9695 lines
269 KiB
JSON
9695 lines
269 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 7.0,
|
|
"eval_steps": 500,
|
|
"global_step": 4389,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.007974481658692184,
|
|
"grad_norm": 18.18803099603151,
|
|
"learning_rate": 3.644646924829157e-07,
|
|
"loss": 0.7198,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.7252004146575928,
|
|
"step": 5,
|
|
"valid_targets_mean": 3434.4,
|
|
"valid_targets_min": 864
|
|
},
|
|
{
|
|
"epoch": 0.01594896331738437,
|
|
"grad_norm": 14.307158106996866,
|
|
"learning_rate": 8.200455580865605e-07,
|
|
"loss": 0.67,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.6224847435951233,
|
|
"step": 10,
|
|
"valid_targets_mean": 4096.6,
|
|
"valid_targets_min": 510
|
|
},
|
|
{
|
|
"epoch": 0.023923444976076555,
|
|
"grad_norm": 14.459528130720114,
|
|
"learning_rate": 1.2756264236902052e-06,
|
|
"loss": 0.6655,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.6342157125473022,
|
|
"step": 15,
|
|
"valid_targets_mean": 5213.8,
|
|
"valid_targets_min": 1714
|
|
},
|
|
{
|
|
"epoch": 0.03189792663476874,
|
|
"grad_norm": 13.998960975547485,
|
|
"learning_rate": 1.7312072892938498e-06,
|
|
"loss": 0.6358,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.6497429013252258,
|
|
"step": 20,
|
|
"valid_targets_mean": 3823.2,
|
|
"valid_targets_min": 1156
|
|
},
|
|
{
|
|
"epoch": 0.03987240829346093,
|
|
"grad_norm": 7.703819741872236,
|
|
"learning_rate": 2.1867881548974945e-06,
|
|
"loss": 0.5962,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.5347366333007812,
|
|
"step": 25,
|
|
"valid_targets_mean": 4505.9,
|
|
"valid_targets_min": 1707
|
|
},
|
|
{
|
|
"epoch": 0.04784688995215311,
|
|
"grad_norm": 4.528783755997337,
|
|
"learning_rate": 2.642369020501139e-06,
|
|
"loss": 0.5417,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.48101523518562317,
|
|
"step": 30,
|
|
"valid_targets_mean": 4406.1,
|
|
"valid_targets_min": 1431
|
|
},
|
|
{
|
|
"epoch": 0.05582137161084529,
|
|
"grad_norm": 2.971933510909393,
|
|
"learning_rate": 3.0979498861047843e-06,
|
|
"loss": 0.4918,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4876139760017395,
|
|
"step": 35,
|
|
"valid_targets_mean": 3874.3,
|
|
"valid_targets_min": 1363
|
|
},
|
|
{
|
|
"epoch": 0.06379585326953748,
|
|
"grad_norm": 1.836628593469571,
|
|
"learning_rate": 3.5535307517084285e-06,
|
|
"loss": 0.4547,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.47373178601264954,
|
|
"step": 40,
|
|
"valid_targets_mean": 3826.8,
|
|
"valid_targets_min": 1465
|
|
},
|
|
{
|
|
"epoch": 0.07177033492822966,
|
|
"grad_norm": 1.2423251085316844,
|
|
"learning_rate": 4.009111617312073e-06,
|
|
"loss": 0.4167,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4141938090324402,
|
|
"step": 45,
|
|
"valid_targets_mean": 4673.0,
|
|
"valid_targets_min": 1811
|
|
},
|
|
{
|
|
"epoch": 0.07974481658692185,
|
|
"grad_norm": 1.1976629955780103,
|
|
"learning_rate": 4.464692482915718e-06,
|
|
"loss": 0.4112,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.4462246000766754,
|
|
"step": 50,
|
|
"valid_targets_mean": 3402.2,
|
|
"valid_targets_min": 1417
|
|
},
|
|
{
|
|
"epoch": 0.08771929824561403,
|
|
"grad_norm": 0.9374607824761864,
|
|
"learning_rate": 4.920273348519363e-06,
|
|
"loss": 0.4168,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.39574891328811646,
|
|
"step": 55,
|
|
"valid_targets_mean": 3730.4,
|
|
"valid_targets_min": 1185
|
|
},
|
|
{
|
|
"epoch": 0.09569377990430622,
|
|
"grad_norm": 0.8345026125790815,
|
|
"learning_rate": 5.375854214123008e-06,
|
|
"loss": 0.4056,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3709059953689575,
|
|
"step": 60,
|
|
"valid_targets_mean": 4410.6,
|
|
"valid_targets_min": 1877
|
|
},
|
|
{
|
|
"epoch": 0.10366826156299841,
|
|
"grad_norm": 0.7251088865875112,
|
|
"learning_rate": 5.831435079726651e-06,
|
|
"loss": 0.3957,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.414184033870697,
|
|
"step": 65,
|
|
"valid_targets_mean": 4409.2,
|
|
"valid_targets_min": 1579
|
|
},
|
|
{
|
|
"epoch": 0.11164274322169059,
|
|
"grad_norm": 0.6361706459112471,
|
|
"learning_rate": 6.287015945330297e-06,
|
|
"loss": 0.3825,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3978373408317566,
|
|
"step": 70,
|
|
"valid_targets_mean": 4352.1,
|
|
"valid_targets_min": 629
|
|
},
|
|
{
|
|
"epoch": 0.11961722488038277,
|
|
"grad_norm": 0.8778851407966174,
|
|
"learning_rate": 6.742596810933942e-06,
|
|
"loss": 0.3551,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3310779929161072,
|
|
"step": 75,
|
|
"valid_targets_mean": 3677.3,
|
|
"valid_targets_min": 1683
|
|
},
|
|
{
|
|
"epoch": 0.12759170653907495,
|
|
"grad_norm": 0.5499827496003175,
|
|
"learning_rate": 7.1981776765375854e-06,
|
|
"loss": 0.3597,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.34589529037475586,
|
|
"step": 80,
|
|
"valid_targets_mean": 4806.0,
|
|
"valid_targets_min": 1519
|
|
},
|
|
{
|
|
"epoch": 0.13556618819776714,
|
|
"grad_norm": 0.6376269307263297,
|
|
"learning_rate": 7.65375854214123e-06,
|
|
"loss": 0.3658,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3655011057853699,
|
|
"step": 85,
|
|
"valid_targets_mean": 3583.9,
|
|
"valid_targets_min": 2175
|
|
},
|
|
{
|
|
"epoch": 0.14354066985645933,
|
|
"grad_norm": 0.5742252753369051,
|
|
"learning_rate": 8.109339407744875e-06,
|
|
"loss": 0.343,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.385811448097229,
|
|
"step": 90,
|
|
"valid_targets_mean": 4250.4,
|
|
"valid_targets_min": 1863
|
|
},
|
|
{
|
|
"epoch": 0.15151515151515152,
|
|
"grad_norm": 0.5753097344872611,
|
|
"learning_rate": 8.564920273348521e-06,
|
|
"loss": 0.3558,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3747674226760864,
|
|
"step": 95,
|
|
"valid_targets_mean": 4264.9,
|
|
"valid_targets_min": 1876
|
|
},
|
|
{
|
|
"epoch": 0.1594896331738437,
|
|
"grad_norm": 0.5354986309600116,
|
|
"learning_rate": 9.020501138952164e-06,
|
|
"loss": 0.3392,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29349711537361145,
|
|
"step": 100,
|
|
"valid_targets_mean": 4301.4,
|
|
"valid_targets_min": 1022
|
|
},
|
|
{
|
|
"epoch": 0.1674641148325359,
|
|
"grad_norm": 0.5042094725304733,
|
|
"learning_rate": 9.47608200455581e-06,
|
|
"loss": 0.3269,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3074486553668976,
|
|
"step": 105,
|
|
"valid_targets_mean": 4285.6,
|
|
"valid_targets_min": 949
|
|
},
|
|
{
|
|
"epoch": 0.17543859649122806,
|
|
"grad_norm": 0.5342534363482105,
|
|
"learning_rate": 9.931662870159453e-06,
|
|
"loss": 0.3345,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.33433669805526733,
|
|
"step": 110,
|
|
"valid_targets_mean": 4667.8,
|
|
"valid_targets_min": 1007
|
|
},
|
|
{
|
|
"epoch": 0.18341307814992025,
|
|
"grad_norm": 0.5664117732647538,
|
|
"learning_rate": 1.03872437357631e-05,
|
|
"loss": 0.3034,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3258150815963745,
|
|
"step": 115,
|
|
"valid_targets_mean": 4168.5,
|
|
"valid_targets_min": 2399
|
|
},
|
|
{
|
|
"epoch": 0.19138755980861244,
|
|
"grad_norm": 0.6195050405082234,
|
|
"learning_rate": 1.0842824601366744e-05,
|
|
"loss": 0.3134,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.317843496799469,
|
|
"step": 120,
|
|
"valid_targets_mean": 4331.0,
|
|
"valid_targets_min": 311
|
|
},
|
|
{
|
|
"epoch": 0.19936204146730463,
|
|
"grad_norm": 0.7106286580810014,
|
|
"learning_rate": 1.1298405466970387e-05,
|
|
"loss": 0.3052,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.345772385597229,
|
|
"step": 125,
|
|
"valid_targets_mean": 3779.2,
|
|
"valid_targets_min": 1227
|
|
},
|
|
{
|
|
"epoch": 0.20733652312599682,
|
|
"grad_norm": 0.5142652660619057,
|
|
"learning_rate": 1.1753986332574032e-05,
|
|
"loss": 0.3078,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2607839107513428,
|
|
"step": 130,
|
|
"valid_targets_mean": 4497.3,
|
|
"valid_targets_min": 809
|
|
},
|
|
{
|
|
"epoch": 0.215311004784689,
|
|
"grad_norm": 0.5952097660485908,
|
|
"learning_rate": 1.2209567198177677e-05,
|
|
"loss": 0.293,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.31705808639526367,
|
|
"step": 135,
|
|
"valid_targets_mean": 4538.1,
|
|
"valid_targets_min": 1387
|
|
},
|
|
{
|
|
"epoch": 0.22328548644338117,
|
|
"grad_norm": 0.5105105591440974,
|
|
"learning_rate": 1.2665148063781323e-05,
|
|
"loss": 0.2989,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2607859969139099,
|
|
"step": 140,
|
|
"valid_targets_mean": 4369.7,
|
|
"valid_targets_min": 1806
|
|
},
|
|
{
|
|
"epoch": 0.23125996810207336,
|
|
"grad_norm": 0.503655334834791,
|
|
"learning_rate": 1.3120728929384968e-05,
|
|
"loss": 0.2945,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26573601365089417,
|
|
"step": 145,
|
|
"valid_targets_mean": 4547.8,
|
|
"valid_targets_min": 1503
|
|
},
|
|
{
|
|
"epoch": 0.23923444976076555,
|
|
"grad_norm": 0.5331226450754283,
|
|
"learning_rate": 1.357630979498861e-05,
|
|
"loss": 0.3136,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29404884576797485,
|
|
"step": 150,
|
|
"valid_targets_mean": 4822.4,
|
|
"valid_targets_min": 465
|
|
},
|
|
{
|
|
"epoch": 0.24720893141945774,
|
|
"grad_norm": 0.5583950119493155,
|
|
"learning_rate": 1.4031890660592255e-05,
|
|
"loss": 0.2926,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28105857968330383,
|
|
"step": 155,
|
|
"valid_targets_mean": 4380.1,
|
|
"valid_targets_min": 1890
|
|
},
|
|
{
|
|
"epoch": 0.2551834130781499,
|
|
"grad_norm": 0.5248261173296397,
|
|
"learning_rate": 1.4487471526195902e-05,
|
|
"loss": 0.3019,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.35021817684173584,
|
|
"step": 160,
|
|
"valid_targets_mean": 5202.3,
|
|
"valid_targets_min": 1672
|
|
},
|
|
{
|
|
"epoch": 0.2631578947368421,
|
|
"grad_norm": 0.4720997343296031,
|
|
"learning_rate": 1.4943052391799546e-05,
|
|
"loss": 0.287,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3036966323852539,
|
|
"step": 165,
|
|
"valid_targets_mean": 5538.6,
|
|
"valid_targets_min": 635
|
|
},
|
|
{
|
|
"epoch": 0.2711323763955343,
|
|
"grad_norm": 0.5595618392183499,
|
|
"learning_rate": 1.539863325740319e-05,
|
|
"loss": 0.2909,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.30375349521636963,
|
|
"step": 170,
|
|
"valid_targets_mean": 4671.9,
|
|
"valid_targets_min": 1260
|
|
},
|
|
{
|
|
"epoch": 0.27910685805422647,
|
|
"grad_norm": 0.5555511001381592,
|
|
"learning_rate": 1.5854214123006836e-05,
|
|
"loss": 0.2989,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.31066784262657166,
|
|
"step": 175,
|
|
"valid_targets_mean": 3623.2,
|
|
"valid_targets_min": 404
|
|
},
|
|
{
|
|
"epoch": 0.28708133971291866,
|
|
"grad_norm": 0.49946036712405906,
|
|
"learning_rate": 1.630979498861048e-05,
|
|
"loss": 0.2777,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25820064544677734,
|
|
"step": 180,
|
|
"valid_targets_mean": 4328.4,
|
|
"valid_targets_min": 2082
|
|
},
|
|
{
|
|
"epoch": 0.29505582137161085,
|
|
"grad_norm": 0.6264471320724355,
|
|
"learning_rate": 1.6765375854214125e-05,
|
|
"loss": 0.3004,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28867748379707336,
|
|
"step": 185,
|
|
"valid_targets_mean": 4590.6,
|
|
"valid_targets_min": 621
|
|
},
|
|
{
|
|
"epoch": 0.30303030303030304,
|
|
"grad_norm": 0.5759630894438817,
|
|
"learning_rate": 1.722095671981777e-05,
|
|
"loss": 0.3039,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.315102219581604,
|
|
"step": 190,
|
|
"valid_targets_mean": 4509.0,
|
|
"valid_targets_min": 1333
|
|
},
|
|
{
|
|
"epoch": 0.31100478468899523,
|
|
"grad_norm": 0.5294255834259166,
|
|
"learning_rate": 1.7676537585421415e-05,
|
|
"loss": 0.2705,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27122050523757935,
|
|
"step": 195,
|
|
"valid_targets_mean": 4544.4,
|
|
"valid_targets_min": 1545
|
|
},
|
|
{
|
|
"epoch": 0.3189792663476874,
|
|
"grad_norm": 0.6154065761188445,
|
|
"learning_rate": 1.813211845102506e-05,
|
|
"loss": 0.2654,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24356196820735931,
|
|
"step": 200,
|
|
"valid_targets_mean": 3598.1,
|
|
"valid_targets_min": 1584
