9101 lines
252 KiB
JSON
9101 lines
252 KiB
JSON
|
{
|
||
|
|
"best_global_step": null,
|
||
|
|
"best_metric": null,
|
||
|
|
"best_model_checkpoint": null,
|
||
|
|
"epoch": 7.0,
|
||
|
|
"eval_steps": 500,
|
||
|
|
"global_step": 4116,
|
||
|
|
"is_hyper_param_search": false,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 0.008503401360544218,
|
||
|
|
"grad_norm": 16.212205368967272,
|
||
|
|
"learning_rate": 3.883495145631068e-07,
|
||
|
|
"loss": 0.6801,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6412181854248047,
|
||
|
|
"step": 5,
|
||
|
|
"valid_targets_mean": 5690.9,
|
||
|
|
"valid_targets_min": 710
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.017006802721088437,
|
||
|
|
"grad_norm": 15.842373298977977,
|
||
|
|
"learning_rate": 8.737864077669904e-07,
|
||
|
|
"loss": 0.653,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.692336916923523,
|
||
|
|
"step": 10,
|
||
|
|
"valid_targets_mean": 5249.8,
|
||
|
|
"valid_targets_min": 583
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.025510204081632654,
|
||
|
|
"grad_norm": 12.903429989220449,
|
||
|
|
"learning_rate": 1.359223300970874e-06,
|
||
|
|
"loss": 0.6461,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.6253611445426941,
|
||
|
|
"step": 15,
|
||
|
|
"valid_targets_mean": 5825.1,
|
||
|
|
"valid_targets_min": 2428
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.034013605442176874,
|
||
|
|
"grad_norm": 9.715570377570725,
|
||
|
|
"learning_rate": 1.8446601941747574e-06,
|
||
|
|
"loss": 0.6176,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5475209951400757,
|
||
|
|
"step": 20,
|
||
|
|
"valid_targets_mean": 5133.3,
|
||
|
|
"valid_targets_min": 384
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04251700680272109,
|
||
|
|
"grad_norm": 6.236119721297098,
|
||
|
|
"learning_rate": 2.330097087378641e-06,
|
||
|
|
"loss": 0.5723,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5103737115859985,
|
||
|
|
"step": 25,
|
||
|
|
"valid_targets_mean": 5818.4,
|
||
|
|
"valid_targets_min": 400
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05102040816326531,
|
||
|
|
"grad_norm": 4.382783869913738,
|
||
|
|
"learning_rate": 2.8155339805825245e-06,
|
||
|
|
"loss": 0.5255,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.5426304340362549,
|
||
|
|
"step": 30,
|
||
|
|
"valid_targets_mean": 4920.6,
|
||
|
|
"valid_targets_min": 586
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05952380952380952,
|
||
|
|
"grad_norm": 2.390655945177813,
|
||
|
|
"learning_rate": 3.300970873786408e-06,
|
||
|
|
"loss": 0.4911,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4295525848865509,
|
||
|
|
"step": 35,
|
||
|
|
"valid_targets_mean": 5206.9,
|
||
|
|
"valid_targets_min": 2542
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06802721088435375,
|
||
|
|
"grad_norm": 1.374954647693056,
|
||
|
|
"learning_rate": 3.7864077669902915e-06,
|
||
|
|
"loss": 0.4598,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4327910244464874,
|
||
|
|
"step": 40,
|
||
|
|
"valid_targets_mean": 6174.8,
|
||
|
|
"valid_targets_min": 2007
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07653061224489796,
|
||
|
|
"grad_norm": 1.1866115615214397,
|
||
|
|
"learning_rate": 4.271844660194175e-06,
|
||
|
|
"loss": 0.4169,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4375618100166321,
|
||
|
|
"step": 45,
|
||
|
|
"valid_targets_mean": 4256.1,
|
||
|
|
"valid_targets_min": 659
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08503401360544217,
|
||
|
|
"grad_norm": 3.2070636208531953,
|
||
|
|
"learning_rate": 4.7572815533980585e-06,
|
||
|
|
"loss": 0.4148,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.4211738109588623,
|
||
|
|
"step": 50,
|
||
|
|
"valid_targets_mean": 5053.5,
|
||
|
|
"valid_targets_min": 408
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0935374149659864,
|
||
|
|
"grad_norm": 0.7712176598442968,
|
||
|
|
"learning_rate": 5.242718446601942e-06,
|
||
|
|
"loss": 0.4079,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.39906126260757446,
|
||
|
|
"step": 55,
|
||
|
|
"valid_targets_mean": 5180.8,
|
||
|
|
"valid_targets_min": 961
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.10204081632653061,
|
||
|
|
"grad_norm": 0.7348089964608792,
|
||
|
|
"learning_rate": 5.728155339805825e-06,
|
||
|
|
"loss": 0.4261,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.42760056257247925,
|
||
|
|
"step": 60,
|
||
|
|
"valid_targets_mean": 5789.3,
|
||
|
|
"valid_targets_min": 836
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11054421768707483,
|
||
|
|
"grad_norm": 1.2138211786074047,
|
||
|
|
"learning_rate": 6.213592233009709e-06,
|
||
|
|
"loss": 0.3852,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.35905641317367554,
|
||
|
|
"step": 65,
|
||
|
|
"valid_targets_mean": 5922.7,
|
||
|
|
"valid_targets_min": 3339
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11904761904761904,
|
||
|
|
"grad_norm": 0.5383757591717289,
|
||
|
|
"learning_rate": 6.6990291262135935e-06,
|
||
|
|
"loss": 0.3784,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.35616347193717957,
|
||
|
|
"step": 70,
|
||
|
|
"valid_targets_mean": 5891.1,
|
||
|
|
"valid_targets_min": 2685
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12755102040816327,
|
||
|
|
"grad_norm": 0.5413947490705122,
|
||
|
|
"learning_rate": 7.184466019417476e-06,
|
||
|
|
"loss": 0.3729,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.354912668466568,
|
||
|
|
"step": 75,
|
||
|
|
"valid_targets_mean": 5267.7,
|
||
|
|
"valid_targets_min": 327
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1360544217687075,
|
||
|
|
"grad_norm": 0.48711465651262886,
|
||
|
|
"learning_rate": 7.66990291262136e-06,
|
||
|
|
"loss": 0.3755,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.35380858182907104,
|
||
|
|
"step": 80,
|
||
|
|
"valid_targets_mean": 5820.1,
|
||
|
|
"valid_targets_min": 441
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1445578231292517,
|
||
|
|
"grad_norm": 0.4967846896224688,
|
||
|
|
"learning_rate": 8.155339805825243e-06,
|
||
|
|
"loss": 0.3572,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.337882936000824,
|
||
|
|
"step": 85,
|
||
|
|
"valid_targets_mean": 5691.3,
|
||
|
|
"valid_targets_min": 2886
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15306122448979592,
|
||
|
|
"grad_norm": 0.5072333662568151,
|
||
|
|
"learning_rate": 8.640776699029127e-06,
|
||
|
|
"loss": 0.3369,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3258446156978607,
|
||
|
|
"step": 90,
|
||
|
|
"valid_targets_mean": 6011.1,
|
||
|
|
"valid_targets_min": 627
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16156462585034015,
|
||
|
|
"grad_norm": 0.5315888354481622,
|
||
|
|
"learning_rate": 9.12621359223301e-06,
|
||
|
|
"loss": 0.3422,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3361511826515198,
|
||
|
|
"step": 95,
|
||
|
|
"valid_targets_mean": 5238.4,
|
||
|
|
"valid_targets_min": 3017
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17006802721088435,
|
||
|
|
"grad_norm": 0.4805018570155199,
|
||
|
|
"learning_rate": 9.611650485436894e-06,
|
||
|
|
"loss": 0.3538,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.33803626894950867,
|
||
|
|
"step": 100,
|
||
|
|
"valid_targets_mean": 6288.4,
|
||
|
|
"valid_targets_min": 2664
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17857142857142858,
|
||
|
|
"grad_norm": 0.5373504958233697,
|
||
|
|
"learning_rate": 1.0097087378640778e-05,
|
||
|
|
"loss": 0.3292,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28921476006507874,
|
||
|
|
"step": 105,
|
||
|
|
"valid_targets_mean": 4569.5,
|
||
|
|
"valid_targets_min": 434
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1870748299319728,
|
||
|
|
"grad_norm": 0.5643618333301738,
|
||
|
|
"learning_rate": 1.0582524271844662e-05,
|
||
|
|
"loss": 0.3515,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.379652738571167,
|
||
|
|
"step": 110,
|
||
|
|
"valid_targets_mean": 4639.3,
|
||
|
|
"valid_targets_min": 371
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.195578231292517,
|
||
|
|
"grad_norm": 0.5502830651998034,
|
||
|
|
"learning_rate": 1.1067961165048544e-05,
|
||
|
|
"loss": 0.3143,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2940191328525543,
|
||
|
|
"step": 115,
|
||
|
|
"valid_targets_mean": 4894.9,
|
||
|
|
"valid_targets_min": 525
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.20408163265306123,
|
||
|
|
"grad_norm": 0.6239959565095892,
|
||
|
|
"learning_rate": 1.1553398058252427e-05,
|
||
|
|
"loss": 0.334,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.32048025727272034,
|
||
|
|
"step": 120,
|
||
|
|
"valid_targets_mean": 4110.2,
|
||
|
|
"valid_targets_min": 314
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21258503401360543,
|
||
|
|
"grad_norm": 0.5131713065712218,
|
||
|
|
"learning_rate": 1.2038834951456311e-05,
|
||
|
|
"loss": 0.3186,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28931280970573425,
|
||
|
|
"step": 125,
|
||
|
|
"valid_targets_mean": 5004.6,
|
||
|
|
"valid_targets_min": 611
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22108843537414966,
|
||
|
|
"grad_norm": 0.48621148375035866,
|
||
|
|
"learning_rate": 1.2524271844660197e-05,
|
||
|
|
"loss": 0.3143,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.32320141792297363,
|
||
|
|
"step": 130,
|
||
|
|
"valid_targets_mean": 5466.6,
|
||
|
|
"valid_targets_min": 1906
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22959183673469388,
|
||
|
|
"grad_norm": 0.4866711162190797,
|
||
|
|
"learning_rate": 1.300970873786408e-05,
|
||
|
|
"loss": 0.3094,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28861409425735474,
|
||
|
|
"step": 135,
|
||
|
|
"valid_targets_mean": 5561.9,
|
||
|
|
"valid_targets_min": 685
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23809523809523808,
|
||
|
|
"grad_norm": 0.5328418386572757,
|
||
|
|
"learning_rate": 1.3495145631067962e-05,
|
||
|
|
"loss": 0.3182,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3140959143638611,
|
||
|
|
"step": 140,
|
||
|
|
"valid_targets_mean": 5382.1,
|
||
|
|
"valid_targets_min": 394
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2465986394557823,
|
||
|
|
"grad_norm": 0.5161857783793827,
|
||
|
|
"learning_rate": 1.3980582524271846e-05,
|
||
|
|
"loss": 0.3221,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2937636971473694,
|
||
|
|
"step": 145,
|
||
|
|
"valid_targets_mean": 5471.7,
|
||
|
|
"valid_targets_min": 524
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25510204081632654,
|
||
|
|
"grad_norm": 0.6456188785969973,
|
||
|
|
"learning_rate": 1.446601941747573e-05,
|
||
|
|
"loss": 0.3119,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3476308584213257,
|
||
|
|
"step": 150,
|
||
|
|
"valid_targets_mean": 5098.4,
|
||
|
|
"valid_targets_min": 1290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.26360544217687076,
|
||
|
|
"grad_norm": 0.4742590945164516,
|
||
|
|
"learning_rate": 1.4951456310679614e-05,
|
||
|
|
"loss": 0.3086,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.308721661567688,
|
||
|
|
"step": 155,
|
||
|
|
"valid_targets_mean": 6538.9,
|
||
|
|
"valid_targets_min": 370
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.272108843537415,
|
||
|
|
"grad_norm": 0.46129544290249297,
|
||
|
|
"learning_rate": 1.5436893203883496e-05,
|
||
|
|
"loss": 0.3204,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27554184198379517,
|
||
|
|
"step": 160,
|
||
|
|
"valid_targets_mean": 5502.1,
|
||
|
|
"valid_targets_min": 1782
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.28061224489795916,
|
||
|
|
"grad_norm": 0.5531859687558784,
|
||
|
|
"learning_rate": 1.592233009708738e-05,
|
||
|
|
"loss": 0.2973,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2988378703594208,
|
||
|
|
"step": 165,
|
||
|
|
"valid_targets_mean": 5475.5,
|
||
|
|
"valid_targets_min": 312
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2891156462585034,
|
||
|
|
"grad_norm": 0.6355507270146583,
|
||
|
|
"learning_rate": 1.6407766990291263e-05,
|
||
|
|
"loss": 0.3212,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.318202942609787,
|
||
|
|
"step": 170,
|
||
|
|
"valid_targets_mean": 4374.1,
|
||
|
|
"valid_targets_min": 772
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2976190476190476,
|
||
|
|
"grad_norm": 1.1079788331454845,
|
||
|
|
"learning_rate": 1.6893203883495145e-05,
|
||
|
|
"loss": 0.2981,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28919780254364014,
|
||
|
|
"step": 175,
|
||
|
|
"valid_targets_mean": 4159.5,
|
||
|
|
"valid_targets_min": 313
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.30612244897959184,
|
||
|
|
"grad_norm": 0.6235845287727317,
|
||
|
|
"learning_rate": 1.737864077669903e-05,
|
||
|
|
"loss": 0.317,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2841646373271942,
|
||
|
|
"step": 180,
|
||
|
|
"valid_targets_mean": 4122.2,
|
||
|
|
"valid_targets_min": 337
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.31462585034013607,
|
||
|
|
"grad_norm": 0.5901599076058308,
|
||
|
|
"learning_rate": 1.7864077669902916e-05,
|
||
|
|
"loss": 0.3243,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3466235101222992,
|
||
|
|
"step": 185,
|
||
|
|
"valid_targets_mean": 4435.6,
|
||
|
|
"valid_targets_min": 364
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3231292517006803,
|
||
|
|
"grad_norm": 0.5490681931677598,
|
||
|
|
"learning_rate": 1.8349514563106798e-05,
|
||
|
|
"loss": 0.2964,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3018025755882263,
|
||
|
|
"step": 190,
|
||
|
|
"valid_targets_mean": 4696.1,
|
||
|
|
"valid_targets_min": 414
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.33163265306122447,
|
||
|
|
"grad_norm": 0.5623459523103859,
|
||
|
|
"learning_rate": 1.883495145631068e-05,
|
||
|
|
"loss": 0.3013,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28652340173721313,
|
||
|
|
"step": 195,
|
||
|
|
"valid_targets_mean": 4946.8,
|
||
|
|
"valid_targets_min": 333
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3401360544217687,
|
||
|
|
"grad_norm": 0.5760085134024041,
|
||
|
|
"learning_rate": 1.9320388349514565e-05,
|
||
|
|
"loss": 0.3007,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2942473888397217,
|
||
|
|
"step": 200,
|
||
|
|
"valid_targets_mean": 4843.6,
|
||
|
|
"valid_targets_min": 509
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3486394557823129,
|
||
|
|
"grad_norm": 0.5157725752831408,
|
||
|
|
"learning_rate": 1.9805825242718447e-05,
|
||
|
|
"loss": 0.2896,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2663571834564209,
|
||
|
|
"step": 205,
|
||
|
|
"valid_targets_mean": 5218.6,
|
||
|
|
"valid_targets_min": 887
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.35714285714285715,
|
||
|
|
"grad_norm": 0.46995942803497975,
|
||
|
|
"learning_rate": 2.0291262135922333e-05,
|
||
|
|
"loss": 0.2792,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25146517157554626,
|
||
|
|
"step": 210,
|
||
|
|
"valid_targets_mean": 5228.5,
|
||
|
|
"valid_targets_min": 519
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3656462585034014,
|
||
|
|
"grad_norm": 0.4563152185302371,
|
||
|
|
"learning_rate": 2.0776699029126215e-05,
|
||
|
|
"loss": 0.2784,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30271318554878235,
|
||
|
|
"step": 215,
|
||
|
|
"valid_targets_mean": 6504.3,
|
||
|
|
"valid_targets_min": 3941
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3741496598639456,
|
||
|
|
"grad_norm": 0.5039103715374049,
|
||
|
|
"learning_rate": 2.1262135922330097e-05,
|
||
|
|
"loss": 0.2791,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26329848170280457,
|
||
|
|
"step": 220,
|
||
|
|
"valid_targets_mean": 5225.2,
|
||
|
|
"valid_targets_min": 2285
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3826530612244898,
|
||
|
|
"grad_norm": 0.49381520150707864,
|
||
|
|
"learning_rate": 2.1747572815533982e-05,
|
||
|
|
"loss": 0.2877,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26511698961257935,
|
||
|
|
"step": 225,
|
||
|
|
"valid_targets_mean": 5408.8,
|
||
|
|
"valid_targets_min": 765
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.391156462585034,
|
||
|
|
"grad_norm": 0.5806381202734253,
|
||
|
|
"learning_rate": 2.2233009708737864e-05,
|
||
|
|
"loss": 0.2772,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29458144307136536,
|
||
|
|
"step": 230,
|
||
|
|
"valid_targets_mean": 4349.2,
|
||
|
|
"valid_targets_min": 299
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39965986394557823,
|
||
|
|
"grad_norm": 0.5504764645155861,
|
||
|
|
"learning_rate": 2.271844660194175e-05,
|
||
|
|
"loss": 0.2864,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.32245898246765137,
|
||
|
|
"step": 235,
|
||
|
|
"valid_targets_mean": 5735.0,
|
||
|
|
"valid_targets_min": 903
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.40816326530612246,
|
||
|
|
"grad_norm": 0.4790743722282976,
|
||
|
|
"learning_rate": 2.3203883495145632e-05,
|
||
|
|
"loss": 0.2837,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2824363112449646,
|
||
|
|
"step": 240,
|
||
|
|
"valid_targets_mean": 5662.0,
|
||
|
|
"valid_targets_min": 425
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4166666666666667,
|
||
|
|
"grad_norm": 0.5851203608074932,
|
||
|
|
"learning_rate": 2.3689320388349514e-05,
|
||
|
|
"loss": 0.27,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24771523475646973,
|
||
|
|
"step": 245,
|
||
|
|
"valid_targets_mean": 5077.2,
|
||
|
|
"valid_targets_min": 430
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.42517006802721086,
|
||
|
|
"grad_norm": 0.6510658662088212,
|
||
|
|
"learning_rate": 2.41747572815534e-05,
|
||
|
|
"loss": 0.2859,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30241885781288147,
|
||
|
|
"step": 250,
|
||
|
|
"valid_targets_mean": 4475.7,
|
||
|
|
"valid_targets_min": 505
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4336734693877551,
|
||
|
|
"grad_norm": 0.510290702920634,
|
||
|
|
"learning_rate": 2.466019417475728e-05,
|
||
|
|
"loss": 0.2782,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2837359309196472,
|
||
|
|
"step": 255,
|
||
|
|
"valid_targets_mean": 6328.8,
|
||
|
|
"valid_targets_min": 2729
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4421768707482993,
|
||
|
|
"grad_norm": 0.585268749606104,
|
||
|
|
"learning_rate": 2.514563106796117e-05,
|
||
|
|
"loss": 0.2709,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2812116742134094,
|
||
|
|
"step": 260,
|
||
|
|
"valid_targets_mean": 4890.2,
|
||
|
|
"valid_targets_min": 406
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45068027210884354,
|
||
|
|
"grad_norm": 0.5345677901656546,
|
||
|
|
"learning_rate": 2.5631067961165052e-05,
|
||
|
|
"loss": 0.2754,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2353639304637909,
|
||
|
|
"step": 265,
|
||
|
|
"valid_targets_mean": 5539.2,
|
||
|
|
"valid_targets_min": 283
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45918367346938777,
|
||
|
|
"grad_norm": 0.6185441068286449,
|
||
|
|
"learning_rate": 2.6116504854368934e-05,
|
||
|
|
"loss": 0.296,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.32710278034210205,
|
||
|
|
"step": 270,
|
||
|
|
"valid_targets_mean": 4608.0,
|
||
|
|
"valid_targets_min": 865
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.467687074829932,
|
||
|
|
"grad_norm": 0.5155457113492166,
|
||
|
|
"learning_rate": 2.660194174757282e-05,
|
||
|
|
"loss": 0.2862,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28434938192367554,
|
||
|
|
"step": 275,
|
||
|
|
"valid_targets_mean": 5376.1,
|
||
|
|
"valid_targets_min": 791
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47619047619047616,
|
||
|
|
"grad_norm": 0.6795869917034234,
|
||
|
|
"learning_rate": 2.7087378640776702e-05,
|
||
|
|
"loss": 0.2799,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27224963903427124,
|
||
|
|
"step": 280,
|
||
|
|
"valid_targets_mean": 4397.5,
|
||
|
|
"valid_targets_min": 381
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4846938775510204,
|
||
|
|
"grad_norm": 0.48837782033581506,
|
||
|
|
"learning_rate": 2.7572815533980587e-05,
|
||
|
|
"loss": 0.2717,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28289148211479187,
|
||
|
|
"step": 285,
|
||
|
|
"valid_targets_mean": 5553.6,
|
||
|
|
"valid_targets_min": 995
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4931972789115646,
|
||
|
|
"grad_norm": 0.4934805021963428,
|
||
|
|
"learning_rate": 2.805825242718447e-05,
|
||
|
|
"loss": 0.2586,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24568292498588562,
|
||
|
|
"step": 290,
|
||
|
|
"valid_targets_mean": 5729.4,
|
||
|
|
"valid_targets_min": 2301
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5017006802721088,
|
||
|
|
"grad_norm": 0.5282463409598596,
|
||
|
|
"learning_rate": 2.854368932038835e-05,
|
||
|
|
"loss": 0.28,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24484096467494965,
|
||
|
|
"step": 295,
|
||
|
|
"valid_targets_mean": 5248.2,
|
||
|
|
"valid_targets_min": 1929
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5102040816326531,
|
||
|
|
"grad_norm": 0.7625290518327503,
|
||
|
|
"learning_rate": 2.9029126213592237e-05,
|
||
|
|
"loss": 0.2884,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2803293466567993,
|
||
|
|
"step": 300,
|
||
|
|
"valid_targets_mean": 5017.3,
|
||
|
|
"valid_targets_min": 427
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5187074829931972,
|
||
|
|
"grad_norm": 0.5233549084880463,
|
||
|
|
"learning_rate": 2.951456310679612e-05,
|
||
|
|
"loss": 0.2719,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2852564752101898,
|
||
|
|
"step": 305,
|
||
|
|
"valid_targets_mean": 5715.1,
|
||
|
|
"valid_targets_min": 1476
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5272108843537415,
|
||
|
|
"grad_norm": 0.5830259308259479,
|
||
|
|
"learning_rate": 3.0000000000000004e-05,
|
||
|
|
"loss": 0.2805,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27544933557510376,
|
||
|
|
"step": 310,
|
||
|
|
"valid_targets_mean": 4726.8,
|
||
|
|
"valid_targets_min": 478
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5357142857142857,
|
||
|
|
"grad_norm": 0.5069745228460134,
|
||
|
|
"learning_rate": 3.0485436893203886e-05,
|
||
|
|
"loss": 0.271,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29192113876342773,
|
||
|
|
"step": 315,
|
||
|
|
"valid_targets_mean": 5274.2,
|
||
|
|
"valid_targets_min": 365
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.54421768707483,
|
||
|
|
"grad_norm": 0.6046054401073275,
|
||
|
|
"learning_rate": 3.097087378640777e-05,
|
||
|
|
"loss": 0.2694,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2385951280593872,
|
||
|
|
"step": 320,
|
||
|
|
"valid_targets_mean": 4602.9,
|
||
|
|
"valid_targets_min": 506
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5527210884353742,
|
||
|
|
"grad_norm": 0.49505739709447727,
|
||
|
|
"learning_rate": 3.1456310679611654e-05,
|
||
|
|
"loss": 0.2669,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2404285967350006,
|
||
|
|
"step": 325,
|
||
|
|
"valid_targets_mean": 5299.4,
|
||
|
|
"valid_targets_min": 2585
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5612244897959183,
|
||
|
|
"grad_norm": 0.5731965988979448,
|
||
|
|
"learning_rate": 3.194174757281554e-05,
|
||
|
|
"loss": 0.2687,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2439209371805191,
|
||
|
|
"step": 330,
|
||
|
|
"valid_targets_mean": 3687.1,
|
||
|
|
"valid_targets_min": 337
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5697278911564626,
|
||
|
|
"grad_norm": 0.4835304573292077,
|
||
|
|
"learning_rate": 3.242718446601942e-05,
|
||
|
|
"loss": 0.2534,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2265438735485077,
|
||
|
|
"step": 335,
|
||
|
|
"valid_targets_mean": 6027.6,
|
||
|
|
"valid_targets_min": 2709
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5782312925170068,
|
||
|
|
"grad_norm": 0.4708764477854292,
|
||
|
|
"learning_rate": 3.29126213592233e-05,
|
||
|
|
"loss": 0.2688,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2452189177274704,
|
||
|
|
"step": 340,
|
||
|
|
"valid_targets_mean": 5955.9,
|
||
|
|
"valid_targets_min": 2163
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5867346938775511,
|
||
|
|
"grad_norm": 0.5640776181987645,
|
||
|
|
"learning_rate": 3.339805825242719e-05,
|
||
|
|
"loss": 0.2672,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2595880329608917,
|
||
|
|
"step": 345,
|
||
|
|
"valid_targets_mean": 4005.2,
|
||
|
|
"valid_targets_min": 334
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5952380952380952,
|
||
|
|
"grad_norm": 0.54852635814411,
|
||
|
|
"learning_rate": 3.388349514563107e-05,
|
||
|
|
"loss": 0.2594,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24560917913913727,
|
||
|
|
"step": 350,
|
||
|
|
"valid_targets_mean": 4649.9,
|
||
|
|
"valid_targets_min": 501
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6037414965986394,
|
||
|
|
"grad_norm": 0.6564594447948926,
|
||
|
|
"learning_rate": 3.436893203883495e-05,
|
||
|
|
"loss": 0.265,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2703518867492676,
|
||
|
|
"step": 355,
|
||
|
|
"valid_targets_mean": 4632.9,
|
||
|
|
"valid_targets_min": 349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6122448979591837,
|
||
|
|
"grad_norm": 0.6842985090301186,
|
||
|
|
"learning_rate": 3.485436893203884e-05,
|
||
|
|
"loss": 0.2826,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26784777641296387,
|
||
|
|
"step": 360,
|
||
|
|
"valid_targets_mean": 5690.6,
|
||
|
|
"valid_targets_min": 1324
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6207482993197279,
|
||
|
|
"grad_norm": 0.7495655689907519,
|
||
|
|
"learning_rate": 3.5339805825242724e-05,
|
||
|
|
"loss": 0.2497,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2552953064441681,
|
||
|
|
"step": 365,
|
||
|
|
"valid_targets_mean": 4928.0,
|
||
|
|
"valid_targets_min": 478
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6292517006802721,
|
||
|
|
"grad_norm": 0.5681207660340626,
|
||
|
|
"learning_rate": 3.582524271844661e-05,
|
||
|
|
"loss": 0.2689,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27484530210494995,
|
||
|
|
"step": 370,
|
||
|
|
"valid_targets_mean": 5191.2,
|
||
|
|
"valid_targets_min": 458
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6377551020408163,
|
||
|
|
"grad_norm": 0.4679434844815811,
|
||
|
|
"learning_rate": 3.631067961165049e-05,
|
||
|
|
"loss": 0.2529,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2521194815635681,
|
||
|
|
"step": 375,
|
||
|
|
"valid_targets_mean": 5558.8,
|
||
|
|
"valid_targets_min": 359
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6462585034013606,
|
||
|
|
"grad_norm": 0.5479389467634336,
|
||
|
|
"learning_rate": 3.679611650485437e-05,
|
||
|
|
"loss": 0.2595,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24181511998176575,
|
||
|
|
"step": 380,
|
||
|
|
"valid_targets_mean": 4694.6,
|
||
|
|
"valid_targets_min": 836
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6547619047619048,
|
||
|
|
"grad_norm": 0.48604498113655703,
|
||
|
|
"learning_rate": 3.728155339805826e-05,
|
||
|
|
"loss": 0.2472,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23702500760555267,
|
||
|
|
"step": 385,
|
||
|
|
"valid_targets_mean": 5850.6,
|
||
|
|
"valid_targets_min": 1970
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6632653061224489,
|
||
|
|
"grad_norm": 0.5475925588407128,
|
||
|
|
"learning_rate": 3.776699029126214e-05,
|
||
|
|
"loss": 0.2682,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26854199171066284,
|
||
|
|
"step": 390,
|
||
|
|
"valid_targets_mean": 4486.4,
|
||
|
|
"valid_targets_min": 286
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6717687074829932,
|
||
|
|
"grad_norm": 0.5247416056617925,
|
||
|
|
"learning_rate": 3.825242718446602e-05,
|
||
|
|
"loss": 0.2691,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25068363547325134,
|
||
|
|
"step": 395,
|
||
|
|
"valid_targets_mean": 5590.1,
|
||
|
|
"valid_targets_min": 2582
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6802721088435374,
|
||
|
|
"grad_norm": 0.48478226890904935,
|
||
|
|
"learning_rate": 3.873786407766991e-05,
|
||
|
|
"loss": 0.2539,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.261441171169281,
|
||
|
|
"step": 400,
|
||
|
|
"valid_targets_mean": 5585.8,
|
||
|
|
"valid_targets_min": 2802
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6887755102040817,
|
||
|
|
"grad_norm": 0.5397048376380215,
|
||
|
|
"learning_rate": 3.9223300970873787e-05,
|
||
|
|
"loss": 0.2598,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25783365964889526,
|
||
|
|
"step": 405,
|
||
|
|
"valid_targets_mean": 4574.6,
|
||
|
|
"valid_targets_min": 434
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6972789115646258,
|
||
|
|
"grad_norm": 0.5069769493372015,
|
||
|
|
"learning_rate": 3.970873786407767e-05,
|
||
|
|
"loss": 0.2424,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23376122117042542,
|
||
|
|
"step": 410,
|
||
|
|
"valid_targets_mean": 5286.1,
|
||
|
|
"valid_targets_min": 2536
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.70578231292517,
|
||
|
|
"grad_norm": 0.65118075980262,
|
||
|
|
"learning_rate": 3.9999971224844676e-05,
|
||
|
|
"loss": 0.2615,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28898805379867554,
|
||
|
|
"step": 415,
|
||
|
|
"valid_targets_mean": 3882.5,
|
||
|
|
"valid_targets_min": 260
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7142857142857143,
|
||
|
|
"grad_norm": 0.48077467829708165,
|
||
|
|
"learning_rate": 3.9999647505298214e-05,
|
||
|
|
"loss": 0.2535,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23364660143852234,
|
||
|
|
"step": 420,
|
||
|
|
"valid_targets_mean": 5486.5,
|
||
|
|
"valid_targets_min": 554
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7227891156462585,
|
||
|
|
"grad_norm": 0.5611327067894291,
|
||
|
|
"learning_rate": 3.999896410310244e-05,
|
||
|
|
"loss": 0.2614,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28746485710144043,
|
||
|
|
"step": 425,
|
||
|
|
"valid_targets_mean": 5517.4,
|
||
|
|
"valid_targets_min": 552
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7312925170068028,
|
||
|
|
"grad_norm": 0.4977137457808474,
|
||
|
|
"learning_rate": 3.9997921030547975e-05,
|
||
|
|
"loss": 0.2477,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22914865612983704,
|
||
|
|
"step": 430,
|
||
|
|
"valid_targets_mean": 5473.6,
|
||
|
|
"valid_targets_min": 1880
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7397959183673469,
|
||
|
|
"grad_norm": 0.5011757518547365,
|
||
|
|
"learning_rate": 3.9996518306393906e-05,
|
||
|
|
"loss": 0.2437,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21840690076351166,
|
||
|
|
"step": 435,
|
||
|
|
"valid_targets_mean": 5157.0,
|
||
|
|
"valid_targets_min": 423
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7482993197278912,
|
||
|
|
"grad_norm": 0.555592853915884,
|
||
|
|
"learning_rate": 3.999475595586746e-05,
|
||
|
|
"loss": 0.2681,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29114052653312683,
|
||
|
|
"step": 440,
|
||
|
|
"valid_targets_mean": 5807.2,
|
||
|
|
"valid_targets_min": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7568027210884354,
|
||
|
|
"grad_norm": 0.5111589074437468,
|
||
|
|
"learning_rate": 3.9992634010663535e-05,
|
||
|
|
"loss": 0.2667,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2571555972099304,
|
||
|
|
"step": 445,
|
||
|
|
"valid_targets_mean": 6047.1,
|
||
|
|
"valid_targets_min": 4080
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7653061224489796,
|
||
|
|
"grad_norm": 0.6007728114916696,
|
||
|
|
"learning_rate": 3.999015250894415e-05,
|
||
|
|
"loss": 0.2571,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2505347728729248,
|
||
|
|
"step": 450,
|
||
|
|
"valid_targets_mean": 4910.6,
|
||
|
|
"valid_targets_min": 985
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7738095238095238,
|
||
|
|
"grad_norm": 0.6053846092239997,
|
||
|
|
"learning_rate": 3.998731149533774e-05,
|
||
|
|
"loss": 0.2572,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24147354066371918,
|
||
|
|
"step": 455,
|
||
|
|
"valid_targets_mean": 4135.2,
|
||
|
|
"valid_targets_min": 303
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.782312925170068,
|
||
|
|
"grad_norm": 0.49756458026003453,
|
||
|
|
"learning_rate": 3.9984111020938376e-05,
|
||
|
|
"loss": 0.2295,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23897001147270203,
|
||
|
|
"step": 460,
|
||
|
|
"valid_targets_mean": 4759.7,
|
||
|
|
"valid_targets_min": 746
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7908163265306123,
|
||
|
|
"grad_norm": 0.5524914217101976,
|
||
|
|
"learning_rate": 3.998055114330483e-05,
|
||
|
|
"loss": 0.2501,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2931286096572876,
|
||
|
|
"step": 465,
|
||
|
|
"valid_targets_mean": 5170.7,
|
||
|
|
"valid_targets_min": 808
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7993197278911565,
|
||
|
|
"grad_norm": 0.5336306086484336,
|
||
|
|
"learning_rate": 3.9976631926459536e-05,
|
||
|
|
"loss": 0.2444,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2402130365371704,
|
||
|
|
"step": 470,
|
||
|
|
"valid_targets_mean": 5531.8,
|
||
|
|
"valid_targets_min": 522
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8078231292517006,
|
||
|
|
"grad_norm": 0.4820768482572941,
|
||
|
|
"learning_rate": 3.997235344088747e-05,
|
||
|
|
"loss": 0.255,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2625737488269806,
|
||
|
|
"step": 475,
|
||
|
|
"valid_targets_mean": 5446.8,
|
||
|
|
"valid_targets_min": 2158
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8163265306122449,
|
||
|
|
"grad_norm": 0.46972229321153836,
|
||
|
|
"learning_rate": 3.9967715763534805e-05,
|
||
|
|
"loss": 0.2471,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21720445156097412,
|
||
|
|
"step": 480,
|
||
|
|
"valid_targets_mean": 5028.1,
|
||
|
|
"valid_targets_min": 1146
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8248299319727891,
|
||
|
|
"grad_norm": 0.4881651347469552,
|
||
|
|
"learning_rate": 3.9962718977807637e-05,
|
||
|
|
"loss": 0.2418,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23963674902915955,
|
||
|
|
"step": 485,
|
||
|
|
"valid_targets_mean": 4634.7,
|
||
|
|
"valid_targets_min": 792
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8333333333333334,
|
||
|
|
"grad_norm": 0.5802649358804667,
|
||
|
|
"learning_rate": 3.995736317357041e-05,
|
||
|
|
"loss": 0.2394,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2413857877254486,
|
||
|
|
"step": 490,
|
||
|
|
"valid_targets_mean": 4176.6,
|
||
|
|
"valid_targets_min": 316
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8418367346938775,
|
||
|
|
"grad_norm": 0.6235712790183929,
|
||
|
|
"learning_rate": 3.9951648447144305e-05,
|
||
|
|
"loss": 0.2535,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.28814971446990967,
|
||
|
|
"step": 495,
|
||
|
|
"valid_targets_mean": 4807.1,
|
||
|
|
"valid_targets_min": 341
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8503401360544217,
|
||
|
|
"grad_norm": 0.48717989065979084,
|
||
|
|
"learning_rate": 3.9945574901305534e-05,
|
||
|
|
"loss": 0.2452,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24348053336143494,
|
||
|
|
"step": 500,
|
||
|
|
"valid_targets_mean": 5724.1,
|
||
|
|
"valid_targets_min": 1837
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.858843537414966,
|
||
|
|
"grad_norm": 0.5704527175621606,
|
||
|
|
"learning_rate": 3.9939142645283475e-05,
|
||
|
|
"loss": 0.2761,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2759855389595032,
|
||
|
|
"step": 505,
|
||
|
|
"valid_targets_mean": 4416.4,
|
||
|
|
"valid_targets_min": 375
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8673469387755102,
|
||
|
|
"grad_norm": 0.5075135308688633,
|
||
|
|
"learning_rate": 3.9932351794758705e-05,
|
||
|
|
"loss": 0.2409,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24038991332054138,
|
||
|
|
"step": 510,
|
||
|
|
"valid_targets_mean": 4855.7,
|
||
|
|
"valid_targets_min": 257
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8758503401360545,
|
||
|
|
"grad_norm": 0.5873777799417169,
|
||
|
|
"learning_rate": 3.992520247186094e-05,
|
||
|
|
