9988 lines
277 KiB
JSON
9988 lines
277 KiB
JSON
|
{
|
||
|
|
"best_global_step": null,
|
||
|
|
"best_metric": null,
|
||
|
|
"best_model_checkpoint": null,
|
||
|
|
"epoch": 7.0,
|
||
|
|
"eval_steps": 500,
|
||
|
|
"global_step": 4522,
|
||
|
|
"is_hyper_param_search": false,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 0.007739938080495356,
|
||
|
|
"grad_norm": 27.556927518982825,
|
||
|
|
"learning_rate": 3.532008830022075e-07,
|
||
|
|
"loss": 0.892,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.8904160261154175,
|
||
|
|
"step": 5,
|
||
|
|
"valid_targets_mean": 3787.4,
|
||
|
|
"valid_targets_min": 443
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.015479876160990712,
|
||
|
|
"grad_norm": 25.83740578639817,
|
||
|
|
"learning_rate": 7.94701986754967e-07,
|
||
|
|
"loss": 0.8816,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.8632640242576599,
|
||
|
|
"step": 10,
|
||
|
|
"valid_targets_mean": 3744.1,
|
||
|
|
"valid_targets_min": 445
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.02321981424148607,
|
||
|
|
"grad_norm": 25.005085818906228,
|
||
|
|
"learning_rate": 1.2362030905077265e-06,
|
||
|
|
"loss": 0.8637,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.8470054864883423,
|
||
|
|
"step": 15,
|
||
|
|
"valid_targets_mean": 3806.1,
|
||
|
|
"valid_targets_min": 364
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.030959752321981424,
|
||
|
|
"grad_norm": 18.660280284469277,
|
||
|
|
"learning_rate": 1.6777041942604857e-06,
|
||
|
|
"loss": 0.7919,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.7651952505111694,
|
||
|
|
"step": 20,
|
||
|
|
"valid_targets_mean": 3883.8,
|
||
|
|
"valid_targets_min": 423
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03869969040247678,
|
||
|
|
"grad_norm": 11.255870766169258,
|
||
|
|
"learning_rate": 2.1192052980132454e-06,
|
||
|
|
"loss": 0.7138,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6832537651062012,
|
||
|
|
"step": 25,
|
||
|
|
"valid_targets_mean": 3872.5,
|
||
|
|
"valid_targets_min": 249
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04643962848297214,
|
||
|
|
"grad_norm": 6.110117995797416,
|
||
|
|
"learning_rate": 2.5607064017660042e-06,
|
||
|
|
"loss": 0.6415,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6171088218688965,
|
||
|
|
"step": 30,
|
||
|
|
"valid_targets_mean": 3780.5,
|
||
|
|
"valid_targets_min": 372
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05417956656346749,
|
||
|
|
"grad_norm": 3.273509381094363,
|
||
|
|
"learning_rate": 3.002207505518764e-06,
|
||
|
|
"loss": 0.5883,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.570875346660614,
|
||
|
|
"step": 35,
|
||
|
|
"valid_targets_mean": 3663.5,
|
||
|
|
"valid_targets_min": 460
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06191950464396285,
|
||
|
|
"grad_norm": 1.9089002542885252,
|
||
|
|
"learning_rate": 3.443708609271523e-06,
|
||
|
|
"loss": 0.5497,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5394970774650574,
|
||
|
|
"step": 40,
|
||
|
|
"valid_targets_mean": 3721.1,
|
||
|
|
"valid_targets_min": 448
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0696594427244582,
|
||
|
|
"grad_norm": 1.4749856282916256,
|
||
|
|
"learning_rate": 3.885209713024282e-06,
|
||
|
|
"loss": 0.5268,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5216352939605713,
|
||
|
|
"step": 45,
|
||
|
|
"valid_targets_mean": 3723.2,
|
||
|
|
"valid_targets_min": 391
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07739938080495357,
|
||
|
|
"grad_norm": 1.2456728516608595,
|
||
|
|
"learning_rate": 4.3267108167770425e-06,
|
||
|
|
"loss": 0.5085,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5022653937339783,
|
||
|
|
"step": 50,
|
||
|
|
"valid_targets_mean": 3877.2,
|
||
|
|
"valid_targets_min": 462
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08513931888544891,
|
||
|
|
"grad_norm": 0.9344457472412111,
|
||
|
|
"learning_rate": 4.768211920529802e-06,
|
||
|
|
"loss": 0.4942,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4939752221107483,
|
||
|
|
"step": 55,
|
||
|
|
"valid_targets_mean": 3845.5,
|
||
|
|
"valid_targets_min": 497
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09287925696594428,
|
||
|
|
"grad_norm": 1.0508961596972632,
|
||
|
|
"learning_rate": 5.209713024282561e-06,
|
||
|
|
"loss": 0.5425,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5407856702804565,
|
||
|
|
"step": 60,
|
||
|
|
"valid_targets_mean": 4920.5,
|
||
|
|
"valid_targets_min": 2219
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.10061919504643962,
|
||
|
|
"grad_norm": 2.404524847584156,
|
||
|
|
"learning_rate": 5.65121412803532e-06,
|
||
|
|
"loss": 0.6234,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.771496057510376,
|
||
|
|
"step": 65,
|
||
|
|
"valid_targets_mean": 965.5,
|
||
|
|
"valid_targets_min": 456
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.10835913312693499,
|
||
|
|
"grad_norm": 0.7500210796142174,
|
||
|
|
"learning_rate": 6.09271523178808e-06,
|
||
|
|
"loss": 0.5762,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5049552917480469,
|
||
|
|
"step": 70,
|
||
|
|
"valid_targets_mean": 4698.2,
|
||
|
|
"valid_targets_min": 3199
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11609907120743033,
|
||
|
|
"grad_norm": 1.6871500381953828,
|
||
|
|
"learning_rate": 6.5342163355408396e-06,
|
||
|
|
"loss": 0.6308,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.7972051501274109,
|
||
|
|
"step": 75,
|
||
|
|
"valid_targets_mean": 877.1,
|
||
|
|
"valid_targets_min": 427
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1238390092879257,
|
||
|
|
"grad_norm": 0.546360990583992,
|
||
|
|
"learning_rate": 6.975717439293599e-06,
|
||
|
|
"loss": 0.4788,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.471763551235199,
|
||
|
|
"step": 80,
|
||
|
|
"valid_targets_mean": 4635.7,
|
||
|
|
"valid_targets_min": 3134
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13157894736842105,
|
||
|
|
"grad_norm": 1.4972099630902207,
|
||
|
|
"learning_rate": 7.417218543046358e-06,
|
||
|
|
"loss": 0.5921,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.7793874740600586,
|
||
|
|
"step": 85,
|
||
|
|
"valid_targets_mean": 855.6,
|
||
|
|
"valid_targets_min": 374
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1393188854489164,
|
||
|
|
"grad_norm": 0.4329830336064823,
|
||
|
|
"learning_rate": 7.858719646799117e-06,
|
||
|
|
"loss": 0.5311,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4547600746154785,
|
||
|
|
"step": 90,
|
||
|
|
"valid_targets_mean": 4777.3,
|
||
|
|
"valid_targets_min": 2671
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14705882352941177,
|
||
|
|
"grad_norm": 1.2092863822799789,
|
||
|
|
"learning_rate": 8.300220750551877e-06,
|
||
|
|
"loss": 0.509,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.7532879114151001,
|
||
|
|
"step": 95,
|
||
|
|
"valid_targets_mean": 898.8,
|
||
|
|
"valid_targets_min": 299
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15479876160990713,
|
||
|
|
"grad_norm": 0.3364633308240861,
|
||
|
|
"learning_rate": 8.741721854304637e-06,
|
||
|
|
"loss": 0.5294,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4063645005226135,
|
||
|
|
"step": 100,
|
||
|
|
"valid_targets_mean": 4618.4,
|
||
|
|
"valid_targets_min": 2204
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16253869969040247,
|
||
|
|
"grad_norm": 0.30813702113400704,
|
||
|
|
"learning_rate": 9.183222958057397e-06,
|
||
|
|
"loss": 0.4197,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.42718881368637085,
|
||
|
|
"step": 105,
|
||
|
|
"valid_targets_mean": 3491.6,
|
||
|
|
"valid_targets_min": 313
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17027863777089783,
|
||
|
|
"grad_norm": 0.2868877308884401,
|
||
|
|
"learning_rate": 9.624724061810156e-06,
|
||
|
|
"loss": 0.5495,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3857671022415161,
|
||
|
|
"step": 110,
|
||
|
|
"valid_targets_mean": 4612.6,
|
||
|
|
"valid_targets_min": 3000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1780185758513932,
|
||
|
|
"grad_norm": 0.2731374867607047,
|
||
|
|
"learning_rate": 1.0066225165562915e-05,
|
||
|
|
"loss": 0.3943,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.39293545484542847,
|
||
|
|
"step": 115,
|
||
|
|
"valid_targets_mean": 3892.6,
|
||
|
|
"valid_targets_min": 238
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18575851393188855,
|
||
|
|
"grad_norm": 0.3285651905194498,
|
||
|
|
"learning_rate": 1.0507726269315674e-05,
|
||
|
|
"loss": 0.5759,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3785663843154907,
|
||
|
|
"step": 120,
|
||
|
|
"valid_targets_mean": 4844.6,
|
||
|
|
"valid_targets_min": 2731
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19349845201238391,
|
||
|
|
"grad_norm": 0.7354604849034491,
|
||
|
|
"learning_rate": 1.0949227373068434e-05,
|
||
|
|
"loss": 0.4273,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6366130709648132,
|
||
|
|
"step": 125,
|
||
|
|
"valid_targets_mean": 984.1,
|
||
|
|
"valid_targets_min": 428
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.20123839009287925,
|
||
|
|
"grad_norm": 0.2636158119865519,
|
||
|
|
"learning_rate": 1.1390728476821193e-05,
|
||
|
|
"loss": 0.5083,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3731876611709595,
|
||
|
|
"step": 130,
|
||
|
|
"valid_targets_mean": 4991.7,
|
||
|
|
"valid_targets_min": 3133
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2089783281733746,
|
||
|
|
"grad_norm": 0.297450354902991,
|
||
|
|
"learning_rate": 1.1832229580573952e-05,
|
||
|
|
"loss": 0.3767,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.39807814359664917,
|
||
|
|
"step": 135,
|
||
|
|
"valid_targets_mean": 3466.8,
|
||
|
|
"valid_targets_min": 300
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21671826625386997,
|
||
|
|
"grad_norm": 0.48109530125201033,
|
||
|
|
"learning_rate": 1.2273730684326713e-05,
|
||
|
|
"loss": 0.5755,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4340590536594391,
|
||
|
|
"step": 140,
|
||
|
|
"valid_targets_mean": 5756.4,
|
||
|
|
"valid_targets_min": 472
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22445820433436534,
|
||
|
|
"grad_norm": 0.3275210496136407,
|
||
|
|
"learning_rate": 1.2715231788079472e-05,
|
||
|
|
"loss": 0.4283,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4109799265861511,
|
||
|
|
"step": 145,
|
||
|
|
"valid_targets_mean": 5488.7,
|
||
|
|
"valid_targets_min": 504
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23219814241486067,
|
||
|
|
"grad_norm": 0.30796738066453233,
|
||
|
|
"learning_rate": 1.3156732891832232e-05,
|
||
|
|
"loss": 0.4393,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4262891411781311,
|
||
|
|
"step": 150,
|
||
|
|
"valid_targets_mean": 5880.8,
|
||
|
|
"valid_targets_min": 492
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23993808049535603,
|
||
|
|
"grad_norm": 0.2545325339807838,
|
||
|
|
"learning_rate": 1.3598233995584991e-05,
|
||
|
|
"loss": 0.4167,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4104169011116028,
|
||
|
|
"step": 155,
|
||
|
|
"valid_targets_mean": 5631.9,
|
||
|
|
"valid_targets_min": 1034
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2476780185758514,
|
||
|
|
"grad_norm": 0.2446617971912774,
|
||
|
|
"learning_rate": 1.403973509933775e-05,
|
||
|
|
"loss": 0.423,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.39815640449523926,
|
||
|
|
"step": 160,
|
||
|
|
"valid_targets_mean": 5617.9,
|
||
|
|
"valid_targets_min": 391
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25541795665634676,
|
||
|
|
"grad_norm": 0.22284165617378673,
|
||
|
|
"learning_rate": 1.448123620309051e-05,
|
||
|
|
"loss": 0.4097,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3992321193218231,
|
||
|
|
"step": 165,
|
||
|
|
"valid_targets_mean": 6436.5,
|
||
|
|
"valid_targets_min": 560
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2631578947368421,
|
||
|
|
"grad_norm": 0.23998966886321677,
|
||
|
|
"learning_rate": 1.4922737306843269e-05,
|
||
|
|
"loss": 0.4083,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.421750009059906,
|
||
|
|
"step": 170,
|
||
|
|
"valid_targets_mean": 5309.5,
|
||
|
|
"valid_targets_min": 482
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2708978328173375,
|
||
|
|
"grad_norm": 0.2652460886376332,
|
||
|
|
"learning_rate": 1.5364238410596028e-05,
|
||
|
|
"loss": 0.3977,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3830735683441162,
|
||
|
|
"step": 175,
|
||
|
|
"valid_targets_mean": 5691.1,
|
||
|
|
"valid_targets_min": 844
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2786377708978328,
|
||
|
|
"grad_norm": 0.20818519699549254,
|
||
|
|
"learning_rate": 1.5805739514348787e-05,
|
||
|
|
"loss": 0.3919,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.38981950283050537,
|
||
|
|
"step": 180,
|
||
|
|
"valid_targets_mean": 5925.6,
|
||
|
|
"valid_targets_min": 500
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.28637770897832815,
|
||
|
|
"grad_norm": 0.30671822619741035,
|
||
|
|
"learning_rate": 1.6247240618101546e-05,
|
||
|
|
"loss": 0.3898,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3816753029823303,
|
||
|
|
"step": 185,
|
||
|
|
"valid_targets_mean": 5658.9,
|
||
|
|
"valid_targets_min": 491
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.29411764705882354,
|
||
|
|
"grad_norm": 0.22176353887772493,
|
||
|
|
"learning_rate": 1.6688741721854306e-05,
|
||
|
|
"loss": 0.3846,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4071277379989624,
|
||
|
|
"step": 190,
|
||
|
|
"valid_targets_mean": 5660.4,
|
||
|
|
"valid_targets_min": 422
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3018575851393189,
|
||
|
|
"grad_norm": 0.20407166925251616,
|
||
|
|
"learning_rate": 1.7130242825607065e-05,
|
||
|
|
"loss": 0.3738,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3546677529811859,
|
||
|
|
"step": 195,
|
||
|
|
"valid_targets_mean": 6256.5,
|
||
|
|
"valid_targets_min": 590
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.30959752321981426,
|
||
|
|
"grad_norm": 0.21994328177215697,
|
||
|
|
"learning_rate": 1.7571743929359824e-05,
|
||
|
|
"loss": 0.3791,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.38544416427612305,
|
||
|
|
"step": 200,
|
||
|
|
"valid_targets_mean": 5626.4,
|
||
|
|
"valid_targets_min": 524
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3173374613003096,
|
||
|
|
"grad_norm": 0.25321655786045555,
|
||
|
|
"learning_rate": 1.8013245033112587e-05,
|
||
|
|
"loss": 0.3722,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.37200507521629333,
|
||
|
|
"step": 205,
|
||
|
|
"valid_targets_mean": 5776.4,
|
||
|
|
"valid_targets_min": 921
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.32507739938080493,
|
||
|
|
"grad_norm": 0.2529712556833944,
|
||
|
|
"learning_rate": 1.8454746136865346e-05,
|
||
|
|
"loss": 0.3708,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.369222491979599,
|
||
|
|
"step": 210,
|
||
|
|
"valid_targets_mean": 5494.7,
|
||
|
|
"valid_targets_min": 1092
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3328173374613003,
|
||
|
|
"grad_norm": 0.27145223441537497,
|
||
|
|
"learning_rate": 1.8896247240618105e-05,
|
||
|
|
"loss": 0.3678,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3673415184020996,
|
||
|
|
"step": 215,
|
||
|
|
"valid_targets_mean": 6014.0,
|
||
|
|
"valid_targets_min": 234
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.34055727554179566,
|
||
|
|
"grad_norm": 0.24180531829358837,
|
||
|
|
"learning_rate": 1.9337748344370864e-05,
|
||
|
|
"loss": 0.3826,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.38699570298194885,
|
||
|
|
"step": 220,
|
||
|
|
"valid_targets_mean": 6489.6,
|
||
|
|
"valid_targets_min": 487
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.34829721362229105,
|
||
|
|
"grad_norm": 0.35281465810969276,
|
||
|
|
"learning_rate": 1.9779249448123624e-05,
|
||
|
|
"loss": 0.5078,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4161370098590851,
|
||
|
|
"step": 225,
|
||
|
|
"valid_targets_mean": 3702.6,
|
||
|
|
"valid_targets_min": 466
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3560371517027864,
|
||
|
|
"grad_norm": 0.2818358878789778,
|
||
|
|
"learning_rate": 2.0220750551876383e-05,
|
||
|
|
"loss": 0.3645,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.36115381121635437,
|
||
|
|
"step": 230,
|
||
|
|
"valid_targets_mean": 5997.3,
|
||
|
|
"valid_targets_min": 240
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3637770897832817,
|
||
|
|
"grad_norm": 0.37869304711480956,
|
||
|
|
"learning_rate": 2.066225165562914e-05,
|
||
|
|
"loss": 0.4724,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3962448835372925,
|
||
|
|
"step": 235,
|
||
|
|
"valid_targets_mean": 4160.1,
|
||
|
|
"valid_targets_min": 473
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3715170278637771,
|
||
|
|
"grad_norm": 0.24128327528471735,
|
||
|
|
"learning_rate": 2.11037527593819e-05,
|
||
|
|
"loss": 0.361,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3546305000782013,
|
||
|
|
"step": 240,
|
||
|
|
"valid_targets_mean": 6533.9,
|
||
|
|
"valid_targets_min": 1876
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.37925696594427244,
|
||
|
|
"grad_norm": 0.2934150431724293,
|
||
|
|
"learning_rate": 2.1545253863134657e-05,
|
||
|
|
"loss": 0.4663,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.35832321643829346,
|
||
|
|
"step": 245,
|
||
|
|
"valid_targets_mean": 6688.6,
|
||
|
|
"valid_targets_min": 298
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38699690402476783,
|
||
|
|
"grad_norm": 0.24569929082029643,
|
||
|
|
"learning_rate": 2.198675496688742e-05,
|
||
|
|
"loss": 0.365,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3805839419364929,
|
||
|
|
"step": 250,
|
||
|
|
"valid_targets_mean": 6724.0,
|
||
|
|
"valid_targets_min": 549
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39473684210526316,
|
||
|
|
"grad_norm": 0.2676607413364154,
|
||
|
|
"learning_rate": 2.242825607064018e-05,
|
||
|
|
"loss": 0.5116,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3619913160800934,
|
||
|
|
"step": 255,
|
||
|
|
"valid_targets_mean": 6117.0,
|
||
|
|
"valid_targets_min": 219
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4024767801857585,
|
||
|
|
"grad_norm": 0.21831069654210952,
|
||
|
|
"learning_rate": 2.286975717439294e-05,
|
||
|
|
"loss": 0.3526,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3447280526161194,
|
||
|
|
"step": 260,
|
||
|
|
"valid_targets_mean": 6469.6,
|
||
|
|
"valid_targets_min": 209
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4102167182662539,
|
||
|
|
"grad_norm": 0.323008560170421,
|
||
|
|
"learning_rate": 2.3311258278145698e-05,
|
||
|
|
"loss": 0.468,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3715176284313202,
|
||
|
|
"step": 265,
|
||
|
|
"valid_targets_mean": 6510.8,
|
||
|
|
"valid_targets_min": 206
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4179566563467492,
|
||
|
|
"grad_norm": 0.26909093178554144,
|
||
|
|
"learning_rate": 2.3752759381898457e-05,
|
||
|
|
"loss": 0.3478,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3583526611328125,
|
||
|
|
"step": 270,
|
||
|
|
"valid_targets_mean": 5183.3,
|
||
|
|
"valid_targets_min": 399
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.42569659442724456,
|
||
|
|
"grad_norm": 0.30810837217677667,
|
||
|
|
"learning_rate": 2.4194260485651216e-05,
|
||
|
|
"loss": 0.5027,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3488261103630066,
|
||
|
|
"step": 275,
|
||
|
|
"valid_targets_mean": 6249.5,
|
||
|
|
"valid_targets_min": 284
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.43343653250773995,
|
||
|
|
"grad_norm": 0.26894531256040477,
|
||
|
|
"learning_rate": 2.4635761589403975e-05,
|
||
|
|
"loss": 0.3467,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.35679471492767334,
|
||
|
|
"step": 280,
|
||
|
|
"valid_targets_mean": 5410.6,
|
||
|
|
"valid_targets_min": 265
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4411764705882353,
|
||
|
|
"grad_norm": 0.3312324996386416,
|
||
|
|
"learning_rate": 2.5077262693156735e-05,
|
||
|
|
"loss": 0.4927,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.34337306022644043,
|
||
|
|
"step": 285,
|
||
|
|
"valid_targets_mean": 6375.5,
|
||
|
|
"valid_targets_min": 401
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.44891640866873067,
|
||
|
|
"grad_norm": 0.39494218666902076,
|
||
|
|
"learning_rate": 2.5518763796909497e-05,
|
||
|
|
"loss": 0.3531,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3893047571182251,
|
||
|
|
"step": 290,
|
||
|
|
"valid_targets_mean": 3327.1,
|
||
|
|
"valid_targets_min": 428
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.456656346749226,
|
||
|
|
"grad_norm": 1.0882850361109044,
|
||
|
|
"learning_rate": 2.5960264900662253e-05,
|
||
|
|
"loss": 0.699,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6760176420211792,
|
||
|
|
"step": 295,
|
||
|
|
"valid_targets_mean": 3046.5,
|
||
|
|
"valid_targets_min": 249
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.46439628482972134,
|
||
|
|
"grad_norm": 0.5746604494320542,
|
||
|
|
"learning_rate": 2.6401766004415016e-05,
|
||
|
|
"loss": 0.6483,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6252200603485107,
|
||
|
|
"step": 300,
|
||
|
|
"valid_targets_mean": 3748.6,
|
||
|
|
"valid_targets_min": 339
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47213622291021673,
|
||
|
|
"grad_norm": 0.41455786824108815,
|
||
|
|
"learning_rate": 2.684326710816777e-05,
|
||
|
|
"loss": 0.6354,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6014313697814941,
|
||
|
|
"step": 305,
|
||
|
|
"valid_targets_mean": 3423.9,
|
||
|
|
"valid_targets_min": 455
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47987616099071206,
|
||
|
|
"grad_norm": 0.34007638232435167,
|
||
|
|
"learning_rate": 2.7284768211920534e-05,
|
||
|
|
"loss": 0.623,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5637083649635315,
|
||
|
|
"step": 310,
|
||
|
|
"valid_targets_mean": 4002.1,
|
||
|
|
"valid_targets_min": 292
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.48761609907120745,
|
||
|
|
"grad_norm": 0.382117908367287,
|
||
|
|
"learning_rate": 2.772626931567329e-05,
|
||
|
|
"loss": 0.6261,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6549586653709412,
|
||
|
|
"step": 315,
|
||
|
|
"valid_targets_mean": 3102.1,
|
||
|
|
"valid_targets_min": 317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4953560371517028,
|
||
|
|
"grad_norm": 0.3940906758062528,
|
||
|
|
"learning_rate": 2.8167770419426053e-05,
|
||
|
|
"loss": 0.5969,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6056185364723206,
|
||
|
|
"step": 320,
|
||
|
|
"valid_targets_mean": 2004.2,
|
||
|
|
"valid_targets_min": 192
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5030959752321982,
|
||
|
|
"grad_norm": 0.375885909620112,
|
||
|
|
"learning_rate": 2.860927152317881e-05,
|
||
|
|
"loss": 0.615,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6279362440109253,
|
||
|
|
"step": 325,
|
||
|
|
"valid_targets_mean": 2664.8,
|
||
|
|
"valid_targets_min": 247
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5108359133126935,
|
||
|
|
"grad_norm": 0.33582760402597334,
|
||
|
|
"learning_rate": 2.905077262693157e-05,
|
||
|
|
"loss": 0.5966,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6175297498703003,
|
||
|
|
"step": 330,
|
||
|
|
"valid_targets_mean": 3044.0,
|
||
|
|
"valid_targets_min": 298
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5185758513931888,
|
||
|
|
"grad_norm": 0.27046783261811397,
|
||
|
|
"learning_rate": 2.9492273730684327e-05,
|
||
|
|
"loss": 0.584,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5971802473068237,
|
||
|
|
"step": 335,
|
||
|
|
"valid_targets_mean": 3938.2,
|
||
|
|
"valid_targets_min": 941
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5263157894736842,
|
||
|
|
"grad_norm": 0.28776196940498977,
|
||
|
|
"learning_rate": 2.993377483443709e-05,
|
||
|
|
"loss": 0.5947,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5944355726242065,
|
||
|
|
"step": 340,
|
||
|
|
"valid_targets_mean": 3717.0,
|
||
|
|
"valid_targets_min": 403
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5340557275541795,
|
||
|
|
"grad_norm": 0.2561673068863187,
|
||
|
|
"learning_rate": 3.037527593818985e-05,
|
||
|
|
"loss": 0.569,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5431984663009644,
|
||
|
|
"step": 345,
|
||
|
|
"valid_targets_mean": 4406.2,
|
||
|
|
"valid_targets_min": 387
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.541795665634675,
|
||
|
|
"grad_norm": 0.2876476408420966,
|
||
|
|
"learning_rate": 3.081677704194261e-05,
|
||
|
|
"loss": 0.5948,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.631894588470459,
|
||
|
|
"step": 350,
|
||
|
|
"valid_targets_mean": 3612.3,
|
||
|
|
"valid_targets_min": 371
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5495356037151703,
|
||
|
|
"grad_norm": 0.3467225994463012,
|
||
|
|
"learning_rate": 3.1258278145695364e-05,
|
||
|
|
"loss": 0.5648,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5806475877761841,
|
||
|
|
"step": 355,
|
||
|
|
"valid_targets_mean": 2256.9,
|
||
|
|
"valid_targets_min": 283
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5572755417956656,
|
||
|
|
"grad_norm": 0.2978415489318983,
|
||
|
|
"learning_rate": 3.169977924944813e-05,
|
||
|
|
"loss": 0.5789,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.593154788017273,
|
||
|
|
"step": 360,
|
||
|
|
"valid_targets_mean": 2972.2,
|
||
|
|
"valid_targets_min": 318
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.565015479876161,
|
||
|
|
"grad_norm": 0.49143740553944854,
|
||
|
|
"learning_rate": 3.214128035320088e-05,
|
||
|
|
"loss": 0.6138,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6824336051940918,
|
||
|
|
"step": 365,
|
||
|
|
"valid_targets_mean": 3760.0,
|
||
|
|
"valid_targets_min": 1056
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5727554179566563,
|
||
|
|
"grad_norm": 0.3948478797898095,
|
||
|
|
"learning_rate": 3.2582781456953645e-05,
|
||
|
|
"loss": 0.6943,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.641958475112915,
|
||
|
|
"step": 370,
|
||
|
|
"valid_targets_mean": 4103.2,
|
||
|
|
"valid_targets_min": 797
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5804953560371517,
|
||
|
|
"grad_norm": 0.402162923706298,
|
||
|
|
"learning_rate": 3.30242825607064e-05,
|
||
|
|
"loss": 0.7164,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.7169279456138611,
|
||
|
|
"step": 375,
|
||
|
|
"valid_targets_mean": 3301.9,
|
||
|
|
"valid_targets_min": 791
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5882352941176471,
|
||
|
|
"grad_norm": 0.3206597394354228,
|
||
|
|
"learning_rate": 3.3465783664459164e-05,
|
||
|
|
"loss": 0.6949,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6718679070472717,
|
||
|
|
"step": 380,
|
||
|
|
"valid_targets_mean": 4432.9,
|
||
|
|
"valid_targets_min": 877
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5959752321981424,
|
||
|
|
"grad_norm": 0.3040223910621384,
|
||
|
|
"learning_rate": 3.3907284768211926e-05,
|
||
|
|
"loss": 0.6864,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6755282878875732,
|
||
|
|
"step": 385,
|
||
|
|
"valid_targets_mean": 4270.7,
|
||
|
|
"valid_targets_min": 907
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6037151702786377,
|
||
|
|
"grad_norm": 0.28120388810976427,
|
||
|
|
"learning_rate": 3.434878587196468e-05,
|
||
|
|
"loss": 0.6843,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6924828290939331,
|
||
|
|
"step": 390,
|
||
|
|
"valid_targets_mean": 4288.7,
|
||
|
|
"valid_targets_min": 758
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6114551083591331,
|
||
|
|
"grad_norm": 0.2751222830388332,
|
||
|
|
"learning_rate": 3.4790286975717445e-05,
|
||
|
|
"loss": 0.6884,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6589582562446594,
|
||
|
|
"step": 395,
|
||
|
|
"valid_targets_mean": 4493.8,
|
||
|
|
"valid_targets_min": 1317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6191950464396285,
|
||
|
|
"grad_norm": 0.24002550970370776,
|
||
|
|
"learning_rate": 3.52317880794702e-05,
|
||
|
|
"loss": 0.6668,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.619488000869751,
|
||
|
|
"step": 400,
|
||
|
|
"valid_targets_mean": 4781.4,
|
||
|
|
"valid_targets_min": 671
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6269349845201239,
|
||
|
|
"grad_norm": 0.30164597012451416,
|
||
|
|
"learning_rate": 3.567328918322296e-05,
|
||
|
|
"loss": 0.6829,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6519378423690796,
|
||
|
|
"step": 405,
|
||
|
|
"valid_targets_mean": 4028.6,
|
||
|
|
"valid_targets_min": 1025
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6346749226006192,
|
||
|
|
"grad_norm": 0.32111469384583546,
|
||
|
|
"learning_rate": 3.611479028697572e-05,
|
||
|
|
"loss": 0.6674,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.7031950950622559,
|
||
|
|
"step": 410,
|
||
|
|
"valid_targets_mean": 3116.2,
|
||
|
|
"valid_targets_min": 442
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6424148606811145,
|
||
|
|
"grad_norm": 0.2712892665192101,
|
||
|
|
"learning_rate": 3.655629139072848e-05,
|
||
|
|
"loss": 0.6834,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.7068458795547485,
|
||
|
|
"step": 415,
|
||
|
|
"valid_targets_mean": 4106.4,
|
||
|
|
"valid_targets_min": 854
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6501547987616099,
|
||
|
|
"grad_norm": 0.3005853741441399,
|
||
|
|
"learning_rate": 3.699779249448124e-05,
|
||
|
|
"loss": 0.6785,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.668350100517273,
|
||
|
|
"step": 420,
|
||
|
|
"valid_targets_mean": 3609.7,
|
||
|
|
"valid_targets_min": 1029
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6578947368421053,
|
||
|
|
"grad_norm": 0.27670923688419646,
|
||
|
|
"learning_rate": 3.7439293598234e-05,
|
||
|
|
"loss": 0.6842,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.673589825630188,
|
||
|
|
"step": 425,
|
||
|
|
"valid_targets_mean": 4175.3,
|
||
|
|
"valid_targets_min": 1010
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6656346749226006,
|
||
|
|
"grad_norm": 0.34726927864442714,
|
||
|
|
"learning_rate": 3.7880794701986756e-05,
|
||
|
|
"loss": 0.6614,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6509352326393127,
|
||
|
|
"step": 430,
|
||
|
|
"valid_targets_mean": 3942.0,
|
||
|
|
"valid_targets_min": 1104
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.673374613003096,
|
||
|
|
"grad_norm": 0.3108276325016972,
|
||
|
|
"learning_rate": 3.832229580573952e-05,
|
||
|
|
"loss": 0.6797,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6784442067146301,
|
||
|
|
"step": 435,
|
||
|
|
"valid_targets_mean": 3787.9,
|
||
|
|
"valid_targets_min": 776
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6811145510835913,
|
||
|
|
"grad_norm": 0.4223824064369445,
|
||
|
|
"learning_rate": 3.8763796909492275e-05,
|
||
|
|
"loss": 0.629,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5819063782691956,
|
||
|
|
"step": 440,
|
||
|
|
"valid_targets_mean": 3564.7,
|
||
|
|
"valid_targets_min": 900
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6888544891640866,
|
||
|
|
"grad_norm": 0.3139755574168235,
|
||
|
|
"learning_rate": 3.920529801324504e-05,
|
||
|
|
"loss": 0.5674,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5353734493255615,
|
||
|
|
"step": 445,
|
||
|
|
"valid_targets_mean": 4363.4,
|
||
|
|
"valid_targets_min": 750
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6965944272445821,
|
||
|
|
"grad_norm": 0.3352957129000239,
|
||
|
|
"learning_rate": 3.964679911699779e-05,
|
||
|
|
"loss": 0.5623,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5364030599594116,
|
||
|
|
"step": 450,
|
||
|
|
"valid_targets_mean": 3516.3,
|
||
|
|
"valid_targets_min": 767
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7043343653250774,
|
||
|
|
"grad_norm": 0.27671133483601845,
|
||
|
|
"learning_rate": 3.999999403892832e-05,
|
||
|
|
"loss": 0.5615,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.535529613494873,
|
||
|
|
"step": 455,
|
||
|
|
"valid_targets_mean": 4421.9,
|
||
|
|
"valid_targets_min": 880
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7120743034055728,
|
||
|
|
"grad_norm": 0.33159154931191204,
|
||
|
|
"learning_rate": 3.999978540179264e-05,
|
||
|
|
"loss": 0.5692,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5733986496925354,
|
||
|
|
"step": 460,
|
||
|
|
"valid_targets_mean": 3265.3,
|
||
|
|
"valid_targets_min": 906
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7198142414860681,
|
||
|
|
"grad_norm": 0.27863339298056705,
|
||
|
|
"learning_rate": 3.9999278714626385e-05,
|
||
|
|
"loss": 0.5396,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5022937059402466,
|
||
|
|
"step": 465,
|
||
|
|
"valid_targets_mean": 4266.4,
|
||
|
|
"valid_targets_min": 1047
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7275541795665634,
|
||
|
|
"grad_norm": 0.3215020958488482,
|
||
|
|
"learning_rate": 3.999847398498055e-05,
|
||
|
|
"loss": 0.5677,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5704535245895386,
|
||
|
|
"step": 470,
|
||
|
|
"valid_targets_mean": 3498.2,
|
||
|
|
"valid_targets_min": 1036
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7352941176470589,
|
||
|
|
"grad_norm": 0.24782635183685015,
|
||
|
|
"learning_rate": 3.9997371224847754e-05,
|
||
|
|
"loss": 0.5454,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5425188541412354,
|
||
|
|
"step": 475,
|
||
|
|
"valid_targets_mean": 4387.9,
|
||
|
|
"valid_targets_min": 635
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7430340557275542,
|
||
|
|
"grad_norm": 0.2954436790597722,
|
||
|
|
"learning_rate": 3.999597045066204e-05,
|
||
|
|
"loss": 0.5519,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.545947253704071,
|
||
|
|
"step": 480,
|
||
|
|
"valid_targets_mean": 3467.2,
|
||
|
|
"valid_targets_min": 695
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7507739938080495,
|
||
|
|
"grad_norm": 0.23719265876687168,
|
||
|
|
"learning_rate": 3.999427168329869e-05,
|
||
|
|
"loss": 0.5362,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5340520739555359,
|
||
|
|
"step": 485,
|
||
|
|
"valid_targets_mean": 4467.1,
|
||
|
|
"valid_targets_min": 613
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7585139318885449,
|
||
|
|
"grad_norm": 0.29791010866331696,
|
||
|
|
"learning_rate": 3.9992274948073856e-05,
|
||
|
|
"loss": 0.5578,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5808913707733154,
|
||
|
|
"step": 490,
|
||
|
|
"valid_targets_mean": 3536.6,
|
||
|
|
"valid_targets_min": 834
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7662538699690402,
|
||
|
|
"grad_norm": 0.2766837442133623,
|
||
|
|
"learning_rate": 3.99899802747442e-05,
|
||
|
|
"loss": 0.5458,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5579825043678284,
|
||
|
|
"step": 495,
|
||
|
|
"valid_targets_mean": 4088.9,
|
||
|
|
"valid_targets_min": 837
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7739938080495357,
|
||
|
|
"grad_norm": 0.30874766806242304,
|
||
|
|
"learning_rate": 3.998738769750647e-05,
|
||
|
|
"loss": 0.5506,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5537128448486328,
|
||
|
|
"step": 500,
|
||
|
|
"valid_targets_mean": 3560.9,
|
||
|
|
"valid_targets_min": 1032
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.781733746130031,
|
||
|
|
"grad_norm": 0.24295732633093,
|
||
|
|
"learning_rate": 3.998449725499696e-05,
|
||
|
|
"loss": 0.5412,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5296654105186462,
|
||
|
|
"step": 505,
|
||
|
|
"valid_targets_mean": 4477.0,
|
||
|
|
"valid_targets_min": 502
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7894736842105263,
|
||
|
|
"grad_norm": 0.31515505271622524,
|
||
|
|
"learning_rate": 3.9981308990290964e-05,
|
||
|
|
"loss": 0.5434,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.553435742855072,
|
||
|
|
"step": 510,
|
||
|
|
"valid_targets_mean": 3193.1,
|
||
|
|
"valid_targets_min": 883
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7972136222910217,
|
||
|
|
"grad_norm": 0.2599867754508887,
|
||
|
|
"learning_rate": 3.997782295090209e-05,
|
||
|
|
"loss": 0.5508,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5340229272842407,
|
||
|
|
"step": 515,
|
||
|
|
"valid_targets_mean": 4601.6,
|
||
|
|
"valid_targets_min": 602
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.804953560371517,
|
||
|
|
"grad_norm": 0.28684289447261996,
|
||
|
|
"learning_rate": 3.9974039188781635e-05,
|
||
|
|
"loss": 0.5436,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5592387318611145,
|
||
|
|
"step": 520,
|
||
|
|
"valid_targets_mean": 3430.9,
|
||
|
|
"valid_targets_min": 1017
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8126934984520123,
|
||
|
|
"grad_norm": 0.35078102252774024,
|
||
|
|
"learning_rate": 3.99699577603177e-05,
|
||
|
|
"loss": 0.5297,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5260142683982849,
|
||
|
|
"step": 525,
|
||
|
|
"valid_targets_mean": 3828.6,
|
||
|
|
"valid_targets_min": 1071
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8204334365325078,
|
||
|
|
"grad_norm": 0.3210958740139956,
|
||
|
|
"learning_rate": 3.996557872633444e-05,
|
||
|
|
"loss": 0.6326,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5367951393127441,
|
||
|
|
"step": 530,
|
||
|
|
"valid_targets_mean": 3390.7,
|
||
|
|
"valid_targets_min": 996
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8281733746130031,
|
||
|
|
"grad_norm": 0.2815815654951537,
|
||
|
|
"learning_rate": 3.996090215209112e-05,
|
||
|
|
"loss": 0.537,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5194051265716553,
|
||
|
|
"step": 535,
|
||
|
|
"valid_targets_mean": 3718.5,
|
||
|
|
"valid_targets_min": 819
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8359133126934984,
|
||
|
|
"grad_norm": 0.3797984831660266,
|
||
|
|
"learning_rate": 3.995592810728114e-05,
|
||
|
|
"loss": 0.646,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5256558656692505,
|
||
|
|
"step": 540,
|
||
|
|
"valid_targets_mean": 3193.4,
|
||
|
|
"valid_targets_min": 558
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8436532507739938,
|
||
|
|
"grad_norm": 0.27779660118942445,
|
||
|
|
"learning_rate": 3.9950656666030995e-05,
|
||
|
|
"loss": 0.5302,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5189313292503357,
|
||
|
|
"step": 545,
|
||
|
|
"valid_targets_mean": 3690.3,
|
||
|
|
"valid_targets_min": 434
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8513931888544891,
|
||
|
|
"grad_norm": 0.3389855210231446,
|
||
|
|
"learning_rate": 3.99450879068992e-05,
|
||
|
|
"loss": 0.5942,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5512515902519226,
|
||
|
|
"step": 550,
|
||
|
|
"valid_targets_mean": 3144.1,
|
||
|
|
"valid_targets_min": 742
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8591331269349846,
|
||
|
|
"grad_norm": 0.3482199450875854,
|
||
|
|
"learning_rate": 3.993922191287509e-05,
|
||
|
|
"loss": 0.5273,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5323893427848816,
|
||
|
|
"step": 555,
|
||
|
|
"valid_targets_mean": 3871.4,
|
||
|
|
"valid_targets_min": 502
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8668730650154799,
|
||
|
|
"grad_norm": 0.35141532816152465,
|
||
|
|
"learning_rate": 3.9933058771377574e-05,
|
||
|
|
"loss": 0.5875,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5391024351119995,
|
||
|
|
"step": 560,
|
||
|
|
"valid_targets_mean": 3256.5,
|
||
|
|
"valid_targets_min": 999
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8746130030959752,
|
||
|
|
"grad_norm": 0.32138910916724966,
|
||
|
|
"learning_rate": 3.992659857425389e-05,
|
||
|
|
"loss": 0.5292,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5394055247306824,
|
||
|
|
"step": 565,
|
||
|
|
"valid_targets_mean": 3665.0,
