7682 lines
213 KiB
JSON
7682 lines
213 KiB
JSON
|
{
|
||
|
|
"best_global_step": null,
|
||
|
|
"best_metric": null,
|
||
|
|
"best_model_checkpoint": null,
|
||
|
|
"epoch": 7.0,
|
||
|
|
"eval_steps": 500,
|
||
|
|
"global_step": 3472,
|
||
|
|
"is_hyper_param_search": false,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 0.010080645161290322,
|
||
|
|
"grad_norm": 14.449500155173903,
|
||
|
|
"learning_rate": 4.5977011494252875e-07,
|
||
|
|
"loss": 0.7993,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.38909852504730225,
|
||
|
|
"step": 5,
|
||
|
|
"valid_targets_mean": 2174.2,
|
||
|
|
"valid_targets_min": 331
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.020161290322580645,
|
||
|
|
"grad_norm": 11.371968376917225,
|
||
|
|
"learning_rate": 1.0344827586206898e-06,
|
||
|
|
"loss": 0.7727,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3847242593765259,
|
||
|
|
"step": 10,
|
||
|
|
"valid_targets_mean": 2946.0,
|
||
|
|
"valid_targets_min": 1336
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03024193548387097,
|
||
|
|
"grad_norm": 9.159480193371408,
|
||
|
|
"learning_rate": 1.6091954022988506e-06,
|
||
|
|
"loss": 0.7594,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.33918458223342896,
|
||
|
|
"step": 15,
|
||
|
|
"valid_targets_mean": 3145.9,
|
||
|
|
"valid_targets_min": 1099
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04032258064516129,
|
||
|
|
"grad_norm": 5.650620714196684,
|
||
|
|
"learning_rate": 2.1839080459770117e-06,
|
||
|
|
"loss": 0.7429,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.36300647258758545,
|
||
|
|
"step": 20,
|
||
|
|
"valid_targets_mean": 4150.5,
|
||
|
|
"valid_targets_min": 1342
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05040322580645161,
|
||
|
|
"grad_norm": 5.2774452546548964,
|
||
|
|
"learning_rate": 2.7586206896551725e-06,
|
||
|
|
"loss": 0.6975,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.31926649808883667,
|
||
|
|
"step": 25,
|
||
|
|
"valid_targets_mean": 2899.0,
|
||
|
|
"valid_targets_min": 1639
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06048387096774194,
|
||
|
|
"grad_norm": 3.639767775759478,
|
||
|
|
"learning_rate": 3.3333333333333333e-06,
|
||
|
|
"loss": 0.6738,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3609458804130554,
|
||
|
|
"step": 30,
|
||
|
|
"valid_targets_mean": 3702.1,
|
||
|
|
"valid_targets_min": 1394
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07056451612903226,
|
||
|
|
"grad_norm": 2.707051134946833,
|
||
|
|
"learning_rate": 3.908045977011495e-06,
|
||
|
|
"loss": 0.6767,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24730262160301208,
|
||
|
|
"step": 35,
|
||
|
|
"valid_targets_mean": 2623.6,
|
||
|
|
"valid_targets_min": 1398
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08064516129032258,
|
||
|
|
"grad_norm": 1.655229386846628,
|
||
|
|
"learning_rate": 4.482758620689656e-06,
|
||
|
|
"loss": 0.5759,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3255453109741211,
|
||
|
|
"step": 40,
|
||
|
|
"valid_targets_mean": 3583.9,
|
||
|
|
"valid_targets_min": 1555
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0907258064516129,
|
||
|
|
"grad_norm": 1.3397655726170778,
|
||
|
|
"learning_rate": 5.057471264367817e-06,
|
||
|
|
"loss": 0.5662,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.30872970819473267,
|
||
|
|
"step": 45,
|
||
|
|
"valid_targets_mean": 3672.2,
|
||
|
|
"valid_targets_min": 2485
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.10080645161290322,
|
||
|
|
"grad_norm": 1.1770090510235234,
|
||
|
|
"learning_rate": 5.6321839080459775e-06,
|
||
|
|
"loss": 0.5804,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20463469624519348,
|
||
|
|
"step": 50,
|
||
|
|
"valid_targets_mean": 2331.9,
|
||
|
|
"valid_targets_min": 1387
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11088709677419355,
|
||
|
|
"grad_norm": 1.0547778252669509,
|
||
|
|
"learning_rate": 6.206896551724138e-06,
|
||
|
|
"loss": 0.5211,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22625142335891724,
|
||
|
|
"step": 55,
|
||
|
|
"valid_targets_mean": 2296.1,
|
||
|
|
"valid_targets_min": 1504
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12096774193548387,
|
||
|
|
"grad_norm": 0.9064416672530017,
|
||
|
|
"learning_rate": 6.781609195402299e-06,
|
||
|
|
"loss": 0.5056,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20368868112564087,
|
||
|
|
"step": 60,
|
||
|
|
"valid_targets_mean": 2607.8,
|
||
|
|
"valid_targets_min": 1266
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1310483870967742,
|
||
|
|
"grad_norm": 0.8598717693967117,
|
||
|
|
"learning_rate": 7.35632183908046e-06,
|
||
|
|
"loss": 0.4575,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20426565408706665,
|
||
|
|
"step": 65,
|
||
|
|
"valid_targets_mean": 2609.6,
|
||
|
|
"valid_targets_min": 1512
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14112903225806453,
|
||
|
|
"grad_norm": 0.7584820401140192,
|
||
|
|
"learning_rate": 7.93103448275862e-06,
|
||
|
|
"loss": 0.4856,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.29571622610092163,
|
||
|
|
"step": 70,
|
||
|
|
"valid_targets_mean": 4183.1,
|
||
|
|
"valid_targets_min": 1517
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15120967741935484,
|
||
|
|
"grad_norm": 0.8524803684360264,
|
||
|
|
"learning_rate": 8.505747126436782e-06,
|
||
|
|
"loss": 0.4237,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.25059381127357483,
|
||
|
|
"step": 75,
|
||
|
|
"valid_targets_mean": 3100.4,
|
||
|
|
"valid_targets_min": 1335
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16129032258064516,
|
||
|
|
"grad_norm": 0.7805803147748024,
|
||
|
|
"learning_rate": 9.080459770114942e-06,
|
||
|
|
"loss": 0.4562,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24174298346042633,
|
||
|
|
"step": 80,
|
||
|
|
"valid_targets_mean": 3600.0,
|
||
|
|
"valid_targets_min": 1219
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17137096774193547,
|
||
|
|
"grad_norm": 0.7677664733946342,
|
||
|
|
"learning_rate": 9.655172413793105e-06,
|
||
|
|
"loss": 0.4464,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.3260878324508667,
|
||
|
|
"step": 85,
|
||
|
|
"valid_targets_mean": 2888.0,
|
||
|
|
"valid_targets_min": 387
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1814516129032258,
|
||
|
|
"grad_norm": 0.6501384820600805,
|
||
|
|
"learning_rate": 1.0229885057471264e-05,
|
||
|
|
"loss": 0.4659,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.198003888130188,
|
||
|
|
"step": 90,
|
||
|
|
"valid_targets_mean": 3045.2,
|
||
|
|
"valid_targets_min": 1166
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19153225806451613,
|
||
|
|
"grad_norm": 0.6974668397814775,
|
||
|
|
"learning_rate": 1.0804597701149427e-05,
|
||
|
|
"loss": 0.4604,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17032751441001892,
|
||
|
|
"step": 95,
|
||
|
|
"valid_targets_mean": 2550.2,
|
||
|
|
"valid_targets_min": 1045
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.20161290322580644,
|
||
|
|
"grad_norm": 0.7505539806585207,
|
||
|
|
"learning_rate": 1.1379310344827587e-05,
|
||
|
|
"loss": 0.4469,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19169579446315765,
|
||
|
|
"step": 100,
|
||
|
|
"valid_targets_mean": 2602.4,
|
||
|
|
"valid_targets_min": 1579
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21169354838709678,
|
||
|
|
"grad_norm": 0.7184919283440877,
|
||
|
|
"learning_rate": 1.1954022988505748e-05,
|
||
|
|
"loss": 0.4077,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2819514572620392,
|
||
|
|
"step": 105,
|
||
|
|
"valid_targets_mean": 3276.5,
|
||
|
|
"valid_targets_min": 277
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2217741935483871,
|
||
|
|
"grad_norm": 0.5576782128700126,
|
||
|
|
"learning_rate": 1.2528735632183907e-05,
|
||
|
|
"loss": 0.4011,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16589918732643127,
|
||
|
|
"step": 110,
|
||
|
|
"valid_targets_mean": 3989.9,
|
||
|
|
"valid_targets_min": 1483
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2318548387096774,
|
||
|
|
"grad_norm": 0.6771993510690117,
|
||
|
|
"learning_rate": 1.310344827586207e-05,
|
||
|
|
"loss": 0.4197,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24091395735740662,
|
||
|
|
"step": 115,
|
||
|
|
"valid_targets_mean": 4278.8,
|
||
|
|
"valid_targets_min": 2431
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24193548387096775,
|
||
|
|
"grad_norm": 0.6395511331566243,
|
||
|
|
"learning_rate": 1.367816091954023e-05,
|
||
|
|
"loss": 0.3845,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1813727170228958,
|
||
|
|
"step": 120,
|
||
|
|
"valid_targets_mean": 3667.6,
|
||
|
|
"valid_targets_min": 324
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25201612903225806,
|
||
|
|
"grad_norm": 0.6464135043314241,
|
||
|
|
"learning_rate": 1.4252873563218392e-05,
|
||
|
|
"loss": 0.3881,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2489478588104248,
|
||
|
|
"step": 125,
|
||
|
|
"valid_targets_mean": 3894.1,
|
||
|
|
"valid_targets_min": 1564
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2620967741935484,
|
||
|
|
"grad_norm": 0.6891314564767776,
|
||
|
|
"learning_rate": 1.4827586206896554e-05,
|
||
|
|
"loss": 0.3831,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19056200981140137,
|
||
|
|
"step": 130,
|
||
|
|
"valid_targets_mean": 2912.5,
|
||
|
|
"valid_targets_min": 1355
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2721774193548387,
|
||
|
|
"grad_norm": 0.6543759012481299,
|
||
|
|
"learning_rate": 1.540229885057471e-05,
|
||
|
|
"loss": 0.4173,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19089123606681824,
|
||
|
|
"step": 135,
|
||
|
|
"valid_targets_mean": 3529.0,
|
||
|
|
"valid_targets_min": 494
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.28225806451612906,
|
||
|
|
"grad_norm": 0.6848493902218407,
|
||
|
|
"learning_rate": 1.5977011494252876e-05,
|
||
|
|
"loss": 0.4348,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23437771201133728,
|
||
|
|
"step": 140,
|
||
|
|
"valid_targets_mean": 3498.0,
|
||
|
|
"valid_targets_min": 1183
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2923387096774194,
|
||
|
|
"grad_norm": 0.6569306106351943,
|
||
|
|
"learning_rate": 1.6551724137931037e-05,
|
||
|
|
"loss": 0.4223,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2546037435531616,
|
||
|
|
"step": 145,
|
||
|
|
"valid_targets_mean": 3890.2,
|
||
|
|
"valid_targets_min": 1966
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3024193548387097,
|
||
|
|
"grad_norm": 0.6390804316522537,
|
||
|
|
"learning_rate": 1.7126436781609197e-05,
|
||
|
|
"loss": 0.4105,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22756895422935486,
|
||
|
|
"step": 150,
|
||
|
|
"valid_targets_mean": 3296.8,
|
||
|
|
"valid_targets_min": 1320
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3125,
|
||
|
|
"grad_norm": 0.7960814569838984,
|
||
|
|
"learning_rate": 1.770114942528736e-05,
|
||
|
|
"loss": 0.3663,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24670492112636566,
|
||
|
|
"step": 155,
|
||
|
|
"valid_targets_mean": 3182.8,
|
||
|
|
"valid_targets_min": 346
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3225806451612903,
|
||
|
|
"grad_norm": 0.6886915621414088,
|
||
|
|
"learning_rate": 1.827586206896552e-05,
|
||
|
|
"loss": 0.3741,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16217997670173645,
|
||
|
|
"step": 160,
|
||
|
|
"valid_targets_mean": 2663.0,
|
||
|
|
"valid_targets_min": 304
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3326612903225806,
|
||
|
|
"grad_norm": 0.8210518452320089,
|
||
|
|
"learning_rate": 1.885057471264368e-05,
|
||
|
|
"loss": 0.3957,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15700814127922058,
|
||
|
|
"step": 165,
|
||
|
|
"valid_targets_mean": 2601.8,
|
||
|
|
"valid_targets_min": 1335
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.34274193548387094,
|
||
|
|
"grad_norm": 0.6629968262849919,
|
||
|
|
"learning_rate": 1.942528735632184e-05,
|
||
|
|
"loss": 0.3483,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17296138405799866,
|
||
|
|
"step": 170,
|
||
|
|
"valid_targets_mean": 2878.5,
|
||
|
|
"valid_targets_min": 390
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3528225806451613,
|
||
|
|
"grad_norm": 0.6578514413264408,
|
||
|
|
"learning_rate": 2e-05,
|
||
|
|
"loss": 0.3593,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23956651985645294,
|
||
|
|
"step": 175,
|
||
|
|
"valid_targets_mean": 3449.1,
|
||
|
|
"valid_targets_min": 1179
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3629032258064516,
|
||
|
|
"grad_norm": 0.633793791298061,
|
||
|
|
"learning_rate": 2.0574712643678162e-05,
|
||
|
|
"loss": 0.3761,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1691775918006897,
|
||
|
|
"step": 180,
|
||
|
|
"valid_targets_mean": 3694.9,
|
||
|
|
"valid_targets_min": 1359
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.37298387096774194,
|
||
|
|
"grad_norm": 0.6817986950821238,
|
||
|
|
"learning_rate": 2.1149425287356323e-05,
|
||
|
|
"loss": 0.3638,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20259128510951996,
|
||
|
|
"step": 185,
|
||
|
|
"valid_targets_mean": 3364.4,
|
||
|
|
"valid_targets_min": 1742
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38306451612903225,
|
||
|
|
"grad_norm": 0.6285532936794296,
|
||
|
|
"learning_rate": 2.1724137931034484e-05,
|
||
|
|
"loss": 0.3828,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16747918725013733,
|
||
|
|
"step": 190,
|
||
|
|
"valid_targets_mean": 3639.4,
|
||
|
|
"valid_targets_min": 1133
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39314516129032256,
|
||
|
|
"grad_norm": 0.6569010008623565,
|
||
|
|
"learning_rate": 2.229885057471265e-05,
|
||
|
|
"loss": 0.3846,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18582244217395782,
|
||
|
|
"step": 195,
|
||
|
|
"valid_targets_mean": 3434.5,
|
||
|
|
"valid_targets_min": 1547
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4032258064516129,
|
||
|
|
"grad_norm": 0.7356979400678477,
|
||
|
|
"learning_rate": 2.287356321839081e-05,
|
||
|
|
"loss": 0.3676,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22198417782783508,
|
||
|
|
"step": 200,
|
||
|
|
"valid_targets_mean": 2737.1,
|
||
|
|
"valid_targets_min": 349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.41330645161290325,
|
||
|
|
"grad_norm": 0.6831108834771312,
|
||
|
|
"learning_rate": 2.3448275862068967e-05,
|
||
|
|
"loss": 0.3633,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21049314737319946,
|
||
|
|
"step": 205,
|
||
|
|
"valid_targets_mean": 3560.1,
|
||
|
|
"valid_targets_min": 1480
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.42338709677419356,
|
||
|
|
"grad_norm": 0.905812465923658,
|
||
|
|
"learning_rate": 2.4022988505747127e-05,
|
||
|
|
"loss": 0.3754,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1583234816789627,
|
||
|
|
"step": 210,
|
||
|
|
"valid_targets_mean": 3059.8,
|
||
|
|
"valid_targets_min": 1518
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4334677419354839,
|
||
|
|
"grad_norm": 0.6698258748474574,
|
||
|
|
"learning_rate": 2.4597701149425288e-05,
|
||
|
|
"loss": 0.3611,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20641851425170898,
|
||
|
|
"step": 215,
|
||
|
|
"valid_targets_mean": 3792.4,
|
||
|
|
"valid_targets_min": 360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4435483870967742,
|
||
|
|
"grad_norm": 0.764865915875837,
|
||
|
|
"learning_rate": 2.517241379310345e-05,
|
||
|
|
"loss": 0.3856,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2644338011741638,
|
||
|
|
"step": 220,
|
||
|
|
"valid_targets_mean": 4845.0,
|
||
|
|
"valid_targets_min": 1887
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4536290322580645,
|
||
|
|
"grad_norm": 0.9530476057657241,
|
||
|
|
"learning_rate": 2.574712643678161e-05,
|
||
|
|
"loss": 0.3767,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2127152532339096,
|
||
|
|
"step": 225,
|
||
|
|
"valid_targets_mean": 3657.1,
|
||
|
|
"valid_targets_min": 1407
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4637096774193548,
|
||
|
|
"grad_norm": 0.6932442589053148,
|
||
|
|
"learning_rate": 2.632183908045977e-05,
|
||
|
|
"loss": 0.3682,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2231195867061615,
|
||
|
|
"step": 230,
|
||
|
|
"valid_targets_mean": 3219.0,
|
||
|
|
"valid_targets_min": 440
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4737903225806452,
|
||
|
|
"grad_norm": 0.6545007052655867,
|
||
|
|
"learning_rate": 2.6896551724137935e-05,
|
||
|
|
"loss": 0.3766,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15893417596817017,
|
||
|
|
"step": 235,
|
||
|
|
"valid_targets_mean": 3689.9,
|
||
|
|
"valid_targets_min": 1200
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4838709677419355,
|
||
|
|
"grad_norm": 0.6885228862772315,
|
||
|
|
"learning_rate": 2.7471264367816096e-05,
|
||
|
|
"loss": 0.3538,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18170298635959625,
|
||
|
|
"step": 240,
|
||
|
|
"valid_targets_mean": 3758.2,
|
||
|
|
"valid_targets_min": 2198
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4939516129032258,
|
||
|
|
"grad_norm": 0.611196128823349,
|
||
|
|
"learning_rate": 2.8045977011494257e-05,
|
||
|
|
"loss": 0.3372,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15938988327980042,
|
||
|
|
"step": 245,
|
||
|
|
"valid_targets_mean": 3654.6,
|
||
|
|
"valid_targets_min": 1674
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5040322580645161,
|
||
|
|
"grad_norm": 0.7029642143807657,
|
||
|
|
"learning_rate": 2.8620689655172417e-05,
|
||
|
|
"loss": 0.3322,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1400284320116043,
|
||
|
|
"step": 250,
|
||
|
|
"valid_targets_mean": 3100.2,
|
||
|
|
"valid_targets_min": 345
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5141129032258065,
|
||
|
|
"grad_norm": 0.7922421599556757,
|
||
|
|
"learning_rate": 2.919540229885058e-05,
|
||
|
|
"loss": 0.3765,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17966391146183014,
|
||
|
|
"step": 255,
|
||
|
|
"valid_targets_mean": 3128.8,
|
||
|
|
"valid_targets_min": 1917
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5241935483870968,
|
||
|
|
"grad_norm": 0.6749789286867194,
|
||
|
|
"learning_rate": 2.9770114942528736e-05,
|
||
|
|
"loss": 0.346,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21566540002822876,
|
||
|
|
"step": 260,
|
||
|
|
"valid_targets_mean": 3843.4,
|
||
|
|
"valid_targets_min": 1531
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5342741935483871,
|
||
|
|
"grad_norm": 0.7423662123632774,
|
||
|
|
"learning_rate": 3.0344827586206897e-05,
|
||
|
|
"loss": 0.3812,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1832074522972107,
|
||
|
|
"step": 265,
|
||
|
|
"valid_targets_mean": 2404.9,
|
||
|
|
"valid_targets_min": 1260
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5443548387096774,
|
||
|
|
"grad_norm": 0.7648800632210384,
|
||
|
|
"learning_rate": 3.091954022988506e-05,
|
||
|
|
"loss": 0.3876,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12029941380023956,
|
||
|
|
"step": 270,
|
||
|
|
"valid_targets_mean": 2275.0,
|
||
|
|
"valid_targets_min": 363
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5544354838709677,
|
||
|
|
"grad_norm": 0.597002868682288,
|
||
|
|
"learning_rate": 3.149425287356322e-05,
|
||
|
|
"loss": 0.3859,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12832964956760406,
|
||
|
|
"step": 275,
|
||
|
|
"valid_targets_mean": 3785.6,
|
||
|
|
"valid_targets_min": 2169
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5645161290322581,
|
||
|
|
"grad_norm": 1.0226325341786853,
|
||
|
|
"learning_rate": 3.206896551724138e-05,
|
||
|
|
"loss": 0.3599,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17569780349731445,
|
||
|
|
"step": 280,
|
||
|
|
"valid_targets_mean": 3895.8,
|
||
|
|
"valid_targets_min": 1809
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5745967741935484,
|
||
|
|
"grad_norm": 0.6850736494306995,
|
||
|
|
"learning_rate": 3.264367816091954e-05,
|
||
|
|
"loss": 0.3307,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16676917672157288,
|
||
|
|
"step": 285,
|
||
|
|
"valid_targets_mean": 2524.2,
|
||
|
|
"valid_targets_min": 1707
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5846774193548387,
|
||
|
|
"grad_norm": 0.6383658411094468,
|
||
|
|
"learning_rate": 3.321839080459771e-05,
|
||
|
|
"loss": 0.3376,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14162276685237885,
|
||
|
|
"step": 290,
|
||
|
|
"valid_targets_mean": 2319.9,
|
||
|
|
"valid_targets_min": 285
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.594758064516129,
|
||
|
|
"grad_norm": 0.6331928910150095,
|
||
|
|
"learning_rate": 3.3793103448275865e-05,
|
||
|
|
"loss": 0.3506,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11673640459775925,
|
||
|
|
"step": 295,
|
||
|
|
"valid_targets_mean": 2269.0,
|
||
|
|
"valid_targets_min": 1432
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6048387096774194,
|
||
|
|
"grad_norm": 0.699330352070807,
|
||
|
|
"learning_rate": 3.436781609195403e-05,
|
||
|
|
"loss": 0.3333,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1328885406255722,
|
||
|
|
"step": 300,
|
||
|
|
"valid_targets_mean": 3166.9,
|
||
|
|
"valid_targets_min": 1289
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6149193548387096,
|
||
|
|
"grad_norm": 0.7892062586321906,
|
||
|
|
"learning_rate": 3.4942528735632187e-05,
|
||
|
|
"loss": 0.3274,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1452142894268036,
|
||
|
|
"step": 305,
|
||
|
|
"valid_targets_mean": 3436.1,
|
||
|
|
"valid_targets_min": 1474
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.625,
|
||
|
|
"grad_norm": 0.7567197596368975,
|
||
|
|
"learning_rate": 3.551724137931035e-05,
|
||
|
|
"loss": 0.3496,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1691838800907135,
|
||
|
|
"step": 310,
|
||
|
|
"valid_targets_mean": 2426.5,
|
||
|
|
"valid_targets_min": 456
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6350806451612904,
|
||
|
|
"grad_norm": 0.6387687877365441,
|
||
|
|
"learning_rate": 3.609195402298851e-05,
|
||
|
|
"loss": 0.3347,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14392763376235962,
|
||
|
|
"step": 315,
|
||
|
|
"valid_targets_mean": 2922.0,
|
||
|
|
"valid_targets_min": 1786
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6451612903225806,
|
||
|
|
"grad_norm": 0.697595246140018,
|
||
|
|
"learning_rate": 3.6666666666666666e-05,
|
||
|
|
"loss": 0.3619,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20114536583423615,
|
||
|
|
"step": 320,
|
||
|
|
"valid_targets_mean": 3001.2,
|
||
|
|
"valid_targets_min": 1426
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.655241935483871,
|
||
|
|
"grad_norm": 0.7981270589836555,
|
||
|
|
"learning_rate": 3.724137931034483e-05,
|
||
|
|
"loss": 0.347,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10544638335704803,
|
||
|
|
"step": 325,
|
||
|
|
"valid_targets_mean": 2380.5,
|
||
|
|
"valid_targets_min": 1062
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6653225806451613,
|
||
|
|
"grad_norm": 0.6810979428387121,
|
||
|
|
"learning_rate": 3.7816091954022994e-05,
|
||
|
|
"loss": 0.3721,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.23621079325675964,
|
||
|
|
"step": 330,
|
||
|
|
"valid_targets_mean": 3790.4,
|
||
|
|
"valid_targets_min": 981
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6754032258064516,
|
||
|
|
"grad_norm": 0.6462545622479263,
|
||
|
|
"learning_rate": 3.839080459770115e-05,
|
||
|
|
"loss": 0.3748,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21756233274936676,
|
||
|
|
"step": 335,
|
||
|
|
"valid_targets_mean": 3862.8,
|
||
|
|
"valid_targets_min": 2624
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6854838709677419,
|
||
|
|
"grad_norm": 0.6402667640269576,
|
||
|
|
"learning_rate": 3.8965517241379316e-05,
|
||
|
|
"loss": 0.3639,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19513845443725586,
|
||
|
|
"step": 340,
|
||
|
|
"valid_targets_mean": 4006.6,
|
||
|
|
"valid_targets_min": 2357
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6955645161290323,
|
||
|
|
"grad_norm": 0.5958290929438533,
|
||
|
|
"learning_rate": 3.954022988505747e-05,
|
||
|
|
"loss": 0.3545,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22164441645145416,
|
||
|
|
"step": 345,
|
||
|
|
"valid_targets_mean": 4368.1,
|
||
|
|
"valid_targets_min": 2200
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7056451612903226,
|
||
|
|
"grad_norm": 0.6080525191892164,
|
||
|
|
"learning_rate": 3.99999898870547e-05,
|
||
|
|
"loss": 0.3113,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22130124270915985,
|
||
|
|
"step": 350,
|
||
|
|
"valid_targets_mean": 4355.2,
|
||
|
|
"valid_targets_min": 2703
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7157258064516129,
|
||
|
|
"grad_norm": 0.6864525148845186,
|
||
|
|
"learning_rate": 3.99996359350429e-05,
|
||
|
|
"loss": 0.3377,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13324949145317078,
|
||
|
|
"step": 355,
|
||
|
|
"valid_targets_mean": 2617.1,
|
||
|
|
"valid_targets_min": 1609
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7258064516129032,
|
||
|
|
"grad_norm": 0.6266334317635329,
|
||
|
|
"learning_rate": 3.999877634599299e-05,
|
||
|
|
"loss": 0.3145,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1214652732014656,
|
||
|
|
"step": 360,
|
||
|
|
"valid_targets_mean": 2246.9,
|
||
|
|
"valid_targets_min": 390
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7358870967741935,
|
||
|
|
"grad_norm": 0.7481942159429219,
|
||
|
|
"learning_rate": 3.999741114163738e-05,
|
||
|
|
"loss": 0.3248,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1819283664226532,
|
||
|
|
"step": 365,
|
||
|
|
"valid_targets_mean": 2264.9,
|
||
|
|
"valid_targets_min": 322
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7459677419354839,
|
||
|
|
"grad_norm": 0.6898667518454638,
|
||
|
|
"learning_rate": 3.999554035649159e-05,
|
||
|
|
"loss": 0.3275,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1737220585346222,
|
||
|
|
"step": 370,
|
||
|
|
"valid_targets_mean": 3956.5,
|
||
|
|
"valid_targets_min": 2480
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7560483870967742,
|
||
|
|
"grad_norm": 0.623426624804504,
|
||
|
|
"learning_rate": 3.99931640378534e-05,
|
||
|
|
"loss": 0.3875,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1841704100370407,
|
||
|
|
"step": 375,
|
||
|
|
"valid_targets_mean": 3527.2,
|
||
|
|
"valid_targets_min": 1474
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7661290322580645,
|
||
|
|
"grad_norm": 0.5898127264312745,
|
||
|
|
"learning_rate": 3.999028224580163e-05,
|
||
|
|
"loss": 0.3521,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18726076185703278,
|
||
|
|
"step": 380,
|
||
|
|
"valid_targets_mean": 3584.4,
|
||
|
|
"valid_targets_min": 1465
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7762096774193549,
|
||
|
|
"grad_norm": 0.6250921322087949,
|
||
|
|
"learning_rate": 3.9986895053194646e-05,
|
||
|
|
"loss": 0.3396,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1958356499671936,
|
||
|
|
"step": 385,
|
||
|
|
"valid_targets_mean": 3088.4,
|
||
|
|
"valid_targets_min": 1258
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7862903225806451,
|
||
|
|
"grad_norm": 0.661234325735104,
|
||
|
|
"learning_rate": 3.998300254566852e-05,
|
||
|
|
"loss": 0.3471,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2639602720737457,
|
||
|
|
"step": 390,
|
||
|
|
"valid_targets_mean": 4186.8,
|
||
|
|
"valid_targets_min": 1389
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7963709677419355,
|
||
|
|
"grad_norm": 0.6447546831327706,
|
||
|
|
"learning_rate": 3.997860482163482e-05,
|
||
|
|
"loss": 0.3562,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1819433867931366,
|
||
|
|
"step": 395,
|
||
|
|
"valid_targets_mean": 3570.2,
|
||
|
|
"valid_targets_min": 1618
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8064516129032258,
|
||
|
|
"grad_norm": 0.613167058722557,
|
||
|
|
"learning_rate": 3.99737019922782e-05,
|
||
|
|
"loss": 0.3607,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14388181269168854,
|
||
|
|
"step": 400,
|
||
|
|
"valid_targets_mean": 3024.6,
|
||
|
|
"valid_targets_min": 1945
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8165322580645161,
|
||
|
|
"grad_norm": 0.6635526325808375,
|
||
|
|
"learning_rate": 3.9968294181553515e-05,
|
||
|
|
"loss": 0.3471,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1845952868461609,
|
||
|
|
"step": 405,
|
||
|
|
"valid_targets_mean": 3271.5,
|
||
|
|
"valid_targets_min": 1506
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8266129032258065,
|
||
|
|
"grad_norm": 0.9024241548907097,
|
||
|
|
"learning_rate": 3.9962381526182725e-05,
|
||
|
|
"loss": 0.2926,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17348361015319824,
|
||
|
|
"step": 410,
|
||
|
|
"valid_targets_mean": 3015.5,
|
||
|
|
"valid_targets_min": 757
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8366935483870968,
|
||
|
|
"grad_norm": 0.5719043080218256,
|
||
|
|
"learning_rate": 3.995596417565143e-05,
|
||
|
|
"loss": 0.3207,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15138940513134003,
|
||
|
|
"step": 415,
|
||
|
|
"valid_targets_mean": 3349.9,
|
||
|
|
"valid_targets_min": 408
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8467741935483871,
|
||
|
|
"grad_norm": 0.6313666469410305,
|
||
|
|
"learning_rate": 3.994904229220507e-05,
|
||
|
|
"loss": 0.317,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2121599018573761,
|
||
|
|
"step": 420,
|
||
|
|
"valid_targets_mean": 4051.1,
|
||
|
|
"valid_targets_min": 1292
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8568548387096774,
|
||
|
|
"grad_norm": 0.6150577948254076,
|
||
|
|
"learning_rate": 3.994161605084489e-05,
|
||
|
|
"loss": 0.3243,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13537941873073578,
|
||
|
|
"step": 425,
|
||
|
|
"valid_targets_mean": 2550.8,
|
||
|
|
"valid_targets_min": 1231
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8669354838709677,
|
||
|
|
"grad_norm": 0.6894670400731433,
|
||
|
|
"learning_rate": 3.993368563932343e-05,
|
||
|
|
"loss": 0.34,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1476442515850067,
|
||
|
|
"step": 430,
|
||
|
|
"valid_targets_mean": 2535.0,
|
||
|
|
"valid_targets_min": 1438
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8770161290322581,
|
||
|
|
"grad_norm": 0.5750482412844541,
|
||
|
|
"learning_rate": 3.9925251258139824e-05,
