|
|
"grad_norm": 0.3757054816538643,
|
||
|
|
"learning_rate": 2.8848202935703505e-05,
|
||
|
|
"loss": 0.2941,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11628149449825287,
|
||
|
|
"step": 1575,
|
||
|
|
"valid_targets_mean": 6525.2,
|
||
|
|
"valid_targets_min": 727
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9431500465983227,
|
||
|
|
"grad_norm": 0.5028815332185801,
|
||
|
|
"learning_rate": 2.8764826133517045e-05,
|
||
|
|
"loss": 0.3001,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20181849598884583,
|
||
|
|
"step": 1580,
|
||
|
|
"valid_targets_mean": 6108.0,
|
||
|
|
"valid_targets_min": 1088
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9524697110904006,
|
||
|
|
"grad_norm": 0.6013603697094064,
|
||
|
|
"learning_rate": 2.8681260367674237e-05,
|
||
|
|
"loss": 0.2898,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13791179656982422,
|
||
|
|
"step": 1585,
|
||
|
|
"valid_targets_mean": 4386.8,
|
||
|
|
"valid_targets_min": 539
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.961789375582479,
|
||
|
|
"grad_norm": 0.6055794123287405,
|
||
|
|
"learning_rate": 2.8597507439795845e-05,
|
||
|
|
"loss": 0.3045,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15686023235321045,
|
||
|
|
"step": 1590,
|
||
|
|
"valid_targets_mean": 3235.8,
|
||
|
|
"valid_targets_min": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9711090400745572,
|
||
|
|
"grad_norm": 0.6744139972566192,
|
||
|
|
"learning_rate": 2.8513569155537698e-05,
|
||
|
|
"loss": 0.3061,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17476236820220947,
|
||
|
|
"step": 1595,
|
||
|
|
"valid_targets_mean": 4473.0,
|
||
|
|
"valid_targets_min": 552
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9804287045666356,
|
||
|
|
"grad_norm": 0.5310968626569939,
|
||
|
|
"learning_rate": 2.8429447324551805e-05,
|
||
|
|
"loss": 0.3034,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16782844066619873,
|
||
|
|
"step": 1600,
|
||
|
|
"valid_targets_mean": 6201.1,
|
||
|
|
"valid_targets_min": 1702
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.989748369058714,
|
||
|
|
"grad_norm": 0.6556645098152435,
|
||
|
|
"learning_rate": 2.834514376044728e-05,
|
||
|
|
"loss": 0.289,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16085655987262726,
|
||
|
|
"step": 1605,
|
||
|
|
"valid_targets_mean": 3211.0,
|
||
|
|
"valid_targets_min": 777
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.999068033550792,
|
||
|
|
"grad_norm": 0.5002052254593363,
|
||
|
|
"learning_rate": 2.8260660280751326e-05,
|
||
|
|
"loss": 0.3064,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12481134384870529,
|
||
|
|
"step": 1610,
|
||
|
|
"valid_targets_mean": 4778.5,
|
||
|
|
"valid_targets_min": 687
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0074557315936628,
|
||
|
|
"grad_norm": 0.5292889403810667,
|
||
|
|
"learning_rate": 2.8175998706869964e-05,
|
||
|
|
"loss": 0.2806,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1518377661705017,
|
||
|
|
"step": 1615,
|
||
|
|
"valid_targets_mean": 5447.9,
|
||
|
|
"valid_targets_min": 922
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.016775396085741,
|
||
|
|
"grad_norm": 0.46964718996519694,
|
||
|
|
"learning_rate": 2.809116086404882e-05,
|
||
|
|
"loss": 0.2535,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11578822880983353,
|
||
|
|
"step": 1620,
|
||
|
|
"valid_targets_mean": 6830.5,
|
||
|
|
"valid_targets_min": 584
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0260950605778194,
|
||
|
|
"grad_norm": 0.5846447353564823,
|
||
|
|
"learning_rate": 2.8006148581333766e-05,
|
||
|
|
"loss": 0.2783,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2010546624660492,
|
||
|
|
"step": 1625,
|
||
|
|
"valid_targets_mean": 5073.8,
|
||
|
|
"valid_targets_min": 605
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0354147250698973,
|
||
|
|
"grad_norm": 0.4946670622778176,
|
||
|
|
"learning_rate": 2.792096369153146e-05,
|
||
|
|
"loss": 0.2862,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17162825167179108,
|
||
|
|
"step": 1630,
|
||
|
|
"valid_targets_mean": 6312.6,
|
||
|
|
"valid_targets_min": 655
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0447343895619756,
|
||
|
|
"grad_norm": 0.5617475466654164,
|
||
|
|
"learning_rate": 2.7835608031169864e-05,
|
||
|
|
"loss": 0.2629,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11132988333702087,
|
||
|
|
"step": 1635,
|
||
|
|
"valid_targets_mean": 3892.8,
|
||
|
|
"valid_targets_min": 625
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.054054054054054,
|
||
|
|
"grad_norm": 0.5239034909783888,
|
||
|
|
"learning_rate": 2.7750083440458634e-05,
|
||
|
|
"loss": 0.2512,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1619134247303009,
|
||
|
|
"step": 1640,
|
||
|
|
"valid_targets_mean": 4180.4,
|
||
|
|
"valid_targets_min": 821
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0633737185461323,
|
||
|
|
"grad_norm": 0.5465173702868134,
|
||
|
|
"learning_rate": 2.7664391763249443e-05,
|
||
|
|
"loss": 0.2813,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20818915963172913,
|
||
|
|
"step": 1645,
|
||
|
|
"valid_targets_mean": 5546.6,
|
||
|
|
"valid_targets_min": 791
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0726933830382106,
|
||
|
|
"grad_norm": 0.49669743734249666,
|
||
|
|
"learning_rate": 2.757853484699624e-05,
|
||
|
|
"loss": 0.2787,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1643545925617218,
|
||
|
|
"step": 1650,
|
||
|
|
"valid_targets_mean": 6022.4,
|
||
|
|
"valid_targets_min": 610
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.082013047530289,
|
||
|
|
"grad_norm": 0.5506375718836168,
|
||
|
|
"learning_rate": 2.7492514542715422e-05,
|
||
|
|
"loss": 0.2637,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.123469777405262,
|
||
|
|
"step": 1655,
|
||
|
|
"valid_targets_mean": 4580.1,
|
||
|
|
"valid_targets_min": 877
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0913327120223673,
|
||
|
|
"grad_norm": 0.5320475797062215,
|
||
|
|
"learning_rate": 2.7406332704945894e-05,
|
||
|
|
"loss": 0.2625,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1467972993850708,
|
||
|
|
"step": 1660,
|
||
|
|
"valid_targets_mean": 5559.6,
|
||
|
|
"valid_targets_min": 652
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1006523765144456,
|
||
|
|
"grad_norm": 0.6289577640306172,
|
||
|
|
"learning_rate": 2.7319991191709148e-05,
|
||
|
|
"loss": 0.2952,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1240549087524414,
|
||
|
|
"step": 1665,
|
||
|
|
"valid_targets_mean": 4148.8,
|
||
|
|
"valid_targets_min": 842
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.109972041006524,
|
||
|
|
"grad_norm": 0.5354141375667172,
|
||
|
|
"learning_rate": 2.7233491864469142e-05,
|
||
|
|
"loss": 0.2796,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09915781021118164,
|
||
|
|
"step": 1670,
|
||
|
|
"valid_targets_mean": 3954.1,
|
||
|
|
"valid_targets_min": 568
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1192917054986022,
|
||
|
|
"grad_norm": 0.5089575112348431,
|
||
|
|
"learning_rate": 2.7146836588092215e-05,
|
||
|
|
"loss": 0.2673,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18762989342212677,
|
||
|
|
"step": 1675,
|
||
|
|
"valid_targets_mean": 6694.0,
|
||
|
|
"valid_targets_min": 710
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.12861136999068,
|
||
|
|
"grad_norm": 0.5821508808001727,
|
||
|
|
"learning_rate": 2.706002723080684e-05,
|
||
|
|
"loss": 0.2634,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14423713088035583,
|
||
|
|
"step": 1680,
|
||
|
|
"valid_targets_mean": 4937.8,
|
||
|
|
"valid_targets_min": 558
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1379310344827585,
|
||
|
|
"grad_norm": 0.5330186143321833,
|
||
|
|
"learning_rate": 2.6973065664163405e-05,
|
||
|
|
"loss": 0.2742,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12571248412132263,
|
||
|
|
"step": 1685,
|
||
|
|
"valid_targets_mean": 5132.9,
|
||
|
|
"valid_targets_min": 1226
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.147250698974837,
|
||
|
|
"grad_norm": 0.49558717785187245,
|
||
|
|
"learning_rate": 2.688595376299378e-05,
|
||
|
|
"loss": 0.2706,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12781086564064026,
|
||
|
|
"step": 1690,
|
||
|
|
"valid_targets_mean": 5655.2,
|
||
|
|
"valid_targets_min": 861
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.156570363466915,
|
||
|
|
"grad_norm": 0.6449835237590265,
|
||
|
|
"learning_rate": 2.6798693405370987e-05,
|
||
|
|
"loss": 0.2779,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16559576988220215,
|
||
|
|
"step": 1695,
|
||
|
|
"valid_targets_mean": 3675.6,
|
||
|
|
"valid_targets_min": 518
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1658900279589934,
|
||
|
|
"grad_norm": 0.5092581986261165,
|
||
|
|
"learning_rate": 2.6711286472568644e-05,
|
||
|
|
"loss": 0.2836,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16142301261425018,
|
||
|
|
"step": 1700,
|
||
|
|
"valid_targets_mean": 5692.0,
|
||
|
|
"valid_targets_min": 596
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1752096924510718,
|
||
|
|
"grad_norm": 0.5106227916202737,
|
||
|
|
"learning_rate": 2.6623734849020436e-05,
|
||
|
|
"loss": 0.2679,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14139044284820557,
|
||
|
|
"step": 1705,
|
||
|
|
"valid_targets_mean": 5081.5,
|
||
|
|
"valid_targets_min": 606
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.18452935694315,
|
||
|
|
"grad_norm": 0.7035586437323843,
|
||
|
|
"learning_rate": 2.653604042227949e-05,
|
||
|
|
"loss": 0.2686,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08263888210058212,
|
||
|
|
"step": 1710,
|
||
|
|
"valid_targets_mean": 3034.6,
|
||
|
|
"valid_targets_min": 614
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1938490214352284,
|
||
|
|
"grad_norm": 0.44634067410199846,
|
||
|
|
"learning_rate": 2.644820508297765e-05,
|
||
|
|
"loss": 0.2372,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08851374685764313,
|
||
|
|
"step": 1715,
|
||
|
|
"valid_targets_mean": 7097.5,
|
||
|
|
"valid_targets_min": 796
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2031686859273067,
|
||
|
|
"grad_norm": 0.5992203960349496,
|
||
|
|
"learning_rate": 2.6360230724784766e-05,
|
||
|
|
"loss": 0.2658,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14227928221225739,
|
||
|
|
"step": 1720,
|
||
|
|
"valid_targets_mean": 5119.8,
|
||
|
|
"valid_targets_min": 808
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.212488350419385,
|
||
|
|
"grad_norm": 0.5650228131359676,
|
||
|
|
"learning_rate": 2.627211924436782e-05,
|
||
|
|
"loss": 0.2731,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13075804710388184,
|
||
|
|
"step": 1725,
|
||
|
|
"valid_targets_mean": 3759.1,
|
||
|
|
"valid_targets_min": 653
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2218080149114634,
|
||
|
|
"grad_norm": 0.45430033521167756,
|
||
|
|
"learning_rate": 2.6183872541350068e-05,
|
||
|
|
"loss": 0.269,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1148616373538971,
|
||
|
|
"step": 1730,
|
||
|
|
"valid_targets_mean": 5176.9,
|
||
|
|
"valid_targets_min": 832
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2311276794035413,
|
||
|
|
"grad_norm": 0.5323131508674257,
|
||
|
|
"learning_rate": 2.609549251827005e-05,
|
||
|
|
"loss": 0.2764,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21254248917102814,
|
||
|
|
"step": 1735,
|
||
|
|
"valid_targets_mean": 7102.0,
|
||
|
|
"valid_targets_min": 540
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2404473438956196,
|
||
|
|
"grad_norm": 0.5677874994727236,
|
||
|
|
"learning_rate": 2.6006981080540638e-05,
|
||
|
|
"loss": 0.2949,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13414950668811798,
|
||
|
|
"step": 1740,
|
||
|
|
"valid_targets_mean": 6328.2,
|
||
|
|
"valid_targets_min": 3955
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.249767008387698,
|
||
|
|
"grad_norm": 0.5705375005102685,
|
||
|
|
"learning_rate": 2.5918340136407865e-05,
|
||
|
|
"loss": 0.296,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1413242071866989,
|
||
|
|
"step": 1745,
|
||
|
|
"valid_targets_mean": 3831.9,
|
||
|
|
"valid_targets_min": 583
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2590866728797763,
|
||
|
|
"grad_norm": 0.5845399820876519,
|
||
|
|
"learning_rate": 2.5829571596909862e-05,
|
||
|
|
"loss": 0.2687,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15089261531829834,
|
||
|
|
"step": 1750,
|
||
|
|
"valid_targets_mean": 4379.6,
|
||
|
|
"valid_targets_min": 681
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2684063373718546,
|
||
|
|
"grad_norm": 0.5418554412409033,
|
||
|
|
"learning_rate": 2.5740677375835637e-05,
|
||
|
|
"loss": 0.2823,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12447644770145416,
|
||
|
|
"step": 1755,
|
||
|
|
"valid_targets_mean": 3708.0,
|
||
|
|
"valid_targets_min": 754
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.277726001863933,
|
||
|
|
"grad_norm": 0.49278733191903185,
|
||
|
|
"learning_rate": 2.5651659389683777e-05,
|
||
|
|
"loss": 0.2666,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12036804854869843,
|
||
|
|
"step": 1760,
|
||
|
|
"valid_targets_mean": 5812.8,
|
||
|
|
"valid_targets_min": 3080
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2870456663560113,
|
||
|
|
"grad_norm": 0.4800168474409703,
|
||
|
|
"learning_rate": 2.5562519557621183e-05,
|
||
|
|
"loss": 0.2749,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09417456388473511,
|
||
|
|
"step": 1765,
|
||
|
|
"valid_targets_mean": 4699.4,
|
||
|
|
"valid_targets_min": 593
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2963653308480896,
|
||
|
|
"grad_norm": 0.6100330268116317,
|
||
|
|
"learning_rate": 2.5473259801441663e-05,
|
||
|
|
"loss": 0.2595,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1364918351173401,
|
||
|
|
"step": 1770,
|
||
|
|
"valid_targets_mean": 4840.5,
|
||
|
|
"valid_targets_min": 627
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.305684995340168,
|
||
|
|
"grad_norm": 0.5955885898771116,
|
||
|
|
"learning_rate": 2.53838820455245e-05,
|
||
|
|
"loss": 0.2679,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13605651259422302,
|
||
|
|
"step": 1775,
|
||
|
|
"valid_targets_mean": 4633.5,
|
||
|
|
"valid_targets_min": 666
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.315004659832246,
|
||
|
|
"grad_norm": 0.5961555441355574,
|
||
|
|
"learning_rate": 2.5294388216792987e-05,
|
||
|
|
"loss": 0.2646,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11901737004518509,
|
||
|
|
"step": 1780,
|
||
|
|
"valid_targets_mean": 5800.9,
|
||
|
|
"valid_targets_min": 826
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3243243243243246,
|
||
|
|
"grad_norm": 0.4741774046736269,
|
||
|
|
"learning_rate": 2.5204780244672858e-05,
|
||
|
|
"loss": 0.2973,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1427319198846817,
|
||
|
|
"step": 1785,
|
||
|
|
"valid_targets_mean": 5902.4,
|
||
|
|
"valid_targets_min": 589
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3336439888164024,
|
||
|
|
"grad_norm": 0.5299819257259158,
|
||
|
|
"learning_rate": 2.5115060061050693e-05,
|
||
|
|
"loss": 0.2544,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13748562335968018,
|
||
|
|
"step": 1790,
|
||
|
|
"valid_targets_mean": 5136.4,
|
||
|
|
"valid_targets_min": 1020
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3429636533084808,
|
||
|
|
"grad_norm": 0.7595798680321887,
|
||
|
|
"learning_rate": 2.5025229600232285e-05,
|
||
|
|
"loss": 0.3141,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1914379745721817,
|
||
|
|
"step": 1795,
|
||
|
|
"valid_targets_mean": 2987.5,
|
||
|
|
"valid_targets_min": 365
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.352283317800559,
|
||
|
|
"grad_norm": 0.5535604657254797,
|
||
|
|
"learning_rate": 2.493529079890093e-05,
|
||
|
|
"loss": 0.2666,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13937866687774658,
|
||
|
|
"step": 1800,
|
||
|
|
"valid_targets_mean": 4819.0,
|
||
|
|
"valid_targets_min": 542
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3616029822926374,
|
||
|
|
"grad_norm": 0.6044743151809133,
|
||
|
|
"learning_rate": 2.4845245596075666e-05,
|
||
|
|
"loss": 0.2766,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11259683966636658,
|
||
|
|
"step": 1805,
|
||
|
|
"valid_targets_mean": 3044.8,
|
||
|
|
"valid_targets_min": 706
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3709226467847158,
|
||
|
|
"grad_norm": 0.5529570008404465,
|
||
|
|
"learning_rate": 2.475509593306947e-05,
|
||
|
|
"loss": 0.2737,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21021634340286255,
|
||
|
|
"step": 1810,
|
||
|
|
"valid_targets_mean": 5082.1,
|
||
|
|
"valid_targets_min": 580
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.380242311276794,
|
||
|
|
"grad_norm": 0.6196979670212198,
|
||
|
|
"learning_rate": 2.4664843753447423e-05,
|
||
|
|
"loss": 0.2783,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14590218663215637,
|
||
|
|
"step": 1815,
|
||
|
|
"valid_targets_mean": 3901.1,
|
||
|
|
"valid_targets_min": 709
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3895619757688724,
|
||
|
|
"grad_norm": 0.7490264310504905,
|
||
|
|
"learning_rate": 2.4574491002984777e-05,
|
||
|
|
"loss": 0.295,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1642751693725586,
|
||
|
|
"step": 1820,
|
||
|
|
"valid_targets_mean": 2777.9,
|
||
|
|
"valid_targets_min": 709
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3988816402609507,
|
||
|
|
"grad_norm": 0.5212402217327484,
|
||
|
|
"learning_rate": 2.448403962962504e-05,
|
||
|
|
"loss": 0.2952,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14331313967704773,
|
||
|
|
"step": 1825,
|
||
|
|
"valid_targets_mean": 6343.5,
|
||
|
|
"valid_targets_min": 962
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.408201304753029,
|
||
|
|
"grad_norm": 0.6407403857989383,
|
||
|
|
"learning_rate": 2.4393491583437946e-05,
|
||
|
|
"loss": 0.2918,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22689306735992432,
|
||
|
|
"step": 1830,
|
||
|
|
"valid_targets_mean": 5847.9,
|
||
|
|
"valid_targets_min": 614
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.417520969245107,
|
||
|
|
"grad_norm": 0.6484608475863175,
|
||
|
|
"learning_rate": 2.430284881657744e-05,
|
||
|
|
"loss": 0.2874,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14794909954071045,
|
||
|
|
"step": 1835,
|
||
|
|
"valid_targets_mean": 5091.9,
|
||
|
|
"valid_targets_min": 764
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4268406337371853,
|
||
|
|
"grad_norm": 0.5984917696785221,
|
||
|
|
"learning_rate": 2.421211328323957e-05,
|
||
|
|
"loss": 0.2773,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12450984120368958,
|
||
|
|
"step": 1840,
|
||
|
|
"valid_targets_mean": 3853.8,
|
||
|
|
"valid_targets_min": 654
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4361602982292636,
|
||
|
|
"grad_norm": 0.6876372814409679,
|
||
|
|
"learning_rate": 2.4121286939620385e-05,
|
||
|
|
"loss": 0.2858,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15968386828899384,
|
||
|
|
"step": 1845,
|
||
|
|
"valid_targets_mean": 4041.8,
|
||
|
|
"valid_targets_min": 509
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.445479962721342,