"loss": 0.2383,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23789118230342865,
|
||
|
|
"step": 515,
|
||
|
|
"valid_targets_mean": 4291.9,
|
||
|
|
"valid_targets_min": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8843537414965986,
|
||
|
|
"grad_norm": 0.5715901285900169,
|
||
|
|
"learning_rate": 3.991769480516681e-05,
|
||
|
|
"loss": 0.251,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.26526695489883423,
|
||
|
|
"step": 520,
|
||
|
|
"valid_targets_mean": 4385.2,
|
||
|
|
"valid_targets_min": 477
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8928571428571429,
|
||
|
|
"grad_norm": 0.4687521725115409,
|
||
|
|
"learning_rate": 3.9909828929697575e-05,
|
||
|
|
"loss": 0.2501,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2208852469921112,
|
||
|
|
"step": 525,
|
||
|
|
"valid_targets_mean": 5751.4,
|
||
|
|
"valid_targets_min": 685
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9013605442176871,
|
||
|
|
"grad_norm": 0.5214967334875037,
|
||
|
|
"learning_rate": 3.9901604986916666e-05,
|
||
|
|
"loss": 0.2543,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.27274268865585327,
|
||
|
|
"step": 530,
|
||
|
|
"valid_targets_mean": 5532.2,
|
||
|
|
"valid_targets_min": 1819
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9098639455782312,
|
||
|
|
"grad_norm": 0.7819545600315054,
|
||
|
|
"learning_rate": 3.9893023124727176e-05,
|
||
|
|
"loss": 0.2385,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2478397935628891,
|
||
|
|
"step": 535,
|
||
|
|
"valid_targets_mean": 4693.9,
|
||
|
|
"valid_targets_min": 425
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9183673469387755,
|
||
|
|
"grad_norm": 0.5629555644653216,
|
||
|
|
"learning_rate": 3.988408349746916e-05,
|
||
|
|
"loss": 0.2367,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2255719006061554,
|
||
|
|
"step": 540,
|
||
|
|
"valid_targets_mean": 4195.3,
|
||
|
|
"valid_targets_min": 278
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9268707482993197,
|
||
|
|
"grad_norm": 0.4831228933640306,
|
||
|
|
"learning_rate": 3.9874786265916895e-05,
|
||
|
|
"loss": 0.2577,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22280704975128174,
|
||
|
|
"step": 545,
|
||
|
|
"valid_targets_mean": 5950.4,
|
||
|
|
"valid_targets_min": 2833
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.935374149659864,
|
||
|
|
"grad_norm": 0.5237918966472492,
|
||
|
|
"learning_rate": 3.986513159727598e-05,
|
||
|
|
"loss": 0.2431,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2568817734718323,
|
||
|
|
"step": 550,
|
||
|
|
"valid_targets_mean": 4737.4,
|
||
|
|
"valid_targets_min": 313
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9438775510204082,
|
||
|
|
"grad_norm": 0.4778343709264419,
|
||
|
|
"learning_rate": 3.985511966518031e-05,
|
||
|
|
"loss": 0.2308,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22147250175476074,
|
||
|
|
"step": 555,
|
||
|
|
"valid_targets_mean": 5475.8,
|
||
|
|
"valid_targets_min": 642
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9523809523809523,
|
||
|
|
"grad_norm": 0.5126825038505003,
|
||
|
|
"learning_rate": 3.984475064968897e-05,
|
||
|
|
"loss": 0.2532,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.242509663105011,
|
||
|
|
"step": 560,
|
||
|
|
"valid_targets_mean": 4754.2,
|
||
|
|
"valid_targets_min": 307
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9608843537414966,
|
||
|
|
"grad_norm": 0.5023131295917506,
|
||
|
|
"learning_rate": 3.9834024737282994e-05,
|
||
|
|
"loss": 0.2338,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22972457110881805,
|
||
|
|
"step": 565,
|
||
|
|
"valid_targets_mean": 4817.7,
|
||
|
|
"valid_targets_min": 746
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9693877551020408,
|
||
|
|
"grad_norm": 0.5113251620543043,
|
||
|
|
"learning_rate": 3.9822942120862016e-05,
|
||
|
|
"loss": 0.2443,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23256660997867584,
|
||
|
|
"step": 570,
|
||
|
|
"valid_targets_mean": 5196.7,
|
||
|
|
"valid_targets_min": 367
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9778911564625851,
|
||
|
|
"grad_norm": 0.49802820327858593,
|
||
|
|
"learning_rate": 3.981150299974079e-05,
|
||
|
|
"loss": 0.2446,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2441190779209137,
|
||
|
|
"step": 575,
|
||
|
|
"valid_targets_mean": 6362.8,
|
||
|
|
"valid_targets_min": 3477
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9863945578231292,
|
||
|
|
"grad_norm": 0.4871147367096653,
|
||
|
|
"learning_rate": 3.979970757964561e-05,
|
||
|
|
"loss": 0.236,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23753859102725983,
|
||
|
|
"step": 580,
|
||
|
|
"valid_targets_mean": 5088.4,
|
||
|
|
"valid_targets_min": 501
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9948979591836735,
|
||
|
|
"grad_norm": 0.4814043848594895,
|
||
|
|
"learning_rate": 3.978755607271061e-05,
|
||
|
|
"loss": 0.2339,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22681377828121185,
|
||
|
|
"step": 585,
|
||
|
|
"valid_targets_mean": 5906.8,
|
||
|
|
"valid_targets_min": 769
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0034013605442176,
|
||
|
|
"grad_norm": 0.5041002639607305,
|
||
|
|
"learning_rate": 3.977504869747395e-05,
|
||
|
|
"loss": 0.2437,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21633821725845337,
|
||
|
|
"step": 590,
|
||
|
|
"valid_targets_mean": 5083.9,
|
||
|
|
"valid_targets_min": 473
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0119047619047619,
|
||
|
|
"grad_norm": 0.5248414566248647,
|
||
|
|
"learning_rate": 3.976218567887389e-05,
|
||
|
|
"loss": 0.2302,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24877363443374634,
|
||
|
|
"step": 595,
|
||
|
|
"valid_targets_mean": 5060.4,
|
||
|
|
"valid_targets_min": 299
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0204081632653061,
|
||
|
|
"grad_norm": 0.4627153326228623,
|
||
|
|
"learning_rate": 3.974896724824475e-05,
|
||
|
|
"loss": 0.2186,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.206419438123703,
|
||
|
|
"step": 600,
|
||
|
|
"valid_targets_mean": 5692.1,
|
||
|
|
"valid_targets_min": 542
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0289115646258504,
|
||
|
|
"grad_norm": 0.49603632483484655,
|
||
|
|
"learning_rate": 3.97353936433127e-05,
|
||
|
|
"loss": 0.2217,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21563109755516052,
|
||
|
|
"step": 605,
|
||
|
|
"valid_targets_mean": 5386.2,
|
||
|
|
"valid_targets_min": 313
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0374149659863945,
|
||
|
|
"grad_norm": 0.49644652956510316,
|
||
|
|
"learning_rate": 3.972146510819158e-05,
|
||
|
|
"loss": 0.2344,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2030610293149948,
|
||
|
|
"step": 610,
|
||
|
|
"valid_targets_mean": 5466.8,
|
||
|
|
"valid_targets_min": 749
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0459183673469388,
|
||
|
|
"grad_norm": 0.5383639948743353,
|
||
|
|
"learning_rate": 3.970718189337841e-05,
|
||
|
|
"loss": 0.228,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2408256232738495,
|
||
|
|
"step": 615,
|
||
|
|
"valid_targets_mean": 4846.4,
|
||
|
|
"valid_targets_min": 427
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.054421768707483,
|
||
|
|
"grad_norm": 0.5372974253718178,
|
||
|
|
"learning_rate": 3.969254425574895e-05,
|
||
|
|
"loss": 0.2323,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21638332307338715,
|
||
|
|
"step": 620,
|
||
|
|
"valid_targets_mean": 4916.7,
|
||
|
|
"valid_targets_min": 526
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0629251700680271,
|
||
|
|
"grad_norm": 0.5138048113048331,
|
||
|
|
"learning_rate": 3.9677552458553046e-05,
|
||
|
|
"loss": 0.2479,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2712721824645996,
|
||
|
|
"step": 625,
|
||
|
|
"valid_targets_mean": 5917.1,
|
||
|
|
"valid_targets_min": 1621
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0714285714285714,
|
||
|
|
"grad_norm": 0.7049486920370831,
|
||
|
|
"learning_rate": 3.9662206771409925e-05,
|
||
|
|
"loss": 0.2375,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2678952217102051,
|
||
|
|
"step": 630,
|
||
|
|
"valid_targets_mean": 4505.2,
|
||
|
|
"valid_targets_min": 353
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0799319727891157,
|
||
|
|
"grad_norm": 0.5122947938749143,
|
||
|
|
"learning_rate": 3.964650747030332e-05,
|
||
|
|
"loss": 0.2341,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2512699067592621,
|
||
|
|
"step": 635,
|
||
|
|
"valid_targets_mean": 4917.4,
|
||
|
|
"valid_targets_min": 1056
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.08843537414966,
|
||
|
|
"grad_norm": 0.5488947084860933,
|
||
|
|
"learning_rate": 3.963045483757652e-05,
|
||
|
|
"loss": 0.2394,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.262661874294281,
|
||
|
|
"step": 640,
|
||
|
|
"valid_targets_mean": 5650.9,
|
||
|
|
"valid_targets_min": 1599
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.096938775510204,
|
||
|
|
"grad_norm": 0.5126893930327161,
|
||
|
|
"learning_rate": 3.961404916192728e-05,
|
||
|
|
"loss": 0.2257,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22291921079158783,
|
||
|
|
"step": 645,
|
||
|
|
"valid_targets_mean": 4957.9,
|
||
|
|
"valid_targets_min": 270
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1054421768707483,
|
||
|
|
"grad_norm": 0.5663483864272819,
|
||
|
|
"learning_rate": 3.959729073840265e-05,
|
||
|
|
"loss": 0.2254,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22292132675647736,
|
||
|
|
"step": 650,
|
||
|
|
"valid_targets_mean": 5356.8,
|
||
|
|
"valid_targets_min": 341
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1139455782312926,
|
||
|
|
"grad_norm": 0.4683224045320832,
|
||
|
|
"learning_rate": 3.958017986839364e-05,
|
||
|
|
"loss": 0.2114,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22692638635635376,
|
||
|
|
"step": 655,
|
||
|
|
"valid_targets_mean": 6229.6,
|
||
|
|
"valid_targets_min": 694
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1224489795918366,
|
||
|
|
"grad_norm": 0.5062671953623455,
|
||
|
|
"learning_rate": 3.956271685962985e-05,
|
||
|
|
"loss": 0.2183,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19825373589992523,
|
||
|
|
"step": 660,
|
||
|
|
"valid_targets_mean": 5436.9,
|
||
|
|
"valid_targets_min": 2273
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.130952380952381,
|
||
|
|
"grad_norm": 0.551797861331001,
|
||
|
|
"learning_rate": 3.954490202617385e-05,
|
||
|
|
"loss": 0.2085,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22262123227119446,
|
||
|
|
"step": 665,
|
||
|
|
"valid_targets_mean": 5162.3,
|
||
|
|
"valid_targets_min": 444
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1394557823129252,
|
||
|
|
"grad_norm": 0.46838767310672824,
|
||
|
|
"learning_rate": 3.9526735688415624e-05,
|
||
|
|
"loss": 0.2177,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20003627240657806,
|
||
|
|
"step": 670,
|
||
|
|
"valid_targets_mean": 4588.8,
|
||
|
|
"valid_targets_min": 681
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1479591836734695,
|
||
|
|
"grad_norm": 0.5540754917306583,
|
||
|
|
"learning_rate": 3.9508218173066766e-05,
|
||
|
|
"loss": 0.2215,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21115349233150482,
|
||
|
|
"step": 675,
|
||
|
|
"valid_targets_mean": 5819.2,
|
||
|
|
"valid_targets_min": 2819
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1564625850340136,
|
||
|
|
"grad_norm": 0.47451094142681177,
|
||
|
|
"learning_rate": 3.948934981315457e-05,
|
||
|
|
"loss": 0.2203,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21901313960552216,
|
||
|
|
"step": 680,
|
||
|
|
"valid_targets_mean": 5100.9,
|
||
|
|
"valid_targets_min": 333
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1649659863945578,
|
||
|
|
"grad_norm": 0.46540556367924035,
|
||
|
|
"learning_rate": 3.947013094801612e-05,
|
||
|
|
"loss": 0.2126,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20066355168819427,
|
||
|
|
"step": 685,
|
||
|
|
"valid_targets_mean": 5669.0,
|
||
|
|
"valid_targets_min": 1302
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1734693877551021,
|
||
|
|
"grad_norm": 0.4712816695900599,
|
||
|
|
"learning_rate": 3.945056192329212e-05,
|
||
|
|
"loss": 0.2187,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2249683439731598,
|
||
|
|
"step": 690,
|
||
|
|
"valid_targets_mean": 5267.4,
|
||
|
|
"valid_targets_min": 1243
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1819727891156462,
|
||
|
|
"grad_norm": 0.6584654093292903,
|
||
|
|
"learning_rate": 3.9430643090920705e-05,
|
||
|
|
"loss": 0.2304,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23535579442977905,
|
||
|
|
"step": 695,
|
||
|
|
"valid_targets_mean": 4195.1,
|
||
|
|
"valid_targets_min": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1904761904761905,
|
||
|
|
"grad_norm": 0.5354712434221329,
|
||
|
|
"learning_rate": 3.941037480913111e-05,
|
||
|
|
"loss": 0.2262,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22008852660655975,
|
||
|
|
"step": 700,
|
||
|
|
"valid_targets_mean": 5621.8,
|
||
|
|
"valid_targets_min": 541
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1989795918367347,
|
||
|
|
"grad_norm": 0.5255636583449317,
|
||
|
|
"learning_rate": 3.9389757442437214e-05,
|
||
|
|
"loss": 0.2217,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19580186903476715,
|
||
|
|
"step": 705,
|
||
|
|
"valid_targets_mean": 4256.7,
|
||
|
|
"valid_targets_min": 560
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2074829931972788,
|
||
|
|
"grad_norm": 0.6406794533044734,
|
||
|
|
"learning_rate": 3.936879136163102e-05,
|
||
|
|
"loss": 0.2246,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24956966936588287,
|
||
|
|
"step": 710,
|
||
|
|
"valid_targets_mean": 5905.5,
|
||
|
|
"valid_targets_min": 808
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.215986394557823,
|
||
|
|
"grad_norm": 0.49711584769585276,
|
||
|
|
"learning_rate": 3.934747694377594e-05,
|
||
|
|
"loss": 0.218,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2028270959854126,
|
||
|
|
"step": 715,
|
||
|
|
"valid_targets_mean": 5860.9,
|
||
|
|
"valid_targets_min": 2369
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2244897959183674,
|
||
|
|
"grad_norm": 0.5264547948207293,
|
||
|
|
"learning_rate": 3.932581457220004e-05,
|
||
|
|
"loss": 0.2176,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22405561804771423,
|
||
|
|
"step": 720,
|
||
|
|
"valid_targets_mean": 5497.6,
|
||
|
|
"valid_targets_min": 1697
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2329931972789117,
|
||
|
|
"grad_norm": 0.6083163123744523,
|
||
|
|
"learning_rate": 3.9303804636489165e-05,
|
||
|
|
"loss": 0.2148,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21123895049095154,
|
||
|
|
"step": 725,
|
||
|
|
"valid_targets_mean": 5074.1,
|
||
|
|
"valid_targets_min": 475
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2414965986394557,
|
||
|
|
"grad_norm": 0.5466663773950268,
|
||
|
|
"learning_rate": 3.928144753247987e-05,
|
||
|
|
"loss": 0.2163,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2176574170589447,
|
||
|
|
"step": 730,
|
||
|
|
"valid_targets_mean": 4256.7,
|
||
|
|
"valid_targets_min": 264
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.25,
|
||
|
|
"grad_norm": 0.4717649434078562,
|
||
|
|
"learning_rate": 3.925874366225237e-05,
|
||
|
|
"loss": 0.2301,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2158469259738922,
|
||
|
|
"step": 735,
|
||
|
|
"valid_targets_mean": 5555.6,
|
||
|
|
"valid_targets_min": 2361
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2585034013605443,
|
||
|
|
"grad_norm": 0.48874129793361043,
|
||
|
|
"learning_rate": 3.923569343412327e-05,
|
||
|
|
"loss": 0.2062,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2069219946861267,
|
||
|
|
"step": 740,
|
||
|
|
"valid_targets_mean": 4796.9,
|
||
|
|
"valid_targets_min": 326
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2670068027210886,
|
||
|
|
"grad_norm": 1.225304701472566,
|
||
|
|
"learning_rate": 3.921229726263824e-05,
|
||
|
|
"loss": 0.2163,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2326240837574005,
|
||
|
|
"step": 745,
|
||
|
|
"valid_targets_mean": 5964.2,
|
||
|
|
"valid_targets_min": 1239
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2755102040816326,
|
||
|
|
"grad_norm": 0.5031013289502423,
|
||
|
|
"learning_rate": 3.918855556856453e-05,
|
||
|
|
"loss": 0.2245,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2155681848526001,
|
||
|
|
"step": 750,
|
||
|
|
"valid_targets_mean": 5542.2,
|
||
|
|
"valid_targets_min": 585
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.284013605442177,
|
||
|
|
"grad_norm": 0.583184712506713,
|
||
|
|
"learning_rate": 3.916446877888345e-05,
|
||
|
|
"loss": 0.2275,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23659196496009827,
|
||
|
|
"step": 755,
|
||
|
|
"valid_targets_mean": 5372.1,
|
||
|
|
"valid_targets_min": 786
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2925170068027212,
|
||
|
|
"grad_norm": 0.47328660809645895,
|
||
|
|
"learning_rate": 3.914003732678264e-05,
|
||
|
|
"loss": 0.2338,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21961551904678345,
|
||
|
|
"step": 760,
|
||
|
|
"valid_targets_mean": 4733.2,
|
||
|
|
"valid_targets_min": 440
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3010204081632653,
|
||
|
|
"grad_norm": 0.5136294896801605,
|
||
|
|
"learning_rate": 3.911526165164831e-05,
|
||
|
|
"loss": 0.2213,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21073564887046814,
|
||
|
|
"step": 765,
|
||
|
|
"valid_targets_mean": 4688.6,
|
||
|
|
"valid_targets_min": 408
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3095238095238095,
|
||
|
|
"grad_norm": 0.5527736877444719,
|
||
|
|
"learning_rate": 3.909014219905736e-05,
|
||
|
|
"loss": 0.2334,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24369236826896667,
|
||
|
|
"step": 770,
|
||
|
|
"valid_targets_mean": 5702.6,
|
||
|
|
"valid_targets_min": 916
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3180272108843538,
|
||
|
|
"grad_norm": 0.49545440438113,
|
||
|
|
"learning_rate": 3.906467942076929e-05,
|
||
|
|
"loss": 0.2268,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2232738882303238,
|
||
|
|
"step": 775,
|
||
|
|
"valid_targets_mean": 5145.0,
|
||
|
|
"valid_targets_min": 2598
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3265306122448979,
|
||
|
|
"grad_norm": 0.5970072929782432,
|
||
|
|
"learning_rate": 3.903887377471816e-05,
|
||
|
|
"loss": 0.2137,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23750390112400055,
|
||
|
|
"step": 780,
|
||
|
|
"valid_targets_mean": 4200.5,
|
||
|
|
"valid_targets_min": 347
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3350340136054422,
|
||
|
|
"grad_norm": 0.4577691072379611,
|
||
|
|
"learning_rate": 3.901272572500431e-05,
|
||
|
|
"loss": 0.2167,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2235446721315384,
|
||
|
|
"step": 785,
|
||
|
|
"valid_targets_mean": 5234.2,
|
||
|
|
"valid_targets_min": 287
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3435374149659864,
|
||
|
|
"grad_norm": 0.5226812348998608,
|
||
|
|
"learning_rate": 3.8986235741886016e-05,
|
||
|
|
"loss": 0.2161,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20838627219200134,
|
||
|
|
"step": 790,
|
||
|
|
"valid_targets_mean": 4426.8,
|
||
|
|
"valid_targets_min": 626
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3520408163265305,
|
||
|
|
"grad_norm": 0.5368263263004966,
|
||
|
|
"learning_rate": 3.8959404301771035e-05,
|
||
|
|
"loss": 0.2151,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21472008526325226,
|
||
|
|
"step": 795,
|
||
|
|
"valid_targets_mean": 5159.5,
|
||
|
|
"valid_targets_min": 2173
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3605442176870748,
|
||
|
|
"grad_norm": 3.0875210337820707,
|
||
|
|
"learning_rate": 3.893223188720804e-05,
|
||
|
|
"loss": 0.2052,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2349500060081482,
|
||
|
|
"step": 800,
|
||
|
|
"valid_targets_mean": 5517.8,
|
||
|
|
"valid_targets_min": 924
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.369047619047619,
|
||
|
|
"grad_norm": 0.4767596812909599,
|
||
|
|
"learning_rate": 3.890471898687797e-05,
|
||
|
|
"loss": 0.2152,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20095720887184143,
|
||
|
|
"step": 805,
|
||
|
|
"valid_targets_mean": 5096.4,
|
||
|
|
"valid_targets_min": 340
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3775510204081631,
|
||
|
|
"grad_norm": 0.5508686107530064,
|
||
|
|
"learning_rate": 3.887686609558516e-05,
|
||
|
|
"loss": 0.2197,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2549709677696228,
|
||
|
|
"step": 810,
|
||
|
|
"valid_targets_mean": 4902.7,
|
||
|
|
"valid_targets_min": 393
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3860544217687074,
|
||
|
|
"grad_norm": 0.5945732852629755,
|
||
|
|
"learning_rate": 3.8848673714248545e-05,
|
||
|
|
"loss": 0.206,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1957624852657318,
|
||
|
|
"step": 815,
|
||
|
|
"valid_targets_mean": 5539.9,
|
||
|
|
"valid_targets_min": 825
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3945578231292517,
|
||
|
|
"grad_norm": 0.499017099479234,
|
||
|
|
"learning_rate": 3.882014234989258e-05,
|
||
|
|
"loss": 0.2108,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20013388991355896,
|
||
|
|
"step": 820,
|
||
|
|
"valid_targets_mean": 4803.4,
|
||
|
|
"valid_targets_min": 365
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.403061224489796,
|
||
|
|
"grad_norm": 0.5492108236569415,
|
||
|
|
"learning_rate": 3.879127251563814e-05,
|
||
|
|
"loss": 0.2176,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23466426134109497,
|
||
|
|
"step": 825,
|
||
|
|
"valid_targets_mean": 4707.1,
|
||
|
|
"valid_targets_min": 351
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4115646258503403,
|
||
|
|
"grad_norm": 0.4632328855589951,
|
||
|
|
"learning_rate": 3.87620647306933e-05,
|
||
|
|
"loss": 0.2034,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18950562179088593,
|
||
|
|
"step": 830,
|
||
|
|
"valid_targets_mean": 4996.4,
|
||
|
|
"valid_targets_min": 544
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4200680272108843,
|
||
|
|
"grad_norm": 0.5469272265173081,
|
||
|
|
"learning_rate": 3.873251952034398e-05,
|
||
|
|
"loss": 0.2153,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21564960479736328,
|
||
|
|
"step": 835,
|
||
|
|
"valid_targets_mean": 4719.8,
|
||
|
|
"valid_targets_min": 443
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4285714285714286,
|
||
|
|
"grad_norm": 0.42498495314442347,
|
||
|
|
"learning_rate": 3.8702637415944554e-05,
|
||
|
|
"loss": 0.2094,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19484342634677887,
|
||
|
|
"step": 840,
|
||
|
|
"valid_targets_mean": 5729.6,
|
||
|
|
"valid_targets_min": 553
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4370748299319729,
|
||
|
|
"grad_norm": 0.476498159004368,
|
||
|
|
"learning_rate": 3.867241895490818e-05,
|
||
|
|
"loss": 0.1979,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2057555913925171,
|
||
|
|
"step": 845,
|
||
|
|
"valid_targets_mean": 4989.2,
|
||
|
|
"valid_targets_min": 405
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.445578231292517,
|
||
|
|
"grad_norm": 0.5473941467600217,
|
||
|
|
"learning_rate": 3.864186468069727e-05,
|
||
|
|
"loss": 0.221,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22331508994102478,
|
||
|
|
"step": 850,
|
||
|
|
"valid_targets_mean": 5107.7,
|
||
|
|
"valid_targets_min": 3410
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4540816326530612,
|
||
|
|
"grad_norm": 0.48985947700276483,
|
||
|
|
"learning_rate": 3.86109751428136e-05,
|
||
|
|
"loss": 0.2203,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2061440348625183,
|
||
|
|
"step": 855,
|
||
|
|
"valid_targets_mean": 5563.9,
|
||
|
|
"valid_targets_min": 3066
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4625850340136055,
|
||
|
|
"grad_norm": 0.45735993747543985,
|
||
|
|
"learning_rate": 3.8579750896788526e-05,
|
||
|
|
"loss": 0.2228,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22413288056850433,
|
||
|
|
"step": 860,
|
||
|
|
"valid_targets_mean": 5768.6,
|
||
|
|
"valid_targets_min": 2754
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4710884353741496,
|
||
|
|
"grad_norm": 0.48363124415160685,
|
||
|
|
"learning_rate": 3.8548192504172905e-05,
|
||
|
|
"loss": 0.2133,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20641928911209106,
|
||
|
|
"step": 865,
|
||
|
|
"valid_targets_mean": 5050.6,
|
||
|
|
"valid_targets_min": 728
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4795918367346939,
|
||
|
|
"grad_norm": 0.4942217654884803,
|
||
|
|
"learning_rate": 3.851630053252706e-05,
|
||
|
|
"loss": 0.2185,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20422926545143127,
|
||
|
|
"step": 870,
|
||
|
|
"valid_targets_mean": 4567.1,
|
||
|
|
"valid_targets_min": 362
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4880952380952381,
|
||
|
|
"grad_norm": 0.517633245754593,
|
||
|
|
"learning_rate": 3.848407555541054e-05,
|
||
|
|
"loss": 0.2086,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19889432191848755,
|
||
|
|
"step": 875,
|
||
|
|
"valid_targets_mean": 5322.9,
|
||
|
|
"valid_targets_min": 320
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4965986394557822,
|
||
|
|
"grad_norm": 0.4590808466890335,
|
||
|
|
"learning_rate": 3.845151815237183e-05,
|
||
|
|
"loss": 0.2069,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19725415110588074,
|
||
|
|
"step": 880,
|
||
|
|
"valid_targets_mean": 5180.2,
|
||
|
|
"valid_targets_min": 280
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5051020408163265,
|
||
|
|
"grad_norm": 0.5301105087756887,
|
||
|
|
"learning_rate": 3.841862890893787e-05,
|
||
|
|
"loss": 0.2251,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22895437479019165,
|
||
|
|
"step": 885,
|
||
|
|
"valid_targets_mean": 4465.4,
|
||
|
|
"valid_targets_min": 290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5136054421768708,
|
||
|
|
"grad_norm": 0.4803844596014545,
|
||
|
|
"learning_rate": 3.838540841660363e-05,
|
||
|
|
"loss": 0.2016,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19924038648605347,
|
||
|
|
"step": 890,
|
||
|
|
"valid_targets_mean": 5804.7,
|
||
|
|
"valid_targets_min": 915
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5221088435374148,
|
||
|
|
"grad_norm": 0.46388594875311256,
|
||
|
|
"learning_rate": 3.835185727282135e-05,
|
||
|
|
"loss": 0.2007,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19634197652339935,
|
||
|
|
"step": 895,
|
||
|
|
"valid_targets_mean": 4971.6,
|
||
|
|
"valid_targets_min": 319
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5306122448979593,
|
||
|
|
"grad_norm": 0.49922429094048293,
|
||
|
|
"learning_rate": 3.8317976080989884e-05,
|
||
|
|
"loss": 0.214,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19395598769187927,
|
||
|
|
"step": 900,
|
||
|
|
"valid_targets_mean": 5364.1,
|
||
|
|
"valid_targets_min": 351
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5391156462585034,
|
||
|
|
"grad_norm": 0.5232545338127721,
|
||
|
|
"learning_rate": 3.828376545044381e-05,
|
||
|
|
"loss": 0.2239,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24333474040031433,
|
||
|
|
"step": 905,
|
||
|
|
"valid_targets_mean": 4739.3,
|
||
|
|
"valid_targets_min": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5476190476190477,
|
||
|
|
"grad_norm": 0.5072930731695229,
|
||
|
|
"learning_rate": 3.824922599644249e-05,
|
||
|
|
"loss": 0.216,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21110276877880096,
|
||
|
|
"step": 910,
|
||
|
|
"valid_targets_mean": 5527.6,
|
||
|
|
"valid_targets_min": 1938
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.556122448979592,
|
||
|
|
"grad_norm": 0.44843482845791327,
|
||
|
|
"learning_rate": 3.8214358340159e-05,
|
||
|
|
"loss": 0.2208,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24277737736701965,
|
||
|
|
"step": 915,
|
||
|
|
"valid_targets_mean": 6230.3,
|
||
|
|
"valid_targets_min": 2664
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.564625850340136,
|
||
|
|
"grad_norm": 0.5381399540955741,
|
||
|
|
"learning_rate": 3.817916310866892e-05,
|
||
|
|
"loss": 0.2032,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2023087441921234,
|
||
|
|
"step": 920,
|
||
|
|
"valid_targets_mean": 4653.0,
|
||
|
|
"valid_targets_min": 433
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5731292517006803,
|
||
|
|
"grad_norm": 0.6537391559814443,
|
||
|
|
"learning_rate": 3.814364093493912e-05,
|
||
|
|
"loss": 0.2115,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2325202375650406,
|
||
|
|
"step": 925,
|
||
|
|
"valid_targets_mean": 3981.6,
|
||
|
|
"valid_targets_min": 608
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5816326530612246,
|
||
|
|
"grad_norm": 0.44773070706362733,
|
||
|
|
"learning_rate": 3.810779245781633e-05,
|
||
|
|
"loss": 0.2088,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20989757776260376,
|
||
|
|
"step": 930,
|
||
|
|
"valid_targets_mean": 5905.4,
|
||
|
|
"valid_targets_min": 326
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5901360544217686,
|
||
|
|
"grad_norm": 0.4705077477618086,
|
||
|
|
"learning_rate": 3.807161832201569e-05,
|
||
|
|
"loss": 0.2133,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20571091771125793,
|
||
|
|
"step": 935,
|
||
|
|
"valid_targets_mean": 5233.2,
|
||
|
|
"valid_targets_min": 432
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.598639455782313,
|
||
|
|
"grad_norm": 0.5398884047357821,
|
||
|
|
"learning_rate": 3.80351191781091e-05,
|
||
|
|
"loss": 0.1985,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18788626790046692,
|
||
|
|
"step": 940,
|
||
|
|
"valid_targets_mean": 4194.9,
|
||
|
|
"valid_targets_min": 417
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6071428571428572,
|
||
|
|
"grad_norm": 0.5257768990556408,
|
||
|
|
"learning_rate": 3.799829568251358e-05,
|
||
|
|
"loss": 0.2173,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22379785776138306,
|
||
|
|
"step": 945,
|
||
|
|
"valid_targets_mean": 5133.4,
|
||
|
|
"valid_targets_min": 333
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6156462585034013,
|
||
|
|
"grad_norm": 0.433893009068408,
|
||
|
|
"learning_rate": 3.7961148497479414e-05,
|
||
|
|
"loss": 0.2194,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18847689032554626,
|
||
|
|
"step": 950,
|
||
|
|
"valid_targets_mean": 5614.2,
|
||
|
|
"valid_targets_min": 480
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6241496598639455,
|
||
|
|
"grad_norm": 0.5135686330087356,
|
||
|
|
"learning_rate": 3.792367829107828e-05,
|
||
|
|
"loss": 0.2152,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21871328353881836,
|
||
|
|
"step": 955,
|
||
|
|
"valid_targets_mean": 4853.5,
|
||
|
|
"valid_targets_min": 731
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6326530612244898,
|
||
|
|
"grad_norm": 0.539730763074639,
|
||
|
|
"learning_rate": 3.7885885737191214e-05,
|
||
|
|
"loss": 0.2108,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2223399132490158,
|
||
|
|
"step": 960,
|
||
|
|
"valid_targets_mean": 5840.2,
|
||
|
|
"valid_targets_min": 520
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.641156462585034,
|
||
|
|
"grad_norm": 0.42135759448972815,
|
||
|
|
"learning_rate": 3.784777151549648e-05,
|
||
|
|
"loss": 0.2178,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18741963803768158,
|
||
|
|
"step": 965,
|
||
|
|
"valid_targets_mean": 5679.9,
|
||
|
|
"valid_targets_min": 2862
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6496598639455784,
|
||
|
|
"grad_norm": 0.6589967444221846,
|
||
|
|
"learning_rate": 3.780933631145738e-05,
|
||
|
|
"loss": 0.2164,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2285858392715454,
|
||
|
|
"step": 970,
|
||
|
|
"valid_targets_mean": 5193.8,
|
||
|
|
"valid_targets_min": 425
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6581632653061225,
|
||
|
|
"grad_norm": 0.4868453224916295,
|
||
|
|
"learning_rate": 3.777058081630989e-05,
|
||
|
|
"loss": 0.2146,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20116983354091644,
|
||
|
|
"step": 975,
|
||
|
|
"valid_targets_mean": 5680.8,
|
||
|
|
"valid_targets_min": 703
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6666666666666665,
|
||
|
|
"grad_norm": 0.5769718589047669,
|
||
|
|
"learning_rate": 3.773150572705026e-05,
|
||
|
|
"loss": 0.2064,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1806698441505432,
|
||
|
|
"step": 980,
|
||
|
|
"valid_targets_mean": 5046.1,
|
||
|
|
"valid_targets_min": 2693
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.675170068027211,
|
||
|
|
"grad_norm": 0.4765294691850201,
|
||
|
|
"learning_rate": 3.769211174642245e-05,
|
||
|
|
"loss": 0.2075,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22181227803230286,
|
||
|
|
"step": 985,
|
||
|
|
"valid_targets_mean": 4921.1,
|
||
|
|
"valid_targets_min": 349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.683673469387755,
|
||
|
|
"grad_norm": 0.5055633432270628,
|
||
|
|
"learning_rate": 3.765239958290551e-05,
|
||
|
|
"loss": 0.2183,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20720846951007843,
|
||
|
|
"step": 990,
|
||
|
|
"valid_targets_mean": 4407.1,
|
||
|
|
"valid_targets_min": 307
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6921768707482994,
|
||
|
|
"grad_norm": 0.5212127298387436,
|
||
|
|
"learning_rate": 3.761236995070086e-05,
|
||
|
|
"loss": 0.2274,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22654619812965393,
|
||
|
|
"step": 995,
|
||
|
|
"valid_targets_mean": 4863.2,
|
||
|
|
"valid_targets_min": 365
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7006802721088436,
|
||
|
|
"grad_norm": 0.4391797278249632,
|
||
|
|
"learning_rate": 3.757202356971937e-05,
|
||
|
|
"loss": 0.2083,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2114083170890808,
|
||
|
|
"step": 1000,
|
||
|
|
"valid_targets_mean": 5378.8,
|
||
|
|
"valid_targets_min": 757
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7091836734693877,
|
||
|
|
"grad_norm": 0.48454215394808475,
|
||
|
|
"learning_rate": 3.75313611655685e-05,
|
||
|
|
"loss": 0.2102,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22386911511421204,
|
||
|
|
"step": 1005,
|
||
|
|
"valid_targets_mean": 4752.9,
|
||
|
|
"valid_targets_min": 385
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.717687074829932,
|
||
|
|
"grad_norm": 0.41628497294945443,
|
||
|
|
"learning_rate": 3.74903834695392e-05,
|
||
|
|
"loss": 0.2254,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20033439993858337,
|
||
|
|
"step": 1010,
|
||
|
|
"valid_targets_mean": 6052.3,
|
||
|
|
"valid_targets_min": 4185
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7261904761904763,
|
||
|
|
"grad_norm": 0.4503523065828619,
|
||
|
|
"learning_rate": 3.7449091218592765e-05,
|
||
|
|
"loss": 0.2143,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23498773574829102,
|
||
|
|
"step": 1015,
|
||
|
|
"valid_targets_mean": 5659.1,
|
||
|
|
"valid_targets_min": 665
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7346938775510203,
|
||
|
|
"grad_norm": 0.5122324540051815,
|
||
|
|
"learning_rate": 3.740748515534761e-05,
|
||
|
|
"loss": 0.2183,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22834497690200806,
|
||
|
|
"step": 1020,
|
||
|
|
"valid_targets_mean": 5980.3,
|
||
|
|
"valid_targets_min": 646
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7431972789115646,
|
||
|
|
"grad_norm": 0.4532316939484903,
|
||
|
|
"learning_rate": 3.736556602806587e-05,
|
||
|
|
"loss": 0.2104,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20753628015518188,
|
||
|
|
"step": 1025,
|
||
|
|
"valid_targets_mean": 5217.2,
|
||
|
|
"valid_targets_min": 1008
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.751700680272109,
|
||
|
|
"grad_norm": 0.4465999279934645,
|
||
|
|
"learning_rate": 3.732333459063995e-05,
|
||
|
|
"loss": 0.2163,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23834779858589172,
|
||
|
|
"step": 1030,
|
||
|
|
"valid_targets_mean": 6413.3,
|
||
|
|
"valid_targets_min": 3063