|
||
|
|
"loss": 0.3352,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19290408492088318,
|
||
|
|
"step": 435,
|
||
|
|
"valid_targets_mean": 3563.2,
|
||
|
|
"valid_targets_min": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8870967741935484,
|
||
|
|
"grad_norm": 0.646205776364604,
|
||
|
|
"learning_rate": 3.991631312053474e-05,
|
||
|
|
"loss": 0.3198,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11438087373971939,
|
||
|
|
"step": 440,
|
||
|
|
"valid_targets_mean": 2495.2,
|
||
|
|
"valid_targets_min": 339
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8971774193548387,
|
||
|
|
"grad_norm": 0.5420872512419563,
|
||
|
|
"learning_rate": 3.9906871452484956e-05,
|
||
|
|
"loss": 0.3505,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1990976333618164,
|
||
|
|
"step": 445,
|
||
|
|
"valid_targets_mean": 4355.4,
|
||
|
|
"valid_targets_min": 1552
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.907258064516129,
|
||
|
|
"grad_norm": 0.563835722962377,
|
||
|
|
"learning_rate": 3.989692649269768e-05,
|
||
|
|
"loss": 0.3363,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19511142373085022,
|
||
|
|
"step": 450,
|
||
|
|
"valid_targets_mean": 3751.9,
|
||
|
|
"valid_targets_min": 1369
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9173387096774194,
|
||
|
|
"grad_norm": 0.6779471818931806,
|
||
|
|
"learning_rate": 3.988647849260447e-05,
|
||
|
|
"loss": 0.3295,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12462078779935837,
|
||
|
|
"step": 455,
|
||
|
|
"valid_targets_mean": 2648.4,
|
||
|
|
"valid_targets_min": 1182
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9274193548387096,
|
||
|
|
"grad_norm": 0.7157513704263145,
|
||
|
|
"learning_rate": 3.987552771635494e-05,
|
||
|
|
"loss": 0.3276,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1252136379480362,
|
||
|
|
"step": 460,
|
||
|
|
"valid_targets_mean": 1898.0,
|
||
|
|
"valid_targets_min": 353
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9375,
|
||
|
|
"grad_norm": 0.6111715003003196,
|
||
|
|
"learning_rate": 3.986407444081004e-05,
|
||
|
|
"loss": 0.3747,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10861244052648544,
|
||
|
|
"step": 465,
|
||
|
|
"valid_targets_mean": 2788.2,
|
||
|
|
"valid_targets_min": 1383
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9475806451612904,
|
||
|
|
"grad_norm": 0.6316800924159856,
|
||
|
|
"learning_rate": 3.985211895553504e-05,
|
||
|
|
"loss": 0.3469,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1829114854335785,
|
||
|
|
"step": 470,
|
||
|
|
"valid_targets_mean": 2946.9,
|
||
|
|
"valid_targets_min": 393
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9576612903225806,
|
||
|
|
"grad_norm": 0.641123491516597,
|
||
|
|
"learning_rate": 3.983966156279226e-05,
|
||
|
|
"loss": 0.321,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13300415873527527,
|
||
|
|
"step": 475,
|
||
|
|
"valid_targets_mean": 2532.9,
|
||
|
|
"valid_targets_min": 1582
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.967741935483871,
|
||
|
|
"grad_norm": 0.5833099259121834,
|
||
|
|
"learning_rate": 3.9826702577533384e-05,
|
||
|
|
"loss": 0.3406,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17960479855537415,
|
||
|
|
"step": 480,
|
||
|
|
"valid_targets_mean": 3628.5,
|
||
|
|
"valid_targets_min": 1527
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9778225806451613,
|
||
|
|
"grad_norm": 0.6574133518231456,
|
||
|
|
"learning_rate": 3.981324232739154e-05,
|
||
|
|
"loss": 0.3092,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15223054587841034,
|
||
|
|
"step": 485,
|
||
|
|
"valid_targets_mean": 3036.4,
|
||
|
|
"valid_targets_min": 995
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9879032258064516,
|
||
|
|
"grad_norm": 0.6319952606161152,
|
||
|
|
"learning_rate": 3.979928115267295e-05,
|
||
|
|
"loss": 0.3326,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12126471102237701,
|
||
|
|
"step": 490,
|
||
|
|
"valid_targets_mean": 2551.1,
|
||
|
|
"valid_targets_min": 338
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9979838709677419,
|
||
|
|
"grad_norm": 0.5758082434151766,
|
||
|
|
"learning_rate": 3.978481940634841e-05,
|
||
|
|
"loss": 0.3364,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18527314066886902,
|
||
|
|
"step": 495,
|
||
|
|
"valid_targets_mean": 3776.9,
|
||
|
|
"valid_targets_min": 1652
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0080645161290323,
|
||
|
|
"grad_norm": 0.9495672179628695,
|
||
|
|
"learning_rate": 3.976985745404429e-05,
|
||
|
|
"loss": 0.3136,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15025606751441956,
|
||
|
|
"step": 500,
|
||
|
|
"valid_targets_mean": 2923.8,
|
||
|
|
"valid_targets_min": 1274
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0181451612903225,
|
||
|
|
"grad_norm": 0.6078727672977905,
|
||
|
|
"learning_rate": 3.975439567403335e-05,
|
||
|
|
"loss": 0.3148,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1373104453086853,
|
||
|
|
"step": 505,
|
||
|
|
"valid_targets_mean": 3079.1,
|
||
|
|
"valid_targets_min": 1205
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.028225806451613,
|
||
|
|
"grad_norm": 0.6283065155905574,
|
||
|
|
"learning_rate": 3.973843445722513e-05,
|
||
|
|
"loss": 0.3366,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11961142718791962,
|
||
|
|
"step": 510,
|
||
|
|
"valid_targets_mean": 2468.0,
|
||
|
|
"valid_targets_min": 298
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0383064516129032,
|
||
|
|
"grad_norm": 0.6304226530367547,
|
||
|
|
"learning_rate": 3.9721974207156096e-05,
|
||
|
|
"loss": 0.2923,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09555420279502869,
|
||
|
|
"step": 515,
|
||
|
|
"valid_targets_mean": 2101.6,
|
||
|
|
"valid_targets_min": 1342
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0483870967741935,
|
||
|
|
"grad_norm": 0.538625747707866,
|
||
|
|
"learning_rate": 3.9705015339979436e-05,
|
||
|
|
"loss": 0.3375,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12457385659217834,
|
||
|
|
"step": 520,
|
||
|
|
"valid_targets_mean": 3551.1,
|
||
|
|
"valid_targets_min": 1754
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0584677419354838,
|
||
|
|
"grad_norm": 0.5865283856897597,
|
||
|
|
"learning_rate": 3.9687558284454515e-05,
|
||
|
|
"loss": 0.2944,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1649620532989502,
|
||
|
|
"step": 525,
|
||
|
|
"valid_targets_mean": 4226.4,
|
||
|
|
"valid_targets_min": 1989
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0685483870967742,
|
||
|
|
"grad_norm": 0.6558608299673332,
|
||
|
|
"learning_rate": 3.9669603481936064e-05,
|
||
|
|
"loss": 0.3219,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14547483623027802,
|
||
|
|
"step": 530,
|
||
|
|
"valid_targets_mean": 2827.2,
|
||
|
|
"valid_targets_min": 1335
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0786290322580645,
|
||
|
|
"grad_norm": 0.6702199344850266,
|
||
|
|
"learning_rate": 3.9651151386363e-05,
|
||
|
|
"loss": 0.3199,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1180150955915451,
|
||
|
|
"step": 535,
|
||
|
|
"valid_targets_mean": 2292.6,
|
||
|
|
"valid_targets_min": 1323
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0887096774193548,
|
||
|
|
"grad_norm": 0.7208280310479842,
|
||
|
|
"learning_rate": 3.963220246424696e-05,
|
||
|
|
"loss": 0.2949,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1448603719472885,
|
||
|
|
"step": 540,
|
||
|
|
"valid_targets_mean": 2840.2,
|
||
|
|
"valid_targets_min": 1336
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0987903225806452,
|
||
|
|
"grad_norm": 0.637061132389325,
|
||
|
|
"learning_rate": 3.961275719466052e-05,
|
||
|
|
"loss": 0.3336,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17886832356452942,
|
||
|
|
"step": 545,
|
||
|
|
"valid_targets_mean": 3777.1,
|
||
|
|
"valid_targets_min": 1259
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1088709677419355,
|
||
|
|
"grad_norm": 0.5909230138630466,
|
||
|
|
"learning_rate": 3.9592816069225043e-05,
|
||
|
|
"loss": 0.2839,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1421118676662445,
|
||
|
|
"step": 550,
|
||
|
|
"valid_targets_mean": 3384.0,
|
||
|
|
"valid_targets_min": 976
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1189516129032258,
|
||
|
|
"grad_norm": 0.5692871788434832,
|
||
|
|
"learning_rate": 3.9572379592098294e-05,
|
||
|
|
"loss": 0.3015,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13936136662960052,
|
||
|
|
"step": 555,
|
||
|
|
"valid_targets_mean": 3265.2,
|
||
|
|
"valid_targets_min": 1584
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.129032258064516,
|
||
|
|
"grad_norm": 0.5657346443983022,
|
||
|
|
"learning_rate": 3.955144827996166e-05,
|
||
|
|
"loss": 0.316,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1900048553943634,
|
||
|
|
"step": 560,
|
||
|
|
"valid_targets_mean": 3915.8,
|
||
|
|
"valid_targets_min": 2745
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1391129032258065,
|
||
|
|
"grad_norm": 0.5782917302554,
|
||
|
|
"learning_rate": 3.9530022662007114e-05,
|
||
|
|
"loss": 0.3025,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14163415133953094,
|
||
|
|
"step": 565,
|
||
|
|
"valid_targets_mean": 3317.1,
|
||
|
|
"valid_targets_min": 1155
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1491935483870968,
|
||
|
|
"grad_norm": 0.5339721175078633,
|
||
|
|
"learning_rate": 3.950810327992382e-05,
|
||
|
|
"loss": 0.3047,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17530983686447144,
|
||
|
|
"step": 570,
|
||
|
|
"valid_targets_mean": 4576.4,
|
||
|
|
"valid_targets_min": 3479
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.159274193548387,
|
||
|
|
"grad_norm": 0.5897624274737416,
|
||
|
|
"learning_rate": 3.948569068788442e-05,
|
||
|
|
"loss": 0.2877,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16287142038345337,
|
||
|
|
"step": 575,
|
||
|
|
"valid_targets_mean": 3053.9,
|
||
|
|
"valid_targets_min": 2014
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1693548387096775,
|
||
|
|
"grad_norm": 0.6062396033776388,
|
||
|
|
"learning_rate": 3.9462785452531076e-05,
|
||
|
|
"loss": 0.3173,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14661666750907898,
|
||
|
|
"step": 580,
|
||
|
|
"valid_targets_mean": 3045.1,
|
||
|
|
"valid_targets_min": 1254
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1794354838709677,
|
||
|
|
"grad_norm": 0.6078068955149878,
|
||
|
|
"learning_rate": 3.9439388152961093e-05,
|
||
|
|
"loss": 0.3106,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1872226595878601,
|
||
|
|
"step": 585,
|
||
|
|
"valid_targets_mean": 3270.1,
|
||
|
|
"valid_targets_min": 1265
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.189516129032258,
|
||
|
|
"grad_norm": 0.6157740081966554,
|
||
|
|
"learning_rate": 3.941549938071231e-05,
|
||
|
|
"loss": 0.3379,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22470320761203766,
|
||
|
|
"step": 590,
|
||
|
|
"valid_targets_mean": 3804.6,
|
||
|
|
"valid_targets_min": 1281
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1995967741935485,
|
||
|
|
"grad_norm": 0.4986097868632788,
|
||
|
|
"learning_rate": 3.939111973974811e-05,
|
||
|
|
"loss": 0.3325,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1838417500257492,
|
||
|
|
"step": 595,
|
||
|
|
"valid_targets_mean": 4358.6,
|
||
|
|
"valid_targets_min": 1559
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2096774193548387,
|
||
|
|
"grad_norm": 0.5944107669366764,
|
||
|
|
"learning_rate": 3.936624984644218e-05,
|
||
|
|
"loss": 0.3085,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16719096899032593,
|
||
|
|
"step": 600,
|
||
|
|
"valid_targets_mean": 3411.1,
|
||
|
|
"valid_targets_min": 1386
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.219758064516129,
|
||
|
|
"grad_norm": 0.6386545230945542,
|
||
|
|
"learning_rate": 3.934089032956295e-05,
|
||
|
|
"loss": 0.3,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1210237666964531,
|
||
|
|
"step": 605,
|
||
|
|
"valid_targets_mean": 1967.9,
|
||
|
|
"valid_targets_min": 364
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2298387096774193,
|
||
|
|
"grad_norm": 0.7023140502598928,
|
||
|
|
"learning_rate": 3.931504183025762e-05,
|
||
|
|
"loss": 0.2938,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1287229061126709,
|
||
|
|
"step": 610,
|
||
|
|
"valid_targets_mean": 2255.1,
|
||
|
|
"valid_targets_min": 1321
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2399193548387097,
|
||
|
|
"grad_norm": 0.5628057715319447,
|
||
|
|
"learning_rate": 3.928870500203603e-05,
|
||
|
|
"loss": 0.2881,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1529722362756729,
|
||
|
|
"step": 615,
|
||
|
|
"valid_targets_mean": 3989.5,
|
||
|
|
"valid_targets_min": 2436
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.25,
|
||
|
|
"grad_norm": 0.6229379671562058,
|
||
|
|
"learning_rate": 3.926188051075409e-05,
|
||
|
|
"loss": 0.2939,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15819160640239716,
|
||
|
|
"step": 620,
|
||
|
|
"valid_targets_mean": 2794.4,
|
||
|
|
"valid_targets_min": 1519
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2600806451612903,
|
||
|
|
"grad_norm": 0.6595755333923846,
|
||
|
|
"learning_rate": 3.923456903459695e-05,
|
||
|
|
"loss": 0.3204,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1298087239265442,
|
||
|
|
"step": 625,
|
||
|
|
"valid_targets_mean": 2613.9,
|
||
|
|
"valid_targets_min": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2701612903225805,
|
||
|
|
"grad_norm": 0.5917981051139445,
|
||
|
|
"learning_rate": 3.9206771264061884e-05,
|
||
|
|
"loss": 0.298,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.22833240032196045,
|
||
|
|
"step": 630,
|
||
|
|
"valid_targets_mean": 3583.5,
|
||
|
|
"valid_targets_min": 2028
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.280241935483871,
|
||
|
|
"grad_norm": 0.6389994081977598,
|
||
|
|
"learning_rate": 3.91784879019408e-05,
|
||
|
|
"loss": 0.3312,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19841280579566956,
|
||
|
|
"step": 635,
|
||
|
|
"valid_targets_mean": 3476.6,
|
||
|
|
"valid_targets_min": 1951
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2903225806451613,
|
||
|
|
"grad_norm": 0.5698647201595781,
|
||
|
|
"learning_rate": 3.9149719663302496e-05,
|
||
|
|
"loss": 0.2895,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1601460874080658,
|
||
|
|
"step": 640,
|
||
|
|
"valid_targets_mean": 3440.1,
|
||
|
|
"valid_targets_min": 1317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3004032258064515,
|
||
|
|
"grad_norm": 0.6139737482638931,
|
||
|
|
"learning_rate": 3.9120467275474544e-05,
|
||
|
|
"loss": 0.3015,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17951342463493347,
|
||
|
|
"step": 645,
|
||
|
|
"valid_targets_mean": 3460.4,
|
||
|
|
"valid_targets_min": 1703
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.310483870967742,
|
||
|
|
"grad_norm": 0.6168063964346713,
|
||
|
|
"learning_rate": 3.909073147802496e-05,
|
||
|
|
"loss": 0.3216,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1675456166267395,
|
||
|
|
"step": 650,
|
||
|
|
"valid_targets_mean": 3176.8,
|
||
|
|
"valid_targets_min": 1323
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3205645161290323,
|
||
|
|
"grad_norm": 0.5945947066053701,
|
||
|
|
"learning_rate": 3.906051302274344e-05,
|
||
|
|
"loss": 0.2831,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16125710308551788,
|
||
|
|
"step": 655,
|
||
|
|
"valid_targets_mean": 3592.8,
|
||
|
|
"valid_targets_min": 1874
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3306451612903225,
|
||
|
|
"grad_norm": 0.6366258222761411,
|
||
|
|
"learning_rate": 3.902981267362242e-05,
|
||
|
|
"loss": 0.3148,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20425012707710266,
|
||
|
|
"step": 660,
|
||
|
|
"valid_targets_mean": 3301.2,
|
||
|
|
"valid_targets_min": 345
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.340725806451613,
|
||
|
|
"grad_norm": 0.617178070667225,
|
||
|
|
"learning_rate": 3.89986312068377e-05,
|
||
|
|
"loss": 0.2911,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13732275366783142,
|
||
|
|
"step": 665,
|
||
|
|
"valid_targets_mean": 2800.2,
|
||
|
|
"valid_targets_min": 1880
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3508064516129032,
|
||
|
|
"grad_norm": 0.6428778374501827,
|
||
|
|
"learning_rate": 3.8966969410728854e-05,
|
||
|
|
"loss": 0.2988,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1693982034921646,
|
||
|
|
"step": 670,
|
||
|
|
"valid_targets_mean": 2846.6,
|
||
|
|
"valid_targets_min": 1494
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3608870967741935,
|
||
|
|
"grad_norm": 0.6153857780244045,
|
||
|
|
"learning_rate": 3.893482808577929e-05,
|
||
|
|
"loss": 0.3083,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15399247407913208,
|
||
|
|
"step": 675,
|
||
|
|
"valid_targets_mean": 2930.9,
|
||
|
|
"valid_targets_min": 276
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.370967741935484,
|
||
|
|
"grad_norm": 0.5468392817702072,
|
||
|
|
"learning_rate": 3.8902208044596024e-05,
|
||
|
|
"loss": 0.3065,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2326115071773529,
|
||
|
|
"step": 680,
|
||
|
|
"valid_targets_mean": 5143.5,
|
||
|
|
"valid_targets_min": 1293
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3810483870967742,
|
||
|
|
"grad_norm": 0.6309848320705175,
|
||
|
|
"learning_rate": 3.8869110111889126e-05,
|
||
|
|
"loss": 0.3281,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24701620638370514,
|
||
|
|
"step": 685,
|
||
|
|
"valid_targets_mean": 4289.4,
|
||
|
|
"valid_targets_min": 1965
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3911290322580645,
|
||
|
|
"grad_norm": 0.5810848023168461,
|
||
|
|
"learning_rate": 3.8835535124450845e-05,
|
||
|
|
"loss": 0.2897,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11355516314506531,
|
||
|
|
"step": 690,
|
||
|
|
"valid_targets_mean": 2546.0,
|
||
|
|
"valid_targets_min": 1045
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4012096774193548,
|
||
|
|
"grad_norm": 0.5520816804914948,
|
||
|
|
"learning_rate": 3.88014839311345e-05,
|
||
|
|
"loss": 0.3073,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16475120186805725,
|
||
|
|
"step": 695,
|
||
|
|
"valid_targets_mean": 3843.6,
|
||
|
|
"valid_targets_min": 1449
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4112903225806452,
|
||
|
|
"grad_norm": 0.6492478311719185,
|
||
|
|
"learning_rate": 3.8766957392832996e-05,
|
||
|
|
"loss": 0.3106,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14411963522434235,
|
||
|
|
"step": 700,
|
||
|
|
"valid_targets_mean": 2730.1,
|
||
|
|
"valid_targets_min": 1290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4213709677419355,
|
||
|
|
"grad_norm": 0.5734158826495738,
|
||
|
|
"learning_rate": 3.873195638245705e-05,
|
||
|
|
"loss": 0.3111,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10562250763177872,
|
||
|
|
"step": 705,
|
||
|
|
"valid_targets_mean": 2661.0,
|
||
|
|
"valid_targets_min": 1415
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4314516129032258,
|
||
|
|
"grad_norm": 0.5521253625818489,
|
||
|
|
"learning_rate": 3.869648178491311e-05,
|
||
|
|
"loss": 0.3008,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0762757807970047,
|
||
|
|
"step": 710,
|
||
|
|
"valid_targets_mean": 2243.5,
|
||
|
|
"valid_targets_min": 1336
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.441532258064516,
|
||
|
|
"grad_norm": 0.599682787447533,
|
||
|
|
"learning_rate": 3.866053449708105e-05,
|
||
|
|
"loss": 0.3073,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11245685815811157,
|
||
|
|
"step": 715,
|
||
|
|
"valid_targets_mean": 2342.9,
|
||
|
|
"valid_targets_min": 1541
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4516129032258065,
|
||
|
|
"grad_norm": 0.5929783118351118,
|
||
|
|
"learning_rate": 3.862411542779141e-05,
|
||
|
|
"loss": 0.2898,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1838771104812622,
|
||
|
|
"step": 720,
|
||
|
|
"valid_targets_mean": 3953.0,
|
||
|
|
"valid_targets_min": 2028
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4616935483870968,
|
||
|
|
"grad_norm": 0.6634591421777337,
|
||
|
|
"learning_rate": 3.858722549780246e-05,
|
||
|
|
"loss": 0.3384,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1225997805595398,
|
||
|
|
"step": 725,
|
||
|
|
"valid_targets_mean": 2619.6,
|
||
|
|
"valid_targets_min": 320
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.471774193548387,
|
||
|
|
"grad_norm": 0.6070473432659903,
|
||
|
|
"learning_rate": 3.8549865639776934e-05,
|
||
|
|
"loss": 0.289,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11902737617492676,
|
||
|
|
"step": 730,
|
||
|
|
"valid_targets_mean": 2846.2,
|
||
|
|
"valid_targets_min": 1665
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4818548387096775,
|
||
|
|
"grad_norm": 0.5605852162478396,
|
||
|
|
"learning_rate": 3.851203679825843e-05,
|
||
|
|
"loss": 0.2831,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14411628246307373,
|
||
|
|
"step": 735,
|
||
|
|
"valid_targets_mean": 3102.5,
|
||
|
|
"valid_targets_min": 1641
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4919354838709677,
|
||
|
|
"grad_norm": 0.541887084524303,
|
||
|
|
"learning_rate": 3.847373992964751e-05,
|
||
|
|
"loss": 0.3046,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12645664811134338,
|
||
|
|
"step": 740,
|
||
|
|
"valid_targets_mean": 3280.5,
|
||
|
|
"valid_targets_min": 1416
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.502016129032258,
|
||
|
|
"grad_norm": 0.5655834584416893,
|
||
|
|
"learning_rate": 3.843497600217757e-05,
|
||
|
|
"loss": 0.3001,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1514502465724945,
|
||
|
|
"step": 745,
|
||
|
|
"valid_targets_mean": 3661.1,
|
||
|
|
"valid_targets_min": 1133
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5120967741935485,
|
||
|
|
"grad_norm": 0.6585666737978816,
|
||
|
|
"learning_rate": 3.839574599589032e-05,
|
||
|
|
"loss": 0.2866,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1817966252565384,
|
||
|
|
"step": 750,
|
||
|
|
"valid_targets_mean": 2366.5,
|
||
|
|
"valid_targets_min": 1064
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5221774193548387,
|
||
|
|
"grad_norm": 0.6400032883157108,
|
||
|
|
"learning_rate": 3.835605090261103e-05,
|
||
|
|
"loss": 0.2808,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12336242198944092,
|
||
|
|
"step": 755,
|
||
|
|
"valid_targets_mean": 2274.1,
|
||
|
|
"valid_targets_min": 332
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.532258064516129,
|
||
|
|
"grad_norm": 0.6628619015388243,
|
||
|
|
"learning_rate": 3.831589172592342e-05,
|
||
|
|
"loss": 0.306,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15280428528785706,
|
||
|
|
"step": 760,
|
||
|
|
"valid_targets_mean": 3704.4,
|
||
|
|
"valid_targets_min": 2132
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5423387096774195,
|
||
|
|
"grad_norm": 0.6129730169084127,
|
||
|
|
"learning_rate": 3.827526948114436e-05,
|
||
|
|
"loss": 0.276,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14778190851211548,
|
||
|
|
"step": 765,
|
||
|
|
"valid_targets_mean": 3633.5,
|
||
|
|
"valid_targets_min": 2037
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5524193548387095,
|
||
|
|
"grad_norm": 0.6251091481580988,
|
||
|
|
"learning_rate": 3.823418519529809e-05,
|
||
|
|
"loss": 0.3076,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1652298867702484,
|
||
|
|
"step": 770,
|
||
|
|
"valid_targets_mean": 3264.1,
|
||
|
|
"valid_targets_min": 411
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5625,
|
||
|
|
"grad_norm": 0.6566767774002054,
|
||
|
|
"learning_rate": 3.819263990709037e-05,
|
||
|
|
"loss": 0.2931,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19226676225662231,
|
||
|
|
"step": 775,
|
||
|
|
"valid_targets_mean": 3214.6,
|
||
|
|
"valid_targets_min": 1254
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5725806451612905,
|
||
|
|
"grad_norm": 0.5713085684059585,
|
||
|
|
"learning_rate": 3.815063466688213e-05,
|
||
|
|
"loss": 0.2921,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1409255564212799,
|
||
|
|
"step": 780,
|
||
|
|
"valid_targets_mean": 3301.4,
|
||
|
|
"valid_targets_min": 1122
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5826612903225805,
|
||
|
|
"grad_norm": 0.5415988229419266,
|
||
|
|
"learning_rate": 3.810817053666298e-05,
|
||
|
|
"loss": 0.2738,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12441530078649521,
|
||
|
|
"step": 785,
|
||
|
|
"valid_targets_mean": 3753.4,
|
||
|
|
"valid_targets_min": 2306
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.592741935483871,
|
||
|
|
"grad_norm": 0.5455112683899996,
|
||
|
|
"learning_rate": 3.80652485900243e-05,
|
||
|
|
"loss": 0.2873,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2546766698360443,
|
||
|
|
"step": 790,
|
||
|
|
"valid_targets_mean": 4565.8,
|
||
|
|
"valid_targets_min": 2183
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6028225806451613,
|
||
|
|
"grad_norm": 0.5844075231273836,
|
||
|
|
"learning_rate": 3.802186991213214e-05,
|
||
|
|
"loss": 0.3359,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11077812314033508,
|
||
|
|
"step": 795,
|
||
|
|
"valid_targets_mean": 2542.4,
|
||
|
|
"valid_targets_min": 1462
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6129032258064515,
|
||
|
|
"grad_norm": 0.5396228505416735,
|
||
|
|
"learning_rate": 3.797803559969979e-05,
|
||
|
|
"loss": 0.2888,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16104665398597717,
|
||
|
|
"step": 800,
|
||
|
|
"valid_targets_mean": 3929.0,
|
||
|
|
"valid_targets_min": 1658
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.622983870967742,
|
||
|
|
"grad_norm": 0.5415640995646658,
|
||
|
|
"learning_rate": 3.7933746760960005e-05,
|
||
|
|
"loss": 0.3012,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13626790046691895,
|
||
|
|
"step": 805,
|
||
|
|
"valid_targets_mean": 3165.4,
|
||
|
|
"valid_targets_min": 1771
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6330645161290323,
|
||
|
|
"grad_norm": 0.5733038792595481,
|
||
|
|
"learning_rate": 3.788900451563703e-05,
|
||
|
|
"loss": 0.2949,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11801113933324814,
|
||
|
|
"step": 810,
|
||
|
|
"valid_targets_mean": 2827.5,
|
||
|
|
"valid_targets_min": 1397
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6431451612903225,
|
||
|
|
"grad_norm": 0.5957312339869926,
|
||
|
|
"learning_rate": 3.784380999491829e-05,
|
||
|
|
"loss": 0.2795,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12770137190818787,
|
||
|
|
"step": 815,
|
||
|
|
"valid_targets_mean": 2705.8,
|
||
|
|
"valid_targets_min": 1072
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.653225806451613,
|
||
|
|
"grad_norm": 0.6419793649421601,
|
||
|
|
"learning_rate": 3.779816434142574e-05,
|
||
|
|
"loss": 0.2958,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12950441241264343,
|
||
|
|
"step": 820,
|
||
|
|
"valid_targets_mean": 2315.6,
|
||
|
|
"valid_targets_min": 349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6633064516129032,
|
||
|
|
"grad_norm": 0.6704287059994095,
|
||
|
|
"learning_rate": 3.775206870918707e-05,
|
||
|
|
"loss": 0.2799,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1239781379699707,
|
||
|
|
"step": 825,
|
||
|
|
"valid_targets_mean": 2324.4,
|
||
|
|
"valid_targets_min": 1263
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6733870967741935,
|
||
|
|
"grad_norm": 0.7904219186441301,
|
||
|
|
"learning_rate": 3.770552426360642e-05,
|
||
|
|
"loss": 0.3028,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14967873692512512,
|
||
|
|
"step": 830,
|
||
|
|
"valid_targets_mean": 2720.8,
|
||
|
|
"valid_targets_min": 1431
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.683467741935484,
|
||
|
|
"grad_norm": 0.5633263028669034,
|
||
|
|
"learning_rate": 3.7658532181434985e-05,
|
||
|
|
"loss": 0.3186,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18387183547019958,
|
||
|
|
"step": 835,
|
||
|
|
"valid_targets_mean": 3591.6,
|
||
|
|
"valid_targets_min": 1990
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6935483870967742,
|
||
|
|
"grad_norm": 0.683660085227521,
|
||
|
|
"learning_rate": 3.761109365074127e-05,
|
||
|
|
"loss": 0.3065,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10762707889080048,
|
||
|
|
"step": 840,
|
||
|
|
"valid_targets_mean": 2037.8,
|
||
|
|
"valid_targets_min": 1180
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7036290322580645,
|
||
|
|
"grad_norm": 0.5896378683559472,
|
||
|
|
"learning_rate": 3.756320987088102e-05,
|
||
|
|
"loss": 0.2783,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1592298150062561,
|
||
|
|
"step": 845,
|
||
|
|
"valid_targets_mean": 3112.8,
|
||
|
|
"valid_targets_min": 1257
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.713709677419355,
|
||
|
|
"grad_norm": 0.556399317751764,
|
||
|
|
"learning_rate": 3.751488205246688e-05,
|
||
|
|
"loss": 0.2828,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15031422674655914,
|
||
|
|
"step": 850,
|
||
|
|
"valid_targets_mean": 3392.6,
|
||
|
|
"valid_targets_min": 1468
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.723790322580645,
|
||
|
|
"grad_norm": 0.6687752553220462,
|
||
|
|
"learning_rate": 3.7466111417337857e-05,
|
||
|
|
"loss": 0.3016,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15977172553539276,
|
||
|
|
"step": 855,
|
||
|
|
"valid_targets_mean": 2286.5,
|
||
|
|
"valid_targets_min": 460
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7338709677419355,
|
||
|
|
"grad_norm": 0.5757315760427534,
|
||
|
|
"learning_rate": 3.741689919852837e-05,
|
||
|
|
"loss": 0.2938,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1219063550233841,
|
||
|
|
"step": 860,
|
||
|
|
"valid_targets_mean": 3427.9,
|
||
|
|
"valid_targets_min": 1395
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7439516129032258,
|
||
|
|
"grad_norm": 0.6985465024959804,
|
||
|
|
"learning_rate": 3.7367246640237086e-05,
|
||
|
|
"loss": 0.2869,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1495954841375351,
|
||
|
|
"step": 865,
|
||
|
|
"valid_targets_mean": 2516.1,
|
||
|
|
"valid_targets_min": 402
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.754032258064516,
|
||
|
|
"grad_norm": 0.5710037760995264,
|
||
|
|
"learning_rate": 3.731715499779549e-05,
|
||
|
|
"loss": 0.2851,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10619024187326431,
|
||
|
|
"step": 870,
|
||
|
|