|
||
|
|
"grad_norm": 0.5833739226210374,
|
||
|
|
"learning_rate": 2.4030371743873713e-05,
|
||
|
|
"loss": 0.2831,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18069790303707123,
|
||
|
|
"step": 1850,
|
||
|
|
"valid_targets_mean": 5125.1,
|
||
|
|
"valid_targets_min": 747
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4547996272134203,
|
||
|
|
"grad_norm": 0.5367210801375877,
|
||
|
|
"learning_rate": 2.3939369656069005e-05,
|
||
|
|
"loss": 0.2951,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15029451251029968,
|
||
|
|
"step": 1855,
|
||
|
|
"valid_targets_mean": 4438.8,
|
||
|
|
"valid_targets_min": 593
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4641192917054986,
|
||
|
|
"grad_norm": 0.5771972670174873,
|
||
|
|
"learning_rate": 2.3848282638149012e-05,
|
||
|
|
"loss": 0.2644,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13196644186973572,
|
||
|
|
"step": 1860,
|
||
|
|
"valid_targets_mean": 4663.0,
|
||
|
|
"valid_targets_min": 509
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.473438956197577,
|
||
|
|
"grad_norm": 0.6958778835646237,
|
||
|
|
"learning_rate": 2.3757112653887554e-05,
|
||
|
|
"loss": 0.2682,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15621313452720642,
|
||
|
|
"step": 1865,
|
||
|
|
"valid_targets_mean": 5268.4,
|
||
|
|
"valid_targets_min": 1146
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4827586206896552,
|
||
|
|
"grad_norm": 0.7008735846998657,
|
||
|
|
"learning_rate": 2.366586166884712e-05,
|
||
|
|
"loss": 0.2943,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1513005495071411,
|
||
|
|
"step": 1870,
|
||
|
|
"valid_targets_mean": 4922.9,
|
||
|
|
"valid_targets_min": 923
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4920782851817336,
|
||
|
|
"grad_norm": 0.587312892114291,
|
||
|
|
"learning_rate": 2.3574531650336536e-05,
|
||
|
|
"loss": 0.2723,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17634403705596924,
|
||
|
|
"step": 1875,
|
||
|
|
"valid_targets_mean": 5135.4,
|
||
|
|
"valid_targets_min": 575
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.501397949673812,
|
||
|
|
"grad_norm": 0.6004867780009553,
|
||
|
|
"learning_rate": 2.3483124567368534e-05,
|
||
|
|
"loss": 0.296,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1237824410200119,
|
||
|
|
"step": 1880,
|
||
|
|
"valid_targets_mean": 4112.2,
|
||
|
|
"valid_targets_min": 825
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5107176141658902,
|
||
|
|
"grad_norm": 0.5118431043376553,
|
||
|
|
"learning_rate": 2.33916423906173e-05,
|
||
|
|
"loss": 0.2682,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12049616128206253,
|
||
|
|
"step": 1885,
|
||
|
|
"valid_targets_mean": 5118.1,
|
||
|
|
"valid_targets_min": 684
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.520037278657968,
|
||
|
|
"grad_norm": 0.5126355608601477,
|
||
|
|
"learning_rate": 2.330008709237599e-05,
|
||
|
|
"loss": 0.2595,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09314330667257309,
|
||
|
|
"step": 1890,
|
||
|
|
"valid_targets_mean": 3217.1,
|
||
|
|
"valid_targets_min": 601
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5293569431500464,
|
||
|
|
"grad_norm": 0.6127057285208951,
|
||
|
|
"learning_rate": 2.3208460646514208e-05,
|
||
|
|
"loss": 0.28,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15857522189617157,
|
||
|
|
"step": 1895,
|
||
|
|
"valid_targets_mean": 4296.1,
|
||
|
|
"valid_targets_min": 626
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5386766076421248,
|
||
|
|
"grad_norm": 0.6407537982213708,
|
||
|
|
"learning_rate": 2.3116765028435458e-05,
|
||
|
|
"loss": 0.2614,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17790746688842773,
|
||
|
|
"step": 1900,
|
||
|
|
"valid_targets_mean": 4053.0,
|
||
|
|
"valid_targets_min": 638
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.547996272134203,
|
||
|
|
"grad_norm": 0.49172550249638125,
|
||
|
|
"learning_rate": 2.302500221503455e-05,
|
||
|
|
"loss": 0.28,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14827083051204681,
|
||
|
|
"step": 1905,
|
||
|
|
"valid_targets_mean": 5737.0,
|
||
|
|
"valid_targets_min": 837
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5573159366262814,
|
||
|
|
"grad_norm": 0.48065153682336825,
|
||
|
|
"learning_rate": 2.2933174184654973e-05,
|
||
|
|
"loss": 0.2626,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11452475190162659,
|
||
|
|
"step": 1910,
|
||
|
|
"valid_targets_mean": 5597.1,
|
||
|
|
"valid_targets_min": 1214
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5666356011183598,
|
||
|
|
"grad_norm": 0.5795134626041417,
|
||
|
|
"learning_rate": 2.2841282917046246e-05,
|
||
|
|
"loss": 0.2718,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13477712869644165,
|
||
|
|
"step": 1915,
|
||
|
|
"valid_targets_mean": 5005.1,
|
||
|
|
"valid_targets_min": 619
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.575955265610438,
|
||
|
|
"grad_norm": 0.5067593251735485,
|
||
|
|
"learning_rate": 2.274933039332125e-05,
|
||
|
|
"loss": 0.2666,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09253917634487152,
|
||
|
|
"step": 1920,
|
||
|
|
"valid_targets_mean": 4068.8,
|
||
|
|
"valid_targets_min": 744
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5852749301025164,
|
||
|
|
"grad_norm": 0.5619645505197108,
|
||
|
|
"learning_rate": 2.2657318595913506e-05,
|
||
|
|
"loss": 0.2742,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12507201731204987,
|
||
|
|
"step": 1925,
|
||
|
|
"valid_targets_mean": 7303.0,
|
||
|
|
"valid_targets_min": 690
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5945945945945947,
|
||
|
|
"grad_norm": 0.5419369319567007,
|
||
|
|
"learning_rate": 2.256524950853443e-05,
|
||
|
|
"loss": 0.2724,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12294361740350723,
|
||
|
|
"step": 1930,
|
||
|
|
"valid_targets_mean": 5198.2,
|
||
|
|
"valid_targets_min": 820
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6039142590866726,
|
||
|
|
"grad_norm": 0.4842573555827678,
|
||
|
|
"learning_rate": 2.2473125116130557e-05,
|
||
|
|
"loss": 0.2856,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10169918835163116,
|
||
|
|
"step": 1935,
|
||
|
|
"valid_targets_mean": 4681.8,
|
||
|
|
"valid_targets_min": 749
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6132339235787514,
|
||
|
|
"grad_norm": 0.7654131628481188,
|
||
|
|
"learning_rate": 2.2380947404840793e-05,
|
||
|
|
"loss": 0.2453,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12937580049037933,
|
||
|
|
"step": 1940,
|
||
|
|
"valid_targets_mean": 5884.4,
|
||
|
|
"valid_targets_min": 536
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6225535880708293,
|
||
|
|
"grad_norm": 0.5049681711378494,
|
||
|
|
"learning_rate": 2.2288718361953534e-05,
|
||
|
|
"loss": 0.2981,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13497032225131989,
|
||
|
|
"step": 1945,
|
||
|
|
"valid_targets_mean": 6440.6,
|
||
|
|
"valid_targets_min": 4910
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6318732525629076,
|
||
|
|
"grad_norm": 0.5652419330537398,
|
||
|
|
"learning_rate": 2.2196439975863867e-05,
|
||
|
|
"loss": 0.2891,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.184743270277977,
|
||
|
|
"step": 1950,
|
||
|
|
"valid_targets_mean": 6100.0,
|
||
|
|
"valid_targets_min": 865
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.641192917054986,
|
||
|
|
"grad_norm": 0.4654749587743219,
|
||
|
|
"learning_rate": 2.2104114236030666e-05,
|
||
|
|
"loss": 0.2801,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1262211799621582,
|
||
|
|
"step": 1955,
|
||
|
|
"valid_targets_mean": 5130.6,
|
||
|
|
"valid_targets_min": 647
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6505125815470643,
|
||
|
|
"grad_norm": 0.5404326509158578,
|
||
|
|
"learning_rate": 2.2011743132933746e-05,
|
||
|
|
"loss": 0.2859,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14590272307395935,
|
||
|
|
"step": 1960,
|
||
|
|
"valid_targets_mean": 5844.5,
|
||
|
|
"valid_targets_min": 874
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6598322460391426,
|
||
|
|
"grad_norm": 0.5263616478096891,
|
||
|
|
"learning_rate": 2.19193286580309e-05,
|
||
|
|
"loss": 0.2745,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1100989580154419,
|
||
|
|
"step": 1965,
|
||
|
|
"valid_targets_mean": 3534.8,
|
||
|
|
"valid_targets_min": 626
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.669151910531221,
|
||
|
|
"grad_norm": 0.49412985027996864,
|
||
|
|
"learning_rate": 2.1826872803714997e-05,
|
||
|
|
"loss": 0.2631,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12736055254936218,
|
||
|
|
"step": 1970,
|
||
|
|
"valid_targets_mean": 5909.6,
|
||
|
|
"valid_targets_min": 840
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6784715750232992,
|
||
|
|
"grad_norm": 0.5458384212482188,
|
||
|
|
"learning_rate": 2.173437756327102e-05,
|
||
|
|
"loss": 0.2666,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1414768397808075,
|
||
|
|
"step": 1975,
|
||
|
|
"valid_targets_mean": 5901.9,
|
||
|
|
"valid_targets_min": 603
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6877912395153776,
|
||
|
|
"grad_norm": 0.5040953710850293,
|
||
|
|
"learning_rate": 2.1641844930833078e-05,
|
||
|
|
"loss": 0.2659,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1042417585849762,
|
||
|
|
"step": 1980,
|
||
|
|
"valid_targets_mean": 4592.0,
|
||
|
|
"valid_targets_min": 625
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.697110904007456,
|
||
|
|
"grad_norm": 0.5312945350339939,
|
||
|
|
"learning_rate": 2.154927690134145e-05,
|
||
|
|
"loss": 0.2868,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20565815269947052,
|
||
|
|
"step": 1985,
|
||
|
|
"valid_targets_mean": 6693.9,
|
||
|
|
"valid_targets_min": 4100
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.706430568499534,
|
||
|
|
"grad_norm": 0.5632199869855236,
|
||
|
|
"learning_rate": 2.1456675470499523e-05,
|
||
|
|
"loss": 0.2722,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11075471341609955,
|
||
|
|
"step": 1990,
|
||
|
|
"valid_targets_mean": 3691.5,
|
||
|
|
"valid_targets_min": 713
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7157502329916126,
|
||
|
|
"grad_norm": 0.5986344150569787,
|
||
|
|
"learning_rate": 2.1364042634730806e-05,
|
||
|
|
"loss": 0.2678,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1491662561893463,
|
||
|
|
"step": 1995,
|
||
|
|
"valid_targets_mean": 4528.8,
|
||
|
|
"valid_targets_min": 632
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7250698974836904,
|
||
|
|
"grad_norm": 0.5253906197959318,
|
||
|
|
"learning_rate": 2.127138039113589e-05,
|
||
|
|
"loss": 0.2858,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16810084879398346,
|
||
|
|
"step": 2000,
|
||
|
|
"valid_targets_mean": 5004.4,
|
||
|
|
"valid_targets_min": 702
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7343895619757688,
|
||
|
|
"grad_norm": 0.5014510804983215,
|
||
|
|
"learning_rate": 2.1178690737449354e-05,
|
||
|
|
"loss": 0.2498,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08236116170883179,
|
||
|
|
"step": 2005,
|
||
|
|
"valid_targets_mean": 4569.4,
|
||
|
|
"valid_targets_min": 603
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.743709226467847,
|
||
|
|
"grad_norm": 0.5109341986195375,
|
||
|
|
"learning_rate": 2.1085975671996736e-05,
|
||
|
|
"loss": 0.269,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13857677578926086,
|
||
|
|
"step": 2010,
|
||
|
|
"valid_targets_mean": 5565.0,
|
||
|
|
"valid_targets_min": 648
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7530288909599254,
|
||
|
|
"grad_norm": 0.520696037379353,
|
||
|
|
"learning_rate": 2.0993237193651436e-05,
|
||
|
|
"loss": 0.2774,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11594519019126892,
|
||
|
|
"step": 2015,
|
||
|
|
"valid_targets_mean": 2930.2,
|
||
|
|
"valid_targets_min": 739
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7623485554520038,
|
||
|
|
"grad_norm": 0.5721420492569583,
|
||
|
|
"learning_rate": 2.0900477301791606e-05,
|
||
|
|
"loss": 0.2862,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1227039098739624,
|
||
|
|
"step": 2020,
|
||
|
|
"valid_targets_mean": 3563.9,
|
||
|
|
"valid_targets_min": 574
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.771668219944082,
|
||
|
|
"grad_norm": 0.5495188887219793,
|
||
|
|
"learning_rate": 2.0807697996257072e-05,
|
||
|
|
"loss": 0.2712,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1720818281173706,
|
||
|
|
"step": 2025,
|
||
|
|
"valid_targets_mean": 6046.1,
|
||
|
|
"valid_targets_min": 888
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7809878844361604,
|
||
|
|
"grad_norm": 0.6224268073531435,
|
||
|
|
"learning_rate": 2.0714901277306203e-05,
|
||
|
|
"loss": 0.3067,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2149972915649414,
|
||
|
|
"step": 2030,
|
||
|
|
"valid_targets_mean": 5714.6,
|
||
|
|
"valid_targets_min": 836
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7903075489282387,
|
||
|
|
"grad_norm": 0.5101186043959527,
|
||
|
|
"learning_rate": 2.062208914557278e-05,
|
||
|
|
"loss": 0.2618,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15443718433380127,
|
||
|
|
"step": 2035,
|
||
|
|
"valid_targets_mean": 4860.4,
|
||
|
|
"valid_targets_min": 613
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.799627213420317,
|
||
|
|
"grad_norm": 0.465730308391406,
|
||
|
|
"learning_rate": 2.052926360202289e-05,
|
||
|
|
"loss": 0.2712,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10560670495033264,
|
||
|
|
"step": 2040,
|
||
|
|
"valid_targets_mean": 5708.8,
|
||
|
|
"valid_targets_min": 675
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.808946877912395,
|
||
|
|
"grad_norm": 0.5648143056467078,
|
||
|
|
"learning_rate": 2.0436426647911748e-05,
|
||
|
|
"loss": 0.2792,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1117134541273117,
|
||
|
|
"step": 2045,
|
||
|
|
"valid_targets_mean": 4367.9,
|
||
|
|
"valid_targets_min": 625
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8182665424044733,
|
||
|
|
"grad_norm": 0.5362184151499179,
|
||
|
|
"learning_rate": 2.034358028474059e-05,
|
||
|
|
"loss": 0.2558,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09404626488685608,
|
||
|
|
"step": 2050,
|
||
|
|
"valid_targets_mean": 4299.8,
|
||
|
|
"valid_targets_min": 809
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8275862068965516,
|
||
|
|
"grad_norm": 0.5260334627888182,
|
||
|
|
"learning_rate": 2.0250726514213506e-05,
|
||
|
|
"loss": 0.2773,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12250328063964844,
|
||
|
|
"step": 2055,
|
||
|
|
"valid_targets_mean": 4901.0,
|
||
|
|
"valid_targets_min": 644
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.83690587138863,
|
||
|
|
"grad_norm": 0.6455746155499706,
|
||
|
|
"learning_rate": 2.015786733819427e-05,
|
||
|
|
"loss": 0.2811,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18054530024528503,
|
||
|
|
"step": 2060,
|
||
|
|
"valid_targets_mean": 4682.0,
|
||
|
|
"valid_targets_min": 546
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8462255358807083,
|
||
|
|
"grad_norm": 0.5333129526661133,
|
||
|
|
"learning_rate": 2.0065004758663202e-05,
|
||
|
|
"loss": 0.2978,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14793553948402405,
|
||
|
|
"step": 2065,
|
||
|
|
"valid_targets_mean": 5393.5,
|
||
|
|
"valid_targets_min": 412
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8555452003727866,
|
||
|
|
"grad_norm": 0.5355855129703568,
|
||
|
|
"learning_rate": 1.9972140777673997e-05,
|
||
|
|
"loss": 0.2824,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22068268060684204,
|
||
|
|
"step": 2070,
|
||
|
|
"valid_targets_mean": 6896.8,
|
||
|
|
"valid_targets_min": 630
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.864864864864865,
|
||
|
|
"grad_norm": 0.5615545329452644,
|
||
|
|
"learning_rate": 1.9879277397310574e-05,
|
||
|
|
"loss": 0.2589,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1188819482922554,
|
||
|
|
"step": 2075,
|
||
|
|
"valid_targets_mean": 6302.5,
|
||
|
|
"valid_targets_min": 2993
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8741845293569432,
|
||
|
|
"grad_norm": 0.5455018795916219,
|
||
|
|
"learning_rate": 1.9786416619643888e-05,
|
||
|
|
"loss": 0.2768,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14095181226730347,
|
||
|
|
"step": 2080,
|
||
|
|
"valid_targets_mean": 5563.6,
|
||
|
|
"valid_targets_min": 786
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8835041938490216,
|
||
|
|
"grad_norm": 0.5014806599649799,
|
||
|
|
"learning_rate": 1.9693560446688796e-05,
|
||
|
|
"loss": 0.2941,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1475827395915985,
|
||
|
|
"step": 2085,
|
||
|
|
"valid_targets_mean": 6287.4,
|
||
|
|
"valid_targets_min": 643
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8928238583410995,
|
||
|
|
"grad_norm": 0.4696191174847012,
|
||
|
|
"learning_rate": 1.960071088036087e-05,
|
||
|
|
"loss": 0.2907,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1551734209060669,
|
||
|
|
"step": 2090,
|
||
|
|
"valid_targets_mean": 6234.9,
|
||
|
|
"valid_targets_min": 1123
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9021435228331782,
|
||
|
|
"grad_norm": 0.6982443969786916,
|
||
|
|
"learning_rate": 1.9507869922433244e-05,
|
||
|
|
"loss": 0.2634,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1464214026927948,
|
||
|
|
"step": 2095,
|
||
|
|
"valid_targets_mean": 5309.9,
|
||
|
|
"valid_targets_min": 787
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.911463187325256,
|
||
|
|
"grad_norm": 0.6480676914512138,
|
||
|
|
"learning_rate": 1.9415039574493482e-05,
|
||
|
|
"loss": 0.2727,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11281028389930725,
|
||
|
|
"step": 2100,
|
||
|
|
"valid_targets_mean": 4017.0,
|
||
|
|
"valid_targets_min": 547
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9207828518173344,
|
||
|
|
"grad_norm": 0.4826199943728114,
|
||
|
|
"learning_rate": 1.9322221837900387e-05,
|
||
|
|
"loss": 0.2815,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17623111605644226,
|
||
|
|
"step": 2105,
|
||
|
|
"valid_targets_mean": 6391.0,
|
||
|
|
"valid_targets_min": 883
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9301025163094128,
|
||
|
|
"grad_norm": 0.6058017347410541,
|
||
|
|
"learning_rate": 1.922941871374087e-05,
|
||
|
|
"loss": 0.2753,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11884057521820068,
|
||
|
|
"step": 2110,
|
||
|
|
"valid_targets_mean": 3350.4,
|
||
|
|
"valid_targets_min": 707
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.939422180801491,
|
||
|
|
"grad_norm": 0.4451921771186091,
|
||
|
|
"learning_rate": 1.9136632202786806e-05,
|
||
|
|
"loss": 0.2683,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11219266802072525,
|
||
|
|
"step": 2115,
|
||
|
|
"valid_targets_mean": 5268.8,
|
||
|
|
"valid_targets_min": 563
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9487418452935694,
|
||
|
|
"grad_norm": 0.6134382871279838,
|
||
|
|
"learning_rate": 1.9043864305451913e-05,
|
||
|
|
"loss": 0.2882,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14519618451595306,
|
||
|
|
"step": 2120,
|
||
|
|
"valid_targets_mean": 4630.2,
|
||
|
|
"valid_targets_min": 705
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9580615097856477,
