7957 lines
221 KiB
JSON
7957 lines
221 KiB
JSON
|
{
|
||
|
|
"best_global_step": null,
|
||
|
|
"best_metric": null,
|
||
|
|
"best_model_checkpoint": null,
|
||
|
|
"epoch": 7.0,
|
||
|
|
"eval_steps": 500,
|
||
|
|
"global_step": 3598,
|
||
|
|
"is_hyper_param_search": false,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 0.009737098344693282,
|
||
|
|
"grad_norm": 6.659023843744882,
|
||
|
|
"learning_rate": 4.444444444444445e-07,
|
||
|
|
"loss": 0.708,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3341186046600342,
|
||
|
|
"step": 5,
|
||
|
|
"valid_targets_mean": 3131.6,
|
||
|
|
"valid_targets_min": 1131
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.019474196689386564,
|
||
|
|
"grad_norm": 7.096902024555233,
|
||
|
|
"learning_rate": 1.0000000000000002e-06,
|
||
|
|
"loss": 0.7467,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3632035255432129,
|
||
|
|
"step": 10,
|
||
|
|
"valid_targets_mean": 3640.4,
|
||
|
|
"valid_targets_min": 1212
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.029211295034079845,
|
||
|
|
"grad_norm": 8.390948057011439,
|
||
|
|
"learning_rate": 1.5555555555555558e-06,
|
||
|
|
"loss": 0.723,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3690454959869385,
|
||
|
|
"step": 15,
|
||
|
|
"valid_targets_mean": 2841.9,
|
||
|
|
"valid_targets_min": 1460
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03894839337877313,
|
||
|
|
"grad_norm": 4.941585314582268,
|
||
|
|
"learning_rate": 2.1111111111111114e-06,
|
||
|
|
"loss": 0.7416,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.34727782011032104,
|
||
|
|
"step": 20,
|
||
|
|
"valid_targets_mean": 2777.1,
|
||
|
|
"valid_targets_min": 816
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04868549172346641,
|
||
|
|
"grad_norm": 3.910822741404409,
|
||
|
|
"learning_rate": 2.666666666666667e-06,
|
||
|
|
"loss": 0.7126,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3296957314014435,
|
||
|
|
"step": 25,
|
||
|
|
"valid_targets_mean": 2570.0,
|
||
|
|
"valid_targets_min": 819
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05842259006815969,
|
||
|
|
"grad_norm": 3.6743426354854005,
|
||
|
|
"learning_rate": 3.2222222222222227e-06,
|
||
|
|
"loss": 0.7041,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30102282762527466,
|
||
|
|
"step": 30,
|
||
|
|
"valid_targets_mean": 1553.8,
|
||
|
|
"valid_targets_min": 919
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06815968841285297,
|
||
|
|
"grad_norm": 2.6901440323068724,
|
||
|
|
"learning_rate": 3.777777777777778e-06,
|
||
|
|
"loss": 0.6225,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3363445997238159,
|
||
|
|
"step": 35,
|
||
|
|
"valid_targets_mean": 2740.1,
|
||
|
|
"valid_targets_min": 956
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07789678675754626,
|
||
|
|
"grad_norm": 1.5292949966980411,
|
||
|
|
"learning_rate": 4.333333333333334e-06,
|
||
|
|
"loss": 0.6163,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2927810847759247,
|
||
|
|
"step": 40,
|
||
|
|
"valid_targets_mean": 3039.4,
|
||
|
|
"valid_targets_min": 1511
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08763388510223953,
|
||
|
|
"grad_norm": 1.3434308671523765,
|
||
|
|
"learning_rate": 4.888888888888889e-06,
|
||
|
|
"loss": 0.5893,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3337903916835785,
|
||
|
|
"step": 45,
|
||
|
|
"valid_targets_mean": 3930.4,
|
||
|
|
"valid_targets_min": 1675
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09737098344693282,
|
||
|
|
"grad_norm": 0.9373741509167212,
|
||
|
|
"learning_rate": 5.444444444444445e-06,
|
||
|
|
"loss": 0.5577,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18909615278244019,
|
||
|
|
"step": 50,
|
||
|
|
"valid_targets_mean": 3237.6,
|
||
|
|
"valid_targets_min": 1383
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.10710808179162609,
|
||
|
|
"grad_norm": 1.3352877614003678,
|
||
|
|
"learning_rate": 6e-06,
|
||
|
|
"loss": 0.5395,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2800602316856384,
|
||
|
|
"step": 55,
|
||
|
|
"valid_targets_mean": 1897.5,
|
||
|
|
"valid_targets_min": 989
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11684518013631938,
|
||
|
|
"grad_norm": 0.7395266220632808,
|
||
|
|
"learning_rate": 6.555555555555556e-06,
|
||
|
|
"loss": 0.492,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2359597086906433,
|
||
|
|
"step": 60,
|
||
|
|
"valid_targets_mean": 4924.5,
|
||
|
|
"valid_targets_min": 880
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12658227848101267,
|
||
|
|
"grad_norm": 0.9102371594210115,
|
||
|
|
"learning_rate": 7.111111111111112e-06,
|
||
|
|
"loss": 0.5166,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2459840178489685,
|
||
|
|
"step": 65,
|
||
|
|
"valid_targets_mean": 2506.4,
|
||
|
|
"valid_targets_min": 590
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13631937682570594,
|
||
|
|
"grad_norm": 0.8136377348280819,
|
||
|
|
"learning_rate": 7.666666666666667e-06,
|
||
|
|
"loss": 0.4982,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24337783455848694,
|
||
|
|
"step": 70,
|
||
|
|
"valid_targets_mean": 3078.9,
|
||
|
|
"valid_targets_min": 1886
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1460564751703992,
|
||
|
|
"grad_norm": 0.989281570628736,
|
||
|
|
"learning_rate": 8.222222222222222e-06,
|
||
|
|
"loss": 0.5096,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2819697856903076,
|
||
|
|
"step": 75,
|
||
|
|
"valid_targets_mean": 2441.6,
|
||
|
|
"valid_targets_min": 1560
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15579357351509251,
|
||
|
|
"grad_norm": 0.9036044445773288,
|
||
|
|
"learning_rate": 8.777777777777778e-06,
|
||
|
|
"loss": 0.4988,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25402408838272095,
|
||
|
|
"step": 80,
|
||
|
|
"valid_targets_mean": 3471.0,
|
||
|
|
"valid_targets_min": 1593
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1655306718597858,
|
||
|
|
"grad_norm": 0.7988252984615115,
|
||
|
|
"learning_rate": 9.333333333333334e-06,
|
||
|
|
"loss": 0.4763,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23960669338703156,
|
||
|
|
"step": 85,
|
||
|
|
"valid_targets_mean": 2990.8,
|
||
|
|
"valid_targets_min": 609
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17526777020447906,
|
||
|
|
"grad_norm": 0.7571243365246795,
|
||
|
|
"learning_rate": 9.88888888888889e-06,
|
||
|
|
"loss": 0.4758,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28586897253990173,
|
||
|
|
"step": 90,
|
||
|
|
"valid_targets_mean": 3950.6,
|
||
|
|
"valid_targets_min": 843
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18500486854917234,
|
||
|
|
"grad_norm": 0.7339048901803402,
|
||
|
|
"learning_rate": 1.0444444444444445e-05,
|
||
|
|
"loss": 0.4865,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2568185329437256,
|
||
|
|
"step": 95,
|
||
|
|
"valid_targets_mean": 3648.8,
|
||
|
|
"valid_targets_min": 2224
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19474196689386564,
|
||
|
|
"grad_norm": 0.7948184895802777,
|
||
|
|
"learning_rate": 1.1000000000000001e-05,
|
||
|
|
"loss": 0.4653,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25145047903060913,
|
||
|
|
"step": 100,
|
||
|
|
"valid_targets_mean": 3199.6,
|
||
|
|
"valid_targets_min": 1424
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2044790652385589,
|
||
|
|
"grad_norm": 0.7168429039256338,
|
||
|
|
"learning_rate": 1.1555555555555556e-05,
|
||
|
|
"loss": 0.4581,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26357921957969666,
|
||
|
|
"step": 105,
|
||
|
|
"valid_targets_mean": 3663.4,
|
||
|
|
"valid_targets_min": 1249
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21421616358325218,
|
||
|
|
"grad_norm": 0.7547313993147083,
|
||
|
|
"learning_rate": 1.211111111111111e-05,
|
||
|
|
"loss": 0.4502,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23485320806503296,
|
||
|
|
"step": 110,
|
||
|
|
"valid_targets_mean": 4071.5,
|
||
|
|
"valid_targets_min": 328
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22395326192794549,
|
||
|
|
"grad_norm": 0.5836593751970055,
|
||
|
|
"learning_rate": 1.2666666666666667e-05,
|
||
|
|
"loss": 0.4342,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17011061310768127,
|
||
|
|
"step": 115,
|
||
|
|
"valid_targets_mean": 3115.8,
|
||
|
|
"valid_targets_min": 1423
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23369036027263876,
|
||
|
|
"grad_norm": 0.7449951697430623,
|
||
|
|
"learning_rate": 1.3222222222222223e-05,
|
||
|
|
"loss": 0.451,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21737040579319,
|
||
|
|
"step": 120,
|
||
|
|
"valid_targets_mean": 2913.0,
|
||
|
|
"valid_targets_min": 662
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24342745861733203,
|
||
|
|
"grad_norm": 0.8242919386330488,
|
||
|
|
"learning_rate": 1.377777777777778e-05,
|
||
|
|
"loss": 0.4221,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2685893177986145,
|
||
|
|
"step": 125,
|
||
|
|
"valid_targets_mean": 3352.2,
|
||
|
|
"valid_targets_min": 1471
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25316455696202533,
|
||
|
|
"grad_norm": 0.6620888315425654,
|
||
|
|
"learning_rate": 1.4333333333333334e-05,
|
||
|
|
"loss": 0.4111,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2265009582042694,
|
||
|
|
"step": 130,
|
||
|
|
"valid_targets_mean": 3928.9,
|
||
|
|
"valid_targets_min": 1108
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2629016553067186,
|
||
|
|
"grad_norm": 0.7782866385397257,
|
||
|
|
"learning_rate": 1.488888888888889e-05,
|
||
|
|
"loss": 0.4129,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19729581475257874,
|
||
|
|
"step": 135,
|
||
|
|
"valid_targets_mean": 2165.6,
|
||
|
|
"valid_targets_min": 952
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2726387536514119,
|
||
|
|
"grad_norm": 0.9252799291147483,
|
||
|
|
"learning_rate": 1.5444444444444446e-05,
|
||
|
|
"loss": 0.4197,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22347350418567657,
|
||
|
|
"step": 140,
|
||
|
|
"valid_targets_mean": 1940.4,
|
||
|
|
"valid_targets_min": 949
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.28237585199610515,
|
||
|
|
"grad_norm": 0.7315249415292264,
|
||
|
|
"learning_rate": 1.6000000000000003e-05,
|
||
|
|
"loss": 0.3976,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2102898508310318,
|
||
|
|
"step": 145,
|
||
|
|
"valid_targets_mean": 3084.5,
|
||
|
|
"valid_targets_min": 1159
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2921129503407984,
|
||
|
|
"grad_norm": 1.02712502502217,
|
||
|
|
"learning_rate": 1.6555555555555556e-05,
|
||
|
|
"loss": 0.42,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2322973608970642,
|
||
|
|
"step": 150,
|
||
|
|
"valid_targets_mean": 1999.4,
|
||
|
|
"valid_targets_min": 507
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3018500486854917,
|
||
|
|
"grad_norm": 0.6139566424152655,
|
||
|
|
"learning_rate": 1.7111111111111112e-05,
|
||
|
|
"loss": 0.392,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1920212060213089,
|
||
|
|
"step": 155,
|
||
|
|
"valid_targets_mean": 4469.4,
|
||
|
|
"valid_targets_min": 1199
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.31158714703018503,
|
||
|
|
"grad_norm": 0.8626688830616602,
|
||
|
|
"learning_rate": 1.7666666666666668e-05,
|
||
|
|
"loss": 0.3874,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15205690264701843,
|
||
|
|
"step": 160,
|
||
|
|
"valid_targets_mean": 2269.1,
|
||
|
|
"valid_targets_min": 1221
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3213242453748783,
|
||
|
|
"grad_norm": 0.7369325509923896,
|
||
|
|
"learning_rate": 1.8222222222222224e-05,
|
||
|
|
"loss": 0.3808,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1633809208869934,
|
||
|
|
"step": 165,
|
||
|
|
"valid_targets_mean": 3191.4,
|
||
|
|
"valid_targets_min": 693
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3310613437195716,
|
||
|
|
"grad_norm": 0.9261185610744254,
|
||
|
|
"learning_rate": 1.877777777777778e-05,
|
||
|
|
"loss": 0.4041,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2576698362827301,
|
||
|
|
"step": 170,
|
||
|
|
"valid_targets_mean": 2714.6,
|
||
|
|
"valid_targets_min": 793
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.34079844206426485,
|
||
|
|
"grad_norm": 0.6904663234626722,
|
||
|
|
"learning_rate": 1.9333333333333333e-05,
|
||
|
|
"loss": 0.379,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19274544715881348,
|
||
|
|
"step": 175,
|
||
|
|
"valid_targets_mean": 3690.5,
|
||
|
|
"valid_targets_min": 1294
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3505355404089581,
|
||
|
|
"grad_norm": 0.8172102038691861,
|
||
|
|
"learning_rate": 1.988888888888889e-05,
|
||
|
|
"loss": 0.4065,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20005688071250916,
|
||
|
|
"step": 180,
|
||
|
|
"valid_targets_mean": 3082.0,
|
||
|
|
"valid_targets_min": 1208
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3602726387536514,
|
||
|
|
"grad_norm": 0.7380067514394181,
|
||
|
|
"learning_rate": 2.0444444444444446e-05,
|
||
|
|
"loss": 0.4028,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.171134814620018,
|
||
|
|
"step": 185,
|
||
|
|
"valid_targets_mean": 2461.5,
|
||
|
|
"valid_targets_min": 1228
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.37000973709834467,
|
||
|
|
"grad_norm": 0.7600421793317651,
|
||
|
|
"learning_rate": 2.1000000000000002e-05,
|
||
|
|
"loss": 0.3993,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23319822549819946,
|
||
|
|
"step": 190,
|
||
|
|
"valid_targets_mean": 3521.8,
|
||
|
|
"valid_targets_min": 993
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.379746835443038,
|
||
|
|
"grad_norm": 0.6656904832221395,
|
||
|
|
"learning_rate": 2.1555555555555555e-05,
|
||
|
|
"loss": 0.4064,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21867111325263977,
|
||
|
|
"step": 195,
|
||
|
|
"valid_targets_mean": 4448.8,
|
||
|
|
"valid_targets_min": 1679
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3894839337877313,
|
||
|
|
"grad_norm": 0.8878862701932538,
|
||
|
|
"learning_rate": 2.2111111111111115e-05,
|
||
|
|
"loss": 0.3681,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20832327008247375,
|
||
|
|
"step": 200,
|
||
|
|
"valid_targets_mean": 2900.9,
|
||
|
|
"valid_targets_min": 967
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39922103213242455,
|
||
|
|
"grad_norm": 0.5989319061194426,
|
||
|
|
"learning_rate": 2.2666666666666668e-05,
|
||
|
|
"loss": 0.3461,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14120624959468842,
|
||
|
|
"step": 205,
|
||
|
|
"valid_targets_mean": 3533.9,
|
||
|
|
"valid_targets_min": 808
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4089581304771178,
|
||
|
|
"grad_norm": 0.6157267635559596,
|
||
|
|
"learning_rate": 2.3222222222222227e-05,
|
||
|
|
"loss": 0.3701,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1939328908920288,
|
||
|
|
"step": 210,
|
||
|
|
"valid_targets_mean": 4087.9,
|
||
|
|
"valid_targets_min": 1256
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4186952288218111,
|
||
|
|
"grad_norm": 0.6990114686440374,
|
||
|
|
"learning_rate": 2.377777777777778e-05,
|
||
|
|
"loss": 0.3819,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18667009472846985,
|
||
|
|
"step": 215,
|
||
|
|
"valid_targets_mean": 2944.9,
|
||
|
|
"valid_targets_min": 709
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.42843232716650437,
|
||
|
|
"grad_norm": 0.8210463913714243,
|
||
|
|
"learning_rate": 2.4333333333333333e-05,
|
||
|
|
"loss": 0.403,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2342182993888855,
|
||
|
|
"step": 220,
|
||
|
|
"valid_targets_mean": 3265.4,
|
||
|
|
"valid_targets_min": 940
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.43816942551119764,
|
||
|
|
"grad_norm": 0.8024376262887541,
|
||
|
|
"learning_rate": 2.4888888888888893e-05,
|
||
|
|
"loss": 0.3901,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20045126974582672,
|
||
|
|
"step": 225,
|
||
|
|
"valid_targets_mean": 3143.4,
|
||
|
|
"valid_targets_min": 751
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.44790652385589097,
|
||
|
|
"grad_norm": 0.824782738636105,
|
||
|
|
"learning_rate": 2.5444444444444446e-05,
|
||
|
|
"loss": 0.4028,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15007048845291138,
|
||
|
|
"step": 230,
|
||
|
|
"valid_targets_mean": 2354.4,
|
||
|
|
"valid_targets_min": 1504
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45764362220058424,
|
||
|
|
"grad_norm": 1.0330846998899283,
|
||
|
|
"learning_rate": 2.6000000000000002e-05,
|
||
|
|
"loss": 0.3932,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18381167948246002,
|
||
|
|
"step": 235,
|
||
|
|
"valid_targets_mean": 1898.6,
|
||
|
|
"valid_targets_min": 781
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4673807205452775,
|
||
|
|
"grad_norm": 0.882733313530884,
|
||
|
|
"learning_rate": 2.6555555555555558e-05,
|
||
|
|
"loss": 0.396,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17795206606388092,
|
||
|
|
"step": 240,
|
||
|
|
"valid_targets_mean": 2748.1,
|
||
|
|
"valid_targets_min": 1364
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4771178188899708,
|
||
|
|
"grad_norm": 0.7790421961369038,
|
||
|
|
"learning_rate": 2.7111111111111114e-05,
|
||
|
|
"loss": 0.3861,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1974773406982422,
|
||
|
|
"step": 245,
|
||
|
|
"valid_targets_mean": 3469.5,
|
||
|
|
"valid_targets_min": 1085
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.48685491723466406,
|
||
|
|
"grad_norm": 0.9770266940397602,
|
||
|
|
"learning_rate": 2.7666666666666667e-05,
|
||
|
|
"loss": 0.3841,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2519018054008484,
|
||
|
|
"step": 250,
|
||
|
|
"valid_targets_mean": 2484.8,
|
||
|
|
"valid_targets_min": 1504
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.49659201557935734,
|
||
|
|
"grad_norm": 0.868298351235294,
|
||
|
|
"learning_rate": 2.8222222222222227e-05,
|
||
|
|
"loss": 0.3897,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2724197208881378,
|
||
|
|
"step": 255,
|
||
|
|
"valid_targets_mean": 3457.8,
|
||
|
|
"valid_targets_min": 1065
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5063291139240507,
|
||
|
|
"grad_norm": 0.7596864734968185,
|
||
|
|
"learning_rate": 2.877777777777778e-05,
|
||
|
|
"loss": 0.3422,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23726016283035278,
|
||
|
|
"step": 260,
|
||
|
|
"valid_targets_mean": 4169.0,
|
||
|
|
"valid_targets_min": 2438
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5160662122687439,
|
||
|
|
"grad_norm": 0.8718334359023358,
|
||
|
|
"learning_rate": 2.9333333333333333e-05,
|
||
|
|
"loss": 0.3726,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2140669971704483,
|
||
|
|
"step": 265,
|
||
|
|
"valid_targets_mean": 3537.9,
|
||
|
|
"valid_targets_min": 992
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5258033106134372,
|
||
|
|
"grad_norm": 0.6989164218061054,
|
||
|
|
"learning_rate": 2.9888888888888892e-05,
|
||
|
|
"loss": 0.3541,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20658528804779053,
|
||
|
|
"step": 270,
|
||
|
|
"valid_targets_mean": 4186.6,
|
||
|
|
"valid_targets_min": 1458
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5355404089581305,
|
||
|
|
"grad_norm": 0.874757587964136,
|
||
|
|
"learning_rate": 3.0444444444444445e-05,
|
||
|
|
"loss": 0.3688,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19685882329940796,
|
||
|
|
"step": 275,
|
||
|
|
"valid_targets_mean": 2467.0,
|
||
|
|
"valid_targets_min": 1414
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5452775073028238,
|
||
|
|
"grad_norm": 0.8053998992861866,
|
||
|
|
"learning_rate": 3.1e-05,
|
||
|
|
"loss": 0.3908,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15753547847270966,
|
||
|
|
"step": 280,
|
||
|
|
"valid_targets_mean": 2525.6,
|
||
|
|
"valid_targets_min": 1493
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.555014605647517,
|
||
|
|
"grad_norm": 0.805790874423672,
|
||
|
|
"learning_rate": 3.155555555555556e-05,
|
||
|
|
"loss": 0.3506,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16402557492256165,
|
||
|
|
"step": 285,
|
||
|
|
"valid_targets_mean": 2415.4,
|
||
|
|
"valid_targets_min": 1408
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5647517039922103,
|
||
|
|
"grad_norm": 0.9207062014606001,
|
||
|
|
"learning_rate": 3.2111111111111114e-05,
|
||
|
|
"loss": 0.3528,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13046719133853912,
|
||
|
|
"step": 290,
|
||
|
|
"valid_targets_mean": 1499.0,
|
||
|
|
"valid_targets_min": 871
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5744888023369036,
|
||
|
|
"grad_norm": 0.7580062459322378,
|
||
|
|
"learning_rate": 3.266666666666667e-05,
|
||
|
|
"loss": 0.3548,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1915191262960434,
|
||
|
|
"step": 295,
|
||
|
|
"valid_targets_mean": 2890.1,
|
||
|
|
"valid_targets_min": 562
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5842259006815969,
|
||
|
|
"grad_norm": 0.8333886934844644,
|
||
|
|
"learning_rate": 3.3222222222222226e-05,
|
||
|
|
"loss": 0.3643,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1909688115119934,
|
||
|
|
"step": 300,
|
||
|
|
"valid_targets_mean": 2758.6,
|
||
|
|
"valid_targets_min": 1306
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5939629990262901,
|
||
|
|
"grad_norm": 0.8714843579233928,
|
||
|
|
"learning_rate": 3.377777777777778e-05,
|
||
|
|
"loss": 0.3757,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1878107488155365,
|
||
|
|
"step": 305,
|
||
|
|
"valid_targets_mean": 3332.0,
|
||
|
|
"valid_targets_min": 1009
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6037000973709834,
|
||
|
|
"grad_norm": 0.8947388864456415,
|
||
|
|
"learning_rate": 3.433333333333333e-05,
|
||
|
|
"loss": 0.3929,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21496254205703735,
|
||
|
|
"step": 310,
|
||
|
|
"valid_targets_mean": 2662.9,
|
||
|
|
"valid_targets_min": 763
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6134371957156767,
|
||
|
|
"grad_norm": 0.8537461910420446,
|
||
|
|
"learning_rate": 3.4888888888888895e-05,
|
||
|
|
"loss": 0.3531,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20809201896190643,
|
||
|
|
"step": 315,
|
||
|
|
"valid_targets_mean": 2802.1,
|
||
|
|
"valid_targets_min": 936
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6231742940603701,
|
||
|
|
"grad_norm": 0.6719251437783023,
|
||
|
|
"learning_rate": 3.5444444444444445e-05,
|
||
|
|
"loss": 0.3503,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16037863492965698,
|
||
|
|
"step": 320,
|
||
|
|
"valid_targets_mean": 3631.4,
|
||
|
|
"valid_targets_min": 1046
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6329113924050633,
|
||
|
|
"grad_norm": 0.7848437937857925,
|
||
|
|
"learning_rate": 3.6e-05,
|
||
|
|
"loss": 0.3651,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18178021907806396,
|
||
|
|
"step": 325,
|
||
|
|
"valid_targets_mean": 3496.6,
|
||
|
|
"valid_targets_min": 589
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6426484907497566,
|
||
|
|
"grad_norm": 0.9302516156723694,
|
||
|
|
"learning_rate": 3.655555555555556e-05,
|
||
|
|
"loss": 0.3494,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17631950974464417,
|
||
|
|
"step": 330,
|
||
|
|
"valid_targets_mean": 2446.1,
|
||
|
|
"valid_targets_min": 1254
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6523855890944499,
|
||
|
|
"grad_norm": 0.9528588397406593,
|
||
|
|
"learning_rate": 3.7111111111111113e-05,
|
||
|
|
"loss": 0.3732,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16663740575313568,
|
||
|
|
"step": 335,
|
||
|
|
"valid_targets_mean": 2470.2,
|
||
|
|
"valid_targets_min": 761
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6621226874391432,
|
||
|
|
"grad_norm": 0.660648347651872,
|
||
|
|
"learning_rate": 3.766666666666667e-05,
|
||
|
|
"loss": 0.3491,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23378151655197144,
|
||
|
|
"step": 340,
|
||
|
|
"valid_targets_mean": 5074.6,
|
||
|
|
"valid_targets_min": 1774
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6718597857838364,
|
||
|
|
"grad_norm": 0.7317060175922973,
|
||
|
|
"learning_rate": 3.8222222222222226e-05,
|
||
|
|
"loss": 0.3508,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18020671606063843,
|
||
|
|
"step": 345,
|
||
|
|
"valid_targets_mean": 3391.1,
|
||
|
|
"valid_targets_min": 747
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6815968841285297,
|
||
|
|
"grad_norm": 0.7458049884451803,
|
||
|
|
"learning_rate": 3.877777777777778e-05,
|
||
|
|
"loss": 0.3635,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22611746191978455,
|
||
|
|
"step": 350,
|
||
|
|
"valid_targets_mean": 4497.8,
|
||
|
|
"valid_targets_min": 560
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.691333982473223,
|
||
|
|
"grad_norm": 0.7028549167777147,
|
||
|
|
"learning_rate": 3.933333333333333e-05,
|
||
|
|
"loss": 0.3387,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18253403902053833,
|
||
|
|
"step": 355,
|
||
|
|
"valid_targets_mean": 3703.0,
|
||
|
|
"valid_targets_min": 1396
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7010710808179162,
|
||
|
|
"grad_norm": 0.8105132831889875,
|
||
|
|
"learning_rate": 3.9888888888888895e-05,
|
||
|
|
"loss": 0.3615,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23544040322303772,
|
||
|
|
"step": 360,
|
||
|
|
"valid_targets_mean": 4297.5,
|
||
|
|
"valid_targets_min": 1307
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7108081791626095,
|
||
|
|
"grad_norm": 0.9310499702219079,
|
||
|
|
"learning_rate": 3.9999849385947476e-05,
|
||
|
|
"loss": 0.359,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2279876470565796,
|
||
|
|
"step": 365,
|
||
|
|
"valid_targets_mean": 2907.0,
|
||
|
|
"valid_targets_min": 754
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7205452775073028,
|
||
|
|
"grad_norm": 0.677986808450289,
|
||
|
|
"learning_rate": 3.999923752024691e-05,
|
||
|
|
"loss": 0.3444,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1297803521156311,
|
||
|
|
"step": 370,
|
||
|
|
"valid_targets_mean": 2690.5,
|
||
|
|
"valid_targets_min": 1284
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7302823758519961,
|
||
|
|
"grad_norm": 0.8247430862079875,
|
||
|
|
"learning_rate": 3.999815500390824e-05,
|
||
|
|
"loss": 0.3404,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13174360990524292,
|
||
|
|
"step": 375,
|
||
|
|
"valid_targets_mean": 1874.4,
|
||
|
|
"valid_targets_min": 1231
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7400194741966893,
|
||
|
|
"grad_norm": 0.7296077001600262,
|
||
|
|
"learning_rate": 3.9996601862406804e-05,
|
||
|
|
"loss": 0.3359,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1771552413702011,
|
||
|
|
"step": 380,
|
||
|
|
"valid_targets_mean": 3262.9,
|
||
|
|
"valid_targets_min": 1149
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7497565725413826,
|
||
|
|
"grad_norm": 0.8158122022730464,
|
||
|
|
"learning_rate": 3.999457813229333e-05,
|
||
|
|
"loss": 0.3512,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11544027179479599,
|
||
|
|
"step": 385,
|
||
|
|
"valid_targets_mean": 1960.2,
|
||
|
|
"valid_targets_min": 1035
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.759493670886076,
|
||
|
|
"grad_norm": 0.7322606730903284,
|
||
|
|
"learning_rate": 3.999208386119314e-05,
|
||
|
|
"loss": 0.336,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.143889918923378,
|
||
|
|
"step": 390,
|
||
|
|
"valid_targets_mean": 3042.1,
|
||
|
|
"valid_targets_min": 973
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7692307692307693,
|
||
|
|
"grad_norm": 0.7701997355670589,
|
||
|
|
"learning_rate": 3.998911910780497e-05,
|
||
|
|
"loss": 0.3252,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10486048460006714,
|
||
|
|
"step": 395,
|
||
|
|
"valid_targets_mean": 2053.1,
|
||
|
|
"valid_targets_min": 797
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7789678675754625,
|
||
|
|
"grad_norm": 0.7202330659166157,
|
||
|
|
"learning_rate": 3.9985683941899654e-05,
|
||
|
|
"loss": 0.322,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18157202005386353,
|
||
|
|
"step": 400,
|
||
|
|
"valid_targets_mean": 4142.8,
|
||
|
|
"valid_targets_min": 1497
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7887049659201558,
|
||
|
|
"grad_norm": 0.8024411032059268,
|
||
|
|
"learning_rate": 3.9981778444318414e-05,
|
||
|
|
"loss": 0.3468,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1389027237892151,
|
||
|
|
"step": 405,
|
||
|
|
"valid_targets_mean": 3080.4,
|
||
|
|
"valid_targets_min": 895
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7984420642648491,
|
||
|
|
"grad_norm": 0.7764211510246699,
|
||
|
|
"learning_rate": 3.9977402706971e-05,
|
||
|
|
"loss": 0.3257,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1333743929862976,
|
||
|
|
"step": 410,
|
||
|
|
"valid_targets_mean": 2548.8,
|
||
|
|
"valid_targets_min": 1103
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8081791626095424,
|
||
|
|
"grad_norm": 0.7770749008950054,
|
||
|
|
"learning_rate": 3.997255683283352e-05,
|
||
|
|
"loss": 0.3419,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16055616736412048,
|
||
|
|
"step": 415,
|
||
|
|
"valid_targets_mean": 3037.5,
|
||
|
|
"valid_targets_min": 774
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8179162609542356,
|
||
|
|
"grad_norm": 0.9875453975255327,
|
||
|
|
"learning_rate": 3.9967240935946004e-05,
|
||
|
|
"loss": 0.3449,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2241707742214203,
|
||
|
|
"step": 420,
|
||
|
|
"valid_targets_mean": 2675.0,
|
||
|
|
"valid_targets_min": 1080
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8276533592989289,
|
||
|
|
"grad_norm": 0.8008854116862701,
|
||
|
|
"learning_rate": 3.996145514140974e-05,
|
||
|
|
"loss": 0.3324,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21498018503189087,
|
||
|
|
"step": 425,
|
||
|
|
"valid_targets_mean": 3361.9,
|
||
|
|
"valid_targets_min": 557
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8373904576436222,
|
||
|
|
"grad_norm": 0.8712307832624219,
|
||
|
|
"learning_rate": 3.995519958538431e-05,
|
||
|
|
"loss": 0.3372,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18727466464042664,
|
||
|
|
"step": 430,
|
||
|
|
"valid_targets_mean": 2693.6,
|
||
|
|
"valid_targets_min": 1161
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8471275559883155,
|
||
|
|
"grad_norm": 0.7823837544455564,
|
||
|
|
"learning_rate": 3.99484744150844e-05,
|
||
|
|
"loss": 0.3503,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18069647252559662,
|
||
|
|
"step": 435,
|
||
|
|
"valid_targets_mean": 3102.6,
|
||
|
|
"valid_targets_min": 889
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8568646543330087,
|
||
|
|
"grad_norm": 0.6463271995153775,
|
||
|
|
"learning_rate": 3.9941279788776324e-05,
|
||
|
|
"loss": 0.3271,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18949922919273376,
|
||
|
|
"step": 440,
|
||
|
|
"valid_targets_mean": 5080.8,
|
||
|
|
"valid_targets_min": 1045
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.866601752677702,
|
||
|
|
"grad_norm": 0.9414900911940838,
|
||
|
|
"learning_rate": 3.9933615875774314e-05,
|
||
|
|
"loss": 0.3223,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.199024498462677,
|
||
|
|
"step": 445,
|
||
|
|
"valid_targets_mean": 2462.9,
|
||
|
|
"valid_targets_min": 764
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8763388510223953,
|
||
|
|
"grad_norm": 0.8197953554869378,
|
||
|
|
"learning_rate": 3.992548285643653e-05,
|
||
|
|
"loss": 0.318,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1533162146806717,
|
||
|
|
"step": 450,
|
||
|
|