"valid_targets_mean": 3481.4,
|
||
|
|
"valid_targets_min": 1656
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7641129032258065,
|
||
|
|
"grad_norm": 0.6269061801904894,
|
||
|
|
"learning_rate": 3.726662553763611e-05,
|
||
|
|
"loss": 0.3127,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16063417494297028,
|
||
|
|
"step": 875,
|
||
|
|
"valid_targets_mean": 2911.5,
|
||
|
|
"valid_targets_min": 1783
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7741935483870968,
|
||
|
|
"grad_norm": 0.6253710266197406,
|
||
|
|
"learning_rate": 3.721565953726055e-05,
|
||
|
|
"loss": 0.2804,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18021218478679657,
|
||
|
|
"step": 880,
|
||
|
|
"valid_targets_mean": 3110.1,
|
||
|
|
"valid_targets_min": 1286
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.784274193548387,
|
||
|
|
"grad_norm": 0.5411711391901217,
|
||
|
|
"learning_rate": 3.716425828520712e-05,
|
||
|
|
"loss": 0.2938,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1553705781698227,
|
||
|
|
"step": 885,
|
||
|
|
"valid_targets_mean": 3150.5,
|
||
|
|
"valid_targets_min": 1739
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7943548387096775,
|
||
|
|
"grad_norm": 0.5682608892409724,
|
||
|
|
"learning_rate": 3.7112423081018326e-05,
|
||
|
|
"loss": 0.2917,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17028126120567322,
|
||
|
|
"step": 890,
|
||
|
|
"valid_targets_mean": 3763.6,
|
||
|
|
"valid_targets_min": 1488
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8044354838709677,
|
||
|
|
"grad_norm": 0.5898527400440658,
|
||
|
|
"learning_rate": 3.7060155235207976e-05,
|
||
|
|
"loss": 0.3118,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12904928624629974,
|
||
|
|
"step": 895,
|
||
|
|
"valid_targets_mean": 2568.0,
|
||
|
|
"valid_targets_min": 1327
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.814516129032258,
|
||
|
|
"grad_norm": 0.5778165584891043,
|
||
|
|
"learning_rate": 3.7007456069228064e-05,
|
||
|
|
"loss": 0.3069,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.20194143056869507,
|
||
|
|
"step": 900,
|
||
|
|
"valid_targets_mean": 3282.2,
|
||
|
|
"valid_targets_min": 1163
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8245967741935485,
|
||
|
|
"grad_norm": 0.6817185728094591,
|
||
|
|
"learning_rate": 3.6954326915435354e-05,
|
||
|
|
"loss": 0.2971,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10493286699056625,
|
||
|
|
"step": 905,
|
||
|
|
"valid_targets_mean": 1949.0,
|
||
|
|
"valid_targets_min": 968
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8346774193548387,
|
||
|
|
"grad_norm": 0.6344978996331193,
|
||
|
|
"learning_rate": 3.6900769117057696e-05,
|
||
|
|
"loss": 0.2991,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11940060555934906,
|
||
|
|
"step": 910,
|
||
|
|
"valid_targets_mean": 2556.0,
|
||
|
|
"valid_targets_min": 924
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.844758064516129,
|
||
|
|
"grad_norm": 0.6051889288090552,
|
||
|
|
"learning_rate": 3.684678402816006e-05,
|
||
|
|
"loss": 0.2965,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11885014176368713,
|
||
|
|
"step": 915,
|
||
|
|
"valid_targets_mean": 3416.1,
|
||
|
|
"valid_targets_min": 2588
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8548387096774195,
|
||
|
|
"grad_norm": 0.6084362506112826,
|
||
|
|
"learning_rate": 3.679237301361032e-05,
|
||
|
|
"loss": 0.3094,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1991499960422516,
|
||
|
|
"step": 920,
|
||
|
|
"valid_targets_mean": 3730.0,
|
||
|
|
"valid_targets_min": 1164
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8649193548387095,
|
||
|
|
"grad_norm": 0.5684848168481426,
|
||
|
|
"learning_rate": 3.6737537449044725e-05,
|
||
|
|
"loss": 0.3162,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16222137212753296,
|
||
|
|
"step": 925,
|
||
|
|
"valid_targets_mean": 3704.6,
|
||
|
|
"valid_targets_min": 1533
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.875,
|
||
|
|
"grad_norm": 0.5466954105965599,
|
||
|
|
"learning_rate": 3.6682278720833135e-05,
|
||
|
|
"loss": 0.3014,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14616164565086365,
|
||
|
|
"step": 930,
|
||
|
|
"valid_targets_mean": 3506.9,
|
||
|
|
"valid_targets_min": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8850806451612905,
|
||
|
|
"grad_norm": 0.5606631218352714,
|
||
|
|
"learning_rate": 3.662659822604396e-05,
|
||
|
|
"loss": 0.3059,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11968483030796051,
|
||
|
|
"step": 935,
|
||
|
|
"valid_targets_mean": 2919.9,
|
||
|
|
"valid_targets_min": 1260
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8951612903225805,
|
||
|
|
"grad_norm": 0.5824908141498345,
|
||
|
|
"learning_rate": 3.657049737240886e-05,
|
||
|
|
"loss": 0.3154,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15521743893623352,
|
||
|
|
"step": 940,
|
||
|
|
"valid_targets_mean": 3377.1,
|
||
|
|
"valid_targets_min": 1194
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.905241935483871,
|
||
|
|
"grad_norm": 0.5178937923625422,
|
||
|
|
"learning_rate": 3.651397757828711e-05,
|
||
|
|
"loss": 0.3176,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13206562399864197,
|
||
|
|
"step": 945,
|
||
|
|
"valid_targets_mean": 3180.6,
|
||
|
|
"valid_targets_min": 432
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9153225806451613,
|
||
|
|
"grad_norm": 0.5209978876514817,
|
||
|
|
"learning_rate": 3.645704027262981e-05,
|
||
|
|
"loss": 0.3049,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13803443312644958,
|
||
|
|
"step": 950,
|
||
|
|
"valid_targets_mean": 3622.2,
|
||
|
|
"valid_targets_min": 1662
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9254032258064515,
|
||
|
|
"grad_norm": 0.9102324584673673,
|
||
|
|
"learning_rate": 3.639968689494367e-05,
|
||
|
|
"loss": 0.3283,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17182065546512604,
|
||
|
|
"step": 955,
|
||
|
|
"valid_targets_mean": 3440.8,
|
||
|
|
"valid_targets_min": 760
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.935483870967742,
|
||
|
|
"grad_norm": 0.6985231469759166,
|
||
|
|
"learning_rate": 3.63419188952547e-05,
|
||
|
|
"loss": 0.3032,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12019950151443481,
|
||
|
|
"step": 960,
|
||
|
|
"valid_targets_mean": 2839.2,
|
||
|
|
"valid_targets_min": 1269
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9455645161290323,
|
||
|
|
"grad_norm": 0.5105293005958725,
|
||
|
|
"learning_rate": 3.628373773407149e-05,
|
||
|
|
"loss": 0.2916,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11004840582609177,
|
||
|
|
"step": 965,
|
||
|
|
"valid_targets_mean": 3024.9,
|
||
|
|
"valid_targets_min": 1317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9556451612903225,
|
||
|
|
"grad_norm": 0.5031004562041399,
|
||
|
|
"learning_rate": 3.622514488234832e-05,
|
||
|
|
"loss": 0.29,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1211269348859787,
|
||
|
|
"step": 970,
|
||
|
|
"valid_targets_mean": 3122.9,
|
||
|
|
"valid_targets_min": 1205
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.965725806451613,
|
||
|
|
"grad_norm": 0.5334028525432333,
|
||
|
|
"learning_rate": 3.6166141821447956e-05,
|
||
|
|
"loss": 0.2951,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09518647193908691,
|
||
|
|
"step": 975,
|
||
|
|
"valid_targets_mean": 2709.9,
|
||
|
|
"valid_targets_min": 1182
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9758064516129032,
|
||
|
|
"grad_norm": 0.5900614507494186,
|
||
|
|
"learning_rate": 3.6106730043104205e-05,
|
||
|
|
"loss": 0.3072,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17060106992721558,
|
||
|
|
"step": 980,
|
||
|
|
"valid_targets_mean": 3403.8,
|
||
|
|
"valid_targets_min": 1462
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9858870967741935,
|
||
|
|
"grad_norm": 0.503596051885693,
|
||
|
|
"learning_rate": 3.604691104938419e-05,
|
||
|
|
"loss": 0.307,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17102748155593872,
|
||
|
|
"step": 985,
|
||
|
|
"valid_targets_mean": 4337.2,
|
||
|
|
"valid_targets_min": 1290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.995967741935484,
|
||
|
|
"grad_norm": 0.5880890005511348,
|
||
|
|
"learning_rate": 3.598668635265037e-05,
|
||
|
|
"loss": 0.2792,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12174057960510254,
|
||
|
|
"step": 990,
|
||
|
|
"valid_targets_mean": 2455.6,
|
||
|
|
"valid_targets_min": 1020
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.006048387096774,
|
||
|
|
"grad_norm": 0.6211016825696782,
|
||
|
|
"learning_rate": 3.5926057475522335e-05,
|
||
|
|
"loss": 0.2737,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11613931506872177,
|
||
|
|
"step": 995,
|
||
|
|
"valid_targets_mean": 2418.6,
|
||
|
|
"valid_targets_min": 332
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0161290322580645,
|
||
|
|
"grad_norm": 0.578400328341469,
|
||
|
|
"learning_rate": 3.586502595083828e-05,
|
||
|
|
"loss": 0.2883,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16484898328781128,
|
||
|
|
"step": 1000,
|
||
|
|
"valid_targets_mean": 3390.9,
|
||
|
|
"valid_targets_min": 324
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.026209677419355,
|
||
|
|
"grad_norm": 0.6213286467703952,
|
||
|
|
"learning_rate": 3.580359332161627e-05,
|
||
|
|
"loss": 0.2829,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18482288718223572,
|
||
|
|
"step": 1005,
|
||
|
|
"valid_targets_mean": 3653.5,
|
||
|
|
"valid_targets_min": 1182
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.036290322580645,
|
||
|
|
"grad_norm": 0.5282634389941855,
|
||
|
|
"learning_rate": 3.574176114101519e-05,
|
||
|
|
"loss": 0.2831,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0990750789642334,
|
||
|
|
"step": 1010,
|
||
|
|
"valid_targets_mean": 3522.1,
|
||
|
|
"valid_targets_min": 366
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0463709677419355,
|
||
|
|
"grad_norm": 0.5781186476264257,
|
||
|
|
"learning_rate": 3.567953097229554e-05,
|
||
|
|
"loss": 0.239,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12393483519554138,
|
||
|
|
"step": 1015,
|
||
|
|
"valid_targets_mean": 3234.8,
|
||
|
|
"valid_targets_min": 2022
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.056451612903226,
|
||
|
|
"grad_norm": 0.539811650072706,
|
||
|
|
"learning_rate": 3.561690438877988e-05,
|
||
|
|
"loss": 0.2444,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10210546851158142,
|
||
|
|
"step": 1020,
|
||
|
|
"valid_targets_mean": 3349.9,
|
||
|
|
"valid_targets_min": 1584
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.066532258064516,
|
||
|
|
"grad_norm": 0.6019523861196648,
|
||
|
|
"learning_rate": 3.5553882973813026e-05,
|
||
|
|
"loss": 0.2335,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13868030905723572,
|
||
|
|
"step": 1025,
|
||
|
|
"valid_targets_mean": 3342.5,
|
||
|
|
"valid_targets_min": 1491
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0766129032258065,
|
||
|
|
"grad_norm": 0.5749282019614553,
|
||
|
|
"learning_rate": 3.549046832072207e-05,
|
||
|
|
"loss": 0.2604,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11710874736309052,
|
||
|
|
"step": 1030,
|
||
|
|
"valid_targets_mean": 2953.2,
|
||
|
|
"valid_targets_min": 1320
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.086693548387097,
|
||
|
|
"grad_norm": 0.5913137241144276,
|
||
|
|
"learning_rate": 3.542666203277606e-05,
|
||
|
|
"loss": 0.282,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17050395905971527,
|
||
|
|
"step": 1035,
|
||
|
|
"valid_targets_mean": 3180.8,
|
||
|
|
"valid_targets_min": 1374
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.096774193548387,
|
||
|
|
"grad_norm": 0.6803043503170255,
|
||
|
|
"learning_rate": 3.536246572314549e-05,
|
||
|
|
"loss": 0.2555,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1644517183303833,
|
||
|
|
"step": 1040,
|
||
|
|
"valid_targets_mean": 3274.1,
|
||
|
|
"valid_targets_min": 1762
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1068548387096775,
|
||
|
|
"grad_norm": 0.5213587044425169,
|
||
|
|
"learning_rate": 3.529788101486149e-05,
|
||
|
|
"loss": 0.2537,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11054035276174545,
|
||
|
|
"step": 1045,
|
||
|
|
"valid_targets_mean": 3712.4,
|
||
|
|
"valid_targets_min": 1323
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1169354838709675,
|
||
|
|
"grad_norm": 0.6842724047110571,
|
||
|
|
"learning_rate": 3.5232909540774826e-05,
|
||
|
|
"loss": 0.2337,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17232675850391388,
|
||
|
|
"step": 1050,
|
||
|
|
"valid_targets_mean": 3646.6,
|
||
|
|
"valid_targets_min": 331
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.127016129032258,
|
||
|
|
"grad_norm": 0.5689392798834595,
|
||
|
|
"learning_rate": 3.516755294351457e-05,
|
||
|
|
"loss": 0.2612,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11836811900138855,
|
||
|
|
"step": 1055,
|
||
|
|
"valid_targets_mean": 3378.0,
|
||
|
|
"valid_targets_min": 1421
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1370967741935485,
|
||
|
|
"grad_norm": 0.5962649101875634,
|
||
|
|
"learning_rate": 3.5101812875446616e-05,
|
||
|
|
"loss": 0.2803,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11792893707752228,
|
||
|
|
"step": 1060,
|
||
|
|
"valid_targets_mean": 3032.0,
|
||
|
|
"valid_targets_min": 1336
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1471774193548385,
|
||
|
|
"grad_norm": 0.580814768775117,
|
||
|
|
"learning_rate": 3.5035690998631895e-05,
|
||
|
|
"loss": 0.2745,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12068000435829163,
|
||
|
|
"step": 1065,
|
||
|
|
"valid_targets_mean": 3185.5,
|
||
|
|
"valid_targets_min": 1776
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.157258064516129,
|
||
|
|
"grad_norm": 0.5343890649027909,
|
||
|
|
"learning_rate": 3.496918898478432e-05,
|
||
|
|
"loss": 0.271,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11779273301362991,
|
||
|
|
"step": 1070,
|
||
|
|
"valid_targets_mean": 3638.0,
|
||
|
|
"valid_targets_min": 1262
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1673387096774195,
|
||
|
|
"grad_norm": 0.5918405006879289,
|
||
|
|
"learning_rate": 3.4902308515228574e-05,
|
||
|
|
"loss": 0.261,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15182001888751984,
|
||
|
|
"step": 1075,
|
||
|
|
"valid_targets_mean": 3606.5,
|
||
|
|
"valid_targets_min": 1137
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1774193548387095,
|
||
|
|
"grad_norm": 0.6430482707680586,
|
||
|
|
"learning_rate": 3.483505128085756e-05,
|
||
|
|
"loss": 0.2693,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12436778843402863,
|
||
|
|
"step": 1080,
|
||
|
|
"valid_targets_mean": 2416.8,
|
||
|
|
"valid_targets_min": 1415
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1875,
|
||
|
|
"grad_norm": 0.5724531942180483,
|
||
|
|
"learning_rate": 3.4767418982089665e-05,
|
||
|
|
"loss": 0.2769,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11897970736026764,
|
||
|
|
"step": 1085,
|
||
|
|
"valid_targets_mean": 2878.0,
|
||
|
|
"valid_targets_min": 2021
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1975806451612905,
|
||
|
|
"grad_norm": 0.5539109317192396,
|
||
|
|
"learning_rate": 3.4699413328825784e-05,
|
||
|
|
"loss": 0.2504,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13403059542179108,
|
||
|
|
"step": 1090,
|
||
|
|
"valid_targets_mean": 4146.5,
|
||
|
|
"valid_targets_min": 1449
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2076612903225805,
|
||
|
|
"grad_norm": 0.620581105860266,
|
||
|
|
"learning_rate": 3.463103604040606e-05,
|
||
|
|
"loss": 0.2607,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09869108349084854,
|
||
|
|
"step": 1095,
|
||
|
|
"valid_targets_mean": 2461.8,
|
||
|
|
"valid_targets_min": 1190
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.217741935483871,
|
||
|
|
"grad_norm": 0.6649558815064723,
|
||
|
|
"learning_rate": 3.456228884556645e-05,
|
||
|
|
"loss": 0.2885,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11505790054798126,
|
||
|
|
"step": 1100,
|
||
|
|
"valid_targets_mean": 2470.1,
|
||
|
|
"valid_targets_min": 1159
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2278225806451615,
|
||
|
|
"grad_norm": 0.5593391642161272,
|
||
|
|
"learning_rate": 3.4493173482394984e-05,
|
||
|
|
"loss": 0.2554,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13512714207172394,
|
||
|
|
"step": 1105,
|
||
|
|
"valid_targets_mean": 3664.0,
|
||
|
|
"valid_targets_min": 2074
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2379032258064515,
|
||
|
|
"grad_norm": 0.5569583113500925,
|
||
|
|
"learning_rate": 3.442369169828786e-05,
|
||
|
|
"loss": 0.2872,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10272161662578583,
|
||
|
|
"step": 1110,
|
||
|
|
"valid_targets_mean": 3004.1,
|
||
|
|
"valid_targets_min": 1263
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.247983870967742,
|
||
|
|
"grad_norm": 0.585944719353325,
|
||
|
|
"learning_rate": 3.435384524990522e-05,
|
||
|
|
"loss": 0.2672,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13769933581352234,
|
||
|
|
"step": 1115,
|
||
|
|
"valid_targets_mean": 3702.1,
|
||
|
|
"valid_targets_min": 1619
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.258064516129032,
|
||
|
|
"grad_norm": 0.6048369829848832,
|
||
|
|
"learning_rate": 3.428363590312676e-05,
|
||
|
|
"loss": 0.2675,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12044262886047363,
|
||
|
|
"step": 1120,
|
||
|
|
"valid_targets_mean": 3748.9,
|
||
|
|
"valid_targets_min": 395
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2681451612903225,
|
||
|
|
"grad_norm": 0.6194701815132484,
|
||
|
|
"learning_rate": 3.421306543300713e-05,
|
||
|
|
"loss": 0.2578,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16619104146957397,
|
||
|
|
"step": 1125,
|
||
|
|
"valid_targets_mean": 3142.2,
|
||
|
|
"valid_targets_min": 1276
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.278225806451613,
|
||
|
|
"grad_norm": 0.571985462506431,
|
||
|
|
"learning_rate": 3.4142135623730954e-05,
|
||
|
|
"loss": 0.2638,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07842089235782623,
|
||
|
|
"step": 1130,
|
||
|
|
"valid_targets_mean": 2162.5,
|
||
|
|
"valid_targets_min": 370
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.288306451612903,
|
||
|
|
"grad_norm": 0.6006273643605213,
|
||
|
|
"learning_rate": 3.4070848268567826e-05,
|
||
|
|
"loss": 0.2539,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13216403126716614,
|
||
|
|
"step": 1135,
|
||
|
|
"valid_targets_mean": 3703.0,
|
||
|
|
"valid_targets_min": 1660
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2983870967741935,
|
||
|
|
"grad_norm": 0.6265284656020031,
|
||
|
|
"learning_rate": 3.3999205169826906e-05,
|
||
|
|
"loss": 0.2415,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12116280943155289,
|
||
|
|
"step": 1140,
|
||
|
|
"valid_targets_mean": 2788.2,
|
||
|
|
"valid_targets_min": 253
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.308467741935484,
|
||
|
|
"grad_norm": 0.577329024533845,
|
||
|
|
"learning_rate": 3.392720813881137e-05,
|
||
|
|
"loss": 0.268,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11659111827611923,
|
||
|
|
"step": 1145,
|
||
|
|
"valid_targets_mean": 3074.2,
|
||
|
|
"valid_targets_min": 1596
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.318548387096774,
|
||
|
|
"grad_norm": 0.5446977663688667,
|
||
|
|
"learning_rate": 3.385485899577264e-05,
|
||
|
|
"loss": 0.246,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11271137744188309,
|
||
|
|
"step": 1150,
|
||
|
|
"valid_targets_mean": 3104.2,
|
||
|
|
"valid_targets_min": 1675
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3286290322580645,
|
||
|
|
"grad_norm": 0.5146450196513005,
|
||
|
|
"learning_rate": 3.378215956986433e-05,
|
||
|
|
"loss": 0.2686,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0944584310054779,
|
||
|
|
"step": 1155,
|
||
|
|
"valid_targets_mean": 3398.5,
|
||
|
|
"valid_targets_min": 579
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.338709677419355,
|
||
|
|
"grad_norm": 0.6128603075013848,
|
||
|
|
"learning_rate": 3.370911169909601e-05,
|
||
|
|
"loss": 0.2507,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15895679593086243,
|
||
|
|
"step": 1160,
|
||
|
|
"valid_targets_mean": 3274.9,
|
||
|
|
"valid_targets_min": 1271
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.348790322580645,
|
||
|
|
"grad_norm": 0.6446347060811517,
|
||
|
|
"learning_rate": 3.363571723028676e-05,
|
||
|
|
"loss": 0.2769,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11649549752473831,
|
||
|
|
"step": 1165,
|
||
|
|
"valid_targets_mean": 2543.1,
|
||
|
|
"valid_targets_min": 1086
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3588709677419355,
|
||
|
|
"grad_norm": 0.6547317596651695,
|
||
|
|
"learning_rate": 3.356197801901843e-05,
|
||
|
|
"loss": 0.2599,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12316887825727463,
|
||
|
|
"step": 1170,
|
||
|
|
"valid_targets_mean": 2334.2,
|
||
|
|
"valid_targets_min": 338
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.368951612903226,
|
||
|
|
"grad_norm": 0.5999208493637858,
|
||
|
|
"learning_rate": 3.34878959295888e-05,
|
||
|
|
"loss": 0.3013,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1776617467403412,
|
||
|
|
"step": 1175,
|
||
|
|
"valid_targets_mean": 3795.9,
|
||
|
|
"valid_targets_min": 1369
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.379032258064516,
|
||
|
|
"grad_norm": 0.6085698682101284,
|
||
|
|
"learning_rate": 3.341347283496435e-05,
|
||
|
|
"loss": 0.2867,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08039484173059464,
|
||
|
|
"step": 1180,
|
||
|
|
"valid_targets_mean": 2364.5,
|
||
|
|
"valid_targets_min": 1443
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3891129032258065,
|
||
|
|
"grad_norm": 0.5828710866114005,
|
||
|
|
"learning_rate": 3.333871061673302e-05,
|
||
|
|
"loss": 0.2723,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10385999828577042,
|
||
|
|
"step": 1185,
|
||
|
|
"valid_targets_mean": 2703.1,
|
||
|
|
"valid_targets_min": 1218
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.399193548387097,
|
||
|
|
"grad_norm": 0.8353825836697509,
|
||
|
|
"learning_rate": 3.326361116505652e-05,
|
||
|
|
"loss": 0.294,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16143205761909485,
|
||
|
|
"step": 1190,
|
||
|
|
"valid_targets_mean": 4019.1,
|
||
|
|
"valid_targets_min": 2348
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.409274193548387,
|
||
|
|
"grad_norm": 0.558757792902701,
|
||
|
|
"learning_rate": 3.318817637862264e-05,
|
||
|
|
"loss": 0.2562,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10586412250995636,
|
||
|
|
"step": 1195,
|
||
|
|
"valid_targets_mean": 3296.5,
|
||
|
|
"valid_targets_min": 1411
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4193548387096775,
|
||
|
|
"grad_norm": 0.5864759501797366,
|
||
|
|
"learning_rate": 3.3112408164597184e-05,
|
||
|
|
"loss": 0.2851,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12110777199268341,
|
||
|
|
"step": 1200,
|
||
|
|
"valid_targets_mean": 3508.5,
|
||
|
|
"valid_targets_min": 2069
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.429435483870968,
|
||
|
|
"grad_norm": 0.5415595242896772,
|
||
|
|
"learning_rate": 3.303630843857581e-05,
|
||
|
|
"loss": 0.2601,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1720709502696991,
|
||
|
|
"step": 1205,
|
||
|
|
"valid_targets_mean": 4223.1,
|
||
|
|
"valid_targets_min": 1702
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.439516129032258,
|
||
|
|
"grad_norm": 0.5997651694999233,
|
||
|
|
"learning_rate": 3.295987912453552e-05,
|
||
|
|
"loss": 0.2757,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09011685103178024,
|
||
|
|
"step": 1210,
|
||
|
|
"valid_targets_mean": 2850.0,
|
||
|
|
"valid_targets_min": 1072
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4495967741935485,
|
||
|
|
"grad_norm": 0.6130961464369015,
|
||
|
|
"learning_rate": 3.28831221547861e-05,
|
||
|
|
"loss": 0.2702,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16433356702327728,
|
||
|
|
"step": 1215,
|
||
|
|
"valid_targets_mean": 3341.5,
|
||
|
|
"valid_targets_min": 1084
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4596774193548385,
|
||
|
|
"grad_norm": 0.5468980347753639,
|
||
|
|
"learning_rate": 3.280603946992122e-05,
|
||
|
|
"loss": 0.2736,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1779639720916748,
|
||
|
|
"step": 1220,
|
||
|
|
"valid_targets_mean": 3742.9,
|
||
|
|
"valid_targets_min": 1109
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.469758064516129,
|
||
|
|
"grad_norm": 0.5403422803725355,
|
||
|
|
"learning_rate": 3.272863301876937e-05,
|
||
|
|
"loss": 0.2808,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1515682190656662,
|
||
|
|
"step": 1225,
|
||
|
|
"valid_targets_mean": 3297.0,
|
||
|
|
"valid_targets_min": 1415
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4798387096774195,
|
||
|
|
"grad_norm": 0.6145470960691523,
|
||
|
|
"learning_rate": 3.265090475834461e-05,
|
||
|
|
"loss": 0.2725,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21551278233528137,
|
||
|
|
"step": 1230,
|
||
|
|
"valid_targets_mean": 3862.6,
|
||
|
|
"valid_targets_min": 320
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4899193548387095,
|
||
|
|
"grad_norm": 0.5444833553289726,
|
||
|
|
"learning_rate": 3.257285665379708e-05,
|
||
|
|
"loss": 0.2561,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13659034669399261,
|
||
|
|
"step": 1235,
|
||
|
|
"valid_targets_mean": 4359.4,
|
||
|
|
"valid_targets_min": 472
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5,
|
||
|
|
"grad_norm": 0.620901927484836,
|
||
|
|
"learning_rate": 3.249449067836332e-05,
|
||
|
|
"loss": 0.2472,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12945063412189484,
|
||
|
|
"step": 1240,
|
||
|
|
"valid_targets_mean": 2810.6,
|
||
|
|
"valid_targets_min": 1212
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5100806451612905,
|
||
|
|
"grad_norm": 0.5860127611331303,
|
||
|
|
"learning_rate": 3.241580881331639e-05,
|
||
|
|
"loss": 0.2549,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16003957390785217,
|
||
|
|
"step": 1245,
|
||
|
|
"valid_targets_mean": 3824.4,
|
||
|
|
"valid_targets_min": 1982
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5201612903225805,
|
||
|
|
"grad_norm": 0.5990336315744222,
|
||
|
|
"learning_rate": 3.233681304791575e-05,
|
||
|
|
"loss": 0.274,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17913851141929626,
|
||
|
|
"step": 1250,
|
||
|
|
"valid_targets_mean": 2950.5,
|
||
|
|
"valid_targets_min": 1124
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.530241935483871,
|
||
|
|
"grad_norm": 0.5450810353474514,
|
||
|
|
"learning_rate": 3.2257505379357015e-05,
|
||
|
|
"loss": 0.2713,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1434987187385559,
|
||
|
|
"step": 1255,
|
||
|
|
"valid_targets_mean": 3871.0,
|
||
|
|
"valid_targets_min": 2018
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.540322580645161,
|
||
|
|
"grad_norm": 0.569367431192682,
|
||
|
|
"learning_rate": 3.2177887812721384e-05,
|
||
|
|
"loss": 0.2728,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1444566249847412,
|
||
|
|
"step": 1260,
|
||
|
|
"valid_targets_mean": 3413.8,
|
||
|
|
"valid_targets_min": 1257
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5504032258064515,
|
||
|
|
"grad_norm": 0.611719932482177,
|
||
|
|
"learning_rate": 3.2097962360925055e-05,
|
||
|
|
"loss": 0.2817,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08206669241189957,
|
||
|
|
"step": 1265,
|
||
|
|
"valid_targets_mean": 2167.6,
|
||
|
|
"valid_targets_min": 1111
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.560483870967742,
|
||
|
|
"grad_norm": 0.5962394286362083,
|
||
|
|
"learning_rate": 3.201773104466823e-05,
|
||
|
|
"loss": 0.2509,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07353386282920837,
|
||
|
|
"step": 1270,
|
||
|
|
"valid_targets_mean": 2128.2,
|
||
|
|
"valid_targets_min": 343
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.570564516129032,
|
||
|
|
"grad_norm": 0.5408851361404984,
|
||
|
|
"learning_rate": 3.1937195892384104e-05,
|
||
|
|
"loss": 0.2832,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12658008933067322,
|
||
|
|
"step": 1275,
|
||
|
|
"valid_targets_mean": 3592.2,
|
||
|
|
"valid_targets_min": 1874
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5806451612903225,
|
||
|
|
"grad_norm": 0.6078132242018582,
|
||
|
|
"learning_rate": 3.1856358940187524e-05,
|
||
|
|
"loss": 0.2605,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1171221137046814,
|
||
|
|
"step": 1280,
|
||
|
|
"valid_targets_mean": 3559.9,
|
||
|
|
"valid_targets_min": 1188
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.590725806451613,
|
||
|
|
"grad_norm": 0.5215848738012204,
|
||
|
|
"learning_rate": 3.1775222231823547e-05,
|
||
|
|
"loss": 0.2597,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08176010102033615,
|
||
|
|
"step": 1285,
|
||
|
|
"valid_targets_mean": 2901.9,
|
||
|
|
"valid_targets_min": 434
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.600806451612903,
|
||
|
|
"grad_norm": 0.534585604248551,
|
||
|
|
"learning_rate": 3.169378781861576e-05,
|
||
|
|
"loss": 0.2649,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11587056517601013,
|
||
|
|
"step": 1290,
|
||
|
|
"valid_targets_mean": 3437.4,
|
||
|
|
"valid_targets_min": 1395
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6108870967741935,
|
||
|
|
"grad_norm": 0.617467386537604,
|
||
|
|
"learning_rate": 3.161205775941442e-05,
|
||
|
|
"loss": 0.2619,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13246780633926392,
|
||
|
|
"step": 1295,
|
||
|
|
"valid_targets_mean": 2994.2,
|
||
|
|
"valid_targets_min": 363
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.620967741935484,
|
||
|
|
"grad_norm": 0.5424119141053696,
|
||
|
|
"learning_rate": 3.153003412054439e-05,
|
||
|
|
"loss": 0.2672,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12123814225196838,
|
||
|
|
"step": 1300,
|
||
|
|
"valid_targets_mean": 3541.8,
|
||
|
|
"valid_targets_min": 1480
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.631048387096774,
|
||
|
|
"grad_norm": 0.5348548260735864,
|
||
|
|
"learning_rate": 3.14477189757529e-05,
|
||
|
|
"loss": 0.2536,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10437840223312378,
|
||
|
|
"step": 1305,
|
||
|
|
"valid_targets_mean": 3128.1,
|
||
|
|
"valid_targets_min": 1493