|
||
|
|
"valid_targets_min": 640
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8823529411764706,
|
||
|
|
"grad_norm": 0.3825214069962551,
|
||
|
|
"learning_rate": 3.991984141777813e-05,
|
||
|
|
"loss": 0.6388,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.552566647529602,
|
||
|
|
"step": 570,
|
||
|
|
"valid_targets_mean": 3457.3,
|
||
|
|
"valid_targets_min": 1110
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8900928792569659,
|
||
|
|
"grad_norm": 0.34481419478575126,
|
||
|
|
"learning_rate": 3.9912787402649945e-05,
|
||
|
|
"loss": 0.5203,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5106000900268555,
|
||
|
|
"step": 575,
|
||
|
|
"valid_targets_mean": 3141.0,
|
||
|
|
"valid_targets_min": 566
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8978328173374613,
|
||
|
|
"grad_norm": 0.3739562669842394,
|
||
|
|
"learning_rate": 3.990543663399291e-05,
|
||
|
|
"loss": 0.622,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5330145955085754,
|
||
|
|
"step": 580,
|
||
|
|
"valid_targets_mean": 3509.7,
|
||
|
|
"valid_targets_min": 780
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9055727554179567,
|
||
|
|
"grad_norm": 0.3348505399780836,
|
||
|
|
"learning_rate": 3.989778922135305e-05,
|
||
|
|
"loss": 0.5337,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5145993232727051,
|
||
|
|
"step": 585,
|
||
|
|
"valid_targets_mean": 3106.4,
|
||
|
|
"valid_targets_min": 794
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.913312693498452,
|
||
|
|
"grad_norm": 0.3873353157680655,
|
||
|
|
"learning_rate": 3.988984527869718e-05,
|
||
|
|
"loss": 0.5691,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5175759196281433,
|
||
|
|
"step": 590,
|
||
|
|
"valid_targets_mean": 3530.0,
|
||
|
|
"valid_targets_min": 744
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9210526315789473,
|
||
|
|
"grad_norm": 0.3359366524024443,
|
||
|
|
"learning_rate": 3.988160492441116e-05,
|
||
|
|
"loss": 0.5303,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5438101291656494,
|
||
|
|
"step": 595,
|
||
|
|
"valid_targets_mean": 3219.6,
|
||
|
|
"valid_targets_min": 435
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9287925696594427,
|
||
|
|
"grad_norm": 1.7207426883364212,
|
||
|
|
"learning_rate": 3.987306828129823e-05,
|
||
|
|
"loss": 0.5296,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3982352018356323,
|
||
|
|
"step": 600,
|
||
|
|
"valid_targets_mean": 4948.2,
|
||
|
|
"valid_targets_min": 1383
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9365325077399381,
|
||
|
|
"grad_norm": 0.5343111344299726,
|
||
|
|
"learning_rate": 3.986423547657706e-05,
|
||
|
|
"loss": 0.3516,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.31627655029296875,
|
||
|
|
"step": 605,
|
||
|
|
"valid_targets_mean": 4927.9,
|
||
|
|
"valid_targets_min": 1102
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9442724458204335,
|
||
|
|
"grad_norm": 0.4950142470043242,
|
||
|
|
"learning_rate": 3.985510664187998e-05,
|
||
|
|
"loss": 0.3171,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3267781734466553,
|
||
|
|
"step": 610,
|
||
|
|
"valid_targets_mean": 4602.5,
|
||
|
|
"valid_targets_min": 1297
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9520123839009288,
|
||
|
|
"grad_norm": 0.37176927038082724,
|
||
|
|
"learning_rate": 3.9845681913250905e-05,
|
||
|
|
"loss": 0.2737,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27025002241134644,
|
||
|
|
"step": 615,
|
||
|
|
"valid_targets_mean": 4983.0,
|
||
|
|
"valid_targets_min": 938
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9597523219814241,
|
||
|
|
"grad_norm": 0.36396182688735873,
|
||
|
|
"learning_rate": 3.983596143114337e-05,
|
||
|
|
"loss": 0.2785,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28528472781181335,
|
||
|
|
"step": 620,
|
||
|
|
"valid_targets_mean": 4184.5,
|
||
|
|
"valid_targets_min": 1158
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9674922600619195,
|
||
|
|
"grad_norm": 0.3557625588894437,
|
||
|
|
"learning_rate": 3.9825945340418446e-05,
|
||
|
|
"loss": 0.2831,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27326470613479614,
|
||
|
|
"step": 625,
|
||
|
|
"valid_targets_mean": 4258.2,
|
||
|
|
"valid_targets_min": 662
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9752321981424149,
|
||
|
|
"grad_norm": 0.30812730643038905,
|
||
|
|
"learning_rate": 3.9815633790342525e-05,
|
||
|
|
"loss": 0.2744,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2685595154762268,
|
||
|
|
"step": 630,
|
||
|
|
"valid_targets_mean": 4959.4,
|
||
|
|
"valid_targets_min": 2049
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9829721362229102,
|
||
|
|
"grad_norm": 0.4825624237839052,
|
||
|
|
"learning_rate": 3.980502693458516e-05,
|
||
|
|
"loss": 0.2893,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28646308183670044,
|
||
|
|
"step": 635,
|
||
|
|
"valid_targets_mean": 3119.1,
|
||
|
|
"valid_targets_min": 1116
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9907120743034056,
|
||
|
|
"grad_norm": 0.4111094435742784,
|
||
|
|
"learning_rate": 3.9794124931216724e-05,
|
||
|
|
"loss": 0.2805,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2639378309249878,
|
||
|
|
"step": 640,
|
||
|
|
"valid_targets_mean": 2866.9,
|
||
|
|
"valid_targets_min": 1189
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9984520123839009,
|
||
|
|
"grad_norm": 0.46785914905775106,
|
||
|
|
"learning_rate": 3.9782927942706086e-05,
|
||
|
|
"loss": 0.2661,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2755046486854553,
|
||
|
|
"step": 645,
|
||
|
|
"valid_targets_mean": 3248.7,
|
||
|
|
"valid_targets_min": 1746
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0061919504643964,
|
||
|
|
"grad_norm": 0.5319148163563938,
|
||
|
|
"learning_rate": 3.977143613591817e-05,
|
||
|
|
"loss": 0.3265,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.32363638281822205,
|
||
|
|
"step": 650,
|
||
|
|
"valid_targets_mean": 3799.7,
|
||
|
|
"valid_targets_min": 470
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0139318885448916,
|
||
|
|
"grad_norm": 0.4246865898408867,
|
||
|
|
"learning_rate": 3.9759649682111484e-05,
|
||
|
|
"loss": 0.3096,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30781614780426025,
|
||
|
|
"step": 655,
|
||
|
|
"valid_targets_mean": 3723.8,
|
||
|
|
"valid_targets_min": 388
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.021671826625387,
|
||
|
|
"grad_norm": 0.34137841867743135,
|
||
|
|
"learning_rate": 3.9747568756935555e-05,
|
||
|
|
"loss": 0.2986,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30192291736602783,
|
||
|
|
"step": 660,
|
||
|
|
"valid_targets_mean": 3488.4,
|
||
|
|
"valid_targets_min": 338
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0294117647058822,
|
||
|
|
"grad_norm": 0.2645686351341279,
|
||
|
|
"learning_rate": 3.973519354042833e-05,
|
||
|
|
"loss": 0.2921,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28123363852500916,
|
||
|
|
"step": 665,
|
||
|
|
"valid_targets_mean": 3832.2,
|
||
|
|
"valid_targets_min": 506
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0371517027863777,
|
||
|
|
"grad_norm": 0.25230597151294065,
|
||
|
|
"learning_rate": 3.972252421701346e-05,
|
||
|
|
"loss": 0.2853,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2776462733745575,
|
||
|
|
"step": 670,
|
||
|
|
"valid_targets_mean": 3689.5,
|
||
|
|
"valid_targets_min": 397
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0448916408668731,
|
||
|
|
"grad_norm": 0.24245916938203588,
|
||
|
|
"learning_rate": 3.97095609754976e-05,
|
||
|
|
"loss": 0.2813,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27049681544303894,
|
||
|
|
"step": 675,
|
||
|
|
"valid_targets_mean": 3592.6,
|
||
|
|
"valid_targets_min": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0526315789473684,
|
||
|
|
"grad_norm": 0.2419732498578009,
|
||
|
|
"learning_rate": 3.969630400906753e-05,
|
||
|
|
"loss": 0.2782,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2769536077976227,
|
||
|
|
"step": 680,
|
||
|
|
"valid_targets_mean": 3646.3,
|
||
|
|
"valid_targets_min": 428
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0603715170278638,
|
||
|
|
"grad_norm": 0.22727538707855877,
|
||
|
|
"learning_rate": 3.9682753515287334e-05,
|
||
|
|
"loss": 0.2744,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2753305733203888,
|
||
|
|
"step": 685,
|
||
|
|
"valid_targets_mean": 3811.8,
|
||
|
|
"valid_targets_min": 430
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.068111455108359,
|
||
|
|
"grad_norm": 0.2533839742087906,
|
||
|
|
"learning_rate": 3.9668909696095444e-05,
|
||
|
|
"loss": 0.27,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2791718542575836,
|
||
|
|
"step": 690,
|
||
|
|
"valid_targets_mean": 3824.1,
|
||
|
|
"valid_targets_min": 349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0758513931888545,
|
||
|
|
"grad_norm": 0.2322787444698477,
|
||
|
|
"learning_rate": 3.9654772757801606e-05,
|
||
|
|
"loss": 0.2692,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2597501873970032,
|
||
|
|
"step": 695,
|
||
|
|
"valid_targets_mean": 3658.0,
|
||
|
|
"valid_targets_min": 444
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.08359133126935,
|
||
|
|
"grad_norm": 0.22745117941648865,
|
||
|
|
"learning_rate": 3.964034291108382e-05,
|
||
|
|
"loss": 0.2704,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2703322470188141,
|
||
|
|
"step": 700,
|
||
|
|
"valid_targets_mean": 3583.1,
|
||
|
|
"valid_targets_min": 457
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0913312693498451,
|
||
|
|
"grad_norm": 0.4034079172121534,
|
||
|
|
"learning_rate": 3.962562037098521e-05,
|
||
|
|
"loss": 0.3174,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.332100510597229,
|
||
|
|
"step": 705,
|
||
|
|
"valid_targets_mean": 4880.6,
|
||
|
|
"valid_targets_min": 3320
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0990712074303406,
|
||
|
|
"grad_norm": 0.7668348372190322,
|
||
|
|
"learning_rate": 3.961060535691081e-05,
|
||
|
|
"loss": 0.3621,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5471717119216919,
|
||
|
|
"step": 710,
|
||
|
|
"valid_targets_mean": 978.9,
|
||
|
|
"valid_targets_min": 200
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1068111455108358,
|
||
|
|
"grad_norm": 0.3877763233460875,
|
||
|
|
"learning_rate": 3.959529809262429e-05,
|
||
|
|
"loss": 0.4204,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3108642101287842,
|
||
|
|
"step": 715,
|
||
|
|
"valid_targets_mean": 4807.5,
|
||
|
|
"valid_targets_min": 2887
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1145510835913313,
|
||
|
|
"grad_norm": 1.0159786257283314,
|
||
|
|
"learning_rate": 3.957969880624462e-05,
|
||
|
|
"loss": 0.3737,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5812292098999023,
|
||
|
|
"step": 720,
|
||
|
|
"valid_targets_mean": 870.8,
|
||
|
|
"valid_targets_min": 239
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1222910216718267,
|
||
|
|
"grad_norm": 0.4372131955086727,
|
||
|
|
"learning_rate": 3.9563807730242684e-05,
|
||
|
|
"loss": 0.3599,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30259042978286743,
|
||
|
|
"step": 725,
|
||
|
|
"valid_targets_mean": 4695.3,
|
||
|
|
"valid_targets_min": 3169
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.130030959752322,
|
||
|
|
"grad_norm": 1.1927759066847647,
|
||
|
|
"learning_rate": 3.954762510143782e-05,
|
||
|
|
"loss": 0.3597,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5620971918106079,
|
||
|
|
"step": 730,
|
||
|
|
"valid_targets_mean": 994.2,
|
||
|
|
"valid_targets_min": 414
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1377708978328174,
|
||
|
|
"grad_norm": 0.3704712851502889,
|
||
|
|
"learning_rate": 3.9531151160994256e-05,
|
||
|
|
"loss": 0.4258,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2940506041049957,
|
||
|
|
"step": 735,
|
||
|
|
"valid_targets_mean": 4788.1,
|
||
|
|
"valid_targets_min": 2297
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1455108359133126,
|
||
|
|
"grad_norm": 0.38181505163325646,
|
||
|
|
"learning_rate": 3.951438615441755e-05,
|
||
|
|
"loss": 0.3064,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3326740264892578,
|
||
|
|
"step": 740,
|
||
|
|
"valid_targets_mean": 3206.7,
|
||
|
|
"valid_targets_min": 447
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.153250773993808,
|
||
|
|
"grad_norm": 0.48852229064634584,
|
||
|
|
"learning_rate": 3.949733033155092e-05,
|
||
|
|
"loss": 0.4566,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3101053833961487,
|
||
|
|
"step": 745,
|
||
|
|
"valid_targets_mean": 4860.4,
|
||
|
|
"valid_targets_min": 3021
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1609907120743035,
|
||
|
|
"grad_norm": 0.32193766133136864,
|
||
|
|
"learning_rate": 3.9479983946571524e-05,
|
||
|
|
"loss": 0.2961,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2978580594062805,
|
||
|
|
"step": 750,
|
||
|
|
"valid_targets_mean": 4770.2,
|
||
|
|
"valid_targets_min": 2896
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1687306501547987,
|
||
|
|
"grad_norm": 0.6029948299606571,
|
||
|
|
"learning_rate": 3.946234725798666e-05,
|
||
|
|
"loss": 0.4336,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3056912124156952,
|
||
|
|
"step": 755,
|
||
|
|
"valid_targets_mean": 4724.9,
|
||
|
|
"valid_targets_min": 3054
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1764705882352942,
|
||
|
|
"grad_norm": 0.37902448510655284,
|
||
|
|
"learning_rate": 3.944442052862992e-05,
|
||
|
|
"loss": 0.2927,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3016030192375183,
|
||
|
|
"step": 760,
|
||
|
|
"valid_targets_mean": 4654.0,
|
||
|
|
"valid_targets_min": 1994
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1842105263157894,
|
||
|
|
"grad_norm": 0.6875511279833298,
|
||
|
|
"learning_rate": 3.9426204025657295e-05,
|
||
|
|
"loss": 0.458,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.297335147857666,
|
||
|
|
"step": 765,
|
||
|
|
"valid_targets_mean": 4692.8,
|
||
|
|
"valid_targets_min": 742
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1919504643962848,
|
||
|
|
"grad_norm": 0.4012339607775376,
|
||
|
|
"learning_rate": 3.940769802054315e-05,
|
||
|
|
"loss": 0.292,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.295739084482193,
|
||
|
|
"step": 770,
|
||
|
|
"valid_targets_mean": 5052.9,
|
||
|
|
"valid_targets_min": 2772
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1996904024767803,
|
||
|
|
"grad_norm": 0.6629268941860235,
|
||
|
|
"learning_rate": 3.938890278907621e-05,
|
||
|
|
"loss": 0.4621,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30522918701171875,
|
||
|
|
"step": 775,
|
||
|
|
"valid_targets_mean": 5030.0,
|
||
|
|
"valid_targets_min": 3232
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2074303405572755,
|
||
|
|
"grad_norm": 0.4100318866856934,
|
||
|
|
"learning_rate": 3.9369818611355475e-05,
|
||
|
|
"loss": 0.2955,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28987500071525574,
|
||
|
|
"step": 780,
|
||
|
|
"valid_targets_mean": 5112.5,
|
||
|
|
"valid_targets_min": 3319
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.215170278637771,
|
||
|
|
"grad_norm": 0.46319313137857115,
|
||
|
|
"learning_rate": 3.9350445771785954e-05,
|
||
|
|
"loss": 0.4707,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3385385274887085,
|
||
|
|
"step": 785,
|
||
|
|
"valid_targets_mean": 5787.9,
|
||
|
|
"valid_targets_min": 628
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2229102167182662,
|
||
|
|
"grad_norm": 0.30199081920889614,
|
||
|
|
"learning_rate": 3.9330784559074525e-05,
|
||
|
|
"loss": 0.3403,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.34751012921333313,
|
||
|
|
"step": 790,
|
||
|
|
"valid_targets_mean": 6091.0,
|
||
|
|
"valid_targets_min": 590
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2306501547987616,
|
||
|
|
"grad_norm": 0.4629861297903929,
|
||
|
|
"learning_rate": 3.931083526622559e-05,
|
||
|
|
"loss": 0.3443,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3463875949382782,
|
||
|
|
"step": 795,
|
||
|
|
"valid_targets_mean": 5457.9,
|
||
|
|
"valid_targets_min": 537
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.238390092879257,
|
||
|
|
"grad_norm": 0.2514668027759007,
|
||
|
|
"learning_rate": 3.92905981905367e-05,
|
||
|
|
"loss": 0.3336,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.333392471075058,
|
||
|
|
"step": 800,
|
||
|
|
"valid_targets_mean": 5283.4,
|
||
|
|
"valid_targets_min": 556
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2461300309597523,
|
||
|
|
"grad_norm": 0.21309261347399663,
|
||
|
|
"learning_rate": 3.9270073633594144e-05,
|
||
|
|
"loss": 0.3419,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3198739290237427,
|
||
|
|
"step": 805,
|
||
|
|
"valid_targets_mean": 6079.6,
|
||
|
|
"valid_targets_min": 531
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2538699690402477,
|
||
|
|
"grad_norm": 0.2115270600832779,
|
||
|
|
"learning_rate": 3.924926190126845e-05,
|
||
|
|
"loss": 0.3318,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3539857864379883,
|
||
|
|
"step": 810,
|
||
|
|
"valid_targets_mean": 5938.4,
|
||
|
|
"valid_targets_min": 588
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.261609907120743,
|
||
|
|
"grad_norm": 0.20396437320544156,
|
||
|
|
"learning_rate": 3.9228163303709806e-05,
|
||
|
|
"loss": 0.3297,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.32634931802749634,
|
||
|
|
"step": 815,
|
||
|
|
"valid_targets_mean": 5991.7,
|
||
|
|
"valid_targets_min": 605
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2693498452012384,
|
||
|
|
"grad_norm": 0.204049798669846,
|
||
|
|
"learning_rate": 3.920677815534348e-05,
|
||
|
|
"loss": 0.3331,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3222689926624298,
|
||
|
|
"step": 820,
|
||
|
|
"valid_targets_mean": 5768.7,
|
||
|
|
"valid_targets_min": 589
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2770897832817338,
|
||
|
|
"grad_norm": 0.22100682244251071,
|
||
|
|
"learning_rate": 3.91851067748651e-05,
|
||
|
|
"loss": 0.3247,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3211948275566101,
|
||
|
|
"step": 825,
|
||
|
|
"valid_targets_mean": 5973.1,
|
||
|
|
"valid_targets_min": 598
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.284829721362229,
|
||
|
|
"grad_norm": 0.22772819459778706,
|
||
|
|
"learning_rate": 3.9163149485235894e-05,
|
||
|
|
"loss": 0.3272,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.32974010705947876,
|
||
|
|
"step": 830,
|
||
|
|
"valid_targets_mean": 5782.9,
|
||
|
|
"valid_targets_min": 427
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2925696594427245,
|
||
|
|
"grad_norm": 0.20650135830015126,
|
||
|
|
"learning_rate": 3.914090661367794e-05,
|
||
|
|
"loss": 0.3193,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.31986042857170105,
|
||
|
|
"step": 835,
|
||
|
|
"valid_targets_mean": 5733.2,
|
||
|
|
"valid_targets_min": 768
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3003095975232197,
|
||
|
|
"grad_norm": 0.21947617652357646,
|
||
|
|
"learning_rate": 3.9118378491669207e-05,
|
||
|
|
"loss": 0.3252,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3288787007331848,
|
||
|
|
"step": 840,
|
||
|
|
"valid_targets_mean": 5674.5,
|
||
|
|
"valid_targets_min": 597
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3080495356037152,
|
||
|
|
"grad_norm": 0.23527245926277268,
|
||
|
|
"learning_rate": 3.909556545493867e-05,
|
||
|
|
"loss": 0.3168,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.331102579832077,
|
||
|
|
"step": 845,
|
||
|
|
"valid_targets_mean": 5429.0,
|
||
|
|
"valid_targets_min": 755
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3157894736842106,
|
||
|
|
"grad_norm": 0.21280208465320555,
|
||
|
|
"learning_rate": 3.9072467843461286e-05,
|
||
|
|
"loss": 0.3201,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30824142694473267,
|
||
|
|
"step": 850,
|
||
|
|
"valid_targets_mean": 5852.8,
|
||
|
|
"valid_targets_min": 1304
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3235294117647058,
|
||
|
|
"grad_norm": 0.23383298673593644,
|
||
|
|
"learning_rate": 3.904908600145295e-05,
|
||
|
|
"loss": 0.3182,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.33378487825393677,
|
||
|
|
"step": 855,
|
||
|
|
"valid_targets_mean": 5466.6,
|
||
|
|
"valid_targets_min": 1012
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3312693498452013,
|
||
|
|
"grad_norm": 0.22216412170410357,
|
||
|
|
"learning_rate": 3.9025420277365345e-05,
|
||
|
|
"loss": 0.3173,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30657580494880676,
|
||
|
|
"step": 860,
|
||
|
|
"valid_targets_mean": 5432.9,
|
||
|
|
"valid_targets_min": 495
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3390092879256965,
|
||
|
|
"grad_norm": 0.20453560860787573,
|
||
|
|
"learning_rate": 3.900147102388073e-05,
|
||
|
|
"loss": 0.3215,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.33301278948783875,
|
||
|
|
"step": 865,
|
||
|
|
"valid_targets_mean": 6355.9,
|
||
|
|
"valid_targets_min": 629
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.346749226006192,
|
||
|
|
"grad_norm": 1.5776928817171278,
|
||
|
|
"learning_rate": 3.897723859790672e-05,
|
||
|
|
"loss": 0.4367,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6464313864707947,
|
||
|
|
"step": 870,
|
||
|
|
"valid_targets_mean": 805.4,
|
||
|
|
"valid_targets_min": 438
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3544891640866874,
|
||
|
|
"grad_norm": 0.41455119164204957,
|
||
|
|
"learning_rate": 3.895272336057096e-05,
|
||
|
|
"loss": 0.3224,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3198009133338928,
|
||
|
|
"step": 875,
|
||
|
|
"valid_targets_mean": 6133.8,
|
||
|
|
"valid_targets_min": 229
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3622291021671826,
|
||
|
|
"grad_norm": 1.431305106171426,
|
||
|
|
"learning_rate": 3.8927925677215735e-05,
|
||
|
|
"loss": 0.4051,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6523476839065552,
|
||
|
|
"step": 880,
|
||
|
|
"valid_targets_mean": 817.7,
|
||
|
|
"valid_targets_min": 373
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.369969040247678,
|
||
|
|
"grad_norm": 0.22202095087530208,
|
||
|
|
"learning_rate": 3.890284591739252e-05,
|
||
|
|
"loss": 0.3195,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30774980783462524,
|
||
|
|
"step": 885,
|
||
|
|
"valid_targets_mean": 6547.3,
|
||
|
|
"valid_targets_min": 1808
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3777089783281733,
|
||
|
|
"grad_norm": 0.27218719354449,
|
||
|
|
"learning_rate": 3.887748445485649e-05,
|
||
|
|
"loss": 0.4054,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.31560736894607544,
|
||
|
|
"step": 890,
|
||
|
|
"valid_targets_mean": 6934.5,
|
||
|
|
"valid_targets_min": 376
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3854489164086687,
|
||
|
|
"grad_norm": 0.20507818751159052,
|
||
|
|
"learning_rate": 3.885184166756092e-05,
|
||
|
|
"loss": 0.3138,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3145754337310791,
|
||
|
|
"step": 895,
|
||
|
|
"valid_targets_mean": 6416.0,
|
||
|
|
"valid_targets_min": 451
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3931888544891642,
|
||
|
|
"grad_norm": 0.276745519274244,
|
||
|
|
"learning_rate": 3.8825917937651605e-05,
|
||
|
|
"loss": 0.4522,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3206539452075958,
|
||
|
|
"step": 900,
|
||
|
|
"valid_targets_mean": 6050.4,
|
||
|
|
"valid_targets_min": 298
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4009287925696594,
|
||
|
|
"grad_norm": 0.21712786584843402,
|
||
|
|
"learning_rate": 3.8799713651461106e-05,
|
||
|
|
"loss": 0.3108,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3032996356487274,
|
||
|
|
"step": 905,
|
||
|
|
"valid_targets_mean": 5863.9,
|
||
|
|
"valid_targets_min": 212
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4086687306501549,
|
||
|
|
"grad_norm": 0.38088615829663824,
|
||
|
|
"learning_rate": 3.877322919950303e-05,
|
||
|
|
"loss": 0.4089,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3200809061527252,
|
||
|
|
"step": 910,
|
||
|
|
"valid_targets_mean": 6129.1,
|
||
|
|
"valid_targets_min": 324
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.41640866873065,
|
||
|
|
"grad_norm": 0.22320754420667796,
|
||
|
|
"learning_rate": 3.8746464976466196e-05,
|
||
|
|
"loss": 0.3076,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30708760023117065,
|
||
|
|
"step": 915,
|
||
|
|
"valid_targets_mean": 6139.4,
|
||
|
|
"valid_targets_min": 478
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4241486068111455,
|
||
|
|
"grad_norm": 0.34575592687439827,
|
||
|
|
"learning_rate": 3.8719421381208755e-05,
|
||
|
|
"loss": 0.4495,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.32162410020828247,
|
||
|
|
"step": 920,
|
||
|
|
"valid_targets_mean": 6069.3,
|
||
|
|
"valid_targets_min": 230
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.431888544891641,
|
||
|
|
"grad_norm": 0.22823315786752674,
|
||
|
|
"learning_rate": 3.869209881675227e-05,
|
||
|
|
"loss": 0.3047,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3064790368080139,
|
||
|
|
"step": 925,
|
||
|
|
"valid_targets_mean": 6344.3,
|
||
|
|
"valid_targets_min": 600
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4396284829721362,
|
||
|
|
"grad_norm": 0.3713080551571927,
|
||
|
|
"learning_rate": 3.866449769027564e-05,
|
||
|
|
"loss": 0.4448,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3074290454387665,
|
||
|
|
"step": 930,
|
||
|
|
"valid_targets_mean": 6771.4,
|
||
|
|
"valid_targets_min": 486
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4473684210526316,
|
||
|
|
"grad_norm": 0.2704341788510641,
|
||
|
|
"learning_rate": 3.8636618413109134e-05,
|
||
|
|
"loss": 0.3048,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.304415762424469,
|
||
|
|
"step": 935,
|
||
|
|
"valid_targets_mean": 6569.9,
|
||
|
|
"valid_targets_min": 353
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4551083591331269,
|
||
|
|
"grad_norm": 1.0257560788628433,
|
||
|
|
"learning_rate": 3.8608461400728165e-05,
|
||
|
|
"loss": 0.5503,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5682530999183655,
|
||
|
|
"step": 940,
|
||
|
|
"valid_targets_mean": 2007.4,
|
||
|
|
"valid_targets_min": 246
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4628482972136223,
|
||
|
|
"grad_norm": 0.4387060803991135,
|
||
|
|
"learning_rate": 3.8580027072747154e-05,
|
||
|
|
"loss": 0.5419,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.55305016040802,
|
||
|
|
"step": 945,
|
||
|
|
"valid_targets_mean": 3351.7,
|
||
|
|
"valid_targets_min": 267
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4705882352941178,
|
||
|
|
"grad_norm": 0.35202685819782087,
|
||
|
|
"learning_rate": 3.855131585291327e-05,
|
||
|
|
"loss": 0.5337,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5412226319313049,
|
||
|
|
"step": 950,
|
||
|
|
"valid_targets_mean": 3195.7,
|
||
|
|
"valid_targets_min": 296
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.478328173374613,
|
||
|
|
"grad_norm": 0.316033059237561,
|
||
|
|
"learning_rate": 3.852232816910011e-05,
|
||
|
|
"loss": 0.5299,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5056208968162537,
|
||
|
|
"step": 955,
|
||
|
|
"valid_targets_mean": 3282.1,
|
||
|
|
"valid_targets_min": 296
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4860681114551084,
|
||
|
|
"grad_norm": 0.28464766873460606,
|
||
|
|
"learning_rate": 3.84930644533013e-05,
|
||
|
|
"loss": 0.5141,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48853540420532227,
|
||
|
|
"step": 960,
|
||
|
|
"valid_targets_mean": 3273.5,
|
||
|
|
"valid_targets_min": 426
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4938080495356036,
|
||
|
|
"grad_norm": 0.2537715497805479,
|
||
|
|
"learning_rate": 3.846352514162408e-05,
|
||
|
|
"loss": 0.519,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4617595374584198,
|
||
|
|
"step": 965,
|
||
|
|
"valid_targets_mean": 3904.3,
|
||
|
|
"valid_targets_min": 286
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.501547987616099,
|
||
|
|
"grad_norm": 0.28277744582484216,
|
||
|
|
"learning_rate": 3.843371067428283e-05,
|
||
|
|
"loss": 0.5259,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5558550953865051,
|
||
|
|
"step": 970,
|
||
|
|
"valid_targets_mean": 3417.9,
|
||
|
|
"valid_targets_min": 431
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5092879256965945,
|
||
|
|
"grad_norm": 0.319771845595299,
|
||
|
|
"learning_rate": 3.840362149559246e-05,
|
||
|
|
"loss": 0.5182,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5502806901931763,
|
||
|
|
"step": 975,
|
||
|
|
"valid_targets_mean": 2151.9,
|
||
|
|
"valid_targets_min": 306
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5170278637770898,
|
||
|
|
"grad_norm": 0.2807493202880763,
|
||
|
|
"learning_rate": 3.83732580539618e-05,
|
||
|
|
"loss": 0.5101,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.491119384765625,
|
||
|
|
"step": 980,
|
||
|
|
"valid_targets_mean": 2757.0,
|
||
|
|
"valid_targets_min": 366
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.524767801857585,
|
||
|
|
"grad_norm": 0.30373133574076155,
|
||
|
|
"learning_rate": 3.8342620801886954e-05,
|
||
|
|
"loss": 0.5192,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5305666327476501,
|
||
|
|
"step": 985,
|
||
|
|
"valid_targets_mean": 2699.5,
|
||
|
|
"valid_targets_min": 297
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5325077399380804,
|
||
|
|
"grad_norm": 0.2489893446264647,
|
||
|
|
"learning_rate": 3.831171019594451e-05,
|
||
|
|
"loss": 0.5079,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4916561543941498,
|
||
|
|
"step": 990,
|
||
|
|
"valid_targets_mean": 3419.5,
|
||
|
|
"valid_targets_min": 281
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5402476780185759,
|
||
|
|
"grad_norm": 0.2776483187268575,
|
||
|
|
"learning_rate": 3.8280526696784764e-05,
|
||
|
|
"loss": 0.5081,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5029506683349609,
|
||
|
|
"step": 995,
|
||
|
|
"valid_targets_mean": 3438.5,
|
||
|
|
"valid_targets_min": 301
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5479876160990713,
|
||
|
|
"grad_norm": 0.2510226682854113,
|
||
|
|
"learning_rate": 3.8249070769124846e-05,
|
||
|
|
"loss": 0.5092,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.486871600151062,
|
||
|
|
"step": 1000,
|
||
|
|
"valid_targets_mean": 4330.6,
|
||
|
|
"valid_targets_min": 247
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5557275541795665,
|
||
|
|
"grad_norm": 0.31141018821112487,
|
||
|
|
"learning_rate": 3.8217342881741795e-05,
|
||
|
|
"loss": 0.5092,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5455483198165894,
|
||
|
|
"step": 1005,
|
||
|
|
"valid_targets_mean": 2712.3,
|
||
|
|
"valid_targets_min": 262
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5634674922600618,
|
||
|
|
"grad_norm": 0.5861860309851681,
|
||
|
|
"learning_rate": 3.818534350746557e-05,
|
||
|
|
"loss": 0.533,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.668022632598877,
|
||
|
|
"step": 1010,
|
||
|
|
"valid_targets_mean": 3619.0,
|
||
|
|
"valid_targets_min": 1099
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5712074303405572,
|
||
|
|
"grad_norm": 0.4009985038183046,
|
||
|
|
"learning_rate": 3.8153073123172026e-05,
|
||
|
|
"loss": 0.639,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6261304616928101,
|
||
|
|
"step": 1015,
|
||
|
|
"valid_targets_mean": 3834.9,
|
||
|
|
"valid_targets_min": 1026
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5789473684210527,
|
||
|
|
"grad_norm": 0.33257332827465463,
|
||
|
|
"learning_rate": 3.812053220977577e-05,
|
||
|
|
"loss": 0.639,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6216365098953247,
|
||
|
|
"step": 1020,
|
||
|
|
"valid_targets_mean": 3756.7,
|
||
|
|
"valid_targets_min": 1083
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.586687306501548,
|
||
|
|
"grad_norm": 0.29326553190726723,
|
||
|
|
"learning_rate": 3.808772125222301e-05,
|
||
|
|
"loss": 0.6418,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6320463418960571,
|
||
|
|
"step": 1025,
|
||
|
|
"valid_targets_mean": 3851.6,
|
||
|
|
"valid_targets_min": 1063
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5944272445820433,
|
||
|
|
"grad_norm": 0.29083135754041894,
|
||
|
|
"learning_rate": 3.805464073948433e-05,
|
||
|
|
"loss": 0.6257,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6506540179252625,
|
||
|
|
"step": 1030,
|
||
|
|
"valid_targets_mean": 3610.4,
|
||
|
|
"valid_targets_min": 1092
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6021671826625385,
|
||
|
|
"grad_norm": 0.25760367030021886,
|
||
|
|
"learning_rate": 3.8021291164547436e-05,
|
||
|
|
"loss": 0.6222,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5811383724212646,
|
||
|
|
"step": 1035,
|
||
|
|
"valid_targets_mean": 4324.1,
|
||
|
|
"valid_targets_min": 791
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.609907120743034,
|
||
|
|
"grad_norm": 0.26663244411969533,
|
||
|
|
"learning_rate": 3.7987673024409736e-05,
|
||
|
|
"loss": 0.6369,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6155533790588379,
|
||
|
|
"step": 1040,
|
||
|
|
"valid_targets_mean": 3782.9,
|
||
|
|
"valid_targets_min": 1167
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6176470588235294,
|
||
|
|
"grad_norm": 0.28885115085965224,
|
||
|
|
"learning_rate": 3.7953786820070985e-05,
|
||
|
|
"loss": 0.6179,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5951216816902161,
|
||
|
|
"step": 1045,
|
||
|
|
"valid_targets_mean": 4237.8,
|
||
|
|
"valid_targets_min": 636
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6253869969040249,
|
||
|
|
"grad_norm": 0.323857777141142,
|
||
|
|
"learning_rate": 3.791963305652583e-05,
|
||
|
|
"loss": 0.621,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6057307720184326,
|
||
|
|
"step": 1050,
|
||
|
|
"valid_targets_mean": 4379.6,
|
||
|
|
"valid_targets_min": 417
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.63312693498452,
|
||
|
|
"grad_norm": 0.2359587775817818,
|
||
|
|
"learning_rate": 3.7885212242756235e-05,
|
||
|
|
"loss": 0.605,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5471640825271606,
|
||
|
|
"step": 1055,
|
||
|
|
"valid_targets_mean": 4701.0,
|
||
|
|
"valid_targets_min": 826
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6408668730650153,
|
||
|
|
"grad_norm": 0.2700917467415719,
|
||
|
|
"learning_rate": 3.7850524891723935e-05,
|
||
|
|
"loss": 0.6286,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5909395217895508,
|
||
|
|
"step": 1060,
|
||
|
|
"valid_targets_mean": 4621.8,
|
||
|
|
"valid_targets_min": 983
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6486068111455108,
|
||
|
|
"grad_norm": 0.2928847055917006,
|
||
|
|
"learning_rate": 3.7815571520362783e-05,
|
||
|
|
"loss": 0.6339,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6801971197128296,
|
||
|
|
"step": 1065,
|
||
|
|
"valid_targets_mean": 3181.1,
|
||
|
|
"valid_targets_min": 492
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6563467492260062,
|
||
|
|
"grad_norm": 0.2798984740901887,
|
||
|
|
"learning_rate": 3.778035264957104e-05,
|
||
|
|
"loss": 0.6323,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6776220798492432,
|
||
|
|
"step": 1070,
|
||
|
|
"valid_targets_mean": 4021.6,
|
||
|
|
"valid_targets_min": 665
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6640866873065017,
|
||
|
|
"grad_norm": 0.30219051113140105,
|
||
|
|
"learning_rate": 3.77448688042036e-05,
|
||
|
|
"loss": 0.6169,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6817198991775513,
|
||
|
|
"step": 1075,
|
||
|
|
"valid_targets_mean": 3290.7,
|
||
|
|
"valid_targets_min": 280
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.671826625386997,
|
||
|
|
"grad_norm": 0.26303876045049596,
|
||
|
|
"learning_rate": 3.7709120513064196e-05,
|
||
|
|
"loss": 0.6262,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6559048891067505,
|
||
|
|
"step": 1080,
|
||
|
|
"valid_targets_mean": 4297.4,
|
||
|
|
"valid_targets_min": 1038
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6795665634674921,
|
||
|
|
"grad_norm": 0.5036205376425451,
|
||
|
|
"learning_rate": 3.767310830889753e-05,
|
||
|
|
"loss": 0.6021,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5323740839958191,
|
||
|
|
"step": 1085,
|
||
|
|
"valid_targets_mean": 3676.8,
|
||
|
|
"valid_targets_min": 1023
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6873065015479876,
|
||
|
|
"grad_norm": 0.3904934068144397,
|
||
|
|
"learning_rate": 3.7636832728381276e-05,
|
||
|
|
"loss": 0.5348,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5480472445487976,
|
||
|
|
"step": 1090,
|
||
|
|
"valid_targets_mean": 3877.6,
|
||
|
|
"valid_targets_min": 891
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.695046439628483,
|
||
|
|
"grad_norm": 0.3028491652210317,
|
||
|
|
"learning_rate": 3.760029431211812e-05,
|
||
|
|
"loss": 0.5197,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5557341575622559,
|
||
|
|
"step": 1095,
|
||
|
|
"valid_targets_mean": 3730.8,
|
||
|
|
"valid_targets_min": 921
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7027863777089784,
|
||
|
|
"grad_norm": 0.33031589942205014,
|
||
|
|
"learning_rate": 3.756349360462772e-05,
|
||
|
|
"loss": 0.5196,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5021288394927979,
|
||
|
|
"step": 1100,
|
||
|
|
"valid_targets_mean": 3764.0,
|
||
|
|
"valid_targets_min": 771
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7105263157894737,
|
||
|
|
"grad_norm": 0.2771094090671581,
|
||
|
|
"learning_rate": 3.752643115433856e-05,
|
||
|
|
"loss": 0.52,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5006573796272278,
|
||
|
|
"step": 1105,
|
||
|
|
"valid_targets_mean": 3619.3,
|
||
|
|
"valid_targets_min": 981
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.718266253869969,
|
||
|
|
"grad_norm": 0.2932821415250669,
|
||
|
|
"learning_rate": 3.7489107513579785e-05,
|
||
|
|
"loss": 0.5126,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5139281749725342,
|
||
|
|
"step": 1110,
|
||
|
|
"valid_targets_mean": 3383.6,
|
||
|
|
"valid_targets_min": 747
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7260061919504643,
|
||
|
|
"grad_norm": 0.29429266053364417,
|
||
|
|
"learning_rate": 3.745152323857297e-05,
|
||
|
|
"loss": 0.5142,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5317479968070984,
|
||
|
|
"step": 1115,
|
||
|
|
"valid_targets_mean": 3617.2,
|
||
|
|
"valid_targets_min": 877
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7337461300309598,
|
||
|
|
"grad_norm": 0.272377877948682,
|
||
|
|
"learning_rate": 3.741367888942385e-05,
|
||
|
|
"loss": 0.5116,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4985455572605133,
|
||
|
|
"step": 1120,
|
||
|
|
"valid_targets_mean": 3628.0,
|
||
|
|
"valid_targets_min": 1027
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7414860681114552,
|
||
|
|
"grad_norm": 0.2727651244150089,
|
||
|
|
"learning_rate": 3.737557503011395e-05,
|
||
|
|
"loss": 0.5125,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5064275860786438,
|
||
|
|
"step": 1125,
|
||
|
|
"valid_targets_mean": 3666.4,
|
||
|
|
"valid_targets_min": 1044
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7492260061919505,
|
||
|
|
"grad_norm": 0.3037283056381341,
|
||
|
|
"learning_rate": 3.733721222849218e-05,
|
||
|
|
"loss": 0.5,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4965721368789673,
|
||
|
|
"step": 1130,
|
||
|
|
"valid_targets_mean": 3281.1,
|
||
|
|
"valid_targets_min": 851
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7569659442724457,
|
||
|
|
"grad_norm": 0.27151606619735624,
|
||
|
|
"learning_rate": 3.729859105626638e-05,
|
||
|
|