|
||
|
|
"grad_norm": 0.48706548411604855,
|
||
|
|
"learning_rate": 1.8951117021748596e-05,
|
||
|
|
"loss": 0.2605,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15150219202041626,
|
||
|
|
"step": 2125,
|
||
|
|
"valid_targets_mean": 6955.8,
|
||
|
|
"valid_targets_min": 1116
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.967381174277726,
|
||
|
|
"grad_norm": 0.5186021315691236,
|
||
|
|
"learning_rate": 1.8858392351244875e-05,
|
||
|
|
"loss": 0.2737,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11331013590097427,
|
||
|
|
"step": 2130,
|
||
|
|
"valid_targets_mean": 4412.8,
|
||
|
|
"valid_targets_min": 688
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9767008387698044,
|
||
|
|
"grad_norm": 0.48786881769879664,
|
||
|
|
"learning_rate": 1.8765692293021196e-05,
|
||
|
|
"loss": 0.2608,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15675842761993408,
|
||
|
|
"step": 2135,
|
||
|
|
"valid_targets_mean": 7727.8,
|
||
|
|
"valid_targets_min": 694
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9860205032618827,
|
||
|
|
"grad_norm": 0.47320013953711443,
|
||
|
|
"learning_rate": 1.8673018845627416e-05,
|
||
|
|
"loss": 0.2558,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10385037958621979,
|
||
|
|
"step": 2140,
|
||
|
|
"valid_targets_mean": 4898.1,
|
||
|
|
"valid_targets_min": 596
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9953401677539606,
|
||
|
|
"grad_norm": 0.5062929181924658,
|
||
|
|
"learning_rate": 1.8580374007039682e-05,
|
||
|
|
"loss": 0.2543,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1441090703010559,
|
||
|
|
"step": 2145,
|
||
|
|
"valid_targets_mean": 5699.0,
|
||
|
|
"valid_targets_min": 655
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.003727865796831,
|
||
|
|
"grad_norm": 0.5803333218296234,
|
||
|
|
"learning_rate": 1.8487759774617343e-05,
|
||
|
|
"loss": 0.2803,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15436740219593048,
|
||
|
|
"step": 2150,
|
||
|
|
"valid_targets_mean": 5990.1,
|
||
|
|
"valid_targets_min": 826
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.01304753028891,
|
||
|
|
"grad_norm": 0.5587261690943286,
|
||
|
|
"learning_rate": 1.8395178145059894e-05,
|
||
|
|
"loss": 0.2792,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1152942031621933,
|
||
|
|
"step": 2155,
|
||
|
|
"valid_targets_mean": 4924.8,
|
||
|
|
"valid_targets_min": 618
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.022367194780988,
|
||
|
|
"grad_norm": 0.5320679194006216,
|
||
|
|
"learning_rate": 1.8302631114363947e-05,
|
||
|
|
"loss": 0.2567,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11048678308725357,
|
||
|
|
"step": 2160,
|
||
|
|
"valid_targets_mean": 4801.8,
|
||
|
|
"valid_targets_min": 573
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.031686859273067,
|
||
|
|
"grad_norm": 0.6354452547746888,
|
||
|
|
"learning_rate": 1.8210120677780184e-05,
|
||
|
|
"loss": 0.2747,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1954927146434784,
|
||
|
|
"step": 2165,
|
||
|
|
"valid_targets_mean": 6579.2,
|
||
|
|
"valid_targets_min": 893
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.0410065237651445,
|
||
|
|
"grad_norm": 0.7206844586350641,
|
||
|
|
"learning_rate": 1.8117648829770343e-05,
|
||
|
|
"loss": 0.2408,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12859874963760376,
|
||
|
|
"step": 2170,
|
||
|
|
"valid_targets_mean": 3069.2,
|
||
|
|
"valid_targets_min": 503
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.050326188257222,
|
||
|
|
"grad_norm": 0.5478254602639251,
|
||
|
|
"learning_rate": 1.8025217563964207e-05,
|
||
|
|
"loss": 0.2602,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15957587957382202,
|
||
|
|
"step": 2175,
|
||
|
|
"valid_targets_mean": 4061.1,
|
||
|
|
"valid_targets_min": 950
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.059645852749301,
|
||
|
|
"grad_norm": 0.6324021250650365,
|
||
|
|
"learning_rate": 1.793282887311665e-05,
|
||
|
|
"loss": 0.2802,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10431281477212906,
|
||
|
|
"step": 2180,
|
||
|
|
"valid_targets_mean": 3180.2,
|
||
|
|
"valid_targets_min": 777
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.068965517241379,
|
||
|
|
"grad_norm": 0.5251835665032747,
|
||
|
|
"learning_rate": 1.784048474906466e-05,
|
||
|
|
"loss": 0.2443,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11237050592899323,
|
||
|
|
"step": 2185,
|
||
|
|
"valid_targets_mean": 5693.2,
|
||
|
|
"valid_targets_min": 484
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.078285181733458,
|
||
|
|
"grad_norm": 0.44863755207833583,
|
||
|
|
"learning_rate": 1.7748187182684375e-05,
|
||
|
|
"loss": 0.2467,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09372510015964508,
|
||
|
|
"step": 2190,
|
||
|
|
"valid_targets_mean": 6607.4,
|
||
|
|
"valid_targets_min": 689
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.087604846225536,
|
||
|
|
"grad_norm": 0.5024788065936043,
|
||
|
|
"learning_rate": 1.7655938163848216e-05,
|
||
|
|
"loss": 0.2689,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09910226613283157,
|
||
|
|
"step": 2195,
|
||
|
|
"valid_targets_mean": 4260.9,
|
||
|
|
"valid_targets_min": 942
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.096924510717614,
|
||
|
|
"grad_norm": 0.5110069509227418,
|
||
|
|
"learning_rate": 1.7563739681381908e-05,
|
||
|
|
"loss": 0.2615,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12914255261421204,
|
||
|
|
"step": 2200,
|
||
|
|
"valid_targets_mean": 5826.6,
|
||
|
|
"valid_targets_min": 866
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.106244175209692,
|
||
|
|
"grad_norm": 0.576624608091089,
|
||
|
|
"learning_rate": 1.7471593723021684e-05,
|
||
|
|
"loss": 0.2574,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13167554140090942,
|
||
|
|
"step": 2205,
|
||
|
|
"valid_targets_mean": 3742.0,
|
||
|
|
"valid_targets_min": 528
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.115563839701771,
|
||
|
|
"grad_norm": 0.4827733206808235,
|
||
|
|
"learning_rate": 1.737950227537137e-05,
|
||
|
|
"loss": 0.2686,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18647488951683044,
|
||
|
|
"step": 2210,
|
||
|
|
"valid_targets_mean": 7610.8,
|
||
|
|
"valid_targets_min": 760
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.124883504193849,
|
||
|
|
"grad_norm": 0.5551894678935408,
|
||
|
|
"learning_rate": 1.7287467323859598e-05,
|
||
|
|
"loss": 0.2374,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14511461555957794,
|
||
|
|
"step": 2215,
|
||
|
|
"valid_targets_mean": 5611.1,
|
||
|
|
"valid_targets_min": 621
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.134203168685927,
|
||
|
|
"grad_norm": 0.5049555029902162,
|
||
|
|
"learning_rate": 1.7195490852696963e-05,
|
||
|
|
"loss": 0.2447,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13741421699523926,
|
||
|
|
"step": 2220,
|
||
|
|
"valid_targets_mean": 6353.0,
|
||
|
|
"valid_targets_min": 802
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.143522833178006,
|
||
|
|
"grad_norm": 0.579866821796645,
|
||
|
|
"learning_rate": 1.7103574844833266e-05,
|
||
|
|
"loss": 0.2516,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14753815531730652,
|
||
|
|
"step": 2225,
|
||
|
|
"valid_targets_mean": 4859.0,
|
||
|
|
"valid_targets_min": 929
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.1528424976700835,
|
||
|
|
"grad_norm": 0.5464896113948133,
|
||
|
|
"learning_rate": 1.701172128191478e-05,
|
||
|
|
"loss": 0.2546,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1342930942773819,
|
||
|
|
"step": 2230,
|
||
|
|
"valid_targets_mean": 4386.4,
|
||
|
|
"valid_targets_min": 694
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.162162162162162,
|
||
|
|
"grad_norm": 0.5898872394291468,
|
||
|
|
"learning_rate": 1.6919932144241493e-05,
|
||
|
|
"loss": 0.2773,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09483079612255096,
|
||
|
|
"step": 2235,
|
||
|
|
"valid_targets_mean": 4791.9,
|
||
|
|
"valid_targets_min": 576
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.17148182665424,
|
||
|
|
"grad_norm": 0.6705325965489486,
|
||
|
|
"learning_rate": 1.6828209410724423e-05,
|
||
|
|
"loss": 0.2669,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10487116873264313,
|
||
|
|
"step": 2240,
|
||
|
|
"valid_targets_mean": 2065.9,
|
||
|
|
"valid_targets_min": 473
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.180801491146319,
|
||
|
|
"grad_norm": 0.5340251209992367,
|
||
|
|
"learning_rate": 1.6736555058842967e-05,
|
||
|
|
"loss": 0.2528,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11475913971662521,
|
||
|
|
"step": 2245,
|
||
|
|
"valid_targets_mean": 5041.6,
|
||
|
|
"valid_targets_min": 568
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.190121155638397,
|
||
|
|
"grad_norm": 0.48570444100634097,
|
||
|
|
"learning_rate": 1.6644971064602266e-05,
|
||
|
|
"loss": 0.2369,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09356614202260971,
|
||
|
|
"step": 2250,
|
||
|
|
"valid_targets_mean": 5655.6,
|
||
|
|
"valid_targets_min": 904
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.199440820130476,
|
||
|
|
"grad_norm": 0.47392278033116964,
|
||
|
|
"learning_rate": 1.655345940249059e-05,
|
||
|
|
"loss": 0.2541,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08585932850837708,
|
||
|
|
"step": 2255,
|
||
|
|
"valid_targets_mean": 4776.4,
|
||
|
|
"valid_targets_min": 597
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.2087604846225535,
|
||
|
|
"grad_norm": 0.5624368723816141,
|
||
|
|
"learning_rate": 1.6462022045436778e-05,
|
||
|
|
"loss": 0.2493,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1162760853767395,
|
||
|
|
"step": 2260,
|
||
|
|
"valid_targets_mean": 4441.5,
|
||
|
|
"valid_targets_min": 873
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.218080149114632,
|
||
|
|
"grad_norm": 0.5711086611688477,
|
||
|
|
"learning_rate": 1.6370660964767704e-05,
|
||
|
|
"loss": 0.2615,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10824091732501984,
|
||
|
|
"step": 2265,
|
||
|
|
"valid_targets_mean": 4291.0,
|
||
|
|
"valid_targets_min": 605
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.22739981360671,
|
||
|
|
"grad_norm": 0.4906487814744683,
|
||
|
|
"learning_rate": 1.627937813016578e-05,
|
||
|
|
"loss": 0.2303,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13659150898456573,
|
||
|
|
"step": 2270,
|
||
|
|
"valid_targets_mean": 4575.6,
|
||
|
|
"valid_targets_min": 796
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.236719478098788,
|
||
|
|
"grad_norm": 0.4622484471541839,
|
||
|
|
"learning_rate": 1.6188175509626484e-05,
|
||
|
|
"loss": 0.2723,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07075522840023041,
|
||
|
|
"step": 2275,
|
||
|
|
"valid_targets_mean": 3664.4,
|
||
|
|
"valid_targets_min": 642
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.246039142590867,
|
||
|
|
"grad_norm": 0.6201432717928734,
|
||
|
|
"learning_rate": 1.6097055069415938e-05,
|
||
|
|
"loss": 0.2603,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.131620854139328,
|
||
|
|
"step": 2280,
|
||
|
|
"valid_targets_mean": 3895.9,
|
||
|
|
"valid_targets_min": 547
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.255358807082945,
|
||
|
|
"grad_norm": 0.61239815803087,
|
||
|
|
"learning_rate": 1.6006018774028497e-05,
|
||
|
|
"loss": 0.2666,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1612945795059204,
|
||
|
|
"step": 2285,
|
||
|
|
"valid_targets_mean": 3411.0,
|
||
|
|
"valid_targets_min": 526
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.264678471575023,
|
||
|
|
"grad_norm": 1.0951629398221896,
|
||
|
|
"learning_rate": 1.5915068586144426e-05,
|
||
|
|
"loss": 0.2594,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17076720297336578,
|
||
|
|
"step": 2290,
|
||
|
|
"valid_targets_mean": 6258.9,
|
||
|
|
"valid_targets_min": 5239
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.273998136067101,
|
||
|
|
"grad_norm": 0.5599955819041673,
|
||
|
|
"learning_rate": 1.5824206466587563e-05,
|
||
|
|
"loss": 0.2484,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16024473309516907,
|
||
|
|
"step": 2295,
|
||
|
|
"valid_targets_mean": 6416.5,
|
||
|
|
"valid_targets_min": 674
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.28331780055918,
|
||
|
|
"grad_norm": 0.5409203848430278,
|
||
|
|
"learning_rate": 1.5733434374283067e-05,
|
||
|
|
"loss": 0.2639,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1788056194782257,
|
||
|
|
"step": 2300,
|
||
|
|
"valid_targets_mean": 6314.5,
|
||
|
|
"valid_targets_min": 977
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.292637465051258,
|
||
|
|
"grad_norm": 0.5525340181441826,
|
||
|
|
"learning_rate": 1.5642754266215147e-05,
|
||
|
|
"loss": 0.249,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10343192517757416,
|
||
|
|
"step": 2305,
|
||
|
|
"valid_targets_mean": 4025.0,
|
||
|
|
"valid_targets_min": 581
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.301957129543337,
|
||
|
|
"grad_norm": 0.5639085873042597,
|
||
|
|
"learning_rate": 1.555216809738491e-05,
|
||
|
|
"loss": 0.2597,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13496622443199158,
|
||
|
|
"step": 2310,
|
||
|
|
"valid_targets_mean": 5392.5,
|
||
|
|
"valid_targets_min": 920
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.311276794035415,
|
||
|
|
"grad_norm": 0.4882279556351354,
|
||
|
|
"learning_rate": 1.5461677820768195e-05,
|
||
|
|
"loss": 0.2525,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10598690807819366,
|
||
|
|
"step": 2315,
|
||
|
|
"valid_targets_mean": 5273.5,
|
||
|
|
"valid_targets_min": 473
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.320596458527493,
|
||
|
|
"grad_norm": 0.49598199720276676,
|
||
|
|
"learning_rate": 1.537128538727349e-05,
|
||
|
|
"loss": 0.2595,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1132953092455864,
|
||
|
|
"step": 2320,
|
||
|
|
"valid_targets_mean": 5409.5,
|
||
|
|
"valid_targets_min": 596
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.329916123019571,
|
||
|
|
"grad_norm": 0.5169699688433299,
|
||
|
|
"learning_rate": 1.5280992745699807e-05,
|
||
|
|
"loss": 0.2684,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16021116077899933,
|
||
|
|
"step": 2325,
|
||
|
|
"valid_targets_mean": 6086.0,
|
||
|
|
"valid_targets_min": 867
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.339235787511649,
|
||
|
|
"grad_norm": 0.4504942343977889,
|
||
|
|
"learning_rate": 1.519080184269475e-05,
|
||
|
|
"loss": 0.2671,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12180132418870926,
|
||
|
|
"step": 2330,
|
||
|
|
"valid_targets_mean": 6874.4,
|
||
|
|
"valid_targets_min": 1069
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.348555452003728,
|
||
|
|
"grad_norm": 0.5686064279702613,
|
||
|
|
"learning_rate": 1.5100714622712503e-05,
|
||
|
|
"loss": 0.2576,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1288069486618042,
|
||
|
|
"step": 2335,
|
||
|
|
"valid_targets_mean": 3521.8,
|
||
|
|
"valid_targets_min": 743
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.357875116495806,
|
||
|
|
"grad_norm": 0.5806812624141529,
|
||
|
|
"learning_rate": 1.5010733027971905e-05,
|
||
|
|
"loss": 0.2892,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2506939172744751,
|
||
|
|
"step": 2340,
|
||
|
|
"valid_targets_mean": 6094.5,
|
||
|
|
"valid_targets_min": 791
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.367194780987885,
|
||
|
|
"grad_norm": 0.5446395068811772,
|
||
|
|
"learning_rate": 1.4920858998414584e-05,
|
||
|
|
"loss": 0.2434,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14984920620918274,
|
||
|
|
"step": 2345,
|
||
|
|
"valid_targets_mean": 6253.6,
|
||
|
|
"valid_targets_min": 838
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.3765144454799625,
|
||
|
|
"grad_norm": 0.5322905769855798,
|
||
|
|
"learning_rate": 1.4831094471663154e-05,
|
||
|
|
"loss": 0.2829,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16243310272693634,
|
||
|
|
"step": 2350,
|
||
|
|
"valid_targets_mean": 6074.1,
|
||
|
|
"valid_targets_min": 883
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.385834109972041,
|
||
|
|
"grad_norm": 0.5163708930094104,
|
||
|
|
"learning_rate": 1.4741441382979406e-05,
|
||
|
|
"loss": 0.2685,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10942017287015915,
|
||
|
|
"step": 2355,
|
||
|
|
"valid_targets_mean": 5099.8,
|
||
|
|
"valid_targets_min": 922
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.395153774464119,
|
||
|
|
"grad_norm": 0.7217313569207575,
|
||
|
|
"learning_rate": 1.4651901665222616e-05,
|
||
|
|
"loss": 0.2668,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13996443152427673,
|
||
|
|
"step": 2360,
|
||
|
|
"valid_targets_mean": 6237.0,
|
||
|
|
"valid_targets_min": 750
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.404473438956198,
|
||
|
|
"grad_norm": 0.7142568849785025,
|
||
|
|
"learning_rate": 1.456247724880786e-05,
|
||
|
|
"loss": 0.2809,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15958957374095917,
|
||
|
|
"step": 2365,
|
||
|
|
"valid_targets_mean": 3937.8,
|
||
|
|
"valid_targets_min": 412
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.413793103448276,
|
||
|
|
"grad_norm": 0.5317089283187401,
|
||
|
|
"learning_rate": 1.4473170061664371e-05,
|
||
|
|
"loss": 0.2627,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11956043541431427,
|
||
|
|
"step": 2370,
|
||
|
|
"valid_targets_mean": 4924.1,
|
||
|
|
"valid_targets_min": 702
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.423112767940355,
|
||
|
|
"grad_norm": 0.5526526615118859,
|
||
|
|
"learning_rate": 1.4383982029194035e-05,
|
||
|
|
"loss": 0.2638,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08924585580825806,
|
||
|
|
"step": 2375,
|
||
|
|
"valid_targets_mean": 2993.5,
|
||
|
|
"valid_targets_min": 658
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.4324324324324325,
|
||
|
|
"grad_norm": 0.5047233519724696,
|
||
|
|
"learning_rate": 1.4294915074229817e-05,
|
||
|
|
"loss": 0.2371,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14239254593849182,
|
||
|
|
"step": 2380,
|
||
|
|
"valid_targets_mean": 6312.4,
|
||
|
|
"valid_targets_min": 1003
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.44175209692451,
|
||
|
|
"grad_norm": 0.5326904699485265,
|
||
|
|
"learning_rate": 1.4205971116994357e-05,
|
||
|
|
"loss": 0.2746,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13368207216262817,
|
||
|
|
"step": 2385,
|
||
|
|
"valid_targets_mean": 6602.8,
|
||
|
|
"valid_targets_min": 4127
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.451071761416589,
|
||
|
|
"grad_norm": 0.5326319777790417,
|
||
|
|
"learning_rate": 1.4117152075058522e-05,
|
||
|
|
"loss": 0.232,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10195570439100266,
|
||
|
|
"step": 2390,
|
||
|
|
"valid_targets_mean": 5973.5,
|
||
|
|
"valid_targets_min": 680
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.460391425908667,
|
||
|
|
"grad_norm": 0.6127049935852104,
|
||
|
|
"learning_rate": 1.4028459863300112e-05,
|
||
|
|
"loss": 0.2432,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10925943404436111,
|
||
|
|
"step": 2395,
|
||
|
|
"valid_targets_mean": 4009.9,
|
||
|
|
"valid_targets_min": 688