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6411290322580645,
|
||
|
|
"grad_norm": 0.6064463368832572,
|
||
|
|
"learning_rate": 3.136511440615715e-05,
|
||
|
|
"loss": 0.2566,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09684427827596664,
|
||
|
|
"step": 1310,
|
||
|
|
"valid_targets_mean": 2283.6,
|
||
|
|
"valid_targets_min": 1151
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.651209677419355,
|
||
|
|
"grad_norm": 0.5526449220235825,
|
||
|
|
"learning_rate": 3.128222250019164e-05,
|
||
|
|
"loss": 0.2554,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14107558131217957,
|
||
|
|
"step": 1315,
|
||
|
|
"valid_targets_mean": 3755.9,
|
||
|
|
"valid_targets_min": 1549
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.661290322580645,
|
||
|
|
"grad_norm": 0.5543380832321381,
|
||
|
|
"learning_rate": 3.119904535355542e-05,
|
||
|
|
"loss": 0.2699,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15373364090919495,
|
||
|
|
"step": 1320,
|
||
|
|
"valid_targets_mean": 3278.4,
|
||
|
|
"valid_targets_min": 1909
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6713709677419355,
|
||
|
|
"grad_norm": 0.7402856431959683,
|
||
|
|
"learning_rate": 3.1115585069159064e-05,
|
||
|
|
"loss": 0.2576,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14970913529396057,
|
||
|
|
"step": 1325,
|
||
|
|
"valid_targets_mean": 2711.2,
|
||
|
|
"valid_targets_min": 391
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.681451612903226,
|
||
|
|
"grad_norm": 0.5964117860025188,
|
||
|
|
"learning_rate": 3.103184375707153e-05,
|
||
|
|
"loss": 0.2616,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11866350471973419,
|
||
|
|
"step": 1330,
|
||
|
|
"valid_targets_mean": 3497.8,
|
||
|
|
"valid_targets_min": 1379
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.691532258064516,
|
||
|
|
"grad_norm": 0.5651922938953524,
|
||
|
|
"learning_rate": 3.0947823534466824e-05,
|
||
|
|
"loss": 0.2627,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09416229277849197,
|
||
|
|
"step": 1335,
|
||
|
|
"valid_targets_mean": 2966.6,
|
||
|
|
"valid_targets_min": 359
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7016129032258065,
|
||
|
|
"grad_norm": 0.5810423389474473,
|
||
|
|
"learning_rate": 3.0863526525570416e-05,
|
||
|
|
"loss": 0.256,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15101176500320435,
|
||
|
|
"step": 1340,
|
||
|
|
"valid_targets_mean": 3474.4,
|
||
|
|
"valid_targets_min": 1651
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.711693548387097,
|
||
|
|
"grad_norm": 0.6129936176857649,
|
||
|
|
"learning_rate": 3.0778954861605606e-05,
|
||
|
|
"loss": 0.2757,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1488267332315445,
|
||
|
|
"step": 1345,
|
||
|
|
"valid_targets_mean": 3804.8,
|
||
|
|
"valid_targets_min": 1903
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.721774193548387,
|
||
|
|
"grad_norm": 0.5390254074552373,
|
||
|
|
"learning_rate": 3.069411068073961e-05,
|
||
|
|
"loss": 0.2639,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15416580438613892,
|
||
|
|
"step": 1350,
|
||
|
|
"valid_targets_mean": 4647.1,
|
||
|
|
"valid_targets_min": 3126
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7318548387096775,
|
||
|
|
"grad_norm": 0.5862240616844475,
|
||
|
|
"learning_rate": 3.060899612802947e-05,
|
||
|
|
"loss": 0.2571,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11063511669635773,
|
||
|
|
"step": 1355,
|
||
|
|
"valid_targets_mean": 2562.8,
|
||
|
|
"valid_targets_min": 757
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.741935483870968,
|
||
|
|
"grad_norm": 0.6009282129840543,
|
||
|
|
"learning_rate": 3.05236133553679e-05,
|
||
|
|
"loss": 0.2758,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10215406864881516,
|
||
|
|
"step": 1360,
|
||
|
|
"valid_targets_mean": 2616.4,
|
||
|
|
"valid_targets_min": 1540
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.752016129032258,
|
||
|
|
"grad_norm": 0.5019479621411045,
|
||
|
|
"learning_rate": 3.0437964521428783e-05,
|
||
|
|
"loss": 0.2482,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09111334383487701,
|
||
|
|
"step": 1365,
|
||
|
|
"valid_targets_mean": 3227.1,
|
||
|
|
"valid_targets_min": 1062
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7620967741935485,
|
||
|
|
"grad_norm": 0.6039902083563039,
|
||
|
|
"learning_rate": 3.0352051791612686e-05,
|
||
|
|
"loss": 0.2524,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12219463288784027,
|
||
|
|
"step": 1370,
|
||
|
|
"valid_targets_mean": 3549.9,
|
||
|
|
"valid_targets_min": 460
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.772177419354839,
|
||
|
|
"grad_norm": 0.6264619916598958,
|
||
|
|
"learning_rate": 3.0265877337992056e-05,
|
||
|
|
"loss": 0.2423,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09153392910957336,
|
||
|
|
"step": 1375,
|
||
|
|
"valid_targets_mean": 2143.8,
|
||
|
|
"valid_targets_min": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.782258064516129,
|
||
|
|
"grad_norm": 0.5885020761203044,
|
||
|
|
"learning_rate": 3.017944333925633e-05,
|
||
|
|
"loss": 0.2543,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09630314260721207,
|
||
|
|
"step": 1380,
|
||
|
|
"valid_targets_mean": 2232.8,
|
||
|
|
"valid_targets_min": 399
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7923387096774195,
|
||
|
|
"grad_norm": 0.6414281985977918,
|
||
|
|
"learning_rate": 3.0092751980656837e-05,
|
||
|
|
"loss": 0.2524,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13684843480587006,
|
||
|
|
"step": 1385,
|
||
|
|
"valid_targets_mean": 3041.1,
|
||
|
|
"valid_targets_min": 1113
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8024193548387095,
|
||
|
|
"grad_norm": 0.6645335019906936,
|
||
|
|
"learning_rate": 3.0005805453951567e-05,
|
||
|
|
"loss": 0.25,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11580963432788849,
|
||
|
|
"step": 1390,
|
||
|
|
"valid_targets_mean": 2602.4,
|
||
|
|
"valid_targets_min": 941
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8125,
|
||
|
|
"grad_norm": 0.6050375320287946,
|
||
|
|
"learning_rate": 2.9918605957349745e-05,
|
||
|
|
"loss": 0.2736,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15735115110874176,
|
||
|
|
"step": 1395,
|
||
|
|
"valid_targets_mean": 3238.0,
|
||
|
|
"valid_targets_min": 1315
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8225806451612905,
|
||
|
|
"grad_norm": 0.5540619168976046,
|
||
|
|
"learning_rate": 2.9831155695456263e-05,
|
||
|
|
"loss": 0.2431,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13711747527122498,
|
||
|
|
"step": 1400,
|
||
|
|
"valid_targets_mean": 3780.6,
|
||
|
|
"valid_targets_min": 1758
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8326612903225805,
|
||
|
|
"grad_norm": 0.5744608291282568,
|
||
|
|
"learning_rate": 2.9743456879215934e-05,
|
||
|
|
"loss": 0.263,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16890239715576172,
|
||
|
|
"step": 1405,
|
||
|
|
"valid_targets_mean": 3838.4,
|
||
|
|
"valid_targets_min": 911
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.842741935483871,
|
||
|
|
"grad_norm": 0.5745835883100003,
|
||
|
|
"learning_rate": 2.9655511725857607e-05,
|
||
|
|
"loss": 0.258,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1245071217417717,
|
||
|
|
"step": 1410,
|
||
|
|
"valid_targets_mean": 3271.0,
|
||
|
|
"valid_targets_min": 1670
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.852822580645161,
|
||
|
|
"grad_norm": 0.5475488920799221,
|
||
|
|
"learning_rate": 2.9567322458838097e-05,
|
||
|
|
"loss": 0.2544,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11185264587402344,
|
||
|
|
"step": 1415,
|
||
|
|
"valid_targets_mean": 3503.8,
|
||
|
|
"valid_targets_min": 2532
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8629032258064515,
|
||
|
|
"grad_norm": 0.5577848226800461,
|
||
|
|
"learning_rate": 2.9478891307785974e-05,
|
||
|
|
"loss": 0.2653,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10932277143001556,
|
||
|
|
"step": 1420,
|
||
|
|
"valid_targets_mean": 3250.8,
|
||
|
|
"valid_targets_min": 1856
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.872983870967742,
|
||
|
|
"grad_norm": 0.5245535685909258,
|
||
|
|
"learning_rate": 2.9390220508445215e-05,
|
||
|
|
"loss": 0.2521,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.135254368185997,
|
||
|
|
"step": 1425,
|
||
|
|
"valid_targets_mean": 4234.2,
|
||
|
|
"valid_targets_min": 2131
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.883064516129032,
|
||
|
|
"grad_norm": 0.6192900329890271,
|
||
|
|
"learning_rate": 2.9301312302618634e-05,
|
||
|
|
"loss": 0.2996,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18384113907814026,
|
||
|
|
"step": 1430,
|
||
|
|
"valid_targets_mean": 3253.5,
|
||
|
|
"valid_targets_min": 1519
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8931451612903225,
|
||
|
|
"grad_norm": 0.5954676115372128,
|
||
|
|
"learning_rate": 2.9212168938111243e-05,
|
||
|
|
"loss": 0.28,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21537812054157257,
|
||
|
|
"step": 1435,
|
||
|
|
"valid_targets_mean": 3931.5,
|
||
|
|
"valid_targets_min": 1257
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.903225806451613,
|
||
|
|
"grad_norm": 0.6135352867407852,
|
||
|
|
"learning_rate": 2.912279266867341e-05,
|
||
|
|
"loss": 0.2586,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11614513397216797,
|
||
|
|
"step": 1440,
|
||
|
|
"valid_targets_mean": 2637.1,
|
||
|
|
"valid_targets_min": 1830
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.913306451612903,
|
||
|
|
"grad_norm": 0.5438776470905065,
|
||
|
|
"learning_rate": 2.9033185753943868e-05,
|
||
|
|
"loss": 0.2743,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18999066948890686,
|
||
|
|
"step": 1445,
|
||
|
|
"valid_targets_mean": 4511.1,
|
||
|
|
"valid_targets_min": 1841
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9233870967741935,
|
||
|
|
"grad_norm": 0.544867986133032,
|
||
|
|
"learning_rate": 2.894335045939261e-05,
|
||
|
|
"loss": 0.2657,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.21123896539211273,
|
||
|
|
"step": 1450,
|
||
|
|
"valid_targets_mean": 4936.5,
|
||
|
|
"valid_targets_min": 1545
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.933467741935484,
|
||
|
|
"grad_norm": 0.6081189521438323,
|
||
|
|
"learning_rate": 2.8853289056263595e-05,
|
||
|
|
"loss": 0.2588,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11442147195339203,
|
||
|
|
"step": 1455,
|
||
|
|
"valid_targets_mean": 2803.4,
|
||
|
|
"valid_targets_min": 1462
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.943548387096774,
|
||
|
|
"grad_norm": 0.6814139024869164,
|
||
|
|
"learning_rate": 2.8763003821517324e-05,
|
||
|
|
"loss": 0.2555,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10975331813097,
|
||
|
|
"step": 1460,
|
||
|
|
"valid_targets_mean": 2592.1,
|
||
|
|
"valid_targets_min": 1288
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9536290322580645,
|
||
|
|
"grad_norm": 0.589340005352357,
|
||
|
|
"learning_rate": 2.8672497037773284e-05,
|
||
|
|
"loss": 0.2732,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12053504586219788,
|
||
|
|
"step": 1465,
|
||
|
|
"valid_targets_mean": 3578.4,
|
||
|
|
"valid_targets_min": 1473
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.963709677419355,
|
||
|
|
"grad_norm": 0.5694348566111326,
|
||
|
|
"learning_rate": 2.8581770993252226e-05,
|
||
|
|
"loss": 0.2822,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1630900353193283,
|
||
|
|
"step": 1470,
|
||
|
|
"valid_targets_mean": 4398.6,
|
||
|
|
"valid_targets_min": 1284
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.973790322580645,
|
||
|
|
"grad_norm": 0.5800833368540554,
|
||
|
|
"learning_rate": 2.849082798171833e-05,
|
||
|
|
"loss": 0.2882,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10425388067960739,
|
||
|
|
"step": 1475,
|
||
|
|
"valid_targets_mean": 3350.4,
|
||
|
|
"valid_targets_min": 1584
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9838709677419355,
|
||
|
|
"grad_norm": 0.5511529492499462,
|
||
|
|
"learning_rate": 2.8399670302421198e-05,
|
||
|
|
"loss": 0.2599,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0935806930065155,
|
||
|
|
"step": 1480,
|
||
|
|
"valid_targets_mean": 2860.0,
|
||
|
|
"valid_targets_min": 1056
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.993951612903226,
|
||
|
|
"grad_norm": 0.5545792368426485,
|
||
|
|
"learning_rate": 2.8308300260037734e-05,
|
||
|
|
"loss": 0.2455,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12762320041656494,
|
||
|
|
"step": 1485,
|
||
|
|
"valid_targets_mean": 3613.5,
|
||
|
|
"valid_targets_min": 2010
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.004032258064516,
|
||
|
|
"grad_norm": 0.6196980261026537,
|
||
|
|
"learning_rate": 2.821672016461386e-05,
|
||
|
|
"loss": 0.245,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10218601673841476,
|
||
|
|
"step": 1490,
|
||
|
|
"valid_targets_mean": 2679.9,
|
||
|
|
"valid_targets_min": 426
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0141129032258065,
|
||
|
|
"grad_norm": 0.6480683189464084,
|
||
|
|
"learning_rate": 2.812493233150614e-05,
|
||
|
|
"loss": 0.2443,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09944482892751694,
|
||
|
|
"step": 1495,
|
||
|
|
"valid_targets_mean": 3006.2,
|
||
|
|
"valid_targets_min": 1910
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.024193548387097,
|
||
|
|
"grad_norm": 0.568616906010806,
|
||
|
|
"learning_rate": 2.803293908132322e-05,
|
||
|
|
"loss": 0.228,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11579032242298126,
|
||
|
|
"step": 1500,
|
||
|
|
"valid_targets_mean": 3835.2,
|
||
|
|
"valid_targets_min": 1543
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.034274193548387,
|
||
|
|
"grad_norm": 0.5583673452039545,
|
||
|
|
"learning_rate": 2.7940742739867165e-05,
|
||
|
|
"loss": 0.2066,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09896944463253021,
|
||
|
|
"step": 1505,
|
||
|
|
"valid_targets_mean": 3283.8,
|
||
|
|
"valid_targets_min": 1232
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0443548387096775,
|
||
|
|
"grad_norm": 0.6445914246590435,
|
||
|
|
"learning_rate": 2.7848345638074644e-05,
|
||
|
|
"loss": 0.2275,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10756675899028778,
|
||
|
|
"step": 1510,
|
||
|
|
"valid_targets_mean": 2379.1,
|
||
|
|
"valid_targets_min": 395
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0544354838709675,
|
||
|
|
"grad_norm": 0.5768530851180089,
|
||
|
|
"learning_rate": 2.7755750111958043e-05,
|
||
|
|
"loss": 0.2045,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10050986707210541,
|
||
|
|
"step": 1515,
|
||
|
|
"valid_targets_mean": 3304.0,
|
||
|
|
"valid_targets_min": 2431
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.064516129032258,
|
||
|
|
"grad_norm": 0.5912909450401477,
|
||
|
|
"learning_rate": 2.766295850254635e-05,
|
||
|
|
"loss": 0.2232,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12583386898040771,
|
||
|
|
"step": 1520,
|
||
|
|
"valid_targets_mean": 3370.2,
|
||
|
|
"valid_targets_min": 1574
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0745967741935485,
|
||
|
|
"grad_norm": 0.5779722774640857,
|
||
|
|
"learning_rate": 2.7569973155825983e-05,
|
||
|
|
"loss": 0.2285,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13876058161258698,
|
||
|
|
"step": 1525,
|
||
|
|
"valid_targets_mean": 3821.9,
|
||
|
|
"valid_targets_min": 1758
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0846774193548385,
|
||
|
|
"grad_norm": 0.6084551162321066,
|
||
|
|
"learning_rate": 2.7476796422681516e-05,
|
||
|
|
"loss": 0.2329,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08827266842126846,
|
||
|
|
"step": 1530,
|
||
|
|
"valid_targets_mean": 2669.5,
|
||
|
|
"valid_targets_min": 370
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.094758064516129,
|
||
|
|
"grad_norm": 0.5753793470137871,
|
||
|
|
"learning_rate": 2.738343065883619e-05,
|
||
|
|
"loss": 0.2412,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08057849109172821,
|
||
|
|
"step": 1535,
|
||
|
|
"valid_targets_mean": 3226.8,
|
||
|
|
"valid_targets_min": 1647
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1048387096774195,
|
||
|
|
"grad_norm": 0.6031701385073922,
|
||
|
|
"learning_rate": 2.7289878224792398e-05,
|
||
|
|
"loss": 0.2278,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08894006162881851,
|
||
|
|
"step": 1540,
|
||
|
|
"valid_targets_mean": 2933.9,
|
||
|
|
"valid_targets_min": 1442
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1149193548387095,
|
||
|
|
"grad_norm": 0.6582341729338994,
|
||
|
|
"learning_rate": 2.7196141485771966e-05,
|
||
|
|
"loss": 0.2347,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13643427193164825,
|
||
|
|
"step": 1545,
|
||
|
|
"valid_targets_mean": 2450.0,
|
||
|
|
"valid_targets_min": 395
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.125,
|
||
|
|
"grad_norm": 0.6016298129552541,
|
||
|
|
"learning_rate": 2.7102222811656392e-05,
|
||
|
|
"loss": 0.2529,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1077808290719986,
|
||
|
|
"step": 1550,
|
||
|
|
"valid_targets_mean": 3460.5,
|
||
|
|
"valid_targets_min": 1485
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1350806451612905,
|
||
|
|
"grad_norm": 0.5861031124344813,
|
||
|
|
"learning_rate": 2.700812457692691e-05,
|
||
|
|
"loss": 0.2123,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09558555483818054,
|
||
|
|
"step": 1555,
|
||
|
|
"valid_targets_mean": 3426.8,
|
||
|
|
"valid_targets_min": 1241
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1451612903225805,
|
||
|
|
"grad_norm": 0.6572670415510821,
|
||
|
|
"learning_rate": 2.6913849160604467e-05,
|
||
|
|
"loss": 0.2361,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09726105630397797,
|
||
|
|
"step": 1560,
|
||
|
|
"valid_targets_mean": 2654.1,
|
||
|
|
"valid_targets_min": 1482
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.155241935483871,
|
||
|
|
"grad_norm": 0.6716333136075975,
|
||
|
|
"learning_rate": 2.6819398946189555e-05,
|
||
|
|
"loss": 0.2395,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1290110945701599,
|
||
|
|
"step": 1565,
|
||
|
|
"valid_targets_mean": 2667.6,
|
||
|
|
"valid_targets_min": 329
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1653225806451615,
|
||
|
|
"grad_norm": 0.5614932563669551,
|
||
|
|
"learning_rate": 2.6724776321601984e-05,
|
||
|
|
"loss": 0.2439,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12074358761310577,
|
||
|
|
"step": 1570,
|
||
|
|
"valid_targets_mean": 3776.2,
|
||
|
|
"valid_targets_min": 1720
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1754032258064515,
|
||
|
|
"grad_norm": 0.554963070526007,
|
||
|
|
"learning_rate": 2.6629983679120474e-05,
|
||
|
|
"loss": 0.2256,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10534197837114334,
|
||
|
|
"step": 1575,
|
||
|
|
"valid_targets_mean": 4121.2,
|
||
|
|
"valid_targets_min": 2721
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.185483870967742,
|
||
|
|
"grad_norm": 0.6047252989899893,
|
||
|
|
"learning_rate": 2.65350234153222e-05,
|
||
|
|
"loss": 0.2386,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1304488331079483,
|
||
|
|
"step": 1580,
|
||
|
|
"valid_targets_mean": 3193.6,
|
||
|
|
"valid_targets_min": 1151
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.195564516129032,
|
||
|
|
"grad_norm": 0.5598425771271849,
|
||
|
|
"learning_rate": 2.6439897931022192e-05,
|
||
|
|
"loss": 0.2296,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0967915952205658,
|
||
|
|
"step": 1585,
|
||
|
|
"valid_targets_mean": 3729.4,
|
||
|
|
"valid_targets_min": 356
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2056451612903225,
|
||
|
|
"grad_norm": 0.6425736188896479,
|
||
|
|
"learning_rate": 2.6344609631212645e-05,
|
||
|
|
"loss": 0.243,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16863906383514404,
|
||
|
|
"step": 1590,
|
||
|
|
"valid_targets_mean": 3783.5,
|
||
|
|
"valid_targets_min": 1342
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.215725806451613,
|
||
|
|
"grad_norm": 0.8494800208578204,
|
||
|
|
"learning_rate": 2.6249160925002083e-05,
|
||
|
|
"loss": 0.235,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13157713413238525,
|
||
|
|
"step": 1595,
|
||
|
|
"valid_targets_mean": 3489.2,
|
||
|
|
"valid_targets_min": 1602
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.225806451612903,
|
||
|
|
"grad_norm": 0.574931407354099,
|
||
|
|
"learning_rate": 2.6153554225554497e-05,
|
||
|
|
"loss": 0.2274,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09766871482133865,
|
||
|
|
"step": 1600,
|
||
|
|
"valid_targets_mean": 2872.1,
|
||
|
|
"valid_targets_min": 360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2358870967741935,
|
||
|
|
"grad_norm": 0.6105844603302589,
|
||
|
|
"learning_rate": 2.6057791950028303e-05,
|
||
|
|
"loss": 0.2362,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08750882744789124,
|
||
|
|
"step": 1605,
|
||
|
|
"valid_targets_mean": 2416.2,
|
||
|
|
"valid_targets_min": 1062
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.245967741935484,
|
||
|
|
"grad_norm": 0.5383359600827379,
|
||
|
|
"learning_rate": 2.5961876519515238e-05,
|
||
|
|
"loss": 0.2412,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11167649924755096,
|
||
|
|
"step": 1610,
|
||
|
|
"valid_targets_mean": 3386.5,
|
||
|
|
"valid_targets_min": 1378
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.256048387096774,
|
||
|
|
"grad_norm": 0.6806866407417329,
|
||
|
|
"learning_rate": 2.5865810358979146e-05,
|
||
|
|
"loss": 0.2656,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.2206556797027588,
|
||
|
|
"step": 1615,
|
||
|
|
"valid_targets_mean": 4999.8,
|
||
|
|
"valid_targets_min": 1266
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2661290322580645,
|
||
|
|
"grad_norm": 0.5535494321375118,
|
||
|
|
"learning_rate": 2.5769595897194693e-05,
|
||
|
|
"loss": 0.2204,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08556301146745682,
|
||
|
|
"step": 1620,
|
||
|
|
"valid_targets_mean": 3029.5,
|
||
|
|
"valid_targets_min": 403
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.276209677419355,
|
||
|
|
"grad_norm": 0.5759204508775331,
|
||
|
|
"learning_rate": 2.5673235566685935e-05,
|
||
|
|
"loss": 0.2155,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09597568213939667,
|
||
|
|
"step": 1625,
|
||
|
|
"valid_targets_mean": 3134.8,
|
||
|
|
"valid_targets_min": 1350
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.286290322580645,
|
||
|
|
"grad_norm": 0.6420977509076261,
|
||
|
|
"learning_rate": 2.557673180366482e-05,
|
||
|
|
"loss": 0.2226,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10002493858337402,
|
||
|
|
"step": 1630,
|
||
|
|
"valid_targets_mean": 2703.8,
|
||
|
|
"valid_targets_min": 1334
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.2963709677419355,
|
||
|
|
"grad_norm": 0.5696214702956488,
|
||
|
|
"learning_rate": 2.5480087047969605e-05,
|
||
|
|
"loss": 0.224,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07859362661838531,
|
||
|
|
"step": 1635,
|
||
|
|
"valid_targets_mean": 2919.1,
|
||
|
|
"valid_targets_min": 387
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.306451612903226,
|
||
|
|
"grad_norm": 0.5852826845010795,
|
||
|
|
"learning_rate": 2.538330374300317e-05,
|
||
|
|
"loss": 0.2309,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13316066563129425,
|
||
|
|
"step": 1640,
|
||
|
|
"valid_targets_mean": 4385.6,
|
||
|
|
"valid_targets_min": 2018
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.316532258064516,
|
||
|
|
"grad_norm": 0.6177874966838176,
|
||
|
|
"learning_rate": 2.5286384335671237e-05,
|
||
|
|
"loss": 0.2518,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09096910804510117,
|
||
|
|
"step": 1645,
|
||
|
|
"valid_targets_mean": 2630.8,
|
||
|
|
"valid_targets_min": 873
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3266129032258065,
|
||
|
|
"grad_norm": 0.6294526575969375,
|
||
|
|
"learning_rate": 2.5189331276320515e-05,
|
||
|
|
"loss": 0.2363,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12998098134994507,
|
||
|
|
"step": 1650,
|
||
|
|
"valid_targets_mean": 3522.4,
|
||
|
|
"valid_targets_min": 1772
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.336693548387097,
|
||
|
|
"grad_norm": 0.6325054232663175,
|
||
|
|
"learning_rate": 2.509214701867675e-05,
|
||
|
|
"loss": 0.234,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10132864117622375,
|
||
|
|
"step": 1655,
|
||
|
|
"valid_targets_mean": 2656.9,
|
||
|
|
"valid_targets_min": 914
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.346774193548387,
|
||
|
|
"grad_norm": 0.5977821999065315,
|
||
|
|
"learning_rate": 2.499483401978266e-05,
|
||
|
|
"loss": 0.2209,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11908560991287231,
|
||
|
|
"step": 1660,
|
||
|
|
"valid_targets_mean": 3584.4,
|
||
|
|
"valid_targets_min": 1926
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3568548387096775,
|
||
|
|
"grad_norm": 0.6022539036633235,
|
||
|
|
"learning_rate": 2.4897394739935877e-05,
|
||
|
|
"loss": 0.2418,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09482988715171814,
|
||
|
|
"step": 1665,
|
||
|
|
"valid_targets_mean": 2684.9,
|
||
|
|
"valid_targets_min": 349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.366935483870968,
|
||
|
|
"grad_norm": 0.665981082259953,
|
||
|
|
"learning_rate": 2.479983164262667e-05,
|
||
|
|
"loss": 0.2126,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11829803884029388,
|
||
|
|
"step": 1670,
|
||
|
|
"valid_targets_mean": 2976.6,
|
||
|
|
"valid_targets_min": 1763
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.377016129032258,
|
||
|
|
"grad_norm": 0.5640253887796919,
|
||
|
|
"learning_rate": 2.470214719447574e-05,
|
||
|
|
"loss": 0.2423,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13566476106643677,
|
||
|
|
"step": 1675,
|
||
|
|
"valid_targets_mean": 4445.0,
|
||
|
|
"valid_targets_min": 1399
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3870967741935485,
|
||
|
|
"grad_norm": 0.5930299898066264,
|
||
|
|
"learning_rate": 2.460434386517177e-05,
|
||
|
|
"loss": 0.2388,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12184359133243561,
|
||
|
|
"step": 1680,
|
||
|
|
"valid_targets_mean": 3574.5,
|
||
|
|
"valid_targets_min": 1833
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.3971774193548385,
|
||
|
|
"grad_norm": 0.5922228986584855,
|
||
|
|
"learning_rate": 2.4506424127409068e-05,
|
||
|
|
"loss": 0.2264,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07741091400384903,
|
||
|
|
"step": 1685,
|
||
|
|
"valid_targets_mean": 2708.0,
|
||
|
|
"valid_targets_min": 487
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.407258064516129,
|
||
|
|
"grad_norm": 0.6275564285618686,
|
||
|
|
"learning_rate": 2.4408390456825005e-05,
|
||
|
|
"loss": 0.2295,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0991937592625618,
|
||
|
|
"step": 1690,
|
||
|
|
"valid_targets_mean": 3040.6,
|
||
|
|
"valid_targets_min": 1462
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4173387096774195,
|
||
|
|
"grad_norm": 0.5420900311230281,
|
||
|
|
"learning_rate": 2.4310245331937453e-05,
|
||
|
|
"loss": 0.2286,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1279710829257965,
|
||
|
|
"step": 1695,
|
||
|
|
"valid_targets_mean": 4013.6,
|
||
|
|
"valid_targets_min": 2154
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4274193548387095,
|
||
|
|
"grad_norm": 0.6262040533793154,
|
||
|
|
"learning_rate": 2.421199123408207e-05,
|
||
|
|
"loss": 0.2248,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0829395204782486,
|
||
|
|
"step": 1700,
|
||
|
|
"valid_targets_mean": 2760.6,
|
||
|
|
"valid_targets_min": 1164
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4375,
|
||
|
|
"grad_norm": 0.5464495205298482,
|
||
|
|
"learning_rate": 2.4113630647349647e-05,
|
||
|
|
"loss": 0.2392,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12593115866184235,
|
||
|
|
"step": 1705,
|
||
|
|
"valid_targets_mean": 3775.5,
|
||
|
|
"valid_targets_min": 1140
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4475806451612905,
|
||
|
|
"grad_norm": 0.7109547985857375,
|
||
|
|
"learning_rate": 2.401516605852323e-05,
|
||
|
|
"loss": 0.2544,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09599505364894867,
|
||
|
|
"step": 1710,
|
||
|
|
"valid_targets_mean": 2103.2,
|
||
|
|
"valid_targets_min": 440
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4576612903225805,
|
||
|
|
"grad_norm": 0.6425695725469902,
|
||
|
|
"learning_rate": 2.3916599957015284e-05,
|
||
|
|
"loss": 0.2114,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09730251133441925,
|
||
|
|
"step": 1715,
|
||
|
|
"valid_targets_mean": 2826.1,
|
||
|
|
"valid_targets_min": 1664
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.467741935483871,
|
||
|
|
"grad_norm": 1.6719140053069148,
|
||
|
|
"learning_rate": 2.3817934834804754e-05,
|
||
|
|
"loss": 0.2289,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10053573548793793,
|
||
|
|
"step": 1720,
|
||
|
|
"valid_targets_mean": 3630.6,
|
||
|
|
"valid_targets_min": 1767
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4778225806451615,
|
||
|
|
"grad_norm": 0.6349187273410546,
|
||
|
|
"learning_rate": 2.371917318637406e-05,
|
||
|
|
"loss": 0.2634,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1196623146533966,
|
||
|
|
"step": 1725,
|
||
|
|
"valid_targets_mean": 2914.8,
|
||
|
|
"valid_targets_min": 1095
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.4879032258064515,
|
||
|
|
"grad_norm": 0.5642721667982462,
|
||
|
|
"learning_rate": 2.3620317508646014e-05,
|
||
|
|
"loss": 0.2225,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08153025805950165,
|
||
|
|
"step": 1730,
|
||
|
|
"valid_targets_mean": 2913.2,
|
||
|
|
"valid_targets_min": 1483
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.497983870967742,
|
||
|
|
"grad_norm": 0.6427403893933606,
|
||
|
|
"learning_rate": 2.3521370300920726e-05,
|
||
|
|
"loss": 0.2382,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09541168808937073,
|
||
|
|
"step": 1735,
|
||
|
|
"valid_targets_mean": 2953.4,
|
||
|
|
"valid_targets_min": 1314
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.508064516129032,
|
||
|
|
"grad_norm": 0.6565712750827525,
|
||
|
|
"learning_rate": 2.3422334064812383e-05,
|
||
|
|
"loss": 0.2368,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09855005145072937,
|
||
|
|
"step": 1740,
|
||
|
|
"valid_targets_mean": 3127.4,
|
||
|
|
"valid_targets_min": 1669