|
|
},
|
|
{
|
|
"epoch": 0.3269537480063796,
|
|
"grad_norm": 0.5224082452829883,
|
|
"learning_rate": 1.8587699316628704e-05,
|
|
"loss": 0.2894,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.284517765045166,
|
|
"step": 205,
|
|
"valid_targets_mean": 5208.4,
|
|
"valid_targets_min": 1290
|
|
},
|
|
{
|
|
"epoch": 0.3349282296650718,
|
|
"grad_norm": 0.5756342094108516,
|
|
"learning_rate": 1.904328018223235e-05,
|
|
"loss": 0.3041,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3382299542427063,
|
|
"step": 210,
|
|
"valid_targets_mean": 4352.9,
|
|
"valid_targets_min": 1068
|
|
},
|
|
{
|
|
"epoch": 0.34290271132376393,
|
|
"grad_norm": 0.5209919174559063,
|
|
"learning_rate": 1.9498861047835993e-05,
|
|
"loss": 0.284,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3093770444393158,
|
|
"step": 215,
|
|
"valid_targets_mean": 4669.0,
|
|
"valid_targets_min": 1859
|
|
},
|
|
{
|
|
"epoch": 0.3508771929824561,
|
|
"grad_norm": 0.5564760822604873,
|
|
"learning_rate": 1.9954441913439638e-05,
|
|
"loss": 0.3043,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.41471871733665466,
|
|
"step": 220,
|
|
"valid_targets_mean": 5771.4,
|
|
"valid_targets_min": 467
|
|
},
|
|
{
|
|
"epoch": 0.3588516746411483,
|
|
"grad_norm": 0.4401205513624318,
|
|
"learning_rate": 2.0410022779043283e-05,
|
|
"loss": 0.2672,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23879970610141754,
|
|
"step": 225,
|
|
"valid_targets_mean": 5340.2,
|
|
"valid_targets_min": 1166
|
|
},
|
|
{
|
|
"epoch": 0.3668261562998405,
|
|
"grad_norm": 0.548529089679332,
|
|
"learning_rate": 2.0865603644646927e-05,
|
|
"loss": 0.255,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2805835008621216,
|
|
"step": 230,
|
|
"valid_targets_mean": 4598.0,
|
|
"valid_targets_min": 1647
|
|
},
|
|
{
|
|
"epoch": 0.3748006379585327,
|
|
"grad_norm": 0.6106048799267036,
|
|
"learning_rate": 2.1321184510250572e-05,
|
|
"loss": 0.2804,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.260797917842865,
|
|
"step": 235,
|
|
"valid_targets_mean": 3621.2,
|
|
"valid_targets_min": 1345
|
|
},
|
|
{
|
|
"epoch": 0.3827751196172249,
|
|
"grad_norm": 0.5381663991323422,
|
|
"learning_rate": 2.1776765375854217e-05,
|
|
"loss": 0.2794,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2917918562889099,
|
|
"step": 240,
|
|
"valid_targets_mean": 4667.2,
|
|
"valid_targets_min": 2104
|
|
},
|
|
{
|
|
"epoch": 0.39074960127591707,
|
|
"grad_norm": 0.7581595258947841,
|
|
"learning_rate": 2.223234624145786e-05,
|
|
"loss": 0.2657,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28909996151924133,
|
|
"step": 245,
|
|
"valid_targets_mean": 4371.6,
|
|
"valid_targets_min": 1739
|
|
},
|
|
{
|
|
"epoch": 0.39872408293460926,
|
|
"grad_norm": 0.5806802069339,
|
|
"learning_rate": 2.2687927107061506e-05,
|
|
"loss": 0.2573,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2631956934928894,
|
|
"step": 250,
|
|
"valid_targets_mean": 4192.1,
|
|
"valid_targets_min": 1587
|
|
},
|
|
{
|
|
"epoch": 0.40669856459330145,
|
|
"grad_norm": 0.6263257130474685,
|
|
"learning_rate": 2.314350797266515e-05,
|
|
"loss": 0.2803,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25036656856536865,
|
|
"step": 255,
|
|
"valid_targets_mean": 4222.9,
|
|
"valid_targets_min": 2196
|
|
},
|
|
{
|
|
"epoch": 0.41467304625199364,
|
|
"grad_norm": 0.5344890255498245,
|
|
"learning_rate": 2.3599088838268792e-05,
|
|
"loss": 0.2723,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26563674211502075,
|
|
"step": 260,
|
|
"valid_targets_mean": 4978.1,
|
|
"valid_targets_min": 1867
|
|
},
|
|
{
|
|
"epoch": 0.4226475279106858,
|
|
"grad_norm": 0.5058620232957759,
|
|
"learning_rate": 2.4054669703872436e-05,
|
|
"loss": 0.2592,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25486892461776733,
|
|
"step": 265,
|
|
"valid_targets_mean": 5157.2,
|
|
"valid_targets_min": 1801
|
|
},
|
|
{
|
|
"epoch": 0.430622009569378,
|
|
"grad_norm": 0.6522226250750656,
|
|
"learning_rate": 2.4510250569476085e-05,
|
|
"loss": 0.2691,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27999359369277954,
|
|
"step": 270,
|
|
"valid_targets_mean": 4255.8,
|
|
"valid_targets_min": 1735
|
|
},
|
|
{
|
|
"epoch": 0.43859649122807015,
|
|
"grad_norm": 0.5746725160756948,
|
|
"learning_rate": 2.496583143507973e-05,
|
|
"loss": 0.2798,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23281586170196533,
|
|
"step": 275,
|
|
"valid_targets_mean": 5547.3,
|
|
"valid_targets_min": 1335
|
|
},
|
|
{
|
|
"epoch": 0.44657097288676234,
|
|
"grad_norm": 0.5241130870299622,
|
|
"learning_rate": 2.5421412300683374e-05,
|
|
"loss": 0.2733,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28230002522468567,
|
|
"step": 280,
|
|
"valid_targets_mean": 4519.5,
|
|
"valid_targets_min": 1043
|
|
},
|
|
{
|
|
"epoch": 0.45454545454545453,
|
|
"grad_norm": 0.566135695536269,
|
|
"learning_rate": 2.587699316628702e-05,
|
|
"loss": 0.2549,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25908809900283813,
|
|
"step": 285,
|
|
"valid_targets_mean": 3886.4,
|
|
"valid_targets_min": 1509
|
|
},
|
|
{
|
|
"epoch": 0.4625199362041467,
|
|
"grad_norm": 0.5438761096776028,
|
|
"learning_rate": 2.6332574031890663e-05,
|
|
"loss": 0.256,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2712545394897461,
|
|
"step": 290,
|
|
"valid_targets_mean": 4487.6,
|
|
"valid_targets_min": 736
|
|
},
|
|
{
|
|
"epoch": 0.4704944178628389,
|
|
"grad_norm": 0.559648132337892,
|
|
"learning_rate": 2.6788154897494308e-05,
|
|
"loss": 0.2619,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24313515424728394,
|
|
"step": 295,
|
|
"valid_targets_mean": 4423.8,
|
|
"valid_targets_min": 686
|
|
},
|
|
{
|
|
"epoch": 0.4784688995215311,
|
|
"grad_norm": 0.5966418677079639,
|
|
"learning_rate": 2.7243735763097953e-05,
|
|
"loss": 0.2602,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23721837997436523,
|
|
"step": 300,
|
|
"valid_targets_mean": 3706.7,
|
|
"valid_targets_min": 1173
|
|
},
|
|
{
|
|
"epoch": 0.4864433811802233,
|
|
"grad_norm": 0.5815566112988382,
|
|
"learning_rate": 2.7699316628701597e-05,
|
|
"loss": 0.2378,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23788884282112122,
|
|
"step": 305,
|
|
"valid_targets_mean": 3594.3,
|
|
"valid_targets_min": 1815
|
|
},
|
|
{
|
|
"epoch": 0.4944178628389155,
|
|
"grad_norm": 0.5045506008502036,
|
|
"learning_rate": 2.815489749430524e-05,
|
|
"loss": 0.2546,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25513797998428345,
|
|
"step": 310,
|
|
"valid_targets_mean": 5347.7,
|
|
"valid_targets_min": 1221
|
|
},
|
|
{
|
|
"epoch": 0.5023923444976076,
|
|
"grad_norm": 0.5701555342264036,
|
|
"learning_rate": 2.8610478359908883e-05,
|
|
"loss": 0.2425,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24246780574321747,
|
|
"step": 315,
|
|
"valid_targets_mean": 3548.3,
|
|
"valid_targets_min": 1136
|
|
},
|
|
{
|
|
"epoch": 0.5103668261562998,
|
|
"grad_norm": 0.5182367732301608,
|
|
"learning_rate": 2.906605922551253e-05,
|
|
"loss": 0.2587,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23432882130146027,
|
|
"step": 320,
|
|
"valid_targets_mean": 4655.8,
|
|
"valid_targets_min": 1400
|
|
},
|
|
{
|
|
"epoch": 0.518341307814992,
|
|
"grad_norm": 0.581611039848372,
|
|
"learning_rate": 2.9521640091116176e-05,
|
|
"loss": 0.2491,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24655424058437347,
|
|
"step": 325,
|
|
"valid_targets_mean": 4075.2,
|
|
"valid_targets_min": 1087
|
|
},
|
|
{
|
|
"epoch": 0.5263157894736842,
|
|
"grad_norm": 0.5119496690813441,
|
|
"learning_rate": 2.997722095671982e-05,
|
|
"loss": 0.2542,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28360801935195923,
|
|
"step": 330,
|
|
"valid_targets_mean": 4828.7,
|
|
"valid_targets_min": 2185
|
|
},
|
|
{
|
|
"epoch": 0.5342902711323764,
|
|
"grad_norm": 0.4842764810320044,
|
|
"learning_rate": 3.0432801822323465e-05,
|
|
"loss": 0.2553,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.3004698157310486,
|
|
"step": 335,
|
|
"valid_targets_mean": 5661.6,
|
|
"valid_targets_min": 1938
|
|
},
|
|
{
|
|
"epoch": 0.5422647527910686,
|
|
"grad_norm": 0.5349587768564628,
|
|
"learning_rate": 3.088838268792711e-05,
|
|
"loss": 0.2468,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2438666820526123,
|
|
"step": 340,
|
|
"valid_targets_mean": 4150.4,
|
|
"valid_targets_min": 1564
|
|
},
|
|
{
|
|
"epoch": 0.5502392344497608,
|
|
"grad_norm": 0.5402803413388506,
|
|
"learning_rate": 3.1343963553530755e-05,
|
|
"loss": 0.2517,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25511205196380615,
|
|
"step": 345,
|
|
"valid_targets_mean": 4414.5,
|
|
"valid_targets_min": 2053
|
|
},
|
|
{
|
|
"epoch": 0.5582137161084529,
|
|
"grad_norm": 0.7357121532770818,
|
|
"learning_rate": 3.17995444191344e-05,
|
|
"loss": 0.2943,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.30022668838500977,
|
|
"step": 350,
|
|
"valid_targets_mean": 3161.2,
|
|
"valid_targets_min": 655
|
|
},
|
|
{
|
|
"epoch": 0.5661881977671451,
|
|
"grad_norm": 0.6635034938265247,
|
|
"learning_rate": 3.2255125284738044e-05,
|
|
"loss": 0.2474,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27807778120040894,
|
|
"step": 355,
|
|
"valid_targets_mean": 3865.3,
|
|
"valid_targets_min": 2029
|
|
},
|
|
{
|
|
"epoch": 0.5741626794258373,
|
|
"grad_norm": 0.5759574185615652,
|
|
"learning_rate": 3.271070615034169e-05,
|
|
"loss": 0.2462,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2875663936138153,
|
|
"step": 360,
|
|
"valid_targets_mean": 4687.4,
|
|
"valid_targets_min": 1812
|
|
},
|
|
{
|
|
"epoch": 0.5821371610845295,
|
|
"grad_norm": 0.5603633405311572,
|
|
"learning_rate": 3.316628701594533e-05,
|
|
"loss": 0.2512,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25267183780670166,
|
|
"step": 365,
|
|
"valid_targets_mean": 4408.4,
|
|
"valid_targets_min": 1131
|
|
},
|
|
{
|
|
"epoch": 0.5901116427432217,
|
|
"grad_norm": 0.6869663265290378,
|
|
"learning_rate": 3.362186788154898e-05,
|
|
"loss": 0.2635,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2470800280570984,
|
|
"step": 370,
|
|
"valid_targets_mean": 3387.9,
|
|
"valid_targets_min": 1615
|
|
},
|
|
{
|
|
"epoch": 0.5980861244019139,
|
|
"grad_norm": 0.578708204023255,
|
|
"learning_rate": 3.407744874715262e-05,
|
|
"loss": 0.2602,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25405412912368774,
|
|
"step": 375,
|
|
"valid_targets_mean": 3955.7,
|
|
"valid_targets_min": 1637
|
|
},
|
|
{
|
|
"epoch": 0.6060606060606061,
|
|
"grad_norm": 0.5959034995071885,
|
|
"learning_rate": 3.453302961275627e-05,
|
|
"loss": 0.2278,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2276259809732437,
|
|
"step": 380,
|
|
"valid_targets_mean": 3714.9,
|
|
"valid_targets_min": 1249
|
|
},
|
|
{
|
|
"epoch": 0.6140350877192983,
|
|
"grad_norm": 0.7236145579514683,
|
|
"learning_rate": 3.498861047835991e-05,
|
|
"loss": 0.2493,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23672303557395935,
|
|
"step": 385,
|
|
"valid_targets_mean": 3127.5,
|
|
"valid_targets_min": 440
|
|
},
|
|
{
|
|
"epoch": 0.6220095693779905,
|
|
"grad_norm": 0.4956883624100392,
|
|
"learning_rate": 3.5444191343963557e-05,
|
|
"loss": 0.2586,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2489890456199646,
|
|
"step": 390,
|
|
"valid_targets_mean": 4754.2,
|
|
"valid_targets_min": 1637
|
|
},
|
|
{
|
|
"epoch": 0.6299840510366826,
|
|
"grad_norm": 0.8505599140008155,
|
|
"learning_rate": 3.58997722095672e-05,
|
|
"loss": 0.2573,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2957842946052551,
|
|
"step": 395,
|
|
"valid_targets_mean": 4341.0,
|
|
"valid_targets_min": 975
|
|
},
|
|
{
|
|
"epoch": 0.6379585326953748,
|
|
"grad_norm": 0.4755589673221355,
|
|
"learning_rate": 3.6355353075170846e-05,
|
|
"loss": 0.2488,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2880033254623413,
|
|
"step": 400,
|
|
"valid_targets_mean": 5651.7,
|
|
"valid_targets_min": 2397
|
|
},
|
|
{
|
|
"epoch": 0.645933014354067,
|
|
"grad_norm": 0.5917983081159787,
|
|
"learning_rate": 3.681093394077449e-05,
|
|