"loss": 0.5105,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5014647245407104,
|
||
|
|
"step": 1135,
|
||
|
|
"valid_targets_mean": 3577.5,
|
||
|
|
"valid_targets_min": 1157
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7647058823529411,
|
||
|
|
"grad_norm": 0.2799259205452004,
|
||
|
|
"learning_rate": 3.725971208899481e-05,
|
||
|
|
"loss": 0.5131,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.476270854473114,
|
||
|
|
"step": 1140,
|
||
|
|
"valid_targets_mean": 3314.5,
|
||
|
|
"valid_targets_min": 706
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7724458204334366,
|
||
|
|
"grad_norm": 0.2700644491689463,
|
||
|
|
"learning_rate": 3.7220575906077546e-05,
|
||
|
|
"loss": 0.5147,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5040243864059448,
|
||
|
|
"step": 1145,
|
||
|
|
"valid_targets_mean": 3562.3,
|
||
|
|
"valid_targets_min": 1014
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.780185758513932,
|
||
|
|
"grad_norm": 0.29196613518133757,
|
||
|
|
"learning_rate": 3.718118309074789e-05,
|
||
|
|
"loss": 0.5101,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.507203221321106,
|
||
|
|
"step": 1150,
|
||
|
|
"valid_targets_mean": 3536.9,
|
||
|
|
"valid_targets_min": 792
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7879256965944272,
|
||
|
|
"grad_norm": 0.2733785186987948,
|
||
|
|
"learning_rate": 3.714153423006361e-05,
|
||
|
|
"loss": 0.5033,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5127211213111877,
|
||
|
|
"step": 1155,
|
||
|
|
"valid_targets_mean": 3570.4,
|
||
|
|
"valid_targets_min": 1059
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7956656346749225,
|
||
|
|
"grad_norm": 0.27888755380256725,
|
||
|
|
"learning_rate": 3.710162991489828e-05,
|
||
|
|
"loss": 0.5194,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5064829587936401,
|
||
|
|
"step": 1160,
|
||
|
|
"valid_targets_mean": 3647.4,
|
||
|
|
"valid_targets_min": 784
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.803405572755418,
|
||
|
|
"grad_norm": 0.2801140370328727,
|
||
|
|
"learning_rate": 3.7061470739932376e-05,
|
||
|
|
"loss": 0.5041,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5027486085891724,
|
||
|
|
"step": 1165,
|
||
|
|
"valid_targets_mean": 3439.6,
|
||
|
|
"valid_targets_min": 1012
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8111455108359134,
|
||
|
|
"grad_norm": 0.25781419832614627,
|
||
|
|
"learning_rate": 3.7021057303644485e-05,
|
||
|
|
"loss": 0.5014,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.45872068405151367,
|
||
|
|
"step": 1170,
|
||
|
|
"valid_targets_mean": 3415.4,
|
||
|
|
"valid_targets_min": 555
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8188854489164088,
|
||
|
|
"grad_norm": 0.30686560628574633,
|
||
|
|
"learning_rate": 3.698039020830238e-05,
|
||
|
|
"loss": 0.5923,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4970795512199402,
|
||
|
|
"step": 1175,
|
||
|
|
"valid_targets_mean": 3413.9,
|
||
|
|
"valid_targets_min": 973
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.826625386996904,
|
||
|
|
"grad_norm": 0.2607802633996197,
|
||
|
|
"learning_rate": 3.693947005995401e-05,
|
||
|
|
"loss": 0.507,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4885653555393219,
|
||
|
|
"step": 1180,
|
||
|
|
"valid_targets_mean": 3740.7,
|
||
|
|
"valid_targets_min": 849
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8343653250773992,
|
||
|
|
"grad_norm": 0.31517914947777337,
|
||
|
|
"learning_rate": 3.689829746841848e-05,
|
||
|
|
"loss": 0.6057,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.530507504940033,
|
||
|
|
"step": 1185,
|
||
|
|
"valid_targets_mean": 3384.0,
|
||
|
|
"valid_targets_min": 870
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8421052631578947,
|
||
|
|
"grad_norm": 0.30778705038262993,
|
||
|
|
"learning_rate": 3.6856873047277e-05,
|
||
|
|
"loss": 0.4989,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.49664801359176636,
|
||
|
|
"step": 1190,
|
||
|
|
"valid_targets_mean": 3341.3,
|
||
|
|
"valid_targets_min": 1113
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8498452012383901,
|
||
|
|
"grad_norm": 0.3482703179318802,
|
||
|
|
"learning_rate": 3.681519741386368e-05,
|
||
|
|
"loss": 0.5529,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5027183294296265,
|
||
|
|
"step": 1195,
|
||
|
|
"valid_targets_mean": 3684.2,
|
||
|
|
"valid_targets_min": 1026
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8575851393188856,
|
||
|
|
"grad_norm": 0.28341496023003704,
|
||
|
|
"learning_rate": 3.6773271189256376e-05,
|
||
|
|
"loss": 0.4994,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48868709802627563,
|
||
|
|
"step": 1200,
|
||
|
|
"valid_targets_mean": 3157.4,
|
||
|
|
"valid_targets_min": 525
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8653250773993808,
|
||
|
|
"grad_norm": 0.3459453566898097,
|
||
|
|
"learning_rate": 3.673109499826743e-05,
|
||
|
|
"loss": 0.5524,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5330445766448975,
|
||
|
|
"step": 1205,
|
||
|
|
"valid_targets_mean": 3240.1,
|
||
|
|
"valid_targets_min": 984
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.873065015479876,
|
||
|
|
"grad_norm": 0.31995691659479647,
|
||
|
|
"learning_rate": 3.668866946943432e-05,
|
||
|
|
"loss": 0.4978,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.49783432483673096,
|
||
|
|
"step": 1210,
|
||
|
|
"valid_targets_mean": 3223.1,
|
||
|
|
"valid_targets_min": 328
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8808049535603715,
|
||
|
|
"grad_norm": 0.3540740257592273,
|
||
|
|
"learning_rate": 3.664599523501035e-05,
|
||
|
|
"loss": 0.6005,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5141074657440186,
|
||
|
|
"step": 1215,
|
||
|
|
"valid_targets_mean": 3842.8,
|
||
|
|
"valid_targets_min": 811
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.888544891640867,
|
||
|
|
"grad_norm": 0.3342693153663899,
|
||
|
|
"learning_rate": 3.660307293095518e-05,
|
||
|
|
"loss": 0.4986,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.50550377368927,
|
||
|
|
"step": 1220,
|
||
|
|
"valid_targets_mean": 3401.7,
|
||
|
|
"valid_targets_min": 358
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8962848297213624,
|
||
|
|
"grad_norm": 0.32620690960822574,
|
||
|
|
"learning_rate": 3.6559903196925386e-05,
|
||
|
|
"loss": 0.5838,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5673384070396423,
|
||
|
|
"step": 1225,
|
||
|
|
"valid_targets_mean": 3138.4,
|
||
|
|
"valid_targets_min": 426
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9040247678018576,
|
||
|
|
"grad_norm": 0.27239871659976117,
|
||
|
|
"learning_rate": 3.6516486676264895e-05,
|
||
|
|
"loss": 0.5065,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5245934724807739,
|
||
|
|
"step": 1230,
|
||
|
|
"valid_targets_mean": 3528.3,
|
||
|
|
"valid_targets_min": 285
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9117647058823528,
|
||
|
|
"grad_norm": 0.35982319998862494,
|
||
|
|
"learning_rate": 3.6472824015995404e-05,
|
||
|
|
"loss": 0.5369,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6063302755355835,
|
||
|
|
"step": 1235,
|
||
|
|
"valid_targets_mean": 2719.7,
|
||
|
|
"valid_targets_min": 533
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9195046439628483,
|
||
|
|
"grad_norm": 0.2826114999814448,
|
||
|
|
"learning_rate": 3.642891586680676e-05,
|
||
|
|
"loss": 0.496,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5089360475540161,
|
||
|
|
"step": 1240,
|
||
|
|
"valid_targets_mean": 3570.4,
|
||
|
|
"valid_targets_min": 564
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9272445820433437,
|
||
|
|
"grad_norm": 1.3430187245686194,
|
||
|
|
"learning_rate": 3.638476288304726e-05,
|
||
|
|
"loss": 0.5239,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5505675077438354,
|
||
|
|
"step": 1245,
|
||
|
|
"valid_targets_mean": 2341.9,
|
||
|
|
"valid_targets_min": 417
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9349845201238391,
|
||
|
|
"grad_norm": 0.5111336826720022,
|
||
|
|
"learning_rate": 3.634036572271385e-05,
|
||
|
|
"loss": 0.3167,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29680368304252625,
|
||
|
|
"step": 1250,
|
||
|
|
"valid_targets_mean": 4556.8,
|
||
|
|
"valid_targets_min": 1546
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9427244582043344,
|
||
|
|
"grad_norm": 0.5734316816453808,
|
||
|
|
"learning_rate": 3.62957250474424e-05,
|
||
|
|
"loss": 0.2675,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28456592559814453,
|
||
|
|
"step": 1255,
|
||
|
|
"valid_targets_mean": 4626.2,
|
||
|
|
"valid_targets_min": 925
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9504643962848296,
|
||
|
|
"grad_norm": 0.3279083661231235,
|
||
|
|
"learning_rate": 3.6250841522497756e-05,
|
||
|
|
"loss": 0.2473,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24205255508422852,
|
||
|
|
"step": 1260,
|
||
|
|
"valid_targets_mean": 4938.1,
|
||
|
|
"valid_targets_min": 1451
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.958204334365325,
|
||
|
|
"grad_norm": 0.38856871328008913,
|
||
|
|
"learning_rate": 3.620571581676392e-05,
|
||
|
|
"loss": 0.2449,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2540673017501831,
|
||
|
|
"step": 1265,
|
||
|
|
"valid_targets_mean": 4161.5,
|
||
|
|
"valid_targets_min": 1249
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9659442724458205,
|
||
|
|
"grad_norm": 0.4964409706125973,
|
||
|
|
"learning_rate": 3.6160348602733986e-05,
|
||
|
|
"loss": 0.258,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2558837831020355,
|
||
|
|
"step": 1270,
|
||
|
|
"valid_targets_mean": 4068.4,
|
||
|
|
"valid_targets_min": 1239
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.973684210526316,
|
||
|
|
"grad_norm": 0.3365178870057779,
|
||
|
|
"learning_rate": 3.61147405565002e-05,
|
||
|
|
"loss": 0.2526,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24560043215751648,
|
||
|
|
"step": 1275,
|
||
|
|
"valid_targets_mean": 4988.2,
|
||
|
|
"valid_targets_min": 1593
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9814241486068112,
|
||
|
|
"grad_norm": 0.4980914239459907,
|
||
|
|
"learning_rate": 3.6068892357743824e-05,
|
||
|
|
"loss": 0.2528,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24032273888587952,
|
||
|
|
"step": 1280,
|
||
|
|
"valid_targets_mean": 2636.5,
|
||
|
|
"valid_targets_min": 274
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9891640866873064,
|
||
|
|
"grad_norm": 0.40866270249525627,
|
||
|
|
"learning_rate": 3.602280468972504e-05,
|
||
|
|
"loss": 0.2507,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25480788946151733,
|
||
|
|
"step": 1285,
|
||
|
|
"valid_targets_mean": 2712.8,
|
||
|
|
"valid_targets_min": 982
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9969040247678018,
|
||
|
|
"grad_norm": 0.3061496368374394,
|
||
|
|
"learning_rate": 3.597647823927277e-05,
|
||
|
|
"loss": 0.2347,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23438185453414917,
|
||
|
|
"step": 1290,
|
||
|
|
"valid_targets_mean": 3597.4,
|
||
|
|
"valid_targets_min": 1869
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0046439628482973,
|
||
|
|
"grad_norm": 0.48887551388962647,
|
||
|
|
"learning_rate": 3.592991369677441e-05,
|
||
|
|
"loss": 0.2752,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2809794545173645,
|
||
|
|
"step": 1295,
|
||
|
|
"valid_targets_mean": 3714.5,
|
||
|
|
"valid_targets_min": 313
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0123839009287927,
|
||
|
|
"grad_norm": 0.3973862995199013,
|
||
|
|
"learning_rate": 3.588311175616558e-05,
|
||
|
|
"loss": 0.2751,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27006852626800537,
|
||
|
|
"step": 1300,
|
||
|
|
"valid_targets_mean": 3769.2,
|
||
|
|
"valid_targets_min": 428
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0201238390092877,
|
||
|
|
"grad_norm": 0.3258988746910927,
|
||
|
|
"learning_rate": 3.583607311491975e-05,
|
||
|
|
"loss": 0.2662,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26744651794433594,
|
||
|
|
"step": 1305,
|
||
|
|
"valid_targets_mean": 3845.8,
|
||
|
|
"valid_targets_min": 444
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.027863777089783,
|
||
|
|
"grad_norm": 0.28182483680323517,
|
||
|
|
"learning_rate": 3.578879847403787e-05,
|
||
|
|
"loss": 0.2647,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26419389247894287,
|
||
|
|
"step": 1310,
|
||
|
|
"valid_targets_mean": 3735.0,
|
||
|
|
"valid_targets_min": 259
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0356037151702786,
|
||
|
|
"grad_norm": 0.2648680231069796,
|
||
|
|
"learning_rate": 3.57412885380379e-05,
|
||
|
|
"loss": 0.2571,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24876728653907776,
|
||
|
|
"step": 1315,
|
||
|
|
"valid_targets_mean": 3725.4,
|
||
|
|
"valid_targets_min": 386
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.043343653250774,
|
||
|
|
"grad_norm": 0.28303745142040815,
|
||
|
|
"learning_rate": 3.569354401494433e-05,
|
||
|
|
"loss": 0.2554,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2525283396244049,
|
||
|
|
"step": 1320,
|
||
|
|
"valid_targets_mean": 3718.9,
|
||
|
|
"valid_targets_min": 463
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0510835913312695,
|
||
|
|
"grad_norm": 0.23647880766650375,
|
||
|
|
"learning_rate": 3.564556561627764e-05,
|
||
|
|
"loss": 0.2515,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2512373924255371,
|
||
|
|
"step": 1325,
|
||
|
|
"valid_targets_mean": 3712.7,
|
||
|
|
"valid_targets_min": 434
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0588235294117645,
|
||
|
|
"grad_norm": 0.23555452307251556,
|
||
|
|
"learning_rate": 3.5597354057043644e-05,
|
||
|
|
"loss": 0.2506,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25014546513557434,
|
||
|
|
"step": 1330,
|
||
|
|
"valid_targets_mean": 3745.0,
|
||
|
|
"valid_targets_min": 433
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.06656346749226,
|
||
|
|
"grad_norm": 0.23287509332785636,
|
||
|
|
"learning_rate": 3.5548910055722884e-05,
|
||
|
|
"loss": 0.2462,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23879434168338776,
|
||
|
|
"step": 1335,
|
||
|
|
"valid_targets_mean": 3624.5,
|
||
|
|
"valid_targets_min": 397
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0743034055727554,
|
||
|
|
"grad_norm": 0.2618478645057261,
|
||
|
|
"learning_rate": 3.5500234334259925e-05,
|
||
|
|
"loss": 0.2514,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2599344849586487,
|
||
|
|
"step": 1340,
|
||
|
|
"valid_targets_mean": 3754.9,
|
||
|
|
"valid_targets_min": 269
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.082043343653251,
|
||
|
|
"grad_norm": 0.24365346227436066,
|
||
|
|
"learning_rate": 3.545132761805256e-05,
|
||
|
|
"loss": 0.2477,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24797475337982178,
|
||
|
|
"step": 1345,
|
||
|
|
"valid_targets_mean": 3859.0,
|
||
|
|
"valid_targets_min": 463
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0897832817337463,
|
||
|
|
"grad_norm": 0.34401890593680245,
|
||
|
|
"learning_rate": 3.540219063594101e-05,
|
||
|
|
"loss": 0.2756,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2905760407447815,
|
||
|
|
"step": 1350,
|
||
|
|
"valid_targets_mean": 4864.0,
|
||
|
|
"valid_targets_min": 2389
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0975232198142413,
|
||
|
|
"grad_norm": 0.31575301820877294,
|
||
|
|
"learning_rate": 3.53528241201971e-05,
|
||
|
|
"loss": 0.2871,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2831037640571594,
|
||
|
|
"step": 1355,
|
||
|
|
"valid_targets_mean": 4532.4,
|
||
|
|
"valid_targets_min": 468
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1052631578947367,
|
||
|
|
"grad_norm": 0.48014293305490685,
|
||
|
|
"learning_rate": 3.530322880651329e-05,
|
||
|
|
"loss": 0.4246,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2933792471885681,
|
||
|
|
"step": 1360,
|
||
|
|
"valid_targets_mean": 5055.2,
|
||
|
|
"valid_targets_min": 2995
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.113003095975232,
|
||
|
|
"grad_norm": 0.35510711588647426,
|
||
|
|
"learning_rate": 3.5253405433991754e-05,
|
||
|
|
"loss": 0.2926,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.32099688053131104,
|
||
|
|
"step": 1365,
|
||
|
|
"valid_targets_mean": 2780.0,
|
||
|
|
"valid_targets_min": 448
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1207430340557276,
|
||
|
|
"grad_norm": 0.3170903834561464,
|
||
|
|
"learning_rate": 3.520335474513334e-05,
|
||
|
|
"loss": 0.3813,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2778979539871216,
|
||
|
|
"step": 1370,
|
||
|
|
"valid_targets_mean": 4765.8,
|
||
|
|
"valid_targets_min": 2211
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.128482972136223,
|
||
|
|
"grad_norm": 0.3646764652856628,
|
||
|
|
"learning_rate": 3.515307748582652e-05,
|
||
|
|
"loss": 0.2847,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3122536838054657,
|
||
|
|
"step": 1375,
|
||
|
|
"valid_targets_mean": 3243.5,
|
||
|
|
"valid_targets_min": 412
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.136222910216718,
|
||
|
|
"grad_norm": 0.5272987838751837,
|
||
|
|
"learning_rate": 3.510257440533626e-05,
|
||
|
|
"loss": 0.4405,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2815614938735962,
|
||
|
|
"step": 1380,
|
||
|
|
"valid_targets_mean": 4668.5,
|
||
|
|
"valid_targets_min": 2456
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1439628482972135,
|
||
|
|
"grad_norm": 0.3095689708837096,
|
||
|
|
"learning_rate": 3.5051846256292886e-05,
|
||
|
|
"loss": 0.2779,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27151453495025635,
|
||
|
|
"step": 1385,
|
||
|
|
"valid_targets_mean": 4707.3,
|
||
|
|
"valid_targets_min": 2483
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.151702786377709,
|
||
|
|
"grad_norm": 0.6644423354296274,
|
||
|
|
"learning_rate": 3.5000893794680815e-05,
|
||
|
|
"loss": 0.426,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2865567207336426,
|
||
|
|
"step": 1390,
|
||
|
|
"valid_targets_mean": 4803.9,
|
||
|
|
"valid_targets_min": 3123
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1594427244582044,
|
||
|
|
"grad_norm": 0.3687511141636036,
|
||
|
|
"learning_rate": 3.4949717779827336e-05,
|
||
|
|
"loss": 0.2805,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27756816148757935,
|
||
|
|
"step": 1395,
|
||
|
|
"valid_targets_mean": 4720.2,
|
||
|
|
"valid_targets_min": 2652
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1671826625387,
|
||
|
|
"grad_norm": 0.7225507721638017,
|
||
|
|
"learning_rate": 3.489831897439127e-05,
|
||
|
|
"loss": 0.4041,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29848140478134155,
|
||
|
|
"step": 1400,
|
||
|
|
"valid_targets_mean": 3505.7,
|
||
|
|
"valid_targets_min": 429
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.174922600619195,
|
||
|
|
"grad_norm": 0.38897287955846077,
|
||
|
|
"learning_rate": 3.484669814435161e-05,
|
||
|
|
"loss": 0.278,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27883386611938477,
|
||
|
|
"step": 1405,
|
||
|
|
"valid_targets_mean": 4646.0,
|
||
|
|
"valid_targets_min": 2402
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1826625386996903,
|
||
|
|
"grad_norm": 0.5904052821062835,
|
||
|
|
"learning_rate": 3.4794856058996123e-05,
|
||
|
|
"loss": 0.4266,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5118474364280701,
|
||
|
|
"step": 1410,
|
||
|
|
"valid_targets_mean": 973.6,
|
||
|
|
"valid_targets_min": 281
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1904024767801857,
|
||
|
|
"grad_norm": 0.29658573203399535,
|
||
|
|
"learning_rate": 3.474279349090984e-05,
|
||
|
|
"loss": 0.2742,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2676442861557007,
|
||
|
|
"step": 1415,
|
||
|
|
"valid_targets_mean": 4744.8,
|
||
|
|
"valid_targets_min": 2671
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.198142414860681,
|
||
|
|
"grad_norm": 0.47920469483113043,
|
||
|
|
"learning_rate": 3.469051121596359e-05,
|
||
|
|
"loss": 0.4281,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3080791234970093,
|
||
|
|
"step": 1420,
|
||
|
|
"valid_targets_mean": 3931.4,
|
||
|
|
"valid_targets_min": 474
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2058823529411766,
|
||
|
|
"grad_norm": 0.2983318679863187,
|
||
|
|
"learning_rate": 3.463801001330242e-05,
|
||
|
|
"loss": 0.2819,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27747613191604614,
|
||
|
|
"step": 1425,
|
||
|
|
"valid_targets_mean": 5010.2,
|
||
|
|
"valid_targets_min": 2642
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2136222910216716,
|
||
|
|
"grad_norm": 0.9084353126798798,
|
||
|
|
"learning_rate": 3.4585290665333973e-05,
|
||
|
|
"loss": 0.4325,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5517691373825073,
|
||
|
|
"step": 1430,
|
||
|
|
"valid_targets_mean": 812.3,
|
||
|
|
"valid_targets_min": 129
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.221362229102167,
|
||
|
|
"grad_norm": 0.2779292183039381,
|
||
|
|
"learning_rate": 3.453235395771684e-05,
|
||
|
|
"loss": 0.3122,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3111690282821655,
|
||
|
|
"step": 1435,
|
||
|
|
"valid_targets_mean": 5991.4,
|
||
|
|
"valid_targets_min": 913
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2291021671826625,
|
||
|
|
"grad_norm": 0.26003521710729083,
|
||
|
|
"learning_rate": 3.4479200679348864e-05,
|
||
|
|
"loss": 0.3186,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.31395748257637024,
|
||
|
|
"step": 1440,
|
||
|
|
"valid_targets_mean": 5973.2,
|
||
|
|
"valid_targets_min": 542
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.236842105263158,
|
||
|
|
"grad_norm": 0.24457858576148117,
|
||
|
|
"learning_rate": 3.442583162235535e-05,
|
||
|
|
"loss": 0.3117,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3054529130458832,
|
||
|
|
"step": 1445,
|
||
|
|
"valid_targets_mean": 5461.6,
|
||
|
|
"valid_targets_min": 360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2445820433436534,
|
||
|
|
"grad_norm": 0.24004213325496904,
|
||
|
|
"learning_rate": 3.437224758207728e-05,
|
||
|
|
"loss": 0.3199,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3216618597507477,
|
||
|
|
"step": 1450,
|
||
|
|
"valid_targets_mean": 5406.7,
|
||
|
|
"valid_targets_min": 310
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2523219814241484,
|
||
|
|
"grad_norm": 0.21241014618626178,
|
||
|
|
"learning_rate": 3.431844935705947e-05,
|
||
|
|
"loss": 0.3024,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2961379289627075,
|
||
|
|
"step": 1455,
|
||
|
|
"valid_targets_mean": 5525.3,
|
||
|
|
"valid_targets_min": 987
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.260061919504644,
|
||
|
|
"grad_norm": 0.22152056824200905,
|
||
|
|
"learning_rate": 3.4264437749038646e-05,
|
||
|
|
"loss": 0.3129,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3125271797180176,
|
||
|
|
"step": 1460,
|
||
|
|
"valid_targets_mean": 5494.2,
|
||
|
|
"valid_targets_min": 803
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2678018575851393,
|
||
|
|
"grad_norm": 0.19374454211250638,
|
||
|
|
"learning_rate": 3.421021356293153e-05,
|
||
|
|
"loss": 0.3122,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3092339038848877,
|
||
|
|
"step": 1465,
|
||
|
|
"valid_targets_mean": 5719.7,
|
||
|
|
"valid_targets_min": 988
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2755417956656347,
|
||
|
|
"grad_norm": 0.22834211854210285,
|
||
|
|
"learning_rate": 3.41557776068228e-05,
|
||
|
|
"loss": 0.3044,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.32388952374458313,
|
||
|
|
"step": 1470,
|
||
|
|
"valid_targets_mean": 5339.7,
|
||
|
|
"valid_targets_min": 596
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.28328173374613,
|
||
|
|
"grad_norm": 0.20465907806901326,
|
||
|
|
"learning_rate": 3.410113069195308e-05,
|
||
|
|
"loss": 0.3059,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.32336580753326416,
|
||
|
|
"step": 1475,
|
||
|
|
"valid_targets_mean": 5534.7,
|
||
|
|
"valid_targets_min": 575
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.291021671826625,
|
||
|
|
"grad_norm": 0.2063190951057895,
|
||
|
|
"learning_rate": 3.404627363270684e-05,
|
||
|
|
"loss": 0.3019,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30562636256217957,
|
||
|
|
"step": 1480,
|
||
|
|
"valid_targets_mean": 5558.2,
|
||
|
|
"valid_targets_min": 256
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2987616099071206,
|
||
|
|
"grad_norm": 0.21467967984382336,
|
||
|
|
"learning_rate": 3.3991207246600266e-05,
|
||
|
|
"loss": 0.3044,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30534613132476807,
|
||
|
|
"step": 1485,
|
||
|
|
"valid_targets_mean": 5328.4,
|
||
|
|
"valid_targets_min": 524
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.306501547987616,
|
||
|
|
"grad_norm": 0.2012666852914282,
|
||
|
|
"learning_rate": 3.393593235426905e-05,
|
||
|
|
"loss": 0.2981,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30283552408218384,
|
||
|
|
"step": 1490,
|
||
|
|
"valid_targets_mean": 5696.6,
|
||
|
|
"valid_targets_min": 614
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3142414860681115,
|
||
|
|
"grad_norm": 0.20278482725474747,
|
||
|
|
"learning_rate": 3.388044977945622e-05,
|
||
|
|
"loss": 0.3067,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29829442501068115,
|
||
|
|
"step": 1495,
|
||
|
|
"valid_targets_mean": 5787.3,
|
||
|
|
"valid_targets_min": 496
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.321981424148607,
|
||
|
|
"grad_norm": 0.22029551441625908,
|
||
|
|
"learning_rate": 3.38247603489998e-05,
|
||
|
|
"loss": 0.2961,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30468475818634033,
|
||
|
|
"step": 1500,
|
||
|
|
"valid_targets_mean": 5574.5,
|
||
|
|
"valid_targets_min": 736
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.329721362229102,
|
||
|
|
"grad_norm": 0.1975905695562398,
|
||
|
|
"learning_rate": 3.376886489282051e-05,
|
||
|
|
"loss": 0.3051,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3025839328765869,
|
||
|
|
"step": 1505,
|
||
|
|
"valid_targets_mean": 5946.5,
|
||
|
|
"valid_targets_min": 670
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3374613003095974,
|
||
|
|
"grad_norm": 0.18840491243619079,
|
||
|
|
"learning_rate": 3.371276424390941e-05,
|
||
|
|
"loss": 0.2981,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.300428569316864,
|
||
|
|
"step": 1510,
|
||
|
|
"valid_targets_mean": 6433.6,
|
||
|
|
"valid_targets_min": 353
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.345201238390093,
|
||
|
|
"grad_norm": 1.1139901439943023,
|
||
|
|
"learning_rate": 3.365645923831547e-05,
|
||
|
|
"loss": 0.3514,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5291757583618164,
|
||
|
|
"step": 1515,
|
||
|
|
"valid_targets_mean": 1094.5,
|
||
|
|
"valid_targets_min": 286
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3529411764705883,
|
||
|
|
"grad_norm": 0.22641563503300632,
|
||
|
|
"learning_rate": 3.359995071513314e-05,
|
||
|
|
"loss": 0.3641,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2923022508621216,
|
||
|
|
"step": 1520,
|
||
|
|
"valid_targets_mean": 6291.4,
|
||
|
|
"valid_targets_min": 338
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3606811145510838,
|
||
|
|
"grad_norm": 0.8435882996564606,
|
||
|
|
"learning_rate": 3.3543239516489806e-05,
|
||
|
|
"loss": 0.3199,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.43032994866371155,
|
||
|
|
"step": 1525,
|
||
|
|
"valid_targets_mean": 1675.7,
|
||
|
|
"valid_targets_min": 527
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3684210526315788,
|
||
|
|
"grad_norm": 0.24179443879767235,
|
||
|
|
"learning_rate": 3.348632648753325e-05,
|
||
|
|
"loss": 0.3669,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2968432903289795,
|
||
|
|
"step": 1530,
|
||
|
|
"valid_targets_mean": 6662.4,
|
||
|
|
"valid_targets_min": 1694
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.376160990712074,
|
||
|
|
"grad_norm": 0.3188601331382926,
|
||
|
|
"learning_rate": 3.342921247641908e-05,
|
||
|
|
"loss": 0.3804,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3978138566017151,
|
||
|
|
"step": 1535,
|
||
|
|
"valid_targets_mean": 2726.5,
|
||
|
|
"valid_targets_min": 474
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3839009287925697,
|
||
|
|
"grad_norm": 0.21191486331117068,
|
||
|
|
"learning_rate": 3.337189833429808e-05,
|
||
|
|
"loss": 0.2983,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2926940321922302,
|
||
|
|
"step": 1540,
|
||
|
|
"valid_targets_mean": 6219.8,
|
||
|
|
"valid_targets_min": 602
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.391640866873065,
|
||
|
|
"grad_norm": 0.767333077770554,
|
||
|
|
"learning_rate": 3.3314384915303474e-05,
|
||
|
|
"loss": 0.4243,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5755810737609863,
|
||
|
|
"step": 1545,
|
||
|
|
"valid_targets_mean": 844.2,
|
||
|
|
"valid_targets_min": 290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3993808049535605,
|
||
|
|
"grad_norm": 0.22948341213402115,
|
||
|
|
"learning_rate": 3.3256673076538306e-05,
|
||
|
|
"loss": 0.2988,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30840983986854553,
|
||
|
|
"step": 1550,
|
||
|
|
"valid_targets_mean": 5912.4,
|
||
|
|
"valid_targets_min": 277
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4071207430340555,
|
||
|
|
"grad_norm": 0.42228954926883366,
|
||
|
|
"learning_rate": 3.3198763678062556e-05,
|
||
|
|
"loss": 0.3829,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4217245578765869,
|
||
|
|
"step": 1555,
|
||
|
|
"valid_targets_mean": 1662.7,
|
||
|
|
"valid_targets_min": 313
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.414860681114551,
|
||
|
|
"grad_norm": 0.2073997058373611,
|
||
|
|
"learning_rate": 3.314065758288038e-05,
|
||
|
|
"loss": 0.295,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2824292778968811,
|
||
|
|
"step": 1560,
|
||
|
|
"valid_targets_mean": 6104.8,
|
||
|
|
"valid_targets_min": 435
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4226006191950464,
|
||
|
|
"grad_norm": 0.30844072012315227,
|
||
|
|
"learning_rate": 3.308235565692724e-05,
|
||
|
|
"loss": 0.4224,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.32070478796958923,
|
||
|
|
"step": 1565,
|
||
|
|
"valid_targets_mean": 2831.9,
|
||
|
|
"valid_targets_min": 265
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.430340557275542,
|
||
|
|
"grad_norm": 0.21689549134132569,
|
||
|
|
"learning_rate": 3.302385876905701e-05,
|
||
|
|
"loss": 0.2934,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29463163018226624,
|
||
|
|
"step": 1570,
|
||
|
|
"valid_targets_mean": 6694.6,
|
||
|
|
"valid_targets_min": 250
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4380804953560373,
|
||
|
|
"grad_norm": 0.31437531938149327,
|
||
|
|
"learning_rate": 3.2965167791028996e-05,
|
||
|
|
"loss": 0.4221,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.33279407024383545,
|
||
|
|
"step": 1575,
|
||
|
|
"valid_targets_mean": 3286.7,
|
||
|
|
"valid_targets_min": 441
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4458204334365323,
|
||
|
|
"grad_norm": 0.21591415805086642,
|
||
|
|
"learning_rate": 3.290628359749498e-05,
|
||
|
|
"loss": 0.2911,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29980888962745667,
|
||
|
|
"step": 1580,
|
||
|
|
"valid_targets_mean": 6532.3,
|
||
|
|
"valid_targets_min": 694
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4535603715170278,
|
||
|
|
"grad_norm": 0.9171936690789186,
|
||
|
|
"learning_rate": 3.284720706598615e-05,
|
||
|
|
"loss": 0.4732,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5318858623504639,
|
||
|
|
"step": 1585,
|
||
|
|
"valid_targets_mean": 1694.4,
|
||
|
|
"valid_targets_min": 257
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.461300309597523,
|
||
|
|
"grad_norm": 0.38731811756957063,
|
||
|
|
"learning_rate": 3.2787939076900066e-05,
|
||
|
|
"loss": 0.5154,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.527919590473175,
|
||
|
|
"step": 1590,
|
||
|
|
"valid_targets_mean": 2654.1,
|
||
|
|
"valid_targets_min": 261
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4690402476780187,
|
||
|
|
"grad_norm": 0.4171753480169824,
|
||
|
|
"learning_rate": 3.27284805134875e-05,
|
||
|
|
"loss": 0.5075,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4914923310279846,
|
||
|
|
"step": 1595,
|
||
|
|
"valid_targets_mean": 2251.2,
|
||
|
|
"valid_targets_min": 337
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.476780185758514,
|
||
|
|
"grad_norm": 0.3216725921959662,
|
||
|
|
"learning_rate": 3.2668832261839286e-05,
|
||
|
|
"loss": 0.5088,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.507831335067749,
|
||
|
|
"step": 1600,
|
||
|
|
"valid_targets_mean": 3427.8,
|
||
|
|
"valid_targets_min": 298
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.484520123839009,
|
||
|
|
"grad_norm": 0.2881166520422749,
|
||
|
|
"learning_rate": 3.2608995210873133e-05,
|
||
|
|
"loss": 0.4895,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4918453097343445,
|
||
|
|
"step": 1605,
|
||
|
|
"valid_targets_mean": 3676.3,
|
||
|
|
"valid_targets_min": 415
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4922600619195046,
|
||
|
|
"grad_norm": 0.2958170811269623,
|
||
|
|
"learning_rate": 3.254897025232034e-05,
|
||
|
|
"loss": 0.4977,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4771663248538971,
|
||
|
|
"step": 1610,
|
||
|
|
"valid_targets_mean": 3749.4,
|
||
|
|
"valid_targets_min": 319
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5,
|
||
|
|
"grad_norm": 0.2837362902945792,
|
||
|
|
"learning_rate": 3.248875828071255e-05,
|
||
|
|
"loss": 0.4807,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.47112274169921875,
|
||
|
|
"step": 1615,
|
||
|
|
"valid_targets_mean": 3321.1,
|
||
|
|
"valid_targets_min": 266
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5077399380804954,
|
||
|
|
"grad_norm": 0.27666577838284534,
|
||
|
|
"learning_rate": 3.242836019336839e-05,
|
||
|
|
"loss": 0.4933,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46771207451820374,
|
||
|
|
"step": 1620,
|
||
|
|
"valid_targets_mean": 3494.8,
|
||
|
|
"valid_targets_min": 260
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.515479876160991,
|
||
|
|
"grad_norm": 0.2736109868976536,
|
||
|
|
"learning_rate": 3.2367776890380116e-05,
|
||
|
|
"loss": 0.4956,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4948537349700928,
|
||
|
|
"step": 1625,
|
||
|
|
"valid_targets_mean": 3470.5,
|
||
|
|
"valid_targets_min": 217
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.523219814241486,
|
||
|
|
"grad_norm": 0.34656851977562886,
|
||
|
|
"learning_rate": 3.2307009274600166e-05,
|
||
|
|
"loss": 0.4846,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5057222247123718,
|
||
|
|
"step": 1630,
|
||
|
|
"valid_targets_mean": 2027.0,
|
||
|
|
"valid_targets_min": 249
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5309597523219813,
|
||
|
|
"grad_norm": 0.3123862808012193,
|
||
|
|
"learning_rate": 3.224605825162774e-05,
|
||
|
|
"loss": 0.4904,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4758598506450653,
|
||
|
|
"step": 1635,
|
||
|
|
"valid_targets_mean": 2672.5,
|
||
|
|
"valid_targets_min": 243
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.538699690402477,
|
||
|
|
"grad_norm": 0.26930289579873684,
|
||
|
|
"learning_rate": 3.218492472979531e-05,
|
||
|
|
"loss": 0.481,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48779118061065674,
|
||
|
|
"step": 1640,
|
||
|
|
"valid_targets_mean": 3140.0,
|
||
|
|
"valid_targets_min": 302
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5464396284829722,
|
||
|
|
"grad_norm": 0.23750045721199364,
|
||
|
|
"learning_rate": 3.2123609620155054e-05,
|
||
|
|
"loss": 0.4887,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4799521863460541,
|
||
|
|
"step": 1645,
|
||
|
|
"valid_targets_mean": 3984.5,
|
||
|
|
"valid_targets_min": 258
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5541795665634677,
|
||
|
|
"grad_norm": 0.23865030053456518,
|
||
|
|
"learning_rate": 3.2062113836465284e-05,
|
||
|
|
"loss": 0.4724,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.45580023527145386,
|
||
|
|
"step": 1650,
|
||
|
|
"valid_targets_mean": 3525.9,
|
||
|
|
"valid_targets_min": 242
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5619195046439627,
|
||
|
|
"grad_norm": 0.2237430503251543,
|
||
|
|
"learning_rate": 3.200043829517685e-05,
|
||
|
|
"loss": 0.4848,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46315300464630127,
|
||
|
|
"step": 1655,
|
||
|
|
"valid_targets_mean": 4537.2,
|
||
|
|
"valid_targets_min": 394
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.569659442724458,
|
||
|
|
"grad_norm": 0.3930034774926644,
|
||
|
|
"learning_rate": 3.1938583915419466e-05,
|
||
|
|
"loss": 0.6249,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6780774593353271,
|
||
|
|
"step": 1660,
|
||
|
|
"valid_targets_mean": 3944.9,
|
||
|
|
"valid_targets_min": 626
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5773993808049536,
|
||
|
|
"grad_norm": 0.34453504911087135,
|
||
|
|
"learning_rate": 3.1876551618988e-05,
|
||
|
|
"loss": 0.617,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.753798246383667,
|
||
|
|
"step": 1665,
|
||
|
|
"valid_targets_mean": 3327.8,
|
||
|
|
"valid_targets_min": 423
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.585139318885449,
|
||
|
|
"grad_norm": 0.2838881011156283,
|
||
|
|
"learning_rate": 3.181434233032877e-05,
|
||
|
|
"loss": 0.6165,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6676381826400757,
|
||
|
|
"step": 1670,
|
||
|
|
"valid_targets_mean": 4198.4,
|
||
|
|
"valid_targets_min": 570
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5928792569659445,
|
||
|
|
"grad_norm": 0.2790899111099941,
|
||
|
|
"learning_rate": 3.1751956976525735e-05,
|
||
|
|
"loss": 0.5993,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5974432826042175,
|
||
|
|
"step": 1675,
|
||
|
|
"valid_targets_mean": 3893.0,
|
||
|
|
"valid_targets_min": 885
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6006191950464395,
|
||
|
|
"grad_norm": 0.2794378550756072,
|
||
|
|
"learning_rate": 3.1689396487286715e-05,
|
||
|
|
"loss": 0.6127,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6124075055122375,
|
||
|
|
"step": 1680,
|
||
|
|
"valid_targets_mean": 3704.1,
|
||
|
|
"valid_targets_min": 891
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.608359133126935,
|
||
|
|
"grad_norm": 0.2789151336324382,
|
||
|
|
"learning_rate": 3.16266617949295e-05,
|
||
|
|
"loss": 0.6065,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5840175151824951,
|
||
|
|
"step": 1685,
|
||
|
|
"valid_targets_mean": 3793.6,
|
||
|
|
"valid_targets_min": 1124
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6160990712074303,
|
||
|
|
"grad_norm": 0.2734596668158036,
|
||
|
|
"learning_rate": 3.1563753834367964e-05,
|
||
|
|
"loss": 0.5993,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5768203735351562,
|
||
|
|
"step": 1690,
|
||
|
|
"valid_targets_mean": 3647.5,
|
||
|
|
"valid_targets_min": 775
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.623839009287926,
|
||
|
|
"grad_norm": 0.2629657965107749,
|
||
|
|
"learning_rate": 3.150067354309813e-05,
|
||
|
|
"loss": 0.5964,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6152141094207764,
|
||
|
|
"step": 1695,
|
||
|
|
"valid_targets_mean": 3801.7,
|
||
|
|
"valid_targets_min": 835
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6315789473684212,
|
||
|
|
"grad_norm": 0.2422647569007029,
|
||
|
|
"learning_rate": 3.143742186118425e-05,
|
||
|
|
"loss": 0.595,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5769956707954407,
|
||
|
|
"step": 1700,
|
||
|
|
"valid_targets_mean": 4259.5,
|
||
|
|
"valid_targets_min": 886