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5181451612903225,
|
||
|
|
"grad_norm": 0.6693692674056112,
|
||
|
|
"learning_rate": 2.3323211304186012e-05,
|
||
|
|
"loss": 0.239,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09766016900539398,
|
||
|
|
"step": 1745,
|
||
|
|
"valid_targets_mean": 2913.9,
|
||
|
|
"valid_targets_min": 1335
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.528225806451613,
|
||
|
|
"grad_norm": 0.6046941165705318,
|
||
|
|
"learning_rate": 2.3224004525094194e-05,
|
||
|
|
"loss": 0.2406,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12182774394750595,
|
||
|
|
"step": 1750,
|
||
|
|
"valid_targets_mean": 3334.9,
|
||
|
|
"valid_targets_min": 1643
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.538306451612903,
|
||
|
|
"grad_norm": 0.652594536790013,
|
||
|
|
"learning_rate": 2.3124716235713676e-05,
|
||
|
|
"loss": 0.2102,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08517846465110779,
|
||
|
|
"step": 1755,
|
||
|
|
"valid_targets_mean": 4519.8,
|
||
|
|
"valid_targets_min": 2051
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5483870967741935,
|
||
|
|
"grad_norm": 0.6855407204775368,
|
||
|
|
"learning_rate": 2.302534894628199e-05,
|
||
|
|
"loss": 0.2385,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10159999877214432,
|
||
|
|
"step": 1760,
|
||
|
|
"valid_targets_mean": 2390.9,
|
||
|
|
"valid_targets_min": 421
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.558467741935484,
|
||
|
|
"grad_norm": 0.655770211524336,
|
||
|
|
"learning_rate": 2.2925905169033946e-05,
|
||
|
|
"loss": 0.2489,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1329360008239746,
|
||
|
|
"step": 1765,
|
||
|
|
"valid_targets_mean": 3074.8,
|
||
|
|
"valid_targets_min": 1312
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.568548387096774,
|
||
|
|
"grad_norm": 0.5830849028335179,
|
||
|
|
"learning_rate": 2.282638741813818e-05,
|
||
|
|
"loss": 0.2579,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10126137733459473,
|
||
|
|
"step": 1770,
|
||
|
|
"valid_targets_mean": 3158.2,
|
||
|
|
"valid_targets_min": 2155
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.5786290322580645,
|
||
|
|
"grad_norm": 0.7053507701211764,
|
||
|
|
"learning_rate": 2.2726798209633514e-05,
|
||
|
|
"loss": 0.2703,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1637437492609024,
|
||
|
|
"step": 1775,
|
||
|
|
"valid_targets_mean": 3134.9,
|
||
|
|
"valid_targets_min": 1244
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.588709677419355,
|
||
|
|
"grad_norm": 0.6397603823448965,
|
||
|
|
"learning_rate": 2.262714006136542e-05,
|
||
|
|
"loss": 0.2566,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17420338094234467,
|
||
|
|
"step": 1780,
|
||
|
|
"valid_targets_mean": 3090.2,
|
||
|
|
"valid_targets_min": 370
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.598790322580645,
|
||
|
|
"grad_norm": 0.5281054185200672,
|
||
|
|
"learning_rate": 2.2527415492922287e-05,
|
||
|
|
"loss": 0.2223,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08755169063806534,
|
||
|
|
"step": 1785,
|
||
|
|
"valid_targets_mean": 4280.8,
|
||
|
|
"valid_targets_min": 1825
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6088709677419355,
|
||
|
|
"grad_norm": 0.5641139956036748,
|
||
|
|
"learning_rate": 2.2427627025571795e-05,
|
||
|
|
"loss": 0.2242,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07409320771694183,
|
||
|
|
"step": 1790,
|
||
|
|
"valid_targets_mean": 2658.5,
|
||
|
|
"valid_targets_min": 1393
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.618951612903226,
|
||
|
|
"grad_norm": 0.6577233492704598,
|
||
|
|
"learning_rate": 2.2327777182197116e-05,
|
||
|
|
"loss": 0.2147,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16011226177215576,
|
||
|
|
"step": 1795,
|
||
|
|
"valid_targets_mean": 3605.6,
|
||
|
|
"valid_targets_min": 2221
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.629032258064516,
|
||
|
|
"grad_norm": 0.6581332141707709,
|
||
|
|
"learning_rate": 2.2227868487233165e-05,
|
||
|
|
"loss": 0.2166,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12629841268062592,
|
||
|
|
"step": 1800,
|
||
|
|
"valid_targets_mean": 3402.9,
|
||
|
|
"valid_targets_min": 277
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6391129032258065,
|
||
|
|
"grad_norm": 0.5530267785490928,
|
||
|
|
"learning_rate": 2.2127903466602736e-05,
|
||
|
|
"loss": 0.2433,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08644229918718338,
|
||
|
|
"step": 1805,
|
||
|
|
"valid_targets_mean": 2980.9,
|
||
|
|
"valid_targets_min": 893
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.649193548387097,
|
||
|
|
"grad_norm": 0.6168149871774915,
|
||
|
|
"learning_rate": 2.2027884647652704e-05,
|
||
|
|
"loss": 0.2483,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08936543762683868,
|
||
|
|
"step": 1810,
|
||
|
|
"valid_targets_mean": 2522.4,
|
||
|
|
"valid_targets_min": 1817
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.659274193548387,
|
||
|
|
"grad_norm": 0.6039554586395655,
|
||
|
|
"learning_rate": 2.1927814559090052e-05,
|
||
|
|
"loss": 0.2311,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11215218901634216,
|
||
|
|
"step": 1815,
|
||
|
|
"valid_targets_mean": 3582.6,
|
||
|
|
"valid_targets_min": 2162
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6693548387096775,
|
||
|
|
"grad_norm": 0.5919572216362302,
|
||
|
|
"learning_rate": 2.1827695730917995e-05,
|
||
|
|
"loss": 0.2661,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10347072035074234,
|
||
|
|
"step": 1820,
|
||
|
|
"valid_targets_mean": 3614.0,
|
||
|
|
"valid_targets_min": 2330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.679435483870968,
|
||
|
|
"grad_norm": 0.6002963488223565,
|
||
|
|
"learning_rate": 2.1727530694371993e-05,
|
||
|
|
"loss": 0.24,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12887421250343323,
|
||
|
|
"step": 1825,
|
||
|
|
"valid_targets_mean": 3667.4,
|
||
|
|
"valid_targets_min": 2244
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.689516129032258,
|
||
|
|
"grad_norm": 0.585802064588058,
|
||
|
|
"learning_rate": 2.162732198185578e-05,
|
||
|
|
"loss": 0.2423,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10119467228651047,
|
||
|
|
"step": 1830,
|
||
|
|
"valid_targets_mean": 3594.5,
|
||
|
|
"valid_targets_min": 1639
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6995967741935485,
|
||
|
|
"grad_norm": 0.6386022482382743,
|
||
|
|
"learning_rate": 2.1527072126877288e-05,
|
||
|
|
"loss": 0.2424,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10957943648099899,
|
||
|
|
"step": 1835,
|
||
|
|
"valid_targets_mean": 2851.9,
|
||
|
|
"valid_targets_min": 1898
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.709677419354839,
|
||
|
|
"grad_norm": 0.5686855466856149,
|
||
|
|
"learning_rate": 2.1426783663984648e-05,
|
||
|
|
"loss": 0.2499,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16990455985069275,
|
||
|
|
"step": 1840,
|
||
|
|
"valid_targets_mean": 3819.4,
|
||
|
|
"valid_targets_min": 340
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.719758064516129,
|
||
|
|
"grad_norm": 0.6159316797474549,
|
||
|
|
"learning_rate": 2.132645912870208e-05,
|
||
|
|
"loss": 0.2213,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13322120904922485,
|
||
|
|
"step": 1845,
|
||
|
|
"valid_targets_mean": 3916.6,
|
||
|
|
"valid_targets_min": 1994
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7298387096774195,
|
||
|
|
"grad_norm": 0.7207974173801004,
|
||
|
|
"learning_rate": 2.1226101057465804e-05,
|
||
|
|
"loss": 0.2444,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1405506283044815,
|
||
|
|
"step": 1850,
|
||
|
|
"valid_targets_mean": 2705.8,
|
||
|
|
"valid_targets_min": 1286
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7399193548387095,
|
||
|
|
"grad_norm": 0.6108512002599794,
|
||
|
|
"learning_rate": 2.1125711987559887e-05,
|
||
|
|
"loss": 0.2526,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07237091660499573,
|
||
|
|
"step": 1855,
|
||
|
|
"valid_targets_mean": 2300.8,
|
||
|
|
"valid_targets_min": 1283
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.75,
|
||
|
|
"grad_norm": 0.6511419389637578,
|
||
|
|
"learning_rate": 2.1025294457052145e-05,
|
||
|
|
"loss": 0.2294,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11400432139635086,
|
||
|
|
"step": 1860,
|
||
|
|
"valid_targets_mean": 2622.4,
|
||
|
|
"valid_targets_min": 555
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7600806451612905,
|
||
|
|
"grad_norm": 0.6120672421834176,
|
||
|
|
"learning_rate": 2.092485100472991e-05,
|
||
|
|
"loss": 0.2205,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10978744179010391,
|
||
|
|
"step": 1865,
|
||
|
|
"valid_targets_mean": 2725.6,
|
||
|
|
"valid_targets_min": 861
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.7701612903225805,
|
||
|
|
"grad_norm": 0.6324375125709412,
|
||
|
|
"learning_rate": 2.082438417003591e-05,
|
||
|
|
"loss": 0.2385,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12156516313552856,
|
||
|
|
"step": 1870,
|
||
|
|
"valid_targets_mean": 2576.8,
|
||
|
|
"valid_targets_min": 393
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.780241935483871,
|
||
|
|
"grad_norm": 0.6449683380605613,
|
||
|
|
"learning_rate": 2.0723896493004013e-05,
|
||
|
|
"loss": 0.2381,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16230705380439758,
|
||
|
|
"step": 1875,
|
||
|
|
"valid_targets_mean": 3485.1,
|
||
|
|
"valid_targets_min": 1517
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.790322580645161,
|
||
|
|
"grad_norm": 0.5925684008434925,
|
||
|
|
"learning_rate": 2.0623390514195036e-05,
|
||
|
|
"loss": 0.2266,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.134189635515213,
|
||
|
|
"step": 1880,
|
||
|
|
"valid_targets_mean": 3793.2,
|
||
|
|
"valid_targets_min": 2289
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8004032258064515,
|
||
|
|
"grad_norm": 0.5469268437637105,
|
||
|
|
"learning_rate": 2.05228687746325e-05,
|
||
|
|
"loss": 0.2275,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11094589531421661,
|
||
|
|
"step": 1885,
|
||
|
|
"valid_targets_mean": 2929.2,
|
||
|
|
"valid_targets_min": 359
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.810483870967742,
|
||
|
|
"grad_norm": 0.6128697613399711,
|
||
|
|
"learning_rate": 2.0422333815738403e-05,
|
||
|
|
"loss": 0.2259,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1311267614364624,
|
||
|
|
"step": 1890,
|
||
|
|
"valid_targets_mean": 3038.1,
|
||
|
|
"valid_targets_min": 1189
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.820564516129032,
|
||
|
|
"grad_norm": 0.6383949647469853,
|
||
|
|
"learning_rate": 2.0321788179268953e-05,
|
||
|
|
"loss": 0.2316,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10558299720287323,
|
||
|
|
"step": 1895,
|
||
|
|
"valid_targets_mean": 2551.6,
|
||
|
|
"valid_targets_min": 322
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8306451612903225,
|
||
|
|
"grad_norm": 0.6130282977042798,
|
||
|
|
"learning_rate": 2.022123440725032e-05,
|
||
|
|
"loss": 0.2366,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.084941066801548,
|
||
|
|
"step": 1900,
|
||
|
|
"valid_targets_mean": 2756.0,
|
||
|
|
"valid_targets_min": 1506
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.840725806451613,
|
||
|
|
"grad_norm": 0.5837381112008512,
|
||
|
|
"learning_rate": 2.0120675041914345e-05,
|
||
|
|
"loss": 0.2473,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07631230354309082,
|
||
|
|
"step": 1905,
|
||
|
|
"valid_targets_mean": 2594.1,
|
||
|
|
"valid_targets_min": 1467
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.850806451612903,
|
||
|
|
"grad_norm": 0.6585556336162205,
|
||
|
|
"learning_rate": 2.0020112625634298e-05,
|
||
|
|
"loss": 0.2257,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13330991566181183,
|
||
|
|
"step": 1910,
|
||
|
|
"valid_targets_mean": 2681.0,
|
||
|
|
"valid_targets_min": 1135
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8608870967741935,
|
||
|
|
"grad_norm": 0.62580665087514,
|
||
|
|
"learning_rate": 1.9919549700860575e-05,
|
||
|
|
"loss": 0.2452,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09314854443073273,
|
||
|
|
"step": 1915,
|
||
|
|
"valid_targets_mean": 3280.6,
|
||
|
|
"valid_targets_min": 1261
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.870967741935484,
|
||
|
|
"grad_norm": 0.6466009849243044,
|
||
|
|
"learning_rate": 1.981898881005642e-05,
|
||
|
|
"loss": 0.2167,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09541472792625427,
|
||
|
|
"step": 1920,
|
||
|
|
"valid_targets_mean": 2459.8,
|
||
|
|
"valid_targets_min": 397
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.881048387096774,
|
||
|
|
"grad_norm": 0.6311572797908637,
|
||
|
|
"learning_rate": 1.971843249563367e-05,
|
||
|
|
"loss": 0.2258,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09598266333341599,
|
||
|
|
"step": 1925,
|
||
|
|
"valid_targets_mean": 2833.8,
|
||
|
|
"valid_targets_min": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.8911290322580645,
|
||
|
|
"grad_norm": 0.6468042826489239,
|
||
|
|
"learning_rate": 1.9617883299888448e-05,
|
||
|
|
"loss": 0.2301,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16761614382266998,
|
||
|
|
"step": 1930,
|
||
|
|
"valid_targets_mean": 3247.1,
|
||
|
|
"valid_targets_min": 1576
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.901209677419355,
|
||
|
|
"grad_norm": 0.6219301830686003,
|
||
|
|
"learning_rate": 1.9517343764936913e-05,
|
||
|
|
"loss": 0.2436,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14241242408752441,
|
||
|
|
"step": 1935,
|
||
|
|
"valid_targets_mean": 3839.8,
|
||
|
|
"valid_targets_min": 1965
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.911290322580645,
|
||
|
|
"grad_norm": 0.6420104864862746,
|
||
|
|
"learning_rate": 1.941681643265096e-05,
|
||
|
|
"loss": 0.231,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09952138364315033,
|
||
|
|
"step": 1940,
|
||
|
|
"valid_targets_mean": 2448.6,
|
||
|
|
"valid_targets_min": 1134
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9213709677419355,
|
||
|
|
"grad_norm": 0.5999830651340256,
|
||
|
|
"learning_rate": 1.931630384459398e-05,
|
||
|
|
"loss": 0.224,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06938642263412476,
|
||
|
|
"step": 1945,
|
||
|
|
"valid_targets_mean": 2313.2,
|
||
|
|
"valid_targets_min": 1322
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.931451612903226,
|
||
|
|
"grad_norm": 0.5599170849645915,
|
||
|
|
"learning_rate": 1.9215808541956598e-05,
|
||
|
|
"loss": 0.2512,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13666555285453796,
|
||
|
|
"step": 1950,
|
||
|
|
"valid_targets_mean": 4009.2,
|
||
|
|
"valid_targets_min": 1504
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.941532258064516,
|
||
|
|
"grad_norm": 0.6393948739988566,
|
||
|
|
"learning_rate": 1.9115333065492428e-05,
|
||
|
|
"loss": 0.2261,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10924457013607025,
|
||
|
|
"step": 1955,
|
||
|
|
"valid_targets_mean": 2666.2,
|
||
|
|
"valid_targets_min": 1192
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9516129032258065,
|
||
|
|
"grad_norm": 0.6217378350750355,
|
||
|
|
"learning_rate": 1.901487995545382e-05,
|
||
|
|
"loss": 0.2572,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13093984127044678,
|
||
|
|
"step": 1960,
|
||
|
|
"valid_targets_mean": 3649.9,
|
||
|
|
"valid_targets_min": 1564
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.961693548387097,
|
||
|
|
"grad_norm": 0.6402957850458644,
|
||
|
|
"learning_rate": 1.891445175152766e-05,
|
||
|
|
"loss": 0.2364,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12494450807571411,
|
||
|
|
"step": 1965,
|
||
|
|
"valid_targets_mean": 3567.4,
|
||
|
|
"valid_targets_min": 396
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.971774193548387,
|
||
|
|
"grad_norm": 0.6022918098664621,
|
||
|
|
"learning_rate": 1.8814050992771146e-05,
|
||
|
|
"loss": 0.2299,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06463226675987244,
|
||
|
|
"step": 1970,
|
||
|
|
"valid_targets_mean": 2425.1,
|
||
|
|
"valid_targets_min": 374
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.9818548387096775,
|
||
|
|
"grad_norm": 0.5886862169023838,
|
||
|
|
"learning_rate": 1.8713680217547597e-05,
|
||
|
|
"loss": 0.2134,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08834889531135559,
|
||
|
|
"step": 1975,
|
||
|
|
"valid_targets_mean": 2680.4,
|
||
|
|
"valid_targets_min": 1076
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.991935483870968,
|
||
|
|
"grad_norm": 0.7277023253149896,
|
||
|
|
"learning_rate": 1.8613341963462277e-05,
|
||
|
|
"loss": 0.2294,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13720348477363586,
|
||
|
|
"step": 1980,
|
||
|
|
"valid_targets_mean": 2581.6,
|
||
|
|
"valid_targets_min": 364
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.002016129032258,
|
||
|
|
"grad_norm": 0.6088699709458538,
|
||
|
|
"learning_rate": 1.8513038767298254e-05,
|
||
|
|
"loss": 0.2337,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07562052458524704,
|
||
|
|
"step": 1985,
|
||
|
|
"valid_targets_mean": 2229.4,
|
||
|
|
"valid_targets_min": 399
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.012096774193548,
|
||
|
|
"grad_norm": 0.7035284909210588,
|
||
|
|
"learning_rate": 1.8412773164952233e-05,
|
||
|
|
"loss": 0.2074,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09331750124692917,
|
||
|
|
"step": 1990,
|
||
|
|
"valid_targets_mean": 2995.2,
|
||
|
|
"valid_targets_min": 1384
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.022177419354839,
|
||
|
|
"grad_norm": 0.660811520874278,
|
||
|
|
"learning_rate": 1.8312547691370467e-05,
|
||
|
|
"loss": 0.2097,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08929209411144257,
|
||
|
|
"step": 1995,
|
||
|
|
"valid_targets_mean": 2813.5,
|
||
|
|
"valid_targets_min": 942
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.032258064516129,
|
||
|
|
"grad_norm": 0.5488695873603645,
|
||
|
|
"learning_rate": 1.8212364880484664e-05,
|
||
|
|
"loss": 0.1971,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0736876130104065,
|
||
|
|
"step": 2000,
|
||
|
|
"valid_targets_mean": 3529.0,
|
||
|
|
"valid_targets_min": 1194
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.042338709677419,
|
||
|
|
"grad_norm": 0.6347569432286869,
|
||
|
|
"learning_rate": 1.811222726514793e-05,
|
||
|
|
"loss": 0.203,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0888785794377327,
|
||
|
|
"step": 2005,
|
||
|
|
"valid_targets_mean": 3000.4,
|
||
|
|
"valid_targets_min": 1877
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.05241935483871,
|
||
|
|
"grad_norm": 0.5725029195818679,
|
||
|
|
"learning_rate": 1.8012137377070683e-05,
|
||
|
|
"loss": 0.1789,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0814841091632843,
|
||
|
|
"step": 2010,
|
||
|
|
"valid_targets_mean": 2873.2,
|
||
|
|
"valid_targets_min": 1671
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.0625,
|
||
|
|
"grad_norm": 0.6686548355616967,
|
||
|
|
"learning_rate": 1.791209774675674e-05,
|
||
|
|
"loss": 0.1924,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10090295225381851,
|
||
|
|
"step": 2015,
|
||
|
|
"valid_targets_mean": 2488.5,
|
||
|
|
"valid_targets_min": 1060
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.07258064516129,
|
||
|
|
"grad_norm": 0.6355501797130095,
|
||
|
|
"learning_rate": 1.781211090343925e-05,
|
||
|
|
"loss": 0.2015,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08359242975711823,
|
||
|
|
"step": 2020,
|
||
|
|
"valid_targets_mean": 3068.4,
|
||
|
|
"valid_targets_min": 1943
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.082661290322581,
|
||
|
|
"grad_norm": 0.6598682839958637,
|
||
|
|
"learning_rate": 1.7712179375016792e-05,
|
||
|
|
"loss": 0.2118,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10339263826608658,
|
||
|
|
"step": 2025,
|
||
|
|
"valid_targets_mean": 3350.5,
|
||
|
|
"valid_targets_min": 1844
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.092741935483871,
|
||
|
|
"grad_norm": 0.6632899860464143,
|
||
|
|
"learning_rate": 1.7612305687989465e-05,
|
||
|
|
"loss": 0.2184,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11771737784147263,
|
||
|
|
"step": 2030,
|
||
|
|
"valid_targets_mean": 3088.5,
|
||
|
|
"valid_targets_min": 1364
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.102822580645161,
|
||
|
|
"grad_norm": 0.5455949449267282,
|
||
|
|
"learning_rate": 1.7512492367394988e-05,
|
||
|
|
"loss": 0.2106,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0710437074303627,
|
||
|
|
"step": 2035,
|
||
|
|
"valid_targets_mean": 3906.1,
|
||
|
|
"valid_targets_min": 1941
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.112903225806452,
|
||
|
|
"grad_norm": 0.5870371714663924,
|
||
|
|
"learning_rate": 1.74127419367449e-05,
|
||
|
|
"loss": 0.2027,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08473192900419235,
|
||
|
|
"step": 2040,
|
||
|
|
"valid_targets_mean": 3041.8,
|
||
|
|
"valid_targets_min": 1566
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.122983870967742,
|
||
|
|
"grad_norm": 0.8185178407447274,
|
||
|
|
"learning_rate": 1.731305691796071e-05,
|
||
|
|
"loss": 0.1878,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0766078531742096,
|
||
|
|
"step": 2045,
|
||
|
|
"valid_targets_mean": 2362.0,
|
||
|
|
"valid_targets_min": 377
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.133064516129032,
|
||
|
|
"grad_norm": 0.7391684981445084,
|
||
|
|
"learning_rate": 1.7213439831310186e-05,
|
||
|
|
"loss": 0.2042,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1130332350730896,
|
||
|
|
"step": 2050,
|
||
|
|
"valid_targets_mean": 3482.2,
|
||
|
|
"valid_targets_min": 1196
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.143145161290323,
|
||
|
|
"grad_norm": 0.635862670761652,
|
||
|
|
"learning_rate": 1.7113893195343602e-05,
|
||
|
|
"loss": 0.2289,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1211572214961052,
|
||
|
|
"step": 2055,
|
||
|
|
"valid_targets_mean": 3921.5,
|
||
|
|
"valid_targets_min": 1463
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.153225806451613,
|
||
|
|
"grad_norm": 0.7238833565302708,
|
||
|
|
"learning_rate": 1.7014419526830078e-05,
|
||
|
|
"loss": 0.204,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10985315591096878,
|
||
|
|
"step": 2060,
|
||
|
|
"valid_targets_mean": 2456.0,
|
||
|
|
"valid_targets_min": 1446
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.163306451612903,
|
||
|
|
"grad_norm": 0.6395469252565582,
|
||
|
|
"learning_rate": 1.6915021340693952e-05,
|
||
|
|
"loss": 0.2154,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08048843592405319,
|
||
|
|
"step": 2065,
|
||
|
|
"valid_targets_mean": 2882.8,
|
||
|
|
"valid_targets_min": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.173387096774194,
|
||
|
|
"grad_norm": 0.5271310503583636,
|
||
|
|
"learning_rate": 1.6815701149951193e-05,
|
||
|
|
"loss": 0.1936,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08474685251712799,
|
||
|
|
"step": 2070,
|
||
|
|
"valid_targets_mean": 4114.2,
|
||
|
|
"valid_targets_min": 1742
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.183467741935484,
|
||
|
|
"grad_norm": 0.6050852712197098,
|
||
|
|
"learning_rate": 1.6716461465645868e-05,
|
||
|
|
"loss": 0.2082,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12912030518054962,
|
||
|
|
"step": 2075,
|
||
|
|
"valid_targets_mean": 4300.2,
|
||
|
|
"valid_targets_min": 2352
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.193548387096774,
|
||
|
|
"grad_norm": 0.6096579930090852,
|
||
|
|
"learning_rate": 1.6617304796786636e-05,
|
||
|
|
"loss": 0.2218,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07606025040149689,
|
||
|
|
"step": 2080,
|
||
|
|
"valid_targets_mean": 2741.6,
|
||
|
|
"valid_targets_min": 332
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.203629032258065,
|
||
|
|
"grad_norm": 0.5756357647128211,
|
||
|
|
"learning_rate": 1.6518233650283356e-05,
|
||
|
|
"loss": 0.2136,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08158040046691895,
|
||
|
|
"step": 2085,
|
||
|
|
"valid_targets_mean": 3955.4,
|
||
|
|
"valid_targets_min": 2360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.213709677419355,
|
||
|
|
"grad_norm": 0.7014520641033342,
|
||
|
|
"learning_rate": 1.6419250530883677e-05,
|
||
|
|
"loss": 0.1935,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09476025402545929,
|
||
|
|
"step": 2090,
|
||
|
|
"valid_targets_mean": 2671.6,
|
||
|
|
"valid_targets_min": 439
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.223790322580645,
|
||
|
|
"grad_norm": 0.7071591474124187,
|
||
|
|
"learning_rate": 1.6320357941109718e-05,
|
||
|
|
"loss": 0.2189,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.19258907437324524,
|
||
|
|
"step": 2095,
|
||
|
|
"valid_targets_mean": 3923.0,
|
||
|
|
"valid_targets_min": 1685
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.233870967741935,
|
||
|
|
"grad_norm": 0.6377243428638062,
|
||
|
|
"learning_rate": 1.6221558381194804e-05,
|
||
|
|
"loss": 0.2004,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06718167662620544,
|
||
|
|
"step": 2100,
|
||
|
|
"valid_targets_mean": 2454.1,
|
||
|
|
"valid_targets_min": 917
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.243951612903226,
|
||
|
|
"grad_norm": 0.6019582322268854,
|
||
|
|
"learning_rate": 1.612285434902024e-05,
|
||
|
|
"loss": 0.198,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0957595631480217,
|
||
|
|
"step": 2105,
|
||
|
|
"valid_targets_mean": 3524.9,
|
||
|
|
"valid_targets_min": 1196
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.254032258064516,
|
||
|
|
"grad_norm": 0.6269637000916031,
|
||
|
|
"learning_rate": 1.602424834005218e-05,
|
||
|
|
"loss": 0.1885,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1165969967842102,
|
||
|
|
"step": 2110,
|
||
|
|
"valid_targets_mean": 4072.8,
|
||
|
|
"valid_targets_min": 1293
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.264112903225806,
|
||
|
|
"grad_norm": 0.6668835567850124,
|
||
|
|
"learning_rate": 1.592574284727852e-05,
|
||
|
|
"loss": 0.2214,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09571275115013123,
|
||
|
|
"step": 2115,
|
||
|
|
"valid_targets_mean": 2841.2,
|
||
|
|
"valid_targets_min": 327
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.274193548387097,
|
||
|
|
"grad_norm": 0.6544301355753144,
|
||
|
|
"learning_rate": 1.5827340361145873e-05,
|
||
|
|
"loss": 0.2017,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08974354714155197,
|
||
|
|
"step": 2120,
|
||
|
|
"valid_targets_mean": 3352.1,
|
||
|
|
"valid_targets_min": 2235
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.284274193548387,
|
||
|
|
"grad_norm": 0.7132421819665873,
|
||
|
|
"learning_rate": 1.572904336949661e-05,
|
||
|
|
"loss": 0.2097,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0757712721824646,
|
||
|
|
"step": 2125,
|
||
|
|
"valid_targets_mean": 1831.4,
|
||
|
|
"valid_targets_min": 1135
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.294354838709677,
|
||
|
|
"grad_norm": 0.602624249250874,
|
||
|
|
"learning_rate": 1.563085435750595e-05,
|
||
|
|
"loss": 0.1901,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08245712518692017,
|
||
|
|
"step": 2130,
|
||
|
|
"valid_targets_mean": 3184.0,
|
||
|
|
"valid_targets_min": 1668
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.304435483870968,
|
||
|
|
"grad_norm": 0.672902690661134,
|
||
|
|
"learning_rate": 1.553277580761915e-05,
|
||
|
|
"loss": 0.2016,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07349157333374023,
|
||
|
|
"step": 2135,
|
||
|
|
"valid_targets_mean": 2635.1,
|
||
|
|
"valid_targets_min": 1407
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.314516129032258,
|
||
|
|
"grad_norm": 0.5465394558442096,
|
||
|
|
"learning_rate": 1.5434810199488708e-05,
|
||
|
|
"loss": 0.2278,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07390094548463821,
|
||
|
|
"step": 2140,
|
||
|
|
"valid_targets_mean": 4155.2,
|
||
|
|
"valid_targets_min": 1583
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.324596774193548,
|
||
|
|
"grad_norm": 0.7245490208166988,
|
||
|
|
"learning_rate": 1.533696000991172e-05,
|
||
|
|
"loss": 0.228,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09360332041978836,
|
||
|
|
"step": 2145,
|
||
|
|
"valid_targets_mean": 3448.9,
|
||
|
|
"valid_targets_min": 358
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.334677419354839,
|
||
|
|
"grad_norm": 0.6977389351544049,
|
||
|
|
"learning_rate": 1.5239227712767216e-05,
|
||
|
|
"loss": 0.2015,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.111058309674263,
|
||
|
|
"step": 2150,
|
||
|
|
"valid_targets_mean": 2823.1,
|
||
|
|
"valid_targets_min": 1263
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.344758064516129,
|
||
|
|
"grad_norm": 0.5559162121552728,
|
||
|
|
"learning_rate": 1.5141615778953626e-05,
|
||
|
|
"loss": 0.2114,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09068315476179123,
|
||
|
|
"step": 2155,
|
||
|
|
"valid_targets_mean": 4450.2,
|
||
|
|
"valid_targets_min": 1212
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.354838709677419,
|
||
|
|
"grad_norm": 0.6293904288777087,
|
||
|
|
"learning_rate": 1.5044126676326345e-05,
|
||
|
|
"loss": 0.2272,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08106718957424164,
|
||
|
|
"step": 2160,
|
||
|
|
"valid_targets_mean": 2731.4,
|
||
|
|
"valid_targets_min": 393
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.36491935483871,
|
||
|
|
"grad_norm": 0.6051514109494363,
|
||
|
|
"learning_rate": 1.4946762869635283e-05,
|
||
|
|
"loss": 0.2092,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12616080045700073,
|
||
|
|
"step": 2165,
|
||
|
|
"valid_targets_mean": 3888.4,
|
||
|
|
"valid_targets_min": 472
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.375,
|
||
|
|
"grad_norm": 0.6947857280248805,
|
||
|
|
"learning_rate": 1.4849526820462591e-05,
|
||
|
|
"loss": 0.2032,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16071775555610657,
|
||
|
|
"step": 2170,
|
||
|
|
"valid_targets_mean": 3320.1,
|
||
|
|
"valid_targets_min": 1269
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.38508064516129,
|
||
|
|
"grad_norm": 0.6364941132247152,
|
||
|
|
"learning_rate": 1.4752420987160423e-05,
|
||
|
|
"loss": 0.1889,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0816536396741867,
|
||
|
|
"step": 2175,
|
||
|
|
"valid_targets_mean": 2365.0,
|
||
|
|
"valid_targets_min": 1462