"valid_targets_mean": 2308.9,
|
||
|
|
"valid_targets_min": 973
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8860759493670886,
|
||
|
|
"grad_norm": 0.8282523432739521,
|
||
|
|
"learning_rate": 3.99168809221608e-05,
|
||
|
|
"loss": 0.3219,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18272504210472107,
|
||
|
|
"step": 455,
|
||
|
|
"valid_targets_mean": 2762.1,
|
||
|
|
"valid_targets_min": 767
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8958130477117819,
|
||
|
|
"grad_norm": 0.652073664820321,
|
||
|
|
"learning_rate": 3.9907810275380133e-05,
|
||
|
|
"loss": 0.3377,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12357011437416077,
|
||
|
|
"step": 460,
|
||
|
|
"valid_targets_mean": 2613.4,
|
||
|
|
"valid_targets_min": 1247
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9055501460564752,
|
||
|
|
"grad_norm": 0.7821036719182141,
|
||
|
|
"learning_rate": 3.989827112955794e-05,
|
||
|
|
"loss": 0.3134,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19240909814834595,
|
||
|
|
"step": 465,
|
||
|
|
"valid_targets_mean": 3378.9,
|
||
|
|
"valid_targets_min": 1163
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9152872444011685,
|
||
|
|
"grad_norm": 0.7504672949569424,
|
||
|
|
"learning_rate": 3.988826370918306e-05,
|
||
|
|
"loss": 0.3374,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13087493181228638,
|
||
|
|
"step": 470,
|
||
|
|
"valid_targets_mean": 2295.5,
|
||
|
|
"valid_targets_min": 1017
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9250243427458618,
|
||
|
|
"grad_norm": 0.7954921854314398,
|
||
|
|
"learning_rate": 3.987778824976439e-05,
|
||
|
|
"loss": 0.3327,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22488024830818176,
|
||
|
|
"step": 475,
|
||
|
|
"valid_targets_mean": 3514.4,
|
||
|
|
"valid_targets_min": 1207
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.934761441090555,
|
||
|
|
"grad_norm": 0.7890591234039893,
|
||
|
|
"learning_rate": 3.9866844997825415e-05,
|
||
|
|
"loss": 0.3029,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2003348469734192,
|
||
|
|
"step": 480,
|
||
|
|
"valid_targets_mean": 3809.4,
|
||
|
|
"valid_targets_min": 1026
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9444985394352483,
|
||
|
|
"grad_norm": 0.7469531975242553,
|
||
|
|
"learning_rate": 3.985543421089839e-05,
|
||
|
|
"loss": 0.3513,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16702266037464142,
|
||
|
|
"step": 485,
|
||
|
|
"valid_targets_mean": 3206.9,
|
||
|
|
"valid_targets_min": 879
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9542356377799416,
|
||
|
|
"grad_norm": 0.8923246595958493,
|
||
|
|
"learning_rate": 3.984355615751824e-05,
|
||
|
|
"loss": 0.33,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1329648494720459,
|
||
|
|
"step": 490,
|
||
|
|
"valid_targets_mean": 2201.8,
|
||
|
|
"valid_targets_min": 1126
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9639727361246349,
|
||
|
|
"grad_norm": 0.8616912426190275,
|
||
|
|
"learning_rate": 3.983121111721631e-05,
|
||
|
|
"loss": 0.3458,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21117722988128662,
|
||
|
|
"step": 495,
|
||
|
|
"valid_targets_mean": 2512.8,
|
||
|
|
"valid_targets_min": 1080
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9737098344693281,
|
||
|
|
"grad_norm": 1.4342599215630012,
|
||
|
|
"learning_rate": 3.981839938051372e-05,
|
||
|
|
"loss": 0.3138,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16730377078056335,
|
||
|
|
"step": 500,
|
||
|
|
"valid_targets_mean": 2778.1,
|
||
|
|
"valid_targets_min": 1517
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9834469328140214,
|
||
|
|
"grad_norm": 0.6581607561348938,
|
||
|
|
"learning_rate": 3.980512124891458e-05,
|
||
|
|
"loss": 0.3228,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13223397731781006,
|
||
|
|
"step": 505,
|
||
|
|
"valid_targets_mean": 2865.0,
|
||
|
|
"valid_targets_min": 388
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9931840311587147,
|
||
|
|
"grad_norm": 0.7372438054026565,
|
||
|
|
"learning_rate": 3.979137703489883e-05,
|
||
|
|
"loss": 0.3199,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17514921724796295,
|
||
|
|
"step": 510,
|
||
|
|
"valid_targets_mean": 3584.5,
|
||
|
|
"valid_targets_min": 822
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0019474196689386,
|
||
|
|
"grad_norm": 0.7858899547464105,
|
||
|
|
"learning_rate": 3.9777167061915e-05,
|
||
|
|
"loss": 0.3175,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16728319227695465,
|
||
|
|
"step": 515,
|
||
|
|
"valid_targets_mean": 3270.8,
|
||
|
|
"valid_targets_min": 944
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.011684518013632,
|
||
|
|
"grad_norm": 0.9705259234628123,
|
||
|
|
"learning_rate": 3.976249166437243e-05,
|
||
|
|
"loss": 0.2926,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15755394101142883,
|
||
|
|
"step": 520,
|
||
|
|
"valid_targets_mean": 2423.5,
|
||
|
|
"valid_targets_min": 355
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0214216163583252,
|
||
|
|
"grad_norm": 0.6806683609325453,
|
||
|
|
"learning_rate": 3.974735118763359e-05,
|
||
|
|
"loss": 0.301,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15417352318763733,
|
||
|
|
"step": 525,
|
||
|
|
"valid_targets_mean": 3597.0,
|
||
|
|
"valid_targets_min": 1566
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0311587147030186,
|
||
|
|
"grad_norm": 0.8739140435404488,
|
||
|
|
"learning_rate": 3.97317459880058e-05,
|
||
|
|
"loss": 0.302,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11484783887863159,
|
||
|
|
"step": 530,
|
||
|
|
"valid_targets_mean": 1932.6,
|
||
|
|
"valid_targets_min": 793
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0408958130477117,
|
||
|
|
"grad_norm": 0.8457500312718982,
|
||
|
|
"learning_rate": 3.971567643273292e-05,
|
||
|
|
"loss": 0.2973,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16622629761695862,
|
||
|
|
"step": 535,
|
||
|
|
"valid_targets_mean": 2864.4,
|
||
|
|
"valid_targets_min": 815
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0506329113924051,
|
||
|
|
"grad_norm": 0.7340752528530782,
|
||
|
|
"learning_rate": 3.9699142899986685e-05,
|
||
|
|
"loss": 0.2905,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11975589394569397,
|
||
|
|
"step": 540,
|
||
|
|
"valid_targets_mean": 3145.0,
|
||
|
|
"valid_targets_min": 1661
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0603700097370983,
|
||
|
|
"grad_norm": 0.6371936175766221,
|
||
|
|
"learning_rate": 3.968214577885782e-05,
|
||
|
|
"loss": 0.2761,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1377580165863037,
|
||
|
|
"step": 545,
|
||
|
|
"valid_targets_mean": 3580.8,
|
||
|
|
"valid_targets_min": 684
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0701071080817917,
|
||
|
|
"grad_norm": 0.7077946139993203,
|
||
|
|
"learning_rate": 3.966468546934687e-05,
|
||
|
|
"loss": 0.3011,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12562993168830872,
|
||
|
|
"step": 550,
|
||
|
|
"valid_targets_mean": 3473.5,
|
||
|
|
"valid_targets_min": 1306
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0798442064264848,
|
||
|
|
"grad_norm": 0.7019427006203095,
|
||
|
|
"learning_rate": 3.964676238235479e-05,
|
||
|
|
"loss": 0.2989,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14723308384418488,
|
||
|
|
"step": 555,
|
||
|
|
"valid_targets_mean": 3323.2,
|
||
|
|
"valid_targets_min": 291
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0895813047711782,
|
||
|
|
"grad_norm": 0.6320463460477083,
|
||
|
|
"learning_rate": 3.9628376939673275e-05,
|
||
|
|
"loss": 0.2786,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06414983421564102,
|
||
|
|
"step": 560,
|
||
|
|
"valid_targets_mean": 1992.5,
|
||
|
|
"valid_targets_min": 625
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0993184031158714,
|
||
|
|
"grad_norm": 0.7782569308843855,
|
||
|
|
"learning_rate": 3.960952957397484e-05,
|
||
|
|
"loss": 0.2856,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14603619277477264,
|
||
|
|
"step": 565,
|
||
|
|
"valid_targets_mean": 2900.5,
|
||
|
|
"valid_targets_min": 975
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1090555014605648,
|
||
|
|
"grad_norm": 0.7063719109391404,
|
||
|
|
"learning_rate": 3.9590220728802625e-05,
|
||
|
|
"loss": 0.3081,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17670801281929016,
|
||
|
|
"step": 570,
|
||
|
|
"valid_targets_mean": 4031.0,
|
||
|
|
"valid_targets_min": 875
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1187925998052581,
|
||
|
|
"grad_norm": 0.7654501871062778,
|
||
|
|
"learning_rate": 3.9570450858559975e-05,
|
||
|
|
"loss": 0.2826,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13301081955432892,
|
||
|
|
"step": 575,
|
||
|
|
"valid_targets_mean": 2697.2,
|
||
|
|
"valid_targets_min": 991
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1285296981499513,
|
||
|
|
"grad_norm": 0.8053526181867805,
|
||
|
|
"learning_rate": 3.9550220428499725e-05,
|
||
|
|
"loss": 0.2952,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11767854541540146,
|
||
|
|
"step": 580,
|
||
|
|
"valid_targets_mean": 2463.9,
|
||
|
|
"valid_targets_min": 1309
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1382667964946447,
|
||
|
|
"grad_norm": 0.7055726427752043,
|
||
|
|
"learning_rate": 3.9529529914713266e-05,
|
||
|
|
"loss": 0.3074,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17200462520122528,
|
||
|
|
"step": 585,
|
||
|
|
"valid_targets_mean": 3473.1,
|
||
|
|
"valid_targets_min": 1285
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1480038948393378,
|
||
|
|
"grad_norm": 0.7796259491074381,
|
||
|
|
"learning_rate": 3.950837980411934e-05,
|
||
|
|
"loss": 0.2979,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1489405483007431,
|
||
|
|
"step": 590,
|
||
|
|
"valid_targets_mean": 2696.5,
|
||
|
|
"valid_targets_min": 1311
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1577409931840312,
|
||
|
|
"grad_norm": 0.6673286671425366,
|
||
|
|
"learning_rate": 3.948677059445257e-05,
|
||
|
|
"loss": 0.2733,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12524479627609253,
|
||
|
|
"step": 595,
|
||
|
|
"valid_targets_mean": 3187.4,
|
||
|
|
"valid_targets_min": 1479
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1674780915287244,
|
||
|
|
"grad_norm": 0.7241193327707911,
|
||
|
|
"learning_rate": 3.946470279425177e-05,
|
||
|
|
"loss": 0.2954,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13313235342502594,
|
||
|
|
"step": 600,
|
||
|
|
"valid_targets_mean": 3273.0,
|
||
|
|
"valid_targets_min": 1199
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1772151898734178,
|
||
|
|
"grad_norm": 0.7914833715074661,
|
||
|
|
"learning_rate": 3.944217692284794e-05,
|
||
|
|
"loss": 0.2866,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13487054407596588,
|
||
|
|
"step": 605,
|
||
|
|
"valid_targets_mean": 2979.8,
|
||
|
|
"valid_targets_min": 1731
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.186952288218111,
|
||
|
|
"grad_norm": 0.7822511631725727,
|
||
|
|
"learning_rate": 3.941919351035208e-05,
|
||
|
|
"loss": 0.2902,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13327060639858246,
|
||
|
|
"step": 610,
|
||
|
|
"valid_targets_mean": 2522.2,
|
||
|
|
"valid_targets_min": 672
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1966893865628043,
|
||
|
|
"grad_norm": 0.734117202012451,
|
||
|
|
"learning_rate": 3.9395753097642703e-05,
|
||
|
|
"loss": 0.307,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15425297617912292,
|
||
|
|
"step": 615,
|
||
|
|
"valid_targets_mean": 3389.9,
|
||
|
|
"valid_targets_min": 1414
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2064264849074975,
|
||
|
|
"grad_norm": 0.8355556162478627,
|
||
|
|
"learning_rate": 3.937185623635309e-05,
|
||
|
|
"loss": 0.3072,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15557613968849182,
|
||
|
|
"step": 620,
|
||
|
|
"valid_targets_mean": 2553.2,
|
||
|
|
"valid_targets_min": 749
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2161635832521909,
|
||
|
|
"grad_norm": 0.7047142278427146,
|
||
|
|
"learning_rate": 3.934750348885835e-05,
|
||
|
|
"loss": 0.2815,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19004464149475098,
|
||
|
|
"step": 625,
|
||
|
|
"valid_targets_mean": 4268.2,
|
||
|
|
"valid_targets_min": 1013
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.225900681596884,
|
||
|
|
"grad_norm": 0.7237181867450329,
|
||
|
|
"learning_rate": 3.932269542826211e-05,
|
||
|
|
"loss": 0.2897,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16163642704486847,
|
||
|
|
"step": 630,
|
||
|
|
"valid_targets_mean": 3763.0,
|
||
|
|
"valid_targets_min": 705
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2356377799415774,
|
||
|
|
"grad_norm": 0.6541968706944473,
|
||
|
|
"learning_rate": 3.9297432638383133e-05,
|
||
|
|
"loss": 0.2998,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15847386419773102,
|
||
|
|
"step": 635,
|
||
|
|
"valid_targets_mean": 3743.4,
|
||
|
|
"valid_targets_min": 1411
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2453748782862708,
|
||
|
|
"grad_norm": 0.7551346347644923,
|
||
|
|
"learning_rate": 3.9271715713741484e-05,
|
||
|
|
"loss": 0.2892,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17334289848804474,
|
||
|
|
"step": 640,
|
||
|
|
"valid_targets_mean": 3369.0,
|
||
|
|
"valid_targets_min": 1937
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.255111976630964,
|
||
|
|
"grad_norm": 0.7989231073679077,
|
||
|
|
"learning_rate": 3.9245545259544575e-05,
|
||
|
|
"loss": 0.2914,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13173508644104004,
|
||
|
|
"step": 645,
|
||
|
|
"valid_targets_mean": 2341.4,
|
||
|
|
"valid_targets_min": 654
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2648490749756571,
|
||
|
|
"grad_norm": 0.6713664995677392,
|
||
|
|
"learning_rate": 3.9218921891672955e-05,
|
||
|
|
"loss": 0.2937,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1702912300825119,
|
||
|
|
"step": 650,
|
||
|
|
"valid_targets_mean": 4714.0,
|
||
|
|
"valid_targets_min": 1272
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2745861733203505,
|
||
|
|
"grad_norm": 0.839712903367956,
|
||
|
|
"learning_rate": 3.919184623666575e-05,
|
||
|
|
"loss": 0.2806,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15325814485549927,
|
||
|
|
"step": 655,
|
||
|
|
"valid_targets_mean": 2320.9,
|
||
|
|
"valid_targets_min": 822
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.284323271665044,
|
||
|
|
"grad_norm": 0.6641240558454146,
|
||
|
|
"learning_rate": 3.916431893170596e-05,
|
||
|
|
"loss": 0.2996,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13758313655853271,
|
||
|
|
"step": 660,
|
||
|
|
"valid_targets_mean": 3853.6,
|
||
|
|
"valid_targets_min": 1057
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.294060370009737,
|
||
|
|
"grad_norm": 0.6883117123729979,
|
||
|
|
"learning_rate": 3.913634062460548e-05,
|
||
|
|
"loss": 0.2867,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13828516006469727,
|
||
|
|
"step": 665,
|
||
|
|
"valid_targets_mean": 3092.4,
|
||
|
|
"valid_targets_min": 1104
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3037974683544304,
|
||
|
|
"grad_norm": 0.644971806948973,
|
||
|
|
"learning_rate": 3.910791197378982e-05,
|
||
|
|
"loss": 0.3031,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09337528049945831,
|
||
|
|
"step": 670,
|
||
|
|
"valid_targets_mean": 2732.9,
|
||
|
|
"valid_targets_min": 664
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3135345666991236,
|
||
|
|
"grad_norm": 0.8550709015445958,
|
||
|
|
"learning_rate": 3.9079033648282595e-05,
|
||
|
|
"loss": 0.3224,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1829337328672409,
|
||
|
|
"step": 675,
|
||
|
|
"valid_targets_mean": 2290.4,
|
||
|
|
"valid_targets_min": 683
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.323271665043817,
|
||
|
|
"grad_norm": 0.8135393271299947,
|
||
|
|
"learning_rate": 3.9049706327689844e-05,
|
||
|
|
"loss": 0.2723,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11390693485736847,
|
||
|
|
"step": 680,
|
||
|
|
"valid_targets_mean": 2498.4,
|
||
|
|
"valid_targets_min": 1532
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3330087633885102,
|
||
|
|
"grad_norm": 0.6943074012950569,
|
||
|
|
"learning_rate": 3.9019930702183995e-05,
|
||
|
|
"loss": 0.2744,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1324174702167511,
|
||
|
|
"step": 685,
|
||
|
|
"valid_targets_mean": 3205.0,
|
||
|
|
"valid_targets_min": 622
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3427458617332035,
|
||
|
|
"grad_norm": 0.6843356225052816,
|
||
|
|
"learning_rate": 3.8989707472487605e-05,
|
||
|
|
"loss": 0.2835,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1364578902721405,
|
||
|
|
"step": 690,
|
||
|
|
"valid_targets_mean": 3158.2,
|
||
|
|
"valid_targets_min": 1224
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.352482960077897,
|
||
|
|
"grad_norm": 0.819539606411039,
|
||
|
|
"learning_rate": 3.8959037349856915e-05,
|
||
|
|
"loss": 0.2898,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10035207122564316,
|
||
|
|
"step": 695,
|
||
|
|
"valid_targets_mean": 2056.0,
|
||
|
|
"valid_targets_min": 580
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.36222005842259,
|
||
|
|
"grad_norm": 0.776933978950489,
|
||
|
|
"learning_rate": 3.8927921056065085e-05,
|
||
|
|
"loss": 0.2916,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10031980276107788,
|
||
|
|
"step": 700,
|
||
|
|
"valid_targets_mean": 1935.1,
|
||
|
|
"valid_targets_min": 1267
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3719571567672832,
|
||
|
|
"grad_norm": 0.7965490972208926,
|
||
|
|
"learning_rate": 3.8896359323385196e-05,
|
||
|
|
"loss": 0.296,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13414305448532104,
|
||
|
|
"step": 705,
|
||
|
|
"valid_targets_mean": 2537.6,
|
||
|
|
"valid_targets_min": 1122
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3816942551119766,
|
||
|
|
"grad_norm": 0.8531863308487779,
|
||
|
|
"learning_rate": 3.886435289457306e-05,
|
||
|
|
"loss": 0.2983,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1738170087337494,
|
||
|
|
"step": 710,
|
||
|
|
"valid_targets_mean": 2457.0,
|
||
|
|
"valid_targets_min": 779
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.39143135345667,
|
||
|
|
"grad_norm": 0.731101519911782,
|
||
|
|
"learning_rate": 3.883190252284969e-05,
|
||
|
|
"loss": 0.2662,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11143383383750916,
|
||
|
|
"step": 715,
|
||
|
|
"valid_targets_mean": 2316.5,
|
||
|
|
"valid_targets_min": 317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4011684518013632,
|
||
|
|
"grad_norm": 0.6662025242467065,
|
||
|
|
"learning_rate": 3.879900897188361e-05,
|
||
|
|
"loss": 0.2824,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1794198453426361,
|
||
|
|
"step": 720,
|
||
|
|
"valid_targets_mean": 4965.1,
|
||
|
|
"valid_targets_min": 1986
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4109055501460563,
|
||
|
|
"grad_norm": 0.8360114160936927,
|
||
|
|
"learning_rate": 3.876567301577287e-05,
|
||
|
|
"loss": 0.267,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15071365237236023,
|
||
|
|
"step": 725,
|
||
|
|
"valid_targets_mean": 2652.9,
|
||
|
|
"valid_targets_min": 973
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4206426484907497,
|
||
|
|
"grad_norm": 0.6719303787687317,
|
||
|
|
"learning_rate": 3.873189543902683e-05,
|
||
|
|
"loss": 0.2857,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17039050161838531,
|
||
|
|
"step": 730,
|
||
|
|
"valid_targets_mean": 3710.0,
|
||
|
|
"valid_targets_min": 1899
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4303797468354431,
|
||
|
|
"grad_norm": 0.8200941586506575,
|
||
|
|
"learning_rate": 3.86976770365477e-05,
|
||
|
|
"loss": 0.2859,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17211097478866577,
|
||
|
|
"step": 735,
|
||
|
|
"valid_targets_mean": 3249.9,
|
||
|
|
"valid_targets_min": 1512
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4401168451801363,
|
||
|
|
"grad_norm": 0.7647285702362657,
|
||
|
|
"learning_rate": 3.8663018613611824e-05,
|
||
|
|
"loss": 0.2832,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1194126158952713,
|
||
|
|
"step": 740,
|
||
|
|
"valid_targets_mean": 2530.5,
|
||
|
|
"valid_targets_min": 1285
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4498539435248297,
|
||
|
|
"grad_norm": 0.7598532952082641,
|
||
|
|
"learning_rate": 3.862792098585074e-05,
|
||
|
|
"loss": 0.2751,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1360061913728714,
|
||
|
|
"step": 745,
|
||
|
|
"valid_targets_mean": 3239.8,
|
||
|
|
"valid_targets_min": 909
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4595910418695228,
|
||
|
|
"grad_norm": 0.8903098563066778,
|
||
|
|
"learning_rate": 3.8592384979231984e-05,
|
||
|
|
"loss": 0.2623,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11483582854270935,
|
||
|
|
"step": 750,
|
||
|
|
"valid_targets_mean": 2928.1,
|
||
|
|
"valid_targets_min": 1210
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4693281402142162,
|
||
|
|
"grad_norm": 0.8495650293717296,
|
||
|
|
"learning_rate": 3.8556411430039656e-05,
|
||
|
|
"loss": 0.2949,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19118264317512512,
|
||
|
|
"step": 755,
|
||
|
|
"valid_targets_mean": 2806.5,
|
||
|
|
"valid_targets_min": 936
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4790652385589094,
|
||
|
|
"grad_norm": 0.9628487252912299,
|
||
|
|
"learning_rate": 3.852000118485471e-05,
|
||
|
|
"loss": 0.3034,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1747865378856659,
|
||
|
|
"step": 760,
|
||
|
|
"valid_targets_mean": 2188.1,
|
||
|
|
"valid_targets_min": 1392
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4888023369036028,
|
||
|
|
"grad_norm": 0.7004325923802018,
|
||
|
|
"learning_rate": 3.8483155100535074e-05,
|
||
|
|
"loss": 0.2768,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12580302357673645,
|
||
|
|
"step": 765,
|
||
|
|
"valid_targets_mean": 3205.1,
|
||
|
|
"valid_targets_min": 1514
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4985394352482961,
|
||
|
|
"grad_norm": 0.74062192186481,
|
||
|
|
"learning_rate": 3.8445874044195475e-05,
|
||
|
|
"loss": 0.2734,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11894209682941437,
|
||
|
|
"step": 770,
|
||
|
|
"valid_targets_mean": 2435.5,
|
||
|
|
"valid_targets_min": 1201
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5082765335929893,
|
||
|
|
"grad_norm": 0.8816120657903092,
|
||
|
|
"learning_rate": 3.840815889318699e-05,
|
||
|
|
"loss": 0.2707,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16805878281593323,
|
||
|
|
"step": 775,
|
||
|
|
"valid_targets_mean": 2607.0,
|
||
|
|
"valid_targets_min": 1426
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5180136319376825,
|
||
|
|
"grad_norm": 0.6447763694072602,
|
||
|
|
"learning_rate": 3.837001053507645e-05,
|
||
|
|
"loss": 0.2731,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11964346468448639,
|
||
|
|
"step": 780,
|
||
|
|
"valid_targets_mean": 3388.4,
|
||
|
|
"valid_targets_min": 568
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5277507302823758,
|
||
|
|
"grad_norm": 0.7624960908344781,
|
||
|
|
"learning_rate": 3.833142986762556e-05,
|
||
|
|
"loss": 0.2886,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1300625503063202,
|
||
|
|
"step": 785,
|
||
|
|
"valid_targets_mean": 2600.6,
|
||
|
|
"valid_targets_min": 1059
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5374878286270692,
|
||
|
|
"grad_norm": 0.8002057073411786,
|
||
|
|
"learning_rate": 3.82924177987697e-05,
|
||
|
|
"loss": 0.2832,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.121633380651474,
|
||
|
|
"step": 790,
|
||
|
|
"valid_targets_mean": 1773.9,
|
||
|
|
"valid_targets_min": 603
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5472249269717624,
|
||
|
|
"grad_norm": 0.6878158152271249,
|
||
|
|
"learning_rate": 3.8252975246596634e-05,
|
||
|
|
"loss": 0.284,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11953762918710709,
|
||
|
|
"step": 795,
|
||
|
|
"valid_targets_mean": 2672.4,
|
||
|
|
"valid_targets_min": 304
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5569620253164556,
|
||
|
|
"grad_norm": 0.6397890876132153,
|
||
|
|
"learning_rate": 3.821310313932486e-05,
|
||
|
|
"loss": 0.3103,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1216995120048523,
|
||
|
|
"step": 800,
|
||
|
|
"valid_targets_mean": 3221.5,
|
||
|
|
"valid_targets_min": 1079
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.566699123661149,
|
||
|
|
"grad_norm": 0.7994201671438256,
|
||
|
|
"learning_rate": 3.8172802415281796e-05,
|
||
|
|
"loss": 0.2919,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1517755687236786,
|
||
|
|
"step": 805,
|
||
|
|
"valid_targets_mean": 2454.0,
|
||
|
|
"valid_targets_min": 1178
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5764362220058423,
|
||
|
|
"grad_norm": 0.8475574220657818,
|
||
|
|
"learning_rate": 3.813207402288167e-05,
|
||
|
|
"loss": 0.2841,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12581583857536316,
|
||
|
|
"step": 810,
|
||
|
|
"valid_targets_mean": 2915.9,
|
||
|
|
"valid_targets_min": 1215
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5861733203505355,
|
||
|
|
"grad_norm": 0.78684352089434,
|
||
|
|
"learning_rate": 3.8090918920603214e-05,
|
||
|
|
"loss": 0.2652,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10042424499988556,
|
||
|
|
"step": 815,
|
||
|
|
"valid_targets_mean": 2571.6,
|
||
|
|
"valid_targets_min": 989
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5959104186952289,
|
||
|
|
"grad_norm": 0.7306518474461766,
|
||
|
|
"learning_rate": 3.804933807696712e-05,
|
||
|
|
"loss": 0.2753,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11414317041635513,
|
||
|
|
"step": 820,
|
||
|
|
"valid_targets_mean": 2278.1,
|
||
|
|
"valid_targets_min": 961
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6056475170399223,
|
||
|
|
"grad_norm": 0.6843854426395394,
|
||
|
|
"learning_rate": 3.80073324705132e-05,
|
||
|
|
"loss": 0.295,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12278768420219421,
|
||
|
|
"step": 825,
|
||
|
|
"valid_targets_mean": 3124.6,
|
||
|
|
"valid_targets_min": 1350
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6153846153846154,
|
||
|
|
"grad_norm": 0.7572889779601448,
|
||
|
|
"learning_rate": 3.7964903089777443e-05,
|
||
|
|
"loss": 0.2655,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11568583548069,
|
||
|
|
"step": 830,
|
||
|
|
"valid_targets_mean": 2210.9,
|
||
|
|
"valid_targets_min": 560
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6251217137293086,
|
||
|
|
"grad_norm": 0.6441562182965196,
|
||
|
|
"learning_rate": 3.792205093326865e-05,
|
||
|
|
"loss": 0.276,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11288474500179291,
|
||
|
|
"step": 835,
|
||
|
|
"valid_targets_mean": 2595.0,
|
||
|
|
"valid_targets_min": 392
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.634858812074002,
|
||
|
|
"grad_norm": 0.6518505921760807,
|
||
|
|
"learning_rate": 3.7878777009445025e-05,
|
||
|
|
"loss": 0.284,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11960506439208984,
|
||
|
|
"step": 840,
|
||
|
|
"valid_targets_mean": 3584.9,
|
||
|
|
"valid_targets_min": 1431
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6445959104186954,
|
||
|
|
"grad_norm": 0.8324832568062053,
|
||
|
|
"learning_rate": 3.7835082336690376e-05,
|
||
|
|
"loss": 0.2742,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1086902916431427,
|
||
|
|
"step": 845,
|
||
|
|
"valid_targets_mean": 1682.0,
|
||
|
|
"valid_targets_min": 715
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6543330087633885,
|
||
|
|
"grad_norm": 0.899466301772704,
|
||
|
|
"learning_rate": 3.7790967943290186e-05,
|
||
|
|
"loss": 0.3167,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19043229520320892,
|
||
|
|
"step": 850,
|
||
|
|
"valid_targets_mean": 2796.4,
|
||
|
|
"valid_targets_min": 300
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6640701071080817,
|
||
|
|
"grad_norm": 0.6833506865776794,
|
||
|
|
"learning_rate": 3.7746434867407405e-05,
|
||
|
|
"loss": 0.2733,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1205308735370636,
|
||
|
|
"step": 855,
|
||
|
|
"valid_targets_mean": 3405.9,
|
||
|
|
"valid_targets_min": 1141
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.673807205452775,
|
||
|
|
"grad_norm": 0.6735391267174385,
|
||
|
|
"learning_rate": 3.7701484157058016e-05,
|
||
|
|
"loss": 0.2809,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14146654307842255,
|
||
|
|
"step": 860,
|
||
|
|
"valid_targets_mean": 3473.8,
|
||
|
|
"valid_targets_min": 1788
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6835443037974684,
|
||
|
|
"grad_norm": 0.7478096289483627,
|
||
|
|
"learning_rate": 3.7656116870086376e-05,
|
||
|
|
"loss": 0.2884,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13985690474510193,
|
||
|
|
"step": 865,
|
||
|
|
"valid_targets_mean": 2415.1,
|
||
|
|
"valid_targets_min": 1055
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6932814021421616,
|
||
|
|
"grad_norm": 0.6917897462416348,
|
||
|
|
"learning_rate": 3.761033407414032e-05,
|
||
|
|
"loss": 0.2819,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13799864053726196,
|
||
|
|
"step": 870,
|
||
|
|
"valid_targets_mean": 3341.5,
|
||
|
|
"valid_targets_min": 1898
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7030185004868548,
|
||
|
|
"grad_norm": 0.7825581458353967,
|
||
|
|
"learning_rate": 3.756413684664602e-05,
|
||
|
|
"loss": 0.2881,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17821389436721802,
|
||
|
|
"step": 875,
|
||
|
|
"valid_targets_mean": 3530.2,
|
||
|
|
"valid_targets_min": 1414
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7127555988315482,
|
||
|
|
"grad_norm": 0.6296436510408459,
|
||
|
|
"learning_rate": 3.751752627478266e-05,
|
||
|
|
"loss": 0.266,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09756700694561005,
|
||
|
|
"step": 880,
|
||
|
|
"valid_targets_mean": 2915.2,
|
||
|
|
"valid_targets_min": 1017
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7224926971762415,
|
||
|
|
"grad_norm": 0.6844544440869792,
|
||
|
|
"learning_rate": 3.747050345545683e-05,
|
||
|
|
"loss": 0.273,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11896862089633942,
|
||
|
|
"step": 885,
|
||
|
|
"valid_targets_mean": 2729.0,
|
||
|
|
"valid_targets_min": 641
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7322297955209347,
|
||
|
|
"grad_norm": 0.6474680504011459,
|
||
|
|
"learning_rate": 3.7423069495276705e-05,
|
||
|
|
"loss": 0.283,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15866486728191376,
|
||
|
|
"step": 890,
|
||
|
|
"valid_targets_mean": 4134.0,
|
||
|
|
"valid_targets_min": 1708
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.741966893865628,
|
||
|
|
"grad_norm": 0.7020327824367882,
|
||
|
|
"learning_rate": 3.737522551052603e-05,
|
||
|
|
"loss": 0.2665,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10592299699783325,
|
||
|
|
"step": 895,
|
||
|
|
"valid_targets_mean": 2089.9,
|
||
|
|
"valid_targets_min": 800
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7517039922103215,
|
||
|
|
"grad_norm": 0.6709837312667878,
|
||
|
|
"learning_rate": 3.732697262713783e-05,
|
||
|
|
"loss": 0.2681,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17397888004779816,
|
||
|
|
"step": 900,
|
||
|
|
"valid_targets_mean": 4001.9,
|
||
|
|
"valid_targets_min": 968
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7614410905550146,
|
||
|
|
"grad_norm": 0.7029093037875962,
|
||
|
|
"learning_rate": 3.727831198066791e-05,