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6393188854489162,
|
||
|
|
"grad_norm": 0.4788947094554788,
|
||
|
|
"learning_rate": 3.137399973124471e-05,
|
||
|
|
"loss": 0.5978,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6065722703933716,
|
||
|
|
"step": 1705,
|
||
|
|
"valid_targets_mean": 4118.3,
|
||
|
|
"valid_targets_min": 1310
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6470588235294117,
|
||
|
|
"grad_norm": 0.22889775320047,
|
||
|
|
"learning_rate": 3.1310408098438035e-05,
|
||
|
|
"loss": 0.5946,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5836830139160156,
|
||
|
|
"step": 1710,
|
||
|
|
"valid_targets_mean": 4160.2,
|
||
|
|
"valid_targets_min": 795
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.654798761609907,
|
||
|
|
"grad_norm": 0.22473322357744355,
|
||
|
|
"learning_rate": 3.124664791044881e-05,
|
||
|
|
"loss": 0.6107,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5740764141082764,
|
||
|
|
"step": 1715,
|
||
|
|
"valid_targets_mean": 4790.0,
|
||
|
|
"valid_targets_min": 1104
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6625386996904026,
|
||
|
|
"grad_norm": 0.2966932065747707,
|
||
|
|
"learning_rate": 3.118272011747351e-05,
|
||
|
|
"loss": 0.5952,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5603684186935425,
|
||
|
|
"step": 1720,
|
||
|
|
"valid_targets_mean": 4677.6,
|
||
|
|
"valid_targets_min": 628
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.670278637770898,
|
||
|
|
"grad_norm": 0.23231334161121603,
|
||
|
|
"learning_rate": 3.1118625672206416e-05,
|
||
|
|
"loss": 0.6097,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5626471638679504,
|
||
|
|
"step": 1725,
|
||
|
|
"valid_targets_mean": 3947.1,
|
||
|
|
"valid_targets_min": 1088
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.678018575851393,
|
||
|
|
"grad_norm": 0.4117228703939541,
|
||
|
|
"learning_rate": 3.105436552982533e-05,
|
||
|
|
"loss": 0.6053,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6307675838470459,
|
||
|
|
"step": 1730,
|
||
|
|
"valid_targets_mean": 2494.4,
|
||
|
|
"valid_targets_min": 400
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6857585139318885,
|
||
|
|
"grad_norm": 0.3574754903244721,
|
||
|
|
"learning_rate": 3.098994064797741e-05,
|
||
|
|
"loss": 0.5138,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5313445329666138,
|
||
|
|
"step": 1735,
|
||
|
|
"valid_targets_mean": 3803.9,
|
||
|
|
"valid_targets_min": 341
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.693498452012384,
|
||
|
|
"grad_norm": 0.33886970691605184,
|
||
|
|
"learning_rate": 3.0925351986764835e-05,
|
||
|
|
"loss": 0.4993,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5782281756401062,
|
||
|
|
"step": 1740,
|
||
|
|
"valid_targets_mean": 2460.7,
|
||
|
|
"valid_targets_min": 507
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7012383900928794,
|
||
|
|
"grad_norm": 0.28990040169452735,
|
||
|
|
"learning_rate": 3.0860600508730584e-05,
|
||
|
|
"loss": 0.5112,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5253264904022217,
|
||
|
|
"step": 1745,
|
||
|
|
"valid_targets_mean": 3748.4,
|
||
|
|
"valid_targets_min": 754
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.708978328173375,
|
||
|
|
"grad_norm": 0.34065402581608445,
|
||
|
|
"learning_rate": 3.0795687178843985e-05,
|
||
|
|
"loss": 0.501,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6163568496704102,
|
||
|
|
"step": 1750,
|
||
|
|
"valid_targets_mean": 2391.1,
|
||
|
|
"valid_targets_min": 521
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.71671826625387,
|
||
|
|
"grad_norm": 0.2766460932239942,
|
||
|
|
"learning_rate": 3.0730612964486425e-05,
|
||
|
|
"loss": 0.4911,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5015374422073364,
|
||
|
|
"step": 1755,
|
||
|
|
"valid_targets_mean": 3659.7,
|
||
|
|
"valid_targets_min": 806
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7244582043343653,
|
||
|
|
"grad_norm": 0.2787890495183026,
|
||
|
|
"learning_rate": 3.06653788354369e-05,
|
||
|
|
"loss": 0.492,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5921324491500854,
|
||
|
|
"step": 1760,
|
||
|
|
"valid_targets_mean": 3227.0,
|
||
|
|
"valid_targets_min": 450
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7321981424148607,
|
||
|
|
"grad_norm": 0.27694944731950677,
|
||
|
|
"learning_rate": 3.059998576385754e-05,
|
||
|
|
"loss": 0.4994,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5102630257606506,
|
||
|
|
"step": 1765,
|
||
|
|
"valid_targets_mean": 3305.4,
|
||
|
|
"valid_targets_min": 621
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.739938080495356,
|
||
|
|
"grad_norm": 0.27204346790929906,
|
||
|
|
"learning_rate": 3.0534434724279154e-05,
|
||
|
|
"loss": 0.4925,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5549046993255615,
|
||
|
|
"step": 1770,
|
||
|
|
"valid_targets_mean": 3149.7,
|
||
|
|
"valid_targets_min": 462
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7476780185758516,
|
||
|
|
"grad_norm": 0.2527886963134821,
|
||
|
|
"learning_rate": 3.0468726693586677e-05,
|
||
|
|
"loss": 0.4835,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5114155411720276,
|
||
|
|
"step": 1775,
|
||
|
|
"valid_targets_mean": 3709.8,
|
||
|
|
"valid_targets_min": 263
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7554179566563466,
|
||
|
|
"grad_norm": 0.2707841045336601,
|
||
|
|
"learning_rate": 3.0402862651004664e-05,
|
||
|
|
"loss": 0.4911,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5209786891937256,
|
||
|
|
"step": 1780,
|
||
|
|
"valid_targets_mean": 3436.5,
|
||
|
|
"valid_targets_min": 523
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.763157894736842,
|
||
|
|
"grad_norm": 0.22844304332076776,
|
||
|
|
"learning_rate": 3.0336843578082634e-05,
|
||
|
|
"loss": 0.5002,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5332248210906982,
|
||
|
|
"step": 1785,
|
||
|
|
"valid_targets_mean": 4309.8,
|
||
|
|
"valid_targets_min": 269
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7708978328173375,
|
||
|
|
"grad_norm": 0.26064350286425514,
|
||
|
|
"learning_rate": 3.0270670458680477e-05,
|
||
|
|
"loss": 0.491,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5058698654174805,
|
||
|
|
"step": 1790,
|
||
|
|
"valid_targets_mean": 3643.2,
|
||
|
|
"valid_targets_min": 373
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.778637770897833,
|
||
|
|
"grad_norm": 0.2463274711183827,
|
||
|
|
"learning_rate": 3.0204344278953783e-05,
|
||
|
|
"loss": 0.4916,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5086028575897217,
|
||
|
|
"step": 1795,
|
||
|
|
"valid_targets_mean": 3911.8,
|
||
|
|
"valid_targets_min": 1193
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7863777089783284,
|
||
|
|
"grad_norm": 0.23836230016162377,
|
||
|
|
"learning_rate": 3.013786602733916e-05,
|
||
|
|
"loss": 0.4844,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5128570199012756,
|
||
|
|
"step": 1800,
|
||
|
|
"valid_targets_mean": 3986.0,
|
||
|
|
"valid_targets_min": 1074
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7941176470588234,
|
||
|
|
"grad_norm": 0.24363120488345796,
|
||
|
|
"learning_rate": 3.0071236694539484e-05,
|
||
|
|
"loss": 0.503,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5001965165138245,
|
||
|
|
"step": 1805,
|
||
|
|
"valid_targets_mean": 4058.6,
|
||
|
|
"valid_targets_min": 964
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.801857585139319,
|
||
|
|
"grad_norm": 0.23968393576042865,
|
||
|
|
"learning_rate": 3.000445727350914e-05,
|
||
|
|
"loss": 0.4874,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.47443175315856934,
|
||
|
|
"step": 1810,
|
||
|
|
"valid_targets_mean": 3589.4,
|
||
|
|
"valid_targets_min": 1066
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8095975232198143,
|
||
|
|
"grad_norm": 0.25812573903793573,
|
||
|
|
"learning_rate": 2.9937528759439236e-05,
|
||
|
|
"loss": 0.4925,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5002115964889526,
|
||
|
|
"step": 1815,
|
||
|
|
"valid_targets_mean": 3394.9,
|
||
|
|
"valid_targets_min": 311
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8173374613003097,
|
||
|
|
"grad_norm": 0.2670121513113001,
|
||
|
|
"learning_rate": 2.9870452149742754e-05,
|
||
|
|
"loss": 0.5637,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5274510383605957,
|
||
|
|
"step": 1820,
|
||
|
|
"valid_targets_mean": 3667.1,
|
||
|
|
"valid_targets_min": 855
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.825077399380805,
|
||
|
|
"grad_norm": 0.26737218697905935,
|
||
|
|
"learning_rate": 2.9803228444039695e-05,
|
||
|
|
"loss": 0.4918,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5100606679916382,
|
||
|
|
"step": 1825,
|
||
|
|
"valid_targets_mean": 3378.0,
|
||
|
|
"valid_targets_min": 549
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8328173374613,
|
||
|
|
"grad_norm": 0.2924958072815413,
|
||
|
|
"learning_rate": 2.9735858644142186e-05,
|
||
|
|
"loss": 0.5766,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5242557525634766,
|
||
|
|
"step": 1830,
|
||
|
|
"valid_targets_mean": 2999.8,
|
||
|
|
"valid_targets_min": 301
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8405572755417956,
|
||
|
|
"grad_norm": 0.2657127896678461,
|
||
|
|
"learning_rate": 2.9668343754039547e-05,
|
||
|
|
"loss": 0.4882,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.500880241394043,
|
||
|
|
"step": 1835,
|
||
|
|
"valid_targets_mean": 3465.6,
|
||
|
|
"valid_targets_min": 544
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.848297213622291,
|
||
|
|
"grad_norm": 0.31895585928274195,
|
||
|
|
"learning_rate": 2.9600684779883332e-05,
|
||
|
|
"loss": 0.5327,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5870118141174316,
|
||
|
|
"step": 1840,
|
||
|
|
"valid_targets_mean": 2587.6,
|
||
|
|
"valid_targets_min": 369
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8560371517027865,
|
||
|
|
"grad_norm": 0.2947087376516324,
|
||
|
|
"learning_rate": 2.9532882729972324e-05,
|
||
|
|
"loss": 0.486,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5051394701004028,
|
||
|
|
"step": 1845,
|
||
|
|
"valid_targets_mean": 3274.3,
|
||
|
|
"valid_targets_min": 346
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.863777089783282,
|
||
|
|
"grad_norm": 0.35596997480807346,
|
||
|
|
"learning_rate": 2.9464938614737516e-05,
|
||
|
|
"loss": 0.5258,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6333910822868347,
|
||
|
|
"step": 1850,
|
||
|
|
"valid_targets_mean": 2213.1,
|
||
|
|
"valid_targets_min": 520
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.871517027863777,
|
||
|
|
"grad_norm": 0.268815107437887,
|
||
|
|
"learning_rate": 2.9396853446727038e-05,
|
||
|
|
"loss": 0.4882,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48987096548080444,
|
||
|
|
"step": 1855,
|
||
|
|
"valid_targets_mean": 3187.1,
|
||
|
|
"valid_targets_min": 454
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8792569659442724,
|
||
|
|
"grad_norm": 0.6097764254204212,
|
||
|
|
"learning_rate": 2.9328628240591125e-05,
|
||
|
|
"loss": 0.5772,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.9093445539474487,
|
||
|
|
"step": 1860,
|
||
|
|
"valid_targets_mean": 1165.0,
|
||
|
|
"valid_targets_min": 257
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.886996904024768,
|
||
|
|
"grad_norm": 0.27264167148390517,
|
||
|
|
"learning_rate": 2.926026401306691e-05,
|
||
|
|
"loss": 0.4846,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4717441201210022,
|
||
|
|
"step": 1865,
|
||
|
|
"valid_targets_mean": 3309.6,
|
||
|
|
"valid_targets_min": 624
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8947368421052633,
|
||
|
|
"grad_norm": 0.6110346071645671,
|
||
|
|
"learning_rate": 2.9191761782963328e-05,
|
||
|
|
"loss": 0.5526,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.908230185508728,
|
||
|
|
"step": 1870,
|
||
|
|
"valid_targets_mean": 1005.5,
|
||
|
|
"valid_targets_min": 350
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9024767801857587,
|
||
|
|
"grad_norm": 0.27103733941142416,
|
||
|
|
"learning_rate": 2.9123122571145927e-05,
|
||
|
|
"loss": 0.4986,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46968722343444824,
|
||
|
|
"step": 1875,
|
||
|
|
"valid_targets_mean": 3382.6,
|
||
|
|
"valid_targets_min": 693
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9102167182662537,
|
||
|
|
"grad_norm": 0.4078092587308404,
|
||
|
|
"learning_rate": 2.905434740052164e-05,
|
||
|
|
"loss": 0.5031,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6267281174659729,
|
||
|
|
"step": 1880,
|
||
|
|
"valid_targets_mean": 1660.4,
|
||
|
|
"valid_targets_min": 441
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.917956656346749,
|
||
|
|
"grad_norm": 0.27217237815118095,
|
||
|
|
"learning_rate": 2.8985437296023548e-05,
|
||
|
|
"loss": 0.4992,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46388787031173706,
|
||
|
|
"step": 1885,
|
||
|
|
"valid_targets_mean": 3356.6,
|
||
|
|
"valid_targets_min": 296
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9256965944272446,
|
||
|
|
"grad_norm": 0.3389404842302698,
|
||
|
|
"learning_rate": 2.8916393284595617e-05,
|
||
|
|
"loss": 0.4991,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5835111737251282,
|
||
|
|
"step": 1890,
|
||
|
|
"valid_targets_mean": 2242.9,
|
||
|
|
"valid_targets_min": 530
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.93343653250774,
|
||
|
|
"grad_norm": 0.6828246789401863,
|
||
|
|
"learning_rate": 2.8847216395177358e-05,
|
||
|
|
"loss": 0.3413,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2925789952278137,
|
||
|
|
"step": 1895,
|
||
|
|
"valid_targets_mean": 4536.9,
|
||
|
|
"valid_targets_min": 1128
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9411764705882355,
|
||
|
|
"grad_norm": 0.3769733982281982,
|
||
|
|
"learning_rate": 2.877790765868854e-05,
|
||
|
|
"loss": 0.2536,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24766355752944946,
|
||
|
|
"step": 1900,
|
||
|
|
"valid_targets_mean": 4706.6,
|
||
|
|
"valid_targets_min": 957
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9489164086687305,
|
||
|
|
"grad_norm": 0.3489263472505753,
|
||
|
|
"learning_rate": 2.8708468108013808e-05,
|
||
|
|
"loss": 0.2412,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.217962384223938,
|
||
|
|
"step": 1905,
|
||
|
|
"valid_targets_mean": 5179.4,
|
||
|
|
"valid_targets_min": 1060
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.956656346749226,
|
||
|
|
"grad_norm": 0.34320738226175634,
|
||
|
|
"learning_rate": 2.863889877798726e-05,
|
||
|
|
"loss": 0.2315,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2418029010295868,
|
||
|
|
"step": 1910,
|
||
|
|
"valid_targets_mean": 4519.4,
|
||
|
|
"valid_targets_min": 1164
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9643962848297214,
|
||
|
|
"grad_norm": 0.3437270072909334,
|
||
|
|
"learning_rate": 2.8569200705377066e-05,
|
||
|
|
"loss": 0.2468,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24452561140060425,
|
||
|
|
"step": 1915,
|
||
|
|
"valid_targets_mean": 4008.1,
|
||
|
|
"valid_targets_min": 1280
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.972136222910217,
|
||
|
|
"grad_norm": 0.41905262685644057,
|
||
|
|
"learning_rate": 2.8499374928870024e-05,
|
||
|
|
"loss": 0.2509,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24626369774341583,
|
||
|
|
"step": 1920,
|
||
|
|
"valid_targets_mean": 4820.1,
|
||
|
|
"valid_targets_min": 1337
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9798761609907123,
|
||
|
|
"grad_norm": 0.7922007857648102,
|
||
|
|
"learning_rate": 2.8429422489056023e-05,
|
||
|
|
"loss": 0.2431,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24944472312927246,
|
||
|
|
"step": 1925,
|
||
|
|
"valid_targets_mean": 2952.3,
|
||
|
|
"valid_targets_min": 187
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9876160990712073,
|
||
|
|
"grad_norm": 0.3952237220664012,
|
||
|
|
"learning_rate": 2.8359344428412593e-05,
|
||
|
|
"loss": 0.2353,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23666620254516602,
|
||
|
|
"step": 1930,
|
||
|
|
"valid_targets_mean": 2867.5,
|
||
|
|
"valid_targets_min": 1216
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9953560371517027,
|
||
|
|
"grad_norm": 0.2845565759709267,
|
||
|
|
"learning_rate": 2.8289141791289352e-05,
|
||
|
|
"loss": 0.227,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2118290364742279,
|
||
|
|
"step": 1935,
|
||
|
|
"valid_targets_mean": 3687.5,
|
||
|
|
"valid_targets_min": 1785
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.003095975232198,
|
||
|
|
"grad_norm": 0.7305362958203637,
|
||
|
|
"learning_rate": 2.8218815623892423e-05,
|
||
|
|
"loss": 0.2513,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2713759243488312,
|
||
|
|
"step": 1940,
|
||
|
|
"valid_targets_mean": 3785.3,
|
||
|
|
"valid_targets_min": 449
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0108359133126936,
|
||
|
|
"grad_norm": 0.4746075790560174,
|
||
|
|
"learning_rate": 2.814836697426888e-05,
|
||
|
|
"loss": 0.263,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2620471715927124,
|
||
|
|
"step": 1945,
|
||
|
|
"valid_targets_mean": 3751.2,
|
||
|
|
"valid_targets_min": 410
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.018575851393189,
|
||
|
|
"grad_norm": 0.3403079763185503,
|
||
|
|
"learning_rate": 2.8077796892291093e-05,
|
||
|
|
"loss": 0.2531,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24807900190353394,
|
||
|
|
"step": 1950,
|
||
|
|
"valid_targets_mean": 3883.6,
|
||
|
|
"valid_targets_min": 451
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.026315789473684,
|
||
|
|
"grad_norm": 0.3002256864648988,
|
||
|
|
"learning_rate": 2.8007106429641092e-05,
|
||
|
|
"loss": 0.252,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25025856494903564,
|
||
|
|
"step": 1955,
|
||
|
|
"valid_targets_mean": 3835.6,
|
||
|
|
"valid_targets_min": 388
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0340557275541795,
|
||
|
|
"grad_norm": 0.26678605333498767,
|
||
|
|
"learning_rate": 2.7936296639794917e-05,
|
||
|
|
"loss": 0.2474,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2513650059700012,
|
||
|
|
"step": 1960,
|
||
|
|
"valid_targets_mean": 3575.0,
|
||
|
|
"valid_targets_min": 600
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.041795665634675,
|
||
|
|
"grad_norm": 0.2336652691036555,
|
||
|
|
"learning_rate": 2.7865368578006893e-05,
|
||
|
|
"loss": 0.2421,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24708211421966553,
|
||
|
|
"step": 1965,
|
||
|
|
"valid_targets_mean": 3773.7,
|
||
|
|
"valid_targets_min": 435
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0495356037151704,
|
||
|
|
"grad_norm": 0.22195813868047298,
|
||
|
|
"learning_rate": 2.7794323301293907e-05,
|
||
|
|
"loss": 0.2398,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23768779635429382,
|
||
|
|
"step": 1970,
|
||
|
|
"valid_targets_mean": 3777.3,
|
||
|
|
"valid_targets_min": 370
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.057275541795666,
|
||
|
|
"grad_norm": 0.22702183078369795,
|
||
|
|
"learning_rate": 2.7723161868419664e-05,
|
||
|
|
"loss": 0.2388,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23351342976093292,
|
||
|
|
"step": 1975,
|
||
|
|
"valid_targets_mean": 3814.9,
|
||
|
|
"valid_targets_min": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.065015479876161,
|
||
|
|
"grad_norm": 0.23703672199968612,
|
||
|
|
"learning_rate": 2.7651885339878913e-05,
|
||
|
|
"loss": 0.2368,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2429065853357315,
|
||
|
|
"step": 1980,
|
||
|
|
"valid_targets_mean": 3898.8,
|
||
|
|
"valid_targets_min": 488
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0727554179566563,
|
||
|
|
"grad_norm": 0.2236570194271787,
|
||
|
|
"learning_rate": 2.758049477788162e-05,
|
||
|
|
"loss": 0.2355,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23221372067928314,
|
||
|
|
"step": 1985,
|
||
|
|
"valid_targets_mean": 3729.5,
|
||
|
|
"valid_targets_min": 427
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0804953560371517,
|
||
|
|
"grad_norm": 0.2251840393421153,
|
||
|
|
"learning_rate": 2.7508991246337157e-05,
|
||
|
|
"loss": 0.2389,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24171897768974304,
|
||
|
|
"step": 1990,
|
||
|
|
"valid_targets_mean": 3804.0,
|
||
|
|
"valid_targets_min": 405
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.088235294117647,
|
||
|
|
"grad_norm": 0.3836604007383393,
|
||
|
|
"learning_rate": 2.7437375810838448e-05,
|
||
|
|
"loss": 0.255,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28773391246795654,
|
||
|
|
"step": 1995,
|
||
|
|
"valid_targets_mean": 4964.8,
|
||
|
|
"valid_targets_min": 2437
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0959752321981426,
|
||
|
|
"grad_norm": 0.2834164141706284,
|
||
|
|
"learning_rate": 2.7365649538646085e-05,
|
||
|
|
"loss": 0.2752,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2704768776893616,
|
||
|
|
"step": 2000,
|
||
|
|
"valid_targets_mean": 4907.5,
|
||
|
|
"valid_targets_min": 3098
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1037151702786376,
|
||
|
|
"grad_norm": 0.6386956455535773,
|
||
|
|
"learning_rate": 2.7293813498672418e-05,
|
||
|
|
"loss": 0.4013,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.273630827665329,
|
||
|
|
"step": 2005,
|
||
|
|
"valid_targets_mean": 4839.3,
|
||
|
|
"valid_targets_min": 2419
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.111455108359133,
|
||
|
|
"grad_norm": 0.3246002724374112,
|
||
|
|
"learning_rate": 2.7221868761465618e-05,
|
||
|
|
"loss": 0.2761,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27676087617874146,
|
||
|
|
"step": 2010,
|
||
|
|
"valid_targets_mean": 4763.2,
|
||
|
|
"valid_targets_min": 2304
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1191950464396285,
|
||
|
|
"grad_norm": 0.4429751785043121,
|
||
|
|
"learning_rate": 2.7149816399193773e-05,
|
||
|
|
"loss": 0.3715,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2722010314464569,
|
||
|
|
"step": 2015,
|
||
|
|
"valid_targets_mean": 4746.2,
|
||
|
|
"valid_targets_min": 2753
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.126934984520124,
|
||
|
|
"grad_norm": 0.26919114159701113,
|
||
|
|
"learning_rate": 2.7077657485628816e-05,
|
||
|
|
"loss": 0.2676,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2598717212677002,
|
||
|
|
"step": 2020,
|
||
|
|
"valid_targets_mean": 4622.8,
|
||
|
|
"valid_targets_min": 2830
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1346749226006194,
|
||
|
|
"grad_norm": 0.6999216592748773,
|
||
|
|
"learning_rate": 2.700539309613061e-05,
|
||
|
|
"loss": 0.4265,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2868514657020569,
|
||
|
|
"step": 2025,
|
||
|
|
"valid_targets_mean": 4741.5,
|
||
|
|
"valid_targets_min": 801
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1424148606811144,
|
||
|
|
"grad_norm": 0.41535413594486353,
|
||
|
|
"learning_rate": 2.6933024307630894e-05,
|
||
|
|
"loss": 0.2715,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26744377613067627,
|
||
|
|
"step": 2030,
|
||
|
|
"valid_targets_mean": 4781.2,
|
||
|
|
"valid_targets_min": 3327
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.15015479876161,
|
||
|
|
"grad_norm": 0.6015174551389255,
|
||
|
|
"learning_rate": 2.6860552198617226e-05,
|
||
|
|
"loss": 0.4058,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.39680948853492737,
|
||
|
|
"step": 2035,
|
||
|
|
"valid_targets_mean": 1474.7,
|
||
|
|
"valid_targets_min": 260
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1578947368421053,
|
||
|
|
"grad_norm": 0.38179431322717006,
|
||
|
|
"learning_rate": 2.678797784911689e-05,
|
||
|
|
"loss": 0.2741,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2740272879600525,
|
||
|
|
"step": 2040,
|
||
|
|
"valid_targets_mean": 4764.6,
|
||
|
|
"valid_targets_min": 2478
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1656346749226008,
|
||
|
|
"grad_norm": 0.9479439463673041,
|
||
|
|
"learning_rate": 2.6715302340680863e-05,
|
||
|
|
"loss": 0.3832,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5201065540313721,
|
||
|
|
"step": 2045,
|
||
|
|
"valid_targets_mean": 905.6,
|
||
|
|
"valid_targets_min": 532
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.173374613003096,
|
||
|
|
"grad_norm": 0.30597801360783256,
|
||
|
|
"learning_rate": 2.664252675636763e-05,
|
||
|
|
"loss": 0.2728,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2585655450820923,
|
||
|
|
"step": 2050,
|
||
|
|
"valid_targets_mean": 4740.0,
|
||
|
|
"valid_targets_min": 2461
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.181114551083591,
|
||
|
|
"grad_norm": 0.6137377992368781,
|
||
|
|
"learning_rate": 2.6569652180727096e-05,
|
||
|
|
"loss": 0.3638,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4728262424468994,
|
||
|
|
"step": 2055,
|
||
|
|
"valid_targets_mean": 878.2,
|
||
|
|
"valid_targets_min": 159
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1888544891640866,
|
||
|
|
"grad_norm": 0.32945637544943984,
|
||
|
|
"learning_rate": 2.6496679699784376e-05,
|
||
|
|
"loss": 0.312,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2642161250114441,
|
||
|
|
"step": 2060,
|
||
|
|
"valid_targets_mean": 4672.1,
|
||
|
|
"valid_targets_min": 2558
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.196594427244582,
|
||
|
|
"grad_norm": 0.5593566963415534,
|
||
|
|
"learning_rate": 2.6423610401023662e-05,
|
||
|
|
"loss": 0.4001,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46090978384017944,
|
||
|
|
"step": 2065,
|
||
|
|
"valid_targets_mean": 998.0,
|
||
|
|
"valid_targets_min": 390
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2043343653250775,
|
||
|
|
"grad_norm": 0.28497657143593647,
|
||
|
|
"learning_rate": 2.635044537337196e-05,
|
||
|
|
"loss": 0.2771,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2618706524372101,
|
||
|
|
"step": 2070,
|
||
|
|
"valid_targets_mean": 4944.3,
|
||
|
|
"valid_targets_min": 3172
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.212074303405573,
|
||
|
|
"grad_norm": 0.8514008375510441,
|
||
|
|
"learning_rate": 2.6277185707182917e-05,
|
||
|
|
"loss": 0.3596,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.45843955874443054,
|
||
|
|
"step": 2075,
|
||
|
|
"valid_targets_mean": 983.4,
|
||
|
|
"valid_targets_min": 297
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.219814241486068,
|
||
|
|
"grad_norm": 0.2643279770220927,
|
||
|
|
"learning_rate": 2.620383249422053e-05,
|
||
|
|
"loss": 0.3426,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3013157248497009,
|
||
|
|
"step": 2080,
|
||
|
|
"valid_targets_mean": 5441.6,
|
||
|
|
"valid_targets_min": 628
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2275541795665634,
|
||
|
|
"grad_norm": 0.24756362863422254,
|
||
|
|
"learning_rate": 2.6130386827642894e-05,
|
||
|
|
"loss": 0.3042,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3151610195636749,
|
||
|
|
"step": 2085,
|
||
|
|
"valid_targets_mean": 5756.2,
|
||
|
|
"valid_targets_min": 530
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.235294117647059,
|
||
|
|
"grad_norm": 0.22296485931533286,
|
||
|
|
"learning_rate": 2.6056849801985914e-05,
|
||
|
|
"loss": 0.3,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.286399245262146,
|
||
|
|
"step": 2090,
|
||
|
|
"valid_targets_mean": 5889.1,
|
||
|
|
"valid_targets_min": 741
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2430340557275543,
|
||
|
|
"grad_norm": 0.21494967510647828,
|
||
|
|
"learning_rate": 2.5983222513146995e-05,
|
||
|
|
"loss": 0.3038,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3223365545272827,
|
||
|
|
"step": 2095,
|
||
|
|
"valid_targets_mean": 5961.3,
|
||
|
|
"valid_targets_min": 958
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2507739938080498,
|
||
|
|
"grad_norm": 0.20016387351953638,
|
||
|
|
"learning_rate": 2.5909506058368677e-05,
|
||
|
|
"loss": 0.2951,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30144166946411133,
|
||
|
|
"step": 2100,
|
||
|
|
"valid_targets_mean": 5714.0,
|
||
|
|
"valid_targets_min": 801
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2585139318885448,
|
||
|
|
"grad_norm": 0.22123613093372393,
|
||
|
|
"learning_rate": 2.5835701536222342e-05,
|
||
|
|
"loss": 0.2975,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30729299783706665,
|
||
|
|
"step": 2105,
|
||
|
|
"valid_targets_mean": 5337.6,
|
||
|
|
"valid_targets_min": 408
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.26625386996904,
|
||
|
|
"grad_norm": 0.2127231104377584,
|
||
|
|
"learning_rate": 2.576181004659179e-05,
|
||
|
|
"loss": 0.3007,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2999975085258484,
|
||
|
|
"step": 2110,
|
||
|
|
"valid_targets_mean": 5438.8,
|
||
|
|
"valid_targets_min": 895
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2739938080495357,
|
||
|
|
"grad_norm": 0.21486183155251595,
|
||
|
|
"learning_rate": 2.5687832690656875e-05,
|
||
|
|
"loss": 0.2898,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3005903363227844,
|
||
|
|
"step": 2115,
|
||
|
|
"valid_targets_mean": 5660.5,
|
||
|
|
"valid_targets_min": 541
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.281733746130031,
|
||
|
|
"grad_norm": 0.19699008434077442,
|
||
|
|
"learning_rate": 2.561377057087709e-05,
|
||
|
|
"loss": 0.2947,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2977728843688965,
|
||
|
|
"step": 2120,
|
||
|
|
"valid_targets_mean": 5774.7,
|
||
|
|
"valid_targets_min": 547
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2894736842105265,
|
||
|
|
"grad_norm": 0.18701344681295348,
|
||
|
|
"learning_rate": 2.5539624790975125e-05,
|
||
|
|
"loss": 0.294,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28735631704330444,
|
||
|
|
"step": 2125,
|
||
|
|
"valid_targets_mean": 5829.5,
|
||
|
|
"valid_targets_min": 628
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2972136222910216,
|
||
|
|
"grad_norm": 0.20856754132693325,
|
||
|
|
"learning_rate": 2.5465396455920428e-05,
|
||
|
|
"loss": 0.2938,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2794041931629181,
|
||
|
|
"step": 2130,
|
||
|
|
"valid_targets_mean": 5642.9,
|
||
|
|
"valid_targets_min": 252
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.304953560371517,
|
||
|
|
"grad_norm": 0.21326325421872644,
|
||
|
|
"learning_rate": 2.5391086671912765e-05,
|
||
|
|
"loss": 0.2884,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2905271649360657,
|
||
|
|
"step": 2135,
|
||
|
|
"valid_targets_mean": 6218.2,
|
||
|
|
"valid_targets_min": 273
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3126934984520124,
|
||
|
|
"grad_norm": 0.1935923560586963,
|
||
|
|
"learning_rate": 2.5316696546365674e-05,
|
||
|
|
"loss": 0.2968,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29446861147880554,
|
||
|
|
"step": 2140,
|
||
|
|
"valid_targets_mean": 5509.6,
|
||
|
|
"valid_targets_min": 897
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.320433436532508,
|
||
|
|
"grad_norm": 0.20447572380000706,
|
||
|
|
"learning_rate": 2.5242227187890008e-05,
|
||
|
|
"loss": 0.2847,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27835071086883545,
|
||
|
|
"step": 2145,
|
||
|
|
"valid_targets_mean": 5951.1,
|
||
|
|
"valid_targets_min": 625
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3281733746130033,
|
||
|
|
"grad_norm": 0.1970968487755153,
|
||
|
|
"learning_rate": 2.516767970627741e-05,
|
||
|
|
"loss": 0.295,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28273922204971313,
|
||
|
|
"step": 2150,
|
||
|
|
"valid_targets_mean": 5832.3,
|
||
|
|
"valid_targets_min": 642
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3359133126934983,
|
||
|
|
"grad_norm": 0.2188915881932126,
|
||
|
|
"learning_rate": 2.5093055212483757e-05,
|
||
|
|
"loss": 0.2884,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28982633352279663,
|
||
|
|
"step": 2155,
|
||
|
|
"valid_targets_mean": 6418.6,
|
||
|
|
"valid_targets_min": 475
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.343653250773994,
|
||
|
|
"grad_norm": 0.18790031582217906,
|
||
|
|
"learning_rate": 2.5018354818612615e-05,
|
||
|
|
"loss": 0.295,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2886435389518738,
|
||
|
|
"step": 2160,
|
||
|
|
"valid_targets_mean": 6394.8,
|
||
|
|
"valid_targets_min": 868
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3513931888544892,
|
||
|
|
"grad_norm": 0.22674808208203867,
|
||
|
|
"learning_rate": 2.4943579637898668e-05,
|
||
|
|
"loss": 0.3932,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2859913110733032,
|
||
|
|
"step": 2165,
|
||
|
|
"valid_targets_mean": 6349.0,
|
||
|
|
"valid_targets_min": 242
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3591331269349847,
|
||
|
|
"grad_norm": 0.47306572073094305,
|
||
|
|
"learning_rate": 2.4868730784691105e-05,
|
||
|
|
"loss": 0.2823,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2770742177963257,
|
||
|
|
"step": 2170,
|
||
|
|
"valid_targets_mean": 6417.1,
|
||
|
|
"valid_targets_min": 691
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.36687306501548,
|
||
|
|
"grad_norm": 0.1930232964245055,
|
||
|
|
"learning_rate": 2.479380937443705e-05,
|
||
|
|
"loss": 0.3776,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.291863352060318,
|
||
|
|
"step": 2175,
|
||
|
|
"valid_targets_mean": 7150.5,
|
||
|
|
"valid_targets_min": 1264
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.374613003095975,
|
||
|
|
"grad_norm": 0.7616879880042922,
|
||
|
|
"learning_rate": 2.4718816523664925e-05,
|
||
|
|
"loss": 0.3466,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5495944023132324,
|
||
|
|
"step": 2180,
|
||
|
|
"valid_targets_mean": 834.6,
|
||
|
|
"valid_targets_min": 421
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3823529411764706,
|
||
|
|
"grad_norm": 0.200054274335465,
|
||
|
|
"learning_rate": 2.4643753349967777e-05,
|
||
|
|
"loss": 0.3091,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29339897632598877,
|
||
|
|
"step": 2185,
|
||
|
|
"valid_targets_mean": 6313.1,
|
||
|
|
"valid_targets_min": 324
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.390092879256966,
|
||
|
|
"grad_norm": 0.7723738196288834,
|
||
|
|
"learning_rate": 2.456862097198668e-05,
|
||
|
|
"loss": 0.3542,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5526779294013977,
|
||
|
|
"step": 2190,
|
||
|
|
"valid_targets_mean": 796.4,
|
||
|
|
"valid_targets_min": 421
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3978328173374615,
|
||
|
|
"grad_norm": 0.20609983860967704,
|
||
|
|
"learning_rate": 2.4493420509394027e-05,
|
||
|
|
"loss": 0.34,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2773880064487457,
|
||
|
|
"step": 2195,
|
||
|
|
"valid_targets_mean": 6093.4,
|
||
|
|
"valid_targets_min": 238
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.405572755417957,
|
||
|
|
"grad_norm": 0.9519323619088927,
|
||
|
|
"learning_rate": 2.441815308287685e-05,
|
||
|
|
"loss": 0.3475,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5886904001235962,
|
||
|
|
"step": 2200,
|
||
|
|
"valid_targets_mean": 830.1,
|
||
|
|
"valid_targets_min": 365
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.413312693498452,
|
||
|
|
"grad_norm": 0.19313810368068124,
|
||
|
|
"learning_rate": 2.4342819814120112e-05,
|
||
|
|
"loss": 0.3122,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28064554929733276,
|
||
|
|
"step": 2205,
|
||
|
|
"valid_targets_mean": 6361.8,
|
||
|
|
"valid_targets_min": 544
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4210526315789473,
|
||
|
|
"grad_norm": 0.6543529924027404,
|
||
|
|
"learning_rate": 2.4267421825790027e-05,
|
||
|
|
"loss": 0.3988,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5514394640922546,
|
||
|
|
"step": 2210,
|
||
|
|
"valid_targets_mean": 802.3,
|
||
|
|
"valid_targets_min": 238
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.428792569659443,
|
||
|
|
"grad_norm": 0.20722717237366967,
|
||
|
|
"learning_rate": 2.419196024151728e-05,
|
||
|
|
"loss": 0.2892,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2848924398422241,
|
||
|
|
"step": 2215,
|
||
|
|
"valid_targets_mean": 6045.7,
|
||
|
|
"valid_targets_min": 317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4365325077399382,
|
||
|
|
"grad_norm": 0.9660788801642334,
|
||
|
|
"learning_rate": 2.411643618588032e-05,
|
||
|
|
"loss": 0.3988,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5620875358581543,
|
||
|
|
"step": 2220,
|
||
|
|
"valid_targets_mean": 778.7,
|
||
|
|
"valid_targets_min": 530
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4442724458204337,
|
||
|
|
"grad_norm": 0.20176515115664123,
|
||
|
|
"learning_rate": 2.4040850784388568e-05,
|
||
|
|
"loss": 0.2883,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28532230854034424,
|
||
|
|
"step": 2225,
|
||
|
|
"valid_targets_mean": 6171.3,
|
||
|
|
"valid_targets_min": 397
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4520123839009287,
|
||
|
|
"grad_norm": 0.588483355564494,
|
||
|
|
"learning_rate": 2.3965205163465683e-05,
|
||
|
|
"loss": 0.4104,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5912326574325562,
|
||
|
|
"step": 2230,
|
||
|
|
"valid_targets_mean": 944.8,
|
||
|
|
"valid_targets_min": 326
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.459752321981424,
|
||
|
|
"grad_norm": 0.3609237825565516,
|
||
|
|
"learning_rate": 2.3889500450432727e-05,
|
||
|
|
"loss": 0.4971,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4894105792045593,
|
||
|
|
"step": 2235,
|
||
|
|
"valid_targets_mean": 3094.3,
|
||
|
|
"valid_targets_min": 251
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4674922600619196,
|
||
|
|
"grad_norm": 0.442201899240148,
|
||
|
|
"learning_rate": 2.3813737773491426e-05,
|
||
|
|
"loss": 0.4975,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5362527966499329,
|
||
|
|
"step": 2240,
|
||
|
|
"valid_targets_mean": 1920.0,
|
||
|
|
"valid_targets_min": 260
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.475232198142415,
|
||
|
|
"grad_norm": 0.3427233322917521,
|
||
|
|
"learning_rate": 2.3737918261707283e-05,
|
||
|
|
"loss": 0.489,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.49932098388671875,
|
||
|
|
"step": 2245,
|
||
|
|
"valid_targets_mean": 2706.2,
|
||
|
|
"valid_targets_min": 244
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4829721362229105,
|
||
|
|
"grad_norm": 0.3563035591537088,
|
||
|
|
"learning_rate": 2.3662043044992827e-05,
|
||
|
|
"loss": 0.4763,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48159676790237427,
|
||
|
|
"step": 2250,
|
||
|
|
"valid_targets_mean": 2330.8,
|
||
|
|
"valid_targets_min": 245
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4907120743034055,
|
||
|
|
"grad_norm": 0.2902229754967154,
|
||
|
|
"learning_rate": 2.3586113254090724e-05,
|
||
|
|
"loss": 0.4856,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4765833020210266,
|
||
|
|
"step": 2255,
|
||
|
|
"valid_targets_mean": 3534.0,
|
||
|
|
"valid_targets_min": 323
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.498452012383901,
|
||
|
|
"grad_norm": 0.2945760890553162,
|
||
|
|
"learning_rate": 2.3510130020556937e-05,
|
||
|
|
"loss": 0.4665,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48470738530158997,
|
||
|
|
"step": 2260,
|
||
|
|
"valid_targets_mean": 3620.2,
|
||
|
|
"valid_targets_min": 335
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5061919504643964,
|
||
|
|
"grad_norm": 0.26115831646363025,
|
||
|
|
"learning_rate": 2.3434094476743865e-05,
|
||
|
|
"loss": 0.4789,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.45804664492607117,
|
||
|
|
"step": 2265,
|
||
|
|
"valid_targets_mean": 3821.6,
|
||
|
|
"valid_targets_min": 260
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.513931888544892,
|
||
|
|
"grad_norm": 0.2978855346411902,
|
||
|
|