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.395161290322581,
|
||
|
|
"grad_norm": 0.6584814716982702,
|
||
|
|
"learning_rate": 1.4655447824788751e-05,
|
||
|
|
"loss": 0.2429,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13508567214012146,
|
||
|
|
"step": 2180,
|
||
|
|
"valid_targets_mean": 3173.0,
|
||
|
|
"valid_targets_min": 1236
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.405241935483871,
|
||
|
|
"grad_norm": 0.6181434861945387,
|
||
|
|
"learning_rate": 1.455860978505335e-05,
|
||
|
|
"loss": 0.1892,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10209818184375763,
|
||
|
|
"step": 2185,
|
||
|
|
"valid_targets_mean": 2873.9,
|
||
|
|
"valid_targets_min": 1385
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.415322580645161,
|
||
|
|
"grad_norm": 0.5703762962329544,
|
||
|
|
"learning_rate": 1.4461909316243756e-05,
|
||
|
|
"loss": 0.194,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1135522723197937,
|
||
|
|
"step": 2190,
|
||
|
|
"valid_targets_mean": 4035.5,
|
||
|
|
"valid_targets_min": 1647
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.425403225806452,
|
||
|
|
"grad_norm": 0.6801687111771926,
|
||
|
|
"learning_rate": 1.4365348863171406e-05,
|
||
|
|
"loss": 0.1992,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09940125048160553,
|
||
|
|
"step": 2195,
|
||
|
|
"valid_targets_mean": 2376.0,
|
||
|
|
"valid_targets_min": 391
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.435483870967742,
|
||
|
|
"grad_norm": 0.6058048930044343,
|
||
|
|
"learning_rate": 1.4268930867107815e-05,
|
||
|
|
"loss": 0.1928,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08014991134405136,
|
||
|
|
"step": 2200,
|
||
|
|
"valid_targets_mean": 3025.1,
|
||
|
|
"valid_targets_min": 1809
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.445564516129032,
|
||
|
|
"grad_norm": 0.6076019554921235,
|
||
|
|
"learning_rate": 1.4172657765722854e-05,
|
||
|
|
"loss": 0.2253,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12928083539009094,
|
||
|
|
"step": 2205,
|
||
|
|
"valid_targets_mean": 3551.6,
|
||
|
|
"valid_targets_min": 1658
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.455645161290323,
|
||
|
|
"grad_norm": 0.6230752799781244,
|
||
|
|
"learning_rate": 1.4076531993023119e-05,
|
||
|
|
"loss": 0.2254,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1294112652540207,
|
||
|
|
"step": 2210,
|
||
|
|
"valid_targets_mean": 3246.5,
|
||
|
|
"valid_targets_min": 1509
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.465725806451613,
|
||
|
|
"grad_norm": 0.6122080531351842,
|
||
|
|
"learning_rate": 1.3980555979290411e-05,
|
||
|
|
"loss": 0.2108,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10016797482967377,
|
||
|
|
"step": 2215,
|
||
|
|
"valid_targets_mean": 3908.2,
|
||
|
|
"valid_targets_min": 1769
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.475806451612903,
|
||
|
|
"grad_norm": 0.6517180321108508,
|
||
|
|
"learning_rate": 1.3884732151020244e-05,
|
||
|
|
"loss": 0.1864,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0860491544008255,
|
||
|
|
"step": 2220,
|
||
|
|
"valid_targets_mean": 2572.1,
|
||
|
|
"valid_targets_min": 400
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.485887096774194,
|
||
|
|
"grad_norm": 0.6300860804677118,
|
||
|
|
"learning_rate": 1.3789062930860568e-05,
|
||
|
|
"loss": 0.2266,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11208859086036682,
|
||
|
|
"step": 2225,
|
||
|
|
"valid_targets_mean": 2997.4,
|
||
|
|
"valid_targets_min": 1537
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.495967741935484,
|
||
|
|
"grad_norm": 0.7066162504275811,
|
||
|
|
"learning_rate": 1.3693550737550461e-05,
|
||
|
|
"loss": 0.2155,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09464829415082932,
|
||
|
|
"step": 2230,
|
||
|
|
"valid_targets_mean": 2797.9,
|
||
|
|
"valid_targets_min": 1305
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.506048387096774,
|
||
|
|
"grad_norm": 0.6164914557556402,
|
||
|
|
"learning_rate": 1.3598197985859015e-05,
|
||
|
|
"loss": 0.2015,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09443396329879761,
|
||
|
|
"step": 2235,
|
||
|
|
"valid_targets_mean": 3473.5,
|
||
|
|
"valid_targets_min": 2019
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.516129032258064,
|
||
|
|
"grad_norm": 0.6976355172985539,
|
||
|
|
"learning_rate": 1.3503007086524246e-05,
|
||
|
|
"loss": 0.2035,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10510143637657166,
|
||
|
|
"step": 2240,
|
||
|
|
"valid_targets_mean": 2421.9,
|
||
|
|
"valid_targets_min": 1328
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.526209677419355,
|
||
|
|
"grad_norm": 0.6725361556340994,
|
||
|
|
"learning_rate": 1.34079804461922e-05,
|
||
|
|
"loss": 0.1956,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0974196195602417,
|
||
|
|
"step": 2245,
|
||
|
|
"valid_targets_mean": 3222.0,
|
||
|
|
"valid_targets_min": 417
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.536290322580645,
|
||
|
|
"grad_norm": 0.5584171280441081,
|
||
|
|
"learning_rate": 1.331312046735604e-05,
|
||
|
|
"loss": 0.1926,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07046773284673691,
|
||
|
|
"step": 2250,
|
||
|
|
"valid_targets_mean": 3709.0,
|
||
|
|
"valid_targets_min": 404
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.546370967741936,
|
||
|
|
"grad_norm": 0.6841167084366355,
|
||
|
|
"learning_rate": 1.3218429548295377e-05,
|
||
|
|
"loss": 0.2271,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07062976062297821,
|
||
|
|
"step": 2255,
|
||
|
|
"valid_targets_mean": 2267.9,
|
||
|
|
"valid_targets_min": 361
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.556451612903226,
|
||
|
|
"grad_norm": 0.6646274595202382,
|
||
|
|
"learning_rate": 1.3123910083015572e-05,
|
||
|
|
"loss": 0.2034,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1463637799024582,
|
||
|
|
"step": 2260,
|
||
|
|
"valid_targets_mean": 3185.2,
|
||
|
|
"valid_targets_min": 356
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.566532258064516,
|
||
|
|
"grad_norm": 0.6943503019804396,
|
||
|
|
"learning_rate": 1.3029564461187245e-05,
|
||
|
|
"loss": 0.2176,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09197193384170532,
|
||
|
|
"step": 2265,
|
||
|
|
"valid_targets_mean": 2933.0,
|
||
|
|
"valid_targets_min": 1378
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.576612903225806,
|
||
|
|
"grad_norm": 0.6383279293104014,
|
||
|
|
"learning_rate": 1.2935395068085863e-05,
|
||
|
|
"loss": 0.2146,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09128080308437347,
|
||
|
|
"step": 2270,
|
||
|
|
"valid_targets_mean": 2425.9,
|
||
|
|
"valid_targets_min": 324
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.586693548387097,
|
||
|
|
"grad_norm": 0.7004171993662113,
|
||
|
|
"learning_rate": 1.284140428453141e-05,
|
||
|
|
"loss": 0.2101,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10103457421064377,
|
||
|
|
"step": 2275,
|
||
|
|
"valid_targets_mean": 2708.1,
|
||
|
|
"valid_targets_min": 1470
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.596774193548387,
|
||
|
|
"grad_norm": 0.6675458017112398,
|
||
|
|
"learning_rate": 1.2747594486828217e-05,
|
||
|
|
"loss": 0.2063,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08124849200248718,
|
||
|
|
"step": 2280,
|
||
|
|
"valid_targets_mean": 2318.4,
|
||
|
|
"valid_targets_min": 318
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.606854838709677,
|
||
|
|
"grad_norm": 0.6232978478968111,
|
||
|
|
"learning_rate": 1.265396804670487e-05,
|
||
|
|
"loss": 0.2068,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10344327241182327,
|
||
|
|
"step": 2285,
|
||
|
|
"valid_targets_mean": 2956.5,
|
||
|
|
"valid_targets_min": 1405
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.616935483870968,
|
||
|
|
"grad_norm": 0.6919673063369659,
|
||
|
|
"learning_rate": 1.256052733125424e-05,
|
||
|
|
"loss": 0.2168,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14700259268283844,
|
||
|
|
"step": 2290,
|
||
|
|
"valid_targets_mean": 4126.4,
|
||
|
|
"valid_targets_min": 1642
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.627016129032258,
|
||
|
|
"grad_norm": 0.6554048558960205,
|
||
|
|
"learning_rate": 1.2467274702873673e-05,
|
||
|
|
"loss": 0.2135,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08968253433704376,
|
||
|
|
"step": 2295,
|
||
|
|
"valid_targets_mean": 3072.6,
|
||
|
|
"valid_targets_min": 430
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.637096774193548,
|
||
|
|
"grad_norm": 0.682758861747768,
|
||
|
|
"learning_rate": 1.2374212519205222e-05,
|
||
|
|
"loss": 0.2118,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09823716431856155,
|
||
|
|
"step": 2300,
|
||
|
|
"valid_targets_mean": 3030.9,
|
||
|
|
"valid_targets_min": 1482
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.647177419354839,
|
||
|
|
"grad_norm": 0.6645266200440272,
|
||
|
|
"learning_rate": 1.228134313307606e-05,
|
||
|
|
"loss": 0.2063,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1754535585641861,
|
||
|
|
"step": 2305,
|
||
|
|
"valid_targets_mean": 3793.4,
|
||
|
|
"valid_targets_min": 1336
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.657258064516129,
|
||
|
|
"grad_norm": 0.6471086948112743,
|
||
|
|
"learning_rate": 1.2188668892438997e-05,
|
||
|
|
"loss": 0.2299,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17917540669441223,
|
||
|
|
"step": 2310,
|
||
|
|
"valid_targets_mean": 4111.9,
|
||
|
|
"valid_targets_min": 1064
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.667338709677419,
|
||
|
|
"grad_norm": 0.611240952774229,
|
||
|
|
"learning_rate": 1.2096192140313106e-05,
|
||
|
|
"loss": 0.1984,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.057186633348464966,
|
||
|
|
"step": 2315,
|
||
|
|
"valid_targets_mean": 2643.0,
|
||
|
|
"valid_targets_min": 1326
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.67741935483871,
|
||
|
|
"grad_norm": 0.6721726312255223,
|
||
|
|
"learning_rate": 1.2003915214724504e-05,
|
||
|
|
"loss": 0.2328,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14121165871620178,
|
||
|
|
"step": 2320,
|
||
|
|
"valid_targets_mean": 3566.2,
|
||
|
|
"valid_targets_min": 1229
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.6875,
|
||
|
|
"grad_norm": 0.5951813798752076,
|
||
|
|
"learning_rate": 1.191184044864722e-05,
|
||
|
|
"loss": 0.1954,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06060823053121567,
|
||
|
|
"step": 2325,
|
||
|
|
"valid_targets_mean": 2410.5,
|
||
|
|
"valid_targets_min": 350
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.69758064516129,
|
||
|
|
"grad_norm": 0.6634119476207369,
|
||
|
|
"learning_rate": 1.1819970169944237e-05,
|
||
|
|
"loss": 0.2247,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09355494379997253,
|
||
|
|
"step": 2330,
|
||
|
|
"valid_targets_mean": 2947.2,
|
||
|
|
"valid_targets_min": 1471
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.707661290322581,
|
||
|
|
"grad_norm": 0.6527161084627628,
|
||
|
|
"learning_rate": 1.172830670130861e-05,
|
||
|
|
"loss": 0.2013,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05121547356247902,
|
||
|
|
"step": 2335,
|
||
|
|
"valid_targets_mean": 1780.5,
|
||
|
|
"valid_targets_min": 395
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.717741935483871,
|
||
|
|
"grad_norm": 0.5923457947699236,
|
||
|
|
"learning_rate": 1.1636852360204765e-05,
|
||
|
|
"loss": 0.2251,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08893446624279022,
|
||
|
|
"step": 2340,
|
||
|
|
"valid_targets_mean": 3808.9,
|
||
|
|
"valid_targets_min": 1282
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.727822580645161,
|
||
|
|
"grad_norm": 0.6134900108472701,
|
||
|
|
"learning_rate": 1.1545609458809897e-05,
|
||
|
|
"loss": 0.1967,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11972342431545258,
|
||
|
|
"step": 2345,
|
||
|
|
"valid_targets_mean": 3859.8,
|
||
|
|
"valid_targets_min": 1724
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.737903225806452,
|
||
|
|
"grad_norm": 0.6977629368220143,
|
||
|
|
"learning_rate": 1.1454580303955527e-05,
|
||
|
|
"loss": 0.1978,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08183122426271439,
|
||
|
|
"step": 2350,
|
||
|
|
"valid_targets_mean": 3006.8,
|
||
|
|
"valid_targets_min": 1988
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.747983870967742,
|
||
|
|
"grad_norm": 0.6703891952346482,
|
||
|
|
"learning_rate": 1.1363767197069153e-05,
|
||
|
|
"loss": 0.2218,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0819975957274437,
|
||
|
|
"step": 2355,
|
||
|
|
"valid_targets_mean": 2956.8,
|
||
|
|
"valid_targets_min": 370
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.758064516129032,
|
||
|
|
"grad_norm": 0.6538324155398517,
|
||
|
|
"learning_rate": 1.1273172434116084e-05,
|
||
|
|
"loss": 0.2098,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11679733544588089,
|
||
|
|
"step": 2360,
|
||
|
|
"valid_targets_mean": 3381.8,
|
||
|
|
"valid_targets_min": 1245
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.768145161290323,
|
||
|
|
"grad_norm": 0.5845412336568625,
|
||
|
|
"learning_rate": 1.1182798305541398e-05,
|
||
|
|
"loss": 0.2095,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10125453025102615,
|
||
|
|
"step": 2365,
|
||
|
|
"valid_targets_mean": 3555.9,
|
||
|
|
"valid_targets_min": 1783
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.778225806451613,
|
||
|
|
"grad_norm": 0.7733291737221234,
|
||
|
|
"learning_rate": 1.1092647096212024e-05,
|
||
|
|
"loss": 0.2498,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08022960275411606,
|
||
|
|
"step": 2370,
|
||
|
|
"valid_targets_mean": 2148.0,
|
||
|
|
"valid_targets_min": 416
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.788306451612903,
|
||
|
|
"grad_norm": 0.6240433081593019,
|
||
|
|
"learning_rate": 1.1002721085358981e-05,
|
||
|
|
"loss": 0.2155,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14964377880096436,
|
||
|
|
"step": 2375,
|
||
|
|
"valid_targets_mean": 4125.1,
|
||
|
|
"valid_targets_min": 332
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.798387096774194,
|
||
|
|
"grad_norm": 0.6271445364830281,
|
||
|
|
"learning_rate": 1.0913022546519727e-05,
|
||
|
|
"loss": 0.1951,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1267169713973999,
|
||
|
|
"step": 2380,
|
||
|
|
"valid_targets_mean": 3932.6,
|
||
|
|
"valid_targets_min": 2294
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.808467741935484,
|
||
|
|
"grad_norm": 0.5791473764633525,
|
||
|
|
"learning_rate": 1.0823553747480727e-05,
|
||
|
|
"loss": 0.2017,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1334831416606903,
|
||
|
|
"step": 2385,
|
||
|
|
"valid_targets_mean": 4574.9,
|
||
|
|
"valid_targets_min": 2043
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.818548387096774,
|
||
|
|
"grad_norm": 0.705562192823843,
|
||
|
|
"learning_rate": 1.0734316950220092e-05,
|
||
|
|
"loss": 0.2157,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10326686501502991,
|
||
|
|
"step": 2390,
|
||
|
|
"valid_targets_mean": 3475.9,
|
||
|
|
"valid_targets_min": 1453
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.828629032258064,
|
||
|
|
"grad_norm": 0.606426743557359,
|
||
|
|
"learning_rate": 1.0645314410850368e-05,
|
||
|
|
"loss": 0.2218,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14200296998023987,
|
||
|
|
"step": 2395,
|
||
|
|
"valid_targets_mean": 3981.2,
|
||
|
|
"valid_targets_min": 478
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.838709677419355,
|
||
|
|
"grad_norm": 0.6872920140461152,
|
||
|
|
"learning_rate": 1.0556548379561541e-05,
|
||
|
|
"loss": 0.2023,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11578333377838135,
|
||
|
|
"step": 2400,
|
||
|
|
"valid_targets_mean": 3194.6,
|
||
|
|
"valid_targets_min": 1596
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.848790322580645,
|
||
|
|
"grad_norm": 0.6228292302829015,
|
||
|
|
"learning_rate": 1.0468021100564125e-05,
|
||
|
|
"loss": 0.2221,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0996391624212265,
|
||
|
|
"step": 2405,
|
||
|
|
"valid_targets_mean": 3179.1,
|
||
|
|
"valid_targets_min": 1669
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.858870967741936,
|
||
|
|
"grad_norm": 0.5368708028608827,
|
||
|
|
"learning_rate": 1.037973481203242e-05,
|
||
|
|
"loss": 0.2169,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0917302742600441,
|
||
|
|
"step": 2410,
|
||
|
|
"valid_targets_mean": 4231.2,
|
||
|
|
"valid_targets_min": 1325
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.868951612903226,
|
||
|
|
"grad_norm": 0.6285450272064274,
|
||
|
|
"learning_rate": 1.029169174604792e-05,
|
||
|
|
"loss": 0.1943,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0975906029343605,
|
||
|
|
"step": 2415,
|
||
|
|
"valid_targets_mean": 3037.1,
|
||
|
|
"valid_targets_min": 1690
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.879032258064516,
|
||
|
|
"grad_norm": 0.6917603574873773,
|
||
|
|
"learning_rate": 1.0203894128542905e-05,
|
||
|
|
"loss": 0.193,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09343405812978745,
|
||
|
|
"step": 2420,
|
||
|
|
"valid_targets_mean": 2223.4,
|
||
|
|
"valid_targets_min": 968
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.889112903225806,
|
||
|
|
"grad_norm": 0.5655800765748986,
|
||
|
|
"learning_rate": 1.011634417924415e-05,
|
||
|
|
"loss": 0.1946,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07445259392261505,
|
||
|
|
"step": 2425,
|
||
|
|
"valid_targets_mean": 2644.2,
|
||
|
|
"valid_targets_min": 387
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.899193548387097,
|
||
|
|
"grad_norm": 0.6378304004625874,
|
||
|
|
"learning_rate": 1.0029044111616795e-05,
|
||
|
|
"loss": 0.2208,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10037689656019211,
|
||
|
|
"step": 2430,
|
||
|
|
"valid_targets_mean": 3578.8,
|
||
|
|
"valid_targets_min": 2622
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.909274193548387,
|
||
|
|
"grad_norm": 0.6951335771452958,
|
||
|
|
"learning_rate": 9.941996132808391e-06,
|
||
|
|
"loss": 0.2006,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12032778561115265,
|
||
|
|
"step": 2435,
|
||
|
|
"valid_targets_mean": 3168.0,
|
||
|
|
"valid_targets_min": 1749
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.919354838709677,
|
||
|
|
"grad_norm": 0.7789494916306356,
|
||
|
|
"learning_rate": 9.855202443593113e-06,
|
||
|
|
"loss": 0.2109,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10983292013406754,
|
||
|
|
"step": 2440,
|
||
|
|
"valid_targets_mean": 2799.5,
|
||
|
|
"valid_targets_min": 363
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.929435483870968,
|
||
|
|
"grad_norm": 0.7155346643978607,
|
||
|
|
"learning_rate": 9.768665238316104e-06,
|
||
|
|
"loss": 0.2176,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1286262571811676,
|
||
|
|
"step": 2445,
|
||
|
|
"valid_targets_mean": 3508.2,
|
||
|
|
"valid_targets_min": 1492
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.939516129032258,
|
||
|
|
"grad_norm": 0.6016868479355055,
|
||
|
|
"learning_rate": 9.682386704838007e-06,
|
||
|
|
"loss": 0.2165,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11881684511899948,
|
||
|
|
"step": 2450,
|
||
|
|
"valid_targets_mean": 4665.4,
|
||
|
|
"valid_targets_min": 2281
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.949596774193548,
|
||
|
|
"grad_norm": 0.7281422020987302,
|
||
|
|
"learning_rate": 9.596369024479617e-06,
|
||
|
|
"loss": 0.2214,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11448834091424942,
|
||
|
|
"step": 2455,
|
||
|
|
"valid_targets_mean": 3200.2,
|
||
|
|
"valid_targets_min": 1359
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.959677419354839,
|
||
|
|
"grad_norm": 0.6563548460503863,
|
||
|
|
"learning_rate": 9.510614371966787e-06,
|
||
|
|
"loss": 0.2207,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10339592397212982,
|
||
|
|
"step": 2460,
|
||
|
|
"valid_targets_mean": 4006.2,
|
||
|
|
"valid_targets_min": 1749
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.969758064516129,
|
||
|
|
"grad_norm": 0.7134556286133679,
|
||
|
|
"learning_rate": 9.425124915375407e-06,
|
||
|
|
"loss": 0.2219,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12733376026153564,
|
||
|
|
"step": 2465,
|
||
|
|
"valid_targets_mean": 3450.2,
|
||
|
|
"valid_targets_min": 1688
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.979838709677419,
|
||
|
|
"grad_norm": 0.7200966145445753,
|
||
|
|
"learning_rate": 9.33990281607661e-06,
|
||
|
|
"loss": 0.2116,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09558679908514023,
|
||
|
|
"step": 2470,
|
||
|
|
"valid_targets_mean": 3284.0,
|
||
|
|
"valid_targets_min": 1852
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.98991935483871,
|
||
|
|
"grad_norm": 0.6342851557173954,
|
||
|
|
"learning_rate": 9.254950228682094e-06,
|
||
|
|
"loss": 0.2365,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.24668997526168823,
|
||
|
|
"step": 2475,
|
||
|
|
"valid_targets_mean": 5133.2,
|
||
|
|
"valid_targets_min": 2527
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.0,
|
||
|
|
"grad_norm": 0.5691209549248927,
|
||
|
|
"learning_rate": 9.1702693009897e-06,
|
||
|
|
"loss": 0.2034,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10690069198608398,
|
||
|
|
"step": 2480,
|
||
|
|
"valid_targets_mean": 4330.8,
|
||
|
|
"valid_targets_min": 2120
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.01008064516129,
|
||
|
|
"grad_norm": 0.5865611302629445,
|
||
|
|
"learning_rate": 9.085862173929076e-06,
|
||
|
|
"loss": 0.1971,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.058146946132183075,
|
||
|
|
"step": 2485,
|
||
|
|
"valid_targets_mean": 2996.4,
|
||
|
|
"valid_targets_min": 1363
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.020161290322581,
|
||
|
|
"grad_norm": 0.6637323317759176,
|
||
|
|
"learning_rate": 9.00173098150755e-06,
|
||
|
|
"loss": 0.1817,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09070940315723419,
|
||
|
|
"step": 2490,
|
||
|
|
"valid_targets_mean": 3229.9,
|
||
|
|
"valid_targets_min": 2005
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.030241935483871,
|
||
|
|
"grad_norm": 0.6526110558448533,
|
||
|
|
"learning_rate": 8.91787785075618e-06,
|
||
|
|
"loss": 0.193,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07960911095142365,
|
||
|
|
"step": 2495,
|
||
|
|
"valid_targets_mean": 3764.6,
|
||
|
|
"valid_targets_min": 1986
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.040322580645161,
|
||
|
|
"grad_norm": 0.6459991888086781,
|
||
|
|
"learning_rate": 8.834304901675999e-06,
|
||
|
|
"loss": 0.1678,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09934863448143005,
|
||
|
|
"step": 2500,
|
||
|
|
"valid_targets_mean": 4030.2,
|
||
|
|
"valid_targets_min": 2193
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.050403225806452,
|
||
|
|
"grad_norm": 0.7622956241451976,
|
||
|
|
"learning_rate": 8.751014247184388e-06,
|
||
|
|
"loss": 0.1918,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11245658993721008,
|
||
|
|
"step": 2505,
|
||
|
|
"valid_targets_mean": 2618.0,
|
||
|
|
"valid_targets_min": 1113
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.060483870967742,
|
||
|
|
"grad_norm": 1.058627809029913,
|
||
|
|
"learning_rate": 8.66800799306168e-06,
|
||
|
|
"loss": 0.1995,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07308492064476013,
|
||
|
|
"step": 2510,
|
||
|
|
"valid_targets_mean": 3026.8,
|
||
|
|
"valid_targets_min": 1608
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.070564516129032,
|
||
|
|
"grad_norm": 0.8084934387362966,
|
||
|
|
"learning_rate": 8.585288237897885e-06,
|
||
|
|
"loss": 0.2071,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06541739404201508,
|
||
|
|
"step": 2515,
|
||
|
|
"valid_targets_mean": 2353.1,
|
||
|
|
"valid_targets_min": 1586
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.080645161290323,
|
||
|
|
"grad_norm": 1.693490881065617,
|
||
|
|
"learning_rate": 8.502857073039678e-06,
|
||
|
|
"loss": 0.1905,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09127959609031677,
|
||
|
|
"step": 2520,
|
||
|
|
"valid_targets_mean": 3504.2,
|
||
|
|
"valid_targets_min": 1279
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.090725806451613,
|
||
|
|
"grad_norm": 0.6924869486528952,
|
||
|
|
"learning_rate": 8.420716582537508e-06,
|
||
|
|
"loss": 0.2145,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1090395450592041,
|
||
|
|
"step": 2525,
|
||
|
|
"valid_targets_mean": 3337.1,
|
||
|
|
"valid_targets_min": 1874
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.100806451612903,
|
||
|
|
"grad_norm": 0.6148523548993665,
|
||
|
|
"learning_rate": 8.33886884309288e-06,
|
||
|
|
"loss": 0.2103,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10533513873815536,
|
||
|
|
"step": 2530,
|
||
|
|
"valid_targets_mean": 3188.9,
|
||
|
|
"valid_targets_min": 1290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.110887096774194,
|
||
|
|
"grad_norm": 0.6660747383316608,
|
||
|
|
"learning_rate": 8.257315924005891e-06,
|
||
|
|
"loss": 0.1748,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06165033578872681,
|
||
|
|
"step": 2535,
|
||
|
|
"valid_targets_mean": 2354.8,
|
||
|
|
"valid_targets_min": 1405
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.120967741935484,
|
||
|
|
"grad_norm": 0.7001866892074488,
|
||
|
|
"learning_rate": 8.176059887122902e-06,
|
||
|
|
"loss": 0.1939,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08742007613182068,
|
||
|
|
"step": 2540,
|
||
|
|
"valid_targets_mean": 2981.0,
|
||
|
|
"valid_targets_min": 332
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.131048387096774,
|
||
|
|
"grad_norm": 0.6332719563250582,
|
||
|
|
"learning_rate": 8.095102786784404e-06,
|
||
|
|
"loss": 0.219,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0933089479804039,
|
||
|
|
"step": 2545,
|
||
|
|
"valid_targets_mean": 3998.8,
|
||
|
|
"valid_targets_min": 1997
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.141129032258065,
|
||
|
|
"grad_norm": 0.6638199862464099,
|
||
|
|
"learning_rate": 8.014446669773061e-06,
|
||
|
|
"loss": 0.1748,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10286946594715118,
|
||
|
|
"step": 2550,
|
||
|
|
"valid_targets_mean": 2951.2,
|
||
|
|
"valid_targets_min": 1735
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.151209677419355,
|
||
|
|
"grad_norm": 0.7824224121364516,
|
||
|
|
"learning_rate": 7.93409357526201e-06,
|
||
|
|
"loss": 0.2029,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07972052693367004,
|
||
|
|
"step": 2555,
|
||
|
|
"valid_targets_mean": 2489.6,
|
||
|
|
"valid_targets_min": 1449
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.161290322580645,
|
||
|
|
"grad_norm": 0.8784052596939906,
|
||
|
|
"learning_rate": 7.854045534763268e-06,
|
||
|
|
"loss": 0.1809,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11835326254367828,
|
||
|
|
"step": 2560,
|
||
|
|
"valid_targets_mean": 2337.1,
|
||
|
|
"valid_targets_min": 488
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.171370967741935,
|
||
|
|
"grad_norm": 0.6390754162733329,
|
||
|
|
"learning_rate": 7.774304572076376e-06,
|
||
|
|
"loss": 0.1787,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10689078271389008,
|
||
|
|
"step": 2565,
|
||
|
|
"valid_targets_mean": 3769.2,
|
||
|
|
"valid_targets_min": 2519
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.181451612903226,
|
||
|
|
"grad_norm": 0.732437011627482,
|
||
|
|
"learning_rate": 7.694872703237236e-06,
|
||
|
|
"loss": 0.1743,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09571628272533417,
|
||
|
|
"step": 2570,
|
||
|
|
"valid_targets_mean": 3143.8,
|
||
|
|
"valid_targets_min": 1288
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.191532258064516,
|
||
|
|
"grad_norm": 0.6543010588636728,
|
||
|
|
"learning_rate": 7.615751936467151e-06,
|
||
|
|
"loss": 0.2064,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14224469661712646,
|
||
|
|
"step": 2575,
|
||
|
|
"valid_targets_mean": 4731.5,
|
||
|
|
"valid_targets_min": 1407
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.201612903225806,
|
||
|
|
"grad_norm": 0.6676555607328929,
|
||
|
|
"learning_rate": 7.536944272122047e-06,
|
||
|
|
"loss": 0.2156,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0936073511838913,
|
||
|
|
"step": 2580,
|
||
|
|
"valid_targets_mean": 3390.5,
|
||
|
|
"valid_targets_min": 1178
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.211693548387097,
|
||
|
|
"grad_norm": 0.6510475569401915,
|
||
|
|
"learning_rate": 7.458451702641895e-06,
|
||
|
|
"loss": 0.2062,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07718747109174728,
|
||
|
|
"step": 2585,
|
||
|
|
"valid_targets_mean": 3298.9,
|
||
|
|
"valid_targets_min": 1542
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.221774193548387,
|
||
|
|
"grad_norm": 0.7493304371423593,
|
||
|
|
"learning_rate": 7.380276212500325e-06,
|
||
|
|
"loss": 0.2001,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07538608461618423,
|
||
|
|
"step": 2590,
|
||
|
|
"valid_targets_mean": 2927.5,
|
||
|
|
"valid_targets_min": 1537
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.231854838709677,
|
||
|
|
"grad_norm": 0.7246220633185994,
|
||
|
|
"learning_rate": 7.302419778154488e-06,
|
||
|
|
"loss": 0.1895,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08335809409618378,
|
||
|
|
"step": 2595,
|
||
|
|
"valid_targets_mean": 2870.4,
|
||
|
|
"valid_targets_min": 1320
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.241935483870968,
|
||
|
|
"grad_norm": 0.6181278857725739,
|
||
|
|
"learning_rate": 7.2248843679950534e-06,
|
||
|
|
"loss": 0.21,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12479527294635773,
|
||
|
|
"step": 2600,
|
||
|
|
"valid_targets_mean": 3470.0,
|
||
|
|
"valid_targets_min": 360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.252016129032258,
|
||
|
|
"grad_norm": 0.7256966133733808,
|
||
|
|
"learning_rate": 7.147671942296468e-06,
|
||
|
|
"loss": 0.1981,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1743425875902176,
|
||
|
|
"step": 2605,
|
||
|
|
"valid_targets_mean": 4168.5,
|
||
|
|
"valid_targets_min": 1100
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.262096774193548,
|
||
|
|
"grad_norm": 0.7386031453280537,
|
||
|
|
"learning_rate": 7.07078445316737e-06,
|
||
|
|
"loss": 0.1993,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12011529505252838,
|
||
|
|
"step": 2610,
|
||
|
|
"valid_targets_mean": 2726.5,
|
||
|
|