"loss": 0.2422,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2151985466480255,
|
|
"step": 405,
|
|
"valid_targets_mean": 3261.0,
|
|
"valid_targets_min": 961
|
|
},
|
|
{
|
|
"epoch": 0.6539074960127592,
|
|
"grad_norm": 0.5368476047325583,
|
|
"learning_rate": 3.7266514806378135e-05,
|
|
"loss": 0.2415,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22232232987880707,
|
|
"step": 410,
|
|
"valid_targets_mean": 4013.4,
|
|
"valid_targets_min": 1506
|
|
},
|
|
{
|
|
"epoch": 0.6618819776714514,
|
|
"grad_norm": 0.5211632420663286,
|
|
"learning_rate": 3.772209567198178e-05,
|
|
"loss": 0.2347,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23664991557598114,
|
|
"step": 415,
|
|
"valid_targets_mean": 4907.0,
|
|
"valid_targets_min": 1902
|
|
},
|
|
{
|
|
"epoch": 0.6698564593301436,
|
|
"grad_norm": 0.4767758351048497,
|
|
"learning_rate": 3.8177676537585425e-05,
|
|
"loss": 0.2493,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23484618961811066,
|
|
"step": 420,
|
|
"valid_targets_mean": 4859.4,
|
|
"valid_targets_min": 1921
|
|
},
|
|
{
|
|
"epoch": 0.6778309409888357,
|
|
"grad_norm": 1.2417309029902845,
|
|
"learning_rate": 3.863325740318907e-05,
|
|
"loss": 0.2442,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26523351669311523,
|
|
"step": 425,
|
|
"valid_targets_mean": 4098.0,
|
|
"valid_targets_min": 1888
|
|
},
|
|
{
|
|
"epoch": 0.6858054226475279,
|
|
"grad_norm": 0.6004123621981486,
|
|
"learning_rate": 3.9088838268792714e-05,
|
|
"loss": 0.2533,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2325078547000885,
|
|
"step": 430,
|
|
"valid_targets_mean": 3585.0,
|
|
"valid_targets_min": 1821
|
|
},
|
|
{
|
|
"epoch": 0.69377990430622,
|
|
"grad_norm": 0.5537664296614766,
|
|
"learning_rate": 3.954441913439636e-05,
|
|
"loss": 0.2253,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23680134117603302,
|
|
"step": 435,
|
|
"valid_targets_mean": 3540.8,
|
|
"valid_targets_min": 416
|
|
},
|
|
{
|
|
"epoch": 0.7017543859649122,
|
|
"grad_norm": 0.5101795039873435,
|
|
"learning_rate": 4e-05,
|
|
"loss": 0.2372,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21456380188465118,
|
|
"step": 440,
|
|
"valid_targets_mean": 3978.2,
|
|
"valid_targets_min": 1659
|
|
},
|
|
{
|
|
"epoch": 0.7097288676236044,
|
|
"grad_norm": 0.5031434314037209,
|
|
"learning_rate": 3.9999841858814384e-05,
|
|
"loss": 0.252,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23804379999637604,
|
|
"step": 445,
|
|
"valid_targets_mean": 4688.0,
|
|
"valid_targets_min": 2147
|
|
},
|
|
{
|
|
"epoch": 0.7177033492822966,
|
|
"grad_norm": 0.53053495346675,
|
|
"learning_rate": 3.999936743775839e-05,
|
|
"loss": 0.2612,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.32655003666877747,
|
|
"step": 450,
|
|
"valid_targets_mean": 5120.7,
|
|
"valid_targets_min": 528
|
|
},
|
|
{
|
|
"epoch": 0.7256778309409888,
|
|
"grad_norm": 0.5915625545140855,
|
|
"learning_rate": 3.9998576744334574e-05,
|
|
"loss": 0.2356,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22716417908668518,
|
|
"step": 455,
|
|
"valid_targets_mean": 3332.4,
|
|
"valid_targets_min": 377
|
|
},
|
|
{
|
|
"epoch": 0.733652312599681,
|
|
"grad_norm": 0.5945946110181048,
|
|
"learning_rate": 3.999746979104705e-05,
|
|
"loss": 0.2292,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26090529561042786,
|
|
"step": 460,
|
|
"valid_targets_mean": 3924.7,
|
|
"valid_targets_min": 1501
|
|
},
|
|
{
|
|
"epoch": 0.7416267942583732,
|
|
"grad_norm": 0.6362179372108026,
|
|
"learning_rate": 3.999604659540131e-05,
|
|
"loss": 0.2348,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22476539015769958,
|
|
"step": 465,
|
|
"valid_targets_mean": 4122.6,
|
|
"valid_targets_min": 2180
|
|
},
|
|
{
|
|
"epoch": 0.7496012759170654,
|
|
"grad_norm": 0.6471807631332203,
|
|
"learning_rate": 3.999430717990395e-05,
|
|
"loss": 0.2426,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23010434210300446,
|
|
"step": 470,
|
|
"valid_targets_mean": 3640.4,
|
|
"valid_targets_min": 1361
|
|
},
|
|
{
|
|
"epoch": 0.7575757575757576,
|
|
"grad_norm": 0.5703865355046819,
|
|
"learning_rate": 3.999225157206228e-05,
|
|
"loss": 0.2282,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23672768473625183,
|
|
"step": 475,
|
|
"valid_targets_mean": 4066.2,
|
|
"valid_targets_min": 1061
|
|
},
|
|
{
|
|
"epoch": 0.7655502392344498,
|
|
"grad_norm": 0.5858989027495684,
|
|
"learning_rate": 3.998987980438393e-05,
|
|
"loss": 0.256,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.32610398530960083,
|
|
"step": 480,
|
|
"valid_targets_mean": 4883.6,
|
|
"valid_targets_min": 1424
|
|
},
|
|
{
|
|
"epoch": 0.773524720893142,
|
|
"grad_norm": 0.5813017376748988,
|
|
"learning_rate": 3.9987191914376306e-05,
|
|
"loss": 0.2518,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2823824882507324,
|
|
"step": 485,
|
|
"valid_targets_mean": 4392.2,
|
|
"valid_targets_min": 359
|
|
},
|
|
{
|
|
"epoch": 0.7814992025518341,
|
|
"grad_norm": 0.5636916408209902,
|
|
"learning_rate": 3.998418794454604e-05,
|
|
"loss": 0.2332,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25734850764274597,
|
|
"step": 490,
|
|
"valid_targets_mean": 4360.2,
|
|
"valid_targets_min": 1272
|
|
},
|
|
{
|
|
"epoch": 0.7894736842105263,
|
|
"grad_norm": 0.4972880813638341,
|
|
"learning_rate": 3.998086794239825e-05,
|
|
"loss": 0.232,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22650626301765442,
|
|
"step": 495,
|
|
"valid_targets_mean": 5009.5,
|
|
"valid_targets_min": 1342
|
|
},
|
|
{
|
|
"epoch": 0.7974481658692185,
|
|
"grad_norm": 0.6083655046569959,
|
|
"learning_rate": 3.997723196043585e-05,
|
|
"loss": 0.2392,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.254880428314209,
|
|
"step": 500,
|
|
"valid_targets_mean": 3642.2,
|
|
"valid_targets_min": 1547
|
|
},
|
|
{
|
|
"epoch": 0.8054226475279107,
|
|
"grad_norm": 0.4854760864042363,
|
|
"learning_rate": 3.9973280056158695e-05,
|
|
"loss": 0.2498,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2242877185344696,
|
|
"step": 505,
|
|
"valid_targets_mean": 4663.0,
|
|
"valid_targets_min": 1162
|
|
},
|
|
{
|
|
"epoch": 0.8133971291866029,
|
|
"grad_norm": 0.4838878799701098,
|
|
"learning_rate": 3.9969012292062655e-05,
|
|
"loss": 0.2495,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2093224823474884,
|
|
"step": 510,
|
|
"valid_targets_mean": 4763.6,
|
|
"valid_targets_min": 1913
|
|
},
|
|
{
|
|
"epoch": 0.8213716108452951,
|
|
"grad_norm": 0.6268372836484649,
|
|
"learning_rate": 3.996442873563866e-05,
|
|
"loss": 0.2491,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.28269273042678833,
|
|
"step": 515,
|
|
"valid_targets_mean": 4417.9,
|
|
"valid_targets_min": 445
|
|
},
|
|
{
|
|
"epoch": 0.8293460925039873,
|
|
"grad_norm": 0.5129765312718306,
|
|
"learning_rate": 3.9959529459371624e-05,
|
|
"loss": 0.2351,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26153725385665894,
|
|
"step": 520,
|
|
"valid_targets_mean": 5101.4,
|
|
"valid_targets_min": 1506
|
|
},
|
|
{
|
|
"epoch": 0.8373205741626795,
|
|
"grad_norm": 0.614028186365327,
|
|
"learning_rate": 3.9954314540739284e-05,
|
|
"loss": 0.2273,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21306970715522766,
|
|
"step": 525,
|
|
"valid_targets_mean": 4610.0,
|
|
"valid_targets_min": 1604
|
|
},
|
|
{
|
|
"epoch": 0.8452950558213717,
|
|
"grad_norm": 0.5183329707466268,
|
|
"learning_rate": 3.994878406221097e-05,
|
|
"loss": 0.2447,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22915929555892944,
|
|
"step": 530,
|
|
"valid_targets_mean": 3992.6,
|
|
"valid_targets_min": 1596
|
|
},
|
|
{
|
|
"epoch": 0.8532695374800638,
|
|
"grad_norm": 0.48748891646891646,
|
|
"learning_rate": 3.994293811124632e-05,
|
|
"loss": 0.2446,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21784503757953644,
|
|
"step": 535,
|
|
"valid_targets_mean": 4734.7,
|
|
"valid_targets_min": 2272
|
|
},
|
|
{
|
|
"epoch": 0.861244019138756,
|
|
"grad_norm": 0.5577470057219691,
|
|
"learning_rate": 3.993677678029392e-05,
|
|
"loss": 0.2441,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.261586457490921,
|
|
"step": 540,
|
|
"valid_targets_mean": 4207.6,
|
|
"valid_targets_min": 2094
|
|
},
|
|
{
|
|
"epoch": 0.8692185007974481,
|
|
"grad_norm": 0.5361831016352179,
|
|
"learning_rate": 3.9930300166789765e-05,
|
|
"loss": 0.2246,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24090802669525146,
|
|
"step": 545,
|
|
"valid_targets_mean": 4649.8,
|
|
"valid_targets_min": 1477
|
|
},
|
|
{
|
|
"epoch": 0.8771929824561403,
|
|
"grad_norm": 0.5535412241281055,
|
|
"learning_rate": 3.992350837315581e-05,
|
|
"loss": 0.2477,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29971128702163696,
|
|
"step": 550,
|
|
"valid_targets_mean": 4245.1,
|
|
"valid_targets_min": 489
|
|
},
|
|
{
|
|
"epoch": 0.8851674641148325,
|
|
"grad_norm": 0.5925574761855791,
|
|
"learning_rate": 3.991640150679826e-05,
|
|
"loss": 0.2321,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22118769586086273,
|
|
"step": 555,
|
|
"valid_targets_mean": 4004.8,
|
|
"valid_targets_min": 447
|
|
},
|
|
{
|
|
"epoch": 0.8931419457735247,
|
|
"grad_norm": 0.5767617153732819,
|
|
"learning_rate": 3.990897968010596e-05,
|
|
"loss": 0.2275,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20498380064964294,
|
|
"step": 560,
|
|
"valid_targets_mean": 4139.0,
|
|
"valid_targets_min": 838
|
|
},
|
|
{
|
|
"epoch": 0.9011164274322169,
|
|
"grad_norm": 0.4835231868725431,
|
|
"learning_rate": 3.990124301044855e-05,
|
|
"loss": 0.2155,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1946374922990799,
|
|
"step": 565,
|
|
"valid_targets_mean": 4275.1,
|
|
"valid_targets_min": 2198
|
|
},
|
|
{
|
|
"epoch": 0.9090909090909091,
|
|
"grad_norm": 0.5296996222002296,
|
|
"learning_rate": 3.989319162017465e-05,
|
|
"loss": 0.2375,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2601141035556793,
|
|
"step": 570,
|
|
"valid_targets_mean": 3926.8,
|
|
"valid_targets_min": 763
|
|
},
|
|
{
|
|
"epoch": 0.9170653907496013,
|
|
"grad_norm": 0.4471460846434403,
|
|
"learning_rate": 3.988482563660989e-05,
|
|
"loss": 0.2377,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24986547231674194,
|
|
"step": 575,
|
|
"valid_targets_mean": 5239.9,
|
|
"valid_targets_min": 883
|
|
},
|
|
{
|
|
"epoch": 0.9250398724082934,
|
|
"grad_norm": 0.4777445171386324,
|
|
"learning_rate": 3.987614519205493e-05,
|
|
"loss": 0.2451,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22537899017333984,
|
|
"step": 580,
|
|
"valid_targets_mean": 4626.4,
|
|
"valid_targets_min": 632
|
|
},
|
|
{
|
|
"epoch": 0.9330143540669856,
|
|
"grad_norm": 0.49075192060803235,
|
|
"learning_rate": 3.986715042378334e-05,
|
|
"loss": 0.2346,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23568783700466156,
|
|
"step": 585,
|
|
"valid_targets_mean": 5212.7,
|
|
"valid_targets_min": 1512
|
|
},
|
|
{
|
|
"epoch": 0.9409888357256778,
|
|
"grad_norm": 0.5462851495643627,
|
|
"learning_rate": 3.985784147403947e-05,
|
|
"loss": 0.2376,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23025473952293396,
|
|
"step": 590,
|
|
"valid_targets_mean": 3878.6,
|
|
"valid_targets_min": 794
|
|
},
|
|
{
|
|
"epoch": 0.94896331738437,
|
|
"grad_norm": 0.5130018222856094,
|
|
"learning_rate": 3.9848218490036144e-05,
|
|
"loss": 0.2436,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21813419461250305,
|
|
"step": 595,
|
|
"valid_targets_mean": 4043.2,
|
|
"valid_targets_min": 1178
|
|
},
|
|
{
|
|
"epoch": 0.9569377990430622,
|
|
"grad_norm": 0.4970815254000969,
|
|
"learning_rate": 3.983828162395238e-05,
|
|
"loss": 0.2438,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24089229106903076,
|
|
"step": 600,
|
|
"valid_targets_mean": 4810.4,
|
|
"valid_targets_min": 985
|
|
},
|
|
{
|
|
"epoch": 0.9649122807017544,
|
|
"grad_norm": 0.5721802181838589,
|
|
"learning_rate": 3.9828031032930944e-05,
|
|
"loss": 0.2342,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2031339704990387,
|
|
"step": 605,
|
|
"valid_targets_mean": 3739.4,
|
|
"valid_targets_min": 1367
|
|
},
|
|
{
|
|
"epoch": 0.9728867623604466,
|
|
"grad_norm": 0.5188176743792337,