|
||
|
|
"loss": 0.2891,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18118925392627716,
|
||
|
|
"step": 905,
|
||
|
|
"valid_targets_mean": 3871.5,
|
||
|
|
"valid_targets_min": 993
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7711781888997078,
|
||
|
|
"grad_norm": 0.6670349375759711,
|
||
|
|
"learning_rate": 3.722924471626815e-05,
|
||
|
|
"loss": 0.2733,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11348405480384827,
|
||
|
|
"step": 910,
|
||
|
|
"valid_targets_mean": 2493.2,
|
||
|
|
"valid_targets_min": 1225
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7809152872444012,
|
||
|
|
"grad_norm": 0.857871388664443,
|
||
|
|
"learning_rate": 3.7179771988659526e-05,
|
||
|
|
"loss": 0.2766,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1357949823141098,
|
||
|
|
"step": 915,
|
||
|
|
"valid_targets_mean": 1648.1,
|
||
|
|
"valid_targets_min": 411
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7906523855890946,
|
||
|
|
"grad_norm": 0.8701206332552374,
|
||
|
|
"learning_rate": 3.712989496210497e-05,
|
||
|
|
"loss": 0.2827,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11637701839208603,
|
||
|
|
"step": 920,
|
||
|
|
"valid_targets_mean": 1869.1,
|
||
|
|
"valid_targets_min": 1118
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8003894839337877,
|
||
|
|
"grad_norm": 0.7733174755020938,
|
||
|
|
"learning_rate": 3.707961481038196e-05,
|
||
|
|
"loss": 0.2852,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16073478758335114,
|
||
|
|
"step": 925,
|
||
|
|
"valid_targets_mean": 2269.2,
|
||
|
|
"valid_targets_min": 614
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.810126582278481,
|
||
|
|
"grad_norm": 0.9844262069470177,
|
||
|
|
"learning_rate": 3.702893271675487e-05,
|
||
|
|
"loss": 0.2849,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13788703083992004,
|
||
|
|
"step": 930,
|
||
|
|
"valid_targets_mean": 1586.2,
|
||
|
|
"valid_targets_min": 750
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8198636806231743,
|
||
|
|
"grad_norm": 0.8214724111954528,
|
||
|
|
"learning_rate": 3.697784987394716e-05,
|
||
|
|
"loss": 0.2681,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18131202459335327,
|
||
|
|
"step": 935,
|
||
|
|
"valid_targets_mean": 2752.5,
|
||
|
|
"valid_targets_min": 1511
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8296007789678677,
|
||
|
|
"grad_norm": 0.6915571590018944,
|
||
|
|
"learning_rate": 3.692636748411329e-05,
|
||
|
|
"loss": 0.2617,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1040545403957367,
|
||
|
|
"step": 940,
|
||
|
|
"valid_targets_mean": 2429.0,
|
||
|
|
"valid_targets_min": 949
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8393378773125608,
|
||
|
|
"grad_norm": 0.6643771603137818,
|
||
|
|
"learning_rate": 3.687448675881043e-05,
|
||
|
|
"loss": 0.2695,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10466742515563965,
|
||
|
|
"step": 945,
|
||
|
|
"valid_targets_mean": 2981.1,
|
||
|
|
"valid_targets_min": 1571
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.849074975657254,
|
||
|
|
"grad_norm": 0.7594173087993755,
|
||
|
|
"learning_rate": 3.682220891896995e-05,
|
||
|
|
"loss": 0.276,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1362399160861969,
|
||
|
|
"step": 950,
|
||
|
|
"valid_targets_mean": 3018.8,
|
||
|
|
"valid_targets_min": 1849
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8588120740019474,
|
||
|
|
"grad_norm": 0.8226705248019373,
|
||
|
|
"learning_rate": 3.676953519486868e-05,
|
||
|
|
"loss": 0.2941,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1400984525680542,
|
||
|
|
"step": 955,
|
||
|
|
"valid_targets_mean": 1782.9,
|
||
|
|
"valid_targets_min": 819
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8685491723466408,
|
||
|
|
"grad_norm": 0.716706641470173,
|
||
|
|
"learning_rate": 3.671646682609999e-05,
|
||
|
|
"loss": 0.2916,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16295509040355682,
|
||
|
|
"step": 960,
|
||
|
|
"valid_targets_mean": 3783.8,
|
||
|
|
"valid_targets_min": 1576
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.878286270691334,
|
||
|
|
"grad_norm": 0.6680265782602849,
|
||
|
|
"learning_rate": 3.666300506154455e-05,
|
||
|
|
"loss": 0.2929,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16173237562179565,
|
||
|
|
"step": 965,
|
||
|
|
"valid_targets_mean": 4008.1,
|
||
|
|
"valid_targets_min": 1997
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8880233690360273,
|
||
|
|
"grad_norm": 0.7900756008320604,
|
||
|
|
"learning_rate": 3.660915115934103e-05,
|
||
|
|
"loss": 0.2989,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14800488948822021,
|
||
|
|
"step": 970,
|
||
|
|
"valid_targets_mean": 2644.2,
|
||
|
|
"valid_targets_min": 980
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8977604673807207,
|
||
|
|
"grad_norm": 0.6749709974662765,
|
||
|
|
"learning_rate": 3.6554906386856394e-05,
|
||
|
|
"loss": 0.2749,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12733563780784607,
|
||
|
|
"step": 975,
|
||
|
|
"valid_targets_mean": 3121.5,
|
||
|
|
"valid_targets_min": 1292
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9074975657254138,
|
||
|
|
"grad_norm": 0.6981467453263979,
|
||
|
|
"learning_rate": 3.650027202065617e-05,
|
||
|
|
"loss": 0.2905,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1113966852426529,
|
||
|
|
"step": 980,
|
||
|
|
"valid_targets_mean": 2686.6,
|
||
|
|
"valid_targets_min": 1086
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.917234664070107,
|
||
|
|
"grad_norm": 0.5759410728202508,
|
||
|
|
"learning_rate": 3.644524934647432e-05,
|
||
|
|
"loss": 0.2736,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11122055351734161,
|
||
|
|
"step": 985,
|
||
|
|
"valid_targets_mean": 3869.5,
|
||
|
|
"valid_targets_min": 974
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9269717624148004,
|
||
|
|
"grad_norm": 0.6088691920046002,
|
||
|
|
"learning_rate": 3.6389839659183064e-05,
|
||
|
|
"loss": 0.2846,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1156475841999054,
|
||
|
|
"step": 990,
|
||
|
|
"valid_targets_mean": 3328.2,
|
||
|
|
"valid_targets_min": 1268
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9367088607594938,
|
||
|
|
"grad_norm": 0.6429286727874445,
|
||
|
|
"learning_rate": 3.633404426276234e-05,
|
||
|
|
"loss": 0.2777,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12231488525867462,
|
||
|
|
"step": 995,
|
||
|
|
"valid_targets_mean": 3249.6,
|
||
|
|
"valid_targets_min": 1012
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.946445959104187,
|
||
|
|
"grad_norm": 0.7228518779689982,
|
||
|
|
"learning_rate": 3.6277864470269164e-05,
|
||
|
|
"loss": 0.2712,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12440205365419388,
|
||
|
|
"step": 1000,
|
||
|
|
"valid_targets_mean": 2283.8,
|
||
|
|
"valid_targets_min": 1294
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.95618305744888,
|
||
|
|
"grad_norm": 0.6995191271444386,
|
||
|
|
"learning_rate": 3.622130160380668e-05,
|
||
|
|
"loss": 0.2662,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1588570922613144,
|
||
|
|
"step": 1005,
|
||
|
|
"valid_targets_mean": 3194.8,
|
||
|
|
"valid_targets_min": 1316
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9659201557935735,
|
||
|
|
"grad_norm": 0.6375267751555264,
|
||
|
|
"learning_rate": 3.616435699449311e-05,
|
||
|
|
"loss": 0.2711,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08528415113687515,
|
||
|
|
"step": 1010,
|
||
|
|
"valid_targets_mean": 2210.2,
|
||
|
|
"valid_targets_min": 729
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9756572541382669,
|
||
|
|
"grad_norm": 0.639572278608358,
|
||
|
|
"learning_rate": 3.6107031982430374e-05,
|
||
|
|
"loss": 0.2974,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.193544402718544,
|
||
|
|
"step": 1015,
|
||
|
|
"valid_targets_mean": 4963.1,
|
||
|
|
"valid_targets_min": 1505
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.98539435248296,
|
||
|
|
"grad_norm": 0.865702036273128,
|
||
|
|
"learning_rate": 3.604932791667258e-05,
|
||
|
|
"loss": 0.2775,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15228410065174103,
|
||
|
|
"step": 1020,
|
||
|
|
"valid_targets_mean": 3556.8,
|
||
|
|
"valid_targets_min": 858
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9951314508276532,
|
||
|
|
"grad_norm": 0.6528761878848074,
|
||
|
|
"learning_rate": 3.599124615519427e-05,
|
||
|
|
"loss": 0.2423,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11325448751449585,
|
||
|
|
"step": 1025,
|
||
|
|
"valid_targets_mean": 2563.0,
|
||
|
|
"valid_targets_min": 1192
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0038948393378773,
|
||
|
|
"grad_norm": 0.7377205804499293,
|
||
|
|
"learning_rate": 3.5932788064858454e-05,
|
||
|
|
"loss": 0.2514,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1272452026605606,
|
||
|
|
"step": 1030,
|
||
|
|
"valid_targets_mean": 2710.0,
|
||
|
|
"valid_targets_min": 931
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0136319376825704,
|
||
|
|
"grad_norm": 0.8091893084525955,
|
||
|
|
"learning_rate": 3.5873955021384464e-05,
|
||
|
|
"loss": 0.2485,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11636292934417725,
|
||
|
|
"step": 1035,
|
||
|
|
"valid_targets_mean": 2944.6,
|
||
|
|
"valid_targets_min": 1276
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.023369036027264,
|
||
|
|
"grad_norm": 0.7003258404063661,
|
||
|
|
"learning_rate": 3.5814748409315545e-05,
|
||
|
|
"loss": 0.2476,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14736023545265198,
|
||
|
|
"step": 1040,
|
||
|
|
"valid_targets_mean": 3605.8,
|
||
|
|
"valid_targets_min": 810
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.033106134371957,
|
||
|
|
"grad_norm": 0.6474036924563122,
|
||
|
|
"learning_rate": 3.575516962198632e-05,
|
||
|
|
"loss": 0.242,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07607968896627426,
|
||
|
|
"step": 1045,
|
||
|
|
"valid_targets_mean": 2740.0,
|
||
|
|
"valid_targets_min": 1238
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0428432327166504,
|
||
|
|
"grad_norm": 0.824017837195956,
|
||
|
|
"learning_rate": 3.5695220061489935e-05,
|
||
|
|
"loss": 0.2429,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12290918081998825,
|
||
|
|
"step": 1050,
|
||
|
|
"valid_targets_mean": 2670.5,
|
||
|
|
"valid_targets_min": 1133
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0525803310613435,
|
||
|
|
"grad_norm": 0.7035984369000724,
|
||
|
|
"learning_rate": 3.563490113864514e-05,
|
||
|
|
"loss": 0.2464,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09453026950359344,
|
||
|
|
"step": 1055,
|
||
|
|
"valid_targets_mean": 2812.5,
|
||
|
|
"valid_targets_min": 1225
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.062317429406037,
|
||
|
|
"grad_norm": 0.7983219444234277,
|
||
|
|
"learning_rate": 3.557421427296302e-05,
|
||
|
|
"loss": 0.2339,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0978788286447525,
|
||
|
|
"step": 1060,
|
||
|
|
"valid_targets_mean": 2292.6,
|
||
|
|
"valid_targets_min": 1485
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0720545277507303,
|
||
|
|
"grad_norm": 0.7973735559663347,
|
||
|
|
"learning_rate": 3.551316089261363e-05,
|
||
|
|
"loss": 0.2394,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13901948928833008,
|
||
|
|
"step": 1065,
|
||
|
|
"valid_targets_mean": 3392.8,
|
||
|
|
"valid_targets_min": 1411
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0817916260954235,
|
||
|
|
"grad_norm": 0.9870859614285062,
|
||
|
|
"learning_rate": 3.545174243439236e-05,
|
||
|
|
"loss": 0.228,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13286170363426208,
|
||
|
|
"step": 1070,
|
||
|
|
"valid_targets_mean": 3008.1,
|
||
|
|
"valid_targets_min": 297
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.091528724440117,
|
||
|
|
"grad_norm": 0.7644411930451773,
|
||
|
|
"learning_rate": 3.538996034368615e-05,
|
||
|
|
"loss": 0.2415,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16460761427879333,
|
||
|
|
"step": 1075,
|
||
|
|
"valid_targets_mean": 4027.2,
|
||
|
|
"valid_targets_min": 1542
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1012658227848102,
|
||
|
|
"grad_norm": 0.7956734974517503,
|
||
|
|
"learning_rate": 3.5327816074439445e-05,
|
||
|
|
"loss": 0.243,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11771304905414581,
|
||
|
|
"step": 1080,
|
||
|
|
"valid_targets_mean": 2529.5,
|
||
|
|
"valid_targets_min": 879
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1110029211295034,
|
||
|
|
"grad_norm": 0.6436131645357464,
|
||
|
|
"learning_rate": 3.526531108912e-05,
|
||
|
|
"loss": 0.2331,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12277089059352875,
|
||
|
|
"step": 1085,
|
||
|
|
"valid_targets_mean": 3293.4,
|
||
|
|
"valid_targets_min": 591
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1207400194741965,
|
||
|
|
"grad_norm": 0.9138861433788087,
|
||
|
|
"learning_rate": 3.520244685868446e-05,
|
||
|
|
"loss": 0.2261,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1006382405757904,
|
||
|
|
"step": 1090,
|
||
|
|
"valid_targets_mean": 1795.1,
|
||
|
|
"valid_targets_min": 295
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.13047711781889,
|
||
|
|
"grad_norm": 0.7538499512907332,
|
||
|
|
"learning_rate": 3.513922486254373e-05,
|
||
|
|
"loss": 0.237,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0947801023721695,
|
||
|
|
"step": 1095,
|
||
|
|
"valid_targets_mean": 2311.6,
|
||
|
|
"valid_targets_min": 1122
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1402142161635833,
|
||
|
|
"grad_norm": 0.6105399013697741,
|
||
|
|
"learning_rate": 3.507564658852817e-05,
|
||
|
|
"loss": 0.2408,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1116853579878807,
|
||
|
|
"step": 1100,
|
||
|
|
"valid_targets_mean": 4266.0,
|
||
|
|
"valid_targets_min": 1560
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1499513145082765,
|
||
|
|
"grad_norm": 0.8544955534319482,
|
||
|
|
"learning_rate": 3.501171353285258e-05,
|
||
|
|
"loss": 0.2488,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.116363525390625,
|
||
|
|
"step": 1105,
|
||
|
|
"valid_targets_mean": 2383.8,
|
||
|
|
"valid_targets_min": 1061
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1596884128529696,
|
||
|
|
"grad_norm": 0.7574503962153785,
|
||
|
|
"learning_rate": 3.494742720008102e-05,
|
||
|
|
"loss": 0.24,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09310120344161987,
|
||
|
|
"step": 1110,
|
||
|
|
"valid_targets_mean": 2279.9,
|
||
|
|
"valid_targets_min": 323
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1694255111976632,
|
||
|
|
"grad_norm": 0.7054565876739526,
|
||
|
|
"learning_rate": 3.488278910309131e-05,
|
||
|
|
"loss": 0.2348,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11796418577432632,
|
||
|
|
"step": 1115,
|
||
|
|
"valid_targets_mean": 2481.4,
|
||
|
|
"valid_targets_min": 1256
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1791626095423564,
|
||
|
|
"grad_norm": 0.8624432463998876,
|
||
|
|
"learning_rate": 3.481780076303953e-05,
|
||
|
|
"loss": 0.2477,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13654282689094543,
|
||
|
|
"step": 1120,
|
||
|
|
"valid_targets_mean": 3006.1,
|
||
|
|
"valid_targets_min": 1040
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1888997078870496,
|
||
|
|
"grad_norm": 0.7729213149253524,
|
||
|
|
"learning_rate": 3.475246370932419e-05,
|
||
|
|
"loss": 0.2484,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1039193719625473,
|
||
|
|
"step": 1125,
|
||
|
|
"valid_targets_mean": 2482.0,
|
||
|
|
"valid_targets_min": 880
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1986368062317427,
|
||
|
|
"grad_norm": 0.8306922531898471,
|
||
|
|
"learning_rate": 3.468677947955019e-05,
|
||
|
|
"loss": 0.2347,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09656044840812683,
|
||
|
|
"step": 1130,
|
||
|
|
"valid_targets_mean": 1547.6,
|
||
|
|
"valid_targets_min": 734
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2083739045764363,
|
||
|
|
"grad_norm": 0.6807291213869415,
|
||
|
|
"learning_rate": 3.4620749619492704e-05,
|
||
|
|
"loss": 0.2236,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11662845313549042,
|
||
|
|
"step": 1135,
|
||
|
|
"valid_targets_mean": 3677.1,
|
||
|
|
"valid_targets_min": 1293
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2181110029211295,
|
||
|
|
"grad_norm": 0.773377055591104,
|
||
|
|
"learning_rate": 3.455437568306076e-05,
|
||
|
|
"loss": 0.2428,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13956855237483978,
|
||
|
|
"step": 1140,
|
||
|
|
"valid_targets_mean": 3065.8,
|
||
|
|
"valid_targets_min": 753
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2278481012658227,
|
||
|
|
"grad_norm": 0.6679841589471326,
|
||
|
|
"learning_rate": 3.448765923226069e-05,
|
||
|
|
"loss": 0.2152,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10934606194496155,
|
||
|
|
"step": 1145,
|
||
|
|
"valid_targets_mean": 4065.2,
|
||
|
|
"valid_targets_min": 727
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2375851996105163,
|
||
|
|
"grad_norm": 0.9727047265171258,
|
||
|
|
"learning_rate": 3.442060183715935e-05,
|
||
|
|
"loss": 0.2371,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13547798991203308,
|
||
|
|
"step": 1150,
|
||
|
|
"valid_targets_mean": 1850.2,
|
||
|
|
"valid_targets_min": 1051
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2473222979552094,
|
||
|
|
"grad_norm": 1.4802333281654472,
|
||
|
|
"learning_rate": 3.4353205075847195e-05,
|
||
|
|
"loss": 0.2553,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1198016032576561,
|
||
|
|
"step": 1155,
|
||
|
|
"valid_targets_mean": 2603.4,
|
||
|
|
"valid_targets_min": 903
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2570593962999026,
|
||
|
|
"grad_norm": 0.6955103270734062,
|
||
|
|
"learning_rate": 3.428547053440113e-05,
|
||
|
|
"loss": 0.2327,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1287013739347458,
|
||
|
|
"step": 1160,
|
||
|
|
"valid_targets_mean": 3495.5,
|
||
|
|
"valid_targets_min": 1050
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2667964946445958,
|
||
|
|
"grad_norm": 0.7088985386321863,
|
||
|
|
"learning_rate": 3.421739980684718e-05,
|
||
|
|
"loss": 0.2368,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10113312304019928,
|
||
|
|
"step": 1165,
|
||
|
|
"valid_targets_mean": 2847.8,
|
||
|
|
"valid_targets_min": 1382
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2765335929892894,
|
||
|
|
"grad_norm": 0.7514819406593808,
|
||
|
|
"learning_rate": 3.4148994495123e-05,
|
||
|
|
"loss": 0.2442,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1315937042236328,
|
||
|
|
"step": 1170,
|
||
|
|
"valid_targets_mean": 3536.6,
|
||
|
|
"valid_targets_min": 1575
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2862706913339825,
|
||
|
|
"grad_norm": 0.7145930585200598,
|
||
|
|
"learning_rate": 3.408025620904012e-05,
|
||
|
|
"loss": 0.2294,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13585850596427917,
|
||
|
|
"step": 1175,
|
||
|
|
"valid_targets_mean": 4002.5,
|
||
|
|
"valid_targets_min": 1390
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2960077896786757,
|
||
|
|
"grad_norm": 0.7845317543314521,
|
||
|
|
"learning_rate": 3.401118656624613e-05,
|
||
|
|
"loss": 0.2448,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12422779947519302,
|
||
|
|
"step": 1180,
|
||
|
|
"valid_targets_mean": 3427.6,
|
||
|
|
"valid_targets_min": 1350
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.305744888023369,
|
||
|
|
"grad_norm": 0.6726307033660588,
|
||
|
|
"learning_rate": 3.3941787192186584e-05,
|
||
|
|
"loss": 0.2507,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11506038904190063,
|
||
|
|
"step": 1185,
|
||
|
|
"valid_targets_mean": 3105.5,
|
||
|
|
"valid_targets_min": 762
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3154819863680625,
|
||
|
|
"grad_norm": 0.5479126063821994,
|
||
|
|
"learning_rate": 3.387205972006671e-05,
|
||
|
|
"loss": 0.2282,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1155364066362381,
|
||
|
|
"step": 1190,
|
||
|
|
"valid_targets_mean": 5606.8,
|
||
|
|
"valid_targets_min": 1619
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3252190847127556,
|
||
|
|
"grad_norm": 0.7289512246085008,
|
||
|
|
"learning_rate": 3.3802005790813046e-05,
|
||
|
|
"loss": 0.2157,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08744939416646957,
|
||
|
|
"step": 1195,
|
||
|
|
"valid_targets_mean": 2208.5,
|
||
|
|
"valid_targets_min": 317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.334956183057449,
|
||
|
|
"grad_norm": 0.790510816502957,
|
||
|
|
"learning_rate": 3.373162705303476e-05,
|
||
|
|
"loss": 0.2426,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11945021152496338,
|
||
|
|
"step": 1200,
|
||
|
|
"valid_targets_mean": 2726.6,
|
||
|
|
"valid_targets_min": 1387
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.344693281402142,
|
||
|
|
"grad_norm": 0.7673786848762931,
|
||
|
|
"learning_rate": 3.366092516298489e-05,
|
||
|
|
"loss": 0.2414,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11489750444889069,
|
||
|
|
"step": 1205,
|
||
|
|
"valid_targets_mean": 2555.6,
|
||
|
|
"valid_targets_min": 1796
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3544303797468356,
|
||
|
|
"grad_norm": 0.7113294796141776,
|
||
|
|
"learning_rate": 3.358990178452135e-05,
|
||
|
|
"loss": 0.2357,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11983954906463623,
|
||
|
|
"step": 1210,
|
||
|
|
"valid_targets_mean": 3281.4,
|
||
|
|
"valid_targets_min": 1324
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3641674780915287,
|
||
|
|
"grad_norm": 0.777022529971605,
|
||
|
|
"learning_rate": 3.351855858906778e-05,
|
||
|
|
"loss": 0.2617,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09583482146263123,
|
||
|
|
"step": 1215,
|
||
|
|
"valid_targets_mean": 2159.6,
|
||
|
|
"valid_targets_min": 674
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.373904576436222,
|
||
|
|
"grad_norm": 0.6562812529073736,
|
||
|
|
"learning_rate": 3.344689725557422e-05,
|
||
|
|
"loss": 0.2463,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12501710653305054,
|
||
|
|
"step": 1220,
|
||
|
|
"valid_targets_mean": 4507.2,
|
||
|
|
"valid_targets_min": 1699
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3836416747809155,
|
||
|
|
"grad_norm": 0.8459375828858623,
|
||
|
|
"learning_rate": 3.337491947047757e-05,
|
||
|
|
"loss": 0.2442,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14401216804981232,
|
||
|
|
"step": 1225,
|
||
|
|
"valid_targets_mean": 2357.1,
|
||
|
|
"valid_targets_min": 735
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3933787731256086,
|
||
|
|
"grad_norm": 0.8326039552981002,
|
||
|
|
"learning_rate": 3.330262692766193e-05,
|
||
|
|
"loss": 0.2421,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09895455837249756,
|
||
|
|
"step": 1230,
|
||
|
|
"valid_targets_mean": 2117.0,
|
||
|
|
"valid_targets_min": 1140
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.403115871470302,
|
||
|
|
"grad_norm": 0.7636805929064753,
|
||
|
|
"learning_rate": 3.323002132841875e-05,
|
||
|
|
"loss": 0.2603,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15124696493148804,
|
||
|
|
"step": 1235,
|
||
|
|
"valid_targets_mean": 3200.6,
|
||
|
|
"valid_targets_min": 1942
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.412852969814995,
|
||
|
|
"grad_norm": 0.7622703732144163,
|
||
|
|
"learning_rate": 3.315710438140674e-05,
|
||
|
|
"loss": 0.2351,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0872589498758316,
|
||
|
|
"step": 1240,
|
||
|
|
"valid_targets_mean": 2806.6,
|
||
|
|
"valid_targets_min": 434
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4225900681596886,
|
||
|
|
"grad_norm": 0.8028872445443147,
|
||
|
|
"learning_rate": 3.308387780261169e-05,
|
||
|
|
"loss": 0.2439,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1435239613056183,
|
||
|
|
"step": 1245,
|
||
|
|
"valid_targets_mean": 2922.1,
|
||
|
|
"valid_targets_min": 764
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4323271665043817,
|
||
|
|
"grad_norm": 0.6764755919211342,
|
||
|
|
"learning_rate": 3.3010343315306106e-05,
|
||
|
|
"loss": 0.2336,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1212049052119255,
|
||
|
|
"step": 1250,
|
||
|
|
"valid_targets_mean": 4281.9,
|
||
|
|
"valid_targets_min": 1472
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.442064264849075,
|
||
|
|
"grad_norm": 0.8208423036570491,
|
||
|
|
"learning_rate": 3.293650265000864e-05,
|
||
|
|
"loss": 0.252,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12002964317798615,
|
||
|
|
"step": 1255,
|
||
|
|
"valid_targets_mean": 2410.8,
|
||
|
|
"valid_targets_min": 369
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.451801363193768,
|
||
|
|
"grad_norm": 0.8169817154882845,
|
||
|
|
"learning_rate": 3.286235754444335e-05,
|
||
|
|
"loss": 0.2395,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1182420626282692,
|
||
|
|
"step": 1260,
|
||
|
|
"valid_targets_mean": 2966.0,
|
||
|
|
"valid_targets_min": 1388
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4615384615384617,
|
||
|
|
"grad_norm": 0.7317850822612334,
|
||
|
|
"learning_rate": 3.278790974349882e-05,
|
||
|
|
"loss": 0.2413,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08225102722644806,
|
||
|
|
"step": 1265,
|
||
|
|
"valid_targets_mean": 2597.4,
|
||
|
|
"valid_targets_min": 870
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.471275559883155,
|
||
|
|
"grad_norm": 0.688118158837201,
|
||
|
|
"learning_rate": 3.271316099918709e-05,
|
||
|
|
"loss": 0.2318,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1168079525232315,
|
||
|
|
"step": 1270,
|
||
|
|
"valid_targets_mean": 3529.0,
|
||
|
|
"valid_targets_min": 1027
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.481012658227848,
|
||
|
|
"grad_norm": 0.6621925099955391,
|
||
|
|
"learning_rate": 3.2638113070602436e-05,
|
||
|
|
"loss": 0.237,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10795827209949493,
|
||
|
|
"step": 1275,
|
||
|
|
"valid_targets_mean": 3523.0,
|
||
|
|
"valid_targets_min": 1461
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4907497565725416,
|
||
|
|
"grad_norm": 0.7147418152885615,
|
||
|
|
"learning_rate": 3.256276772387997e-05,
|
||
|
|
"loss": 0.2367,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10587899386882782,
|
||
|
|
"step": 1280,
|
||
|
|
"valid_targets_mean": 3488.6,
|
||
|
|
"valid_targets_min": 406
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5004868549172348,
|
||
|
|
"grad_norm": 0.5826637028854669,
|
||
|
|
"learning_rate": 3.248712673215405e-05,
|
||
|
|
"loss": 0.2395,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08619245886802673,
|
||
|
|
"step": 1285,
|
||
|
|
"valid_targets_mean": 3449.2,
|
||
|
|
"valid_targets_min": 347
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.510223953261928,
|
||
|
|
"grad_norm": 0.7131189592981495,
|
||
|
|
"learning_rate": 3.241119187551661e-05,
|
||
|
|
"loss": 0.2463,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11175684630870819,
|
||
|
|
"step": 1290,
|
||
|
|
"valid_targets_mean": 2802.9,
|
||
|
|
"valid_targets_min": 1079
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.519961051606621,
|
||
|
|
"grad_norm": 0.5797857681941478,
|
||
|
|
"learning_rate": 3.233496494097521e-05,
|
||
|
|
"loss": 0.2416,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1312026083469391,
|
||
|
|
"step": 1295,
|
||
|
|
"valid_targets_mean": 4597.2,
|
||
|
|
"valid_targets_min": 2042
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5296981499513143,
|
||
|
|
"grad_norm": 0.7921243303267592,
|
||
|
|
"learning_rate": 3.2258447722411e-05,
|
||
|
|
"loss": 0.2465,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10868055373430252,
|
||
|
|
"step": 1300,
|
||
|
|
"valid_targets_mean": 2683.8,
|
||
|
|
"valid_targets_min": 716
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.539435248296008,
|
||
|
|
"grad_norm": 0.774825290380697,
|
||
|
|
"learning_rate": 3.2181642020536536e-05,
|
||
|
|
"loss": 0.2383,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12708991765975952,
|
||
|
|
"step": 1305,
|
||
|
|
"valid_targets_mean": 2726.1,
|
||
|
|
"valid_targets_min": 1300
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.549172346640701,
|
||
|
|
"grad_norm": 0.762900017416662,
|
||
|
|
"learning_rate": 3.210454964285333e-05,
|
||
|
|
"loss": 0.2257,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13067397475242615,
|
||
|
|
"step": 1310,
|
||
|
|
"valid_targets_mean": 3118.5,
|
||
|
|
"valid_targets_min": 1296
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.558909444985394,
|
||
|
|
"grad_norm": 0.7072153569488926,
|
||
|
|
"learning_rate": 3.202717240360941e-05,
|
||
|
|
"loss": 0.2169,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.135477215051651,
|
||
|
|
"step": 1315,
|
||
|
|
"valid_targets_mean": 4222.9,
|
||
|
|
"valid_targets_min": 683
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.568646543330088,
|
||
|
|
"grad_norm": 0.6623399885812072,
|
||
|
|
"learning_rate": 3.194951212375654e-05,
|
||
|
|
"loss": 0.2327,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10676338523626328,
|
||
|
|
"step": 1320,
|
||
|
|
"valid_targets_mean": 3114.4,
|
||
|
|
"valid_targets_min": 1354
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.578383641674781,
|
||
|
|
"grad_norm": 0.7492045989504769,
|
||
|
|
"learning_rate": 3.18715706309074e-05,
|
||
|
|
"loss": 0.2357,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13046498596668243,
|
||
|
|
"step": 1325,
|
||
|
|
"valid_targets_mean": 2841.6,
|
||
|
|
"valid_targets_min": 1208
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.588120740019474,
|
||
|
|
"grad_norm": 0.726912759253883,
|
||
|
|
"learning_rate": 3.17933497592926e-05,
|
||
|
|
"loss": 0.2502,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12865900993347168,
|
||
|
|
"step": 1330,
|
||
|
|
"valid_targets_mean": 3095.1,
|
||
|
|
"valid_targets_min": 879
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5978578383641677,
|
||
|
|
"grad_norm": 1.1433374470740283,
|
||
|
|
"learning_rate": 3.171485134971748e-05,
|
||
|
|
"loss": 0.2534,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15513890981674194,
|
||
|
|
"step": 1335,
|
||
|
|
"valid_targets_mean": 2641.1,
|
||
|
|
"valid_targets_min": 1152
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.607594936708861,
|
||
|
|
"grad_norm": 0.7637229854220811,
|
||
|
|
"learning_rate": 3.163607724951877e-05,
|
||
|
|
"loss": 0.2361,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09301097691059113,
|
||
|
|
"step": 1340,
|
||
|
|
"valid_targets_mean": 1989.9,
|
||
|
|
"valid_targets_min": 1201
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.617332035053554,
|
||
|
|
"grad_norm": 0.8127208648335544,
|
||
|
|
"learning_rate": 3.155702931252119e-05,
|
||
|
|
"loss": 0.2381,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1383160650730133,
|
||
|
|
"step": 1345,
|
||
|
|
"valid_targets_mean": 3118.1,
|
||
|
|
"valid_targets_min": 832
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.627069133398247,
|
||
|
|
"grad_norm": 0.6903895599708771,
|
||
|
|
"learning_rate": 3.147770939899376e-05,
|
||
|
|
"loss": 0.2203,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07707555592060089,
|
||
|
|
"step": 1350,
|
||
|
|
"valid_targets_mean": 2399.9,
|
||
|
|
"valid_targets_min": 1426
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6368062317429404,
|
||
|
|
"grad_norm": 0.7897696866609453,
|
||
|
|
"learning_rate": 3.1398119375606046e-05,
|
||
|
|
"loss": 0.2286,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11783762276172638,
|
||
|
|
"step": 1355,
|
||
|
|