"valid_targets_min": 1158
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.272177419354839,
|
||
|
|
"grad_norm": 0.628916059004811,
|
||
|
|
"learning_rate": 6.99422384450126e-06,
|
||
|
|
"loss": 0.1835,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08023014664649963,
|
||
|
|
"step": 2615,
|
||
|
|
"valid_targets_mean": 3292.9,
|
||
|
|
"valid_targets_min": 1038
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.282258064516129,
|
||
|
|
"grad_norm": 0.700393604415949,
|
||
|
|
"learning_rate": 6.9179920519273534e-06,
|
||
|
|
"loss": 0.1746,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09363005310297012,
|
||
|
|
"step": 2620,
|
||
|
|
"valid_targets_mean": 3017.2,
|
||
|
|
"valid_targets_min": 1231
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.292338709677419,
|
||
|
|
"grad_norm": 0.7907257658880886,
|
||
|
|
"learning_rate": 6.842091002761606e-06,
|
||
|
|
"loss": 0.1801,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09486392140388489,
|
||
|
|
"step": 2625,
|
||
|
|
"valid_targets_mean": 3091.9,
|
||
|
|
"valid_targets_min": 370
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.30241935483871,
|
||
|
|
"grad_norm": 0.6222542409675017,
|
||
|
|
"learning_rate": 6.766522615958051e-06,
|
||
|
|
"loss": 0.1786,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06968790292739868,
|
||
|
|
"step": 2630,
|
||
|
|
"valid_targets_mean": 2889.6,
|
||
|
|
"valid_targets_min": 1145
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.3125,
|
||
|
|
"grad_norm": 0.6678478818510593,
|
||
|
|
"learning_rate": 6.691288802060216e-06,
|
||
|
|
"loss": 0.1911,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11301000416278839,
|
||
|
|
"step": 2635,
|
||
|
|
"valid_targets_mean": 3654.6,
|
||
|
|
"valid_targets_min": 1398
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.32258064516129,
|
||
|
|
"grad_norm": 0.6350689658243185,
|
||
|
|
"learning_rate": 6.616391463152872e-06,
|
||
|
|
"loss": 0.1779,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08706215769052505,
|
||
|
|
"step": 2640,
|
||
|
|
"valid_targets_mean": 3271.4,
|
||
|
|
"valid_targets_min": 1567
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.332661290322581,
|
||
|
|
"grad_norm": 0.6343864062771389,
|
||
|
|
"learning_rate": 6.5418324928139244e-06,
|
||
|
|
"loss": 0.1832,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07230157405138016,
|
||
|
|
"step": 2645,
|
||
|
|
"valid_targets_mean": 2072.0,
|
||
|
|
"valid_targets_min": 294
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.342741935483871,
|
||
|
|
"grad_norm": 0.6829275609606014,
|
||
|
|
"learning_rate": 6.4676137760665214e-06,
|
||
|
|
"loss": 0.198,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09237651526927948,
|
||
|
|
"step": 2650,
|
||
|
|
"valid_targets_mean": 3193.6,
|
||
|
|
"valid_targets_min": 1561
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.352822580645161,
|
||
|
|
"grad_norm": 0.7070815489161848,
|
||
|
|
"learning_rate": 6.3937371893314325e-06,
|
||
|
|
"loss": 0.1841,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10177172720432281,
|
||
|
|
"step": 2655,
|
||
|
|
"valid_targets_mean": 3580.9,
|
||
|
|
"valid_targets_min": 1807
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.362903225806452,
|
||
|
|
"grad_norm": 1.194444989183062,
|
||
|
|
"learning_rate": 6.320204600379587e-06,
|
||
|
|
"loss": 0.2145,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11016004532575607,
|
||
|
|
"step": 2660,
|
||
|
|
"valid_targets_mean": 3116.4,
|
||
|
|
"valid_targets_min": 1152
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.372983870967742,
|
||
|
|
"grad_norm": 0.8183678524734058,
|
||
|
|
"learning_rate": 6.247017868284839e-06,
|
||
|
|
"loss": 0.2065,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10302819311618805,
|
||
|
|
"step": 2665,
|
||
|
|
"valid_targets_mean": 2671.6,
|
||
|
|
"valid_targets_min": 1165
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.383064516129032,
|
||
|
|
"grad_norm": 0.6904928811956119,
|
||
|
|
"learning_rate": 6.174178843376999e-06,
|
||
|
|
"loss": 0.204,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1024511381983757,
|
||
|
|
"step": 2670,
|
||
|
|
"valid_targets_mean": 2585.1,
|
||
|
|
"valid_targets_min": 332
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.393145161290323,
|
||
|
|
"grad_norm": 0.8562556068075553,
|
||
|
|
"learning_rate": 6.101689367195027e-06,
|
||
|
|
"loss": 0.1992,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1049245148897171,
|
||
|
|
"step": 2675,
|
||
|
|
"valid_targets_mean": 3106.8,
|
||
|
|
"valid_targets_min": 981
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.403225806451613,
|
||
|
|
"grad_norm": 0.6926061587265497,
|
||
|
|
"learning_rate": 6.0295512724404945e-06,
|
||
|
|
"loss": 0.1924,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09678656607866287,
|
||
|
|
"step": 2680,
|
||
|
|
"valid_targets_mean": 3331.4,
|
||
|
|
"valid_targets_min": 1997
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.413306451612903,
|
||
|
|
"grad_norm": 0.6585891358301418,
|
||
|
|
"learning_rate": 5.957766382931216e-06,
|
||
|
|
"loss": 0.2257,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10457658022642136,
|
||
|
|
"step": 2685,
|
||
|
|
"valid_targets_mean": 3408.2,
|
||
|
|
"valid_targets_min": 1410
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.423387096774194,
|
||
|
|
"grad_norm": 0.6796361799834924,
|
||
|
|
"learning_rate": 5.88633651355518e-06,
|
||
|
|
"loss": 0.1903,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0864211916923523,
|
||
|
|
"step": 2690,
|
||
|
|
"valid_targets_mean": 3498.2,
|
||
|
|
"valid_targets_min": 1548
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.433467741935484,
|
||
|
|
"grad_norm": 0.7530672571838339,
|
||
|
|
"learning_rate": 5.815263470224639e-06,
|
||
|
|
"loss": 0.1811,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07902814447879791,
|
||
|
|
"step": 2695,
|
||
|
|
"valid_targets_mean": 2516.8,
|
||
|
|
"valid_targets_min": 1514
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.443548387096774,
|
||
|
|
"grad_norm": 0.6722226346316681,
|
||
|
|
"learning_rate": 5.744549049830472e-06,
|
||
|
|
"loss": 0.1846,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08121909946203232,
|
||
|
|
"step": 2700,
|
||
|
|
"valid_targets_mean": 3172.6,
|
||
|
|
"valid_targets_min": 1282
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.453629032258064,
|
||
|
|
"grad_norm": 0.7139254584157829,
|
||
|
|
"learning_rate": 5.674195040196702e-06,
|
||
|
|
"loss": 0.198,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0766548365354538,
|
||
|
|
"step": 2705,
|
||
|
|
"valid_targets_mean": 2970.8,
|
||
|
|
"valid_targets_min": 361
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.463709677419355,
|
||
|
|
"grad_norm": 0.6444868116713358,
|
||
|
|
"learning_rate": 5.6042032200353694e-06,
|
||
|
|
"loss": 0.1838,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08874198794364929,
|
||
|
|
"step": 2710,
|
||
|
|
"valid_targets_mean": 3077.6,
|
||
|
|
"valid_targets_min": 1200
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.473790322580645,
|
||
|
|
"grad_norm": 0.687901786486838,
|
||
|
|
"learning_rate": 5.534575358901522e-06,
|
||
|
|
"loss": 0.1788,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12347336858510971,
|
||
|
|
"step": 2715,
|
||
|
|
"valid_targets_mean": 3733.6,
|
||
|
|
"valid_targets_min": 1494
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.483870967741936,
|
||
|
|
"grad_norm": 0.6524135751244532,
|
||
|
|
"learning_rate": 5.465313217148478e-06,
|
||
|
|
"loss": 0.1942,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08183799684047699,
|
||
|
|
"step": 2720,
|
||
|
|
"valid_targets_mean": 3643.6,
|
||
|
|
"valid_targets_min": 2271
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.493951612903226,
|
||
|
|
"grad_norm": 0.6850877730731605,
|
||
|
|
"learning_rate": 5.396418545883318e-06,
|
||
|
|
"loss": 0.2085,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09083373099565506,
|
||
|
|
"step": 2725,
|
||
|
|
"valid_targets_mean": 3266.8,
|
||
|
|
"valid_targets_min": 1912
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.504032258064516,
|
||
|
|
"grad_norm": 0.6091503833935844,
|
||
|
|
"learning_rate": 5.327893086922626e-06,
|
||
|
|
"loss": 0.2013,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09076303988695145,
|
||
|
|
"step": 2730,
|
||
|
|
"valid_targets_mean": 3807.9,
|
||
|
|
"valid_targets_min": 1216
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.514112903225806,
|
||
|
|
"grad_norm": 0.7264968253820917,
|
||
|
|
"learning_rate": 5.259738572748447e-06,
|
||
|
|
"loss": 0.1912,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1065942570567131,
|
||
|
|
"step": 2735,
|
||
|
|
"valid_targets_mean": 2888.5,
|
||
|
|
"valid_targets_min": 888
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.524193548387097,
|
||
|
|
"grad_norm": 0.6870495015252566,
|
||
|
|
"learning_rate": 5.191956726464487e-06,
|
||
|
|
"loss": 0.1837,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09205818921327591,
|
||
|
|
"step": 2740,
|
||
|
|
"valid_targets_mean": 2970.9,
|
||
|
|
"valid_targets_min": 1384
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.534274193548387,
|
||
|
|
"grad_norm": 0.6437426909509353,
|
||
|
|
"learning_rate": 5.124549261752525e-06,
|
||
|
|
"loss": 0.1871,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12780871987342834,
|
||
|
|
"step": 2745,
|
||
|
|
"valid_targets_mean": 3851.4,
|
||
|
|
"valid_targets_min": 1205
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.544354838709677,
|
||
|
|
"grad_norm": 0.6598134508230882,
|
||
|
|
"learning_rate": 5.057517882829132e-06,
|
||
|
|
"loss": 0.202,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0748056024312973,
|
||
|
|
"step": 2750,
|
||
|
|
"valid_targets_mean": 2986.5,
|
||
|
|
"valid_targets_min": 339
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.554435483870968,
|
||
|
|
"grad_norm": 0.6678686679492717,
|
||
|
|
"learning_rate": 4.990864284402559e-06,
|
||
|
|
"loss": 0.1886,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10124251246452332,
|
||
|
|
"step": 2755,
|
||
|
|
"valid_targets_mean": 3355.1,
|
||
|
|
"valid_targets_min": 1847
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.564516129032258,
|
||
|
|
"grad_norm": 0.749326026222191,
|
||
|
|
"learning_rate": 4.924590151629871e-06,
|
||
|
|
"loss": 0.2002,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.17438571155071259,
|
||
|
|
"step": 2760,
|
||
|
|
"valid_targets_mean": 3584.8,
|
||
|
|
"valid_targets_min": 2284
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.574596774193548,
|
||
|
|
"grad_norm": 0.6775931147721376,
|
||
|
|
"learning_rate": 4.858697160074386e-06,
|
||
|
|
"loss": 0.195,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1284763664007187,
|
||
|
|
"step": 2765,
|
||
|
|
"valid_targets_mean": 3360.8,
|
||
|
|
"valid_targets_min": 1190
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.584677419354839,
|
||
|
|
"grad_norm": 0.7420820950139783,
|
||
|
|
"learning_rate": 4.793186975663289e-06,
|
||
|
|
"loss": 0.1994,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0818614661693573,
|
||
|
|
"step": 2770,
|
||
|
|
"valid_targets_mean": 2338.4,
|
||
|
|
"valid_targets_min": 438
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.594758064516129,
|
||
|
|
"grad_norm": 0.6841166352057325,
|
||
|
|
"learning_rate": 4.7280612546454975e-06,
|
||
|
|
"loss": 0.1649,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06983286142349243,
|
||
|
|
"step": 2775,
|
||
|
|
"valid_targets_mean": 3660.2,
|
||
|
|
"valid_targets_min": 1247
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.604838709677419,
|
||
|
|
"grad_norm": 0.7396912950280988,
|
||
|
|
"learning_rate": 4.66332164354983e-06,
|
||
|
|
"loss": 0.1826,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11039724200963974,
|
||
|
|
"step": 2780,
|
||
|
|
"valid_targets_mean": 3117.0,
|
||
|
|
"valid_targets_min": 1044
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.61491935483871,
|
||
|
|
"grad_norm": 0.6582481516163271,
|
||
|
|
"learning_rate": 4.59896977914333e-06,
|
||
|
|
"loss": 0.1824,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09300467371940613,
|
||
|
|
"step": 2785,
|
||
|
|
"valid_targets_mean": 2817.6,
|
||
|
|
"valid_targets_min": 1680
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.625,
|
||
|
|
"grad_norm": 0.6290026064515466,
|
||
|
|
"learning_rate": 4.535007288389923e-06,
|
||
|
|
"loss": 0.1875,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07869211584329605,
|
||
|
|
"step": 2790,
|
||
|
|
"valid_targets_mean": 3965.6,
|
||
|
|
"valid_targets_min": 1736
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.63508064516129,
|
||
|
|
"grad_norm": 0.6321621632961499,
|
||
|
|
"learning_rate": 4.471435788409273e-06,
|
||
|
|
"loss": 0.1885,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07932322472333908,
|
||
|
|
"step": 2795,
|
||
|
|
"valid_targets_mean": 2513.1,
|
||
|
|
"valid_targets_min": 1211
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.645161290322581,
|
||
|
|
"grad_norm": 0.7518398697704604,
|
||
|
|
"learning_rate": 4.408256886435873e-06,
|
||
|
|
"loss": 0.1933,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12717461585998535,
|
||
|
|
"step": 2800,
|
||
|
|
"valid_targets_mean": 3683.9,
|
||
|
|
"valid_targets_min": 1275
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.655241935483871,
|
||
|
|
"grad_norm": 0.6698024376761166,
|
||
|
|
"learning_rate": 4.345472179778445e-06,
|
||
|
|
"loss": 0.1939,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07357233762741089,
|
||
|
|
"step": 2805,
|
||
|
|
"valid_targets_mean": 3326.8,
|
||
|
|
"valid_targets_min": 2077
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.665322580645161,
|
||
|
|
"grad_norm": 0.6563716649630288,
|
||
|
|
"learning_rate": 4.2830832557795435e-06,
|
||
|
|
"loss": 0.1947,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11173593997955322,
|
||
|
|
"step": 2810,
|
||
|
|
"valid_targets_mean": 3727.4,
|
||
|
|
"valid_targets_min": 1962
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.675403225806452,
|
||
|
|
"grad_norm": 0.6357217822816715,
|
||
|
|
"learning_rate": 4.221091691775421e-06,
|
||
|
|
"loss": 0.1908,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06565943360328674,
|
||
|
|
"step": 2815,
|
||
|
|
"valid_targets_mean": 2688.8,
|
||
|
|
"valid_targets_min": 1084
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.685483870967742,
|
||
|
|
"grad_norm": 0.7529099454803696,
|
||
|
|
"learning_rate": 4.159499055056137e-06,
|
||
|
|
"loss": 0.1759,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08579786121845245,
|
||
|
|
"step": 2820,
|
||
|
|
"valid_targets_mean": 2878.9,
|
||
|
|
"valid_targets_min": 1637
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.695564516129032,
|
||
|
|
"grad_norm": 0.7195373176517006,
|
||
|
|
"learning_rate": 4.098306902825966e-06,
|
||
|
|
"loss": 0.1811,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09593765437602997,
|
||
|
|
"step": 2825,
|
||
|
|
"valid_targets_mean": 2664.8,
|
||
|
|
"valid_targets_min": 1269
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.705645161290323,
|
||
|
|
"grad_norm": 0.6654359775221081,
|
||
|
|
"learning_rate": 4.037516782163995e-06,
|
||
|
|
"loss": 0.1878,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11844100058078766,
|
||
|
|
"step": 2830,
|
||
|
|
"valid_targets_mean": 3273.9,
|
||
|
|
"valid_targets_min": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.715725806451613,
|
||
|
|
"grad_norm": 0.6794810700340962,
|
||
|
|
"learning_rate": 3.977130229985038e-06,
|
||
|
|
"loss": 0.1917,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11235381662845612,
|
||
|
|
"step": 2835,
|
||
|
|
"valid_targets_mean": 3560.2,
|
||
|
|
"valid_targets_min": 1354
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.725806451612903,
|
||
|
|
"grad_norm": 0.6287485616422421,
|
||
|
|
"learning_rate": 3.9171487730007496e-06,
|
||
|
|
"loss": 0.1977,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09734977781772614,
|
||
|
|
"step": 2840,
|
||
|
|
"valid_targets_mean": 4227.5,
|
||
|
|
"valid_targets_min": 1949
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.735887096774194,
|
||
|
|
"grad_norm": 0.6671462276412441,
|
||
|
|
"learning_rate": 3.857573927681042e-06,
|
||
|
|
"loss": 0.1876,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11975468695163727,
|
||
|
|
"step": 2845,
|
||
|
|
"valid_targets_mean": 4551.6,
|
||
|
|
"valid_targets_min": 2351
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.745967741935484,
|
||
|
|
"grad_norm": 0.6213637142751531,
|
||
|
|
"learning_rate": 3.798407200215752e-06,
|
||
|
|
"loss": 0.2329,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07167664170265198,
|
||
|
|
"step": 2850,
|
||
|
|
"valid_targets_mean": 2491.8,
|
||
|
|
"valid_targets_min": 366
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.756048387096774,
|
||
|
|
"grad_norm": 0.6909896697573505,
|
||
|
|
"learning_rate": 3.7396500864765585e-06,
|
||
|
|
"loss": 0.203,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1118776947259903,
|
||
|
|
"step": 2855,
|
||
|
|
"valid_targets_mean": 3490.9,
|
||
|
|
"valid_targets_min": 387
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.766129032258064,
|
||
|
|
"grad_norm": 0.5774516637581999,
|
||
|
|
"learning_rate": 3.6813040719791394e-06,
|
||
|
|
"loss": 0.1918,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07964398711919785,
|
||
|
|
"step": 2860,
|
||
|
|
"valid_targets_mean": 3919.1,
|
||
|
|
"valid_targets_min": 2230
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.776209677419355,
|
||
|
|
"grad_norm": 0.549586665574616,
|
||
|
|
"learning_rate": 3.6233706318456508e-06,
|
||
|
|
"loss": 0.176,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09143416583538055,
|
||
|
|
"step": 2865,
|
||
|
|
"valid_targets_mean": 3261.9,
|
||
|
|
"valid_targets_min": 2162
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.786290322580645,
|
||
|
|
"grad_norm": 0.6444476601211784,
|
||
|
|
"learning_rate": 3.5658512307674075e-06,
|
||
|
|
"loss": 0.1876,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08233319222927094,
|
||
|
|
"step": 2870,
|
||
|
|
"valid_targets_mean": 3220.0,
|
||
|
|
"valid_targets_min": 1989
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.796370967741936,
|
||
|
|
"grad_norm": 0.6962832887556154,
|
||
|
|
"learning_rate": 3.508747322967867e-06,
|
||
|
|
"loss": 0.1811,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11017526686191559,
|
||
|
|
"step": 2875,
|
||
|
|
"valid_targets_mean": 2871.8,
|
||
|
|
"valid_targets_min": 1491
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.806451612903226,
|
||
|
|
"grad_norm": 0.6517017014439975,
|
||
|
|
"learning_rate": 3.452060352165838e-06,
|
||
|
|
"loss": 0.1932,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08151370286941528,
|
||
|
|
"step": 2880,
|
||
|
|
"valid_targets_mean": 3313.5,
|
||
|
|
"valid_targets_min": 1335
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.816532258064516,
|
||
|
|
"grad_norm": 0.719484023254342,
|
||
|
|
"learning_rate": 3.395791751539015e-06,
|
||
|
|
"loss": 0.1755,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08000224083662033,
|
||
|
|
"step": 2885,
|
||
|
|
"valid_targets_mean": 3127.6,
|
||
|
|
"valid_targets_min": 358
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.826612903225806,
|
||
|
|
"grad_norm": 0.6369218865217999,
|
||
|
|
"learning_rate": 3.3399429436877285e-06,
|
||
|
|
"loss": 0.1655,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07361101359128952,
|
||
|
|
"step": 2890,
|
||
|
|
"valid_targets_mean": 2926.4,
|
||
|
|
"valid_targets_min": 1446
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.836693548387097,
|
||
|
|
"grad_norm": 0.6772136699540655,
|
||
|
|
"learning_rate": 3.2845153405989592e-06,
|
||
|
|
"loss": 0.1845,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07584335654973984,
|
||
|
|
"step": 2895,
|
||
|
|
"valid_targets_mean": 2697.2,
|
||
|
|
"valid_targets_min": 1609
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.846774193548387,
|
||
|
|
"grad_norm": 0.6459077660177165,
|
||
|
|
"learning_rate": 3.2295103436106736e-06,
|
||
|
|
"loss": 0.1979,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08547687530517578,
|
||
|
|
"step": 2900,
|
||
|
|
"valid_targets_mean": 3031.6,
|
||
|
|
"valid_targets_min": 370
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.856854838709677,
|
||
|
|
"grad_norm": 0.6598690730337423,
|
||
|
|
"learning_rate": 3.174929343376374e-06,
|
||
|
|
"loss": 0.171,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08729280531406403,
|
||
|
|
"step": 2905,
|
||
|
|
"valid_targets_mean": 3624.8,
|
||
|
|
"valid_targets_min": 1177
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.866935483870968,
|
||
|
|
"grad_norm": 0.7156778335113698,
|
||
|
|
"learning_rate": 3.1207737198299503e-06,
|
||
|
|
"loss": 0.1768,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.067437544465065,
|
||
|
|
"step": 2910,
|
||
|
|
"valid_targets_mean": 2517.5,
|
||
|
|
"valid_targets_min": 396
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.877016129032258,
|
||
|
|
"grad_norm": 0.6908451397615742,
|
||
|
|
"learning_rate": 3.0670448421507747e-06,
|
||
|
|
"loss": 0.1915,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08772405982017517,
|
||
|
|
"step": 2915,
|
||
|
|
"valid_targets_mean": 2731.4,
|
||
|
|
"valid_targets_min": 1180
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.887096774193548,
|
||
|
|
"grad_norm": 0.691623307355018,
|
||
|
|
"learning_rate": 3.0137440687291008e-06,
|
||
|
|
"loss": 0.2129,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13214321434497833,
|
||
|
|
"step": 2920,
|
||
|
|
"valid_targets_mean": 3643.8,
|
||
|
|
"valid_targets_min": 329
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.897177419354839,
|
||
|
|
"grad_norm": 0.6079767501684598,
|
||
|
|
"learning_rate": 2.9608727471317177e-06,
|
||
|
|
"loss": 0.1882,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09109807014465332,
|
||
|
|
"step": 2925,
|
||
|
|
"valid_targets_mean": 3781.1,
|
||
|
|
"valid_targets_min": 2027
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.907258064516129,
|
||
|
|
"grad_norm": 0.7098676441409183,
|
||
|
|
"learning_rate": 2.908432214067891e-06,
|
||
|
|
"loss": 0.211,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10125988721847534,
|
||
|
|
"step": 2930,
|
||
|
|
"valid_targets_mean": 2966.9,
|
||
|
|
"valid_targets_min": 1135
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.917338709677419,
|
||
|
|
"grad_norm": 0.6882308021370365,
|
||
|
|
"learning_rate": 2.856423795355532e-06,
|
||
|
|
"loss": 0.2009,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07672426104545593,
|
||
|
|
"step": 2935,
|
||
|
|
"valid_targets_mean": 3166.6,
|
||
|
|
"valid_targets_min": 1298
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.92741935483871,
|
||
|
|
"grad_norm": 0.6792808050314645,
|
||
|
|
"learning_rate": 2.8048488058877233e-06,
|
||
|
|
"loss": 0.2035,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09512323141098022,
|
||
|
|
"step": 2940,
|
||
|
|
"valid_targets_mean": 2997.1,
|
||
|
|
"valid_targets_min": 1397
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.9375,
|
||
|
|
"grad_norm": 0.7144321377908089,
|
||
|
|
"learning_rate": 2.753708549599443e-06,
|
||
|
|
"loss": 0.1828,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10287399590015411,
|
||
|
|
"step": 2945,
|
||
|
|
"valid_targets_mean": 2827.2,
|
||
|
|
"valid_targets_min": 1402
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.94758064516129,
|
||
|
|
"grad_norm": 0.7472729896129943,
|
||
|
|
"learning_rate": 2.7030043194346235e-06,
|
||
|
|
"loss": 0.1939,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07781745493412018,
|
||
|
|
"step": 2950,
|
||
|
|
"valid_targets_mean": 4722.2,
|
||
|
|
"valid_targets_min": 1674
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.957661290322581,
|
||
|
|
"grad_norm": 0.925212297192844,
|
||
|
|
"learning_rate": 2.652737397313421e-06,
|
||
|
|
"loss": 0.1908,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09082068502902985,
|
||
|
|
"step": 2955,
|
||
|
|
"valid_targets_mean": 2822.4,
|
||
|
|
"valid_targets_min": 1463
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.967741935483871,
|
||
|
|
"grad_norm": 0.6251820300218922,
|
||
|
|
"learning_rate": 2.6029090540998602e-06,
|
||
|
|
"loss": 0.1802,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09201687574386597,
|
||
|
|
"step": 2960,
|
||
|
|
"valid_targets_mean": 4578.1,
|
||
|
|
"valid_targets_min": 1868
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.977822580645161,
|
||
|
|
"grad_norm": 0.6474326081901013,
|
||
|
|
"learning_rate": 2.553520549569668e-06,
|
||
|
|
"loss": 0.1981,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07766154408454895,
|
||
|
|
"step": 2965,
|
||
|
|
"valid_targets_mean": 3138.8,
|
||
|
|
"valid_targets_min": 1020
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.987903225806452,
|
||
|
|
"grad_norm": 0.6735863153512439,
|
||
|
|
"learning_rate": 2.5045731323784294e-06,
|
||
|
|
"loss": 0.1894,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08804526180028915,
|
||
|
|
"step": 2970,
|
||
|
|
"valid_targets_mean": 3002.1,
|
||
|
|
"valid_targets_min": 1133
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.997983870967742,
|
||
|
|
"grad_norm": 0.6185967706562228,
|
||
|
|
"learning_rate": 2.4560680400300195e-06,
|
||
|
|
"loss": 0.1986,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09211928397417068,
|
||
|
|
"step": 2975,
|
||
|
|
"valid_targets_mean": 4141.1,
|
||
|
|
"valid_targets_min": 2767
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.008064516129032,
|
||
|
|
"grad_norm": 0.7231117673573182,
|
||
|
|
"learning_rate": 2.4080064988453322e-06,
|
||
|
|
"loss": 0.1851,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07972070574760437,
|
||
|
|
"step": 2980,
|
||
|
|
"valid_targets_mean": 3055.9,
|
||
|
|
"valid_targets_min": 1624
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.018145161290323,
|
||
|
|
"grad_norm": 0.6845063225042552,
|
||
|
|
"learning_rate": 2.3603897239312403e-06,
|
||
|
|
"loss": 0.1939,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12189587950706482,
|
||
|
|
"step": 2985,
|
||
|
|
"valid_targets_mean": 3994.5,
|
||
|
|
"valid_targets_min": 1257
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.028225806451613,
|
||
|
|
"grad_norm": 0.6589735778709775,
|
||
|
|
"learning_rate": 2.313218919149922e-06,
|
||
|
|
"loss": 0.191,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10585145652294159,
|
||
|
|
"step": 2990,
|
||
|
|
"valid_targets_mean": 2971.6,
|
||
|
|
"valid_targets_min": 1330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.038306451612903,
|
||
|
|
"grad_norm": 0.7184798071415125,
|
||
|
|
"learning_rate": 2.2664952770883764e-06,
|
||
|
|
"loss": 0.187,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10647550225257874,
|
||
|
|
"step": 2995,
|
||
|
|
"valid_targets_mean": 2577.1,
|
||
|
|
"valid_targets_min": 1271
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.048387096774194,
|
||
|
|
"grad_norm": 0.6947469001633894,
|
||
|
|
"learning_rate": 2.2202199790283084e-06,
|
||
|
|
"loss": 0.1888,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0668880045413971,
|
||
|
|
"step": 3000,
|
||
|
|
"valid_targets_mean": 2420.0,
|
||
|
|
"valid_targets_min": 1491
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.058467741935484,
|
||
|
|
"grad_norm": 0.6074572853781195,
|
||
|
|
"learning_rate": 2.174394194916252e-06,
|
||
|
|
"loss": 0.2111,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07996638864278793,
|
||
|
|
"step": 3005,
|
||
|
|
"valid_targets_mean": 2873.6,
|
||
|
|
"valid_targets_min": 1165
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.068548387096774,
|
||
|
|
"grad_norm": 0.6099142778729095,
|
||
|
|
"learning_rate": 2.1290190833339852e-06,
|
||
|
|
"loss": 0.1931,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08393967151641846,
|
||
|
|
"step": 3010,
|
||
|
|
"valid_targets_mean": 3295.0,
|
||
|
|
"valid_targets_min": 2151
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.078629032258065,
|
||
|
|
"grad_norm": 0.8917073328163435,
|
||
|
|
"learning_rate": 2.084095791469234e-06,
|
||
|
|
"loss": 0.1842,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1370064616203308,
|
||
|
|
"step": 3015,
|
||
|
|
"valid_targets_mean": 3948.1,
|
||
|
|
"valid_targets_min": 2125
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.088709677419355,
|
||
|
|
"grad_norm": 0.7034243395425919,
|
||
|
|
"learning_rate": 2.0396254550866888e-06,
|
||
|
|
"loss": 0.161,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06838004291057587,
|
||
|
|
"step": 3020,
|
||
|
|
"valid_targets_mean": 2788.1,
|
||
|
|
"valid_targets_min": 332
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.098790322580645,
|
||
|
|
"grad_norm": 0.6383218835109817,
|
||
|
|
"learning_rate": 1.9956091984992778e-06,
|
||
|
|
"loss": 0.1774,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08986695110797882,
|
||
|
|
"step": 3025,
|
||
|
|
"valid_targets_mean": 3469.9,
|
||
|
|
"valid_targets_min": 1439
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.108870967741935,
|
||
|
|
"grad_norm": 0.7317322078355529,
|
||
|
|
"learning_rate": 1.9520481345397345e-06,
|
||
|
|
"loss": 0.1808,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13795316219329834,
|
||
|
|
"step": 3030,
|
||
|
|
"valid_targets_mean": 3459.5,
|
||
|
|
"valid_targets_min": 1364
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.118951612903226,
|
||
|
|
"grad_norm": 0.6411031027784116,
|
||
|
|
"learning_rate": 1.908943364532474e-06,
|
||
|
|
"loss": 0.1811,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.061361946165561676,
|
||
|
|
"step": 3035,
|
||
|
|
"valid_targets_mean": 2649.1,
|
||
|
|
"valid_targets_min": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.129032258064516,
|
||
|
|
"grad_norm": 0.7014456256928921,
|
||
|
|
"learning_rate": 1.8662959782657465e-06,
|
||
|
|
"loss": 0.1903,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0989256203174591,
|
||
|
|
"step": 3040,
|
||
|
|
"valid_targets_mean": 3231.9,
|
||
|
|
"valid_targets_min": 438
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.139112903225806,
|
||
|
|
"grad_norm": 0.6561526885433161,
|
||
|
|
"learning_rate": 1.8241070539640881e-06,
|
||
|
|
"loss": 0.1771,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06555653363466263,
|
||
|
|
"step": 3045,
|
||
|
|
"valid_targets_mean": 2576.1,
|
||
|
|
"valid_targets_min": 392