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.760204081632653,
|
||
|
|
"grad_norm": 0.5151801706610752,
|
||
|
|
"learning_rate": 3.728079160257903e-05,
|
||
|
|
"loss": 0.2042,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2005973607301712,
|
||
|
|
"step": 1035,
|
||
|
|
"valid_targets_mean": 4667.8,
|
||
|
|
"valid_targets_min": 423
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7687074829931972,
|
||
|
|
"grad_norm": 0.4530448395812097,
|
||
|
|
"learning_rate": 3.723793782899531e-05,
|
||
|
|
"loss": 0.2082,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21784581243991852,
|
||
|
|
"step": 1040,
|
||
|
|
"valid_targets_mean": 5188.7,
|
||
|
|
"valid_targets_min": 286
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7772108843537415,
|
||
|
|
"grad_norm": 0.5081119031546925,
|
||
|
|
"learning_rate": 3.719477404059032e-05,
|
||
|
|
"loss": 0.2164,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19356593489646912,
|
||
|
|
"step": 1045,
|
||
|
|
"valid_targets_mean": 4262.4,
|
||
|
|
"valid_targets_min": 366
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7857142857142856,
|
||
|
|
"grad_norm": 0.4542628404275234,
|
||
|
|
"learning_rate": 3.715130101364103e-05,
|
||
|
|
"loss": 0.2062,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18892955780029297,
|
||
|
|
"step": 1050,
|
||
|
|
"valid_targets_mean": 5145.2,
|
||
|
|
"valid_targets_min": 407
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.79421768707483,
|
||
|
|
"grad_norm": 0.44713443661740765,
|
||
|
|
"learning_rate": 3.710751952998591e-05,
|
||
|
|
"loss": 0.207,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18980108201503754,
|
||
|
|
"step": 1055,
|
||
|
|
"valid_targets_mean": 5316.9,
|
||
|
|
"valid_targets_min": 509
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8027210884353742,
|
||
|
|
"grad_norm": 0.40625775559429467,
|
||
|
|
"learning_rate": 3.7063430377010804e-05,
|
||
|
|
"loss": 0.2078,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20597517490386963,
|
||
|
|
"step": 1060,
|
||
|
|
"valid_targets_mean": 5749.0,
|
||
|
|
"valid_targets_min": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8112244897959182,
|
||
|
|
"grad_norm": 0.42636210781384953,
|
||
|
|
"learning_rate": 3.7019034347634884e-05,
|
||
|
|
"loss": 0.206,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19122816622257233,
|
||
|
|
"step": 1065,
|
||
|
|
"valid_targets_mean": 5505.2,
|
||
|
|
"valid_targets_min": 2184
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8197278911564627,
|
||
|
|
"grad_norm": 0.5457431032106774,
|
||
|
|
"learning_rate": 3.6974332240296285e-05,
|
||
|
|
"loss": 0.2262,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23618677258491516,
|
||
|
|
"step": 1070,
|
||
|
|
"valid_targets_mean": 5652.8,
|
||
|
|
"valid_targets_min": 912
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8282312925170068,
|
||
|
|
"grad_norm": 0.4958314209228894,
|
||
|
|
"learning_rate": 3.692932485893779e-05,
|
||
|
|
"loss": 0.211,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2300088107585907,
|
||
|
|
"step": 1075,
|
||
|
|
"valid_targets_mean": 5673.1,
|
||
|
|
"valid_targets_min": 638
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.836734693877551,
|
||
|
|
"grad_norm": 0.4513717313721071,
|
||
|
|
"learning_rate": 3.6884013012992384e-05,
|
||
|
|
"loss": 0.2212,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22713524103164673,
|
||
|
|
"step": 1080,
|
||
|
|
"valid_targets_mean": 5285.3,
|
||
|
|
"valid_targets_min": 441
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8452380952380953,
|
||
|
|
"grad_norm": 0.4561740142455686,
|
||
|
|
"learning_rate": 3.683839751736866e-05,
|
||
|
|
"loss": 0.2111,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2061767429113388,
|
||
|
|
"step": 1085,
|
||
|
|
"valid_targets_mean": 5541.7,
|
||
|
|
"valid_targets_min": 1086
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8537414965986394,
|
||
|
|
"grad_norm": 0.4401617101373109,
|
||
|
|
"learning_rate": 3.67924791924362e-05,
|
||
|
|
"loss": 0.2008,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19806018471717834,
|
||
|
|
"step": 1090,
|
||
|
|
"valid_targets_mean": 5393.2,
|
||
|
|
"valid_targets_min": 1883
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8622448979591837,
|
||
|
|
"grad_norm": 0.4436705452466841,
|
||
|
|
"learning_rate": 3.6746258864010804e-05,
|
||
|
|
"loss": 0.2036,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19894617795944214,
|
||
|
|
"step": 1095,
|
||
|
|
"valid_targets_mean": 5572.6,
|
||
|
|
"valid_targets_min": 2539
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.870748299319728,
|
||
|
|
"grad_norm": 0.44946000363455185,
|
||
|
|
"learning_rate": 3.669973736333961e-05,
|
||
|
|
"loss": 0.2085,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19905972480773926,
|
||
|
|
"step": 1100,
|
||
|
|
"valid_targets_mean": 6042.1,
|
||
|
|
"valid_targets_min": 2517
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.879251700680272,
|
||
|
|
"grad_norm": 0.49415555699716174,
|
||
|
|
"learning_rate": 3.665291552708622e-05,
|
||
|
|
"loss": 0.2069,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20742449164390564,
|
||
|
|
"step": 1105,
|
||
|
|
"valid_targets_mean": 4889.6,
|
||
|
|
"valid_targets_min": 441
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8877551020408163,
|
||
|
|
"grad_norm": 0.43746352958822043,
|
||
|
|
"learning_rate": 3.660579419731556e-05,
|
||
|
|
"loss": 0.2098,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1905226707458496,
|
||
|
|
"step": 1110,
|
||
|
|
"valid_targets_mean": 5531.6,
|
||
|
|
"valid_targets_min": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8962585034013606,
|
||
|
|
"grad_norm": 0.46517380829800536,
|
||
|
|
"learning_rate": 3.655837422147881e-05,
|
||
|
|
"loss": 0.2006,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19196555018424988,
|
||
|
|
"step": 1115,
|
||
|
|
"valid_targets_mean": 5093.4,
|
||
|
|
"valid_targets_min": 384
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9047619047619047,
|
||
|
|
"grad_norm": 0.5208917501400703,
|
||
|
|
"learning_rate": 3.6510656452398124e-05,
|
||
|
|
"loss": 0.1946,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20660817623138428,
|
||
|
|
"step": 1120,
|
||
|
|
"valid_targets_mean": 4252.8,
|
||
|
|
"valid_targets_min": 359
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.913265306122449,
|
||
|
|
"grad_norm": 0.4744574151189743,
|
||
|
|
"learning_rate": 3.646264174825131e-05,
|
||
|
|
"loss": 0.2133,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20245268940925598,
|
||
|
|
"step": 1125,
|
||
|
|
"valid_targets_mean": 4970.0,
|
||
|
|
"valid_targets_min": 966
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9217687074829932,
|
||
|
|
"grad_norm": 0.5284359759001329,
|
||
|
|
"learning_rate": 3.6414330972556396e-05,
|
||
|
|
"loss": 0.1998,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19377276301383972,
|
||
|
|
"step": 1130,
|
||
|
|
"valid_targets_mean": 4326.1,
|
||
|
|
"valid_targets_min": 369
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9302721088435373,
|
||
|
|
"grad_norm": 0.43120238330248883,
|
||
|
|
"learning_rate": 3.6365724994156074e-05,
|
||
|
|
"loss": 0.209,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1820700466632843,
|
||
|
|
"step": 1135,
|
||
|
|
"valid_targets_mean": 5761.3,
|
||
|
|
"valid_targets_min": 368
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9387755102040818,
|
||
|
|
"grad_norm": 0.5201899574631024,
|
||
|
|
"learning_rate": 3.63168246872021e-05,
|
||
|
|
"loss": 0.2217,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2328445315361023,
|
||
|
|
"step": 1140,
|
||
|
|
"valid_targets_mean": 5319.4,
|
||
|
|
"valid_targets_min": 789
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9472789115646258,
|
||
|
|
"grad_norm": 0.5661162186437034,
|
||
|
|
"learning_rate": 3.626763093113959e-05,
|
||
|
|
"loss": 0.1975,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2142610251903534,
|
||
|
|
"step": 1145,
|
||
|
|
"valid_targets_mean": 3262.6,
|
||
|
|
"valid_targets_min": 485
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.95578231292517,
|
||
|
|
"grad_norm": 0.5171198225709942,
|
||
|
|
"learning_rate": 3.621814461069114e-05,
|
||
|
|
"loss": 0.2029,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18397854268550873,
|
||
|
|
"step": 1150,
|
||
|
|
"valid_targets_mean": 4444.0,
|
||
|
|
"valid_targets_min": 375
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9642857142857144,
|
||
|
|
"grad_norm": 0.47295010025271694,
|
||
|
|
"learning_rate": 3.6168366615841e-05,
|
||
|
|
"loss": 0.2102,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20326940715312958,
|
||
|
|
"step": 1155,
|
||
|
|
"valid_targets_mean": 5261.0,
|
||
|
|
"valid_targets_min": 307
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9727891156462585,
|
||
|
|
"grad_norm": 0.5107326889430264,
|
||
|
|
"learning_rate": 3.6118297841818986e-05,
|
||
|
|
"loss": 0.2112,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22397896647453308,
|
||
|
|
"step": 1160,
|
||
|
|
"valid_targets_mean": 4348.3,
|
||
|
|
"valid_targets_min": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9812925170068028,
|
||
|
|
"grad_norm": 0.47103586665188263,
|
||
|
|
"learning_rate": 3.6067939189084446e-05,
|
||
|
|
"loss": 0.2159,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20325933396816254,
|
||
|
|
"step": 1165,
|
||
|
|
"valid_targets_mean": 5047.1,
|
||
|
|
"valid_targets_min": 1317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.989795918367347,
|
||
|
|
"grad_norm": 0.5083687782339495,
|
||
|
|
"learning_rate": 3.6017291563310005e-05,
|
||
|
|
"loss": 0.1954,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20802119374275208,
|
||
|
|
"step": 1170,
|
||
|
|
"valid_targets_mean": 5499.7,
|
||
|
|
"valid_targets_min": 371
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.998299319727891,
|
||
|
|
"grad_norm": 0.6316741795411249,
|
||
|
|
"learning_rate": 3.5966355875365344e-05,
|
||
|
|
"loss": 0.2054,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20001211762428284,
|
||
|
|
"step": 1175,
|
||
|
|
"valid_targets_mean": 2571.2,
|
||
|
|
"valid_targets_min": 279
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.006802721088435,
|
||
|
|
"grad_norm": 0.48876140678397584,
|
||
|
|
"learning_rate": 3.591513304130076e-05,
|
||
|
|
"loss": 0.1917,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18007999658584595,
|
||
|
|
"step": 1180,
|
||
|
|
"valid_targets_mean": 4412.9,
|
||
|
|
"valid_targets_min": 608
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0153061224489797,
|
||
|
|
"grad_norm": 0.4682267011868987,
|
||
|
|
"learning_rate": 3.586362398233072e-05,
|
||
|
|
"loss": 0.1911,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17856481671333313,
|
||
|
|
"step": 1185,
|
||
|
|
"valid_targets_mean": 5045.8,
|
||
|
|
"valid_targets_min": 1682
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0238095238095237,
|
||
|
|
"grad_norm": 0.5624964830657753,
|
||
|
|
"learning_rate": 3.58118296248173e-05,
|
||
|
|
"loss": 0.1874,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19905634224414825,
|
||
|
|
"step": 1190,
|
||
|
|
"valid_targets_mean": 4665.5,
|
||
|
|
"valid_targets_min": 302
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0323129251700682,
|
||
|
|
"grad_norm": 0.49447628676709326,
|
||
|
|
"learning_rate": 3.57597509002535e-05,
|
||
|
|
"loss": 0.1874,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20415085554122925,
|
||
|
|
"step": 1195,
|
||
|
|
"valid_targets_mean": 5528.1,
|
||
|
|
"valid_targets_min": 433
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0408163265306123,
|
||
|
|
"grad_norm": 0.8190865550278484,
|
||
|
|
"learning_rate": 3.57073887452465e-05,
|
||
|
|
"loss": 0.1932,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18466854095458984,
|
||
|
|
"step": 1200,
|
||
|
|
"valid_targets_mean": 4963.3,
|
||
|
|
"valid_targets_min": 800
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0493197278911564,
|
||
|
|
"grad_norm": 0.542099405027282,
|
||
|
|
"learning_rate": 3.565474410150083e-05,
|
||
|
|
"loss": 0.1842,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19095389544963837,
|
||
|
|
"step": 1205,
|
||
|
|
"valid_targets_mean": 5275.3,
|
||
|
|
"valid_targets_min": 440
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.057823129251701,
|
||
|
|
"grad_norm": 0.4913047816208801,
|
||
|
|
"learning_rate": 3.560181791580144e-05,
|
||
|
|
"loss": 0.192,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19801266491413116,
|
||
|
|
"step": 1210,
|
||
|
|
"valid_targets_mean": 4936.1,
|
||
|
|
"valid_targets_min": 439
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.066326530612245,
|
||
|
|
"grad_norm": 0.4830904919123309,
|
||
|
|
"learning_rate": 3.5548611139996634e-05,
|
||
|
|
"loss": 0.1855,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2041804939508438,
|
||
|
|
"step": 1215,
|
||
|
|
"valid_targets_mean": 5764.8,
|
||
|
|
"valid_targets_min": 2289
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.074829931972789,
|
||
|
|
"grad_norm": 0.5978492052502327,
|
||
|
|
"learning_rate": 3.549512473098097e-05,
|
||
|
|
"loss": 0.1754,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1676190197467804,
|
||
|
|
"step": 1220,
|
||
|
|
"valid_targets_mean": 5737.7,
|
||
|
|
"valid_targets_min": 1427
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0833333333333335,
|
||
|
|
"grad_norm": 0.4627818354833916,
|
||
|
|
"learning_rate": 3.544135965067807e-05,
|
||
|
|
"loss": 0.1884,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18280234932899475,
|
||
|
|
"step": 1225,
|
||
|
|
"valid_targets_mean": 4998.4,
|
||
|
|
"valid_targets_min": 498
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0918367346938775,
|
||
|
|
"grad_norm": 0.4699287294157052,
|
||
|
|
"learning_rate": 3.5387316866023314e-05,
|
||
|
|
"loss": 0.182,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16945812106132507,
|
||
|
|
"step": 1230,
|
||
|
|
"valid_targets_mean": 5049.8,
|
||
|
|
"valid_targets_min": 349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1003401360544216,
|
||
|
|
"grad_norm": 0.7264309703198378,
|
||
|
|
"learning_rate": 3.533299734894642e-05,
|
||
|
|
"loss": 0.1804,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18058422207832336,
|
||
|
|
"step": 1235,
|
||
|
|
"valid_targets_mean": 4218.7,
|
||
|
|
"valid_targets_min": 380
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.108843537414966,
|
||
|
|
"grad_norm": 0.5728141339751329,
|
||
|
|
"learning_rate": 3.527840207635399e-05,
|
||
|
|
"loss": 0.1947,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19773359596729279,
|
||
|
|
"step": 1240,
|
||
|
|
"valid_targets_mean": 3877.1,
|
||
|
|
"valid_targets_min": 346
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.11734693877551,
|
||
|
|
"grad_norm": 0.4534327887318085,
|
||
|
|
"learning_rate": 3.522353203011197e-05,
|
||
|
|
"loss": 0.1819,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15077179670333862,
|
||
|
|
"step": 1245,
|
||
|
|
"valid_targets_mean": 4782.4,
|
||
|
|
"valid_targets_min": 283
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1258503401360542,
|
||
|
|
"grad_norm": 0.4140675531543249,
|
||
|
|
"learning_rate": 3.5168388197027894e-05,
|
||
|
|
"loss": 0.1941,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17872223258018494,
|
||
|
|
"step": 1250,
|
||
|
|
"valid_targets_mean": 6284.6,
|
||
|
|
"valid_targets_min": 2455
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1343537414965987,
|
||
|
|
"grad_norm": 0.5204603490570961,
|
||
|
|
"learning_rate": 3.5112971568833265e-05,
|
||
|
|
"loss": 0.1842,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18467211723327637,
|
||
|
|
"step": 1255,
|
||
|
|
"valid_targets_mean": 4161.4,
|
||
|
|
"valid_targets_min": 442
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.142857142857143,
|
||
|
|
"grad_norm": 0.5297267587243909,
|
||
|
|
"learning_rate": 3.505728314216563e-05,
|
||
|
|
"loss": 0.1863,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19348610937595367,
|
||
|
|
"step": 1260,
|
||
|
|
"valid_targets_mean": 4336.3,
|
||
|
|
"valid_targets_min": 334
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1513605442176873,
|
||
|
|
"grad_norm": 0.44387023048331414,
|
||
|
|
"learning_rate": 3.500132391855067e-05,
|
||
|
|
"loss": 0.1948,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19352880120277405,
|
||
|
|
"step": 1265,
|
||
|
|
"valid_targets_mean": 5925.4,
|
||
|
|
"valid_targets_min": 685
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1598639455782314,
|
||
|
|
"grad_norm": 0.6640673868234788,
|
||
|
|
"learning_rate": 3.494509490438422e-05,
|
||
|
|
"loss": 0.1918,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19546517729759216,
|
||
|
|
"step": 1270,
|
||
|
|
"valid_targets_mean": 4086.4,
|
||
|
|
"valid_targets_min": 358
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1683673469387754,
|
||
|
|
"grad_norm": 0.46496036082989145,
|
||
|
|
"learning_rate": 3.4888597110914155e-05,
|
||
|
|
"loss": 0.1814,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1798856556415558,
|
||
|
|
"step": 1275,
|
||
|
|
"valid_targets_mean": 5040.2,
|
||
|
|
"valid_targets_min": 733
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.17687074829932,
|
||
|
|
"grad_norm": 0.4328124919631194,
|
||
|
|
"learning_rate": 3.483183155422216e-05,
|
||
|
|
"loss": 0.1794,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.187459796667099,
|
||
|
|
"step": 1280,
|
||
|
|
"valid_targets_mean": 5586.8,
|
||
|
|
"valid_targets_min": 2173
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.185374149659864,
|
||
|
|
"grad_norm": 0.5185502281400557,
|
||
|
|
"learning_rate": 3.477479925520555e-05,
|
||
|
|
"loss": 0.1778,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18551820516586304,
|
||
|
|
"step": 1285,
|
||
|
|
"valid_targets_mean": 4702.4,
|
||
|
|
"valid_targets_min": 560
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.193877551020408,
|
||
|
|
"grad_norm": 0.5033131167250684,
|
||
|
|
"learning_rate": 3.471750123955878e-05,
|
||
|
|
"loss": 0.1879,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.193929061293602,
|
||
|
|
"step": 1290,
|
||
|
|
"valid_targets_mean": 4453.0,
|
||
|
|
"valid_targets_min": 336
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2023809523809526,
|
||
|
|
"grad_norm": 0.5217630751152134,
|
||
|
|
"learning_rate": 3.465993853775516e-05,
|
||
|
|
"loss": 0.1944,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18472431600093842,
|
||
|
|
"step": 1295,
|
||
|
|
"valid_targets_mean": 5340.3,
|
||
|
|
"valid_targets_min": 279
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2108843537414966,
|
||
|
|
"grad_norm": 0.44926286818850636,
|
||
|
|
"learning_rate": 3.460211218502815e-05,
|
||
|
|
"loss": 0.1916,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15788491070270538,
|
||
|
|
"step": 1300,
|
||
|
|
"valid_targets_mean": 5304.3,
|
||
|
|
"valid_targets_min": 1204
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2193877551020407,
|
||
|
|
"grad_norm": 0.5331367023371473,
|
||
|
|
"learning_rate": 3.45440232213529e-05,
|
||
|
|
"loss": 0.1882,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2080489695072174,
|
||
|
|
"step": 1305,
|
||
|
|
"valid_targets_mean": 4884.9,
|
||
|
|
"valid_targets_min": 467
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.227891156462585,
|
||
|
|
"grad_norm": 0.47156629390071225,
|
||
|
|
"learning_rate": 3.4485672691427416e-05,
|
||
|
|
"loss": 0.1929,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1818922758102417,
|
||
|
|
"step": 1310,
|
||
|
|
"valid_targets_mean": 5560.1,
|
||
|
|
"valid_targets_min": 307
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2363945578231292,
|
||
|
|
"grad_norm": 0.5262171034079817,
|
||
|
|
"learning_rate": 3.4427061644653873e-05,
|
||
|
|
"loss": 0.2048,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19919820129871368,
|
||
|
|
"step": 1315,
|
||
|
|
"valid_targets_mean": 4189.7,
|
||
|
|
"valid_targets_min": 404
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2448979591836733,
|
||
|
|
"grad_norm": 0.40102180477864185,
|
||
|
|
"learning_rate": 3.436819113511967e-05,
|
||
|
|
"loss": 0.1778,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15597829222679138,
|
||
|
|
"step": 1320,
|
||
|
|
"valid_targets_mean": 5941.2,
|
||
|
|
"valid_targets_min": 3078
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.253401360544218,
|
||
|
|
"grad_norm": 0.4713272661391161,
|
||
|
|
"learning_rate": 3.43090622215785e-05,
|
||
|
|
"loss": 0.19,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1802557408809662,
|
||
|
|
"step": 1325,
|
||
|
|
"valid_targets_mean": 4518.2,
|
||
|
|
"valid_targets_min": 469
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.261904761904762,
|
||
|
|
"grad_norm": 0.5900943367804872,
|
||
|
|
"learning_rate": 3.4249675967431325e-05,
|
||
|
|
"loss": 0.1746,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19313806295394897,
|
||
|
|
"step": 1330,
|
||
|
|
"valid_targets_mean": 4941.3,
|
||
|
|
"valid_targets_min": 825
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.270408163265306,
|
||
|
|
"grad_norm": 0.460032628569175,
|
||
|
|
"learning_rate": 3.419003344070723e-05,
|
||
|
|
"loss": 0.185,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18703874945640564,
|
||
|
|
"step": 1335,
|
||
|
|
"valid_targets_mean": 4961.1,
|
||
|
|
"valid_targets_min": 991
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2789115646258504,
|
||
|
|
"grad_norm": 0.44707823142358133,
|
||
|
|
"learning_rate": 3.413013571404423e-05,
|
||
|
|
"loss": 0.1787,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17614120244979858,
|
||
|
|
"step": 1340,
|
||
|
|
"valid_targets_mean": 5576.6,
|
||
|
|
"valid_targets_min": 331
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2874149659863945,
|
||
|
|
"grad_norm": 0.4868809540696579,
|
||
|
|
"learning_rate": 3.406998386466995e-05,
|
||
|
|
"loss": 0.1896,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.201404869556427,
|
||
|
|
"step": 1345,
|
||
|
|
"valid_targets_mean": 5212.1,
|
||
|
|
"valid_targets_min": 1210
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.295918367346939,
|
||
|
|
"grad_norm": 0.5077746609983623,
|
||
|
|
"learning_rate": 3.400957897438229e-05,
|
||
|
|
"loss": 0.181,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18614467978477478,
|
||
|
|
"step": 1350,
|
||
|
|
"valid_targets_mean": 4716.4,
|
||
|
|
"valid_targets_min": 349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.304421768707483,
|
||
|
|
"grad_norm": 1.045064276720603,
|
||
|
|
"learning_rate": 3.3948922129529935e-05,
|
||
|
|
"loss": 0.1984,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21427641808986664,
|
||
|
|
"step": 1355,
|
||
|
|
"valid_targets_mean": 5340.4,
|
||
|
|
"valid_targets_min": 825
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.312925170068027,
|
||
|
|
"grad_norm": 0.48150905319674575,
|
||
|
|
"learning_rate": 3.388801442099286e-05,
|
||
|
|
"loss": 0.1932,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1955031454563141,
|
||
|
|
"step": 1360,
|
||
|
|
"valid_targets_mean": 5783.1,
|
||
|
|
"valid_targets_min": 773
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3214285714285716,
|
||
|
|
"grad_norm": 0.4999416938719426,
|
||
|
|
"learning_rate": 3.3826856944162655e-05,
|
||
|
|
"loss": 0.2012,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2055264711380005,
|
||
|
|
"step": 1365,
|
||
|
|
"valid_targets_mean": 4798.9,
|
||
|
|
"valid_targets_min": 445
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3299319727891157,
|
||
|
|
"grad_norm": 0.610139372863544,
|
||
|
|
"learning_rate": 3.376545079892288e-05,
|
||
|
|
"loss": 0.1766,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17390289902687073,
|
||
|
|
"step": 1370,
|
||
|
|
"valid_targets_mean": 6032.4,
|
||
|
|
"valid_targets_min": 4015
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3384353741496597,
|
||
|
|
"grad_norm": 0.45245488892599056,
|
||
|
|
"learning_rate": 3.3703797089629245e-05,
|
||
|
|
"loss": 0.1833,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19045419991016388,
|
||
|
|
"step": 1375,
|
||
|
|
"valid_targets_mean": 5607.8,
|
||
|
|
"valid_targets_min": 1146
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3469387755102042,
|
||
|
|
"grad_norm": 0.4784557437390841,
|
||
|
|
"learning_rate": 3.364189692508976e-05,
|
||
|
|
"loss": 0.1878,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22167515754699707,
|
||
|
|
"step": 1380,
|
||
|
|
"valid_targets_mean": 5258.6,
|
||
|
|
"valid_targets_min": 471
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3554421768707483,
|
||
|
|
"grad_norm": 0.5149436257886181,
|
||
|
|
"learning_rate": 3.357975141854482e-05,
|
||
|
|
"loss": 0.1844,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21249035000801086,
|
||
|
|
"step": 1385,
|
||
|
|
"valid_targets_mean": 4707.0,
|
||
|
|
"valid_targets_min": 356
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3639455782312924,
|
||
|
|
"grad_norm": 0.5850620816636127,
|
||
|
|
"learning_rate": 3.3517361687647136e-05,
|
||
|
|
"loss": 0.1995,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17860059440135956,
|
||
|
|
"step": 1390,
|
||
|
|
"valid_targets_mean": 5836.6,
|
||
|
|
"valid_targets_min": 1555
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.372448979591837,
|
||
|
|
"grad_norm": 0.48422996605087937,
|
||
|
|
"learning_rate": 3.3454728854441667e-05,
|
||
|
|
"loss": 0.1964,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1890917718410492,
|
||
|
|
"step": 1395,
|
||
|
|
"valid_targets_mean": 4811.6,
|
||
|
|
"valid_targets_min": 684
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.380952380952381,
|
||
|
|
"grad_norm": 0.5020954557267779,
|
||
|
|
"learning_rate": 3.3391854045345437e-05,
|
||
|
|
"loss": 0.2125,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1835050731897354,
|
||
|
|
"step": 1400,
|
||
|
|
"valid_targets_mean": 5382.7,
|
||
|
|
"valid_targets_min": 414
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.389455782312925,
|
||
|
|
"grad_norm": 0.44472072289037046,
|
||
|
|
"learning_rate": 3.332873839112727e-05,
|
||
|
|
"loss": 0.1815,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1824030578136444,
|
||
|
|
"step": 1405,
|
||
|
|
"valid_targets_mean": 5961.9,
|
||
|
|
"valid_targets_min": 1883
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3979591836734695,
|
||
|
|
"grad_norm": 0.47212145560147684,
|
||
|
|
"learning_rate": 3.3265383026887464e-05,
|
||
|
|
"loss": 0.1749,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16177970170974731,
|
||
|
|
"step": 1410,
|
||
|
|
"valid_targets_mean": 4572.0,
|
||
|
|
"valid_targets_min": 318
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4064625850340136,
|
||
|
|
"grad_norm": 0.46214691638217914,
|
||
|
|
"learning_rate": 3.320178909203735e-05,
|
||
|
|
"loss": 0.1919,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1828824281692505,
|
||
|
|
"step": 1415,
|
||
|
|
"valid_targets_mean": 5612.4,
|
||
|
|
"valid_targets_min": 500
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4149659863945576,
|
||
|
|
"grad_norm": 0.6765825065424675,
|
||
|
|
"learning_rate": 3.313795773027885e-05,
|
||
|
|
"loss": 0.1998,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19763198494911194,
|
||
|
|
"step": 1420,
|
||
|
|
"valid_targets_mean": 4614.1,
|
||
|
|
"valid_targets_min": 260
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.423469387755102,
|
||
|
|
"grad_norm": 0.5102976599346384,
|
||
|
|
"learning_rate": 3.307389008958385e-05,
|
||
|
|
"loss": 0.2007,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19278958439826965,
|
||
|
|
"step": 1425,
|
||
|
|
"valid_targets_mean": 5896.1,
|
||
|
|
"valid_targets_min": 347
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.431972789115646,
|
||
|
|
"grad_norm": 0.503581105834159,
|
||
|
|
"learning_rate": 3.30095873221736e-05,
|
||
|
|
"loss": 0.1966,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18435746431350708,
|
||
|
|
"step": 1430,
|
||
|
|
"valid_targets_mean": 5119.2,
|
||
|
|
"valid_targets_min": 366
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4404761904761907,
|
||
|
|
"grad_norm": 0.4474068580869408,
|
||
|
|
"learning_rate": 3.294505058449797e-05,
|
||
|
|
"loss": 0.1788,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18685245513916016,
|
||
|
|
"step": 1435,
|
||
|
|
"valid_targets_mean": 5780.7,
|
||
|
|
"valid_targets_min": 681
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4489795918367347,
|
||
|
|
"grad_norm": 0.41622735583910236,
|
||
|
|
"learning_rate": 3.288028103721465e-05,
|
||
|
|
"loss": 0.1814,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15953834354877472,
|
||
|
|
"step": 1440,
|
||
|
|
"valid_targets_mean": 5816.8,
|
||
|
|
"valid_targets_min": 2871
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.457482993197279,
|
||
|
|
"grad_norm": 0.4465626719984511,
|
||
|
|
"learning_rate": 3.2815279845168296e-05,
|
||
|
|
"loss": 0.1984,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20260348916053772,
|
||
|
|
"step": 1445,
|
||
|
|
"valid_targets_mean": 5325.4,
|
||
|
|
"valid_targets_min": 452
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4659863945578233,
|
||
|
|
"grad_norm": 0.41633690487509545,
|
||
|
|
"learning_rate": 3.2750048177369554e-05,
|
||
|
|
"loss": 0.1787,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18053773045539856,
|
||
|
|
"step": 1450,
|
||
|
|
"valid_targets_mean": 5929.5,
|
||
|
|
"valid_targets_min": 652
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4744897959183674,
|
||
|
|
"grad_norm": 0.42411107206212223,
|
||
|
|
"learning_rate": 3.268458720697406e-05,
|
||
|
|
"loss": 0.1764,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18054378032684326,
|
||
|
|
"step": 1455,
|
||
|
|
"valid_targets_mean": 6285.4,
|
||
|
|
"valid_targets_min": 3092
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4829931972789114,
|
||
|
|
"grad_norm": 0.4801434025546991,
|
||
|
|
"learning_rate": 3.2618898111261316e-05,
|
||
|
|
"loss": 0.1835,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18590672314167023,
|
||
|
|
"step": 1460,
|
||
|
|
"valid_targets_mean": 4688.4,
|
||
|
|
"valid_targets_min": 377
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.491496598639456,
|
||
|
|
"grad_norm": 0.4299683061243853,
|
||
|
|
"learning_rate": 3.2552982071613555e-05,
|
||
|
|
"loss": 0.1848,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1850741058588028,
|
||
|
|
"step": 1465,
|
||
|
|
"valid_targets_mean": 6172.9,
|
||
|
|
"valid_targets_min": 486
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5,
|
||
|
|
"grad_norm": 0.5184342039007735,
|
||
|
|
"learning_rate": 3.248684027349445e-05,
|
||
|
|
"loss": 0.1877,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17551356554031372,
|
||
|
|
"step": 1470,
|
||
|
|
"valid_targets_mean": 5149.6,
|
||
|
|
"valid_targets_min": 1099
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.508503401360544,
|
||
|
|
"grad_norm": 0.4777002183530347,
|
||
|
|
"learning_rate": 3.2420473906427824e-05,
|
||
|
|
"loss": 0.1807,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1735362857580185,
|
||
|
|
"step": 1475,
|
||
|
|
"valid_targets_mean": 5659.2,
|
||
|
|
"valid_targets_min": 912
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5170068027210886,
|
||
|
|
"grad_norm": 0.45495063939680586,
|
||
|
|
"learning_rate": 3.235388416397626e-05,
|
||
|
|
"loss": 0.1952,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20709705352783203,
|
||
|
|
"step": 1480,
|
||
|
|
"valid_targets_mean": 5489.8,
|
||
|
|
"valid_targets_min": 2713
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5255102040816326,
|
||
|
|
"grad_norm": 0.5485887687300974,
|
||
|
|
"learning_rate": 3.228707224371961e-05,
|
||
|
|
"loss": 0.1821,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18838176131248474,
|
||
|
|
"step": 1485,
|
||
|
|
"valid_targets_mean": 5773.9,
|
||
|
|
"valid_targets_min": 371
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.534013605442177,
|
||
|
|
"grad_norm": 0.4695867851253173,
|
||
|
|
"learning_rate": 3.222003934723347e-05,
|
||
|
|
"loss": 0.1892,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1890532225370407,
|
||
|
|
"step": 1490,
|
||
|
|
"valid_targets_mean": 6022.9,
|
||
|
|
"valid_targets_min": 321
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.542517006802721,
|
||
|
|
"grad_norm": 0.47136584934841597,
|
||
|
|
"learning_rate": 3.215278668006757e-05,
|
||
|
|
"loss": 0.1884,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17988421022891998,
|
||
|
|
"step": 1495,
|
||
|
|
"valid_targets_mean": 4950.6,
|
||
|
|
"valid_targets_min": 786
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5510204081632653,
|
||
|
|
"grad_norm": 0.4396466153637515,
|
||
|
|
"learning_rate": 3.208531545172413e-05,
|
||
|
|
"loss": 0.1909,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19453471899032593,
|
||
|
|
"step": 1500,
|
||
|
|
"valid_targets_mean": 5923.2,
|
||
|
|
"valid_targets_min": 3744
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5595238095238093,
|
||
|
|
"grad_norm": 0.45393722871708864,
|
||
|
|
"learning_rate": 3.2017626875636024e-05,
|
||
|
|
"loss": 0.1703,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17787441611289978,
|
||
|
|
"step": 1505,
|
||
|
|
"valid_targets_mean": 5203.3,
|
||
|
|
"valid_targets_min": 2684
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.568027210884354,
|
||
|
|
"grad_norm": 0.43586842764838324,
|
||
|
|
"learning_rate": 3.194972216914502e-05,
|
||
|
|
"loss": 0.1695,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.166270911693573,
|
||
|
|
"step": 1510,
|
||
|
|
"valid_targets_mean": 5435.4,
|
||
|
|
"valid_targets_min": 2379
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.576530612244898,
|
||
|
|
"grad_norm": 0.5080413565752104,
|
||
|
|
"learning_rate": 3.18816025534799e-05,
|
||
|
|
"loss": 0.1913,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18987306952476501,
|
||
|
|
"step": 1515,
|
||
|
|
"valid_targets_mean": 4775.7,
|
||
|
|
"valid_targets_min": 319
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5850340136054424,
|
||
|
|
"grad_norm": 0.47909989222563665,
|
||
|
|
"learning_rate": 3.181326925373443e-05,
|
||
|
|
"loss": 0.177,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17786143720149994,
|
||
|
|
"step": 1520,
|
||
|
|
"valid_targets_mean": 5268.4,
|
||
|
|
"valid_targets_min": 384
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5935374149659864,
|
||
|
|
"grad_norm": 0.47307616162349375,
|
||
|
|
"learning_rate": 3.174472349884539e-05,
|
||
|
|
"loss": 0.1887,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19906634092330933,
|
||
|
|
"step": 1525,
|
||
|
|
"valid_targets_mean": 5265.6,
|
||
|
|
"valid_targets_min": 819
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6020408163265305,
|
||
|
|
"grad_norm": 0.5241685453140259,
|
||
|
|
"learning_rate": 3.1675966521570444e-05,
|
||
|
|
"loss": 0.1969,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.186308354139328,
|
||
|
|
"step": 1530,
|
||
|
|
"valid_targets_mean": 4326.4,
|
||
|
|
"valid_targets_min": 553
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6105442176870746,
|
||
|
|
"grad_norm": 0.4824124789042875,
|
||
|
|
"learning_rate": 3.160699955846597e-05,
|
||
|
|
"loss": 0.208,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1919880509376526,
|
||
|
|
"step": 1535,
|
||
|
|
"valid_targets_mean": 4851.0,
|
||
|
|
"valid_targets_min": 335
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.619047619047619,
|
||
|
|
"grad_norm": 0.47872735083966544,
|
||
|
|
"learning_rate": 3.1537823849864834e-05,
|
||
|
|
"loss": 0.1925,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21181975305080414,
|
||
|
|
"step": 1540,
|
||
|
|
"valid_targets_mean": 4542.9,
|
||
|
|
"valid_targets_min": 806
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.627551020408163,
|
||
|
|
"grad_norm": 0.44425564563286324,
|
||
|
|
"learning_rate": 3.1468440639854056e-05,
|
||
|
|
"loss": 0.1821,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18633927404880524,
|
||
|
|
"step": 1545,
|
||
|
|
"valid_targets_mean": 5434.5,
|
||
|
|
"valid_targets_min": 3098
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6360544217687076,
|
||
|
|