"learning_rate": 2.335800775578348e-05,
|
||
|
|
"loss": 0.4746,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48575156927108765,
|
||
|
|
"step": 2270,
|
||
|
|
"valid_targets_mean": 3024.7,
|
||
|
|
"valid_targets_min": 354
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5216718266253872,
|
||
|
|
"grad_norm": 0.33111655279214086,
|
||
|
|
"learning_rate": 2.3281870991570425e-05,
|
||
|
|
"loss": 0.4673,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46261557936668396,
|
||
|
|
"step": 2275,
|
||
|
|
"valid_targets_mean": 2763.7,
|
||
|
|
"valid_targets_min": 227
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5294117647058822,
|
||
|
|
"grad_norm": 0.30026197614520794,
|
||
|
|
"learning_rate": 2.320568531874511e-05,
|
||
|
|
"loss": 0.4809,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4561736583709717,
|
||
|
|
"step": 2280,
|
||
|
|
"valid_targets_mean": 3144.2,
|
||
|
|
"valid_targets_min": 263
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5371517027863777,
|
||
|
|
"grad_norm": 0.3995092839933271,
|
||
|
|
"learning_rate": 2.3129451872676835e-05,
|
||
|
|
"loss": 0.4635,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4765763282775879,
|
||
|
|
"step": 2285,
|
||
|
|
"valid_targets_mean": 2372.6,
|
||
|
|
"valid_targets_min": 349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.544891640866873,
|
||
|
|
"grad_norm": 0.2763051246985775,
|
||
|
|
"learning_rate": 2.3053171789446832e-05,
|
||
|
|
"loss": 0.4758,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.44267189502716064,
|
||
|
|
"step": 2290,
|
||
|
|
"valid_targets_mean": 2830.0,
|
||
|
|
"valid_targets_min": 279
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5526315789473686,
|
||
|
|
"grad_norm": 0.2705585941320523,
|
||
|
|
"learning_rate": 2.2976846205831357e-05,
|
||
|
|
"loss": 0.4619,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4643164575099945,
|
||
|
|
"step": 2295,
|
||
|
|
"valid_targets_mean": 2977.7,
|
||
|
|
"valid_targets_min": 268
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.560371517027864,
|
||
|
|
"grad_norm": 0.24261510549979376,
|
||
|
|
"learning_rate": 2.290047625928474e-05,
|
||
|
|
"loss": 0.4688,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4681593179702759,
|
||
|
|
"step": 2300,
|
||
|
|
"valid_targets_mean": 3759.2,
|
||
|
|
"valid_targets_min": 275
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.568111455108359,
|
||
|
|
"grad_norm": 0.5673406867117412,
|
||
|
|
"learning_rate": 2.2824063087922436e-05,
|
||
|
|
"loss": 0.5696,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5672910213470459,
|
||
|
|
"step": 2305,
|
||
|
|
"valid_targets_mean": 4735.6,
|
||
|
|
"valid_targets_min": 1023
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5758513931888545,
|
||
|
|
"grad_norm": 0.2942112144701351,
|
||
|
|
"learning_rate": 2.2747607830504066e-05,
|
||
|
|
"loss": 0.5893,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5426371097564697,
|
||
|
|
"step": 2310,
|
||
|
|
"valid_targets_mean": 4437.0,
|
||
|
|
"valid_targets_min": 493
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.58359133126935,
|
||
|
|
"grad_norm": 0.28501048640424526,
|
||
|
|
"learning_rate": 2.267111162641645e-05,
|
||
|
|
"loss": 0.6193,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5572389960289001,
|
||
|
|
"step": 2315,
|
||
|
|
"valid_targets_mean": 4264.1,
|
||
|
|
"valid_targets_min": 970
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5913312693498454,
|
||
|
|
"grad_norm": 0.2922610739607367,
|
||
|
|
"learning_rate": 2.259457561565662e-05,
|
||
|
|
"loss": 0.5993,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6052597761154175,
|
||
|
|
"step": 2320,
|
||
|
|
"valid_targets_mean": 3503.9,
|
||
|
|
"valid_targets_min": 610
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.599071207430341,
|
||
|
|
"grad_norm": 0.24525289479476817,
|
||
|
|
"learning_rate": 2.2518000938814822e-05,
|
||
|
|
"loss": 0.5956,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6306358575820923,
|
||
|
|
"step": 2325,
|
||
|
|
"valid_targets_mean": 4163.0,
|
||
|
|
"valid_targets_min": 848
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.606811145510836,
|
||
|
|
"grad_norm": 0.2640499380752657,
|
||
|
|
"learning_rate": 2.244138873705755e-05,
|
||
|
|
"loss": 0.5977,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6389599442481995,
|
||
|
|
"step": 2330,
|
||
|
|
"valid_targets_mean": 3816.2,
|
||
|
|
"valid_targets_min": 861
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6145510835913313,
|
||
|
|
"grad_norm": 0.25693537778307857,
|
||
|
|
"learning_rate": 2.2364740152110508e-05,
|
||
|
|
"loss": 0.5869,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6063191890716553,
|
||
|
|
"step": 2335,
|
||
|
|
"valid_targets_mean": 4045.8,
|
||
|
|
"valid_targets_min": 341
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6222910216718267,
|
||
|
|
"grad_norm": 0.2824002640117761,
|
||
|
|
"learning_rate": 2.2288056326241594e-05,
|
||
|
|
"loss": 0.5745,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5906564593315125,
|
||
|
|
"step": 2340,
|
||
|
|
"valid_targets_mean": 3426.6,
|
||
|
|
"valid_targets_min": 1103
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6300309597523217,
|
||
|
|
"grad_norm": 0.2779802104429867,
|
||
|
|
"learning_rate": 2.2211338402243915e-05,
|
||
|
|
"loss": 0.5899,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5718535780906677,
|
||
|
|
"step": 2345,
|
||
|
|
"valid_targets_mean": 3857.1,
|
||
|
|
"valid_targets_min": 897
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6377708978328176,
|
||
|
|
"grad_norm": 0.2695545747850322,
|
||
|
|
"learning_rate": 2.213458752341872e-05,
|
||
|
|
"loss": 0.5778,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6076485514640808,
|
||
|
|
"step": 2350,
|
||
|
|
"valid_targets_mean": 3702.2,
|
||
|
|
"valid_targets_min": 1152
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6455108359133126,
|
||
|
|
"grad_norm": 0.23501901473473052,
|
||
|
|
"learning_rate": 2.2057804833558362e-05,
|
||
|
|
"loss": 0.5864,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5793353319168091,
|
||
|
|
"step": 2355,
|
||
|
|
"valid_targets_mean": 4113.9,
|
||
|
|
"valid_targets_min": 1093
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.653250773993808,
|
||
|
|
"grad_norm": 0.24513376998514685,
|
||
|
|
"learning_rate": 2.1980991476929274e-05,
|
||
|
|
"loss": 0.5982,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5926626920700073,
|
||
|
|
"step": 2360,
|
||
|
|
"valid_targets_mean": 4190.3,
|
||
|
|
"valid_targets_min": 328
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6609907120743035,
|
||
|
|
"grad_norm": 0.2177779208099969,
|
||
|
|
"learning_rate": 2.1904148598254902e-05,
|
||
|
|
"loss": 0.5849,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.591439962387085,
|
||
|
|
"step": 2365,
|
||
|
|
"valid_targets_mean": 4339.2,
|
||
|
|
"valid_targets_min": 1005
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6687306501547985,
|
||
|
|
"grad_norm": 0.2251458371882269,
|
||
|
|
"learning_rate": 2.182727734269866e-05,
|
||
|
|
"loss": 0.5966,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5952557325363159,
|
||
|
|
"step": 2370,
|
||
|
|
"valid_targets_mean": 4695.3,
|
||
|
|
"valid_targets_min": 1178
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6764705882352944,
|
||
|
|
"grad_norm": 0.22668390966728172,
|
||
|
|
"learning_rate": 2.1750378855846827e-05,
|
||
|
|
"loss": 0.5788,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5567156076431274,
|
||
|
|
"step": 2375,
|
||
|
|
"valid_targets_mean": 4633.9,
|
||
|
|
"valid_targets_min": 901
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6842105263157894,
|
||
|
|
"grad_norm": 0.27854092500061617,
|
||
|
|
"learning_rate": 2.1673454283691533e-05,
|
||
|
|
"loss": 0.523,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.47287312150001526,
|
||
|
|
"step": 2380,
|
||
|
|
"valid_targets_mean": 4319.0,
|
||
|
|
"valid_targets_min": 854
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.691950464396285,
|
||
|
|
"grad_norm": 0.2743136053610333,
|
||
|
|
"learning_rate": 2.1596504772613622e-05,
|
||
|
|
"loss": 0.4797,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4419233798980713,
|
||
|
|
"step": 2385,
|
||
|
|
"valid_targets_mean": 4809.1,
|
||
|
|
"valid_targets_min": 757
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6996904024767803,
|
||
|
|
"grad_norm": 0.46178389572420325,
|
||
|
|
"learning_rate": 2.1519531469365613e-05,
|
||
|
|
"loss": 0.5101,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4898723363876343,
|
||
|
|
"step": 2390,
|
||
|
|
"valid_targets_mean": 3717.0,
|
||
|
|
"valid_targets_min": 810
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7074303405572753,
|
||
|
|
"grad_norm": 0.24897330694012126,
|
||
|
|
"learning_rate": 2.144253552105458e-05,
|
||
|
|
"loss": 0.4717,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.40806204080581665,
|
||
|
|
"step": 2395,
|
||
|
|
"valid_targets_mean": 4199.5,
|
||
|
|
"valid_targets_min": 645
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.715170278637771,
|
||
|
|
"grad_norm": 0.2536128933284602,
|
||
|
|
"learning_rate": 2.136551807512507e-05,
|
||
|
|
"loss": 0.5017,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4504415988922119,
|
||
|
|
"step": 2400,
|
||
|
|
"valid_targets_mean": 3818.9,
|
||
|
|
"valid_targets_min": 926
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.722910216718266,
|
||
|
|
"grad_norm": 0.22951176860514835,
|
||
|
|
"learning_rate": 2.1288480279342006e-05,
|
||
|
|
"loss": 0.4626,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4290314316749573,
|
||
|
|
"step": 2405,
|
||
|
|
"valid_targets_mean": 4357.0,
|
||
|
|
"valid_targets_min": 617
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7306501547987616,
|
||
|
|
"grad_norm": 0.24553759063701874,
|
||
|
|
"learning_rate": 2.121142328177358e-05,
|
||
|
|
"loss": 0.5038,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4579368829727173,
|
||
|
|
"step": 2410,
|
||
|
|
"valid_targets_mean": 3918.5,
|
||
|
|
"valid_targets_min": 964
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.738390092879257,
|
||
|
|
"grad_norm": 0.23135776365550756,
|
||
|
|
"learning_rate": 2.1134348230774133e-05,
|
||
|
|
"loss": 0.472,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.44658517837524414,
|
||
|
|
"step": 2415,
|
||
|
|
"valid_targets_mean": 4063.1,
|
||
|
|
"valid_targets_min": 575
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.746130030959752,
|
||
|
|
"grad_norm": 0.2500055344912002,
|
||
|
|
"learning_rate": 2.1057256274967056e-05,
|
||
|
|
"loss": 0.4802,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4399147033691406,
|
||
|
|
"step": 2420,
|
||
|
|
"valid_targets_mean": 3417.5,
|
||
|
|
"valid_targets_min": 848
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.753869969040248,
|
||
|
|
"grad_norm": 0.2555411253834385,
|
||
|
|
"learning_rate": 2.0980148563227664e-05,
|
||
|
|
"loss": 0.4775,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.49702179431915283,
|
||
|
|
"step": 2425,
|
||
|
|
"valid_targets_mean": 3247.6,
|
||
|
|
"valid_targets_min": 441
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.761609907120743,
|
||
|
|
"grad_norm": 0.23311295181175976,
|
||
|
|
"learning_rate": 2.0903026244666084e-05,
|
||
|
|
"loss": 0.4861,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4511328339576721,
|
||
|
|
"step": 2430,
|
||
|
|
"valid_targets_mean": 3815.3,
|
||
|
|
"valid_targets_min": 766
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7693498452012384,
|
||
|
|
"grad_norm": 0.24825291652687764,
|
||
|
|
"learning_rate": 2.0825890468610108e-05,
|
||
|
|
"loss": 0.4854,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48870447278022766,
|
||
|
|
"step": 2435,
|
||
|
|
"valid_targets_mean": 3516.8,
|
||
|
|
"valid_targets_min": 553
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.777089783281734,
|
||
|
|
"grad_norm": 0.24344411954794534,
|
||
|
|
"learning_rate": 2.07487423845881e-05,
|
||
|
|
"loss": 0.4794,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.47710180282592773,
|
||
|
|
"step": 2440,
|
||
|
|
"valid_targets_mean": 3548.4,
|
||
|
|
"valid_targets_min": 741
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.784829721362229,
|
||
|
|
"grad_norm": 0.2630649508737684,
|
||
|
|
"learning_rate": 2.067158314231184e-05,
|
||
|
|
"loss": 0.4722,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.47597736120224,
|
||
|
|
"step": 2445,
|
||
|
|
"valid_targets_mean": 3108.8,
|
||
|
|
"valid_targets_min": 546
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7925696594427247,
|
||
|
|
"grad_norm": 0.24168466928553042,
|
||
|
|
"learning_rate": 2.059441389165938e-05,
|
||
|
|
"loss": 0.4939,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5191800594329834,
|
||
|
|
"step": 2450,
|
||
|
|
"valid_targets_mean": 3956.9,
|
||
|
|
"valid_targets_min": 809
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8003095975232197,
|
||
|
|
"grad_norm": 0.2681123708837574,
|
||
|
|
"learning_rate": 2.0517235782657945e-05,
|
||
|
|
"loss": 0.4811,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5111984610557556,
|
||
|
|
"step": 2455,
|
||
|
|
"valid_targets_mean": 3006.1,
|
||
|
|
"valid_targets_min": 506
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.808049535603715,
|
||
|
|
"grad_norm": 0.2753655099377223,
|
||
|
|
"learning_rate": 2.0440049965466762e-05,
|
||
|
|
"loss": 0.4756,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48138660192489624,
|
||
|
|
"step": 2460,
|
||
|
|
"valid_targets_mean": 3219.2,
|
||
|
|
"valid_targets_min": 574
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8157894736842106,
|
||
|
|
"grad_norm": 0.6090992070732313,
|
||
|
|
"learning_rate": 2.0362857590359925e-05,
|
||
|
|
"loss": 0.5443,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.9017142653465271,
|
||
|
|
"step": 2465,
|
||
|
|
"valid_targets_mean": 1110.0,
|
||
|
|
"valid_targets_min": 530
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8235294117647056,
|
||
|
|
"grad_norm": 0.2539597480038725,
|
||
|
|
"learning_rate": 2.0285659807709276e-05,
|
||
|
|
"loss": 0.4839,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4835926592350006,
|
||
|
|
"step": 2470,
|
||
|
|
"valid_targets_mean": 3346.5,
|
||
|
|
"valid_targets_min": 319
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8312693498452015,
|
||
|
|
"grad_norm": 0.5900638146840933,
|
||
|
|
"learning_rate": 2.0208457767967226e-05,
|
||
|
|
"loss": 0.5595,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.9306286573410034,
|
||
|
|
"step": 2475,
|
||
|
|
"valid_targets_mean": 1103.4,
|
||
|
|
"valid_targets_min": 335
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8390092879256965,
|
||
|
|
"grad_norm": 0.24533912534710758,
|
||
|
|
"learning_rate": 2.0131252621649637e-05,
|
||
|
|
"loss": 0.4813,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46409204602241516,
|
||
|
|
"step": 2480,
|
||
|
|
"valid_targets_mean": 3436.7,
|
||
|
|
"valid_targets_min": 820
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.846749226006192,
|
||
|
|
"grad_norm": 0.3835198272780034,
|
||
|
|
"learning_rate": 2.0054045519318663e-05,
|
||
|
|
"loss": 0.5031,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6639645099639893,
|
||
|
|
"step": 2485,
|
||
|
|
"valid_targets_mean": 1655.0,
|
||
|
|
"valid_targets_min": 539
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8544891640866874,
|
||
|
|
"grad_norm": 0.24591656363796577,
|
||
|
|
"learning_rate": 1.9976837611565617e-05,
|
||
|
|
"loss": 0.4911,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.44083571434020996,
|
||
|
|
"step": 2490,
|
||
|
|
"valid_targets_mean": 3273.7,
|
||
|
|
"valid_targets_min": 980
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8622291021671824,
|
||
|
|
"grad_norm": 0.34461073117062446,
|
||
|
|
"learning_rate": 1.9899630048993802e-05,
|
||
|
|
"loss": 0.4883,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.57768714427948,
|
||
|
|
"step": 2495,
|
||
|
|
"valid_targets_mean": 2103.5,
|
||
|
|
"valid_targets_min": 401
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8699690402476783,
|
||
|
|
"grad_norm": 0.2715672112578731,
|
||
|
|
"learning_rate": 1.9822423982201382e-05,
|
||
|
|
"loss": 0.5051,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4588810205459595,
|
||
|
|
"step": 2500,
|
||
|
|
"valid_targets_mean": 3329.1,
|
||
|
|
"valid_targets_min": 306
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8777089783281733,
|
||
|
|
"grad_norm": 0.32183856191089943,
|
||
|
|
"learning_rate": 1.974522056176423e-05,
|
||
|
|
"loss": 0.4818,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5170823335647583,
|
||
|
|
"step": 2505,
|
||
|
|
"valid_targets_mean": 2073.9,
|
||
|
|
"valid_targets_min": 410
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8854489164086687,
|
||
|
|
"grad_norm": 0.2620503767469688,
|
||
|
|
"learning_rate": 1.9668020938218795e-05,
|
||
|
|
"loss": 0.5593,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.45399680733680725,
|
||
|
|
"step": 2510,
|
||
|
|
"valid_targets_mean": 3753.7,
|
||
|
|
"valid_targets_min": 1155
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.893188854489164,
|
||
|
|
"grad_norm": 0.23932474806140933,
|
||
|
|
"learning_rate": 1.959082626204492e-05,
|
||
|
|
"loss": 0.4547,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4369451105594635,
|
||
|
|
"step": 2515,
|
||
|
|
"valid_targets_mean": 3801.1,
|
||
|
|
"valid_targets_min": 296
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.900928792569659,
|
||
|
|
"grad_norm": 0.252485093333616,
|
||
|
|
"learning_rate": 1.9513637683648725e-05,
|
||
|
|
"loss": 0.5726,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48154330253601074,
|
||
|
|
"step": 2520,
|
||
|
|
"valid_targets_mean": 3548.1,
|
||
|
|
"valid_targets_min": 1109
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9086687306501546,
|
||
|
|
"grad_norm": 0.22517768636491495,
|
||
|
|
"learning_rate": 1.9436456353345468e-05,
|
||
|
|
"loss": 0.4609,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4374823570251465,
|
||
|
|
"step": 2525,
|
||
|
|
"valid_targets_mean": 4182.1,
|
||
|
|
"valid_targets_min": 329
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.91640866873065,
|
||
|
|
"grad_norm": 0.255208759251232,
|
||
|
|
"learning_rate": 1.935928342134238e-05,
|
||
|
|
"loss": 0.5195,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.47102975845336914,
|
||
|
|
"step": 2530,
|
||
|
|
"valid_targets_mean": 3701.5,
|
||
|
|
"valid_targets_min": 424
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9241486068111455,
|
||
|
|
"grad_norm": 0.23342082990079394,
|
||
|
|
"learning_rate": 1.928212003772153e-05,
|
||
|
|
"loss": 0.4647,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4411478638648987,
|
||
|
|
"step": 2535,
|
||
|
|
"valid_targets_mean": 3829.9,
|
||
|
|
"valid_targets_min": 252
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.931888544891641,
|
||
|
|
"grad_norm": 0.6593784063280745,
|
||
|
|
"learning_rate": 1.9204967352422708e-05,
|
||
|
|
"loss": 0.3867,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27076905965805054,
|
||
|
|
"step": 2540,
|
||
|
|
"valid_targets_mean": 4718.9,
|
||
|
|
"valid_targets_min": 1551
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.939628482972136,
|
||
|
|
"grad_norm": 0.4321015501415171,
|
||
|
|
"learning_rate": 1.9127826515226267e-05,
|
||
|
|
"loss": 0.2552,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23807236552238464,
|
||
|
|
"step": 2545,
|
||
|
|
"valid_targets_mean": 5331.9,
|
||
|
|
"valid_targets_min": 1956
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9473684210526314,
|
||
|
|
"grad_norm": 0.4696404575331375,
|
||
|
|
"learning_rate": 1.9050698675735982e-05,
|
||
|
|
"loss": 0.2405,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2161053717136383,
|
||
|
|
"step": 2550,
|
||
|
|
"valid_targets_mean": 5134.1,
|
||
|
|
"valid_targets_min": 999
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.955108359133127,
|
||
|
|
"grad_norm": 0.4256206392348206,
|
||
|
|
"learning_rate": 1.897358498336192e-05,
|
||
|
|
"loss": 0.2231,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22363455593585968,
|
||
|
|
"step": 2555,
|
||
|
|
"valid_targets_mean": 4941.0,
|
||
|
|
"valid_targets_min": 1327
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9628482972136223,
|
||
|
|
"grad_norm": 0.34772406326665967,
|
||
|
|
"learning_rate": 1.889648658730335e-05,
|
||
|
|
"loss": 0.2414,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24909768998622894,
|
||
|
|
"step": 2560,
|
||
|
|
"valid_targets_mean": 4170.8,
|
||
|
|
"valid_targets_min": 887
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9705882352941178,
|
||
|
|
"grad_norm": 0.579956181654713,
|
||
|
|
"learning_rate": 1.8819404636531547e-05,
|
||
|
|
"loss": 0.2435,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2532193660736084,
|
||
|
|
"step": 2565,
|
||
|
|
"valid_targets_mean": 4252.7,
|
||
|
|
"valid_targets_min": 964
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9783281733746128,
|
||
|
|
"grad_norm": 0.7392339246968006,
|
||
|
|
"learning_rate": 1.874234027977274e-05,
|
||
|
|
"loss": 0.236,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.243461012840271,
|
||
|
|
"step": 2570,
|
||
|
|
"valid_targets_mean": 3253.8,
|
||
|
|
"valid_targets_min": 229
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.986068111455108,
|
||
|
|
"grad_norm": 0.3907545308997725,
|
||
|
|
"learning_rate": 1.8665294665490947e-05,
|
||
|
|
"loss": 0.2325,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23020660877227783,
|
||
|
|
"step": 2575,
|
||
|
|
"valid_targets_mean": 2460.5,
|
||
|
|
"valid_targets_min": 592
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9938080495356036,
|
||
|
|
"grad_norm": 0.3366995459243321,
|
||
|
|
"learning_rate": 1.8588268941870854e-05,
|
||
|
|
"loss": 0.2255,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22101302444934845,
|
||
|
|
"step": 2580,
|
||
|
|
"valid_targets_mean": 3589.9,
|
||
|
|
"valid_targets_min": 1104
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.001547987616099,
|
||
|
|
"grad_norm": 0.8108085991673352,
|
||
|
|
"learning_rate": 1.8511264256800764e-05,
|
||
|
|
"loss": 0.2314,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26870933175086975,
|
||
|
|
"step": 2585,
|
||
|
|
"valid_targets_mean": 3681.9,
|
||
|
|
"valid_targets_min": 433
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.0092879256965945,
|
||
|
|
"grad_norm": 0.3888707848990401,
|
||
|
|
"learning_rate": 1.8434281757855417e-05,
|
||
|
|
"loss": 0.2556,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25091439485549927,
|
||
|
|
"step": 2590,
|
||
|
|
"valid_targets_mean": 3970.2,
|
||
|
|
"valid_targets_min": 409
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.0170278637770895,
|
||
|
|
"grad_norm": 0.33143335924270506,
|
||
|
|
"learning_rate": 1.835732259227892e-05,
|
||
|
|
"loss": 0.2475,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24228446185588837,
|
||
|
|
"step": 2595,
|
||
|
|
"valid_targets_mean": 3712.5,
|
||
|
|
"valid_targets_min": 478
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.024767801857585,
|
||
|
|
"grad_norm": 0.3270764771550791,
|
||
|
|
"learning_rate": 1.828038790696768e-05,
|
||
|
|
"loss": 0.2436,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24327068030834198,
|
||
|
|
"step": 2600,
|
||
|
|
"valid_targets_mean": 3618.2,
|
||
|
|
"valid_targets_min": 391
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.03250773993808,
|
||
|
|
"grad_norm": 0.2722892758849342,
|
||
|
|
"learning_rate": 1.820347884845326e-05,
|
||
|
|
"loss": 0.2392,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23661786317825317,
|
||
|
|
"step": 2605,
|
||
|
|
"valid_targets_mean": 3822.5,
|
||
|
|
"valid_targets_min": 431
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.040247678018575,
|
||
|
|
"grad_norm": 0.23652163554329414,
|
||
|
|
"learning_rate": 1.8126596562885322e-05,
|
||
|
|
"loss": 0.235,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23628389835357666,
|
||
|
|
"step": 2610,
|
||
|
|
"valid_targets_mean": 3878.1,
|
||
|
|
"valid_targets_min": 448
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.047987616099071,
|
||
|
|
"grad_norm": 0.22654062303440944,
|
||
|
|
"learning_rate": 1.804974219601453e-05,
|
||
|
|
"loss": 0.2338,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23388496041297913,
|
||
|
|
"step": 2615,
|
||
|
|
"valid_targets_mean": 3599.6,
|
||
|
|
"valid_targets_min": 289
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.055727554179566,
|
||
|
|
"grad_norm": 0.2286808775375755,
|
||
|
|
"learning_rate": 1.7972916893175496e-05,
|
||
|
|
"loss": 0.2317,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2317322939634323,
|
||
|
|
"step": 2620,
|
||
|
|
"valid_targets_mean": 3547.5,
|
||
|
|
"valid_targets_min": 421
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.063467492260062,
|
||
|
|
"grad_norm": 0.23333158980389443,
|
||
|
|
"learning_rate": 1.789612179926969e-05,
|
||
|
|
"loss": 0.2273,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22209343314170837,
|
||
|
|
"step": 2625,
|
||
|
|
"valid_targets_mean": 3543.9,
|
||
|
|
"valid_targets_min": 402
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.071207430340557,
|
||
|
|
"grad_norm": 0.2116870281954567,
|
||
|
|
"learning_rate": 1.781935805874842e-05,
|
||
|
|
"loss": 0.23,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22863523662090302,
|
||
|
|
"step": 2630,
|
||
|
|
"valid_targets_mean": 3743.6,
|
||
|
|
"valid_targets_min": 369
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.078947368421052,
|
||
|
|
"grad_norm": 0.2040024460892695,
|
||
|
|
"learning_rate": 1.7742626815595702e-05,
|
||
|
|
"loss": 0.2295,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2306627780199051,
|
||
|
|
"step": 2635,
|
||
|
|
"valid_targets_mean": 3927.4,
|
||
|
|
"valid_targets_min": 377
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.086687306501548,
|
||
|
|
"grad_norm": 0.3910296629790183,
|
||
|
|
"learning_rate": 1.7665929213311274e-05,
|
||
|
|
"loss": 0.2384,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2606777250766754,
|
||
|
|
"step": 2640,
|
||
|
|
"valid_targets_mean": 4279.9,
|
||
|
|
"valid_targets_min": 478
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.094427244582043,
|
||
|
|
"grad_norm": 0.2935733981334276,
|
||
|
|
"learning_rate": 1.758926639489354e-05,
|
||
|
|
"loss": 0.2714,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26189124584198,
|
||
|
|
"step": 2645,
|
||
|
|
"valid_targets_mean": 4793.1,
|
||
|
|
"valid_targets_min": 2022
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.102167182662539,
|
||
|
|
"grad_norm": 0.6318856638450114,
|
||
|
|
"learning_rate": 1.7512639502822534e-05,
|
||
|
|
"loss": 0.3866,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5014207363128662,
|
||
|
|
"step": 2650,
|
||
|
|
"valid_targets_mean": 894.3,
|
||
|
|
"valid_targets_min": 364
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.109907120743034,
|
||
|
|
"grad_norm": 0.29138566436288815,
|
||
|
|
"learning_rate": 1.743604967904287e-05,
|
||
|
|
"loss": 0.2679,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27116087079048157,
|
||
|
|
"step": 2655,
|
||
|
|
"valid_targets_mean": 4972.6,
|
||
|
|
"valid_targets_min": 3047
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.117647058823529,
|
||
|
|
"grad_norm": 0.3813369506421943,
|
||
|
|
"learning_rate": 1.735949806494677e-05,
|
||
|
|
"loss": 0.3597,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2658672332763672,
|
||
|
|
"step": 2660,
|
||
|
|
"valid_targets_mean": 4199.6,
|
||
|
|
"valid_targets_min": 611
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.125386996904025,
|
||
|
|
"grad_norm": 0.26180262358058815,
|
||
|
|
"learning_rate": 1.7282985801357024e-05,
|
||
|
|
"loss": 0.2631,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2672886252403259,
|
||
|
|
"step": 2665,
|
||
|
|
"valid_targets_mean": 4869.9,
|
||
|
|
"valid_targets_min": 2215
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.13312693498452,
|
||
|
|
"grad_norm": 0.579909263714248,
|
||
|
|
"learning_rate": 1.7206514028509988e-05,
|
||
|
|
"loss": 0.4053,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5239918231964111,
|
||
|
|
"step": 2670,
|
||
|
|
"valid_targets_mean": 865.0,
|
||
|
|
"valid_targets_min": 363
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.140866873065016,
|
||
|
|
"grad_norm": 0.25664432553870914,
|
||
|
|
"learning_rate": 1.71300838860386e-05,
|
||
|
|
"loss": 0.2662,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2669444680213928,
|
||
|
|
"step": 2675,
|
||
|
|
"valid_targets_mean": 4827.9,
|
||
|
|
"valid_targets_min": 3031
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.148606811145511,
|
||
|
|
"grad_norm": 0.62825485606549,
|
||
|
|
"learning_rate": 1.70536965129554e-05,
|
||
|
|
"loss": 0.3633,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5174235701560974,
|
||
|
|
"step": 2680,
|
||
|
|
"valid_targets_mean": 951.7,
|
||
|
|
"valid_targets_min": 325
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.156346749226006,
|
||
|
|
"grad_norm": 0.23132536550337102,
|
||
|
|
"learning_rate": 1.697735304763554e-05,
|
||
|
|
"loss": 0.2877,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26538655161857605,
|
||
|
|
"step": 2685,
|
||
|
|
"valid_targets_mean": 4826.3,
|
||
|
|
"valid_targets_min": 2645
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.164086687306502,
|
||
|
|
"grad_norm": 0.8716004891910766,
|
||
|
|
"learning_rate": 1.6901054627799845e-05,
|
||
|
|
"loss": 0.3225,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5482535362243652,
|
||
|
|
"step": 2690,
|
||
|
|
"valid_targets_mean": 823.9,
|
||
|
|
"valid_targets_min": 430
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.171826625386997,
|
||
|
|
"grad_norm": 0.22190804228550204,
|
||
|
|
"learning_rate": 1.682480239049783e-05,
|
||
|
|
"loss": 0.3129,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2635763883590698,
|
||
|
|
"step": 2695,
|
||
|
|
"valid_targets_mean": 4833.1,
|
||
|
|
"valid_targets_min": 2940
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.179566563467493,
|
||
|
|
"grad_norm": 0.7110633405876773,
|
||
|
|
"learning_rate": 1.674859747209077e-05,
|
||
|
|
"loss": 0.3108,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5079576969146729,
|
||
|
|
"step": 2700,
|
||
|
|
"valid_targets_mean": 755.9,
|
||
|
|
"valid_targets_min": 264
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.187306501547988,
|
||
|
|
"grad_norm": 0.3612147194067334,
|
||
|
|
"learning_rate": 1.6672441008234768e-05,
|
||
|
|
"loss": 0.3409,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26029324531555176,
|
||
|
|
"step": 2705,
|
||
|
|
"valid_targets_mean": 4627.0,
|
||
|
|
"valid_targets_min": 841
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.195046439628483,
|
||
|
|
"grad_norm": 1.302432252453704,
|
||
|
|
"learning_rate": 1.6596334133863822e-05,
|
||
|
|
"loss": 0.3491,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5193295478820801,
|
||
|
|
"step": 2710,
|
||
|
|
"valid_targets_mean": 851.7,
|
||
|
|
"valid_targets_min": 286
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.2027863777089784,
|
||
|
|
"grad_norm": 0.23433616507023192,
|
||
|
|
"learning_rate": 1.6520277983172892e-05,
|
||
|
|
"loss": 0.3068,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.273078978061676,
|
||
|
|
"step": 2715,
|
||
|
|
"valid_targets_mean": 4954.3,
|
||
|
|
"valid_targets_min": 2940
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.2105263157894735,
|
||
|
|
"grad_norm": 0.6670286123083791,
|
||
|
|
"learning_rate": 1.6444273689601066e-05,
|
||
|
|
"loss": 0.3106,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.49370327591896057,
|
||
|
|
"step": 2720,
|
||
|
|
"valid_targets_mean": 830.1,
|
||
|
|
"valid_targets_min": 291
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.218266253869969,
|
||
|
|
"grad_norm": 0.27280755147591246,
|
||
|
|
"learning_rate": 1.6368322385814597e-05,
|
||
|
|
"loss": 0.3614,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28182387351989746,
|
||
|
|
"step": 2725,
|
||
|
|
"valid_targets_mean": 5988.3,
|
||
|
|
"valid_targets_min": 536
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.226006191950464,
|
||
|
|
"grad_norm": 0.25219811993073077,
|
||
|
|
"learning_rate": 1.629242520369004e-05,
|
||
|
|
"loss": 0.2932,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3066892623901367,
|
||
|
|
"step": 2730,
|
||
|
|
"valid_targets_mean": 5611.2,
|
||
|
|
"valid_targets_min": 524
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.233746130030959,
|
||
|
|
"grad_norm": 0.23505777627724175,
|
||
|
|
"learning_rate": 1.6216583274297407e-05,
|
||
|
|
"loss": 0.2975,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29743415117263794,
|
||
|
|
"step": 2735,
|
||
|
|
"valid_targets_mean": 5332.3,
|
||
|
|
"valid_targets_min": 1127
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.241486068111455,
|
||
|
|
"grad_norm": 0.23374493818782638,
|
||
|
|
"learning_rate": 1.61407977278833e-05,
|
||
|
|
"loss": 0.2888,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3100311756134033,
|
||
|
|
"step": 2740,
|
||
|
|
"valid_targets_mean": 5595.0,
|
||
|
|
"valid_targets_min": 484
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.24922600619195,
|
||
|
|
"grad_norm": 0.2293031545656493,
|
||
|
|
"learning_rate": 1.6065069693854044e-05,
|
||
|
|
"loss": 0.2914,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2857232689857483,
|
||
|
|
"step": 2745,
|
||
|
|
"valid_targets_mean": 5674.6,
|
||
|
|
"valid_targets_min": 566
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.256965944272446,
|
||
|
|
"grad_norm": 0.19540620188606414,
|
||
|
|
"learning_rate": 1.5989400300758912e-05,
|
||
|
|
"loss": 0.2886,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27938780188560486,
|
||
|
|
"step": 2750,
|
||
|
|
"valid_targets_mean": 5872.0,
|
||
|
|
"valid_targets_min": 1577
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.264705882352941,
|
||
|
|
"grad_norm": 0.19615010314893858,
|
||
|
|
"learning_rate": 1.5913790676273236e-05,
|
||
|
|
"loss": 0.2943,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2912869155406952,
|
||
|
|
"step": 2755,
|
||
|
|
"valid_targets_mean": 5701.9,
|
||
|
|
"valid_targets_min": 1121
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.272445820433436,
|
||
|
|
"grad_norm": 0.21611328192226884,
|
||
|
|
"learning_rate": 1.5838241947181656e-05,
|
||
|
|
"loss": 0.2826,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26908057928085327,
|
||
|
|
"step": 2760,
|
||
|
|
"valid_targets_mean": 5619.1,
|
||
|
|
"valid_targets_min": 611
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.280185758513932,
|
||
|
|
"grad_norm": 0.19729929718412653,
|
||
|
|
"learning_rate": 1.5762755239361304e-05,
|
||
|
|
"loss": 0.288,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27383720874786377,
|
||
|
|
"step": 2765,
|
||
|
|
"valid_targets_mean": 5753.0,
|
||
|
|
"valid_targets_min": 583
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.287925696594427,
|
||
|
|
"grad_norm": 0.2316974734370182,
|
||
|
|
"learning_rate": 1.568733167776502e-05,
|
||
|
|
"loss": 0.2887,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27816247940063477,
|
||
|
|
"step": 2770,
|
||
|
|
"valid_targets_mean": 5811.5,
|
||
|
|
"valid_targets_min": 559
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.295665634674923,
|
||
|
|
"grad_norm": 0.2043005550910197,
|
||
|
|
"learning_rate": 1.561197238640462e-05,
|
||
|
|
"loss": 0.2881,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2836746275424957,
|
||
|
|
"step": 2775,
|
||
|
|
"valid_targets_mean": 5628.8,
|
||
|
|
"valid_targets_min": 299
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.303405572755418,
|
||
|
|
"grad_norm": 0.2096414594089389,
|
||
|
|
"learning_rate": 1.553667848833409e-05,
|
||
|
|
"loss": 0.2791,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2772459387779236,
|
||
|
|
"step": 2780,
|
||
|
|
"valid_targets_mean": 5455.2,
|
||
|
|
"valid_targets_min": 767
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.311145510835913,
|
||
|
|
"grad_norm": 0.20928840498048162,
|
||
|
|
"learning_rate": 1.5461451105632923e-05,
|
||
|
|
"loss": 0.2892,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2892909646034241,
|
||
|
|
"step": 2785,
|
||
|
|
"valid_targets_mean": 5440.1,
|
||
|
|
"valid_targets_min": 1058
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.318885448916409,
|
||
|
|
"grad_norm": 0.19642096425583394,
|
||
|
|
"learning_rate": 1.5386291359389324e-05,
|
||
|
|
"loss": 0.2811,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.280448853969574,
|
||
|
|
"step": 2790,
|
||
|
|
"valid_targets_mean": 5518.4,
|
||
|
|
"valid_targets_min": 521
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.326625386996904,
|
||
|
|
"grad_norm": 0.46248419164739263,
|
||
|
|
"learning_rate": 1.5311200369683542e-05,
|
||
|
|
"loss": 0.2872,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29715824127197266,
|
||
|
|
"step": 2795,
|
||
|
|
"valid_targets_mean": 5511.0,
|
||
|
|
"valid_targets_min": 494
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.3343653250774,
|
||
|
|
"grad_norm": 0.19011956617776618,
|
||
|
|
"learning_rate": 1.523617925557118e-05,
|
||
|
|
"loss": 0.2803,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2842664122581482,
|
||
|
|
"step": 2800,
|
||
|
|
"valid_targets_mean": 6648.1,
|
||
|
|
"valid_targets_min": 462
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.342105263157895,
|
||
|
|
"grad_norm": 0.17634824850125916,
|
||
|
|
"learning_rate": 1.5161229135066496e-05,
|
||
|
|
"loss": 0.2883,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28761130571365356,
|
||
|
|
"step": 2805,
|
||
|
|
"valid_targets_mean": 6246.2,
|
||
|
|
"valid_targets_min": 463
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.34984520123839,
|
||
|
|
"grad_norm": 0.19393815611691653,
|
||
|
|
"learning_rate": 1.5086351125125753e-05,
|
||
|
|
"loss": 0.3807,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27923503518104553,
|
||
|
|
"step": 2810,
|
||
|
|
"valid_targets_mean": 6088.3,
|
||
|
|
"valid_targets_min": 244
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.357585139318886,
|
||
|
|
"grad_norm": 0.1862260873038448,
|
||
|
|
"learning_rate": 1.5011546341630598e-05,
|
||
|
|
"loss": 0.2774,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26851555705070496,
|
||
|
|
"step": 2815,
|
||
|
|
"valid_targets_mean": 6268.0,
|
||
|
|
"valid_targets_min": 246
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.365325077399381,
|
||
|
|
"grad_norm": 0.17512304882027216,
|
||
|
|
"learning_rate": 1.493681589937139e-05,
|
||
|
|
"loss": 0.3637,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28443822264671326,
|
||
|
|
"step": 2820,
|
||
|
|
"valid_targets_mean": 7251.7,
|
||
|
|
"valid_targets_min": 1371
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.3730650154798765,
|
||
|
|
"grad_norm": 0.27171620851182704,
|
||
|
|
"learning_rate": 1.48621609120306e-05,
|
||
|
|
"loss": 0.2881,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3243144750595093,
|
||
|
|
"step": 2825,
|
||
|
|
"valid_targets_mean": 2685.1,
|
||
|
|
"valid_targets_min": 384
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.3808049535603715,
|
||
|
|
"grad_norm": 0.18701425872211058,
|
||
|
|
"learning_rate": 1.4787582492166235e-05,
|
||
|
|
"loss": 0.3509,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.283557653427124,
|
||
|
|
"step": 2830,
|
||
|
|
"valid_targets_mean": 6766.1,
|
||
|
|
"valid_targets_min": 328
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.3885448916408665,
|
||
|
|
"grad_norm": 0.23662683269826953,
|
||
|
|
"learning_rate": 1.4713081751195235e-05,
|
||
|
|
"loss": 0.2958,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.33191239833831787,
