Model: laion/Kimi-K2T-neulab-agenttuning-kg-sandboxes-maxeps-32k Source: Original Platform
7715 lines
214 KiB
JSON
7715 lines
214 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 7.0,
|
|
"eval_steps": 500,
|
|
"global_step": 3486,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.010050251256281407,
|
|
"grad_norm": 16.704252975602397,
|
|
"learning_rate": 4.5845272206303726e-07,
|
|
"loss": 1.1396,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.5883040428161621,
|
|
"step": 5,
|
|
"valid_targets_mean": 2139.4,
|
|
"valid_targets_min": 1654
|
|
},
|
|
{
|
|
"epoch": 0.020100502512562814,
|
|
"grad_norm": 15.581264896681018,
|
|
"learning_rate": 1.031518624641834e-06,
|
|
"loss": 1.1295,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.5863569378852844,
|
|
"step": 10,
|
|
"valid_targets_mean": 2168.6,
|
|
"valid_targets_min": 1609
|
|
},
|
|
{
|
|
"epoch": 0.03015075376884422,
|
|
"grad_norm": 14.579418003435261,
|
|
"learning_rate": 1.6045845272206306e-06,
|
|
"loss": 1.1057,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.5504764914512634,
|
|
"step": 15,
|
|
"valid_targets_mean": 1956.4,
|
|
"valid_targets_min": 1575
|
|
},
|
|
{
|
|
"epoch": 0.04020100502512563,
|
|
"grad_norm": 9.815946485122087,
|
|
"learning_rate": 2.177650429799427e-06,
|
|
"loss": 1.0608,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.5473997592926025,
|
|
"step": 20,
|
|
"valid_targets_mean": 2086.6,
|
|
"valid_targets_min": 1698
|
|
},
|
|
{
|
|
"epoch": 0.05025125628140704,
|
|
"grad_norm": 7.568785171836882,
|
|
"learning_rate": 2.7507163323782236e-06,
|
|
"loss": 0.9959,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.5016837120056152,
|
|
"step": 25,
|
|
"valid_targets_mean": 2053.2,
|
|
"valid_targets_min": 1400
|
|
},
|
|
{
|
|
"epoch": 0.06030150753768844,
|
|
"grad_norm": 6.272210042418966,
|
|
"learning_rate": 3.32378223495702e-06,
|
|
"loss": 0.9317,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4753239154815674,
|
|
"step": 30,
|
|
"valid_targets_mean": 2127.5,
|
|
"valid_targets_min": 1688
|
|
},
|
|
{
|
|
"epoch": 0.07035175879396985,
|
|
"grad_norm": 5.025209680577307,
|
|
"learning_rate": 3.896848137535817e-06,
|
|
"loss": 0.8666,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.36996227502822876,
|
|
"step": 35,
|
|
"valid_targets_mean": 1771.9,
|
|
"valid_targets_min": 1593
|
|
},
|
|
{
|
|
"epoch": 0.08040201005025126,
|
|
"grad_norm": 2.8007336600814137,
|
|
"learning_rate": 4.469914040114613e-06,
|
|
"loss": 0.7738,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3509371876716614,
|
|
"step": 40,
|
|
"valid_targets_mean": 1941.4,
|
|
"valid_targets_min": 1736
|
|
},
|
|
{
|
|
"epoch": 0.09045226130653267,
|
|
"grad_norm": 2.232586006172974,
|
|
"learning_rate": 5.04297994269341e-06,
|
|
"loss": 0.7255,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.33917897939682007,
|
|
"step": 45,
|
|
"valid_targets_mean": 1890.6,
|
|
"valid_targets_min": 1633
|
|
},
|
|
{
|
|
"epoch": 0.10050251256281408,
|
|
"grad_norm": 1.8024357428720437,
|
|
"learning_rate": 5.616045845272207e-06,
|
|
"loss": 0.6825,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2903943657875061,
|
|
"step": 50,
|
|
"valid_targets_mean": 1639.6,
|
|
"valid_targets_min": 411
|
|
},
|
|
{
|
|
"epoch": 0.11055276381909548,
|
|
"grad_norm": 1.5284057095330459,
|
|
"learning_rate": 6.189111747851003e-06,
|
|
"loss": 0.6262,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.33878976106643677,
|
|
"step": 55,
|
|
"valid_targets_mean": 2221.4,
|
|
"valid_targets_min": 1405
|
|
},
|
|
{
|
|
"epoch": 0.12060301507537688,
|
|
"grad_norm": 1.5165979795493318,
|
|
"learning_rate": 6.7621776504298e-06,
|
|
"loss": 0.6073,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3074074387550354,
|
|
"step": 60,
|
|
"valid_targets_mean": 1915.1,
|
|
"valid_targets_min": 1516
|
|
},
|
|
{
|
|
"epoch": 0.1306532663316583,
|
|
"grad_norm": 1.2263977667239911,
|
|
"learning_rate": 7.335243553008596e-06,
|
|
"loss": 0.5642,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3101275563240051,
|
|
"step": 65,
|
|
"valid_targets_mean": 1972.0,
|
|
"valid_targets_min": 1585
|
|
},
|
|
{
|
|
"epoch": 0.1407035175879397,
|
|
"grad_norm": 1.1713356188904516,
|
|
"learning_rate": 7.908309455587392e-06,
|
|
"loss": 0.5129,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27801650762557983,
|
|
"step": 70,
|
|
"valid_targets_mean": 1801.5,
|
|
"valid_targets_min": 1218
|
|
},
|
|
{
|
|
"epoch": 0.1507537688442211,
|
|
"grad_norm": 1.073756116058224,
|
|
"learning_rate": 8.481375358166189e-06,
|
|
"loss": 0.5022,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3032459616661072,
|
|
"step": 75,
|
|
"valid_targets_mean": 2217.1,
|
|
"valid_targets_min": 1776
|
|
},
|
|
{
|
|
"epoch": 0.16080402010050251,
|
|
"grad_norm": 1.0030346347847157,
|
|
"learning_rate": 9.054441260744986e-06,
|
|
"loss": 0.477,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24391627311706543,
|
|
"step": 80,
|
|
"valid_targets_mean": 2076.2,
|
|
"valid_targets_min": 1873
|
|
},
|
|
{
|
|
"epoch": 0.1708542713567839,
|
|
"grad_norm": 0.9081795028326867,
|
|
"learning_rate": 9.627507163323783e-06,
|
|
"loss": 0.4422,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22889365255832672,
|
|
"step": 85,
|
|
"valid_targets_mean": 2021.4,
|
|
"valid_targets_min": 1727
|
|
},
|
|
{
|
|
"epoch": 0.18090452261306533,
|
|
"grad_norm": 0.930947081455514,
|
|
"learning_rate": 1.020057306590258e-05,
|
|
"loss": 0.4355,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2052917629480362,
|
|
"step": 90,
|
|
"valid_targets_mean": 1988.0,
|
|
"valid_targets_min": 1614
|
|
},
|
|
{
|
|
"epoch": 0.19095477386934673,
|
|
"grad_norm": 0.8443825977323414,
|
|
"learning_rate": 1.0773638968481377e-05,
|
|
"loss": 0.4268,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19906312227249146,
|
|
"step": 95,
|
|
"valid_targets_mean": 1859.1,
|
|
"valid_targets_min": 1459
|
|
},
|
|
{
|
|
"epoch": 0.20100502512562815,
|
|
"grad_norm": 0.904418344661789,
|
|
"learning_rate": 1.1346704871060172e-05,
|
|
"loss": 0.4302,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18945840001106262,
|
|
"step": 100,
|
|
"valid_targets_mean": 1878.1,
|
|
"valid_targets_min": 1633
|
|
},
|
|
{
|
|
"epoch": 0.21105527638190955,
|
|
"grad_norm": 0.8484526883139902,
|
|
"learning_rate": 1.191977077363897e-05,
|
|
"loss": 0.4201,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2142827808856964,
|
|
"step": 105,
|
|
"valid_targets_mean": 1948.2,
|
|
"valid_targets_min": 1452
|
|
},
|
|
{
|
|
"epoch": 0.22110552763819097,
|
|
"grad_norm": 0.9080360759652468,
|
|
"learning_rate": 1.2492836676217766e-05,
|
|
"loss": 0.4159,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1805184930562973,
|
|
"step": 110,
|
|
"valid_targets_mean": 1980.8,
|
|
"valid_targets_min": 1583
|
|
},
|
|
{
|
|
"epoch": 0.23115577889447236,
|
|
"grad_norm": 0.8847861818146405,
|
|
"learning_rate": 1.3065902578796563e-05,
|
|
"loss": 0.385,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18697614967823029,
|
|
"step": 115,
|
|
"valid_targets_mean": 1988.5,
|
|
"valid_targets_min": 1656
|
|
},
|
|
{
|
|
"epoch": 0.24120603015075376,
|
|
"grad_norm": 0.903716010241867,
|
|
"learning_rate": 1.3638968481375358e-05,
|
|
"loss": 0.3821,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1984958052635193,
|
|
"step": 120,
|
|
"valid_targets_mean": 2044.6,
|
|
"valid_targets_min": 1667
|
|
},
|
|
{
|
|
"epoch": 0.25125628140703515,
|
|
"grad_norm": 0.9821351482104137,
|
|
"learning_rate": 1.4212034383954157e-05,
|
|
"loss": 0.3799,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20172768831253052,
|
|
"step": 125,
|
|
"valid_targets_mean": 2020.9,
|
|
"valid_targets_min": 1522
|
|
},
|
|
{
|
|
"epoch": 0.2613065326633166,
|
|
"grad_norm": 0.9589835135885825,
|
|
"learning_rate": 1.4785100286532952e-05,
|
|
"loss": 0.3793,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1896677166223526,
|
|
"step": 130,
|
|
"valid_targets_mean": 2033.2,
|
|
"valid_targets_min": 1753
|
|
},
|
|
{
|
|
"epoch": 0.271356783919598,
|
|
"grad_norm": 0.9255959959442394,
|
|
"learning_rate": 1.535816618911175e-05,
|
|
"loss": 0.3915,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18564650416374207,
|
|
"step": 135,
|
|
"valid_targets_mean": 1825.6,
|
|
"valid_targets_min": 1415
|
|
},
|
|
{
|
|
"epoch": 0.2814070351758794,
|
|
"grad_norm": 0.9452855181185389,
|
|
"learning_rate": 1.5931232091690546e-05,
|
|
"loss": 0.3638,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1661728322505951,
|
|
"step": 140,
|
|
"valid_targets_mean": 1941.5,
|
|
"valid_targets_min": 1674
|
|
},
|
|
{
|
|
"epoch": 0.2914572864321608,
|
|
"grad_norm": 0.8881234552641465,
|
|
"learning_rate": 1.6504297994269343e-05,
|
|
"loss": 0.3605,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18108977377414703,
|
|
"step": 145,
|
|
"valid_targets_mean": 1959.0,
|
|
"valid_targets_min": 1572
|
|
},
|
|
{
|
|
"epoch": 0.3015075376884422,
|
|
"grad_norm": 0.9174021472627735,
|
|
"learning_rate": 1.707736389684814e-05,
|
|
"loss": 0.3946,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2209405153989792,
|
|
"step": 150,
|
|
"valid_targets_mean": 2077.6,
|
|
"valid_targets_min": 1653
|
|
},
|
|
{
|
|
"epoch": 0.31155778894472363,
|
|
"grad_norm": 0.9155526125174067,
|
|
"learning_rate": 1.7650429799426937e-05,
|
|
"loss": 0.3678,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21430011093616486,
|
|
"step": 155,
|
|
"valid_targets_mean": 1937.8,
|
|
"valid_targets_min": 1535
|
|
},
|
|
{
|
|
"epoch": 0.32160804020100503,
|
|
"grad_norm": 0.9447846769376781,
|
|
"learning_rate": 1.822349570200573e-05,
|
|
"loss": 0.361,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19075533747673035,
|
|
"step": 160,
|
|
"valid_targets_mean": 2034.6,
|
|
"valid_targets_min": 1752
|
|
},
|
|
{
|
|
"epoch": 0.3316582914572864,
|
|
"grad_norm": 0.9142384614733955,
|
|
"learning_rate": 1.879656160458453e-05,
|
|
"loss": 0.3555,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16853366792201996,
|
|
"step": 165,
|
|
"valid_targets_mean": 2014.1,
|
|
"valid_targets_min": 1777
|
|
},
|
|
{
|
|
"epoch": 0.3417085427135678,
|
|
"grad_norm": 0.935602217423317,
|
|
"learning_rate": 1.9369627507163324e-05,
|
|
"loss": 0.3479,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16409529745578766,
|
|
"step": 170,
|
|
"valid_targets_mean": 1815.2,
|
|
"valid_targets_min": 1503
|
|
},
|
|
{
|
|
"epoch": 0.35175879396984927,
|
|
"grad_norm": 0.8908725699593318,
|
|
"learning_rate": 1.994269340974212e-05,
|
|
"loss": 0.3436,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1572292149066925,
|
|
"step": 175,
|
|
"valid_targets_mean": 1747.9,
|
|
"valid_targets_min": 1401
|
|
},
|
|
{
|
|
"epoch": 0.36180904522613067,
|
|
"grad_norm": 0.9398730198173295,
|
|
"learning_rate": 2.0515759312320918e-05,
|
|
"loss": 0.37,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21750184893608093,
|
|
"step": 180,
|
|
"valid_targets_mean": 2175.1,
|
|
"valid_targets_min": 1853
|
|
},
|
|
{
|
|
"epoch": 0.37185929648241206,
|
|
"grad_norm": 0.9547047364651494,
|
|
"learning_rate": 2.108882521489971e-05,
|
|
"loss": 0.3437,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17827999591827393,
|
|
"step": 185,
|
|
"valid_targets_mean": 1967.2,
|
|
"valid_targets_min": 1371
|
|
},
|
|
{
|
|
"epoch": 0.38190954773869346,
|
|
"grad_norm": 0.9404091211489307,
|
|
"learning_rate": 2.1661891117478512e-05,
|
|
"loss": 0.3441,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16849786043167114,
|
|
"step": 190,
|
|
"valid_targets_mean": 1859.9,
|
|
"valid_targets_min": 1544
|
|
},
|
|
{
|
|
"epoch": 0.39195979899497485,
|
|
"grad_norm": 0.9524162292427605,
|
|
"learning_rate": 2.223495702005731e-05,
|
|
"loss": 0.3487,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1635562926530838,
|
|
"step": 195,
|
|
"valid_targets_mean": 1855.1,
|
|
"valid_targets_min": 1533
|
|
},
|
|
{
|
|
"epoch": 0.4020100502512563,
|
|
"grad_norm": 0.9324443995891041,
|
|
"learning_rate": 2.2808022922636102e-05,
|
|
"loss": 0.3424,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15275134146213531,
|
|
"step": 200,
|
|
"valid_targets_mean": 1901.0,
|
|
"valid_targets_min": 1688
|
|
},
|
|
{
|
|
"epoch": 0.4120603015075377,
|
|
"grad_norm": 1.0134032774772266,
|
|
"learning_rate": 2.3381088825214902e-05,
|
|
"loss": 0.346,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18456706404685974,
|
|
"step": 205,
|
|
"valid_targets_mean": 2067.9,
|
|
"valid_targets_min": 1764
|
|
},
|
|
{
|
|
"epoch": 0.4221105527638191,
|
|
"grad_norm": 1.0364737101281571,
|
|
"learning_rate": 2.39541547277937e-05,
|
|
"loss": 0.3497,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14801688492298126,
|
|
"step": 210,
|
|
"valid_targets_mean": 1902.2,
|
|
"valid_targets_min": 1562
|
|
},
|
|
{
|
|
"epoch": 0.4321608040201005,
|
|
"grad_norm": 0.9637681247243756,
|
|
"learning_rate": 2.4527220630372496e-05,
|
|
"loss": 0.3425,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17429080605506897,
|
|
"step": 215,
|
|
"valid_targets_mean": 2036.0,
|
|
"valid_targets_min": 1588
|
|
},
|
|
{
|
|
"epoch": 0.44221105527638194,
|
|
"grad_norm": 0.8699994825421622,
|
|
"learning_rate": 2.510028653295129e-05,
|
|
"loss": 0.325,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17875926196575165,
|
|
"step": 220,
|
|
"valid_targets_mean": 1954.5,
|
|
"valid_targets_min": 1678
|
|
},
|
|
{
|
|
"epoch": 0.45226130653266333,
|
|
"grad_norm": 0.8998360619023689,
|
|
"learning_rate": 2.567335243553009e-05,
|
|
"loss": 0.3315,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1401105523109436,
|
|
"step": 225,
|
|
"valid_targets_mean": 1679.2,
|
|
"valid_targets_min": 1318
|
|
},
|
|
{
|
|
"epoch": 0.4623115577889447,
|
|
"grad_norm": 0.9371618133591197,
|
|
"learning_rate": 2.6246418338108887e-05,
|
|
"loss": 0.3262,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16612085700035095,
|
|
"step": 230,
|
|
"valid_targets_mean": 1942.6,
|
|
"valid_targets_min": 1613
|
|
},
|
|
{
|
|
"epoch": 0.4723618090452261,
|
|
"grad_norm": 0.9407825971120665,
|
|
"learning_rate": 2.681948424068768e-05,
|
|
"loss": 0.3183,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.151786670088768,
|
|
"step": 235,
|
|
"valid_targets_mean": 1896.2,
|
|
"valid_targets_min": 1555
|
|
},
|
|
{
|
|
"epoch": 0.4824120603015075,
|
|
"grad_norm": 0.8985548093202931,
|
|
"learning_rate": 2.7392550143266478e-05,
|
|
"loss": 0.3179,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1582747995853424,
|
|
"step": 240,
|
|
"valid_targets_mean": 2013.4,
|
|
"valid_targets_min": 1754
|
|
},
|
|
{
|
|
"epoch": 0.49246231155778897,
|
|
"grad_norm": 0.8536046922204864,
|
|
"learning_rate": 2.7965616045845278e-05,
|
|
"loss": 0.3225,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15721815824508667,
|
|
"step": 245,
|
|
"valid_targets_mean": 1898.0,
|
|
"valid_targets_min": 1596
|
|
},
|
|
{
|
|
"epoch": 0.5025125628140703,
|
|
"grad_norm": 0.8922644966092359,
|
|
"learning_rate": 2.853868194842407e-05,
|
|
"loss": 0.3256,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18321126699447632,
|
|
"step": 250,
|
|
"valid_targets_mean": 2099.4,
|
|
"valid_targets_min": 1778
|
|
},
|
|
{
|
|
"epoch": 0.5125628140703518,
|
|
"grad_norm": 0.9489461938593724,
|
|
"learning_rate": 2.911174785100287e-05,
|
|
"loss": 0.32,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19852513074874878,
|
|
"step": 255,
|
|
"valid_targets_mean": 2191.1,
|
|
"valid_targets_min": 1784
|
|
},
|
|
{
|
|
"epoch": 0.5226130653266332,
|
|
"grad_norm": 0.9148126181556434,
|
|
"learning_rate": 2.9684813753581662e-05,
|
|
"loss": 0.3315,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15712934732437134,
|
|
"step": 260,
|
|
"valid_targets_mean": 1970.6,
|
|
"valid_targets_min": 1617
|
|
},
|
|
{
|
|
"epoch": 0.5326633165829145,
|
|
"grad_norm": 0.950417119697935,
|
|
"learning_rate": 3.0257879656160462e-05,
|
|
"loss": 0.3283,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1584141105413437,
|
|
"step": 265,
|
|
"valid_targets_mean": 1823.2,
|
|
"valid_targets_min": 1434
|
|
},
|
|
{
|
|
"epoch": 0.542713567839196,
|
|
"grad_norm": 0.9745824443598577,
|
|
"learning_rate": 3.083094555873926e-05,
|
|
"loss": 0.3324,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1314104050397873,
|
|
"step": 270,
|
|
"valid_targets_mean": 1765.8,
|
|
"valid_targets_min": 1440
|
|
},
|
|
{
|
|
"epoch": 0.5527638190954773,
|
|
"grad_norm": 0.9264963270996217,
|
|
"learning_rate": 3.140401146131805e-05,
|
|
"loss": 0.3114,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15098661184310913,
|
|
"step": 275,
|
|
"valid_targets_mean": 1933.4,
|
|
"valid_targets_min": 1617
|
|
},
|
|
{
|
|
"epoch": 0.5628140703517588,
|
|
"grad_norm": 0.9186727846763542,
|
|
"learning_rate": 3.1977077363896846e-05,
|
|
"loss": 0.3237,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14021429419517517,
|
|
"step": 280,
|
|
"valid_targets_mean": 1885.5,
|
|
"valid_targets_min": 1620
|
|
},
|
|
{
|
|
"epoch": 0.5728643216080402,
|
|
"grad_norm": 0.9954140486312582,
|
|
"learning_rate": 3.2550143266475647e-05,
|
|
"loss": 0.3149,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13806772232055664,
|
|
"step": 285,
|
|
"valid_targets_mean": 1870.0,
|
|
"valid_targets_min": 1647
|
|
},
|
|
{
|
|
"epoch": 0.5829145728643216,
|
|
"grad_norm": 0.9004500479166959,
|
|
"learning_rate": 3.312320916905445e-05,
|
|
"loss": 0.3251,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13797347247600555,
|
|
"step": 290,
|
|
"valid_targets_mean": 1869.8,
|
|
"valid_targets_min": 1547
|
|
},
|
|
{
|
|
"epoch": 0.592964824120603,
|
|
"grad_norm": 0.855596828380846,
|
|
"learning_rate": 3.369627507163324e-05,
|
|
"loss": 0.3132,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15226422250270844,
|
|
"step": 295,
|
|
"valid_targets_mean": 1911.8,
|
|
"valid_targets_min": 1733
|
|
},
|
|
{
|
|
"epoch": 0.6030150753768844,
|
|
"grad_norm": 0.9101274364021121,
|
|
"learning_rate": 3.4269340974212034e-05,
|
|
"loss": 0.3275,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14344167709350586,
|
|
"step": 300,
|
|
"valid_targets_mean": 2022.6,
|
|
"valid_targets_min": 1743
|
|
},
|
|
{
|
|
"epoch": 0.6130653266331658,
|
|
"grad_norm": 0.9048897124062735,
|
|
"learning_rate": 3.4842406876790834e-05,
|
|
"loss": 0.3243,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1544516384601593,
|
|
"step": 305,
|
|
"valid_targets_mean": 1878.5,
|
|
"valid_targets_min": 1693
|
|
},
|
|
{
|
|
"epoch": 0.6231155778894473,
|
|
"grad_norm": 0.9282954510457359,
|
|
"learning_rate": 3.541547277936963e-05,
|
|
"loss": 0.3282,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16127240657806396,
|
|
"step": 310,
|
|
"valid_targets_mean": 1931.5,
|
|
"valid_targets_min": 1523
|
|
},
|
|
{
|
|
"epoch": 0.6331658291457286,
|
|
"grad_norm": 1.0581063158841393,
|
|
"learning_rate": 3.598853868194843e-05,
|
|
"loss": 0.3275,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17590607702732086,
|
|
"step": 315,
|
|
"valid_targets_mean": 2068.2,
|
|
"valid_targets_min": 1374
|
|
},
|
|
{
|
|
"epoch": 0.6432160804020101,
|
|
"grad_norm": 0.8586482505155548,
|
|
"learning_rate": 3.656160458452722e-05,
|
|
"loss": 0.3144,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14930427074432373,
|
|
"step": 320,
|
|
"valid_targets_mean": 1891.5,
|
|
"valid_targets_min": 1551
|
|
},
|
|
{
|
|
"epoch": 0.6532663316582915,
|
|
"grad_norm": 0.9205690454598691,
|
|
"learning_rate": 3.713467048710602e-05,
|
|
"loss": 0.3168,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1630660593509674,
|
|
"step": 325,
|
|
"valid_targets_mean": 1976.4,
|
|
"valid_targets_min": 1756
|
|
},
|
|
{
|
|
"epoch": 0.6633165829145728,
|
|
"grad_norm": 0.870615556679017,
|
|
"learning_rate": 3.7707736389684816e-05,
|
|
"loss": 0.3051,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1486581712961197,
|
|
"step": 330,
|
|
"valid_targets_mean": 1807.9,
|
|
"valid_targets_min": 1311
|
|
},
|
|
{
|
|
"epoch": 0.6733668341708543,
|
|
"grad_norm": 0.9286204564647733,
|
|
"learning_rate": 3.828080229226361e-05,
|
|
"loss": 0.3105,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1487278938293457,
|
|
"step": 335,
|
|
"valid_targets_mean": 1763.9,
|
|
"valid_targets_min": 1534
|
|
},
|
|
{
|
|
"epoch": 0.6834170854271356,
|
|
"grad_norm": 0.8744389968374939,
|
|
"learning_rate": 3.885386819484241e-05,
|
|
"loss": 0.3111,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15776096284389496,
|
|
"step": 340,
|
|
"valid_targets_mean": 2035.8,
|
|
"valid_targets_min": 1722
|
|
},
|
|
{
|
|
"epoch": 0.6934673366834171,
|
|
"grad_norm": 0.9855408500337101,
|
|
"learning_rate": 3.942693409742121e-05,
|
|
"loss": 0.3073,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18504242599010468,
|
|
"step": 345,
|
|
"valid_targets_mean": 1988.2,
|
|
"valid_targets_min": 1353
|
|
},
|
|
{
|
|
"epoch": 0.7035175879396985,
|
|
"grad_norm": 0.8706114263575653,
|
|
"learning_rate": 4e-05,
|
|
"loss": 0.3109,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14136071503162384,
|
|
"step": 350,
|
|
"valid_targets_mean": 1901.2,
|
|
"valid_targets_min": 1554
|
|
},
|
|
{
|
|
"epoch": 0.7135678391959799,
|
|
"grad_norm": 0.8086762821420328,
|
|
"learning_rate": 3.999974926797441e-05,
|
|
"loss": 0.3108,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16270069777965546,
|
|
"step": 355,
|
|
"valid_targets_mean": 2005.9,
|
|
"valid_targets_min": 1632
|
|
},
|
|
{
|
|
"epoch": 0.7236180904522613,
|
|
"grad_norm": 0.9480613670780712,
|
|
"learning_rate": 3.999899707818427e-05,
|
|
"loss": 0.3144,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1576726883649826,
|
|
"step": 360,
|
|
"valid_targets_mean": 1859.2,
|
|
"valid_targets_min": 1513
|
|
},
|
|
{
|
|
"epoch": 0.7336683417085427,
|
|
"grad_norm": 0.9286822057297827,
|
|
"learning_rate": 3.999774344948941e-05,
|
|
"loss": 0.3036,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14612747728824615,
|
|
"step": 365,
|
|
"valid_targets_mean": 1920.5,
|
|
"valid_targets_min": 1732
|
|
},
|
|
{
|
|
"epoch": 0.7437185929648241,
|
|
"grad_norm": 0.8557504013275805,
|
|
"learning_rate": 3.99959884133223e-05,
|
|
"loss": 0.3076,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13518589735031128,
|
|
"step": 370,
|
|
"valid_targets_mean": 1913.9,
|
|
"valid_targets_min": 1696
|
|
},
|
|
{
|
|
"epoch": 0.7537688442211056,
|
|
"grad_norm": 0.8613748730656752,
|
|
"learning_rate": 3.999373201368732e-05,
|
|
"loss": 0.311,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1468210518360138,
|
|
"step": 375,
|
|
"valid_targets_mean": 1869.1,
|
|
"valid_targets_min": 1512
|
|
},
|
|
{
|
|
"epoch": 0.7638190954773869,
|
|
"grad_norm": 0.8513543769815133,
|
|
"learning_rate": 3.9990974307159645e-05,
|
|
"loss": 0.3041,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1705024093389511,
|
|
"step": 380,
|
|
"valid_targets_mean": 2068.9,
|
|
"valid_targets_min": 1693
|
|
},
|
|
{
|
|
"epoch": 0.7738693467336684,
|
|
"grad_norm": 0.8750757050686859,
|
|
"learning_rate": 3.99877153628838e-05,
|
|
"loss": 0.3122,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14878663420677185,
|
|
"step": 385,
|
|
"valid_targets_mean": 1901.0,
|
|
"valid_targets_min": 1423
|
|
},
|
|
{
|
|
"epoch": 0.7839195979899497,
|
|
"grad_norm": 0.8436783918710871,
|
|
"learning_rate": 3.998395526257195e-05,
|
|
"loss": 0.3034,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1522998958826065,
|
|
"step": 390,
|
|
"valid_targets_mean": 2079.6,
|
|
"valid_targets_min": 1662
|
|
},
|
|
{
|
|
"epoch": 0.7939698492462312,
|
|
"grad_norm": 0.8909241139693342,
|
|
"learning_rate": 3.9979694100501865e-05,
|
|
"loss": 0.2992,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.152342289686203,
|
|
"step": 395,
|
|
"valid_targets_mean": 1929.4,
|
|
"valid_targets_min": 1516
|
|
},
|
|
{
|
|
"epoch": 0.8040201005025126,
|
|
"grad_norm": 0.8502190127055751,
|
|
"learning_rate": 3.997493198351452e-05,
|
|
"loss": 0.2954,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16207163035869598,
|
|
"step": 400,
|
|
"valid_targets_mean": 2065.4,
|
|
"valid_targets_min": 1594
|
|
},
|
|
{
|
|
"epoch": 0.8140703517587939,
|
|
"grad_norm": 0.7704433762532498,
|
|
"learning_rate": 3.996966903101144e-05,
|
|
"loss": 0.3127,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1685250699520111,
|
|
"step": 405,
|
|
"valid_targets_mean": 2004.8,
|
|
"valid_targets_min": 1599
|
|
},
|
|
{
|
|
"epoch": 0.8241206030150754,
|
|
"grad_norm": 0.7812447729012929,
|
|
"learning_rate": 3.996390537495169e-05,
|
|
"loss": 0.305,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1564028561115265,
|
|
"step": 410,
|
|
"valid_targets_mean": 1878.9,
|
|
"valid_targets_min": 1483
|
|
},
|
|
{
|
|
"epoch": 0.8341708542713567,
|
|
"grad_norm": 0.7651469462202585,
|
|
"learning_rate": 3.99576411598486e-05,
|
|
"loss": 0.2996,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15438030660152435,
|
|
"step": 415,
|
|
"valid_targets_mean": 1874.8,
|
|
"valid_targets_min": 1445
|
|
},
|
|
{
|
|
"epoch": 0.8442211055276382,
|
|
"grad_norm": 0.7866832467887138,
|
|
"learning_rate": 3.995087654276609e-05,
|
|
"loss": 0.3073,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1518499106168747,
|
|
"step": 420,
|
|
"valid_targets_mean": 1889.2,
|
|
"valid_targets_min": 1622
|
|
},
|
|
{
|
|
"epoch": 0.8542713567839196,
|
|
"grad_norm": 0.8128983954980532,
|
|
"learning_rate": 3.9943611693314775e-05,
|
|
"loss": 0.3095,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16506221890449524,
|
|
"step": 425,
|
|
"valid_targets_mean": 2090.5,
|
|
"valid_targets_min": 1805
|
|
},
|
|
{
|
|
"epoch": 0.864321608040201,
|
|
"grad_norm": 0.805101760020037,
|
|
"learning_rate": 3.993584679364771e-05,
|
|
"loss": 0.3107,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1521514505147934,
|
|
"step": 430,
|
|
"valid_targets_mean": 1945.0,
|
|
"valid_targets_min": 1682
|
|
},
|
|
{
|
|
"epoch": 0.8743718592964824,
|
|
"grad_norm": 0.8044252147307049,
|
|
"learning_rate": 3.9927582038455795e-05,
|
|
"loss": 0.295,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16960017383098602,
|
|
"step": 435,