"valid_targets_mean": 2069.9,
|
||
|
|
"valid_targets_min": 1264
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.646543330087634,
|
||
|
|
"grad_norm": 0.7747773221903235,
|
||
|
|
"learning_rate": 3.13182611153842e-05,
|
||
|
|
"loss": 0.2414,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13851211965084076,
|
||
|
|
"step": 1360,
|
||
|
|
"valid_targets_mean": 3161.2,
|
||
|
|
"valid_targets_min": 1458
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.656280428432327,
|
||
|
|
"grad_norm": 0.8478030732090569,
|
||
|
|
"learning_rate": 3.1238136497666944e-05,
|
||
|
|
"loss": 0.2344,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12053705751895905,
|
||
|
|
"step": 1365,
|
||
|
|
"valid_targets_mean": 3051.6,
|
||
|
|
"valid_targets_min": 947
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6660175267770203,
|
||
|
|
"grad_norm": 0.6743401573821601,
|
||
|
|
"learning_rate": 3.115774740806128e-05,
|
||
|
|
"loss": 0.2388,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1084040105342865,
|
||
|
|
"step": 1370,
|
||
|
|
"valid_targets_mean": 3590.5,
|
||
|
|
"valid_targets_min": 1565
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.675754625121714,
|
||
|
|
"grad_norm": 0.8102210531374561,
|
||
|
|
"learning_rate": 3.1077095738398153e-05,
|
||
|
|
"loss": 0.2362,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10908074676990509,
|
||
|
|
"step": 1375,
|
||
|
|
"valid_targets_mean": 2245.8,
|
||
|
|
"valid_targets_min": 1055
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.685491723466407,
|
||
|
|
"grad_norm": 0.8612666953860721,
|
||
|
|
"learning_rate": 3.099618338668791e-05,
|
||
|
|
"loss": 0.2473,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08933654427528381,
|
||
|
|
"step": 1380,
|
||
|
|
"valid_targets_mean": 1772.0,
|
||
|
|
"valid_targets_min": 761
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6952288218111002,
|
||
|
|
"grad_norm": 0.7294410115256623,
|
||
|
|
"learning_rate": 3.0915012257075635e-05,
|
||
|
|
"loss": 0.2214,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12363948673009872,
|
||
|
|
"step": 1385,
|
||
|
|
"valid_targets_mean": 3711.4,
|
||
|
|
"valid_targets_min": 1066
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.704965920155794,
|
||
|
|
"grad_norm": 0.7820610472085976,
|
||
|
|
"learning_rate": 3.083358425979637e-05,
|
||
|
|
"loss": 0.2472,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08485744148492813,
|
||
|
|
"step": 1390,
|
||
|
|
"valid_targets_mean": 1744.8,
|
||
|
|
"valid_targets_min": 942
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.714703018500487,
|
||
|
|
"grad_norm": 0.8743217316124318,
|
||
|
|
"learning_rate": 3.075190131113009e-05,
|
||
|
|
"loss": 0.2428,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12635160982608795,
|
||
|
|
"step": 1395,
|
||
|
|
"valid_targets_mean": 2463.0,
|
||
|
|
"valid_targets_min": 825
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.72444011684518,
|
||
|
|
"grad_norm": 0.7151099603336041,
|
||
|
|
"learning_rate": 3.0669965333356695e-05,
|
||
|
|
"loss": 0.2436,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0974293202161789,
|
||
|
|
"step": 1400,
|
||
|
|
"valid_targets_mean": 2545.4,
|
||
|
|
"valid_targets_min": 907
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7341772151898733,
|
||
|
|
"grad_norm": 0.8709721011876739,
|
||
|
|
"learning_rate": 3.058777825471071e-05,
|
||
|
|
"loss": 0.2423,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0986693874001503,
|
||
|
|
"step": 1405,
|
||
|
|
"valid_targets_mean": 1826.6,
|
||
|
|
"valid_targets_min": 603
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7439143135345665,
|
||
|
|
"grad_norm": 0.5683439455404877,
|
||
|
|
"learning_rate": 3.0505342009335898e-05,
|
||
|
|
"loss": 0.2221,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04642200842499733,
|
||
|
|
"step": 1410,
|
||
|
|
"valid_targets_mean": 2629.6,
|
||
|
|
"valid_targets_min": 1201
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.75365141187926,
|
||
|
|
"grad_norm": 0.7939068734660043,
|
||
|
|
"learning_rate": 3.0422658537239823e-05,
|
||
|
|
"loss": 0.2509,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10494433343410492,
|
||
|
|
"step": 1415,
|
||
|
|
"valid_targets_mean": 2548.2,
|
||
|
|
"valid_targets_min": 1609
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7633885102239533,
|
||
|
|
"grad_norm": 0.728119544785679,
|
||
|
|
"learning_rate": 3.0339729784248103e-05,
|
||
|
|
"loss": 0.2306,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10721099376678467,
|
||
|
|
"step": 1420,
|
||
|
|
"valid_targets_mean": 3043.4,
|
||
|
|
"valid_targets_min": 415
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7731256085686464,
|
||
|
|
"grad_norm": 0.6664958468000649,
|
||
|
|
"learning_rate": 3.025655770195866e-05,
|
||
|
|
"loss": 0.246,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10482774674892426,
|
||
|
|
"step": 1425,
|
||
|
|
"valid_targets_mean": 3292.0,
|
||
|
|
"valid_targets_min": 828
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.78286270691334,
|
||
|
|
"grad_norm": 0.7866347340028299,
|
||
|
|
"learning_rate": 3.0173144247695796e-05,
|
||
|
|
"loss": 0.2365,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08502174913883209,
|
||
|
|
"step": 1430,
|
||
|
|
"valid_targets_mean": 2277.9,
|
||
|
|
"valid_targets_min": 1290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.792599805258033,
|
||
|
|
"grad_norm": 0.7338395704422648,
|
||
|
|
"learning_rate": 3.008949138446413e-05,
|
||
|
|
"loss": 0.2161,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1111348569393158,
|
||
|
|
"step": 1435,
|
||
|
|
"valid_targets_mean": 2493.5,
|
||
|
|
"valid_targets_min": 1220
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8023369036027264,
|
||
|
|
"grad_norm": 0.6653501894540796,
|
||
|
|
"learning_rate": 3.0005601080902376e-05,
|
||
|
|
"loss": 0.2319,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13429823517799377,
|
||
|
|
"step": 1440,
|
||
|
|
"valid_targets_mean": 4067.4,
|
||
|
|
"valid_targets_min": 726
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8120740019474195,
|
||
|
|
"grad_norm": 0.6643398691845159,
|
||
|
|
"learning_rate": 2.9921475311237055e-05,
|
||
|
|
"loss": 0.2283,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1200399249792099,
|
||
|
|
"step": 1445,
|
||
|
|
"valid_targets_mean": 3554.1,
|
||
|
|
"valid_targets_min": 353
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8218111002921127,
|
||
|
|
"grad_norm": 0.7001948444160422,
|
||
|
|
"learning_rate": 2.9837116055236e-05,
|
||
|
|
"loss": 0.2414,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07412829995155334,
|
||
|
|
"step": 1450,
|
||
|
|
"valid_targets_mean": 2387.1,
|
||
|
|
"valid_targets_min": 1514
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8315481986368063,
|
||
|
|
"grad_norm": 0.8364387339858684,
|
||
|
|
"learning_rate": 2.975252529816178e-05,
|
||
|
|
"loss": 0.2334,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12055306136608124,
|
||
|
|
"step": 1455,
|
||
|
|
"valid_targets_mean": 2205.9,
|
||
|
|
"valid_targets_min": 490
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8412852969814995,
|
||
|
|
"grad_norm": 0.8070899662966771,
|
||
|
|
"learning_rate": 2.9667705030724976e-05,
|
||
|
|
"loss": 0.2257,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11533784866333008,
|
||
|
|
"step": 1460,
|
||
|
|
"valid_targets_mean": 2544.1,
|
||
|
|
"valid_targets_min": 1081
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8510223953261926,
|
||
|
|
"grad_norm": 0.7655668406337863,
|
||
|
|
"learning_rate": 2.9582657249037335e-05,
|
||
|
|
"loss": 0.225,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09084555506706238,
|
||
|
|
"step": 1465,
|
||
|
|
"valid_targets_mean": 2362.1,
|
||
|
|
"valid_targets_min": 1568
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8607594936708862,
|
||
|
|
"grad_norm": 0.682547283492675,
|
||
|
|
"learning_rate": 2.9497383954564807e-05,
|
||
|
|
"loss": 0.2405,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12286686152219772,
|
||
|
|
"step": 1470,
|
||
|
|
"valid_targets_mean": 2804.1,
|
||
|
|
"valid_targets_min": 688
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8704965920155794,
|
||
|
|
"grad_norm": 0.6403813646612935,
|
||
|
|
"learning_rate": 2.941188715408041e-05,
|
||
|
|
"loss": 0.2342,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10485698282718658,
|
||
|
|
"step": 1475,
|
||
|
|
"valid_targets_mean": 3492.6,
|
||
|
|
"valid_targets_min": 873
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8802336903602725,
|
||
|
|
"grad_norm": 0.834361603882908,
|
||
|
|
"learning_rate": 2.9326168859617044e-05,
|
||
|
|
"loss": 0.2276,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09768801182508469,
|
||
|
|
"step": 1480,
|
||
|
|
"valid_targets_mean": 1849.0,
|
||
|
|
"valid_targets_min": 698
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.889970788704966,
|
||
|
|
"grad_norm": 0.8142513172358198,
|
||
|
|
"learning_rate": 2.9240231088420112e-05,
|
||
|
|
"loss": 0.2319,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16129133105278015,
|
||
|
|
"step": 1485,
|
||
|
|
"valid_targets_mean": 3350.1,
|
||
|
|
"valid_targets_min": 1870
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8997078870496593,
|
||
|
|
"grad_norm": 0.6938599150985236,
|
||
|
|
"learning_rate": 2.9154075862900047e-05,
|
||
|
|
"loss": 0.2321,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1012616902589798,
|
||
|
|
"step": 1490,
|
||
|
|
"valid_targets_mean": 3434.6,
|
||
|
|
"valid_targets_min": 1459
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9094449853943525,
|
||
|
|
"grad_norm": 0.8285559001767225,
|
||
|
|
"learning_rate": 2.9067705210584764e-05,
|
||
|
|
"loss": 0.2476,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14174655079841614,
|
||
|
|
"step": 1495,
|
||
|
|
"valid_targets_mean": 3295.6,
|
||
|
|
"valid_targets_min": 1011
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9191820837390456,
|
||
|
|
"grad_norm": 0.8908683413393621,
|
||
|
|
"learning_rate": 2.898112116407186e-05,
|
||
|
|
"loss": 0.2307,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1453523486852646,
|
||
|
|
"step": 1500,
|
||
|
|
"valid_targets_mean": 2463.2,
|
||
|
|
"valid_targets_min": 676
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.928919182083739,
|
||
|
|
"grad_norm": 0.7010616312835074,
|
||
|
|
"learning_rate": 2.889432576098087e-05,
|
||
|
|
"loss": 0.2436,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1279774010181427,
|
||
|
|
"step": 1505,
|
||
|
|
"valid_targets_mean": 3304.6,
|
||
|
|
"valid_targets_min": 1697
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9386562804284324,
|
||
|
|
"grad_norm": 0.6383882094213501,
|
||
|
|
"learning_rate": 2.8807321043905253e-05,
|
||
|
|
"loss": 0.2196,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0935613363981247,
|
||
|
|
"step": 1510,
|
||
|
|
"valid_targets_mean": 3910.6,
|
||
|
|
"valid_targets_min": 1412
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9483933787731256,
|
||
|
|
"grad_norm": 0.6063463816177003,
|
||
|
|
"learning_rate": 2.872010906036435e-05,
|
||
|
|
"loss": 0.2466,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10311927646398544,
|
||
|
|
"step": 1515,
|
||
|
|
"valid_targets_mean": 3975.9,
|
||
|
|
"valid_targets_min": 1086
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9581304771178187,
|
||
|
|
"grad_norm": 0.765783427016532,
|
||
|
|
"learning_rate": 2.863269186275519e-05,
|
||
|
|
"loss": 0.2383,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14790189266204834,
|
||
|
|
"step": 1520,
|
||
|
|
"valid_targets_mean": 2920.8,
|
||
|
|
"valid_targets_min": 1026
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9678675754625123,
|
||
|
|
"grad_norm": 0.6912276152303378,
|
||
|
|
"learning_rate": 2.8545071508304193e-05,
|
||
|
|
"loss": 0.2358,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14214438199996948,
|
||
|
|
"step": 1525,
|
||
|
|
"valid_targets_mean": 3831.4,
|
||
|
|
"valid_targets_min": 1235
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9776046738072055,
|
||
|
|
"grad_norm": 0.6832518900842646,
|
||
|
|
"learning_rate": 2.8457250059018765e-05,
|
||
|
|
"loss": 0.2333,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13680899143218994,
|
||
|
|
"step": 1530,
|
||
|
|
"valid_targets_mean": 3673.4,
|
||
|
|
"valid_targets_min": 1012
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9873417721518987,
|
||
|
|
"grad_norm": 0.6389373131378439,
|
||
|
|
"learning_rate": 2.836922958163875e-05,
|
||
|
|
"loss": 0.2399,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09092959761619568,
|
||
|
|
"step": 1535,
|
||
|
|
"valid_targets_mean": 2189.8,
|
||
|
|
"valid_targets_min": 986
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9970788704965923,
|
||
|
|
"grad_norm": 0.7702760592823965,
|
||
|
|
"learning_rate": 2.8281012147587816e-05,
|
||
|
|
"loss": 0.2403,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15108105540275574,
|
||
|
|
"step": 1540,
|
||
|
|
"valid_targets_mean": 3401.4,
|
||
|
|
"valid_targets_min": 1125
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.005842259006816,
|
||
|
|
"grad_norm": 0.7884417020648354,
|
||
|
|
"learning_rate": 2.8192599832924677e-05,
|
||
|
|
"loss": 0.2116,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11154694110155106,
|
||
|
|
"step": 1545,
|
||
|
|
"valid_targets_mean": 2741.4,
|
||
|
|
"valid_targets_min": 1141
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.015579357351509,
|
||
|
|
"grad_norm": 0.932462463859451,
|
||
|
|
"learning_rate": 2.810399471829429e-05,
|
||
|
|
"loss": 0.2033,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1208382397890091,
|
||
|
|
"step": 1550,
|
||
|
|
"valid_targets_mean": 2476.8,
|
||
|
|
"valid_targets_min": 790
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0253164556962027,
|
||
|
|
"grad_norm": 0.8686517940140047,
|
||
|
|
"learning_rate": 2.8015198888878835e-05,
|
||
|
|
"loss": 0.2044,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08823367953300476,
|
||
|
|
"step": 1555,
|
||
|
|
"valid_targets_mean": 2029.2,
|
||
|
|
"valid_targets_min": 832
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.035053554040896,
|
||
|
|
"grad_norm": 0.7193788183596941,
|
||
|
|
"learning_rate": 2.7926214434348645e-05,
|
||
|
|
"loss": 0.2216,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11511079967021942,
|
||
|
|
"step": 1560,
|
||
|
|
"valid_targets_mean": 3413.9,
|
||
|
|
"valid_targets_min": 805
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.044790652385589,
|
||
|
|
"grad_norm": 0.8078546054093075,
|
||
|
|
"learning_rate": 2.78370434488131e-05,
|
||
|
|
"loss": 0.2148,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07618220895528793,
|
||
|
|
"step": 1565,
|
||
|
|
"valid_targets_mean": 1835.6,
|
||
|
|
"valid_targets_min": 331
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0545277507302826,
|
||
|
|
"grad_norm": 0.66507455067397,
|
||
|
|
"learning_rate": 2.7747688030771253e-05,
|
||
|
|
"loss": 0.2039,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0778750628232956,
|
||
|
|
"step": 1570,
|
||
|
|
"valid_targets_mean": 3440.4,
|
||
|
|
"valid_targets_min": 1193
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0642648490749758,
|
||
|
|
"grad_norm": 0.9655649886690748,
|
||
|
|
"learning_rate": 2.765815028306249e-05,
|
||
|
|
"loss": 0.2026,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.130184605717659,
|
||
|
|
"step": 1575,
|
||
|
|
"valid_targets_mean": 2262.8,
|
||
|
|
"valid_targets_min": 889
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.074001947419669,
|
||
|
|
"grad_norm": 0.7563092839122303,
|
||
|
|
"learning_rate": 2.756843231281707e-05,
|
||
|
|
"loss": 0.2062,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11092890053987503,
|
||
|
|
"step": 1580,
|
||
|
|
"valid_targets_mean": 3115.4,
|
||
|
|
"valid_targets_min": 1238
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.083739045764362,
|
||
|
|
"grad_norm": 1.0075562492746872,
|
||
|
|
"learning_rate": 2.747853623140646e-05,
|
||
|
|
"loss": 0.2066,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11350446939468384,
|
||
|
|
"step": 1585,
|
||
|
|
"valid_targets_mean": 3914.6,
|
||
|
|
"valid_targets_min": 1350
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0934761441090557,
|
||
|
|
"grad_norm": 0.8524230873224183,
|
||
|
|
"learning_rate": 2.7388464154393735e-05,
|
||
|
|
"loss": 0.2087,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11565576493740082,
|
||
|
|
"step": 1590,
|
||
|
|
"valid_targets_mean": 2432.1,
|
||
|
|
"valid_targets_min": 1141
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.103213242453749,
|
||
|
|
"grad_norm": 0.9501433370814879,
|
||
|
|
"learning_rate": 2.7298218201483725e-05,
|
||
|
|
"loss": 0.2075,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11040070652961731,
|
||
|
|
"step": 1595,
|
||
|
|
"valid_targets_mean": 2233.0,
|
||
|
|
"valid_targets_min": 1012
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.112950340798442,
|
||
|
|
"grad_norm": 0.7471389146890169,
|
||
|
|
"learning_rate": 2.7207800496473157e-05,
|
||
|
|
"loss": 0.2003,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08324768394231796,
|
||
|
|
"step": 1600,
|
||
|
|
"valid_targets_mean": 2868.4,
|
||
|
|
"valid_targets_min": 1412
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.122687439143135,
|
||
|
|
"grad_norm": 0.8740749170157414,
|
||
|
|
"learning_rate": 2.7117213167200696e-05,
|
||
|
|
"loss": 0.2042,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13211554288864136,
|
||
|
|
"step": 1605,
|
||
|
|
"valid_targets_mean": 2986.2,
|
||
|
|
"valid_targets_min": 1504
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.132424537487829,
|
||
|
|
"grad_norm": 0.7082769479304671,
|
||
|
|
"learning_rate": 2.7026458345496826e-05,
|
||
|
|
"loss": 0.2217,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08424953371286392,
|
||
|
|
"step": 1610,
|
||
|
|
"valid_targets_mean": 3818.8,
|
||
|
|
"valid_targets_min": 1278
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.142161635832522,
|
||
|
|
"grad_norm": 1.1956579863090437,
|
||
|
|
"learning_rate": 2.6935538167133702e-05,
|
||
|
|
"loss": 0.2121,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10671257227659225,
|
||
|
|
"step": 1615,
|
||
|
|
"valid_targets_mean": 2870.0,
|
||
|
|
"valid_targets_min": 874
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.151898734177215,
|
||
|
|
"grad_norm": 0.7418229133223673,
|
||
|
|
"learning_rate": 2.6844454771774898e-05,
|
||
|
|
"loss": 0.2083,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10029250383377075,
|
||
|
|
"step": 1620,
|
||
|
|
"valid_targets_mean": 3097.9,
|
||
|
|
"valid_targets_min": 879
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1616358325219083,
|
||
|
|
"grad_norm": 0.834880643912381,
|
||
|
|
"learning_rate": 2.6753210302925044e-05,
|
||
|
|
"loss": 0.201,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11291643977165222,
|
||
|
|
"step": 1625,
|
||
|
|
"valid_targets_mean": 2799.5,
|
||
|
|
"valid_targets_min": 760
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.171372930866602,
|
||
|
|
"grad_norm": 0.8248332557311356,
|
||
|
|
"learning_rate": 2.6661806907879385e-05,
|
||
|
|
"loss": 0.2071,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06772299110889435,
|
||
|
|
"step": 1630,
|
||
|
|
"valid_targets_mean": 1523.8,
|
||
|
|
"valid_targets_min": 784
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.181110029211295,
|
||
|
|
"grad_norm": 0.850479033976409,
|
||
|
|
"learning_rate": 2.6570246737673216e-05,
|
||
|
|
"loss": 0.2029,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11443410068750381,
|
||
|
|
"step": 1635,
|
||
|
|
"valid_targets_mean": 2366.2,
|
||
|
|
"valid_targets_min": 996
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.190847127555988,
|
||
|
|
"grad_norm": 0.8214978220609186,
|
||
|
|
"learning_rate": 2.6478531947031325e-05,
|
||
|
|
"loss": 0.2102,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08211231231689453,
|
||
|
|
"step": 1640,
|
||
|
|
"valid_targets_mean": 2102.9,
|
||
|
|
"valid_targets_min": 848
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2005842259006814,
|
||
|
|
"grad_norm": 0.7642459302531467,
|
||
|
|
"learning_rate": 2.638666469431723e-05,
|
||
|
|
"loss": 0.2055,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09310499578714371,
|
||
|
|
"step": 1645,
|
||
|
|
"valid_targets_mean": 2598.9,
|
||
|
|
"valid_targets_min": 996
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.210321324245375,
|
||
|
|
"grad_norm": 0.7279570609997414,
|
||
|
|
"learning_rate": 2.6294647141482413e-05,
|
||
|
|
"loss": 0.2032,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10722191631793976,
|
||
|
|
"step": 1650,
|
||
|
|
"valid_targets_mean": 3704.6,
|
||
|
|
"valid_targets_min": 1141
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.220058422590068,
|
||
|
|
"grad_norm": 0.6325703321144482,
|
||
|
|
"learning_rate": 2.6202481454015426e-05,
|
||
|
|
"loss": 0.1842,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08083665370941162,
|
||
|
|
"step": 1655,
|
||
|
|
"valid_targets_mean": 2974.0,
|
||
|
|
"valid_targets_min": 1032
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2297955209347613,
|
||
|
|
"grad_norm": 0.8206360624533692,
|
||
|
|
"learning_rate": 2.6110169800890937e-05,
|
||
|
|
"loss": 0.2165,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10948295146226883,
|
||
|
|
"step": 1660,
|
||
|
|
"valid_targets_mean": 2686.6,
|
||
|
|
"valid_targets_min": 1341
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.239532619279455,
|
||
|
|
"grad_norm": 0.6636697940759875,
|
||
|
|
"learning_rate": 2.6017714354518696e-05,
|
||
|
|
"loss": 0.2061,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11225951462984085,
|
||
|
|
"step": 1665,
|
||
|
|
"valid_targets_mean": 4094.9,
|
||
|
|
"valid_targets_min": 1625
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.249269717624148,
|
||
|
|
"grad_norm": 0.6399187741843992,
|
||
|
|
"learning_rate": 2.592511729069239e-05,
|
||
|
|
"loss": 0.1873,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11166227608919144,
|
||
|
|
"step": 1670,
|
||
|
|
"valid_targets_mean": 4579.6,
|
||
|
|
"valid_targets_min": 888
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2590068159688412,
|
||
|
|
"grad_norm": 1.0378728175418352,
|
||
|
|
"learning_rate": 2.5832380788538457e-05,
|
||
|
|
"loss": 0.2108,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1347440779209137,
|
||
|
|
"step": 1675,
|
||
|
|
"valid_targets_mean": 4024.6,
|
||
|
|
"valid_targets_min": 930
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2687439143135344,
|
||
|
|
"grad_norm": 0.7778994020586357,
|
||
|
|
"learning_rate": 2.573950703046479e-05,
|
||
|
|
"loss": 0.2054,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09186933189630508,
|
||
|
|
"step": 1680,
|
||
|
|
"valid_targets_mean": 2595.4,
|
||
|
|
"valid_targets_min": 1197
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.278481012658228,
|
||
|
|
"grad_norm": 0.6882678466561416,
|
||
|
|
"learning_rate": 2.564649820210939e-05,
|
||
|
|
"loss": 0.2147,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09283961355686188,
|
||
|
|
"step": 1685,
|
||
|
|
"valid_targets_mean": 3646.6,
|
||
|
|
"valid_targets_min": 1005
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.288218111002921,
|
||
|
|
"grad_norm": 0.8323723770495097,
|
||
|
|
"learning_rate": 2.5553356492288933e-05,
|
||
|
|
"loss": 0.208,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14615213871002197,
|
||
|
|
"step": 1690,
|
||
|
|
"valid_targets_mean": 3665.2,
|
||
|
|
"valid_targets_min": 828
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2979552093476143,
|
||
|
|
"grad_norm": 0.8864387309698119,
|
||
|
|
"learning_rate": 2.5460084092947233e-05,
|
||
|
|
"loss": 0.191,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10318636894226074,
|
||
|
|
"step": 1695,
|
||
|
|
"valid_targets_mean": 2623.8,
|
||
|
|
"valid_targets_min": 290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3076923076923075,
|
||
|
|
"grad_norm": 0.8413830842891964,
|
||
|
|
"learning_rate": 2.5366683199103692e-05,
|
||
|
|
"loss": 0.1957,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11743028461933136,
|
||
|
|
"step": 1700,
|
||
|
|
"valid_targets_mean": 2832.2,
|
||
|
|
"valid_targets_min": 1831
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.317429406037001,
|
||
|
|
"grad_norm": 0.8730154800545424,
|
||
|
|
"learning_rate": 2.5273156008801614e-05,
|
||
|
|
"loss": 0.2008,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09757040441036224,
|
||
|
|
"step": 1705,
|
||
|
|
"valid_targets_mean": 1879.2,
|
||
|
|
"valid_targets_min": 562
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3271665043816943,
|
||
|
|
"grad_norm": 0.6255478267813253,
|
||
|
|
"learning_rate": 2.517950472305651e-05,
|
||
|
|
"loss": 0.1971,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08264683187007904,
|
||
|
|
"step": 1710,
|
||
|
|
"valid_targets_mean": 4320.8,
|
||
|
|
"valid_targets_min": 1713
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3369036027263874,
|
||
|
|
"grad_norm": 0.7830805843533527,
|
||
|
|
"learning_rate": 2.508573154580427e-05,
|
||
|
|
"loss": 0.206,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08903951942920685,
|
||
|
|
"step": 1715,
|
||
|
|
"valid_targets_mean": 2675.0,
|
||
|
|
"valid_targets_min": 1307
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.346640701071081,
|
||
|
|
"grad_norm": 0.7058135583859145,
|
||
|
|
"learning_rate": 2.4991838683849306e-05,
|
||
|
|
"loss": 0.2081,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16063743829727173,
|
||
|
|
"step": 1720,
|
||
|
|
"valid_targets_mean": 4538.2,
|
||
|
|
"valid_targets_min": 1549
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.356377799415774,
|
||
|
|
"grad_norm": 0.8558203857749066,
|
||
|
|
"learning_rate": 2.4897828346812637e-05,
|
||
|
|
"loss": 0.2063,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08233465254306793,
|
||
|
|
"step": 1725,
|
||
|
|
"valid_targets_mean": 2467.9,
|
||
|
|
"valid_targets_min": 1055
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3661148977604674,
|
||
|
|
"grad_norm": 0.8346203949558786,
|
||
|
|
"learning_rate": 2.4803702747079858e-05,
|
||
|
|
"loss": 0.2123,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09843271225690842,
|
||
|
|
"step": 1730,
|
||
|
|
"valid_targets_mean": 2434.5,
|
||
|
|
"valid_targets_min": 579
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3758519961051605,
|
||
|
|
"grad_norm": 0.6408101681065131,
|
||
|
|
"learning_rate": 2.4709464099749093e-05,
|
||
|
|
"loss": 0.1934,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11571668833494186,
|
||
|
|
"step": 1735,
|
||
|
|
"valid_targets_mean": 4195.4,
|
||
|
|
"valid_targets_min": 627
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.385589094449854,
|
||
|
|
"grad_norm": 0.7510417907281103,
|
||
|
|
"learning_rate": 2.4615114622578875e-05,
|
||
|
|
"loss": 0.2036,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08446644991636276,
|
||
|
|
"step": 1740,
|
||
|
|
"valid_targets_mean": 3063.1,
|
||
|
|
"valid_targets_min": 896
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3953261927945473,
|
||
|
|
"grad_norm": 0.6380803714432992,
|
||
|
|
"learning_rate": 2.4520656535935927e-05,
|
||
|
|
"loss": 0.2045,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06864303350448608,
|
||
|
|
"step": 1745,
|
||
|
|
"valid_targets_mean": 2743.0,
|
||
|
|
"valid_targets_min": 1665
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4050632911392404,
|
||
|
|
"grad_norm": 0.7549085112117736,
|
||
|
|
"learning_rate": 2.4426092062742933e-05,
|
||
|
|
"loss": 0.1952,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10727421194314957,
|
||
|
|
"step": 1750,
|
||
|
|
"valid_targets_mean": 2954.8,
|
||
|
|
"valid_targets_min": 1493
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4148003894839336,
|
||
|
|
"grad_norm": 0.9668943886987618,
|
||
|
|
"learning_rate": 2.4331423428426227e-05,
|
||
|
|
"loss": 0.1975,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12314222753047943,
|
||
|
|
"step": 1755,
|
||
|
|
"valid_targets_mean": 2293.2,
|
||
|
|
"valid_targets_min": 1388
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.424537487828627,
|
||
|
|
"grad_norm": 0.7866864607213474,
|
||
|
|
"learning_rate": 2.4236652860863396e-05,
|
||
|
|
"loss": 0.2202,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10131599009037018,
|
||
|
|
"step": 1760,
|
||
|
|
"valid_targets_mean": 2515.0,
|
||
|
|
"valid_targets_min": 815
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4342745861733204,
|
||
|
|
"grad_norm": 0.6962218941955951,
|
||
|
|
"learning_rate": 2.414178259033087e-05,
|
||
|
|
"loss": 0.2102,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09250883758068085,
|
||
|
|
"step": 1765,
|
||
|
|
"valid_targets_mean": 2914.4,
|
||
|
|
"valid_targets_min": 1009
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4440116845180135,
|
||
|
|
"grad_norm": 0.8444040237557682,
|
||
|
|
"learning_rate": 2.404681484945143e-05,
|
||
|
|
"loss": 0.2097,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09887482225894928,
|
||
|
|
"step": 1770,
|
||
|
|
"valid_targets_mean": 2721.4,
|
||
|
|
"valid_targets_min": 1022
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.453748782862707,
|
||
|
|
"grad_norm": 0.6780117367085235,
|
||
|
|
"learning_rate": 2.3951751873141683e-05,
|
||
|
|
"loss": 0.1976,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08050332963466644,
|
||
|
|
"step": 1775,
|
||
|
|
"valid_targets_mean": 3568.2,
|
||
|
|
"valid_targets_min": 1270
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4634858812074003,
|
||
|
|
"grad_norm": 0.8098629508625784,
|
||
|
|
"learning_rate": 2.385659589855944e-05,
|
||
|
|
"loss": 0.2127,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1392337828874588,
|
||
|
|
"step": 1780,
|
||
|
|
"valid_targets_mean": 4161.6,
|
||
|
|
"valid_targets_min": 1532
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4732229795520935,
|
||
|
|
"grad_norm": 0.6886434383200422,
|
||
|
|
"learning_rate": 2.3761349165051075e-05,
|
||
|
|
"loss": 0.1887,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10259492695331573,
|
||
|
|
"step": 1785,
|
||
|
|
"valid_targets_mean": 3790.5,
|
||
|
|
"valid_targets_min": 1148
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4829600778967866,
|
||
|
|
"grad_norm": 0.7575123339265709,
|
||
|
|
"learning_rate": 2.366601391409884e-05,
|
||
|
|
"loss": 0.2166,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10067093372344971,
|
||
|
|
"step": 1790,
|
||
|
|
"valid_targets_mean": 2686.6,
|
||
|
|
"valid_targets_min": 1520
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.49269717624148,
|
||
|
|
"grad_norm": 0.806811733165281,
|
||
|
|
"learning_rate": 2.35705923892681e-05,
|
||
|
|
"loss": 0.2002,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09538982808589935,
|
||
|
|
"step": 1795,
|
||
|
|
"valid_targets_mean": 2691.8,
|
||
|
|
"valid_targets_min": 1233
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5024342745861734,
|
||
|
|
"grad_norm": 0.7246311962378337,
|
||
|
|
"learning_rate": 2.3475086836154547e-05,
|
||
|
|
"loss": 0.2096,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0840829387307167,
|
||
|
|
"step": 1800,
|
||
|
|
"valid_targets_mean": 2930.0,
|
||
|
|
"valid_targets_min": 1572
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5121713729308666,
|
||
|
|
"grad_norm": 0.6538910297866242,
|
||
|
|
"learning_rate": 2.3379499502331347e-05,
|
||
|
|
"loss": 0.1772,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10728005319833755,
|
||
|
|
"step": 1805,
|
||
|
|
"valid_targets_mean": 3528.4,
|
||
|
|
"valid_targets_min": 1381