"grad_norm": 0.4330232416286868,
|
||
|
|
"learning_rate": 3.139885117625249e-05,
|
||
|
|
"loss": 0.1758,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18250718712806702,
|
||
|
|
"step": 1550,
|
||
|
|
"valid_targets_mean": 5508.0,
|
||
|
|
"valid_targets_min": 404
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6445578231292517,
|
||
|
|
"grad_norm": 0.46662688241184896,
|
||
|
|
"learning_rate": 3.132905671058831e-05,
|
||
|
|
"loss": 0.1705,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17651695013046265,
|
||
|
|
"step": 1555,
|
||
|
|
"valid_targets_mean": 4875.1,
|
||
|
|
"valid_targets_min": 368
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6530612244897958,
|
||
|
|
"grad_norm": 0.4174835476472129,
|
||
|
|
"learning_rate": 3.1259058498076554e-05,
|
||
|
|
"loss": 0.1877,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18197211623191833,
|
||
|
|
"step": 1560,
|
||
|
|
"valid_targets_mean": 5360.8,
|
||
|
|
"valid_targets_min": 340
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6615646258503403,
|
||
|
|
"grad_norm": 0.4450510722392302,
|
||
|
|
"learning_rate": 3.1188857797596545e-05,
|
||
|
|
"loss": 0.1819,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17218464612960815,
|
||
|
|
"step": 1565,
|
||
|
|
"valid_targets_mean": 5491.8,
|
||
|
|
"valid_targets_min": 482
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6700680272108843,
|
||
|
|
"grad_norm": 0.4484479475490854,
|
||
|
|
"learning_rate": 3.111845587166922e-05,
|
||
|
|
"loss": 0.1842,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17864450812339783,
|
||
|
|
"step": 1570,
|
||
|
|
"valid_targets_mean": 5240.9,
|
||
|
|
"valid_targets_min": 552
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.678571428571429,
|
||
|
|
"grad_norm": 0.47587737802717933,
|
||
|
|
"learning_rate": 3.104785398643447e-05,
|
||
|
|
"loss": 0.1926,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20044058561325073,
|
||
|
|
"step": 1575,
|
||
|
|
"valid_targets_mean": 5136.2,
|
||
|
|
"valid_targets_min": 872
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.687074829931973,
|
||
|
|
"grad_norm": 0.5018235508884444,
|
||
|
|
"learning_rate": 3.0977053411628315e-05,
|
||
|
|
"loss": 0.1912,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19904190301895142,
|
||
|
|
"step": 1580,
|
||
|
|
"valid_targets_mean": 4798.7,
|
||
|
|
"valid_targets_min": 323
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.695578231292517,
|
||
|
|
"grad_norm": 0.4459205617333467,
|
||
|
|
"learning_rate": 3.090605542056013e-05,
|
||
|
|
"loss": 0.2067,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20209312438964844,
|
||
|
|
"step": 1585,
|
||
|
|
"valid_targets_mean": 5350.9,
|
||
|
|
"valid_targets_min": 753
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.704081632653061,
|
||
|
|
"grad_norm": 0.5105473118050233,
|
||
|
|
"learning_rate": 3.083486129008967e-05,
|
||
|
|
"loss": 0.186,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1771620810031891,
|
||
|
|
"step": 1590,
|
||
|
|
"valid_targets_mean": 4776.2,
|
||
|
|
"valid_targets_min": 278
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7125850340136055,
|
||
|
|
"grad_norm": 0.5131779127702178,
|
||
|
|
"learning_rate": 3.07634723006042e-05,
|
||
|
|
"loss": 0.1837,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18170857429504395,
|
||
|
|
"step": 1595,
|
||
|
|
"valid_targets_mean": 5518.4,
|
||
|
|
"valid_targets_min": 2342
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7210884353741496,
|
||
|
|
"grad_norm": 0.4151455966543216,
|
||
|
|
"learning_rate": 3.069188973599538e-05,
|
||
|
|
"loss": 0.1895,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17690356075763702,
|
||
|
|
"step": 1600,
|
||
|
|
"valid_targets_mean": 6046.2,
|
||
|
|
"valid_targets_min": 3324
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.729591836734694,
|
||
|
|
"grad_norm": 0.4410817525461484,
|
||
|
|
"learning_rate": 3.062011488363623e-05,
|
||
|
|
"loss": 0.177,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15733939409255981,
|
||
|
|
"step": 1605,
|
||
|
|
"valid_targets_mean": 5203.9,
|
||
|
|
"valid_targets_min": 307
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.738095238095238,
|
||
|
|
"grad_norm": 0.47581985380691993,
|
||
|
|
"learning_rate": 3.054814903435794e-05,
|
||
|
|
"loss": 0.2033,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18917620182037354,
|
||
|
|
"step": 1610,
|
||
|
|
"valid_targets_mean": 5622.0,
|
||
|
|
"valid_targets_min": 677
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.746598639455782,
|
||
|
|
"grad_norm": 0.4661124543367342,
|
||
|
|
"learning_rate": 3.0475993482426695e-05,
|
||
|
|
"loss": 0.1838,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19890253245830536,
|
||
|
|
"step": 1615,
|
||
|
|
"valid_targets_mean": 5409.4,
|
||
|
|
"valid_targets_min": 491
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7551020408163263,
|
||
|
|
"grad_norm": 0.4485450833066948,
|
||
|
|
"learning_rate": 3.0403649525520374e-05,
|
||
|
|
"loss": 0.1894,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1755256950855255,
|
||
|
|
"step": 1620,
|
||
|
|
"valid_targets_mean": 5249.2,
|
||
|
|
"valid_targets_min": 435
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7636054421768708,
|
||
|
|
"grad_norm": 0.4980136434775561,
|
||
|
|
"learning_rate": 3.033111846470521e-05,
|
||
|
|
"loss": 0.2049,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19441989064216614,
|
||
|
|
"step": 1625,
|
||
|
|
"valid_targets_mean": 4450.1,
|
||
|
|
"valid_targets_min": 277
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.772108843537415,
|
||
|
|
"grad_norm": 0.46687820297214816,
|
||
|
|
"learning_rate": 3.02584016044124e-05,
|
||
|
|
"loss": 0.1736,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1853276491165161,
|
||
|
|
"step": 1630,
|
||
|
|
"valid_targets_mean": 5128.4,
|
||
|
|
"valid_targets_min": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7806122448979593,
|
||
|
|
"grad_norm": 0.408499685211329,
|
||
|
|
"learning_rate": 3.0185500252414633e-05,
|
||
|
|
"loss": 0.1885,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19249539077281952,
|
||
|
|
"step": 1635,
|
||
|
|
"valid_targets_mean": 6238.9,
|
||
|
|
"valid_targets_min": 394
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7891156462585034,
|
||
|
|
"grad_norm": 0.4327158558319995,
|
||
|
|
"learning_rate": 3.01124157198026e-05,
|
||
|
|
"loss": 0.1784,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1649228185415268,
|
||
|
|
"step": 1640,
|
||
|
|
"valid_targets_mean": 5835.8,
|
||
|
|
"valid_targets_min": 3349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7976190476190474,
|
||
|
|
"grad_norm": 0.4645255528815955,
|
||
|
|
"learning_rate": 3.003914932096137e-05,
|
||
|
|
"loss": 0.1872,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17619681358337402,
|
||
|
|
"step": 1645,
|
||
|
|
"valid_targets_mean": 6008.9,
|
||
|
|
"valid_targets_min": 498
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.806122448979592,
|
||
|
|
"grad_norm": 0.525027023066297,
|
||
|
|
"learning_rate": 2.99657023735468e-05,
|
||
|
|
"loss": 0.189,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17193004488945007,
|
||
|
|
"step": 1650,
|
||
|
|
"valid_targets_mean": 5420.2,
|
||
|
|
"valid_targets_min": 483
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.814625850340136,
|
||
|
|
"grad_norm": 0.4217497888835726,
|
||
|
|
"learning_rate": 2.9892076198461794e-05,
|
||
|
|
"loss": 0.1919,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17490792274475098,
|
||
|
|
"step": 1655,
|
||
|
|
"valid_targets_mean": 5399.8,
|
||
|
|
"valid_targets_min": 577
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8231292517006805,
|
||
|
|
"grad_norm": 0.5434713380140986,
|
||
|
|
"learning_rate": 2.9818272119832576e-05,
|
||
|
|
"loss": 0.19,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2084447145462036,
|
||
|
|
"step": 1660,
|
||
|
|
"valid_targets_mean": 4128.0,
|
||
|
|
"valid_targets_min": 347
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8316326530612246,
|
||
|
|
"grad_norm": 0.46008996016815673,
|
||
|
|
"learning_rate": 2.974429146498487e-05,
|
||
|
|
"loss": 0.1955,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17945440113544464,
|
||
|
|
"step": 1665,
|
||
|
|
"valid_targets_mean": 5591.0,
|
||
|
|
"valid_targets_min": 2685
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8401360544217686,
|
||
|
|
"grad_norm": 0.46394397208564736,
|
||
|
|
"learning_rate": 2.9670135564420016e-05,
|
||
|
|
"loss": 0.1814,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18872299790382385,
|
||
|
|
"step": 1670,
|
||
|
|
"valid_targets_mean": 5334.8,
|
||
|
|
"valid_targets_min": 326
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8486394557823127,
|
||
|
|
"grad_norm": 0.43187937504915086,
|
||
|
|
"learning_rate": 2.9595805751791068e-05,
|
||
|
|
"loss": 0.1846,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17939303815364838,
|
||
|
|
"step": 1675,
|
||
|
|
"valid_targets_mean": 5890.7,
|
||
|
|
"valid_targets_min": 2163
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.857142857142857,
|
||
|
|
"grad_norm": 0.4602559934490223,
|
||
|
|
"learning_rate": 2.9521303363878774e-05,
|
||
|
|
"loss": 0.181,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18438182771205902,
|
||
|
|
"step": 1680,
|
||
|
|
"valid_targets_mean": 5229.4,
|
||
|
|
"valid_targets_min": 358
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8656462585034013,
|
||
|
|
"grad_norm": 0.42417646096790995,
|
||
|
|
"learning_rate": 2.9446629740567566e-05,
|
||
|
|
"loss": 0.1751,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17440176010131836,
|
||
|
|
"step": 1685,
|
||
|
|
"valid_targets_mean": 5195.2,
|
||
|
|
"valid_targets_min": 995
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8741496598639458,
|
||
|
|
"grad_norm": 0.45505682272764436,
|
||
|
|
"learning_rate": 2.9371786224821447e-05,
|
||
|
|
"loss": 0.179,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1680985689163208,
|
||
|
|
"step": 1690,
|
||
|
|
"valid_targets_mean": 5355.5,
|
||
|
|
"valid_targets_min": 297
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.88265306122449,
|
||
|
|
"grad_norm": 0.4189097375828769,
|
||
|
|
"learning_rate": 2.9296774162659836e-05,
|
||
|
|
"loss": 0.1817,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18291297554969788,
|
||
|
|
"step": 1695,
|
||
|
|
"valid_targets_mean": 5323.9,
|
||
|
|
"valid_targets_min": 333
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.891156462585034,
|
||
|
|
"grad_norm": 0.4539774003906082,
|
||
|
|
"learning_rate": 2.9221594903133385e-05,
|
||
|
|
"loss": 0.1915,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.201238214969635,
|
||
|
|
"step": 1700,
|
||
|
|
"valid_targets_mean": 5677.1,
|
||
|
|
"valid_targets_min": 1133
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8996598639455784,
|
||
|
|
"grad_norm": 0.5017624906979566,
|
||
|
|
"learning_rate": 2.9146249798299672e-05,
|
||
|
|
"loss": 0.1887,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1967279613018036,
|
||
|
|
"step": 1705,
|
||
|
|
"valid_targets_mean": 5102.3,
|
||
|
|
"valid_targets_min": 379
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9081632653061225,
|
||
|
|
"grad_norm": 0.4737316469014641,
|
||
|
|
"learning_rate": 2.9070740203198927e-05,
|
||
|
|
"loss": 0.1851,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1842687726020813,
|
||
|
|
"step": 1710,
|
||
|
|
"valid_targets_mean": 4767.0,
|
||
|
|
"valid_targets_min": 343
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9166666666666665,
|
||
|
|
"grad_norm": 0.5055407945367937,
|
||
|
|
"learning_rate": 2.899506747582965e-05,
|
||
|
|
"loss": 0.1894,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17843550443649292,
|
||
|
|
"step": 1715,
|
||
|
|
"valid_targets_mean": 4355.5,
|
||
|
|
"valid_targets_min": 418
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.925170068027211,
|
||
|
|
"grad_norm": 0.45597283974698105,
|
||
|
|
"learning_rate": 2.8919232977124177e-05,
|
||
|
|
"loss": 0.1744,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16779714822769165,
|
||
|
|
"step": 1720,
|
||
|
|
"valid_targets_mean": 5379.9,
|
||
|
|
"valid_targets_min": 648
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.933673469387755,
|
||
|
|
"grad_norm": 0.42176217801227517,
|
||
|
|
"learning_rate": 2.884323807092423e-05,
|
||
|
|
"loss": 0.1809,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16843809187412262,
|
||
|
|
"step": 1725,
|
||
|
|
"valid_targets_mean": 5274.2,
|
||
|
|
"valid_targets_min": 504
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.942176870748299,
|
||
|
|
"grad_norm": 0.44551785180298836,
|
||
|
|
"learning_rate": 2.8767084123956334e-05,
|
||
|
|
"loss": 0.1856,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17556804418563843,
|
||
|
|
"step": 1730,
|
||
|
|
"valid_targets_mean": 5447.4,
|
||
|
|
"valid_targets_min": 276
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9506802721088436,
|
||
|
|
"grad_norm": 0.4438002809067087,
|
||
|
|
"learning_rate": 2.8690772505807307e-05,
|
||
|
|
"loss": 0.2012,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21305593848228455,
|
||
|
|
"step": 1735,
|
||
|
|
"valid_targets_mean": 6458.3,
|
||
|
|
"valid_targets_min": 336
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9591836734693877,
|
||
|
|
"grad_norm": 0.448551732212738,
|
||
|
|
"learning_rate": 2.8614304588899588e-05,
|
||
|
|
"loss": 0.1824,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16192924976348877,
|
||
|
|
"step": 1740,
|
||
|
|
"valid_targets_mean": 4411.8,
|
||
|
|
"valid_targets_min": 331
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.967687074829932,
|
||
|
|
"grad_norm": 0.5026562063952194,
|
||
|
|
"learning_rate": 2.8537681748466557e-05,
|
||
|
|
"loss": 0.1923,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18430952727794647,
|
||
|
|
"step": 1745,
|
||
|
|
"valid_targets_mean": 5175.2,
|
||
|
|
"valid_targets_min": 292
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9761904761904763,
|
||
|
|
"grad_norm": 0.4741010643319806,
|
||
|
|
"learning_rate": 2.8460905362527832e-05,
|
||
|
|
"loss": 0.1826,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21873563528060913,
|
||
|
|
"step": 1750,
|
||
|
|
"valid_targets_mean": 5759.1,
|
||
|
|
"valid_targets_min": 362
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9846938775510203,
|
||
|
|
"grad_norm": 0.4382554485503363,
|
||
|
|
"learning_rate": 2.8383976811864416e-05,
|
||
|
|
"loss": 0.1756,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18096034228801727,
|
||
|
|
"step": 1755,
|
||
|
|
"valid_targets_mean": 5076.9,
|
||
|
|
"valid_targets_min": 319
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9931972789115644,
|
||
|
|
"grad_norm": 0.4660310311841204,
|
||
|
|
"learning_rate": 2.8306897479993954e-05,
|
||
|
|
"loss": 0.1812,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18245714902877808,
|
||
|
|
"step": 1760,
|
||
|
|
"valid_targets_mean": 4965.2,
|
||
|
|
"valid_targets_min": 672
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.001700680272109,
|
||
|
|
"grad_norm": 0.4520397265541428,
|
||
|
|
"learning_rate": 2.8229668753145797e-05,
|
||
|
|
"loss": 0.1851,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17070698738098145,
|
||
|
|
"step": 1765,
|
||
|
|
"valid_targets_mean": 5591.6,
|
||
|
|
"valid_targets_min": 2714
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.010204081632653,
|
||
|
|
"grad_norm": 0.5577514109244567,
|
||
|
|
"learning_rate": 2.815229202023607e-05,
|
||
|
|
"loss": 0.1669,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16992975771427155,
|
||
|
|
"step": 1770,
|
||
|
|
"valid_targets_mean": 4508.6,
|
||
|
|
"valid_targets_min": 389
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0187074829931975,
|
||
|
|
"grad_norm": 0.48100122627937225,
|
||
|
|
"learning_rate": 2.8074768672842716e-05,
|
||
|
|
"loss": 0.184,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18181568384170532,
|
||
|
|
"step": 1775,
|
||
|
|
"valid_targets_mean": 4869.8,
|
||
|
|
"valid_targets_min": 644
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0272108843537415,
|
||
|
|
"grad_norm": 0.4183061720775051,
|
||
|
|
"learning_rate": 2.7997100105180455e-05,
|
||
|
|
"loss": 0.1698,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14422762393951416,
|
||
|
|
"step": 1780,
|
||
|
|
"valid_targets_mean": 5742.6,
|
||
|
|
"valid_targets_min": 278
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0357142857142856,
|
||
|
|
"grad_norm": 0.44953367048759907,
|
||
|
|
"learning_rate": 2.7919287714075714e-05,
|
||
|
|
"loss": 0.1577,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17385989427566528,
|
||
|
|
"step": 1785,
|
||
|
|
"valid_targets_mean": 5713.5,
|
||
|
|
"valid_targets_min": 495
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.04421768707483,
|
||
|
|
"grad_norm": 0.46732012607735846,
|
||
|
|
"learning_rate": 2.7841332898941513e-05,
|
||
|
|
"loss": 0.1762,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19512367248535156,
|
||
|
|
"step": 1790,
|
||
|
|
"valid_targets_mean": 6563.8,
|
||
|
|
"valid_targets_min": 4414
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.052721088435374,
|
||
|
|
"grad_norm": 0.496506747312761,
|
||
|
|
"learning_rate": 2.7763237061752276e-05,
|
||
|
|
"loss": 0.1587,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13929255306720734,
|
||
|
|
"step": 1795,
|
||
|
|
"valid_targets_mean": 5757.0,
|
||
|
|
"valid_targets_min": 2009
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.061224489795918,
|
||
|
|
"grad_norm": 0.458039951586885,
|
||
|
|
"learning_rate": 2.768500160701864e-05,
|
||
|
|
"loss": 0.1721,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16377782821655273,
|
||
|
|
"step": 1800,
|
||
|
|
"valid_targets_mean": 5703.6,
|
||
|
|
"valid_targets_min": 2857
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0697278911564627,
|
||
|
|
"grad_norm": 0.5381137925495079,
|
||
|
|
"learning_rate": 2.7606627941762186e-05,
|
||
|
|
"loss": 0.1624,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17157962918281555,
|
||
|
|
"step": 1805,
|
||
|
|
"valid_targets_mean": 4385.7,
|
||
|
|
"valid_targets_min": 480
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.078231292517007,
|
||
|
|
"grad_norm": 0.4947846490207794,
|
||
|
|
"learning_rate": 2.752811747549013e-05,
|
||
|
|
"loss": 0.1843,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22358940541744232,
|
||
|
|
"step": 1810,
|
||
|
|
"valid_targets_mean": 6425.8,
|
||
|
|
"valid_targets_min": 1231
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.086734693877551,
|
||
|
|
"grad_norm": 0.5485066342584352,
|
||
|
|
"learning_rate": 2.7449471620169988e-05,
|
||
|
|
"loss": 0.1624,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17057745158672333,
|
||
|
|
"step": 1815,
|
||
|
|
"valid_targets_mean": 5166.1,
|
||
|
|
"valid_targets_min": 1988
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0952380952380953,
|
||
|
|
"grad_norm": 0.5080968971922526,
|
||
|
|
"learning_rate": 2.7370691790204173e-05,
|
||
|
|
"loss": 0.1651,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15060283243656158,
|
||
|
|
"step": 1820,
|
||
|
|
"valid_targets_mean": 5219.8,
|
||
|
|
"valid_targets_min": 493
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1037414965986394,
|
||
|
|
"grad_norm": 0.5058105967341384,
|
||
|
|
"learning_rate": 2.7291779402404538e-05,
|
||
|
|
"loss": 0.1633,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15751765668392181,
|
||
|
|
"step": 1825,
|
||
|
|
"valid_targets_mean": 4031.5,
|
||
|
|
"valid_targets_min": 452
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1122448979591835,
|
||
|
|
"grad_norm": 0.45993653404856305,
|
||
|
|
"learning_rate": 2.7212735875966946e-05,
|
||
|
|
"loss": 0.1715,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15872247517108917,
|
||
|
|
"step": 1830,
|
||
|
|
"valid_targets_mean": 5667.1,
|
||
|
|
"valid_targets_min": 2631
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.120748299319728,
|
||
|
|
"grad_norm": 0.49712217901684974,
|
||
|
|
"learning_rate": 2.713356263244569e-05,
|
||
|
|
"loss": 0.181,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18740703165531158,
|
||
|
|
"step": 1835,
|
||
|
|
"valid_targets_mean": 5154.7,
|
||
|
|
"valid_targets_min": 680
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.129251700680272,
|
||
|
|
"grad_norm": 0.47800675048159197,
|
||
|
|
"learning_rate": 2.7054261095727977e-05,
|
||
|
|
"loss": 0.1679,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15372514724731445,
|
||
|
|
"step": 1840,
|
||
|
|
"valid_targets_mean": 5205.6,
|
||
|
|
"valid_targets_min": 783
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.137755102040816,
|
||
|
|
"grad_norm": 0.47024291506371513,
|
||
|
|
"learning_rate": 2.6974832692008278e-05,
|
||
|
|
"loss": 0.1653,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16864655911922455,
|
||
|
|
"step": 1845,
|
||
|
|
"valid_targets_mean": 5267.2,
|
||
|
|
"valid_targets_min": 769
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1462585034013606,
|
||
|
|
"grad_norm": 0.4706237870585836,
|
||
|
|
"learning_rate": 2.6895278849762694e-05,
|
||
|
|
"loss": 0.1593,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14508825540542603,
|
||
|
|
"step": 1850,
|
||
|
|
"valid_targets_mean": 4849.2,
|
||
|
|
"valid_targets_min": 387
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1547619047619047,
|
||
|
|
"grad_norm": 0.5016546638182298,
|
||
|
|
"learning_rate": 2.68156009997233e-05,
|
||
|
|
"loss": 0.1586,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.173442542552948,
|
||
|
|
"step": 1855,
|
||
|
|
"valid_targets_mean": 4968.6,
|
||
|
|
"valid_targets_min": 328
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.163265306122449,
|
||
|
|
"grad_norm": 0.5135357192107041,
|
||
|
|
"learning_rate": 2.673580057485234e-05,
|
||
|
|
"loss": 0.172,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16546142101287842,
|
||
|
|
"step": 1860,
|
||
|
|
"valid_targets_mean": 4211.1,
|
||
|
|
"valid_targets_min": 504
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.171768707482993,
|
||
|
|
"grad_norm": 0.5003107178447966,
|
||
|
|
"learning_rate": 2.6655879010316535e-05,
|
||
|
|
"loss": 0.1666,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17069391906261444,
|
||
|
|
"step": 1865,
|
||
|
|
"valid_targets_mean": 4661.6,
|
||
|
|
"valid_targets_min": 473
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1802721088435373,
|
||
|
|
"grad_norm": 0.47584841798120603,
|
||
|
|
"learning_rate": 2.657583774346121e-05,
|
||
|
|
"loss": 0.1673,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18579839169979095,
|
||
|
|
"step": 1870,
|
||
|
|
"valid_targets_mean": 5495.0,
|
||
|
|
"valid_targets_min": 2740
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.188775510204082,
|
||
|
|
"grad_norm": 0.4931209999265287,
|
||
|
|
"learning_rate": 2.649567821378449e-05,
|
||
|
|
"loss": 0.1578,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14383836090564728,
|
||
|
|
"step": 1875,
|
||
|
|
"valid_targets_mean": 4991.2,
|
||
|
|
"valid_targets_min": 648
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.197278911564626,
|
||
|
|
"grad_norm": 0.49044425967551464,
|
||
|
|
"learning_rate": 2.641540186291138e-05,
|
||
|
|
"loss": 0.1619,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18027395009994507,
|
||
|
|
"step": 1880,
|
||
|
|
"valid_targets_mean": 4987.1,
|
||
|
|
"valid_targets_min": 529
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.20578231292517,
|
||
|
|
"grad_norm": 0.5143241255675114,
|
||
|
|
"learning_rate": 2.6335010134567852e-05,
|
||
|
|
"loss": 0.1649,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16928228735923767,
|
||
|
|
"step": 1885,
|
||
|
|
"valid_targets_mean": 4937.5,
|
||
|
|
"valid_targets_min": 353
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2142857142857144,
|
||
|
|
"grad_norm": 0.5375678401526168,
|
||
|
|
"learning_rate": 2.6254504474554883e-05,
|
||
|
|
"loss": 0.1643,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16120317578315735,
|
||
|
|
"step": 1890,
|
||
|
|
"valid_targets_mean": 3813.0,
|
||
|
|
"valid_targets_min": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2227891156462585,
|
||
|
|
"grad_norm": 0.4641228961390609,
|
||
|
|
"learning_rate": 2.617388633072244e-05,
|
||
|
|
"loss": 0.1695,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16579824686050415,
|
||
|
|
"step": 1895,
|
||
|
|
"valid_targets_mean": 6118.4,
|
||
|
|
"valid_targets_min": 2369
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2312925170068025,
|
||
|
|
"grad_norm": 0.4558014809555061,
|
||
|
|
"learning_rate": 2.609315715294346e-05,
|
||
|
|
"loss": 0.1618,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16386950016021729,
|
||
|
|
"step": 1900,
|
||
|
|
"valid_targets_mean": 5869.9,
|
||
|
|
"valid_targets_min": 607
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.239795918367347,
|
||
|
|
"grad_norm": 0.44271035237391254,
|
||
|
|
"learning_rate": 2.6012318393087754e-05,
|
||
|
|
"loss": 0.1626,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.160554438829422,
|
||
|
|
"step": 1905,
|
||
|
|
"valid_targets_mean": 4916.1,
|
||
|
|
"valid_targets_min": 2662
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.248299319727891,
|
||
|
|
"grad_norm": 0.4715222249414142,
|
||
|
|
"learning_rate": 2.593137150499591e-05,
|
||
|
|
"loss": 0.1797,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1527043730020523,
|
||
|
|
"step": 1910,
|
||
|
|
"valid_targets_mean": 4481.6,
|
||
|
|
"valid_targets_min": 424
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2568027210884356,
|
||
|
|
"grad_norm": 0.441189404780352,
|
||
|
|
"learning_rate": 2.5850317944453153e-05,
|
||
|
|
"loss": 0.1624,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1682853400707245,
|
||
|
|
"step": 1915,
|
||
|
|
"valid_targets_mean": 6198.2,
|
||
|
|
"valid_targets_min": 3863
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2653061224489797,
|
||
|
|
"grad_norm": 1.3052707770766574,
|
||
|
|
"learning_rate": 2.5769159169163137e-05,
|
||
|
|
"loss": 0.1677,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15218862891197205,
|
||
|
|
"step": 1920,
|
||
|
|
"valid_targets_mean": 5376.6,
|
||
|
|
"valid_targets_min": 878
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2738095238095237,
|
||
|
|
"grad_norm": 0.45441857407345054,
|
||
|
|
"learning_rate": 2.568789663872175e-05,
|
||
|
|
"loss": 0.1531,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.158279687166214,
|
||
|
|
"step": 1925,
|
||
|
|
"valid_targets_mean": 5163.6,
|
||
|
|
"valid_targets_min": 430
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.282312925170068,
|
||
|
|
"grad_norm": 0.4783488385173733,
|
||
|
|
"learning_rate": 2.5606531814590868e-05,
|
||
|
|
"loss": 0.1692,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16506332159042358,
|
||
|
|
"step": 1930,
|
||
|
|
"valid_targets_mean": 4505.9,
|
||
|
|
"valid_targets_min": 341
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2908163265306123,
|
||
|
|
"grad_norm": 0.4555733556088053,
|
||
|
|
"learning_rate": 2.552506616007205e-05,
|
||
|
|
"loss": 0.1665,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17181992530822754,
|
||
|
|
"step": 1935,
|
||
|
|
"valid_targets_mean": 5269.8,
|
||
|
|
"valid_targets_min": 591
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2993197278911564,
|
||
|
|
"grad_norm": 0.5076558428269352,
|
||
|
|
"learning_rate": 2.544350114028025e-05,
|
||
|
|
"loss": 0.1621,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19029831886291504,
|
||
|
|
"step": 1940,
|
||
|
|
"valid_targets_mean": 6025.2,
|
||
|
|
"valid_targets_min": 375
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.307823129251701,
|
||
|
|
"grad_norm": 0.46104761793655025,
|
||
|
|
"learning_rate": 2.5361838222117435e-05,
|
||
|
|
"loss": 0.1712,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17312756180763245,
|
||
|
|
"step": 1945,
|
||
|
|
"valid_targets_mean": 5350.4,
|
||
|
|
"valid_targets_min": 344
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.316326530612245,
|
||
|
|
"grad_norm": 0.485840354945783,
|
||
|
|
"learning_rate": 2.5280078874246227e-05,
|
||
|
|
"loss": 0.1763,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18720793724060059,
|
||
|
|
"step": 1950,
|
||
|
|
"valid_targets_mean": 5318.8,
|
||
|
|
"valid_targets_min": 416
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.324829931972789,
|
||
|
|
"grad_norm": 0.4686384486704096,
|
||
|
|
"learning_rate": 2.519822456706349e-05,
|
||
|
|
"loss": 0.1644,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17492544651031494,
|
||
|
|
"step": 1955,
|
||
|
|
"valid_targets_mean": 5278.1,
|
||
|
|
"valid_targets_min": 459
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3333333333333335,
|
||
|
|
"grad_norm": 0.5326463889293587,
|
||
|
|
"learning_rate": 2.5116276772673863e-05,
|
||
|
|
"loss": 0.172,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15641175210475922,
|
||
|
|
"step": 1960,
|
||
|
|
"valid_targets_mean": 4587.9,
|
||
|
|
"valid_targets_min": 1001
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3418367346938775,
|
||
|
|
"grad_norm": 0.47154166690173754,
|
||
|
|
"learning_rate": 2.503423696486332e-05,
|
||
|
|
"loss": 0.1648,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17395135760307312,
|
||
|
|
"step": 1965,
|
||
|
|
"valid_targets_mean": 5421.9,
|
||
|
|
"valid_targets_min": 2278
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3503401360544216,
|
||
|
|
"grad_norm": 0.6315325902013319,
|
||
|
|
"learning_rate": 2.495210661907263e-05,
|
||
|
|
"loss": 0.1529,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15730559825897217,
|
||
|
|
"step": 1970,
|
||
|
|
"valid_targets_mean": 5285.7,
|
||
|
|
"valid_targets_min": 315
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.358843537414966,
|
||
|
|
"grad_norm": 0.4657843180141234,
|
||
|
|
"learning_rate": 2.4869887212370847e-05,
|
||
|
|
"loss": 0.1777,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1763974279165268,
|
||
|
|
"step": 1975,
|
||
|
|
"valid_targets_mean": 5252.2,
|
||
|
|
"valid_targets_min": 553
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.36734693877551,
|
||
|
|
"grad_norm": 0.5676003662112784,
|
||
|
|
"learning_rate": 2.4787580223428736e-05,
|
||
|
|
"loss": 0.1693,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18310853838920593,
|
||
|
|
"step": 1980,
|
||
|
|
"valid_targets_mean": 5984.9,
|
||
|
|
"valid_targets_min": 3224
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3758503401360542,
|
||
|
|
"grad_norm": 0.4786440340216272,
|
||
|
|
"learning_rate": 2.470518713249217e-05,
|
||
|
|
"loss": 0.1616,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14558261632919312,
|
||
|
|
"step": 1985,
|
||
|
|
"valid_targets_mean": 4911.1,
|
||
|
|
"valid_targets_min": 520
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3843537414965987,
|
||
|
|
"grad_norm": 0.4356605850128712,
|
||
|
|
"learning_rate": 2.4622709421355535e-05,
|
||
|
|
"loss": 0.1636,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16069340705871582,
|
||
|
|
"step": 1990,
|
||
|
|
"valid_targets_mean": 5421.4,
|
||
|
|
"valid_targets_min": 348
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.392857142857143,
|
||
|
|
"grad_norm": 0.4316837663637552,
|
||
|
|
"learning_rate": 2.4540148573335057e-05,
|
||
|
|
"loss": 0.166,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15478411316871643,
|
||
|
|
"step": 1995,
|
||
|
|
"valid_targets_mean": 5890.4,
|
||
|
|
"valid_targets_min": 2004
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4013605442176873,
|
||
|
|
"grad_norm": 0.45536684902236574,
|
||
|
|
"learning_rate": 2.445750607324212e-05,
|
||
|
|
"loss": 0.1797,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18176986277103424,
|
||
|
|
"step": 2000,
|
||
|
|
"valid_targets_mean": 5820.0,
|
||
|
|
"valid_targets_min": 571
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4098639455782314,
|
||
|
|
"grad_norm": 0.4902862102973783,
|
||
|
|
"learning_rate": 2.4374783407356602e-05,
|
||
|
|
"loss": 0.1567,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15701062977313995,
|
||
|
|
"step": 2005,
|
||
|
|
"valid_targets_mean": 4827.1,
|
||
|
|
"valid_targets_min": 629
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4183673469387754,
|
||
|
|
"grad_norm": 0.48845385555850057,
|
||
|
|
"learning_rate": 2.4291982063400096e-05,
|
||
|
|
"loss": 0.1678,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16644497215747833,
|
||
|
|
"step": 2010,
|
||
|
|
"valid_targets_mean": 5014.6,
|
||
|
|
"valid_targets_min": 284
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4268707482993195,
|
||
|
|
"grad_norm": 0.45383224445693654,
|
||
|
|
"learning_rate": 2.4209103530509196e-05,
|
||
|
|
"loss": 0.164,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17515942454338074,
|
||
|
|
"step": 2015,
|
||
|
|
"valid_targets_mean": 5743.1,
|
||
|
|
"valid_targets_min": 406
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.435374149659864,
|
||
|
|
"grad_norm": 0.5269027225996431,
|
||
|
|
"learning_rate": 2.412614929920868e-05,
|
||
|
|
"loss": 0.1767,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2476090043783188,
|
||
|
|
"step": 2020,
|
||
|
|
"valid_targets_mean": 5020.2,
|
||
|
|
"valid_targets_min": 327
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.443877551020408,
|
||
|
|
"grad_norm": 0.5990196131401615,
|
||
|
|
"learning_rate": 2.4043120861384723e-05,
|
||
|
|
"loss": 0.1587,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16078883409500122,
|
||
|
|
"step": 2025,
|
||
|
|
"valid_targets_mean": 3829.6,
|
||
|
|
"valid_targets_min": 229
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4523809523809526,
|
||
|
|
"grad_norm": 0.4394621781684443,
|
||
|
|
"learning_rate": 2.3960019710258068e-05,
|
||
|
|
"loss": 0.1855,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21002984046936035,
|
||
|
|
"step": 2030,
|
||
|
|
"valid_targets_mean": 5720.8,
|
||
|
|
"valid_targets_min": 2441
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4608843537414966,
|
||
|
|
"grad_norm": 0.4612558726720703,
|
||
|
|
"learning_rate": 2.3876847340357164e-05,
|
||
|
|
"loss": 0.1662,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1559475064277649,
|
||
|
|
"step": 2035,
|
||
|
|
"valid_targets_mean": 5102.2,
|
||
|
|
"valid_targets_min": 745
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4693877551020407,
|
||
|
|
"grad_norm": 0.48397637882784506,
|
||
|
|
"learning_rate": 2.3793605247491303e-05,
|
||
|
|
"loss": 0.1697,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17857158184051514,
|
||
|
|
"step": 2040,
|
||
|
|
"valid_targets_mean": 5443.9,
|
||
|
|
"valid_targets_min": 1099
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.477891156462585,
|
||
|
|
"grad_norm": 0.5347861503218174,
|
||
|
|
"learning_rate": 2.371029492872369e-05,
|
||
|
|
"loss": 0.1663,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1758844405412674,
|
||
|
|
"step": 2045,
|
||
|
|
"valid_targets_mean": 5166.1,
|
||
|
|
"valid_targets_min": 238
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4863945578231292,
|
||
|
|
"grad_norm": 0.51108946728499,
|
||
|
|
"learning_rate": 2.362691788234453e-05,
|
||
|
|
"loss": 0.1748,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1522696614265442,
|
||
|
|
"step": 2050,
|
||
|
|
"valid_targets_mean": 5479.4,
|
||
|
|
"valid_targets_min": 361
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4948979591836733,
|
||
|
|
"grad_norm": 0.45763921438001226,
|
||
|
|
"learning_rate": 2.3543475607844105e-05,
|
||
|
|
"loss": 0.1674,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15340670943260193,
|
||
|
|
"step": 2055,
|
||
|
|
"valid_targets_mean": 5066.4,
|
||
|
|
"valid_targets_min": 1830
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.503401360544218,
|
||
|
|
"grad_norm": 0.47066613154061715,
|
||
|
|
"learning_rate": 2.3459969605885785e-05,
|
||
|
|
"loss": 0.1591,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16512474417686462,
|
||
|
|
"step": 2060,
|
||
|
|
"valid_targets_mean": 5034.3,
|
||
|
|
"valid_targets_min": 631
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.511904761904762,
|
||
|
|
"grad_norm": 0.545372946661658,