|
||
|
|
"step": 2835,
|
||
|
|
"valid_targets_mean": 3729.0,
|
||
|
|
"valid_targets_min": 270
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.396284829721362,
|
||
|
|
"grad_norm": 0.17539875611037736,
|
||
|
|
"learning_rate": 1.4638659799376912e-05,
|
||
|
|
"loss": 0.384,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2772039473056793,
|
||
|
|
"step": 2840,
|
||
|
|
"valid_targets_mean": 5790.3,
|
||
|
|
"valid_targets_min": 271
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.404024767801857,
|
||
|
|
"grad_norm": 0.21454227953258442,
|
||
|
|
"learning_rate": 1.456431774579643e-05,
|
||
|
|
"loss": 0.279,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2875722646713257,
|
||
|
|
"step": 2845,
|
||
|
|
"valid_targets_mean": 4450.3,
|
||
|
|
"valid_targets_min": 246
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.411764705882353,
|
||
|
|
"grad_norm": 0.18286374824235943,
|
||
|
|
"learning_rate": 1.4490056698348241e-05,
|
||
|
|
"loss": 0.3646,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2715464234352112,
|
||
|
|
"step": 2850,
|
||
|
|
"valid_targets_mean": 6372.9,
|
||
|
|
"valid_targets_min": 281
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.419504643962848,
|
||
|
|
"grad_norm": 0.634680305738102,
|
||
|
|
"learning_rate": 1.4415877763719582e-05,
|
||
|
|
"loss": 0.3359,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.576937198638916,
|
||
|
|
"step": 2855,
|
||
|
|
"valid_targets_mean": 906.0,
|
||
|
|
"valid_targets_min": 317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.427244582043343,
|
||
|
|
"grad_norm": 0.17171803068750183,
|
||
|
|
"learning_rate": 1.4341782047374015e-05,
|
||
|
|
"loss": 0.3336,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2643846571445465,
|
||
|
|
"step": 2860,
|
||
|
|
"valid_targets_mean": 6534.2,
|
||
|
|
"valid_targets_min": 244
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.434984520123839,
|
||
|
|
"grad_norm": 0.7451349217815492,
|
||
|
|
"learning_rate": 1.4267770653534907e-05,
|
||
|
|
"loss": 0.3356,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5564426183700562,
|
||
|
|
"step": 2865,
|
||
|
|
"valid_targets_mean": 814.0,
|
||
|
|
"valid_targets_min": 414
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.442724458204334,
|
||
|
|
"grad_norm": 0.17649111159783717,
|
||
|
|
"learning_rate": 1.4193844685168993e-05,
|
||
|
|
"loss": 0.3353,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2630617022514343,
|
||
|
|
"step": 2870,
|
||
|
|
"valid_targets_mean": 6517.6,
|
||
|
|
"valid_targets_min": 301
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.45046439628483,
|
||
|
|
"grad_norm": 0.6660055717120287,
|
||
|
|
"learning_rate": 1.4120005243969965e-05,
|
||
|
|
"loss": 0.3407,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5513697862625122,
|
||
|
|
"step": 2875,
|
||
|
|
"valid_targets_mean": 796.5,
|
||
|
|
"valid_targets_min": 171
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.458204334365325,
|
||
|
|
"grad_norm": 0.49915231308382485,
|
||
|
|
"learning_rate": 1.4046253430342019e-05,
|
||
|
|
"loss": 0.506,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46099716424942017,
|
||
|
|
"step": 2880,
|
||
|
|
"valid_targets_mean": 3823.5,
|
||
|
|
"valid_targets_min": 296
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.46594427244582,
|
||
|
|
"grad_norm": 0.2574313216412615,
|
||
|
|
"learning_rate": 1.3972590343383454e-05,
|
||
|
|
"loss": 0.4782,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.44467753171920776,
|
||
|
|
"step": 2885,
|
||
|
|
"valid_targets_mean": 4368.6,
|
||
|
|
"valid_targets_min": 289
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.473684210526316,
|
||
|
|
"grad_norm": 0.329293406547403,
|
||
|
|
"learning_rate": 1.3899017080870313e-05,
|
||
|
|
"loss": 0.4865,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5082035064697266,
|
||
|
|
"step": 2890,
|
||
|
|
"valid_targets_mean": 3313.4,
|
||
|
|
"valid_targets_min": 239
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.481424148606811,
|
||
|
|
"grad_norm": 0.3135054984337207,
|
||
|
|
"learning_rate": 1.3825534739240027e-05,
|
||
|
|
"loss": 0.4699,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.49610787630081177,
|
||
|
|
"step": 2895,
|
||
|
|
"valid_targets_mean": 2229.5,
|
||
|
|
"valid_targets_min": 275
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.489164086687307,
|
||
|
|
"grad_norm": 0.3080950894434962,
|
||
|
|
"learning_rate": 1.3752144413575041e-05,
|
||
|
|
"loss": 0.4767,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.494039922952652,
|
||
|
|
"step": 2900,
|
||
|
|
"valid_targets_mean": 2805.3,
|
||
|
|
"valid_targets_min": 269
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.496904024767802,
|
||
|
|
"grad_norm": 0.2853829576139309,
|
||
|
|
"learning_rate": 1.3678847197586546e-05,
|
||
|
|
"loss": 0.4551,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48569631576538086,
|
||
|
|
"step": 2905,
|
||
|
|
"valid_targets_mean": 2612.6,
|
||
|
|
"valid_targets_min": 314
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.504643962848297,
|
||
|
|
"grad_norm": 0.2305119291965777,
|
||
|
|
"learning_rate": 1.3605644183598128e-05,
|
||
|
|
"loss": 0.4753,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4643450975418091,
|
||
|
|
"step": 2910,
|
||
|
|
"valid_targets_mean": 3845.4,
|
||
|
|
"valid_targets_min": 276
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.512383900928793,
|
||
|
|
"grad_norm": 0.23900971003829202,
|
||
|
|
"learning_rate": 1.3532536462529522e-05,
|
||
|
|
"loss": 0.4593,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.43229028582572937,
|
||
|
|
"step": 2915,
|
||
|
|
"valid_targets_mean": 3683.9,
|
||
|
|
"valid_targets_min": 264
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.520123839009288,
|
||
|
|
"grad_norm": 0.22682537790108437,
|
||
|
|
"learning_rate": 1.3459525123880344e-05,
|
||
|
|
"loss": 0.4628,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4533507227897644,
|
||
|
|
"step": 2920,
|
||
|
|
"valid_targets_mean": 3909.4,
|
||
|
|
"valid_targets_min": 327
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.527863777089784,
|
||
|
|
"grad_norm": 0.26308066445119704,
|
||
|
|
"learning_rate": 1.3386611255713853e-05,
|
||
|
|
"loss": 0.4725,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4806407690048218,
|
||
|
|
"step": 2925,
|
||
|
|
"valid_targets_mean": 2976.2,
|
||
|
|
"valid_targets_min": 358
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.535603715170279,
|
||
|
|
"grad_norm": 0.31980656296788024,
|
||
|
|
"learning_rate": 1.3313795944640732e-05,
|
||
|
|
"loss": 0.4501,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.47049209475517273,
|
||
|
|
"step": 2930,
|
||
|
|
"valid_targets_mean": 2441.2,
|
||
|
|
"valid_targets_min": 278
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.543343653250774,
|
||
|
|
"grad_norm": 0.25426584596736523,
|
||
|
|
"learning_rate": 1.3241080275802923e-05,
|
||
|
|
"loss": 0.4733,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46209028363227844,
|
||
|
|
"step": 2935,
|
||
|
|
"valid_targets_mean": 2866.4,
|
||
|
|
"valid_targets_min": 278
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.5510835913312695,
|
||
|
|
"grad_norm": 0.2823259325001304,
|
||
|
|
"learning_rate": 1.3168465332857414e-05,
|
||
|
|
"loss": 0.4481,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.45362043380737305,
|
||
|
|
"step": 2940,
|
||
|
|
"valid_targets_mean": 2253.2,
|
||
|
|
"valid_targets_min": 354
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.5588235294117645,
|
||
|
|
"grad_norm": 0.2309948397719081,
|
||
|
|
"learning_rate": 1.3095952197960116e-05,
|
||
|
|
"loss": 0.4588,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.43498915433883667,
|
||
|
|
"step": 2945,
|
||
|
|
"valid_targets_mean": 3370.6,
|
||
|
|
"valid_targets_min": 247
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.56656346749226,
|
||
|
|
"grad_norm": 0.4944613890105304,
|
||
|
|
"learning_rate": 1.3023541951749723e-05,
|
||
|
|
"loss": 0.5422,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.606234073638916,
|
||
|
|
"step": 2950,
|
||
|
|
"valid_targets_mean": 4023.8,
|
||
|
|
"valid_targets_min": 935
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.574303405572755,
|
||
|
|
"grad_norm": 0.3896573784813356,
|
||
|
|
"learning_rate": 1.2951235673331626e-05,
|
||
|
|
"loss": 0.5869,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5932326316833496,
|
||
|
|
"step": 2955,
|
||
|
|
"valid_targets_mean": 4263.8,
|
||
|
|
"valid_targets_min": 1031
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.58204334365325,
|
||
|
|
"grad_norm": 0.2637326299879724,
|
||
|
|
"learning_rate": 1.28790344402618e-05,
|
||
|
|
"loss": 0.6079,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5984759330749512,
|
||
|
|
"step": 2960,
|
||
|
|
"valid_targets_mean": 4282.4,
|
||
|
|
"valid_targets_min": 1234
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.589783281733746,
|
||
|
|
"grad_norm": 0.26444558338532514,
|
||
|
|
"learning_rate": 1.280693932853079e-05,
|
||
|
|
"loss": 0.5815,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5566880702972412,
|
||
|
|
"step": 2965,
|
||
|
|
"valid_targets_mean": 4329.7,
|
||
|
|
"valid_targets_min": 1251
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.597523219814241,
|
||
|
|
"grad_norm": 0.24212177588144612,
|
||
|
|
"learning_rate": 1.2734951412547628e-05,
|
||
|
|
"loss": 0.5816,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5645322203636169,
|
||
|
|
"step": 2970,
|
||
|
|
"valid_targets_mean": 4413.9,
|
||
|
|
"valid_targets_min": 1191
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.605263157894737,
|
||
|
|
"grad_norm": 0.2853345642123513,
|
||
|
|
"learning_rate": 1.266307176512386e-05,
|
||
|
|
"loss": 0.5875,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5966958999633789,
|
||
|
|
"step": 2975,
|
||
|
|
"valid_targets_mean": 3331.2,
|
||
|
|
"valid_targets_min": 506
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.613003095975232,
|
||
|
|
"grad_norm": 0.23002754365858166,
|
||
|
|
"learning_rate": 1.2591301457457531e-05,
|
||
|
|
"loss": 0.5849,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.601931631565094,
|
||
|
|
"step": 2980,
|
||
|
|
"valid_targets_mean": 4073.6,
|
||
|
|
"valid_targets_min": 858
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.620743034055727,
|
||
|
|
"grad_norm": 0.2967277481804825,
|
||
|
|
"learning_rate": 1.2519641559117233e-05,
|
||
|
|
"loss": 0.5688,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.610668420791626,
|
||
|
|
"step": 2985,
|
||
|
|
"valid_targets_mean": 3107.3,
|
||
|
|
"valid_targets_min": 514
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.628482972136223,
|
||
|
|
"grad_norm": 0.23482040823362543,
|
||
|
|
"learning_rate": 1.2448093138026176e-05,
|
||
|
|
"loss": 0.5851,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6307334303855896,
|
||
|
|
"step": 2990,
|
||
|
|
"valid_targets_mean": 4157.3,
|
||
|
|
"valid_targets_min": 328
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.636222910216718,
|
||
|
|
"grad_norm": 0.2549241532491689,
|
||
|
|
"learning_rate": 1.2376657260446252e-05,
|
||
|
|
"loss": 0.562,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5880957245826721,
|
||
|
|
"step": 2995,
|
||
|
|
"valid_targets_mean": 3665.2,
|
||
|
|
"valid_targets_min": 827
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.643962848297214,
|
||
|
|
"grad_norm": 0.2428043503401644,
|
||
|
|
"learning_rate": 1.2305334990962165e-05,
|
||
|
|
"loss": 0.5838,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.569591760635376,
|
||
|
|
"step": 3000,
|
||
|
|
"valid_targets_mean": 3729.1,
|
||
|
|
"valid_targets_min": 1186
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.651702786377709,
|
||
|
|
"grad_norm": 0.23984381729142804,
|
||
|
|
"learning_rate": 1.2234127392465544e-05,
|
||
|
|
"loss": 0.5866,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6036013960838318,
|
||
|
|
"step": 3005,
|
||
|
|
"valid_targets_mean": 3811.2,
|
||
|
|
"valid_targets_min": 1187
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.659442724458204,
|
||
|
|
"grad_norm": 0.2142553650624317,
|
||
|
|
"learning_rate": 1.2163035526139119e-05,
|
||
|
|
"loss": 0.5769,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5330759882926941,
|
||
|
|
"step": 3010,
|
||
|
|
"valid_targets_mean": 4278.7,
|
||
|
|
"valid_targets_min": 859
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.6671826625387,
|
||
|
|
"grad_norm": 0.26331734859785466,
|
||
|
|
"learning_rate": 1.2092060451440912e-05,
|
||
|
|
"loss": 0.5871,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6167346835136414,
|
||
|
|
"step": 3015,
|
||
|
|
"valid_targets_mean": 3975.3,
|
||
|
|
"valid_targets_min": 963
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.674922600619195,
|
||
|
|
"grad_norm": 0.2219499889213089,
|
||
|
|
"learning_rate": 1.2021203226088416e-05,
|
||
|
|
"loss": 0.5784,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.564358115196228,
|
||
|
|
"step": 3020,
|
||
|
|
"valid_targets_mean": 4346.6,
|
||
|
|
"valid_targets_min": 555
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.682662538699691,
|
||
|
|
"grad_norm": 0.4073813341909736,
|
||
|
|
"learning_rate": 1.1950464906042888e-05,
|
||
|
|
"loss": 0.5328,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.49985259771347046,
|
||
|
|
"step": 3025,
|
||
|
|
"valid_targets_mean": 3941.8,
|
||
|
|
"valid_targets_min": 883
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.690402476780186,
|
||
|
|
"grad_norm": 0.30837815762079446,
|
||
|
|
"learning_rate": 1.1879846545493553e-05,
|
||
|
|
"loss": 0.4796,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.44186288118362427,
|
||
|
|
"step": 3030,
|
||
|
|
"valid_targets_mean": 4111.9,
|
||
|
|
"valid_targets_min": 902
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.698142414860681,
|
||
|
|
"grad_norm": 0.2645806220054804,
|
||
|
|
"learning_rate": 1.1809349196841937e-05,
|
||
|
|
"loss": 0.4932,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5013681650161743,
|
||
|
|
"step": 3035,
|
||
|
|
"valid_targets_mean": 3929.4,
|
||
|
|
"valid_targets_min": 1072
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.705882352941177,
|
||
|
|
"grad_norm": 0.2581919225262186,
|
||
|
|
"learning_rate": 1.1738973910686152e-05,
|
||
|
|
"loss": 0.4812,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48816511034965515,
|
||
|
|
"step": 3040,
|
||
|
|
"valid_targets_mean": 4184.5,
|
||
|
|
"valid_targets_min": 829
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.713622291021672,
|
||
|
|
"grad_norm": 0.24109939456814972,
|
||
|
|
"learning_rate": 1.1668721735805261e-05,
|
||
|
|
"loss": 0.4855,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4825838804244995,
|
||
|
|
"step": 3045,
|
||
|
|
"valid_targets_mean": 4404.0,
|
||
|
|
"valid_targets_min": 385
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.7213622291021675,
|
||
|
|
"grad_norm": 0.22779491277948177,
|
||
|
|
"learning_rate": 1.1598593719143634e-05,
|
||
|
|
"loss": 0.4593,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4659040570259094,
|
||
|
|
"step": 3050,
|
||
|
|
"valid_targets_mean": 4238.4,
|
||
|
|
"valid_targets_min": 1125
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.7291021671826625,
|
||
|
|
"grad_norm": 0.19958878516457226,
|
||
|
|
"learning_rate": 1.1528590905795376e-05,
|
||
|
|
"loss": 0.4904,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.47717684507369995,
|
||
|
|
"step": 3055,
|
||
|
|
"valid_targets_mean": 4663.1,
|
||
|
|
"valid_targets_min": 1147
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.7368421052631575,
|
||
|
|
"grad_norm": 0.2232370082711173,
|
||
|
|
"learning_rate": 1.1458714338988699e-05,
|
||
|
|
"loss": 0.4673,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4640059471130371,
|
||
|
|
"step": 3060,
|
||
|
|
"valid_targets_mean": 4196.7,
|
||
|
|
"valid_targets_min": 868
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.744582043343653,
|
||
|
|
"grad_norm": 0.2115959492445532,
|
||
|
|
"learning_rate": 1.1388965060070435e-05,
|
||
|
|
"loss": 0.4737,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46135616302490234,
|
||
|
|
"step": 3065,
|
||
|
|
"valid_targets_mean": 4613.0,
|
||
|
|
"valid_targets_min": 1130
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.752321981424148,
|
||
|
|
"grad_norm": 0.2466030680419363,
|
||
|
|
"learning_rate": 1.1319344108490462e-05,
|
||
|
|
"loss": 0.4585,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.44720178842544556,
|
||
|
|
"step": 3070,
|
||
|
|
"valid_targets_mean": 4339.9,
|
||
|
|
"valid_targets_min": 917
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.760061919504644,
|
||
|
|
"grad_norm": 0.21622363064066494,
|
||
|
|
"learning_rate": 1.1249852521786257e-05,
|
||
|
|
"loss": 0.4872,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48151838779449463,
|
||
|
|
"step": 3075,
|
||
|
|
"valid_targets_mean": 4427.8,
|
||
|
|
"valid_targets_min": 986
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.767801857585139,
|
||
|
|
"grad_norm": 0.2261750134889117,
|
||
|
|
"learning_rate": 1.1180491335567402e-05,
|
||
|
|
"loss": 0.4704,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46817654371261597,
|
||
|
|
"step": 3080,
|
||
|
|
"valid_targets_mean": 4592.4,
|
||
|
|
"valid_targets_min": 855
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.775541795665634,
|
||
|
|
"grad_norm": 0.21343235013182196,
|
||
|
|
"learning_rate": 1.1111261583500169e-05,
|
||
|
|
"loss": 0.4741,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.45930105447769165,
|
||
|
|
"step": 3085,
|
||
|
|
"valid_targets_mean": 4412.0,
|
||
|
|
"valid_targets_min": 1345
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.78328173374613,
|
||
|
|
"grad_norm": 0.20236951284938198,
|
||
|
|
"learning_rate": 1.1042164297292128e-05,
|
||
|
|
"loss": 0.4651,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.43615061044692993,
|
||
|
|
"step": 3090,
|
||
|
|
"valid_targets_mean": 4575.8,
|
||
|
|
"valid_targets_min": 1126
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.791021671826625,
|
||
|
|
"grad_norm": 0.20482977157799043,
|
||
|
|
"learning_rate": 1.097320050667673e-05,
|
||
|
|
"loss": 0.4774,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4736917018890381,
|
||
|
|
"step": 3095,
|
||
|
|
"valid_targets_mean": 4447.5,
|
||
|
|
"valid_targets_min": 918
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.798761609907121,
|
||
|
|
"grad_norm": 0.22133634754436274,
|
||
|
|
"learning_rate": 1.0904371239397997e-05,
|
||
|
|
"loss": 0.4756,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.44803208112716675,
|
||
|
|
"step": 3100,
|
||
|
|
"valid_targets_mean": 4437.8,
|
||
|
|
"valid_targets_min": 1123
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.806501547987616,
|
||
|
|
"grad_norm": 0.23152333509355782,
|
||
|
|
"learning_rate": 1.0835677521195204e-05,
|
||
|
|
"loss": 0.4738,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46068382263183594,
|
||
|
|
"step": 3105,
|
||
|
|
"valid_targets_mean": 3840.9,
|
||
|
|
"valid_targets_min": 952
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.814241486068111,
|
||
|
|
"grad_norm": 0.2272765789169338,
|
||
|
|
"learning_rate": 1.0767120375787577e-05,
|
||
|
|
"loss": 0.4529,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4283614158630371,
|
||
|
|
"step": 3110,
|
||
|
|
"valid_targets_mean": 3756.0,
|
||
|
|
"valid_targets_min": 409
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.821981424148607,
|
||
|
|
"grad_norm": 0.25648636917307394,
|
||
|
|
"learning_rate": 1.0698700824859025e-05,
|
||
|
|
"loss": 0.5585,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46974149346351624,
|
||
|
|
"step": 3115,
|
||
|
|
"valid_targets_mean": 3755.8,
|
||
|
|
"valid_targets_min": 776
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.829721362229102,
|
||
|
|
"grad_norm": 0.24451697329527045,
|
||
|
|
"learning_rate": 1.0630419888042973e-05,
|
||
|
|
"loss": 0.4632,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4404352307319641,
|
||
|
|
"step": 3120,
|
||
|
|
"valid_targets_mean": 3807.7,
|
||
|
|
"valid_targets_min": 395
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.837461300309598,
|
||
|
|
"grad_norm": 0.3346159167672239,
|
||
|
|
"learning_rate": 1.0562278582907095e-05,
|
||
|
|
"loss": 0.565,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.45883190631866455,
|
||
|
|
"step": 3125,
|
||
|
|
"valid_targets_mean": 3717.1,
|
||
|
|
"valid_targets_min": 369
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.845201238390093,
|
||
|
|
"grad_norm": 0.24806934992934893,
|
||
|
|
"learning_rate": 1.049427792493819e-05,
|
||
|
|
"loss": 0.4559,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.42593497037887573,
|
||
|
|
"step": 3130,
|
||
|
|
"valid_targets_mean": 3834.6,
|
||
|
|
"valid_targets_min": 1051
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.852941176470588,
|
||
|
|
"grad_norm": 0.23414129599125166,
|
||
|
|
"learning_rate": 1.0426418927527035e-05,
|
||
|
|
"loss": 0.527,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.467929869890213,
|
||
|
|
"step": 3135,
|
||
|
|
"valid_targets_mean": 3716.2,
|
||
|
|
"valid_targets_min": 382
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.860681114551084,
|
||
|
|
"grad_norm": 0.22717158441765975,
|
||
|
|
"learning_rate": 1.0358702601953309e-05,
|
||
|
|
"loss": 0.4536,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4281642436981201,
|
||
|
|
"step": 3140,
|
||
|
|
"valid_targets_mean": 3947.7,
|
||
|
|
"valid_targets_min": 665
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.868421052631579,
|
||
|
|
"grad_norm": 0.25262814135792955,
|
||
|
|
"learning_rate": 1.0291129957370467e-05,
|
||
|
|
"loss": 0.5205,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.45696771144866943,
|
||
|
|
"step": 3145,
|
||
|
|
"valid_targets_mean": 3347.8,
|
||
|
|
"valid_targets_min": 1039
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.876160990712075,
|
||
|
|
"grad_norm": 0.23645986881373657,
|
||
|
|
"learning_rate": 1.0223702000790774e-05,
|
||
|
|
"loss": 0.463,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4532109200954437,
|
||
|
|
"step": 3150,
|
||
|
|
"valid_targets_mean": 3933.3,
|
||
|
|
"valid_targets_min": 821
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.88390092879257,
|
||
|
|
"grad_norm": 0.23161716780466984,
|
||
|
|
"learning_rate": 1.0156419737070224e-05,
|
||
|
|
"loss": 0.5626,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4650879502296448,
|
||
|
|
"step": 3155,
|
||
|
|
"valid_targets_mean": 3639.2,
|
||
|
|
"valid_targets_min": 360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.891640866873065,
|
||
|
|
"grad_norm": 0.2352621025858952,
|
||
|
|
"learning_rate": 1.0089284168893618e-05,
|
||
|
|
"loss": 0.4511,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4373045861721039,
|
||
|
|
"step": 3160,
|
||
|
|
"valid_targets_mean": 3762.1,
|
||
|
|
"valid_targets_min": 427
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.8993808049535605,
|
||
|
|
"grad_norm": 0.26731619438338716,
|
||
|
|
"learning_rate": 1.0022296296759599e-05,
|
||
|
|
"loss": 0.5547,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.47625628113746643,
|
||
|
|
"step": 3165,
|
||
|
|
"valid_targets_mean": 3373.5,
|
||
|
|
"valid_targets_min": 795
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.9071207430340555,
|
||
|
|
"grad_norm": 0.23969184232947452,
|
||
|
|
"learning_rate": 9.955457118965734e-06,
|
||
|
|
"loss": 0.4625,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4491170346736908,
|
||
|
|
"step": 3170,
|
||
|
|
"valid_targets_mean": 3545.1,
|
||
|
|
"valid_targets_min": 458
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.914860681114551,
|
||
|
|
"grad_norm": 0.26375183217064135,
|
||
|
|
"learning_rate": 9.88876763159368e-06,
|
||
|
|
"loss": 0.5049,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4723615050315857,
|
||
|
|
"step": 3175,
|
||
|
|
"valid_targets_mean": 3247.4,
|
||
|
|
"valid_targets_min": 905
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.922600619195046,
|
||
|
|
"grad_norm": 0.23641821535896607,
|
||
|
|
"learning_rate": 9.822228828494271e-06,
|
||
|
|
"loss": 0.4638,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.45484021306037903,
|
||
|
|
"step": 3180,
|
||
|
|
"valid_targets_mean": 3632.9,
|
||
|
|
"valid_targets_min": 506
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.930340557275541,
|
||
|
|
"grad_norm": 0.800585951475013,
|
||
|
|
"learning_rate": 9.755841701272776e-06,
|
||
|
|
"loss": 0.4107,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2813580632209778,
|
||
|
|
"step": 3185,
|
||
|
|
"valid_targets_mean": 4943.5,
|
||
|
|
"valid_targets_min": 1635
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.938080495356037,
|
||
|
|
"grad_norm": 0.4021972708835097,
|
||
|
|
"learning_rate": 9.68960723927407e-06,
|
||
|
|
"loss": 0.2558,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23448145389556885,
|
||
|
|
"step": 3190,
|
||
|
|
"valid_targets_mean": 5463.8,
|
||
|
|
"valid_targets_min": 2499
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.945820433436532,
|
||
|
|
"grad_norm": 0.41570879872796196,
|
||
|
|
"learning_rate": 9.623526429567914e-06,
|
||
|
|
"loss": 0.2411,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22293655574321747,
|
||
|
|
"step": 3195,
|
||
|
|
"valid_targets_mean": 5126.4,
|
||
|
|
"valid_targets_min": 842
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.953560371517028,
|
||
|
|
"grad_norm": 0.5301017492952698,
|
||
|
|
"learning_rate": 9.55760025693424e-06,
|
||
|
|
"loss": 0.2173,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22266924381256104,
|
||
|
|
"step": 3200,
|
||
|
|
"valid_targets_mean": 4619.5,
|
||
|
|
"valid_targets_min": 1759
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.961300309597523,
|
||
|
|
"grad_norm": 0.43462714559850535,
|
||
|
|
"learning_rate": 9.491829703848476e-06,
|
||
|
|
"loss": 0.2326,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23762568831443787,
|
||
|
|
"step": 3205,
|
||
|
|
"valid_targets_mean": 4067.3,
|
||
|
|
"valid_targets_min": 1730
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.969040247678018,
|
||
|
|
"grad_norm": 0.32250360313981175,
|
||
|
|
"learning_rate": 9.426215750466918e-06,
|
||
|
|
"loss": 0.2371,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2375468909740448,
|
||
|
|
"step": 3210,
|
||
|
|
"valid_targets_mean": 4208.2,
|
||
|
|
"valid_targets_min": 1326
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.976780185758514,
|
||
|
|
"grad_norm": 0.4024473565699988,
|
||
|
|
"learning_rate": 9.360759374612084e-06,
|
||
|
|
"loss": 0.2326,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.227864608168602,
|
||
|
|
"step": 3215,
|
||
|
|
"valid_targets_mean": 4078.0,
|
||
|
|
"valid_targets_min": 2275
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.984520123839009,
|
||
|
|
"grad_norm": 0.44875854807460835,
|
||
|
|
"learning_rate": 9.295461551758198e-06,
|
||
|
|
"loss": 0.2308,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22797873616218567,
|
||
|
|
"step": 3220,
|
||
|
|
"valid_targets_mean": 3467.1,
|
||
|
|
"valid_targets_min": 1777
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.992260061919505,
|
||
|
|
"grad_norm": 0.35347425663288073,
|
||
|
|
"learning_rate": 9.230323255016597e-06,
|
||
|
|
"loss": 0.2237,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22162383794784546,
|
||
|
|
"step": 3225,
|
||
|
|
"valid_targets_mean": 2943.9,
|
||
|
|
"valid_targets_min": 1774
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0,
|
||
|
|
"grad_norm": 0.5222633437217012,
|
||
|
|
"learning_rate": 9.165345455121268e-06,
|
||
|
|
"loss": 0.2183,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22755886614322662,
|
||
|
|
"step": 3230,
|
||
|
|
"valid_targets_mean": 2756.4,
|
||
|
|
"valid_targets_min": 182
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.007739938080495,
|
||
|
|
"grad_norm": 0.46175174059703594,
|
||
|
|
"learning_rate": 9.100529120414353e-06,
|
||
|
|
"loss": 0.2533,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2465011179447174,
|
||
|
|
"step": 3235,
|
||
|
|
"valid_targets_mean": 3787.4,
|
||
|
|
"valid_targets_min": 443
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.015479876160991,
|
||
|
|
"grad_norm": 0.35320780998971385,
|
||
|
|
"learning_rate": 9.035875216831755e-06,
|
||
|
|
"loss": 0.2445,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23728284239768982,
|
||
|
|
"step": 3240,
|
||
|
|
"valid_targets_mean": 3744.1,
|
||
|
|
"valid_targets_min": 445
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.023219814241486,
|
||
|
|
"grad_norm": 0.32626821411586115,
|
||
|
|
"learning_rate": 8.971384707888707e-06,
|
||
|
|
"loss": 0.2389,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23758280277252197,
|
||
|
|
"step": 3245,
|
||
|
|
"valid_targets_mean": 3806.1,
|
||
|
|
"valid_targets_min": 364
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.030959752321982,
|
||
|
|
"grad_norm": 0.2668998324674633,
|
||
|
|
"learning_rate": 8.907058554665426e-06,
|
||
|
|
"loss": 0.2356,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23749437928199768,
|
||
|
|
"step": 3250,
|
||
|
|
"valid_targets_mean": 3883.8,
|
||
|
|
"valid_targets_min": 423
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.038699690402477,
|
||
|
|
"grad_norm": 0.24519449071981902,
|
||
|
|
"learning_rate": 8.842897715792791e-06,
|
||
|
|
"loss": 0.2303,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23347638547420502,
|
||
|
|
"step": 3255,
|
||
|
|
"valid_targets_mean": 3872.5,
|
||
|
|
"valid_targets_min": 249
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.046439628482972,
|
||
|
|
"grad_norm": 0.21564104070745477,
|
||
|
|
"learning_rate": 8.778903147438071e-06,
|
||
|
|
"loss": 0.2295,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23272711038589478,
|
||
|
|
"step": 3260,
|
||
|
|
"valid_targets_mean": 3780.5,
|
||
|
|
"valid_targets_min": 372
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.054179566563468,
|
||
|
|
"grad_norm": 0.21977507086750261,
|
||
|
|
"learning_rate": 8.715075803290642e-06,
|
||
|
|
"loss": 0.2274,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22935795783996582,
|
||
|
|
"step": 3265,
|
||
|
|
"valid_targets_mean": 3663.5,
|
||
|
|
"valid_targets_min": 460
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.061919504643963,
|
||
|
|
"grad_norm": 0.23019901014613497,
|
||
|
|
"learning_rate": 8.651416634547818e-06,
|
||
|
|
"loss": 0.2244,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22157183289527893,
|
||
|
|
"step": 3270,
|
||
|
|
"valid_targets_mean": 3721.1,
|
||
|
|
"valid_targets_min": 448
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.069659442724459,
|
||
|
|
"grad_norm": 0.22100108126667986,
|
||
|
|
"learning_rate": 8.587926589900627e-06,
|
||
|
|
"loss": 0.2239,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22390300035476685,
|
||
|
|
"step": 3275,
|
||
|
|
"valid_targets_mean": 3723.2,
|
||
|
|
"valid_targets_min": 391
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.077399380804954,
|
||
|
|
"grad_norm": 0.20174630767505958,
|
||
|
|
"learning_rate": 8.524606615519708e-06,
|
||
|
|
"loss": 0.2245,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22582106292247772,
|
||
|
|
"step": 3280,
|
||
|
|
"valid_targets_mean": 3877.2,
|
||
|
|
"valid_targets_min": 462
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.085139318885449,
|
||
|
|
"grad_norm": 0.21643913543323,
|
||
|
|
"learning_rate": 8.461457655041208e-06,
|
||
|
|
"loss": 0.2277,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23050709068775177,
|
||
|
|
"step": 3285,
|
||
|
|
"valid_targets_mean": 3845.5,
|
||
|
|
"valid_targets_min": 497
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0928792569659445,
|
||
|
|
"grad_norm": 0.2786879558674102,
|
||
|
|
"learning_rate": 8.398480649552689e-06,
|
||
|
|
"loss": 0.2661,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26511168479919434,
|
||
|
|
"step": 3290,
|
||
|
|
"valid_targets_mean": 4920.5,
|
||
|
|
"valid_targets_min": 2219
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.1006191950464395,
|
||
|
|
"grad_norm": 0.6740160489683154,
|
||
|
|
"learning_rate": 8.335676537579158e-06,
|
||
|
|
"loss": 0.3311,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.44381558895111084,
|
||
|
|
"step": 3295,
|
||
|
|
"valid_targets_mean": 965.5,
|
||
|
|
"valid_targets_min": 456
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.108359133126935,
|
||
|
|
"grad_norm": 0.22954004116622237,
|
||
|
|
"learning_rate": 8.273046255069033e-06,
|
||
|
|
"loss": 0.3068,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2575819492340088,
|
||
|
|
"step": 3300,
|
||
|
|
"valid_targets_mean": 4698.2,
|
||
|
|
"valid_targets_min": 3199
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.11609907120743,
|
||
|
|
"grad_norm": 0.5744850476579234,
|
||
|
|
"learning_rate": 8.210590735380201e-06,
|
||
|
|
"loss": 0.352,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4695858955383301,
|
||
|
|
"step": 3305,
|
||
|
|
"valid_targets_mean": 877.1,
|
||
|
|
"valid_targets_min": 427
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.123839009287925,
|
||
|
|
"grad_norm": 0.21320232656799015,
|
||
|
|
"learning_rate": 8.148310909266135e-06,
|
||
|
|
"loss": 0.2578,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2552856206893921,
|
||
|
|
"step": 3310,
|
||
|
|
"valid_targets_mean": 4635.7,
|
||
|
|
"valid_targets_min": 3134
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.131578947368421,
|
||
|
|
"grad_norm": 0.5369217114379193,
|
||
|
|
"learning_rate": 8.08620770486199e-06,
|
||
|
|
"loss": 0.3455,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.47632044553756714,
|
||
|
|
"step": 3315,
|
||
|
|
"valid_targets_mean": 855.6,
|
||
|
|
"valid_targets_min": 374
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.139318885448916,
|
||
|
|
"grad_norm": 0.20559409162279968,
|
||
|
|
"learning_rate": 8.024282047670797e-06,
|
||
|
|
"loss": 0.3109,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2635243833065033,
|
||
|
|
"step": 3320,
|
||
|
|
"valid_targets_mean": 4777.3,
|
||
|
|
"valid_targets_min": 2671
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.147058823529412,
|
||
|
|
"grad_norm": 0.5632801377087874,
|
||
|
|
"learning_rate": 7.962534860549655e-06,
|
||
|
|
"loss": 0.3066,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4839676022529602,
|
||
|
|
"step": 3325,
|
||
|
|
"valid_targets_mean": 898.8,
|
||
|
|
"valid_targets_min": 299
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.154798761609907,
|
||
|
|
"grad_norm": 0.21683357441885256,
|
||
|
|
"learning_rate": 7.900967063696e-06,
|
||
|
|
"loss": 0.3305,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24795609712600708,
|
||
|
|
"step": 3330,
|
||
|
|
"valid_targets_mean": 4618.4,
|
||
|
|
"valid_targets_min": 2204
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.162538699690402,
|
||
|
|
"grad_norm": 0.22317526667844584,
|
||
|
|
"learning_rate": 7.839579574633846e-06,
|
||
|
|
"loss": 0.2615,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26636117696762085,
|
||
|
|
"step": 3335,
|
||
|
|
"valid_targets_mean": 3491.6,
|
||
|
|
"valid_targets_min": 313
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.170278637770898,
|
||
|
|
"grad_norm": 0.20519786410349783,
|
||
|
|
"learning_rate": 7.778373308200178e-06,
|
||
|
|
"loss": 0.3613,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25080955028533936,
|
||
|
|
"step": 3340,
|
||
|
|
"valid_targets_mean": 4612.6,
|
||
|
|
"valid_targets_min": 3000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.178018575851393,
|
||
|
|
"grad_norm": 0.2312999809015304,
|
||
|
|
"learning_rate": 7.717349176531255e-06,
|
||
|
|
"loss": 0.2579,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25787150859832764,
|
||
|
|
"step": 3345,
|
||
|
|
"valid_targets_mean": 3892.6,
|
||
|
|
"valid_targets_min": 238
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.185758513931889,
|
||
|
|
"grad_norm": 0.24968468292165155,
|
||
|
|
"learning_rate": 7.656508089049047e-06,
|
||
|
|
"loss": 0.3806,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25625646114349365,
|
||
|
|
"step": 3350,
|
||
|
|
"valid_targets_mean": 4844.6,
|
||
|
|
"valid_targets_min": 2731
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.193498452012384,
|
||
|
|
"grad_norm": 0.5108274166408217,
|
||
|
|
"learning_rate": 7.595850952447677e-06,
|
||
|
|
"loss": 0.2915,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.43939924240112305,
|
||
|
|
"step": 3355,
|
||
|
|
"valid_targets_mean": 984.1,
|
||
|
|
"valid_targets_min": 428
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.201238390092879,
|
||
|
|
"grad_norm": 0.21786468874882947,
|
||
|
|
"learning_rate": 7.535378670679929e-06,
|
||
|
|
"loss": 0.3494,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25967565178871155,
|
||
|
|
"step": 3360,
|
||
|
|
"valid_targets_mean": 4991.7,
|
||
|
|
"valid_targets_min": 3133
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.208978328173375,
|
||
|
|
"grad_norm": 0.23486937258838944,
|
||
|
|
"learning_rate": 7.475092144943739e-06,
|
||
|
|
"loss": 0.2627,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27736467123031616,
|
||
|
|
"step": 3365,
|
||
|
|
"valid_targets_mean": 3466.8,
|
||
|
|
"valid_targets_min": 300
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.21671826625387,
|
||
|
|
"grad_norm": 0.28398621066808244,
|
||
|
|
"learning_rate": 7.414992273668778e-06,
|
||
|
|
"loss": 0.394,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2862158417701721,
|
||
|
|
"step": 3370,
|
||
|
|
"valid_targets_mean": 5756.4,
|
||
|
|
"valid_targets_min": 472
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.224458204334366,
|
||
|
|
"grad_norm": 0.259890669293052,
|
||
|
|
"learning_rate": 7.355079952503097e-06,
|
||
|
|
"loss": 0.2828,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2718966007232666,
|
||
|
|
"step": 3375,
|
||
|
|
"valid_targets_mean": 5488.7,
|
||
|
|
"valid_targets_min": 504
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.232198142414861,
|
||
|
|
"grad_norm": 0.2242016308855682,
|
||
|
|
"learning_rate": 7.295356074299726e-06,
|
||
|
|
"loss": 0.2945,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2871878743171692,
|
||
|
|
"step": 3380,
|
||
|
|
"valid_targets_mean": 5880.8,
|
||
|
|
"valid_targets_min": 492
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.239938080495356,
|
||
|
|
"grad_norm": 0.19303123902326239,
|
||
|
|
"learning_rate": 7.2358215291034014e-06,
|
||
|
|
"loss": 0.2813,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2771328389644623,
|
||
|
|
"step": 3385,
|
||
|
|
"valid_targets_mean": 5631.9,
|
||
|
|
"valid_targets_min": 1034
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.247678018575852,
|
||
|
|
"grad_norm": 0.2061645273001123,
|
||
|
|
"learning_rate": 7.176477204137287e-06,
|
||
|
|
"loss": 0.2914,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27410978078842163,
|
||
|
|
"step": 3390,
|
||
|
|
"valid_targets_mean": 5617.9,
|
||
|
|
"valid_targets_min": 391
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.255417956656347,
|
||
|
|
"grad_norm": 0.18324443765294485,
|
||
|
|
"learning_rate": 7.117323983789779e-06,
|
||
|
|
"loss": 0.2854,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2797078490257263,
|
||
|
|
"step": 3395,
|
||
|
|
"valid_targets_mean": 6436.5,
|
||
|
|
"valid_targets_min": 560
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.2631578947368425,
|
||
|
|
"grad_norm": 0.20754166183738834,
|
||
|
|
"learning_rate": 7.058362749601286e-06,
|
||
|
|
"loss": 0.2873,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2978689968585968,
|
||
|
|
"step": 3400,
|
||
|
|
"valid_targets_mean": 5309.5,
|
||