|
|
"valid_targets_mean": 2031.5,
|
|
"valid_targets_min": 1475
|
|
},
|
|
{
|
|
"epoch": 0.8844221105527639,
|
|
"grad_norm": 0.8004321905309837,
|
|
"learning_rate": 3.99188176349629e-05,
|
|
"loss": 0.3087,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15131238102912903,
|
|
"step": 440,
|
|
"valid_targets_mean": 1969.0,
|
|
"valid_targets_min": 1635
|
|
},
|
|
{
|
|
"epoch": 0.8944723618090452,
|
|
"grad_norm": 0.7607167761398257,
|
|
"learning_rate": 3.99095538029207e-05,
|
|
"loss": 0.3014,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14829257130622864,
|
|
"step": 445,
|
|
"valid_targets_mean": 2025.6,
|
|
"valid_targets_min": 1523
|
|
},
|
|
{
|
|
"epoch": 0.9045226130653267,
|
|
"grad_norm": 0.7858978397046998,
|
|
"learning_rate": 3.989979077460312e-05,
|
|
"loss": 0.2997,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15068481862545013,
|
|
"step": 450,
|
|
"valid_targets_mean": 1822.4,
|
|
"valid_targets_min": 1385
|
|
},
|
|
{
|
|
"epoch": 0.914572864321608,
|
|
"grad_norm": 0.8866620125997927,
|
|
"learning_rate": 3.988952879480057e-05,
|
|
"loss": 0.3011,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1392013430595398,
|
|
"step": 455,
|
|
"valid_targets_mean": 1947.6,
|
|
"valid_targets_min": 1501
|
|
},
|
|
{
|
|
"epoch": 0.9246231155778895,
|
|
"grad_norm": 0.7226941516453836,
|
|
"learning_rate": 3.987876812081371e-05,
|
|
"loss": 0.3153,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.140838623046875,
|
|
"step": 460,
|
|
"valid_targets_mean": 1941.8,
|
|
"valid_targets_min": 1520
|
|
},
|
|
{
|
|
"epoch": 0.9346733668341709,
|
|
"grad_norm": 0.8494740903438865,
|
|
"learning_rate": 3.9867509022447134e-05,
|
|
"loss": 0.3117,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18316376209259033,
|
|
"step": 465,
|
|
"valid_targets_mean": 2172.4,
|
|
"valid_targets_min": 1944
|
|
},
|
|
{
|
|
"epoch": 0.9447236180904522,
|
|
"grad_norm": 0.8122621816644038,
|
|
"learning_rate": 3.9855751782002485e-05,
|
|
"loss": 0.2984,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1252681314945221,
|
|
"step": 470,
|
|
"valid_targets_mean": 1803.1,
|
|
"valid_targets_min": 1344
|
|
},
|
|
{
|
|
"epoch": 0.9547738693467337,
|
|
"grad_norm": 0.8808576305845461,
|
|
"learning_rate": 3.984349669427143e-05,
|
|
"loss": 0.2999,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1422131508588791,
|
|
"step": 475,
|
|
"valid_targets_mean": 1844.4,
|
|
"valid_targets_min": 1549
|
|
},
|
|
{
|
|
"epoch": 0.964824120603015,
|
|
"grad_norm": 0.7605267960008191,
|
|
"learning_rate": 3.983074406652826e-05,
|
|
"loss": 0.302,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16110354661941528,
|
|
"step": 480,
|
|
"valid_targets_mean": 2031.4,
|
|
"valid_targets_min": 1434
|
|
},
|
|
{
|
|
"epoch": 0.9748743718592965,
|
|
"grad_norm": 0.7548024675005609,
|
|
"learning_rate": 3.9817494218522215e-05,
|
|
"loss": 0.3114,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15635475516319275,
|
|
"step": 485,
|
|
"valid_targets_mean": 2117.9,
|
|
"valid_targets_min": 1459
|
|
},
|
|
{
|
|
"epoch": 0.9849246231155779,
|
|
"grad_norm": 0.7465223549270914,
|
|
"learning_rate": 3.98037474824694e-05,
|
|
"loss": 0.2903,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14686113595962524,
|
|
"step": 490,
|
|
"valid_targets_mean": 1969.4,
|
|
"valid_targets_min": 1772
|
|
},
|
|
{
|
|
"epoch": 0.9949748743718593,
|
|
"grad_norm": 0.8186869768334977,
|
|
"learning_rate": 3.978950420304451e-05,
|
|
"loss": 0.2835,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13193729519844055,
|
|
"step": 495,
|
|
"valid_targets_mean": 1678.5,
|
|
"valid_targets_min": 374
|
|
},
|
|
{
|
|
"epoch": 1.0040201005025127,
|
|
"grad_norm": 0.7196686161600628,
|
|
"learning_rate": 3.977476473737219e-05,
|
|
"loss": 0.2977,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15454989671707153,
|
|
"step": 500,
|
|
"valid_targets_mean": 1980.1,
|
|
"valid_targets_min": 1799
|
|
},
|
|
{
|
|
"epoch": 1.014070351758794,
|
|
"grad_norm": 0.7173201150906622,
|
|
"learning_rate": 3.975952945501804e-05,
|
|
"loss": 0.2879,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15298259258270264,
|
|
"step": 505,
|
|
"valid_targets_mean": 1976.4,
|
|
"valid_targets_min": 1385
|
|
},
|
|
{
|
|
"epoch": 1.0241206030150753,
|
|
"grad_norm": 0.7519117703969496,
|
|
"learning_rate": 3.974379873797938e-05,
|
|
"loss": 0.2978,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12313634902238846,
|
|
"step": 510,
|
|
"valid_targets_mean": 1868.5,
|
|
"valid_targets_min": 1495
|
|
},
|
|
{
|
|
"epoch": 1.0341708542713568,
|
|
"grad_norm": 0.812037089331207,
|
|
"learning_rate": 3.972757298067566e-05,
|
|
"loss": 0.2896,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14630839228630066,
|
|
"step": 515,
|
|
"valid_targets_mean": 1919.4,
|
|
"valid_targets_min": 1526
|
|
},
|
|
{
|
|
"epoch": 1.0442211055276382,
|
|
"grad_norm": 0.7396594889501839,
|
|
"learning_rate": 3.9710852589938595e-05,
|
|
"loss": 0.2763,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13931876420974731,
|
|
"step": 520,
|
|
"valid_targets_mean": 1887.4,
|
|
"valid_targets_min": 1502
|
|
},
|
|
{
|
|
"epoch": 1.0542713567839197,
|
|
"grad_norm": 0.7759252051703647,
|
|
"learning_rate": 3.969363798500192e-05,
|
|
"loss": 0.2857,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14481891691684723,
|
|
"step": 525,
|
|
"valid_targets_mean": 1891.6,
|
|
"valid_targets_min": 1592
|
|
},
|
|
{
|
|
"epoch": 1.064321608040201,
|
|
"grad_norm": 0.7141473732724476,
|
|
"learning_rate": 3.96759295974909e-05,
|
|
"loss": 0.2939,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13978004455566406,
|
|
"step": 530,
|
|
"valid_targets_mean": 1850.0,
|
|
"valid_targets_min": 1579
|
|
},
|
|
{
|
|
"epoch": 1.0743718592964824,
|
|
"grad_norm": 0.7711560563366033,
|
|
"learning_rate": 3.9657727871411543e-05,
|
|
"loss": 0.2828,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11955422163009644,
|
|
"step": 535,
|
|
"valid_targets_mean": 1804.8,
|
|
"valid_targets_min": 1579
|
|
},
|
|
{
|
|
"epoch": 1.0844221105527638,
|
|
"grad_norm": 0.6838252813246216,
|
|
"learning_rate": 3.96390332631394e-05,
|
|
"loss": 0.2932,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14464765787124634,
|
|
"step": 540,
|
|
"valid_targets_mean": 1880.8,
|
|
"valid_targets_min": 1475
|
|
},
|
|
{
|
|
"epoch": 1.0944723618090453,
|
|
"grad_norm": 0.7518363198045526,
|
|
"learning_rate": 3.961984624140818e-05,
|
|
"loss": 0.2966,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15483617782592773,
|
|
"step": 545,
|
|
"valid_targets_mean": 2044.0,
|
|
"valid_targets_min": 1523
|
|
},
|
|
{
|
|
"epoch": 1.1045226130653267,
|
|
"grad_norm": 1.0058802387412142,
|
|
"learning_rate": 3.9600167287297954e-05,
|
|
"loss": 0.2965,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11946623772382736,
|
|
"step": 550,
|
|
"valid_targets_mean": 1841.1,
|
|
"valid_targets_min": 1446
|
|
},
|
|
{
|
|
"epoch": 1.114572864321608,
|
|
"grad_norm": 0.7521570188169158,
|
|
"learning_rate": 3.957999689422313e-05,
|
|
"loss": 0.2863,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11747627705335617,
|
|
"step": 555,
|
|
"valid_targets_mean": 1813.8,
|
|
"valid_targets_min": 1420
|
|
},
|
|
{
|
|
"epoch": 1.1246231155778894,
|
|
"grad_norm": 0.764886586834361,
|
|
"learning_rate": 3.955933556792007e-05,
|
|
"loss": 0.2733,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1453430950641632,
|
|
"step": 560,
|
|
"valid_targets_mean": 2059.6,
|
|
"valid_targets_min": 1554
|
|
},
|
|
{
|
|
"epoch": 1.1346733668341709,
|
|
"grad_norm": 0.774016017218095,
|
|
"learning_rate": 3.9538183826434374e-05,
|
|
"loss": 0.2954,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16331876814365387,
|
|
"step": 565,
|
|
"valid_targets_mean": 2151.4,
|
|
"valid_targets_min": 1619
|
|
},
|
|
{
|
|
"epoch": 1.1447236180904523,
|
|
"grad_norm": 0.8156602703420109,
|
|
"learning_rate": 3.951654220010795e-05,
|
|
"loss": 0.2813,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15402823686599731,
|
|
"step": 570,
|
|
"valid_targets_mean": 2026.8,
|
|
"valid_targets_min": 1758
|
|
},
|
|
{
|
|
"epoch": 1.1547738693467338,
|
|
"grad_norm": 0.7177974676824063,
|
|
"learning_rate": 3.949441123156569e-05,
|
|
"loss": 0.2873,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14421150088310242,
|
|
"step": 575,
|
|
"valid_targets_mean": 1979.0,
|
|
"valid_targets_min": 1625
|
|
},
|
|
{
|
|
"epoch": 1.164824120603015,
|
|
"grad_norm": 0.798190565074642,
|
|
"learning_rate": 3.9471791475701834e-05,
|
|
"loss": 0.281,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10651945322751999,
|
|
"step": 580,
|
|
"valid_targets_mean": 1744.4,
|
|
"valid_targets_min": 1625
|
|
},
|
|
{
|
|
"epoch": 1.1748743718592964,
|
|
"grad_norm": 0.783838518807519,
|
|
"learning_rate": 3.9448683499666105e-05,
|
|
"loss": 0.2964,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15353409945964813,
|
|
"step": 585,
|
|
"valid_targets_mean": 2084.2,
|
|
"valid_targets_min": 1780
|
|
},
|
|
{
|
|
"epoch": 1.184924623115578,
|
|
"grad_norm": 0.7478054264786744,
|
|
"learning_rate": 3.942508788284948e-05,
|
|
"loss": 0.2874,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14684703946113586,
|
|
"step": 590,
|
|
"valid_targets_mean": 2000.8,
|
|
"valid_targets_min": 1643
|
|
},
|
|
{
|
|
"epoch": 1.1949748743718593,
|
|
"grad_norm": 0.7471869729538552,
|
|
"learning_rate": 3.940100521686962e-05,
|
|
"loss": 0.2834,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13836337625980377,
|
|
"step": 595,
|
|
"valid_targets_mean": 1951.6,
|
|
"valid_targets_min": 1683
|
|
},
|
|
{
|
|
"epoch": 1.2050251256281408,
|
|
"grad_norm": 0.7899927223062734,
|
|
"learning_rate": 3.9376436105556104e-05,
|
|
"loss": 0.2774,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12772196531295776,
|
|
"step": 600,
|
|
"valid_targets_mean": 1685.8,
|
|
"valid_targets_min": 1280
|
|
},
|
|
{
|
|
"epoch": 1.215075376884422,
|
|
"grad_norm": 0.7530975203032924,
|
|
"learning_rate": 3.935138116493522e-05,
|
|
"loss": 0.2818,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12249584496021271,
|
|
"step": 605,
|
|
"valid_targets_mean": 1794.6,
|
|
"valid_targets_min": 1647
|
|
},
|
|
{
|
|
"epoch": 1.2251256281407035,
|
|
"grad_norm": 0.735864445448524,
|
|
"learning_rate": 3.9325841023214586e-05,
|
|
"loss": 0.2788,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13310131430625916,
|
|
"step": 610,
|
|
"valid_targets_mean": 1937.5,
|
|
"valid_targets_min": 1637
|
|
},
|
|
{
|
|
"epoch": 1.235175879396985,
|
|
"grad_norm": 0.717575736297205,
|
|
"learning_rate": 3.929981632076734e-05,
|
|
"loss": 0.2842,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12135225534439087,
|
|
"step": 615,
|
|
"valid_targets_mean": 1720.8,
|
|
"valid_targets_min": 1372
|
|
},
|
|
{
|
|
"epoch": 1.2452261306532664,
|
|
"grad_norm": 0.7467122446434402,
|
|
"learning_rate": 3.9273307710116124e-05,
|
|
"loss": 0.2898,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13005143404006958,
|
|
"step": 620,
|
|
"valid_targets_mean": 1878.0,
|
|
"valid_targets_min": 1739
|
|
},
|
|
{
|
|
"epoch": 1.2552763819095478,
|
|
"grad_norm": 0.8119440899631103,
|
|
"learning_rate": 3.9246315855916695e-05,
|
|
"loss": 0.2982,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16680994629859924,
|
|
"step": 625,
|
|
"valid_targets_mean": 1951.4,
|
|
"valid_targets_min": 315
|
|
},
|
|
{
|
|
"epoch": 1.265326633165829,
|
|
"grad_norm": 0.7538943076927731,
|
|
"learning_rate": 3.9218841434941285e-05,
|
|
"loss": 0.2787,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13128474354743958,
|
|
"step": 630,
|
|
"valid_targets_mean": 1866.5,
|
|
"valid_targets_min": 1675
|
|
},
|
|
{
|
|
"epoch": 1.2753768844221105,
|
|
"grad_norm": 0.7145158577635828,
|
|
"learning_rate": 3.9190885136061616e-05,
|
|
"loss": 0.2751,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13453172147274017,
|
|
"step": 635,
|
|
"valid_targets_mean": 2023.9,
|
|
"valid_targets_min": 1644
|
|
},
|
|
{
|
|
"epoch": 1.285427135678392,
|
|
"grad_norm": 0.7152458086707898,
|
|
"learning_rate": 3.9162447660231626e-05,
|
|
"loss": 0.2902,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14699062705039978,
|
|
"step": 640,
|
|
"valid_targets_mean": 1997.6,
|
|
"valid_targets_min": 1666
|
|
},
|
|
{
|
|
"epoch": 1.2954773869346734,
|
|
"grad_norm": 0.7282360851595434,
|
|
"learning_rate": 3.913352972046992e-05,
|
|
"loss": 0.2769,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15252172946929932,
|
|
"step": 645,
|
|
"valid_targets_mean": 2013.2,
|
|
"valid_targets_min": 1655
|
|
},
|
|
{
|
|
"epoch": 1.3055276381909549,
|
|
"grad_norm": 0.8342452501507404,
|
|
"learning_rate": 3.910413204184185e-05,
|
|
"loss": 0.2826,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14294087886810303,
|
|
"step": 650,
|
|
"valid_targets_mean": 1944.5,
|
|
"valid_targets_min": 1460
|
|
},
|
|
{
|
|
"epoch": 1.315577889447236,
|
|
"grad_norm": 0.7224693446402728,
|
|
"learning_rate": 3.907425536144136e-05,
|
|
"loss": 0.2776,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13652563095092773,
|
|
"step": 655,
|
|
"valid_targets_mean": 1872.4,
|
|
"valid_targets_min": 1476
|
|
},
|
|
{
|
|
"epoch": 1.3256281407035175,
|
|
"grad_norm": 0.7115241210248913,
|
|
"learning_rate": 3.9043900428372535e-05,
|
|
"loss": 0.2852,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11880982667207718,
|
|
"step": 660,
|
|
"valid_targets_mean": 1844.6,
|
|
"valid_targets_min": 1618
|
|
},
|
|
{
|
|
"epoch": 1.335678391959799,
|
|
"grad_norm": 0.711141447495574,
|
|
"learning_rate": 3.901306800373073e-05,
|
|
"loss": 0.2817,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13360953330993652,
|
|
"step": 665,
|
|
"valid_targets_mean": 1966.9,
|
|
"valid_targets_min": 1488
|
|
},
|
|
{
|
|
"epoch": 1.3457286432160804,
|
|
"grad_norm": 0.7315738627309077,
|
|
"learning_rate": 3.8981758860583596e-05,
|
|
"loss": 0.2912,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17795708775520325,
|
|
"step": 670,
|
|
"valid_targets_mean": 2202.2,
|
|
"valid_targets_min": 1845
|
|
},
|
|
{
|
|
"epoch": 1.3557788944723619,
|
|
"grad_norm": 0.7230039633760416,
|
|
"learning_rate": 3.894997378395161e-05,
|
|
"loss": 0.2826,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13923469185829163,
|
|
"step": 675,
|
|
"valid_targets_mean": 1911.8,
|
|
"valid_targets_min": 1727
|
|
},
|
|
{
|
|
"epoch": 1.3658291457286431,
|
|
"grad_norm": 0.7350754069200873,
|
|
"learning_rate": 3.8917713570788436e-05,
|
|
"loss": 0.2667,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14508052170276642,
|
|
"step": 680,
|
|
"valid_targets_mean": 1901.6,
|
|
"valid_targets_min": 1587
|
|
},
|
|
{
|
|
"epoch": 1.3758793969849246,
|
|
"grad_norm": 0.7198273494996336,
|
|
"learning_rate": 3.8884979029960945e-05,
|
|
"loss": 0.281,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16034308075904846,
|
|
"step": 685,
|
|
"valid_targets_mean": 2041.9,
|
|
"valid_targets_min": 1426
|
|
},
|
|
{
|
|
"epoch": 1.385929648241206,
|
|
"grad_norm": 0.7432451206302603,
|
|
"learning_rate": 3.88517709822289e-05,
|
|
"loss": 0.2964,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15086729824543,
|
|
"step": 690,
|
|
"valid_targets_mean": 1975.0,
|
|
"valid_targets_min": 1577
|
|
},
|
|
{
|
|
"epoch": 1.3959798994974875,
|
|
"grad_norm": 0.7318700335819568,
|
|
"learning_rate": 3.8818090260224404e-05,
|
|
"loss": 0.2887,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14579252898693085,
|
|
"step": 695,
|
|
"valid_targets_mean": 1931.9,
|
|
"valid_targets_min": 1611
|
|
},
|
|
{
|
|
"epoch": 1.406030150753769,
|
|
"grad_norm": 0.7273270744814034,
|
|
"learning_rate": 3.878393770843102e-05,
|
|
"loss": 0.2804,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13960416615009308,
|
|
"step": 700,
|
|
"valid_targets_mean": 1983.5,
|
|
"valid_targets_min": 1684
|
|
},
|
|
{
|
|
"epoch": 1.4160804020100501,
|
|
"grad_norm": 0.7560940500658831,
|
|
"learning_rate": 3.874931418316262e-05,
|
|
"loss": 0.2745,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13562002778053284,
|
|
"step": 705,
|
|
"valid_targets_mean": 1979.4,
|
|
"valid_targets_min": 1628
|
|
},
|
|
{
|
|
"epoch": 1.4261306532663316,
|
|
"grad_norm": 0.7413136574634417,
|
|
"learning_rate": 3.871422055254185e-05,
|
|
"loss": 0.2889,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13405956327915192,
|
|
"step": 710,
|
|
"valid_targets_mean": 1935.9,
|
|
"valid_targets_min": 1419
|
|
},
|
|
{
|
|
"epoch": 1.436180904522613,
|
|
"grad_norm": 0.7183405647179412,
|
|
"learning_rate": 3.867865769647841e-05,
|
|
"loss": 0.2837,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1300911009311676,
|
|
"step": 715,
|
|
"valid_targets_mean": 1806.2,
|
|
"valid_targets_min": 967
|
|
},
|
|
{
|
|
"epoch": 1.4462311557788945,
|
|
"grad_norm": 0.7205997857820704,
|
|
"learning_rate": 3.864262650664701e-05,
|
|
"loss": 0.2786,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14690859615802765,
|
|
"step": 720,
|
|
"valid_targets_mean": 1994.8,
|
|
"valid_targets_min": 1620
|
|
},
|
|
{
|
|
"epoch": 1.456281407035176,
|
|
"grad_norm": 0.6921676546022655,
|
|
"learning_rate": 3.8606127886464964e-05,
|
|
"loss": 0.2776,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12897585332393646,
|
|
"step": 725,
|
|
"valid_targets_mean": 1909.2,
|
|
"valid_targets_min": 1300
|
|
},
|
|
{
|
|
"epoch": 1.4663316582914572,
|
|
"grad_norm": 0.684365651530659,
|
|
"learning_rate": 3.856916275106958e-05,
|
|
"loss": 0.2768,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1374325305223465,
|
|
"step": 730,
|
|
"valid_targets_mean": 1996.9,
|
|
"valid_targets_min": 1550
|
|
},
|
|
{
|
|
"epoch": 1.4763819095477386,
|
|
"grad_norm": 0.7467285482016467,
|
|
"learning_rate": 3.8531732027295166e-05,
|
|
"loss": 0.285,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1550510972738266,
|
|
"step": 735,
|
|
"valid_targets_mean": 1983.6,
|
|
"valid_targets_min": 1495
|
|
},
|
|
{
|
|
"epoch": 1.48643216080402,
|
|
"grad_norm": 0.6492054270811952,
|
|
"learning_rate": 3.849383665364985e-05,
|
|
"loss": 0.274,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12660571932792664,
|
|
"step": 740,
|
|
"valid_targets_mean": 1994.5,
|
|
"valid_targets_min": 1644
|
|
},
|
|
{
|
|
"epoch": 1.4964824120603015,
|
|
"grad_norm": 0.7018848075154865,
|
|
"learning_rate": 3.8455477580292015e-05,
|
|
"loss": 0.2784,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12711700797080994,
|
|
"step": 745,
|
|
"valid_targets_mean": 1943.2,
|
|
"valid_targets_min": 1349
|
|
},
|
|
{
|
|
"epoch": 1.506532663316583,
|
|
"grad_norm": 0.6808337554303464,
|
|
"learning_rate": 3.841665576900647e-05,
|
|
"loss": 0.2752,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1464816927909851,
|
|
"step": 750,
|
|
"valid_targets_mean": 2009.9,
|
|
"valid_targets_min": 1601
|
|
},
|
|
{
|
|
"epoch": 1.5165829145728642,
|
|
"grad_norm": 0.7242146066013057,
|
|
"learning_rate": 3.8377372193180364e-05,
|
|
"loss": 0.2823,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14681029319763184,
|
|
"step": 755,
|
|
"valid_targets_mean": 2006.0,
|
|
"valid_targets_min": 1708
|
|
},
|
|
{
|
|
"epoch": 1.5266331658291459,
|
|
"grad_norm": 0.7122657580231347,
|
|
"learning_rate": 3.833762783777875e-05,
|
|
"loss": 0.2787,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12979522347450256,
|
|
"step": 760,
|
|
"valid_targets_mean": 1914.8,
|
|
"valid_targets_min": 1611
|
|
},
|
|
{
|
|
"epoch": 1.536683417085427,
|
|
"grad_norm": 0.7427265591780403,
|
|
"learning_rate": 3.829742369931989e-05,
|
|
"loss": 0.2901,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14869844913482666,
|
|
"step": 765,
|
|
"valid_targets_mean": 2014.4,
|
|
"valid_targets_min": 1603
|
|
},
|
|
{
|
|
"epoch": 1.5467336683417086,
|
|
"grad_norm": 0.7803531741310764,
|
|
"learning_rate": 3.8256760785850294e-05,
|
|
"loss": 0.2878,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14112207293510437,
|
|
"step": 770,
|
|
"valid_targets_mean": 2075.5,
|
|
"valid_targets_min": 1613
|
|
},
|
|
{
|
|
"epoch": 1.55678391959799,
|
|
"grad_norm": 0.6993536102058309,
|
|
"learning_rate": 3.821564011691944e-05,
|
|
"loss": 0.2731,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13066790997982025,
|
|
"step": 775,
|
|
"valid_targets_mean": 1963.9,
|
|
"valid_targets_min": 1576
|
|
},
|
|
{
|
|
"epoch": 1.5668341708542712,
|
|
"grad_norm": 0.6900901162736238,
|
|
"learning_rate": 3.817406272355417e-05,
|
|
"loss": 0.2795,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1374991238117218,
|
|
"step": 780,
|
|
"valid_targets_mean": 1985.2,
|
|
"valid_targets_min": 1467
|
|
},
|
|
{
|
|
"epoch": 1.576884422110553,
|
|
"grad_norm": 0.6493336247461317,
|
|
"learning_rate": 3.813202964823291e-05,
|
|
"loss": 0.2815,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1204734519124031,
|
|
"step": 785,
|
|
"valid_targets_mean": 1838.6,
|
|
"valid_targets_min": 1596
|
|
},
|
|
{
|
|
"epoch": 1.5869346733668341,
|
|
"grad_norm": 0.7585515215352826,
|
|
"learning_rate": 3.808954194485946e-05,
|
|
"loss": 0.2764,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13259339332580566,
|
|
"step": 790,
|
|
"valid_targets_mean": 1963.4,
|
|
"valid_targets_min": 1692
|
|
},
|
|
{
|
|
"epoch": 1.5969849246231156,
|
|
"grad_norm": 0.6977495466804854,
|
|
"learning_rate": 3.804660067873661e-05,
|
|
"loss": 0.2771,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13366694748401642,
|
|
"step": 795,
|
|
"valid_targets_mean": 1815.0,
|
|
"valid_targets_min": 1674
|
|
},
|
|
{
|
|
"epoch": 1.607035175879397,
|
|
"grad_norm": 0.673279184510467,
|
|
"learning_rate": 3.800320692653944e-05,
|
|
"loss": 0.2652,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1157965362071991,
|
|
"step": 800,
|
|
"valid_targets_mean": 1920.8,
|
|
"valid_targets_min": 1565
|
|
},
|
|
{
|
|
"epoch": 1.6170854271356783,
|
|
"grad_norm": 0.7257340553685114,
|
|
"learning_rate": 3.795936177628827e-05,
|
|
"loss": 0.2747,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12203636020421982,
|
|
"step": 805,
|
|
"valid_targets_mean": 1807.4,
|
|
"valid_targets_min": 1643
|
|
},
|
|
{
|
|
"epoch": 1.62713567839196,
|
|
"grad_norm": 0.7058775062635547,
|
|
"learning_rate": 3.791506632732144e-05,
|
|
"loss": 0.2921,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14820978045463562,
|
|
"step": 810,
|
|
"valid_targets_mean": 1882.8,
|
|
"valid_targets_min": 1466
|
|
},
|
|
{
|
|
"epoch": 1.6371859296482412,
|
|
"grad_norm": 0.7378265007459175,
|
|
"learning_rate": 3.787032169026772e-05,
|
|
"loss": 0.279,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12492141872644424,
|
|
"step": 815,
|
|
"valid_targets_mean": 1849.9,
|
|
"valid_targets_min": 1670
|
|
},
|
|
{
|
|
"epoch": 1.6472361809045226,
|
|
"grad_norm": 0.7135000910479742,
|
|
"learning_rate": 3.7825128987018465e-05,
|
|
"loss": 0.2777,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15317213535308838,
|
|
"step": 820,
|
|
"valid_targets_mean": 1897.0,
|
|
"valid_targets_min": 1752
|
|
},
|
|
{
|
|
"epoch": 1.657286432160804,
|
|
"grad_norm": 0.6390258404762654,
|
|
"learning_rate": 3.7779489350699446e-05,
|
|
"loss": 0.2738,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1252155900001526,
|
|
"step": 825,
|
|
"valid_targets_mean": 1882.5,
|
|
"valid_targets_min": 1562
|
|
},
|
|
{
|
|
"epoch": 1.6673366834170853,
|
|
"grad_norm": 0.7610669983203394,
|
|
"learning_rate": 3.7733403925642546e-05,
|
|
"loss": 0.2744,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1321600377559662,
|
|
"step": 830,
|
|
"valid_targets_mean": 1878.5,
|
|
"valid_targets_min": 1558
|
|
},
|
|
{
|
|
"epoch": 1.677386934673367,
|
|
"grad_norm": 0.6710787487746116,
|
|
"learning_rate": 3.768687386735694e-05,
|
|
"loss": 0.2818,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13134528696537018,
|
|
"step": 835,
|
|
"valid_targets_mean": 2050.1,
|
|
"valid_targets_min": 1705
|
|
},
|
|
{
|
|
"epoch": 1.6874371859296482,
|
|
"grad_norm": 0.661888625855343,
|
|
"learning_rate": 3.763990034250021e-05,
|
|
"loss": 0.2813,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14188703894615173,
|
|
"step": 840,
|
|
"valid_targets_mean": 2169.8,
|
|
"valid_targets_min": 1872
|
|
},
|
|
{
|
|
"epoch": 1.6974874371859296,
|
|
"grad_norm": 0.6576164901346973,
|
|
"learning_rate": 3.7592484528849065e-05,
|
|
"loss": 0.2814,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12151030451059341,
|
|
"step": 845,
|
|
"valid_targets_mean": 1808.1,
|
|
"valid_targets_min": 1619
|
|
},
|
|
{
|
|
"epoch": 1.707537688442211,
|
|
"grad_norm": 0.6557113695055623,
|
|
"learning_rate": 3.75446276152698e-05,
|
|
"loss": 0.2745,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1294020116329193,
|
|
"step": 850,
|
|
"valid_targets_mean": 1903.0,
|
|
"valid_targets_min": 1617
|
|
},
|
|
{
|
|
"epoch": 1.7175879396984923,
|
|
"grad_norm": 0.72724275282573,
|
|
"learning_rate": 3.74963308016885e-05,
|
|
"loss": 0.2806,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14017245173454285,
|
|
"step": 855,
|
|
"valid_targets_mean": 2030.0,
|
|
"valid_targets_min": 1472
|
|
},
|
|
{
|
|
"epoch": 1.727638190954774,
|
|
"grad_norm": 0.6353385183186154,
|
|
"learning_rate": 3.7447595299060964e-05,
|
|
"loss": 0.2856,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13149696588516235,
|
|
"step": 860,
|
|
"valid_targets_mean": 1884.1,
|
|
"valid_targets_min": 1532
|
|
},
|
|
{
|
|
"epoch": 1.7376884422110552,
|
|
"grad_norm": 0.6682270011097987,
|
|
"learning_rate": 3.739842232934231e-05,
|
|
"loss": 0.271,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12547338008880615,
|
|
"step": 865,
|
|
"valid_targets_mean": 1758.2,
|
|
"valid_targets_min": 1244
|
|
},
|
|
{
|
|
"epoch": 1.7477386934673367,
|
|
"grad_norm": 0.7005021734281794,
|
|
"learning_rate": 3.734881312545638e-05,
|
|
"loss": 0.2834,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12836101651191711,
|
|
"step": 870,
|
|
"valid_targets_mean": 1941.2,
|
|
"valid_targets_min": 1529
|
|
},
|
|
{
|
|
"epoch": 1.7577889447236181,
|
|
"grad_norm": 0.6413905433199246,
|
|
"learning_rate": 3.7298768931264785e-05,
|
|