|
||
|
|
"learning_rate": 2.3376401378279037e-05,
|
||
|
|
"loss": 0.1597,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14831149578094482,
|
||
|
|
"step": 2065,
|
||
|
|
"valid_targets_mean": 3900.6,
|
||
|
|
"valid_targets_min": 411
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.520408163265306,
|
||
|
|
"grad_norm": 0.496735169684081,
|
||
|
|
"learning_rate": 2.329277242795243e-05,
|
||
|
|
"loss": 0.1694,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17439019680023193,
|
||
|
|
"step": 2070,
|
||
|
|
"valid_targets_mean": 5492.9,
|
||
|
|
"valid_targets_min": 2860
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5289115646258504,
|
||
|
|
"grad_norm": 0.477761468890142,
|
||
|
|
"learning_rate": 2.320908425892658e-05,
|
||
|
|
"loss": 0.1743,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15913477540016174,
|
||
|
|
"step": 2075,
|
||
|
|
"valid_targets_mean": 4576.9,
|
||
|
|
"valid_targets_min": 577
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5374149659863945,
|
||
|
|
"grad_norm": 0.48551019574331766,
|
||
|
|
"learning_rate": 2.3125338376287137e-05,
|
||
|
|
"loss": 0.1552,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14921371638774872,
|
||
|
|
"step": 2080,
|
||
|
|
"valid_targets_mean": 5279.2,
|
||
|
|
"valid_targets_min": 694
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.545918367346939,
|
||
|
|
"grad_norm": 0.46948062378291805,
|
||
|
|
"learning_rate": 2.3041536286157706e-05,
|
||
|
|
"loss": 0.1771,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.179522305727005,
|
||
|
|
"step": 2085,
|
||
|
|
"valid_targets_mean": 5614.4,
|
||
|
|
"valid_targets_min": 850
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.554421768707483,
|
||
|
|
"grad_norm": 0.492702049755144,
|
||
|
|
"learning_rate": 2.295767949567272e-05,
|
||
|
|
"loss": 0.1611,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1650063842535019,
|
||
|
|
"step": 2090,
|
||
|
|
"valid_targets_mean": 4799.0,
|
||
|
|
"valid_targets_min": 765
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.562925170068027,
|
||
|
|
"grad_norm": 0.5154772550707454,
|
||
|
|
"learning_rate": 2.28737695129504e-05,
|
||
|
|
"loss": 0.1692,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15846556425094604,
|
||
|
|
"step": 2095,
|
||
|
|
"valid_targets_mean": 4471.6,
|
||
|
|
"valid_targets_min": 408
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.571428571428571,
|
||
|
|
"grad_norm": 0.4564428688516693,
|
||
|
|
"learning_rate": 2.2789807847065574e-05,
|
||
|
|
"loss": 0.1619,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15885907411575317,
|
||
|
|
"step": 2100,
|
||
|
|
"valid_targets_mean": 5255.8,
|
||
|
|
"valid_targets_min": 433
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5799319727891157,
|
||
|
|
"grad_norm": 0.4222558582383424,
|
||
|
|
"learning_rate": 2.2705796008022592e-05,
|
||
|
|
"loss": 0.1701,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15134954452514648,
|
||
|
|
"step": 2105,
|
||
|
|
"valid_targets_mean": 5776.4,
|
||
|
|
"valid_targets_min": 1733
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5884353741496597,
|
||
|
|
"grad_norm": 0.46490484815807,
|
||
|
|
"learning_rate": 2.2621735506728126e-05,
|
||
|
|
"loss": 0.1677,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1756136566400528,
|
||
|
|
"step": 2110,
|
||
|
|
"valid_targets_mean": 5899.6,
|
||
|
|
"valid_targets_min": 478
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5969387755102042,
|
||
|
|
"grad_norm": 0.7286914783210093,
|
||
|
|
"learning_rate": 2.2537627854963994e-05,
|
||
|
|
"loss": 0.1704,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16937755048274994,
|
||
|
|
"step": 2115,
|
||
|
|
"valid_targets_mean": 5804.9,
|
||
|
|
"valid_targets_min": 785
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6054421768707483,
|
||
|
|
"grad_norm": 0.516314011172263,
|
||
|
|
"learning_rate": 2.2453474565360018e-05,
|
||
|
|
"loss": 0.1732,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18747320771217346,
|
||
|
|
"step": 2120,
|
||
|
|
"valid_targets_mean": 4925.5,
|
||
|
|
"valid_targets_min": 413
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6139455782312924,
|
||
|
|
"grad_norm": 0.5306295275754309,
|
||
|
|
"learning_rate": 2.2369277151366776e-05,
|
||
|
|
"loss": 0.1647,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1703931987285614,
|
||
|
|
"step": 2125,
|
||
|
|
"valid_targets_mean": 3619.0,
|
||
|
|
"valid_targets_min": 303
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.622448979591837,
|
||
|
|
"grad_norm": 0.6806423333618543,
|
||
|
|
"learning_rate": 2.228503712722841e-05,
|
||
|
|
"loss": 0.1711,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15308068692684174,
|
||
|
|
"step": 2130,
|
||
|
|
"valid_targets_mean": 4672.9,
|
||
|
|
"valid_targets_min": 455
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.630952380952381,
|
||
|
|
"grad_norm": 0.42208826156580315,
|
||
|
|
"learning_rate": 2.220075600795537e-05,
|
||
|
|
"loss": 0.1609,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15407651662826538,
|
||
|
|
"step": 2135,
|
||
|
|
"valid_targets_mean": 5448.2,
|
||
|
|
"valid_targets_min": 652
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6394557823129254,
|
||
|
|
"grad_norm": 0.4710117254586844,
|
||
|
|
"learning_rate": 2.2116435309297177e-05,
|
||
|
|
"loss": 0.1556,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17985662817955017,
|
||
|
|
"step": 2140,
|
||
|
|
"valid_targets_mean": 5338.1,
|
||
|
|
"valid_targets_min": 649
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6479591836734695,
|
||
|
|
"grad_norm": 0.4311477385659195,
|
||
|
|
"learning_rate": 2.203207654771519e-05,
|
||
|
|
"loss": 0.1603,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15320411324501038,
|
||
|
|
"step": 2145,
|
||
|
|
"valid_targets_mean": 5767.8,
|
||
|
|
"valid_targets_min": 3271
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6564625850340136,
|
||
|
|
"grad_norm": 0.45065236562479644,
|
||
|
|
"learning_rate": 2.1947681240355266e-05,
|
||
|
|
"loss": 0.1735,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17802459001541138,
|
||
|
|
"step": 2150,
|
||
|
|
"valid_targets_mean": 5735.2,
|
||
|
|
"valid_targets_min": 2095
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6649659863945576,
|
||
|
|
"grad_norm": 0.4679055285720377,
|
||
|
|
"learning_rate": 2.1863250905020566e-05,
|
||
|
|
"loss": 0.1615,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1417907327413559,
|
||
|
|
"step": 2155,
|
||
|
|
"valid_targets_mean": 4572.4,
|
||
|
|
"valid_targets_min": 472
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.673469387755102,
|
||
|
|
"grad_norm": 0.4646082226154742,
|
||
|
|
"learning_rate": 2.177878706014418e-05,
|
||
|
|
"loss": 0.1571,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13954725861549377,
|
||
|
|
"step": 2160,
|
||
|
|
"valid_targets_mean": 4848.2,
|
||
|
|
"valid_targets_min": 346
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.681972789115646,
|
||
|
|
"grad_norm": 0.5064425845227202,
|
||
|
|
"learning_rate": 2.1694291224761845e-05,
|
||
|
|
"loss": 0.1675,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16030800342559814,
|
||
|
|
"step": 2165,
|
||
|
|
"valid_targets_mean": 4531.7,
|
||
|
|
"valid_targets_min": 313
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6904761904761907,
|
||
|
|
"grad_norm": 0.4067402697324011,
|
||
|
|
"learning_rate": 2.160976491848465e-05,
|
||
|
|
"loss": 0.1722,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21204915642738342,
|
||
|
|
"step": 2170,
|
||
|
|
"valid_targets_mean": 6558.9,
|
||
|
|
"valid_targets_min": 1602
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6989795918367347,
|
||
|
|
"grad_norm": 0.5037466189353319,
|
||
|
|
"learning_rate": 2.1525209661471677e-05,
|
||
|
|
"loss": 0.1525,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15384790301322937,
|
||
|
|
"step": 2175,
|
||
|
|
"valid_targets_mean": 3843.7,
|
||
|
|
"valid_targets_min": 423
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.707482993197279,
|
||
|
|
"grad_norm": 0.42884641012064423,
|
||
|
|
"learning_rate": 2.1440626974402664e-05,
|
||
|
|
"loss": 0.1722,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16450552642345428,
|
||
|
|
"step": 2180,
|
||
|
|
"valid_targets_mean": 5757.4,
|
||
|
|
"valid_targets_min": 488
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.715986394557823,
|
||
|
|
"grad_norm": 0.415844885766819,
|
||
|
|
"learning_rate": 2.135601837845068e-05,
|
||
|
|
"loss": 0.1648,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16901686787605286,
|
||
|
|
"step": 2185,
|
||
|
|
"valid_targets_mean": 5592.8,
|
||
|
|
"valid_targets_min": 305
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7244897959183674,
|
||
|
|
"grad_norm": 0.4434158100547713,
|
||
|
|
"learning_rate": 2.1271385395254728e-05,
|
||
|
|
"loss": 0.1654,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16155441105365753,
|
||
|
|
"step": 2190,
|
||
|
|
"valid_targets_mean": 5630.5,
|
||
|
|
"valid_targets_min": 440
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7329931972789114,
|
||
|
|
"grad_norm": 0.481228187292928,
|
||
|
|
"learning_rate": 2.118672954689242e-05,
|
||
|
|
"loss": 0.1617,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17072485387325287,
|
||
|
|
"step": 2195,
|
||
|
|
"valid_targets_mean": 4525.3,
|
||
|
|
"valid_targets_min": 423
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.741496598639456,
|
||
|
|
"grad_norm": 0.48334148504745406,
|
||
|
|
"learning_rate": 2.1102052355852586e-05,
|
||
|
|
"loss": 0.1665,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16323237121105194,
|
||
|
|
"step": 2200,
|
||
|
|
"valid_targets_mean": 4942.8,
|
||
|
|
"valid_targets_min": 477
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.75,
|
||
|
|
"grad_norm": 0.44903660302342335,
|
||
|
|
"learning_rate": 2.1017355345007883e-05,
|
||
|
|
"loss": 0.1687,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1585753709077835,
|
||
|
|
"step": 2205,
|
||
|
|
"valid_targets_mean": 4774.4,
|
||
|
|
"valid_targets_min": 351
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.758503401360544,
|
||
|
|
"grad_norm": 0.49189814602587245,
|
||
|
|
"learning_rate": 2.093264003758742e-05,
|
||
|
|
"loss": 0.1695,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1656944900751114,
|
||
|
|
"step": 2210,
|
||
|
|
"valid_targets_mean": 4491.1,
|
||
|
|
"valid_targets_min": 367
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7670068027210886,
|
||
|
|
"grad_norm": 0.5202489708129102,
|
||
|
|
"learning_rate": 2.0847907957149348e-05,
|
||
|
|
"loss": 0.1725,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1556740701198578,
|
||
|
|
"step": 2215,
|
||
|
|
"valid_targets_mean": 4057.2,
|
||
|
|
"valid_targets_min": 290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7755102040816326,
|
||
|
|
"grad_norm": 0.8118612378047526,
|
||
|
|
"learning_rate": 2.0763160627553502e-05,
|
||
|
|
"loss": 0.1675,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13766740262508392,
|
||
|
|
"step": 2220,
|
||
|
|
"valid_targets_mean": 4582.0,
|
||
|
|
"valid_targets_min": 410
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.784013605442177,
|
||
|
|
"grad_norm": 0.6910628490888107,
|
||
|
|
"learning_rate": 2.067839957293394e-05,
|
||
|
|
"loss": 0.1574,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16262155771255493,
|
||
|
|
"step": 2225,
|
||
|
|
"valid_targets_mean": 5520.4,
|
||
|
|
"valid_targets_min": 337
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.792517006802721,
|
||
|
|
"grad_norm": 0.48495411705911967,
|
||
|
|
"learning_rate": 2.059362631767156e-05,
|
||
|
|
"loss": 0.1719,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15017281472682953,
|
||
|
|
"step": 2230,
|
||
|
|
"valid_targets_mean": 5267.6,
|
||
|
|
"valid_targets_min": 377
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8010204081632653,
|
||
|
|
"grad_norm": 0.43447923224165097,
|
||
|
|
"learning_rate": 2.0508842386366687e-05,
|
||
|
|
"loss": 0.1643,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14844268560409546,
|
||
|
|
"step": 2235,
|
||
|
|
"valid_targets_mean": 5612.4,
|
||
|
|
"valid_targets_min": 2161
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8095238095238093,
|
||
|
|
"grad_norm": 0.4458128002265448,
|
||
|
|
"learning_rate": 2.0424049303811637e-05,
|
||
|
|
"loss": 0.1726,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16279703378677368,
|
||
|
|
"step": 2240,
|
||
|
|
"valid_targets_mean": 6102.6,
|
||
|
|
"valid_targets_min": 2302
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.818027210884354,
|
||
|
|
"grad_norm": 0.501567414107054,
|
||
|
|
"learning_rate": 2.0339248594963326e-05,
|
||
|
|
"loss": 0.1616,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18073803186416626,
|
||
|
|
"step": 2245,
|
||
|
|
"valid_targets_mean": 4899.9,
|
||
|
|
"valid_targets_min": 666
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.826530612244898,
|
||
|
|
"grad_norm": 0.5345803823631814,
|
||
|
|
"learning_rate": 2.025444178491582e-05,
|
||
|
|
"loss": 0.1725,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21028247475624084,
|
||
|
|
"step": 2250,
|
||
|
|
"valid_targets_mean": 5144.6,
|
||
|
|
"valid_targets_min": 1731
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8350340136054424,
|
||
|
|
"grad_norm": 0.4456078631873231,
|
||
|
|
"learning_rate": 2.0169630398872897e-05,
|
||
|
|
"loss": 0.1747,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2198677808046341,
|
||
|
|
"step": 2255,
|
||
|
|
"valid_targets_mean": 6189.5,
|
||
|
|
"valid_targets_min": 292
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8435374149659864,
|
||
|
|
"grad_norm": 0.5925031735704271,
|
||
|
|
"learning_rate": 2.0084815962120644e-05,
|
||
|
|
"loss": 0.1649,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16646525263786316,
|
||
|
|
"step": 2260,
|
||
|
|
"valid_targets_mean": 4327.4,
|
||
|
|
"valid_targets_min": 331
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8520408163265305,
|
||
|
|
"grad_norm": 0.4544039016029427,
|
||
|
|
"learning_rate": 2e-05,
|
||
|
|
"loss": 0.1703,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15249738097190857,
|
||
|
|
"step": 2265,
|
||
|
|
"valid_targets_mean": 4656.6,
|
||
|
|
"valid_targets_min": 337
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8605442176870746,
|
||
|
|
"grad_norm": 0.4582707423967414,
|
||
|
|
"learning_rate": 1.9915184037879362e-05,
|
||
|
|
"loss": 0.1696,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17427833378314972,
|
||
|
|
"step": 2270,
|
||
|
|
"valid_targets_mean": 5367.1,
|
||
|
|
"valid_targets_min": 358
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.869047619047619,
|
||
|
|
"grad_norm": 0.46411555713344427,
|
||
|
|
"learning_rate": 1.983036960112711e-05,
|
||
|
|
"loss": 0.1728,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16276660561561584,
|
||
|
|
"step": 2275,
|
||
|
|
"valid_targets_mean": 4722.9,
|
||
|
|
"valid_targets_min": 464
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.877551020408163,
|
||
|
|
"grad_norm": 0.4713060579125166,
|
||
|
|
"learning_rate": 1.9745558215084184e-05,
|
||
|
|
"loss": 0.1709,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17453616857528687,
|
||
|
|
"step": 2280,
|
||
|
|
"valid_targets_mean": 5223.8,
|
||
|
|
"valid_targets_min": 324
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8860544217687076,
|
||
|
|
"grad_norm": 0.46550114829857997,
|
||
|
|
"learning_rate": 1.9660751405036677e-05,
|
||
|
|
"loss": 0.1714,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1718387007713318,
|
||
|
|
"step": 2285,
|
||
|
|
"valid_targets_mean": 4657.7,
|
||
|
|
"valid_targets_min": 898
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8945578231292517,
|
||
|
|
"grad_norm": 0.530135067500812,
|
||
|
|
"learning_rate": 1.9575950696188367e-05,
|
||
|
|
"loss": 0.1697,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19820010662078857,
|
||
|
|
"step": 2290,
|
||
|
|
"valid_targets_mean": 5563.4,
|
||
|
|
"valid_targets_min": 381
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9030612244897958,
|
||
|
|
"grad_norm": 0.39299392181893433,
|
||
|
|
"learning_rate": 1.949115761363332e-05,
|
||
|
|
"loss": 0.1676,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15993238985538483,
|
||
|
|
"step": 2295,
|
||
|
|
"valid_targets_mean": 6103.0,
|
||
|
|
"valid_targets_min": 3884
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9115646258503403,
|
||
|
|
"grad_norm": 0.4902746179025179,
|
||
|
|
"learning_rate": 1.9406373682328445e-05,
|
||
|
|
"loss": 0.1695,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16460567712783813,
|
||
|
|
"step": 2300,
|
||
|
|
"valid_targets_mean": 5573.0,
|
||
|
|
"valid_targets_min": 369
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9200680272108843,
|
||
|
|
"grad_norm": 0.45011872843091116,
|
||
|
|
"learning_rate": 1.932160042706607e-05,
|
||
|
|
"loss": 0.1762,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18999271094799042,
|
||
|
|
"step": 2305,
|
||
|
|
"valid_targets_mean": 5767.9,
|
||
|
|
"valid_targets_min": 552
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.928571428571429,
|
||
|
|
"grad_norm": 0.5411827208262425,
|
||
|
|
"learning_rate": 1.9236839372446504e-05,
|
||
|
|
"loss": 0.1731,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1514822244644165,
|
||
|
|
"step": 2310,
|
||
|
|
"valid_targets_mean": 4316.9,
|
||
|
|
"valid_targets_min": 389
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.937074829931973,
|
||
|
|
"grad_norm": 0.42015294768820016,
|
||
|
|
"learning_rate": 1.9152092042850655e-05,
|
||
|
|
"loss": 0.1658,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15763172507286072,
|
||
|
|
"step": 2315,
|
||
|
|
"valid_targets_mean": 5343.4,
|
||
|
|
"valid_targets_min": 2106
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.945578231292517,
|
||
|
|
"grad_norm": 0.4425637527831149,
|
||
|
|
"learning_rate": 1.9067359962412588e-05,
|
||
|
|
"loss": 0.157,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15353673696517944,
|
||
|
|
"step": 2320,
|
||
|
|
"valid_targets_mean": 5375.4,
|
||
|
|
"valid_targets_min": 819
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.954081632653061,
|
||
|
|
"grad_norm": 0.4145524472746335,
|
||
|
|
"learning_rate": 1.8982644654992123e-05,
|
||
|
|
"loss": 0.1711,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16244058310985565,
|
||
|
|
"step": 2325,
|
||
|
|
"valid_targets_mean": 6221.2,
|
||
|
|
"valid_targets_min": 1722
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9625850340136055,
|
||
|
|
"grad_norm": 0.48121261318525693,
|
||
|
|
"learning_rate": 1.8897947644147424e-05,
|
||
|
|
"loss": 0.1844,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17649000883102417,
|
||
|
|
"step": 2330,
|
||
|
|
"valid_targets_mean": 5373.4,
|
||
|
|
"valid_targets_min": 586
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9710884353741496,
|
||
|
|
"grad_norm": 0.4930394096095763,
|
||
|
|
"learning_rate": 1.8813270453107586e-05,
|
||
|
|
"loss": 0.1936,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18251213431358337,
|
||
|
|
"step": 2335,
|
||
|
|
"valid_targets_mean": 4796.8,
|
||
|
|
"valid_targets_min": 631
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.979591836734694,
|
||
|
|
"grad_norm": 0.5306238489330735,
|
||
|
|
"learning_rate": 1.872861460474528e-05,
|
||
|
|
"loss": 0.1686,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17390327155590057,
|
||
|
|
"step": 2340,
|
||
|
|
"valid_targets_mean": 5086.6,
|
||
|
|
"valid_targets_min": 437
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.988095238095238,
|
||
|
|
"grad_norm": 0.46421774260921983,
|
||
|
|
"learning_rate": 1.8643981621549325e-05,
|
||
|
|
"loss": 0.1588,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16242510080337524,
|
||
|
|
"step": 2345,
|
||
|
|
"valid_targets_mean": 4933.1,
|
||
|
|
"valid_targets_min": 865
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.996598639455782,
|
||
|
|
"grad_norm": 0.5185311491929223,
|
||
|
|
"learning_rate": 1.855937302559734e-05,
|
||
|
|
"loss": 0.1662,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17603228986263275,
|
||
|
|
"step": 2350,
|
||
|
|
"valid_targets_mean": 4795.0,
|
||
|
|
"valid_targets_min": 1003
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.005102040816326,
|
||
|
|
"grad_norm": 0.5119100337747748,
|
||
|
|
"learning_rate": 1.8474790338528336e-05,
|
||
|
|
"loss": 0.1617,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15747244656085968,
|
||
|
|
"step": 2355,
|
||
|
|
"valid_targets_mean": 4015.2,
|
||
|
|
"valid_targets_min": 361
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.01360544217687,
|
||
|
|
"grad_norm": 0.4975961537541007,
|
||
|
|
"learning_rate": 1.839023508151536e-05,
|
||
|
|
"loss": 0.1645,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14870601892471313,
|
||
|
|
"step": 2360,
|
||
|
|
"valid_targets_mean": 5503.6,
|
||
|
|
"valid_targets_min": 423
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.022108843537415,
|
||
|
|
"grad_norm": 0.5105477973235613,
|
||
|
|
"learning_rate": 1.8305708775238162e-05,
|
||
|
|
"loss": 0.1587,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12657307088375092,
|
||
|
|
"step": 2365,
|
||
|
|
"valid_targets_mean": 4909.1,
|
||
|
|
"valid_targets_min": 526
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.030612244897959,
|
||
|
|
"grad_norm": 0.5333336831686714,
|
||
|
|
"learning_rate": 1.822121293985583e-05,
|
||
|
|
"loss": 0.152,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1470189392566681,
|
||
|
|
"step": 2370,
|
||
|
|
"valid_targets_mean": 5212.4,
|
||
|
|
"valid_targets_min": 1983
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.039115646258503,
|
||
|
|
"grad_norm": 0.4783778377746073,
|
||
|
|
"learning_rate": 1.813674909497944e-05,
|
||
|
|
"loss": 0.1376,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14330455660820007,
|
||
|
|
"step": 2375,
|
||
|
|
"valid_targets_mean": 4687.0,
|
||
|
|
"valid_targets_min": 286
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.0476190476190474,
|
||
|
|
"grad_norm": 0.43585053622611664,
|
||
|
|
"learning_rate": 1.8052318759644744e-05,
|
||
|
|
"loss": 0.1397,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1439986228942871,
|
||
|
|
"step": 2380,
|
||
|
|
"valid_targets_mean": 5710.4,
|
||
|
|
"valid_targets_min": 349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.0561224489795915,
|
||
|
|
"grad_norm": 0.49010138506359086,
|
||
|
|
"learning_rate": 1.7967923452284824e-05,
|
||
|
|
"loss": 0.1515,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13188686966896057,
|
||
|
|
"step": 2385,
|
||
|
|
"valid_targets_mean": 4638.7,
|
||
|
|
"valid_targets_min": 455
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.0646258503401365,
|
||
|
|
"grad_norm": 0.5442999416689708,
|
||
|
|
"learning_rate": 1.7883564690702826e-05,
|
||
|
|
"loss": 0.1515,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15161950886249542,
|
||
|
|
"step": 2390,
|
||
|
|
"valid_targets_mean": 4190.1,
|
||
|
|
"valid_targets_min": 2020
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.0731292517006805,
|
||
|
|
"grad_norm": 0.49076017657420506,
|
||
|
|
"learning_rate": 1.7799243992044634e-05,
|
||
|
|
"loss": 0.1592,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16888359189033508,
|
||
|
|
"step": 2395,
|
||
|
|
"valid_targets_mean": 5936.6,
|
||
|
|
"valid_targets_min": 2919
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.081632653061225,
|
||
|
|
"grad_norm": 0.5433540601221294,
|
||
|
|
"learning_rate": 1.7714962872771593e-05,
|
||
|
|
"loss": 0.1505,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14678245782852173,
|
||
|
|
"step": 2400,
|
||
|
|
"valid_targets_mean": 5064.6,
|
||
|
|
"valid_targets_min": 384
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.090136054421769,
|
||
|
|
"grad_norm": 0.448130014078134,
|
||
|
|
"learning_rate": 1.7630722848633234e-05,
|
||
|
|
"loss": 0.1419,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15183424949645996,
|
||
|
|
"step": 2405,
|
||
|
|
"valid_targets_mean": 6097.4,
|
||
|
|
"valid_targets_min": 433
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.098639455782313,
|
||
|
|
"grad_norm": 0.46025448014115466,
|
||
|
|
"learning_rate": 1.7546525434639992e-05,
|
||
|
|
"loss": 0.1584,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13818961381912231,
|
||
|
|
"step": 2410,
|
||
|
|
"valid_targets_mean": 5115.4,
|
||
|
|
"valid_targets_min": 789
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.107142857142857,
|
||
|
|
"grad_norm": 0.5710305391772,
|
||
|
|
"learning_rate": 1.7462372145036012e-05,
|
||
|
|
"loss": 0.153,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15131168067455292,
|
||
|
|
"step": 2415,
|
||
|
|
"valid_targets_mean": 4072.5,
|
||
|
|
"valid_targets_min": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.115646258503402,
|
||
|
|
"grad_norm": 0.5338737757068109,
|
||
|
|
"learning_rate": 1.737826449327188e-05,
|
||
|
|
"loss": 0.1595,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1464684009552002,
|
||
|
|
"step": 2420,
|
||
|
|
"valid_targets_mean": 5709.5,
|
||
|
|
"valid_targets_min": 810
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.124149659863946,
|
||
|
|
"grad_norm": 0.4665266406077432,
|
||
|
|
"learning_rate": 1.729420399197741e-05,
|
||
|
|
"loss": 0.1649,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18957260251045227,
|
||
|
|
"step": 2425,
|
||
|
|
"valid_targets_mean": 5518.6,
|
||
|
|
"valid_targets_min": 270
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.13265306122449,
|
||
|
|
"grad_norm": 0.46485834796910214,
|
||
|
|
"learning_rate": 1.7210192152934423e-05,
|
||
|
|
"loss": 0.1364,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1315125972032547,
|
||
|
|
"step": 2430,
|
||
|
|
"valid_targets_mean": 5049.2,
|
||
|
|
"valid_targets_min": 338
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.141156462585034,
|
||
|
|
"grad_norm": 0.4426622430443458,
|
||
|
|
"learning_rate": 1.712623048704961e-05,
|
||
|
|
"loss": 0.1544,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12669606506824493,
|
||
|
|
"step": 2435,
|
||
|
|
"valid_targets_mean": 5076.4,
|
||
|
|
"valid_targets_min": 2004
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.149659863945578,
|
||
|
|
"grad_norm": 0.5024505808377839,
|
||
|
|
"learning_rate": 1.7042320504327285e-05,
|
||
|
|
"loss": 0.1566,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15898427367210388,
|
||
|
|
"step": 2440,
|
||
|
|
"valid_targets_mean": 4983.1,
|
||
|
|
"valid_targets_min": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.158163265306122,
|
||
|
|
"grad_norm": 0.4998064648780282,
|
||
|
|
"learning_rate": 1.6958463713842298e-05,
|
||
|
|
"loss": 0.1564,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15977466106414795,
|
||
|
|
"step": 2445,
|
||
|
|
"valid_targets_mean": 5054.6,
|
||
|
|
"valid_targets_min": 580
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.166666666666667,
|
||
|
|
"grad_norm": 0.5527420401105752,
|
||
|
|
"learning_rate": 1.687466162371286e-05,
|
||
|
|
"loss": 0.1617,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15538963675498962,
|
||
|
|
"step": 2450,
|
||
|
|
"valid_targets_mean": 4966.9,
|
||
|
|
"valid_targets_min": 398
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.175170068027211,
|
||
|
|
"grad_norm": 0.45641378939370597,
|
||
|
|
"learning_rate": 1.6790915741073418e-05,
|
||
|
|
"loss": 0.1518,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1592978537082672,
|
||
|
|
"step": 2455,
|
||
|
|
"valid_targets_mean": 5755.1,
|
||
|
|
"valid_targets_min": 2484
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.183673469387755,
|
||
|
|
"grad_norm": 0.532160905088476,
|
||
|
|
"learning_rate": 1.670722757204758e-05,
|
||
|
|
"loss": 0.1582,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15307070314884186,
|
||
|
|
"step": 2460,
|
||
|
|
"valid_targets_mean": 4880.3,
|
||
|
|
"valid_targets_min": 791
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.192176870748299,
|
||
|
|
"grad_norm": 0.484272946777653,
|
||
|
|
"learning_rate": 1.6623598621720966e-05,
|
||
|
|
"loss": 0.1526,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16526508331298828,
|
||
|
|
"step": 2465,
|
||
|
|
"valid_targets_mean": 5719.8,
|
||
|
|
"valid_targets_min": 634
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.200680272108843,
|
||
|
|
"grad_norm": 0.5056415188335692,
|
||
|
|
"learning_rate": 1.6540030394114218e-05,
|
||
|
|
"loss": 0.1514,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13783684372901917,
|
||
|
|
"step": 2470,
|
||
|
|
"valid_targets_mean": 4630.0,
|
||
|
|
"valid_targets_min": 413
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.209183673469388,
|
||
|
|
"grad_norm": 0.45776442126978983,
|
||
|
|
"learning_rate": 1.64565243921559e-05,
|
||
|
|
"loss": 0.147,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15790414810180664,
|
||
|
|
"step": 2475,
|
||
|
|
"valid_targets_mean": 5720.7,
|
||
|
|
"valid_targets_min": 3696
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.217687074829932,
|
||
|
|
"grad_norm": 0.4619571995875309,
|
||
|
|
"learning_rate": 1.6373082117655473e-05,
|
||
|
|
"loss": 0.1539,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14637413620948792,
|
||
|
|
"step": 2480,
|
||
|
|
"valid_targets_mean": 5422.7,
|
||
|
|
"valid_targets_min": 1054
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.226190476190476,
|
||
|
|
"grad_norm": 0.475074179990662,
|
||
|
|
"learning_rate": 1.6289705071276323e-05,
|
||
|
|
"loss": 0.1487,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14552085101604462,
|
||
|
|
"step": 2485,
|
||
|
|
"valid_targets_mean": 5148.8,
|
||
|
|
"valid_targets_min": 633
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.23469387755102,
|
||
|
|
"grad_norm": 0.5128661520361073,
|
||
|
|
"learning_rate": 1.62063947525087e-05,
|
||
|
|
"loss": 0.1612,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15531602501869202,
|
||
|
|
"step": 2490,
|
||
|
|
"valid_targets_mean": 5132.4,
|
||
|
|
"valid_targets_min": 394
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.243197278911564,
|
||
|
|
"grad_norm": 0.47359274629055687,
|
||
|
|
"learning_rate": 1.612315265964284e-05,
|
||
|
|
"loss": 0.1522,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1517346203327179,
|
||
|
|
"step": 2495,
|
||
|
|
"valid_targets_mean": 5182.2,
|
||
|
|
"valid_targets_min": 746
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.2517006802721085,
|
||
|
|
"grad_norm": 0.5229889319203138,
|
||
|
|
"learning_rate": 1.6039980289741935e-05,
|
||
|
|
"loss": 0.1577,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1641138643026352,
|
||
|
|
"step": 2500,
|
||
|
|
"valid_targets_mean": 5705.8,
|
||
|
|
"valid_targets_min": 311
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.260204081632653,
|
||
|
|
"grad_norm": 0.49518033491896063,
|
||
|
|
"learning_rate": 1.595687913861528e-05,
|
||
|
|
"loss": 0.1547,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1696288287639618,
|
||
|
|
"step": 2505,
|
||
|
|
"valid_targets_mean": 4899.6,
|
||
|
|
"valid_targets_min": 423
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.2687074829931975,
|
||
|
|
"grad_norm": 0.5342932345061655,
|
||
|
|
"learning_rate": 1.587385070079133e-05,
|
||
|
|
"loss": 0.149,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1331261396408081,
|
||
|
|
"step": 2510,
|
||
|
|
"valid_targets_mean": 5110.5,
|
||
|
|
"valid_targets_min": 769
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.2772108843537415,
|
||
|
|
"grad_norm": 0.5062564145960146,
|
||
|
|
"learning_rate": 1.5790896469490807e-05,
|
||
|
|
"loss": 0.1495,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.149044007062912,
|
||
|
|
"step": 2515,
|
||
|
|
"valid_targets_mean": 4992.7,
|
||
|
|
"valid_targets_min": 337
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.285714285714286,
|
||
|
|
"grad_norm": 0.4274109284832628,
|
||
|
|
"learning_rate": 1.5708017936599908e-05,
|
||
|
|
"loss": 0.1538,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14069831371307373,
|
||
|
|
"step": 2520,
|
||
|
|
"valid_targets_mean": 6227.8,
|
||
|
|
"valid_targets_min": 1883
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.29421768707483,
|
||
|
|
"grad_norm": 0.46912555713175375,
|
||
|
|
"learning_rate": 1.56252165926434e-05,
|
||
|
|
"loss": 0.1511,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14161109924316406,
|
||
|
|
"step": 2525,
|
||
|
|
"valid_targets_mean": 5531.4,
|
||
|
|
"valid_targets_min": 3408
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.302721088435375,
|
||
|
|
"grad_norm": 0.4865811219933871,
|
||
|
|
"learning_rate": 1.554249392675788e-05,
|
||
|
|
"loss": 0.1557,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15894815325737,
|
||
|
|
"step": 2530,
|
||
|
|
"valid_targets_mean": 4926.1,
|
||
|
|
"valid_targets_min": 422
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.311224489795919,
|
||
|
|
"grad_norm": 0.5048170804377767,
|
||
|
|
"learning_rate": 1.5459851426664956e-05,
|
||
|
|
"loss": 0.149,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1395973414182663,
|
||
|
|
"step": 2535,
|
||
|
|
"valid_targets_mean": 4769.4,
|
||
|
|
"valid_targets_min": 347
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.319727891156463,
|
||
|
|
"grad_norm": 0.4424357727794296,
|
||
|
|
"learning_rate": 1.5377290578644468e-05,
|
||
|
|
"loss": 0.1625,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16818508505821228,
|
||
|
|
"step": 2540,
|
||
|
|
"valid_targets_mean": 6576.6,
|
||
|
|
"valid_targets_min": 4533
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.328231292517007,
|
||
|
|
"grad_norm": 0.5030992297531525,
|
||
|
|
"learning_rate": 1.5294812867507835e-05,
|
||
|
|
"loss": 0.1479,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14952751994132996,
|
||
|
|
"step": 2545,
|
||
|
|
"valid_targets_mean": 5295.3,
|
||
|
|
"valid_targets_min": 366
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.336734693877551,
|
||
|
|
"grad_norm": 0.5219093775587884,
|
||
|
|
"learning_rate": 1.5212419776571267e-05,
|
||
|
|
"loss": 0.1501,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14514940977096558,
|
||
|
|
"step": 2550,
|
||
|
|
"valid_targets_mean": 4355.4,
|
||
|
|
"valid_targets_min": 749
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.345238095238095,
|
||
|
|
"grad_norm": 0.4753399327443524,
|
||
|
|
"learning_rate": 1.5130112787629152e-05,
|
||
|
|
"loss": 0.16,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16155847907066345,
|
||
|
|
"step": 2555,
|
||
|
|
"valid_targets_mean": 5289.1,
|
||
|
|
"valid_targets_min": 515
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.35374149659864,
|
||
|
|
"grad_norm": 0.4534672767409702,
|
||
|
|
"learning_rate": 1.5047893380927377e-05,
|
||
|
|
"loss": 0.1424,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.147069051861763,
|
||
|
|
"step": 2560,
|
||
|
|
"valid_targets_mean": 5306.6,
|
||
|
|
"valid_targets_min": 693
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.362244897959184,
|
||
|
|
"grad_norm": 0.976161030429131,
|
||
|
|
"learning_rate": 1.4965763035136687e-05,
|
||
|
|
"loss": 0.1474,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1554407924413681,
|
||
|
|
"step": 2565,
|
||
|
|
"valid_targets_mean": 4044.1,
|
||
|
|
"valid_targets_min": 498
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.370748299319728,
|
||
|
|
"grad_norm": 0.5320170463003601,
|
||
|
|
"learning_rate": 1.4883723227326144e-05,
|
||
|
|
"loss": 0.1553,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17096030712127686,
|
||
|
|
"step": 2570,
|
||
|
|
"valid_targets_mean": 4863.4,
|
||
|
|
"valid_targets_min": 538
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.379251700680272,
|
||
|
|
"grad_norm": 0.530531269711091,
|
||
|
|
"learning_rate": 1.4801775432936516e-05,
|
||
|
|
"loss": 0.1502,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13900458812713623,
|
||
|
|
"step": 2575,
|
||
|
|
"valid_targets_mean": 4932.1,
|
||
|
|
"valid_targets_min": 970
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.387755102040816,
|
||
|
|
"grad_norm": 0.4714210045910267,
|
||
|
|
"learning_rate": 1.4719921125753773e-05,