|
|
"valid_targets_min": 482
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.2708978328173375,
|
||
|
|
"grad_norm": 0.2002353291457852,
|
||
|
|
"learning_rate": 6.9995943802511175e-06,
|
||
|
|
"loss": 0.2825,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27477341890335083,
|
||
|
|
"step": 3405,
|
||
|
|
"valid_targets_mean": 5691.1,
|
||
|
|
"valid_targets_min": 844
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.2786377708978325,
|
||
|
|
"grad_norm": 0.1912165622777591,
|
||
|
|
"learning_rate": 6.9410197515443824e-06,
|
||
|
|
"loss": 0.2827,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2815304100513458,
|
||
|
|
"step": 3410,
|
||
|
|
"valid_targets_mean": 5925.6,
|
||
|
|
"valid_targets_min": 500
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.286377708978328,
|
||
|
|
"grad_norm": 0.2372970623736488,
|
||
|
|
"learning_rate": 6.882639736398951e-06,
|
||
|
|
"loss": 0.2835,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27664849162101746,
|
||
|
|
"step": 3415,
|
||
|
|
"valid_targets_mean": 5658.9,
|
||
|
|
"valid_targets_min": 491
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.294117647058823,
|
||
|
|
"grad_norm": 0.21244228690774647,
|
||
|
|
"learning_rate": 6.824455204832412e-06,
|
||
|
|
"loss": 0.2829,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30230027437210083,
|
||
|
|
"step": 3420,
|
||
|
|
"valid_targets_mean": 5660.4,
|
||
|
|
"valid_targets_min": 422
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.301857585139319,
|
||
|
|
"grad_norm": 0.18975531690153224,
|
||
|
|
"learning_rate": 6.76646702394915e-06,
|
||
|
|
"loss": 0.2764,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2635669708251953,
|
||
|
|
"step": 3425,
|
||
|
|
"valid_targets_mean": 6256.5,
|
||
|
|
"valid_targets_min": 590
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.309597523219814,
|
||
|
|
"grad_norm": 0.22890150480418328,
|
||
|
|
"learning_rate": 6.708676057927388e-06,
|
||
|
|
"loss": 0.2829,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2877359986305237,
|
||
|
|
"step": 3430,
|
||
|
|
"valid_targets_mean": 5626.4,
|
||
|
|
"valid_targets_min": 524
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.317337461300309,
|
||
|
|
"grad_norm": 0.20487876618125223,
|
||
|
|
"learning_rate": 6.6510831680063245e-06,
|
||
|
|
"loss": 0.2788,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27886518836021423,
|
||
|
|
"step": 3435,
|
||
|
|
"valid_targets_mean": 5776.4,
|
||
|
|
"valid_targets_min": 921
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.325077399380805,
|
||
|
|
"grad_norm": 0.29132235875618673,
|
||
|
|
"learning_rate": 6.5936892124732995e-06,
|
||
|
|
"loss": 0.2798,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27785706520080566,
|
||
|
|
"step": 3440,
|
||
|
|
"valid_targets_mean": 5494.7,
|
||
|
|
"valid_targets_min": 1092
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.3328173374613,
|
||
|
|
"grad_norm": 0.18734230400423119,
|
||
|
|
"learning_rate": 6.5364950466509965e-06,
|
||
|
|
"loss": 0.2793,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27842575311660767,
|
||
|
|
"step": 3445,
|
||
|
|
"valid_targets_mean": 6014.0,
|
||
|
|
"valid_targets_min": 234
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.340557275541796,
|
||
|
|
"grad_norm": 0.18382162910149114,
|
||
|
|
"learning_rate": 6.479501522884715e-06,
|
||
|
|
"loss": 0.2833,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2894097864627838,
|
||
|
|
"step": 3450,
|
||
|
|
"valid_targets_mean": 6489.6,
|
||
|
|
"valid_targets_min": 487
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.348297213622291,
|
||
|
|
"grad_norm": 0.2327283325433985,
|
||
|
|
"learning_rate": 6.422709490529633e-06,
|
||
|
|
"loss": 0.3752,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.31111323833465576,
|
||
|
|
"step": 3455,
|
||
|
|
"valid_targets_mean": 3702.6,
|
||
|
|
"valid_targets_min": 466
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.356037151702786,
|
||
|
|
"grad_norm": 0.17672937394671162,
|
||
|
|
"learning_rate": 6.3661197959381905e-06,
|
||
|
|
"loss": 0.2754,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2722792327404022,
|
||
|
|
"step": 3460,
|
||
|
|
"valid_targets_mean": 5997.3,
|
||
|
|
"valid_targets_min": 240
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.363777089783282,
|
||
|
|
"grad_norm": 0.2045483704047087,
|
||
|
|
"learning_rate": 6.309733282447441e-06,
|
||
|
|
"loss": 0.3536,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29910486936569214,
|
||
|
|
"step": 3465,
|
||
|
|
"valid_targets_mean": 4160.1,
|
||
|
|
"valid_targets_min": 473
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.371517027863777,
|
||
|
|
"grad_norm": 0.1710903453756809,
|
||
|
|
"learning_rate": 6.2535507903665045e-06,
|
||
|
|
"loss": 0.2763,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2698843479156494,
|
||
|
|
"step": 3470,
|
||
|
|
"valid_targets_mean": 6533.9,
|
||
|
|
"valid_targets_min": 1876
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.379256965944273,
|
||
|
|
"grad_norm": 0.17268443873673098,
|
||
|
|
"learning_rate": 6.197573156964041e-06,
|
||
|
|
"loss": 0.3526,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.278929203748703,
|
||
|
|
"step": 3475,
|
||
|
|
"valid_targets_mean": 6688.6,
|
||
|
|
"valid_targets_min": 298
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.386996904024768,
|
||
|
|
"grad_norm": 0.17449571184265,
|
||
|
|
"learning_rate": 6.141801216455761e-06,
|
||
|
|
"loss": 0.2823,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2963022291660309,
|
||
|
|
"step": 3480,
|
||
|
|
"valid_targets_mean": 6724.0,
|
||
|
|
"valid_targets_min": 549
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.394736842105263,
|
||
|
|
"grad_norm": 0.18997767838363935,
|
||
|
|
"learning_rate": 6.086235799992024e-06,
|
||
|
|
"loss": 0.3879,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2817431688308716,
|
||
|
|
"step": 3485,
|
||
|
|
"valid_targets_mean": 6117.0,
|
||
|
|
"valid_targets_min": 219
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.402476780185759,
|
||
|
|
"grad_norm": 0.1681458384896858,
|
||
|
|
"learning_rate": 6.030877735645404e-06,
|
||
|
|
"loss": 0.2731,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26748985052108765,
|
||
|
|
"step": 3490,
|
||
|
|
"valid_targets_mean": 6469.6,
|
||
|
|
"valid_targets_min": 209
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.410216718266254,
|
||
|
|
"grad_norm": 0.19523810996095045,
|
||
|
|
"learning_rate": 5.975727848398409e-06,
|
||
|
|
"loss": 0.3616,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29237279295921326,
|
||
|
|
"step": 3495,
|
||
|
|
"valid_targets_mean": 6510.8,
|
||
|
|
"valid_targets_min": 206
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.41795665634675,
|
||
|
|
"grad_norm": 0.18873244720196836,
|
||
|
|
"learning_rate": 5.920786960131126e-06,
|
||
|
|
"loss": 0.2713,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2800244987010956,
|
||
|
|
"step": 3500,
|
||
|
|
"valid_targets_mean": 5183.3,
|
||
|
|
"valid_targets_min": 399
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.425696594427245,
|
||
|
|
"grad_norm": 0.17812665496098537,
|
||
|
|
"learning_rate": 5.866055889609015e-06,
|
||
|
|
"loss": 0.389,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2750130891799927,
|
||
|
|
"step": 3505,
|
||
|
|
"valid_targets_mean": 6249.5,
|
||
|
|
"valid_targets_min": 284
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.43343653250774,
|
||
|
|
"grad_norm": 0.19643662179343646,
|
||
|
|
"learning_rate": 5.811535452470682e-06,
|
||
|
|
"loss": 0.2737,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28120073676109314,
|
||
|
|
"step": 3510,
|
||
|
|
"valid_targets_mean": 5410.6,
|
||
|
|
"valid_targets_min": 265
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.4411764705882355,
|
||
|
|
"grad_norm": 0.17029835946090943,
|
||
|
|
"learning_rate": 5.757226461215754e-06,
|
||
|
|
"loss": 0.3875,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2717994451522827,
|
||
|
|
"step": 3515,
|
||
|
|
"valid_targets_mean": 6375.5,
|
||
|
|
"valid_targets_min": 401
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.4489164086687305,
|
||
|
|
"grad_norm": 0.24845806261913753,
|
||
|
|
"learning_rate": 5.703129725192738e-06,
|
||
|
|
"loss": 0.2795,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.307073175907135,
|
||
|
|
"step": 3520,
|
||
|
|
"valid_targets_mean": 3327.1,
|
||
|
|
"valid_targets_min": 428
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.456656346749226,
|
||
|
|
"grad_norm": 0.7016497148090439,
|
||
|
|
"learning_rate": 5.649246050586965e-06,
|
||
|
|
"loss": 0.5163,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48795342445373535,
|
||
|
|
"step": 3525,
|
||
|
|
"valid_targets_mean": 3046.5,
|
||
|
|
"valid_targets_min": 249
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.464396284829721,
|
||
|
|
"grad_norm": 0.5205091915783154,
|
||
|
|
"learning_rate": 5.595576240408609e-06,
|
||
|
|
"loss": 0.4776,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46614086627960205,
|
||
|
|
"step": 3530,
|
||
|
|
"valid_targets_mean": 3748.6,
|
||
|
|
"valid_targets_min": 339
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.472136222910216,
|
||
|
|
"grad_norm": 0.284266973071161,
|
||
|
|
"learning_rate": 5.542121094480675e-06,
|
||
|
|
"loss": 0.468,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.44249796867370605,
|
||
|
|
"step": 3535,
|
||
|
|
"valid_targets_mean": 3423.9,
|
||
|
|
"valid_targets_min": 455
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.479876160990712,
|
||
|
|
"grad_norm": 0.24996380450181893,
|
||
|
|
"learning_rate": 5.488881409427096e-06,
|
||
|
|
"loss": 0.4667,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.42274272441864014,
|
||
|
|
"step": 3540,
|
||
|
|
"valid_targets_mean": 4002.1,
|
||
|
|
"valid_targets_min": 292
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.487616099071207,
|
||
|
|
"grad_norm": 0.27509887225225904,
|
||
|
|
"learning_rate": 5.435857978660885e-06,
|
||
|
|
"loss": 0.4717,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5059709548950195,
|
||
|
|
"step": 3545,
|
||
|
|
"valid_targets_mean": 3102.1,
|
||
|
|
"valid_targets_min": 317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.495356037151703,
|
||
|
|
"grad_norm": 0.318472082749501,
|
||
|
|
"learning_rate": 5.383051592372277e-06,
|
||
|
|
"loss": 0.4509,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.44909152388572693,
|
||
|
|
"step": 3550,
|
||
|
|
"valid_targets_mean": 2004.2,
|
||
|
|
"valid_targets_min": 192
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.503095975232198,
|
||
|
|
"grad_norm": 0.27891640487290276,
|
||
|
|
"learning_rate": 5.330463037516964e-06,
|
||
|
|
"loss": 0.4737,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.47619572281837463,
|
||
|
|
"step": 3555,
|
||
|
|
"valid_targets_mean": 2664.8,
|
||
|
|
"valid_targets_min": 247
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.510835913312693,
|
||
|
|
"grad_norm": 0.2526652302618136,
|
||
|
|
"learning_rate": 5.278093097804377e-06,
|
||
|
|
"loss": 0.4599,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.47615906596183777,
|
||
|
|
"step": 3560,
|
||
|
|
"valid_targets_mean": 3044.0,
|
||
|
|
"valid_targets_min": 298
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.518575851393189,
|
||
|
|
"grad_norm": 0.22362771763233155,
|
||
|
|
"learning_rate": 5.225942553685987e-06,
|
||
|
|
"loss": 0.4532,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.47099441289901733,
|
||
|
|
"step": 3565,
|
||
|
|
"valid_targets_mean": 3938.2,
|
||
|
|
"valid_targets_min": 941
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.526315789473684,
|
||
|
|
"grad_norm": 0.23022426396991758,
|
||
|
|
"learning_rate": 5.1740121823437085e-06,
|
||
|
|
"loss": 0.4609,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4693618416786194,
|
||
|
|
"step": 3570,
|
||
|
|
"valid_targets_mean": 3717.0,
|
||
|
|
"valid_targets_min": 403
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.534055727554179,
|
||
|
|
"grad_norm": 0.21066099585336917,
|
||
|
|
"learning_rate": 5.1223027576782835e-06,
|
||
|
|
"loss": 0.4466,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.43319055438041687,
|
||
|
|
"step": 3575,
|
||
|
|
"valid_targets_mean": 4406.2,
|
||
|
|
"valid_targets_min": 387
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.541795665634675,
|
||
|
|
"grad_norm": 0.2462994585909714,
|
||
|
|
"learning_rate": 5.070815050297759e-06,
|
||
|
|
"loss": 0.4692,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5124592185020447,
|
||
|
|
"step": 3580,
|
||
|
|
"valid_targets_mean": 3612.3,
|
||
|
|
"valid_targets_min": 371
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.54953560371517,
|
||
|
|
"grad_norm": 0.2882684199682849,
|
||
|
|
"learning_rate": 5.019549827506003e-06,
|
||
|
|
"loss": 0.4441,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4470786154270172,
|
||
|
|
"step": 3585,
|
||
|
|
"valid_targets_mean": 2256.9,
|
||
|
|
"valid_targets_min": 283
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.557275541795666,
|
||
|
|
"grad_norm": 0.3015709751328135,
|
||
|
|
"learning_rate": 4.9685078532912755e-06,
|
||
|
|
"loss": 0.4566,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46981459856033325,
|
||
|
|
"step": 3590,
|
||
|
|
"valid_targets_mean": 2972.2,
|
||
|
|
"valid_targets_min": 318
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.565015479876161,
|
||
|
|
"grad_norm": 0.6162494876554508,
|
||
|
|
"learning_rate": 4.917689888314838e-06,
|
||
|
|
"loss": 0.5034,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5745257139205933,
|
||
|
|
"step": 3595,
|
||
|
|
"valid_targets_mean": 3760.0,
|
||
|
|
"valid_targets_min": 1056
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.572755417956657,
|
||
|
|
"grad_norm": 0.3804133268859842,
|
||
|
|
"learning_rate": 4.8670966898996105e-06,
|
||
|
|
"loss": 0.5871,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5360809564590454,
|
||
|
|
"step": 3600,
|
||
|
|
"valid_targets_mean": 4103.2,
|
||
|
|
"valid_targets_min": 797
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.580495356037152,
|
||
|
|
"grad_norm": 0.3154874738247873,
|
||
|
|
"learning_rate": 4.816729012018912e-06,
|
||
|
|
"loss": 0.603,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5993633270263672,
|
||
|
|
"step": 3605,
|
||
|
|
"valid_targets_mean": 3301.9,
|
||
|
|
"valid_targets_min": 791
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.588235294117647,
|
||
|
|
"grad_norm": 0.40853720271858424,
|
||
|
|
"learning_rate": 4.76658760528518e-06,
|
||
|
|
"loss": 0.5859,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5672604441642761,
|
||
|
|
"step": 3610,
|
||
|
|
"valid_targets_mean": 4432.9,
|
||
|
|
"valid_targets_min": 877
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.595975232198143,
|
||
|
|
"grad_norm": 0.2471719456157069,
|
||
|
|
"learning_rate": 4.716673216938834e-06,
|
||
|
|
"loss": 0.5759,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5674139261245728,
|
||
|
|
"step": 3615,
|
||
|
|
"valid_targets_mean": 4270.7,
|
||
|
|
"valid_targets_min": 907
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.603715170278638,
|
||
|
|
"grad_norm": 0.24661156417659658,
|
||
|
|
"learning_rate": 4.666986590837102e-06,
|
||
|
|
"loss": 0.5768,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.585923969745636,
|
||
|
|
"step": 3620,
|
||
|
|
"valid_targets_mean": 4288.7,
|
||
|
|
"valid_targets_min": 758
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.611455108359133,
|
||
|
|
"grad_norm": 0.2431070480908425,
|
||
|
|
"learning_rate": 4.617528467442947e-06,
|
||
|
|
"loss": 0.5792,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5570114850997925,
|
||
|
|
"step": 3625,
|
||
|
|
"valid_targets_mean": 4493.8,
|
||
|
|
"valid_targets_min": 1317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.6191950464396285,
|
||
|
|
"grad_norm": 0.2062020965348972,
|
||
|
|
"learning_rate": 4.568299583814031e-06,
|
||
|
|
"loss": 0.5627,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5259556174278259,
|
||
|
|
"step": 3630,
|
||
|
|
"valid_targets_mean": 4781.4,
|
||
|
|
"valid_targets_min": 671
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.6269349845201235,
|
||
|
|
"grad_norm": 0.23215009813694382,
|
||
|
|
"learning_rate": 4.519300673591754e-06,
|
||
|
|
"loss": 0.5763,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5511263608932495,
|
||
|
|
"step": 3635,
|
||
|
|
"valid_targets_mean": 4028.6,
|
||
|
|
"valid_targets_min": 1025
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.634674922600619,
|
||
|
|
"grad_norm": 0.2659801170809018,
|
||
|
|
"learning_rate": 4.4705324669902785e-06,
|
||
|
|
"loss": 0.566,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5898749828338623,
|
||
|
|
"step": 3640,
|
||
|
|
"valid_targets_mean": 3116.2,
|
||
|
|
"valid_targets_min": 442
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.642414860681114,
|
||
|
|
"grad_norm": 0.23164692818472982,
|
||
|
|
"learning_rate": 4.42199569078567e-06,
|
||
|
|
"loss": 0.5829,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6067973375320435,
|
||
|
|
"step": 3645,
|
||
|
|
"valid_targets_mean": 4106.4,
|
||
|
|
"valid_targets_min": 854
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.65015479876161,
|
||
|
|
"grad_norm": 0.24650304948341,
|
||
|
|
"learning_rate": 4.373691068305081e-06,
|
||
|
|
"loss": 0.5755,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5628385543823242,
|
||
|
|
"step": 3650,
|
||
|
|
"valid_targets_mean": 3609.7,
|
||
|
|
"valid_targets_min": 1029
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.657894736842105,
|
||
|
|
"grad_norm": 0.2335450138318682,
|
||
|
|
"learning_rate": 4.325619319415943e-06,
|
||
|
|
"loss": 0.5865,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5749136209487915,
|
||
|
|
"step": 3655,
|
||
|
|
"valid_targets_mean": 4175.3,
|
||
|
|
"valid_targets_min": 1010
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.6656346749226,
|
||
|
|
"grad_norm": 0.22867187021217783,
|
||
|
|
"learning_rate": 4.277781160515249e-06,
|
||
|
|
"loss": 0.5659,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5563583970069885,
|
||
|
|
"step": 3660,
|
||
|
|
"valid_targets_mean": 3942.0,
|
||
|
|
"valid_targets_min": 1104
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.673374613003096,
|
||
|
|
"grad_norm": 0.24600419361659334,
|
||
|
|
"learning_rate": 4.230177304518892e-06,
|
||
|
|
"loss": 0.5845,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5792802572250366,
|
||
|
|
"step": 3665,
|
||
|
|
"valid_targets_mean": 3787.9,
|
||
|
|
"valid_targets_min": 776
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.681114551083591,
|
||
|
|
"grad_norm": 0.590460402083745,
|
||
|
|
"learning_rate": 4.18280846085102e-06,
|
||
|
|
"loss": 0.5417,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5060890913009644,
|
||
|
|
"step": 3670,
|
||
|
|
"valid_targets_mean": 3564.7,
|
||
|
|
"valid_targets_min": 900
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.688854489164086,
|
||
|
|
"grad_norm": 0.33161022492475034,
|
||
|
|
"learning_rate": 4.135675335433469e-06,
|
||
|
|
"loss": 0.4888,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4605845808982849,
|
||
|
|
"step": 3675,
|
||
|
|
"valid_targets_mean": 4363.4,
|
||
|
|
"valid_targets_min": 750
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.696594427244582,
|
||
|
|
"grad_norm": 0.27858026043777906,
|
||
|
|
"learning_rate": 4.088778630675252e-06,
|
||
|
|
"loss": 0.4778,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4550124406814575,
|
||
|
|
"step": 3680,
|
||
|
|
"valid_targets_mean": 3516.3,
|
||
|
|
"valid_targets_min": 767
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.704334365325077,
|
||
|
|
"grad_norm": 0.2493805981064756,
|
||
|
|
"learning_rate": 4.042119045462074e-06,
|
||
|
|
"loss": 0.4804,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4591600298881531,
|
||
|
|
"step": 3685,
|
||
|
|
"valid_targets_mean": 4421.9,
|
||
|
|
"valid_targets_min": 880
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.712074303405573,
|
||
|
|
"grad_norm": 0.27835291256825917,
|
||
|
|
"learning_rate": 3.9956972751459515e-06,
|
||
|
|
"loss": 0.4825,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48625510931015015,
|
||
|
|
"step": 3690,
|
||
|
|
"valid_targets_mean": 3265.3,
|
||
|
|
"valid_targets_min": 906
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.719814241486068,
|
||
|
|
"grad_norm": 0.29757773725269804,
|
||
|
|
"learning_rate": 3.949514011534794e-06,
|
||
|
|
"loss": 0.4588,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4274042248725891,
|
||
|
|
"step": 3695,
|
||
|
|
"valid_targets_mean": 4266.4,
|
||
|
|
"valid_targets_min": 1047
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.727554179566564,
|
||
|
|
"grad_norm": 0.24347983792179456,
|
||
|
|
"learning_rate": 3.9035699428821575e-06,
|
||
|
|
"loss": 0.4838,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4861769676208496,
|
||
|
|
"step": 3700,
|
||
|
|
"valid_targets_mean": 3498.2,
|
||
|
|
"valid_targets_min": 1036
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.735294117647059,
|
||
|
|
"grad_norm": 0.22966275501218952,
|
||
|
|
"learning_rate": 3.857865753876935e-06,
|
||
|
|
"loss": 0.4661,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4672721028327942,
|
||
|
|
"step": 3705,
|
||
|
|
"valid_targets_mean": 4387.9,
|
||
|
|
"valid_targets_min": 635
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.743034055727554,
|
||
|
|
"grad_norm": 0.23358947681239817,
|
||
|
|
"learning_rate": 3.812402125633181e-06,
|
||
|
|
"loss": 0.4698,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4631599187850952,
|
||
|
|
"step": 3710,
|
||
|
|
"valid_targets_mean": 3467.2,
|
||
|
|
"valid_targets_min": 695
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.75077399380805,
|
||
|
|
"grad_norm": 0.1987129439176276,
|
||
|
|
"learning_rate": 3.767179735679958e-06,
|
||
|
|
"loss": 0.4574,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46047163009643555,
|
||
|
|
"step": 3715,
|
||
|
|
"valid_targets_mean": 4467.1,
|
||
|
|
"valid_targets_min": 613
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.758513931888545,
|
||
|
|
"grad_norm": 0.2390767036262366,
|
||
|
|
"learning_rate": 3.722199257951222e-06,
|
||
|
|
"loss": 0.476,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.49894654750823975,
|
||
|
|
"step": 3720,
|
||
|
|
"valid_targets_mean": 3536.6,
|
||
|
|
"valid_targets_min": 834
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.76625386996904,
|
||
|
|
"grad_norm": 0.22023687532168928,
|
||
|
|
"learning_rate": 3.6774613627758137e-06,
|
||
|
|
"loss": 0.4692,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4829775393009186,
|
||
|
|
"step": 3725,
|
||
|
|
"valid_targets_mean": 4088.9,
|
||
|
|
"valid_targets_min": 837
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.773993808049536,
|
||
|
|
"grad_norm": 0.2326601342243223,
|
||
|
|
"learning_rate": 3.6329667168674432e-06,
|
||
|
|
"loss": 0.4716,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.47471004724502563,
|
||
|
|
"step": 3730,
|
||
|
|
"valid_targets_mean": 3560.9,
|
||
|
|
"valid_targets_min": 1032
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.781733746130031,
|
||
|
|
"grad_norm": 0.20224485771148334,
|
||
|
|
"learning_rate": 3.5887159833147457e-06,
|
||
|
|
"loss": 0.4661,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.45915916562080383,
|
||
|
|
"step": 3735,
|
||
|
|
"valid_targets_mean": 4477.0,
|
||
|
|
"valid_targets_min": 502
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.7894736842105265,
|
||
|
|
"grad_norm": 0.3197928529621115,
|
||
|
|
"learning_rate": 3.5447098215714216e-06,
|
||
|
|
"loss": 0.4656,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4737987518310547,
|
||
|
|
"step": 3740,
|
||
|
|
"valid_targets_mean": 3193.1,
|
||
|
|
"valid_targets_min": 883
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.7972136222910216,
|
||
|
|
"grad_norm": 0.2087739962429902,
|
||
|
|
"learning_rate": 3.5009488874464027e-06,
|
||
|
|
"loss": 0.4769,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46324607729911804,
|
||
|
|
"step": 3745,
|
||
|
|
"valid_targets_mean": 4601.6,
|
||
|
|
"valid_targets_min": 602
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.804953560371517,
|
||
|
|
"grad_norm": 0.235603717126632,
|
||
|
|
"learning_rate": 3.4574338330940727e-06,
|
||
|
|
"loss": 0.4669,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4816066026687622,
|
||
|
|
"step": 3750,
|
||
|
|
"valid_targets_mean": 3430.9,
|
||
|
|
"valid_targets_min": 1017
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.812693498452012,
|
||
|
|
"grad_norm": 0.22300738921016897,
|
||
|
|
"learning_rate": 3.414165307004551e-06,
|
||
|
|
"loss": 0.4553,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4534761905670166,
|
||
|
|
"step": 3755,
|
||
|
|
"valid_targets_mean": 3828.6,
|
||
|
|
"valid_targets_min": 1071
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.820433436532507,
|
||
|
|
"grad_norm": 0.24840600029239254,
|
||
|
|
"learning_rate": 3.3711439539940384e-06,
|
||
|
|
"loss": 0.545,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46194446086883545,
|
||
|
|
"step": 3760,
|
||
|
|
"valid_targets_mean": 3390.7,
|
||
|
|
"valid_targets_min": 996
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.828173374613003,
|
||
|
|
"grad_norm": 0.2270107975328524,
|
||
|
|
"learning_rate": 3.328370415195188e-06,
|
||
|
|
"loss": 0.4655,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4496748447418213,
|
||
|
|
"step": 3765,
|
||
|
|
"valid_targets_mean": 3718.5,
|
||
|
|
"valid_targets_min": 819
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.835913312693498,
|
||
|
|
"grad_norm": 0.2537842544308086,
|
||
|
|
"learning_rate": 3.2858453280475745e-06,
|
||
|
|
"loss": 0.5559,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4518022835254669,
|
||
|
|
"step": 3770,
|
||
|
|
"valid_targets_mean": 3193.4,
|
||
|
|
"valid_targets_min": 558
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.843653250773993,
|
||
|
|
"grad_norm": 0.22494456032870389,
|
||
|
|
"learning_rate": 3.2435693262881675e-06,
|
||
|
|
"loss": 0.4586,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.44924771785736084,
|
||
|
|
"step": 3775,
|
||
|
|
"valid_targets_mean": 3690.3,
|
||
|
|
"valid_targets_min": 434
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.851393188854489,
|
||
|
|
"grad_norm": 0.2467891393986999,
|
||
|
|
"learning_rate": 3.20154303994191e-06,
|
||
|
|
"loss": 0.514,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.47867903113365173,
|
||
|
|
"step": 3780,
|
||
|
|
"valid_targets_mean": 3144.1,
|
||
|
|
"valid_targets_min": 742
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.859133126934984,
|
||
|
|
"grad_norm": 0.2260680213371111,
|
||
|
|
"learning_rate": 3.1597670953123137e-06,
|
||
|
|
"loss": 0.4576,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4645009934902191,
|
||
|
|
"step": 3785,
|
||
|
|
"valid_targets_mean": 3871.4,
|
||
|
|
"valid_targets_min": 502
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.86687306501548,
|
||
|
|
"grad_norm": 0.2537287338737017,
|
||
|
|
"learning_rate": 3.1182421149721454e-06,
|
||
|
|
"loss": 0.5101,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4672238230705261,
|
||
|
|
"step": 3790,
|
||
|
|
"valid_targets_mean": 3256.5,
|
||
|
|
"valid_targets_min": 999
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.874613003095975,
|
||
|
|
"grad_norm": 0.24822660029616553,
|
||
|
|
"learning_rate": 3.076968717754125e-06,
|
||
|
|
"loss": 0.4597,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.47167205810546875,
|
||
|
|
"step": 3795,
|
||
|
|
"valid_targets_mean": 3665.0,
|
||
|
|
"valid_targets_min": 640
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.882352941176471,
|
||
|
|
"grad_norm": 0.26241269257707805,
|
||
|
|
"learning_rate": 3.035947518741713e-06,
|
||
|
|
"loss": 0.5552,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48451489210128784,
|
||
|
|
"step": 3800,
|
||
|
|
"valid_targets_mean": 3457.3,
|
||
|
|
"valid_targets_min": 1110
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.890092879256966,
|
||
|
|
"grad_norm": 0.24091140151012463,
|
||
|
|
"learning_rate": 2.9951791292599596e-06,
|
||
|
|
"loss": 0.4526,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4411616921424866,
|
||
|
|
"step": 3805,
|
||
|
|
"valid_targets_mean": 3141.0,
|
||
|
|
"valid_targets_min": 566
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.897832817337461,
|
||
|
|
"grad_norm": 0.26035852676543286,
|
||
|
|
"learning_rate": 2.9546641568663625e-06,
|
||
|
|
"loss": 0.5425,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46469932794570923,
|
||
|
|
"step": 3810,
|
||
|
|
"valid_targets_mean": 3509.7,
|
||
|
|
"valid_targets_min": 780
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.905572755417957,
|
||
|
|
"grad_norm": 0.2672929499198019,
|
||
|
|
"learning_rate": 2.914403205341838e-06,
|
||
|
|
"loss": 0.464,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4433569312095642,
|
||
|
|
"step": 3815,
|
||
|
|
"valid_targets_mean": 3106.4,
|
||
|
|
"valid_targets_min": 794
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.913312693498452,
|
||
|
|
"grad_norm": 0.24433056516759502,
|
||
|
|
"learning_rate": 2.874396874681722e-06,
|
||
|
|
"loss": 0.4958,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.45166558027267456,
|
||
|
|
"step": 3820,
|
||
|
|
"valid_targets_mean": 3530.0,
|
||
|
|
"valid_targets_min": 744
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.921052631578947,
|
||
|
|
"grad_norm": 0.2496398970671602,
|
||
|
|
"learning_rate": 2.8346457610868092e-06,
|
||
|
|
"loss": 0.4635,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4738770127296448,
|
||
|
|
"step": 3825,
|
||
|
|
"valid_targets_mean": 3219.6,
|
||
|
|
"valid_targets_min": 435
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.928792569659443,
|
||
|
|
"grad_norm": 0.7918384022233264,
|
||
|
|
"learning_rate": 2.795150456954492e-06,
|
||
|
|
"loss": 0.4397,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28021833300590515,
|
||
|
|
"step": 3830,
|
||
|
|
"valid_targets_mean": 4948.2,
|
||
|
|
"valid_targets_min": 1383
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.936532507739938,
|
||
|
|
"grad_norm": 0.7060298906130598,
|
||
|
|
"learning_rate": 2.755911550869914e-06,
|
||
|
|
"loss": 0.2644,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2414156198501587,
|
||
|
|
"step": 3835,
|
||
|
|
"valid_targets_mean": 4927.9,
|
||
|
|
"valid_targets_min": 1102
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.944272445820434,
|
||
|
|
"grad_norm": 0.5924071782052266,
|
||
|
|
"learning_rate": 2.7169296275972e-06,
|
||
|
|
"loss": 0.2491,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26075243949890137,
|
||
|
|
"step": 3840,
|
||
|
|
"valid_targets_mean": 4602.5,
|
||
|
|
"valid_targets_min": 1297
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.952012383900929,
|
||
|
|
"grad_norm": 0.354719583070335,
|
||
|
|
"learning_rate": 2.6782052680707638e-06,
|
||
|
|
"loss": 0.2196,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21986541152000427,
|
||
|
|
"step": 3845,
|
||
|
|
"valid_targets_mean": 4983.0,
|
||
|
|
"valid_targets_min": 938
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.959752321981425,
|
||
|
|
"grad_norm": 0.32815134501043863,
|
||
|
|
"learning_rate": 2.639739049386625e-06,
|
||
|
|
"loss": 0.2287,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2353232502937317,
|
||
|
|
"step": 3850,
|
||
|
|
"valid_targets_mean": 4184.5,
|
||
|
|
"valid_targets_min": 1158
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.96749226006192,
|
||
|
|
"grad_norm": 0.29354431176592993,
|
||
|
|
"learning_rate": 2.601531544793816e-06,
|
||
|
|
"loss": 0.2349,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.227220356464386,
|
||
|
|
"step": 3855,
|
||
|
|
"valid_targets_mean": 4258.2,
|
||
|
|
"valid_targets_min": 662
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.975232198142415,
|
||
|
|
"grad_norm": 0.408383768585089,
|
||
|
|
"learning_rate": 2.563583323685839e-06,
|
||
|
|
"loss": 0.234,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22872968018054962,
|
||
|
|
"step": 3860,
|
||
|
|
"valid_targets_mean": 4959.4,
|
||
|
|
"valid_targets_min": 2049
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.9829721362229105,
|
||
|
|
"grad_norm": 0.7386918214808669,
|
||
|
|
"learning_rate": 2.5258949515921872e-06,
|
||
|
|
"loss": 0.2301,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23305855691432953,
|
||
|
|
"step": 3865,
|
||
|
|
"valid_targets_mean": 3119.1,
|
||
|
|
"valid_targets_min": 1116
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.9907120743034055,
|
||
|
|
"grad_norm": 0.5201586363397805,
|
||
|
|
"learning_rate": 2.488466990169911e-06,
|
||
|
|
"loss": 0.2303,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2162656933069229,
|
||
|
|
"step": 3870,
|
||
|
|
"valid_targets_mean": 2866.9,
|
||
|
|
"valid_targets_min": 1189
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.9984520123839005,
|
||
|
|
"grad_norm": 0.3768004723743164,
|
||
|
|
"learning_rate": 2.45129999719524e-06,
|
||
|
|
"loss": 0.2205,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23022907972335815,
|
||
|
|
"step": 3875,
|
||
|
|
"valid_targets_mean": 3248.7,
|
||
|
|
"valid_targets_min": 1746
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.006191950464396,
|
||
|
|
"grad_norm": 0.7308253542609516,
|
||
|
|
"learning_rate": 2.4143945265552927e-06,
|
||
|
|
"loss": 0.243,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2502497434616089,
|
||
|
|
"step": 3880,
|
||
|
|
"valid_targets_mean": 3799.7,
|
||
|
|
"valid_targets_min": 470
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.013931888544891,
|
||
|
|
"grad_norm": 0.5156224831580052,
|
||
|
|
"learning_rate": 2.377751128239809e-06,
|
||
|
|
"loss": 0.2454,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24603943526744843,
|
||
|
|
"step": 3885,
|
||
|
|
"valid_targets_mean": 3723.8,
|
||
|
|
"valid_targets_min": 388
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.021671826625387,
|
||
|
|
"grad_norm": 0.37028415928517333,
|
||
|
|
"learning_rate": 2.341370348332941e-06,
|
||
|
|
"loss": 0.2385,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2414015829563141,
|
||
|
|
"step": 3890,
|
||
|
|
"valid_targets_mean": 3488.4,
|
||
|
|
"valid_targets_min": 338
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.029411764705882,
|
||
|
|
"grad_norm": 0.321020586374499,
|
||
|
|
"learning_rate": 2.305252729005134e-06,
|
||
|
|
"loss": 0.2351,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.226506307721138,
|
||
|
|
"step": 3895,
|
||
|
|
"valid_targets_mean": 3832.2,
|
||
|
|
"valid_targets_min": 506
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.037151702786378,
|
||
|
|
"grad_norm": 0.27588207136124954,
|
||
|
|
"learning_rate": 2.2693988085050435e-06,
|
||
|
|
"loss": 0.23,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22368520498275757,
|
||
|
|
"step": 3900,
|
||
|
|
"valid_targets_mean": 3689.5,
|
||
|
|
"valid_targets_min": 397
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.044891640866873,
|
||
|
|
"grad_norm": 0.24698080813322737,
|
||
|
|
"learning_rate": 2.2338091211515025e-06,
|
||
|
|
"loss": 0.2282,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21921437978744507,
|
||
|
|
"step": 3905,
|
||
|
|
"valid_targets_mean": 3592.6,
|
||
|
|
"valid_targets_min": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.052631578947368,
|
||
|
|
"grad_norm": 0.24160586592370403,
|
||
|
|
"learning_rate": 2.1984841973255787e-06,
|
||
|
|
"loss": 0.2265,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2258760929107666,
|
||
|
|
"step": 3910,
|
||
|
|
"valid_targets_mean": 3646.3,
|
||
|
|
"valid_targets_min": 428
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.060371517027864,
|
||
|
|
"grad_norm": 0.20852630625256058,
|
||
|
|
"learning_rate": 2.1634245634626504e-06,
|
||
|
|
"loss": 0.2242,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22488918900489807,
|
||
|
|
"step": 3915,
|
||
|
|
"valid_targets_mean": 3811.8,
|
||
|
|
"valid_targets_min": 430
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.068111455108359,
|
||
|
|
"grad_norm": 0.21750025216024813,
|
||
|
|
"learning_rate": 2.128630742044564e-06,
|
||
|
|
"loss": 0.2218,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2304970920085907,
|
||
|
|
"step": 3920,
|
||
|
|
"valid_targets_mean": 3824.1,
|
||
|
|
"valid_targets_min": 349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.075851393188855,
|
||
|
|
"grad_norm": 0.20376927741543333,
|
||
|
|
"learning_rate": 2.0941032515918702e-06,
|
||
|
|
"loss": 0.2224,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21400514245033264,
|
||
|
|
"step": 3925,
|
||
|
|
"valid_targets_mean": 3658.0,
|
||
|
|
"valid_targets_min": 444
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.08359133126935,
|
||
|
|
"grad_norm": 0.22239360261588229,
|
||
|
|
"learning_rate": 2.0598426066560663e-06,
|
||
|
|
"loss": 0.225,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22495588660240173,
|
||
|
|
"step": 3930,
|
||
|
|
"valid_targets_mean": 3583.1,
|
||
|
|
"valid_targets_min": 457
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.091331269349845,
|
||
|
|
"grad_norm": 0.4783966342138252,
|
||
|
|
"learning_rate": 2.0258493178119364e-06,
|
||
|
|
"loss": 0.2569,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26994526386260986,
|
||
|
|
"step": 3935,
|
||
|
|
"valid_targets_mean": 4880.6,
|
||
|
|
"valid_targets_min": 3320
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.099071207430341,
|
||
|
|
"grad_norm": 0.7133255962783707,
|
||
|
|
"learning_rate": 1.9921238916499664e-06,
|
||
|
|
"loss": 0.2932,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4310452342033386,
|
||
|
|
"step": 3940,
|
||
|
|
"valid_targets_mean": 978.9,
|
||
|
|
"valid_targets_min": 200
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.106811145510836,
|
||
|
|
"grad_norm": 0.28513898616323047,
|
||
|
|
"learning_rate": 1.9586668307687605e-06,
|
||
|
|
"loss": 0.3405,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25778090953826904,
|
||
|
|
"step": 3945,
|
||
|
|
"valid_targets_mean": 4807.5,
|
||
|
|
"valid_targets_min": 2887
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.114551083591332,
|
||
|
|
"grad_norm": 0.6196759705156598,
|
||
|
|
"learning_rate": 1.925478633767566e-06,
|
||
|
|
"loss": 0.3071,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4638613164424896,
|
||
|
|
"step": 3950,
|
||
|
|
"valid_targets_mean": 870.8,
|
||
|
|
"valid_targets_min": 239
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.122291021671827,
|
||
|
|
"grad_norm": 0.19699579106490298,
|
||
|
|
"learning_rate": 1.8925597952388464e-06,
|
||
|
|
"loss": 0.2978,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2559766173362732,
|
||
|
|
"step": 3955,
|
||
|
|
"valid_targets_mean": 4695.3,
|
||
|
|
"valid_targets_min": 3169
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.130030959752322,
|
||
|
|
"grad_norm": 0.6051662401240115,
|
||
|
|
"learning_rate": 1.8599108057609116e-06,
|
||
|
|
"loss": 0.2986,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4490961730480194,
|
||
|
|
"step": 3960,
|
||
|
|
"valid_targets_mean": 994.2,
|
||
|
|
"valid_targets_min": 414
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.137770897832818,
|
||
|
|
"grad_norm": 0.19747305131471982,
|
||
|
|
"learning_rate": 1.8275321518905943e-06,
|
||
|
|
"loss": 0.35,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24954885244369507,
|
||
|
|
"step": 3965,
|
||
|
|
"valid_targets_mean": 4788.1,
|
||
|
|
"valid_targets_min": 2297
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.145510835913313,
|
||
|
|
"grad_norm": 0.2407667586566885,