"loss": 0.2626,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1462337076663971,
|
|
"step": 875,
|
|
"valid_targets_mean": 2001.6,
|
|
"valid_targets_min": 1699
|
|
},
|
|
{
|
|
"epoch": 1.7678391959798994,
|
|
"grad_norm": 0.6753763718219002,
|
|
"learning_rate": 3.724829100153574e-05,
|
|
"loss": 0.2832,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1251835823059082,
|
|
"step": 880,
|
|
"valid_targets_mean": 1688.6,
|
|
"valid_targets_min": 469
|
|
},
|
|
{
|
|
"epoch": 1.777889447236181,
|
|
"grad_norm": 0.6879790591393932,
|
|
"learning_rate": 3.719738060191261e-05,
|
|
"loss": 0.2771,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12924236059188843,
|
|
"step": 885,
|
|
"valid_targets_mean": 1859.8,
|
|
"valid_targets_min": 1518
|
|
},
|
|
{
|
|
"epoch": 1.7879396984924623,
|
|
"grad_norm": 0.6713679303498401,
|
|
"learning_rate": 3.714603900888215e-05,
|
|
"loss": 0.2834,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12691420316696167,
|
|
"step": 890,
|
|
"valid_targets_mean": 1899.2,
|
|
"valid_targets_min": 1724
|
|
},
|
|
{
|
|
"epoch": 1.7979899497487437,
|
|
"grad_norm": 0.6285372671632026,
|
|
"learning_rate": 3.7094267509742535e-05,
|
|
"loss": 0.2788,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11492075026035309,
|
|
"step": 895,
|
|
"valid_targets_mean": 1780.6,
|
|
"valid_targets_min": 1619
|
|
},
|
|
{
|
|
"epoch": 1.8080402010050252,
|
|
"grad_norm": 0.6749813833829235,
|
|
"learning_rate": 3.704206740257103e-05,
|
|
"loss": 0.2773,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12661439180374146,
|
|
"step": 900,
|
|
"valid_targets_mean": 1918.4,
|
|
"valid_targets_min": 1655
|
|
},
|
|
{
|
|
"epoch": 1.8180904522613064,
|
|
"grad_norm": 0.6620991395459196,
|
|
"learning_rate": 3.6989439996191514e-05,
|
|
"loss": 0.2736,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15227796137332916,
|
|
"step": 905,
|
|
"valid_targets_mean": 2056.1,
|
|
"valid_targets_min": 1380
|
|
},
|
|
{
|
|
"epoch": 1.828140703517588,
|
|
"grad_norm": 0.6714934537155701,
|
|
"learning_rate": 3.6936386610141585e-05,
|
|
"loss": 0.2776,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15056657791137695,
|
|
"step": 910,
|
|
"valid_targets_mean": 1960.9,
|
|
"valid_targets_min": 1764
|
|
},
|
|
{
|
|
"epoch": 1.8381909547738693,
|
|
"grad_norm": 0.6353364040125997,
|
|
"learning_rate": 3.6882908574639566e-05,
|
|
"loss": 0.2871,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15501555800437927,
|
|
"step": 915,
|
|
"valid_targets_mean": 2174.6,
|
|
"valid_targets_min": 1771
|
|
},
|
|
{
|
|
"epoch": 1.8482412060301507,
|
|
"grad_norm": 0.6603064463776511,
|
|
"learning_rate": 3.6829007230551054e-05,
|
|
"loss": 0.2788,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1511480212211609,
|
|
"step": 920,
|
|
"valid_targets_mean": 1987.2,
|
|
"valid_targets_min": 1622
|
|
},
|
|
{
|
|
"epoch": 1.8582914572864322,
|
|
"grad_norm": 0.6867490260724767,
|
|
"learning_rate": 3.6774683929355374e-05,
|
|
"loss": 0.2639,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13275766372680664,
|
|
"step": 925,
|
|
"valid_targets_mean": 1943.4,
|
|
"valid_targets_min": 1605
|
|
},
|
|
{
|
|
"epoch": 1.8683417085427134,
|
|
"grad_norm": 0.6281476059052868,
|
|
"learning_rate": 3.671994003311165e-05,
|
|
"loss": 0.2784,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13019250333309174,
|
|
"step": 930,
|
|
"valid_targets_mean": 1938.0,
|
|
"valid_targets_min": 1827
|
|
},
|
|
{
|
|
"epoch": 1.878391959798995,
|
|
"grad_norm": 0.620820239950261,
|
|
"learning_rate": 3.66647769144247e-05,
|
|
"loss": 0.2718,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12289062887430191,
|
|
"step": 935,
|
|
"valid_targets_mean": 1825.9,
|
|
"valid_targets_min": 1614
|
|
},
|
|
{
|
|
"epoch": 1.8884422110552763,
|
|
"grad_norm": 0.6408145658291278,
|
|
"learning_rate": 3.6609195956410564e-05,
|
|
"loss": 0.2777,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1174338161945343,
|
|
"step": 940,
|
|
"valid_targets_mean": 1904.1,
|
|
"valid_targets_min": 1699
|
|
},
|
|
{
|
|
"epoch": 1.8984924623115578,
|
|
"grad_norm": 0.6289373075761888,
|
|
"learning_rate": 3.6553198552661855e-05,
|
|
"loss": 0.276,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1556728184223175,
|
|
"step": 945,
|
|
"valid_targets_mean": 2098.9,
|
|
"valid_targets_min": 1700
|
|
},
|
|
{
|
|
"epoch": 1.9085427135678392,
|
|
"grad_norm": 0.6721306869593799,
|
|
"learning_rate": 3.649678610721283e-05,
|
|
"loss": 0.2612,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13502375781536102,
|
|
"step": 950,
|
|
"valid_targets_mean": 1885.4,
|
|
"valid_targets_min": 1516
|
|
},
|
|
{
|
|
"epoch": 1.9185929648241205,
|
|
"grad_norm": 0.6399475509949627,
|
|
"learning_rate": 3.643996003450416e-05,
|
|
"loss": 0.2807,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13162364065647125,
|
|
"step": 955,
|
|
"valid_targets_mean": 1937.5,
|
|
"valid_targets_min": 1715
|
|
},
|
|
{
|
|
"epoch": 1.9286432160804021,
|
|
"grad_norm": 0.6484383090102106,
|
|
"learning_rate": 3.638272175934746e-05,
|
|
"loss": 0.2814,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14255571365356445,
|
|
"step": 960,
|
|
"valid_targets_mean": 2027.4,
|
|
"valid_targets_min": 1789
|
|
},
|
|
{
|
|
"epoch": 1.9386934673366834,
|
|
"grad_norm": 0.6844201818433822,
|
|
"learning_rate": 3.632507271688962e-05,
|
|
"loss": 0.2644,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1373480260372162,
|
|
"step": 965,
|
|
"valid_targets_mean": 1904.0,
|
|
"valid_targets_min": 1444
|
|
},
|
|
{
|
|
"epoch": 1.9487437185929648,
|
|
"grad_norm": 0.7019606611650521,
|
|
"learning_rate": 3.626701435257674e-05,
|
|
"loss": 0.2799,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1427798569202423,
|
|
"step": 970,
|
|
"valid_targets_mean": 1930.2,
|
|
"valid_targets_min": 1712
|
|
},
|
|
{
|
|
"epoch": 1.9587939698492463,
|
|
"grad_norm": 0.6909414786539319,
|
|
"learning_rate": 3.620854812211797e-05,
|
|
"loss": 0.2666,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11826258897781372,
|
|
"step": 975,
|
|
"valid_targets_mean": 1774.6,
|
|
"valid_targets_min": 1479
|
|
},
|
|
{
|
|
"epoch": 1.9688442211055275,
|
|
"grad_norm": 0.6276077163244973,
|
|
"learning_rate": 3.614967549144893e-05,
|
|
"loss": 0.2812,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14596611261367798,
|
|
"step": 980,
|
|
"valid_targets_mean": 2121.4,
|
|
"valid_targets_min": 1733
|
|
},
|
|
{
|
|
"epoch": 1.9788944723618092,
|
|
"grad_norm": 0.6602503602897184,
|
|
"learning_rate": 3.609039793669502e-05,
|
|
"loss": 0.2825,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14816994965076447,
|
|
"step": 985,
|
|
"valid_targets_mean": 2045.0,
|
|
"valid_targets_min": 1686
|
|
},
|
|
{
|
|
"epoch": 1.9889447236180904,
|
|
"grad_norm": 0.656463668548065,
|
|
"learning_rate": 3.603071694413438e-05,
|
|
"loss": 0.2832,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1370791792869568,
|
|
"step": 990,
|
|
"valid_targets_mean": 1928.4,
|
|
"valid_targets_min": 1460
|
|
},
|
|
{
|
|
"epoch": 1.9989949748743718,
|
|
"grad_norm": 0.6357735414404356,
|
|
"learning_rate": 3.597063401016062e-05,
|
|
"loss": 0.2819,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1692444384098053,
|
|
"step": 995,
|
|
"valid_targets_mean": 2108.0,
|
|
"valid_targets_min": 1684
|
|
},
|
|
{
|
|
"epoch": 2.0080402010050253,
|
|
"grad_norm": 0.6425471565377394,
|
|
"learning_rate": 3.591015064124532e-05,
|
|
"loss": 0.2628,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13621094822883606,
|
|
"step": 1000,
|
|
"valid_targets_mean": 2013.2,
|
|
"valid_targets_min": 1565
|
|
},
|
|
{
|
|
"epoch": 2.0180904522613066,
|
|
"grad_norm": 0.62954499152394,
|
|
"learning_rate": 3.584926835390023e-05,
|
|
"loss": 0.2666,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1283409297466278,
|
|
"step": 1005,
|
|
"valid_targets_mean": 2001.6,
|
|
"valid_targets_min": 1773
|
|
},
|
|
{
|
|
"epoch": 2.028140703517588,
|
|
"grad_norm": 0.6619594668703057,
|
|
"learning_rate": 3.57879886746393e-05,
|
|
"loss": 0.2677,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14071981608867645,
|
|
"step": 1010,
|
|
"valid_targets_mean": 2026.1,
|
|
"valid_targets_min": 1565
|
|
},
|
|
{
|
|
"epoch": 2.0381909547738695,
|
|
"grad_norm": 0.626915100125346,
|
|
"learning_rate": 3.5726313139940305e-05,
|
|
"loss": 0.2679,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1505325734615326,
|
|
"step": 1015,
|
|
"valid_targets_mean": 2115.1,
|
|
"valid_targets_min": 1649
|
|
},
|
|
{
|
|
"epoch": 2.0482412060301507,
|
|
"grad_norm": 0.6407113566574509,
|
|
"learning_rate": 3.566424329620645e-05,
|
|
"loss": 0.2671,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10944472998380661,
|
|
"step": 1020,
|
|
"valid_targets_mean": 1880.6,
|
|
"valid_targets_min": 1657
|
|
},
|
|
{
|
|
"epoch": 2.0582914572864324,
|
|
"grad_norm": 0.6701245966688966,
|
|
"learning_rate": 3.5601780699727474e-05,
|
|
"loss": 0.2586,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11160222440958023,
|
|
"step": 1025,
|
|
"valid_targets_mean": 1840.0,
|
|
"valid_targets_min": 1570
|
|
},
|
|
{
|
|
"epoch": 2.0683417085427136,
|
|
"grad_norm": 0.6378202318058058,
|
|
"learning_rate": 3.553892691664072e-05,
|
|
"loss": 0.2581,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12387561798095703,
|
|
"step": 1030,
|
|
"valid_targets_mean": 1978.5,
|
|
"valid_targets_min": 1598
|
|
},
|
|
{
|
|
"epoch": 2.078391959798995,
|
|
"grad_norm": 0.6612123339696645,
|
|
"learning_rate": 3.547568352289183e-05,
|
|
"loss": 0.2596,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12964512407779694,
|
|
"step": 1035,
|
|
"valid_targets_mean": 1919.0,
|
|
"valid_targets_min": 1613
|
|
},
|
|
{
|
|
"epoch": 2.0884422110552765,
|
|
"grad_norm": 0.6262576438121411,
|
|
"learning_rate": 3.5412052104195225e-05,
|
|
"loss": 0.2563,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13986793160438538,
|
|
"step": 1040,
|
|
"valid_targets_mean": 1985.0,
|
|
"valid_targets_min": 1601
|
|
},
|
|
{
|
|
"epoch": 2.0984924623115577,
|
|
"grad_norm": 0.689544791477732,
|
|
"learning_rate": 3.534803425599435e-05,
|
|
"loss": 0.2673,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12940703332424164,
|
|
"step": 1045,
|
|
"valid_targets_mean": 1960.1,
|
|
"valid_targets_min": 1638
|
|
},
|
|
{
|
|
"epoch": 2.1085427135678394,
|
|
"grad_norm": 0.6278057721979309,
|
|
"learning_rate": 3.5283631583421685e-05,
|
|
"loss": 0.257,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10866901278495789,
|
|
"step": 1050,
|
|
"valid_targets_mean": 1862.4,
|
|
"valid_targets_min": 1645
|
|
},
|
|
{
|
|
"epoch": 2.1185929648241206,
|
|
"grad_norm": 0.6560565345460811,
|
|
"learning_rate": 3.521884570125848e-05,
|
|
"loss": 0.2743,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14260122179985046,
|
|
"step": 1055,
|
|
"valid_targets_mean": 2141.5,
|
|
"valid_targets_min": 1868
|
|
},
|
|
{
|
|
"epoch": 2.128643216080402,
|
|
"grad_norm": 0.6630291051665875,
|
|
"learning_rate": 3.5153678233894286e-05,
|
|
"loss": 0.2578,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1296427845954895,
|
|
"step": 1060,
|
|
"valid_targets_mean": 2008.6,
|
|
"valid_targets_min": 1599
|
|
},
|
|
{
|
|
"epoch": 2.1386934673366835,
|
|
"grad_norm": 0.6085943303856464,
|
|
"learning_rate": 3.50881308152862e-05,
|
|
"loss": 0.2666,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12743543088436127,
|
|
"step": 1065,
|
|
"valid_targets_mean": 1855.6,
|
|
"valid_targets_min": 1307
|
|
},
|
|
{
|
|
"epoch": 2.1487437185929648,
|
|
"grad_norm": 0.6334171563021671,
|
|
"learning_rate": 3.502220508891795e-05,
|
|
"loss": 0.2605,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14165031909942627,
|
|
"step": 1070,
|
|
"valid_targets_mean": 2056.4,
|
|
"valid_targets_min": 1831
|
|
},
|
|
{
|
|
"epoch": 2.1587939698492464,
|
|
"grad_norm": 0.6751921247405751,
|
|
"learning_rate": 3.495590270775861e-05,
|
|
"loss": 0.2596,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1233670562505722,
|
|
"step": 1075,
|
|
"valid_targets_mean": 1940.5,
|
|
"valid_targets_min": 1617
|
|
},
|
|
{
|
|
"epoch": 2.1688442211055277,
|
|
"grad_norm": 0.6586550623918564,
|
|
"learning_rate": 3.488922533422121e-05,
|
|
"loss": 0.266,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13168886303901672,
|
|
"step": 1080,
|
|
"valid_targets_mean": 1975.5,
|
|
"valid_targets_min": 1552
|
|
},
|
|
{
|
|
"epoch": 2.178894472361809,
|
|
"grad_norm": 0.6413425021249286,
|
|
"learning_rate": 3.4822174640121056e-05,
|
|
"loss": 0.2657,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13131633400917053,
|
|
"step": 1085,
|
|
"valid_targets_mean": 1999.2,
|
|
"valid_targets_min": 1691
|
|
},
|
|
{
|
|
"epoch": 2.1889447236180906,
|
|
"grad_norm": 0.6825410712421459,
|
|
"learning_rate": 3.475475230663377e-05,
|
|
"loss": 0.2616,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1298404037952423,
|
|
"step": 1090,
|
|
"valid_targets_mean": 1850.8,
|
|
"valid_targets_min": 1421
|
|
},
|
|
{
|
|
"epoch": 2.198994974874372,
|
|
"grad_norm": 0.7039998586056406,
|
|
"learning_rate": 3.468696002425318e-05,
|
|
"loss": 0.2614,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11604462563991547,
|
|
"step": 1095,
|
|
"valid_targets_mean": 1813.8,
|
|
"valid_targets_min": 1496
|
|
},
|
|
{
|
|
"epoch": 2.2090452261306535,
|
|
"grad_norm": 0.6245234379262704,
|
|
"learning_rate": 3.461879949274893e-05,
|
|
"loss": 0.2629,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13127946853637695,
|
|
"step": 1100,
|
|
"valid_targets_mean": 1998.4,
|
|
"valid_targets_min": 1743
|
|
},
|
|
{
|
|
"epoch": 2.2190954773869347,
|
|
"grad_norm": 0.6542326269088536,
|
|
"learning_rate": 3.4550272421123806e-05,
|
|
"loss": 0.2568,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1270817220211029,
|
|
"step": 1105,
|
|
"valid_targets_mean": 1940.2,
|
|
"valid_targets_min": 1611
|
|
},
|
|
{
|
|
"epoch": 2.229145728643216,
|
|
"grad_norm": 0.6001275878330509,
|
|
"learning_rate": 3.448138052757098e-05,
|
|
"loss": 0.2633,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1243213415145874,
|
|
"step": 1110,
|
|
"valid_targets_mean": 1955.0,
|
|
"valid_targets_min": 1364
|
|
},
|
|
{
|
|
"epoch": 2.2391959798994976,
|
|
"grad_norm": 0.6387617199605115,
|
|
"learning_rate": 3.4412125539430836e-05,
|
|
"loss": 0.263,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13336977362632751,
|
|
"step": 1115,
|
|
"valid_targets_mean": 1978.2,
|
|
"valid_targets_min": 1642
|
|
},
|
|
{
|
|
"epoch": 2.249246231155779,
|
|
"grad_norm": 0.6540337659695888,
|
|
"learning_rate": 3.434250919314773e-05,
|
|
"loss": 0.2673,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13685552775859833,
|
|
"step": 1120,
|
|
"valid_targets_mean": 1862.2,
|
|
"valid_targets_min": 1674
|
|
},
|
|
{
|
|
"epoch": 2.2592964824120605,
|
|
"grad_norm": 0.5985014800434116,
|
|
"learning_rate": 3.427253323422641e-05,
|
|
"loss": 0.265,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13089847564697266,
|
|
"step": 1125,
|
|
"valid_targets_mean": 1939.4,
|
|
"valid_targets_min": 1570
|
|
},
|
|
{
|
|
"epoch": 2.2693467336683417,
|
|
"grad_norm": 0.692692928410177,
|
|
"learning_rate": 3.420219941718826e-05,
|
|
"loss": 0.263,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1436246931552887,
|
|
"step": 1130,
|
|
"valid_targets_mean": 2045.0,
|
|
"valid_targets_min": 1482
|
|
},
|
|
{
|
|
"epoch": 2.279396984924623,
|
|
"grad_norm": 0.6294661003433245,
|
|
"learning_rate": 3.413150950552734e-05,
|
|
"loss": 0.2681,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1222902163863182,
|
|
"step": 1135,
|
|
"valid_targets_mean": 1902.1,
|
|
"valid_targets_min": 1344
|
|
},
|
|
{
|
|
"epoch": 2.2894472361809046,
|
|
"grad_norm": 0.6829315031838317,
|
|
"learning_rate": 3.406046527166612e-05,
|
|
"loss": 0.26,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1552143096923828,
|
|
"step": 1140,
|
|
"valid_targets_mean": 1961.6,
|
|
"valid_targets_min": 1675
|
|
},
|
|
{
|
|
"epoch": 2.299497487437186,
|
|
"grad_norm": 0.6159368854803465,
|
|
"learning_rate": 3.3989068496911054e-05,
|
|
"loss": 0.2472,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1455489993095398,
|
|
"step": 1145,
|
|
"valid_targets_mean": 2109.8,
|
|
"valid_targets_min": 1642
|
|
},
|
|
{
|
|
"epoch": 2.3095477386934675,
|
|
"grad_norm": 0.6642521354586247,
|
|
"learning_rate": 3.391732097140795e-05,
|
|
"loss": 0.2635,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12750039994716644,
|
|
"step": 1150,
|
|
"valid_targets_mean": 1868.2,
|
|
"valid_targets_min": 602
|
|
},
|
|
{
|
|
"epoch": 2.3195979899497488,
|
|
"grad_norm": 0.6454262025628663,
|
|
"learning_rate": 3.384522449409704e-05,
|
|
"loss": 0.2631,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11974665522575378,
|
|
"step": 1155,
|
|
"valid_targets_mean": 1827.2,
|
|
"valid_targets_min": 1440
|
|
},
|
|
{
|
|
"epoch": 2.32964824120603,
|
|
"grad_norm": 0.6885980282135313,
|
|
"learning_rate": 3.37727808726679e-05,
|
|
"loss": 0.2697,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1286284625530243,
|
|
"step": 1160,
|
|
"valid_targets_mean": 1934.6,
|
|
"valid_targets_min": 1521
|
|
},
|
|
{
|
|
"epoch": 2.3396984924623117,
|
|
"grad_norm": 0.6870260068241173,
|
|
"learning_rate": 3.3699991923514135e-05,
|
|
"loss": 0.2582,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14052963256835938,
|
|
"step": 1165,
|
|
"valid_targets_mean": 2010.8,
|
|
"valid_targets_min": 1454
|
|
},
|
|
{
|
|
"epoch": 2.349748743718593,
|
|
"grad_norm": 0.6412936300219414,
|
|
"learning_rate": 3.36268594716878e-05,
|
|
"loss": 0.2617,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13330818712711334,
|
|
"step": 1170,
|
|
"valid_targets_mean": 1950.0,
|
|
"valid_targets_min": 1703
|
|
},
|
|
{
|
|
"epoch": 2.3597989949748746,
|
|
"grad_norm": 0.6499185682065258,
|
|
"learning_rate": 3.3553385350853686e-05,
|
|
"loss": 0.2687,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1161436066031456,
|
|
"step": 1175,
|
|
"valid_targets_mean": 1956.2,
|
|
"valid_targets_min": 1625
|
|
},
|
|
{
|
|
"epoch": 2.369849246231156,
|
|
"grad_norm": 0.6741478456713295,
|
|
"learning_rate": 3.34795714032433e-05,
|
|
"loss": 0.2636,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15226960182189941,
|
|
"step": 1180,
|
|
"valid_targets_mean": 2083.1,
|
|
"valid_targets_min": 1495
|
|
},
|
|
{
|
|
"epoch": 2.379899497487437,
|
|
"grad_norm": 0.6343620369077736,
|
|
"learning_rate": 3.340541947960871e-05,
|
|
"loss": 0.2651,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12335594743490219,
|
|
"step": 1185,
|
|
"valid_targets_mean": 1796.5,
|
|
"valid_targets_min": 1379
|
|
},
|
|
{
|
|
"epoch": 2.3899497487437187,
|
|
"grad_norm": 0.6590074152258247,
|
|
"learning_rate": 3.33309314391761e-05,
|
|
"loss": 0.2612,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12662598490715027,
|
|
"step": 1190,
|
|
"valid_targets_mean": 1924.2,
|
|
"valid_targets_min": 1681
|
|
},
|
|
{
|
|
"epoch": 2.4,
|
|
"grad_norm": 0.636427549779774,
|
|
"learning_rate": 3.325610914959922e-05,
|
|
"loss": 0.2681,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11519978940486908,
|
|
"step": 1195,
|
|
"valid_targets_mean": 1903.0,
|
|
"valid_targets_min": 1694
|
|
},
|
|
{
|
|
"epoch": 2.4100502512562816,
|
|
"grad_norm": 0.6589766266394599,
|
|
"learning_rate": 3.3180954486912465e-05,
|
|
"loss": 0.2662,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13555380702018738,
|
|
"step": 1200,
|
|
"valid_targets_mean": 1978.8,
|
|
"valid_targets_min": 1756
|
|
},
|
|
{
|
|
"epoch": 2.420100502512563,
|
|
"grad_norm": 0.6309382193896093,
|
|
"learning_rate": 3.310546933548394e-05,
|
|
"loss": 0.28,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13470861315727234,
|
|
"step": 1205,
|
|
"valid_targets_mean": 1913.2,
|
|
"valid_targets_min": 1593
|
|
},
|
|
{
|
|
"epoch": 2.430150753768844,
|
|
"grad_norm": 0.7026787740666961,
|
|
"learning_rate": 3.302965558796812e-05,
|
|
"loss": 0.2562,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13509687781333923,
|
|
"step": 1210,
|
|
"valid_targets_mean": 2025.9,
|
|
"valid_targets_min": 1464
|
|
},
|
|
{
|
|
"epoch": 2.4402010050251257,
|
|
"grad_norm": 0.7217664434939466,
|
|
"learning_rate": 3.295351514525847e-05,
|
|
"loss": 0.2656,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13727162778377533,
|
|
"step": 1215,
|
|
"valid_targets_mean": 1900.1,
|
|
"valid_targets_min": 1462
|
|
},
|
|
{
|
|
"epoch": 2.450251256281407,
|
|
"grad_norm": 0.6513251212853165,
|
|
"learning_rate": 3.287704991643972e-05,
|
|
"loss": 0.2516,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13785016536712646,
|
|
"step": 1220,
|
|
"valid_targets_mean": 2037.9,
|
|
"valid_targets_min": 1828
|
|
},
|
|
{
|
|
"epoch": 2.4603015075376886,
|
|
"grad_norm": 0.6183730614968596,
|
|
"learning_rate": 3.280026181874004e-05,
|
|
"loss": 0.2653,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12823602557182312,
|
|
"step": 1225,
|
|
"valid_targets_mean": 1932.0,
|
|
"valid_targets_min": 1502
|
|
},
|
|
{
|
|
"epoch": 2.47035175879397,
|
|
"grad_norm": 0.6625983274499935,
|
|
"learning_rate": 3.2723152777482964e-05,
|
|
"loss": 0.2518,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1229400709271431,
|
|
"step": 1230,
|
|
"valid_targets_mean": 1897.5,
|
|
"valid_targets_min": 1453
|
|
},
|
|
{
|
|
"epoch": 2.480402010050251,
|
|
"grad_norm": 0.6941902383142403,
|
|
"learning_rate": 3.2645724726039106e-05,
|
|
"loss": 0.2683,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14125655591487885,
|
|
"step": 1235,
|
|
"valid_targets_mean": 2079.4,
|
|
"valid_targets_min": 1741
|
|
},
|
|
{
|
|
"epoch": 2.4904522613065327,
|
|
"grad_norm": 0.6210859835387814,
|
|
"learning_rate": 3.2567979605777676e-05,
|
|
"loss": 0.2663,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13878211379051208,
|
|
"step": 1240,
|
|
"valid_targets_mean": 1904.8,
|
|
"valid_targets_min": 1613
|
|
},
|
|
{
|
|
"epoch": 2.500502512562814,
|
|
"grad_norm": 0.6695104827983686,
|
|
"learning_rate": 3.2489919366017825e-05,
|
|
"loss": 0.2478,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12152104824781418,
|
|
"step": 1245,
|
|
"valid_targets_mean": 1897.6,
|
|
"valid_targets_min": 1508
|
|
},
|
|
{
|
|
"epoch": 2.5105527638190956,
|
|
"grad_norm": 0.6519486178952696,
|
|
"learning_rate": 3.241154596397975e-05,
|
|
"loss": 0.2657,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14372539520263672,
|
|
"step": 1250,
|
|
"valid_targets_mean": 2076.8,
|
|
"valid_targets_min": 1636
|
|
},
|
|
{
|
|
"epoch": 2.520603015075377,
|
|
"grad_norm": 0.5571963001965639,
|
|
"learning_rate": 3.233286136473565e-05,
|
|
"loss": 0.2547,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1333080232143402,
|
|
"step": 1255,
|
|
"valid_targets_mean": 2029.2,
|
|
"valid_targets_min": 1609
|
|
},
|
|
{
|
|
"epoch": 2.530653266331658,
|
|
"grad_norm": 0.649439324697005,
|
|
"learning_rate": 3.2253867541160415e-05,
|
|
"loss": 0.2643,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11850276589393616,
|
|
"step": 1260,
|
|
"valid_targets_mean": 1898.0,
|
|
"valid_targets_min": 1741
|
|
},
|
|
{
|
|
"epoch": 2.5407035175879398,
|
|
"grad_norm": 0.6065230851074052,
|
|
"learning_rate": 3.217456647388217e-05,
|
|
"loss": 0.2621,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10484498739242554,
|
|
"step": 1265,
|
|
"valid_targets_mean": 1816.8,
|
|
"valid_targets_min": 1394
|
|
},
|
|
{
|
|
"epoch": 2.550753768844221,
|
|
"grad_norm": 0.6277850375614765,
|
|
"learning_rate": 3.209496015123266e-05,
|
|
"loss": 0.2599,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11773302406072617,
|
|
"step": 1270,
|
|
"valid_targets_mean": 1848.2,
|
|
"valid_targets_min": 1582
|
|
},
|
|
{
|
|
"epoch": 2.5608040201005027,
|
|
"grad_norm": 0.6323388699610939,
|
|
"learning_rate": 3.2015050569197326e-05,
|
|
"loss": 0.2602,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12265069782733917,
|
|
"step": 1275,
|
|
"valid_targets_mean": 1898.8,
|
|
"valid_targets_min": 1554
|
|
},
|
|
{
|
|
"epoch": 2.570854271356784,
|
|
"grad_norm": 0.6167972202418391,
|
|
"learning_rate": 3.1934839731365304e-05,
|
|
"loss": 0.254,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1358634978532791,
|
|
"step": 1280,
|
|
"valid_targets_mean": 1851.9,
|
|
"valid_targets_min": 1678
|
|
},
|
|
{
|
|
"epoch": 2.580904522613065,
|
|
"grad_norm": 0.6213433721128977,
|
|
"learning_rate": 3.185432964887918e-05,
|
|
"loss": 0.2613,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13280673325061798,
|
|
"step": 1285,
|
|
"valid_targets_mean": 2021.1,
|
|
"valid_targets_min": 1657
|
|
},
|
|
{
|
|
"epoch": 2.590954773869347,
|
|
"grad_norm": 0.6195187467854718,
|
|
"learning_rate": 3.1773522340384564e-05,
|
|
"loss": 0.2573,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15085375308990479,
|
|
"step": 1290,
|
|
"valid_targets_mean": 2189.6,
|
|
"valid_targets_min": 1719
|
|
},
|
|
{
|
|
"epoch": 2.601005025125628,
|
|
"grad_norm": 0.6320639775257483,
|
|
"learning_rate": 3.1692419831979464e-05,
|
|
"loss": 0.2645,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1646469235420227,
|
|
"step": 1295,
|
|
"valid_targets_mean": 2113.8,
|
|
"valid_targets_min": 1693
|
|
},
|
|
{
|
|
"epoch": 2.6110552763819097,
|
|
"grad_norm": 0.6510337505123375,
|
|
"learning_rate": 3.161102415716351e-05,
|
|
"loss": 0.2711,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11679929494857788,
|
|
"step": 1300,
|
|
"valid_targets_mean": 1857.4,
|
|
"valid_targets_min": 1582
|
|
},
|
|
{
|
|
"epoch": 2.621105527638191,
|
|
"grad_norm": 0.6303881484028009,
|
|
"learning_rate": 3.152933735678693e-05,
|
|
"loss": 0.2578,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13267043232917786,
|
|
"step": 1305,
|
|
"valid_targets_mean": 2001.0,
|
|
"valid_targets_min": 1696
|
|
},
|
|
{
|
|
"epoch": 2.631155778894472,
|
|
"grad_norm": 0.6351202817609105,
|
|
"learning_rate": 3.144736147899943e-05,
|
|
"loss": 0.2566,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13581131398677826,
|
|
"step": 1310,
|
|
"valid_targets_mean": 2019.5,
|
|
"valid_targets_min": 1757
|
|
},
|
|
{
|