|
||
|
|
"loss": 0.1547,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15984129905700684,
|
||
|
|
"step": 2580,
|
||
|
|
"valid_targets_mean": 6184.6,
|
||
|
|
"valid_targets_min": 504
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.39625850340136,
|
||
|
|
"grad_norm": 0.4686231350838581,
|
||
|
|
"learning_rate": 1.4638161777882573e-05,
|
||
|
|
"loss": 0.1419,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13544218242168427,
|
||
|
|
"step": 2585,
|
||
|
|
"valid_targets_mean": 4723.6,
|
||
|
|
"valid_targets_min": 2670
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.404761904761905,
|
||
|
|
"grad_norm": 0.461580824294726,
|
||
|
|
"learning_rate": 1.4556498859719756e-05,
|
||
|
|
"loss": 0.1592,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15786665678024292,
|
||
|
|
"step": 2590,
|
||
|
|
"valid_targets_mean": 5468.1,
|
||
|
|
"valid_targets_min": 307
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.413265306122449,
|
||
|
|
"grad_norm": 0.5091179099294835,
|
||
|
|
"learning_rate": 1.4474933839927956e-05,
|
||
|
|
"loss": 0.1578,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16065721213817596,
|
||
|
|
"step": 2595,
|
||
|
|
"valid_targets_mean": 4445.5,
|
||
|
|
"valid_targets_min": 477
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.421768707482993,
|
||
|
|
"grad_norm": 0.5524920434835537,
|
||
|
|
"learning_rate": 1.4393468185409139e-05,
|
||
|
|
"loss": 0.15,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15395322442054749,
|
||
|
|
"step": 2600,
|
||
|
|
"valid_targets_mean": 3875.0,
|
||
|
|
"valid_targets_min": 324
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.430272108843537,
|
||
|
|
"grad_norm": 0.5294500377375655,
|
||
|
|
"learning_rate": 1.4312103361278254e-05,
|
||
|
|
"loss": 0.162,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1423143893480301,
|
||
|
|
"step": 2605,
|
||
|
|
"valid_targets_mean": 4127.9,
|
||
|
|
"valid_targets_min": 369
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.438775510204081,
|
||
|
|
"grad_norm": 0.5321239023601794,
|
||
|
|
"learning_rate": 1.4230840830836875e-05,
|
||
|
|
"loss": 0.1458,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15434446930885315,
|
||
|
|
"step": 2610,
|
||
|
|
"valid_targets_mean": 5638.4,
|
||
|
|
"valid_targets_min": 496
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.447278911564625,
|
||
|
|
"grad_norm": 0.4415839977903979,
|
||
|
|
"learning_rate": 1.4149682055546854e-05,
|
||
|
|
"loss": 0.1457,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13892988860607147,
|
||
|
|
"step": 2615,
|
||
|
|
"valid_targets_mean": 5697.5,
|
||
|
|
"valid_targets_min": 2358
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.45578231292517,
|
||
|
|
"grad_norm": 0.4833091388897899,
|
||
|
|
"learning_rate": 1.4068628495004095e-05,
|
||
|
|
"loss": 0.1512,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14210116863250732,
|
||
|
|
"step": 2620,
|
||
|
|
"valid_targets_mean": 5138.7,
|
||
|
|
"valid_targets_min": 527
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.464285714285714,
|
||
|
|
"grad_norm": 1.6240477567407294,
|
||
|
|
"learning_rate": 1.3987681606912254e-05,
|
||
|
|
"loss": 0.1473,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17049244046211243,
|
||
|
|
"step": 2625,
|
||
|
|
"valid_targets_mean": 4175.8,
|
||
|
|
"valid_targets_min": 337
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.4727891156462585,
|
||
|
|
"grad_norm": 0.5629947877878578,
|
||
|
|
"learning_rate": 1.3906842847056546e-05,
|
||
|
|
"loss": 0.1589,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16542261838912964,
|
||
|
|
"step": 2630,
|
||
|
|
"valid_targets_mean": 3810.1,
|
||
|
|
"valid_targets_min": 408
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.4812925170068025,
|
||
|
|
"grad_norm": 0.5353633665528811,
|
||
|
|
"learning_rate": 1.382611366927757e-05,
|
||
|
|
"loss": 0.1416,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15019583702087402,
|
||
|
|
"step": 2635,
|
||
|
|
"valid_targets_mean": 4086.1,
|
||
|
|
"valid_targets_min": 358
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.489795918367347,
|
||
|
|
"grad_norm": 0.474652025790846,
|
||
|
|
"learning_rate": 1.3745495525445126e-05,
|
||
|
|
"loss": 0.1588,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16043367981910706,
|
||
|
|
"step": 2640,
|
||
|
|
"valid_targets_mean": 5114.6,
|
||
|
|
"valid_targets_min": 499
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.4982993197278915,
|
||
|
|
"grad_norm": 0.4950874177981697,
|
||
|
|
"learning_rate": 1.3664989865432157e-05,
|
||
|
|
"loss": 0.1613,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1527048796415329,
|
||
|
|
"step": 2645,
|
||
|
|
"valid_targets_mean": 5332.1,
|
||
|
|
"valid_targets_min": 617
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.506802721088436,
|
||
|
|
"grad_norm": 0.4839071438386304,
|
||
|
|
"learning_rate": 1.3584598137088625e-05,
|
||
|
|
"loss": 0.141,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.125609889626503,
|
||
|
|
"step": 2650,
|
||
|
|
"valid_targets_mean": 4625.4,
|
||
|
|
"valid_targets_min": 444
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.51530612244898,
|
||
|
|
"grad_norm": 0.44719699241656785,
|
||
|
|
"learning_rate": 1.3504321786215514e-05,
|
||
|
|
"loss": 0.1415,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13784818351268768,
|
||
|
|
"step": 2655,
|
||
|
|
"valid_targets_mean": 6044.3,
|
||
|
|
"valid_targets_min": 2822
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.523809523809524,
|
||
|
|
"grad_norm": 0.47246871896701914,
|
||
|
|
"learning_rate": 1.3424162256538797e-05,
|
||
|
|
"loss": 0.1517,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15130752325057983,
|
||
|
|
"step": 2660,
|
||
|
|
"valid_targets_mean": 5843.5,
|
||
|
|
"valid_targets_min": 705
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.532312925170068,
|
||
|
|
"grad_norm": 0.551839525748048,
|
||
|
|
"learning_rate": 1.3344120989683473e-05,
|
||
|
|
"loss": 0.16,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17199057340621948,
|
||
|
|
"step": 2665,
|
||
|
|
"valid_targets_mean": 3781.0,
|
||
|
|
"valid_targets_min": 447
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.540816326530612,
|
||
|
|
"grad_norm": 0.5290089694264519,
|
||
|
|
"learning_rate": 1.3264199425147667e-05,
|
||
|
|
"loss": 0.1701,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14920049905776978,
|
||
|
|
"step": 2670,
|
||
|
|
"valid_targets_mean": 4531.3,
|
||
|
|
"valid_targets_min": 293
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.549319727891157,
|
||
|
|
"grad_norm": 0.5006472169583304,
|
||
|
|
"learning_rate": 1.3184399000276708e-05,
|
||
|
|
"loss": 0.1551,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1673208475112915,
|
||
|
|
"step": 2675,
|
||
|
|
"valid_targets_mean": 4833.3,
|
||
|
|
"valid_targets_min": 321
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.557823129251701,
|
||
|
|
"grad_norm": 0.49417722534201963,
|
||
|
|
"learning_rate": 1.3104721150237305e-05,
|
||
|
|
"loss": 0.1518,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16881683468818665,
|
||
|
|
"step": 2680,
|
||
|
|
"valid_targets_mean": 5458.8,
|
||
|
|
"valid_targets_min": 412
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.566326530612245,
|
||
|
|
"grad_norm": 0.464241599389466,
|
||
|
|
"learning_rate": 1.3025167307991734e-05,
|
||
|
|
"loss": 0.1485,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1441410481929779,
|
||
|
|
"step": 2685,
|
||
|
|
"valid_targets_mean": 5621.9,
|
||
|
|
"valid_targets_min": 351
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.574829931972789,
|
||
|
|
"grad_norm": 0.4897483165209782,
|
||
|
|
"learning_rate": 1.294573890427203e-05,
|
||
|
|
"loss": 0.1498,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12957708537578583,
|
||
|
|
"step": 2690,
|
||
|
|
"valid_targets_mean": 4299.4,
|
||
|
|
"valid_targets_min": 377
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.583333333333333,
|
||
|
|
"grad_norm": 0.4848260598365829,
|
||
|
|
"learning_rate": 1.2866437367554313e-05,
|
||
|
|
"loss": 0.1526,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13792306184768677,
|
||
|
|
"step": 2695,
|
||
|
|
"valid_targets_mean": 5020.5,
|
||
|
|
"valid_targets_min": 349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.591836734693878,
|
||
|
|
"grad_norm": 0.46310220634278043,
|
||
|
|
"learning_rate": 1.278726412403306e-05,
|
||
|
|
"loss": 0.1487,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1504206657409668,
|
||
|
|
"step": 2700,
|
||
|
|
"valid_targets_mean": 4835.1,
|
||
|
|
"valid_targets_min": 263
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.600340136054422,
|
||
|
|
"grad_norm": 0.48946489868653864,
|
||
|
|
"learning_rate": 1.2708220597595462e-05,
|
||
|
|
"loss": 0.1496,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16788050532341003,
|
||
|
|
"step": 2705,
|
||
|
|
"valid_targets_mean": 5502.5,
|
||
|
|
"valid_targets_min": 1903
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.608843537414966,
|
||
|
|
"grad_norm": 0.4854888998352304,
|
||
|
|
"learning_rate": 1.2629308209795834e-05,
|
||
|
|
"loss": 0.1508,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14552603662014008,
|
||
|
|
"step": 2710,
|
||
|
|
"valid_targets_mean": 5004.9,
|
||
|
|
"valid_targets_min": 471
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.61734693877551,
|
||
|
|
"grad_norm": 0.5615487814060942,
|
||
|
|
"learning_rate": 1.2550528379830019e-05,
|
||
|
|
"loss": 0.1498,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15200123190879822,
|
||
|
|
"step": 2715,
|
||
|
|
"valid_targets_mean": 4244.0,
|
||
|
|
"valid_targets_min": 368
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.625850340136054,
|
||
|
|
"grad_norm": 0.42362377467809786,
|
||
|
|
"learning_rate": 1.2471882524509873e-05,
|
||
|
|
"loss": 0.1573,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1435042917728424,
|
||
|
|
"step": 2720,
|
||
|
|
"valid_targets_mean": 5792.3,
|
||
|
|
"valid_targets_min": 1685
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.634353741496598,
|
||
|
|
"grad_norm": 0.42097430140148373,
|
||
|
|
"learning_rate": 1.2393372058237819e-05,
|
||
|
|
"loss": 0.1497,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14636842906475067,
|
||
|
|
"step": 2725,
|
||
|
|
"valid_targets_mean": 6146.5,
|
||
|
|
"valid_targets_min": 1577
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.642857142857143,
|
||
|
|
"grad_norm": 0.4725687083845275,
|
||
|
|
"learning_rate": 1.2314998392981364e-05,
|
||
|
|
"loss": 0.143,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13137871026992798,
|
||
|
|
"step": 2730,
|
||
|
|
"valid_targets_mean": 5022.7,
|
||
|
|
"valid_targets_min": 335
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.651360544217687,
|
||
|
|
"grad_norm": 0.5219126088318514,
|
||
|
|
"learning_rate": 1.2236762938247729e-05,
|
||
|
|
"loss": 0.1471,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15758682787418365,
|
||
|
|
"step": 2735,
|
||
|
|
"valid_targets_mean": 5415.1,
|
||
|
|
"valid_targets_min": 367
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.659863945578231,
|
||
|
|
"grad_norm": 0.506196633401625,
|
||
|
|
"learning_rate": 1.2158667101058497e-05,
|
||
|
|
"loss": 0.1556,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16656053066253662,
|
||
|
|
"step": 2740,
|
||
|
|
"valid_targets_mean": 4654.1,
|
||
|
|
"valid_targets_min": 381
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.668367346938775,
|
||
|
|
"grad_norm": 0.47252770675122696,
|
||
|
|
"learning_rate": 1.208071228592429e-05,
|
||
|
|
"loss": 0.153,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14869409799575806,
|
||
|
|
"step": 2745,
|
||
|
|
"valid_targets_mean": 5609.6,
|
||
|
|
"valid_targets_min": 720
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.6768707482993195,
|
||
|
|
"grad_norm": 0.43344108682478294,
|
||
|
|
"learning_rate": 1.2002899894819548e-05,
|
||
|
|
"loss": 0.1449,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15214882791042328,
|
||
|
|
"step": 2750,
|
||
|
|
"valid_targets_mean": 6224.9,
|
||
|
|
"valid_targets_min": 1733
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.685374149659864,
|
||
|
|
"grad_norm": 0.4672621908455977,
|
||
|
|
"learning_rate": 1.1925231327157288e-05,
|
||
|
|
"loss": 0.1445,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15910692512989044,
|
||
|
|
"step": 2755,
|
||
|
|
"valid_targets_mean": 5304.6,
|
||
|
|
"valid_targets_min": 2086
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.6938775510204085,
|
||
|
|
"grad_norm": 0.4737806694964509,
|
||
|
|
"learning_rate": 1.1847707979763934e-05,
|
||
|
|
"loss": 0.1506,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15179231762886047,
|
||
|
|
"step": 2760,
|
||
|
|
"valid_targets_mean": 5679.9,
|
||
|
|
"valid_targets_min": 798
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.7023809523809526,
|
||
|
|
"grad_norm": 0.4894878288475145,
|
||
|
|
"learning_rate": 1.1770331246854211e-05,
|
||
|
|
"loss": 0.1539,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13271725177764893,
|
||
|
|
"step": 2765,
|
||
|
|
"valid_targets_mean": 4533.1,
|
||
|
|
"valid_targets_min": 307
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.710884353741497,
|
||
|
|
"grad_norm": 0.4909434159407714,
|
||
|
|
"learning_rate": 1.169310252000605e-05,
|
||
|
|
"loss": 0.1555,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16060566902160645,
|
||
|
|
"step": 2770,
|
||
|
|
"valid_targets_mean": 4942.4,
|
||
|
|
"valid_targets_min": 684
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.719387755102041,
|
||
|
|
"grad_norm": 0.6709218549373176,
|
||
|
|
"learning_rate": 1.1616023188135594e-05,
|
||
|
|
"loss": 0.1587,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1664205640554428,
|
||
|
|
"step": 2775,
|
||
|
|
"valid_targets_mean": 4404.5,
|
||
|
|
"valid_targets_min": 601
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.727891156462585,
|
||
|
|
"grad_norm": 0.43328533481769743,
|
||
|
|
"learning_rate": 1.1539094637472181e-05,
|
||
|
|
"loss": 0.1435,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12563464045524597,
|
||
|
|
"step": 2780,
|
||
|
|
"valid_targets_mean": 5376.3,
|
||
|
|
"valid_targets_min": 525
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.736394557823129,
|
||
|
|
"grad_norm": 0.4355550307068364,
|
||
|
|
"learning_rate": 1.1462318251533442e-05,
|
||
|
|
"loss": 0.1491,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13416925072669983,
|
||
|
|
"step": 2785,
|
||
|
|
"valid_targets_mean": 5780.0,
|
||
|
|
"valid_targets_min": 950
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.744897959183674,
|
||
|
|
"grad_norm": 0.5150641562490116,
|
||
|
|
"learning_rate": 1.1385695411100424e-05,
|
||
|
|
"loss": 0.1621,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16665419936180115,
|
||
|
|
"step": 2790,
|
||
|
|
"valid_targets_mean": 4723.9,
|
||
|
|
"valid_targets_min": 324
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.753401360544218,
|
||
|
|
"grad_norm": 0.4394685181595614,
|
||
|
|
"learning_rate": 1.1309227494192703e-05,
|
||
|
|
"loss": 0.1568,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15591369569301605,
|
||
|
|
"step": 2795,
|
||
|
|
"valid_targets_mean": 5538.1,
|
||
|
|
"valid_targets_min": 887
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.761904761904762,
|
||
|
|
"grad_norm": 0.4702929825968214,
|
||
|
|
"learning_rate": 1.1232915876043673e-05,
|
||
|
|
"loss": 0.1608,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14338061213493347,
|
||
|
|
"step": 2800,
|
||
|
|
"valid_targets_mean": 4658.1,
|
||
|
|
"valid_targets_min": 341
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.770408163265306,
|
||
|
|
"grad_norm": 0.4878751152682348,
|
||
|
|
"learning_rate": 1.1156761929075777e-05,
|
||
|
|
"loss": 0.1529,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15373317897319794,
|
||
|
|
"step": 2805,
|
||
|
|
"valid_targets_mean": 5382.1,
|
||
|
|
"valid_targets_min": 993
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.77891156462585,
|
||
|
|
"grad_norm": 0.46506323414525,
|
||
|
|
"learning_rate": 1.1080767022875816e-05,
|
||
|
|
"loss": 0.1606,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17930753529071808,
|
||
|
|
"step": 2810,
|
||
|
|
"valid_targets_mean": 5957.2,
|
||
|
|
"valid_targets_min": 2612
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.787414965986395,
|
||
|
|
"grad_norm": 0.46335099362541593,
|
||
|
|
"learning_rate": 1.100493252417035e-05,
|
||
|
|
"loss": 0.1566,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13600006699562073,
|
||
|
|
"step": 2815,
|
||
|
|
"valid_targets_mean": 5025.2,
|
||
|
|
"valid_targets_min": 303
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.795918367346939,
|
||
|
|
"grad_norm": 0.5157629324536205,
|
||
|
|
"learning_rate": 1.0929259796801075e-05,
|
||
|
|
"loss": 0.154,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16718797385692596,
|
||
|
|
"step": 2820,
|
||
|
|
"valid_targets_mean": 4806.3,
|
||
|
|
"valid_targets_min": 485
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.804421768707483,
|
||
|
|
"grad_norm": 0.465742926373782,
|
||
|
|
"learning_rate": 1.085375020170034e-05,
|
||
|
|
"loss": 0.1672,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16306686401367188,
|
||
|
|
"step": 2825,
|
||
|
|
"valid_targets_mean": 5816.1,
|
||
|
|
"valid_targets_min": 2782
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.812925170068027,
|
||
|
|
"grad_norm": 0.5222784528768896,
|
||
|
|
"learning_rate": 1.0778405096866624e-05,
|
||
|
|
"loss": 0.1633,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17221535742282867,
|
||
|
|
"step": 2830,
|
||
|
|
"valid_targets_mean": 5624.4,
|
||
|
|
"valid_targets_min": 3600
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.821428571428571,
|
||
|
|
"grad_norm": 0.44419388370269125,
|
||
|
|
"learning_rate": 1.0703225837340166e-05,
|
||
|
|
"loss": 0.1447,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1237671971321106,
|
||
|
|
"step": 2835,
|
||
|
|
"valid_targets_mean": 6006.3,
|
||
|
|
"valid_targets_min": 2523
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.829931972789115,
|
||
|
|
"grad_norm": 0.48399843783613755,
|
||
|
|
"learning_rate": 1.0628213775178561e-05,
|
||
|
|
"loss": 0.162,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17268820106983185,
|
||
|
|
"step": 2840,
|
||
|
|
"valid_targets_mean": 5256.6,
|
||
|
|
"valid_targets_min": 695
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.83843537414966,
|
||
|
|
"grad_norm": 0.5133202801662862,
|
||
|
|
"learning_rate": 1.055337025943244e-05,
|
||
|
|
"loss": 0.1453,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.155470073223114,
|
||
|
|
"step": 2845,
|
||
|
|
"valid_targets_mean": 4387.2,
|
||
|
|
"valid_targets_min": 356
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.846938775510204,
|
||
|
|
"grad_norm": 0.514603343004657,
|
||
|
|
"learning_rate": 1.0478696636121231e-05,
|
||
|
|
"loss": 0.1559,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13981027901172638,
|
||
|
|
"step": 2850,
|
||
|
|
"valid_targets_mean": 4391.6,
|
||
|
|
"valid_targets_min": 716
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.855442176870748,
|
||
|
|
"grad_norm": 0.5540512056030009,
|
||
|
|
"learning_rate": 1.0404194248208934e-05,
|
||
|
|
"loss": 0.1501,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16355203092098236,
|
||
|
|
"step": 2855,
|
||
|
|
"valid_targets_mean": 4231.9,
|
||
|
|
"valid_targets_min": 279
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.863945578231292,
|
||
|
|
"grad_norm": 0.4459196217546899,
|
||
|
|
"learning_rate": 1.0329864435579982e-05,
|
||
|
|
"loss": 0.1508,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12583604454994202,
|
||
|
|
"step": 2860,
|
||
|
|
"valid_targets_mean": 5498.4,
|
||
|
|
"valid_targets_min": 806
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.872448979591836,
|
||
|
|
"grad_norm": 0.515575755806559,
|
||
|
|
"learning_rate": 1.0255708535015137e-05,
|
||
|
|
"loss": 0.1579,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1577465981245041,
|
||
|
|
"step": 2865,
|
||
|
|
"valid_targets_mean": 4604.1,
|
||
|
|
"valid_targets_min": 318
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.880952380952381,
|
||
|
|
"grad_norm": 0.5341113668588352,
|
||
|
|
"learning_rate": 1.0181727880167428e-05,
|
||
|
|
"loss": 0.1584,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1562502086162567,
|
||
|
|
"step": 2870,
|
||
|
|
"valid_targets_mean": 4223.8,
|
||
|
|
"valid_targets_min": 307
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.889455782312925,
|
||
|
|
"grad_norm": 0.5463005421899149,
|
||
|
|
"learning_rate": 1.0107923801538215e-05,
|
||
|
|
"loss": 0.1645,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14711858332157135,
|
||
|
|
"step": 2875,
|
||
|
|
"valid_targets_mean": 4793.5,
|
||
|
|
"valid_targets_min": 427
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.8979591836734695,
|
||
|
|
"grad_norm": 0.49522933093311056,
|
||
|
|
"learning_rate": 1.003429762645321e-05,
|
||
|
|
"loss": 0.1467,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14953123033046722,
|
||
|
|
"step": 2880,
|
||
|
|
"valid_targets_mean": 5598.2,
|
||
|
|
"valid_targets_min": 761
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.906462585034014,
|
||
|
|
"grad_norm": 0.5120400003927019,
|
||
|
|
"learning_rate": 9.960850679038632e-06,
|
||
|
|
"loss": 0.1502,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1599283367395401,
|
||
|
|
"step": 2885,
|
||
|
|
"valid_targets_mean": 4908.4,
|
||
|
|
"valid_targets_min": 746
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.914965986394558,
|
||
|
|
"grad_norm": 0.543233892077237,
|
||
|
|
"learning_rate": 9.887584280197411e-06,
|
||
|
|
"loss": 0.1417,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13952985405921936,
|
||
|
|
"step": 2890,
|
||
|
|
"valid_targets_mean": 4356.1,
|
||
|
|
"valid_targets_min": 353
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.923469387755102,
|
||
|
|
"grad_norm": 0.5297374158037707,
|
||
|
|
"learning_rate": 9.814499747585373e-06,
|
||
|
|
"loss": 0.1519,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15328747034072876,
|
||
|
|
"step": 2895,
|
||
|
|
"valid_targets_mean": 5036.3,
|
||
|
|
"valid_targets_min": 2399
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.931972789115647,
|
||
|
|
"grad_norm": 0.4970548413757221,
|
||
|
|
"learning_rate": 9.741598395587606e-06,
|
||
|
|
"loss": 0.1482,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15605121850967407,
|
||
|
|
"step": 2900,
|
||
|
|
"valid_targets_mean": 5420.9,
|
||
|
|
"valid_targets_min": 811
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.940476190476191,
|
||
|
|
"grad_norm": 0.4767488208993672,
|
||
|
|
"learning_rate": 9.668881535294792e-06,
|
||
|
|
"loss": 0.1367,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12133751809597015,
|
||
|
|
"step": 2905,
|
||
|
|
"valid_targets_mean": 4799.2,
|
||
|
|
"valid_targets_min": 896
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.948979591836735,
|
||
|
|
"grad_norm": 0.45922070374085383,
|
||
|
|
"learning_rate": 9.596350474479625e-06,
|
||
|
|
"loss": 0.1457,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14983788132667542,
|
||
|
|
"step": 2910,
|
||
|
|
"valid_targets_mean": 5252.6,
|
||
|
|
"valid_targets_min": 379
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.957482993197279,
|
||
|
|
"grad_norm": 0.4772926326192319,
|
||
|
|
"learning_rate": 9.52400651757331e-06,
|
||
|
|
"loss": 0.1657,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18660974502563477,
|
||
|
|
"step": 2915,
|
||
|
|
"valid_targets_mean": 5886.1,
|
||
|
|
"valid_targets_min": 1548
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.965986394557823,
|
||
|
|
"grad_norm": 0.5038182578753977,
|
||
|
|
"learning_rate": 9.451850965642073e-06,
|
||
|
|
"loss": 0.1552,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16633275151252747,
|
||
|
|
"step": 2920,
|
||
|
|
"valid_targets_mean": 4575.6,
|
||
|
|
"valid_targets_min": 278
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.974489795918368,
|
||
|
|
"grad_norm": 0.5494432401778041,
|
||
|
|
"learning_rate": 9.379885116363786e-06,
|
||
|
|
"loss": 0.1596,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1750616729259491,
|
||
|
|
"step": 2925,
|
||
|
|
"valid_targets_mean": 4728.2,
|
||
|
|
"valid_targets_min": 697
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.982993197278912,
|
||
|
|
"grad_norm": 0.5112582591633633,
|
||
|
|
"learning_rate": 9.308110264004625e-06,
|
||
|
|
"loss": 0.1629,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14117315411567688,
|
||
|
|
"step": 2930,
|
||
|
|
"valid_targets_mean": 4149.5,
|
||
|
|
"valid_targets_min": 290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.991496598639456,
|
||
|
|
"grad_norm": 0.47004392514505877,
|
||
|
|
"learning_rate": 9.236527699395803e-06,
|
||
|
|
"loss": 0.1596,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19359202682971954,
|
||
|
|
"step": 2935,
|
||
|
|
"valid_targets_mean": 5871.9,
|
||
|
|
"valid_targets_min": 2868
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0,
|
||
|
|
"grad_norm": 0.47376830034428064,
|
||
|
|
"learning_rate": 9.165138709910338e-06,
|
||
|
|
"loss": 0.1441,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12866127490997314,
|
||
|
|
"step": 2940,
|
||
|
|
"valid_targets_mean": 4660.9,
|
||
|
|
"valid_targets_min": 357
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.008503401360544,
|
||
|
|
"grad_norm": 0.45902527541249266,
|
||
|
|
"learning_rate": 9.093944579439882e-06,
|
||
|
|
"loss": 0.1369,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12167897820472717,
|
||
|
|
"step": 2945,
|
||
|
|
"valid_targets_mean": 5229.8,
|
||
|
|
"valid_targets_min": 472
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.017006802721088,
|
||
|
|
"grad_norm": 0.5174431151677498,
|
||
|
|
"learning_rate": 9.022946588371688e-06,
|
||
|
|
"loss": 0.1489,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13627609610557556,
|
||
|
|
"step": 2950,
|
||
|
|
"valid_targets_mean": 4858.8,
|
||
|
|
"valid_targets_min": 260
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.025510204081633,
|
||
|
|
"grad_norm": 0.48061429549970536,
|
||
|
|
"learning_rate": 8.952146013565532e-06,
|
||
|
|
"loss": 0.1398,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13307006657123566,
|
||
|
|
"step": 2955,
|
||
|
|
"valid_targets_mean": 5370.2,
|
||
|
|
"valid_targets_min": 278
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.034013605442177,
|
||
|
|
"grad_norm": 0.5187416550663642,
|
||
|
|
"learning_rate": 8.881544128330777e-06,
|
||
|
|
"loss": 0.143,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1468641310930252,
|
||
|
|
"step": 2960,
|
||
|
|
"valid_targets_mean": 4802.4,
|
||
|
|
"valid_targets_min": 437
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.042517006802721,
|
||
|
|
"grad_norm": 0.4992347394577296,
|
||
|
|
"learning_rate": 8.81114220240346e-06,
|
||
|
|
"loss": 0.1506,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14591436088085175,
|
||
|
|
"step": 2965,
|
||
|
|
"valid_targets_mean": 4722.5,
|
||
|
|
"valid_targets_min": 395
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.051020408163265,
|
||
|
|
"grad_norm": 0.5270256740976864,
|
||
|
|
"learning_rate": 8.740941501923454e-06,
|
||
|
|
"loss": 0.1354,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12389400601387024,
|
||
|
|
"step": 2970,
|
||
|
|
"valid_targets_mean": 4252.3,
|
||
|
|
"valid_targets_min": 307
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.059523809523809,
|
||
|
|
"grad_norm": 0.4595245232520078,
|
||
|
|
"learning_rate": 8.6709432894117e-06,
|
||
|
|
"loss": 0.1436,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1578424721956253,
|
||
|
|
"step": 2975,
|
||
|
|
"valid_targets_mean": 6190.2,
|
||
|
|
"valid_targets_min": 2400
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.068027210884353,
|
||
|
|
"grad_norm": 0.47842477391507077,
|
||
|
|
"learning_rate": 8.601148823747516e-06,
|
||
|
|
"loss": 0.1516,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13707157969474792,
|
||
|
|
"step": 2980,
|
||
|
|
"valid_targets_mean": 5153.8,
|
||
|
|
"valid_targets_min": 312
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.076530612244898,
|
||
|
|
"grad_norm": 0.5348760834743891,
|
||
|
|
"learning_rate": 8.531559360145942e-06,
|
||
|
|
"loss": 0.1571,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16174763441085815,
|
||
|
|
"step": 2985,
|
||
|
|
"valid_targets_mean": 4613.1,
|
||
|
|
"valid_targets_min": 469
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.085034013605442,
|
||
|
|
"grad_norm": 0.48225259051771846,
|
||
|
|
"learning_rate": 8.462176150135168e-06,
|
||
|
|
"loss": 0.1409,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1347518265247345,
|
||
|
|
"step": 2990,
|
||
|
|
"valid_targets_mean": 5236.9,
|
||
|
|
"valid_targets_min": 1020
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.093537414965986,
|
||
|
|
"grad_norm": 0.47397536574841354,
|
||
|
|
"learning_rate": 8.393000441534036e-06,
|
||
|
|
"loss": 0.1321,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1289183497428894,
|
||
|
|
"step": 2995,
|
||
|
|
"valid_targets_mean": 5292.2,
|
||
|
|
"valid_targets_min": 551
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.1020408163265305,
|
||
|
|
"grad_norm": 0.4773907855838482,
|
||
|
|
"learning_rate": 8.324033478429559e-06,
|
||
|
|
"loss": 0.1443,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1419016718864441,
|
||
|
|
"step": 3000,
|
||
|
|
"valid_targets_mean": 5277.8,
|
||
|
|
"valid_targets_min": 473
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.110544217687075,
|
||
|
|
"grad_norm": 0.5387689505630772,
|
||
|
|
"learning_rate": 8.25527650115461e-06,
|
||
|
|
"loss": 0.1493,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18222028017044067,
|
||
|
|
"step": 3005,
|
||
|
|
"valid_targets_mean": 5219.9,
|
||
|
|
"valid_targets_min": 362
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.119047619047619,
|
||
|
|
"grad_norm": 0.5453427377420597,
|
||
|
|
"learning_rate": 8.186730746265577e-06,
|
||
|
|
"loss": 0.1614,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18752656877040863,
|
||
|
|
"step": 3010,
|
||
|
|
"valid_targets_mean": 4983.4,
|
||
|
|
"valid_targets_min": 285
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.127551020408164,
|
||
|
|
"grad_norm": 0.5656859363443553,
|
||
|
|
"learning_rate": 8.118397446520103e-06,
|
||
|
|
"loss": 0.1422,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1395868957042694,
|
||
|
|
"step": 3015,
|
||
|
|
"valid_targets_mean": 4306.9,
|
||
|
|
"valid_targets_min": 257
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.136054421768708,
|
||
|
|
"grad_norm": 0.4468193657671686,
|
||
|
|
"learning_rate": 8.050277830854984e-06,
|
||
|
|
"loss": 0.149,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1499105989933014,
|
||
|
|
"step": 3020,
|
||
|
|
"valid_targets_mean": 6021.9,
|
||
|
|
"valid_targets_min": 581
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.144557823129252,
|
||
|
|
"grad_norm": 0.516813675205649,
|
||
|
|
"learning_rate": 7.982373124363984e-06,
|
||
|
|
"loss": 0.1421,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12954609096050262,
|
||
|
|
"step": 3025,
|
||
|
|
"valid_targets_mean": 4634.7,
|
||
|
|
"valid_targets_min": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.153061224489796,
|
||
|
|
"grad_norm": 0.5563720136759567,
|
||
|
|
"learning_rate": 7.914684548275875e-06,
|
||
|
|
"loss": 0.1446,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14864128828048706,
|
||
|
|
"step": 3030,
|
||
|
|
"valid_targets_mean": 4747.5,
|
||
|
|
"valid_targets_min": 287
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.16156462585034,
|
||
|
|
"grad_norm": 0.4500140519621652,
|
||
|
|
"learning_rate": 7.847213319932425e-06,
|
||
|
|
"loss": 0.1385,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12891381978988647,
|
||
|
|
"step": 3035,
|
||
|
|
"valid_targets_mean": 6201.4,
|
||
|
|
"valid_targets_min": 340
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.170068027210885,
|
||
|
|
"grad_norm": 0.4526679444742861,
|
||
|
|
"learning_rate": 7.779960652766534e-06,
|
||
|
|
"loss": 0.1395,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1276322454214096,
|
||
|
|
"step": 3040,
|
||
|
|
"valid_targets_mean": 6369.3,
|
||
|
|
"valid_targets_min": 4135
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.178571428571429,
|
||
|
|
"grad_norm": 0.5196825459277832,
|
||
|
|
"learning_rate": 7.7129277562804e-06,
|
||
|
|
"loss": 0.1458,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1399431824684143,
|
||
|
|
"step": 3045,
|
||
|
|
"valid_targets_mean": 5197.4,
|
||
|
|
"valid_targets_min": 2685
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.187074829931973,
|
||
|
|
"grad_norm": 0.4841467876229569,
|
||
|
|
"learning_rate": 7.646115836023744e-06,
|
||
|
|
"loss": 0.1484,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15069720149040222,
|
||
|
|
"step": 3050,
|
||
|
|
"valid_targets_mean": 5769.4,
|
||
|
|
"valid_targets_min": 311
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.195578231292517,
|
||
|
|
"grad_norm": 0.49543390868779,
|
||
|
|
"learning_rate": 7.5795260935721755e-06,
|
||
|
|
"loss": 0.1457,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16008317470550537,
|
||
|
|
"step": 3055,
|
||
|
|
"valid_targets_mean": 5052.4,
|
||
|
|
"valid_targets_min": 371
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.204081632653061,
|
||
|
|
"grad_norm": 0.5121538762422064,
|
||
|
|
"learning_rate": 7.5131597265055585e-06,
|
||
|
|
"loss": 0.1405,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1398521065711975,
|
||
|
|
"step": 3060,
|
||
|
|
"valid_targets_mean": 5110.4,
|
||
|
|
"valid_targets_min": 360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.212585034013605,
|
||
|
|
"grad_norm": 0.4899564989718241,
|
||
|
|
"learning_rate": 7.447017928386453e-06,
|
||
|
|
"loss": 0.1397,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1291321963071823,
|
||
|
|
"step": 3065,
|
||
|
|
"valid_targets_mean": 5174.0,
|
||
|
|
"valid_targets_min": 411
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.22108843537415,
|
||
|
|
"grad_norm": 0.5293211427074018,
|
||
|
|
"learning_rate": 7.381101888738693e-06,
|
||
|
|
"loss": 0.1452,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14836561679840088,
|
||
|
|
"step": 3070,
|
||
|
|
"valid_targets_mean": 5932.8,
|
||
|
|
"valid_targets_min": 610
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.229591836734694,
|
||
|
|
"grad_norm": 0.5094442827883986,
|
||
|
|
"learning_rate": 7.315412793025951e-06,
|
||
|
|
"loss": 0.1378,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14766162633895874,
|
||
|
|
"step": 3075,
|
||
|
|
"valid_targets_mean": 6110.9,
|
||
|
|
"valid_targets_min": 1440
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.238095238095238,
|
||
|
|
"grad_norm": 0.5464753831292435,
|
||
|
|
"learning_rate": 7.24995182263045e-06,
|
||
|
|
"loss": 0.1348,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1328192502260208,
|
||
|
|
"step": 3080,
|
||
|
|
"valid_targets_mean": 4637.1,
|
||
|
|
"valid_targets_min": 277
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.246598639455782,
|
||
|
|
"grad_norm": 0.4894888950055084,
|
||
|
|
"learning_rate": 7.184720154831706e-06,
|
||
|
|
"loss": 0.1529,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18481044471263885,
|
||
|
|
"step": 3085,
|
||
|
|
"valid_targets_mean": 5611.0,
|
||
|
|
"valid_targets_min": 1453
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.255102040816326,
|
||
|
|
"grad_norm": 0.5481582236297604,
|
||
|
|
"learning_rate": 7.119718962785349e-06,
|
||
|
|
"loss": 0.1378,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.146115243434906,
|
||
|
|
"step": 3090,
|
||
|
|
"valid_targets_mean": 4766.4,
|
||
|
|
"valid_targets_min": 280
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.263605442176871,
|
||
|
|
"grad_norm": 0.5381069244801238,
|
||
|
|
"learning_rate": 7.054949415502037e-06,
|
||
|
|