|
|
"learning_rate": 3.98174668790759e-05,
|
|
"loss": 0.2304,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.27065911889076233,
|
|
"step": 610,
|
|
"valid_targets_mean": 4289.9,
|
|
"valid_targets_min": 644
|
|
},
|
|
{
|
|
"epoch": 0.9808612440191388,
|
|
"grad_norm": 0.5016486734647293,
|
|
"learning_rate": 3.9806589329450045e-05,
|
|
"loss": 0.2333,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2198173850774765,
|
|
"step": 615,
|
|
"valid_targets_mean": 4156.7,
|
|
"valid_targets_min": 861
|
|
},
|
|
{
|
|
"epoch": 0.988835725677831,
|
|
"grad_norm": 0.495182154320868,
|
|
"learning_rate": 3.979539855607222e-05,
|
|
"loss": 0.2451,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22900980710983276,
|
|
"step": 620,
|
|
"valid_targets_mean": 4202.0,
|
|
"valid_targets_min": 1404
|
|
},
|
|
{
|
|
"epoch": 0.9968102073365231,
|
|
"grad_norm": 0.4977265786366298,
|
|
"learning_rate": 3.9783894735914646e-05,
|
|
"loss": 0.2291,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20909643173217773,
|
|
"step": 625,
|
|
"valid_targets_mean": 4351.4,
|
|
"valid_targets_min": 1625
|
|
},
|
|
{
|
|
"epoch": 1.0047846889952152,
|
|
"grad_norm": 0.5032203543453754,
|
|
"learning_rate": 3.9772078050900105e-05,
|
|
"loss": 0.2276,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22026538848876953,
|
|
"step": 630,
|
|
"valid_targets_mean": 3949.2,
|
|
"valid_targets_min": 1519
|
|
},
|
|
{
|
|
"epoch": 1.0127591706539074,
|
|
"grad_norm": 0.5229641736587483,
|
|
"learning_rate": 3.9759948687899055e-05,
|
|
"loss": 0.2418,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25218960642814636,
|
|
"step": 635,
|
|
"valid_targets_mean": 3920.1,
|
|
"valid_targets_min": 667
|
|
},
|
|
{
|
|
"epoch": 1.0207336523125996,
|
|
"grad_norm": 0.5502168871278199,
|
|
"learning_rate": 3.974750683872667e-05,
|
|
"loss": 0.2176,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21615633368492126,
|
|
"step": 640,
|
|
"valid_targets_mean": 3966.8,
|
|
"valid_targets_min": 1387
|
|
},
|
|
{
|
|
"epoch": 1.0287081339712918,
|
|
"grad_norm": 0.5482724060330286,
|
|
"learning_rate": 3.973475270013984e-05,
|
|
"loss": 0.2195,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22589367628097534,
|
|
"step": 645,
|
|
"valid_targets_mean": 4249.8,
|
|
"valid_targets_min": 1645
|
|
},
|
|
{
|
|
"epoch": 1.036682615629984,
|
|
"grad_norm": 0.48681150157765296,
|
|
"learning_rate": 3.972168647383402e-05,
|
|
"loss": 0.2162,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18984422087669373,
|
|
"step": 650,
|
|
"valid_targets_mean": 4701.6,
|
|
"valid_targets_min": 1897
|
|
},
|
|
{
|
|
"epoch": 1.0446570972886762,
|
|
"grad_norm": 0.5286726205605536,
|
|
"learning_rate": 3.970830836644006e-05,
|
|
"loss": 0.2313,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2323208749294281,
|
|
"step": 655,
|
|
"valid_targets_mean": 3915.1,
|
|
"valid_targets_min": 2094
|
|
},
|
|
{
|
|
"epoch": 1.0526315789473684,
|
|
"grad_norm": 0.5463582558960633,
|
|
"learning_rate": 3.9694618589520945e-05,
|
|
"loss": 0.2122,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24363505840301514,
|
|
"step": 660,
|
|
"valid_targets_mean": 3746.0,
|
|
"valid_targets_min": 1271
|
|
},
|
|
{
|
|
"epoch": 1.0606060606060606,
|
|
"grad_norm": 0.4692304274685834,
|
|
"learning_rate": 3.9680617359568414e-05,
|
|
"loss": 0.2155,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22661146521568298,
|
|
"step": 665,
|
|
"valid_targets_mean": 4613.2,
|
|
"valid_targets_min": 811
|
|
},
|
|
{
|
|
"epoch": 1.0685805422647527,
|
|
"grad_norm": 0.5076971387573798,
|
|
"learning_rate": 3.966630489799959e-05,
|
|
"loss": 0.2113,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2089880257844925,
|
|
"step": 670,
|
|
"valid_targets_mean": 3869.4,
|
|
"valid_targets_min": 1038
|
|
},
|
|
{
|
|
"epoch": 1.076555023923445,
|
|
"grad_norm": 0.4498347073038136,
|
|
"learning_rate": 3.9651681431153445e-05,
|
|
"loss": 0.2237,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21558557450771332,
|
|
"step": 675,
|
|
"valid_targets_mean": 4482.1,
|
|
"valid_targets_min": 1271
|
|
},
|
|
{
|
|
"epoch": 1.0845295055821371,
|
|
"grad_norm": 0.5064153829663555,
|
|
"learning_rate": 3.96367471902872e-05,
|
|
"loss": 0.2412,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2504901587963104,
|
|
"step": 680,
|
|
"valid_targets_mean": 5206.8,
|
|
"valid_targets_min": 1827
|
|
},
|
|
{
|
|
"epoch": 1.0925039872408293,
|
|
"grad_norm": 0.5259888970380884,
|
|
"learning_rate": 3.9621502411572705e-05,
|
|
"loss": 0.2168,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2569437623023987,
|
|
"step": 685,
|
|
"valid_targets_mean": 4280.9,
|
|
"valid_targets_min": 655
|
|
},
|
|
{
|
|
"epoch": 1.1004784688995215,
|
|
"grad_norm": 0.6046706105924078,
|
|
"learning_rate": 3.960594733609273e-05,
|
|
"loss": 0.2104,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21926787495613098,
|
|
"step": 690,
|
|
"valid_targets_mean": 3736.1,
|
|
"valid_targets_min": 1596
|
|
},
|
|
{
|
|
"epoch": 1.1084529505582137,
|
|
"grad_norm": 0.48006122244847677,
|
|
"learning_rate": 3.9590082209837054e-05,
|
|
"loss": 0.232,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23299375176429749,
|
|
"step": 695,
|
|
"valid_targets_mean": 4608.6,
|
|
"valid_targets_min": 1949
|
|
},
|
|
{
|
|
"epoch": 1.1164274322169059,
|
|
"grad_norm": 0.553902049161527,
|
|
"learning_rate": 3.957390728369867e-05,
|
|
"loss": 0.2196,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2619227170944214,
|
|
"step": 700,
|
|
"valid_targets_mean": 3900.5,
|
|
"valid_targets_min": 1523
|
|
},
|
|
{
|
|
"epoch": 1.124401913875598,
|
|
"grad_norm": 0.5567781830902339,
|
|
"learning_rate": 3.955742281346979e-05,
|
|
"loss": 0.2306,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22066161036491394,
|
|
"step": 705,
|
|
"valid_targets_mean": 3926.1,
|
|
"valid_targets_min": 818
|
|
},
|
|
{
|
|
"epoch": 1.1323763955342903,
|
|
"grad_norm": 0.5030489445709994,
|
|
"learning_rate": 3.9540629059837767e-05,
|
|
"loss": 0.2085,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21337543427944183,
|
|
"step": 710,
|
|
"valid_targets_mean": 4503.9,
|
|
"valid_targets_min": 1611
|
|
},
|
|
{
|
|
"epoch": 1.1403508771929824,
|
|
"grad_norm": 0.4521443857332383,
|
|
"learning_rate": 3.952352628838102e-05,
|
|
"loss": 0.2205,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2025744616985321,
|
|
"step": 715,
|
|
"valid_targets_mean": 4265.0,
|
|
"valid_targets_min": 1616
|
|
},
|
|
{
|
|
"epoch": 1.1483253588516746,
|
|
"grad_norm": 0.5501529406774962,
|
|
"learning_rate": 3.95061147695648e-05,
|
|
"loss": 0.2436,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26300299167633057,
|
|
"step": 720,
|
|
"valid_targets_mean": 5110.0,
|
|
"valid_targets_min": 1602
|
|
},
|
|
{
|
|
"epoch": 1.1562998405103668,
|
|
"grad_norm": 0.5721986878872145,
|
|
"learning_rate": 3.9488394778736935e-05,
|
|
"loss": 0.2363,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2308548390865326,
|
|
"step": 725,
|
|
"valid_targets_mean": 4003.0,
|
|
"valid_targets_min": 1036
|
|
},
|
|
{
|
|
"epoch": 1.164274322169059,
|
|
"grad_norm": 0.4803368155501433,
|
|
"learning_rate": 3.947036659612345e-05,
|
|
"loss": 0.222,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23096925020217896,
|
|
"step": 730,
|
|
"valid_targets_mean": 4496.0,
|
|
"valid_targets_min": 1436
|
|
},
|
|
{
|
|
"epoch": 1.1722488038277512,
|
|
"grad_norm": 0.5066273119569031,
|
|
"learning_rate": 3.945203050682418e-05,
|
|
"loss": 0.2175,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23388348519802094,
|
|
"step": 735,
|
|
"valid_targets_mean": 3819.8,
|
|
"valid_targets_min": 1237
|
|
},
|
|
{
|
|
"epoch": 1.1802232854864434,
|
|
"grad_norm": 0.4830554741716881,
|
|
"learning_rate": 3.94333868008082e-05,
|
|
"loss": 0.2426,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25258368253707886,
|
|
"step": 740,
|
|
"valid_targets_mean": 4634.4,
|
|
"valid_targets_min": 916
|
|
},
|
|
{
|
|
"epoch": 1.1881977671451356,
|
|
"grad_norm": 0.4843244438344029,
|
|
"learning_rate": 3.94144357729093e-05,
|
|
"loss": 0.22,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.206687331199646,
|
|
"step": 745,
|
|
"valid_targets_mean": 4439.3,
|
|
"valid_targets_min": 458
|
|
},
|
|
{
|
|
"epoch": 1.1961722488038278,
|
|
"grad_norm": 0.5243718793861886,
|
|
"learning_rate": 3.939517772282127e-05,
|
|
"loss": 0.2269,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2652336359024048,
|
|
"step": 750,
|
|
"valid_targets_mean": 4485.8,
|
|
"valid_targets_min": 891
|
|
},
|
|
{
|
|
"epoch": 1.20414673046252,
|
|
"grad_norm": 0.5788052182039267,
|
|
"learning_rate": 3.93756129550932e-05,
|
|
"loss": 0.2148,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20965895056724548,
|
|
"step": 755,
|
|
"valid_targets_mean": 3390.4,
|
|
"valid_targets_min": 278
|
|
},
|
|
{
|
|
"epoch": 1.2121212121212122,
|
|
"grad_norm": 0.620883837268961,
|
|
"learning_rate": 3.935574177912465e-05,
|
|
"loss": 0.2143,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23990340530872345,
|
|
"step": 760,
|
|
"valid_targets_mean": 4009.7,
|
|
"valid_targets_min": 2077
|
|
},
|
|
{
|
|
"epoch": 1.2200956937799043,
|
|
"grad_norm": 0.5871274762207975,
|
|
"learning_rate": 3.9335564509160746e-05,
|
|
"loss": 0.2273,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22472716867923737,
|
|
"step": 765,
|
|
"valid_targets_mean": 4286.5,
|
|
"valid_targets_min": 319
|
|
},
|
|
{
|
|
"epoch": 1.2280701754385965,
|
|
"grad_norm": 0.5010639217355422,
|
|
"learning_rate": 3.931508146428724e-05,
|
|
"loss": 0.2218,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21770870685577393,
|
|
"step": 770,
|
|
"valid_targets_mean": 4047.3,
|
|
"valid_targets_min": 1735
|
|
},
|
|
{
|
|
"epoch": 1.2360446570972887,
|
|
"grad_norm": 0.5039635008586113,
|
|
"learning_rate": 3.929429296842542e-05,
|
|
"loss": 0.2272,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19232256710529327,
|
|
"step": 775,
|
|
"valid_targets_mean": 3779.6,
|
|
"valid_targets_min": 1637
|
|
},
|
|
{
|
|
"epoch": 1.244019138755981,
|
|
"grad_norm": 0.5189905433600639,
|
|
"learning_rate": 3.927319935032703e-05,
|
|
"loss": 0.2304,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23623117804527283,
|
|
"step": 780,
|
|
"valid_targets_mean": 4330.4,
|
|
"valid_targets_min": 342
|
|
},
|
|
{
|
|
"epoch": 1.251993620414673,
|
|
"grad_norm": 0.6627491973095729,
|
|
"learning_rate": 3.925180094356905e-05,
|
|
"loss": 0.2394,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24548950791358948,
|
|
"step": 785,
|
|
"valid_targets_mean": 4172.2,
|
|
"valid_targets_min": 1251
|
|
},
|
|
{
|
|
"epoch": 1.2599681020733653,
|
|
"grad_norm": 0.5079001254806608,
|
|
"learning_rate": 3.9230098086548414e-05,
|
|
"loss": 0.2202,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21201512217521667,
|
|
"step": 790,
|
|
"valid_targets_mean": 4065.2,
|
|
"valid_targets_min": 585
|
|
},
|
|
{
|
|
"epoch": 1.2679425837320575,
|
|
"grad_norm": 0.47150540150141773,
|
|
"learning_rate": 3.920809112247668e-05,
|
|
"loss": 0.2112,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20915411412715912,
|
|
"step": 795,
|
|
"valid_targets_mean": 4409.1,
|
|
"valid_targets_min": 1091
|
|
},
|
|
{
|
|
"epoch": 1.2759170653907497,
|
|
"grad_norm": 0.5514140187109288,
|
|
"learning_rate": 3.918578039937459e-05,
|
|
"loss": 0.2214,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23035244643688202,
|
|
"step": 800,
|
|
"valid_targets_mean": 4182.2,
|
|
"valid_targets_min": 2198
|
|
},
|
|
{
|
|
"epoch": 1.2838915470494419,
|
|
"grad_norm": 0.4500995102298934,
|
|
"learning_rate": 3.916316627006656e-05,
|
|
"loss": 0.2108,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2097713053226471,
|
|
"step": 805,
|
|
"valid_targets_mean": 4795.3,
|
|
"valid_targets_min": 1902
|
|
},
|
|
{
|
|
"epoch": 1.291866028708134,
|
|
"grad_norm": 0.52633246142109,
|
|
"learning_rate": 3.914024909217511e-05,
|
|
"loss": 0.2264,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19991585612297058,
|
|
"step": 810,
|
|
"valid_targets_mean": 3731.4,
|
|