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.469711090400746,
|
||
|
|
"grad_norm": 0.5773699620291571,
|
||
|
|
"learning_rate": 1.3939896393862544e-05,
|
||
|
|
"loss": 0.2661,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12614724040031433,
|
||
|
|
"step": 2400,
|
||
|
|
"valid_targets_mean": 5077.9,
|
||
|
|
"valid_targets_min": 591
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.479030754892824,
|
||
|
|
"grad_norm": 0.574995171906055,
|
||
|
|
"learning_rate": 1.3851463576113655e-05,
|
||
|
|
"loss": 0.264,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1420738846063614,
|
||
|
|
"step": 2405,
|
||
|
|
"valid_targets_mean": 5132.0,
|
||
|
|
"valid_targets_min": 1035
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.488350419384902,
|
||
|
|
"grad_norm": 0.6391897539788196,
|
||
|
|
"learning_rate": 1.3763163316604486e-05,
|
||
|
|
"loss": 0.2422,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09364959597587585,
|
||
|
|
"step": 2410,
|
||
|
|
"valid_targets_mean": 2804.4,
|
||
|
|
"valid_targets_min": 794
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.49767008387698,
|
||
|
|
"grad_norm": 0.5792051865469358,
|
||
|
|
"learning_rate": 1.367499751902825e-05,
|
||
|
|
"loss": 0.2449,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12628024816513062,
|
||
|
|
"step": 2415,
|
||
|
|
"valid_targets_mean": 5057.5,
|
||
|
|
"valid_targets_min": 839
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.506989748369059,
|
||
|
|
"grad_norm": 0.547089228099698,
|
||
|
|
"learning_rate": 1.3586968084179235e-05,
|
||
|
|
"loss": 0.2712,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1598655879497528,
|
||
|
|
"step": 2420,
|
||
|
|
"valid_targets_mean": 5351.6,
|
||
|
|
"valid_targets_min": 516
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.516309412861137,
|
||
|
|
"grad_norm": 0.6125735959640675,
|
||
|
|
"learning_rate": 1.3499076909911842e-05,
|
||
|
|
"loss": 0.2685,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14988814294338226,
|
||
|
|
"step": 2425,
|
||
|
|
"valid_targets_mean": 4922.1,
|
||
|
|
"valid_targets_min": 1129
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.525629077353216,
|
||
|
|
"grad_norm": 0.41810085251351536,
|
||
|
|
"learning_rate": 1.3411325891099679e-05,
|
||
|
|
"loss": 0.2558,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10689861327409744,
|
||
|
|
"step": 2430,
|
||
|
|
"valid_targets_mean": 6746.2,
|
||
|
|
"valid_targets_min": 1436
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.534948741845294,
|
||
|
|
"grad_norm": 0.5497466690305944,
|
||
|
|
"learning_rate": 1.3323716919594678e-05,
|
||
|
|
"loss": 0.268,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17994636297225952,
|
||
|
|
"step": 2435,
|
||
|
|
"valid_targets_mean": 6152.4,
|
||
|
|
"valid_targets_min": 752
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.5442684063373715,
|
||
|
|
"grad_norm": 0.5808022675012567,
|
||
|
|
"learning_rate": 1.3236251884186348e-05,
|
||
|
|
"loss": 0.2788,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1272009313106537,
|
||
|
|
"step": 2440,
|
||
|
|
"valid_targets_mean": 4654.0,
|
||
|
|
"valid_targets_min": 852
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.55358807082945,
|
||
|
|
"grad_norm": 0.5496064306151587,
|
||
|
|
"learning_rate": 1.3148932670561024e-05,
|
||
|
|
"loss": 0.2641,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.120310939848423,
|
||
|
|
"step": 2445,
|
||
|
|
"valid_targets_mean": 4881.2,
|
||
|
|
"valid_targets_min": 635
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.562907735321528,
|
||
|
|
"grad_norm": 0.5697997486658825,
|
||
|
|
"learning_rate": 1.3061761161261237e-05,
|
||
|
|
"loss": 0.2509,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15462273359298706,
|
||
|
|
"step": 2450,
|
||
|
|
"valid_targets_mean": 6361.9,
|
||
|
|
"valid_targets_min": 855
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.572227399813607,
|
||
|
|
"grad_norm": 0.4576103911543905,
|
||
|
|
"learning_rate": 1.2974739235645082e-05,
|
||
|
|
"loss": 0.2337,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12618505954742432,
|
||
|
|
"step": 2455,
|
||
|
|
"valid_targets_mean": 6699.5,
|
||
|
|
"valid_targets_min": 5017
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.581547064305685,
|
||
|
|
"grad_norm": 0.5551831601285778,
|
||
|
|
"learning_rate": 1.2887868769845769e-05,
|
||
|
|
"loss": 0.2716,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13530126214027405,
|
||
|
|
"step": 2460,
|
||
|
|
"valid_targets_mean": 5537.5,
|
||
|
|
"valid_targets_min": 883
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.590866728797764,
|
||
|
|
"grad_norm": 0.628765313956734,
|
||
|
|
"learning_rate": 1.2801151636731115e-05,
|
||
|
|
"loss": 0.2784,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15660974383354187,
|
||
|
|
"step": 2465,
|
||
|
|
"valid_targets_mean": 5022.2,
|
||
|
|
"valid_targets_min": 317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.6001863932898415,
|
||
|
|
"grad_norm": 0.5170668685745933,
|
||
|
|
"learning_rate": 1.2714589705863202e-05,
|
||
|
|
"loss": 0.257,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11045550554990768,
|
||
|
|
"step": 2470,
|
||
|
|
"valid_targets_mean": 4511.1,
|
||
|
|
"valid_targets_min": 760
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.609506057781919,
|
||
|
|
"grad_norm": 0.4552614956187894,
|
||
|
|
"learning_rate": 1.2628184843458043e-05,
|
||
|
|
"loss": 0.2608,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12863695621490479,
|
||
|
|
"step": 2475,
|
||
|
|
"valid_targets_mean": 7712.0,
|
||
|
|
"valid_targets_min": 4184
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.618825722273998,
|
||
|
|
"grad_norm": 0.5722470997865292,
|
||
|
|
"learning_rate": 1.2541938912345378e-05,
|
||
|
|
"loss": 0.2701,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12313509732484818,
|
||
|
|
"step": 2480,
|
||
|
|
"valid_targets_mean": 4727.0,
|
||
|
|
"valid_targets_min": 660
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.628145386766077,
|
||
|
|
"grad_norm": 0.7546610065867149,
|
||
|
|
"learning_rate": 1.2455853771928479e-05,
|
||
|
|
"loss": 0.2536,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11356036365032196,
|
||
|
|
"step": 2485,
|
||
|
|
"valid_targets_mean": 3108.6,
|
||
|
|
"valid_targets_min": 549
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.637465051258155,
|
||
|
|
"grad_norm": 0.6298143138160671,
|
||
|
|
"learning_rate": 1.2369931278144112e-05,
|
||
|
|
"loss": 0.2426,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08632612973451614,
|
||
|
|
"step": 2490,
|
||
|
|
"valid_targets_mean": 2548.5,
|
||
|
|
"valid_targets_min": 605
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.646784715750233,
|
||
|
|
"grad_norm": 0.5531529536272812,
|
||
|
|
"learning_rate": 1.2284173283422453e-05,
|
||
|
|
"loss": 0.2682,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1050497367978096,
|
||
|
|
"step": 2495,
|
||
|
|
"valid_targets_mean": 3490.9,
|
||
|
|
"valid_targets_min": 546
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.656104380242311,
|
||
|
|
"grad_norm": 0.545680745287173,
|
||
|
|
"learning_rate": 1.2198581636647214e-05,
|
||
|
|
"loss": 0.2306,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14393538236618042,
|
||
|
|
"step": 2500,
|
||
|
|
"valid_targets_mean": 6216.5,
|
||
|
|
"valid_targets_min": 778
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.665424044734389,
|
||
|
|
"grad_norm": 0.5353351507615532,
|
||
|
|
"learning_rate": 1.2113158183115763e-05,
|
||
|
|
"loss": 0.2914,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12321586161851883,
|
||
|
|
"step": 2505,
|
||
|
|
"valid_targets_mean": 6577.4,
|
||
|
|
"valid_targets_min": 710
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.674743709226468,
|
||
|
|
"grad_norm": 0.4507369283788035,
|
||
|
|
"learning_rate": 1.2027904764499324e-05,
|
||
|
|
"loss": 0.2442,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10029973834753036,
|
||
|
|
"step": 2510,
|
||
|
|
"valid_targets_mean": 4831.0,
|
||
|
|
"valid_targets_min": 589
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.684063373718546,
|
||
|
|
"grad_norm": 0.6475263125429359,
|
||
|
|
"learning_rate": 1.1942823218803293e-05,
|
||
|
|
"loss": 0.2445,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1443321406841278,
|
||
|
|
"step": 2515,
|
||
|
|
"valid_targets_mean": 3275.8,
|
||
|
|
"valid_targets_min": 661
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.693383038210625,
|
||
|
|
"grad_norm": 0.6099769872480744,
|
||
|
|
"learning_rate": 1.1857915380327593e-05,
|
||
|
|
"loss": 0.2436,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1476421356201172,
|
||
|
|
"step": 2520,
|
||
|
|
"valid_targets_mean": 4954.4,
|
||
|
|
"valid_targets_min": 604
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.702702702702703,
|
||
|
|
"grad_norm": 0.5997685585877369,
|
||
|
|
"learning_rate": 1.1773183079627148e-05,
|
||
|
|
"loss": 0.2567,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11305484920740128,
|
||
|
|
"step": 2525,
|
||
|
|
"valid_targets_mean": 3508.6,
|
||
|
|
"valid_targets_min": 615
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.7120223671947805,
|
||
|
|
"grad_norm": 0.5027009500992287,
|
||
|
|
"learning_rate": 1.1688628143472402e-05,
|
||
|
|
"loss": 0.2468,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1407850682735443,
|
||
|
|
"step": 2530,
|
||
|
|
"valid_targets_mean": 6297.2,
|
||
|
|
"valid_targets_min": 600
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.721342031686859,
|
||
|
|
"grad_norm": 0.6934690389597127,
|
||
|
|
"learning_rate": 1.1604252394809957e-05,
|
||
|
|
"loss": 0.2525,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12067228555679321,
|
||
|
|
"step": 2535,
|
||
|
|
"valid_targets_mean": 4050.5,
|
||
|
|
"valid_targets_min": 521
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.730661696178937,
|
||
|
|
"grad_norm": 0.49554451898144447,
|
||
|
|
"learning_rate": 1.1520057652723224e-05,
|
||
|
|
"loss": 0.2668,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12555411458015442,
|
||
|
|
"step": 2540,
|
||
|
|
"valid_targets_mean": 6002.0,
|
||
|
|
"valid_targets_min": 687
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.739981360671016,
|
||
|
|
"grad_norm": 0.5389152540890626,
|
||
|
|
"learning_rate": 1.1436045732393264e-05,
|
||
|
|
"loss": 0.2709,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12667539715766907,
|
||
|
|
"step": 2545,
|
||
|
|
"valid_targets_mean": 4982.9,
|
||
|
|
"valid_targets_min": 630
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.749301025163094,
|
||
|
|
"grad_norm": 0.4734334503201834,
|
||
|
|
"learning_rate": 1.135221844505961e-05,
|
||
|
|
"loss": 0.267,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12821856141090393,
|
||
|
|
"step": 2550,
|
||
|
|
"valid_targets_mean": 5849.5,
|
||
|
|
"valid_targets_min": 722
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.758620689655173,
|
||
|
|
"grad_norm": 0.5356576713862258,
|
||
|
|
"learning_rate": 1.1268577597981252e-05,
|
||
|
|
"loss": 0.2916,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23741666972637177,
|
||
|
|
"step": 2555,
|
||
|
|
"valid_targets_mean": 6091.5,
|
||
|
|
"valid_targets_min": 772
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.7679403541472505,
|
||
|
|
"grad_norm": 0.6426042544488947,
|
||
|
|
"learning_rate": 1.1185124994397625e-05,
|
||
|
|
"loss": 0.2588,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09970158338546753,
|
||
|
|
"step": 2560,
|
||
|
|
"valid_targets_mean": 2620.5,
|
||
|
|
"valid_targets_min": 583
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.777260018639329,
|
||
|
|
"grad_norm": 0.5099988161800889,
|
||
|
|
"learning_rate": 1.110186243348978e-05,
|
||
|
|
"loss": 0.2575,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15777605772018433,
|
||
|
|
"step": 2565,
|
||
|
|
"valid_targets_mean": 5695.2,
|
||
|
|
"valid_targets_min": 666
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.786579683131407,
|
||
|
|
"grad_norm": 0.5162035760502668,
|
||
|
|
"learning_rate": 1.1018791710341597e-05,
|
||
|
|
"loss": 0.2583,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13159963488578796,
|
||
|
|
"step": 2570,
|
||
|
|
"valid_targets_mean": 6662.0,
|
||
|
|
"valid_targets_min": 736
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.795899347623486,
|
||
|
|
"grad_norm": 0.4834235261543621,
|
||
|
|
"learning_rate": 1.0935914615901044e-05,
|
||
|
|
"loss": 0.265,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1334283947944641,
|
||
|
|
"step": 2575,
|
||
|
|
"valid_targets_mean": 6577.5,
|
||
|
|
"valid_targets_min": 933
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.805219012115564,
|
||
|
|
"grad_norm": 0.5924749815858056,
|
||
|
|
"learning_rate": 1.0853232936941579e-05,
|
||
|
|
"loss": 0.2586,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18895485997200012,
|
||
|
|
"step": 2580,
|
||
|
|
"valid_targets_mean": 5526.6,
|
||
|
|
"valid_targets_min": 806
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.814538676607642,
|
||
|
|
"grad_norm": 0.5031605235161631,
|
||
|
|
"learning_rate": 1.0770748456023651e-05,
|
||
|
|
"loss": 0.2333,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13236019015312195,
|
||
|
|
"step": 2585,
|
||
|
|
"valid_targets_mean": 6033.2,
|
||
|
|
"valid_targets_min": 694
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.8238583410997204,
|
||
|
|
"grad_norm": 0.5554082837244384,
|
||
|
|
"learning_rate": 1.0688462951456249e-05,
|
||
|
|
"loss": 0.2571,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12403716146945953,
|
||
|
|
"step": 2590,
|
||
|
|
"valid_targets_mean": 5450.1,
|
||
|
|
"valid_targets_min": 678
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.833178005591798,
|
||
|
|
"grad_norm": 0.501646945025073,
|
||
|
|
"learning_rate": 1.0606378197258563e-05,
|
||
|
|
"loss": 0.256,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1218869537115097,
|
||
|
|
"step": 2595,
|
||
|
|
"valid_targets_mean": 5977.0,
|
||
|
|
"valid_targets_min": 4813
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.842497670083877,
|
||
|
|
"grad_norm": 0.6220416913413397,
|
||
|
|
"learning_rate": 1.0524495963121767e-05,
|
||
|
|
"loss": 0.2626,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18620820343494415,
|
||
|
|
"step": 2600,
|
||
|
|
"valid_targets_mean": 3820.9,
|
||
|
|
"valid_targets_min": 664
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.851817334575955,
|
||
|
|
"grad_norm": 0.6109079432245437,
|
||
|
|
"learning_rate": 1.0442818014370805e-05,
|
||
|
|
"loss": 0.2621,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10935916006565094,
|
||
|
|
"step": 2605,
|
||
|
|
"valid_targets_mean": 5523.4,
|
||
|
|
"valid_targets_min": 622
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.861136999068034,
|
||
|
|
"grad_norm": 0.5514684366851011,
|
||
|
|
"learning_rate": 1.0361346111926391e-05,
|
||
|
|
"loss": 0.2535,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.146936297416687,
|
||
|
|
"step": 2610,
|
||
|
|
"valid_targets_mean": 5350.1,
|
||
|
|
"valid_targets_min": 574
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.870456663560112,
|
||
|
|
"grad_norm": 0.5933429343561408,
|
||
|
|
"learning_rate": 1.0280082012267015e-05,
|
||
|
|
"loss": 0.2376,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09740177541971207,
|
||
|
|
"step": 2615,
|
||
|
|
"valid_targets_mean": 4646.4,
|
||
|
|
"valid_targets_min": 676
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.87977632805219,
|
||
|
|
"grad_norm": 0.6427623493642532,
|
||
|
|
"learning_rate": 1.0199027467391077e-05,
|
||
|
|
"loss": 0.2879,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11310213804244995,
|
||
|
|
"step": 2620,
|
||
|
|
"valid_targets_mean": 3542.9,
|
||
|
|
"valid_targets_min": 657
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.889095992544268,
|
||
|
|
"grad_norm": 0.5413840594214038,
|
||
|
|
"learning_rate": 1.0118184224779126e-05,
|
||
|
|
"loss": 0.25,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1330389380455017,
|
||
|
|
"step": 2625,
|
||
|
|
"valid_targets_mean": 6137.1,
|
||
|
|
"valid_targets_min": 670
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.898415657036347,
|
||
|
|
"grad_norm": 0.492235295336191,
|
||
|
|
"learning_rate": 1.0037554027356177e-05,
|
||
|
|
"loss": 0.2616,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1202717050909996,
|
||
|
|
"step": 2630,
|
||
|
|
"valid_targets_mean": 5540.6,
|
||
|
|
"valid_targets_min": 866
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.907735321528425,
|
||
|
|
"grad_norm": 0.4813682371638276,
|
||
|
|
"learning_rate": 9.957138613454138e-06,
|
||
|
|
"loss": 0.2807,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11466509103775024,
|
||
|
|
"step": 2635,
|
||
|
|
"valid_targets_mean": 6431.2,
|
||
|
|
"valid_targets_min": 613
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.917054986020503,
|
||
|
|
"grad_norm": 0.5593137951298424,
|
||
|
|
"learning_rate": 9.876939716774332e-06,
|
||
|
|
"loss": 0.2622,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1478707492351532,
|
||
|
|
"step": 2640,
|
||
|
|
"valid_targets_mean": 5670.5,
|
||
|
|
"valid_targets_min": 961
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.926374650512582,
|
||
|
|
"grad_norm": 0.6708686631904994,
|
||
|
|
"learning_rate": 9.796959066350104e-06,
|
||
|
|
"loss": 0.2755,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25317347049713135,
|
||
|
|
"step": 2645,
|
||
|
|
"valid_targets_mean": 4340.0,
|
||
|
|
"valid_targets_min": 755
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.9356943150046595,
|
||
|
|
"grad_norm": 0.7008506954237591,
|
||
|
|
"learning_rate": 9.717198386509572e-06,
|
||
|
|
"loss": 0.2832,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20241795480251312,
|
||
|
|
"step": 2650,
|
||
|
|
"valid_targets_mean": 4713.8,
|
||
|
|
"valid_targets_min": 493
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.945013979496738,
|
||
|
|
"grad_norm": 0.5191217083194473,
|
||
|
|
"learning_rate": 9.63765939683845e-06,
|
||
|
|
"loss": 0.2493,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13980895280838013,
|
||
|
|
"step": 2655,
|
||
|
|
"valid_targets_mean": 6232.8,
|
||
|
|
"valid_targets_min": 638
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.954333643988816,
|
||
|
|
"grad_norm": 0.5901841551481329,
|
||
|
|
"learning_rate": 9.558343812142942e-06,
|
||
|
|
"loss": 0.2536,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12289948016405106,
|
||
|
|
"step": 2660,
|
||
|
|
"valid_targets_mean": 4783.9,
|
||
|
|
"valid_targets_min": 524
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.963653308480895,
|
||
|
|
"grad_norm": 0.5075172419995063,
|
||
|
|
"learning_rate": 9.479253342412815e-06,
|
||
|
|
"loss": 0.2507,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12621164321899414,
|
||
|
|
"step": 2665,
|
||
|
|
"valid_targets_mean": 5941.8,
|
||
|
|
"valid_targets_min": 603
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.972972972972973,
|
||
|
|
"grad_norm": 0.6574309970681059,
|
||
|
|
"learning_rate": 9.400389692784479e-06,
|
||
|
|
"loss": 0.2668,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13879956305027008,
|
||
|
|
"step": 2670,
|
||
|
|
"valid_targets_mean": 4276.4,
|
||
|
|
"valid_targets_min": 428