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5219084712755597,
|
||
|
|
"grad_norm": 0.7855769486272002,
|
||
|
|
"learning_rate": 2.328383263729624e-05,
|
||
|
|
"loss": 0.1856,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09734708815813065,
|
||
|
|
"step": 1810,
|
||
|
|
"valid_targets_mean": 2871.0,
|
||
|
|
"valid_targets_min": 1470
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5316455696202533,
|
||
|
|
"grad_norm": 0.8783046030095799,
|
||
|
|
"learning_rate": 2.318808849241861e-05,
|
||
|
|
"loss": 0.2185,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11639495193958282,
|
||
|
|
"step": 1815,
|
||
|
|
"valid_targets_mean": 2588.6,
|
||
|
|
"valid_targets_min": 1005
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5413826679649465,
|
||
|
|
"grad_norm": 0.7676279432819073,
|
||
|
|
"learning_rate": 2.3092269320886497e-05,
|
||
|
|
"loss": 0.2051,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08277629315853119,
|
||
|
|
"step": 1820,
|
||
|
|
"valid_targets_mean": 2333.9,
|
||
|
|
"valid_targets_min": 1259
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5511197663096397,
|
||
|
|
"grad_norm": 0.7720477396893514,
|
||
|
|
"learning_rate": 2.299637737765358e-05,
|
||
|
|
"loss": 0.2096,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11269618570804596,
|
||
|
|
"step": 1825,
|
||
|
|
"valid_targets_mean": 3290.0,
|
||
|
|
"valid_targets_min": 1317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5608568646543333,
|
||
|
|
"grad_norm": 0.6419534081897562,
|
||
|
|
"learning_rate": 2.290041491938612e-05,
|
||
|
|
"loss": 0.1907,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06976529955863953,
|
||
|
|
"step": 1830,
|
||
|
|
"valid_targets_mean": 2259.8,
|
||
|
|
"valid_targets_min": 324
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5705939629990264,
|
||
|
|
"grad_norm": 3.0439876407749433,
|
||
|
|
"learning_rate": 2.280438420440981e-05,
|
||
|
|
"loss": 0.2121,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1164056658744812,
|
||
|
|
"step": 1835,
|
||
|
|
"valid_targets_mean": 4026.6,
|
||
|
|
"valid_targets_min": 1391
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5803310613437196,
|
||
|
|
"grad_norm": 0.8720642252712207,
|
||
|
|
"learning_rate": 2.2708287492656675e-05,
|
||
|
|
"loss": 0.2186,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12395752966403961,
|
||
|
|
"step": 1840,
|
||
|
|
"valid_targets_mean": 2965.1,
|
||
|
|
"valid_targets_min": 875
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5900681596884128,
|
||
|
|
"grad_norm": 0.7036801924845678,
|
||
|
|
"learning_rate": 2.2612127045611867e-05,
|
||
|
|
"loss": 0.2088,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09772051870822906,
|
||
|
|
"step": 1845,
|
||
|
|
"valid_targets_mean": 2779.5,
|
||
|
|
"valid_targets_min": 953
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.599805258033106,
|
||
|
|
"grad_norm": 0.8487045108395053,
|
||
|
|
"learning_rate": 2.2515905126260455e-05,
|
||
|
|
"loss": 0.2022,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1014087051153183,
|
||
|
|
"step": 1850,
|
||
|
|
"valid_targets_mean": 2140.2,
|
||
|
|
"valid_targets_min": 546
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6095423563777995,
|
||
|
|
"grad_norm": 0.8643947332086274,
|
||
|
|
"learning_rate": 2.2419623999034146e-05,
|
||
|
|
"loss": 0.2092,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10153505206108093,
|
||
|
|
"step": 1855,
|
||
|
|
"valid_targets_mean": 2834.2,
|
||
|
|
"valid_targets_min": 1513
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6192794547224927,
|
||
|
|
"grad_norm": 0.833500456749334,
|
||
|
|
"learning_rate": 2.232328592975802e-05,
|
||
|
|
"loss": 0.1893,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08012451976537704,
|
||
|
|
"step": 1860,
|
||
|
|
"valid_targets_mean": 2420.8,
|
||
|
|
"valid_targets_min": 1175
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.629016553067186,
|
||
|
|
"grad_norm": 0.7209308141157239,
|
||
|
|
"learning_rate": 2.2226893185597204e-05,
|
||
|
|
"loss": 0.1946,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08700944483280182,
|
||
|
|
"step": 1865,
|
||
|
|
"valid_targets_mean": 2764.1,
|
||
|
|
"valid_targets_min": 1449
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6387536514118795,
|
||
|
|
"grad_norm": 0.7333785716703014,
|
||
|
|
"learning_rate": 2.213044803500351e-05,
|
||
|
|
"loss": 0.196,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10017083585262299,
|
||
|
|
"step": 1870,
|
||
|
|
"valid_targets_mean": 3329.2,
|
||
|
|
"valid_targets_min": 1173
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6484907497565726,
|
||
|
|
"grad_norm": 0.6004357753526836,
|
||
|
|
"learning_rate": 2.2033952747662052e-05,
|
||
|
|
"loss": 0.2029,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04709317535161972,
|
||
|
|
"step": 1875,
|
||
|
|
"valid_targets_mean": 2364.6,
|
||
|
|
"valid_targets_min": 1576
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6582278481012658,
|
||
|
|
"grad_norm": 0.6759321598294998,
|
||
|
|
"learning_rate": 2.193740959443782e-05,
|
||
|
|
"loss": 0.1914,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05771394073963165,
|
||
|
|
"step": 1880,
|
||
|
|
"valid_targets_mean": 2007.0,
|
||
|
|
"valid_targets_min": 942
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.667964946445959,
|
||
|
|
"grad_norm": 0.8065156038990903,
|
||
|
|
"learning_rate": 2.1840820847322274e-05,
|
||
|
|
"loss": 0.2055,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0800083726644516,
|
||
|
|
"step": 1885,
|
||
|
|
"valid_targets_mean": 2349.8,
|
||
|
|
"valid_targets_min": 1212
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6777020447906525,
|
||
|
|
"grad_norm": 0.8633743003683529,
|
||
|
|
"learning_rate": 2.1744188779379844e-05,
|
||
|
|
"loss": 0.2075,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09043803811073303,
|
||
|
|
"step": 1890,
|
||
|
|
"valid_targets_mean": 1872.9,
|
||
|
|
"valid_targets_min": 741
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6874391431353457,
|
||
|
|
"grad_norm": 0.9285052612789028,
|
||
|
|
"learning_rate": 2.1647515664694455e-05,
|
||
|
|
"loss": 0.2103,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09387189149856567,
|
||
|
|
"step": 1895,
|
||
|
|
"valid_targets_mean": 1954.0,
|
||
|
|
"valid_targets_min": 1135
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.697176241480039,
|
||
|
|
"grad_norm": 0.765084208446341,
|
||
|
|
"learning_rate": 2.155080377831599e-05,
|
||
|
|
"loss": 0.1933,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1057450920343399,
|
||
|
|
"step": 1900,
|
||
|
|
"valid_targets_mean": 2513.1,
|
||
|
|
"valid_targets_min": 947
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.706913339824732,
|
||
|
|
"grad_norm": 0.773037410109213,
|
||
|
|
"learning_rate": 2.145405539620677e-05,
|
||
|
|
"loss": 0.2161,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12816593050956726,
|
||
|
|
"step": 1905,
|
||
|
|
"valid_targets_mean": 4268.6,
|
||
|
|
"valid_targets_min": 1579
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7166504381694256,
|
||
|
|
"grad_norm": 0.7251729159914246,
|
||
|
|
"learning_rate": 2.135727279518798e-05,
|
||
|
|
"loss": 0.2129,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08925525844097137,
|
||
|
|
"step": 1910,
|
||
|
|
"valid_targets_mean": 2894.1,
|
||
|
|
"valid_targets_min": 1667
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.726387536514119,
|
||
|
|
"grad_norm": 0.8781856514746874,
|
||
|
|
"learning_rate": 2.1260458252886117e-05,
|
||
|
|
"loss": 0.2094,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0762854665517807,
|
||
|
|
"step": 1915,
|
||
|
|
"valid_targets_mean": 1900.8,
|
||
|
|
"valid_targets_min": 514
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.736124634858812,
|
||
|
|
"grad_norm": 0.7861955365536748,
|
||
|
|
"learning_rate": 2.1163614047679322e-05,
|
||
|
|
"loss": 0.2045,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12471576035022736,
|
||
|
|
"step": 1920,
|
||
|
|
"valid_targets_mean": 3699.2,
|
||
|
|
"valid_targets_min": 1566
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7458617332035056,
|
||
|
|
"grad_norm": 0.6865624618114795,
|
||
|
|
"learning_rate": 2.1066742458643853e-05,
|
||
|
|
"loss": 0.2036,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08267700672149658,
|
||
|
|
"step": 1925,
|
||
|
|
"valid_targets_mean": 3012.9,
|
||
|
|
"valid_targets_min": 936
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7555988315481987,
|
||
|
|
"grad_norm": 1.0641850208530184,
|
||
|
|
"learning_rate": 2.096984576550037e-05,
|
||
|
|
"loss": 0.2101,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13514497876167297,
|
||
|
|
"step": 1930,
|
||
|
|
"valid_targets_mean": 2025.6,
|
||
|
|
"valid_targets_min": 969
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.765335929892892,
|
||
|
|
"grad_norm": 0.7327848270033929,
|
||
|
|
"learning_rate": 2.087292624856033e-05,
|
||
|
|
"loss": 0.1981,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07251231372356415,
|
||
|
|
"step": 1935,
|
||
|
|
"valid_targets_mean": 1982.9,
|
||
|
|
"valid_targets_min": 610
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.775073028237585,
|
||
|
|
"grad_norm": 0.8591447330742223,
|
||
|
|
"learning_rate": 2.0775986188672324e-05,
|
||
|
|
"loss": 0.2074,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10901618003845215,
|
||
|
|
"step": 1940,
|
||
|
|
"valid_targets_mean": 2344.4,
|
||
|
|
"valid_targets_min": 1603
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7848101265822782,
|
||
|
|
"grad_norm": 0.6459578970389677,
|
||
|
|
"learning_rate": 2.067902786716836e-05,
|
||
|
|
"loss": 0.1962,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09305711090564728,
|
||
|
|
"step": 1945,
|
||
|
|
"valid_targets_mean": 3765.6,
|
||
|
|
"valid_targets_min": 1404
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.794547224926972,
|
||
|
|
"grad_norm": 0.7623980367866929,
|
||
|
|
"learning_rate": 2.0582053565810242e-05,
|
||
|
|
"loss": 0.2023,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1069234162569046,
|
||
|
|
"step": 1950,
|
||
|
|
"valid_targets_mean": 3184.4,
|
||
|
|
"valid_targets_min": 1089
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.804284323271665,
|
||
|
|
"grad_norm": 0.6865944868768693,
|
||
|
|
"learning_rate": 2.0485065566735795e-05,
|
||
|
|
"loss": 0.1887,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0835491418838501,
|
||
|
|
"step": 1955,
|
||
|
|
"valid_targets_mean": 2863.6,
|
||
|
|
"valid_targets_min": 1303
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.814021421616358,
|
||
|
|
"grad_norm": 0.8714331071602269,
|
||
|
|
"learning_rate": 2.038806615240523e-05,
|
||
|
|
"loss": 0.2155,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12258565425872803,
|
||
|
|
"step": 1960,
|
||
|
|
"valid_targets_mean": 2687.0,
|
||
|
|
"valid_targets_min": 820
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8237585199610518,
|
||
|
|
"grad_norm": 0.6495564993161029,
|
||
|
|
"learning_rate": 2.0291057605547393e-05,
|
||
|
|
"loss": 0.1964,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07098618894815445,
|
||
|
|
"step": 1965,
|
||
|
|
"valid_targets_mean": 3116.6,
|
||
|
|
"valid_targets_min": 1017
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.833495618305745,
|
||
|
|
"grad_norm": 1.3323923060061573,
|
||
|
|
"learning_rate": 2.0194042209106023e-05,
|
||
|
|
"loss": 0.2138,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12420941144227982,
|
||
|
|
"step": 1970,
|
||
|
|
"valid_targets_mean": 2603.6,
|
||
|
|
"valid_targets_min": 1056
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.843232716650438,
|
||
|
|
"grad_norm": 0.7575990456041161,
|
||
|
|
"learning_rate": 2.0097022246186092e-05,
|
||
|
|
"loss": 0.1954,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09626314043998718,
|
||
|
|
"step": 1975,
|
||
|
|
"valid_targets_mean": 2141.5,
|
||
|
|
"valid_targets_min": 764
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8529698149951317,
|
||
|
|
"grad_norm": 0.7947575483779744,
|
||
|
|
"learning_rate": 2e-05,
|
||
|
|
"loss": 0.2095,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09834505617618561,
|
||
|
|
"step": 1980,
|
||
|
|
"valid_targets_mean": 3005.2,
|
||
|
|
"valid_targets_min": 1381
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.862706913339825,
|
||
|
|
"grad_norm": 0.7163570110860416,
|
||
|
|
"learning_rate": 1.9902977753813918e-05,
|
||
|
|
"loss": 0.2051,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10482780635356903,
|
||
|
|
"step": 1985,
|
||
|
|
"valid_targets_mean": 3228.2,
|
||
|
|
"valid_targets_min": 903
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.872444011684518,
|
||
|
|
"grad_norm": 0.6547426313348985,
|
||
|
|
"learning_rate": 1.980595779089398e-05,
|
||
|
|
"loss": 0.2073,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08434221148490906,
|
||
|
|
"step": 1990,
|
||
|
|
"valid_targets_mean": 2802.6,
|
||
|
|
"valid_targets_min": 862
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.882181110029211,
|
||
|
|
"grad_norm": 0.8338109259355849,
|
||
|
|
"learning_rate": 1.9708942394452613e-05,
|
||
|
|
"loss": 0.2031,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09361699223518372,
|
||
|
|
"step": 1995,
|
||
|
|
"valid_targets_mean": 2393.4,
|
||
|
|
"valid_targets_min": 1246
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8919182083739043,
|
||
|
|
"grad_norm": 0.7966795271064488,
|
||
|
|
"learning_rate": 1.9611933847594775e-05,
|
||
|
|
"loss": 0.1919,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11265958845615387,
|
||
|
|
"step": 2000,
|
||
|
|
"valid_targets_mean": 2988.9,
|
||
|
|
"valid_targets_min": 1391
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.901655306718598,
|
||
|
|
"grad_norm": 0.717835201092483,
|
||
|
|
"learning_rate": 1.951493443326421e-05,
|
||
|
|
"loss": 0.1994,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09257732331752777,
|
||
|
|
"step": 2005,
|
||
|
|
"valid_targets_mean": 3231.4,
|
||
|
|
"valid_targets_min": 1164
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.911392405063291,
|
||
|
|
"grad_norm": 0.7700485323990515,
|
||
|
|
"learning_rate": 1.9417946434189768e-05,
|
||
|
|
"loss": 0.2083,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10518110543489456,
|
||
|
|
"step": 2010,
|
||
|
|
"valid_targets_mean": 3154.5,
|
||
|
|
"valid_targets_min": 565
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9211295034079843,
|
||
|
|
"grad_norm": 0.8320741111362749,
|
||
|
|
"learning_rate": 1.9320972132831643e-05,
|
||
|
|
"loss": 0.2127,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10657656937837601,
|
||
|
|
"step": 2015,
|
||
|
|
"valid_targets_mean": 2612.1,
|
||
|
|
"valid_targets_min": 1660
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.930866601752678,
|
||
|
|
"grad_norm": 0.7083370381604185,
|
||
|
|
"learning_rate": 1.9224013811327686e-05,
|
||
|
|
"loss": 0.1982,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08739273250102997,
|
||
|
|
"step": 2020,
|
||
|
|
"valid_targets_mean": 3379.9,
|
||
|
|
"valid_targets_min": 1267
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.940603700097371,
|
||
|
|
"grad_norm": 0.8258382734735102,
|
||
|
|
"learning_rate": 1.912707375143967e-05,
|
||
|
|
"loss": 0.1968,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10398921370506287,
|
||
|
|
"step": 2025,
|
||
|
|
"valid_targets_mean": 3194.2,
|
||
|
|
"valid_targets_min": 441
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.950340798442064,
|
||
|
|
"grad_norm": 1.0294293750196222,
|
||
|
|
"learning_rate": 1.9030154234499635e-05,
|
||
|
|
"loss": 0.1988,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07358142733573914,
|
||
|
|
"step": 2030,
|
||
|
|
"valid_targets_mean": 1779.0,
|
||
|
|
"valid_targets_min": 727
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.960077896786758,
|
||
|
|
"grad_norm": 0.6679677522098728,
|
||
|
|
"learning_rate": 1.8933257541356154e-05,
|
||
|
|
"loss": 0.2011,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1221453845500946,
|
||
|
|
"step": 2035,
|
||
|
|
"valid_targets_mean": 3645.1,
|
||
|
|
"valid_targets_min": 1627
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.969814995131451,
|
||
|
|
"grad_norm": 0.7182017298628506,
|
||
|
|
"learning_rate": 1.8836385952320685e-05,
|
||
|
|
"loss": 0.2216,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07548783719539642,
|
||
|
|
"step": 2040,
|
||
|
|
"valid_targets_mean": 2344.6,
|
||
|
|
"valid_targets_min": 1120
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.979552093476144,
|
||
|
|
"grad_norm": 0.7840004882557964,
|
||
|
|
"learning_rate": 1.873954174711389e-05,
|
||
|
|
"loss": 0.1897,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12413585186004639,
|
||
|
|
"step": 2045,
|
||
|
|
"valid_targets_mean": 3969.9,
|
||
|
|
"valid_targets_min": 2425
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9892891918208373,
|
||
|
|
"grad_norm": 0.7584589071211272,
|
||
|
|
"learning_rate": 1.8642727204812025e-05,
|
||
|
|
"loss": 0.1966,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10364937037229538,
|
||
|
|
"step": 2050,
|
||
|
|
"valid_targets_mean": 3282.2,
|
||
|
|
"valid_targets_min": 1595
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9990262901655305,
|
||
|
|
"grad_norm": 0.7077299563219889,
|
||
|
|
"learning_rate": 1.854594460379324e-05,
|
||
|
|
"loss": 0.1898,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09577010571956635,
|
||
|
|
"step": 2055,
|
||
|
|
"valid_targets_mean": 4010.4,
|
||
|
|
"valid_targets_min": 1128
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.0077896786757545,
|
||
|
|
"grad_norm": 0.6714162811777626,
|
||
|
|
"learning_rate": 1.8449196221684015e-05,
|
||
|
|
"loss": 0.1806,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07479497790336609,
|
||
|
|
"step": 2060,
|
||
|
|
"valid_targets_mean": 3739.5,
|
||
|
|
"valid_targets_min": 1494
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.017526777020448,
|
||
|
|
"grad_norm": 0.8151018805872865,
|
||
|
|
"learning_rate": 1.835248433530555e-05,
|
||
|
|
"loss": 0.1745,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07561913132667542,
|
||
|
|
"step": 2065,
|
||
|
|
"valid_targets_mean": 2534.4,
|
||
|
|
"valid_targets_min": 1118
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.027263875365141,
|
||
|
|
"grad_norm": 0.6788404341948994,
|
||
|
|
"learning_rate": 1.825581122062016e-05,
|
||
|
|
"loss": 0.1803,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08429465442895889,
|
||
|
|
"step": 2070,
|
||
|
|
"valid_targets_mean": 4065.4,
|
||
|
|
"valid_targets_min": 317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.0370009737098345,
|
||
|
|
"grad_norm": 0.7313335639788372,
|
||
|
|
"learning_rate": 1.815917915267773e-05,
|
||
|
|
"loss": 0.1797,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07114744186401367,
|
||
|
|
"step": 2075,
|
||
|
|
"valid_targets_mean": 3094.9,
|
||
|
|
"valid_targets_min": 1534
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.046738072054528,
|
||
|
|
"grad_norm": 0.7934180657619316,
|
||
|
|
"learning_rate": 1.8062590405562185e-05,
|
||
|
|
"loss": 0.175,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09695077687501907,
|
||
|
|
"step": 2080,
|
||
|
|
"valid_targets_mean": 3541.2,
|
||
|
|
"valid_targets_min": 1576
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.056475170399221,
|
||
|
|
"grad_norm": 0.94573730100302,
|
||
|
|
"learning_rate": 1.7966047252337958e-05,
|
||
|
|
"loss": 0.1788,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10533531755208969,
|
||
|
|
"step": 2085,
|
||
|
|
"valid_targets_mean": 2514.2,
|
||
|
|
"valid_targets_min": 956
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.066212268743914,
|
||
|
|
"grad_norm": 0.7362176811630884,
|
||
|
|
"learning_rate": 1.78695519649965e-05,
|
||
|
|
"loss": 0.1825,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07893982529640198,
|
||
|
|
"step": 2090,
|
||
|
|
"valid_targets_mean": 3586.1,
|
||
|
|
"valid_targets_min": 989
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.075949367088608,
|
||
|
|
"grad_norm": 0.7981259595246765,
|
||
|
|
"learning_rate": 1.7773106814402796e-05,
|
||
|
|
"loss": 0.1801,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08646465092897415,
|
||
|
|
"step": 2095,
|
||
|
|
"valid_targets_mean": 3420.0,
|
||
|
|
"valid_targets_min": 1093
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.085686465433301,
|
||
|
|
"grad_norm": 0.725014254374623,
|
||
|
|
"learning_rate": 1.7676714070241987e-05,
|
||
|
|
"loss": 0.1579,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07800125330686569,
|
||
|
|
"step": 2100,
|
||
|
|
"valid_targets_mean": 3832.5,
|
||
|
|
"valid_targets_min": 963
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.095423563777994,
|
||
|
|
"grad_norm": 0.7915625107935261,
|
||
|
|
"learning_rate": 1.7580376000965868e-05,
|
||
|
|
"loss": 0.1745,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09259450435638428,
|
||
|
|
"step": 2105,
|
||
|
|
"valid_targets_mean": 3586.2,
|
||
|
|
"valid_targets_min": 1209
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.105160662122687,
|
||
|
|
"grad_norm": 0.6952194628686748,
|
||
|
|
"learning_rate": 1.748409487373955e-05,
|
||
|
|
"loss": 0.1724,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09008082747459412,
|
||
|
|
"step": 2110,
|
||
|
|
"valid_targets_mean": 3660.9,
|
||
|
|
"valid_targets_min": 1594
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.114897760467381,
|
||
|
|
"grad_norm": 0.7269176114716939,
|
||
|
|
"learning_rate": 1.7387872954388136e-05,
|
||
|
|
"loss": 0.1797,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10437388718128204,
|
||
|
|
"step": 2115,
|
||
|
|
"valid_targets_mean": 4212.4,
|
||
|
|
"valid_targets_min": 745
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.124634858812074,
|
||
|
|
"grad_norm": 0.8316587832578521,
|
||
|
|
"learning_rate": 1.729171250734333e-05,
|
||
|
|
"loss": 0.1781,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07686236500740051,
|
||
|
|
"step": 2120,
|
||
|
|
"valid_targets_mean": 2780.6,
|
||
|
|
"valid_targets_min": 2031
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.134371957156767,
|
||
|
|
"grad_norm": 0.854473682482018,
|
||
|
|
"learning_rate": 1.71956157955902e-05,
|
||
|
|
"loss": 0.1902,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07336435467004776,
|
||
|
|
"step": 2125,
|
||
|
|
"valid_targets_mean": 2311.0,
|
||
|
|
"valid_targets_min": 726
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.144109055501461,
|
||
|
|
"grad_norm": 0.929705165504535,
|
||
|
|
"learning_rate": 1.7099585080613884e-05,
|
||
|
|
"loss": 0.1846,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10245804488658905,
|
||
|
|
"step": 2130,
|
||
|
|
"valid_targets_mean": 2843.9,
|
||
|
|
"valid_targets_min": 803
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.153846153846154,
|
||
|
|
"grad_norm": 0.9340385662936084,
|
||
|
|
"learning_rate": 1.7003622622346424e-05,
|
||
|
|
"loss": 0.166,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10968220233917236,
|
||
|
|
"step": 2135,
|
||
|
|
"valid_targets_mean": 2766.9,
|
||
|
|
"valid_targets_min": 1151
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.163583252190847,
|
||
|
|
"grad_norm": 0.7442177721335224,
|
||
|
|
"learning_rate": 1.6907730679113513e-05,
|
||
|
|
"loss": 0.1665,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0798567458987236,
|
||
|
|
"step": 2140,
|
||
|
|
"valid_targets_mean": 2954.2,
|
||
|
|
"valid_targets_min": 1231
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.1733203505355405,
|
||
|
|
"grad_norm": 0.8795685109074634,
|
||
|
|
"learning_rate": 1.6811911507581394e-05,
|
||
|
|
"loss": 0.1691,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07578116655349731,
|
||
|
|
"step": 2145,
|
||
|
|
"valid_targets_mean": 2428.5,
|
||
|
|
"valid_targets_min": 734
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.183057448880234,
|
||
|
|
"grad_norm": 0.6600606559095514,
|
||
|
|
"learning_rate": 1.6716167362703764e-05,
|
||
|
|
"loss": 0.1743,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07112875580787659,
|
||
|
|
"step": 2150,
|
||
|
|
"valid_targets_mean": 2903.2,
|
||
|
|
"valid_targets_min": 994
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.192794547224927,
|
||
|
|
"grad_norm": 0.8807610133387657,
|
||
|
|
"learning_rate": 1.6620500497668656e-05,
|
||
|
|
"loss": 0.1895,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06985434889793396,
|
||
|
|
"step": 2155,
|
||
|
|
"valid_targets_mean": 2146.6,
|
||
|
|
"valid_targets_min": 1199
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.2025316455696204,
|
||
|
|
"grad_norm": 0.7491123700968002,
|
||
|
|
"learning_rate": 1.6524913163845456e-05,
|
||
|
|
"loss": 0.1656,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10740687698125839,
|
||
|
|
"step": 2160,
|
||
|
|
"valid_targets_mean": 3712.4,
|
||
|
|
"valid_targets_min": 1252
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.212268743914313,
|
||
|
|
"grad_norm": 0.8696828626973055,
|
||
|
|
"learning_rate": 1.6429407610731902e-05,
|
||
|
|
"loss": 0.1892,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11889979243278503,
|
||
|
|
"step": 2165,
|
||
|
|
"valid_targets_mean": 3022.0,
|
||
|
|
"valid_targets_min": 1606
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.222005842259007,
|
||
|
|
"grad_norm": 0.9981169218888921,
|
||
|
|
"learning_rate": 1.6333986085901167e-05,
|
||
|
|
"loss": 0.1775,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08759283274412155,
|
||
|
|
"step": 2170,
|
||
|
|
"valid_targets_mean": 1711.8,
|
||
|
|
"valid_targets_min": 727
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.2317429406037,
|
||
|
|
"grad_norm": 0.793574430063338,
|
||
|
|
"learning_rate": 1.6238650834948935e-05,
|
||
|
|
"loss": 0.1723,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08245506882667542,
|
||
|
|
"step": 2175,
|
||
|
|
"valid_targets_mean": 2374.1,
|
||
|
|
"valid_targets_min": 1013
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.241480038948393,
|
||
|
|
"grad_norm": 0.851176631900123,
|
||
|
|
"learning_rate": 1.614340410144056e-05,
|
||
|
|
"loss": 0.1879,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11328478157520294,
|
||
|
|
"step": 2180,
|
||
|
|
"valid_targets_mean": 3784.9,
|
||
|
|
"valid_targets_min": 1349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.251217137293087,
|
||
|
|
"grad_norm": 0.8926459640954965,
|
||
|
|
"learning_rate": 1.604824812685832e-05,
|
||
|
|
"loss": 0.1729,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13632264733314514,
|
||
|
|
"step": 2185,
|
||
|
|
"valid_targets_mean": 3229.4,
|
||
|
|
"valid_targets_min": 1558
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.26095423563778,
|
||
|
|
"grad_norm": 0.8828024040800031,
|
||
|
|
"learning_rate": 1.5953185150548574e-05,
|
||
|
|
"loss": 0.1829,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11008234322071075,
|
||
|
|
"step": 2190,
|
||
|
|
"valid_targets_mean": 3296.6,
|
||
|
|
"valid_targets_min": 1173
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.270691333982473,
|
||
|
|
"grad_norm": 0.6219157530963444,
|
||
|
|
"learning_rate": 1.5858217409669138e-05,
|
||
|
|
"loss": 0.1753,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09567482769489288,
|
||
|
|
"step": 2195,
|
||
|
|
"valid_targets_mean": 4047.8,
|
||
|
|
"valid_targets_min": 1394
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.280428432327167,
|
||
|
|
"grad_norm": 0.5984832035464692,
|
||
|
|
"learning_rate": 1.576334713913661e-05,
|
||
|
|
"loss": 0.1718,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08540554344654083,
|
||
|
|
"step": 2200,
|
||
|
|
"valid_targets_mean": 4458.5,
|
||
|
|
"valid_targets_min": 1064
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.29016553067186,
|
||
|
|
"grad_norm": 1.0053890597595496,
|
||
|
|
"learning_rate": 1.566857657157378e-05,
|
||
|
|
"loss": 0.1777,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10111808776855469,
|
||
|
|
"step": 2205,
|
||
|
|
"valid_targets_mean": 2122.9,
|
||
|
|
"valid_targets_min": 905
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.299902629016553,
|
||
|
|
"grad_norm": 0.7592102668598462,
|
||
|
|
"learning_rate": 1.557390793725707e-05,
|
||
|
|
"loss": 0.1692,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09460680186748505,
|
||
|
|
"step": 2210,
|
||
|
|
"valid_targets_mean": 3360.1,
|
||
|
|
"valid_targets_min": 903
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.309639727361247,
|
||
|
|
"grad_norm": 0.8125751626562816,
|
||
|
|
"learning_rate": 1.5479343464064077e-05,
|
||
|
|
"loss": 0.1867,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08566434681415558,
|
||
|
|
"step": 2215,
|
||
|
|
"valid_targets_mean": 2385.2,
|
||
|
|
"valid_targets_min": 992
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.319376825705939,
|
||
|
|
"grad_norm": 0.9700126039051541,
|
||
|
|
"learning_rate": 1.5384885377421132e-05,
|
||
|
|
"loss": 0.1711,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0993308424949646,
|
||
|
|
"step": 2220,
|
||
|
|
"valid_targets_mean": 2004.8,
|
||
|
|
"valid_targets_min": 782
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.329113924050633,
|
||
|
|
"grad_norm": 0.8320811442092875,
|
||
|
|
"learning_rate": 1.5290535900250914e-05,
|
||
|
|
"loss": 0.176,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08704739809036255,
|
||
|
|
"step": 2225,
|
||
|
|
"valid_targets_mean": 2471.2,
|
||
|
|
"valid_targets_min": 706
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.3388510223953265,
|
||
|
|
"grad_norm": 0.8370133721221221,
|
||
|
|
"learning_rate": 1.5196297252920142e-05,
|
||
|
|
"loss": 0.1898,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07865718752145767,
|
||
|
|
"step": 2230,
|
||
|
|
"valid_targets_mean": 1931.5,
|
||
|
|
"valid_targets_min": 412
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.348588120740019,
|
||
|
|
"grad_norm": 0.7705056096763342,
|
||
|
|
"learning_rate": 1.5102171653187367e-05,
|
||
|
|
"loss": 0.1777,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08623531460762024,
|
||
|
|
"step": 2235,
|
||
|
|
"valid_targets_mean": 2429.9,
|
||
|
|
"valid_targets_min": 771
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.358325219084713,
|
||
|
|
"grad_norm": 0.9845962520737138,
|
||
|
|
"learning_rate": 1.5008161316150699e-05,
|
||
|
|
"loss": 0.1856,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10871316492557526,
|
||
|
|
"step": 2240,
|
||
|
|
"valid_targets_mean": 2547.9,
|
||
|
|
"valid_targets_min": 612
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.368062317429406,
|
||
|
|
"grad_norm": 0.8289568853379979,
|
||
|
|
"learning_rate": 1.4914268454195737e-05,
|
||
|
|
"loss": 0.1752,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06594522297382355,
|
||
|
|
"step": 2245,
|
||
|
|
"valid_targets_mean": 2254.0,
|
||
|
|
"valid_targets_min": 753
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.377799415774099,
|
||
|
|
"grad_norm": 0.8395243990768753,
|
||
|
|
"learning_rate": 1.4820495276943491e-05,
|
||
|
|
"loss": 0.1808,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09926756471395493,
|
||
|
|
"step": 2250,
|
||
|
|
"valid_targets_mean": 2749.0,
|
||
|
|
"valid_targets_min": 847
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.387536514118793,
|
||
|
|
"grad_norm": 0.7530052418500517,
|
||
|
|
"learning_rate": 1.4726843991198389e-05,
|
||
|
|
"loss": 0.1743,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10601849108934402,
|
||
|
|
"step": 2255,
|
||
|
|
"valid_targets_mean": 3593.2,
|
||
|
|
"valid_targets_min": 816
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.3972736124634855,
|
||
|
|
"grad_norm": 0.8654911495042591,