"valid_targets_min": 1883
|
|
},
|
|
{
|
|
"epoch": 1.2998405103668262,
|
|
"grad_norm": 0.4467294156911823,
|
|
"learning_rate": 3.911702922811522e-05,
|
|
"loss": 0.2144,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.203420490026474,
|
|
"step": 815,
|
|
"valid_targets_mean": 4653.8,
|
|
"valid_targets_min": 2174
|
|
},
|
|
{
|
|
"epoch": 1.3078149920255182,
|
|
"grad_norm": 0.5081778926888767,
|
|
"learning_rate": 3.909350704508856e-05,
|
|
"loss": 0.2137,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20174889266490936,
|
|
"step": 820,
|
|
"valid_targets_mean": 3439.1,
|
|
"valid_targets_min": 1350
|
|
},
|
|
{
|
|
"epoch": 1.3157894736842106,
|
|
"grad_norm": 0.9310514979548232,
|
|
"learning_rate": 3.906968291507773e-05,
|
|
"loss": 0.2228,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2513682246208191,
|
|
"step": 825,
|
|
"valid_targets_mean": 4917.4,
|
|
"valid_targets_min": 1283
|
|
},
|
|
{
|
|
"epoch": 1.3237639553429026,
|
|
"grad_norm": 0.5555892851581838,
|
|
"learning_rate": 3.904555721484034e-05,
|
|
"loss": 0.2026,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20496833324432373,
|
|
"step": 830,
|
|
"valid_targets_mean": 4021.7,
|
|
"valid_targets_min": 248
|
|
},
|
|
{
|
|
"epoch": 1.331738437001595,
|
|
"grad_norm": 0.47232094168845945,
|
|
"learning_rate": 3.9021130325903076e-05,
|
|
"loss": 0.2119,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19785736501216888,
|
|
"step": 835,
|
|
"valid_targets_mean": 5302.4,
|
|
"valid_targets_min": 2353
|
|
},
|
|
{
|
|
"epoch": 1.339712918660287,
|
|
"grad_norm": 0.5812182577655588,
|
|
"learning_rate": 3.899640263455566e-05,
|
|
"loss": 0.2062,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18643829226493835,
|
|
"step": 840,
|
|
"valid_targets_mean": 3905.0,
|
|
"valid_targets_min": 961
|
|
},
|
|
{
|
|
"epoch": 1.3476874003189794,
|
|
"grad_norm": 0.5500888468056214,
|
|
"learning_rate": 3.897137453184472e-05,
|
|
"loss": 0.1961,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18745486438274384,
|
|
"step": 845,
|
|
"valid_targets_mean": 3677.1,
|
|
"valid_targets_min": 574
|
|
},
|
|
{
|
|
"epoch": 1.3556618819776713,
|
|
"grad_norm": 0.6847492160049518,
|
|
"learning_rate": 3.894604641356767e-05,
|
|
"loss": 0.216,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2598264813423157,
|
|
"step": 850,
|
|
"valid_targets_mean": 4733.8,
|
|
"valid_targets_min": 1433
|
|
},
|
|
{
|
|
"epoch": 1.3636363636363638,
|
|
"grad_norm": 0.45372699194812466,
|
|
"learning_rate": 3.8920418680266346e-05,
|
|
"loss": 0.2196,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19950619339942932,
|
|
"step": 855,
|
|
"valid_targets_mean": 4372.9,
|
|
"valid_targets_min": 933
|
|
},
|
|
{
|
|
"epoch": 1.3716108452950557,
|
|
"grad_norm": 0.49007066533357946,
|
|
"learning_rate": 3.889449173722077e-05,
|
|
"loss": 0.2219,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20979920029640198,
|
|
"step": 860,
|
|
"valid_targets_mean": 3954.1,
|
|
"valid_targets_min": 1402
|
|
},
|
|
{
|
|
"epoch": 1.3795853269537481,
|
|
"grad_norm": 0.5091212381448587,
|
|
"learning_rate": 3.8868265994442694e-05,
|
|
"loss": 0.2061,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22372718155384064,
|
|
"step": 865,
|
|
"valid_targets_mean": 4045.6,
|
|
"valid_targets_min": 600
|
|
},
|
|
{
|
|
"epoch": 1.38755980861244,
|
|
"grad_norm": 0.49225432870048075,
|
|
"learning_rate": 3.8841741866669126e-05,
|
|
"loss": 0.1956,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1865185797214508,
|
|
"step": 870,
|
|
"valid_targets_mean": 3703.1,
|
|
"valid_targets_min": 1619
|
|
},
|
|
{
|
|
"epoch": 1.3955342902711323,
|
|
"grad_norm": 0.47745581156193523,
|
|
"learning_rate": 3.881491977335577e-05,
|
|
"loss": 0.2085,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2252838909626007,
|
|
"step": 875,
|
|
"valid_targets_mean": 4296.0,
|
|
"valid_targets_min": 1796
|
|
},
|
|
{
|
|
"epoch": 1.4035087719298245,
|
|
"grad_norm": 0.43928492337677577,
|
|
"learning_rate": 3.878780013867038e-05,
|
|
"loss": 0.2232,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20441752672195435,
|
|
"step": 880,
|
|
"valid_targets_mean": 4535.2,
|
|
"valid_targets_min": 1824
|
|
},
|
|
{
|
|
"epoch": 1.4114832535885167,
|
|
"grad_norm": 0.5033914360127287,
|
|
"learning_rate": 3.8760383391486074e-05,
|
|
"loss": 0.2246,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.30052030086517334,
|
|
"step": 885,
|
|
"valid_targets_mean": 4486.1,
|
|
"valid_targets_min": 531
|
|
},
|
|
{
|
|
"epoch": 1.4194577352472089,
|
|
"grad_norm": 0.46617057902823866,
|
|
"learning_rate": 3.873266996537456e-05,
|
|
"loss": 0.2118,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19758670032024384,
|
|
"step": 890,
|
|
"valid_targets_mean": 4368.9,
|
|
"valid_targets_min": 588
|
|
},
|
|
{
|
|
"epoch": 1.427432216905901,
|
|
"grad_norm": 0.43206619780636607,
|
|
"learning_rate": 3.8704660298599225e-05,
|
|
"loss": 0.223,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2040582001209259,
|
|
"step": 895,
|
|
"valid_targets_mean": 6326.5,
|
|
"valid_targets_min": 1686
|
|
},
|
|
{
|
|
"epoch": 1.4354066985645932,
|
|
"grad_norm": 0.48207222426509794,
|
|
"learning_rate": 3.867635483410827e-05,
|
|
"loss": 0.2178,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2013099491596222,
|
|
"step": 900,
|
|
"valid_targets_mean": 4192.6,
|
|
"valid_targets_min": 1486
|
|
},
|
|
{
|
|
"epoch": 1.4433811802232854,
|
|
"grad_norm": 0.4666337478470858,
|
|
"learning_rate": 3.864775401952767e-05,
|
|
"loss": 0.2141,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22802595794200897,
|
|
"step": 905,
|
|
"valid_targets_mean": 4553.6,
|
|
"valid_targets_min": 376
|
|
},
|
|
{
|
|
"epoch": 1.4513556618819776,
|
|
"grad_norm": 0.44722531533793,
|
|
"learning_rate": 3.8618858307154085e-05,
|
|
"loss": 0.2123,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20279376208782196,
|
|
"step": 910,
|
|
"valid_targets_mean": 4846.8,
|
|
"valid_targets_min": 1507
|
|
},
|
|
{
|
|
"epoch": 1.4593301435406698,
|
|
"grad_norm": 0.5371093701872409,
|
|
"learning_rate": 3.8589668153947743e-05,
|
|
"loss": 0.2151,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23264798521995544,
|
|
"step": 915,
|
|
"valid_targets_mean": 4581.4,
|
|
"valid_targets_min": 386
|
|
},
|
|
{
|
|
"epoch": 1.467304625199362,
|
|
"grad_norm": 0.5165471090234898,
|
|
"learning_rate": 3.8560184021525194e-05,
|
|
"loss": 0.2096,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2197292149066925,
|
|
"step": 920,
|
|
"valid_targets_mean": 4447.9,
|
|
"valid_targets_min": 2488
|
|
},
|
|
{
|
|
"epoch": 1.4752791068580542,
|
|
"grad_norm": 0.4798137016348713,
|
|
"learning_rate": 3.853040637615199e-05,
|
|
"loss": 0.2136,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22513622045516968,
|
|
"step": 925,
|
|
"valid_targets_mean": 4678.5,
|
|
"valid_targets_min": 286
|
|
},
|
|
{
|
|
"epoch": 1.4832535885167464,
|
|
"grad_norm": 0.4979070884147898,
|
|
"learning_rate": 3.850033568873536e-05,
|
|
"loss": 0.2098,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23755387961864471,
|
|
"step": 930,
|
|
"valid_targets_mean": 4021.5,
|
|
"valid_targets_min": 1577
|
|
},
|
|
{
|
|
"epoch": 1.4912280701754386,
|
|
"grad_norm": 0.4572253404809861,
|
|
"learning_rate": 3.8469972434816706e-05,
|
|
"loss": 0.2153,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23543182015419006,
|
|
"step": 935,
|
|
"valid_targets_mean": 5077.9,
|
|
"valid_targets_min": 1116
|
|
},
|
|
{
|
|
"epoch": 1.4992025518341308,
|
|
"grad_norm": 0.47248195445727853,
|
|
"learning_rate": 3.843931709456414e-05,
|
|
"loss": 0.2099,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20499253273010254,
|
|
"step": 940,
|
|
"valid_targets_mean": 3970.7,
|
|
"valid_targets_min": 1530
|
|
},
|
|
{
|
|
"epoch": 1.507177033492823,
|
|
"grad_norm": 0.42598999174275803,
|
|
"learning_rate": 3.840837015276483e-05,
|
|
"loss": 0.2266,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2627623975276947,
|
|
"step": 945,
|
|
"valid_targets_mean": 5831.6,
|
|
"valid_targets_min": 2288
|
|
},
|
|
{
|
|
"epoch": 1.5151515151515151,
|
|
"grad_norm": 0.6269538410575813,
|
|
"learning_rate": 3.83771320988174e-05,
|
|
"loss": 0.207,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18554210662841797,
|
|
"step": 950,
|
|
"valid_targets_mean": 4548.8,
|
|
"valid_targets_min": 1487
|
|
},
|
|
{
|
|
"epoch": 1.5231259968102073,
|
|
"grad_norm": 0.500690066016818,
|
|
"learning_rate": 3.834560342672413e-05,
|
|
"loss": 0.2255,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22179262340068817,
|
|
"step": 955,
|
|
"valid_targets_mean": 3794.7,
|
|
"valid_targets_min": 1451
|
|
},
|
|
{
|
|
"epoch": 1.5311004784688995,
|
|
"grad_norm": 0.4844197425861743,
|
|
"learning_rate": 3.831378463508318e-05,
|
|
"loss": 0.2044,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1850634515285492,
|
|
"step": 960,
|
|
"valid_targets_mean": 4430.5,
|
|
"valid_targets_min": 1689
|
|
},
|
|
{
|
|
"epoch": 1.5390749601275917,
|
|
"grad_norm": 0.49254151371909743,
|
|
"learning_rate": 3.8281676227080694e-05,
|
|
"loss": 0.227,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2288564294576645,
|
|
"step": 965,
|
|
"valid_targets_mean": 3998.4,
|
|
"valid_targets_min": 291
|
|
},
|
|
{
|
|
"epoch": 1.547049441786284,
|
|
"grad_norm": 0.5742983155284956,
|
|
"learning_rate": 3.824927871048284e-05,
|
|
"loss": 0.2122,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2189641147851944,
|
|
"step": 970,
|
|
"valid_targets_mean": 3550.6,
|
|
"valid_targets_min": 1509
|
|
},
|
|
{
|
|
"epoch": 1.555023923444976,
|
|
"grad_norm": 0.5433468243901337,
|
|
"learning_rate": 3.8216592597627797e-05,
|
|
"loss": 0.2134,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.223488450050354,
|
|
"step": 975,
|
|
"valid_targets_mean": 4082.4,
|
|
"valid_targets_min": 1704
|
|
},
|
|
{
|
|
"epoch": 1.5629984051036683,
|
|
"grad_norm": 0.5196900072582149,
|
|
"learning_rate": 3.818361840541761e-05,
|
|
"loss": 0.2059,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20587456226348877,
|
|
"step": 980,
|
|
"valid_targets_mean": 4621.8,
|
|
"valid_targets_min": 815
|
|
},
|
|
{
|
|
"epoch": 1.5709728867623605,
|
|
"grad_norm": 0.5135498979104908,
|
|
"learning_rate": 3.815035665531008e-05,
|
|
"loss": 0.2257,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24312788248062134,
|
|
"step": 985,
|
|
"valid_targets_mean": 3476.9,
|
|
"valid_targets_min": 462
|
|
},
|
|
{
|
|
"epoch": 1.5789473684210527,
|
|
"grad_norm": 0.7898137054288659,
|
|
"learning_rate": 3.811680787331047e-05,
|
|
"loss": 0.2321,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22189228236675262,
|
|
"step": 990,
|
|
"valid_targets_mean": 3913.2,
|
|
"valid_targets_min": 1367
|
|
},
|
|
{
|
|
"epoch": 1.5869218500797448,
|
|
"grad_norm": 0.43969010537708586,
|
|
"learning_rate": 3.8082972589963175e-05,
|
|
"loss": 0.2182,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20273138582706451,
|
|
"step": 995,
|
|
"valid_targets_mean": 4583.5,
|
|
"valid_targets_min": 1345
|
|
},
|
|
{
|
|
"epoch": 1.594896331738437,
|
|
"grad_norm": 0.5555222366008191,
|
|
"learning_rate": 3.80488513403434e-05,
|
|
"loss": 0.2225,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21825510263442993,
|
|
"step": 1000,
|
|
"valid_targets_mean": 3541.6,
|
|
"valid_targets_min": 1470
|
|
},
|
|
{
|
|
"epoch": 1.6028708133971292,
|
|
"grad_norm": 0.4321959152756272,
|
|
"learning_rate": 3.8014444664048616e-05,
|
|
"loss": 0.2096,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1973963975906372,
|
|
"step": 1005,
|
|
"valid_targets_mean": 4639.2,
|
|
"valid_targets_min": 1503
|
|
},
|
|
{
|
|
"epoch": 1.6108452950558214,
|
|
"grad_norm": 0.46446966286351066,
|
|
"learning_rate": 3.797975310519009e-05,
|
|
"loss": 0.21,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2148362100124359,
|
|
"step": 1010,
|
|
"valid_targets_mean": 4174.1,
|
|
"valid_targets_min": 1381
|
|
},
|
|
{
|
|
"epoch": 1.6188197767145136,
|
|
"grad_norm": 0.48913950514237353,
|
|
"learning_rate": 3.794477721238425e-05,
|
|