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.982292637465052,
|
||
|
|
"grad_norm": 0.6193899195087238,
|
||
|
|
"learning_rate": 9.321754563504288e-06,
|
||
|
|
"loss": 0.2741,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08522278070449829,
|
||
|
|
"step": 2675,
|
||
|
|
"valid_targets_mean": 2287.9,
|
||
|
|
"valid_targets_min": 621
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.9916123019571295,
|
||
|
|
"grad_norm": 0.5808948179763211,
|
||
|
|
"learning_rate": 9.243349649891842e-06,
|
||
|
|
"loss": 0.2459,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12116767466068268,
|
||
|
|
"step": 2680,
|
||
|
|
"valid_targets_mean": 4430.6,
|
||
|
|
"valid_targets_min": 524
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0,
|
||
|
|
"grad_norm": 1.5826164762201274,
|
||
|
|
"learning_rate": 9.16517664230345e-06,
|
||
|
|
"loss": 0.2727,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3509882688522339,
|
||
|
|
"step": 2685,
|
||
|
|
"valid_targets_mean": 2484.4,
|
||
|
|
"valid_targets_min": 574
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.009319664492078,
|
||
|
|
"grad_norm": 0.5446928802512481,
|
||
|
|
"learning_rate": 9.087237226095687e-06,
|
||
|
|
"loss": 0.2427,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13190782070159912,
|
||
|
|
"step": 2690,
|
||
|
|
"valid_targets_mean": 4848.4,
|
||
|
|
"valid_targets_min": 650
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.018639328984157,
|
||
|
|
"grad_norm": 0.4339716992060176,
|
||
|
|
"learning_rate": 9.009533081589055e-06,
|
||
|
|
"loss": 0.2333,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10168427973985672,
|
||
|
|
"step": 2695,
|
||
|
|
"valid_targets_mean": 7177.0,
|
||
|
|
"valid_targets_min": 373
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0279589934762345,
|
||
|
|
"grad_norm": 0.6155382174057061,
|
||
|
|
"learning_rate": 8.93206588403176e-06,
|
||
|
|
"loss": 0.2422,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14961549639701843,
|
||
|
|
"step": 2700,
|
||
|
|
"valid_targets_mean": 5373.0,
|
||
|
|
"valid_targets_min": 838
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.037278657968313,
|
||
|
|
"grad_norm": 0.5732292781063013,
|
||
|
|
"learning_rate": 8.854837303563594e-06,
|
||
|
|
"loss": 0.2548,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14827558398246765,
|
||
|
|
"step": 2705,
|
||
|
|
"valid_targets_mean": 5972.9,
|
||
|
|
"valid_targets_min": 583
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.046598322460391,
|
||
|
|
"grad_norm": 0.5741762163204598,
|
||
|
|
"learning_rate": 8.77784900517993e-06,
|
||
|
|
"loss": 0.2476,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13954892754554749,
|
||
|
|
"step": 2710,
|
||
|
|
"valid_targets_mean": 5267.4,
|
||
|
|
"valid_targets_min": 1036
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.05591798695247,
|
||
|
|
"grad_norm": 0.4849018199228435,
|
||
|
|
"learning_rate": 8.701102648695821e-06,
|
||
|
|
"loss": 0.2362,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10622663795948029,
|
||
|
|
"step": 2715,
|
||
|
|
"valid_targets_mean": 4957.6,
|
||
|
|
"valid_targets_min": 899
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.065237651444548,
|
||
|
|
"grad_norm": 0.5548921003454289,
|
||
|
|
"learning_rate": 8.624599888710217e-06,
|
||
|
|
"loss": 0.2512,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12010187655687332,
|
||
|
|
"step": 2720,
|
||
|
|
"valid_targets_mean": 4979.1,
|
||
|
|
"valid_targets_min": 677
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.074557315936627,
|
||
|
|
"grad_norm": 0.47455323629666346,
|
||
|
|
"learning_rate": 8.548342374570304e-06,
|
||
|
|
"loss": 0.242,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09629485011100769,
|
||
|
|
"step": 2725,
|
||
|
|
"valid_targets_mean": 6260.6,
|
||
|
|
"valid_targets_min": 1202
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0838769804287045,
|
||
|
|
"grad_norm": 0.57257158055579,
|
||
|
|
"learning_rate": 8.472331750335913e-06,
|
||
|
|
"loss": 0.25,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19756776094436646,
|
||
|
|
"step": 2730,
|
||
|
|
"valid_targets_mean": 6789.9,
|
||
|
|
"valid_targets_min": 850
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.093196644920783,
|
||
|
|
"grad_norm": 0.5302109963359262,
|
||
|
|
"learning_rate": 8.396569654744117e-06,
|
||
|
|
"loss": 0.2608,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.135857492685318,
|
||
|
|
"step": 2735,
|
||
|
|
"valid_targets_mean": 4630.2,
|
||
|
|
"valid_targets_min": 933
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.102516309412861,
|
||
|
|
"grad_norm": 0.5898112029863783,
|
||
|
|
"learning_rate": 8.321057721173873e-06,
|
||
|
|
"loss": 0.2551,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1367460936307907,
|
||
|
|
"step": 2740,
|
||
|
|
"valid_targets_mean": 5449.5,
|
||
|
|
"valid_targets_min": 734
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.111835973904939,
|
||
|
|
"grad_norm": 0.5588460554852376,
|
||
|
|
"learning_rate": 8.24579757761083e-06,
|
||
|
|
"loss": 0.2408,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1399080753326416,
|
||
|
|
"step": 2745,
|
||
|
|
"valid_targets_mean": 5290.4,
|
||
|
|
"valid_targets_min": 615
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.121155638397018,
|
||
|
|
"grad_norm": 0.4950169386526473,
|
||
|
|
"learning_rate": 8.170790846612205e-06,
|
||
|
|
"loss": 0.2751,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1232457160949707,
|
||
|
|
"step": 2750,
|
||
|
|
"valid_targets_mean": 6646.0,
|
||
|
|
"valid_targets_min": 5512
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.130475302889096,
|
||
|
|
"grad_norm": 0.5475198167495757,
|
||
|
|
"learning_rate": 8.096039145271802e-06,
|
||
|
|
"loss": 0.2339,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10603508353233337,
|
||
|
|
"step": 2755,
|
||
|
|
"valid_targets_mean": 4575.8,
|
||
|
|
"valid_targets_min": 888
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.1397949673811745,
|
||
|
|
"grad_norm": 0.5152217482451276,
|
||
|
|
"learning_rate": 8.021544085185178e-06,
|
||
|
|
"loss": 0.2436,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13730637729167938,
|
||
|
|
"step": 2760,
|
||
|
|
"valid_targets_mean": 6309.9,
|
||
|
|
"valid_targets_min": 869
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.149114631873252,
|
||
|
|
"grad_norm": 0.6992152016366882,
|
||
|
|
"learning_rate": 7.947307272414874e-06,
|
||
|
|
"loss": 0.2293,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12406228482723236,
|
||
|
|
"step": 2765,
|
||
|
|
"valid_targets_mean": 3112.2,
|
||
|
|
"valid_targets_min": 796
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.158434296365331,
|
||
|
|
"grad_norm": 0.5385720151621929,
|
||
|
|
"learning_rate": 7.873330307455797e-06,
|
||
|
|
"loss": 0.2316,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12931911647319794,
|
||
|
|
"step": 2770,
|
||
|
|
"valid_targets_mean": 4058.1,
|
||
|
|
"valid_targets_min": 763
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.167753960857409,
|
||
|
|
"grad_norm": 0.6318395992403971,
|
||
|
|
"learning_rate": 7.799614785200713e-06,
|
||
|
|
"loss": 0.2439,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13588440418243408,
|
||
|
|
"step": 2775,
|
||
|
|
"valid_targets_mean": 6541.0,
|
||
|
|
"valid_targets_min": 2347
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.177073625349488,
|
||
|
|
"grad_norm": 0.5473103801716807,
|
||
|
|
"learning_rate": 7.726162294905857e-06,
|
||
|
|
"loss": 0.261,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14540940523147583,
|
||
|
|
"step": 2780,
|
||
|
|
"valid_targets_mean": 5968.9,
|
||
|
|
"valid_targets_min": 471
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.186393289841566,
|
||
|
|
"grad_norm": 0.6175682183493182,
|
||
|
|
"learning_rate": 7.65297442015668e-06,
|
||
|
|
"loss": 0.2568,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10746894776821136,
|
||
|
|
"step": 2785,
|
||
|
|
"valid_targets_mean": 4411.5,
|
||
|
|
"valid_targets_min": 959
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.1957129543336436,
|
||
|
|
"grad_norm": 0.556555407177781,
|
||
|
|
"learning_rate": 7.58005273883371e-06,
|
||
|
|
"loss": 0.2378,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10153622925281525,
|
||
|
|
"step": 2790,
|
||
|
|
"valid_targets_mean": 5584.4,
|
||
|
|
"valid_targets_min": 771
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.205032618825722,
|
||
|
|
"grad_norm": 0.5693309729815054,
|
||
|
|
"learning_rate": 7.507398823078495e-06,
|
||
|
|
"loss": 0.262,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1400681734085083,
|
||
|
|
"step": 2795,
|
||
|
|
"valid_targets_mean": 4346.9,
|
||
|
|
"valid_targets_min": 452
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.2143522833178,
|
||
|
|
"grad_norm": 0.632076256556578,
|
||
|
|
"learning_rate": 7.4350142392597855e-06,
|
||
|
|
"loss": 0.2381,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14469283819198608,
|
||
|
|
"step": 2800,
|
||
|
|
"valid_targets_mean": 4177.5,
|
||
|
|
"valid_targets_min": 290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.223671947809879,
|
||
|
|
"grad_norm": 0.5353830035069947,
|
||
|
|
"learning_rate": 7.362900547939693e-06,
|
||
|
|
"loss": 0.2343,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11445467919111252,
|
||
|
|
"step": 2805,
|
||
|
|
"valid_targets_mean": 6580.9,
|
||
|
|
"valid_targets_min": 458
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.232991612301957,
|
||
|
|
"grad_norm": 0.6188342364606748,
|
||
|
|
"learning_rate": 7.291059303840082e-06,
|
||
|
|
"loss": 0.2417,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10043007135391235,
|
||
|
|
"step": 2810,
|
||
|
|
"valid_targets_mean": 3242.0,
|
||
|
|
"valid_targets_min": 724
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.242311276794036,
|
||
|
|
"grad_norm": 1.5052539510364185,
|
||
|
|
"learning_rate": 7.219492055809023e-06,
|
||
|
|
"loss": 0.2439,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10451781004667282,
|
||
|
|
"step": 2815,
|
||
|
|
"valid_targets_mean": 3295.0,
|
||
|
|
"valid_targets_min": 896
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.2516309412861135,
|
||
|
|
"grad_norm": 0.6464045132518985,
|
||
|
|
"learning_rate": 7.148200346787437e-06,
|
||
|
|
"loss": 0.2318,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13841761648654938,
|
||
|
|
"step": 2820,
|
||
|
|
"valid_targets_mean": 4806.8,
|
||
|
|
"valid_targets_min": 550
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.260950605778192,
|
||
|
|
"grad_norm": 0.5709213910319424,
|
||
|
|
"learning_rate": 7.0771857137758025e-06,
|
||
|
|
"loss": 0.2513,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1252596378326416,
|
||
|
|
"step": 2825,
|
||
|
|
"valid_targets_mean": 5760.1,
|
||
|
|
"valid_targets_min": 739
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.27027027027027,
|
||
|
|
"grad_norm": 0.47491495993571964,
|
||
|
|
"learning_rate": 7.0064496878010466e-06,
|
||
|
|
"loss": 0.2566,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15263473987579346,
|
||
|
|
"step": 2830,
|
||
|
|
"valid_targets_mean": 7142.2,
|
||
|
|
"valid_targets_min": 687
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.279589934762349,
|
||
|
|
"grad_norm": 0.6086165618513141,
|
||
|
|
"learning_rate": 6.935993793883509e-06,
|
||
|
|
"loss": 0.2515,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11325597763061523,
|
||
|
|
"step": 2835,
|
||
|
|
"valid_targets_mean": 3494.4,
|
||
|
|
"valid_targets_min": 628
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.288909599254427,
|
||
|
|
"grad_norm": 0.5376603467083659,
|
||
|
|
"learning_rate": 6.865819551004051e-06,
|
||
|
|
"loss": 0.2487,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09230548143386841,
|
||
|
|
"step": 2840,
|
||
|
|
"valid_targets_mean": 4799.1,
|
||
|
|
"valid_targets_min": 614
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.298229263746505,
|
||
|
|
"grad_norm": 0.6311049604068126,
|
||
|
|
"learning_rate": 6.795928472071363e-06,
|
||
|
|
"loss": 0.2292,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11850068718194962,
|
||
|
|
"step": 2845,
|
||
|
|
"valid_targets_mean": 3971.4,
|
||
|
|
"valid_targets_min": 820
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.3075489282385835,
|
||
|
|
"grad_norm": 0.6290010758898869,
|
||
|
|
"learning_rate": 6.726322063889299e-06,
|
||
|
|
"loss": 0.2406,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12725819647312164,
|
||
|
|
"step": 2850,
|
||
|
|
"valid_targets_mean": 3878.8,
|
||
|
|
"valid_targets_min": 641
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.316868592730661,
|
||
|
|
"grad_norm": 0.6228366269635568,
|
||
|
|
"learning_rate": 6.6570018271244096e-06,
|
||
|
|
"loss": 0.2315,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14573904871940613,
|
||
|
|
"step": 2855,
|
||
|
|
"valid_targets_mean": 5305.1,
|
||
|
|
"valid_targets_min": 682
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.32618825722274,
|
||
|
|
"grad_norm": 0.5163619404979787,
|
||
|
|
"learning_rate": 6.5879692562735854e-06,
|
||
|
|
"loss": 0.2571,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11391949653625488,
|
||
|
|
"step": 2860,
|
||
|
|
"valid_targets_mean": 4657.1,
|
||
|
|
"valid_targets_min": 562
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.335507921714818,
|
||
|
|
"grad_norm": 0.5413857533427078,
|
||
|
|
"learning_rate": 6.519225839631833e-06,
|
||
|
|
"loss": 0.2567,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06618350744247437,
|
||
|
|
"step": 2865,
|
||
|
|
"valid_targets_mean": 3183.0,
|
||
|
|
"valid_targets_min": 581
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.344827586206897,
|
||
|
|
"grad_norm": 0.6297096633253135,
|
||
|
|
"learning_rate": 6.4507730592602e-06,
|
||
|
|
"loss": 0.2517,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11674400418996811,
|
||
|
|
"step": 2870,
|
||
|
|
"valid_targets_mean": 3327.0,
|
||
|
|
"valid_targets_min": 757
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.354147250698975,
|
||
|
|
"grad_norm": 0.6199033224305049,
|
||
|
|
"learning_rate": 6.382612390953813e-06,
|
||
|
|
"loss": 0.2613,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1019492894411087,
|
||
|
|
"step": 2875,
|
||
|
|
"valid_targets_mean": 3417.9,
|
||
|
|
"valid_targets_min": 569
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.3634669151910535,
|
||
|
|
"grad_norm": 0.5644776654906836,
|
||
|
|
"learning_rate": 6.3147453042100395e-06,
|
||
|
|
"loss": 0.2454,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11583049595355988,
|
||
|
|
"step": 2880,
|
||
|
|
"valid_targets_mean": 4623.0,
|
||
|
|
"valid_targets_min": 785
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.372786579683131,
|
||
|
|
"grad_norm": 0.7092931571501191,
|
||
|
|
"learning_rate": 6.247173262196871e-06,
|
||
|
|
"loss": 0.2508,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10310398042201996,
|
||
|
|
"step": 2885,
|
||
|
|
"valid_targets_mean": 3071.5,
|
||
|
|
"valid_targets_min": 737
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.38210624417521,
|
||
|
|
"grad_norm": 0.5044054391940849,
|
||
|
|
"learning_rate": 6.179897721721304e-06,
|
||
|
|
"loss": 0.222,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12634506821632385,
|
||
|
|
"step": 2890,
|
||
|
|
"valid_targets_mean": 6382.5,
|
||
|
|
"valid_targets_min": 509
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.391425908667288,
|
||
|
|
"grad_norm": 0.5769425650928719,
|
||
|
|
"learning_rate": 6.112920133197979e-06,
|
||
|
|
"loss": 0.2527,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09064634889364243,
|
||
|
|
"step": 2895,
|
||
|
|
"valid_targets_mean": 3814.9,
|
||
|
|
"valid_targets_min": 699
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.400745573159366,
|
||
|
|
"grad_norm": 0.6222024265922106,
|
||
|
|
"learning_rate": 6.046241940617896e-06,
|
||
|
|
"loss": 0.2272,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06750698387622833,
|
||
|
|
"step": 2900,
|
||
|
|
"valid_targets_mean": 2168.8,
|
||
|
|
"valid_targets_min": 530
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.410065237651445,
|
||
|
|
"grad_norm": 0.5282166275724683,
|
||
|
|
"learning_rate": 5.979864581517267e-06,
|
||
|
|
"loss": 0.2288,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13910692930221558,
|
||
|
|
"step": 2905,
|
||
|
|
"valid_targets_mean": 6013.1,
|
||
|
|
"valid_targets_min": 721
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.4193849021435225,
|
||
|
|
"grad_norm": 0.5787890337888052,
|
||
|
|
"learning_rate": 5.913789486946555e-06,
|
||
|
|
"loss": 0.2466,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08422854542732239,
|
||
|
|
"step": 2910,
|
||
|
|
"valid_targets_mean": 3924.2,
|
||
|
|
"valid_targets_min": 754
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.428704566635601,
|
||
|
|
"grad_norm": 0.6526291444606194,
|
||
|
|
"learning_rate": 5.848018081439612e-06,
|
||
|
|
"loss": 0.2582,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1500834822654724,
|
||
|
|
"step": 2915,
|
||
|
|
"valid_targets_mean": 5496.5,
|
||
|
|
"valid_targets_min": 853
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.438024231127679,
|
||
|
|
"grad_norm": 0.49432005935121437,
|
||
|
|
"learning_rate": 5.782551782982959e-06,
|
||
|
|
"loss": 0.2542,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1052955761551857,
|
||
|
|
"step": 2920,
|
||
|
|
"valid_targets_mean": 6369.8,
|
||
|
|
"valid_targets_min": 658
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.447343895619758,
|
||
|
|
"grad_norm": 0.5732140371454704,
|
||
|
|
"learning_rate": 5.7173920029851934e-06,
|
||
|
|
"loss": 0.2278,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09518011659383774,
|
||
|
|
"step": 2925,
|
||
|
|
"valid_targets_mean": 4578.8,
|
||
|
|
"valid_targets_min": 575
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.456663560111836,
|
||
|
|
"grad_norm": 0.5443785584942344,
|
||
|
|
"learning_rate": 5.652540146246615e-06,
|
||
|
|
"loss": 0.2418,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10775406658649445,
|
||
|
|
"step": 2930,
|
||
|
|
"valid_targets_mean": 4634.6,
|
||
|
|
"valid_targets_min": 685
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.465983224603915,
|
||
|
|
"grad_norm": 0.5842793535588633,
|
||
|
|
"learning_rate": 5.587997610928893e-06,
|
||
|
|
"loss": 0.2579,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15116369724273682,
|
||
|
|
"step": 2935,
|
||
|
|
"valid_targets_mean": 5165.0,
|
||
|
|
"valid_targets_min": 624
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.4753028890959925,
|
||
|
|
"grad_norm": 0.5368233154307472,
|
||
|
|
"learning_rate": 5.523765788524941e-06,
|
||
|
|
"loss": 0.2247,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13675621151924133,
|
||
|
|
"step": 2940,
|
||
|
|
"valid_targets_mean": 5429.4,
|
||
|
|
"valid_targets_min": 590
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.484622553588071,
|
||
|
|
"grad_norm": 0.5374013058103315,
|
||
|
|
"learning_rate": 5.459846063828922e-06,
|
||
|
|
"loss": 0.2551,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14966246485710144,
|
||
|
|
"step": 2945,
|
||
|
|
"valid_targets_mean": 6183.8,
|
||
|
|
"valid_targets_min": 815