"loss": 0.1378,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1564604789018631,
|
||
|
|
"step": 3095,
|
||
|
|
"valid_targets_mean": 4250.7,
|
||
|
|
"valid_targets_min": 307
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.272108843537415,
|
||
|
|
"grad_norm": 0.45195076820007074,
|
||
|
|
"learning_rate": 6.990412677826404e-06,
|
||
|
|
"loss": 0.1491,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16626204550266266,
|
||
|
|
"step": 3100,
|
||
|
|
"valid_targets_mean": 6312.3,
|
||
|
|
"valid_targets_min": 2693
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.280612244897959,
|
||
|
|
"grad_norm": 0.478527567964568,
|
||
|
|
"learning_rate": 6.926109910416153e-06,
|
||
|
|
"loss": 0.1433,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15561316907405853,
|
||
|
|
"step": 3105,
|
||
|
|
"valid_targets_mean": 5479.2,
|
||
|
|
"valid_targets_min": 781
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.289115646258503,
|
||
|
|
"grad_norm": 0.5290621867563297,
|
||
|
|
"learning_rate": 6.862042269721158e-06,
|
||
|
|
"loss": 0.1419,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16806018352508545,
|
||
|
|
"step": 3110,
|
||
|
|
"valid_targets_mean": 4655.8,
|
||
|
|
"valid_targets_min": 283
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.2976190476190474,
|
||
|
|
"grad_norm": 0.6611755108699091,
|
||
|
|
"learning_rate": 6.798210907962655e-06,
|
||
|
|
"loss": 0.1455,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14334535598754883,
|
||
|
|
"step": 3115,
|
||
|
|
"valid_targets_mean": 5053.7,
|
||
|
|
"valid_targets_min": 367
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.3061224489795915,
|
||
|
|
"grad_norm": 0.49948323103410336,
|
||
|
|
"learning_rate": 6.734616973112549e-06,
|
||
|
|
"loss": 0.1347,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13360539078712463,
|
||
|
|
"step": 3120,
|
||
|
|
"valid_targets_mean": 6116.2,
|
||
|
|
"valid_targets_min": 3063
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.3146258503401365,
|
||
|
|
"grad_norm": 0.5007993595708619,
|
||
|
|
"learning_rate": 6.671261608872737e-06,
|
||
|
|
"loss": 0.1413,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16443252563476562,
|
||
|
|
"step": 3125,
|
||
|
|
"valid_targets_mean": 5418.4,
|
||
|
|
"valid_targets_min": 416
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.3231292517006805,
|
||
|
|
"grad_norm": 0.5495889068968108,
|
||
|
|
"learning_rate": 6.60814595465457e-06,
|
||
|
|
"loss": 0.1449,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1657184362411499,
|
||
|
|
"step": 3130,
|
||
|
|
"valid_targets_mean": 4555.6,
|
||
|
|
"valid_targets_min": 278
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.331632653061225,
|
||
|
|
"grad_norm": 0.5171114512673033,
|
||
|
|
"learning_rate": 6.545271145558336e-06,
|
||
|
|
"loss": 0.1484,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14910860359668732,
|
||
|
|
"step": 3135,
|
||
|
|
"valid_targets_mean": 5366.7,
|
||
|
|
"valid_targets_min": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.340136054421769,
|
||
|
|
"grad_norm": 0.5073726280162751,
|
||
|
|
"learning_rate": 6.482638312352867e-06,
|
||
|
|
"loss": 0.1394,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13826128840446472,
|
||
|
|
"step": 3140,
|
||
|
|
"valid_targets_mean": 5216.6,
|
||
|
|
"valid_targets_min": 441
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.348639455782313,
|
||
|
|
"grad_norm": 0.5102309431410341,
|
||
|
|
"learning_rate": 6.420248581455184e-06,
|
||
|
|
"loss": 0.1441,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13723532855510712,
|
||
|
|
"step": 3145,
|
||
|
|
"valid_targets_mean": 4826.5,
|
||
|
|
"valid_targets_min": 326
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.357142857142857,
|
||
|
|
"grad_norm": 0.5043030572769284,
|
||
|
|
"learning_rate": 6.358103074910238e-06,
|
||
|
|
"loss": 0.1351,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.131906658411026,
|
||
|
|
"step": 3150,
|
||
|
|
"valid_targets_mean": 4873.4,
|
||
|
|
"valid_targets_min": 400
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.365646258503402,
|
||
|
|
"grad_norm": 0.5160679487955135,
|
||
|
|
"learning_rate": 6.296202910370757e-06,
|
||
|
|
"loss": 0.1345,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14111362397670746,
|
||
|
|
"step": 3155,
|
||
|
|
"valid_targets_mean": 5654.6,
|
||
|
|
"valid_targets_min": 2862
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.374149659863946,
|
||
|
|
"grad_norm": 0.49803097869248497,
|
||
|
|
"learning_rate": 6.234549201077124e-06,
|
||
|
|
"loss": 0.1449,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13837489485740662,
|
||
|
|
"step": 3160,
|
||
|
|
"valid_targets_mean": 4965.4,
|
||
|
|
"valid_targets_min": 603
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.38265306122449,
|
||
|
|
"grad_norm": 0.43242884447337027,
|
||
|
|
"learning_rate": 6.173143055837345e-06,
|
||
|
|
"loss": 0.1427,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12490431219339371,
|
||
|
|
"step": 3165,
|
||
|
|
"valid_targets_mean": 5918.2,
|
||
|
|
"valid_targets_min": 501
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.391156462585034,
|
||
|
|
"grad_norm": 0.5811229387339615,
|
||
|
|
"learning_rate": 6.111985579007149e-06,
|
||
|
|
"loss": 0.1426,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12798233330249786,
|
||
|
|
"step": 3170,
|
||
|
|
"valid_targets_mean": 4023.1,
|
||
|
|
"valid_targets_min": 323
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.399659863945578,
|
||
|
|
"grad_norm": 0.5167158675333369,
|
||
|
|
"learning_rate": 6.051077870470068e-06,
|
||
|
|
"loss": 0.1391,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14932399988174438,
|
||
|
|
"step": 3175,
|
||
|
|
"valid_targets_mean": 4918.9,
|
||
|
|
"valid_targets_min": 366
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.408163265306122,
|
||
|
|
"grad_norm": 0.49396427007555976,
|
||
|
|
"learning_rate": 5.990421025617716e-06,
|
||
|
|
"loss": 0.1428,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15247943997383118,
|
||
|
|
"step": 3180,
|
||
|
|
"valid_targets_mean": 5532.8,
|
||
|
|
"valid_targets_min": 1821
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.416666666666667,
|
||
|
|
"grad_norm": 0.5162541753941031,
|
||
|
|
"learning_rate": 5.930016135330052e-06,
|
||
|
|
"loss": 0.1424,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1384769082069397,
|
||
|
|
"step": 3185,
|
||
|
|
"valid_targets_mean": 5053.3,
|
||
|
|
"valid_targets_min": 827
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.425170068027211,
|
||
|
|
"grad_norm": 0.5336292062218522,
|
||
|
|
"learning_rate": 5.869864285955771e-06,
|
||
|
|
"loss": 0.1456,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14821632206439972,
|
||
|
|
"step": 3190,
|
||
|
|
"valid_targets_mean": 5111.1,
|
||
|
|
"valid_targets_min": 341
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.433673469387755,
|
||
|
|
"grad_norm": 0.4766784558708454,
|
||
|
|
"learning_rate": 5.80996655929277e-06,
|
||
|
|
"loss": 0.1381,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12869378924369812,
|
||
|
|
"step": 3195,
|
||
|
|
"valid_targets_mean": 5456.2,
|
||
|
|
"valid_targets_min": 222
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.442176870748299,
|
||
|
|
"grad_norm": 0.514415295414767,
|
||
|
|
"learning_rate": 5.750324032568677e-06,
|
||
|
|
"loss": 0.141,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15982869267463684,
|
||
|
|
"step": 3200,
|
||
|
|
"valid_targets_mean": 5050.7,
|
||
|
|
"valid_targets_min": 2235
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.450680272108843,
|
||
|
|
"grad_norm": 0.49337149459706553,
|
||
|
|
"learning_rate": 5.6909377784215035e-06,
|
||
|
|
"loss": 0.1469,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12663033604621887,
|
||
|
|
"step": 3205,
|
||
|
|
"valid_targets_mean": 5384.1,
|
||
|
|
"valid_targets_min": 586
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.459183673469388,
|
||
|
|
"grad_norm": 0.48955384070368113,
|
||
|
|
"learning_rate": 5.631808864880342e-06,
|
||
|
|
"loss": 0.1399,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12822724878787994,
|
||
|
|
"step": 3210,
|
||
|
|
"valid_targets_mean": 4940.7,
|
||
|
|
"valid_targets_min": 380
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.467687074829932,
|
||
|
|
"grad_norm": 0.5083351008241171,
|
||
|
|
"learning_rate": 5.5729383553461315e-06,
|
||
|
|
"loss": 0.1374,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13713547587394714,
|
||
|
|
"step": 3215,
|
||
|
|
"valid_targets_mean": 4815.9,
|
||
|
|
"valid_targets_min": 406
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.476190476190476,
|
||
|
|
"grad_norm": 0.5261639760270334,
|
||
|
|
"learning_rate": 5.5143273085725894e-06,
|
||
|
|
"loss": 0.1279,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12472152709960938,
|
||
|
|
"step": 3220,
|
||
|
|
"valid_targets_mean": 4350.4,
|
||
|
|
"valid_targets_min": 394
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.48469387755102,
|
||
|
|
"grad_norm": 0.44781562645348955,
|
||
|
|
"learning_rate": 5.45597677864711e-06,
|
||
|
|
"loss": 0.14,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11936494708061218,
|
||
|
|
"step": 3225,
|
||
|
|
"valid_targets_mean": 5790.3,
|
||
|
|
"valid_targets_min": 2666
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.493197278911564,
|
||
|
|
"grad_norm": 0.480564343122821,
|
||
|
|
"learning_rate": 5.397887814971854e-06,
|
||
|
|
"loss": 0.1362,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13624978065490723,
|
||
|
|
"step": 3230,
|
||
|
|
"valid_targets_mean": 5409.4,
|
||
|
|
"valid_targets_min": 2343
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.5017006802721085,
|
||
|
|
"grad_norm": 0.5722822078183675,
|
||
|
|
"learning_rate": 5.34006146224485e-06,
|
||
|
|
"loss": 0.1531,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16155627369880676,
|
||
|
|
"step": 3235,
|
||
|
|
"valid_targets_mean": 4847.9,
|
||
|
|
"valid_targets_min": 307
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.510204081632653,
|
||
|
|
"grad_norm": 0.48281919394514866,
|
||
|
|
"learning_rate": 5.282498760441219e-06,
|
||
|
|
"loss": 0.1445,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14231206476688385,
|
||
|
|
"step": 3240,
|
||
|
|
"valid_targets_mean": 5389.1,
|
||
|
|
"valid_targets_min": 1741
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.5187074829931975,
|
||
|
|
"grad_norm": 0.4907195483558696,
|
||
|
|
"learning_rate": 5.225200744794465e-06,
|
||
|
|
"loss": 0.1404,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12120344489812851,
|
||
|
|
"step": 3245,
|
||
|
|
"valid_targets_mean": 5018.6,
|
||
|
|
"valid_targets_min": 1310
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.5272108843537415,
|
||
|
|
"grad_norm": 0.5013342819687768,
|
||
|
|
"learning_rate": 5.168168445777839e-06,
|
||
|
|
"loss": 0.1405,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13457141816616058,
|
||
|
|
"step": 3250,
|
||
|
|
"valid_targets_mean": 5196.7,
|
||
|
|
"valid_targets_min": 1146
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.535714285714286,
|
||
|
|
"grad_norm": 0.49250420826254065,
|
||
|
|
"learning_rate": 5.111402889085852e-06,
|
||
|
|
"loss": 0.1379,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.139724463224411,
|
||
|
|
"step": 3255,
|
||
|
|
"valid_targets_mean": 4751.3,
|
||
|
|
"valid_targets_min": 452
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.54421768707483,
|
||
|
|
"grad_norm": 0.47016772046712996,
|
||
|
|
"learning_rate": 5.054905095615779e-06,
|
||
|
|
"loss": 0.1375,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14522133767604828,
|
||
|
|
"step": 3260,
|
||
|
|
"valid_targets_mean": 5891.9,
|
||
|
|
"valid_targets_min": 307
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.552721088435375,
|
||
|
|
"grad_norm": 0.5319011173148278,
|
||
|
|
"learning_rate": 4.9986760814493315e-06,
|
||
|
|
"loss": 0.126,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12562274932861328,
|
||
|
|
"step": 3265,
|
||
|
|
"valid_targets_mean": 5123.5,
|
||
|
|
"valid_targets_min": 407
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.561224489795919,
|
||
|
|
"grad_norm": 0.4618195593377142,
|
||
|
|
"learning_rate": 4.94271685783438e-06,
|
||
|
|
"loss": 0.1397,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14663738012313843,
|
||
|
|
"step": 3270,
|
||
|
|
"valid_targets_mean": 5932.4,
|
||
|
|
"valid_targets_min": 444
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.569727891156463,
|
||
|
|
"grad_norm": 0.5310698617901279,
|
||
|
|
"learning_rate": 4.88702843116674e-06,
|
||
|
|
"loss": 0.1365,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14301025867462158,
|
||
|
|
"step": 3275,
|
||
|
|
"valid_targets_mean": 4878.2,
|
||
|
|
"valid_targets_min": 500
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.578231292517007,
|
||
|
|
"grad_norm": 0.5168080006758261,
|
||
|
|
"learning_rate": 4.831611802972112e-06,
|
||
|
|
"loss": 0.1379,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14914800226688385,
|
||
|
|
"step": 3280,
|
||
|
|
"valid_targets_mean": 4904.2,
|
||
|
|
"valid_targets_min": 445
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.586734693877551,
|
||
|
|
"grad_norm": 0.4256918287937025,
|
||
|
|
"learning_rate": 4.776467969888041e-06,
|
||
|
|
"loss": 0.1338,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11805153638124466,
|
||
|
|
"step": 3285,
|
||
|
|
"valid_targets_mean": 5997.4,
|
||
|
|
"valid_targets_min": 2611
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.595238095238095,
|
||
|
|
"grad_norm": 0.47782104658900454,
|
||
|
|
"learning_rate": 4.721597923646008e-06,
|
||
|
|
"loss": 0.1537,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15221372246742249,
|
||
|
|
"step": 3290,
|
||
|
|
"valid_targets_mean": 5877.9,
|
||
|
|
"valid_targets_min": 348
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.603741496598639,
|
||
|
|
"grad_norm": 0.5566130737079859,
|
||
|
|
"learning_rate": 4.667002651053582e-06,
|
||
|
|
"loss": 0.1434,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1397498995065689,
|
||
|
|
"step": 3295,
|
||
|
|
"valid_targets_mean": 4083.9,
|
||
|
|
"valid_targets_min": 404
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.612244897959184,
|
||
|
|
"grad_norm": 0.513815224416179,
|
||
|
|
"learning_rate": 4.612683133976692e-06,
|
||
|
|
"loss": 0.1393,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14056096971035004,
|
||
|
|
"step": 3300,
|
||
|
|
"valid_targets_mean": 5121.2,
|
||
|
|
"valid_targets_min": 337
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.620748299319728,
|
||
|
|
"grad_norm": 0.5104160536335748,
|
||
|
|
"learning_rate": 4.5586403493219365e-06,
|
||
|
|
"loss": 0.144,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15800310671329498,
|
||
|
|
"step": 3305,
|
||
|
|
"valid_targets_mean": 4920.6,
|
||
|
|
"valid_targets_min": 308
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.629251700680272,
|
||
|
|
"grad_norm": 0.5164372692400362,
|
||
|
|
"learning_rate": 4.504875269019038e-06,
|
||
|
|
"loss": 0.1426,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13907302916049957,
|
||
|
|
"step": 3310,
|
||
|
|
"valid_targets_mean": 4771.6,
|
||
|
|
"valid_targets_min": 694
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.637755102040816,
|
||
|
|
"grad_norm": 0.5463248360767734,
|
||
|
|
"learning_rate": 4.451388860003374e-06,
|
||
|
|
"loss": 0.1626,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18626932799816132,
|
||
|
|
"step": 3315,
|
||
|
|
"valid_targets_mean": 4912.5,
|
||
|
|
"valid_targets_min": 306
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.646258503401361,
|
||
|
|
"grad_norm": 0.4853059796846416,
|
||
|
|
"learning_rate": 4.398182084198561e-06,
|
||
|
|
"loss": 0.1313,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.137210875749588,
|
||
|
|
"step": 3320,
|
||
|
|
"valid_targets_mean": 5425.7,
|
||
|
|
"valid_targets_min": 648
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.654761904761905,
|
||
|
|
"grad_norm": 0.4690508162986458,
|
||
|
|
"learning_rate": 4.345255898499172e-06,
|
||
|
|
"loss": 0.1441,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13499334454536438,
|
||
|
|
"step": 3325,
|
||
|
|
"valid_targets_mean": 5606.1,
|
||
|
|
"valid_targets_min": 458
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.663265306122449,
|
||
|
|
"grad_norm": 0.4790090968477585,
|
||
|
|
"learning_rate": 4.292611254753509e-06,
|
||
|
|
"loss": 0.1364,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11853757500648499,
|
||
|
|
"step": 3330,
|
||
|
|
"valid_targets_mean": 5180.9,
|
||
|
|
"valid_targets_min": 708
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.671768707482993,
|
||
|
|
"grad_norm": 0.4602747411549058,
|
||
|
|
"learning_rate": 4.24024909974651e-06,
|
||
|
|
"loss": 0.147,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1369526982307434,
|
||
|
|
"step": 3335,
|
||
|
|
"valid_targets_mean": 5759.1,
|
||
|
|
"valid_targets_min": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.680272108843537,
|
||
|
|
"grad_norm": 0.5603766441095728,
|
||
|
|
"learning_rate": 4.188170375182705e-06,
|
||
|
|
"loss": 0.1492,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1390063762664795,
|
||
|
|
"step": 3340,
|
||
|
|
"valid_targets_mean": 5108.5,
|
||
|
|
"valid_targets_min": 744
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.688775510204081,
|
||
|
|
"grad_norm": 0.49151411698757796,
|
||
|
|
"learning_rate": 4.136376017669281e-06,
|
||
|
|
"loss": 0.1451,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13594385981559753,
|
||
|
|
"step": 3345,
|
||
|
|
"valid_targets_mean": 5231.4,
|
||
|
|
"valid_targets_min": 464
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.697278911564625,
|
||
|
|
"grad_norm": 0.558682348437814,
|
||
|
|
"learning_rate": 4.084866958699247e-06,
|
||
|
|
"loss": 0.1502,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1658667027950287,
|
||
|
|
"step": 3350,
|
||
|
|
"valid_targets_mean": 4843.1,
|
||
|
|
"valid_targets_min": 493
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.70578231292517,
|
||
|
|
"grad_norm": 0.5549347106657229,
|
||
|
|
"learning_rate": 4.0336441246346684e-06,
|
||
|
|
"loss": 0.1406,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14279866218566895,
|
||
|
|
"step": 3355,
|
||
|
|
"valid_targets_mean": 4391.8,
|
||
|
|
"valid_targets_min": 441
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.714285714285714,
|
||
|
|
"grad_norm": 0.4379105912506674,
|
||
|
|
"learning_rate": 3.982708436690001e-06,
|
||
|
|
"loss": 0.1414,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13956327736377716,
|
||
|
|
"step": 3360,
|
||
|
|
"valid_targets_mean": 6376.7,
|
||
|
|
"valid_targets_min": 718
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.7227891156462585,
|
||
|
|
"grad_norm": 0.460430740096498,
|
||
|
|
"learning_rate": 3.932060810915563e-06,
|
||
|
|
"loss": 0.1365,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13277477025985718,
|
||
|
|
"step": 3365,
|
||
|
|
"valid_targets_mean": 5134.2,
|
||
|
|
"valid_targets_min": 823
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.7312925170068025,
|
||
|
|
"grad_norm": 0.48473569844710945,
|
||
|
|
"learning_rate": 3.881702158181015e-06,
|
||
|
|
"loss": 0.1431,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14500927925109863,
|
||
|
|
"step": 3370,
|
||
|
|
"valid_targets_mean": 5358.2,
|
||
|
|
"valid_targets_min": 477
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.739795918367347,
|
||
|
|
"grad_norm": 0.5219700417543464,
|
||
|
|
"learning_rate": 3.831633384159006e-06,
|
||
|
|
"loss": 0.1386,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1320541799068451,
|
||
|
|
"step": 3375,
|
||
|
|
"valid_targets_mean": 5338.3,
|
||
|
|
"valid_targets_min": 427
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.7482993197278915,
|
||
|
|
"grad_norm": 0.49672502079339886,
|
||
|
|
"learning_rate": 3.7818553893088596e-06,
|
||
|
|
"loss": 0.1432,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13686417043209076,
|
||
|
|
"step": 3380,
|
||
|
|
"valid_targets_mean": 5314.4,
|
||
|
|
"valid_targets_min": 619
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.756802721088436,
|
||
|
|
"grad_norm": 0.5321181900624181,
|
||
|
|
"learning_rate": 3.7323690688604153e-06,
|
||
|
|
"loss": 0.1444,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16239497065544128,
|
||
|
|
"step": 3385,
|
||
|
|
"valid_targets_mean": 5048.8,
|
||
|
|
"valid_targets_min": 321
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.76530612244898,
|
||
|
|
"grad_norm": 0.49480829021226763,
|
||
|
|
"learning_rate": 3.6831753127978974e-06,
|
||
|
|
"loss": 0.1453,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14300841093063354,
|
||
|
|
"step": 3390,
|
||
|
|
"valid_targets_mean": 5319.1,
|
||
|
|
"valid_targets_min": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.773809523809524,
|
||
|
|
"grad_norm": 0.5165731709005764,
|
||
|
|
"learning_rate": 3.634275005843935e-06,
|
||
|
|
"loss": 0.1457,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14028236269950867,
|
||
|
|
"step": 3395,
|
||
|
|
"valid_targets_mean": 4979.1,
|
||
|
|
"valid_targets_min": 529
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.782312925170068,
|
||
|
|
"grad_norm": 0.5470187530155836,
|
||
|
|
"learning_rate": 3.585669027443608e-06,
|
||
|
|
"loss": 0.1482,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13687995076179504,
|
||
|
|
"step": 3400,
|
||
|
|
"valid_targets_mean": 4610.4,
|
||
|
|
"valid_targets_min": 413
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.790816326530612,
|
||
|
|
"grad_norm": 0.4998481274571078,
|
||
|
|
"learning_rate": 3.537358251748695e-06,
|
||
|
|
"loss": 0.1507,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1899339258670807,
|
||
|
|
"step": 3405,
|
||
|
|
"valid_targets_mean": 6369.5,
|
||
|
|
"valid_targets_min": 3996
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.799319727891157,
|
||
|
|
"grad_norm": 0.5200039819551425,
|
||
|
|
"learning_rate": 3.489343547601882e-06,
|
||
|
|
"loss": 0.1463,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12981639802455902,
|
||
|
|
"step": 3410,
|
||
|
|
"valid_targets_mean": 5217.9,
|
||
|
|
"valid_targets_min": 404
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.807823129251701,
|
||
|
|
"grad_norm": 0.8738773964177159,
|
||
|
|
"learning_rate": 3.4416257785211983e-06,
|
||
|
|
"loss": 0.1396,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13396048545837402,
|
||
|
|
"step": 3415,
|
||
|
|
"valid_targets_mean": 5708.7,
|
||
|
|
"valid_targets_min": 469
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.816326530612245,
|
||
|
|
"grad_norm": 0.46498845281354323,
|
||
|
|
"learning_rate": 3.3942058026844472e-06,
|
||
|
|
"loss": 0.1394,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16067393124103546,
|
||
|
|
"step": 3420,
|
||
|
|
"valid_targets_mean": 6423.2,
|
||
|
|
"valid_targets_min": 3060
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.824829931972789,
|
||
|
|
"grad_norm": 0.515819163348875,
|
||
|
|
"learning_rate": 3.3470844729137886e-06,
|
||
|
|
"loss": 0.1376,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13283105194568634,
|
||
|
|
"step": 3425,
|
||
|
|
"valid_targets_mean": 4719.8,
|
||
|
|
"valid_targets_min": 337
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.833333333333333,
|
||
|
|
"grad_norm": 0.4907102526087931,
|
||
|
|
"learning_rate": 3.3002626366603896e-06,
|
||
|
|
"loss": 0.1523,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15237513184547424,
|
||
|
|
"step": 3430,
|
||
|
|
"valid_targets_mean": 6599.6,
|
||
|
|
"valid_targets_min": 4691
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.841836734693878,
|
||
|
|
"grad_norm": 0.4754898273292018,
|
||
|
|
"learning_rate": 3.253741135989199e-06,
|
||
|
|
"loss": 0.1332,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12499884516000748,
|
||
|
|
"step": 3435,
|
||
|
|
"valid_targets_mean": 5234.6,
|
||
|
|
"valid_targets_min": 313
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.850340136054422,
|
||
|
|
"grad_norm": 0.5723368085617931,
|
||
|
|
"learning_rate": 3.2075208075637953e-06,
|
||
|
|
"loss": 0.1477,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1293329894542694,
|
||
|
|
"step": 3440,
|
||
|
|
"valid_targets_mean": 4070.1,
|
||
|
|
"valid_targets_min": 657
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.858843537414966,
|
||
|
|
"grad_norm": 0.4912216817983367,
|
||
|
|
"learning_rate": 3.161602482631343e-06,
|
||
|
|
"loss": 0.1413,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1555420309305191,
|
||
|
|
"step": 3445,
|
||
|
|
"valid_targets_mean": 5352.4,
|
||
|
|
"valid_targets_min": 588
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.86734693877551,
|
||
|
|
"grad_norm": 0.5398843285061182,
|
||
|
|
"learning_rate": 3.115986987007622e-06,
|
||
|
|
"loss": 0.1409,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1407453864812851,
|
||
|
|
"step": 3450,
|
||
|
|
"valid_targets_mean": 4912.4,
|
||
|
|
"valid_targets_min": 659
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.875850340136054,
|
||
|
|
"grad_norm": 0.4204471798778164,
|
||
|
|
"learning_rate": 3.0706751410622184e-06,
|
||
|
|
"loss": 0.141,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1201876774430275,
|
||
|
|
"step": 3455,
|
||
|
|
"valid_targets_mean": 6040.9,
|
||
|
|
"valid_targets_min": 866
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.884353741496598,
|
||
|
|
"grad_norm": 0.4645162241834613,
|
||
|
|
"learning_rate": 3.0256677597037253e-06,
|
||
|
|
"loss": 0.1392,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11338837444782257,
|
||
|
|
"step": 3460,
|
||
|
|
"valid_targets_mean": 5154.6,
|
||
|
|
"valid_targets_min": 449
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.892857142857143,
|
||
|
|
"grad_norm": 0.4750999899280527,
|
||
|
|
"learning_rate": 2.980965652365122e-06,
|
||
|
|
"loss": 0.1415,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12996798753738403,
|
||
|
|
"step": 3465,
|
||
|
|
"valid_targets_mean": 5549.3,
|
||
|
|
"valid_targets_min": 591
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.901360544217687,
|
||
|
|
"grad_norm": 0.5185949051667461,
|
||
|
|
"learning_rate": 2.936569622989198e-06,
|
||
|
|
"loss": 0.1344,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12973017990589142,
|
||
|
|
"step": 3470,
|
||
|
|
"valid_targets_mean": 4620.1,
|
||
|
|
"valid_targets_min": 299
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.909863945578231,
|
||
|
|
"grad_norm": 0.5289073007170707,
|
||
|
|
"learning_rate": 2.892480470014101e-06,
|
||
|
|
"loss": 0.1406,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1633252650499344,
|
||
|
|
"step": 3475,
|
||
|
|
"valid_targets_mean": 5262.0,
|
||
|
|
"valid_targets_min": 345
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.918367346938775,
|
||
|
|
"grad_norm": 0.5079511491752337,
|
||
|
|
"learning_rate": 2.8486989863589665e-06,
|
||
|
|
"loss": 0.133,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13232657313346863,
|
||
|
|
"step": 3480,
|
||
|
|
"valid_targets_mean": 5314.4,
|
||
|
|
"valid_targets_min": 368
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.9268707482993195,
|
||
|
|
"grad_norm": 0.4453001608365813,
|
||
|
|
"learning_rate": 2.8052259594096786e-06,
|
||
|
|
"loss": 0.1411,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12859085202217102,
|
||
|
|
"step": 3485,
|
||
|
|
"valid_targets_mean": 5410.6,
|
||
|
|
"valid_targets_min": 825
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.935374149659864,
|
||
|
|
"grad_norm": 0.5254969295724934,
|
||
|
|
"learning_rate": 2.762062171004689e-06,
|
||
|
|
"loss": 0.1413,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12653842568397522,
|
||
|
|
"step": 3490,
|
||
|
|
"valid_targets_mean": 4721.9,
|
||
|
|
"valid_targets_min": 389
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.9438775510204085,
|
||
|
|
"grad_norm": 0.46089738829885674,
|
||
|
|
"learning_rate": 2.719208397420976e-06,
|
||
|
|
"loss": 0.1307,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12686572968959808,
|
||
|
|
"step": 3495,
|
||
|
|
"valid_targets_mean": 5558.4,
|
||
|
|
"valid_targets_min": 314
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.9523809523809526,
|
||
|
|
"grad_norm": 0.4948698154460524,
|
||
|
|
"learning_rate": 2.6766654093600554e-06,
|
||
|
|
"loss": 0.1326,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12483316659927368,
|
||
|
|
"step": 3500,
|
||
|
|
"valid_targets_mean": 5254.2,
|
||
|
|
"valid_targets_min": 495
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.960884353741497,
|
||
|
|
"grad_norm": 0.49622987411537656,
|
||
|
|
"learning_rate": 2.634433971934145e-06,
|
||
|
|
"loss": 0.1482,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14717312157154083,
|
||
|
|
"step": 3505,
|
||
|
|
"valid_targets_mean": 5452.4,
|
||
|
|
"valid_targets_min": 608
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.969387755102041,
|
||
|
|
"grad_norm": 0.5186252275117647,
|
||
|
|
"learning_rate": 2.592514844652396e-06,
|
||
|
|
"loss": 0.1503,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14417871832847595,
|
||
|
|
"step": 3510,
|
||
|
|
"valid_targets_mean": 5608.8,
|
||
|
|
"valid_targets_min": 1099
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.977891156462585,
|
||
|
|
"grad_norm": 0.5460175318749779,
|
||
|
|
"learning_rate": 2.5509087814072373e-06,
|
||
|
|
"loss": 0.1425,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16436246037483215,
|
||
|
|
"step": 3515,
|
||
|
|
"valid_targets_mean": 4484.2,
|
||
|
|
"valid_targets_min": 693
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.986394557823129,
|
||
|
|
"grad_norm": 0.5356539077275947,
|
||
|
|
"learning_rate": 2.5096165304608055e-06,
|
||
|
|
"loss": 0.1416,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12753883004188538,
|
||
|
|
"step": 3520,
|
||
|
|
"valid_targets_mean": 4510.6,
|
||
|
|
"valid_targets_min": 483
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.994897959183674,
|
||
|
|
"grad_norm": 0.4667795154831103,
|
||
|
|
"learning_rate": 2.468638834431507e-06,
|
||
|
|
"loss": 0.1485,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13350117206573486,
|
||
|
|
"step": 3525,
|
||
|
|
"valid_targets_mean": 5574.2,
|
||
|
|
"valid_targets_min": 2459
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.003401360544218,
|
||
|
|
"grad_norm": 0.5412238598115819,
|
||
|
|
"learning_rate": 2.427976430280634e-06,
|
||
|
|
"loss": 0.1304,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14351561665534973,
|
||
|
|
"step": 3530,
|
||
|
|
"valid_targets_mean": 4752.4,
|
||
|
|
"valid_targets_min": 264
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.011904761904762,
|
||
|
|
"grad_norm": 0.5149971558600521,
|
||
|
|
"learning_rate": 2.3876300492991456e-06,
|
||
|
|
"loss": 0.1311,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12468758225440979,
|
||
|
|
"step": 3535,
|
||
|
|
"valid_targets_mean": 4646.8,
|
||
|
|
"valid_targets_min": 407
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.020408163265306,
|
||
|
|
"grad_norm": 0.5073952749995921,
|
||
|
|
"learning_rate": 2.3476004170944843e-06,
|
||
|
|
"loss": 0.1412,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14289984107017517,
|
||
|
|
"step": 3540,
|
||
|
|
"valid_targets_mean": 5517.3,
|
||
|
|
"valid_targets_min": 1489
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.02891156462585,
|
||
|
|
"grad_norm": 0.4516104506241501,
|
||
|
|
"learning_rate": 2.307888253577557e-06,
|
||
|
|
"loss": 0.1332,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13306070864200592,
|
||
|
|
"step": 3545,
|
||
|
|
"valid_targets_mean": 6236.2,
|
||
|
|
"valid_targets_min": 1649
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.037414965986395,
|
||
|
|
"grad_norm": 0.4761706064871336,
|
||
|
|
"learning_rate": 2.2684942729497505e-06,
|
||
|
|
"loss": 0.1363,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1302957683801651,
|
||
|
|
"step": 3550,
|
||
|
|
"valid_targets_mean": 5306.4,
|
||
|
|
"valid_targets_min": 449
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.045918367346939,
|
||
|
|
"grad_norm": 0.4946476959883977,
|
||
|
|
"learning_rate": 2.229419183690118e-06,
|
||
|
|
"loss": 0.1412,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1318894326686859,
|
||
|
|
"step": 3555,
|
||
|
|
"valid_targets_mean": 5418.2,
|
||
|
|
"valid_targets_min": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.054421768707483,
|
||
|
|
"grad_norm": 0.5428212486088545,
|
||
|
|
"learning_rate": 2.1906636885426293e-06,
|
||
|
|
"loss": 0.1345,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1406325250864029,
|
||
|
|
"step": 3560,
|
||
|
|
"valid_targets_mean": 4837.9,
|
||
|
|
"valid_targets_min": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.062925170068027,
|
||
|
|
"grad_norm": 0.6048614025819129,
|
||
|
|
"learning_rate": 2.1522284845035246e-06,
|
||
|
|
"loss": 0.1361,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1283036470413208,
|
||
|
|
"step": 3565,
|
||
|
|
"valid_targets_mean": 5303.8,
|
||
|
|
"valid_targets_min": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.071428571428571,
|
||
|
|
"grad_norm": 0.5661195711362011,
|
||
|
|
"learning_rate": 2.1141142628087908e-06,
|
||
|
|
"loss": 0.1335,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.132005512714386,
|
||
|
|
"step": 3570,
|
||
|
|
"valid_targets_mean": 4992.3,
|
||
|
|
"valid_targets_min": 308
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.079931972789115,
|
||
|
|
"grad_norm": 0.5447933130459522,
|
||
|
|
"learning_rate": 2.0763217089217204e-06,
|
||
|
|
"loss": 0.1409,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14106619358062744,
|
||
|
|
"step": 3575,
|
||
|
|
"valid_targets_mean": 4824.6,
|
||
|
|
"valid_targets_min": 441
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.08843537414966,
|
||
|
|
"grad_norm": 0.5135629691496373,
|
||
|
|
"learning_rate": 2.0388515025205912e-06,
|
||
|
|
"loss": 0.1429,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1345629245042801,
|
||
|
|
"step": 3580,
|
||
|
|
"valid_targets_mean": 5311.8,
|
||
|
|
"valid_targets_min": 576
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.096938775510204,
|
||
|
|
"grad_norm": 0.602352933197083,
|
||
|
|
"learning_rate": 2.0017043174864257e-06,
|
||
|
|
"loss": 0.1321,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14123421907424927,
|
||
|
|
"step": 3585,
|
||
|
|
"valid_targets_mean": 5272.3,
|
||
|
|
"valid_targets_min": 290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.105442176870748,
|
||
|
|
"grad_norm": 0.535406505242492,
|
||
|
|
"learning_rate": 1.9648808218909023e-06,
|
||
|
|
"loss": 0.1469,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.170871764421463,
|
||
|
|
"step": 3590,
|
||
|
|
"valid_targets_mean": 5110.8,
|
||
|
|
"valid_targets_min": 300
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.113945578231292,
|
||
|
|
"grad_norm": 0.46409344023603655,
|
||
|
|
"learning_rate": 1.928381677984319e-06,
|
||
|
|
"loss": 0.1355,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15017318725585938,
|
||
|
|
"step": 3595,
|
||
|
|
"valid_targets_mean": 6270.8,
|
||
|
|
"valid_targets_min": 2067
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.122448979591836,
|
||
|
|
"grad_norm": 0.4992904761375449,
|
||
|
|
"learning_rate": 1.8922075421836729e-06,
|
||
|
|
"loss": 0.1313,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13552138209342957,
|
||
|
|
"step": 3600,
|
||
|
|
"valid_targets_mean": 5018.8,
|
||
|
|
"valid_targets_min": 806
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.130952380952381,
|
||
|
|
"grad_norm": 0.5818489961015932,
|
||
|
|
"learning_rate": 1.8563590650608908e-06,
|
||
|
|
"loss": 0.1313,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13316436111927032,
|
||
|
|
"step": 3605,
|
||
|
|
"valid_targets_mean": 5251.4,
|
||
|
|
"valid_targets_min": 348
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.139455782312925,
|
||
|
|
"grad_norm": 0.5791124026160988,
|
||
|
|
"learning_rate": 1.8208368913310881e-06,
|
||
|
|
"loss": 0.1447,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1683848798274994,
|
||
|
|
"step": 3610,