"loss": 0.2153,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2184276580810547,
|
|
"step": 1015,
|
|
"valid_targets_mean": 4014.8,
|
|
"valid_targets_min": 1722
|
|
},
|
|
{
|
|
"epoch": 1.6267942583732058,
|
|
"grad_norm": 0.4303667207370021,
|
|
"learning_rate": 3.7909517538744e-05,
|
|
"loss": 0.2129,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2015509009361267,
|
|
"step": 1020,
|
|
"valid_targets_mean": 4448.6,
|
|
"valid_targets_min": 1590
|
|
},
|
|
{
|
|
"epoch": 1.6347687400318978,
|
|
"grad_norm": 0.5289047871909957,
|
|
"learning_rate": 3.7873974641870006e-05,
|
|
"loss": 0.2167,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21874231100082397,
|
|
"step": 1025,
|
|
"valid_targets_mean": 3820.2,
|
|
"valid_targets_min": 1733
|
|
},
|
|
{
|
|
"epoch": 1.6427432216905902,
|
|
"grad_norm": 0.4762507453851103,
|
|
"learning_rate": 3.7838149083841856e-05,
|
|
"loss": 0.2232,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2580573558807373,
|
|
"step": 1030,
|
|
"valid_targets_mean": 5352.6,
|
|
"valid_targets_min": 1846
|
|
},
|
|
{
|
|
"epoch": 1.6507177033492821,
|
|
"grad_norm": 0.47198599267721897,
|
|
"learning_rate": 3.7802041431209166e-05,
|
|
"loss": 0.2052,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.205755352973938,
|
|
"step": 1035,
|
|
"valid_targets_mean": 4287.4,
|
|
"valid_targets_min": 1811
|
|
},
|
|
{
|
|
"epoch": 1.6586921850079746,
|
|
"grad_norm": 0.5028769618064659,
|
|
"learning_rate": 3.776565225498264e-05,
|
|
"loss": 0.2138,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20170894265174866,
|
|
"step": 1040,
|
|
"valid_targets_mean": 3809.2,
|
|
"valid_targets_min": 1688
|
|
},
|
|
{
|
|
"epoch": 1.6666666666666665,
|
|
"grad_norm": 0.504718091730453,
|
|
"learning_rate": 3.7728982130625025e-05,
|
|
"loss": 0.219,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2416345477104187,
|
|
"step": 1045,
|
|
"valid_targets_mean": 4589.6,
|
|
"valid_targets_min": 1648
|
|
},
|
|
{
|
|
"epoch": 1.674641148325359,
|
|
"grad_norm": 0.554866907792144,
|
|
"learning_rate": 3.769203163804202e-05,
|
|
"loss": 0.217,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22144190967082977,
|
|
"step": 1050,
|
|
"valid_targets_mean": 4044.3,
|
|
"valid_targets_min": 867
|
|
},
|
|
{
|
|
"epoch": 1.682615629984051,
|
|
"grad_norm": 0.4912047494949895,
|
|
"learning_rate": 3.7654801361573076e-05,
|
|
"loss": 0.1908,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19729235768318176,
|
|
"step": 1055,
|
|
"valid_targets_mean": 4379.2,
|
|
"valid_targets_min": 1476
|
|
},
|
|
{
|
|
"epoch": 1.6905901116427433,
|
|
"grad_norm": 0.5033895385309353,
|
|
"learning_rate": 3.761729188998222e-05,
|
|
"loss": 0.207,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19748783111572266,
|
|
"step": 1060,
|
|
"valid_targets_mean": 4624.8,
|
|
"valid_targets_min": 2190
|
|
},
|
|
{
|
|
"epoch": 1.6985645933014353,
|
|
"grad_norm": 0.5473790729590564,
|
|
"learning_rate": 3.757950381644868e-05,
|
|
"loss": 0.2199,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25469958782196045,
|
|
"step": 1065,
|
|
"valid_targets_mean": 3925.2,
|
|
"valid_targets_min": 599
|
|
},
|
|
{
|
|
"epoch": 1.7065390749601277,
|
|
"grad_norm": 0.5180414304101282,
|
|
"learning_rate": 3.7541437738557524e-05,
|
|
"loss": 0.2124,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19136464595794678,
|
|
"step": 1070,
|
|
"valid_targets_mean": 3960.3,
|
|
"valid_targets_min": 1645
|
|
},
|
|
{
|
|
"epoch": 1.7145135566188197,
|
|
"grad_norm": 0.5003214798788846,
|
|
"learning_rate": 3.750309425829022e-05,
|
|
"loss": 0.2144,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19182105362415314,
|
|
"step": 1075,
|
|
"valid_targets_mean": 3812.0,
|
|
"valid_targets_min": 1303
|
|
},
|
|
{
|
|
"epoch": 1.722488038277512,
|
|
"grad_norm": 0.49690764989147096,
|
|
"learning_rate": 3.746447398201512e-05,
|
|
"loss": 0.2172,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19056028127670288,
|
|
"step": 1080,
|
|
"valid_targets_mean": 4023.6,
|
|
"valid_targets_min": 1050
|
|
},
|
|
{
|
|
"epoch": 1.730462519936204,
|
|
"grad_norm": 0.5016898081489446,
|
|
"learning_rate": 3.7425577520477846e-05,
|
|
"loss": 0.2186,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21159327030181885,
|
|
"step": 1085,
|
|
"valid_targets_mean": 4200.7,
|
|
"valid_targets_min": 1525
|
|
},
|
|
{
|
|
"epoch": 1.7384370015948964,
|
|
"grad_norm": 0.5659690695722234,
|
|
"learning_rate": 3.738640548879166e-05,
|
|
"loss": 0.2191,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2166757434606552,
|
|
"step": 1090,
|
|
"valid_targets_mean": 4174.4,
|
|
"valid_targets_min": 1178
|
|
},
|
|
{
|
|
"epoch": 1.7464114832535884,
|
|
"grad_norm": 0.4955709318314084,
|
|
"learning_rate": 3.7346958506427696e-05,
|
|
"loss": 0.2176,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26833972334861755,
|
|
"step": 1095,
|
|
"valid_targets_mean": 4771.3,
|
|
"valid_targets_min": 338
|
|
},
|
|
{
|
|
"epoch": 1.7543859649122808,
|
|
"grad_norm": 0.38960186550719944,
|
|
"learning_rate": 3.730723719720523e-05,
|
|
"loss": 0.2002,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1986948400735855,
|
|
"step": 1100,
|
|
"valid_targets_mean": 5489.8,
|
|
"valid_targets_min": 2149
|
|
},
|
|
{
|
|
"epoch": 1.7623604465709728,
|
|
"grad_norm": 0.4629656896157057,
|
|
"learning_rate": 3.7267242189281746e-05,
|
|
"loss": 0.2138,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21994957327842712,
|
|
"step": 1105,
|
|
"valid_targets_mean": 4140.4,
|
|
"valid_targets_min": 1083
|
|
},
|
|
{
|
|
"epoch": 1.7703349282296652,
|
|
"grad_norm": 0.49029157020426084,
|
|
"learning_rate": 3.722697411514305e-05,
|
|
"loss": 0.1988,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22220617532730103,
|
|
"step": 1110,
|
|
"valid_targets_mean": 4312.5,
|
|
"valid_targets_min": 1783
|
|
},
|
|
{
|
|
"epoch": 1.7783094098883572,
|
|
"grad_norm": 0.47257303125367206,
|
|
"learning_rate": 3.7186433611593225e-05,
|
|
"loss": 0.2166,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20860454440116882,
|
|
"step": 1115,
|
|
"valid_targets_mean": 4489.4,
|
|
"valid_targets_min": 1180
|
|
},
|
|
{
|
|
"epoch": 1.7862838915470496,
|
|
"grad_norm": 0.4881940956993811,
|
|
"learning_rate": 3.7145621319744614e-05,
|
|
"loss": 0.2088,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22178339958190918,
|
|
"step": 1120,
|
|
"valid_targets_mean": 4007.1,
|
|
"valid_targets_min": 1672
|
|
},
|
|
{
|
|
"epoch": 1.7942583732057416,
|
|
"grad_norm": 0.49017938696756835,
|
|
"learning_rate": 3.7104537885007635e-05,
|
|
"loss": 0.2365,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2537485361099243,
|
|
"step": 1125,
|
|
"valid_targets_mean": 4017.6,
|
|
"valid_targets_min": 1443
|
|
},
|
|
{
|
|
"epoch": 1.802232854864434,
|
|
"grad_norm": 0.4709602958627766,
|
|
"learning_rate": 3.7063183957080594e-05,
|
|
"loss": 0.209,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22469566762447357,
|
|
"step": 1130,
|
|
"valid_targets_mean": 4085.2,
|
|
"valid_targets_min": 793
|
|
},
|
|
{
|
|
"epoch": 1.810207336523126,
|
|
"grad_norm": 0.4928808140478705,
|
|
"learning_rate": 3.7021560189939416e-05,
|
|
"loss": 0.2256,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20846673846244812,
|
|
"step": 1135,
|
|
"valid_targets_mean": 3571.0,
|
|
"valid_targets_min": 1270
|
|
},
|
|
{
|
|
"epoch": 1.8181818181818183,
|
|
"grad_norm": 0.4479091986972175,
|
|
"learning_rate": 3.697966724182729e-05,
|
|
"loss": 0.2371,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25488775968551636,
|
|
"step": 1140,
|
|
"valid_targets_mean": 5469.8,
|
|
"valid_targets_min": 2538
|
|
},
|
|
{
|
|
"epoch": 1.8261562998405103,
|
|
"grad_norm": 0.5036402613565975,
|
|
"learning_rate": 3.6937505775244246e-05,
|
|
"loss": 0.2096,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22394904494285583,
|
|
"step": 1145,
|
|
"valid_targets_mean": 3642.4,
|
|
"valid_targets_min": 1295
|
|
},
|
|
{
|
|
"epoch": 1.8341307814992025,
|
|
"grad_norm": 0.4780135062253511,
|
|
"learning_rate": 3.689507645693674e-05,
|
|
"loss": 0.2287,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2610904574394226,
|
|
"step": 1150,
|
|
"valid_targets_mean": 4336.4,
|
|
"valid_targets_min": 1326
|
|
},
|
|
{
|
|
"epoch": 1.8421052631578947,
|
|
"grad_norm": 0.44381256880644715,
|
|
"learning_rate": 3.6852379957887025e-05,
|
|
"loss": 0.2159,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2414180189371109,
|
|
"step": 1155,
|
|
"valid_targets_mean": 4727.1,
|
|
"valid_targets_min": 1465
|
|
},
|
|
{
|
|
"epoch": 1.8500797448165869,
|
|
"grad_norm": 0.5083384797022545,
|
|
"learning_rate": 3.6809416953302606e-05,
|
|
"loss": 0.195,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19842639565467834,
|
|
"step": 1160,
|
|
"valid_targets_mean": 3549.2,
|
|
"valid_targets_min": 331
|
|
},
|
|
{
|
|
"epoch": 1.858054226475279,
|
|
"grad_norm": 0.49199658884829844,
|
|
"learning_rate": 3.676618812260553e-05,
|
|
"loss": 0.2152,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2503410577774048,
|
|
"step": 1165,
|
|
"valid_targets_mean": 5109.9,
|
|
"valid_targets_min": 2149
|
|
},
|
|
{
|
|
"epoch": 1.8660287081339713,
|
|
"grad_norm": 0.55008339033863,
|
|
"learning_rate": 3.672269414942166e-05,
|
|
"loss": 0.2141,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22753937542438507,
|
|
"step": 1170,
|
|
"valid_targets_mean": 4630.0,
|
|
"valid_targets_min": 976
|
|
},
|
|
{
|
|
"epoch": 1.8740031897926634,
|
|
"grad_norm": 0.4586261326556948,
|
|
"learning_rate": 3.6678935721569825e-05,
|
|
"loss": 0.2033,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18508616089820862,
|
|
"step": 1175,
|
|
"valid_targets_mean": 4012.5,
|
|
"valid_targets_min": 1306
|
|
},
|
|
{
|
|
"epoch": 1.8819776714513556,
|
|
"grad_norm": 0.5027511997863094,
|
|
"learning_rate": 3.663491353105101e-05,
|
|
"loss": 0.2077,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2094116061925888,
|
|
"step": 1180,
|
|
"valid_targets_mean": 4146.3,
|
|
"valid_targets_min": 1831
|
|
},
|
|
{
|
|
"epoch": 1.8899521531100478,
|
|
"grad_norm": 0.44258348252135243,
|
|
"learning_rate": 3.659062827403735e-05,
|
|
"loss": 0.2054,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21222475171089172,
|
|
"step": 1185,
|
|
"valid_targets_mean": 4740.8,
|
|
"valid_targets_min": 484
|
|
},
|
|
{
|
|
"epoch": 1.89792663476874,
|
|
"grad_norm": 0.4593536791251397,
|
|
"learning_rate": 3.654608065086115e-05,
|
|
"loss": 0.2114,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21094900369644165,
|
|
"step": 1190,
|
|
"valid_targets_mean": 4180.3,
|
|
"valid_targets_min": 1467
|
|
},
|
|
{
|
|
"epoch": 1.9059011164274322,
|
|
"grad_norm": 0.3864590115928493,
|
|
"learning_rate": 3.650127136600379e-05,
|
|
"loss": 0.2242,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2283993363380432,
|
|
"step": 1195,
|
|
"valid_targets_mean": 6030.2,
|
|
"valid_targets_min": 2216
|
|
},
|
|
{
|
|
"epoch": 1.9138755980861244,
|
|
"grad_norm": 0.4493029875433048,
|
|
"learning_rate": 3.645620112808464e-05,
|
|
"loss": 0.2184,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20390526950359344,
|
|
"step": 1200,
|
|
"valid_targets_mean": 4438.4,
|
|
"valid_targets_min": 1179
|
|
},
|
|
{
|
|
"epoch": 1.9218500797448166,
|
|
"grad_norm": 0.4717808228916783,
|
|
"learning_rate": 3.641087064984977e-05,
|
|
"loss": 0.2168,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.29960161447525024,
|
|
"step": 1205,
|
|
"valid_targets_mean": 5767.2,
|
|
"valid_targets_min": 2064
|
|
},
|
|
{
|
|
"epoch": 1.9298245614035088,
|
|
"grad_norm": 0.4408153072153271,
|
|
"learning_rate": 3.636528064816073e-05,
|
|
"loss": 0.2137,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21097023785114288,
|
|
"step": 1210,
|
|
"valid_targets_mean": 4657.0,
|
|
"valid_targets_min": 1295
|
|
},
|
|
{
|
|
"epoch": 1.937799043062201,
|
|
"grad_norm": 0.5308216721572256,
|
|
"learning_rate": 3.6319431843983223e-05,
|
|
"loss": 0.2116,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.26519426703453064,
|
|
"step": 1215,
|
|
"valid_targets_mean": 4505.1,
|
|
"valid_targets_min": 777
|
|
},