|
||
|
|
"learning_rate": 1.4633316800896316e-05,
|
||
|
|
"loss": 0.1727,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08109898865222931,
|
||
|
|
"step": 2260,
|
||
|
|
"valid_targets_mean": 2410.8,
|
||
|
|
"valid_targets_min": 1440
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.407010710808179,
|
||
|
|
"grad_norm": 0.738439387301167,
|
||
|
|
"learning_rate": 1.453991590705277e-05,
|
||
|
|
"loss": 0.1779,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06938911974430084,
|
||
|
|
"step": 2265,
|
||
|
|
"valid_targets_mean": 2660.4,
|
||
|
|
"valid_targets_min": 1141
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.416747809152873,
|
||
|
|
"grad_norm": 0.7377441598826128,
|
||
|
|
"learning_rate": 1.4446643507711068e-05,
|
||
|
|
"loss": 0.1667,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08617185056209564,
|
||
|
|
"step": 2270,
|
||
|
|
"valid_targets_mean": 3686.0,
|
||
|
|
"valid_targets_min": 1726
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.426484907497565,
|
||
|
|
"grad_norm": 0.8555737565964598,
|
||
|
|
"learning_rate": 1.4353501797890617e-05,
|
||
|
|
"loss": 0.1853,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08488436043262482,
|
||
|
|
"step": 2275,
|
||
|
|
"valid_targets_mean": 2369.0,
|
||
|
|
"valid_targets_min": 1055
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.436222005842259,
|
||
|
|
"grad_norm": 0.8121396018190382,
|
||
|
|
"learning_rate": 1.4260492969535218e-05,
|
||
|
|
"loss": 0.1761,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11650624871253967,
|
||
|
|
"step": 2280,
|
||
|
|
"valid_targets_mean": 2974.4,
|
||
|
|
"valid_targets_min": 1081
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.445959104186953,
|
||
|
|
"grad_norm": 0.739472764979638,
|
||
|
|
"learning_rate": 1.416761921146155e-05,
|
||
|
|
"loss": 0.1866,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08772695064544678,
|
||
|
|
"step": 2285,
|
||
|
|
"valid_targets_mean": 3347.6,
|
||
|
|
"valid_targets_min": 755
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.455696202531645,
|
||
|
|
"grad_norm": 0.94241742553367,
|
||
|
|
"learning_rate": 1.4074882709307617e-05,
|
||
|
|
"loss": 0.1825,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10868126899003983,
|
||
|
|
"step": 2290,
|
||
|
|
"valid_targets_mean": 2952.5,
|
||
|
|
"valid_targets_min": 1445
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.465433300876339,
|
||
|
|
"grad_norm": 0.8620357853018356,
|
||
|
|
"learning_rate": 1.398228564548131e-05,
|
||
|
|
"loss": 0.1776,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08930197358131409,
|
||
|
|
"step": 2295,
|
||
|
|
"valid_targets_mean": 2496.9,
|
||
|
|
"valid_targets_min": 1159
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.4751703992210325,
|
||
|
|
"grad_norm": 0.8238165602040978,
|
||
|
|
"learning_rate": 1.3889830199109064e-05,
|
||
|
|
"loss": 0.1734,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06826405972242355,
|
||
|
|
"step": 2300,
|
||
|
|
"valid_targets_mean": 2244.8,
|
||
|
|
"valid_targets_min": 1102
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.484907497565725,
|
||
|
|
"grad_norm": 0.7448771421831222,
|
||
|
|
"learning_rate": 1.3797518545984577e-05,
|
||
|
|
"loss": 0.1775,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08596190810203552,
|
||
|
|
"step": 2305,
|
||
|
|
"valid_targets_mean": 3395.1,
|
||
|
|
"valid_targets_min": 1302
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.494644595910419,
|
||
|
|
"grad_norm": 0.6459287688321765,
|
||
|
|
"learning_rate": 1.3705352858517596e-05,
|
||
|
|
"loss": 0.1676,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06672492623329163,
|
||
|
|
"step": 2310,
|
||
|
|
"valid_targets_mean": 3345.8,
|
||
|
|
"valid_targets_min": 1065
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.504381694255112,
|
||
|
|
"grad_norm": 0.937622099826767,
|
||
|
|
"learning_rate": 1.3613335305682769e-05,
|
||
|
|
"loss": 0.1875,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13400760293006897,
|
||
|
|
"step": 2315,
|
||
|
|
"valid_targets_mean": 2776.0,
|
||
|
|
"valid_targets_min": 1585
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.514118792599805,
|
||
|
|
"grad_norm": 0.7440001368095495,
|
||
|
|
"learning_rate": 1.3521468052968682e-05,
|
||
|
|
"loss": 0.1664,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08430090546607971,
|
||
|
|
"step": 2320,
|
||
|
|
"valid_targets_mean": 3821.0,
|
||
|
|
"valid_targets_min": 1560
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.523855890944499,
|
||
|
|
"grad_norm": 0.9340245915074052,
|
||
|
|
"learning_rate": 1.3429753262326793e-05,
|
||
|
|
"loss": 0.1729,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08000458776950836,
|
||
|
|
"step": 2325,
|
||
|
|
"valid_targets_mean": 2148.0,
|
||
|
|
"valid_targets_min": 676
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.5335929892891915,
|
||
|
|
"grad_norm": 0.8249789412174031,
|
||
|
|
"learning_rate": 1.3338193092120628e-05,
|
||
|
|
"loss": 0.1775,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11837929487228394,
|
||
|
|
"step": 2330,
|
||
|
|
"valid_targets_mean": 2965.6,
|
||
|
|
"valid_targets_min": 411
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.543330087633885,
|
||
|
|
"grad_norm": 0.6986712399643566,
|
||
|
|
"learning_rate": 1.3246789697074958e-05,
|
||
|
|
"loss": 0.1708,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07143793255090714,
|
||
|
|
"step": 2335,
|
||
|
|
"valid_targets_mean": 3674.0,
|
||
|
|
"valid_targets_min": 1680
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.553067185978579,
|
||
|
|
"grad_norm": 0.8287706359385277,
|
||
|
|
"learning_rate": 1.3155545228225104e-05,
|
||
|
|
"loss": 0.1806,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09657485783100128,
|
||
|
|
"step": 2340,
|
||
|
|
"valid_targets_mean": 2945.1,
|
||
|
|
"valid_targets_min": 973
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.5628042843232715,
|
||
|
|
"grad_norm": 0.6624365352547787,
|
||
|
|
"learning_rate": 1.3064461832866304e-05,
|
||
|
|
"loss": 0.1765,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08598685264587402,
|
||
|
|
"step": 2345,
|
||
|
|
"valid_targets_mean": 5114.0,
|
||
|
|
"valid_targets_min": 1194
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.572541382667965,
|
||
|
|
"grad_norm": 0.7677101034035914,
|
||
|
|
"learning_rate": 1.297354165450318e-05,
|
||
|
|
"loss": 0.1689,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08426067978143692,
|
||
|
|
"step": 2350,
|
||
|
|
"valid_targets_mean": 3179.9,
|
||
|
|
"valid_targets_min": 749
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.582278481012658,
|
||
|
|
"grad_norm": 1.1268953134679656,
|
||
|
|
"learning_rate": 1.2882786832799305e-05,
|
||
|
|
"loss": 0.1845,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08948741853237152,
|
||
|
|
"step": 2355,
|
||
|
|
"valid_targets_mean": 1726.6,
|
||
|
|
"valid_targets_min": 727
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.592015579357351,
|
||
|
|
"grad_norm": 0.7420321706926357,
|
||
|
|
"learning_rate": 1.2792199503526848e-05,
|
||
|
|
"loss": 0.1814,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06331359595060349,
|
||
|
|
"step": 2360,
|
||
|
|
"valid_targets_mean": 2387.6,
|
||
|
|
"valid_targets_min": 339
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.601752677702045,
|
||
|
|
"grad_norm": 0.8931184358236555,
|
||
|
|
"learning_rate": 1.2701781798516288e-05,
|
||
|
|
"loss": 0.1835,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10443165898323059,
|
||
|
|
"step": 2365,
|
||
|
|
"valid_targets_mean": 2841.5,
|
||
|
|
"valid_targets_min": 1609
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.611489776046738,
|
||
|
|
"grad_norm": 0.8297066622004157,
|
||
|
|
"learning_rate": 1.2611535845606273e-05,
|
||
|
|
"loss": 0.1777,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07365868240594864,
|
||
|
|
"step": 2370,
|
||
|
|
"valid_targets_mean": 1751.5,
|
||
|
|
"valid_targets_min": 944
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.621226874391431,
|
||
|
|
"grad_norm": 0.7460221119378259,
|
||
|
|
"learning_rate": 1.2521463768593545e-05,
|
||
|
|
"loss": 0.1719,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05629142001271248,
|
||
|
|
"step": 2375,
|
||
|
|
"valid_targets_mean": 2663.4,
|
||
|
|
"valid_targets_min": 1070
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.630963972736125,
|
||
|
|
"grad_norm": 0.7931377268019347,
|
||
|
|
"learning_rate": 1.2431567687182939e-05,
|
||
|
|
"loss": 0.1782,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10078290104866028,
|
||
|
|
"step": 2380,
|
||
|
|
"valid_targets_mean": 3410.9,
|
||
|
|
"valid_targets_min": 1454
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.640701071080818,
|
||
|
|
"grad_norm": 0.7596015233716497,
|
||
|
|
"learning_rate": 1.234184971693751e-05,
|
||
|
|
"loss": 0.1763,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0730675607919693,
|
||
|
|
"step": 2385,
|
||
|
|
"valid_targets_mean": 2975.4,
|
||
|
|
"valid_targets_min": 1671
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.650438169425511,
|
||
|
|
"grad_norm": 1.0400260562656607,
|
||
|
|
"learning_rate": 1.2252311969228754e-05,
|
||
|
|
"loss": 0.184,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07677740603685379,
|
||
|
|
"step": 2390,
|
||
|
|
"valid_targets_mean": 2467.1,
|
||
|
|
"valid_targets_min": 1192
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.660175267770205,
|
||
|
|
"grad_norm": 0.7959513719005028,
|
||
|
|
"learning_rate": 1.2162956551186904e-05,
|
||
|
|
"loss": 0.1777,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0791003629565239,
|
||
|
|
"step": 2395,
|
||
|
|
"valid_targets_mean": 2879.6,
|
||
|
|
"valid_targets_min": 922
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.669912366114898,
|
||
|
|
"grad_norm": 0.9105837151780105,
|
||
|
|
"learning_rate": 1.2073785565651362e-05,
|
||
|
|
"loss": 0.1732,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09069986641407013,
|
||
|
|
"step": 2400,
|
||
|
|
"valid_targets_mean": 2383.4,
|
||
|
|
"valid_targets_min": 669
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.679649464459591,
|
||
|
|
"grad_norm": 0.84640137944273,
|
||
|
|
"learning_rate": 1.1984801111121173e-05,
|
||
|
|
"loss": 0.1698,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07074621319770813,
|
||
|
|
"step": 2405,
|
||
|
|
"valid_targets_mean": 2490.0,
|
||
|
|
"valid_targets_min": 727
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.689386562804284,
|
||
|
|
"grad_norm": 0.8136174798674647,
|
||
|
|
"learning_rate": 1.1896005281705717e-05,
|
||
|
|
"loss": 0.1814,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09965146332979202,
|
||
|
|
"step": 2410,
|
||
|
|
"valid_targets_mean": 2786.1,
|
||
|
|
"valid_targets_min": 1413
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.6991236611489775,
|
||
|
|
"grad_norm": 0.8586390881863123,
|
||
|
|
"learning_rate": 1.180740016707533e-05,
|
||
|
|
"loss": 0.1842,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09565109759569168,
|
||
|
|
"step": 2415,
|
||
|
|
"valid_targets_mean": 3106.9,
|
||
|
|
"valid_targets_min": 1276
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.708860759493671,
|
||
|
|
"grad_norm": 0.9158651878377554,
|
||
|
|
"learning_rate": 1.1718987852412193e-05,
|
||
|
|
"loss": 0.1739,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09477080404758453,
|
||
|
|
"step": 2420,
|
||
|
|
"valid_targets_mean": 3704.9,
|
||
|
|
"valid_targets_min": 1847
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.718597857838364,
|
||
|
|
"grad_norm": 1.0263917100943944,
|
||
|
|
"learning_rate": 1.1630770418361255e-05,
|
||
|
|
"loss": 0.1767,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10718078911304474,
|
||
|
|
"step": 2425,
|
||
|
|
"valid_targets_mean": 2840.9,
|
||
|
|
"valid_targets_min": 361
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.728334956183057,
|
||
|
|
"grad_norm": 0.9774967088484693,
|
||
|
|
"learning_rate": 1.154274994098124e-05,
|
||
|
|
"loss": 0.1828,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11131705343723297,
|
||
|
|
"step": 2430,
|
||
|
|
"valid_targets_mean": 2695.8,
|
||
|
|
"valid_targets_min": 839
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.738072054527751,
|
||
|
|
"grad_norm": 1.0630175252221974,
|
||
|
|
"learning_rate": 1.1454928491695803e-05,
|
||
|
|
"loss": 0.1603,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08804786950349808,
|
||
|
|
"step": 2435,
|
||
|
|
"valid_targets_mean": 1911.8,
|
||
|
|
"valid_targets_min": 533
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.747809152872444,
|
||
|
|
"grad_norm": 0.7434551987653799,
|
||
|
|
"learning_rate": 1.1367308137244817e-05,
|
||
|
|
"loss": 0.1797,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06759445369243622,
|
||
|
|
"step": 2440,
|
||
|
|
"valid_targets_mean": 1998.9,
|
||
|
|
"valid_targets_min": 641
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.757546251217137,
|
||
|
|
"grad_norm": 0.7688618376697021,
|
||
|
|
"learning_rate": 1.1279890939635659e-05,
|
||
|
|
"loss": 0.1758,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09798973798751831,
|
||
|
|
"step": 2445,
|
||
|
|
"valid_targets_mean": 3507.9,
|
||
|
|
"valid_targets_min": 546
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.767283349561831,
|
||
|
|
"grad_norm": 0.9291345312132047,
|
||
|
|
"learning_rate": 1.1192678956094753e-05,
|
||
|
|
"loss": 0.1702,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0608474463224411,
|
||
|
|
"step": 2450,
|
||
|
|
"valid_targets_mean": 1835.0,
|
||
|
|
"valid_targets_min": 749
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.777020447906524,
|
||
|
|
"grad_norm": 0.924147763117448,
|
||
|
|
"learning_rate": 1.1105674239019132e-05,
|
||
|
|
"loss": 0.193,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0942969024181366,
|
||
|
|
"step": 2455,
|
||
|
|
"valid_targets_mean": 2208.0,
|
||
|
|
"valid_targets_min": 1245
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.786757546251217,
|
||
|
|
"grad_norm": 0.7370127844190928,
|
||
|
|
"learning_rate": 1.1018878835928142e-05,
|
||
|
|
"loss": 0.1864,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09319327771663666,
|
||
|
|
"step": 2460,
|
||
|
|
"valid_targets_mean": 3291.6,
|
||
|
|
"valid_targets_min": 1340
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.79649464459591,
|
||
|
|
"grad_norm": 0.7645770396885143,
|
||
|
|
"learning_rate": 1.093229478941525e-05,
|
||
|
|
"loss": 0.169,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07110965996980667,
|
||
|
|
"step": 2465,
|
||
|
|
"valid_targets_mean": 2410.2,
|
||
|
|
"valid_targets_min": 930
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.806231742940604,
|
||
|
|
"grad_norm": 0.6581970408354032,
|
||
|
|
"learning_rate": 1.0845924137099953e-05,
|
||
|
|
"loss": 0.1568,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08824768662452698,
|
||
|
|
"step": 2470,
|
||
|
|
"valid_targets_mean": 4368.1,
|
||
|
|
"valid_targets_min": 1502
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.815968841285297,
|
||
|
|
"grad_norm": 0.7513095850383397,
|
||
|
|
"learning_rate": 1.0759768911579896e-05,
|
||
|
|
"loss": 0.1896,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09617326408624649,
|
||
|
|
"step": 2475,
|
||
|
|
"valid_targets_mean": 3501.6,
|
||
|
|
"valid_targets_min": 892
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.82570593962999,
|
||
|
|
"grad_norm": 0.9663199577680193,
|
||
|
|
"learning_rate": 1.0673831140382962e-05,
|
||
|
|
"loss": 0.1786,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09308472275733948,
|
||
|
|
"step": 2480,
|
||
|
|
"valid_targets_mean": 2080.4,
|
||
|
|
"valid_targets_min": 793
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.8354430379746836,
|
||
|
|
"grad_norm": 0.7754613234125508,
|
||
|
|
"learning_rate": 1.0588112845919594e-05,
|
||
|
|
"loss": 0.1823,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0970308929681778,
|
||
|
|
"step": 2485,
|
||
|
|
"valid_targets_mean": 2594.6,
|
||
|
|
"valid_targets_min": 361
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.845180136319377,
|
||
|
|
"grad_norm": 0.925597519137906,
|
||
|
|
"learning_rate": 1.0502616045435202e-05,
|
||
|
|
"loss": 0.1756,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09768900275230408,
|
||
|
|
"step": 2490,
|
||
|
|
"valid_targets_mean": 2550.8,
|
||
|
|
"valid_targets_min": 590
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.85491723466407,
|
||
|
|
"grad_norm": 0.8492111273552944,
|
||
|
|
"learning_rate": 1.0417342750962671e-05,
|
||
|
|
"loss": 0.1774,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06129820644855499,
|
||
|
|
"step": 2495,
|
||
|
|
"valid_targets_mean": 1947.1,
|
||
|
|
"valid_targets_min": 1349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.8646543330087635,
|
||
|
|
"grad_norm": 0.7507411206504069,
|
||
|
|
"learning_rate": 1.0332294969275032e-05,
|
||
|
|
"loss": 0.1812,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06962542235851288,
|
||
|
|
"step": 2500,
|
||
|
|
"valid_targets_mean": 2455.6,
|
||
|
|
"valid_targets_min": 1232
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.874391431353457,
|
||
|
|
"grad_norm": 0.6262081583851395,
|
||
|
|
"learning_rate": 1.0247474701838222e-05,
|
||
|
|
"loss": 0.1753,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08543452620506287,
|
||
|
|
"step": 2505,
|
||
|
|
"valid_targets_mean": 5083.9,
|
||
|
|
"valid_targets_min": 1348
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.88412852969815,
|
||
|
|
"grad_norm": 0.934766709104454,
|
||
|
|
"learning_rate": 1.0162883944763999e-05,
|
||
|
|
"loss": 0.1799,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0959821417927742,
|
||
|
|
"step": 2510,
|
||
|
|
"valid_targets_mean": 2501.4,
|
||
|
|
"valid_targets_min": 1468
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.893865628042843,
|
||
|
|
"grad_norm": 0.8238816881001866,
|
||
|
|
"learning_rate": 1.0078524688762955e-05,
|
||
|
|
"loss": 0.1725,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09571737051010132,
|
||
|
|
"step": 2515,
|
||
|
|
"valid_targets_mean": 3524.8,
|
||
|
|
"valid_targets_min": 1513
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.903602726387536,
|
||
|
|
"grad_norm": 0.7956379233948672,
|
||
|
|
"learning_rate": 9.994398919097634e-06,
|
||
|
|
"loss": 0.1811,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10006579011678696,
|
||
|
|
"step": 2520,
|
||
|
|
"valid_targets_mean": 3144.8,
|
||
|
|
"valid_targets_min": 1350
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.91333982473223,
|
||
|
|
"grad_norm": 1.0031505931340245,
|
||
|
|
"learning_rate": 9.910508615535876e-06,
|
||
|
|
"loss": 0.1808,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1176973432302475,
|
||
|
|
"step": 2525,
|
||
|
|
"valid_targets_mean": 3230.6,
|
||
|
|
"valid_targets_min": 1483
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.923076923076923,
|
||
|
|
"grad_norm": 0.7961411053530816,
|
||
|
|
"learning_rate": 9.826855752304209e-06,
|
||
|
|
"loss": 0.1763,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06230532005429268,
|
||
|
|
"step": 2530,
|
||
|
|
"valid_targets_mean": 2015.2,
|
||
|
|
"valid_targets_min": 761
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.932814021421616,
|
||
|
|
"grad_norm": 0.8159616941356099,
|
||
|
|
"learning_rate": 9.743442298041348e-06,
|
||
|
|
"loss": 0.1804,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08193934708833694,
|
||
|
|
"step": 2535,
|
||
|
|
"valid_targets_mean": 2771.4,
|
||
|
|
"valid_targets_min": 968
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.94255111976631,
|
||
|
|
"grad_norm": 0.7849148331720305,
|
||
|
|
"learning_rate": 9.660270215751905e-06,
|
||
|
|
"loss": 0.1838,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09219025075435638,
|
||
|
|
"step": 2540,
|
||
|
|
"valid_targets_mean": 3240.4,
|
||
|
|
"valid_targets_min": 1143
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.952288218111003,
|
||
|
|
"grad_norm": 0.8024739199361195,
|
||
|
|
"learning_rate": 9.577341462760182e-06,
|
||
|
|
"loss": 0.1819,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1043066531419754,
|
||
|
|
"step": 2545,
|
||
|
|
"valid_targets_mean": 3759.1,
|
||
|
|
"valid_targets_min": 427
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.962025316455696,
|
||
|
|
"grad_norm": 0.7191457501116008,
|
||
|
|
"learning_rate": 9.494657990664105e-06,
|
||
|
|
"loss": 0.1795,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.061282359063625336,
|
||
|
|
"step": 2550,
|
||
|
|
"valid_targets_mean": 2609.5,
|
||
|
|
"valid_targets_min": 1201
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.97176241480039,
|
||
|
|
"grad_norm": 0.7332575836465272,
|
||
|
|
"learning_rate": 9.412221745289296e-06,
|
||
|
|
"loss": 0.1837,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08472214639186859,
|
||
|
|
"step": 2555,
|
||
|
|
"valid_targets_mean": 2834.2,
|
||
|
|
"valid_targets_min": 784
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.981499513145083,
|
||
|
|
"grad_norm": 0.9439541516349363,
|
||
|
|
"learning_rate": 9.330034666643304e-06,
|
||
|
|
"loss": 0.1748,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09852138161659241,
|
||
|
|
"step": 2560,
|
||
|
|
"valid_targets_mean": 2607.1,
|
||
|
|
"valid_targets_min": 672
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.991236611489776,
|
||
|
|
"grad_norm": 0.8536336779010893,
|
||
|
|
"learning_rate": 9.248098688869917e-06,
|
||
|
|
"loss": 0.1809,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09487920999526978,
|
||
|
|
"step": 2565,
|
||
|
|
"valid_targets_mean": 2370.9,
|
||
|
|
"valid_targets_min": 1042
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0,
|
||
|
|
"grad_norm": 1.0434807392539245,
|
||
|
|
"learning_rate": 9.166415740203644e-06,
|
||
|
|
"loss": 0.1823,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18129360675811768,
|
||
|
|
"step": 2570,
|
||
|
|
"valid_targets_mean": 3821.0,
|
||
|
|
"valid_targets_min": 992
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.009737098344694,
|
||
|
|
"grad_norm": 0.8788739617158123,
|
||
|
|
"learning_rate": 9.084987742924365e-06,
|
||
|
|
"loss": 0.1469,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07336068898439407,
|
||
|
|
"step": 2575,
|
||
|
|
"valid_targets_mean": 2166.5,
|
||
|
|
"valid_targets_min": 667
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.019474196689386,
|
||
|
|
"grad_norm": 0.7132646052444165,
|
||
|
|
"learning_rate": 9.003816613312097e-06,
|
||
|
|
"loss": 0.1381,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07973846793174744,
|
||
|
|
"step": 2580,
|
||
|
|
"valid_targets_mean": 4526.4,
|
||
|
|
"valid_targets_min": 2500
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.02921129503408,
|
||
|
|
"grad_norm": 0.8584741060701055,
|
||
|
|
"learning_rate": 8.922904261601853e-06,
|
||
|
|
"loss": 0.1565,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06728353351354599,
|
||
|
|
"step": 2585,
|
||
|
|
"valid_targets_mean": 2732.1,
|
||
|
|
"valid_targets_min": 1147
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0389483933787735,
|
||
|
|
"grad_norm": 0.7731473340495661,
|
||
|
|
"learning_rate": 8.842252591938725e-06,
|
||
|
|
"loss": 0.1567,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08202247321605682,
|
||
|
|
"step": 2590,
|
||
|
|
"valid_targets_mean": 3961.6,
|
||
|
|
"valid_targets_min": 760
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.048685491723466,
|
||
|
|
"grad_norm": 0.8694597268692199,
|
||
|
|
"learning_rate": 8.76186350233306e-06,
|
||
|
|
"loss": 0.1518,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10540521144866943,
|
||
|
|
"step": 2595,
|
||
|
|
"valid_targets_mean": 2925.5,
|
||
|
|
"valid_targets_min": 1548
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.05842259006816,
|
||
|
|
"grad_norm": 1.040810076249955,
|
||
|
|
"learning_rate": 8.681738884615805e-06,
|
||
|
|
"loss": 0.1638,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07151223719120026,
|
||
|
|
"step": 2600,
|
||
|
|
"valid_targets_mean": 1545.4,
|
||
|
|
"valid_targets_min": 816
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.068159688412853,
|
||
|
|
"grad_norm": 0.7793137982906794,
|
||
|
|
"learning_rate": 8.601880624393964e-06,
|
||
|
|
"loss": 0.1596,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07597843557596207,
|
||
|
|
"step": 2605,
|
||
|
|
"valid_targets_mean": 3288.5,
|
||
|
|
"valid_targets_min": 890
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.077896786757546,
|
||
|
|
"grad_norm": 0.8369156935610657,
|
||
|
|
"learning_rate": 8.522290601006245e-06,
|
||
|
|
"loss": 0.164,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06252319365739822,
|
||
|
|
"step": 2610,
|
||
|
|
"valid_targets_mean": 2056.4,
|
||
|
|
"valid_targets_min": 1449
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.08763388510224,
|
||
|
|
"grad_norm": 0.9475812466325354,
|
||
|
|
"learning_rate": 8.442970687478815e-06,
|
||
|
|
"loss": 0.1575,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0576450489461422,
|
||
|
|
"step": 2615,
|
||
|
|
"valid_targets_mean": 2248.6,
|
||
|
|
"valid_targets_min": 1311
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0973709834469325,
|
||
|
|
"grad_norm": 0.7916662534612631,
|
||
|
|
"learning_rate": 8.363922750481239e-06,
|
||
|
|
"loss": 0.1572,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06718336790800095,
|
||
|
|
"step": 2620,
|
||
|
|
"valid_targets_mean": 2878.2,
|
||
|
|
"valid_targets_min": 1456
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.107108081791626,
|
||
|
|
"grad_norm": 0.9983523544701802,
|
||
|
|
"learning_rate": 8.28514865028253e-06,
|
||
|
|
"loss": 0.1532,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06368249654769897,
|
||
|
|
"step": 2625,
|
||
|
|
"valid_targets_mean": 2964.4,
|
||
|
|
"valid_targets_min": 1247
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.11684518013632,
|
||
|
|
"grad_norm": 0.8540179789550341,
|
||
|
|
"learning_rate": 8.2066502407074e-06,
|
||
|
|
"loss": 0.1631,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08504103124141693,
|
||
|
|
"step": 2630,
|
||
|
|
"valid_targets_mean": 2839.8,
|
||
|
|
"valid_targets_min": 1418
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.1265822784810124,
|
||
|
|
"grad_norm": 0.7428245796076182,
|
||
|
|
"learning_rate": 8.128429369092599e-06,
|
||
|
|
"loss": 0.1534,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07919472455978394,
|
||
|
|
"step": 2635,
|
||
|
|
"valid_targets_mean": 3592.6,
|
||
|
|
"valid_targets_min": 1514
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.136319376825706,
|
||
|
|
"grad_norm": 0.9250428955359546,
|
||
|
|
"learning_rate": 8.050487876243473e-06,
|
||
|
|
"loss": 0.1597,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07924531400203705,
|
||
|
|
"step": 2640,
|
||
|
|
"valid_targets_mean": 2436.1,
|
||
|
|
"valid_targets_min": 592
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.146056475170399,
|
||
|
|
"grad_norm": 0.9951745935177713,
|
||
|
|
"learning_rate": 7.972827596390594e-06,
|
||
|
|
"loss": 0.161,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.061942871659994125,
|
||
|
|
"step": 2645,
|
||
|
|
"valid_targets_mean": 1668.6,
|
||
|
|
"valid_targets_min": 674
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.155793573515092,
|
||
|
|
"grad_norm": 0.9734815345149869,
|
||
|
|
"learning_rate": 7.89545035714667e-06,
|
||
|
|
"loss": 0.163,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09595105051994324,
|
||
|
|
"step": 2650,
|
||
|
|
"valid_targets_mean": 2707.8,
|
||
|
|
"valid_targets_min": 945
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.165530671859786,
|
||
|
|
"grad_norm": 0.788230065562802,
|
||
|
|
"learning_rate": 7.81835797946347e-06,
|
||
|
|
"loss": 0.1562,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06886913627386093,
|
||
|
|
"step": 2655,
|
||
|
|
"valid_targets_mean": 2837.9,
|
||
|
|
"valid_targets_min": 801
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.175267770204479,
|
||
|
|
"grad_norm": 0.8515098494145404,
|
||
|
|
"learning_rate": 7.741552277589e-06,
|
||
|
|
"loss": 0.1532,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06747442483901978,
|
||
|
|
"step": 2660,
|
||
|
|
"valid_targets_mean": 2596.0,
|
||
|
|
"valid_targets_min": 444
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.185004868549172,
|
||
|
|
"grad_norm": 0.7160182068063499,
|
||
|
|
"learning_rate": 7.665035059024794e-06,
|
||
|
|
"loss": 0.1555,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06444238871335983,
|
||
|
|
"step": 2665,
|
||
|
|
"valid_targets_mean": 3043.0,
|
||
|
|
"valid_targets_min": 781
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.194741966893866,
|
||
|
|
"grad_norm": 0.7213961492220646,
|
||
|
|
"learning_rate": 7.5888081244833936e-06,
|
||
|
|
"loss": 0.1467,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0880516916513443,
|
||
|
|
"step": 2670,
|
||
|
|
"valid_targets_mean": 4118.0,
|
||
|
|
"valid_targets_min": 976
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.204479065238559,
|
||
|
|
"grad_norm": 0.8900388790037114,
|
||
|
|
"learning_rate": 7.512873267845948e-06,
|
||
|
|
"loss": 0.1535,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06810447573661804,
|
||
|
|
"step": 2675,
|
||
|
|
"valid_targets_mean": 2510.4,
|
||
|
|
"valid_targets_min": 1402
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.214216163583252,
|
||
|
|
"grad_norm": 0.9916836488767273,
|
||
|
|
"learning_rate": 7.437232276120037e-06,
|
||
|
|
"loss": 0.1598,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09773782640695572,
|
||
|
|
"step": 2680,
|
||
|
|
"valid_targets_mean": 2611.5,
|
||
|
|
"valid_targets_min": 996
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.223953261927946,
|
||
|
|
"grad_norm": 0.8593083409555594,
|
||
|
|
"learning_rate": 7.361886929397568e-06,
|
||
|
|
"loss": 0.1569,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08319692313671112,
|
||
|
|
"step": 2685,
|
||
|
|
"valid_targets_mean": 2737.4,
|
||
|
|
"valid_targets_min": 986
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.233690360272639,
|
||
|
|
"grad_norm": 0.8931821415391766,
|
||
|
|
"learning_rate": 7.286839000812922e-06,
|
||
|
|
"loss": 0.1572,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0685073584318161,
|
||
|
|
"step": 2690,
|
||
|
|
"valid_targets_mean": 2461.8,
|
||
|
|
"valid_targets_min": 1149
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.243427458617332,
|
||
|
|
"grad_norm": 0.9300425765001682,
|
||
|
|
"learning_rate": 7.212090256501185e-06,
|
||
|
|
"loss": 0.1582,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08733557164669037,
|
||
|
|
"step": 2695,
|
||
|
|
"valid_targets_mean": 3048.6,
|
||
|
|
"valid_targets_min": 646
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.253164556962025,
|
||
|
|
"grad_norm": 0.8576363814070205,
|
||
|
|
"learning_rate": 7.137642455556654e-06,
|
||
|
|
"loss": 0.1507,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0786384642124176,
|
||
|
|
"step": 2700,
|
||
|
|
"valid_targets_mean": 2628.5,
|
||
|
|
"valid_targets_min": 1201
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.2629016553067185,
|
||
|
|
"grad_norm": 0.8180043273409469,
|
||
|
|
"learning_rate": 7.063497349991364e-06,
|
||
|
|
"loss": 0.1593,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.043701425194740295,
|
||
|
|
"step": 2705,
|
||
|
|
"valid_targets_mean": 1487.5,
|
||
|
|
"valid_targets_min": 328
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.272638753651412,
|
||
|
|
"grad_norm": 0.904986148543234,
|
||
|
|
"learning_rate": 6.9896566846939e-06,
|
||
|
|