|
||
|
|
"learning_rate": 1.7954243161560136e-06,
|
||
|
|
"loss": 0.2607,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27768316864967346,
|
||
|
|
"step": 3970,
|
||
|
|
"valid_targets_mean": 3206.7,
|
||
|
|
"valid_targets_min": 447
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.153250773993808,
|
||
|
|
"grad_norm": 0.18871312920867617,
|
||
|
|
"learning_rate": 1.7635877770493737e-06,
|
||
|
|
"loss": 0.3738,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26610785722732544,
|
||
|
|
"step": 3975,
|
||
|
|
"valid_targets_mean": 4860.4,
|
||
|
|
"valid_targets_min": 3021
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.1609907120743035,
|
||
|
|
"grad_norm": 0.172809498606094,
|
||
|
|
"learning_rate": 1.7320230090198365e-06,
|
||
|
|
"loss": 0.2561,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2591450810432434,
|
||
|
|
"step": 3980,
|
||
|
|
"valid_targets_mean": 4770.2,
|
||
|
|
"valid_targets_min": 2896
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.1687306501547985,
|
||
|
|
"grad_norm": 0.18948640535205277,
|
||
|
|
"learning_rate": 1.700730482466455e-06,
|
||
|
|
"loss": 0.3608,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2637854218482971,
|
||
|
|
"step": 3985,
|
||
|
|
"valid_targets_mean": 4724.9,
|
||
|
|
"valid_targets_min": 3054
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.176470588235294,
|
||
|
|
"grad_norm": 0.182687504924408,
|
||
|
|
"learning_rate": 1.669710663731159e-06,
|
||
|
|
"loss": 0.2548,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2637596130371094,
|
||
|
|
"step": 3990,
|
||
|
|
"valid_targets_mean": 4654.0,
|
||
|
|
"valid_targets_min": 1994
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.184210526315789,
|
||
|
|
"grad_norm": 0.1888565103072134,
|
||
|
|
"learning_rate": 1.6389640150917973e-06,
|
||
|
|
"loss": 0.3771,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25524628162384033,
|
||
|
|
"step": 3995,
|
||
|
|
"valid_targets_mean": 4692.8,
|
||
|
|
"valid_targets_min": 742
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.191950464396285,
|
||
|
|
"grad_norm": 0.17576705375435045,
|
||
|
|
"learning_rate": 1.6084909947552695e-06,
|
||
|
|
"loss": 0.2531,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25874072313308716,
|
||
|
|
"step": 4000,
|
||
|
|
"valid_targets_mean": 5052.9,
|
||
|
|
"valid_targets_min": 2772
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.19969040247678,
|
||
|
|
"grad_norm": 0.18197073043526465,
|
||
|
|
"learning_rate": 1.5782920568506853e-06,
|
||
|
|
"loss": 0.3816,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26193246245384216,
|
||
|
|
"step": 4005,
|
||
|
|
"valid_targets_mean": 5030.0,
|
||
|
|
"valid_targets_min": 3232
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.207430340557275,
|
||
|
|
"grad_norm": 0.17484204017116525,
|
||
|
|
"learning_rate": 1.5483676514225887e-06,
|
||
|
|
"loss": 0.2577,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25444304943084717,
|
||
|
|
"step": 4010,
|
||
|
|
"valid_targets_mean": 5112.5,
|
||
|
|
"valid_targets_min": 3319
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.215170278637771,
|
||
|
|
"grad_norm": 0.26528355418857497,
|
||
|
|
"learning_rate": 1.5187182244242627e-06,
|
||
|
|
"loss": 0.3882,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28013479709625244,
|
||
|
|
"step": 4015,
|
||
|
|
"valid_targets_mean": 5787.9,
|
||
|
|
"valid_targets_min": 628
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.222910216718266,
|
||
|
|
"grad_norm": 0.28688907001821856,
|
||
|
|
"learning_rate": 1.4893442177110751e-06,
|
||
|
|
"loss": 0.2832,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29207104444503784,
|
||
|
|
"step": 4020,
|
||
|
|
"valid_targets_mean": 6091.0,
|
||
|
|
"valid_targets_min": 590
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.230650154798761,
|
||
|
|
"grad_norm": 0.23953807873931207,
|
||
|
|
"learning_rate": 1.4602460690339082e-06,
|
||
|
|
"loss": 0.2892,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29241496324539185,
|
||
|
|
"step": 4025,
|
||
|
|
"valid_targets_mean": 5457.9,
|
||
|
|
"valid_targets_min": 537
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.238390092879257,
|
||
|
|
"grad_norm": 0.24161066913521997,
|
||
|
|
"learning_rate": 1.431424212032615e-06,
|
||
|
|
"loss": 0.2814,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28175821900367737,
|
||
|
|
"step": 4030,
|
||
|
|
"valid_targets_mean": 5283.4,
|
||
|
|
"valid_targets_min": 556
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.246130030959752,
|
||
|
|
"grad_norm": 0.20258772202152925,
|
||
|
|
"learning_rate": 1.4028790762295773e-06,
|
||
|
|
"loss": 0.2901,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27142903208732605,
|
||
|
|
"step": 4035,
|
||
|
|
"valid_targets_mean": 6079.6,
|
||
|
|
"valid_targets_min": 531
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.253869969040248,
|
||
|
|
"grad_norm": 0.21862880127557435,
|
||
|
|
"learning_rate": 1.3746110870232854e-06,
|
||
|
|
"loss": 0.2825,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3030893802642822,
|
||
|
|
"step": 4040,
|
||
|
|
"valid_targets_mean": 5938.4,
|
||
|
|
"valid_targets_min": 588
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.261609907120743,
|
||
|
|
"grad_norm": 0.23123871639392077,
|
||
|
|
"learning_rate": 1.3466206656820168e-06,
|
||
|
|
"loss": 0.282,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2797814607620239,
|
||
|
|
"step": 4045,
|
||
|
|
"valid_targets_mean": 5991.7,
|
||
|
|
"valid_targets_min": 605
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.269349845201239,
|
||
|
|
"grad_norm": 0.20347940582659751,
|
||
|
|
"learning_rate": 1.318908229337541e-06,
|
||
|
|
"loss": 0.2855,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.276038259267807,
|
||
|
|
"step": 4050,
|
||
|
|
"valid_targets_mean": 5768.7,
|
||
|
|
"valid_targets_min": 589
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.277089783281734,
|
||
|
|
"grad_norm": 0.21753019821614544,
|
||
|
|
"learning_rate": 1.2914741909789164e-06,
|
||
|
|
"loss": 0.2798,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2776251435279846,
|
||
|
|
"step": 4055,
|
||
|
|
"valid_targets_mean": 5973.1,
|
||
|
|
"valid_targets_min": 598
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.284829721362229,
|
||
|
|
"grad_norm": 0.20377603832435986,
|
||
|
|
"learning_rate": 1.26431895944632e-06,
|
||
|
|
"loss": 0.2829,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28434571623802185,
|
||
|
|
"step": 4060,
|
||
|
|
"valid_targets_mean": 5782.9,
|
||
|
|
"valid_targets_min": 427
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.292569659442725,
|
||
|
|
"grad_norm": 0.20969590623223214,
|
||
|
|
"learning_rate": 1.2374429394249865e-06,
|
||
|
|
"loss": 0.2764,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27738112211227417,
|
||
|
|
"step": 4065,
|
||
|
|
"valid_targets_mean": 5733.2,
|
||
|
|
"valid_targets_min": 768
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.30030959752322,
|
||
|
|
"grad_norm": 0.28304297043937127,
|
||
|
|
"learning_rate": 1.2108465314391337e-06,
|
||
|
|
"loss": 0.2825,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28616419434547424,
|
||
|
|
"step": 4070,
|
||
|
|
"valid_targets_mean": 5674.5,
|
||
|
|
"valid_targets_min": 597
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.308049535603715,
|
||
|
|
"grad_norm": 0.20562462269077536,
|
||
|
|
"learning_rate": 1.1845301318460267e-06,
|
||
|
|
"loss": 0.2766,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28878653049468994,
|
||
|
|
"step": 4075,
|
||
|
|
"valid_targets_mean": 5429.0,
|
||
|
|
"valid_targets_min": 755
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.315789473684211,
|
||
|
|
"grad_norm": 0.19661810947933292,
|
||
|
|
"learning_rate": 1.1584941328300615e-06,
|
||
|
|
"loss": 0.2792,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2684420645236969,
|
||
|
|
"step": 4080,
|
||
|
|
"valid_targets_mean": 5852.8,
|
||
|
|
"valid_targets_min": 1304
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.323529411764706,
|
||
|
|
"grad_norm": 0.22774367307594892,
|
||
|
|
"learning_rate": 1.1327389223969143e-06,
|
||
|
|
"loss": 0.2787,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29256555438041687,
|
||
|
|
"step": 4085,
|
||
|
|
"valid_targets_mean": 5466.6,
|
||
|
|
"valid_targets_min": 1012
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.3312693498452015,
|
||
|
|
"grad_norm": 0.2135231971679428,
|
||
|
|
"learning_rate": 1.107264884367767e-06,
|
||
|
|
"loss": 0.2779,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26805704832077026,
|
||
|
|
"step": 4090,
|
||
|
|
"valid_targets_mean": 5432.9,
|
||
|
|
"valid_targets_min": 495
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.3390092879256965,
|
||
|
|
"grad_norm": 0.19695173246043535,
|
||
|
|
"learning_rate": 1.0820723983735837e-06,
|
||
|
|
"loss": 0.2793,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28790634870529175,
|
||
|
|
"step": 4095,
|
||
|
|
"valid_targets_mean": 6355.9,
|
||
|
|
"valid_targets_min": 629
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.346749226006192,
|
||
|
|
"grad_norm": 0.6449801276122118,
|
||
|
|
"learning_rate": 1.0571618398494544e-06,
|
||
|
|
"loss": 0.3675,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5269606113433838,
|
||
|
|
"step": 4100,
|
||
|
|
"valid_targets_mean": 805.4,
|
||
|
|
"valid_targets_min": 438
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.354489164086687,
|
||
|
|
"grad_norm": 0.18394470375657654,
|
||
|
|
"learning_rate": 1.032533580028998e-06,
|
||
|
|
"loss": 0.2814,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2813188433647156,
|
||
|
|
"step": 4105,
|
||
|
|
"valid_targets_mean": 6133.8,
|
||
|
|
"valid_targets_min": 229
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.362229102167182,
|
||
|
|
"grad_norm": 0.608361158967074,
|
||
|
|
"learning_rate": 1.0081879859388332e-06,
|
||
|
|
"loss": 0.3454,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5396931171417236,
|
||
|
|
"step": 4110,
|
||
|
|
"valid_targets_mean": 817.7,
|
||
|
|
"valid_targets_min": 373
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.369969040247678,
|
||
|
|
"grad_norm": 0.17220662803406958,
|
||
|
|
"learning_rate": 9.841254203931028e-07,
|
||
|
|
"loss": 0.2806,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2705855369567871,
|
||
|
|
"step": 4115,
|
||
|
|
"valid_targets_mean": 6547.3,
|
||
|
|
"valid_targets_min": 1808
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.377708978328173,
|
||
|
|
"grad_norm": 0.16483199745916047,
|
||
|
|
"learning_rate": 9.60346241988075e-07,
|
||
|
|
"loss": 0.3482,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2798342704772949,
|
||
|
|
"step": 4120,
|
||
|
|
"valid_targets_mean": 6934.5,
|
||
|
|
"valid_targets_min": 376
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.385448916408668,
|
||
|
|
"grad_norm": 0.17082637483633792,
|
||
|
|
"learning_rate": 9.368508050968005e-07,
|
||
|
|
"loss": 0.2775,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2779572606086731,
|
||
|
|
"step": 4125,
|
||
|
|
"valid_targets_mean": 6416.0,
|
||
|
|
"valid_targets_min": 451
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.393188854489164,
|
||
|
|
"grad_norm": 0.19558536769052057,
|
||
|
|
"learning_rate": 9.136394598638177e-07,
|
||
|
|
"loss": 0.3885,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2838422358036041,
|
||
|
|
"step": 4130,
|
||
|
|
"valid_targets_mean": 6050.4,
|
||
|
|
"valid_targets_min": 298
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.400928792569659,
|
||
|
|
"grad_norm": 0.18042613601020874,
|
||
|
|
"learning_rate": 8.907125521999438e-07,
|
||
|
|
"loss": 0.2745,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26712673902511597,
|
||
|
|
"step": 4135,
|
||
|
|
"valid_targets_mean": 5863.9,
|
||
|
|
"valid_targets_min": 212
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.408668730650155,
|
||
|
|
"grad_norm": 0.19805229902080462,
|
||
|
|
"learning_rate": 8.680704237771209e-07,
|
||
|
|
"loss": 0.3541,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28546249866485596,
|
||
|
|
"step": 4140,
|
||
|
|
"valid_targets_mean": 6129.1,
|
||
|
|
"valid_targets_min": 324
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.41640866873065,
|
||
|
|
"grad_norm": 0.1774281545465225,
|
||
|
|
"learning_rate": 8.457134120233279e-07,
|
||
|
|
"loss": 0.2725,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2717171907424927,
|
||
|
|
"step": 4145,
|
||
|
|
"valid_targets_mean": 6139.4,
|
||
|
|
"valid_targets_min": 478
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.424148606811146,
|
||
|
|
"grad_norm": 0.18363504928784372,
|
||
|
|
"learning_rate": 8.236418501175336e-07,
|
||
|
|
"loss": 0.3875,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28682827949523926,
|
||
|
|
"step": 4150,
|
||
|
|
"valid_targets_mean": 6069.3,
|
||
|
|
"valid_targets_min": 230
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.431888544891641,
|
||
|
|
"grad_norm": 0.1726761626152954,
|
||
|
|
"learning_rate": 8.018560669847564e-07,
|
||
|
|
"loss": 0.2712,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27283775806427,
|
||
|
|
"step": 4155,
|
||
|
|
"valid_targets_mean": 6344.3,
|
||
|
|
"valid_targets_min": 600
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.439628482972136,
|
||
|
|
"grad_norm": 0.20999437151015302,
|
||
|
|
"learning_rate": 7.803563872911524e-07,
|
||
|
|
"loss": 0.3864,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27501577138900757,
|
||
|
|
"step": 4160,
|
||
|
|
"valid_targets_mean": 6771.4,
|
||
|
|
"valid_targets_min": 486
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.447368421052632,
|
||
|
|
"grad_norm": 0.16650565552954363,
|
||
|
|
"learning_rate": 7.591431314391617e-07,
|
||
|
|
"loss": 0.2711,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2717614471912384,
|
||
|
|
"step": 4165,
|
||
|
|
"valid_targets_mean": 6569.9,
|
||
|
|
"valid_targets_min": 353
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.455108359133127,
|
||
|
|
"grad_norm": 0.8442290579421936,
|
||
|
|
"learning_rate": 7.382166155627546e-07,
|
||
|
|
"loss": 0.4757,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4929467737674713,
|
||
|
|
"step": 4170,
|
||
|
|
"valid_targets_mean": 2007.4,
|
||
|
|
"valid_targets_min": 246
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.462848297213622,
|
||
|
|
"grad_norm": 0.6946927154416271,
|
||
|
|
"learning_rate": 7.17577151522717e-07,
|
||
|
|
"loss": 0.4806,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.49507656693458557,
|
||
|
|
"step": 4175,
|
||
|
|
"valid_targets_mean": 3351.7,
|
||
|
|
"valid_targets_min": 267
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.470588235294118,
|
||
|
|
"grad_norm": 0.5873172308180126,
|
||
|
|
"learning_rate": 6.972250469019859e-07,
|
||
|
|
"loss": 0.4739,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48357903957366943,
|
||
|
|
"step": 4180,
|
||
|
|
"valid_targets_mean": 3195.7,
|
||
|
|
"valid_targets_min": 296
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.478328173374613,
|
||
|
|
"grad_norm": 0.5190978804847558,
|
||
|
|
"learning_rate": 6.771606050010925e-07,
|
||
|
|
"loss": 0.4717,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.44882363080978394,
|
||
|
|
"step": 4185,
|
||
|
|
"valid_targets_mean": 3282.1,
|
||
|
|
"valid_targets_min": 296
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.486068111455109,
|
||
|
|
"grad_norm": 0.43125280427707846,
|
||
|
|
"learning_rate": 6.573841248336199e-07,
|
||
|
|
"loss": 0.4552,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.43425577878952026,
|
||
|
|
"step": 4190,
|
||
|
|
"valid_targets_mean": 3273.5,
|
||
|
|
"valid_targets_min": 426
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.493808049535604,
|
||
|
|
"grad_norm": 0.36360654088384836,
|
||
|
|
"learning_rate": 6.378959011217522e-07,
|
||
|
|
"loss": 0.4623,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4070616364479065,
|
||
|
|
"step": 4195,
|
||
|
|
"valid_targets_mean": 3904.3,
|
||
|
|
"valid_targets_min": 286
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.5015479876160995,
|
||
|
|
"grad_norm": 0.34047825272351917,
|
||
|
|
"learning_rate": 6.18696224291897e-07,
|
||
|
|
"loss": 0.4676,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.49775430560112,
|
||
|
|
"step": 4200,
|
||
|
|
"valid_targets_mean": 3417.9,
|
||
|
|
"valid_targets_min": 431
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.5092879256965945,
|
||
|
|
"grad_norm": 0.3604551580884761,
|
||
|
|
"learning_rate": 5.997853804703324e-07,
|
||
|
|
"loss": 0.459,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4818485975265503,
|
||
|
|
"step": 4205,
|
||
|
|
"valid_targets_mean": 2151.9,
|
||
|
|
"valid_targets_min": 306
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.5170278637770895,
|
||
|
|
"grad_norm": 0.2927855563910467,
|
||
|
|
"learning_rate": 5.811636514789598e-07,
|
||
|
|
"loss": 0.4531,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.430344820022583,
|
||
|
|
"step": 4210,
|
||
|
|
"valid_targets_mean": 2757.0,
|
||
|
|
"valid_targets_min": 366
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.524767801857585,
|
||
|
|
"grad_norm": 0.3175657011862788,
|
||
|
|
"learning_rate": 5.628313148311048e-07,
|
||
|
|
"loss": 0.46,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4702930152416229,
|
||
|
|
"step": 4215,
|
||
|
|
"valid_targets_mean": 2699.5,
|
||
|
|
"valid_targets_min": 297
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.53250773993808,
|
||
|
|
"grad_norm": 0.2855051496477021,
|
||
|
|
"learning_rate": 5.447886437273741e-07,
|
||
|
|
"loss": 0.4525,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4379507899284363,
|
||
|
|
"step": 4220,
|
||
|
|
"valid_targets_mean": 3419.5,
|
||
|
|
"valid_targets_min": 281
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.540247678018575,
|
||
|
|
"grad_norm": 0.24201990145313285,
|
||
|
|
"learning_rate": 5.270359070515874e-07,
|
||
|
|
"loss": 0.4518,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.449891060590744,
|
||
|
|
"step": 4225,
|
||
|
|
"valid_targets_mean": 3438.5,
|
||
|
|
"valid_targets_min": 301
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.547987616099071,
|
||
|
|
"grad_norm": 0.23070543563123194,
|
||
|
|
"learning_rate": 5.095733693667626e-07,
|
||
|
|
"loss": 0.4558,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4364064335823059,
|
||
|
|
"step": 4230,
|
||
|
|
"valid_targets_mean": 4330.6,
|
||
|
|
"valid_targets_min": 247
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.555727554179566,
|
||
|
|
"grad_norm": 0.275100284306335,
|
||
|
|
"learning_rate": 4.92401290911193e-07,
|
||
|
|
"loss": 0.4503,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4869125187397003,
|
||
|
|
"step": 4235,
|
||
|
|
"valid_targets_mean": 2712.3,
|
||
|
|
"valid_targets_min": 262
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.563467492260062,
|
||
|
|
"grad_norm": 0.6595214091891206,
|
||
|
|
"learning_rate": 4.7551992759454546e-07,
|
||
|
|
"loss": 0.4812,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6210309863090515,
|
||
|
|
"step": 4240,
|
||
|
|
"valid_targets_mean": 3619.0,
|
||
|
|
"valid_targets_min": 1099
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.571207430340557,
|
||
|
|
"grad_norm": 0.6176884521176255,
|
||
|
|
"learning_rate": 4.589295309940678e-07,
|
||
|
|
"loss": 0.5958,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5824363231658936,
|
||
|
|
"step": 4245,
|
||
|
|
"valid_targets_mean": 3834.9,
|
||
|
|
"valid_targets_min": 1026
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.578947368421053,
|
||
|
|
"grad_norm": 0.5964505132393924,
|
||
|
|
"learning_rate": 4.4263034835081874e-07,
|
||
|
|
"loss": 0.5936,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5758047103881836,
|
||
|
|
"step": 4250,
|
||
|
|
"valid_targets_mean": 3756.7,
|
||
|
|
"valid_targets_min": 1083
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.586687306501548,
|
||
|
|
"grad_norm": 0.5295557204003367,
|
||
|
|
"learning_rate": 4.2662262256600195e-07,
|
||
|
|
"loss": 0.5963,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5851274728775024,
|
||
|
|
"step": 4255,
|
||
|
|
"valid_targets_mean": 3851.6,
|
||
|
|
"valid_targets_min": 1063
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.594427244582043,
|
||
|
|
"grad_norm": 0.5367332940750664,
|
||
|
|
"learning_rate": 4.10906592197331e-07,
|
||
|
|
"loss": 0.5795,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6018767952919006,
|
||
|
|
"step": 4260,
|
||
|
|
"valid_targets_mean": 3610.4,
|
||
|
|
"valid_targets_min": 1092
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.602167182662539,
|
||
|
|
"grad_norm": 0.4162062790860631,
|
||
|
|
"learning_rate": 3.9548249145549e-07,
|
||
|
|
"loss": 0.5765,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5397290587425232,
|
||
|
|
"step": 4265,
|
||
|
|
"valid_targets_mean": 4324.1,
|
||
|
|
"valid_targets_min": 791
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.609907120743034,
|
||
|
|
"grad_norm": 0.4389169389600697,
|
||
|
|
"learning_rate": 3.803505502006255e-07,
|
||
|
|
"loss": 0.5882,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5711245536804199,
|
||
|
|
"step": 4270,
|
||
|
|
"valid_targets_mean": 3782.9,
|
||
|
|
"valid_targets_min": 1167
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.617647058823529,
|
||
|
|
"grad_norm": 0.35409010637457966,
|
||
|
|
"learning_rate": 3.6551099393893787e-07,
|
||
|
|
"loss": 0.5719,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.550596296787262,
|
||
|
|
"step": 4275,
|
||
|
|
"valid_targets_mean": 4237.8,
|
||
|
|
"valid_targets_min": 636
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.625386996904025,
|
||
|
|
"grad_norm": 0.3373229394737517,
|
||
|
|
"learning_rate": 3.509640438193107e-07,
|
||
|
|
"loss": 0.5738,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5639387369155884,
|
||
|
|
"step": 4280,
|
||
|
|
"valid_targets_mean": 4379.6,
|
||
|
|
"valid_targets_min": 417
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.63312693498452,
|
||
|
|
"grad_norm": 0.2822236648052455,
|
||
|
|
"learning_rate": 3.3670991663001587e-07,
|
||
|
|
"loss": 0.5605,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5054062008857727,
|
||
|
|
"step": 4285,
|
||
|
|
"valid_targets_mean": 4701.0,
|
||
|
|
"valid_targets_min": 826
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.640866873065016,
|
||
|
|
"grad_norm": 0.2941747168556465,
|
||
|
|
"learning_rate": 3.2274882479548244e-07,
|
||
|
|
"loss": 0.5817,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5482940077781677,
|
||
|
|
"step": 4290,
|
||
|
|
"valid_targets_mean": 4621.8,
|
||
|
|
"valid_targets_min": 983
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.648606811145511,
|
||
|
|
"grad_norm": 0.33756393689585873,
|
||
|
|
"learning_rate": 3.090809763731395e-07,
|
||
|
|
"loss": 0.5863,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6260626316070557,
|
||
|
|
"step": 4295,
|
||
|
|
"valid_targets_mean": 3181.1,
|
||
|
|
"valid_targets_min": 492
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.656346749226007,
|
||
|
|
"grad_norm": 0.28729815041431667,
|
||
|
|
"learning_rate": 2.9570657505030074e-07,
|
||
|
|
"loss": 0.586,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6321197152137756,
|
||
|
|
"step": 4300,
|
||
|
|
"valid_targets_mean": 4021.6,
|
||
|
|
"valid_targets_min": 665
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.664086687306502,
|
||
|
|
"grad_norm": 0.32651205049372145,
|
||
|
|
"learning_rate": 2.826258201411425e-07,
|
||
|
|
"loss": 0.5715,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6298854947090149,
|
||
|
|
"step": 4305,
|
||
|
|
"valid_targets_mean": 3290.7,
|
||
|
|
"valid_targets_min": 280
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.671826625386997,
|
||
|
|
"grad_norm": 0.27512678800028734,
|
||
|
|
"learning_rate": 2.698389065837237e-07,
|
||
|
|
"loss": 0.5818,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6131622791290283,
|
||
|
|
"step": 4310,
|
||
|
|
"valid_targets_mean": 4297.4,
|
||
|
|
"valid_targets_min": 1038
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.679566563467493,
|
||
|
|
"grad_norm": 0.470687359838086,
|
||
|
|
"learning_rate": 2.5734602493709514e-07,
|
||
|
|
"loss": 0.5562,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.49238407611846924,
|
||
|
|
"step": 4315,
|
||
|
|
"valid_targets_mean": 3676.8,
|
||
|
|
"valid_targets_min": 1023
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.687306501547988,
|
||
|
|
"grad_norm": 0.4812231288166542,
|
||
|
|
"learning_rate": 2.45147361378435e-07,
|
||
|
|
"loss": 0.4948,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5073689222335815,
|
||
|
|
"step": 4320,
|
||
|
|
"valid_targets_mean": 3877.6,
|
||
|
|
"valid_targets_min": 891
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.695046439628483,
|
||
|
|
"grad_norm": 0.47690769473366823,
|
||
|
|
"learning_rate": 2.3324309770029974e-07,
|
||
|
|
"loss": 0.4782,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.512337863445282,
|
||
|
|
"step": 4325,
|
||
|
|
"valid_targets_mean": 3730.8,
|
||
|
|
"valid_targets_min": 921
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.7027863777089784,
|
||
|
|
"grad_norm": 0.4338875758949859,
|
||
|
|
"learning_rate": 2.2163341130789994e-07,
|
||
|
|
"loss": 0.4795,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4618374705314636,
|
||
|
|
"step": 4330,
|
||
|
|
"valid_targets_mean": 3764.0,
|
||
|
|
"valid_targets_min": 771
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.7105263157894735,
|
||
|
|
"grad_norm": 0.41979508979394403,
|
||
|
|
"learning_rate": 2.1031847521645998e-07,
|
||
|
|
"loss": 0.4771,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.45828789472579956,
|
||
|
|
"step": 4335,
|
||
|
|
"valid_targets_mean": 3619.3,
|
||
|
|
"valid_targets_min": 981
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.718266253869969,
|
||
|
|
"grad_norm": 0.41436135114274586,
|
||
|
|
"learning_rate": 1.9929845804864456e-07,
|
||
|
|
"loss": 0.471,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.47076278924942017,
|
||
|
|
"step": 4340,
|
||
|
|
"valid_targets_mean": 3383.6,
|
||
|
|
"valid_targets_min": 747
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.726006191950464,
|
||
|
|
"grad_norm": 0.3999046461087337,
|
||
|
|
"learning_rate": 1.8857352403203632e-07,
|
||
|
|
"loss": 0.4724,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48855504393577576,
|
||
|
|
"step": 4345,
|
||
|
|
"valid_targets_mean": 3617.2,
|
||
|
|
"valid_targets_min": 877
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.73374613003096,
|
||
|
|
"grad_norm": 0.3599932197509348,
|
||
|
|
"learning_rate": 1.7814383299669336e-07,
|
||
|
|
"loss": 0.4701,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.45801666378974915,
|
||
|
|
"step": 4350,
|
||
|
|
"valid_targets_mean": 3628.0,
|
||
|
|
"valid_targets_min": 1027
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.741486068111455,
|
||
|
|
"grad_norm": 0.35977876185020075,
|
||
|
|
"learning_rate": 1.680095403727755e-07,
|
||
|
|
"loss": 0.471,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4639332592487335,
|
||
|
|
"step": 4355,
|
||
|
|
"valid_targets_mean": 3666.4,
|
||
|
|
"valid_targets_min": 1044
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.74922600619195,
|
||
|
|
"grad_norm": 0.3516186058561251,
|
||
|
|
"learning_rate": 1.5817079718821072e-07,
|
||
|
|
"loss": 0.4581,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4535253047943115,
|
||
|
|
"step": 4360,
|
||
|
|
"valid_targets_mean": 3281.1,
|
||
|
|
"valid_targets_min": 851
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.756965944272446,
|
||
|
|
"grad_norm": 0.3205896934626828,
|
||
|
|
"learning_rate": 1.4862775006645902e-07,
|
||
|
|
"loss": 0.4683,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4582814872264862,
|
||
|
|
"step": 4365,
|
||
|
|
"valid_targets_mean": 3577.5,
|
||
|
|
"valid_targets_min": 1157
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.764705882352941,
|
||
|
|
"grad_norm": 0.3333564235852343,
|
||
|
|
"learning_rate": 1.3938054122432098e-07,
|
||
|
|
"loss": 0.4724,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4348883032798767,
|
||
|
|
"step": 4370,
|
||
|
|
"valid_targets_mean": 3314.5,
|
||
|
|
"valid_targets_min": 706
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.772445820433436,
|
||
|
|
"grad_norm": 0.31370387781010467,
|
||
|
|
"learning_rate": 1.3042930846981939e-07,
|
||
|
|
"loss": 0.4731,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4609743356704712,
|
||
|
|
"step": 4375,
|
||
|
|
"valid_targets_mean": 3562.3,
|
||
|
|
"valid_targets_min": 1014
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.780185758513932,
|
||
|
|
"grad_norm": 0.2949175354653084,
|
||
|
|
"learning_rate": 1.2177418520014306e-07,
|
||
|
|
"loss": 0.4691,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46457013487815857,
|
||
|
|
"step": 4380,
|
||
|
|
"valid_targets_mean": 3536.9,
|
||
|
|
"valid_targets_min": 792
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.787925696594427,
|
||
|
|
"grad_norm": 0.3069800185605197,
|
||
|
|
"learning_rate": 1.1341530039966408e-07,
|
||
|
|
"loss": 0.4626,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4700559973716736,
|
||
|
|
"step": 4385,
|
||
|
|
"valid_targets_mean": 3570.4,
|
||
|
|
"valid_targets_min": 1059
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.795665634674923,
|
||
|
|
"grad_norm": 0.2937045282487432,
|
||
|
|
"learning_rate": 1.0535277863801485e-07,
|
||
|
|
"loss": 0.479,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46535611152648926,
|
||
|
|
"step": 4390,
|
||
|
|
"valid_targets_mean": 3647.4,
|
||
|
|
"valid_targets_min": 784
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.803405572755418,
|
||
|
|
"grad_norm": 0.42641556352754384,
|
||
|
|
"learning_rate": 9.758674006822732e-08,
|
||
|
|
"loss": 0.4631,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46037620306015015,
|
||
|
|
"step": 4395,
|
||
|
|
"valid_targets_mean": 3439.6,
|
||
|
|
"valid_targets_min": 1012
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.811145510835914,
|
||
|
|
"grad_norm": 0.2827854697129749,
|
||
|
|
"learning_rate": 9.011730042494338e-08,
|
||
|
|
"loss": 0.4605,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4201684892177582,
|
||
|
|
"step": 4400,
|
||
|
|
"valid_targets_mean": 3415.4,
|
||
|
|
"valid_targets_min": 555
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.818885448916409,
|
||
|
|
"grad_norm": 0.2811826372557516,
|
||
|
|
"learning_rate": 8.294457102269393e-08,
|
||
|
|
"loss": 0.5429,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4572042226791382,
|
||
|
|
"step": 4405,
|
||
|
|
"valid_targets_mean": 3413.9,
|
||
|
|
"valid_targets_min": 973
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.826625386996904,
|
||
|
|
"grad_norm": 0.26138696955244095,
|
||
|
|
"learning_rate": 7.606865875423586e-08,
|
||
|
|
"loss": 0.4677,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.45355531573295593,
|
||
|
|
"step": 4410,
|
||
|
|
"valid_targets_mean": 3740.7,
|
||
|
|
"valid_targets_min": 849
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.8343653250774,
|
||
|
|
"grad_norm": 0.2807067618494636,
|
||
|
|
"learning_rate": 6.948966608896435e-08,
|
||
|
|
"loss": 0.5551,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48855191469192505,
|
||
|
|
"step": 4415,
|
||
|
|
"valid_targets_mean": 3384.0,
|
||
|
|
"valid_targets_min": 870
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.842105263157895,
|
||
|
|
"grad_norm": 0.27975865722061855,
|
||
|
|
"learning_rate": 6.320769107137637e-08,
|
||
|
|
"loss": 0.4589,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4573826789855957,
|
||
|
|
"step": 4420,
|
||
|
|
"valid_targets_mean": 3341.3,
|
||
|
|
"valid_targets_min": 1113
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.84984520123839,
|
||
|
|
"grad_norm": 0.2565515600970192,
|
||
|
|
"learning_rate": 5.72228273196207e-08,
|
||
|
|
"loss": 0.5078,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46407103538513184,
|
||
|
|
"step": 4425,
|
||
|
|
"valid_targets_mean": 3684.2,
|
||
|
|
"valid_targets_min": 1026
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.857585139318886,
|
||
|
|
"grad_norm": 0.2914260522575157,
|
||
|
|
"learning_rate": 5.1535164024094644e-08,
|
||
|
|
"loss": 0.4602,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.44744065403938293,
|
||
|
|
"step": 4430,
|
||
|
|
"valid_targets_mean": 3157.4,
|
||
|
|
"valid_targets_min": 525
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.865325077399381,
|
||
|
|
"grad_norm": 0.30705847695403127,
|
||
|
|
"learning_rate": 4.614478594611837e-08,
|
||
|
|
"loss": 0.5095,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4935224950313568,
|
||
|
|
"step": 4435,
|
||
|
|
"valid_targets_mean": 3240.1,
|
||
|
|
"valid_targets_min": 984
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.8730650154798765,
|
||
|
|
"grad_norm": 0.2735037985206696,
|
||
|
|
"learning_rate": 4.105177341667377e-08,
|
||
|
|
"loss": 0.4585,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4584430754184723,
|
||
|
|
"step": 4440,
|
||
|
|
"valid_targets_mean": 3223.1,
|
||
|
|
"valid_targets_min": 328
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.8808049535603715,
|
||
|
|
"grad_norm": 0.24895930237501906,
|
||
|
|
"learning_rate": 3.625620233520088e-08,
|
||
|
|
"loss": 0.5524,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4749530851840973,
|
||
|
|
"step": 4445,
|
||
|
|
"valid_targets_mean": 3842.8,
|
||
|
|
"valid_targets_min": 811
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.888544891640867,
|
||
|
|
"grad_norm": 0.2722356326871571,
|
||
|
|
"learning_rate": 3.175814416847001e-08,
|
||
|
|
"loss": 0.461,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.46677082777023315,
|
||
|
|
"step": 4450,
|
||
|
|
"valid_targets_mean": 3401.7,
|
||
|
|
"valid_targets_min": 358
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.896284829721362,
|
||
|
|
"grad_norm": 0.28132382958982266,
|
||
|
|
"learning_rate": 2.755766594952025e-08,
|
||
|
|
"loss": 0.5374,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5245473980903625,
|
||
|
|
"step": 4455,
|
||
|
|
"valid_targets_mean": 3138.4,
|
||
|
|
"valid_targets_min": 426
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.904024767801857,
|
||
|
|
"grad_norm": 0.2589814964769566,
|
||
|
|
"learning_rate": 2.3654830276658157e-08,
|
||
|
|
"loss": 0.468,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.48557984828948975,
|
||
|
|
"step": 4460,
|
||
|
|
"valid_targets_mean": 3528.3,
|
||
|
|
"valid_targets_min": 285
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.911764705882353,
|
||
|
|
"grad_norm": 0.293465866905906,
|
||
|
|
"learning_rate": 2.0049695312520657e-08,
|
||
|
|
"loss": 0.4937,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5593820810317993,
|
||
|
|
"step": 4465,
|
||
|
|
"valid_targets_mean": 2719.7,
|
||
|
|
"valid_targets_min": 533
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.919504643962848,
|
||
|
|
"grad_norm": 0.26393938200307865,
|
||
|
|
"learning_rate": 1.674231478321353e-08,
|
||
|
|
"loss": 0.4586,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.47124210000038147,
|
||
|
|
"step": 4470,
|
||
|
|
"valid_targets_mean": 3570.4,
|
||
|
|
"valid_targets_min": 564
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.927244582043343,
|
||
|
|
"grad_norm": 0.5758935458896647,
|
||
|
|
"learning_rate": 1.3732737977509846e-08,
|
||
|
|
"loss": 0.4769,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4738374650478363,
|
||
|
|
"step": 4475,
|
||
|
|
"valid_targets_mean": 2341.9,
|
||
|
|
"valid_targets_min": 417
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.934984520123839,
|
||
|
|
"grad_norm": 0.7795780707603011,
|
||
|
|
"learning_rate": 1.1021009746112755e-08,
|
||
|
|
"loss": 0.2636,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26174822449684143,
|
||
|
|
"step": 4480,
|
||
|
|
"valid_targets_mean": 4556.8,
|
||
|
|
"valid_targets_min": 1546
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.942724458204334,
|
||
|
|
"grad_norm": 0.9176083327797326,
|
||
|
|
"learning_rate": 8.60717050099158e-09,
|
||
|
|
"loss": 0.2444,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26474621891975403,
|
||
|
|
"step": 4485,
|
||
|
|
"valid_targets_mean": 4626.2,
|
||
|
|
"valid_targets_min": 925
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.95046439628483,
|
||
|
|
"grad_norm": 0.7994375185743825,
|
||
|
|
"learning_rate": 6.491256214773422e-09,
|
||
|
|
"loss": 0.2339,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23102757334709167,
|
||
|
|
"step": 4490,
|
||
|
|
"valid_targets_mean": 4938.1,
|
||
|
|
"valid_targets_min": 1451
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.958204334365325,
|
||
|
|
"grad_norm": 0.840595967633915,
|
||
|
|
"learning_rate": 4.673298420214689e-09,
|
||
|
|
"loss": 0.2356,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24519655108451843,
|
||
|
|
"step": 4495,
|
||
|
|
"valid_targets_mean": 4161.5,
|
||
|
|
"valid_targets_min": 1249
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.965944272445821,
|
||
|
|
"grad_norm": 0.968736638594871,
|
||
|
|
"learning_rate": 3.1533242097214754e-09,
|
||
|
|
"loss": 0.2521,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.253293514251709,
|
||
|
|
"step": 4500,
|
||
|
|
"valid_targets_mean": 4068.4,
|
||
|
|
"valid_targets_min": 1239
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.973684210526316,
|
||
|
|
"grad_norm": 0.8945068846444052,
|
||
|
|
"learning_rate": 1.93135623495877e-09,
|
||
|
|
"loss": 0.2507,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24517560005187988,
|
||
|
|
"step": 4505,
|
||
|
|
"valid_targets_mean": 4988.2,
|
||
|
|
"valid_targets_min": 1593
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.981424148606811,
|
||
|
|
"grad_norm": 0.9196850305019586,
|
||
|
|
"learning_rate": 1.0074127064996219e-09,
|
||
|
|
"loss": 0.2359,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22261470556259155,
|
||
|
|
"step": 4510,
|
||
|
|
"valid_targets_mean": 2636.5,
|
||
|
|
"valid_targets_min": 274
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.989164086687307,
|
||
|
|
"grad_norm": 0.9534073928191378,
|
||
|
|
"learning_rate": 3.8150739356090926e-10,
|
||
|
|
"loss": 0.2403,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2474336475133896,
|
||
|
|
"step": 4515,
|
||
|
|
"valid_targets_mean": 2712.8,
|
||
|
|
"valid_targets_min": 982
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.996904024767802,
|
||
|
|
"grad_norm": 0.8589314752272489,
|
||
|
|
"learning_rate": 5.3649623799056917e-11,
|
||
|
|
"loss": 0.2302,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2319791316986084,
|
||
|
|
"step": 4520,
|
||
|
|
"valid_targets_mean": 3597.4,
|
||
|
|
"valid_targets_min": 1869
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 7.0,
|
||
|
|
"step": 4522,
|
||
|
|
"total_flos": 1.326951254196224e+16,
|
||
|
|
"train_loss": 0.0,
|
||
|
|
"train_runtime": 1.6239,
|
||
|
|
"train_samples_per_second": 356216.071,
|
||
|
|
"train_steps_per_second": 2784.656
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 5,
|
||
|
|
"max_steps": 4522,
|
||
|
|
"num_input_tokens_seen": 0,
|
||
|
|
"num_train_epochs": 7,
|
||
|
|
"save_steps": 300,
|
||
|
|
"stateful_callbacks": {
|
||
|
|
"TrainerControl": {
|
||
|
|
"args": {
|
||
|
|
"should_epoch_stop": false,
|
||
|
|
"should_evaluate": false,
|
||
|
|
"should_log": false,
|
||
|
|
"should_save": true,
|
||
|
|
"should_training_stop": true
|
||
|
|
},
|
||
|
|
"attributes": {}
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"total_flos": 1.326951254196224e+16,
|
||
|
|
"train_batch_size": 1,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|