|
|
{
|
|
"epoch": 1.9457735247208932,
|
|
"grad_norm": 0.5036613470499272,
|
|
"learning_rate": 3.6273324962375676e-05,
|
|
"loss": 0.2173,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2110607922077179,
|
|
"step": 1220,
|
|
"valid_targets_mean": 3974.0,
|
|
"valid_targets_min": 1086
|
|
},
|
|
{
|
|
"epoch": 1.9537480063795853,
|
|
"grad_norm": 0.47289512484636687,
|
|
"learning_rate": 3.622696073247777e-05,
|
|
"loss": 0.203,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20814213156700134,
|
|
"step": 1225,
|
|
"valid_targets_mean": 4752.9,
|
|
"valid_targets_min": 445
|
|
},
|
|
{
|
|
"epoch": 1.9617224880382775,
|
|
"grad_norm": 0.49652731607830747,
|
|
"learning_rate": 3.6180339887498953e-05,
|
|
"loss": 0.2177,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2197228968143463,
|
|
"step": 1230,
|
|
"valid_targets_mean": 4165.1,
|
|
"valid_targets_min": 1840
|
|
},
|
|
{
|
|
"epoch": 1.9696969696969697,
|
|
"grad_norm": 0.4303512638873414,
|
|
"learning_rate": 3.613346316470678e-05,
|
|
"loss": 0.2097,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19560399651527405,
|
|
"step": 1235,
|
|
"valid_targets_mean": 4744.9,
|
|
"valid_targets_min": 1290
|
|
},
|
|
{
|
|
"epoch": 1.977671451355662,
|
|
"grad_norm": 0.46027425838155206,
|
|
"learning_rate": 3.60863313054153e-05,
|
|
"loss": 0.1967,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20572498440742493,
|
|
"step": 1240,
|
|
"valid_targets_mean": 4645.9,
|
|
"valid_targets_min": 1393
|
|
},
|
|
{
|
|
"epoch": 1.985645933014354,
|
|
"grad_norm": 0.4179304388168029,
|
|
"learning_rate": 3.6038945054973334e-05,
|
|
"loss": 0.2036,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18564409017562866,
|
|
"step": 1245,
|
|
"valid_targets_mean": 4841.8,
|
|
"valid_targets_min": 2470
|
|
},
|
|
{
|
|
"epoch": 1.9936204146730463,
|
|
"grad_norm": 0.429983595414097,
|
|
"learning_rate": 3.599130516275266e-05,
|
|
"loss": 0.2109,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17854493856430054,
|
|
"step": 1250,
|
|
"valid_targets_mean": 4375.5,
|
|
"valid_targets_min": 738
|
|
},
|
|
{
|
|
"epoch": 2.0015948963317385,
|
|
"grad_norm": 0.5835831287236563,
|
|
"learning_rate": 3.594341238213618e-05,
|
|
"loss": 0.2158,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2125822901725769,
|
|
"step": 1255,
|
|
"valid_targets_mean": 4207.8,
|
|
"valid_targets_min": 1622
|
|
},
|
|
{
|
|
"epoch": 2.0095693779904304,
|
|
"grad_norm": 0.48535029081671693,
|
|
"learning_rate": 3.589526747050601e-05,
|
|
"loss": 0.1937,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18778109550476074,
|
|
"step": 1260,
|
|
"valid_targets_mean": 4847.1,
|
|
"valid_targets_min": 1503
|
|
},
|
|
{
|
|
"epoch": 2.017543859649123,
|
|
"grad_norm": 0.4142330646373312,
|
|
"learning_rate": 3.584687118923149e-05,
|
|
"loss": 0.1803,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16331608593463898,
|
|
"step": 1265,
|
|
"valid_targets_mean": 5046.7,
|
|
"valid_targets_min": 2438
|
|
},
|
|
{
|
|
"epoch": 2.025518341307815,
|
|
"grad_norm": 0.5387717755970093,
|
|
"learning_rate": 3.579822430365714e-05,
|
|
"loss": 0.1921,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19953720271587372,
|
|
"step": 1270,
|
|
"valid_targets_mean": 3987.6,
|
|
"valid_targets_min": 1561
|
|
},
|
|
{
|
|
"epoch": 2.0334928229665072,
|
|
"grad_norm": 0.492124600517885,
|
|
"learning_rate": 3.57493275830906e-05,
|
|
"loss": 0.2113,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2031257301568985,
|
|
"step": 1275,
|
|
"valid_targets_mean": 4217.2,
|
|
"valid_targets_min": 1199
|
|
},
|
|
{
|
|
"epoch": 2.041467304625199,
|
|
"grad_norm": 0.49992210755448996,
|
|
"learning_rate": 3.570018180079037e-05,
|
|
"loss": 0.2059,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1987209916114807,
|
|
"step": 1280,
|
|
"valid_targets_mean": 4681.9,
|
|
"valid_targets_min": 1519
|
|
},
|
|
{
|
|
"epoch": 2.0494417862838916,
|
|
"grad_norm": 0.5505016254551602,
|
|
"learning_rate": 3.5650787733953715e-05,
|
|
"loss": 0.206,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19336707890033722,
|
|
"step": 1285,
|
|
"valid_targets_mean": 4355.0,
|
|
"valid_targets_min": 1653
|
|
},
|
|
{
|
|
"epoch": 2.0574162679425836,
|
|
"grad_norm": 0.579774349638529,
|
|
"learning_rate": 3.560114616370425e-05,
|
|
"loss": 0.1924,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20641644299030304,
|
|
"step": 1290,
|
|
"valid_targets_mean": 4079.1,
|
|
"valid_targets_min": 583
|
|
},
|
|
{
|
|
"epoch": 2.065390749601276,
|
|
"grad_norm": 0.4389607585870106,
|
|
"learning_rate": 3.555125787507964e-05,
|
|
"loss": 0.1832,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18444427847862244,
|
|
"step": 1295,
|
|
"valid_targets_mean": 5314.6,
|
|
"valid_targets_min": 1254
|
|
},
|
|
{
|
|
"epoch": 2.073365231259968,
|
|
"grad_norm": 0.4893113319926811,
|
|
"learning_rate": 3.550112365701921e-05,
|
|
"loss": 0.208,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22254320979118347,
|
|
"step": 1300,
|
|
"valid_targets_mean": 4483.9,
|
|
"valid_targets_min": 1227
|
|
},
|
|
{
|
|
"epoch": 2.0813397129186604,
|
|
"grad_norm": 0.5537767932697284,
|
|
"learning_rate": 3.545074430235142e-05,
|
|
"loss": 0.1835,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18424861133098602,
|
|
"step": 1305,
|
|
"valid_targets_mean": 4432.2,
|
|
"valid_targets_min": 1619
|
|
},
|
|
{
|
|
"epoch": 2.0893141945773523,
|
|
"grad_norm": 0.5182340294860415,
|
|
"learning_rate": 3.540012060778137e-05,
|
|
"loss": 0.1891,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17916396260261536,
|
|
"step": 1310,
|
|
"valid_targets_mean": 4455.2,
|
|
"valid_targets_min": 1876
|
|
},
|
|
{
|
|
"epoch": 2.0972886762360448,
|
|
"grad_norm": 0.4906965964718632,
|
|
"learning_rate": 3.534925337387816e-05,
|
|
"loss": 0.1852,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17498987913131714,
|
|
"step": 1315,
|
|
"valid_targets_mean": 4435.5,
|
|
"valid_targets_min": 1179
|
|
},
|
|
{
|
|
"epoch": 2.1052631578947367,
|
|
"grad_norm": 0.5227104218173729,
|
|
"learning_rate": 3.529814340506226e-05,
|
|
"loss": 0.1992,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19775143265724182,
|
|
"step": 1320,
|
|
"valid_targets_mean": 4533.5,
|
|
"valid_targets_min": 1604
|
|
},
|
|
{
|
|
"epoch": 2.113237639553429,
|
|
"grad_norm": 0.6246358255489487,
|
|
"learning_rate": 3.524679150959277e-05,
|
|
"loss": 0.1844,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18220943212509155,
|
|
"step": 1325,
|
|
"valid_targets_mean": 4121.6,
|
|
"valid_targets_min": 1517
|
|
},
|
|
{
|
|
"epoch": 2.121212121212121,
|
|
"grad_norm": 0.4427948849698209,
|
|
"learning_rate": 3.519519849955466e-05,
|
|
"loss": 0.1887,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17299795150756836,
|
|
"step": 1330,
|
|
"valid_targets_mean": 4151.1,
|
|
"valid_targets_min": 1403
|
|
},
|
|
{
|
|
"epoch": 2.1291866028708135,
|
|
"grad_norm": 0.45665193547767974,
|
|
"learning_rate": 3.514336519084591e-05,
|
|
"loss": 0.1866,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17760662734508514,
|
|
"step": 1335,
|
|
"valid_targets_mean": 5019.9,
|
|
"valid_targets_min": 1252
|
|
},
|
|
{
|
|
"epoch": 2.1371610845295055,
|
|
"grad_norm": 0.5111255448822826,
|
|
"learning_rate": 3.509129240316461e-05,
|
|
"loss": 0.1878,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18380534648895264,
|
|
"step": 1340,
|
|
"valid_targets_mean": 3753.7,
|
|
"valid_targets_min": 1912
|
|
},
|
|
{
|
|
"epoch": 2.145135566188198,
|
|
"grad_norm": 0.4888821553823615,
|
|
"learning_rate": 3.5038980959995985e-05,
|
|
"loss": 0.1939,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18884651362895966,
|
|
"step": 1345,
|
|
"valid_targets_mean": 4299.5,
|
|
"valid_targets_min": 524
|
|
},
|
|
{
|
|
"epoch": 2.15311004784689,
|
|
"grad_norm": 0.5876826648011623,
|
|
"learning_rate": 3.498643168859941e-05,
|
|
"loss": 0.1986,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21021640300750732,
|
|
"step": 1350,
|
|
"valid_targets_mean": 4070.5,
|
|
"valid_targets_min": 1091
|
|
},
|
|
{
|
|
"epoch": 2.1610845295055823,
|
|
"grad_norm": 0.4433617225762594,
|
|
"learning_rate": 3.493364541999529e-05,
|
|
"loss": 0.179,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15240584313869476,
|
|
"step": 1355,
|
|
"valid_targets_mean": 3732.5,
|
|
"valid_targets_min": 1223
|
|
},
|
|
{
|
|
"epoch": 2.1690590111642742,
|
|
"grad_norm": 0.4999003849230528,
|
|
"learning_rate": 3.488062298895194e-05,
|
|
"loss": 0.1967,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22450381517410278,
|
|
"step": 1360,
|
|
"valid_targets_mean": 5037.6,
|
|
"valid_targets_min": 1831
|
|
},
|
|
{
|
|
"epoch": 2.1770334928229667,
|
|
"grad_norm": 0.5515763140814884,
|
|
"learning_rate": 3.482736523397237e-05,
|
|
"loss": 0.1819,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17379963397979736,
|
|
"step": 1365,
|
|
"valid_targets_mean": 4321.4,
|
|
"valid_targets_min": 1374
|
|
},
|
|
{
|
|
"epoch": 2.1850079744816586,
|
|
"grad_norm": 0.5140961821958343,
|
|
"learning_rate": 3.4773872997281026e-05,
|
|
"loss": 0.2001,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21524573862552643,
|
|
"step": 1370,
|
|
"valid_targets_mean": 4404.0,
|
|
"valid_targets_min": 1576
|
|
},
|
|
{
|
|
"epoch": 2.192982456140351,
|
|
"grad_norm": 0.5277729139359486,
|
|
"learning_rate": 3.472014712481048e-05,
|
|
"loss": 0.1926,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22807437181472778,
|
|
"step": 1375,
|
|
"valid_targets_mean": 4321.2,
|
|
"valid_targets_min": 1846
|
|
},
|
|
{
|
|
"epoch": 2.200956937799043,
|
|
"grad_norm": 0.5031333134053864,
|
|
"learning_rate": 3.466618846618806e-05,
|
|
"loss": 0.2122,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22107753157615662,
|
|
"step": 1380,
|
|
"valid_targets_mean": 4169.7,
|
|
"valid_targets_min": 1166
|
|
},
|
|
{
|
|
"epoch": 2.2089314194577354,
|
|
"grad_norm": 0.483311395201815,
|
|
"learning_rate": 3.461199787472238e-05,
|
|
"loss": 0.2041,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19544145464897156,
|
|
"step": 1385,
|
|
"valid_targets_mean": 3844.2,
|
|
"valid_targets_min": 744
|
|
},
|
|
{
|
|
"epoch": 2.2169059011164274,
|
|
"grad_norm": 0.4498139497860062,
|
|
"learning_rate": 3.455757620738989e-05,
|
|
"loss": 0.2111,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21751423180103302,
|
|
"step": 1390,
|
|
"valid_targets_mean": 4871.2,
|
|
"valid_targets_min": 377
|
|
},
|
|
{
|
|
"epoch": 2.22488038277512,
|
|
"grad_norm": 0.46372087812522933,
|
|
"learning_rate": 3.450292432482127e-05,
|
|
"loss": 0.2076,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20186303555965424,
|
|
"step": 1395,
|
|
"valid_targets_mean": 4219.1,
|
|
"valid_targets_min": 1451
|
|
},
|
|
{
|
|
"epoch": 2.2328548644338118,
|
|
"grad_norm": 0.4234330431162755,
|
|
"learning_rate": 3.444804309128789e-05,
|
|
"loss": 0.1932,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16204340755939484,
|
|
"step": 1400,
|
|
"valid_targets_mean": 5115.7,
|
|
"valid_targets_min": 2142
|
|
},
|
|
{
|
|
"epoch": 2.240829346092504,
|
|
"grad_norm": 0.46729513282292884,
|
|
"learning_rate": 3.439293337468808e-05,
|
|
"loss": 0.1936,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1783636510372162,
|
|
"step": 1405,
|
|
"valid_targets_mean": 4136.9,
|
|
"valid_targets_min": 1653
|
|
},
|
|
{
|
|
"epoch": 2.248803827751196,
|
|
"grad_norm": 0.5256776358531895,
|
|
"learning_rate": 3.4337596046533426e-05,
|
|
"loss": 0.1963,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2295759618282318,
|
|
"step": 1410,
|
|
"valid_targets_mean": 4567.9,
|
|
"valid_targets_min": 1222
|
|
},
|
|
{
|
|
"epoch": 2.2567783094098885,
|
|
"grad_norm": 0.4138467943971567,
|
|
"learning_rate": 3.4282031981935e-05,
|
|
"loss": 0.1846,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16236048936843872,
|
|
"step": 1415,
|
|
"valid_targets_mean": 4935.4,
|
|
"valid_targets_min": 1512
|
|
},
|
|
{
|
|
"epoch": 2.2647527910685805,
|
|
"grad_norm": 0.5052303861398547,
|
|
"learning_rate": 3.42262420595895e-05,
|
|