|
"epoch": 2.641206030150754,
|
|
"grad_norm": 0.6014925901916426,
|
|
"learning_rate": 3.1365098579198795e-05,
|
|
"loss": 0.2478,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12695062160491943,
|
|
"step": 1315,
|
|
"valid_targets_mean": 1984.6,
|
|
"valid_targets_min": 1719
|
|
},
|
|
{
|
|
"epoch": 2.651256281407035,
|
|
"grad_norm": 0.656736880078342,
|
|
"learning_rate": 3.128255071997937e-05,
|
|
"loss": 0.2553,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14483338594436646,
|
|
"step": 1320,
|
|
"valid_targets_mean": 2072.8,
|
|
"valid_targets_min": 1701
|
|
},
|
|
{
|
|
"epoch": 2.6613065326633167,
|
|
"grad_norm": 0.672082025018077,
|
|
"learning_rate": 3.119971997108036e-05,
|
|
"loss": 0.2617,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11736010760068893,
|
|
"step": 1325,
|
|
"valid_targets_mean": 1753.8,
|
|
"valid_targets_min": 857
|
|
},
|
|
{
|
|
"epoch": 2.671356783919598,
|
|
"grad_norm": 0.6728470521211699,
|
|
"learning_rate": 3.11166084093339e-05,
|
|
"loss": 0.2566,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12078943848609924,
|
|
"step": 1330,
|
|
"valid_targets_mean": 1868.0,
|
|
"valid_targets_min": 1495
|
|
},
|
|
{
|
|
"epoch": 2.681407035175879,
|
|
"grad_norm": 0.6283285919355515,
|
|
"learning_rate": 3.1033218118613014e-05,
|
|
"loss": 0.256,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1286976933479309,
|
|
"step": 1335,
|
|
"valid_targets_mean": 1942.8,
|
|
"valid_targets_min": 1513
|
|
},
|
|
{
|
|
"epoch": 2.691457286432161,
|
|
"grad_norm": 0.6546498028202181,
|
|
"learning_rate": 3.094955118977936e-05,
|
|
"loss": 0.258,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12801681458950043,
|
|
"step": 1340,
|
|
"valid_targets_mean": 1977.6,
|
|
"valid_targets_min": 1584
|
|
},
|
|
{
|
|
"epoch": 2.701507537688442,
|
|
"grad_norm": 0.6418524471385488,
|
|
"learning_rate": 3.086560972063078e-05,
|
|
"loss": 0.2572,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1460198611021042,
|
|
"step": 1345,
|
|
"valid_targets_mean": 1949.5,
|
|
"valid_targets_min": 1724
|
|
},
|
|
{
|
|
"epoch": 2.7115577889447238,
|
|
"grad_norm": 0.6521259679259458,
|
|
"learning_rate": 3.078139581584875e-05,
|
|
"loss": 0.265,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13334403932094574,
|
|
"step": 1350,
|
|
"valid_targets_mean": 1876.1,
|
|
"valid_targets_min": 1544
|
|
},
|
|
{
|
|
"epoch": 2.721608040201005,
|
|
"grad_norm": 0.5778779786879445,
|
|
"learning_rate": 3.069691158694554e-05,
|
|
"loss": 0.2637,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13477079570293427,
|
|
"step": 1355,
|
|
"valid_targets_mean": 1890.5,
|
|
"valid_targets_min": 1507
|
|
},
|
|
{
|
|
"epoch": 2.7316582914572862,
|
|
"grad_norm": 0.6174488223424366,
|
|
"learning_rate": 3.0612159152211366e-05,
|
|
"loss": 0.2592,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1376238763332367,
|
|
"step": 1360,
|
|
"valid_targets_mean": 1879.2,
|
|
"valid_targets_min": 1430
|
|
},
|
|
{
|
|
"epoch": 2.741708542713568,
|
|
"grad_norm": 0.6679626685834315,
|
|
"learning_rate": 3.052714063666116e-05,
|
|
"loss": 0.263,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12584970891475677,
|
|
"step": 1365,
|
|
"valid_targets_mean": 1821.8,
|
|
"valid_targets_min": 341
|
|
},
|
|
{
|
|
"epoch": 2.751758793969849,
|
|
"grad_norm": 0.5761355734133754,
|
|
"learning_rate": 3.0441858171981406e-05,
|
|
"loss": 0.2462,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10995428264141083,
|
|
"step": 1370,
|
|
"valid_targets_mean": 1901.8,
|
|
"valid_targets_min": 1520
|
|
},
|
|
{
|
|
"epoch": 2.761809045226131,
|
|
"grad_norm": 0.6156256763512853,
|
|
"learning_rate": 3.0356313896476607e-05,
|
|
"loss": 0.2592,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13224874436855316,
|
|
"step": 1375,
|
|
"valid_targets_mean": 1887.4,
|
|
"valid_targets_min": 1625
|
|
},
|
|
{
|
|
"epoch": 2.771859296482412,
|
|
"grad_norm": 0.59180693972894,
|
|
"learning_rate": 3.0270509955015705e-05,
|
|
"loss": 0.2605,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11754358559846878,
|
|
"step": 1380,
|
|
"valid_targets_mean": 1890.8,
|
|
"valid_targets_min": 1627
|
|
},
|
|
{
|
|
"epoch": 2.7819095477386933,
|
|
"grad_norm": 0.590961071360744,
|
|
"learning_rate": 3.018444849897831e-05,
|
|
"loss": 0.2687,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12024317681789398,
|
|
"step": 1385,
|
|
"valid_targets_mean": 1859.1,
|
|
"valid_targets_min": 1495
|
|
},
|
|
{
|
|
"epoch": 2.791959798994975,
|
|
"grad_norm": 0.6089932480541909,
|
|
"learning_rate": 3.0098131686200747e-05,
|
|
"loss": 0.2701,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11352679133415222,
|
|
"step": 1390,
|
|
"valid_targets_mean": 1919.8,
|
|
"valid_targets_min": 1579
|
|
},
|
|
{
|
|
"epoch": 2.802010050251256,
|
|
"grad_norm": 0.6225133532165678,
|
|
"learning_rate": 3.0011561680921942e-05,
|
|
"loss": 0.2621,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1367279291152954,
|
|
"step": 1395,
|
|
"valid_targets_mean": 1992.0,
|
|
"valid_targets_min": 1540
|
|
},
|
|
{
|
|
"epoch": 2.812060301507538,
|
|
"grad_norm": 0.6336598113824402,
|
|
"learning_rate": 2.9924740653729167e-05,
|
|
"loss": 0.2803,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12791481614112854,
|
|
"step": 1400,
|
|
"valid_targets_mean": 1873.8,
|
|
"valid_targets_min": 1518
|
|
},
|
|
{
|
|
"epoch": 2.822110552763819,
|
|
"grad_norm": 0.6636702143322287,
|
|
"learning_rate": 2.9837670781503634e-05,
|
|
"loss": 0.262,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13727378845214844,
|
|
"step": 1405,
|
|
"valid_targets_mean": 1982.2,
|
|
"valid_targets_min": 1679
|
|
},
|
|
{
|
|
"epoch": 2.8321608040201003,
|
|
"grad_norm": 0.6259550650289792,
|
|
"learning_rate": 2.9750354247365875e-05,
|
|
"loss": 0.2585,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14261868596076965,
|
|
"step": 1410,
|
|
"valid_targets_mean": 1927.6,
|
|
"valid_targets_min": 1663
|
|
},
|
|
{
|
|
"epoch": 2.842211055276382,
|
|
"grad_norm": 0.5798568660477534,
|
|
"learning_rate": 2.9662793240621055e-05,
|
|
"loss": 0.2623,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12819689512252808,
|
|
"step": 1415,
|
|
"valid_targets_mean": 1882.1,
|
|
"valid_targets_min": 1396
|
|
},
|
|
{
|
|
"epoch": 2.852261306532663,
|
|
"grad_norm": 0.6447761382097079,
|
|
"learning_rate": 2.9574989956704012e-05,
|
|
"loss": 0.2689,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1498211920261383,
|
|
"step": 1420,
|
|
"valid_targets_mean": 1990.5,
|
|
"valid_targets_min": 1644
|
|
},
|
|
{
|
|
"epoch": 2.862311557788945,
|
|
"grad_norm": 0.6415597606071446,
|
|
"learning_rate": 2.9486946597124282e-05,
|
|
"loss": 0.2621,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14596256613731384,
|
|
"step": 1425,
|
|
"valid_targets_mean": 2007.5,
|
|
"valid_targets_min": 1721
|
|
},
|
|
{
|
|
"epoch": 2.872361809045226,
|
|
"grad_norm": 0.5763319477062853,
|
|
"learning_rate": 2.939866536941085e-05,
|
|
"loss": 0.2564,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13526159524917603,
|
|
"step": 1430,
|
|
"valid_targets_mean": 2127.0,
|
|
"valid_targets_min": 1743
|
|
},
|
|
{
|
|
"epoch": 2.8824120603015073,
|
|
"grad_norm": 0.6352845612049506,
|
|
"learning_rate": 2.9310148487056817e-05,
|
|
"loss": 0.2569,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1475343257188797,
|
|
"step": 1435,
|
|
"valid_targets_mean": 2079.4,
|
|
"valid_targets_min": 1620
|
|
},
|
|
{
|
|
"epoch": 2.892462311557789,
|
|
"grad_norm": 0.6225803831471683,
|
|
"learning_rate": 2.9221398169463912e-05,
|
|
"loss": 0.2676,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13845807313919067,
|
|
"step": 1440,
|
|
"valid_targets_mean": 2215.0,
|
|
"valid_targets_min": 1851
|
|
},
|
|
{
|
|
"epoch": 2.90251256281407,
|
|
"grad_norm": 0.5994942649917321,
|
|
"learning_rate": 2.913241664188682e-05,
|
|
"loss": 0.2634,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13562408089637756,
|
|
"step": 1445,
|
|
"valid_targets_mean": 2086.8,
|
|
"valid_targets_min": 1690
|
|
},
|
|
{
|
|
"epoch": 2.912562814070352,
|
|
"grad_norm": 0.6398231173726743,
|
|
"learning_rate": 2.9043206135377406e-05,
|
|
"loss": 0.2534,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13872802257537842,
|
|
"step": 1450,
|
|
"valid_targets_mean": 1947.9,
|
|
"valid_targets_min": 1486
|
|
},
|
|
{
|
|
"epoch": 2.922613065326633,
|
|
"grad_norm": 0.6330090035411644,
|
|
"learning_rate": 2.895376888672877e-05,
|
|
"loss": 0.2482,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13541609048843384,
|
|
"step": 1455,
|
|
"valid_targets_mean": 2026.1,
|
|
"valid_targets_min": 1579
|
|
},
|
|
{
|
|
"epoch": 2.9326633165829143,
|
|
"grad_norm": 0.647687889862881,
|
|
"learning_rate": 2.886410713841917e-05,
|
|
"loss": 0.2651,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12788647413253784,
|
|
"step": 1460,
|
|
"valid_targets_mean": 1968.1,
|
|
"valid_targets_min": 1434
|
|
},
|
|
{
|
|
"epoch": 2.942713567839196,
|
|
"grad_norm": 0.6025118595408613,
|
|
"learning_rate": 2.8774223138555772e-05,
|
|
"loss": 0.2545,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14202772080898285,
|
|
"step": 1465,
|
|
"valid_targets_mean": 2176.5,
|
|
"valid_targets_min": 1788
|
|
},
|
|
{
|
|
"epoch": 2.9527638190954772,
|
|
"grad_norm": 0.6165132400133987,
|
|
"learning_rate": 2.8684119140818324e-05,
|
|
"loss": 0.2523,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12659062445163727,
|
|
"step": 1470,
|
|
"valid_targets_mean": 1907.2,
|
|
"valid_targets_min": 1456
|
|
},
|
|
{
|
|
"epoch": 2.962814070351759,
|
|
"grad_norm": 0.6022597969752274,
|
|
"learning_rate": 2.8593797404402598e-05,
|
|
"loss": 0.2601,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11887571215629578,
|
|
"step": 1475,
|
|
"valid_targets_mean": 1894.8,
|
|
"valid_targets_min": 1586
|
|
},
|
|
{
|
|
"epoch": 2.97286432160804,
|
|
"grad_norm": 0.6510784894193359,
|
|
"learning_rate": 2.8503260193963802e-05,
|
|
"loss": 0.2527,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12040414661169052,
|
|
"step": 1480,
|
|
"valid_targets_mean": 1964.2,
|
|
"valid_targets_min": 1625
|
|
},
|
|
{
|
|
"epoch": 2.9829145728643214,
|
|
"grad_norm": 0.5663114333152036,
|
|
"learning_rate": 2.8412509779559737e-05,
|
|
"loss": 0.2509,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10809898376464844,
|
|
"step": 1485,
|
|
"valid_targets_mean": 1888.8,
|
|
"valid_targets_min": 1583
|
|
},
|
|
{
|
|
"epoch": 2.992964824120603,
|
|
"grad_norm": 0.603705972511743,
|
|
"learning_rate": 2.8321548436593942e-05,
|
|
"loss": 0.2633,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13096098601818085,
|
|
"step": 1490,
|
|
"valid_targets_mean": 1917.1,
|
|
"valid_targets_min": 1536
|
|
},
|
|
{
|
|
"epoch": 3.0020100502512563,
|
|
"grad_norm": 0.5921608874650711,
|
|
"learning_rate": 2.823037844575859e-05,
|
|
"loss": 0.2521,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12418641895055771,
|
|
"step": 1495,
|
|
"valid_targets_mean": 1946.9,
|
|
"valid_targets_min": 1673
|
|
},
|
|
{
|
|
"epoch": 3.0120603015075376,
|
|
"grad_norm": 0.6203047903182788,
|
|
"learning_rate": 2.8139002092977317e-05,
|
|
"loss": 0.2476,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14137324690818787,
|
|
"step": 1500,
|
|
"valid_targets_mean": 1920.0,
|
|
"valid_targets_min": 1580
|
|
},
|
|
{
|
|
"epoch": 3.0221105527638192,
|
|
"grad_norm": 0.6370731260837448,
|
|
"learning_rate": 2.804742166934794e-05,
|
|
"loss": 0.2481,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13794197142124176,
|
|
"step": 1505,
|
|
"valid_targets_mean": 2015.4,
|
|
"valid_targets_min": 1726
|
|
},
|
|
{
|
|
"epoch": 3.0321608040201005,
|
|
"grad_norm": 0.6173442988372801,
|
|
"learning_rate": 2.795563947108496e-05,
|
|
"loss": 0.2435,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12482940405607224,
|
|
"step": 1510,
|
|
"valid_targets_mean": 2000.0,
|
|
"valid_targets_min": 1718
|
|
},
|
|
{
|
|
"epoch": 3.0422110552763817,
|
|
"grad_norm": 0.629124345327241,
|
|
"learning_rate": 2.786365779946203e-05,
|
|
"loss": 0.2496,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14233480393886566,
|
|
"step": 1515,
|
|
"valid_targets_mean": 2234.5,
|
|
"valid_targets_min": 1800
|
|
},
|
|
{
|
|
"epoch": 3.0522613065326634,
|
|
"grad_norm": 0.6389566537505625,
|
|
"learning_rate": 2.7771478960754235e-05,
|
|
"loss": 0.2504,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11753445863723755,
|
|
"step": 1520,
|
|
"valid_targets_mean": 1870.0,
|
|
"valid_targets_min": 1561
|
|
},
|
|
{
|
|
"epoch": 3.0623115577889446,
|
|
"grad_norm": 0.6515496588672697,
|
|
"learning_rate": 2.7679105266180274e-05,
|
|
"loss": 0.2546,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1324942260980606,
|
|
"step": 1525,
|
|
"valid_targets_mean": 1988.8,
|
|
"valid_targets_min": 1672
|
|
},
|
|
{
|
|
"epoch": 3.0723618090452263,
|
|
"grad_norm": 0.6106496591710592,
|
|
"learning_rate": 2.7586539031844486e-05,
|
|
"loss": 0.242,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11035766452550888,
|
|
"step": 1530,
|
|
"valid_targets_mean": 1916.9,
|
|
"valid_targets_min": 1588
|
|
},
|
|
{
|
|
"epoch": 3.0824120603015075,
|
|
"grad_norm": 0.6172619317644422,
|
|
"learning_rate": 2.7493782578678837e-05,
|
|
"loss": 0.2508,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11244616657495499,
|
|
"step": 1535,
|
|
"valid_targets_mean": 1920.2,
|
|
"valid_targets_min": 1763
|
|
},
|
|
{
|
|
"epoch": 3.0924623115577887,
|
|
"grad_norm": 0.6778610763187426,
|
|
"learning_rate": 2.7400838232384656e-05,
|
|
"loss": 0.2407,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12935179471969604,
|
|
"step": 1540,
|
|
"valid_targets_mean": 1953.8,
|
|
"valid_targets_min": 1523
|
|
},
|
|
{
|
|
"epoch": 3.1025125628140704,
|
|
"grad_norm": 0.6328361613985454,
|
|
"learning_rate": 2.7307708323374355e-05,
|
|
"loss": 0.2435,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12379435449838638,
|
|
"step": 1545,
|
|
"valid_targets_mean": 1873.9,
|
|
"valid_targets_min": 1651
|
|
},
|
|
{
|
|
"epoch": 3.1125628140703516,
|
|
"grad_norm": 0.581115647151899,
|
|
"learning_rate": 2.721439518671302e-05,
|
|
"loss": 0.2351,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11231249570846558,
|
|
"step": 1550,
|
|
"valid_targets_mean": 1858.4,
|
|
"valid_targets_min": 1609
|
|
},
|
|
{
|
|
"epoch": 3.1226130653266333,
|
|
"grad_norm": 0.624234441718617,
|
|
"learning_rate": 2.7120901162059818e-05,
|
|
"loss": 0.2399,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14058652520179749,
|
|
"step": 1555,
|
|
"valid_targets_mean": 2022.9,
|
|
"valid_targets_min": 1598
|
|
},
|
|
{
|
|
"epoch": 3.1326633165829145,
|
|
"grad_norm": 0.6352073441183151,
|
|
"learning_rate": 2.7027228593609378e-05,
|
|
"loss": 0.2407,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12247593700885773,
|
|
"step": 1560,
|
|
"valid_targets_mean": 1990.5,
|
|
"valid_targets_min": 1687
|
|
},
|
|
{
|
|
"epoch": 3.1427135678391958,
|
|
"grad_norm": 0.603641628905667,
|
|
"learning_rate": 2.6933379830032974e-05,
|
|
"loss": 0.2388,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11993648111820221,
|
|
"step": 1565,
|
|
"valid_targets_mean": 1871.5,
|
|
"valid_targets_min": 1596
|
|
},
|
|
{
|
|
"epoch": 3.1527638190954774,
|
|
"grad_norm": 0.6314421515433125,
|
|
"learning_rate": 2.6839357224419673e-05,
|
|
"loss": 0.2378,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12078208476305008,
|
|
"step": 1570,
|
|
"valid_targets_mean": 1985.9,
|
|
"valid_targets_min": 1517
|
|
},
|
|
{
|
|
"epoch": 3.1628140703517587,
|
|
"grad_norm": 0.6323906318827565,
|
|
"learning_rate": 2.6745163134217302e-05,
|
|
"loss": 0.2368,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11989548802375793,
|
|
"step": 1575,
|
|
"valid_targets_mean": 1946.1,
|
|
"valid_targets_min": 1515
|
|
},
|
|
{
|
|
"epoch": 3.1728643216080403,
|
|
"grad_norm": 0.6082065407920635,
|
|
"learning_rate": 2.6650799921173375e-05,
|
|
"loss": 0.2519,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1227322369813919,
|
|
"step": 1580,
|
|
"valid_targets_mean": 2003.4,
|
|
"valid_targets_min": 1772
|
|
},
|
|
{
|
|
"epoch": 3.1829145728643216,
|
|
"grad_norm": 0.6112762833213259,
|
|
"learning_rate": 2.655626995127584e-05,
|
|
"loss": 0.2441,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13183172047138214,
|
|
"step": 1585,
|
|
"valid_targets_mean": 2141.4,
|
|
"valid_targets_min": 1656
|
|
},
|
|
{
|
|
"epoch": 3.192964824120603,
|
|
"grad_norm": 0.6190728959132679,
|
|
"learning_rate": 2.6461575594693774e-05,
|
|
"loss": 0.2467,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11594187468290329,
|
|
"step": 1590,
|
|
"valid_targets_mean": 1882.0,
|
|
"valid_targets_min": 1395
|
|
},
|
|
{
|
|
"epoch": 3.2030150753768845,
|
|
"grad_norm": 0.6768299057154649,
|
|
"learning_rate": 2.6366719225717975e-05,
|
|
"loss": 0.2451,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1447933316230774,
|
|
"step": 1595,
|
|
"valid_targets_mean": 2185.9,
|
|
"valid_targets_min": 1781
|
|
},
|
|
{
|
|
"epoch": 3.2130653266331657,
|
|
"grad_norm": 0.6490069102391672,
|
|
"learning_rate": 2.6271703222701384e-05,
|
|
"loss": 0.2496,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1246681660413742,
|
|
"step": 1600,
|
|
"valid_targets_mean": 1829.9,
|
|
"valid_targets_min": 1457
|
|
},
|
|
{
|
|
"epoch": 3.2231155778894474,
|
|
"grad_norm": 0.6184762831335112,
|
|
"learning_rate": 2.61765299679995e-05,
|
|
"loss": 0.2393,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11580203473567963,
|
|
"step": 1605,
|
|
"valid_targets_mean": 1902.1,
|
|
"valid_targets_min": 1495
|
|
},
|
|
{
|
|
"epoch": 3.2331658291457286,
|
|
"grad_norm": 0.599556550532573,
|
|
"learning_rate": 2.6081201847910612e-05,
|
|
"loss": 0.2421,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13272082805633545,
|
|
"step": 1610,
|
|
"valid_targets_mean": 1951.1,
|
|
"valid_targets_min": 1577
|
|
},
|
|
{
|
|
"epoch": 3.24321608040201,
|
|
"grad_norm": 0.6022271583720907,
|
|
"learning_rate": 2.5985721252615984e-05,
|
|
"loss": 0.2572,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13256467878818512,
|
|
"step": 1615,
|
|
"valid_targets_mean": 2038.5,
|
|
"valid_targets_min": 1602
|
|
},
|
|
{
|
|
"epoch": 3.2532663316582915,
|
|
"grad_norm": 0.6177027874392818,
|
|
"learning_rate": 2.5890090576119926e-05,
|
|
"loss": 0.2546,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12509730458259583,
|
|
"step": 1620,
|
|
"valid_targets_mean": 1926.4,
|
|
"valid_targets_min": 1482
|
|
},
|
|
{
|
|
"epoch": 3.2633165829145727,
|
|
"grad_norm": 0.6003901825151734,
|
|
"learning_rate": 2.5794312216189758e-05,
|
|
"loss": 0.2565,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12230650335550308,
|
|
"step": 1625,
|
|
"valid_targets_mean": 1923.2,
|
|
"valid_targets_min": 1662
|
|
},
|
|
{
|
|
"epoch": 3.2733668341708544,
|
|
"grad_norm": 0.6234979337887453,
|
|
"learning_rate": 2.5698388574295696e-05,
|
|
"loss": 0.24,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13240176439285278,
|
|
"step": 1630,
|
|
"valid_targets_mean": 2129.2,
|
|
"valid_targets_min": 1827
|
|
},
|
|
{
|
|
"epoch": 3.2834170854271356,
|
|
"grad_norm": 0.6008199689808508,
|
|
"learning_rate": 2.5602322055550647e-05,
|
|
"loss": 0.2402,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10845793783664703,
|
|
"step": 1635,
|
|
"valid_targets_mean": 1991.1,
|
|
"valid_targets_min": 1603
|
|
},
|
|
{
|
|
"epoch": 3.293467336683417,
|
|
"grad_norm": 0.6185287996103953,
|
|
"learning_rate": 2.55061150686499e-05,
|
|
"loss": 0.2375,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11505062878131866,
|
|
"step": 1640,
|
|
"valid_targets_mean": 1953.8,
|
|
"valid_targets_min": 1635
|
|
},
|
|
{
|
|
"epoch": 3.3035175879396985,
|
|
"grad_norm": 0.6130492421332225,
|
|
"learning_rate": 2.5409770025810717e-05,
|
|
"loss": 0.2416,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11204279959201813,
|
|
"step": 1645,
|
|
"valid_targets_mean": 1949.2,
|
|
"valid_targets_min": 1619
|
|
},
|
|
{
|
|
"epoch": 3.3135678391959797,
|
|
"grad_norm": 0.6217419166299424,
|
|
"learning_rate": 2.531328934271188e-05,
|
|
"loss": 0.25,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13888216018676758,
|
|
"step": 1650,
|
|
"valid_targets_mean": 2057.2,
|
|
"valid_targets_min": 1606
|
|
},
|
|
{
|
|
"epoch": 3.3236180904522614,
|
|
"grad_norm": 0.6143189424554621,
|
|
"learning_rate": 2.521667543843309e-05,
|
|
"loss": 0.2545,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11148996651172638,
|
|
"step": 1655,
|
|
"valid_targets_mean": 2073.6,
|
|
"valid_targets_min": 1676
|
|
},
|
|
{
|
|
"epoch": 3.3336683417085426,
|
|
"grad_norm": 0.6075381128289719,
|
|
"learning_rate": 2.5119930735394346e-05,
|
|
"loss": 0.2489,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1153491660952568,
|
|
"step": 1660,
|
|
"valid_targets_mean": 1947.4,
|
|
"valid_targets_min": 1720
|
|
},
|
|
{
|
|
"epoch": 3.343718592964824,
|
|
"grad_norm": 0.6176373525461386,
|
|
"learning_rate": 2.5023057659295186e-05,
|
|
"loss": 0.2466,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13310858607292175,
|
|
"step": 1665,
|
|
"valid_targets_mean": 2066.4,
|
|
"valid_targets_min": 1431
|
|
},
|
|
{
|
|
"epoch": 3.3537688442211055,
|
|
"grad_norm": 0.6063837605559068,
|
|
"learning_rate": 2.4926058639053863e-05,
|
|
"loss": 0.2471,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11821518838405609,
|
|
"step": 1670,
|
|
"valid_targets_mean": 2007.5,
|
|
"valid_targets_min": 1572
|
|
},
|
|
{
|
|
"epoch": 3.363819095477387,
|
|
"grad_norm": 0.6067493403741392,
|
|
"learning_rate": 2.482893610674646e-05,
|
|
"loss": 0.2555,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11429301649332047,
|
|
"step": 1675,
|
|
"valid_targets_mean": 1791.6,
|
|
"valid_targets_min": 1407
|
|
},
|
|
{
|
|
"epoch": 3.3738693467336685,
|
|
"grad_norm": 0.6085751692225593,
|
|
"learning_rate": 2.473169249754591e-05,
|
|
"loss": 0.2507,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12028820067644119,
|
|
"step": 1680,
|
|
"valid_targets_mean": 2106.8,
|
|
"valid_targets_min": 1784
|
|
},
|
|
{
|
|
"epoch": 3.3839195979899497,
|
|
"grad_norm": 0.6456071894260242,
|
|
"learning_rate": 2.463433024966091e-05,
|
|
"loss": 0.2523,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11705578118562698,
|
|
"step": 1685,
|
|
"valid_targets_mean": 1852.9,
|
|
"valid_targets_min": 1532
|
|
},
|
|
{
|
|
"epoch": 3.393969849246231,
|
|
"grad_norm": 0.6703119202328979,
|
|
"learning_rate": 2.4536851804274844e-05,
|
|
"loss": 0.2483,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12922021746635437,
|
|
"step": 1690,
|
|
"valid_targets_mean": 1972.2,
|
|
"valid_targets_min": 1268
|
|
},
|
|
{
|
|
"epoch": 3.4040201005025126,
|
|
"grad_norm": 0.7067462163829302,
|
|
"learning_rate": 2.44392596054845e-05,
|
|
"loss": 0.246,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12799987196922302,
|
|
"step": 1695,
|
|
"valid_targets_mean": 1865.4,
|
|
"valid_targets_min": 1541
|
|
},
|
|
{
|
|
"epoch": 3.414070351758794,
|
|
"grad_norm": 0.658632280181476,
|
|
"learning_rate": 2.434155610023885e-05,
|
|
"loss": 0.2464,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13889363408088684,
|
|
"step": 1700,
|
|
"valid_targets_mean": 1930.4,
|
|
"valid_targets_min": 1509
|
|
},
|
|
{
|
|
"epoch": 3.4241206030150755,
|
|
"grad_norm": 0.6616382266521974,
|
|
"learning_rate": 2.424374373827768e-05,
|
|
"loss": 0.2426,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1249459907412529,
|
|
"step": 1705,
|
|
"valid_targets_mean": 1581.4,
|
|
"valid_targets_min": 411
|
|
},
|
|
{
|
|
"epoch": 3.4341708542713567,
|
|
"grad_norm": 0.6354854694248613,
|
|
"learning_rate": 2.4145824972070146e-05,
|
|
"loss": 0.2445,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10950978100299835,
|
|
"step": 1710,
|
|
"valid_targets_mean": 1814.1,
|
|
"valid_targets_min": 1420
|
|
},
|
|
{
|
|
"epoch": 3.4442211055276384,
|
|
"grad_norm": 0.6247648441564573,
|
|
"learning_rate": 2.4047802256753317e-05,
|
|
"loss": 0.2525,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14480818808078766,
|
|
"step": 1715,
|
|
"valid_targets_mean": 2213.0,
|
|
"valid_targets_min": 1798
|
|
},
|
|
{
|
|
"epoch": 3.4542713567839196,
|
|
"grad_norm": 0.623692433740723,
|
|
"learning_rate": 2.394967805007057e-05,
|
|
"loss": 0.2447,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12224388122558594,
|
|
"step": 1720,
|
|
"valid_targets_mean": 2044.4,
|
|
"valid_targets_min": 1692
|
|
},
|
|
{
|
|
"epoch": 3.464321608040201,
|
|
"grad_norm": 0.6818402286314287,
|
|
"learning_rate": 2.385145481231004e-05,
|
|
"loss": 0.2489,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11864765733480453,
|
|
"step": 1725,
|
|
"valid_targets_mean": 1893.0,
|
|
"valid_targets_min": 1454
|
|
},
|
|
{
|
|
"epoch": 3.4743718592964825,
|
|
"grad_norm": 0.6231063468719986,
|
|
"learning_rate": 2.3753135006242847e-05,
|
|
"loss": 0.2449,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12480935454368591,
|
|
"step": 1730,
|
|
"valid_targets_mean": 2019.6,
|
|
"valid_targets_min": 1810
|
|
},
|
|
{
|
|
"epoch": 3.4844221105527637,
|
|
"grad_norm": 0.610994005983243,
|
|
"learning_rate": 2.3654721097061407e-05,
|
|
"loss": 0.2438,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12041659653186798,
|
|
"step": 1735,
|
|
"valid_targets_mean": 1886.2,
|
|
"valid_targets_min": 1593
|
|
},
|
|
{
|
|
"epoch": 3.4944723618090454,
|
|
"grad_norm": 0.6593490811374779,
|
|
"learning_rate": 2.3556215552317606e-05,
|
|
"loss": 0.2498,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12104180455207825,
|
|
"step": 1740,
|
|
"valid_targets_mean": 1846.5,
|
|
"valid_targets_min": 1424
|
|
},
|
|
{
|
|
"epoch": 3.5045226130653266,
|
|
"grad_norm": 0.6610812232581982,
|
|
"learning_rate": 2.3457620841860912e-05,
|
|
"loss": 0.2612,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13691142201423645,
|
|
"step": 1745,
|
|
"valid_targets_mean": 2103.4,
|
|
"valid_targets_min": 1512
|
|
},
|
|
{
|
|
"epoch": 3.5145728643216083,
|
|
"grad_norm": 0.5772588230314603,
|
|
"learning_rate": 2.335893943777648e-05,