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.493942218080149,
|
||
|
|
"grad_norm": 0.6026905257697461,
|
||
|
|
"learning_rate": 5.396239814906374e-06,
|
||
|
|
"loss": 0.2505,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09505718946456909,
|
||
|
|
"step": 2950,
|
||
|
|
"valid_targets_mean": 3829.0,
|
||
|
|
"valid_targets_min": 811
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.503261882572227,
|
||
|
|
"grad_norm": 0.5860851719937075,
|
||
|
|
"learning_rate": 5.332948413064521e-06,
|
||
|
|
"loss": 0.2637,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12456689029932022,
|
||
|
|
"step": 2955,
|
||
|
|
"valid_targets_mean": 4199.9,
|
||
|
|
"valid_targets_min": 647
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.512581547064306,
|
||
|
|
"grad_norm": 0.7718382325159915,
|
||
|
|
"learning_rate": 5.269973222822698e-06,
|
||
|
|
"loss": 0.252,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12590637803077698,
|
||
|
|
"step": 2960,
|
||
|
|
"valid_targets_mean": 3329.8,
|
||
|
|
"valid_targets_min": 582
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.521901211556384,
|
||
|
|
"grad_norm": 0.5285277051856454,
|
||
|
|
"learning_rate": 5.207315601882914e-06,
|
||
|
|
"loss": 0.2676,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16569626331329346,
|
||
|
|
"step": 2965,
|
||
|
|
"valid_targets_mean": 6479.4,
|
||
|
|
"valid_targets_min": 714
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.5312208760484625,
|
||
|
|
"grad_norm": 0.7592176221496588,
|
||
|
|
"learning_rate": 5.14497690110064e-06,
|
||
|
|
"loss": 0.2729,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0808776319026947,
|
||
|
|
"step": 2970,
|
||
|
|
"valid_targets_mean": 2930.8,
|
||
|
|
"valid_targets_min": 526
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.54054054054054,
|
||
|
|
"grad_norm": 0.6374927836054771,
|
||
|
|
"learning_rate": 5.0829584644556186e-06,
|
||
|
|
"loss": 0.25,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14162445068359375,
|
||
|
|
"step": 2975,
|
||
|
|
"valid_targets_mean": 4308.5,
|
||
|
|
"valid_targets_min": 659
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.549860205032619,
|
||
|
|
"grad_norm": 0.6045170148546949,
|
||
|
|
"learning_rate": 5.021261629022924e-06,
|
||
|
|
"loss": 0.2766,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18829691410064697,
|
||
|
|
"step": 2980,
|
||
|
|
"valid_targets_mean": 5733.6,
|
||
|
|
"valid_targets_min": 751
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.559179869524697,
|
||
|
|
"grad_norm": 0.5583993607966227,
|
||
|
|
"learning_rate": 4.959887724944132e-06,
|
||
|
|
"loss": 0.2509,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.145821213722229,
|
||
|
|
"step": 2985,
|
||
|
|
"valid_targets_mean": 5123.0,
|
||
|
|
"valid_targets_min": 762
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.568499534016776,
|
||
|
|
"grad_norm": 0.5091250839065043,
|
||
|
|
"learning_rate": 4.8988380753986245e-06,
|
||
|
|
"loss": 0.2465,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09573192149400711,
|
||
|
|
"step": 2990,
|
||
|
|
"valid_targets_mean": 4664.5,
|
||
|
|
"valid_targets_min": 658
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.577819198508854,
|
||
|
|
"grad_norm": 0.6072728971016825,
|
||
|
|
"learning_rate": 4.838113996575094e-06,
|
||
|
|
"loss": 0.2543,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09102959930896759,
|
||
|
|
"step": 2995,
|
||
|
|
"valid_targets_mean": 3216.6,
|
||
|
|
"valid_targets_min": 613
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.587138863000932,
|
||
|
|
"grad_norm": 0.5784332290816027,
|
||
|
|
"learning_rate": 4.777716797643137e-06,
|
||
|
|
"loss": 0.2419,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16593733429908752,
|
||
|
|
"step": 3000,
|
||
|
|
"valid_targets_mean": 5804.4,
|
||
|
|
"valid_targets_min": 626
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.59645852749301,
|
||
|
|
"grad_norm": 1.4322070132147897,
|
||
|
|
"learning_rate": 4.717647780725072e-06,
|
||
|
|
"loss": 0.2431,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17298297584056854,
|
||
|
|
"step": 3005,
|
||
|
|
"valid_targets_mean": 5308.0,
|
||
|
|
"valid_targets_min": 614
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.605778191985088,
|
||
|
|
"grad_norm": 0.5965030025720491,
|
||
|
|
"learning_rate": 4.657908240867799e-06,
|
||
|
|
"loss": 0.2474,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14895585179328918,
|
||
|
|
"step": 3010,
|
||
|
|
"valid_targets_mean": 6374.2,
|
||
|
|
"valid_targets_min": 3879
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.615097856477167,
|
||
|
|
"grad_norm": 0.5515717035218127,
|
||
|
|
"learning_rate": 4.598499466014939e-06,
|
||
|
|
"loss": 0.2673,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09172220528125763,
|
||
|
|
"step": 3015,
|
||
|
|
"valid_targets_mean": 3808.2,
|
||
|
|
"valid_targets_min": 562
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.624417520969245,
|
||
|
|
"grad_norm": 0.5596904383074023,
|
||
|
|
"learning_rate": 4.539422736979043e-06,
|
||
|
|
"loss": 0.257,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09699073433876038,
|
||
|
|
"step": 3020,
|
||
|
|
"valid_targets_mean": 3534.8,
|
||
|
|
"valid_targets_min": 813
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.633737185461324,
|
||
|
|
"grad_norm": 0.5005101075456648,
|
||
|
|
"learning_rate": 4.480679327413982e-06,
|
||
|
|
"loss": 0.2649,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13157695531845093,
|
||
|
|
"step": 3025,
|
||
|
|
"valid_targets_mean": 6344.0,
|
||
|
|
"valid_targets_min": 697
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.6430568499534015,
|
||
|
|
"grad_norm": 0.629730171356339,
|
||
|
|
"learning_rate": 4.422270503787487e-06,
|
||
|
|
"loss": 0.2399,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09453202784061432,
|
||
|
|
"step": 3030,
|
||
|
|
"valid_targets_mean": 3646.2,
|
||
|
|
"valid_targets_min": 836
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.65237651444548,
|
||
|
|
"grad_norm": 0.5070177733470828,
|
||
|
|
"learning_rate": 4.364197525353842e-06,
|
||
|
|
"loss": 0.2509,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12108378857374191,
|
||
|
|
"step": 3035,
|
||
|
|
"valid_targets_mean": 6703.4,
|
||
|
|
"valid_targets_min": 643
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.661696178937558,
|
||
|
|
"grad_norm": 0.5613589850949523,
|
||
|
|
"learning_rate": 4.30646164412674e-06,
|
||
|
|
"loss": 0.2434,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10456252843141556,
|
||
|
|
"step": 3040,
|
||
|
|
"valid_targets_mean": 5367.0,
|
||
|
|
"valid_targets_min": 942
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.671015843429636,
|
||
|
|
"grad_norm": 0.5537104295242506,
|
||
|
|
"learning_rate": 4.249064104852299e-06,
|
||
|
|
"loss": 0.2291,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11175411939620972,
|
||
|
|
"step": 3045,
|
||
|
|
"valid_targets_mean": 7357.6,
|
||
|
|
"valid_targets_min": 680
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.680335507921715,
|
||
|
|
"grad_norm": 0.6461501759976522,
|
||
|
|
"learning_rate": 4.19200614498219e-06,
|
||
|
|
"loss": 0.2582,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1421043425798416,
|
||
|
|
"step": 3050,
|
||
|
|
"valid_targets_mean": 4899.8,
|
||
|
|
"valid_targets_min": 612
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.689655172413794,
|
||
|
|
"grad_norm": 0.5489666732800522,
|
||
|
|
"learning_rate": 4.1352889946470024e-06,
|
||
|
|
"loss": 0.2451,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.151632621884346,
|
||
|
|
"step": 3055,
|
||
|
|
"valid_targets_mean": 4928.5,
|
||
|
|
"valid_targets_min": 537
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.6989748369058715,
|
||
|
|
"grad_norm": 0.5554761436388534,
|
||
|
|
"learning_rate": 4.078913876629716e-06,
|
||
|
|
"loss": 0.235,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08918242156505585,
|
||
|
|
"step": 3060,
|
||
|
|
"valid_targets_mean": 4231.2,
|
||
|
|
"valid_targets_min": 594
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.708294501397949,
|
||
|
|
"grad_norm": 0.5225377639680258,
|
||
|
|
"learning_rate": 4.0228820063393055e-06,
|
||
|
|
"loss": 0.2468,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15499523282051086,
|
||
|
|
"step": 3065,
|
||
|
|
"valid_targets_mean": 6714.0,
|
||
|
|
"valid_targets_min": 4366
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.717614165890028,
|
||
|
|
"grad_norm": 0.6492690860176612,
|
||
|
|
"learning_rate": 3.967194591784578e-06,
|
||
|
|
"loss": 0.243,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13059458136558533,
|
||
|
|
"step": 3070,
|
||
|
|
"valid_targets_mean": 4396.4,
|
||
|
|
"valid_targets_min": 629
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.726933830382106,
|
||
|
|
"grad_norm": 0.5187176183366539,
|
||
|
|
"learning_rate": 3.911852833548086e-06,
|
||
|
|
"loss": 0.2687,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09858296811580658,
|
||
|
|
"step": 3075,
|
||
|
|
"valid_targets_mean": 5127.8,
|
||
|
|
"valid_targets_min": 774
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.736253494874185,
|
||
|
|
"grad_norm": 0.5833340221223913,
|
||
|
|
"learning_rate": 3.856857924760296e-06,
|
||
|
|
"loss": 0.2271,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10751400887966156,
|
||
|
|
"step": 3080,
|
||
|
|
"valid_targets_mean": 4529.0,
|
||
|
|
"valid_targets_min": 644
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.745573159366263,
|
||
|
|
"grad_norm": 0.5228906113924112,
|
||
|
|
"learning_rate": 3.802211051073814e-06,
|
||
|
|
"loss": 0.2326,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1210096925497055,
|
||
|
|
"step": 3085,
|
||
|
|
"valid_targets_mean": 6070.0,
|
||
|
|
"valid_targets_min": 2414
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.754892823858341,
|
||
|
|
"grad_norm": 0.5627676913557024,
|
||
|
|
"learning_rate": 3.7479133906378783e-06,
|
||
|
|
"loss": 0.2562,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10303059220314026,
|
||
|
|
"step": 3090,
|
||
|
|
"valid_targets_mean": 4547.5,
|
||
|
|
"valid_targets_min": 614
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.764212488350419,
|
||
|
|
"grad_norm": 0.50630008669155,
|
||
|
|
"learning_rate": 3.6939661140728887e-06,
|
||
|
|
"loss": 0.228,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10199972987174988,
|
||
|
|
"step": 3095,
|
||
|
|
"valid_targets_mean": 5480.8,
|
||
|
|
"valid_targets_min": 735
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.773532152842497,
|
||
|
|
"grad_norm": 0.6607120947447211,
|
||
|
|
"learning_rate": 3.6403703844452374e-06,
|
||
|
|
"loss": 0.2514,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12983489036560059,
|
||
|
|
"step": 3100,
|
||
|
|
"valid_targets_mean": 4719.1,
|
||
|
|
"valid_targets_min": 593
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.782851817334576,
|
||
|
|
"grad_norm": 0.5505291318891676,
|
||
|
|
"learning_rate": 3.587127357242193e-06,
|
||
|
|
"loss": 0.2378,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10735077410936356,
|
||
|
|
"step": 3105,
|
||
|
|
"valid_targets_mean": 4468.5,
|
||
|
|
"valid_targets_min": 604
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.792171481826654,
|
||
|
|
"grad_norm": 0.6333852984546299,
|
||
|
|
"learning_rate": 3.534238180347005e-06,
|
||
|
|
"loss": 0.2316,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11015690118074417,
|
||
|
|
"step": 3110,
|
||
|
|
"valid_targets_mean": 3928.5,
|
||
|
|
"valid_targets_min": 696
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.801491146318733,
|
||
|
|
"grad_norm": 0.5309758205740363,
|
||
|
|
"learning_rate": 3.4817039940141517e-06,
|
||
|
|
"loss": 0.2442,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11709916591644287,
|
||
|
|
"step": 3115,
|
||
|
|
"valid_targets_mean": 4573.8,
|
||
|
|
"valid_targets_min": 586
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.8108108108108105,
|
||
|
|
"grad_norm": 0.5698761113334817,
|
||
|
|
"learning_rate": 3.4295259308447593e-06,
|
||
|
|
"loss": 0.2651,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14713452756404877,
|
||
|
|
"step": 3120,
|
||
|
|
"valid_targets_mean": 4855.8,
|
||
|
|
"valid_targets_min": 751
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.820130475302889,
|
||
|
|
"grad_norm": 0.5347371257705741,
|
||
|
|
"learning_rate": 3.3777051157621754e-06,
|
||
|
|
"loss": 0.2478,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12283071875572205,
|
||
|
|
"step": 3125,
|
||
|
|
"valid_targets_mean": 4980.6,
|
||
|
|
"valid_targets_min": 708
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.829450139794967,
|
||
|
|
"grad_norm": 0.4264875314214193,
|
||
|
|
"learning_rate": 3.326242665987738e-06,
|
||
|
|
"loss": 0.2398,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10838539898395538,
|
||
|
|
"step": 3130,
|
||
|
|
"valid_targets_mean": 7263.2,
|
||
|
|
"valid_targets_min": 753
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.838769804287046,
|
||
|
|
"grad_norm": 0.4994473876566269,
|
||
|
|
"learning_rate": 3.2751396910166513e-06,
|
||
|
|
"loss": 0.2686,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12474403530359268,
|
||
|
|
"step": 3135,
|
||
|
|
"valid_targets_mean": 5395.5,
|
||
|
|
"valid_targets_min": 672
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.848089468779124,
|
||
|
|
"grad_norm": 0.491386597496051,
|
||
|
|
"learning_rate": 3.224397292594101e-06,
|
||
|
|
"loss": 0.2491,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10484868288040161,
|
||
|
|
"step": 3140,
|
||
|
|
"valid_targets_mean": 6783.1,
|
||
|
|
"valid_targets_min": 1140
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.857409133271203,
|
||
|
|
"grad_norm": 0.4611921753976029,
|
||
|
|
"learning_rate": 3.1740165646915024e-06,
|
||
|
|
"loss": 0.2509,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09882789850234985,
|
||
|
|
"step": 3145,
|
||
|
|
"valid_targets_mean": 4614.4,
|
||
|
|
"valid_targets_min": 1558
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.8667287977632805,
|
||
|
|
"grad_norm": 0.5445123779369885,
|
||
|
|
"learning_rate": 3.12399859348288e-06,
|
||
|
|
"loss": 0.2478,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12469325959682465,
|
||
|
|
"step": 3150,
|
||
|
|
"valid_targets_mean": 5553.6,
|
||
|
|
"valid_targets_min": 661
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.876048462255358,
|
||
|
|
"grad_norm": 0.5947859099465526,
|
||
|
|
"learning_rate": 3.07434445732149e-06,
|
||
|
|
"loss": 0.2615,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16190886497497559,
|
||
|
|
"step": 3155,
|
||
|
|
"valid_targets_mean": 6681.5,
|
||
|
|
"valid_targets_min": 874
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.885368126747437,
|
||
|
|
"grad_norm": 0.527573448167561,
|
||
|
|
"learning_rate": 3.025055226716531e-06,
|
||
|
|
"loss": 0.2364,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16413170099258423,
|
||
|
|
"step": 3160,
|
||
|
|
"valid_targets_mean": 6850.6,
|
||
|
|
"valid_targets_min": 1314
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.894687791239515,
|
||
|
|
"grad_norm": 0.6006551021052206,
|
||
|
|
"learning_rate": 2.9761319643101094e-06,
|
||
|
|
"loss": 0.2362,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15963928401470184,
|
||
|
|
"step": 3165,
|
||
|
|
"valid_targets_mean": 7288.4,
|
||
|
|
"valid_targets_min": 5620
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.904007455731594,
|
||
|
|
"grad_norm": 0.5605729910517652,
|
||
|
|
"learning_rate": 2.927575724854299e-06,
|
||
|
|
"loss": 0.2609,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20688825845718384,
|
||
|
|
"step": 3170,
|
||
|
|
"valid_targets_mean": 7230.1,
|
||
|
|
"valid_targets_min": 5020
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.913327120223672,
|
||
|
|
"grad_norm": 0.6721385485501952,
|
||
|
|
"learning_rate": 2.879387555188411e-06,
|
||
|
|
"loss": 0.2735,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18570585548877716,
|
||
|
|
"step": 3175,
|
||
|
|
"valid_targets_mean": 4853.8,
|
||
|
|
"valid_targets_min": 455
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.9226467847157505,
|
||
|
|
"grad_norm": 0.5626629539689247,
|
||
|
|
"learning_rate": 2.831568494216419e-06,
|
||
|
|
"loss": 0.2383,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10563358664512634,
|
||
|
|
"step": 3180,
|
||
|
|
"valid_targets_mean": 4321.0,
|
||
|
|
"valid_targets_min": 583
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.931966449207828,
|
||
|
|
"grad_norm": 0.5347964803999372,
|
||
|
|
"learning_rate": 2.784119572884574e-06,
|
||
|
|
"loss": 0.2576,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07239266484975815,
|
||
|
|
"step": 3185,
|
||
|
|
"valid_targets_mean": 3573.4,
|
||
|
|
"valid_targets_min": 669
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.941286113699907,
|
||
|
|
"grad_norm": 0.571705309286221,
|
||
|
|
"learning_rate": 2.7370418141591605e-06,
|
||
|
|
"loss": 0.2527,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08048106729984283,
|
||
|
|
"step": 3190,
|
||
|
|
"valid_targets_mean": 2731.4,
|
||
|
|
"valid_targets_min": 559
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.950605778191985,
|
||
|
|
"grad_norm": 0.5997973413202432,
|
||
|
|
"learning_rate": 2.6903362330044604e-06,
|
||
|
|
"loss": 0.24,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1188001036643982,
|
||
|
|
"step": 3195,
|
||
|
|
"valid_targets_mean": 4488.2,
|
||
|
|
"valid_targets_min": 642
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.959925442684064,
|
||
|
|
"grad_norm": 0.668198664621455,
|
||
|
|
"learning_rate": 2.6440038363608555e-06,
|
||
|
|
"loss": 0.2491,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11177880316972733,
|
||
|
|
"step": 3200,
|
||
|
|
"valid_targets_mean": 3250.4,
|
||
|
|
"valid_targets_min": 806
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.969245107176142,
|
||
|
|
"grad_norm": 0.6531782772123003,
|
||
|
|
"learning_rate": 2.5980456231231242e-06,
|
||
|
|
"loss": 0.2718,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1286979615688324,
|
||
|
|
"step": 3205,
|
||
|
|
"valid_targets_mean": 3771.5,
|
||
|
|
"valid_targets_min": 567
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.9785647716682195,
|
||
|
|
"grad_norm": 0.5686225821315919,
|
||
|
|
"learning_rate": 2.552462584118911e-06,
|
||
|
|
"loss": 0.2209,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11458878964185715,
|
||
|
|
"step": 3210,
|
||
|
|
"valid_targets_mean": 4983.5,
|
||
|
|
"valid_targets_min": 598
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.987884436160298,
|
||
|
|
"grad_norm": 0.6613914944230602,
|
||
|
|
"learning_rate": 2.5072557020873568e-06,
|
||
|
|
"loss": 0.2417,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15523985028266907,
|
||
|
|
"step": 3215,
|
||
|
|
"valid_targets_mean": 4160.5,
|
||
|
|
"valid_targets_min": 583
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.997204100652376,
|
||
|
|
"grad_norm": 0.6549237346089047,
|
||
|
|
"learning_rate": 2.462425951657923e-06,
|
||
|
|
"loss": 0.2627,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1376725435256958,
|
||
|
|
"step": 3220,
|
||
|
|
"valid_targets_mean": 3974.5,
|
||
|
|
"valid_targets_min": 894