"loss": 0.1556,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08402790129184723,
|
||
|
|
"step": 2710,
|
||
|
|
"valid_targets_mean": 2780.1,
|
||
|
|
"valid_targets_min": 1045
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.282375851996105,
|
||
|
|
"grad_norm": 0.8469792592809527,
|
||
|
|
"learning_rate": 6.916122197388322e-06,
|
||
|
|
"loss": 0.1641,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08842235803604126,
|
||
|
|
"step": 2715,
|
||
|
|
"valid_targets_mean": 2929.1,
|
||
|
|
"valid_targets_min": 837
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.292112950340798,
|
||
|
|
"grad_norm": 0.8764054913729791,
|
||
|
|
"learning_rate": 6.842895618593275e-06,
|
||
|
|
"loss": 0.1651,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09535115212202072,
|
||
|
|
"step": 2720,
|
||
|
|
"valid_targets_mean": 3143.9,
|
||
|
|
"valid_targets_min": 1751
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.301850048685492,
|
||
|
|
"grad_norm": 0.8068241577486652,
|
||
|
|
"learning_rate": 6.769978671581257e-06,
|
||
|
|
"loss": 0.1555,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08451789617538452,
|
||
|
|
"step": 2725,
|
||
|
|
"valid_targets_mean": 3457.8,
|
||
|
|
"valid_targets_min": 1521
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.311587147030185,
|
||
|
|
"grad_norm": 1.285539080281442,
|
||
|
|
"learning_rate": 6.697373072338065e-06,
|
||
|
|
"loss": 0.16,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08277589082717896,
|
||
|
|
"step": 2730,
|
||
|
|
"valid_targets_mean": 2291.1,
|
||
|
|
"valid_targets_min": 609
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.321324245374878,
|
||
|
|
"grad_norm": 0.7401126717811027,
|
||
|
|
"learning_rate": 6.625080529522432e-06,
|
||
|
|
"loss": 0.1569,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08714379370212555,
|
||
|
|
"step": 2735,
|
||
|
|
"valid_targets_mean": 4592.0,
|
||
|
|
"valid_targets_min": 977
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.331061343719572,
|
||
|
|
"grad_norm": 0.8569685478901633,
|
||
|
|
"learning_rate": 6.55310274442579e-06,
|
||
|
|
"loss": 0.1636,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1178111732006073,
|
||
|
|
"step": 2740,
|
||
|
|
"valid_targets_mean": 3467.9,
|
||
|
|
"valid_targets_min": 1854
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.340798442064265,
|
||
|
|
"grad_norm": 0.8759406556500658,
|
||
|
|
"learning_rate": 6.481441410932221e-06,
|
||
|
|
"loss": 0.1714,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12362238764762878,
|
||
|
|
"step": 2745,
|
||
|
|
"valid_targets_mean": 4463.8,
|
||
|
|
"valid_targets_min": 946
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.350535540408958,
|
||
|
|
"grad_norm": 0.6505264845417847,
|
||
|
|
"learning_rate": 6.410098215478655e-06,
|
||
|
|
"loss": 0.1583,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.060885533690452576,
|
||
|
|
"step": 2750,
|
||
|
|
"valid_targets_mean": 3237.9,
|
||
|
|
"valid_targets_min": 749
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.360272638753651,
|
||
|
|
"grad_norm": 0.7802849044383483,
|
||
|
|
"learning_rate": 6.339074837015115e-06,
|
||
|
|
"loss": 0.1537,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06824596226215363,
|
||
|
|
"step": 2755,
|
||
|
|
"valid_targets_mean": 3272.1,
|
||
|
|
"valid_targets_min": 1764
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.370009737098345,
|
||
|
|
"grad_norm": 0.701379782837056,
|
||
|
|
"learning_rate": 6.268372946965245e-06,
|
||
|
|
"loss": 0.1645,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07687725126743317,
|
||
|
|
"step": 2760,
|
||
|
|
"valid_targets_mean": 3820.2,
|
||
|
|
"valid_targets_min": 589
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.379746835443038,
|
||
|
|
"grad_norm": 0.7326761676590664,
|
||
|
|
"learning_rate": 6.197994209186959e-06,
|
||
|
|
"loss": 0.157,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05246296897530556,
|
||
|
|
"step": 2765,
|
||
|
|
"valid_targets_mean": 2954.2,
|
||
|
|
"valid_targets_min": 1334
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.389483933787731,
|
||
|
|
"grad_norm": 0.7773206771174859,
|
||
|
|
"learning_rate": 6.1279402799332934e-06,
|
||
|
|
"loss": 0.1522,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06810751557350159,
|
||
|
|
"step": 2770,
|
||
|
|
"valid_targets_mean": 2217.6,
|
||
|
|
"valid_targets_min": 562
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.3992210321324245,
|
||
|
|
"grad_norm": 0.8495851051586549,
|
||
|
|
"learning_rate": 6.058212807813426e-06,
|
||
|
|
"loss": 0.1668,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07084648311138153,
|
||
|
|
"step": 2775,
|
||
|
|
"valid_targets_mean": 2546.6,
|
||
|
|
"valid_targets_min": 1245
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.408958130477118,
|
||
|
|
"grad_norm": 0.9497312924955319,
|
||
|
|
"learning_rate": 5.988813433753869e-06,
|
||
|
|
"loss": 0.1567,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07291065156459808,
|
||
|
|
"step": 2780,
|
||
|
|
"valid_targets_mean": 1611.0,
|
||
|
|
"valid_targets_min": 658
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.418695228821811,
|
||
|
|
"grad_norm": 1.0621931684135297,
|
||
|
|
"learning_rate": 5.919743790959888e-06,
|
||
|
|
"loss": 0.1661,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06611327081918716,
|
||
|
|
"step": 2785,
|
||
|
|
"valid_targets_mean": 1717.2,
|
||
|
|
"valid_targets_min": 545
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.4284323271665045,
|
||
|
|
"grad_norm": 0.8152250708964739,
|
||
|
|
"learning_rate": 5.851005504877012e-06,
|
||
|
|
"loss": 0.1558,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09362748265266418,
|
||
|
|
"step": 2790,
|
||
|
|
"valid_targets_mean": 3392.8,
|
||
|
|
"valid_targets_min": 901
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.438169425511198,
|
||
|
|
"grad_norm": 0.9746220014286787,
|
||
|
|
"learning_rate": 5.782600193152819e-06,
|
||
|
|
"loss": 0.1597,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09429764747619629,
|
||
|
|
"step": 2795,
|
||
|
|
"valid_targets_mean": 2643.0,
|
||
|
|
"valid_targets_min": 981
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.447906523855891,
|
||
|
|
"grad_norm": 0.8351011417902146,
|
||
|
|
"learning_rate": 5.714529465598872e-06,
|
||
|
|
"loss": 0.1663,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09567970037460327,
|
||
|
|
"step": 2800,
|
||
|
|
"valid_targets_mean": 3979.8,
|
||
|
|
"valid_targets_min": 1707
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.457643622200584,
|
||
|
|
"grad_norm": 0.8049061085235286,
|
||
|
|
"learning_rate": 5.646794924152808e-06,
|
||
|
|
"loss": 0.1666,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09440994262695312,
|
||
|
|
"step": 2805,
|
||
|
|
"valid_targets_mean": 3631.0,
|
||
|
|
"valid_targets_min": 627
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.467380720545277,
|
||
|
|
"grad_norm": 0.7490940611455418,
|
||
|
|
"learning_rate": 5.579398162840655e-06,
|
||
|
|
"loss": 0.1544,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07636860013008118,
|
||
|
|
"step": 2810,
|
||
|
|
"valid_targets_mean": 3577.4,
|
||
|
|
"valid_targets_min": 353
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.477117818889971,
|
||
|
|
"grad_norm": 0.9000108148792098,
|
||
|
|
"learning_rate": 5.512340767739315e-06,
|
||
|
|
"loss": 0.1563,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08217882364988327,
|
||
|
|
"step": 2815,
|
||
|
|
"valid_targets_mean": 2592.8,
|
||
|
|
"valid_targets_min": 1436
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.486854917234664,
|
||
|
|
"grad_norm": 0.8305266244336463,
|
||
|
|
"learning_rate": 5.445624316939244e-06,
|
||
|
|
"loss": 0.1641,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08523901551961899,
|
||
|
|
"step": 2820,
|
||
|
|
"valid_targets_mean": 3182.4,
|
||
|
|
"valid_targets_min": 1404
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.496592015579357,
|
||
|
|
"grad_norm": 0.8454176944891794,
|
||
|
|
"learning_rate": 5.379250380507302e-06,
|
||
|
|
"loss": 0.1538,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.061266086995601654,
|
||
|
|
"step": 2825,
|
||
|
|
"valid_targets_mean": 2720.6,
|
||
|
|
"valid_targets_min": 1358
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.506329113924051,
|
||
|
|
"grad_norm": 0.859366502571961,
|
||
|
|
"learning_rate": 5.313220520449811e-06,
|
||
|
|
"loss": 0.1647,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08275482058525085,
|
||
|
|
"step": 2830,
|
||
|
|
"valid_targets_mean": 2100.1,
|
||
|
|
"valid_targets_min": 1103
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.516066212268744,
|
||
|
|
"grad_norm": 0.805421207634601,
|
||
|
|
"learning_rate": 5.247536290675821e-06,
|
||
|
|
"loss": 0.1672,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10261974483728409,
|
||
|
|
"step": 2835,
|
||
|
|
"valid_targets_mean": 4486.8,
|
||
|
|
"valid_targets_min": 1335
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.525803310613437,
|
||
|
|
"grad_norm": 0.844029206448696,
|
||
|
|
"learning_rate": 5.182199236960476e-06,
|
||
|
|
"loss": 0.1661,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10321641713380814,
|
||
|
|
"step": 2840,
|
||
|
|
"valid_targets_mean": 4219.5,
|
||
|
|
"valid_targets_min": 1153
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.535540408958131,
|
||
|
|
"grad_norm": 0.905655733799374,
|
||
|
|
"learning_rate": 5.117210896908702e-06,
|
||
|
|
"loss": 0.1655,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07245047390460968,
|
||
|
|
"step": 2845,
|
||
|
|
"valid_targets_mean": 2070.0,
|
||
|
|
"valid_targets_min": 641
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.545277507302824,
|
||
|
|
"grad_norm": 0.8213330423903124,
|
||
|
|
"learning_rate": 5.0525727999189865e-06,
|
||
|
|
"loss": 0.158,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08456528931856155,
|
||
|
|
"step": 2850,
|
||
|
|
"valid_targets_mean": 3331.4,
|
||
|
|
"valid_targets_min": 415
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.555014605647517,
|
||
|
|
"grad_norm": 0.9709341086367071,
|
||
|
|
"learning_rate": 4.988286467147416e-06,
|
||
|
|
"loss": 0.1617,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08191390335559845,
|
||
|
|
"step": 2855,
|
||
|
|
"valid_targets_mean": 1917.4,
|
||
|
|
"valid_targets_min": 956
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.5647517039922105,
|
||
|
|
"grad_norm": 1.1701719048090364,
|
||
|
|
"learning_rate": 4.924353411471834e-06,
|
||
|
|
"loss": 0.1618,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09931658208370209,
|
||
|
|
"step": 2860,
|
||
|
|
"valid_targets_mean": 2297.6,
|
||
|
|
"valid_targets_min": 1473
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.574488802336903,
|
||
|
|
"grad_norm": 0.9281303675067485,
|
||
|
|
"learning_rate": 4.860775137456275e-06,
|
||
|
|
"loss": 0.1668,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07922400534152985,
|
||
|
|
"step": 2865,
|
||
|
|
"valid_targets_mean": 2203.0,
|
||
|
|
"valid_targets_min": 361
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.584225900681597,
|
||
|
|
"grad_norm": 0.7949441894303411,
|
||
|
|
"learning_rate": 4.797553141315543e-06,
|
||
|
|
"loss": 0.1642,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08693256229162216,
|
||
|
|
"step": 2870,
|
||
|
|
"valid_targets_mean": 2908.5,
|
||
|
|
"valid_targets_min": 1384
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.5939629990262905,
|
||
|
|
"grad_norm": 0.8561484147763357,
|
||
|
|
"learning_rate": 4.734688910880001e-06,
|
||
|
|
"loss": 0.164,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06763464212417603,
|
||
|
|
"step": 2875,
|
||
|
|
"valid_targets_mean": 2343.9,
|
||
|
|
"valid_targets_min": 1005
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.603700097370983,
|
||
|
|
"grad_norm": 0.8806999977127576,
|
||
|
|
"learning_rate": 4.672183925560559e-06,
|
||
|
|
"loss": 0.1572,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07673318684101105,
|
||
|
|
"step": 2880,
|
||
|
|
"valid_targets_mean": 2955.1,
|
||
|
|
"valid_targets_min": 1311
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.613437195715677,
|
||
|
|
"grad_norm": 0.6319972257202355,
|
||
|
|
"learning_rate": 4.610039656313854e-06,
|
||
|
|
"loss": 0.1563,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06642003357410431,
|
||
|
|
"step": 2885,
|
||
|
|
"valid_targets_mean": 4784.6,
|
||
|
|
"valid_targets_min": 3072
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.62317429406037,
|
||
|
|
"grad_norm": 0.916217047093341,
|
||
|
|
"learning_rate": 4.548257565607643e-06,
|
||
|
|
"loss": 0.1541,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08463971316814423,
|
||
|
|
"step": 2890,
|
||
|
|
"valid_targets_mean": 2950.9,
|
||
|
|
"valid_targets_min": 1101
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.632911392405063,
|
||
|
|
"grad_norm": 0.8021495560370561,
|
||
|
|
"learning_rate": 4.486839107386378e-06,
|
||
|
|
"loss": 0.1533,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09788113832473755,
|
||
|
|
"step": 2895,
|
||
|
|
"valid_targets_mean": 3202.8,
|
||
|
|
"valid_targets_min": 992
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.642648490749757,
|
||
|
|
"grad_norm": 0.8588283420614421,
|
||
|
|
"learning_rate": 4.425785727036979e-06,
|
||
|
|
"loss": 0.1683,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07578335702419281,
|
||
|
|
"step": 2900,
|
||
|
|
"valid_targets_mean": 2622.9,
|
||
|
|
"valid_targets_min": 822
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.65238558909445,
|
||
|
|
"grad_norm": 0.9176614969113452,
|
||
|
|
"learning_rate": 4.365098861354862e-06,
|
||
|
|
"loss": 0.1696,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07465335726737976,
|
||
|
|
"step": 2905,
|
||
|
|
"valid_targets_mean": 2558.9,
|
||
|
|
"valid_targets_min": 1013
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.662122687439143,
|
||
|
|
"grad_norm": 0.7619193048169172,
|
||
|
|
"learning_rate": 4.304779938510073e-06,
|
||
|
|
"loss": 0.1768,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0974918007850647,
|
||
|
|
"step": 2910,
|
||
|
|
"valid_targets_mean": 4078.2,
|
||
|
|
"valid_targets_min": 1683
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.671859785783837,
|
||
|
|
"grad_norm": 0.7718442613293865,
|
||
|
|
"learning_rate": 4.244830378013689e-06,
|
||
|
|
"loss": 0.1629,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.084866963326931,
|
||
|
|
"step": 2915,
|
||
|
|
"valid_targets_mean": 2901.4,
|
||
|
|
"valid_targets_min": 1304
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.681596884128529,
|
||
|
|
"grad_norm": 0.8043628746607439,
|
||
|
|
"learning_rate": 4.185251590684458e-06,
|
||
|
|
"loss": 0.1516,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07334727048873901,
|
||
|
|
"step": 2920,
|
||
|
|
"valid_targets_mean": 2889.6,
|
||
|
|
"valid_targets_min": 592
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.691333982473223,
|
||
|
|
"grad_norm": 0.9400775178899726,
|
||
|
|
"learning_rate": 4.126044978615544e-06,
|
||
|
|
"loss": 0.1523,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06582826375961304,
|
||
|
|
"step": 2925,
|
||
|
|
"valid_targets_mean": 3248.0,
|
||
|
|
"valid_targets_min": 484
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.701071080817917,
|
||
|
|
"grad_norm": 0.73244731105116,
|
||
|
|
"learning_rate": 4.067211935141551e-06,
|
||
|
|
"loss": 0.1552,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07959744334220886,
|
||
|
|
"step": 2930,
|
||
|
|
"valid_targets_mean": 3028.8,
|
||
|
|
"valid_targets_min": 859
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.710808179162609,
|
||
|
|
"grad_norm": 0.8805977975136875,
|
||
|
|
"learning_rate": 4.008753844805737e-06,
|
||
|
|
"loss": 0.1727,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07804688066244125,
|
||
|
|
"step": 2935,
|
||
|
|
"valid_targets_mean": 2540.8,
|
||
|
|
"valid_targets_min": 877
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.720545277507303,
|
||
|
|
"grad_norm": 0.7987023815464585,
|
||
|
|
"learning_rate": 3.950672083327425e-06,
|
||
|
|
"loss": 0.1658,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0801348090171814,
|
||
|
|
"step": 2940,
|
||
|
|
"valid_targets_mean": 2479.9,
|
||
|
|
"valid_targets_min": 1414
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.730282375851996,
|
||
|
|
"grad_norm": 0.8299655079384172,
|
||
|
|
"learning_rate": 3.8929680175696315e-06,
|
||
|
|
"loss": 0.1479,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07135441899299622,
|
||
|
|
"step": 2945,
|
||
|
|
"valid_targets_mean": 2472.5,
|
||
|
|
"valid_targets_min": 684
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.740019474196689,
|
||
|
|
"grad_norm": 0.7969720132913237,
|
||
|
|
"learning_rate": 3.835643005506893e-06,
|
||
|
|
"loss": 0.1554,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08523745834827423,
|
||
|
|
"step": 2950,
|
||
|
|
"valid_targets_mean": 3318.8,
|
||
|
|
"valid_targets_min": 1012
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.749756572541383,
|
||
|
|
"grad_norm": 0.7356512604683877,
|
||
|
|
"learning_rate": 3.7786983961933234e-06,
|
||
|
|
"loss": 0.1507,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06636009365320206,
|
||
|
|
"step": 2955,
|
||
|
|
"valid_targets_mean": 3269.4,
|
||
|
|
"valid_targets_min": 930
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.759493670886076,
|
||
|
|
"grad_norm": 0.9844827711902144,
|
||
|
|
"learning_rate": 3.7221355297308483e-06,
|
||
|
|
"loss": 0.1645,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10001415759325027,
|
||
|
|
"step": 2960,
|
||
|
|
"valid_targets_mean": 2881.2,
|
||
|
|
"valid_targets_min": 1582
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.769230769230769,
|
||
|
|
"grad_norm": 1.011006569694185,
|
||
|
|
"learning_rate": 3.665955737237665e-06,
|
||
|
|
"loss": 0.176,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07234857231378555,
|
||
|
|
"step": 2965,
|
||
|
|
"valid_targets_mean": 2019.0,
|
||
|
|
"valid_targets_min": 844
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.778967867575463,
|
||
|
|
"grad_norm": 0.7013473242141751,
|
||
|
|
"learning_rate": 3.6101603408169373e-06,
|
||
|
|
"loss": 0.1526,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07861198484897614,
|
||
|
|
"step": 2970,
|
||
|
|
"valid_targets_mean": 3716.4,
|
||
|
|
"valid_targets_min": 805
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.7887049659201555,
|
||
|
|
"grad_norm": 0.7663032724768525,
|
||
|
|
"learning_rate": 3.5547506535256825e-06,
|
||
|
|
"loss": 0.1481,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07591184973716736,
|
||
|
|
"step": 2975,
|
||
|
|
"valid_targets_mean": 3134.9,
|
||
|
|
"valid_targets_min": 770
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.798442064264849,
|
||
|
|
"grad_norm": 0.9982055407518172,
|
||
|
|
"learning_rate": 3.4997279793438387e-06,
|
||
|
|
"loss": 0.16,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08362538367509842,
|
||
|
|
"step": 2980,
|
||
|
|
"valid_targets_mean": 2185.9,
|
||
|
|
"valid_targets_min": 1351
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.808179162609543,
|
||
|
|
"grad_norm": 0.8369557811189366,
|
||
|
|
"learning_rate": 3.445093613143611e-06,
|
||
|
|
"loss": 0.1663,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06555432081222534,
|
||
|
|
"step": 2985,
|
||
|
|
"valid_targets_mean": 2496.9,
|
||
|
|
"valid_targets_min": 829
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.817916260954235,
|
||
|
|
"grad_norm": 0.7826913703379452,
|
||
|
|
"learning_rate": 3.390848840658978e-06,
|
||
|
|
"loss": 0.1517,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05012332648038864,
|
||
|
|
"step": 2990,
|
||
|
|
"valid_targets_mean": 2353.4,
|
||
|
|
"valid_targets_min": 1369
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.827653359298929,
|
||
|
|
"grad_norm": 0.9167281481100252,
|
||
|
|
"learning_rate": 3.3369949384554514e-06,
|
||
|
|
"loss": 0.1547,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0563075914978981,
|
||
|
|
"step": 2995,
|
||
|
|
"valid_targets_mean": 2255.0,
|
||
|
|
"valid_targets_min": 766
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.837390457643622,
|
||
|
|
"grad_norm": 0.807847869234867,
|
||
|
|
"learning_rate": 3.2835331739000178e-06,
|
||
|
|
"loss": 0.1445,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08572809398174286,
|
||
|
|
"step": 3000,
|
||
|
|
"valid_targets_mean": 3547.1,
|
||
|
|
"valid_targets_min": 931
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.847127555988315,
|
||
|
|
"grad_norm": 0.9042350803251398,
|
||
|
|
"learning_rate": 3.2304648051313146e-06,
|
||
|
|
"loss": 0.1614,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.060835100710392,
|
||
|
|
"step": 3005,
|
||
|
|
"valid_targets_mean": 1769.5,
|
||
|
|
"valid_targets_min": 1011
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.856864654333009,
|
||
|
|
"grad_norm": 0.6067934336851712,
|
||
|
|
"learning_rate": 3.177791081030057e-06,
|
||
|
|
"loss": 0.1567,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.048806332051754,
|
||
|
|
"step": 3010,
|
||
|
|
"valid_targets_mean": 2325.2,
|
||
|
|
"valid_targets_min": 766
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.866601752677702,
|
||
|
|
"grad_norm": 0.8063814297615702,
|
||
|
|
"learning_rate": 3.125513241189575e-06,
|
||
|
|
"loss": 0.1591,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07505717128515244,
|
||
|
|
"step": 3015,
|
||
|
|
"valid_targets_mean": 3204.6,
|
||
|
|
"valid_targets_min": 879
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.876338851022395,
|
||
|
|
"grad_norm": 0.8193782903352972,
|
||
|
|
"learning_rate": 3.073632515886711e-06,
|
||
|
|
"loss": 0.1526,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07415521889925003,
|
||
|
|
"step": 3020,
|
||
|
|
"valid_targets_mean": 3054.9,
|
||
|
|
"valid_targets_min": 629
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.886075949367089,
|
||
|
|
"grad_norm": 0.9230925758861487,
|
||
|
|
"learning_rate": 3.0221501260528406e-06,
|
||
|
|
"loss": 0.1519,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07016010582447052,
|
||
|
|
"step": 3025,
|
||
|
|
"valid_targets_mean": 2294.0,
|
||
|
|
"valid_targets_min": 1459
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.895813047711782,
|
||
|
|
"grad_norm": 1.6643358275637952,
|
||
|
|
"learning_rate": 2.9710672832451303e-06,
|
||
|
|
"loss": 0.1532,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0754982978105545,
|
||
|
|
"step": 3030,
|
||
|
|
"valid_targets_mean": 1577.0,
|
||
|
|
"valid_targets_min": 585
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.905550146056475,
|
||
|
|
"grad_norm": 0.7882969996439173,
|
||
|
|
"learning_rate": 2.9203851896180422e-06,
|
||
|
|
"loss": 0.1693,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.072642020881176,
|
||
|
|
"step": 3035,
|
||
|
|
"valid_targets_mean": 2888.5,
|
||
|
|
"valid_targets_min": 1482
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.915287244401169,
|
||
|
|
"grad_norm": 0.8164072771583177,
|
||
|
|
"learning_rate": 2.8701050378950303e-06,
|
||
|
|
"loss": 0.1614,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07803675532341003,
|
||
|
|
"step": 3040,
|
||
|
|
"valid_targets_mean": 2868.0,
|
||
|
|
"valid_targets_min": 726
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.9250243427458615,
|
||
|
|
"grad_norm": 0.9813071989138026,
|
||
|
|
"learning_rate": 2.820228011340478e-06,
|
||
|
|
"loss": 0.1605,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07952482998371124,
|
||
|
|
"step": 3045,
|
||
|
|
"valid_targets_mean": 1763.4,
|
||
|
|
"valid_targets_min": 722
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.934761441090555,
|
||
|
|
"grad_norm": 0.8810999081460665,
|
||
|
|
"learning_rate": 2.7707552837318584e-06,
|
||
|
|
"loss": 0.1589,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10277026146650314,
|
||
|
|
"step": 3050,
|
||
|
|
"valid_targets_mean": 3749.4,
|
||
|
|
"valid_targets_min": 1459
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.944498539435248,
|
||
|
|
"grad_norm": 0.8466147093454865,
|
||
|
|
"learning_rate": 2.7216880193320915e-06,
|
||
|
|
"loss": 0.1628,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11054025590419769,
|
||
|
|
"step": 3055,
|
||
|
|
"valid_targets_mean": 3374.1,
|
||
|
|
"valid_targets_min": 1362
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.9542356377799415,
|
||
|
|
"grad_norm": 0.8089472605814677,
|
||
|
|
"learning_rate": 2.673027372862178e-06,
|
||
|
|
"loss": 0.1666,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07416535913944244,
|
||
|
|
"step": 3060,
|
||
|
|
"valid_targets_mean": 3063.5,
|
||
|
|
"valid_targets_min": 1618
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.963972736124635,
|
||
|
|
"grad_norm": 0.8188091275314583,
|
||
|
|
"learning_rate": 2.624774489473978e-06,
|
||
|
|
"loss": 0.1535,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07103756070137024,
|
||
|
|
"step": 3065,
|
||
|
|
"valid_targets_mean": 2943.8,
|
||
|
|
"valid_targets_min": 1430
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.973709834469328,
|
||
|
|
"grad_norm": 0.8544686193773303,
|
||
|
|
"learning_rate": 2.5769305047233004e-06,
|
||
|
|
"loss": 0.1594,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08920057862997055,
|
||
|
|
"step": 3070,
|
||
|
|
"valid_targets_mean": 3097.4,
|
||
|
|
"valid_targets_min": 1756
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.983446932814021,
|
||
|
|
"grad_norm": 0.9325324164166047,
|
||
|
|
"learning_rate": 2.5294965445431775e-06,
|
||
|
|
"loss": 0.1599,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07478688657283783,
|
||
|
|
"step": 3075,
|
||
|
|
"valid_targets_mean": 2376.0,
|
||
|
|
"valid_targets_min": 830
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.993184031158715,
|
||
|
|
"grad_norm": 0.7790233347072747,
|
||
|
|
"learning_rate": 2.4824737252173447e-06,
|
||
|
|
"loss": 0.1573,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.058365389704704285,
|
||
|
|
"step": 3080,
|
||
|
|
"valid_targets_mean": 2365.9,
|
||
|
|
"valid_targets_min": 836
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.001947419668939,
|
||
|
|
"grad_norm": 0.7062322366525114,
|
||
|
|
"learning_rate": 2.4358631533539902e-06,
|
||
|
|
"loss": 0.1501,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06631424278020859,
|
||
|
|
"step": 3085,
|
||
|
|
"valid_targets_mean": 3813.0,
|
||
|
|
"valid_targets_min": 837
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.011684518013632,
|
||
|
|
"grad_norm": 0.6129874851931072,
|
||
|
|
"learning_rate": 2.3896659258596877e-06,
|
||
|
|
"loss": 0.1446,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06790415942668915,
|
||
|
|
"step": 3090,
|
||
|
|
"valid_targets_mean": 3828.0,
|
||
|
|
"valid_targets_min": 701
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.021421616358325,
|
||
|
|
"grad_norm": 0.8268875045690981,
|
||
|
|
"learning_rate": 2.3438831299136265e-06,
|
||
|
|
"loss": 0.1506,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08013726770877838,
|
||
|
|
"step": 3095,
|
||
|
|
"valid_targets_mean": 3162.8,
|
||
|
|
"valid_targets_min": 1046
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.031158714703018,
|
||
|
|
"grad_norm": 0.9247544535256333,
|
||
|
|
"learning_rate": 2.2985158429419885e-06,
|
||
|
|
"loss": 0.15,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06007855385541916,
|
||
|
|
"step": 3100,
|
||
|
|
"valid_targets_mean": 2378.1,
|
||
|
|
"valid_targets_min": 1293
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.040895813047712,
|
||
|
|
"grad_norm": 0.7287184304341361,
|
||
|
|
"learning_rate": 2.2535651325926012e-06,
|
||
|
|
"loss": 0.1479,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09027600288391113,
|
||
|
|
"step": 3105,
|
||
|
|
"valid_targets_mean": 5085.2,
|
||
|
|
"valid_targets_min": 1460
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.050632911392405,
|
||
|
|
"grad_norm": 0.7509599811577314,
|
||
|
|
"learning_rate": 2.209032056709821e-06,
|
||
|
|
"loss": 0.1459,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07516921311616898,
|
||
|
|
"step": 3110,
|
||
|
|
"valid_targets_mean": 3181.6,
|
||
|
|
"valid_targets_min": 1125
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.060370009737098,
|
||
|
|
"grad_norm": 0.915474952725836,
|
||
|
|
"learning_rate": 2.1649176633096312e-06,
|
||
|
|
"loss": 0.1688,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10079792141914368,
|
||
|
|
"step": 3115,
|
||
|
|
"valid_targets_mean": 3642.8,
|
||
|
|
"valid_targets_min": 1414
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.070107108081792,
|
||
|
|
"grad_norm": 0.7814533531585784,
|
||
|
|
"learning_rate": 2.121222990554981e-06,
|
||
|
|
"loss": 0.1532,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.062230873852968216,
|
||
|
|
"step": 3120,
|
||
|
|
"valid_targets_mean": 2256.4,
|
||
|
|
"valid_targets_min": 722
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.079844206426485,
|
||
|
|
"grad_norm": 0.7473012000861389,
|
||
|
|
"learning_rate": 2.0779490667313485e-06,
|
||
|
|
"loss": 0.1558,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05086394399404526,
|
||
|
|
"step": 3125,
|
||
|
|
"valid_targets_mean": 2438.0,
|
||
|
|
"valid_targets_min": 1017
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.089581304771178,
|
||
|
|
"grad_norm": 0.8582424039064526,
|
||
|
|
"learning_rate": 2.0350969102225603e-06,
|
||
|
|
"loss": 0.1494,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05906301736831665,
|
||
|
|
"step": 3130,
|
||
|
|
"valid_targets_mean": 2134.6,
|
||
|
|
"valid_targets_min": 735
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.099318403115872,
|
||
|
|
"grad_norm": 0.8437783516310411,
|
||
|
|
"learning_rate": 1.9926675294868024e-06,
|
||
|
|
"loss": 0.1524,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06787307560443878,
|
||
|
|
"step": 3135,
|
||
|
|
"valid_targets_mean": 2764.5,
|
||
|
|
"valid_targets_min": 877
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.109055501460565,
|
||
|
|
"grad_norm": 0.8328816937841937,
|
||
|
|
"learning_rate": 1.9506619230328884e-06,
|
||
|
|
"loss": 0.1457,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0856187641620636,
|
||
|
|
"step": 3140,
|
||
|
|
"valid_targets_mean": 3496.8,
|
||
|
|
"valid_targets_min": 936
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.118792599805258,
|
||
|
|
"grad_norm": 0.9309557209194895,
|
||
|
|
"learning_rate": 1.9090810793967885e-06,
|
||
|
|
"loss": 0.1451,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06270939111709595,
|
||
|
|
"step": 3145,
|
||
|
|
"valid_targets_mean": 1664.8,
|
||
|
|
"valid_targets_min": 289
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.1285296981499515,
|
||
|
|
"grad_norm": 0.8748339479233537,
|
||
|
|
"learning_rate": 1.8679259771183322e-06,
|
||
|
|
"loss": 0.1555,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.054008278995752335,
|
||
|
|
"step": 3150,
|
||
|
|
"valid_targets_mean": 1955.2,
|
||
|
|
"valid_targets_min": 585
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.138266796494644,
|
||
|
|
"grad_norm": 0.8704660174157336,
|
||
|
|
"learning_rate": 1.8271975847182056e-06,
|
||
|
|
"loss": 0.1507,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0936213806271553,
|
||
|
|
"step": 3155,
|
||
|
|
"valid_targets_mean": 3036.5,
|
||
|
|
"valid_targets_min": 1241
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.148003894839338,
|
||
|
|
"grad_norm": 0.9192586335083177,
|
||
|
|
"learning_rate": 1.7868968606751425e-06,
|
||
|
|
"loss": 0.1436,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0694252997636795,
|
||
|
|
"step": 3160,
|
||
|
|