|
|
"loss": 0.2557,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12381799519062042,
|
|
"step": 1750,
|
|
"valid_targets_mean": 1972.5,
|
|
"valid_targets_min": 1660
|
|
},
|
|
{
|
|
"epoch": 3.5246231155778895,
|
|
"grad_norm": 0.6139641852548835,
|
|
"learning_rate": 2.3260173814323138e-05,
|
|
"loss": 0.2429,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12341336905956268,
|
|
"step": 1755,
|
|
"valid_targets_mean": 1923.6,
|
|
"valid_targets_min": 1583
|
|
},
|
|
{
|
|
"epoch": 3.5346733668341708,
|
|
"grad_norm": 0.5773432109485054,
|
|
"learning_rate": 2.3161326447871367e-05,
|
|
"loss": 0.2408,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12881813943386078,
|
|
"step": 1760,
|
|
"valid_targets_mean": 2047.8,
|
|
"valid_targets_min": 1545
|
|
},
|
|
{
|
|
"epoch": 3.544723618090452,
|
|
"grad_norm": 0.664843038452812,
|
|
"learning_rate": 2.306239981684121e-05,
|
|
"loss": 0.2515,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15270757675170898,
|
|
"step": 1765,
|
|
"valid_targets_mean": 2038.0,
|
|
"valid_targets_min": 1603
|
|
},
|
|
{
|
|
"epoch": 3.5547738693467337,
|
|
"grad_norm": 0.6099884121796354,
|
|
"learning_rate": 2.2963396401640133e-05,
|
|
"loss": 0.2468,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14013977348804474,
|
|
"step": 1770,
|
|
"valid_targets_mean": 2067.1,
|
|
"valid_targets_min": 1540
|
|
},
|
|
{
|
|
"epoch": 3.5648241206030153,
|
|
"grad_norm": 0.6555817529218355,
|
|
"learning_rate": 2.2864318684600807e-05,
|
|
"loss": 0.2466,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13876338303089142,
|
|
"step": 1775,
|
|
"valid_targets_mean": 2017.6,
|
|
"valid_targets_min": 1771
|
|
},
|
|
{
|
|
"epoch": 3.5748743718592966,
|
|
"grad_norm": 0.6255419352995575,
|
|
"learning_rate": 2.2765169149918915e-05,
|
|
"loss": 0.2466,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1271287053823471,
|
|
"step": 1780,
|
|
"valid_targets_mean": 1964.9,
|
|
"valid_targets_min": 1582
|
|
},
|
|
{
|
|
"epoch": 3.584924623115578,
|
|
"grad_norm": 0.6331262228956027,
|
|
"learning_rate": 2.2665950283590808e-05,
|
|
"loss": 0.2514,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1088709905743599,
|
|
"step": 1785,
|
|
"valid_targets_mean": 1949.1,
|
|
"valid_targets_min": 1586
|
|
},
|
|
{
|
|
"epoch": 3.594974874371859,
|
|
"grad_norm": 0.5919356394374087,
|
|
"learning_rate": 2.256666457335123e-05,
|
|
"loss": 0.2422,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1088305115699768,
|
|
"step": 1790,
|
|
"valid_targets_mean": 1812.4,
|
|
"valid_targets_min": 1613
|
|
},
|
|
{
|
|
"epoch": 3.6050251256281407,
|
|
"grad_norm": 0.610381915010598,
|
|
"learning_rate": 2.2467314508610897e-05,
|
|
"loss": 0.2498,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13438664376735687,
|
|
"step": 1795,
|
|
"valid_targets_mean": 2138.8,
|
|
"valid_targets_min": 1699
|
|
},
|
|
{
|
|
"epoch": 3.6150753768844224,
|
|
"grad_norm": 0.6010290667169139,
|
|
"learning_rate": 2.2367902580394113e-05,
|
|
"loss": 0.252,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11419950425624847,
|
|
"step": 1800,
|
|
"valid_targets_mean": 1875.8,
|
|
"valid_targets_min": 1531
|
|
},
|
|
{
|
|
"epoch": 3.6251256281407036,
|
|
"grad_norm": 0.6319242982807661,
|
|
"learning_rate": 2.2268431281276294e-05,
|
|
"loss": 0.2438,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11979393661022186,
|
|
"step": 1805,
|
|
"valid_targets_mean": 1902.0,
|
|
"valid_targets_min": 1462
|
|
},
|
|
{
|
|
"epoch": 3.635175879396985,
|
|
"grad_norm": 0.6450829475372098,
|
|
"learning_rate": 2.216890310532146e-05,
|
|
"loss": 0.2445,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12591694295406342,
|
|
"step": 1810,
|
|
"valid_targets_mean": 2020.0,
|
|
"valid_targets_min": 1475
|
|
},
|
|
{
|
|
"epoch": 3.645226130653266,
|
|
"grad_norm": 0.6068850031241749,
|
|
"learning_rate": 2.2069320548019743e-05,
|
|
"loss": 0.2494,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11838498711585999,
|
|
"step": 1815,
|
|
"valid_targets_mean": 1989.9,
|
|
"valid_targets_min": 1544
|
|
},
|
|
{
|
|
"epoch": 3.6552763819095477,
|
|
"grad_norm": 0.6221554316660882,
|
|
"learning_rate": 2.1969686106224758e-05,
|
|
"loss": 0.2481,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14660616219043732,
|
|
"step": 1820,
|
|
"valid_targets_mean": 2165.4,
|
|
"valid_targets_min": 1525
|
|
},
|
|
{
|
|
"epoch": 3.6653266331658294,
|
|
"grad_norm": 0.616071237208613,
|
|
"learning_rate": 2.1870002278091055e-05,
|
|
"loss": 0.2464,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1254953145980835,
|
|
"step": 1825,
|
|
"valid_targets_mean": 1968.8,
|
|
"valid_targets_min": 1460
|
|
},
|
|
{
|
|
"epoch": 3.6753768844221106,
|
|
"grad_norm": 0.6088250422992514,
|
|
"learning_rate": 2.1770271563011444e-05,
|
|
"loss": 0.2451,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11856908351182938,
|
|
"step": 1830,
|
|
"valid_targets_mean": 1858.0,
|
|
"valid_targets_min": 1344
|
|
},
|
|
{
|
|
"epoch": 3.685427135678392,
|
|
"grad_norm": 0.6066966179701787,
|
|
"learning_rate": 2.1670496461554356e-05,
|
|
"loss": 0.2487,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11919589340686798,
|
|
"step": 1835,
|
|
"valid_targets_mean": 1954.8,
|
|
"valid_targets_min": 1544
|
|
},
|
|
{
|
|
"epoch": 3.695477386934673,
|
|
"grad_norm": 0.6259960230250362,
|
|
"learning_rate": 2.1570679475401103e-05,
|
|
"loss": 0.252,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1245955154299736,
|
|
"step": 1840,
|
|
"valid_targets_mean": 1928.5,
|
|
"valid_targets_min": 1584
|
|
},
|
|
{
|
|
"epoch": 3.7055276381909548,
|
|
"grad_norm": 0.6038970547921962,
|
|
"learning_rate": 2.1470823107283212e-05,
|
|
"loss": 0.2429,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14359717071056366,
|
|
"step": 1845,
|
|
"valid_targets_mean": 2152.6,
|
|
"valid_targets_min": 1539
|
|
},
|
|
{
|
|
"epoch": 3.7155778894472364,
|
|
"grad_norm": 0.6146217094463774,
|
|
"learning_rate": 2.1370929860919617e-05,
|
|
"loss": 0.2505,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1139015257358551,
|
|
"step": 1850,
|
|
"valid_targets_mean": 1963.6,
|
|
"valid_targets_min": 1617
|
|
},
|
|
{
|
|
"epoch": 3.7256281407035177,
|
|
"grad_norm": 0.6494104822414303,
|
|
"learning_rate": 2.1271002240953927e-05,
|
|
"loss": 0.2446,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12520886957645416,
|
|
"step": 1855,
|
|
"valid_targets_mean": 1910.1,
|
|
"valid_targets_min": 1582
|
|
},
|
|
{
|
|
"epoch": 3.735678391959799,
|
|
"grad_norm": 0.6387658951725383,
|
|
"learning_rate": 2.1171042752891604e-05,
|
|
"loss": 0.2518,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14562734961509705,
|
|
"step": 1860,
|
|
"valid_targets_mean": 2106.6,
|
|
"valid_targets_min": 1853
|
|
},
|
|
{
|
|
"epoch": 3.7457286432160806,
|
|
"grad_norm": 0.6377407703198521,
|
|
"learning_rate": 2.107105390303712e-05,
|
|
"loss": 0.2372,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11826549470424652,
|
|
"step": 1865,
|
|
"valid_targets_mean": 1966.4,
|
|
"valid_targets_min": 1667
|
|
},
|
|
{
|
|
"epoch": 3.755778894472362,
|
|
"grad_norm": 0.5922288759816562,
|
|
"learning_rate": 2.0971038198431177e-05,
|
|
"loss": 0.2463,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1249329000711441,
|
|
"step": 1870,
|
|
"valid_targets_mean": 2025.9,
|
|
"valid_targets_min": 1621
|
|
},
|
|
{
|
|
"epoch": 3.7658291457286435,
|
|
"grad_norm": 0.6501963314901391,
|
|
"learning_rate": 2.087099814678779e-05,
|
|
"loss": 0.2487,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13852617144584656,
|
|
"step": 1875,
|
|
"valid_targets_mean": 1993.5,
|
|
"valid_targets_min": 1614
|
|
},
|
|
{
|
|
"epoch": 3.7758793969849247,
|
|
"grad_norm": 0.6066542458700422,
|
|
"learning_rate": 2.0770936256431442e-05,
|
|
"loss": 0.2529,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13511061668395996,
|
|
"step": 1880,
|
|
"valid_targets_mean": 1984.5,
|
|
"valid_targets_min": 1664
|
|
},
|
|
{
|
|
"epoch": 3.785929648241206,
|
|
"grad_norm": 0.6223971796090226,
|
|
"learning_rate": 2.0670855036234178e-05,
|
|
"loss": 0.2541,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14860939979553223,
|
|
"step": 1885,
|
|
"valid_targets_mean": 2051.4,
|
|
"valid_targets_min": 1665
|
|
},
|
|
{
|
|
"epoch": 3.7959798994974876,
|
|
"grad_norm": 0.6377333125444395,
|
|
"learning_rate": 2.05707569955527e-05,
|
|
"loss": 0.2546,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13653317093849182,
|
|
"step": 1890,
|
|
"valid_targets_mean": 2064.2,
|
|
"valid_targets_min": 1777
|
|
},
|
|
{
|
|
"epoch": 3.806030150753769,
|
|
"grad_norm": 0.6401529856198249,
|
|
"learning_rate": 2.047064464416546e-05,
|
|
"loss": 0.2548,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.107852503657341,
|
|
"step": 1895,
|
|
"valid_targets_mean": 1914.5,
|
|
"valid_targets_min": 1542
|
|
},
|
|
{
|
|
"epoch": 3.8160804020100505,
|
|
"grad_norm": 0.6204602475711839,
|
|
"learning_rate": 2.0370520492209723e-05,
|
|
"loss": 0.2482,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1351461112499237,
|
|
"step": 1900,
|
|
"valid_targets_mean": 1989.2,
|
|
"valid_targets_min": 1606
|
|
},
|
|
{
|
|
"epoch": 3.8261306532663317,
|
|
"grad_norm": 0.5920592917855331,
|
|
"learning_rate": 2.0270387050118634e-05,
|
|
"loss": 0.2461,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1349911093711853,
|
|
"step": 1905,
|
|
"valid_targets_mean": 2066.9,
|
|
"valid_targets_min": 1627
|
|
},
|
|
{
|
|
"epoch": 3.836180904522613,
|
|
"grad_norm": 0.6123946768977586,
|
|
"learning_rate": 2.017024682855827e-05,
|
|
"loss": 0.2457,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12758906185626984,
|
|
"step": 1910,
|
|
"valid_targets_mean": 2016.1,
|
|
"valid_targets_min": 1586
|
|
},
|
|
{
|
|
"epoch": 3.8462311557788946,
|
|
"grad_norm": 0.6107800156528567,
|
|
"learning_rate": 2.0070102338364683e-05,
|
|
"loss": 0.2439,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1272326409816742,
|
|
"step": 1915,
|
|
"valid_targets_mean": 2020.9,
|
|
"valid_targets_min": 1645
|
|
},
|
|
{
|
|
"epoch": 3.856281407035176,
|
|
"grad_norm": 0.638381883971044,
|
|
"learning_rate": 1.9969956090480964e-05,
|
|
"loss": 0.2467,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11670561134815216,
|
|
"step": 1920,
|
|
"valid_targets_mean": 2032.4,
|
|
"valid_targets_min": 1756
|
|
},
|
|
{
|
|
"epoch": 3.8663316582914575,
|
|
"grad_norm": 0.6338216472700672,
|
|
"learning_rate": 1.986981059589428e-05,
|
|
"loss": 0.2371,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11697655916213989,
|
|
"step": 1925,
|
|
"valid_targets_mean": 1975.4,
|
|
"valid_targets_min": 1628
|
|
},
|
|
{
|
|
"epoch": 3.8763819095477388,
|
|
"grad_norm": 0.6516113020445063,
|
|
"learning_rate": 1.97696683655729e-05,
|
|
"loss": 0.2378,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11896326392889023,
|
|
"step": 1930,
|
|
"valid_targets_mean": 1849.2,
|
|
"valid_targets_min": 1551
|
|
},
|
|
{
|
|
"epoch": 3.88643216080402,
|
|
"grad_norm": 0.6035309484625757,
|
|
"learning_rate": 1.966953191040324e-05,
|
|
"loss": 0.2471,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11365438997745514,
|
|
"step": 1935,
|
|
"valid_targets_mean": 1935.0,
|
|
"valid_targets_min": 1453
|
|
},
|
|
{
|
|
"epoch": 3.8964824120603017,
|
|
"grad_norm": 0.6368980047258589,
|
|
"learning_rate": 1.9569403741126925e-05,
|
|
"loss": 0.2451,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1374650001525879,
|
|
"step": 1940,
|
|
"valid_targets_mean": 2154.6,
|
|
"valid_targets_min": 1909
|
|
},
|
|
{
|
|
"epoch": 3.906532663316583,
|
|
"grad_norm": 0.6221817457625385,
|
|
"learning_rate": 1.9469286368277835e-05,
|
|
"loss": 0.2553,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1260180026292801,
|
|
"step": 1945,
|
|
"valid_targets_mean": 2170.9,
|
|
"valid_targets_min": 1651
|
|
},
|
|
{
|
|
"epoch": 3.9165829145728646,
|
|
"grad_norm": 0.6612618540400514,
|
|
"learning_rate": 1.9369182302119137e-05,
|
|
"loss": 0.236,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11286616325378418,
|
|
"step": 1950,
|
|
"valid_targets_mean": 1827.4,
|
|
"valid_targets_min": 1589
|
|
},
|
|
{
|
|
"epoch": 3.926633165829146,
|
|
"grad_norm": 0.632403163341829,
|
|
"learning_rate": 1.9269094052580348e-05,
|
|
"loss": 0.2507,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12358203530311584,
|
|
"step": 1955,
|
|
"valid_targets_mean": 1900.6,
|
|
"valid_targets_min": 1656
|
|
},
|
|
{
|
|
"epoch": 3.936683417085427,
|
|
"grad_norm": 0.5939374441741363,
|
|
"learning_rate": 1.9169024129194444e-05,
|
|
"loss": 0.2507,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11097219586372375,
|
|
"step": 1960,
|
|
"valid_targets_mean": 1949.8,
|
|
"valid_targets_min": 1540
|
|
},
|
|
{
|
|
"epoch": 3.9467336683417087,
|
|
"grad_norm": 0.6085951215601124,
|
|
"learning_rate": 1.906897504103486e-05,
|
|
"loss": 0.2421,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13879311084747314,
|
|
"step": 1965,
|
|
"valid_targets_mean": 2095.8,
|
|
"valid_targets_min": 1785
|
|
},
|
|
{
|
|
"epoch": 3.95678391959799,
|
|
"grad_norm": 0.6201536475626329,
|
|
"learning_rate": 1.8968949296652664e-05,
|
|
"loss": 0.2534,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13780029118061066,
|
|
"step": 1970,
|
|
"valid_targets_mean": 2114.2,
|
|
"valid_targets_min": 1510
|
|
},
|
|
{
|
|
"epoch": 3.9668341708542716,
|
|
"grad_norm": 0.6263303021009883,
|
|
"learning_rate": 1.88689494040136e-05,
|
|
"loss": 0.2488,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11971112340688705,
|
|
"step": 1975,
|
|
"valid_targets_mean": 1951.4,
|
|
"valid_targets_min": 1600
|
|
},
|
|
{
|
|
"epoch": 3.976884422110553,
|
|
"grad_norm": 0.6078931705106936,
|
|
"learning_rate": 1.8768977870435237e-05,
|
|
"loss": 0.2452,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12903964519500732,
|
|
"step": 1980,
|
|
"valid_targets_mean": 2063.2,
|
|
"valid_targets_min": 1600
|
|
},
|
|
{
|
|
"epoch": 3.986934673366834,
|
|
"grad_norm": 0.6245499335486749,
|
|
"learning_rate": 1.866903720252409e-05,
|
|
"loss": 0.2443,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1270110011100769,
|
|
"step": 1985,
|
|
"valid_targets_mean": 2041.5,
|
|
"valid_targets_min": 1433
|
|
},
|
|
{
|
|
"epoch": 3.9969849246231157,
|
|
"grad_norm": 0.6494118955047276,
|
|
"learning_rate": 1.8569129906112754e-05,
|
|
"loss": 0.2442,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10923419147729874,
|
|
"step": 1990,
|
|
"valid_targets_mean": 1790.6,
|
|
"valid_targets_min": 1532
|
|
},
|
|
{
|
|
"epoch": 4.006030150753769,
|
|
"grad_norm": 0.6023760358961785,
|
|
"learning_rate": 1.846925848619712e-05,
|
|
"loss": 0.2381,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11922505497932434,
|
|
"step": 1995,
|
|
"valid_targets_mean": 2016.6,
|
|
"valid_targets_min": 1548
|
|
},
|
|
{
|
|
"epoch": 4.016080402010051,
|
|
"grad_norm": 0.6267767085413255,
|
|
"learning_rate": 1.8369425446873528e-05,
|
|
"loss": 0.2319,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11328235268592834,
|
|
"step": 2000,
|
|
"valid_targets_mean": 1910.9,
|
|
"valid_targets_min": 1721
|
|
},
|
|
{
|
|
"epoch": 4.026130653266332,
|
|
"grad_norm": 0.6118339532411786,
|
|
"learning_rate": 1.8269633291275998e-05,
|
|
"loss": 0.2343,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11701051890850067,
|
|
"step": 2005,
|
|
"valid_targets_mean": 1949.9,
|
|
"valid_targets_min": 1623
|
|
},
|
|
{
|
|
"epoch": 4.036180904522613,
|
|
"grad_norm": 0.626160053499544,
|
|
"learning_rate": 1.8169884521513462e-05,
|
|
"loss": 0.2341,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1118486076593399,
|
|
"step": 2010,
|
|
"valid_targets_mean": 1934.4,
|
|
"valid_targets_min": 1244
|
|
},
|
|
{
|
|
"epoch": 4.046231155778894,
|
|
"grad_norm": 0.6336238831467331,
|
|
"learning_rate": 1.8070181638607016e-05,
|
|
"loss": 0.2354,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12756919860839844,
|
|
"step": 2015,
|
|
"valid_targets_mean": 1961.1,
|
|
"valid_targets_min": 1385
|
|
},
|
|
{
|
|
"epoch": 4.056281407035176,
|
|
"grad_norm": 0.6004056061762109,
|
|
"learning_rate": 1.7970527142427252e-05,
|
|
"loss": 0.2278,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10966813564300537,
|
|
"step": 2020,
|
|
"valid_targets_mean": 1976.1,
|
|
"valid_targets_min": 1667
|
|
},
|
|
{
|
|
"epoch": 4.066331658291458,
|
|
"grad_norm": 0.6372220433260556,
|
|
"learning_rate": 1.7870923531631537e-05,
|
|
"loss": 0.2347,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11764033138751984,
|
|
"step": 2025,
|
|
"valid_targets_mean": 1943.6,
|
|
"valid_targets_min": 1547
|
|
},
|
|
{
|
|
"epoch": 4.076381909547739,
|
|
"grad_norm": 0.6154798317637824,
|
|
"learning_rate": 1.7771373303601372e-05,
|
|
"loss": 0.2332,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11878792941570282,
|
|
"step": 2030,
|
|
"valid_targets_mean": 1966.8,
|
|
"valid_targets_min": 1457
|
|
},
|
|
{
|
|
"epoch": 4.08643216080402,
|
|
"grad_norm": 0.6309914357384981,
|
|
"learning_rate": 1.767187895437981e-05,
|
|
"loss": 0.2328,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1421642154455185,
|
|
"step": 2035,
|
|
"valid_targets_mean": 2128.5,
|
|
"valid_targets_min": 1635
|
|
},
|
|
{
|
|
"epoch": 4.096482412060301,
|
|
"grad_norm": 0.6028935791555979,
|
|
"learning_rate": 1.75724429786088e-05,
|
|
"loss": 0.2321,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09877996891736984,
|
|
"step": 2040,
|
|
"valid_targets_mean": 1902.8,
|
|
"valid_targets_min": 1700
|
|
},
|
|
{
|
|
"epoch": 4.106532663316583,
|
|
"grad_norm": 0.6393804799082871,
|
|
"learning_rate": 1.747306786946671e-05,
|
|
"loss": 0.234,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11736150085926056,
|
|
"step": 2045,
|
|
"valid_targets_mean": 2036.9,
|
|
"valid_targets_min": 1801
|
|
},
|
|
{
|
|
"epoch": 4.116582914572865,
|
|
"grad_norm": 0.6388806619784981,
|
|
"learning_rate": 1.737375611860578e-05,
|
|
"loss": 0.2343,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10260175168514252,
|
|
"step": 2050,
|
|
"valid_targets_mean": 1721.6,
|
|
"valid_targets_min": 468
|
|
},
|
|
{
|
|
"epoch": 4.126633165829146,
|
|
"grad_norm": 0.6526122757324265,
|
|
"learning_rate": 1.7274510216089665e-05,
|
|
"loss": 0.2364,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11088968813419342,
|
|
"step": 2055,
|
|
"valid_targets_mean": 1870.0,
|
|
"valid_targets_min": 1524
|
|
},
|
|
{
|
|
"epoch": 4.136683417085427,
|
|
"grad_norm": 0.6192198586303513,
|
|
"learning_rate": 1.7175332650330977e-05,
|
|
"loss": 0.2316,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11201795935630798,
|
|
"step": 2060,
|
|
"valid_targets_mean": 1941.4,
|
|
"valid_targets_min": 1737
|
|
},
|
|
{
|
|
"epoch": 4.146733668341708,
|
|
"grad_norm": 0.6300315759623552,
|
|
"learning_rate": 1.707622590802891e-05,
|
|
"loss": 0.2394,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11508876085281372,
|
|
"step": 2065,
|
|
"valid_targets_mean": 1885.0,
|
|
"valid_targets_min": 1660
|
|
},
|
|
{
|
|
"epoch": 4.15678391959799,
|
|
"grad_norm": 0.6699816609645034,
|
|
"learning_rate": 1.697719247410689e-05,
|
|
"loss": 0.2345,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11932380497455597,
|
|
"step": 2070,
|
|
"valid_targets_mean": 1995.8,
|
|
"valid_targets_min": 1802
|
|
},
|
|
{
|
|
"epoch": 4.166834170854272,
|
|
"grad_norm": 0.6100582962272068,
|
|
"learning_rate": 1.687823483165026e-05,
|
|
"loss": 0.2384,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.115172378718853,
|
|
"step": 2075,
|
|
"valid_targets_mean": 1982.0,
|
|
"valid_targets_min": 1557
|
|
},
|
|
{
|
|
"epoch": 4.176884422110553,
|
|
"grad_norm": 0.616881246084034,
|
|
"learning_rate": 1.6779355461844045e-05,
|
|
"loss": 0.2371,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10778307914733887,
|
|
"step": 2080,
|
|
"valid_targets_mean": 1902.1,
|
|
"valid_targets_min": 1682
|
|
},
|
|
{
|
|
"epoch": 4.186934673366834,
|
|
"grad_norm": 0.6151922390723004,
|
|
"learning_rate": 1.6680556843910703e-05,
|
|
"loss": 0.2333,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12145724892616272,
|
|
"step": 2085,
|
|
"valid_targets_mean": 1938.0,
|
|
"valid_targets_min": 1630
|
|
},
|
|
{
|
|
"epoch": 4.1969849246231155,
|
|
"grad_norm": 0.6182529955817782,
|
|
"learning_rate": 1.658184145504801e-05,
|
|
"loss": 0.2382,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10746758431196213,
|
|
"step": 2090,
|
|
"valid_targets_mean": 1875.2,
|
|
"valid_targets_min": 1545
|
|
},
|
|
{
|
|
"epoch": 4.207035175879397,
|
|
"grad_norm": 0.6669737061234823,
|
|
"learning_rate": 1.6483211770366886e-05,
|
|
"loss": 0.2403,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08695043623447418,
|
|
"step": 2095,
|
|
"valid_targets_mean": 1442.2,
|
|
"valid_targets_min": 341
|
|
},
|
|
{
|
|
"epoch": 4.217085427135679,
|
|
"grad_norm": 0.6006873988174043,
|
|
"learning_rate": 1.6384670262829404e-05,
|
|
"loss": 0.2302,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12663252651691437,
|
|
"step": 2100,
|
|
"valid_targets_mean": 2102.9,
|
|
"valid_targets_min": 1613
|
|
},
|
|
{
|
|
"epoch": 4.22713567839196,
|
|
"grad_norm": 0.6083101313800713,
|
|
"learning_rate": 1.6286219403186743e-05,
|
|
"loss": 0.2266,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10597462952136993,
|
|
"step": 2105,
|
|
"valid_targets_mean": 1833.9,
|
|
"valid_targets_min": 1601
|
|
},
|
|
{
|
|
"epoch": 4.237185929648241,
|
|
"grad_norm": 0.6648701832373118,
|
|
"learning_rate": 1.6187861659917245e-05,
|
|
"loss": 0.2354,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1223827674984932,
|
|
"step": 2110,
|
|
"valid_targets_mean": 2070.6,
|
|
"valid_targets_min": 1764
|
|
},
|
|
{
|
|
"epoch": 4.2472361809045225,
|
|
"grad_norm": 0.6376321469888134,
|
|
"learning_rate": 1.608959949916454e-05,
|
|
"loss": 0.2352,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10695086419582367,
|
|
"step": 2115,
|
|
"valid_targets_mean": 1815.2,
|
|
"valid_targets_min": 1609
|
|
},
|
|
{
|
|
"epoch": 4.257286432160804,
|
|
"grad_norm": 0.6520656936914387,
|
|
"learning_rate": 1.5991435384675676e-05,
|
|
"loss": 0.2288,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1230231374502182,
|
|
"step": 2120,
|
|
"valid_targets_mean": 2035.8,
|
|
"valid_targets_min": 1710
|
|
},
|
|
{
|
|
"epoch": 4.267336683417086,
|
|
"grad_norm": 0.6580639695893614,
|
|
"learning_rate": 1.5893371777739388e-05,
|
|
"loss": 0.2415,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12179134786128998,
|
|
"step": 2125,
|
|
"valid_targets_mean": 2063.8,
|
|
"valid_targets_min": 1660
|
|
},
|
|
{
|
|
"epoch": 4.277386934673367,
|
|
"grad_norm": 0.6035532552410974,
|
|
"learning_rate": 1.579541113712435e-05,
|
|
"loss": 0.2239,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10711308568716049,
|
|
"step": 2130,
|
|
"valid_targets_mean": 1812.0,
|
|
"valid_targets_min": 1584
|
|
},
|
|
{
|
|
"epoch": 4.287437185929648,
|
|
"grad_norm": 0.6762629838481116,
|
|
"learning_rate": 1.569755591901755e-05,
|
|
"loss": 0.2286,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12707467377185822,
|
|
"step": 2135,
|
|
"valid_targets_mean": 2009.9,
|
|
"valid_targets_min": 1614
|
|
},
|
|
{
|
|
"epoch": 4.2974874371859295,
|
|
"grad_norm": 0.6028199743368453,
|
|
"learning_rate": 1.5599808576962704e-05,
|
|
"loss": 0.2301,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.114751897752285,
|
|
"step": 2140,
|
|
"valid_targets_mean": 2023.6,
|
|
"valid_targets_min": 1396
|
|
},
|
|
{
|
|
"epoch": 4.307537688442211,
|
|
"grad_norm": 0.6781465131477443,
|
|
"learning_rate": 1.5502171561798705e-05,
|
|
"loss": 0.2388,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14924892783164978,
|
|
"step": 2145,
|
|
"valid_targets_mean": 2245.9,
|
|
"valid_targets_min": 1920
|
|
},
|
|
{
|
|
"epoch": 4.317587939698493,
|
|
"grad_norm": 0.5996885911581359,
|
|
"learning_rate": 1.5404647321598207e-05,
|
|
"loss": 0.2368,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11106342077255249,
|
|
"step": 2150,
|
|
"valid_targets_mean": 2011.4,
|
|
"valid_targets_min": 1678
|
|
},
|
|
{
|
|
"epoch": 4.327638190954774,
|
|
"grad_norm": 0.6303901249729731,
|
|
"learning_rate": 1.5307238301606256e-05,
|
|
"loss": 0.2426,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1307547241449356,
|
|
"step": 2155,
|
|
"valid_targets_mean": 1989.0,
|
|
"valid_targets_min": 1402
|
|
},
|
|
{
|
|
"epoch": 4.337688442211055,
|
|
"grad_norm": 0.6214901841024526,
|
|
"learning_rate": 1.520994694417893e-05,
|
|
"loss": 0.2384,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13225173950195312,
|
|
"step": 2160,
|
|
"valid_targets_mean": 2108.5,
|
|
"valid_targets_min": 1628
|
|
},
|
|
{
|
|
"epoch": 4.3477386934673365,
|
|
"grad_norm": 0.6267190317857583,
|
|
"learning_rate": 1.511277568872215e-05,
|
|
"loss": 0.2352,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12409882247447968,
|
|
"step": 2165,
|
|
"valid_targets_mean": 1988.2,
|
|
"valid_targets_min": 1446
|
|
},
|
|
{
|
|
"epoch": 4.357788944723618,
|
|
"grad_norm": 0.6110047688073623,
|
|
"learning_rate": 1.5015726971630471e-05,
|
|
"loss": 0.2289,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10359127819538116,
|
|
"step": 2170,
|
|
"valid_targets_mean": 1878.1,
|
|
"valid_targets_min": 1537
|
|
},
|
|
{
|
|
"epoch": 4.3678391959799,
|
|
"grad_norm": 0.6135528508078497,
|
|
"learning_rate": 1.491880322622605e-05,
|
|
"loss": 0.242,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1064809113740921,
|
|
"step": 2175,
|
|
"valid_targets_mean": 1911.6,
|
|
"valid_targets_min": 1692
|
|
},
|
|
{
|
|
"epoch": 4.377889447236181,
|
|
"grad_norm": 0.6067878610196109,
|
|
"learning_rate": 1.4822006882697582e-05,
|
|
"loss": 0.2273,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14647342264652252,
|
|
"step": 2180,
|
|
"valid_targets_mean": 2230.4,
|
|
"valid_targets_min": 1478
|
|
},
|
|
{
|
|
"epoch": 4.387939698492462,
|
|
"grad_norm": 0.648124941123338,
|
|
"learning_rate": 1.472534036803939e-05,
|
|
"loss": 0.2402,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11161308735609055,
|
|
"step": 2185,