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.005591798695247,
|
||
|
|
"grad_norm": 0.5154122770646274,
|
||
|
|
"learning_rate": 2.4179742993293552e-06,
|
||
|
|
"loss": 0.2703,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11231924593448639,
|
||
|
|
"step": 3225,
|
||
|
|
"valid_targets_mean": 4495.6,
|
||
|
|
"valid_targets_min": 745
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.0149114631873255,
|
||
|
|
"grad_norm": 0.6309103750160548,
|
||
|
|
"learning_rate": 2.3739017034488756e-06,
|
||
|
|
"loss": 0.2399,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15297481417655945,
|
||
|
|
"step": 3230,
|
||
|
|
"valid_targets_mean": 5364.1,
|
||
|
|
"valid_targets_min": 768
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.024231127679403,
|
||
|
|
"grad_norm": 0.5320408122591563,
|
||
|
|
"learning_rate": 2.3302091141915107e-06,
|
||
|
|
"loss": 0.2402,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11056330054998398,
|
||
|
|
"step": 3235,
|
||
|
|
"valid_targets_mean": 4225.5,
|
||
|
|
"valid_targets_min": 715
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.033550792171482,
|
||
|
|
"grad_norm": 0.5579450997803592,
|
||
|
|
"learning_rate": 2.286897473539602e-06,
|
||
|
|
"loss": 0.2585,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07014621794223785,
|
||
|
|
"step": 3240,
|
||
|
|
"valid_targets_mean": 2739.4,
|
||
|
|
"valid_targets_min": 571
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.04287045666356,
|
||
|
|
"grad_norm": 0.4784261635431962,
|
||
|
|
"learning_rate": 2.2439677152624874e-06,
|
||
|
|
"loss": 0.2305,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08478303253650665,
|
||
|
|
"step": 3245,
|
||
|
|
"valid_targets_mean": 3844.6,
|
||
|
|
"valid_targets_min": 643
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.052190121155639,
|
||
|
|
"grad_norm": 0.772731234411963,
|
||
|
|
"learning_rate": 2.201420764896396e-06,
|
||
|
|
"loss": 0.2409,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13267914950847626,
|
||
|
|
"step": 3250,
|
||
|
|
"valid_targets_mean": 3295.5,
|
||
|
|
"valid_targets_min": 748
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.061509785647717,
|
||
|
|
"grad_norm": 0.6398690385894376,
|
||
|
|
"learning_rate": 2.1592575397244775e-06,
|
||
|
|
"loss": 0.2457,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13361290097236633,
|
||
|
|
"step": 3255,
|
||
|
|
"valid_targets_mean": 3393.0,
|
||
|
|
"valid_targets_min": 651
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.070829450139795,
|
||
|
|
"grad_norm": 0.658689821936898,
|
||
|
|
"learning_rate": 2.1174789487570233e-06,
|
||
|
|
"loss": 0.254,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16041888296604156,
|
||
|
|
"step": 3260,
|
||
|
|
"valid_targets_mean": 4597.0,
|
||
|
|
"valid_targets_min": 726
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.080149114631873,
|
||
|
|
"grad_norm": 0.5078232732321583,
|
||
|
|
"learning_rate": 2.0760858927118833e-06,
|
||
|
|
"loss": 0.235,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10284225642681122,
|
||
|
|
"step": 3265,
|
||
|
|
"valid_targets_mean": 5922.9,
|
||
|
|
"valid_targets_min": 542
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.089468779123951,
|
||
|
|
"grad_norm": 0.5681849314112172,
|
||
|
|
"learning_rate": 2.035079263995028e-06,
|
||
|
|
"loss": 0.2526,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14872777462005615,
|
||
|
|
"step": 3270,
|
||
|
|
"valid_targets_mean": 5084.4,
|
||
|
|
"valid_targets_min": 787
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.09878844361603,
|
||
|
|
"grad_norm": 0.519901516032561,
|
||
|
|
"learning_rate": 1.9944599466813286e-06,
|
||
|
|
"loss": 0.2359,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09071168303489685,
|
||
|
|
"step": 3275,
|
||
|
|
"valid_targets_mean": 4616.6,
|
||
|
|
"valid_targets_min": 837
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.108108108108108,
|
||
|
|
"grad_norm": 0.5859894174952418,
|
||
|
|
"learning_rate": 1.954228816495485e-06,
|
||
|
|
"loss": 0.2202,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1169670820236206,
|
||
|
|
"step": 3280,
|
||
|
|
"valid_targets_mean": 4756.5,
|
||
|
|
"valid_targets_min": 618
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.117427772600187,
|
||
|
|
"grad_norm": 0.5375477780695376,
|
||
|
|
"learning_rate": 1.914386740793133e-06,
|
||
|
|
"loss": 0.2331,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11546333134174347,
|
||
|
|
"step": 3285,
|
||
|
|
"valid_targets_mean": 5241.0,
|
||
|
|
"valid_targets_min": 601
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.1267474370922645,
|
||
|
|
"grad_norm": 0.5650576310115677,
|
||
|
|
"learning_rate": 1.874934578542187e-06,
|
||
|
|
"loss": 0.2621,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12042230367660522,
|
||
|
|
"step": 3290,
|
||
|
|
"valid_targets_mean": 4875.5,
|
||
|
|
"valid_targets_min": 516
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.136067101584343,
|
||
|
|
"grad_norm": 0.5234060750277322,
|
||
|
|
"learning_rate": 1.8358731803042752e-06,
|
||
|
|
"loss": 0.2395,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1171858087182045,
|
||
|
|
"step": 3295,
|
||
|
|
"valid_targets_mean": 5732.6,
|
||
|
|
"valid_targets_min": 675
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.145386766076421,
|
||
|
|
"grad_norm": 0.5471314297616108,
|
||
|
|
"learning_rate": 1.7972033882164264e-06,
|
||
|
|
"loss": 0.2475,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17577917873859406,
|
||
|
|
"step": 3300,
|
||
|
|
"valid_targets_mean": 6558.8,
|
||
|
|
"valid_targets_min": 730
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.154706430568499,
|
||
|
|
"grad_norm": 0.5563934181659101,
|
||
|
|
"learning_rate": 1.7589260359729121e-06,
|
||
|
|
"loss": 0.2298,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11752583831548691,
|
||
|
|
"step": 3305,
|
||
|
|
"valid_targets_mean": 5948.2,
|
||
|
|
"valid_targets_min": 586
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.164026095060578,
|
||
|
|
"grad_norm": 0.6037361447560499,
|
||
|
|
"learning_rate": 1.721041948807256e-06,
|
||
|
|
"loss": 0.2438,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13703708350658417,
|
||
|
|
"step": 3310,
|
||
|
|
"valid_targets_mean": 4745.6,
|
||
|
|
"valid_targets_min": 689
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.173345759552656,
|
||
|
|
"grad_norm": 0.49780638947114875,
|
||
|
|
"learning_rate": 1.6835519434744641e-06,
|
||
|
|
"loss": 0.251,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09111525118350983,
|
||
|
|
"step": 3315,
|
||
|
|
"valid_targets_mean": 3826.9,
|
||
|
|
"valid_targets_min": 652
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.1826654240447345,
|
||
|
|
"grad_norm": 0.5197355893885705,
|
||
|
|
"learning_rate": 1.6464568282334204e-06,
|
||
|
|
"loss": 0.2258,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12354490906000137,
|
||
|
|
"step": 3320,
|
||
|
|
"valid_targets_mean": 6572.2,
|
||
|
|
"valid_targets_min": 5608
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.191985088536812,
|
||
|
|
"grad_norm": 0.5127316606102179,
|
||
|
|
"learning_rate": 1.6097574028294327e-06,
|
||
|
|
"loss": 0.2407,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12837597727775574,
|
||
|
|
"step": 3325,
|
||
|
|
"valid_targets_mean": 5881.0,
|
||
|
|
"valid_targets_min": 637
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.201304753028891,
|
||
|
|
"grad_norm": 0.6381681540587546,
|
||
|
|
"learning_rate": 1.5734544584770039e-06,
|
||
|
|
"loss": 0.236,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1222711056470871,
|
||
|
|
"step": 3330,
|
||
|
|
"valid_targets_mean": 3703.5,
|
||
|
|
"valid_targets_min": 689
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.210624417520969,
|
||
|
|
"grad_norm": 0.6922113370015349,
|
||
|
|
"learning_rate": 1.5375487778427877e-06,
|
||
|
|
"loss": 0.2383,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21385058760643005,
|
||
|
|
"step": 3335,
|
||
|
|
"valid_targets_mean": 5368.2,
|
||
|
|
"valid_targets_min": 696
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.219944082013048,
|
||
|
|
"grad_norm": 0.6049907051447185,
|
||
|
|
"learning_rate": 1.502041135028698e-06,
|
||
|
|
"loss": 0.2347,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09852921217679977,
|
||
|
|
"step": 3340,
|
||
|
|
"valid_targets_mean": 4212.0,
|
||
|
|
"valid_targets_min": 663
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.229263746505126,
|
||
|
|
"grad_norm": 0.5864755878527658,
|
||
|
|
"learning_rate": 1.4669322955552278e-06,
|
||
|
|
"loss": 0.2365,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1378830224275589,
|
||
|
|
"step": 3345,
|
||
|
|
"valid_targets_mean": 4523.9,
|
||
|
|
"valid_targets_min": 754
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.2385834109972045,
|
||
|
|
"grad_norm": 0.5347053622374001,
|
||
|
|
"learning_rate": 1.4322230163449403e-06,
|
||
|
|
"loss": 0.2518,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11226432025432587,
|
||
|
|
"step": 3350,
|
||
|
|
"valid_targets_mean": 5331.0,
|
||
|
|
"valid_targets_min": 674
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.247903075489282,
|
||
|
|
"grad_norm": 0.5331067336280443,
|
||
|
|
"learning_rate": 1.3979140457061568e-06,
|
||
|
|
"loss": 0.2419,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16322267055511475,
|
||
|
|
"step": 3355,
|
||
|
|
"valid_targets_mean": 6979.4,
|
||
|
|
"valid_targets_min": 724
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.25722273998136,
|
||
|
|
"grad_norm": 0.587914114338471,
|
||
|
|
"learning_rate": 1.3640061233168166e-06,
|
||
|
|
"loss": 0.2476,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1671825647354126,
|
||
|
|
"step": 3360,
|
||
|
|
"valid_targets_mean": 4565.1,
|
||
|
|
"valid_targets_min": 694
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.266542404473439,
|
||
|
|
"grad_norm": 0.5453711124872581,
|
||
|
|
"learning_rate": 1.3304999802085371e-06,
|
||
|
|
"loss": 0.2476,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1264810860157013,
|
||
|
|
"step": 3365,
|
||
|
|
"valid_targets_mean": 6392.9,
|
||
|
|
"valid_targets_min": 576
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.275862068965517,
|
||
|
|
"grad_norm": 0.5681387987629308,
|
||
|
|
"learning_rate": 1.2973963387508336e-06,
|
||
|
|
"loss": 0.2356,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09002932161092758,
|
||
|
|
"step": 3370,
|
||
|
|
"valid_targets_mean": 3181.8,
|
||
|
|
"valid_targets_min": 647
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.285181733457596,
|
||
|
|
"grad_norm": 0.5249530824042369,
|
||
|
|
"learning_rate": 1.264695912635583e-06,
|
||
|
|
"loss": 0.2311,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11573824286460876,
|
||
|
|
"step": 3375,
|
||
|
|
"valid_targets_mean": 5553.4,
|
||
|
|
"valid_targets_min": 597
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.294501397949674,
|
||
|
|
"grad_norm": 0.6020141118189879,
|
||
|
|
"learning_rate": 1.2323994068616064e-06,
|
||
|
|
"loss": 0.2379,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14819496870040894,
|
||
|
|
"step": 3380,
|
||
|
|
"valid_targets_mean": 3744.8,
|
||
|
|
"valid_targets_min": 596
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.303821062441752,
|
||
|
|
"grad_norm": 0.5212871817953852,
|
||
|
|
"learning_rate": 1.2005075177194736e-06,
|
||
|
|
"loss": 0.2349,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10104718804359436,
|
||
|
|
"step": 3385,
|
||
|
|
"valid_targets_mean": 4486.9,
|
||
|
|
"valid_targets_min": 946
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.31314072693383,
|
||
|
|
"grad_norm": 0.5406321930636565,
|
||
|
|
"learning_rate": 1.1690209327765033e-06,
|
||
|
|
"loss": 0.2322,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14584863185882568,
|
||
|
|
"step": 3390,
|
||
|
|
"valid_targets_mean": 6095.2,
|
||
|
|
"valid_targets_min": 652
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.322460391425909,
|
||
|
|
"grad_norm": 0.5836538931015579,
|
||
|
|
"learning_rate": 1.1379403308619265e-06,
|
||
|
|
"loss": 0.2336,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09438341856002808,
|
||
|
|
"step": 3395,
|
||
|
|
"valid_targets_mean": 3748.9,
|
||
|
|
"valid_targets_min": 548
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.331780055917987,
|
||
|
|
"grad_norm": 0.4133666815313372,
|
||
|
|
"learning_rate": 1.1072663820522655e-06,
|
||
|
|
"loss": 0.2249,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08744319528341293,
|
||
|
|
"step": 3400,
|
||
|
|
"valid_targets_mean": 6795.9,
|
||
|
|
"valid_targets_min": 493
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.341099720410066,
|
||
|
|
"grad_norm": 0.617926863266978,
|
||
|
|
"learning_rate": 1.0769997476568684e-06,
|
||
|
|
"loss": 0.2452,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14238421618938446,
|
||
|
|
"step": 3405,
|
||
|
|
"valid_targets_mean": 5638.0,
|
||
|
|
"valid_targets_min": 590
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.3504193849021435,
|
||
|
|
"grad_norm": 0.5844355183005746,
|
||
|
|
"learning_rate": 1.0471410802036842e-06,
|
||
|
|
"loss": 0.2305,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16003677248954773,
|
||
|
|
"step": 3410,
|
||
|
|
"valid_targets_mean": 5273.9,
|
||
|
|
"valid_targets_min": 610
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.359739049394221,
|
||
|
|
"grad_norm": 0.4569602874311673,
|
||
|
|
"learning_rate": 1.017691023425147e-06,
|
||
|
|
"loss": 0.2434,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1760561466217041,
|
||
|
|
"step": 3415,
|
||
|
|
"valid_targets_mean": 8134.2,
|
||
|
|
"valid_targets_min": 5571
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.3690587138863,
|
||
|
|
"grad_norm": 0.5855887108117603,
|
||
|
|
"learning_rate": 9.886502122443442e-07,
|
||
|
|
"loss": 0.2166,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07828006148338318,
|
||
|
|
"step": 3420,
|
||
|
|
"valid_targets_mean": 3737.1,
|
||
|
|
"valid_targets_min": 603
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.378378378378378,
|
||
|
|
"grad_norm": 0.611921225241911,
|
||
|
|
"learning_rate": 9.60019272761299e-07,
|
||
|
|
"loss": 0.2599,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09570074081420898,
|
||
|
|
"step": 3425,
|
||
|
|
"valid_targets_mean": 5858.8,
|
||
|
|
"valid_targets_min": 975
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.387698042870457,
|
||
|
|
"grad_norm": 0.6066562183345475,
|
||
|
|
"learning_rate": 9.317988222394825e-07,
|
||
|
|
"loss": 0.2363,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15763790905475616,
|
||
|
|
"step": 3430,
|
||
|
|
"valid_targets_mean": 5263.8,
|
||
|
|
"valid_targets_min": 613
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.397017707362535,
|
||
|
|
"grad_norm": 0.6202638315709991,
|
||
|
|
"learning_rate": 9.039894690925055e-07,
|
||
|
|
"loss": 0.2394,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1113867536187172,
|
||
|
|
"step": 3435,
|
||
|
|
"valid_targets_mean": 4688.0,
|
||
|
|
"valid_targets_min": 617
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.4063373718546135,
|
||
|
|
"grad_norm": 0.5641390903989121,
|
||
|
|
"learning_rate": 8.765918128710016e-07,
|
||
|
|
"loss": 0.2296,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11658851057291031,
|
||
|
|
"step": 3440,
|
||
|
|
"valid_targets_mean": 5362.8,
|
||
|
|
"valid_targets_min": 921
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.415657036346691,
|
||
|
|
"grad_norm": 0.6238322182456596,
|
||
|
|
"learning_rate": 8.496064442496999e-07,
|
||
|
|
"loss": 0.2599,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16795772314071655,
|
||
|
|
"step": 3445,
|
||
|
|
"valid_targets_mean": 5924.2,
|
||
|
|
"valid_targets_min": 738
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.42497670083877,
|
||
|
|
"grad_norm": 0.6897916363559863,
|
||
|
|
"learning_rate": 8.230339450146863e-07,
|
||
|
|
"loss": 0.2712,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19291897118091583,
|
||
|
|
"step": 3450,
|
||
|
|
"valid_targets_mean": 5790.5,
|
||
|
|
"valid_targets_min": 691
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.434296365330848,
|
||
|
|
"grad_norm": 0.5418955721131009,
|
||
|
|
"learning_rate": 7.968748880508759e-07,
|
||
|
|
"loss": 0.2489,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13060906529426575,
|
||
|
|
"step": 3455,
|
||
|
|
"valid_targets_mean": 5471.2,
|
||
|
|
"valid_targets_min": 625
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.443616029822927,
|
||
|
|
"grad_norm": 0.6260625606035904,
|
||
|
|
"learning_rate": 7.711298373296316e-07,
|
||
|
|
"loss": 0.2388,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1072792261838913,
|
||
|
|
"step": 3460,
|
||
|
|
"valid_targets_mean": 3519.0,
|
||
|
|
"valid_targets_min": 572
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.452935694315005,
|
||
|
|
"grad_norm": 0.7695644049011828,
|
||
|
|
"learning_rate": 7.457993478966519e-07,
|
||
|
|
"loss": 0.2676,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11188751459121704,
|
||
|
|
"step": 3465,
|
||
|
|
"valid_targets_mean": 4086.1,
|
||
|
|
"valid_targets_min": 735
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.462255358807083,
|
||
|
|
"grad_norm": 0.5500618381781784,
|
||
|
|
"learning_rate": 7.208839658599531e-07,
|
||
|
|
"loss": 0.2333,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1004505306482315,
|
||
|
|
"step": 3470,
|
||
|
|
"valid_targets_mean": 6739.9,
|
||
|
|
"valid_targets_min": 626
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.471575023299161,
|
||
|
|
"grad_norm": 0.5526996302542733,
|
||
|
|
"learning_rate": 6.963842283781375e-07,
|
||
|
|
"loss": 0.2242,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09276391565799713,
|
||
|
|
"step": 3475,
|
||
|
|
"valid_targets_mean": 3827.2,
|
||
|
|
"valid_targets_min": 843
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.480894687791239,
|
||
|
|
"grad_norm": 0.5649113465048164,
|
||
|
|
"learning_rate": 6.723006636487795e-07,
|
||
|
|
"loss": 0.2475,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08035659790039062,
|
||
|
|
"step": 3480,
|
||
|
|
"valid_targets_mean": 3359.4,
|
||
|
|
"valid_targets_min": 620
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.490214352283318,
|
||
|
|
"grad_norm": 0.5791492190790262,
|
||
|
|
"learning_rate": 6.486337908970663e-07,
|
||
|
|
"loss": 0.2541,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11885468661785126,
|
||
|
|
"step": 3485,
|
||
|
|
"valid_targets_mean": 4618.5,
|
||
|
|
"valid_targets_min": 905
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.499534016775396,
|
||
|
|
"grad_norm": 0.6205061143163002,
|
||
|
|
"learning_rate": 6.253841203645827e-07,
|
||
|
|
"loss": 0.2325,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15617059171199799,
|
||
|
|
"step": 3490,
|
||
|
|
"valid_targets_mean": 4672.6,
|
||
|
|
"valid_targets_min": 597
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.508853681267475,
|
||
|
|
"grad_norm": 0.6028619212942035,
|
||
|
|
"learning_rate": 6.025521532983324e-07,
|
||
|
|
"loss": 0.2432,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10843147337436676,
|
||
|
|
"step": 3495,
|
||
|
|
"valid_targets_mean": 3953.9,
|
||
|
|
"valid_targets_min": 578