"valid_targets_mean": 2225.6,
|
||
|
|
"valid_targets_min": 762
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.1577409931840315,
|
||
|
|
"grad_norm": 0.7622929705588467,
|
||
|
|
"learning_rate": 1.7470247534033703e-06,
|
||
|
|
"loss": 0.1445,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08810773491859436,
|
||
|
|
"step": 3165,
|
||
|
|
"valid_targets_mean": 3285.2,
|
||
|
|
"valid_targets_min": 1078
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.167478091528724,
|
||
|
|
"grad_norm": 0.9722240434806685,
|
||
|
|
"learning_rate": 1.7075822012303022e-06,
|
||
|
|
"loss": 0.1661,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10509815067052841,
|
||
|
|
"step": 3170,
|
||
|
|
"valid_targets_mean": 2533.6,
|
||
|
|
"valid_targets_min": 1122
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.177215189873418,
|
||
|
|
"grad_norm": 0.8410518297648337,
|
||
|
|
"learning_rate": 1.6685701323744386e-06,
|
||
|
|
"loss": 0.1512,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05361660569906235,
|
||
|
|
"step": 3175,
|
||
|
|
"valid_targets_mean": 1968.9,
|
||
|
|
"valid_targets_min": 256
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.186952288218111,
|
||
|
|
"grad_norm": 0.7240547569836189,
|
||
|
|
"learning_rate": 1.6299894649235448e-06,
|
||
|
|
"loss": 0.1548,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07740481942892075,
|
||
|
|
"step": 3180,
|
||
|
|
"valid_targets_mean": 4169.2,
|
||
|
|
"valid_targets_min": 1535
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.196689386562804,
|
||
|
|
"grad_norm": 0.8960600592655978,
|
||
|
|
"learning_rate": 1.591841106813017e-06,
|
||
|
|
"loss": 0.1457,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05455697700381279,
|
||
|
|
"step": 3185,
|
||
|
|
"valid_targets_mean": 1794.1,
|
||
|
|
"valid_targets_min": 339
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.206426484907498,
|
||
|
|
"grad_norm": 0.9394917842854902,
|
||
|
|
"learning_rate": 1.5541259558045308e-06,
|
||
|
|
"loss": 0.1446,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08670862019062042,
|
||
|
|
"step": 3190,
|
||
|
|
"valid_targets_mean": 3084.0,
|
||
|
|
"valid_targets_min": 1261
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.21616358325219,
|
||
|
|
"grad_norm": 0.9303783534345266,
|
||
|
|
"learning_rate": 1.5168448994649266e-06,
|
||
|
|
"loss": 0.1526,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10066616535186768,
|
||
|
|
"step": 3195,
|
||
|
|
"valid_targets_mean": 3411.4,
|
||
|
|
"valid_targets_min": 590
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.225900681596884,
|
||
|
|
"grad_norm": 0.8661424560699829,
|
||
|
|
"learning_rate": 1.479998815145296e-06,
|
||
|
|
"loss": 0.1609,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06600621342658997,
|
||
|
|
"step": 3200,
|
||
|
|
"valid_targets_mean": 2468.0,
|
||
|
|
"valid_targets_min": 1121
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.235637779941578,
|
||
|
|
"grad_norm": 0.7212205370395781,
|
||
|
|
"learning_rate": 1.4435885699603524e-06,
|
||
|
|
"loss": 0.1498,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06191493943333626,
|
||
|
|
"step": 3205,
|
||
|
|
"valid_targets_mean": 3401.9,
|
||
|
|
"valid_targets_min": 1330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.24537487828627,
|
||
|
|
"grad_norm": 0.7318902055656283,
|
||
|
|
"learning_rate": 1.4076150207680185e-06,
|
||
|
|
"loss": 0.1466,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07200630009174347,
|
||
|
|
"step": 3210,
|
||
|
|
"valid_targets_mean": 2824.1,
|
||
|
|
"valid_targets_min": 1297
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.255111976630964,
|
||
|
|
"grad_norm": 0.8089814338499913,
|
||
|
|
"learning_rate": 1.3720790141492657e-06,
|
||
|
|
"loss": 0.1455,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06058313325047493,
|
||
|
|
"step": 3215,
|
||
|
|
"valid_targets_mean": 2880.1,
|
||
|
|
"valid_targets_min": 1603
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.264849074975658,
|
||
|
|
"grad_norm": 0.8956533258704256,
|
||
|
|
"learning_rate": 1.336981386388183e-06,
|
||
|
|
"loss": 0.141,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0910784900188446,
|
||
|
|
"step": 3220,
|
||
|
|
"valid_targets_mean": 2851.8,
|
||
|
|
"valid_targets_min": 973
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.27458617332035,
|
||
|
|
"grad_norm": 0.9983186521247152,
|
||
|
|
"learning_rate": 1.3023229634523027e-06,
|
||
|
|
"loss": 0.1516,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0565696582198143,
|
||
|
|
"step": 3225,
|
||
|
|
"valid_targets_mean": 2139.9,
|
||
|
|
"valid_targets_min": 749
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.284323271665044,
|
||
|
|
"grad_norm": 0.7931983927885173,
|
||
|
|
"learning_rate": 1.2681045609731756e-06,
|
||
|
|
"loss": 0.1476,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.057150743901729584,
|
||
|
|
"step": 3230,
|
||
|
|
"valid_targets_mean": 2575.1,
|
||
|
|
"valid_targets_min": 1146
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.2940603700097375,
|
||
|
|
"grad_norm": 0.8917068940461859,
|
||
|
|
"learning_rate": 1.2343269842271366e-06,
|
||
|
|
"loss": 0.143,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06814391911029816,
|
||
|
|
"step": 3235,
|
||
|
|
"valid_targets_mean": 2241.4,
|
||
|
|
"valid_targets_min": 1401
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.30379746835443,
|
||
|
|
"grad_norm": 0.9725488559567614,
|
||
|
|
"learning_rate": 1.2009910281163962e-06,
|
||
|
|
"loss": 0.1602,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09598910808563232,
|
||
|
|
"step": 3240,
|
||
|
|
"valid_targets_mean": 2813.2,
|
||
|
|
"valid_targets_min": 726
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.313534566699124,
|
||
|
|
"grad_norm": 0.8168973650270686,
|
||
|
|
"learning_rate": 1.168097477150314e-06,
|
||
|
|
"loss": 0.1447,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09560485184192657,
|
||
|
|
"step": 3245,
|
||
|
|
"valid_targets_mean": 3842.6,
|
||
|
|
"valid_targets_min": 2181
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.3232716650438165,
|
||
|
|
"grad_norm": 0.8274857149268511,
|
||
|
|
"learning_rate": 1.1356471054269447e-06,
|
||
|
|
"loss": 0.147,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08190575242042542,
|
||
|
|
"step": 3250,
|
||
|
|
"valid_targets_mean": 3535.4,
|
||
|
|
"valid_targets_min": 331
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.33300876338851,
|
||
|
|
"grad_norm": 0.7704740440870161,
|
||
|
|
"learning_rate": 1.1036406766148033e-06,
|
||
|
|
"loss": 0.1452,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07610350102186203,
|
||
|
|
"step": 3255,
|
||
|
|
"valid_targets_mean": 3209.8,
|
||
|
|
"valid_targets_min": 658
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.342745861733204,
|
||
|
|
"grad_norm": 0.8315281394348871,
|
||
|
|
"learning_rate": 1.0720789439349177e-06,
|
||
|
|
"loss": 0.1418,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06037437915802002,
|
||
|
|
"step": 3260,
|
||
|
|
"valid_targets_mean": 3152.1,
|
||
|
|
"valid_targets_min": 843
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.3524829600778965,
|
||
|
|
"grad_norm": 0.8531153468072658,
|
||
|
|
"learning_rate": 1.0409626501430847e-06,
|
||
|
|
"loss": 0.1502,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0811975970864296,
|
||
|
|
"step": 3265,
|
||
|
|
"valid_targets_mean": 2895.6,
|
||
|
|
"valid_targets_min": 772
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.36222005842259,
|
||
|
|
"grad_norm": 0.9454462934134843,
|
||
|
|
"learning_rate": 1.0102925275123954e-06,
|
||
|
|
"loss": 0.1411,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08641975373029709,
|
||
|
|
"step": 3270,
|
||
|
|
"valid_targets_mean": 2877.4,
|
||
|
|
"valid_targets_min": 1097
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.371957156767284,
|
||
|
|
"grad_norm": 0.9225800431729173,
|
||
|
|
"learning_rate": 9.800692978160064e-07,
|
||
|
|
"loss": 0.1528,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06825999915599823,
|
||
|
|
"step": 3275,
|
||
|
|
"valid_targets_mean": 1949.8,
|
||
|
|
"valid_targets_min": 550
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.381694255111976,
|
||
|
|
"grad_norm": 0.7936902856352361,
|
||
|
|
"learning_rate": 9.502936723101563e-07,
|
||
|
|
"loss": 0.1591,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08225861936807632,
|
||
|
|
"step": 3280,
|
||
|
|
"valid_targets_mean": 3964.6,
|
||
|
|
"valid_targets_min": 1418
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.39143135345667,
|
||
|
|
"grad_norm": 0.7581193556943887,
|
||
|
|
"learning_rate": 9.209663517174094e-07,
|
||
|
|
"loss": 0.1399,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0483221635222435,
|
||
|
|
"step": 3285,
|
||
|
|
"valid_targets_mean": 2562.1,
|
||
|
|
"valid_targets_min": 1065
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.401168451801363,
|
||
|
|
"grad_norm": 0.868024536322805,
|
||
|
|
"learning_rate": 8.920880262101894e-07,
|
||
|
|
"loss": 0.154,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11282053589820862,
|
||
|
|
"step": 3290,
|
||
|
|
"valid_targets_mean": 3496.5,
|
||
|
|
"valid_targets_min": 1513
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.410905550146056,
|
||
|
|
"grad_norm": 0.8888575978246757,
|
||
|
|
"learning_rate": 8.636593753945188e-07,
|
||
|
|
"loss": 0.1563,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06756648421287537,
|
||
|
|
"step": 3295,
|
||
|
|
"valid_targets_mean": 2240.8,
|
||
|
|
"valid_targets_min": 682
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.42064264849075,
|
||
|
|
"grad_norm": 0.8529665918478495,
|
||
|
|
"learning_rate": 8.356810682940386e-07,
|
||
|
|
"loss": 0.1575,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11254268884658813,
|
||
|
|
"step": 3300,
|
||
|
|
"valid_targets_mean": 4025.2,
|
||
|
|
"valid_targets_min": 730
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.430379746835443,
|
||
|
|
"grad_norm": 0.9647205012714647,
|
||
|
|
"learning_rate": 8.081537633342562e-07,
|
||
|
|
"loss": 0.1518,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08208870887756348,
|
||
|
|
"step": 3305,
|
||
|
|
"valid_targets_mean": 3272.1,
|
||
|
|
"valid_targets_min": 1331
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.440116845180136,
|
||
|
|
"grad_norm": 0.964544059777689,
|
||
|
|
"learning_rate": 7.810781083270491e-07,
|
||
|
|
"loss": 0.1487,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.060784801840782166,
|
||
|
|
"step": 3310,
|
||
|
|
"valid_targets_mean": 1944.2,
|
||
|
|
"valid_targets_min": 944
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.44985394352483,
|
||
|
|
"grad_norm": 0.9764587921394783,
|
||
|
|
"learning_rate": 7.544547404554236e-07,
|
||
|
|
"loss": 0.1542,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1391175389289856,
|
||
|
|
"step": 3315,
|
||
|
|
"valid_targets_mean": 2892.4,
|
||
|
|
"valid_targets_min": 1259
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.459591041869523,
|
||
|
|
"grad_norm": 0.9060986854710543,
|
||
|
|
"learning_rate": 7.282842862585204e-07,
|
||
|
|
"loss": 0.1519,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06699791550636292,
|
||
|
|
"step": 3320,
|
||
|
|
"valid_targets_mean": 2136.5,
|
||
|
|
"valid_targets_min": 1072
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.469328140214216,
|
||
|
|
"grad_norm": 0.7302814899655221,
|
||
|
|
"learning_rate": 7.025673616168704e-07,
|
||
|
|
"loss": 0.1493,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08695138990879059,
|
||
|
|
"step": 3325,
|
||
|
|
"valid_targets_mean": 4740.8,
|
||
|
|
"valid_targets_min": 557
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.47906523855891,
|
||
|
|
"grad_norm": 0.996714422784996,
|
||
|
|
"learning_rate": 6.773045717378912e-07,
|
||
|
|
"loss": 0.1613,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10565310716629028,
|
||
|
|
"step": 3330,
|
||
|
|
"valid_targets_mean": 2909.1,
|
||
|
|
"valid_targets_min": 1536
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.4888023369036025,
|
||
|
|
"grad_norm": 0.7573297467786815,
|
||
|
|
"learning_rate": 6.524965111416604e-07,
|
||
|
|
"loss": 0.1466,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09427705407142639,
|
||
|
|
"step": 3335,
|
||
|
|
"valid_targets_mean": 5019.6,
|
||
|
|
"valid_targets_min": 1891
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.498539435248296,
|
||
|
|
"grad_norm": 0.863330378987873,
|
||
|
|
"learning_rate": 6.281437636469135e-07,
|
||
|
|
"loss": 0.1598,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09574584662914276,
|
||
|
|
"step": 3340,
|
||
|
|
"valid_targets_mean": 3279.8,
|
||
|
|
"valid_targets_min": 550
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.508276533592989,
|
||
|
|
"grad_norm": 0.7794865658148458,
|
||
|
|
"learning_rate": 6.042469023573016e-07,
|
||
|
|
"loss": 0.1437,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06418845057487488,
|
||
|
|
"step": 3345,
|
||
|
|
"valid_targets_mean": 3014.4,
|
||
|
|
"valid_targets_min": 683
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.5180136319376825,
|
||
|
|
"grad_norm": 0.9881048043200676,
|
||
|
|
"learning_rate": 5.808064896479227e-07,
|
||
|
|
"loss": 0.1531,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08860848844051361,
|
||
|
|
"step": 3350,
|
||
|
|
"valid_targets_mean": 2982.4,
|
||
|
|
"valid_targets_min": 743
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.527750730282376,
|
||
|
|
"grad_norm": 0.9447565572161739,
|
||
|
|
"learning_rate": 5.578230771520665e-07,
|
||
|
|
"loss": 0.1486,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08905159682035446,
|
||
|
|
"step": 3355,
|
||
|
|
"valid_targets_mean": 3384.9,
|
||
|
|
"valid_targets_min": 862
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.537487828627069,
|
||
|
|
"grad_norm": 0.9284282146812189,
|
||
|
|
"learning_rate": 5.352972057482375e-07,
|
||
|
|
"loss": 0.1419,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.081339031457901,
|
||
|
|
"step": 3360,
|
||
|
|
"valid_targets_mean": 2548.0,
|
||
|
|
"valid_targets_min": 889
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.547224926971762,
|
||
|
|
"grad_norm": 0.6784066397902585,
|
||
|
|
"learning_rate": 5.132294055474307e-07,
|
||
|
|
"loss": 0.1358,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04548973590135574,
|
||
|
|
"step": 3365,
|
||
|
|
"valid_targets_mean": 3270.2,
|
||
|
|
"valid_targets_min": 761
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.556962025316456,
|
||
|
|
"grad_norm": 0.7775427498027039,
|
||
|
|
"learning_rate": 4.916201958806621e-07,
|
||
|
|
"loss": 0.1354,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07671080529689789,
|
||
|
|
"step": 3370,
|
||
|
|
"valid_targets_mean": 3316.1,
|
||
|
|
"valid_targets_min": 1243
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.566699123661149,
|
||
|
|
"grad_norm": 0.6637750998088501,
|
||
|
|
"learning_rate": 4.704700852867361e-07,
|
||
|
|
"loss": 0.1494,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06787963211536407,
|
||
|
|
"step": 3375,
|
||
|
|
"valid_targets_mean": 4622.5,
|
||
|
|
"valid_targets_min": 1390
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.576436222005842,
|
||
|
|
"grad_norm": 1.0260630365672823,
|
||
|
|
"learning_rate": 4.497795715002795e-07,
|
||
|
|
"loss": 0.1508,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06825481355190277,
|
||
|
|
"step": 3380,
|
||
|
|
"valid_targets_mean": 1388.2,
|
||
|
|
"valid_targets_min": 612
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.586173320350536,
|
||
|
|
"grad_norm": 0.9363676312851329,
|
||
|
|
"learning_rate": 4.295491414400288e-07,
|
||
|
|
"loss": 0.1468,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.055219776928424835,
|
||
|
|
"step": 3385,
|
||
|
|
"valid_targets_mean": 1722.1,
|
||
|
|
"valid_targets_min": 1112
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.595910418695229,
|
||
|
|
"grad_norm": 0.8129631138954565,
|
||
|
|
"learning_rate": 4.0977927119737695e-07,
|
||
|
|
"loss": 0.1451,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.047071993350982666,
|
||
|
|
"step": 3390,
|
||
|
|
"valid_targets_mean": 2574.9,
|
||
|
|
"valid_targets_min": 1204
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.605647517039922,
|
||
|
|
"grad_norm": 0.8516137553282794,
|
||
|
|
"learning_rate": 3.9047042602516237e-07,
|
||
|
|
"loss": 0.153,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06578527390956879,
|
||
|
|
"step": 3395,
|
||
|
|
"valid_targets_mean": 3201.1,
|
||
|
|
"valid_targets_min": 1211
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.615384615384615,
|
||
|
|
"grad_norm": 0.772193092512818,
|
||
|
|
"learning_rate": 3.716230603267268e-07,
|
||
|
|
"loss": 0.1397,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06762434542179108,
|
||
|
|
"step": 3400,
|
||
|
|
"valid_targets_mean": 3307.2,
|
||
|
|
"valid_targets_min": 1614
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.625121713729309,
|
||
|
|
"grad_norm": 0.7368043674681918,
|
||
|
|
"learning_rate": 3.532376176452146e-07,
|
||
|
|
"loss": 0.1611,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07038801908493042,
|
||
|
|
"step": 3405,
|
||
|
|
"valid_targets_mean": 3472.5,
|
||
|
|
"valid_targets_min": 1613
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.634858812074002,
|
||
|
|
"grad_norm": 0.6416048814973079,
|
||
|
|
"learning_rate": 3.353145306531347e-07,
|
||
|
|
"loss": 0.1452,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08920369297266006,
|
||
|
|
"step": 3410,
|
||
|
|
"valid_targets_mean": 5517.2,
|
||
|
|
"valid_targets_min": 1445
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.644595910418695,
|
||
|
|
"grad_norm": 0.8370135405169448,
|
||
|
|
"learning_rate": 3.17854221142182e-07,
|
||
|
|
"loss": 0.1577,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07018916308879852,
|
||
|
|
"step": 3415,
|
||
|
|
"valid_targets_mean": 2871.5,
|
||
|
|
"valid_targets_min": 507
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.6543330087633885,
|
||
|
|
"grad_norm": 0.7498795681365913,
|
||
|
|
"learning_rate": 3.008571000133187e-07,
|
||
|
|
"loss": 0.1526,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.057709187269210815,
|
||
|
|
"step": 3420,
|
||
|
|
"valid_targets_mean": 3584.9,
|
||
|
|
"valid_targets_min": 1275
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.664070107108082,
|
||
|
|
"grad_norm": 0.9347846602773136,
|
||
|
|
"learning_rate": 2.843235672670841e-07,
|
||
|
|
"loss": 0.1619,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.058967068791389465,
|
||
|
|
"step": 3425,
|
||
|
|
"valid_targets_mean": 1801.9,
|
||
|
|
"valid_targets_min": 978
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.673807205452775,
|
||
|
|
"grad_norm": 0.8516991908572321,
|
||
|
|
"learning_rate": 2.682540119942023e-07,
|
||
|
|
"loss": 0.148,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06559383124113083,
|
||
|
|
"step": 3430,
|
||
|
|
"valid_targets_mean": 2611.9,
|
||
|
|
"valid_targets_min": 896
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.6835443037974684,
|
||
|
|
"grad_norm": 0.9086726382103076,
|
||
|
|
"learning_rate": 2.5264881236640947e-07,
|
||
|
|
"loss": 0.1451,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07148931175470352,
|
||
|
|
"step": 3435,
|
||
|
|
"valid_targets_mean": 2447.8,
|
||
|
|
"valid_targets_min": 806
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.693281402142162,
|
||
|
|
"grad_norm": 0.7984514670137559,
|
||
|
|
"learning_rate": 2.375083356275676e-07,
|
||
|
|
"loss": 0.1542,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07836025953292847,
|
||
|
|
"step": 3440,
|
||
|
|
"valid_targets_mean": 3478.2,
|
||
|
|
"valid_targets_min": 822
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.703018500486855,
|
||
|
|
"grad_norm": 0.8100452297505701,
|
||
|
|
"learning_rate": 2.2283293808501138e-07,
|
||
|
|
"loss": 0.1493,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07283401489257812,
|
||
|
|
"step": 3445,
|
||
|
|
"valid_targets_mean": 3027.8,
|
||
|
|
"valid_targets_min": 1204
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.712755598831548,
|
||
|
|
"grad_norm": 0.7780103039934312,
|
||
|
|
"learning_rate": 2.086229651011684e-07,
|
||
|
|
"loss": 0.1605,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0804707333445549,
|
||
|
|
"step": 3450,
|
||
|
|
"valid_targets_mean": 3938.6,
|
||
|
|
"valid_targets_min": 1086
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.722492697176241,
|
||
|
|
"grad_norm": 0.6753064529876447,
|
||
|
|
"learning_rate": 1.9487875108542997e-07,
|
||
|
|
"loss": 0.145,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0617658756673336,
|
||
|
|
"step": 3455,
|
||
|
|
"valid_targets_mean": 3657.9,
|
||
|
|
"valid_targets_min": 729
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.732229795520935,
|
||
|
|
"grad_norm": 0.743711913805455,
|
||
|
|
"learning_rate": 1.8160061948628406e-07,
|
||
|
|
"loss": 0.1497,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08515971899032593,
|
||
|
|
"step": 3460,
|
||
|
|
"valid_targets_mean": 3728.2,
|
||
|
|
"valid_targets_min": 1911
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.741966893865628,
|
||
|
|
"grad_norm": 0.9202450054044959,
|
||
|
|
"learning_rate": 1.6878888278369255e-07,
|
||
|
|
"loss": 0.1492,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08222606778144836,
|
||
|
|
"step": 3465,
|
||
|
|
"valid_targets_mean": 2622.2,
|
||
|
|
"valid_targets_min": 1032
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.751703992210321,
|
||
|
|
"grad_norm": 0.8220976561569973,
|
||
|
|
"learning_rate": 1.5644384248176159e-07,
|
||
|
|
"loss": 0.1472,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.057911958545446396,
|
||
|
|
"step": 3470,
|
||
|
|
"valid_targets_mean": 2889.1,
|
||
|
|
"valid_targets_min": 1585
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.761441090555015,
|
||
|
|
"grad_norm": 0.8209075786638838,
|
||
|
|
"learning_rate": 1.445657891016161e-07,
|
||
|
|
"loss": 0.1511,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0630555972456932,
|
||
|
|
"step": 3475,
|
||
|
|
"valid_targets_mean": 3064.6,
|
||
|
|
"valid_targets_min": 1219
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.771178188899708,
|
||
|
|
"grad_norm": 0.7691294952327506,
|
||
|
|
"learning_rate": 1.3315500217458533e-07,
|
||
|
|
"loss": 0.1486,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0932154655456543,
|
||
|
|
"step": 3480,
|
||
|
|
"valid_targets_mean": 4289.0,
|
||
|
|
"valid_targets_min": 1582
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.780915287244401,
|
||
|
|
"grad_norm": 0.815710348733706,
|
||
|
|
"learning_rate": 1.2221175023561238e-07,
|
||
|
|
"loss": 0.1483,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07741531729698181,
|
||
|
|
"step": 3485,
|
||
|
|
"valid_targets_mean": 3159.1,
|
||
|
|
"valid_targets_min": 1049
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.790652385589095,
|
||
|
|
"grad_norm": 1.0470107205372206,
|
||
|
|
"learning_rate": 1.1173629081694392e-07,
|
||
|
|
"loss": 0.1412,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06302547454833984,
|
||
|
|
"step": 3490,
|
||
|
|
"valid_targets_mean": 2552.5,
|
||
|
|
"valid_targets_min": 631
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.800389483933788,
|
||
|
|
"grad_norm": 0.9122606227891553,
|
||
|
|
"learning_rate": 1.0172887044205936e-07,
|
||
|
|
"loss": 0.1635,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1148647665977478,
|
||
|
|
"step": 3495,
|
||
|
|
"valid_targets_mean": 3472.0,
|
||
|
|
"valid_targets_min": 1362
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.810126582278481,
|
||
|
|
"grad_norm": 0.9626562702605224,
|
||
|
|
"learning_rate": 9.21897246198733e-08,
|
||
|
|
"loss": 0.1495,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09417961537837982,
|
||
|
|
"step": 3500,
|
||
|
|
"valid_targets_mean": 1924.6,
|
||
|
|
"valid_targets_min": 622
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.8198636806231745,
|
||
|
|
"grad_norm": 0.6352208483338943,
|
||
|
|
"learning_rate": 8.311907783920437e-08,
|
||
|
|
"loss": 0.1501,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.047910600900650024,
|
||
|
|
"step": 3505,
|
||
|
|
"valid_targets_mean": 3332.6,
|
||
|
|
"valid_targets_min": 1171
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.829600778967867,
|
||
|
|
"grad_norm": 0.9136507988334205,
|
||
|
|
"learning_rate": 7.451714356347062e-08,
|
||
|
|
"loss": 0.1541,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08161959797143936,
|
||
|
|
"step": 3510,
|
||
|
|
"valid_targets_mean": 3096.0,
|
||
|
|
"valid_targets_min": 930
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.839337877312561,
|
||
|
|
"grad_norm": 1.2551160126692982,
|
||
|
|
"learning_rate": 6.638412422568463e-08,
|
||
|
|
"loss": 0.1445,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08053211867809296,
|
||
|
|
"step": 3515,
|
||
|
|
"valid_targets_mean": 2590.6,
|
||
|
|
"valid_targets_min": 1004
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.849074975657254,
|
||
|
|
"grad_norm": 1.0257326889744056,
|
||
|
|
"learning_rate": 5.8720211223677324e-08,
|
||
|
|
"loss": 0.1609,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10879699140787125,
|
||
|
|
"step": 3520,
|
||
|
|
"valid_targets_mean": 2752.0,
|
||
|
|
"valid_targets_min": 891
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.858812074001947,
|
||
|
|
"grad_norm": 0.9127077478996932,
|
||
|
|
"learning_rate": 5.1525584915605996e-08,
|
||
|
|
"loss": 0.1438,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0697256401181221,
|
||
|
|
"step": 3525,
|
||
|
|
"valid_targets_mean": 2522.8,
|
||
|
|
"valid_targets_min": 969
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.868549172346641,
|
||
|
|
"grad_norm": 0.9104831182013515,
|
||
|
|
"learning_rate": 4.480041461569773e-08,
|
||
|
|
"loss": 0.1444,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0774691104888916,
|
||
|
|
"step": 3530,
|
||
|
|
"valid_targets_mean": 2399.1,
|
||
|
|
"valid_targets_min": 1011
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.878286270691334,
|
||
|
|
"grad_norm": 0.8826495691129895,
|
||
|
|
"learning_rate": 3.854485859026591e-08,
|
||
|
|
"loss": 0.1439,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0804499015212059,
|
||
|
|
"step": 3535,
|
||
|
|
"valid_targets_mean": 2748.5,
|
||
|
|
"valid_targets_min": 909
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.888023369036027,
|
||
|
|
"grad_norm": 0.913417678940946,
|
||
|
|
"learning_rate": 3.2759064054002086e-08,
|
||
|
|
"loss": 0.1471,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05592145025730133,
|
||
|
|
"step": 3540,
|
||
|
|
"valid_targets_mean": 2153.2,
|
||
|
|
"valid_targets_min": 1253
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.897760467380721,
|
||
|
|
"grad_norm": 0.8502418717979767,
|
||
|
|
"learning_rate": 2.744316716648543e-08,
|
||
|
|
"loss": 0.1493,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07393450289964676,
|
||
|
|
"step": 3545,
|
||
|
|
"valid_targets_mean": 3059.4,
|
||
|
|
"valid_targets_min": 1200
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.907497565725414,
|
||
|
|
"grad_norm": 0.7870789251167677,
|
||
|
|
"learning_rate": 2.259729302900304e-08,
|
||
|
|
"loss": 0.1481,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06616164743900299,
|
||
|
|
"step": 3550,
|
||
|
|
"valid_targets_mean": 3009.4,
|
||
|
|
"valid_targets_min": 760
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.917234664070107,
|
||
|
|
"grad_norm": 1.1878012555440394,
|
||
|
|
"learning_rate": 1.822155568159012e-08,
|
||
|
|
"loss": 0.1542,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07589468359947205,
|
||
|
|
"step": 3555,
|
||
|
|
"valid_targets_mean": 1543.5,
|
||
|
|
"valid_targets_min": 986
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.926971762414801,
|
||
|
|
"grad_norm": 1.0618609799102148,
|
||
|
|
"learning_rate": 1.431605810034764e-08,
|
||
|
|
"loss": 0.1528,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08335718512535095,
|
||
|
|
"step": 3560,
|
||
|
|
"valid_targets_mean": 1765.9,
|
||
|
|
"valid_targets_min": 698
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.936708860759493,
|
||
|
|
"grad_norm": 0.6868086489492625,
|
||
|
|
"learning_rate": 1.0880892195028747e-08,
|
||
|
|
"loss": 0.135,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06128156557679176,
|
||
|
|
"step": 3565,
|
||
|
|
"valid_targets_mean": 4373.9,
|
||
|
|
"valid_targets_min": 2048
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.946445959104187,
|
||
|
|
"grad_norm": 0.9793651224632165,
|
||
|
|
"learning_rate": 7.916138806864927e-09,
|
||
|
|
"loss": 0.1487,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06675729155540466,
|
||
|
|
"step": 3570,
|
||
|
|
"valid_targets_mean": 3163.9,
|
||
|
|
"valid_targets_min": 1613
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.9561830574488805,
|
||
|
|
"grad_norm": 0.9938042141448199,
|
||
|
|
"learning_rate": 5.421867706671968e-09,
|
||
|
|
"loss": 0.1454,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07806721329689026,
|
||
|
|
"step": 3575,
|
||
|
|
"valid_targets_mean": 2147.5,
|
||
|
|
"valid_targets_min": 879
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.965920155793573,
|
||
|
|
"grad_norm": 0.9088693887620701,
|
||
|
|
"learning_rate": 3.3981375932001792e-09,
|
||
|
|
"loss": 0.157,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08775289356708527,
|
||
|
|
"step": 3580,
|
||
|
|
"valid_targets_mean": 3062.0,
|
||
|
|
"valid_targets_min": 1572
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.975657254138267,
|
||
|
|
"grad_norm": 0.7280671741052036,
|
||
|
|
"learning_rate": 1.8449960917599207e-09,
|
||
|
|
"loss": 0.1479,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06314358115196228,
|
||
|
|
"step": 3585,
|
||
|
|
"valid_targets_mean": 3273.1,
|
||
|
|
"valid_targets_min": 816
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.98539435248296,
|
||
|
|
"grad_norm": 0.9776504526888148,
|
||
|
|
"learning_rate": 7.624797530958461e-10,
|
||
|
|
"loss": 0.1611,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06692498922348022,
|
||
|
|
"step": 3590,
|
||
|
|
"valid_targets_mean": 2156.6,
|
||
|
|
"valid_targets_min": 844
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.995131450827653,
|
||
|
|
"grad_norm": 0.8885675545896405,
|
||
|
|
"learning_rate": 1.5061405252980721e-10,
|
||
|
|
"loss": 0.1441,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08266542851924896,
|
||
|
|
"step": 3595,
|
||
|
|
"valid_targets_mean": 2982.6,
|
||
|
|
"valid_targets_min": 1288
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 7.0,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11300152540206909,
|
||
|
|
"step": 3598,
|
||
|
|
"total_flos": 8.035723720943206e+17,
|
||
|
|
"train_loss": 0.23149373163243941,
|
||
|
|
"train_runtime": 29019.8306,
|
||
|
|
"train_samples_per_second": 1.981,
|
||
|
|
"train_steps_per_second": 0.124,
|
||
|
|
"valid_targets_mean": 2509.4,
|
||
|
|
"valid_targets_min": 1446
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 5,
|
||
|
|
"max_steps": 3598,
|
||
|
|
"num_input_tokens_seen": 0,
|
||
|
|
"num_train_epochs": 7,
|
||
|
|
"save_steps": 600,
|
||
|
|
"stateful_callbacks": {
|
||
|
|
"TrainerControl": {
|
||
|
|
"args": {
|
||
|
|
"should_epoch_stop": false,
|
||
|
|
"should_evaluate": false,
|
||
|
|
"should_log": false,
|
||
|
|
"should_save": true,
|
||
|
|
"should_training_stop": true
|
||
|
|
},
|
||
|
|
"attributes": {}
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"total_flos": 8.035723720943206e+17,
|
||
|
|
"train_batch_size": 1,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|