|
|
"valid_targets_mean": 1850.2,
|
|
"valid_targets_min": 1449
|
|
},
|
|
{
|
|
"epoch": 4.397989949748744,
|
|
"grad_norm": 0.6007008361633944,
|
|
"learning_rate": 1.46288061059906e-05,
|
|
"loss": 0.2333,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11027591675519943,
|
|
"step": 2190,
|
|
"valid_targets_mean": 1881.9,
|
|
"valid_targets_min": 1674
|
|
},
|
|
{
|
|
"epoch": 4.408040201005025,
|
|
"grad_norm": 0.6057394119693348,
|
|
"learning_rate": 1.4532406516974289e-05,
|
|
"loss": 0.2338,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08793096244335175,
|
|
"step": 2195,
|
|
"valid_targets_mean": 1778.6,
|
|
"valid_targets_min": 1518
|
|
},
|
|
{
|
|
"epoch": 4.418090452261307,
|
|
"grad_norm": 0.6459110469450017,
|
|
"learning_rate": 1.4436144018036895e-05,
|
|
"loss": 0.2376,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12819364666938782,
|
|
"step": 2200,
|
|
"valid_targets_mean": 1969.8,
|
|
"valid_targets_min": 1691
|
|
},
|
|
{
|
|
"epoch": 4.428140703517588,
|
|
"grad_norm": 0.6624725743530038,
|
|
"learning_rate": 1.4340021022787545e-05,
|
|
"loss": 0.2346,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13239340484142303,
|
|
"step": 2205,
|
|
"valid_targets_mean": 2207.8,
|
|
"valid_targets_min": 1801
|
|
},
|
|
{
|
|
"epoch": 4.438190954773869,
|
|
"grad_norm": 0.6232937672104663,
|
|
"learning_rate": 1.4244039941337577e-05,
|
|
"loss": 0.2331,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10887771844863892,
|
|
"step": 2210,
|
|
"valid_targets_mean": 2008.4,
|
|
"valid_targets_min": 1646
|
|
},
|
|
{
|
|
"epoch": 4.448241206030151,
|
|
"grad_norm": 0.6299688629812364,
|
|
"learning_rate": 1.4148203180240088e-05,
|
|
"loss": 0.2319,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09231404960155487,
|
|
"step": 2215,
|
|
"valid_targets_mean": 1752.1,
|
|
"valid_targets_min": 1371
|
|
},
|
|
{
|
|
"epoch": 4.458291457286432,
|
|
"grad_norm": 0.6096679722339711,
|
|
"learning_rate": 1.40525131424296e-05,
|
|
"loss": 0.2371,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11810657382011414,
|
|
"step": 2220,
|
|
"valid_targets_mean": 2055.9,
|
|
"valid_targets_min": 1595
|
|
},
|
|
{
|
|
"epoch": 4.468341708542714,
|
|
"grad_norm": 0.6438929286024301,
|
|
"learning_rate": 1.395697222716181e-05,
|
|
"loss": 0.2437,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11675839126110077,
|
|
"step": 2225,
|
|
"valid_targets_mean": 1943.6,
|
|
"valid_targets_min": 1598
|
|
},
|
|
{
|
|
"epoch": 4.478391959798995,
|
|
"grad_norm": 0.6201401199542131,
|
|
"learning_rate": 1.3861582829953435e-05,
|
|
"loss": 0.2282,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11516347527503967,
|
|
"step": 2230,
|
|
"valid_targets_mean": 2056.8,
|
|
"valid_targets_min": 1755
|
|
},
|
|
{
|
|
"epoch": 4.488442211055276,
|
|
"grad_norm": 0.6552448536621228,
|
|
"learning_rate": 1.3766347342522166e-05,
|
|
"loss": 0.2282,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0900382250547409,
|
|
"step": 2235,
|
|
"valid_targets_mean": 1740.6,
|
|
"valid_targets_min": 1554
|
|
},
|
|
{
|
|
"epoch": 4.498492462311558,
|
|
"grad_norm": 0.6337519428230933,
|
|
"learning_rate": 1.367126815272666e-05,
|
|
"loss": 0.2327,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12028815597295761,
|
|
"step": 2240,
|
|
"valid_targets_mean": 2020.5,
|
|
"valid_targets_min": 1584
|
|
},
|
|
{
|
|
"epoch": 4.508542713567839,
|
|
"grad_norm": 0.617662490089928,
|
|
"learning_rate": 1.357634764450672e-05,
|
|
"loss": 0.2276,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11862218379974365,
|
|
"step": 2245,
|
|
"valid_targets_mean": 2001.1,
|
|
"valid_targets_min": 1495
|
|
},
|
|
{
|
|
"epoch": 4.518592964824121,
|
|
"grad_norm": 0.6642253224010579,
|
|
"learning_rate": 1.3481588197823447e-05,
|
|
"loss": 0.2395,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11329816281795502,
|
|
"step": 2250,
|
|
"valid_targets_mean": 1915.5,
|
|
"valid_targets_min": 1805
|
|
},
|
|
{
|
|
"epoch": 4.528643216080402,
|
|
"grad_norm": 0.6133262128156318,
|
|
"learning_rate": 1.3386992188599664e-05,
|
|
"loss": 0.2325,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11062589287757874,
|
|
"step": 2255,
|
|
"valid_targets_mean": 1856.6,
|
|
"valid_targets_min": 1503
|
|
},
|
|
{
|
|
"epoch": 4.538693467336683,
|
|
"grad_norm": 0.6222184386264903,
|
|
"learning_rate": 1.3292561988660265e-05,
|
|
"loss": 0.2308,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1099681556224823,
|
|
"step": 2260,
|
|
"valid_targets_mean": 1841.0,
|
|
"valid_targets_min": 1593
|
|
},
|
|
{
|
|
"epoch": 4.548743718592965,
|
|
"grad_norm": 0.6474925559985863,
|
|
"learning_rate": 1.3198299965672772e-05,
|
|
"loss": 0.2364,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12372187525033951,
|
|
"step": 2265,
|
|
"valid_targets_mean": 1977.8,
|
|
"valid_targets_min": 1548
|
|
},
|
|
{
|
|
"epoch": 4.558793969849246,
|
|
"grad_norm": 0.6519406913414952,
|
|
"learning_rate": 1.3104208483088002e-05,
|
|
"loss": 0.2238,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12047629803419113,
|
|
"step": 2270,
|
|
"valid_targets_mean": 1955.5,
|
|
"valid_targets_min": 1688
|
|
},
|
|
{
|
|
"epoch": 4.568844221105528,
|
|
"grad_norm": 0.6111801258542556,
|
|
"learning_rate": 1.3010289900080737e-05,
|
|
"loss": 0.2391,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11909592151641846,
|
|
"step": 2275,
|
|
"valid_targets_mean": 1995.4,
|
|
"valid_targets_min": 1602
|
|
},
|
|
{
|
|
"epoch": 4.578894472361809,
|
|
"grad_norm": 0.6020462922913059,
|
|
"learning_rate": 1.2916546571490643e-05,
|
|
"loss": 0.2342,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11636243760585785,
|
|
"step": 2280,
|
|
"valid_targets_mean": 1827.4,
|
|
"valid_targets_min": 1551
|
|
},
|
|
{
|
|
"epoch": 4.5889447236180905,
|
|
"grad_norm": 0.648428216562806,
|
|
"learning_rate": 1.2822980847763177e-05,
|
|
"loss": 0.2338,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1061846911907196,
|
|
"step": 2285,
|
|
"valid_targets_mean": 1912.8,
|
|
"valid_targets_min": 1810
|
|
},
|
|
{
|
|
"epoch": 4.598994974874372,
|
|
"grad_norm": 0.6633010914228429,
|
|
"learning_rate": 1.2729595074890696e-05,
|
|
"loss": 0.2351,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11722984164953232,
|
|
"step": 2290,
|
|
"valid_targets_mean": 1970.6,
|
|
"valid_targets_min": 1679
|
|
},
|
|
{
|
|
"epoch": 4.609045226130653,
|
|
"grad_norm": 0.6223947972810909,
|
|
"learning_rate": 1.2636391594353597e-05,
|
|
"loss": 0.2368,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12371756136417389,
|
|
"step": 2295,
|
|
"valid_targets_mean": 1892.6,
|
|
"valid_targets_min": 1460
|
|
},
|
|
{
|
|
"epoch": 4.619095477386935,
|
|
"grad_norm": 0.6430623901401558,
|
|
"learning_rate": 1.2543372743061617e-05,
|
|
"loss": 0.2325,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12313088774681091,
|
|
"step": 2300,
|
|
"valid_targets_mean": 1975.2,
|
|
"valid_targets_min": 1537
|
|
},
|
|
{
|
|
"epoch": 4.629145728643216,
|
|
"grad_norm": 0.6378937057903488,
|
|
"learning_rate": 1.2450540853295258e-05,
|
|
"loss": 0.2362,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12588758766651154,
|
|
"step": 2305,
|
|
"valid_targets_mean": 1898.1,
|
|
"valid_targets_min": 857
|
|
},
|
|
{
|
|
"epoch": 4.6391959798994975,
|
|
"grad_norm": 0.6370836365789156,
|
|
"learning_rate": 1.2357898252647304e-05,
|
|
"loss": 0.2353,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1356721818447113,
|
|
"step": 2310,
|
|
"valid_targets_mean": 2136.4,
|
|
"valid_targets_min": 1808
|
|
},
|
|
{
|
|
"epoch": 4.649246231155779,
|
|
"grad_norm": 0.6031892063385064,
|
|
"learning_rate": 1.2265447263964445e-05,
|
|
"loss": 0.2343,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10438930988311768,
|
|
"step": 2315,
|
|
"valid_targets_mean": 1961.9,
|
|
"valid_targets_min": 1688
|
|
},
|
|
{
|
|
"epoch": 4.65929648241206,
|
|
"grad_norm": 0.6336240507881284,
|
|
"learning_rate": 1.2173190205289038e-05,
|
|
"loss": 0.2365,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11854986846446991,
|
|
"step": 2320,
|
|
"valid_targets_mean": 1907.6,
|
|
"valid_targets_min": 1434
|
|
},
|
|
{
|
|
"epoch": 4.669346733668342,
|
|
"grad_norm": 0.606528964017164,
|
|
"learning_rate": 1.2081129389801018e-05,
|
|
"loss": 0.2318,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11495020240545273,
|
|
"step": 2325,
|
|
"valid_targets_mean": 2125.6,
|
|
"valid_targets_min": 1683
|
|
},
|
|
{
|
|
"epoch": 4.679396984924623,
|
|
"grad_norm": 0.6506409101308817,
|
|
"learning_rate": 1.1989267125759848e-05,
|
|
"loss": 0.2361,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11634145677089691,
|
|
"step": 2330,
|
|
"valid_targets_mean": 1920.5,
|
|
"valid_targets_min": 1696
|
|
},
|
|
{
|
|
"epoch": 4.6894472361809045,
|
|
"grad_norm": 0.6189017654420168,
|
|
"learning_rate": 1.1897605716446682e-05,
|
|
"loss": 0.2398,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10529443621635437,
|
|
"step": 2335,
|
|
"valid_targets_mean": 1834.1,
|
|
"valid_targets_min": 1576
|
|
},
|
|
{
|
|
"epoch": 4.699497487437186,
|
|
"grad_norm": 0.6379783361786449,
|
|
"learning_rate": 1.1806147460106607e-05,
|
|
"loss": 0.2326,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11959750205278397,
|
|
"step": 2340,
|
|
"valid_targets_mean": 2044.4,
|
|
"valid_targets_min": 1610
|
|
},
|
|
{
|
|
"epoch": 4.709547738693467,
|
|
"grad_norm": 0.625281038810434,
|
|
"learning_rate": 1.1714894649891008e-05,
|
|
"loss": 0.2364,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10776595771312714,
|
|
"step": 2345,
|
|
"valid_targets_mean": 1891.5,
|
|
"valid_targets_min": 1433
|
|
},
|
|
{
|
|
"epoch": 4.719597989949749,
|
|
"grad_norm": 0.6493511088385072,
|
|
"learning_rate": 1.162384957380009e-05,
|
|
"loss": 0.2304,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09902386367321014,
|
|
"step": 2350,
|
|
"valid_targets_mean": 1788.8,
|
|
"valid_targets_min": 744
|
|
},
|
|
{
|
|
"epoch": 4.72964824120603,
|
|
"grad_norm": 0.6321668471640397,
|
|
"learning_rate": 1.1533014514625465e-05,
|
|
"loss": 0.2304,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1215641051530838,
|
|
"step": 2355,
|
|
"valid_targets_mean": 1999.5,
|
|
"valid_targets_min": 1621
|
|
},
|
|
{
|
|
"epoch": 4.739698492462312,
|
|
"grad_norm": 0.6580690825369006,
|
|
"learning_rate": 1.1442391749892984e-05,
|
|
"loss": 0.233,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1115509495139122,
|
|
"step": 2360,
|
|
"valid_targets_mean": 1909.8,
|
|
"valid_targets_min": 1502
|
|
},
|
|
{
|
|
"epoch": 4.749748743718593,
|
|
"grad_norm": 0.613893648660376,
|
|
"learning_rate": 1.1351983551805593e-05,
|
|
"loss": 0.2348,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13508017361164093,
|
|
"step": 2365,
|
|
"valid_targets_mean": 2144.0,
|
|
"valid_targets_min": 1466
|
|
},
|
|
{
|
|
"epoch": 4.759798994974874,
|
|
"grad_norm": 0.6440241574004165,
|
|
"learning_rate": 1.126179218718634e-05,
|
|
"loss": 0.2277,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10516525059938431,
|
|
"step": 2370,
|
|
"valid_targets_mean": 1827.8,
|
|
"valid_targets_min": 1492
|
|
},
|
|
{
|
|
"epoch": 4.769849246231156,
|
|
"grad_norm": 0.6331712635156637,
|
|
"learning_rate": 1.117181991742159e-05,
|
|
"loss": 0.2331,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1170034110546112,
|
|
"step": 2375,
|
|
"valid_targets_mean": 2010.0,
|
|
"valid_targets_min": 1624
|
|
},
|
|
{
|
|
"epoch": 4.779899497487437,
|
|
"grad_norm": 0.6159220127110181,
|
|
"learning_rate": 1.1082068998404278e-05,
|
|
"loss": 0.2364,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12923350930213928,
|
|
"step": 2380,
|
|
"valid_targets_mean": 2027.1,
|
|
"valid_targets_min": 1794
|
|
},
|
|
{
|
|
"epoch": 4.789949748743719,
|
|
"grad_norm": 0.6152780656790047,
|
|
"learning_rate": 1.0992541680477388e-05,
|
|
"loss": 0.2402,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1171591579914093,
|
|
"step": 2385,
|
|
"valid_targets_mean": 1989.4,
|
|
"valid_targets_min": 1687
|
|
},
|
|
{
|
|
"epoch": 4.8,
|
|
"grad_norm": 0.6132904670380169,
|
|
"learning_rate": 1.0903240208377483e-05,
|
|
"loss": 0.2316,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10475753247737885,
|
|
"step": 2390,
|
|
"valid_targets_mean": 1889.0,
|
|
"valid_targets_min": 1622
|
|
},
|
|
{
|
|
"epoch": 4.810050251256281,
|
|
"grad_norm": 0.6229891066576655,
|
|
"learning_rate": 1.0814166821178474e-05,
|
|
"loss": 0.2381,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1361134797334671,
|
|
"step": 2395,
|
|
"valid_targets_mean": 2155.6,
|
|
"valid_targets_min": 1435
|
|
},
|
|
{
|
|
"epoch": 4.820100502512563,
|
|
"grad_norm": 0.630981462388972,
|
|
"learning_rate": 1.072532375223544e-05,
|
|
"loss": 0.2287,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11343435943126678,
|
|
"step": 2400,
|
|
"valid_targets_mean": 1961.6,
|
|
"valid_targets_min": 1669
|
|
},
|
|
{
|
|
"epoch": 4.830150753768844,
|
|
"grad_norm": 0.6579805208494299,
|
|
"learning_rate": 1.0636713229128643e-05,
|
|
"loss": 0.2325,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1226939707994461,
|
|
"step": 2405,
|
|
"valid_targets_mean": 2023.2,
|
|
"valid_targets_min": 1673
|
|
},
|
|
{
|
|
"epoch": 4.840201005025126,
|
|
"grad_norm": 0.6456887277675115,
|
|
"learning_rate": 1.0548337473607666e-05,
|
|
"loss": 0.2416,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1285768449306488,
|
|
"step": 2410,
|
|
"valid_targets_mean": 2232.2,
|
|
"valid_targets_min": 1772
|
|
},
|
|
{
|
|
"epoch": 4.850251256281407,
|
|
"grad_norm": 0.6536587569311304,
|
|
"learning_rate": 1.046019870153574e-05,
|
|
"loss": 0.232,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12060318887233734,
|
|
"step": 2415,
|
|
"valid_targets_mean": 1975.6,
|
|
"valid_targets_min": 1684
|
|
},
|
|
{
|
|
"epoch": 4.860301507537688,
|
|
"grad_norm": 0.6122593432905271,
|
|
"learning_rate": 1.0372299122834158e-05,
|
|
"loss": 0.2383,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10870353877544403,
|
|
"step": 2420,
|
|
"valid_targets_mean": 2014.6,
|
|
"valid_targets_min": 1600
|
|
},
|
|
{
|
|
"epoch": 4.87035175879397,
|
|
"grad_norm": 0.6602798668290484,
|
|
"learning_rate": 1.028464094142685e-05,
|
|
"loss": 0.2329,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08966298401355743,
|
|
"step": 2425,
|
|
"valid_targets_mean": 1575.2,
|
|
"valid_targets_min": 283
|
|
},
|
|
{
|
|
"epoch": 4.880402010050251,
|
|
"grad_norm": 0.6311758932035327,
|
|
"learning_rate": 1.019722635518515e-05,
|
|
"loss": 0.2315,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11684532463550568,
|
|
"step": 2430,
|
|
"valid_targets_mean": 2013.5,
|
|
"valid_targets_min": 1518
|
|
},
|
|
{
|
|
"epoch": 4.890452261306533,
|
|
"grad_norm": 0.6106523935899758,
|
|
"learning_rate": 1.0110057555872696e-05,
|
|
"loss": 0.2434,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11216290295124054,
|
|
"step": 2435,
|
|
"valid_targets_mean": 1909.8,
|
|
"valid_targets_min": 1529
|
|
},
|
|
{
|
|
"epoch": 4.900502512562814,
|
|
"grad_norm": 0.5991654198116837,
|
|
"learning_rate": 1.002313672909045e-05,
|
|
"loss": 0.2358,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11757142096757889,
|
|
"step": 2440,
|
|
"valid_targets_mean": 1969.2,
|
|
"valid_targets_min": 1683
|
|
},
|
|
{
|
|
"epoch": 4.910552763819095,
|
|
"grad_norm": 0.6113070651263629,
|
|
"learning_rate": 9.9364660542219e-06,
|
|
"loss": 0.2354,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11280138790607452,
|
|
"step": 2445,
|
|
"valid_targets_mean": 1990.9,
|
|
"valid_targets_min": 1693
|
|
},
|
|
{
|
|
"epoch": 4.920603015075377,
|
|
"grad_norm": 0.6425884878522915,
|
|
"learning_rate": 9.850047704378437e-06,
|
|
"loss": 0.2384,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12504905462265015,
|
|
"step": 2450,
|
|
"valid_targets_mean": 1980.0,
|
|
"valid_targets_min": 1389
|
|
},
|
|
{
|
|
"epoch": 4.9306532663316585,
|
|
"grad_norm": 0.6497139569878582,
|
|
"learning_rate": 9.763883846344862e-06,
|
|
"loss": 0.2365,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12320944666862488,
|
|
"step": 2455,
|
|
"valid_targets_mean": 1914.9,
|
|
"valid_targets_min": 1400
|
|
},
|
|
{
|
|
"epoch": 4.94070351758794,
|
|
"grad_norm": 0.6181067868530401,
|
|
"learning_rate": 9.677976640525031e-06,
|
|
"loss": 0.232,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11560151726007462,
|
|
"step": 2460,
|
|
"valid_targets_mean": 1916.9,
|
|
"valid_targets_min": 1426
|
|
},
|
|
{
|
|
"epoch": 4.950753768844221,
|
|
"grad_norm": 0.6069241836765832,
|
|
"learning_rate": 9.59232824088771e-06,
|
|
"loss": 0.2311,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1040213480591774,
|
|
"step": 2465,
|
|
"valid_targets_mean": 1964.8,
|
|
"valid_targets_min": 1694
|
|
},
|
|
{
|
|
"epoch": 4.960804020100502,
|
|
"grad_norm": 0.6221544554840353,
|
|
"learning_rate": 9.506940794912579e-06,
|
|
"loss": 0.2254,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10236437618732452,
|
|
"step": 2470,
|
|
"valid_targets_mean": 1758.2,
|
|
"valid_targets_min": 1513
|
|
},
|
|
{
|
|
"epoch": 4.970854271356784,
|
|
"grad_norm": 0.639538491259759,
|
|
"learning_rate": 9.421816443536374e-06,
|
|
"loss": 0.2353,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11721061915159225,
|
|
"step": 2475,
|
|
"valid_targets_mean": 1976.2,
|
|
"valid_targets_min": 1493
|
|
},
|
|
{
|
|
"epoch": 4.9809045226130655,
|
|
"grad_norm": 0.6606413365515502,
|
|
"learning_rate": 9.336957321099192e-06,
|
|
"loss": 0.2377,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11795248091220856,
|
|
"step": 2480,
|
|
"valid_targets_mean": 2028.0,
|
|
"valid_targets_min": 1756
|
|
},
|
|
{
|
|
"epoch": 4.990954773869347,
|
|
"grad_norm": 0.630538114085397,
|
|
"learning_rate": 9.252365555290994e-06,
|
|
"loss": 0.2409,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12254156172275543,
|
|
"step": 2485,
|
|
"valid_targets_mean": 1913.0,
|
|
"valid_targets_min": 1708
|
|
},
|
|
{
|
|
"epoch": 5.0,
|
|
"grad_norm": 0.8294402943713904,
|
|
"learning_rate": 9.168043267098266e-06,
|
|
"loss": 0.2334,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23619887232780457,
|
|
"step": 2490,
|
|
"valid_targets_mean": 1875.2,
|
|
"valid_targets_min": 1125
|
|
},
|
|
{
|
|
"epoch": 5.010050251256281,
|
|
"grad_norm": 0.6138694214743897,
|
|
"learning_rate": 9.083992570750827e-06,
|
|
"loss": 0.2196,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10314562916755676,
|
|
"step": 2495,
|
|
"valid_targets_mean": 1827.9,
|
|
"valid_targets_min": 1597
|
|
},
|
|
{
|
|
"epoch": 5.0201005025125625,
|
|
"grad_norm": 0.6564240811779001,
|
|
"learning_rate": 9.000215573668793e-06,
|
|
"loss": 0.2256,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10958938300609589,
|
|
"step": 2500,
|
|
"valid_targets_mean": 1905.2,
|
|
"valid_targets_min": 1431
|
|
},
|
|
{
|
|
"epoch": 5.030150753768845,
|
|
"grad_norm": 0.6550568722465039,
|
|
"learning_rate": 8.916714376409802e-06,
|
|
"loss": 0.225,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11090715229511261,
|
|
"step": 2505,
|
|
"valid_targets_mean": 1860.6,
|
|
"valid_targets_min": 1432
|
|
},
|
|
{
|
|
"epoch": 5.040201005025126,
|
|
"grad_norm": 0.6494928999173536,
|
|
"learning_rate": 8.833491072616272e-06,
|
|
"loss": 0.2248,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12295395135879517,
|
|
"step": 2510,
|
|
"valid_targets_mean": 1951.8,
|
|
"valid_targets_min": 1565
|
|
},
|
|
{
|
|
"epoch": 5.050251256281407,
|
|
"grad_norm": 0.6477026090508842,
|
|
"learning_rate": 8.750547748962968e-06,
|
|
"loss": 0.2166,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12269728630781174,
|
|
"step": 2515,
|
|
"valid_targets_mean": 1932.0,
|
|
"valid_targets_min": 1514
|
|
},
|
|
{
|
|
"epoch": 5.060301507537688,
|
|
"grad_norm": 0.6462395659956239,
|
|
"learning_rate": 8.667886485104633e-06,
|
|
"loss": 0.2236,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10983452945947647,
|
|
"step": 2520,
|
|
"valid_targets_mean": 1943.8,
|
|
"valid_targets_min": 1430
|
|
},
|
|
{
|
|
"epoch": 5.0703517587939695,
|
|
"grad_norm": 0.636964786430829,
|
|
"learning_rate": 8.585509353623884e-06,
|
|
"loss": 0.2131,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09660567343235016,
|
|
"step": 2525,
|
|
"valid_targets_mean": 1747.5,
|
|
"valid_targets_min": 1470
|
|
},
|
|
{
|
|
"epoch": 5.080402010050252,
|
|
"grad_norm": 0.6114137125887,
|
|
"learning_rate": 8.503418419979235e-06,
|
|
"loss": 0.2302,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11162625253200531,
|
|
"step": 2530,
|
|
"valid_targets_mean": 1978.4,
|
|
"valid_targets_min": 1587
|
|
},
|
|
{
|
|
"epoch": 5.090452261306533,
|
|
"grad_norm": 0.6294037479310384,
|
|
"learning_rate": 8.421615742453284e-06,
|
|
"loss": 0.2184,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11897863447666168,
|
|
"step": 2535,
|
|
"valid_targets_mean": 2065.5,
|
|
"valid_targets_min": 1741
|
|
},
|
|
{
|
|
"epoch": 5.100502512562814,
|
|
"grad_norm": 0.6666436919888821,
|
|
"learning_rate": 8.34010337210113e-06,
|
|
"loss": 0.2237,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10252441465854645,
|
|
"step": 2540,
|
|
"valid_targets_mean": 1864.9,
|
|
"valid_targets_min": 1575
|
|
},
|
|
{
|
|
"epoch": 5.110552763819095,
|
|
"grad_norm": 0.6809828871977437,
|
|
"learning_rate": 8.258883352698949e-06,
|
|
"loss": 0.2269,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09401233494281769,
|
|
"step": 2545,
|
|
"valid_targets_mean": 1842.6,
|
|
"valid_targets_min": 1665
|
|
},
|
|
{
|
|
"epoch": 5.1206030150753765,
|
|
"grad_norm": 0.6244237827403878,
|
|
"learning_rate": 8.177957720692753e-06,
|
|
"loss": 0.2216,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10164767503738403,
|
|
"step": 2550,
|
|
"valid_targets_mean": 1794.5,
|
|
"valid_targets_min": 1459
|
|
},
|
|
{
|
|
"epoch": 5.130653266331659,
|
|
"grad_norm": 0.6573790151133837,
|
|
"learning_rate": 8.097328505147285e-06,
|
|
"loss": 0.2221,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12260706722736359,
|
|
"step": 2555,
|
|
"valid_targets_mean": 2011.2,
|
|
"valid_targets_min": 1502
|
|
},
|
|
{
|
|
"epoch": 5.14070351758794,
|
|
"grad_norm": 0.6838041490245526,
|
|
"learning_rate": 8.016997727695215e-06,
|
|
"loss": 0.2277,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11256559938192368,
|
|
"step": 2560,
|
|
"valid_targets_mean": 1943.8,
|
|
"valid_targets_min": 1594
|
|
},
|
|
{
|
|
"epoch": 5.150753768844221,
|
|
"grad_norm": 0.6548063510802185,
|
|
"learning_rate": 7.936967402486384e-06,
|
|
"loss": 0.2305,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11491025239229202,
|
|
"step": 2565,
|
|
"valid_targets_mean": 2111.9,
|
|
"valid_targets_min": 1701
|
|
},
|
|
{
|
|
"epoch": 5.160804020100502,
|
|
"grad_norm": 0.6611733902989801,
|
|
"learning_rate": 7.857239536137363e-06,
|
|
"loss": 0.2246,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10913929343223572,
|
|
"step": 2570,
|
|
"valid_targets_mean": 2041.6,
|
|
"valid_targets_min": 1763
|
|
},
|
|
{
|
|
"epoch": 5.1708542713567835,
|
|
"grad_norm": 0.6622030143425172,
|
|
"learning_rate": 7.777816127681076e-06,
|
|
"loss": 0.2261,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11140120774507523,
|
|
"step": 2575,
|
|
"valid_targets_mean": 1974.4,
|
|
"valid_targets_min": 1618
|
|
},
|
|
{
|
|
"epoch": 5.180904522613066,
|
|
"grad_norm": 0.657869257922356,
|
|
"learning_rate": 7.698699168516743e-06,
|
|
"loss": 0.2168,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09625829756259918,
|
|
"step": 2580,
|
|
"valid_targets_mean": 1849.5,
|
|
"valid_targets_min": 1420
|
|
},
|
|
{
|
|
"epoch": 5.190954773869347,
|
|
"grad_norm": 0.6391929844752074,
|
|
"learning_rate": 7.6198906423599105e-06,
|
|
"loss": 0.229,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11224797368049622,
|
|
"step": 2585,
|
|
"valid_targets_mean": 1977.9,
|
|
"valid_targets_min": 1592
|
|
},
|
|
{
|
|
"epoch": 5.201005025125628,
|
|
"grad_norm": 0.6465693928691969,
|
|
"learning_rate": 7.541392525192712e-06,
|
|
"loss": 0.2245,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10427170246839523,
|
|
"step": 2590,
|
|
"valid_targets_mean": 1881.5,
|
|
"valid_targets_min": 1532
|
|
},
|
|
{
|
|
"epoch": 5.211055276381909,
|
|
"grad_norm": 0.6496393435798226,
|
|
"learning_rate": 7.463206785214334e-06,
|
|
"loss": 0.2262,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11621275544166565,
|
|
"step": 2595,
|
|
"valid_targets_mean": 1869.8,
|
|
"valid_targets_min": 1621
|
|
},
|
|
{
|
|
"epoch": 5.221105527638191,
|
|
"grad_norm": 0.6509077686711011,
|
|
"learning_rate": 7.3853353827916786e-06,
|
|
"loss": 0.2295,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11404542624950409,
|
|
"step": 2600,
|
|
"valid_targets_mean": 2002.9,
|
|
"valid_targets_min": 1688
|
|
},
|
|
{
|
|
"epoch": 5.231155778894473,
|
|
"grad_norm": 0.6559345157202309,
|
|
"learning_rate": 7.307780270410199e-06,
|
|
"loss": 0.2275,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13203291594982147,
|
|
"step": 2605,
|
|
"valid_targets_mean": 2178.0,
|
|
"valid_targets_min": 1635
|
|
},
|
|
{
|
|
"epoch": 5.241206030150754,
|
|
"grad_norm": 0.6285215806241669,
|
|
"learning_rate": 7.230543392624929e-06,
|
|
"loss": 0.2177,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11749070882797241,
|
|
"step": 2610,
|
|
"valid_targets_mean": 2043.9,
|
|
"valid_targets_min": 1626
|
|
},
|
|
{
|
|
"epoch": 5.251256281407035,
|
|
"grad_norm": 0.6485006493251664,
|
|
"learning_rate": 7.1536266860117475e-06,
|
|
"loss": 0.2357,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10736258327960968,
|
|
"step": 2615,
|
|
"valid_targets_mean": 1956.6,
|
|
"valid_targets_min": 1671
|
|
},
|
|
{
|
|
"epoch": 5.261306532663316,
|
|
"grad_norm": 0.6345711432774634,
|
|
"learning_rate": 7.077032079118824e-06,
|
|
"loss": 0.2203,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10026867687702179,
|
|
"step": 2620,
|
|
"valid_targets_mean": 1844.1,
|
|
"valid_targets_min": 1666
|
|
},
|
|
{
|
|
"epoch": 5.271356783919598,
|
|
"grad_norm": 0.6338476583436456,
|
|
"learning_rate": 7.000761492418258e-06,