|
||
|
|
"valid_targets_mean": 5468.9,
|
||
|
|
"valid_targets_min": 437
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.1479591836734695,
|
||
|
|
"grad_norm": 0.5170435646841004,
|
||
|
|
"learning_rate": 1.7856416598410064e-06,
|
||
|
|
"loss": 0.1443,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14764326810836792,
|
||
|
|
"step": 3615,
|
||
|
|
"valid_targets_mean": 4864.4,
|
||
|
|
"valid_targets_min": 895
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.156462585034014,
|
||
|
|
"grad_norm": 0.5445215490541716,
|
||
|
|
"learning_rate": 1.7507740035575049e-06,
|
||
|
|
"loss": 0.1376,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1417514681816101,
|
||
|
|
"step": 3620,
|
||
|
|
"valid_targets_mean": 4410.4,
|
||
|
|
"valid_targets_min": 327
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.164965986394558,
|
||
|
|
"grad_norm": 0.5032291633027232,
|
||
|
|
"learning_rate": 1.7162345495561905e-06,
|
||
|
|
"loss": 0.1372,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15455636382102966,
|
||
|
|
"step": 3625,
|
||
|
|
"valid_targets_mean": 5333.6,
|
||
|
|
"valid_targets_min": 627
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.173469387755102,
|
||
|
|
"grad_norm": 0.5163399690860717,
|
||
|
|
"learning_rate": 1.682023919010125e-06,
|
||
|
|
"loss": 0.1268,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1336623728275299,
|
||
|
|
"step": 3630,
|
||
|
|
"valid_targets_mean": 4367.6,
|
||
|
|
"valid_targets_min": 362
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.181972789115647,
|
||
|
|
"grad_norm": 0.47782407830969,
|
||
|
|
"learning_rate": 1.6481427271786588e-06,
|
||
|
|
"loss": 0.1325,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12816743552684784,
|
||
|
|
"step": 3635,
|
||
|
|
"valid_targets_mean": 5220.8,
|
||
|
|
"valid_targets_min": 786
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.190476190476191,
|
||
|
|
"grad_norm": 0.4945533175368897,
|
||
|
|
"learning_rate": 1.6145915833963788e-06,
|
||
|
|
"loss": 0.1286,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11079257726669312,
|
||
|
|
"step": 3640,
|
||
|
|
"valid_targets_mean": 4806.1,
|
||
|
|
"valid_targets_min": 360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.198979591836735,
|
||
|
|
"grad_norm": 0.5949018727098555,
|
||
|
|
"learning_rate": 1.5813710910621293e-06,
|
||
|
|
"loss": 0.1484,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1445332020521164,
|
||
|
|
"step": 3645,
|
||
|
|
"valid_targets_mean": 4081.7,
|
||
|
|
"valid_targets_min": 515
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.207482993197279,
|
||
|
|
"grad_norm": 0.5261144917667705,
|
||
|
|
"learning_rate": 1.548481847628176e-06,
|
||
|
|
"loss": 0.1429,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14834316074848175,
|
||
|
|
"step": 3650,
|
||
|
|
"valid_targets_mean": 4890.1,
|
||
|
|
"valid_targets_min": 746
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.215986394557823,
|
||
|
|
"grad_norm": 0.5780040697241294,
|
||
|
|
"learning_rate": 1.5159244445894605e-06,
|
||
|
|
"loss": 0.1383,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13335540890693665,
|
||
|
|
"step": 3655,
|
||
|
|
"valid_targets_mean": 3721.8,
|
||
|
|
"valid_targets_min": 333
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.224489795918367,
|
||
|
|
"grad_norm": 0.814175667650355,
|
||
|
|
"learning_rate": 1.4836994674729433e-06,
|
||
|
|
"loss": 0.1427,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12806639075279236,
|
||
|
|
"step": 3660,
|
||
|
|
"valid_targets_mean": 4269.4,
|
||
|
|
"valid_targets_min": 368
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.232993197278912,
|
||
|
|
"grad_norm": 0.4466493097556782,
|
||
|
|
"learning_rate": 1.4518074958271e-06,
|
||
|
|
"loss": 0.1261,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13525915145874023,
|
||
|
|
"step": 3665,
|
||
|
|
"valid_targets_mean": 5949.8,
|
||
|
|
"valid_targets_min": 2319
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.241496598639456,
|
||
|
|
"grad_norm": 0.5404133458819822,
|
||
|
|
"learning_rate": 1.4202491032114797e-06,
|
||
|
|
"loss": 0.1409,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13777552545070648,
|
||
|
|
"step": 3670,
|
||
|
|
"valid_targets_mean": 4552.8,
|
||
|
|
"valid_targets_min": 318
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.25,
|
||
|
|
"grad_norm": 0.5220251057413795,
|
||
|
|
"learning_rate": 1.389024857186403e-06,
|
||
|
|
"loss": 0.1379,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1314641833305359,
|
||
|
|
"step": 3675,
|
||
|
|
"valid_targets_mean": 4993.8,
|
||
|
|
"valid_targets_min": 293
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.258503401360544,
|
||
|
|
"grad_norm": 0.49465475244562024,
|
||
|
|
"learning_rate": 1.3581353193027424e-06,
|
||
|
|
"loss": 0.145,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14749473333358765,
|
||
|
|
"step": 3680,
|
||
|
|
"valid_targets_mean": 5853.1,
|
||
|
|
"valid_targets_min": 717
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.267006802721088,
|
||
|
|
"grad_norm": 0.5425769328744678,
|
||
|
|
"learning_rate": 1.3275810450918257e-06,
|
||
|
|
"loss": 0.1323,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1194867342710495,
|
||
|
|
"step": 3685,
|
||
|
|
"valid_targets_mean": 4467.5,
|
||
|
|
"valid_targets_min": 445
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.275510204081632,
|
||
|
|
"grad_norm": 0.9436303602366533,
|
||
|
|
"learning_rate": 1.2973625840554593e-06,
|
||
|
|
"loss": 0.1301,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13065916299819946,
|
||
|
|
"step": 3690,
|
||
|
|
"valid_targets_mean": 4918.2,
|
||
|
|
"valid_targets_min": 773
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.284013605442177,
|
||
|
|
"grad_norm": 0.5135044261472489,
|
||
|
|
"learning_rate": 1.2674804796560202e-06,
|
||
|
|
"loss": 0.1306,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.140975683927536,
|
||
|
|
"step": 3695,
|
||
|
|
"valid_targets_mean": 5223.2,
|
||
|
|
"valid_targets_min": 2631
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.292517006802721,
|
||
|
|
"grad_norm": 0.6008849716385717,
|
||
|
|
"learning_rate": 1.237935269306707e-06,
|
||
|
|
"loss": 0.1369,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1466502845287323,
|
||
|
|
"step": 3700,
|
||
|
|
"valid_targets_mean": 4270.2,
|
||
|
|
"valid_targets_min": 336
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.301020408163265,
|
||
|
|
"grad_norm": 0.5009466999094054,
|
||
|
|
"learning_rate": 1.2087274843618668e-06,
|
||
|
|
"loss": 0.1481,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16356216371059418,
|
||
|
|
"step": 3705,
|
||
|
|
"valid_targets_mean": 5608.3,
|
||
|
|
"valid_targets_min": 2424
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.309523809523809,
|
||
|
|
"grad_norm": 0.45011371558938584,
|
||
|
|
"learning_rate": 1.1798576501074233e-06,
|
||
|
|
"loss": 0.1317,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12683622539043427,
|
||
|
|
"step": 3710,
|
||
|
|
"valid_targets_mean": 5994.0,
|
||
|
|
"valid_targets_min": 603
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.318027210884353,
|
||
|
|
"grad_norm": 0.5147947059128511,
|
||
|
|
"learning_rate": 1.151326285751455e-06,
|
||
|
|
"loss": 0.1333,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13677933812141418,
|
||
|
|
"step": 3715,
|
||
|
|
"valid_targets_mean": 5408.5,
|
||
|
|
"valid_targets_min": 2630
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.326530612244898,
|
||
|
|
"grad_norm": 0.5276020271978192,
|
||
|
|
"learning_rate": 1.1231339044148393e-06,
|
||
|
|
"loss": 0.1317,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13281160593032837,
|
||
|
|
"step": 3720,
|
||
|
|
"valid_targets_mean": 5316.4,
|
||
|
|
"valid_targets_min": 500
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.335034013605442,
|
||
|
|
"grad_norm": 0.5483754338190715,
|
||
|
|
"learning_rate": 1.095281013122036e-06,
|
||
|
|
"loss": 0.1375,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12652716040611267,
|
||
|
|
"step": 3725,
|
||
|
|
"valid_targets_mean": 5697.8,
|
||
|
|
"valid_targets_min": 344
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.343537414965986,
|
||
|
|
"grad_norm": 0.5132240975977235,
|
||
|
|
"learning_rate": 1.0677681127919581e-06,
|
||
|
|
"loss": 0.1397,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13049758970737457,
|
||
|
|
"step": 3730,
|
||
|
|
"valid_targets_mean": 5565.1,
|
||
|
|
"valid_targets_min": 2574
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.3520408163265305,
|
||
|
|
"grad_norm": 0.5345245556646104,
|
||
|
|
"learning_rate": 1.04059569822897e-06,
|
||
|
|
"loss": 0.1368,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1207992434501648,
|
||
|
|
"step": 3735,
|
||
|
|
"valid_targets_mean": 4756.7,
|
||
|
|
"valid_targets_min": 507
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.360544217687075,
|
||
|
|
"grad_norm": 0.5292919224087456,
|
||
|
|
"learning_rate": 1.0137642581139895e-06,
|
||
|
|
"loss": 0.1396,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1598488986492157,
|
||
|
|
"step": 3740,
|
||
|
|
"valid_targets_mean": 5732.9,
|
||
|
|
"valid_targets_min": 680
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.369047619047619,
|
||
|
|
"grad_norm": 0.4924700791575796,
|
||
|
|
"learning_rate": 9.87274274995693e-07,
|
||
|
|
"loss": 0.1358,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12462534010410309,
|
||
|
|
"step": 3745,
|
||
|
|
"valid_targets_mean": 5150.6,
|
||
|
|
"valid_targets_min": 526
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.377551020408164,
|
||
|
|
"grad_norm": 0.5068020380841414,
|
||
|
|
"learning_rate": 9.61126225281841e-07,
|
||
|
|
"loss": 0.1302,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14289945363998413,
|
||
|
|
"step": 3750,
|
||
|
|
"valid_targets_mean": 5567.8,
|
||
|
|
"valid_targets_min": 2237
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.386054421768708,
|
||
|
|
"grad_norm": 0.5500062379006523,
|
||
|
|
"learning_rate": 9.353205792307163e-07,
|
||
|
|
"loss": 0.1364,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14502450823783875,
|
||
|
|
"step": 3755,
|
||
|
|
"valid_targets_mean": 4606.6,
|
||
|
|
"valid_targets_min": 319
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.394557823129252,
|
||
|
|
"grad_norm": 0.4785409348798175,
|
||
|
|
"learning_rate": 9.098578009426484e-07,
|
||
|
|
"loss": 0.1295,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11811547726392746,
|
||
|
|
"step": 3760,
|
||
|
|
"valid_targets_mean": 5109.4,
|
||
|
|
"valid_targets_min": 491
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.403061224489796,
|
||
|
|
"grad_norm": 0.5713889864007975,
|
||
|
|
"learning_rate": 8.847383483516903e-07,
|
||
|
|
"loss": 0.1284,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12332791090011597,
|
||
|
|
"step": 3765,
|
||
|
|
"valid_targets_mean": 5135.2,
|
||
|
|
"valid_targets_min": 432
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.41156462585034,
|
||
|
|
"grad_norm": 0.47392040091212695,
|
||
|
|
"learning_rate": 8.599626732173671e-07,
|
||
|
|
"loss": 0.1238,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11185071617364883,
|
||
|
|
"step": 3770,
|
||
|
|
"valid_targets_mean": 5058.1,
|
||
|
|
"valid_targets_min": 358
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.420068027210885,
|
||
|
|
"grad_norm": 0.5009157301461793,
|
||
|
|
"learning_rate": 8.355312211165568e-07,
|
||
|
|
"loss": 0.1319,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13047346472740173,
|
||
|
|
"step": 3775,
|
||
|
|
"valid_targets_mean": 5427.2,
|
||
|
|
"valid_targets_min": 2738
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.428571428571429,
|
||
|
|
"grad_norm": 0.4645227497015941,
|
||
|
|
"learning_rate": 8.11444431435473e-07,
|
||
|
|
"loss": 0.1302,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14572149515151978,
|
||
|
|
"step": 3780,
|
||
|
|
"valid_targets_mean": 6111.8,
|
||
|
|
"valid_targets_min": 847
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.437074829931973,
|
||
|
|
"grad_norm": 1.1317354572561027,
|
||
|
|
"learning_rate": 7.87702737361764e-07,
|
||
|
|
"loss": 0.139,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13836367428302765,
|
||
|
|
"step": 3785,
|
||
|
|
"valid_targets_mean": 5487.1,
|
||
|
|
"valid_targets_min": 2643
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.445578231292517,
|
||
|
|
"grad_norm": 0.46776383762726037,
|
||
|
|
"learning_rate": 7.643065658767312e-07,
|
||
|
|
"loss": 0.1391,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14590981602668762,
|
||
|
|
"step": 3790,
|
||
|
|
"valid_targets_mean": 5755.3,
|
||
|
|
"valid_targets_min": 991
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.454081632653061,
|
||
|
|
"grad_norm": 0.5567737825806913,
|
||
|
|
"learning_rate": 7.412563377476312e-07,
|
||
|
|
"loss": 0.1413,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16210207343101501,
|
||
|
|
"step": 3795,
|
||
|
|
"valid_targets_mean": 4803.6,
|
||
|
|
"valid_targets_min": 601
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.462585034013605,
|
||
|
|
"grad_norm": 0.6499553796337002,
|
||
|
|
"learning_rate": 7.185524675201328e-07,
|
||
|
|
"loss": 0.1417,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1397264003753662,
|
||
|
|
"step": 3800,
|
||
|
|
"valid_targets_mean": 4823.9,
|
||
|
|
"valid_targets_min": 983
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.47108843537415,
|
||
|
|
"grad_norm": 0.5137546090897774,
|
||
|
|
"learning_rate": 6.961953635108409e-07,
|
||
|
|
"loss": 0.1381,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1263684332370758,
|
||
|
|
"step": 3805,
|
||
|
|
"valid_targets_mean": 4798.8,
|
||
|
|
"valid_targets_min": 338
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.479591836734694,
|
||
|
|
"grad_norm": 0.45898510326308584,
|
||
|
|
"learning_rate": 6.741854277999583e-07,
|
||
|
|
"loss": 0.1338,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13815146684646606,
|
||
|
|
"step": 3810,
|
||
|
|
"valid_targets_mean": 6198.4,
|
||
|
|
"valid_targets_min": 4405
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.488095238095238,
|
||
|
|
"grad_norm": 0.49633870991064166,
|
||
|
|
"learning_rate": 6.525230562240636e-07,
|
||
|
|
"loss": 0.1353,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12217815220355988,
|
||
|
|
"step": 3815,
|
||
|
|
"valid_targets_mean": 5330.2,
|
||
|
|
"valid_targets_min": 681
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.496598639455782,
|
||
|
|
"grad_norm": 0.6118272707836612,
|
||
|
|
"learning_rate": 6.312086383689831e-07,
|
||
|
|
"loss": 0.1284,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13955502212047577,
|
||
|
|
"step": 3820,
|
||
|
|
"valid_targets_mean": 5343.1,
|
||
|
|
"valid_targets_min": 292
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.505102040816326,
|
||
|
|
"grad_norm": 0.5398809535252362,
|
||
|
|
"learning_rate": 6.10242557562788e-07,
|
||
|
|
"loss": 0.1373,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13160696625709534,
|
||
|
|
"step": 3825,
|
||
|
|
"valid_targets_mean": 5286.6,
|
||
|
|
"valid_targets_min": 670
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.513605442176871,
|
||
|
|
"grad_norm": 0.6152221135870887,
|
||
|
|
"learning_rate": 5.896251908688966e-07,
|
||
|
|
"loss": 0.1336,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1412735879421234,
|
||
|
|
"step": 3830,
|
||
|
|
"valid_targets_mean": 3515.4,
|
||
|
|
"valid_targets_min": 445
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.522108843537415,
|
||
|
|
"grad_norm": 0.493913363327817,
|
||
|
|
"learning_rate": 5.693569090792972e-07,
|
||
|
|
"loss": 0.1371,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12100563943386078,
|
||
|
|
"step": 3835,
|
||
|
|
"valid_targets_mean": 5450.6,
|
||
|
|
"valid_targets_min": 910
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.530612244897959,
|
||
|
|
"grad_norm": 0.6872727843712988,
|
||
|
|
"learning_rate": 5.494380767078822e-07,
|
||
|
|
"loss": 0.1392,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1375819444656372,
|
||
|
|
"step": 3840,
|
||
|
|
"valid_targets_mean": 5006.9,
|
||
|
|
"valid_targets_min": 345
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.539115646258503,
|
||
|
|
"grad_norm": 0.7040593106787781,
|
||
|
|
"learning_rate": 5.298690519838823e-07,
|
||
|
|
"loss": 0.1351,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14189359545707703,
|
||
|
|
"step": 3845,
|
||
|
|
"valid_targets_mean": 4772.2,
|
||
|
|
"valid_targets_min": 1722
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.5476190476190474,
|
||
|
|
"grad_norm": 0.5840836856448359,
|
||
|
|
"learning_rate": 5.106501868454317e-07,
|
||
|
|
"loss": 0.1408,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15618473291397095,
|
||
|
|
"step": 3850,
|
||
|
|
"valid_targets_mean": 3885.8,
|
||
|
|
"valid_targets_min": 291
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.5561224489795915,
|
||
|
|
"grad_norm": 0.5126289992133579,
|
||
|
|
"learning_rate": 4.917818269332442e-07,
|
||
|
|
"loss": 0.1314,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14111900329589844,
|
||
|
|
"step": 3855,
|
||
|
|
"valid_targets_mean": 5815.5,
|
||
|
|
"valid_targets_min": 2955
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.564625850340136,
|
||
|
|
"grad_norm": 0.5723569882255353,
|
||
|
|
"learning_rate": 4.7326431158437826e-07,
|
||
|
|
"loss": 0.1277,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13499674201011658,
|
||
|
|
"step": 3860,
|
||
|
|
"valid_targets_mean": 4710.9,
|
||
|
|
"valid_targets_min": 371
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.5731292517006805,
|
||
|
|
"grad_norm": 0.5249975124188215,
|
||
|
|
"learning_rate": 4.5509797382615517e-07,
|
||
|
|
"loss": 0.138,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14104312658309937,
|
||
|
|
"step": 3865,
|
||
|
|
"valid_targets_mean": 4930.4,
|
||
|
|
"valid_targets_min": 498
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.581632653061225,
|
||
|
|
"grad_norm": 0.538465698250628,
|
||
|
|
"learning_rate": 4.372831403701572e-07,
|
||
|
|
"loss": 0.1434,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16043107211589813,
|
||
|
|
"step": 3870,
|
||
|
|
"valid_targets_mean": 4910.4,
|
||
|
|
"valid_targets_min": 359
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.590136054421769,
|
||
|
|
"grad_norm": 0.6179172711590543,
|
||
|
|
"learning_rate": 4.198201316063566e-07,
|
||
|
|
"loss": 0.1273,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12781232595443726,
|
||
|
|
"step": 3875,
|
||
|
|
"valid_targets_mean": 5813.8,
|
||
|
|
"valid_targets_min": 381
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.598639455782313,
|
||
|
|
"grad_norm": 0.5311532156044606,
|
||
|
|
"learning_rate": 4.02709261597356e-07,
|
||
|
|
"loss": 0.1377,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13504526019096375,
|
||
|
|
"step": 3880,
|
||
|
|
"valid_targets_mean": 5879.4,
|
||
|
|
"valid_targets_min": 321
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.607142857142857,
|
||
|
|
"grad_norm": 0.5708956922729265,
|
||
|
|
"learning_rate": 3.8595083807272393e-07,
|
||
|
|
"loss": 0.1412,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16354067623615265,
|
||
|
|
"step": 3885,
|
||
|
|
"valid_targets_mean": 4698.2,
|
||
|
|
"valid_targets_min": 371
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.615646258503402,
|
||
|
|
"grad_norm": 0.4954734758254675,
|
||
|
|
"learning_rate": 3.695451624234836e-07,
|
||
|
|
"loss": 0.137,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12937995791435242,
|
||
|
|
"step": 3890,
|
||
|
|
"valid_targets_mean": 5635.1,
|
||
|
|
"valid_targets_min": 2801
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.624149659863946,
|
||
|
|
"grad_norm": 0.49901656889565105,
|
||
|
|
"learning_rate": 3.5349252969667956e-07,
|
||
|
|
"loss": 0.1328,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11982685327529907,
|
||
|
|
"step": 3895,
|
||
|
|
"valid_targets_mean": 5422.4,
|
||
|
|
"valid_targets_min": 490
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.63265306122449,
|
||
|
|
"grad_norm": 0.5482594014537372,
|
||
|
|
"learning_rate": 3.3779322859007536e-07,
|
||
|
|
"loss": 0.1412,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13560675084590912,
|
||
|
|
"step": 3900,
|
||
|
|
"valid_targets_mean": 4958.4,
|
||
|
|
"valid_targets_min": 437
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.641156462585034,
|
||
|
|
"grad_norm": 0.5300330627072393,
|
||
|
|
"learning_rate": 3.224475414469552e-07,
|
||
|
|
"loss": 0.124,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1277189552783966,
|
||
|
|
"step": 3905,
|
||
|
|
"valid_targets_mean": 4122.7,
|
||
|
|
"valid_targets_min": 577
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.649659863945578,
|
||
|
|
"grad_norm": 0.5428352334882715,
|
||
|
|
"learning_rate": 3.0745574425105505e-07,
|
||
|
|
"loss": 0.1301,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12326160073280334,
|
||
|
|
"step": 3910,
|
||
|
|
"valid_targets_mean": 4225.0,
|
||
|
|
"valid_targets_min": 366
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.658163265306122,
|
||
|
|
"grad_norm": 0.454378341919778,
|
||
|
|
"learning_rate": 2.928181066215929e-07,
|
||
|
|
"loss": 0.1348,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15102598071098328,
|
||
|
|
"step": 3915,
|
||
|
|
"valid_targets_mean": 6413.6,
|
||
|
|
"valid_targets_min": 2580
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.666666666666667,
|
||
|
|
"grad_norm": 0.5173018715921875,
|
||
|
|
"learning_rate": 2.785348918084241e-07,
|
||
|
|
"loss": 0.1308,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13541677594184875,
|
||
|
|
"step": 3920,
|
||
|
|
"valid_targets_mean": 5312.3,
|
||
|
|
"valid_targets_min": 2133
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.675170068027211,
|
||
|
|
"grad_norm": 0.4831018627745406,
|
||
|
|
"learning_rate": 2.6460635668730027e-07,
|
||
|
|
"loss": 0.1311,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13173210620880127,
|
||
|
|
"step": 3925,
|
||
|
|
"valid_targets_mean": 5105.3,
|
||
|
|
"valid_targets_min": 362
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.683673469387755,
|
||
|
|
"grad_norm": 0.497315872051618,
|
||
|
|
"learning_rate": 2.5103275175526023e-07,
|
||
|
|
"loss": 0.1398,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1208333894610405,
|
||
|
|
"step": 3930,
|
||
|
|
"valid_targets_mean": 5122.9,
|
||
|
|
"valid_targets_min": 353
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.692176870748299,
|
||
|
|
"grad_norm": 0.4848265554748239,
|
||
|
|
"learning_rate": 2.3781432112611303e-07,
|
||
|
|
"loss": 0.142,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12330251187086105,
|
||
|
|
"step": 3935,
|
||
|
|
"valid_targets_mean": 4960.0,
|
||
|
|
"valid_targets_min": 522
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.700680272108843,
|
||
|
|
"grad_norm": 0.5329241591122951,
|
||
|
|
"learning_rate": 2.2495130252605746e-07,
|
||
|
|
"loss": 0.1361,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1381547898054123,
|
||
|
|
"step": 3940,
|
||
|
|
"valid_targets_mean": 4776.6,
|
||
|
|
"valid_targets_min": 556
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.709183673469388,
|
||
|
|
"grad_norm": 0.4454802101455998,
|
||
|
|
"learning_rate": 2.1244392728939857e-07,
|
||
|
|
"loss": 0.1506,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13426657021045685,
|
||
|
|
"step": 3945,
|
||
|
|
"valid_targets_mean": 6203.1,
|
||
|
|
"valid_targets_min": 1466
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.717687074829932,
|
||
|
|
"grad_norm": 0.5537277080619932,
|
||
|
|
"learning_rate": 2.0029242035439768e-07,
|
||
|
|
"loss": 0.1332,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13887690007686615,
|
||
|
|
"step": 3950,
|
||
|
|
"valid_targets_mean": 4429.2,
|
||
|
|
"valid_targets_min": 376
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.726190476190476,
|
||
|
|
"grad_norm": 0.5648912067057678,
|
||
|
|
"learning_rate": 1.8849700025921347e-07,
|
||
|
|
"loss": 0.1475,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16248852014541626,
|
||
|
|
"step": 3955,
|
||
|
|
"valid_targets_mean": 4490.6,
|
||
|
|
"valid_targets_min": 361
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.73469387755102,
|
||
|
|
"grad_norm": 0.5464098818344537,
|
||
|
|
"learning_rate": 1.7705787913798734e-07,
|
||
|
|
"loss": 0.1359,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11722272634506226,
|
||
|
|
"step": 3960,
|
||
|
|
"valid_targets_mean": 5620.1,
|
||
|
|
"valid_targets_min": 2125
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.743197278911564,
|
||
|
|
"grad_norm": 0.45215647197920816,
|
||
|
|
"learning_rate": 1.659752627170086e-07,
|
||
|
|
"loss": 0.1259,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1223878413438797,
|
||
|
|
"step": 3965,
|
||
|
|
"valid_targets_mean": 5482.2,
|
||
|
|
"valid_targets_min": 665
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.7517006802721085,
|
||
|
|
"grad_norm": 0.545893601483813,
|
||
|
|
"learning_rate": 1.5524935031103305e-07,
|
||
|
|
"loss": 0.1494,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1608109176158905,
|
||
|
|
"step": 3970,
|
||
|
|
"valid_targets_mean": 5023.8,
|
||
|
|
"valid_targets_min": 631
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.760204081632653,
|
||
|
|
"grad_norm": 0.4790321655554102,
|
||
|
|
"learning_rate": 1.4488033481969478e-07,
|
||
|
|
"loss": 0.1376,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11648073047399521,
|
||
|
|
"step": 3975,
|
||
|
|
"valid_targets_mean": 5348.5,
|
||
|
|
"valid_targets_min": 752
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.7687074829931975,
|
||
|
|
"grad_norm": 0.5798503030361019,
|
||
|
|
"learning_rate": 1.3486840272402213e-07,
|
||
|
|
"loss": 0.1298,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14990995824337006,
|
||
|
|
"step": 3980,
|
||
|
|
"valid_targets_mean": 4657.9,
|
||
|
|
"valid_targets_min": 317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.7772108843537415,
|
||
|
|
"grad_norm": 0.5391803005965072,
|
||
|
|
"learning_rate": 1.252137340831072e-07,
|
||
|
|
"loss": 0.1379,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1346842348575592,
|
||
|
|
"step": 3985,
|
||
|
|
"valid_targets_mean": 4714.7,
|
||
|
|
"valid_targets_min": 475
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.785714285714286,
|
||
|
|
"grad_norm": 2.3522466924452123,
|
||
|
|
"learning_rate": 1.1591650253084619e-07,
|
||
|
|
"loss": 0.1498,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16637258231639862,
|
||
|
|
"step": 3990,
|
||
|
|
"valid_targets_mean": 5012.9,
|
||
|
|
"valid_targets_min": 2623
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.79421768707483,
|
||
|
|
"grad_norm": 0.4963593510332941,
|
||
|
|
"learning_rate": 1.0697687527283062e-07,
|
||
|
|
"loss": 0.1383,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15907306969165802,
|
||
|
|
"step": 3995,
|
||
|
|
"valid_targets_mean": 5647.0,
|
||
|
|
"valid_targets_min": 483
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.802721088435375,
|
||
|
|
"grad_norm": 0.5375561962162844,
|
||
|
|
"learning_rate": 9.839501308333666e-08,
|
||
|
|
"loss": 0.1349,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1296074092388153,
|
||
|
|
"step": 4000,
|
||
|
|
"valid_targets_mean": 4812.2,
|
||
|
|
"valid_targets_min": 383
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.811224489795919,
|
||
|
|
"grad_norm": 0.5120056947662422,
|
||
|
|
"learning_rate": 9.017107030242944e-08,
|
||
|
|
"loss": 0.1407,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14770393073558807,
|
||
|
|
"step": 4005,
|
||
|
|
"valid_targets_mean": 5055.4,
|
||
|
|
"valid_targets_min": 355
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.819727891156463,
|
||
|
|
"grad_norm": 0.5075547943454766,
|
||
|
|
"learning_rate": 8.230519483319211e-08,
|
||
|
|
"loss": 0.1591,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16366757452487946,
|
||
|
|
"step": 4010,
|
||
|
|
"valid_targets_mean": 4998.1,
|
||
|
|
"valid_targets_min": 534
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.828231292517007,
|
||
|
|
"grad_norm": 0.7097774796072204,
|
||
|
|
"learning_rate": 7.479752813906338e-08,
|
||
|
|
"loss": 0.1396,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12301695346832275,
|
||
|
|
"step": 4015,
|
||
|
|
"valid_targets_mean": 4854.4,
|
||
|
|
"valid_targets_min": 302
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.836734693877551,
|
||
|
|
"grad_norm": 0.5706559495354625,
|
||
|
|
"learning_rate": 6.764820524129745e-08,
|
||
|
|
"loss": 0.1427,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1313580572605133,
|
||
|
|
"step": 4020,
|
||
|
|
"valid_targets_mean": 5001.2,
|
||
|
|
"valid_targets_min": 674
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.845238095238095,
|
||
|
|
"grad_norm": 0.529934098671882,
|
||
|
|
"learning_rate": 6.085735471653032e-08,
|
||
|
|
"loss": 0.1395,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14878013730049133,
|
||
|
|
"step": 4025,
|
||
|
|
"valid_targets_mean": 4574.7,
|
||
|
|
"valid_targets_min": 377
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.853741496598639,
|
||
|
|
"grad_norm": 0.5504701899637366,
|
||
|
|
"learning_rate": 5.442509869446833e-08,
|
||
|
|
"loss": 0.1444,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14428743720054626,
|
||
|
|
"step": 4030,
|
||
|
|
"valid_targets_mean": 4430.9,
|
||
|
|
"valid_targets_min": 365
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.862244897959184,
|
||
|
|
"grad_norm": 0.48835939379977944,
|
||
|
|
"learning_rate": 4.835155285569659e-08,
|
||
|
|
"loss": 0.1378,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1239486038684845,
|
||
|
|
"step": 4035,
|
||
|
|
"valid_targets_mean": 5991.8,
|
||
|
|
"valid_targets_min": 1388
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.870748299319728,
|
||
|
|
"grad_norm": 0.4359534059817426,
|
||
|
|
"learning_rate": 4.263682642959177e-08,
|
||
|
|
"loss": 0.1309,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11775698512792587,
|
||
|
|
"step": 4040,
|
||
|
|
"valid_targets_mean": 6048.2,
|
||
|
|
"valid_targets_min": 2229
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.879251700680272,
|
||
|
|
"grad_norm": 0.49197460386116054,
|
||
|
|
"learning_rate": 3.72810221923614e-08,
|
||
|
|
"loss": 0.136,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14704518020153046,
|
||
|
|
"step": 4045,
|
||
|
|
"valid_targets_mean": 5496.0,
|
||
|
|
"valid_targets_min": 726
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.887755102040816,
|
||
|
|
"grad_norm": 0.5109651656168493,
|
||
|
|
"learning_rate": 3.228423646519652e-08,
|
||
|
|
"loss": 0.153,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16223569214344025,
|
||
|
|
"step": 4050,
|
||
|
|
"valid_targets_mean": 5575.9,
|
||
|
|
"valid_targets_min": 1310
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.896258503401361,
|
||
|
|
"grad_norm": 0.49510738713804237,
|
||
|
|
"learning_rate": 2.764655911253744e-08,
|
||
|
|
"loss": 0.1216,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1253349334001541,
|
||
|
|
"step": 4055,
|
||
|
|
"valid_targets_mean": 5710.8,
|
||
|
|
"valid_targets_min": 318
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.904761904761905,
|
||
|
|
"grad_norm": 0.6017798756303051,
|
||
|
|
"learning_rate": 2.3368073540461778e-08,
|
||
|
|
"loss": 0.1319,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13179939985275269,
|
||
|
|
"step": 4060,
|
||
|
|
"valid_targets_mean": 3800.5,
|
||
|
|
"valid_targets_min": 331
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.913265306122449,
|
||
|
|
"grad_norm": 0.4784736196222179,
|
||
|
|
"learning_rate": 1.944885669517005e-08,
|
||
|
|
"loss": 0.1432,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1561814546585083,
|
||
|
|
"step": 4065,
|
||
|
|
"valid_targets_mean": 5683.8,
|
||
|
|
"valid_targets_min": 532
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.921768707482993,
|
||
|
|
"grad_norm": 0.5613861282082341,
|
||
|
|
"learning_rate": 1.5888979061624566e-08,
|
||
|
|
"loss": 0.1332,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12417840212583542,
|
||
|
|
"step": 4070,
|
||
|
|
"valid_targets_mean": 4445.6,
|
||
|
|
"valid_targets_min": 377
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.930272108843537,
|
||
|
|
"grad_norm": 0.5042685656085569,
|
||
|
|
"learning_rate": 1.268850466226379e-08,
|
||
|
|
"loss": 0.1425,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1308492124080658,
|
||
|
|
"step": 4075,
|
||
|
|
"valid_targets_mean": 5908.4,
|
||
|
|
"valid_targets_min": 1728
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.938775510204081,
|
||
|
|
"grad_norm": 0.5358052595351841,
|
||
|
|
"learning_rate": 9.847491055856583e-09,
|
||
|
|
"loss": 0.1344,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14575347304344177,
|
||
|
|
"step": 4080,
|
||
|
|
"valid_targets_mean": 4705.0,
|
||
|
|
"valid_targets_min": 1708
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.947278911564625,
|
||
|
|
"grad_norm": 0.4651893986560711,
|
||
|
|
"learning_rate": 7.365989336469704e-09,
|
||
|
|
"loss": 0.1356,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13156527280807495,
|
||
|
|
"step": 4085,
|
||
|
|
"valid_targets_mean": 5819.0,
|
||
|
|
"valid_targets_min": 2221
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.95578231292517,
|
||
|
|
"grad_norm": 0.5119377805996788,
|
||
|
|
"learning_rate": 5.244044132544091e-09,
|
||
|
|
"loss": 0.1333,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15422074496746063,
|
||
|
|
"step": 4090,
|
||
|
|
"valid_targets_mean": 5212.2,
|
||
|
|
"valid_targets_min": 2217
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.964285714285714,
|
||
|
|
"grad_norm": 0.5250146217806992,
|
||
|
|
"learning_rate": 3.481693606095515e-09,
|
||
|
|
"loss": 0.1362,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12810224294662476,
|
||
|
|
"step": 4095,
|
||
|
|
"valid_targets_mean": 5086.4,
|
||
|
|
"valid_targets_min": 568
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.9727891156462585,
|
||
|
|
"grad_norm": 0.5211155597675168,
|
||
|
|
"learning_rate": 2.078969452030677e-09,
|
||
|
|
"loss": 0.1457,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19202539324760437,
|
||
|
|
"step": 4100,
|
||
|
|
"valid_targets_mean": 5356.7,
|
||
|
|
"valid_targets_min": 738
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.9812925170068025,
|
||
|
|
"grad_norm": 0.5471446345856446,
|
||
|
|
"learning_rate": 1.0358968975676675e-09,
|
||
|
|
"loss": 0.145,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13759146630764008,
|
||
|
|
"step": 4105,
|
||
|
|
"valid_targets_mean": 4167.4,
|
||
|
|
"valid_targets_min": 349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.989795918367347,
|
||
|
|
"grad_norm": 0.49022153063509466,
|
||
|
|
"learning_rate": 3.524947017941038e-10,
|
||
|
|
"loss": 0.1376,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12410837411880493,
|
||
|
|
"step": 4110,
|
||
|
|
"valid_targets_mean": 5204.0,
|
||
|
|
"valid_targets_min": 440
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.9982993197278915,
|
||
|
|
"grad_norm": 0.5495086795746607,
|
||
|
|
"learning_rate": 2.8775155322957604e-11,
|
||
|
|
"loss": 0.1316,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14458689093589783,
|
||
|
|
"step": 4115,
|
||
|
|
"valid_targets_mean": 4890.0,
|
||
|
|
"valid_targets_min": 591
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 7.0,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15933610498905182,
|
||
|
|
"step": 4116,
|
||
|
|
"total_flos": 1426059575820288.0,
|
||
|
|
"train_loss": 0.1865629164902068,
|
||
|
|
"train_runtime": 19138.755,
|
||
|
|
"train_samples_per_second": 3.44,
|
||
|
|
"train_steps_per_second": 0.215,
|
||
|
|
"valid_targets_mean": 6419.4,
|
||
|
|
"valid_targets_min": 898
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 5,
|
||
|
|
"max_steps": 4116,
|
||
|
|
"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": 1426059575820288.0,
|
||
|
|
"train_batch_size": 1,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|