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.149193548387097,
|
||
|
|
"grad_norm": 0.6293411215123433,
|
||
|
|
"learning_rate": 1.7823776582610474e-06,
|
||
|
|
"loss": 0.1858,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0799480527639389,
|
||
|
|
"step": 3050,
|
||
|
|
"valid_targets_mean": 4497.2,
|
||
|
|
"valid_targets_min": 1195
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.159274193548387,
|
||
|
|
"grad_norm": 0.6794746712197295,
|
||
|
|
"learning_rate": 1.7411088461722281e-06,
|
||
|
|
"loss": 0.1792,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08716973662376404,
|
||
|
|
"step": 3055,
|
||
|
|
"valid_targets_mean": 3232.9,
|
||
|
|
"valid_targets_min": 1828
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.169354838709677,
|
||
|
|
"grad_norm": 0.6930921139214037,
|
||
|
|
"learning_rate": 1.7003016610686197e-06,
|
||
|
|
"loss": 0.1802,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07693126797676086,
|
||
|
|
"step": 3060,
|
||
|
|
"valid_targets_mean": 2518.6,
|
||
|
|
"valid_targets_min": 1290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.179435483870968,
|
||
|
|
"grad_norm": 0.6220499783338116,
|
||
|
|
"learning_rate": 1.6599571346502186e-06,
|
||
|
|
"loss": 0.1746,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05916108936071396,
|
||
|
|
"step": 3065,
|
||
|
|
"valid_targets_mean": 3073.8,
|
||
|
|
"valid_targets_min": 1289
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.189516129032258,
|
||
|
|
"grad_norm": 0.6844041130308192,
|
||
|
|
"learning_rate": 1.620076286919927e-06,
|
||
|
|
"loss": 0.1882,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0733652114868164,
|
||
|
|
"step": 3070,
|
||
|
|
"valid_targets_mean": 2450.0,
|
||
|
|
"valid_targets_min": 919
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.199596774193548,
|
||
|
|
"grad_norm": 0.6863573943090968,
|
||
|
|
"learning_rate": 1.580660126157787e-06,
|
||
|
|
"loss": 0.2161,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08729487657546997,
|
||
|
|
"step": 3075,
|
||
|
|
"valid_targets_mean": 3391.9,
|
||
|
|
"valid_targets_min": 1083
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.209677419354839,
|
||
|
|
"grad_norm": 0.6510811729364249,
|
||
|
|
"learning_rate": 1.5417096488954797e-06,
|
||
|
|
"loss": 0.1866,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09396573901176453,
|
||
|
|
"step": 3080,
|
||
|
|
"valid_targets_mean": 3791.6,
|
||
|
|
"valid_targets_min": 2239
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.219758064516129,
|
||
|
|
"grad_norm": 0.7328195091943156,
|
||
|
|
"learning_rate": 1.5032258398911315e-06,
|
||
|
|
"loss": 0.1936,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07890182733535767,
|
||
|
|
"step": 3085,
|
||
|
|
"valid_targets_mean": 2664.9,
|
||
|
|
"valid_targets_min": 393
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.229838709677419,
|
||
|
|
"grad_norm": 0.6963592412741678,
|
||
|
|
"learning_rate": 1.4652096721044062e-06,
|
||
|
|
"loss": 0.1764,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09602189064025879,
|
||
|
|
"step": 3090,
|
||
|
|
"valid_targets_mean": 3316.9,
|
||
|
|
"valid_targets_min": 1483
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.23991935483871,
|
||
|
|
"grad_norm": 0.7630100008896681,
|
||
|
|
"learning_rate": 1.4276621066719276e-06,
|
||
|
|
"loss": 0.1913,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1523745208978653,
|
||
|
|
"step": 3095,
|
||
|
|
"valid_targets_mean": 3007.5,
|
||
|
|
"valid_targets_min": 1426
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.25,
|
||
|
|
"grad_norm": 0.880419529636551,
|
||
|
|
"learning_rate": 1.3905840928829628e-06,
|
||
|
|
"loss": 0.1675,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06772480905056,
|
||
|
|
"step": 3100,
|
||
|
|
"valid_targets_mean": 2704.9,
|
||
|
|
"valid_targets_min": 1224
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.26008064516129,
|
||
|
|
"grad_norm": 0.6910134351032695,
|
||
|
|
"learning_rate": 1.3539765681554352e-06,
|
||
|
|
"loss": 0.1893,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06466814875602722,
|
||
|
|
"step": 3105,
|
||
|
|
"valid_targets_mean": 3110.1,
|
||
|
|
"valid_targets_min": 526
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.270161290322581,
|
||
|
|
"grad_norm": 0.7048255440383823,
|
||
|
|
"learning_rate": 1.317840458012203e-06,
|
||
|
|
"loss": 0.1812,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06939183175563812,
|
||
|
|
"step": 3110,
|
||
|
|
"valid_targets_mean": 2149.4,
|
||
|
|
"valid_targets_min": 290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.280241935483871,
|
||
|
|
"grad_norm": 0.6784502645153225,
|
||
|
|
"learning_rate": 1.2821766760576847e-06,
|
||
|
|
"loss": 0.1796,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09158091247081757,
|
||
|
|
"step": 3115,
|
||
|
|
"valid_targets_mean": 3475.8,
|
||
|
|
"valid_targets_min": 2136
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.290322580645161,
|
||
|
|
"grad_norm": 0.6049443972017243,
|
||
|
|
"learning_rate": 1.246986123954752e-06,
|
||
|
|
"loss": 0.1785,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09276722371578217,
|
||
|
|
"step": 3120,
|
||
|
|
"valid_targets_mean": 3659.8,
|
||
|
|
"valid_targets_min": 1617
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.300403225806452,
|
||
|
|
"grad_norm": 0.6741527665111027,
|
||
|
|
"learning_rate": 1.2122696914019238e-06,
|
||
|
|
"loss": 0.1935,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12684494256973267,
|
||
|
|
"step": 3125,
|
||
|
|
"valid_targets_mean": 4082.9,
|
||
|
|
"valid_targets_min": 2167
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.310483870967742,
|
||
|
|
"grad_norm": 0.6169712521724912,
|
||
|
|
"learning_rate": 1.1780282561108768e-06,
|
||
|
|
"loss": 0.182,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06459090113639832,
|
||
|
|
"step": 3130,
|
||
|
|
"valid_targets_mean": 2619.1,
|
||
|
|
"valid_targets_min": 1771
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.320564516129032,
|
||
|
|
"grad_norm": 0.7543641728857878,
|
||
|
|
"learning_rate": 1.144262683784274e-06,
|
||
|
|
"loss": 0.1854,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07982587814331055,
|
||
|
|
"step": 3135,
|
||
|
|
"valid_targets_mean": 2501.4,
|
||
|
|
"valid_targets_min": 429
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.330645161290323,
|
||
|
|
"grad_norm": 0.6695157220720194,
|
||
|
|
"learning_rate": 1.1109738280938531e-06,
|
||
|
|
"loss": 0.2019,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06920865923166275,
|
||
|
|
"step": 3140,
|
||
|
|
"valid_targets_mean": 3162.5,
|
||
|
|
"valid_targets_min": 1200
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.340725806451613,
|
||
|
|
"grad_norm": 0.6722956861303685,
|
||
|
|
"learning_rate": 1.078162530658855e-06,
|
||
|
|
"loss": 0.1958,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12872906029224396,
|
||
|
|
"step": 3145,
|
||
|
|
"valid_targets_mean": 3392.8,
|
||
|
|
"valid_targets_min": 1231
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.350806451612903,
|
||
|
|
"grad_norm": 0.6538730115739195,
|
||
|
|
"learning_rate": 1.0458296210247365e-06,
|
||
|
|
"loss": 0.1949,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06740090250968933,
|
||
|
|
"step": 3150,
|
||
|
|
"valid_targets_mean": 3822.1,
|
||
|
|
"valid_targets_min": 2334
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.360887096774194,
|
||
|
|
"grad_norm": 0.6861699382262711,
|
||
|
|
"learning_rate": 1.0139759166422136e-06,
|
||
|
|
"loss": 0.1728,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06144392490386963,
|
||
|
|
"step": 3155,
|
||
|
|
"valid_targets_mean": 2339.0,
|
||
|
|
"valid_targets_min": 406
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.370967741935484,
|
||
|
|
"grad_norm": 0.6536499792705518,
|
||
|
|
"learning_rate": 9.826022228465892e-07,
|
||
|
|
"loss": 0.1972,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1031823456287384,
|
||
|
|
"step": 3160,
|
||
|
|
"valid_targets_mean": 3342.8,
|
||
|
|
"valid_targets_min": 1480
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.381048387096774,
|
||
|
|
"grad_norm": 0.8548828169242808,
|
||
|
|
"learning_rate": 9.517093328373739e-07,
|
||
|
|
"loss": 0.1875,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09356742352247238,
|
||
|
|
"step": 3165,
|
||
|
|
"valid_targets_mean": 2491.9,
|
||
|
|
"valid_targets_min": 1914
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.391129032258064,
|
||
|
|
"grad_norm": 0.7897924673589775,
|
||
|
|
"learning_rate": 9.212980276582573e-07,
|
||
|
|
"loss": 0.1728,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06692245602607727,
|
||
|
|
"step": 3170,
|
||
|
|
"valid_targets_mean": 2233.4,
|
||
|
|
"valid_targets_min": 1319
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.401209677419355,
|
||
|
|
"grad_norm": 0.6062449681408995,
|
||
|
|
"learning_rate": 8.91369076177353e-07,
|
||
|
|
"loss": 0.1798,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07694501429796219,
|
||
|
|
"step": 3175,
|
||
|
|
"valid_targets_mean": 3899.4,
|
||
|
|
"valid_targets_min": 2356
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.411290322580645,
|
||
|
|
"grad_norm": 0.7121364629989388,
|
||
|
|
"learning_rate": 8.61923235067752e-07,
|
||
|
|
"loss": 0.1763,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10031771659851074,
|
||
|
|
"step": 3180,
|
||
|
|
"valid_targets_mean": 3782.0,
|
||
|
|
"valid_targets_min": 2176
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.421370967741936,
|
||
|
|
"grad_norm": 0.6534159995438924,
|
||
|
|
"learning_rate": 8.329612487883976e-07,
|
||
|
|
"loss": 0.1715,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09105800837278366,
|
||
|
|
"step": 3185,
|
||
|
|
"valid_targets_mean": 4004.2,
|
||
|
|
"valid_targets_min": 1578
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.431451612903226,
|
||
|
|
"grad_norm": 0.6488887135113436,
|
||
|
|
"learning_rate": 8.044838495652718e-07,
|
||
|
|
"loss": 0.1864,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07642491906881332,
|
||
|
|
"step": 3190,
|
||
|
|
"valid_targets_mean": 3439.6,
|
||
|
|
"valid_targets_min": 1446
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.441532258064516,
|
||
|
|
"grad_norm": 0.6816899327248644,
|
||
|
|
"learning_rate": 7.764917573728636e-07,
|
||
|
|
"loss": 0.1725,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08198216557502747,
|
||
|
|
"step": 3195,
|
||
|
|
"valid_targets_mean": 3177.2,
|
||
|
|
"valid_targets_min": 1381
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.451612903225806,
|
||
|
|
"grad_norm": 0.6723282932848176,
|
||
|
|
"learning_rate": 7.4898567991599e-07,
|
||
|
|
"loss": 0.1796,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.092200368642807,
|
||
|
|
"step": 3200,
|
||
|
|
"valid_targets_mean": 3759.6,
|
||
|
|
"valid_targets_min": 1238
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.461693548387097,
|
||
|
|
"grad_norm": 0.591383000978237,
|
||
|
|
"learning_rate": 7.219663126118815e-07,
|
||
|
|
"loss": 0.1995,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.18936467170715332,
|
||
|
|
"step": 3205,
|
||
|
|
"valid_targets_mean": 5882.5,
|
||
|
|
"valid_targets_min": 2528
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.471774193548387,
|
||
|
|
"grad_norm": 0.755771507287537,
|
||
|
|
"learning_rate": 6.954343385726203e-07,
|
||
|
|
"loss": 0.204,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.14651596546173096,
|
||
|
|
"step": 3210,
|
||
|
|
"valid_targets_mean": 3786.1,
|
||
|
|
"valid_targets_min": 1951
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.481854838709677,
|
||
|
|
"grad_norm": 0.6644259178536922,
|
||
|
|
"learning_rate": 6.69390428587855e-07,
|
||
|
|
"loss": 0.1903,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15725857019424438,
|
||
|
|
"step": 3215,
|
||
|
|
"valid_targets_mean": 3576.4,
|
||
|
|
"valid_targets_min": 1020
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.491935483870968,
|
||
|
|
"grad_norm": 0.6958664096147635,
|
||
|
|
"learning_rate": 6.438352411078486e-07,
|
||
|
|
"loss": 0.1668,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07040032744407654,
|
||
|
|
"step": 3220,
|
||
|
|
"valid_targets_mean": 2695.2,
|
||
|
|
"valid_targets_min": 1606
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.502016129032258,
|
||
|
|
"grad_norm": 0.6629341946814559,
|
||
|
|
"learning_rate": 6.187694222268215e-07,
|
||
|
|
"loss": 0.1984,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08556666970252991,
|
||
|
|
"step": 3225,
|
||
|
|
"valid_targets_mean": 3309.2,
|
||
|
|
"valid_targets_min": 1304
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.512096774193548,
|
||
|
|
"grad_norm": 0.6639972224014893,
|
||
|
|
"learning_rate": 5.941936056666331e-07,
|
||
|
|
"loss": 0.1701,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10307017713785172,
|
||
|
|
"step": 3230,
|
||
|
|
"valid_targets_mean": 3758.2,
|
||
|
|
"valid_targets_min": 1855
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.522177419354839,
|
||
|
|
"grad_norm": 0.6656899978049949,
|
||
|
|
"learning_rate": 5.701084127607481e-07,
|
||
|
|
"loss": 0.1744,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.062206245958805084,
|
||
|
|
"step": 3235,
|
||
|
|
"valid_targets_mean": 2866.9,
|
||
|
|
"valid_targets_min": 1716
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.532258064516129,
|
||
|
|
"grad_norm": 0.7315080617869197,
|
||
|
|
"learning_rate": 5.465144524385357e-07,
|
||
|
|
"loss": 0.176,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10558101534843445,
|
||
|
|
"step": 3240,
|
||
|
|
"valid_targets_mean": 2902.4,
|
||
|
|
"valid_targets_min": 1973
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.542338709677419,
|
||
|
|
"grad_norm": 0.6874899377596543,
|
||
|
|
"learning_rate": 5.234123212098574e-07,
|
||
|
|
"loss": 0.1963,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07975097000598907,
|
||
|
|
"step": 3245,
|
||
|
|
"valid_targets_mean": 2915.9,
|
||
|
|
"valid_targets_min": 1833
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.55241935483871,
|
||
|
|
"grad_norm": 0.6954834306853704,
|
||
|
|
"learning_rate": 5.008026031500101e-07,
|
||
|
|
"loss": 0.1728,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09110526740550995,
|
||
|
|
"step": 3250,
|
||
|
|
"valid_targets_mean": 2873.6,
|
||
|
|
"valid_targets_min": 1333
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.5625,
|
||
|
|
"grad_norm": 0.699965726307967,
|
||
|
|
"learning_rate": 4.78685869884945e-07,
|
||
|
|
"loss": 0.1843,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0755114033818245,
|
||
|
|
"step": 3255,
|
||
|
|
"valid_targets_mean": 2595.4,
|
||
|
|
"valid_targets_min": 1113
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.57258064516129,
|
||
|
|
"grad_norm": 0.7364476604020108,
|
||
|
|
"learning_rate": 4.570626805768119e-07,
|
||
|
|
"loss": 0.1757,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13106697797775269,
|
||
|
|
"step": 3260,
|
||
|
|
"valid_targets_mean": 3867.9,
|
||
|
|
"valid_targets_min": 1275
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.582661290322581,
|
||
|
|
"grad_norm": 0.6461586754303715,
|
||
|
|
"learning_rate": 4.3593358190983094e-07,
|
||
|
|
"loss": 0.1803,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08801767230033875,
|
||
|
|
"step": 3265,
|
||
|
|
"valid_targets_mean": 3993.0,
|
||
|
|
"valid_targets_min": 1855
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.592741935483871,
|
||
|
|
"grad_norm": 0.6691988271954056,
|
||
|
|
"learning_rate": 4.152991080764723e-07,
|
||
|
|
"loss": 0.1773,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07823152840137482,
|
||
|
|
"step": 3270,
|
||
|
|
"valid_targets_mean": 3073.9,
|
||
|
|
"valid_targets_min": 1431
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.602822580645161,
|
||
|
|
"grad_norm": 0.6579625777069711,
|
||
|
|
"learning_rate": 3.9515978076394293e-07,
|
||
|
|
"loss": 0.1595,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09333357959985733,
|
||
|
|
"step": 3275,
|
||
|
|
"valid_targets_mean": 3750.2,
|
||
|
|
"valid_targets_min": 1408
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.612903225806452,
|
||
|
|
"grad_norm": 0.6482997942194308,
|
||
|
|
"learning_rate": 3.7551610914100535e-07,
|
||
|
|
"loss": 0.1789,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11413640528917313,
|
||
|
|
"step": 3280,
|
||
|
|
"valid_targets_mean": 3580.5,
|
||
|
|
"valid_targets_min": 1266
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.622983870967742,
|
||
|
|
"grad_norm": 0.7274936480035685,
|
||
|
|
"learning_rate": 3.5636858984509083e-07,
|
||
|
|
"loss": 0.1529,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06913761794567108,
|
||
|
|
"step": 3285,
|
||
|
|
"valid_targets_mean": 2631.4,
|
||
|
|
"valid_targets_min": 304
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.633064516129032,
|
||
|
|
"grad_norm": 0.6539620573710183,
|
||
|
|
"learning_rate": 3.377177069697624e-07,
|
||
|
|
"loss": 0.19,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07360495626926422,
|
||
|
|
"step": 3290,
|
||
|
|
"valid_targets_mean": 2682.9,
|
||
|
|
"valid_targets_min": 1687
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.643145161290323,
|
||
|
|
"grad_norm": 0.7033684529150658,
|
||
|
|
"learning_rate": 3.1956393205246237e-07,
|
||
|
|
"loss": 0.1865,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.1134590357542038,
|
||
|
|
"step": 3295,
|
||
|
|
"valid_targets_mean": 3204.1,
|
||
|
|
"valid_targets_min": 430
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.653225806451613,
|
||
|
|
"grad_norm": 0.7122861894560839,
|
||
|
|
"learning_rate": 3.0190772406259117e-07,
|
||
|
|
"loss": 0.1893,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09804700314998627,
|
||
|
|
"step": 3300,
|
||
|
|
"valid_targets_mean": 3464.8,
|
||
|
|
"valid_targets_min": 1264
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.663306451612903,
|
||
|
|
"grad_norm": 0.7572499688342373,
|
||
|
|
"learning_rate": 2.8474952938991387e-07,
|
||
|
|
"loss": 0.2059,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.13579998910427094,
|
||
|
|
"step": 3305,
|
||
|
|
"valid_targets_mean": 3333.2,
|
||
|
|
"valid_targets_min": 1703
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.673387096774194,
|
||
|
|
"grad_norm": 1.2169274066544102,
|
||
|
|
"learning_rate": 2.6808978183326285e-07,
|
||
|
|
"loss": 0.1849,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10292088240385056,
|
||
|
|
"step": 3310,
|
||
|
|
"valid_targets_mean": 3267.2,
|
||
|
|
"valid_targets_min": 1468
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.683467741935484,
|
||
|
|
"grad_norm": 0.7433168347672474,
|
||
|
|
"learning_rate": 2.519289025895755e-07,
|
||
|
|
"loss": 0.1725,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0709567666053772,
|
||
|
|
"step": 3315,
|
||
|
|
"valid_targets_mean": 2685.6,
|
||
|
|
"valid_targets_min": 327
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.693548387096774,
|
||
|
|
"grad_norm": 0.6503726337251372,
|
||
|
|
"learning_rate": 2.3626730024324474e-07,
|
||
|
|
"loss": 0.1788,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09670200198888779,
|
||
|
|
"step": 3320,
|
||
|
|
"valid_targets_mean": 4002.8,
|
||
|
|
"valid_targets_min": 2399
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.703629032258064,
|
||
|
|
"grad_norm": 0.7051357241828445,
|
||
|
|
"learning_rate": 2.2110537075578975e-07,
|
||
|
|
"loss": 0.1883,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.05833191052079201,
|
||
|
|
"step": 3325,
|
||
|
|
"valid_targets_mean": 2070.0,
|
||
|
|
"valid_targets_min": 1183
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.713709677419355,
|
||
|
|
"grad_norm": 0.696574957316745,
|
||
|
|
"learning_rate": 2.064434974558438e-07,
|
||
|
|
"loss": 0.179,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09011270850896835,
|
||
|
|
"step": 3330,
|
||
|
|
"valid_targets_mean": 3172.8,
|
||
|
|
"valid_targets_min": 396
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.723790322580645,
|
||
|
|
"grad_norm": 0.6669526871805014,
|
||
|
|
"learning_rate": 1.922820510294665e-07,
|
||
|
|
"loss": 0.163,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06726178526878357,
|
||
|
|
"step": 3335,
|
||
|
|
"valid_targets_mean": 2805.9,
|
||
|
|
"valid_targets_min": 1369
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.733870967741936,
|
||
|
|
"grad_norm": 0.6110385153107424,
|
||
|
|
"learning_rate": 1.7862138951076246e-07,
|
||
|
|
"loss": 0.2075,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.06413952261209488,
|
||
|
|
"step": 3340,
|
||
|
|
"valid_targets_mean": 3378.9,
|
||
|
|
"valid_targets_min": 2100
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.743951612903226,
|
||
|
|
"grad_norm": 0.6886743175610556,
|
||
|
|
"learning_rate": 1.6546185827284177e-07,
|
||
|
|
"loss": 0.182,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10534577071666718,
|
||
|
|
"step": 3345,
|
||
|
|
"valid_targets_mean": 3057.9,
|
||
|
|
"valid_targets_min": 1269
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.754032258064516,
|
||
|
|
"grad_norm": 0.7193767301226199,
|
||
|
|
"learning_rate": 1.528037900190804e-07,
|
||
|
|
"loss": 0.1713,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0954342931509018,
|
||
|
|
"step": 3350,
|
||
|
|
"valid_targets_mean": 3312.2,
|
||
|
|
"valid_targets_min": 1640
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.764112903225806,
|
||
|
|
"grad_norm": 0.6515301937120946,
|
||
|
|
"learning_rate": 1.4064750477471135e-07,
|
||
|
|
"loss": 0.1963,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09180790930986404,
|
||
|
|
"step": 3355,
|
||
|
|
"valid_targets_mean": 3203.1,
|
||
|
|
"valid_targets_min": 1877
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.774193548387097,
|
||
|
|
"grad_norm": 0.7391398141571515,
|
||
|
|
"learning_rate": 1.2899330987873327e-07,
|
||
|
|
"loss": 0.1903,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07016012072563171,
|
||
|
|
"step": 3360,
|
||
|
|
"valid_targets_mean": 2193.1,
|
||
|
|
"valid_targets_min": 951
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.784274193548387,
|
||
|
|
"grad_norm": 0.6673309441788807,
|
||
|
|
"learning_rate": 1.1784149997613681e-07,
|
||
|
|
"loss": 0.1724,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07068391144275665,
|
||
|
|
"step": 3365,
|
||
|
|
"valid_targets_mean": 3005.1,
|
||
|
|
"valid_targets_min": 343
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.794354838709677,
|
||
|
|
"grad_norm": 0.6853638251678503,
|
||
|
|
"learning_rate": 1.0719235701046382e-07,
|
||
|
|
"loss": 0.1725,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10330955684185028,
|
||
|
|
"step": 3370,
|
||
|
|
"valid_targets_mean": 4084.8,
|
||
|
|
"valid_targets_min": 1135
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.804435483870968,
|
||
|
|
"grad_norm": 0.6396620471208779,
|
||
|
|
"learning_rate": 9.704615021666863e-08,
|
||
|
|
"loss": 0.165,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08056247979402542,
|
||
|
|
"step": 3375,
|
||
|
|
"valid_targets_mean": 3227.0,
|
||
|
|
"valid_targets_min": 1384
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.814516129032258,
|
||
|
|
"grad_norm": 0.6170649707435096,
|
||
|
|
"learning_rate": 8.740313611431683e-08,
|
||
|
|
"loss": 0.1819,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07109205424785614,
|
||
|
|
"step": 3380,
|
||
|
|
"valid_targets_mean": 3095.1,
|
||
|
|
"valid_targets_min": 1477
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.824596774193548,
|
||
|
|
"grad_norm": 0.6949983156075082,
|
||
|
|
"learning_rate": 7.826355850110378e-08,
|
||
|
|
"loss": 0.1771,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.04709267243742943,
|
||
|
|
"step": 3385,
|
||
|
|
"valid_targets_mean": 1935.5,
|
||
|
|
"valid_targets_min": 440
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.834677419354839,
|
||
|
|
"grad_norm": 0.6235945907969527,
|
||
|
|
"learning_rate": 6.962764844668179e-08,
|
||
|
|
"loss": 0.1894,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09936011582612991,
|
||
|
|
"step": 3390,
|
||
|
|
"valid_targets_mean": 3402.4,
|
||
|
|
"valid_targets_min": 2107
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.844758064516129,
|
||
|
|
"grad_norm": 0.6788876550084832,
|
||
|
|
"learning_rate": 6.149562428682477e-08,
|
||
|
|
"loss": 0.1962,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08755327761173248,
|
||
|
|
"step": 3395,
|
||
|
|
"valid_targets_mean": 2905.6,
|
||
|
|
"valid_targets_min": 1119
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.854838709677419,
|
||
|
|
"grad_norm": 0.6630078855332802,
|
||
|
|
"learning_rate": 5.386769161790595e-08,
|
||
|
|
"loss": 0.1768,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07167403399944305,
|
||
|
|
"step": 3400,
|
||
|
|
"valid_targets_mean": 2900.6,
|
||
|
|
"valid_targets_min": 1577
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.86491935483871,
|
||
|
|
"grad_norm": 0.6554140227481854,
|
||
|
|
"learning_rate": 4.674404329169768e-08,
|
||
|
|
"loss": 0.184,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.11994685977697372,
|
||
|
|
"step": 3405,
|
||
|
|
"valid_targets_mean": 4328.9,
|
||
|
|
"valid_targets_min": 365
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.875,
|
||
|
|
"grad_norm": 0.606228374900086,
|
||
|
|
"learning_rate": 4.0124859410504143e-08,
|
||
|
|
"loss": 0.1908,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.15875974297523499,
|
||
|
|
"step": 3410,
|
||
|
|
"valid_targets_mean": 5088.1,
|
||
|
|
"valid_targets_min": 2084
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.88508064516129,
|
||
|
|
"grad_norm": 0.6759801183735287,
|
||
|
|
"learning_rate": 3.401030732259614e-08,
|
||
|
|
"loss": 0.1785,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.0744219422340393,
|
||
|
|
"step": 3415,
|
||
|
|
"valid_targets_mean": 2767.9,
|
||
|
|
"valid_targets_min": 440
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.895161290322581,
|
||
|
|
"grad_norm": 0.6794316224051092,
|
||
|
|
"learning_rate": 2.8400541617987777e-08,
|
||
|
|
"loss": 0.1892,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09825184941291809,
|
||
|
|
"step": 3420,
|
||
|
|
"valid_targets_mean": 3763.2,
|
||
|
|
"valid_targets_min": 1245
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.905241935483871,
|
||
|
|
"grad_norm": 0.7427201942826057,
|
||
|
|
"learning_rate": 2.3295704124528528e-08,
|
||
|
|
"loss": 0.1934,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07533140480518341,
|
||
|
|
"step": 3425,
|
||
|
|
"valid_targets_mean": 3116.1,
|
||
|
|
"valid_targets_min": 1506
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.915322580645161,
|
||
|
|
"grad_norm": 0.7046620362466542,
|
||
|
|
"learning_rate": 1.869592390431052e-08,
|
||
|
|
"loss": 0.1656,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.10719075798988342,
|
||
|
|
"step": 3430,
|
||
|
|
"valid_targets_mean": 2774.6,
|
||
|
|
"valid_targets_min": 305
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.925403225806452,
|
||
|
|
"grad_norm": 0.6507131682418029,
|
||
|
|
"learning_rate": 1.4601317250413361e-08,
|
||
|
|
"loss": 0.1717,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09277139604091644,
|
||
|
|
"step": 3435,
|
||
|
|
"valid_targets_mean": 4052.5,
|
||
|
|
"valid_targets_min": 2361
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.935483870967742,
|
||
|
|
"grad_norm": 0.7038132060457154,
|
||
|
|
"learning_rate": 1.1011987683964276e-08,
|
||
|
|
"loss": 0.1944,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09100843966007233,
|
||
|
|
"step": 3440,
|
||
|
|
"valid_targets_mean": 3273.1,
|
||
|
|
"valid_targets_min": 2018
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.945564516129032,
|
||
|
|
"grad_norm": 0.7544266586646858,
|
||
|
|
"learning_rate": 7.928025951509099e-09,
|
||
|
|
"loss": 0.1907,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.09497767686843872,
|
||
|
|
"step": 3445,
|
||
|
|
"valid_targets_mean": 2332.1,
|
||
|
|
"valid_targets_min": 1370
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.955645161290323,
|
||
|
|
"grad_norm": 0.7129815383500983,
|
||
|
|
"learning_rate": 5.349510022731874e-09,
|
||
|
|
"loss": 0.2087,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.16425669193267822,
|
||
|
|
"step": 3450,
|
||
|
|
"valid_targets_mean": 3887.5,
|
||
|
|
"valid_targets_min": 400
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.965725806451613,
|
||
|
|
"grad_norm": 0.6510703776898255,
|
||
|
|
"learning_rate": 3.276505088478654e-09,
|
||
|
|
"loss": 0.1883,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07020069658756256,
|
||
|
|
"step": 3455,
|
||
|
|
"valid_targets_mean": 3275.1,
|
||
|
|
"valid_targets_min": 1670
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.975806451612903,
|
||
|
|
"grad_norm": 0.630747188451262,
|
||
|
|
"learning_rate": 1.7090635591077153e-09,
|
||
|
|
"loss": 0.1936,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.08311726152896881,
|
||
|
|
"step": 3460,
|
||
|
|
"valid_targets_mean": 3087.1,
|
||
|
|
"valid_targets_min": 1427
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.985887096774194,
|
||
|
|
"grad_norm": 0.625051968711358,
|
||
|
|
"learning_rate": 6.472250631617271e-10,
|
||
|
|
"loss": 0.1887,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07689955830574036,
|
||
|
|
"step": 3465,
|
||
|
|
"valid_targets_mean": 3811.5,
|
||
|
|
"valid_targets_min": 1573
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.995967741935484,
|
||
|
|
"grad_norm": 0.7922178639282371,
|
||
|
|
"learning_rate": 9.101644637743435e-11,
|
||
|
|
"loss": 0.1887,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.07924503087997437,
|
||
|
|
"step": 3470,
|
||
|
|
"valid_targets_mean": 2611.0,
|
||
|
|
"valid_targets_min": 1286
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 7.0,
|
||
|
|
"loss_nan_ranks": 0,
|
||
|
|
"loss_rank_avg": 0.12738347053527832,
|
||
|
|
"step": 3472,
|
||
|
|
"total_flos": 6.410364364720701e+17,
|
||
|
|
"train_loss": 0.25545546521479906,
|
||
|
|
"train_runtime": 14855.1436,
|
||
|
|
"train_samples_per_second": 3.736,
|
||
|
|
"train_steps_per_second": 0.234,
|
||
|
|
"valid_targets_mean": 3825.4,
|
||
|
|
"valid_targets_min": 1564
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 5,
|
||
|
|
"max_steps": 3472,
|
||
|
|
"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": 6.410364364720701e+17,
|
||
|
|
"train_batch_size": 1,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|