|
|
"loss": 0.2226,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10844532400369644,
|
|
"step": 2625,
|
|
"valid_targets_mean": 1876.6,
|
|
"valid_targets_min": 1626
|
|
},
|
|
{
|
|
"epoch": 5.28140703517588,
|
|
"grad_norm": 0.6505841001065518,
|
|
"learning_rate": 6.9248168382579085e-06,
|
|
"loss": 0.233,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13009332120418549,
|
|
"step": 2630,
|
|
"valid_targets_mean": 2151.4,
|
|
"valid_targets_min": 1579
|
|
},
|
|
{
|
|
"epoch": 5.291457286432161,
|
|
"grad_norm": 0.6606491318745418,
|
|
"learning_rate": 6.849200020813482e-06,
|
|
"loss": 0.2347,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11969298124313354,
|
|
"step": 2635,
|
|
"valid_targets_mean": 2017.2,
|
|
"valid_targets_min": 1497
|
|
},
|
|
{
|
|
"epoch": 5.301507537688442,
|
|
"grad_norm": 0.641996210336638,
|
|
"learning_rate": 6.773912936040757e-06,
|
|
"loss": 0.2261,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11677876859903336,
|
|
"step": 2640,
|
|
"valid_targets_mean": 1936.0,
|
|
"valid_targets_min": 1641
|
|
},
|
|
{
|
|
"epoch": 5.311557788944723,
|
|
"grad_norm": 0.6360456919492221,
|
|
"learning_rate": 6.698957471628049e-06,
|
|
"loss": 0.2276,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10789966583251953,
|
|
"step": 2645,
|
|
"valid_targets_mean": 1914.0,
|
|
"valid_targets_min": 1633
|
|
},
|
|
{
|
|
"epoch": 5.321608040201005,
|
|
"grad_norm": 0.6476824550950844,
|
|
"learning_rate": 6.624335506948912e-06,
|
|
"loss": 0.2247,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10909949988126755,
|
|
"step": 2650,
|
|
"valid_targets_mean": 1817.5,
|
|
"valid_targets_min": 1210
|
|
},
|
|
{
|
|
"epoch": 5.331658291457287,
|
|
"grad_norm": 0.6460643365408448,
|
|
"learning_rate": 6.550048913014977e-06,
|
|
"loss": 0.2282,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11689987778663635,
|
|
"step": 2655,
|
|
"valid_targets_mean": 1971.6,
|
|
"valid_targets_min": 1538
|
|
},
|
|
{
|
|
"epoch": 5.341708542713568,
|
|
"grad_norm": 0.6524665678812462,
|
|
"learning_rate": 6.476099552429076e-06,
|
|
"loss": 0.2309,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12329615652561188,
|
|
"step": 2660,
|
|
"valid_targets_mean": 2090.8,
|
|
"valid_targets_min": 1713
|
|
},
|
|
{
|
|
"epoch": 5.351758793969849,
|
|
"grad_norm": 0.6731758801457514,
|
|
"learning_rate": 6.402489279338482e-06,
|
|
"loss": 0.2323,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11523545533418655,
|
|
"step": 2665,
|
|
"valid_targets_mean": 1964.8,
|
|
"valid_targets_min": 1509
|
|
},
|
|
{
|
|
"epoch": 5.36180904522613,
|
|
"grad_norm": 0.6594019717587084,
|
|
"learning_rate": 6.3292199393884916e-06,
|
|
"loss": 0.2233,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12570026516914368,
|
|
"step": 2670,
|
|
"valid_targets_mean": 2048.2,
|
|
"valid_targets_min": 1546
|
|
},
|
|
{
|
|
"epoch": 5.371859296482412,
|
|
"grad_norm": 0.6214913305088532,
|
|
"learning_rate": 6.25629336967611e-06,
|
|
"loss": 0.2305,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10683426260948181,
|
|
"step": 2675,
|
|
"valid_targets_mean": 1872.2,
|
|
"valid_targets_min": 1542
|
|
},
|
|
{
|
|
"epoch": 5.381909547738694,
|
|
"grad_norm": 0.639843071466487,
|
|
"learning_rate": 6.183711398703998e-06,
|
|
"loss": 0.2261,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10204660147428513,
|
|
"step": 2680,
|
|
"valid_targets_mean": 1888.8,
|
|
"valid_targets_min": 1582
|
|
},
|
|
{
|
|
"epoch": 5.391959798994975,
|
|
"grad_norm": 0.640123363694521,
|
|
"learning_rate": 6.111475846334602e-06,
|
|
"loss": 0.2238,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11178840696811676,
|
|
"step": 2685,
|
|
"valid_targets_mean": 1930.5,
|
|
"valid_targets_min": 1739
|
|
},
|
|
{
|
|
"epoch": 5.402010050251256,
|
|
"grad_norm": 0.6605265401268097,
|
|
"learning_rate": 6.0395885237445705e-06,
|
|
"loss": 0.2199,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11413691937923431,
|
|
"step": 2690,
|
|
"valid_targets_mean": 1996.9,
|
|
"valid_targets_min": 1658
|
|
},
|
|
{
|
|
"epoch": 5.4120603015075375,
|
|
"grad_norm": 0.6349270838282373,
|
|
"learning_rate": 5.9680512333793e-06,
|
|
"loss": 0.2274,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11971573531627655,
|
|
"step": 2695,
|
|
"valid_targets_mean": 1958.1,
|
|
"valid_targets_min": 1485
|
|
},
|
|
{
|
|
"epoch": 5.422110552763819,
|
|
"grad_norm": 0.69543673359251,
|
|
"learning_rate": 5.896865768907754e-06,
|
|
"loss": 0.2331,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11075019836425781,
|
|
"step": 2700,
|
|
"valid_targets_mean": 1787.8,
|
|
"valid_targets_min": 381
|
|
},
|
|
{
|
|
"epoch": 5.432160804020101,
|
|
"grad_norm": 0.6249407685389946,
|
|
"learning_rate": 5.826033915177509e-06,
|
|
"loss": 0.2186,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09433281421661377,
|
|
"step": 2705,
|
|
"valid_targets_mean": 1722.5,
|
|
"valid_targets_min": 1456
|
|
},
|
|
{
|
|
"epoch": 5.442211055276382,
|
|
"grad_norm": 0.6666563111524926,
|
|
"learning_rate": 5.755557448169984e-06,
|
|
"loss": 0.2264,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12051384150981903,
|
|
"step": 2710,
|
|
"valid_targets_mean": 2009.5,
|
|
"valid_targets_min": 1841
|
|
},
|
|
{
|
|
"epoch": 5.452261306532663,
|
|
"grad_norm": 0.6139063596928556,
|
|
"learning_rate": 5.685438134955923e-06,
|
|
"loss": 0.2286,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11032386869192123,
|
|
"step": 2715,
|
|
"valid_targets_mean": 2019.1,
|
|
"valid_targets_min": 1482
|
|
},
|
|
{
|
|
"epoch": 5.4623115577889445,
|
|
"grad_norm": 0.6492307637197061,
|
|
"learning_rate": 5.615677733651041e-06,
|
|
"loss": 0.226,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10339774191379547,
|
|
"step": 2720,
|
|
"valid_targets_mean": 1943.2,
|
|
"valid_targets_min": 1612
|
|
},
|
|
{
|
|
"epoch": 5.472361809045226,
|
|
"grad_norm": 0.607496421880328,
|
|
"learning_rate": 5.5462779933720286e-06,
|
|
"loss": 0.2265,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10339631885290146,
|
|
"step": 2725,
|
|
"valid_targets_mean": 1932.9,
|
|
"valid_targets_min": 1601
|
|
},
|
|
{
|
|
"epoch": 5.482412060301508,
|
|
"grad_norm": 0.6190586512003532,
|
|
"learning_rate": 5.477240654192635e-06,
|
|
"loss": 0.2157,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12356507778167725,
|
|
"step": 2730,
|
|
"valid_targets_mean": 2024.9,
|
|
"valid_targets_min": 1710
|
|
},
|
|
{
|
|
"epoch": 5.492462311557789,
|
|
"grad_norm": 0.6804212009349481,
|
|
"learning_rate": 5.4085674471000375e-06,
|
|
"loss": 0.2287,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1158190667629242,
|
|
"step": 2735,
|
|
"valid_targets_mean": 1955.8,
|
|
"valid_targets_min": 1763
|
|
},
|
|
{
|
|
"epoch": 5.50251256281407,
|
|
"grad_norm": 0.6558550063826638,
|
|
"learning_rate": 5.340260093951479e-06,
|
|
"loss": 0.2227,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10619363188743591,
|
|
"step": 2740,
|
|
"valid_targets_mean": 1918.4,
|
|
"valid_targets_min": 1476
|
|
},
|
|
{
|
|
"epoch": 5.5125628140703515,
|
|
"grad_norm": 0.6455831116583547,
|
|
"learning_rate": 5.272320307431051e-06,
|
|
"loss": 0.2222,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12934494018554688,
|
|
"step": 2745,
|
|
"valid_targets_mean": 2134.8,
|
|
"valid_targets_min": 1656
|
|
},
|
|
{
|
|
"epoch": 5.522613065326633,
|
|
"grad_norm": 0.6514162006949176,
|
|
"learning_rate": 5.204749791006796e-06,
|
|
"loss": 0.2217,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11375857889652252,
|
|
"step": 2750,
|
|
"valid_targets_mean": 1926.1,
|
|
"valid_targets_min": 1527
|
|
},
|
|
{
|
|
"epoch": 5.532663316582915,
|
|
"grad_norm": 0.6391707830447325,
|
|
"learning_rate": 5.137550238887943e-06,
|
|
"loss": 0.2251,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09347797185182571,
|
|
"step": 2755,
|
|
"valid_targets_mean": 1818.0,
|
|
"valid_targets_min": 1460
|
|
},
|
|
{
|
|
"epoch": 5.542713567839196,
|
|
"grad_norm": 0.6800372781796856,
|
|
"learning_rate": 5.070723335982484e-06,
|
|
"loss": 0.2236,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11272189021110535,
|
|
"step": 2760,
|
|
"valid_targets_mean": 1933.6,
|
|
"valid_targets_min": 1728
|
|
},
|
|
{
|
|
"epoch": 5.552763819095477,
|
|
"grad_norm": 0.6528678143417211,
|
|
"learning_rate": 5.0042707578548945e-06,
|
|
"loss": 0.2322,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10301244258880615,
|
|
"step": 2765,
|
|
"valid_targets_mean": 1918.4,
|
|
"valid_targets_min": 1474
|
|
},
|
|
{
|
|
"epoch": 5.562814070351759,
|
|
"grad_norm": 0.6497185198171854,
|
|
"learning_rate": 4.938194170684125e-06,
|
|
"loss": 0.2176,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10733124613761902,
|
|
"step": 2770,
|
|
"valid_targets_mean": 2007.0,
|
|
"valid_targets_min": 1689
|
|
},
|
|
{
|
|
"epoch": 5.572864321608041,
|
|
"grad_norm": 0.6341177031116013,
|
|
"learning_rate": 4.872495231221821e-06,
|
|
"loss": 0.2228,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10900768637657166,
|
|
"step": 2775,
|
|
"valid_targets_mean": 2009.2,
|
|
"valid_targets_min": 1539
|
|
},
|
|
{
|
|
"epoch": 5.582914572864322,
|
|
"grad_norm": 0.6457704874711409,
|
|
"learning_rate": 4.807175586750805e-06,
|
|
"loss": 0.2286,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10361837595701218,
|
|
"step": 2780,
|
|
"valid_targets_mean": 1932.2,
|
|
"valid_targets_min": 1601
|
|
},
|
|
{
|
|
"epoch": 5.592964824120603,
|
|
"grad_norm": 0.6472479827021362,
|
|
"learning_rate": 4.742236875043764e-06,
|
|
"loss": 0.2297,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1258879452943802,
|
|
"step": 2785,
|
|
"valid_targets_mean": 2056.1,
|
|
"valid_targets_min": 1635
|
|
},
|
|
{
|
|
"epoch": 5.603015075376884,
|
|
"grad_norm": 0.6518777704345888,
|
|
"learning_rate": 4.677680724322153e-06,
|
|
"loss": 0.2217,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12423891574144363,
|
|
"step": 2790,
|
|
"valid_targets_mean": 1994.9,
|
|
"valid_targets_min": 1496
|
|
},
|
|
{
|
|
"epoch": 5.613065326633166,
|
|
"grad_norm": 0.6503573013771119,
|
|
"learning_rate": 4.6135087532154325e-06,
|
|
"loss": 0.2265,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12951505184173584,
|
|
"step": 2795,
|
|
"valid_targets_mean": 2058.6,
|
|
"valid_targets_min": 1815
|
|
},
|
|
{
|
|
"epoch": 5.623115577889447,
|
|
"grad_norm": 0.6451149807632718,
|
|
"learning_rate": 4.549722570720419e-06,
|
|
"loss": 0.2159,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11863362789154053,
|
|
"step": 2800,
|
|
"valid_targets_mean": 2013.0,
|
|
"valid_targets_min": 1759
|
|
},
|
|
{
|
|
"epoch": 5.633165829145729,
|
|
"grad_norm": 0.6311016933792598,
|
|
"learning_rate": 4.486323776160997e-06,
|
|
"loss": 0.2185,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0973382443189621,
|
|
"step": 2805,
|
|
"valid_targets_mean": 1698.5,
|
|
"valid_targets_min": 1311
|
|
},
|
|
{
|
|
"epoch": 5.64321608040201,
|
|
"grad_norm": 0.650395640631093,
|
|
"learning_rate": 4.423313959147972e-06,
|
|
"loss": 0.2231,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09076140820980072,
|
|
"step": 2810,
|
|
"valid_targets_mean": 1921.6,
|
|
"valid_targets_min": 1638
|
|
},
|
|
{
|
|
"epoch": 5.653266331658291,
|
|
"grad_norm": 0.6330032219047775,
|
|
"learning_rate": 4.360694699539261e-06,
|
|
"loss": 0.2263,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11272220313549042,
|
|
"step": 2815,
|
|
"valid_targets_mean": 2136.6,
|
|
"valid_targets_min": 1660
|
|
},
|
|
{
|
|
"epoch": 5.663316582914573,
|
|
"grad_norm": 0.6149423050844951,
|
|
"learning_rate": 4.2984675674002444e-06,
|
|
"loss": 0.2191,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10571373999118805,
|
|
"step": 2820,
|
|
"valid_targets_mean": 1848.4,
|
|
"valid_targets_min": 1423
|
|
},
|
|
{
|
|
"epoch": 5.673366834170855,
|
|
"grad_norm": 0.6694756571631908,
|
|
"learning_rate": 4.236634122964407e-06,
|
|
"loss": 0.2301,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10646457970142365,
|
|
"step": 2825,
|
|
"valid_targets_mean": 1903.2,
|
|
"valid_targets_min": 1513
|
|
},
|
|
{
|
|
"epoch": 5.683417085427136,
|
|
"grad_norm": 0.6157154091463166,
|
|
"learning_rate": 4.175195916594219e-06,
|
|
"loss": 0.2238,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11080768704414368,
|
|
"step": 2830,
|
|
"valid_targets_mean": 2188.1,
|
|
"valid_targets_min": 1741
|
|
},
|
|
{
|
|
"epoch": 5.693467336683417,
|
|
"grad_norm": 0.6766158558037396,
|
|
"learning_rate": 4.114154488742283e-06,
|
|
"loss": 0.237,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12312798202037811,
|
|
"step": 2835,
|
|
"valid_targets_mean": 2044.4,
|
|
"valid_targets_min": 1724
|
|
},
|
|
{
|
|
"epoch": 5.703517587939698,
|
|
"grad_norm": 0.6377441515400029,
|
|
"learning_rate": 4.053511369912686e-06,
|
|
"loss": 0.229,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11854176968336105,
|
|
"step": 2840,
|
|
"valid_targets_mean": 2101.4,
|
|
"valid_targets_min": 1693
|
|
},
|
|
{
|
|
"epoch": 5.71356783919598,
|
|
"grad_norm": 0.636924833476787,
|
|
"learning_rate": 3.993268080622626e-06,
|
|
"loss": 0.2297,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10837747901678085,
|
|
"step": 2845,
|
|
"valid_targets_mean": 2002.2,
|
|
"valid_targets_min": 1699
|
|
},
|
|
{
|
|
"epoch": 5.723618090452261,
|
|
"grad_norm": 0.6190524919950445,
|
|
"learning_rate": 3.933426131364293e-06,
|
|
"loss": 0.2144,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10250069200992584,
|
|
"step": 2850,
|
|
"valid_targets_mean": 1875.6,
|
|
"valid_targets_min": 1781
|
|
},
|
|
{
|
|
"epoch": 5.733668341708543,
|
|
"grad_norm": 0.6672490862887946,
|
|
"learning_rate": 3.8739870225670075e-06,
|
|
"loss": 0.2213,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11241291463375092,
|
|
"step": 2855,
|
|
"valid_targets_mean": 1861.9,
|
|
"valid_targets_min": 1267
|
|
},
|
|
{
|
|
"epoch": 5.743718592964824,
|
|
"grad_norm": 0.6317991616465451,
|
|
"learning_rate": 3.814952244559589e-06,
|
|
"loss": 0.2187,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10230278968811035,
|
|
"step": 2860,
|
|
"valid_targets_mean": 1882.5,
|
|
"valid_targets_min": 1502
|
|
},
|
|
{
|
|
"epoch": 5.7537688442211055,
|
|
"grad_norm": 0.6557164418153798,
|
|
"learning_rate": 3.756323277532976e-06,
|
|
"loss": 0.2183,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11721843481063843,
|
|
"step": 2865,
|
|
"valid_targets_mean": 1918.4,
|
|
"valid_targets_min": 1490
|
|
},
|
|
{
|
|
"epoch": 5.763819095477387,
|
|
"grad_norm": 0.6171163889209286,
|
|
"learning_rate": 3.698101591503143e-06,
|
|
"loss": 0.2237,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12624111771583557,
|
|
"step": 2870,
|
|
"valid_targets_mean": 2063.9,
|
|
"valid_targets_min": 1371
|
|
},
|
|
{
|
|
"epoch": 5.773869346733669,
|
|
"grad_norm": 0.6345349652922956,
|
|
"learning_rate": 3.640288646274208e-06,
|
|
"loss": 0.2294,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11136958748102188,
|
|
"step": 2875,
|
|
"valid_targets_mean": 1854.8,
|
|
"valid_targets_min": 1326
|
|
},
|
|
{
|
|
"epoch": 5.78391959798995,
|
|
"grad_norm": 0.6547306600857009,
|
|
"learning_rate": 3.582885891401868e-06,
|
|
"loss": 0.2259,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09975391626358032,
|
|
"step": 2880,
|
|
"valid_targets_mean": 1953.5,
|
|
"valid_targets_min": 1493
|
|
},
|
|
{
|
|
"epoch": 5.793969849246231,
|
|
"grad_norm": 0.6458333465841082,
|
|
"learning_rate": 3.5258947661570097e-06,
|
|
"loss": 0.226,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1163594201207161,
|
|
"step": 2885,
|
|
"valid_targets_mean": 1958.8,
|
|
"valid_targets_min": 1589
|
|
},
|
|
{
|
|
"epoch": 5.8040201005025125,
|
|
"grad_norm": 0.6793436896272823,
|
|
"learning_rate": 3.469316699489671e-06,
|
|
"loss": 0.2179,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10323749482631683,
|
|
"step": 2890,
|
|
"valid_targets_mean": 1742.6,
|
|
"valid_targets_min": 1379
|
|
},
|
|
{
|
|
"epoch": 5.814070351758794,
|
|
"grad_norm": 0.6557158436686704,
|
|
"learning_rate": 3.413153109993179e-06,
|
|
"loss": 0.2268,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1238040179014206,
|
|
"step": 2895,
|
|
"valid_targets_mean": 2045.8,
|
|
"valid_targets_min": 1767
|
|
},
|
|
{
|
|
"epoch": 5.824120603015075,
|
|
"grad_norm": 0.626635392594388,
|
|
"learning_rate": 3.3574054058685855e-06,
|
|
"loss": 0.2219,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09461769461631775,
|
|
"step": 2900,
|
|
"valid_targets_mean": 1898.9,
|
|
"valid_targets_min": 1604
|
|
},
|
|
{
|
|
"epoch": 5.834170854271357,
|
|
"grad_norm": 0.6032766987445981,
|
|
"learning_rate": 3.302074984889363e-06,
|
|
"loss": 0.2168,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1111203283071518,
|
|
"step": 2905,
|
|
"valid_targets_mean": 2038.0,
|
|
"valid_targets_min": 1558
|
|
},
|
|
{
|
|
"epoch": 5.844221105527638,
|
|
"grad_norm": 0.6596436959471433,
|
|
"learning_rate": 3.2471632343663704e-06,
|
|
"loss": 0.2284,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11637003719806671,
|
|
"step": 2910,
|
|
"valid_targets_mean": 2032.8,
|
|
"valid_targets_min": 1700
|
|
},
|
|
{
|
|
"epoch": 5.8542713567839195,
|
|
"grad_norm": 0.6455291280466648,
|
|
"learning_rate": 3.192671531113056e-06,
|
|
"loss": 0.223,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11526080220937729,
|
|
"step": 2915,
|
|
"valid_targets_mean": 1958.8,
|
|
"valid_targets_min": 1569
|
|
},
|
|
{
|
|
"epoch": 5.864321608040201,
|
|
"grad_norm": 0.6332221321054743,
|
|
"learning_rate": 3.1386012414109233e-06,
|
|
"loss": 0.2253,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12650589644908905,
|
|
"step": 2920,
|
|
"valid_targets_mean": 2123.8,
|
|
"valid_targets_min": 1679
|
|
},
|
|
{
|
|
"epoch": 5.874371859296483,
|
|
"grad_norm": 0.6191555373039018,
|
|
"learning_rate": 3.08495372097531e-06,
|
|
"loss": 0.2221,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10503631085157394,
|
|
"step": 2925,
|
|
"valid_targets_mean": 1915.2,
|
|
"valid_targets_min": 1631
|
|
},
|
|
{
|
|
"epoch": 5.884422110552764,
|
|
"grad_norm": 0.6320697919537023,
|
|
"learning_rate": 3.0317303149213484e-06,
|
|
"loss": 0.2312,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11310167610645294,
|
|
"step": 2930,
|
|
"valid_targets_mean": 2043.0,
|
|
"valid_targets_min": 1870
|
|
},
|
|
{
|
|
"epoch": 5.894472361809045,
|
|
"grad_norm": 0.6406578360129471,
|
|
"learning_rate": 2.978932357730293e-06,
|
|
"loss": 0.2236,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11449071764945984,
|
|
"step": 2935,
|
|
"valid_targets_mean": 1974.6,
|
|
"valid_targets_min": 1655
|
|
},
|
|
{
|
|
"epoch": 5.9045226130653266,
|
|
"grad_norm": 0.6633164921532272,
|
|
"learning_rate": 2.9265611732160092e-06,
|
|
"loss": 0.2295,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11629427969455719,
|
|
"step": 2940,
|
|
"valid_targets_mean": 2027.4,
|
|
"valid_targets_min": 1739
|
|
},
|
|
{
|
|
"epoch": 5.914572864321608,
|
|
"grad_norm": 0.6526741165394242,
|
|
"learning_rate": 2.8746180744918196e-06,
|
|
"loss": 0.2235,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11561928689479828,
|
|
"step": 2945,
|
|
"valid_targets_mean": 2024.1,
|
|
"valid_targets_min": 1568
|
|
},
|
|
{
|
|
"epoch": 5.924623115577889,
|
|
"grad_norm": 0.6499535176396325,
|
|
"learning_rate": 2.8231043639375634e-06,
|
|
"loss": 0.2232,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09970546513795853,
|
|
"step": 2950,
|
|
"valid_targets_mean": 1870.0,
|
|
"valid_targets_min": 1539
|
|
},
|
|
{
|
|
"epoch": 5.934673366834171,
|
|
"grad_norm": 0.6588779747911337,
|
|
"learning_rate": 2.772021333166934e-06,
|
|
"loss": 0.2207,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1245623305439949,
|
|
"step": 2955,
|
|
"valid_targets_mean": 2110.0,
|
|
"valid_targets_min": 1766
|
|
},
|
|
{
|
|
"epoch": 5.944723618090452,
|
|
"grad_norm": 0.6454183084618251,
|
|
"learning_rate": 2.721370262995109e-06,
|
|
"loss": 0.2241,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12094708532094955,
|
|
"step": 2960,
|
|
"valid_targets_mean": 1964.8,
|
|
"valid_targets_min": 1547
|
|
},
|
|
{
|
|
"epoch": 5.954773869346734,
|
|
"grad_norm": 0.5999153869021547,
|
|
"learning_rate": 2.671152423406631e-06,
|
|
"loss": 0.2248,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11940610408782959,
|
|
"step": 2965,
|
|
"valid_targets_mean": 2101.0,
|
|
"valid_targets_min": 1794
|
|
},
|
|
{
|
|
"epoch": 5.964824120603015,
|
|
"grad_norm": 0.6259042657937979,
|
|
"learning_rate": 2.6213690735235696e-06,
|
|
"loss": 0.2192,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09639428555965424,
|
|
"step": 2970,
|
|
"valid_targets_mean": 1847.9,
|
|
"valid_targets_min": 1573
|
|
},
|
|
{
|
|
"epoch": 5.974874371859297,
|
|
"grad_norm": 0.6170156010891511,
|
|
"learning_rate": 2.5720214615739393e-06,
|
|
"loss": 0.2209,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09657932817935944,
|
|
"step": 2975,
|
|
"valid_targets_mean": 1824.2,
|
|
"valid_targets_min": 1346
|
|
},
|
|
{
|
|
"epoch": 5.984924623115578,
|
|
"grad_norm": 0.6221115067133398,
|
|
"learning_rate": 2.523110824860402e-06,
|
|
"loss": 0.2291,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11124708503484726,
|
|
"step": 2980,
|
|
"valid_targets_mean": 2044.6,
|
|
"valid_targets_min": 1758
|
|
},
|
|
{
|
|
"epoch": 5.994974874371859,
|
|
"grad_norm": 0.6338459095672976,
|
|
"learning_rate": 2.4746383897292647e-06,
|
|
"loss": 0.2282,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12257210165262222,
|
|
"step": 2985,
|
|
"valid_targets_mean": 2121.2,
|
|
"valid_targets_min": 1902
|
|
},
|
|
{
|
|
"epoch": 6.004020100502513,
|
|
"grad_norm": 0.6385888861168111,
|
|
"learning_rate": 2.4266053715397187e-06,
|
|
"loss": 0.2238,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11008796095848083,
|
|
"step": 2990,
|
|
"valid_targets_mean": 1970.8,
|
|
"valid_targets_min": 1512
|
|
},
|
|
{
|
|
"epoch": 6.014070351758794,
|
|
"grad_norm": 0.6166152963536727,
|
|
"learning_rate": 2.379012974633346e-06,
|
|
"loss": 0.2185,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11224382370710373,
|
|
"step": 2995,
|
|
"valid_targets_mean": 2016.2,
|
|
"valid_targets_min": 1665
|
|
},
|
|
{
|
|
"epoch": 6.024120603015075,
|
|
"grad_norm": 0.6614933907364384,
|
|
"learning_rate": 2.331862392303965e-06,
|
|
"loss": 0.2174,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10836464166641235,
|
|
"step": 3000,
|
|
"valid_targets_mean": 1892.2,
|
|
"valid_targets_min": 1617
|
|
},
|
|
{
|
|
"epoch": 6.034170854271356,
|
|
"grad_norm": 0.6766049433745519,
|
|
"learning_rate": 2.2851548067676756e-06,
|
|
"loss": 0.2242,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10418006032705307,
|
|
"step": 3005,
|
|
"valid_targets_mean": 1813.4,
|
|
"valid_targets_min": 815
|
|
},
|
|
{
|
|
"epoch": 6.0442211055276385,
|
|
"grad_norm": 0.6244940331513527,
|
|
"learning_rate": 2.238891389133222e-06,
|
|
"loss": 0.2112,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11208360642194748,
|
|
"step": 3010,
|
|
"valid_targets_mean": 1934.8,
|
|
"valid_targets_min": 1566
|
|
},
|
|
{
|
|
"epoch": 6.05427135678392,
|
|
"grad_norm": 0.6848039818102013,
|
|
"learning_rate": 2.1930732993726543e-06,
|
|
"loss": 0.216,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11770936101675034,
|
|
"step": 3015,
|
|
"valid_targets_mean": 1968.5,
|
|
"valid_targets_min": 1468
|
|
},
|
|
{
|
|
"epoch": 6.064321608040201,
|
|
"grad_norm": 0.6596693714910697,
|
|
"learning_rate": 2.147701686292214e-06,
|
|
"loss": 0.2139,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10197962820529938,
|
|
"step": 3020,
|
|
"valid_targets_mean": 1819.1,
|
|
"valid_targets_min": 1515
|
|
},
|
|
{
|
|
"epoch": 6.074371859296482,
|
|
"grad_norm": 0.6371226669651717,
|
|
"learning_rate": 2.1027776875035544e-06,
|
|
"loss": 0.2239,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.113180011510849,
|
|
"step": 3025,
|
|
"valid_targets_mean": 2004.8,
|
|
"valid_targets_min": 1495
|
|
},
|
|
{
|
|
"epoch": 6.084422110552763,
|
|
"grad_norm": 0.6418535907277048,
|
|
"learning_rate": 2.0583024293951892e-06,
|
|
"loss": 0.2202,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11319401860237122,
|
|
"step": 3030,
|
|
"valid_targets_mean": 1938.0,
|
|
"valid_targets_min": 1746
|
|
},
|
|
{
|
|
"epoch": 6.0944723618090455,
|
|
"grad_norm": 0.6424802756028659,
|
|
"learning_rate": 2.014277027104272e-06,
|
|
"loss": 0.2119,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10676263272762299,
|
|
"step": 3035,
|
|
"valid_targets_mean": 1881.1,
|
|
"valid_targets_min": 1580
|
|
},
|
|
{
|
|
"epoch": 6.104522613065327,
|
|
"grad_norm": 0.6169736559101331,
|
|
"learning_rate": 1.9707025844886394e-06,
|
|
"loss": 0.2194,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1010156199336052,
|
|
"step": 3040,
|
|
"valid_targets_mean": 1872.9,
|
|
"valid_targets_min": 1528
|
|
},
|
|
{
|
|
"epoch": 6.114572864321608,
|
|
"grad_norm": 0.6337912326883276,
|
|
"learning_rate": 1.92758019409911e-06,
|
|
"loss": 0.2179,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09560423344373703,
|
|
"step": 3045,
|
|
"valid_targets_mean": 1633.8,
|
|
"valid_targets_min": 1218
|
|
},
|
|
{
|
|
"epoch": 6.124623115577889,
|
|
"grad_norm": 0.6833401371207766,
|
|
"learning_rate": 1.8849109371521158e-06,
|
|
"loss": 0.2178,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1060650497674942,
|
|
"step": 3050,
|
|
"valid_targets_mean": 1890.5,
|
|
"valid_targets_min": 1598
|
|
},
|
|
{
|
|
"epoch": 6.134673366834171,
|
|
"grad_norm": 0.6399814713740064,
|
|
"learning_rate": 1.8426958835025831e-06,
|
|
"loss": 0.2148,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1128477156162262,
|
|
"step": 3055,
|
|
"valid_targets_mean": 1967.8,
|
|
"valid_targets_min": 1591
|
|
},
|
|
{
|
|
"epoch": 6.1447236180904525,
|
|
"grad_norm": 0.6364684184556513,
|
|
"learning_rate": 1.8009360916171026e-06,
|
|
"loss": 0.2173,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10164594650268555,
|
|
"step": 3060,
|
|
"valid_targets_mean": 1928.9,
|
|