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.5181733457595525,
|
||
|
|
"grad_norm": 0.5574127233241083,
|
||
|
|
"learning_rate": 5.801383819398965e-07,
|
||
|
|
"loss": 0.2394,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1006593257188797,
|
||
|
|
"step": 3500,
|
||
|
|
"valid_targets_mean": 3581.0,
|
||
|
|
"valid_targets_min": 567
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.527493010251631,
|
||
|
|
"grad_norm": 0.8559940748188053,
|
||
|
|
"learning_rate": 5.581432895148608e-07,
|
||
|
|
"loss": 0.2626,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1481635868549347,
|
||
|
|
"step": 3505,
|
||
|
|
"valid_targets_mean": 3525.2,
|
||
|
|
"valid_targets_min": 652
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.536812674743709,
|
||
|
|
"grad_norm": 0.6115204666270448,
|
||
|
|
"learning_rate": 5.365673502223723e-07,
|
||
|
|
"loss": 0.2258,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11971457302570343,
|
||
|
|
"step": 3510,
|
||
|
|
"valid_targets_mean": 4339.6,
|
||
|
|
"valid_targets_min": 662
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.546132339235788,
|
||
|
|
"grad_norm": 0.5061707697153148,
|
||
|
|
"learning_rate": 5.154110292249237e-07,
|
||
|
|
"loss": 0.2378,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13552606105804443,
|
||
|
|
"step": 3515,
|
||
|
|
"valid_targets_mean": 5851.8,
|
||
|
|
"valid_targets_min": 749
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.555452003727866,
|
||
|
|
"grad_norm": 0.5733523704005805,
|
||
|
|
"learning_rate": 4.946747826383269e-07,
|
||
|
|
"loss": 0.2279,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1370592713356018,
|
||
|
|
"step": 3520,
|
||
|
|
"valid_targets_mean": 5941.6,
|
||
|
|
"valid_targets_min": 493
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.564771668219944,
|
||
|
|
"grad_norm": 0.6001847328804019,
|
||
|
|
"learning_rate": 4.743590575218737e-07,
|
||
|
|
"loss": 0.2485,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10027460753917694,
|
||
|
|
"step": 3525,
|
||
|
|
"valid_targets_mean": 3764.9,
|
||
|
|
"valid_targets_min": 637
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.5740913327120225,
|
||
|
|
"grad_norm": 0.6201422965225871,
|
||
|
|
"learning_rate": 4.544642918686992e-07,
|
||
|
|
"loss": 0.2323,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09919697046279907,
|
||
|
|
"step": 3530,
|
||
|
|
"valid_targets_mean": 3370.1,
|
||
|
|
"valid_targets_min": 655
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.5834109972041,
|
||
|
|
"grad_norm": 0.7081499570238575,
|
||
|
|
"learning_rate": 4.3499091459634713e-07,
|
||
|
|
"loss": 0.2308,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1171005517244339,
|
||
|
|
"step": 3535,
|
||
|
|
"valid_targets_mean": 3663.5,
|
||
|
|
"valid_targets_min": 524
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.592730661696179,
|
||
|
|
"grad_norm": 0.539591803636606,
|
||
|
|
"learning_rate": 4.159393455375105e-07,
|
||
|
|
"loss": 0.2229,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10873937606811523,
|
||
|
|
"step": 3540,
|
||
|
|
"valid_targets_mean": 5885.2,
|
||
|
|
"valid_targets_min": 850
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.602050326188257,
|
||
|
|
"grad_norm": 0.4856654919202911,
|
||
|
|
"learning_rate": 3.9730999543098337e-07,
|
||
|
|
"loss": 0.2359,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11818019300699234,
|
||
|
|
"step": 3545,
|
||
|
|
"valid_targets_mean": 6850.5,
|
||
|
|
"valid_targets_min": 543
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.611369990680336,
|
||
|
|
"grad_norm": 0.7661046857521794,
|
||
|
|
"learning_rate": 3.791032659128169e-07,
|
||
|
|
"loss": 0.223,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10599961131811142,
|
||
|
|
"step": 3550,
|
||
|
|
"valid_targets_mean": 1998.4,
|
||
|
|
"valid_targets_min": 622
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.620689655172414,
|
||
|
|
"grad_norm": 0.47029970138362953,
|
||
|
|
"learning_rate": 3.613195495076438e-07,
|
||
|
|
"loss": 0.2418,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08204981684684753,
|
||
|
|
"step": 3555,
|
||
|
|
"valid_targets_mean": 6522.1,
|
||
|
|
"valid_targets_min": 4192
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.630009319664492,
|
||
|
|
"grad_norm": 0.5866433711979756,
|
||
|
|
"learning_rate": 3.439592296202299e-07,
|
||
|
|
"loss": 0.2493,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13587576150894165,
|
||
|
|
"step": 3560,
|
||
|
|
"valid_targets_mean": 6841.0,
|
||
|
|
"valid_targets_min": 636
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.63932898415657,
|
||
|
|
"grad_norm": 0.6292757848173881,
|
||
|
|
"learning_rate": 3.2702268052718924e-07,
|
||
|
|
"loss": 0.2496,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12244050949811935,
|
||
|
|
"step": 3565,
|
||
|
|
"valid_targets_mean": 3719.9,
|
||
|
|
"valid_targets_min": 725
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.648648648648649,
|
||
|
|
"grad_norm": 0.6126574702317991,
|
||
|
|
"learning_rate": 3.1051026736893977e-07,
|
||
|
|
"loss": 0.235,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12401425093412399,
|
||
|
|
"step": 3570,
|
||
|
|
"valid_targets_mean": 5077.0,
|
||
|
|
"valid_targets_min": 690
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.657968313140727,
|
||
|
|
"grad_norm": 0.7462032322237635,
|
||
|
|
"learning_rate": 2.944223461418161e-07,
|
||
|
|
"loss": 0.2436,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11168582737445831,
|
||
|
|
"step": 3575,
|
||
|
|
"valid_targets_mean": 4816.6,
|
||
|
|
"valid_targets_min": 582
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.667287977632805,
|
||
|
|
"grad_norm": 0.5497959797513633,
|
||
|
|
"learning_rate": 2.7875926369039577e-07,
|
||
|
|
"loss": 0.2334,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1527920961380005,
|
||
|
|
"step": 3580,
|
||
|
|
"valid_targets_mean": 6026.9,
|
||
|
|
"valid_targets_min": 783
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.676607642124884,
|
||
|
|
"grad_norm": 0.5459903979732039,
|
||
|
|
"learning_rate": 2.6352135770002283e-07,
|
||
|
|
"loss": 0.2354,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10157076269388199,
|
||
|
|
"step": 3585,
|
||
|
|
"valid_targets_mean": 5246.8,
|
||
|
|
"valid_targets_min": 613
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.6859273066169616,
|
||
|
|
"grad_norm": 0.5358188290860093,
|
||
|
|
"learning_rate": 2.487089566895251e-07,
|
||
|
|
"loss": 0.2225,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10558871924877167,
|
||
|
|
"step": 3590,
|
||
|
|
"valid_targets_mean": 6012.4,
|
||
|
|
"valid_targets_min": 1844
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.69524697110904,
|
||
|
|
"grad_norm": 0.6946920682077415,
|
||
|
|
"learning_rate": 2.3432238000414165e-07,
|
||
|
|
"loss": 0.2449,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12842845916748047,
|
||
|
|
"step": 3595,
|
||
|
|
"valid_targets_mean": 2154.2,
|
||
|
|
"valid_targets_min": 630
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.704566635601118,
|
||
|
|
"grad_norm": 0.6860210918684664,
|
||
|
|
"learning_rate": 2.203619378086197e-07,
|
||
|
|
"loss": 0.2523,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10359951853752136,
|
||
|
|
"step": 3600,
|
||
|
|
"valid_targets_mean": 3469.4,
|
||
|
|
"valid_targets_min": 717
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.713886300093197,
|
||
|
|
"grad_norm": 0.5118220691970857,
|
||
|
|
"learning_rate": 2.0682793108054878e-07,
|
||
|
|
"loss": 0.2369,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10798832774162292,
|
||
|
|
"step": 3605,
|
||
|
|
"valid_targets_mean": 5704.2,
|
||
|
|
"valid_targets_min": 549
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.723205964585275,
|
||
|
|
"grad_norm": 0.5328901006199235,
|
||
|
|
"learning_rate": 1.937206516038548e-07,
|
||
|
|
"loss": 0.2537,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1261061578989029,
|
||
|
|
"step": 3610,
|
||
|
|
"valid_targets_mean": 4658.0,
|
||
|
|
"valid_targets_min": 590
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.732525629077353,
|
||
|
|
"grad_norm": 0.6627114114999452,
|
||
|
|
"learning_rate": 1.8104038196251617e-07,
|
||
|
|
"loss": 0.2526,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14013832807540894,
|
||
|
|
"step": 3615,
|
||
|
|
"valid_targets_mean": 3838.2,
|
||
|
|
"valid_targets_min": 707
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.7418452935694315,
|
||
|
|
"grad_norm": 0.5943523943033668,
|
||
|
|
"learning_rate": 1.6878739553447765e-07,
|
||
|
|
"loss": 0.239,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11525734513998032,
|
||
|
|
"step": 3620,
|
||
|
|
"valid_targets_mean": 4705.2,
|
||
|
|
"valid_targets_min": 432
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.751164958061509,
|
||
|
|
"grad_norm": 0.5963467604063815,
|
||
|
|
"learning_rate": 1.5696195648574385e-07,
|
||
|
|
"loss": 0.2369,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07416298985481262,
|
||
|
|
"step": 3625,
|
||
|
|
"valid_targets_mean": 3635.5,
|
||
|
|
"valid_targets_min": 691
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.760484622553588,
|
||
|
|
"grad_norm": 0.6328033032724268,
|
||
|
|
"learning_rate": 1.4556431976468833e-07,
|
||
|
|
"loss": 0.2637,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09292063117027283,
|
||
|
|
"step": 3630,
|
||
|
|
"valid_targets_mean": 3806.8,
|
||
|
|
"valid_targets_min": 399
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.769804287045666,
|
||
|
|
"grad_norm": 0.5775222189514438,
|
||
|
|
"learning_rate": 1.3459473109656673e-07,
|
||
|
|
"loss": 0.2638,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10743218660354614,
|
||
|
|
"step": 3635,
|
||
|
|
"valid_targets_mean": 4534.9,
|
||
|
|
"valid_targets_min": 728
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.779123951537745,
|
||
|
|
"grad_norm": 0.9749274451927709,
|
||
|
|
"learning_rate": 1.2405342697820787e-07,
|
||
|
|
"loss": 0.2463,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14854204654693604,
|
||
|
|
"step": 3640,
|
||
|
|
"valid_targets_mean": 2509.0,
|
||
|
|
"valid_targets_min": 550
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.788443616029823,
|
||
|
|
"grad_norm": 0.55475132164247,
|
||
|
|
"learning_rate": 1.1394063467291771e-07,
|
||
|
|
"loss": 0.2438,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09817175567150116,
|
||
|
|
"step": 3645,
|
||
|
|
"valid_targets_mean": 4993.0,
|
||
|
|
"valid_targets_min": 546
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.7977632805219015,
|
||
|
|
"grad_norm": 0.6644247463381604,
|
||
|
|
"learning_rate": 1.0425657220557883e-07,
|
||
|
|
"loss": 0.2478,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14776167273521423,
|
||
|
|
"step": 3650,
|
||
|
|
"valid_targets_mean": 4309.2,
|
||
|
|
"valid_targets_min": 732
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.807082945013979,
|
||
|
|
"grad_norm": 0.5543421389868178,
|
||
|
|
"learning_rate": 9.500144835795865e-08,
|
||
|
|
"loss": 0.2482,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17127090692520142,
|
||
|
|
"step": 3655,
|
||
|
|
"valid_targets_mean": 5787.5,
|
||
|
|
"valid_targets_min": 481
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.816402609506058,
|
||
|
|
"grad_norm": 0.5281011173923301,
|
||
|
|
"learning_rate": 8.617546266419307e-08,
|
||
|
|
"loss": 0.251,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11352965235710144,
|
||
|
|
"step": 3660,
|
||
|
|
"valid_targets_mean": 5632.4,
|
||
|
|
"valid_targets_min": 842
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.825722273998136,
|
||
|
|
"grad_norm": 0.6006024721013183,
|
||
|
|
"learning_rate": 7.777880540649652e-08,
|
||
|
|
"loss": 0.2558,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1504848599433899,
|
||
|
|
"step": 3665,
|
||
|
|
"valid_targets_mean": 6240.5,
|
||
|
|
"valid_targets_min": 1057
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.835041938490214,
|
||
|
|
"grad_norm": 0.6038043313902524,
|
||
|
|
"learning_rate": 6.981165761105857e-08,
|
||
|
|
"loss": 0.2299,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11640399694442749,
|
||
|
|
"step": 3670,
|
||
|
|
"valid_targets_mean": 6194.2,
|
||
|
|
"valid_targets_min": 693
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.844361602982293,
|
||
|
|
"grad_norm": 0.5552511455434008,
|
||
|
|
"learning_rate": 6.227419104413601e-08,
|
||
|
|
"loss": 0.2315,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1136164516210556,
|
||
|
|
"step": 3675,
|
||
|
|
"valid_targets_mean": 5095.6,
|
||
|
|
"valid_targets_min": 552
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.853681267474371,
|
||
|
|
"grad_norm": 0.5749126917838003,
|
||
|
|
"learning_rate": 5.516656820835131e-08,
|
||
|
|
"loss": 0.2422,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13571766018867493,
|
||
|
|
"step": 3680,
|
||
|
|
"valid_targets_mean": 4796.4,
|
||
|
|
"valid_targets_min": 640
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.863000931966449,
|
||
|
|
"grad_norm": 0.5660286906801433,
|
||
|
|
"learning_rate": 4.848894233919321e-08,
|
||
|
|
"loss": 0.2638,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12739980220794678,
|
||
|
|
"step": 3685,
|
||
|
|
"valid_targets_mean": 5619.5,
|
||
|
|
"valid_targets_min": 758
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.872320596458527,
|
||
|
|
"grad_norm": 0.5479043859474394,
|
||
|
|
"learning_rate": 4.2241457401703824e-08,
|
||
|
|
"loss": 0.2417,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11300769448280334,
|
||
|
|
"step": 3690,
|
||
|
|
"valid_targets_mean": 4733.5,
|
||
|
|
"valid_targets_min": 626
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.881640260950606,
|
||
|
|
"grad_norm": 0.6564960108171396,
|
||
|
|
"learning_rate": 3.642424808738998e-08,
|
||
|
|
"loss": 0.2563,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11989576369524002,
|
||
|
|
"step": 3695,
|
||
|
|
"valid_targets_mean": 4183.8,
|
||
|
|
"valid_targets_min": 579
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.890959925442684,
|
||
|
|
"grad_norm": 0.5414972818171555,
|
||
|
|
"learning_rate": 3.1037439811303363e-08,
|
||
|
|
"loss": 0.2632,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14904309809207916,
|
||
|
|
"step": 3700,
|
||
|
|
"valid_targets_mean": 5762.8,
|
||
|
|
"valid_targets_min": 750
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.900279589934763,
|
||
|
|
"grad_norm": 0.6346843036138841,
|
||
|
|
"learning_rate": 2.608114870934486e-08,
|
||
|
|
"loss": 0.2429,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13483691215515137,
|
||
|
|
"step": 3705,
|
||
|
|
"valid_targets_mean": 6093.4,
|
||
|
|
"valid_targets_min": 580
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.9095992544268405,
|
||
|
|
"grad_norm": 0.5157660679097174,
|
||
|
|
"learning_rate": 2.1555481635762156e-08,
|
||
|
|
"loss": 0.2372,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12471800297498703,
|
||
|
|
"step": 3710,
|
||
|
|
"valid_targets_mean": 7073.1,
|
||
|
|
"valid_targets_min": 795
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.918918918918919,
|
||
|
|
"grad_norm": 0.5436809190239756,
|
||
|
|
"learning_rate": 1.7460536160840424e-08,
|
||
|
|
"loss": 0.2438,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10760162770748138,
|
||
|
|
"step": 3715,
|
||
|
|
"valid_targets_mean": 5025.2,
|
||
|
|
"valid_targets_min": 595
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.928238583410997,
|
||
|
|
"grad_norm": 0.591019819020007,
|
||
|
|
"learning_rate": 1.3796400568804046e-08,
|
||
|
|
"loss": 0.2618,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11672061681747437,
|
||
|
|
"step": 3720,
|
||
|
|
"valid_targets_mean": 5473.6,
|
||
|
|
"valid_targets_min": 814
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.937558247903075,
|
||
|
|
"grad_norm": 0.6048293000235737,
|
||
|
|
"learning_rate": 1.0563153855911445e-08,
|
||
|
|
"loss": 0.2499,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1153523400425911,
|
||
|
|
"step": 3725,
|
||
|
|
"valid_targets_mean": 4798.6,
|
||
|
|
"valid_targets_min": 710
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.946877912395154,
|
||
|
|
"grad_norm": 0.6030921220983857,
|
||
|
|
"learning_rate": 7.760865728747568e-09,
|
||
|
|
"loss": 0.2407,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11895225942134857,
|
||
|
|
"step": 3730,
|
||
|
|
"valid_targets_mean": 4491.9,
|
||
|
|
"valid_targets_min": 680
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.956197576887232,
|
||
|
|
"grad_norm": 0.5398360264305422,
|
||
|
|
"learning_rate": 5.38959660272953e-09,
|
||
|
|
"loss": 0.2464,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07618951052427292,
|
||
|
|
"step": 3735,
|
||
|
|
"valid_targets_mean": 4349.4,
|
||
|
|
"valid_targets_min": 598
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.9655172413793105,
|
||
|
|
"grad_norm": 0.46636388103852433,
|
||
|
|
"learning_rate": 3.4493976007965445e-09,
|
||
|
|
"loss": 0.2411,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07962162047624588,
|
||
|
|
"step": 3740,
|
||
|
|
"valid_targets_mean": 5633.8,
|
||
|
|
"valid_targets_min": 867
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.974836905871388,
|
||
|
|
"grad_norm": 0.4275784183491297,
|
||
|
|
"learning_rate": 1.9403105523130295e-09,
|
||
|
|
"loss": 0.2367,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10192634165287018,
|
||
|
|
"step": 3745,
|
||
|
|
"valid_targets_mean": 5315.2,
|
||
|
|
"valid_targets_min": 565
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.984156570363467,
|
||
|
|
"grad_norm": 0.788353623302605,
|
||
|
|
"learning_rate": 8.623679921626604e-10,
|
||
|
|
"loss": 0.2461,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08159328252077103,
|
||
|
|
"step": 3750,
|
||
|
|
"valid_targets_mean": 2386.1,
|
||
|
|
"valid_targets_min": 626
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.993476234855545,
|
||
|
|
"grad_norm": 0.6447232617207553,
|
||
|
|
"learning_rate": 2.1559316005115294e-10,
|
||
|
|
"loss": 0.2413,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10851238667964935,
|
||
|
|
"step": 3755,
|
||
|
|
"valid_targets_mean": 4990.6,
|
||
|
|
"valid_targets_min": 754
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 7.0,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23336774110794067,
|
||
|
|
"step": 3759,
|
||
|
|
"total_flos": 1.7308138074936443e+18,
|
||
|
|
"train_loss": 0.2947484547389762,
|
||
|
|
"train_runtime": 44573.6566,
|
||
|
|
"train_samples_per_second": 1.347,
|
||
|
|
"train_steps_per_second": 0.084,
|
||
|
|
"valid_targets_mean": 4439.0,
|
||
|
|
"valid_targets_min": 535
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 5,
|
||
|
|
"max_steps": 3759,
|
||
|
|
"num_input_tokens_seen": 0,
|
||
|
|
"num_train_epochs": 7,
|
||
|
|
"save_steps": 1500,
|
||
|
|
"stateful_callbacks": {
|
||
|
|
"TrainerControl": {
|
||
|
|
"args": {
|
||
|
|
"should_epoch_stop": false,
|
||
|
|
"should_evaluate": false,
|
||
|
|
"should_log": false,
|
||
|
|
"should_save": true,
|
||
|
|
"should_training_stop": true
|
||
|
|
},
|
||
|
|
"attributes": {}
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"total_flos": 1.7308138074936443e+18,
|
||
|
|
"train_batch_size": 1,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|