"valid_targets_min": 1621
|
|
},
|
|
{
|
|
"epoch": 6.154773869346734,
|
|
"grad_norm": 0.6582973525435453,
|
|
"learning_rate": 1.759632608547388e-06,
|
|
"loss": 0.2159,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11537270247936249,
|
|
"step": 3065,
|
|
"valid_targets_mean": 1981.1,
|
|
"valid_targets_min": 1583
|
|
},
|
|
{
|
|
"epoch": 6.164824120603015,
|
|
"grad_norm": 0.6688126193343998,
|
|
"learning_rate": 1.7187864699040413e-06,
|
|
"loss": 0.2178,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10925822705030441,
|
|
"step": 3070,
|
|
"valid_targets_mean": 1955.2,
|
|
"valid_targets_min": 1629
|
|
},
|
|
{
|
|
"epoch": 6.174874371859296,
|
|
"grad_norm": 0.6643540260204456,
|
|
"learning_rate": 1.6783986998305702e-06,
|
|
"loss": 0.2218,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1204906776547432,
|
|
"step": 3075,
|
|
"valid_targets_mean": 2050.5,
|
|
"valid_targets_min": 1540
|
|
},
|
|
{
|
|
"epoch": 6.1849246231155774,
|
|
"grad_norm": 0.6543318004591604,
|
|
"learning_rate": 1.6384703109777222e-06,
|
|
"loss": 0.2271,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11117293685674667,
|
|
"step": 3080,
|
|
"valid_targets_mean": 1878.0,
|
|
"valid_targets_min": 1596
|
|
},
|
|
{
|
|
"epoch": 6.19497487437186,
|
|
"grad_norm": 0.6258242845356509,
|
|
"learning_rate": 1.5990023044780656e-06,
|
|
"loss": 0.2268,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09858259558677673,
|
|
"step": 3085,
|
|
"valid_targets_mean": 1904.2,
|
|
"valid_targets_min": 1482
|
|
},
|
|
{
|
|
"epoch": 6.205025125628141,
|
|
"grad_norm": 0.6294778137265199,
|
|
"learning_rate": 1.5599956699209261e-06,
|
|
"loss": 0.2217,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10270405560731888,
|
|
"step": 3090,
|
|
"valid_targets_mean": 1915.6,
|
|
"valid_targets_min": 1431
|
|
},
|
|
{
|
|
"epoch": 6.215075376884422,
|
|
"grad_norm": 0.6746052281261795,
|
|
"learning_rate": 1.5214513853275592e-06,
|
|
"loss": 0.2141,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11323828995227814,
|
|
"step": 3095,
|
|
"valid_targets_mean": 1972.5,
|
|
"valid_targets_min": 1639
|
|
},
|
|
{
|
|
"epoch": 6.225125628140703,
|
|
"grad_norm": 0.6576852621668436,
|
|
"learning_rate": 1.483370417126615e-06,
|
|
"loss": 0.2273,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12229473888874054,
|
|
"step": 3100,
|
|
"valid_targets_mean": 2271.1,
|
|
"valid_targets_min": 1872
|
|
},
|
|
{
|
|
"epoch": 6.235175879396985,
|
|
"grad_norm": 0.626128074395466,
|
|
"learning_rate": 1.4457537201299254e-06,
|
|
"loss": 0.2192,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0930563360452652,
|
|
"step": 3105,
|
|
"valid_targets_mean": 1817.0,
|
|
"valid_targets_min": 1449
|
|
},
|
|
{
|
|
"epoch": 6.245226130653267,
|
|
"grad_norm": 0.6499195090023381,
|
|
"learning_rate": 1.4086022375085473e-06,
|
|
"loss": 0.2232,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1089339479804039,
|
|
"step": 3110,
|
|
"valid_targets_mean": 2067.6,
|
|
"valid_targets_min": 1540
|
|
},
|
|
{
|
|
"epoch": 6.255276381909548,
|
|
"grad_norm": 0.613305862833866,
|
|
"learning_rate": 1.3719169007691413e-06,
|
|
"loss": 0.2154,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10911812633275986,
|
|
"step": 3115,
|
|
"valid_targets_mean": 1880.4,
|
|
"valid_targets_min": 1572
|
|
},
|
|
{
|
|
"epoch": 6.265326633165829,
|
|
"grad_norm": 0.6562965111535343,
|
|
"learning_rate": 1.3356986297305729e-06,
|
|
"loss": 0.2186,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11662560701370239,
|
|
"step": 3120,
|
|
"valid_targets_mean": 2011.8,
|
|
"valid_targets_min": 1825
|
|
},
|
|
{
|
|
"epoch": 6.27537688442211,
|
|
"grad_norm": 0.6395227531708639,
|
|
"learning_rate": 1.2999483325008976e-06,
|
|
"loss": 0.2125,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09566240012645721,
|
|
"step": 3125,
|
|
"valid_targets_mean": 1840.6,
|
|
"valid_targets_min": 1722
|
|
},
|
|
{
|
|
"epoch": 6.2854271356783915,
|
|
"grad_norm": 0.6471164101234613,
|
|
"learning_rate": 1.2646669054545613e-06,
|
|
"loss": 0.2254,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11017978936433792,
|
|
"step": 3130,
|
|
"valid_targets_mean": 1992.5,
|
|
"valid_targets_min": 1776
|
|
},
|
|
{
|
|
"epoch": 6.295477386934674,
|
|
"grad_norm": 0.6460897939666411,
|
|
"learning_rate": 1.2298552332099244e-06,
|
|
"loss": 0.22,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10727876424789429,
|
|
"step": 3135,
|
|
"valid_targets_mean": 1928.1,
|
|
"valid_targets_min": 1620
|
|
},
|
|
{
|
|
"epoch": 6.305527638190955,
|
|
"grad_norm": 0.6342558935692455,
|
|
"learning_rate": 1.1955141886070965e-06,
|
|
"loss": 0.2195,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10562030971050262,
|
|
"step": 3140,
|
|
"valid_targets_mean": 1932.6,
|
|
"valid_targets_min": 1620
|
|
},
|
|
{
|
|
"epoch": 6.315577889447236,
|
|
"grad_norm": 0.6728899395528547,
|
|
"learning_rate": 1.1616446326860475e-06,
|
|
"loss": 0.2213,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11594712734222412,
|
|
"step": 3145,
|
|
"valid_targets_mean": 1983.5,
|
|
"valid_targets_min": 1629
|
|
},
|
|
{
|
|
"epoch": 6.325628140703517,
|
|
"grad_norm": 0.6118477413305771,
|
|
"learning_rate": 1.1282474146650158e-06,
|
|
"loss": 0.2197,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.111540287733078,
|
|
"step": 3150,
|
|
"valid_targets_mean": 1972.8,
|
|
"valid_targets_min": 1350
|
|
},
|
|
{
|
|
"epoch": 6.335678391959799,
|
|
"grad_norm": 0.6770131452176013,
|
|
"learning_rate": 1.0953233719192102e-06,
|
|
"loss": 0.2217,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11730696260929108,
|
|
"step": 3155,
|
|
"valid_targets_mean": 1984.4,
|
|
"valid_targets_min": 1670
|
|
},
|
|
{
|
|
"epoch": 6.345728643216081,
|
|
"grad_norm": 0.6408832378331694,
|
|
"learning_rate": 1.062873329959826e-06,
|
|
"loss": 0.2163,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10839914530515671,
|
|
"step": 3160,
|
|
"valid_targets_mean": 1968.0,
|
|
"valid_targets_min": 1838
|
|
},
|
|
{
|
|
"epoch": 6.355778894472362,
|
|
"grad_norm": 0.6697635987645005,
|
|
"learning_rate": 1.0308981024133358e-06,
|
|
"loss": 0.2229,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09060360491275787,
|
|
"step": 3165,
|
|
"valid_targets_mean": 1639.1,
|
|
"valid_targets_min": 602
|
|
},
|
|
{
|
|
"epoch": 6.365829145728643,
|
|
"grad_norm": 0.6439207840073027,
|
|
"learning_rate": 9.993984910011001e-07,
|
|
"loss": 0.2104,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08847002685070038,
|
|
"step": 3170,
|
|
"valid_targets_mean": 1806.8,
|
|
"valid_targets_min": 1532
|
|
},
|
|
{
|
|
"epoch": 6.375879396984924,
|
|
"grad_norm": 0.6295009938458105,
|
|
"learning_rate": 9.683752855192496e-07,
|
|
"loss": 0.2136,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.0929751843214035,
|
|
"step": 3175,
|
|
"valid_targets_mean": 1881.4,
|
|
"valid_targets_min": 1424
|
|
},
|
|
{
|
|
"epoch": 6.385929648241206,
|
|
"grad_norm": 0.659091673750488,
|
|
"learning_rate": 9.378292638189057e-07,
|
|
"loss": 0.2173,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10881422460079193,
|
|
"step": 3180,
|
|
"valid_targets_mean": 1960.0,
|
|
"valid_targets_min": 1542
|
|
},
|
|
{
|
|
"epoch": 6.395979899497488,
|
|
"grad_norm": 0.6732141010566015,
|
|
"learning_rate": 9.077611917866602e-07,
|
|
"loss": 0.2197,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10494682192802429,
|
|
"step": 3185,
|
|
"valid_targets_mean": 1755.6,
|
|
"valid_targets_min": 1374
|
|
},
|
|
{
|
|
"epoch": 6.406030150753769,
|
|
"grad_norm": 0.6642444610455142,
|
|
"learning_rate": 8.781718233253688e-07,
|
|
"loss": 0.2146,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13711833953857422,
|
|
"step": 3190,
|
|
"valid_targets_mean": 2197.1,
|
|
"valid_targets_min": 1924
|
|
},
|
|
{
|
|
"epoch": 6.41608040201005,
|
|
"grad_norm": 0.6375293484615818,
|
|
"learning_rate": 8.490619003352596e-07,
|
|
"loss": 0.2146,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11230097711086273,
|
|
"step": 3195,
|
|
"valid_targets_mean": 1955.9,
|
|
"valid_targets_min": 1546
|
|
},
|
|
{
|
|
"epoch": 6.426130653266331,
|
|
"grad_norm": 0.6647257912310426,
|
|
"learning_rate": 8.204321526953296e-07,
|
|
"loss": 0.2205,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10973141342401505,
|
|
"step": 3200,
|
|
"valid_targets_mean": 1937.4,
|
|
"valid_targets_min": 1656
|
|
},
|
|
{
|
|
"epoch": 6.4361809045226135,
|
|
"grad_norm": 0.6820967345847556,
|
|
"learning_rate": 7.922832982450424e-07,
|
|
"loss": 0.2191,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11424146592617035,
|
|
"step": 3205,
|
|
"valid_targets_mean": 2034.2,
|
|
"valid_targets_min": 1719
|
|
},
|
|
{
|
|
"epoch": 6.446231155778895,
|
|
"grad_norm": 0.6398098559797889,
|
|
"learning_rate": 7.646160427663285e-07,
|
|
"loss": 0.2249,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11214125901460648,
|
|
"step": 3210,
|
|
"valid_targets_mean": 1935.2,
|
|
"valid_targets_min": 1513
|
|
},
|
|
{
|
|
"epoch": 6.456281407035176,
|
|
"grad_norm": 0.6239959716005896,
|
|
"learning_rate": 7.374310799658824e-07,
|
|
"loss": 0.2113,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09469613432884216,
|
|
"step": 3215,
|
|
"valid_targets_mean": 1797.1,
|
|
"valid_targets_min": 1604
|
|
},
|
|
{
|
|
"epoch": 6.466331658291457,
|
|
"grad_norm": 0.6396717558272358,
|
|
"learning_rate": 7.107290914577891e-07,
|
|
"loss": 0.2195,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09451834857463837,
|
|
"step": 3220,
|
|
"valid_targets_mean": 1905.0,
|
|
"valid_targets_min": 1582
|
|
},
|
|
{
|
|
"epoch": 6.476381909547738,
|
|
"grad_norm": 0.6611532711101025,
|
|
"learning_rate": 6.84510746746414e-07,
|
|
"loss": 0.2289,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1039697453379631,
|
|
"step": 3225,
|
|
"valid_targets_mean": 1826.6,
|
|
"valid_targets_min": 1346
|
|
},
|
|
{
|
|
"epoch": 6.48643216080402,
|
|
"grad_norm": 0.6612503129151043,
|
|
"learning_rate": 6.587767032096226e-07,
|
|
"loss": 0.218,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1253022849559784,
|
|
"step": 3230,
|
|
"valid_targets_mean": 2138.5,
|
|
"valid_targets_min": 1692
|
|
},
|
|
{
|
|
"epoch": 6.496482412060302,
|
|
"grad_norm": 0.6532918262645208,
|
|
"learning_rate": 6.335276060823047e-07,
|
|
"loss": 0.2171,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1240498349070549,
|
|
"step": 3235,
|
|
"valid_targets_mean": 2099.1,
|
|
"valid_targets_min": 1695
|
|
},
|
|
{
|
|
"epoch": 6.506532663316583,
|
|
"grad_norm": 0.6285760145971682,
|
|
"learning_rate": 6.087640884401835e-07,
|
|
"loss": 0.2178,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11092621833086014,
|
|
"step": 3240,
|
|
"valid_targets_mean": 1973.5,
|
|
"valid_targets_min": 1824
|
|
},
|
|
{
|
|
"epoch": 6.516582914572864,
|
|
"grad_norm": 0.7997470703471016,
|
|
"learning_rate": 5.844867711839541e-07,
|
|
"loss": 0.2131,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12112972140312195,
|
|
"step": 3245,
|
|
"valid_targets_mean": 2050.9,
|
|
"valid_targets_min": 1723
|
|
},
|
|
{
|
|
"epoch": 6.526633165829145,
|
|
"grad_norm": 0.6188474242522309,
|
|
"learning_rate": 5.6069626302371e-07,
|
|
"loss": 0.2208,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09947022795677185,
|
|
"step": 3250,
|
|
"valid_targets_mean": 1933.2,
|
|
"valid_targets_min": 1718
|
|
},
|
|
{
|
|
"epoch": 6.5366834170854275,
|
|
"grad_norm": 0.643975213503449,
|
|
"learning_rate": 5.373931604636817e-07,
|
|
"loss": 0.2246,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09753699600696564,
|
|
"step": 3255,
|
|
"valid_targets_mean": 1932.9,
|
|
"valid_targets_min": 1535
|
|
},
|
|
{
|
|
"epoch": 6.546733668341709,
|
|
"grad_norm": 0.6526742519463627,
|
|
"learning_rate": 5.145780477872798e-07,
|
|
"loss": 0.2172,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11546642333269119,
|
|
"step": 3260,
|
|
"valid_targets_mean": 2046.8,
|
|
"valid_targets_min": 1781
|
|
},
|
|
{
|
|
"epoch": 6.55678391959799,
|
|
"grad_norm": 0.6765638933264958,
|
|
"learning_rate": 4.922514970424464e-07,
|
|
"loss": 0.2224,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12688806653022766,
|
|
"step": 3265,
|
|
"valid_targets_mean": 2027.9,
|
|
"valid_targets_min": 1565
|
|
},
|
|
{
|
|
"epoch": 6.566834170854271,
|
|
"grad_norm": 0.6295372830956298,
|
|
"learning_rate": 4.7041406802730726e-07,
|
|
"loss": 0.2158,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10909399390220642,
|
|
"step": 3270,
|
|
"valid_targets_mean": 1918.5,
|
|
"valid_targets_min": 1643
|
|
},
|
|
{
|
|
"epoch": 6.5768844221105525,
|
|
"grad_norm": 0.6629697573881878,
|
|
"learning_rate": 4.4906630827614707e-07,
|
|
"loss": 0.2157,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10853196680545807,
|
|
"step": 3275,
|
|
"valid_targets_mean": 1915.5,
|
|
"valid_targets_min": 1401
|
|
},
|
|
{
|
|
"epoch": 6.586934673366834,
|
|
"grad_norm": 0.6739593043917333,
|
|
"learning_rate": 4.2820875304567135e-07,
|
|
"loss": 0.2173,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10168344527482986,
|
|
"step": 3280,
|
|
"valid_targets_mean": 1901.9,
|
|
"valid_targets_min": 1447
|
|
},
|
|
{
|
|
"epoch": 6.596984924623116,
|
|
"grad_norm": 0.6163144190284584,
|
|
"learning_rate": 4.0784192530158415e-07,
|
|
"loss": 0.2211,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1096048504114151,
|
|
"step": 3285,
|
|
"valid_targets_mean": 1989.9,
|
|
"valid_targets_min": 1589
|
|
},
|
|
{
|
|
"epoch": 6.607035175879397,
|
|
"grad_norm": 0.649504718247639,
|
|
"learning_rate": 3.8796633570548526e-07,
|
|
"loss": 0.2208,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10920561105012894,
|
|
"step": 3290,
|
|
"valid_targets_mean": 1887.8,
|
|
"valid_targets_min": 1558
|
|
},
|
|
{
|
|
"epoch": 6.617085427135678,
|
|
"grad_norm": 0.7729532474177626,
|
|
"learning_rate": 3.6858248260205565e-07,
|
|
"loss": 0.223,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11699426919221878,
|
|
"step": 3295,
|
|
"valid_targets_mean": 1818.8,
|
|
"valid_targets_min": 1552
|
|
},
|
|
{
|
|
"epoch": 6.6271356783919595,
|
|
"grad_norm": 0.6785904999788005,
|
|
"learning_rate": 3.4969085200657226e-07,
|
|
"loss": 0.2226,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1076914370059967,
|
|
"step": 3300,
|
|
"valid_targets_mean": 1818.0,
|
|
"valid_targets_min": 1516
|
|
},
|
|
{
|
|
"epoch": 6.637185929648242,
|
|
"grad_norm": 0.6395917266149337,
|
|
"learning_rate": 3.312919175927176e-07,
|
|
"loss": 0.2255,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11221298575401306,
|
|
"step": 3305,
|
|
"valid_targets_mean": 1987.1,
|
|
"valid_targets_min": 1595
|
|
},
|
|
{
|
|
"epoch": 6.647236180904523,
|
|
"grad_norm": 0.6442069525803193,
|
|
"learning_rate": 3.1338614068069596e-07,
|
|
"loss": 0.2199,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09851694107055664,
|
|
"step": 3310,
|
|
"valid_targets_mean": 1812.6,
|
|
"valid_targets_min": 1454
|
|
},
|
|
{
|
|
"epoch": 6.657286432160804,
|
|
"grad_norm": 0.6300171565487673,
|
|
"learning_rate": 2.95973970225687e-07,
|
|
"loss": 0.219,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11080724000930786,
|
|
"step": 3315,
|
|
"valid_targets_mean": 1781.5,
|
|
"valid_targets_min": 1366
|
|
},
|
|
{
|
|
"epoch": 6.667336683417085,
|
|
"grad_norm": 0.6316593922923787,
|
|
"learning_rate": 2.790558428065615e-07,
|
|
"loss": 0.2225,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10890462249517441,
|
|
"step": 3320,
|
|
"valid_targets_mean": 1918.8,
|
|
"valid_targets_min": 1564
|
|
},
|
|
{
|
|
"epoch": 6.6773869346733665,
|
|
"grad_norm": 0.6242091518338323,
|
|
"learning_rate": 2.626321826149547e-07,
|
|
"loss": 0.2226,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10438473522663116,
|
|
"step": 3325,
|
|
"valid_targets_mean": 1945.2,
|
|
"valid_targets_min": 1697
|
|
},
|
|
{
|
|
"epoch": 6.687437185929648,
|
|
"grad_norm": 0.6331840999855737,
|
|
"learning_rate": 2.4670340144462565e-07,
|
|
"loss": 0.225,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10541290044784546,
|
|
"step": 3330,
|
|
"valid_targets_mean": 2006.9,
|
|
"valid_targets_min": 1691
|
|
},
|
|
{
|
|
"epoch": 6.69748743718593,
|
|
"grad_norm": 0.6419449269640931,
|
|
"learning_rate": 2.312698986811368e-07,
|
|
"loss": 0.2148,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12872329354286194,
|
|
"step": 3335,
|
|
"valid_targets_mean": 2016.2,
|
|
"valid_targets_min": 1453
|
|
},
|
|
{
|
|
"epoch": 6.707537688442211,
|
|
"grad_norm": 0.6459467924107342,
|
|
"learning_rate": 2.1633206129182404e-07,
|
|
"loss": 0.2148,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11308933794498444,
|
|
"step": 3340,
|
|
"valid_targets_mean": 2025.8,
|
|
"valid_targets_min": 1663
|
|
},
|
|
{
|
|
"epoch": 6.717587939698492,
|
|
"grad_norm": 0.6393525302432572,
|
|
"learning_rate": 2.018902638161091e-07,
|
|
"loss": 0.2241,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10590572655200958,
|
|
"step": 3345,
|
|
"valid_targets_mean": 1888.5,
|
|
"valid_targets_min": 1630
|
|
},
|
|
{
|
|
"epoch": 6.727638190954774,
|
|
"grad_norm": 0.7053788227930644,
|
|
"learning_rate": 1.8794486835610692e-07,
|
|
"loss": 0.221,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11321339011192322,
|
|
"step": 3350,
|
|
"valid_targets_mean": 1799.4,
|
|
"valid_targets_min": 1631
|
|
},
|
|
{
|
|
"epoch": 6.737688442211056,
|
|
"grad_norm": 0.6241706310464993,
|
|
"learning_rate": 1.7449622456754635e-07,
|
|
"loss": 0.2231,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12485706806182861,
|
|
"step": 3355,
|
|
"valid_targets_mean": 2132.2,
|
|
"valid_targets_min": 1588
|
|
},
|
|
{
|
|
"epoch": 6.747738693467337,
|
|
"grad_norm": 0.6372728859900779,
|
|
"learning_rate": 1.6154466965099037e-07,
|
|
"loss": 0.2219,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11582145094871521,
|
|
"step": 3360,
|
|
"valid_targets_mean": 2037.9,
|
|
"valid_targets_min": 1685
|
|
},
|
|
{
|
|
"epoch": 6.757788944723618,
|
|
"grad_norm": 0.6351709431136201,
|
|
"learning_rate": 1.490905283434052e-07,
|
|
"loss": 0.2199,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10975571721792221,
|
|
"step": 3365,
|
|
"valid_targets_mean": 1911.8,
|
|
"valid_targets_min": 1502
|
|
},
|
|
{
|
|
"epoch": 6.767839195979899,
|
|
"grad_norm": 0.6483407285533237,
|
|
"learning_rate": 1.3713411290999346e-07,
|
|
"loss": 0.2256,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11957737803459167,
|
|
"step": 3370,
|
|
"valid_targets_mean": 2024.6,
|
|
"valid_targets_min": 1607
|
|
},
|
|
{
|
|
"epoch": 6.777889447236181,
|
|
"grad_norm": 0.6595032551856087,
|
|
"learning_rate": 1.2567572313638477e-07,
|
|
"loss": 0.223,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12855570018291473,
|
|
"step": 3375,
|
|
"valid_targets_mean": 2055.9,
|
|
"valid_targets_min": 1446
|
|
},
|
|
{
|
|
"epoch": 6.787939698492462,
|
|
"grad_norm": 0.6357877588004213,
|
|
"learning_rate": 1.1471564632110411e-07,
|
|
"loss": 0.217,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11922886967658997,
|
|
"step": 3380,
|
|
"valid_targets_mean": 2047.0,
|
|
"valid_targets_min": 1678
|
|
},
|
|
{
|
|
"epoch": 6.797989949748744,
|
|
"grad_norm": 0.6464780440029113,
|
|
"learning_rate": 1.0425415726837973e-07,
|
|
"loss": 0.2225,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11513714492321014,
|
|
"step": 3385,
|
|
"valid_targets_mean": 2044.1,
|
|
"valid_targets_min": 1406
|
|
},
|
|
{
|
|
"epoch": 6.808040201005025,
|
|
"grad_norm": 0.6407202648706154,
|
|
"learning_rate": 9.42915182812465e-08,
|
|
"loss": 0.2223,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11063450574874878,
|
|
"step": 3390,
|
|
"valid_targets_mean": 1902.8,
|
|
"valid_targets_min": 1551
|
|
},
|
|
{
|
|
"epoch": 6.818090452261306,
|
|
"grad_norm": 0.6362512464108347,
|
|
"learning_rate": 8.482797915496665e-08,
|
|
"loss": 0.2132,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09877845644950867,
|
|
"step": 3395,
|
|
"valid_targets_mean": 1812.5,
|
|
"valid_targets_min": 1420
|
|
},
|
|
{
|
|
"epoch": 6.828140703517588,
|
|
"grad_norm": 0.6285638557618641,
|
|
"learning_rate": 7.586377717077708e-08,
|
|
"loss": 0.22,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10405369102954865,
|
|
"step": 3400,
|
|
"valid_targets_mean": 1881.5,
|
|
"valid_targets_min": 1734
|
|
},
|
|
{
|
|
"epoch": 6.83819095477387,
|
|
"grad_norm": 0.6198567487079395,
|
|
"learning_rate": 6.73991370899274e-08,
|
|
"loss": 0.2164,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10677100718021393,
|
|
"step": 3405,
|
|
"valid_targets_mean": 1935.1,
|
|
"valid_targets_min": 1547
|
|
},
|
|
{
|
|
"epoch": 6.848241206030151,
|
|
"grad_norm": 0.6038831626985319,
|
|
"learning_rate": 5.943427114805334e-08,
|
|
"loss": 0.2213,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10967598855495453,
|
|
"step": 3410,
|
|
"valid_targets_mean": 1946.9,
|
|
"valid_targets_min": 1724
|
|
},
|
|
{
|
|
"epoch": 6.858291457286432,
|
|
"grad_norm": 0.6564914487175532,
|
|
"learning_rate": 5.196937904985211e-08,
|
|
"loss": 0.2219,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14475616812705994,
|
|
"step": 3415,
|
|
"valid_targets_mean": 2265.9,
|
|
"valid_targets_min": 1766
|
|
},
|
|
{
|
|
"epoch": 6.868341708542713,
|
|
"grad_norm": 0.6533520267778749,
|
|
"learning_rate": 4.500464796407533e-08,
|
|
"loss": 0.2115,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11582393944263458,
|
|
"step": 3420,
|
|
"valid_targets_mean": 2013.2,
|
|
"valid_targets_min": 1565
|
|
},
|
|
{
|
|
"epoch": 6.878391959798995,
|
|
"grad_norm": 0.641475057925159,
|
|
"learning_rate": 3.8540252518837195e-08,
|
|
"loss": 0.2227,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11347610503435135,
|
|
"step": 3425,
|
|
"valid_targets_mean": 2026.1,
|
|
"valid_targets_min": 1692
|
|
},
|
|
{
|
|
"epoch": 6.888442211055277,
|
|
"grad_norm": 0.6672978645158152,
|
|
"learning_rate": 3.257635479723353e-08,
|
|
"loss": 0.2176,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10128751397132874,
|
|
"step": 3430,
|
|
"valid_targets_mean": 1844.4,
|
|
"valid_targets_min": 1410
|
|
},
|
|
{
|
|
"epoch": 6.898492462311558,
|
|
"grad_norm": 0.6271347903832479,
|
|
"learning_rate": 2.7113104333280627e-08,
|
|
"loss": 0.219,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11569839715957642,
|
|
"step": 3435,
|
|
"valid_targets_mean": 2062.8,
|
|
"valid_targets_min": 1716
|
|
},
|
|
{
|
|
"epoch": 6.908542713567839,
|
|
"grad_norm": 0.6380723218626885,
|
|
"learning_rate": 2.2150638108162648e-08,
|
|
"loss": 0.216,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11009392142295837,
|
|
"step": 3440,
|
|
"valid_targets_mean": 1932.2,
|
|
"valid_targets_min": 1752
|
|
},
|
|
{
|
|
"epoch": 6.9185929648241205,
|
|
"grad_norm": 0.6327627226488467,
|
|
"learning_rate": 1.7689080546801073e-08,
|
|
"loss": 0.2172,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11736994236707687,
|
|
"step": 3445,
|
|
"valid_targets_mean": 1979.0,
|
|
"valid_targets_min": 1539
|
|
},
|
|
{
|
|
"epoch": 6.928643216080402,
|
|
"grad_norm": 0.6550511609289686,
|
|
"learning_rate": 1.372854351473274e-08,
|
|
"loss": 0.214,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.09947407990694046,
|
|
"step": 3450,
|
|
"valid_targets_mean": 2007.0,
|
|
"valid_targets_min": 1622
|
|
},
|
|
{
|
|
"epoch": 6.938693467336684,
|
|
"grad_norm": 0.6391463329638192,
|
|
"learning_rate": 1.026912631530541e-08,
|
|
"loss": 0.2194,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1035861074924469,
|
|
"step": 3455,
|
|
"valid_targets_mean": 1907.0,
|
|
"valid_targets_min": 1432
|
|
},
|
|
{
|
|
"epoch": 6.948743718592965,
|
|
"grad_norm": 0.6604776128815126,
|
|
"learning_rate": 7.310915687186448e-09,
|
|
"loss": 0.2282,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10321242362260818,
|
|
"step": 3460,
|
|
"valid_targets_mean": 1803.0,
|
|
"valid_targets_min": 967
|
|
},
|
|
{
|
|
"epoch": 6.958793969849246,
|
|
"grad_norm": 0.6579034737961561,
|
|
"learning_rate": 4.8539858021889875e-09,
|
|
"loss": 0.2198,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12899693846702576,
|
|
"step": 3465,
|
|
"valid_targets_mean": 2168.5,
|
|
"valid_targets_min": 1477
|
|
},
|
|
{
|
|
"epoch": 6.9688442211055275,
|
|
"grad_norm": 0.6264526168750094,
|
|
"learning_rate": 2.898398263415647e-09,
|
|
"loss": 0.2257,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10425297915935516,
|
|
"step": 3470,
|
|
"valid_targets_mean": 1905.5,
|
|
"valid_targets_min": 1600
|
|
},
|
|
{
|
|
"epoch": 6.978894472361809,
|
|
"grad_norm": 0.6227196093253512,
|
|
"learning_rate": 1.444202103706438e-09,
|
|
"loss": 0.2168,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10287569463253021,
|
|
"step": 3475,
|
|
"valid_targets_mean": 1907.6,
|
|
"valid_targets_min": 1531
|
|
},
|
|
{
|
|
"epoch": 6.988944723618091,
|
|
"grad_norm": 0.6262261112044841,
|
|
"learning_rate": 4.91433784419737e-10,
|
|
"loss": 0.2164,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.08849115669727325,
|
|
"step": 3480,
|
|
"valid_targets_mean": 1827.2,
|
|
"valid_targets_min": 1522
|
|
},
|
|
{
|
|
"epoch": 6.998994974874372,
|
|
"grad_norm": 0.6588553675866675,
|
|
"learning_rate": 4.0117194506361645e-11,
|
|
"loss": 0.2236,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11387378722429276,
|
|
"step": 3485,
|
|
"valid_targets_mean": 1888.1,
|
|
"valid_targets_min": 1640
|
|
},
|
|
{
|
|
"epoch": 7.0,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20390769839286804,
|
|
"step": 3486,
|
|
"total_flos": 7.25791827817726e+17,
|
|
"train_loss": 0.26778129385640004,
|
|
"train_runtime": 13602.7585,
|
|
"train_samples_per_second": 4.093,
|
|
"train_steps_per_second": 0.256,
|
|
"valid_targets_mean": 1780.6,
|
|
"valid_targets_min": 1241
|
|
}
|
|
],
|
|
"logging_steps": 5,
|
|
"max_steps": 3486,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 7,
|
|
"save_steps": 1500,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": true
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 7.25791827817726e+17,
|
|
"train_batch_size": 1,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|