"loss": 0.1887,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21595962345600128,
|
|
"step": 1420,
|
|
"valid_targets_mean": 4676.1,
|
|
"valid_targets_min": 462
|
|
},
|
|
{
|
|
"epoch": 2.2727272727272725,
|
|
"grad_norm": 0.42027522309214377,
|
|
"learning_rate": 3.417022716176539e-05,
|
|
"loss": 0.1944,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21920445561408997,
|
|
"step": 1425,
|
|
"valid_targets_mean": 5443.5,
|
|
"valid_targets_min": 818
|
|
},
|
|
{
|
|
"epoch": 2.280701754385965,
|
|
"grad_norm": 0.42271888607035824,
|
|
"learning_rate": 3.411398817428889e-05,
|
|
"loss": 0.2142,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21368569135665894,
|
|
"step": 1430,
|
|
"valid_targets_mean": 5346.8,
|
|
"valid_targets_min": 2060
|
|
},
|
|
{
|
|
"epoch": 2.2886762360446573,
|
|
"grad_norm": 0.46762819973280395,
|
|
"learning_rate": 3.4057525986530016e-05,
|
|
"loss": 0.195,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21968646347522736,
|
|
"step": 1435,
|
|
"valid_targets_mean": 4741.2,
|
|
"valid_targets_min": 331
|
|
},
|
|
{
|
|
"epoch": 2.2966507177033493,
|
|
"grad_norm": 0.47510163265424776,
|
|
"learning_rate": 3.400084149138851e-05,
|
|
"loss": 0.1866,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19467958807945251,
|
|
"step": 1440,
|
|
"valid_targets_mean": 4569.6,
|
|
"valid_targets_min": 1837
|
|
},
|
|
{
|
|
"epoch": 2.3046251993620412,
|
|
"grad_norm": 0.4906053699523343,
|
|
"learning_rate": 3.394393558527969e-05,
|
|
"loss": 0.2135,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21481776237487793,
|
|
"step": 1445,
|
|
"valid_targets_mean": 4330.7,
|
|
"valid_targets_min": 1908
|
|
},
|
|
{
|
|
"epoch": 2.3125996810207337,
|
|
"grad_norm": 0.5017709902553857,
|
|
"learning_rate": 3.388680916812031e-05,
|
|
"loss": 0.1935,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20141157507896423,
|
|
"step": 1450,
|
|
"valid_targets_mean": 4137.5,
|
|
"valid_targets_min": 1589
|
|
},
|
|
{
|
|
"epoch": 2.320574162679426,
|
|
"grad_norm": 0.45313626296858717,
|
|
"learning_rate": 3.382946314331429e-05,
|
|
"loss": 0.195,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1925966739654541,
|
|
"step": 1455,
|
|
"valid_targets_mean": 4450.9,
|
|
"valid_targets_min": 397
|
|
},
|
|
{
|
|
"epoch": 2.328548644338118,
|
|
"grad_norm": 0.6495234625151266,
|
|
"learning_rate": 3.377189841773848e-05,
|
|
"loss": 0.1986,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20314140617847443,
|
|
"step": 1460,
|
|
"valid_targets_mean": 4231.6,
|
|
"valid_targets_min": 1037
|
|
},
|
|
{
|
|
"epoch": 2.33652312599681,
|
|
"grad_norm": 0.5436303133229426,
|
|
"learning_rate": 3.371411590172827e-05,
|
|
"loss": 0.1885,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20022675395011902,
|
|
"step": 1465,
|
|
"valid_targets_mean": 4746.3,
|
|
"valid_targets_min": 1710
|
|
},
|
|
{
|
|
"epoch": 2.3444976076555024,
|
|
"grad_norm": 0.4712354546705832,
|
|
"learning_rate": 3.365611650906321e-05,
|
|
"loss": 0.1771,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15993911027908325,
|
|
"step": 1470,
|
|
"valid_targets_mean": 3833.4,
|
|
"valid_targets_min": 1170
|
|
},
|
|
{
|
|
"epoch": 2.352472089314195,
|
|
"grad_norm": 0.4505288458448758,
|
|
"learning_rate": 3.359790115695259e-05,
|
|
"loss": 0.1888,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21294520795345306,
|
|
"step": 1475,
|
|
"valid_targets_mean": 5357.1,
|
|
"valid_targets_min": 1479
|
|
},
|
|
{
|
|
"epoch": 2.360446570972887,
|
|
"grad_norm": 0.5025939693327697,
|
|
"learning_rate": 3.353947076602088e-05,
|
|
"loss": 0.1866,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21563753485679626,
|
|
"step": 1480,
|
|
"valid_targets_mean": 4241.7,
|
|
"valid_targets_min": 858
|
|
},
|
|
{
|
|
"epoch": 2.3684210526315788,
|
|
"grad_norm": 0.4676350895459756,
|
|
"learning_rate": 3.34808262602932e-05,
|
|
"loss": 0.2007,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2031913697719574,
|
|
"step": 1485,
|
|
"valid_targets_mean": 4083.8,
|
|
"valid_targets_min": 884
|
|
},
|
|
{
|
|
"epoch": 2.376395534290271,
|
|
"grad_norm": 0.4563739844772263,
|
|
"learning_rate": 3.342196856718074e-05,
|
|
"loss": 0.1911,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1997055560350418,
|
|
"step": 1490,
|
|
"valid_targets_mean": 4307.0,
|
|
"valid_targets_min": 1402
|
|
},
|
|
{
|
|
"epoch": 2.384370015948963,
|
|
"grad_norm": 0.43749184940511027,
|
|
"learning_rate": 3.336289861746602e-05,
|
|
"loss": 0.19,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18445178866386414,
|
|
"step": 1495,
|
|
"valid_targets_mean": 4663.4,
|
|
"valid_targets_min": 1569
|
|
},
|
|
{
|
|
"epoch": 2.3923444976076556,
|
|
"grad_norm": 0.45977508895019903,
|
|
"learning_rate": 3.330361734528823e-05,
|
|
"loss": 0.1945,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1979564130306244,
|
|
"step": 1500,
|
|
"valid_targets_mean": 4822.2,
|
|
"valid_targets_min": 1832
|
|
},
|
|
{
|
|
"epoch": 2.4003189792663475,
|
|
"grad_norm": 0.5414838734249728,
|
|
"learning_rate": 3.324412568812844e-05,
|
|
"loss": 0.1878,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19212666153907776,
|
|
"step": 1505,
|
|
"valid_targets_mean": 3750.8,
|
|
"valid_targets_min": 777
|
|
},
|
|
{
|
|
"epoch": 2.40829346092504,
|
|
"grad_norm": 0.4358025730158986,
|
|
"learning_rate": 3.318442458679477e-05,
|
|
"loss": 0.18,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1796949803829193,
|
|
"step": 1510,
|
|
"valid_targets_mean": 4733.1,
|
|
"valid_targets_min": 1462
|
|
},
|
|
{
|
|
"epoch": 2.416267942583732,
|
|
"grad_norm": 0.5789725852587035,
|
|
"learning_rate": 3.312451498540751e-05,
|
|
"loss": 0.2148,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22660470008850098,
|
|
"step": 1515,
|
|
"valid_targets_mean": 3926.1,
|
|
"valid_targets_min": 1148
|
|
},
|
|
{
|
|
"epoch": 2.4242424242424243,
|
|
"grad_norm": 0.4810092138228401,
|
|
"learning_rate": 3.306439783138421e-05,
|
|
"loss": 0.2056,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1583409607410431,
|
|
"step": 1520,
|
|
"valid_targets_mean": 3954.2,
|
|
"valid_targets_min": 2144
|
|
},
|
|
{
|
|
"epoch": 2.4322169059011163,
|
|
"grad_norm": 0.480939376721317,
|
|
"learning_rate": 3.3004074075424666e-05,
|
|
"loss": 0.1945,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19095252454280853,
|
|
"step": 1525,
|
|
"valid_targets_mean": 4131.1,
|
|
"valid_targets_min": 636
|
|
},
|
|
{
|
|
"epoch": 2.4401913875598087,
|
|
"grad_norm": 0.44164348513325546,
|
|
"learning_rate": 3.29435446714959e-05,
|
|
"loss": 0.2042,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18615323305130005,
|
|
"step": 1530,
|
|
"valid_targets_mean": 4125.1,
|
|
"valid_targets_min": 909
|
|
},
|
|
{
|
|
"epoch": 2.4481658692185007,
|
|
"grad_norm": 0.49332778067002697,
|
|
"learning_rate": 3.288281057681709e-05,
|
|
"loss": 0.2005,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23022127151489258,
|
|
"step": 1535,
|
|
"valid_targets_mean": 4346.1,
|
|
"valid_targets_min": 1011
|
|
},
|
|
{
|
|
"epoch": 2.456140350877193,
|
|
"grad_norm": 0.5217870366629661,
|
|
"learning_rate": 3.28218727518444e-05,
|
|
"loss": 0.2103,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19895359873771667,
|
|
"step": 1540,
|
|
"valid_targets_mean": 3616.3,
|
|
"valid_targets_min": 1063
|
|
},
|
|
{
|
|
"epoch": 2.464114832535885,
|
|
"grad_norm": 0.44722789365078947,
|
|
"learning_rate": 3.2760732160255835e-05,
|
|
"loss": 0.2128,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16717740893363953,
|
|
"step": 1545,
|
|
"valid_targets_mean": 4226.1,
|
|
"valid_targets_min": 951
|
|
},
|
|
{
|
|
"epoch": 2.4720893141945774,
|
|
"grad_norm": 0.41737884025495914,
|
|
"learning_rate": 3.269938976893595e-05,
|
|
"loss": 0.1834,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2132222056388855,
|
|
"step": 1550,
|
|
"valid_targets_mean": 5227.2,
|
|
"valid_targets_min": 2208
|
|
},
|
|
{
|
|
"epoch": 2.4800637958532694,
|
|
"grad_norm": 0.4677786411136909,
|
|
"learning_rate": 3.2637846547960596e-05,
|
|
"loss": 0.1768,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19589948654174805,
|
|
"step": 1555,
|
|
"valid_targets_mean": 4841.1,
|
|
"valid_targets_min": 1762
|
|
},
|
|
{
|
|
"epoch": 2.488038277511962,
|
|
"grad_norm": 0.4965944170108996,
|
|
"learning_rate": 3.2576103470581564e-05,
|
|
"loss": 0.1885,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17509344220161438,
|
|
"step": 1560,
|
|
"valid_targets_mean": 3434.1,
|
|
"valid_targets_min": 629
|
|
},
|
|
{
|
|
"epoch": 2.496012759170654,
|
|
"grad_norm": 0.492974160023232,
|
|
"learning_rate": 3.25141615132112e-05,
|
|
"loss": 0.1854,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15599842369556427,
|
|
"step": 1565,
|
|
"valid_targets_mean": 4451.3,
|
|
"valid_targets_min": 1194
|
|
},
|
|
{
|
|
"epoch": 2.503987240829346,
|
|
"grad_norm": 0.42642356808242493,
|
|
"learning_rate": 3.245202165540697e-05,
|
|
"loss": 0.1799,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16409683227539062,
|
|
"step": 1570,
|
|
"valid_targets_mean": 4488.6,
|
|
"valid_targets_min": 1827
|
|
},
|
|
{
|
|
"epoch": 2.511961722488038,
|
|
"grad_norm": 0.4616220732618904,
|
|
"learning_rate": 3.238968487985594e-05,
|
|
"loss": 0.1907,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16624875366687775,
|
|
"step": 1575,
|
|
"valid_targets_mean": 3498.1,
|
|
"valid_targets_min": 1233
|
|
},
|
|
{
|
|
"epoch": 2.5199362041467306,
|
|
"grad_norm": 0.45820206047868534,
|
|
"learning_rate": 3.232715217235927e-05,
|
|
"loss": 0.1961,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1971983015537262,
|
|
"step": 1580,
|
|
"valid_targets_mean": 4863.6,
|
|
"valid_targets_min": 385
|
|
},
|
|
{
|
|
"epoch": 2.5279106858054226,
|
|
"grad_norm": 0.42279033871539334,
|
|
"learning_rate": 3.226442452181662e-05,
|
|
"loss": 0.1797,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19996625185012817,
|
|
"step": 1585,
|
|
"valid_targets_mean": 5554.1,
|
|
"valid_targets_min": 2678
|
|
},
|
|
{
|
|
"epoch": 2.535885167464115,
|
|
"grad_norm": 0.46417011052422286,
|
|
"learning_rate": 3.220150292021049e-05,
|
|
"loss": 0.2052,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19248399138450623,
|
|
"step": 1590,
|
|
"valid_targets_mean": 4302.6,
|
|
"valid_targets_min": 1647
|
|
},
|
|
{
|
|
"epoch": 2.543859649122807,
|
|
"grad_norm": 0.42195835961915845,
|
|
"learning_rate": 3.213838836259055e-05,
|
|
"loss": 0.1946,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1905175745487213,
|
|
"step": 1595,
|
|
"valid_targets_mean": 5845.2,
|
|
"valid_targets_min": 2714
|
|
},
|
|
{
|
|
"epoch": 2.5518341307814993,
|
|
"grad_norm": 0.41169073814710955,
|
|
"learning_rate": 3.2075081847057886e-05,
|
|
"loss": 0.2065,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2413644641637802,
|
|
"step": 1600,
|
|
"valid_targets_mean": 6022.2,
|
|
"valid_targets_min": 2591
|
|
},
|
|
{
|
|
"epoch": 2.5598086124401913,
|
|
"grad_norm": 0.42143967615716676,
|
|
"learning_rate": 3.201158437474925e-05,
|
|
"loss": 0.1943,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17829415202140808,
|
|
"step": 1605,
|
|
"valid_targets_mean": 4504.9,
|
|
"valid_targets_min": 2008
|
|
},
|
|
{
|
|
"epoch": 2.5677830940988837,
|
|
"grad_norm": 0.4019639267134053,
|
|
"learning_rate": 3.194789694982119e-05,
|
|
"loss": 0.1808,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19469118118286133,
|
|
"step": 1610,
|
|
"valid_targets_mean": 4991.6,
|
|
"valid_targets_min": 2037
|
|
},
|
|
{
|
|
"epoch": 2.5757575757575757,
|
|
"grad_norm": 0.4967706504496925,
|
|
"learning_rate": 3.1884020579434216e-05,
|
|
"loss": 0.19,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18184614181518555,
|
|
"step": 1615,
|
|
"valid_targets_mean": 3664.2,
|
|
"valid_targets_min": 989
|
|
},
|
|
{
|
|
"epoch": 2.583732057416268,
|
|
"grad_norm": 0.40679294637064134,
|
|
"learning_rate": 3.181995627373679e-05,
|
|
"loss": 0.1844,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17015117406845093,
|
|
"step": 1620,
|
|
"valid_targets_mean": 4571.1,
|
|
"valid_targets_min": 529
|
|
},
|
|
{
|
|
"epoch": 2.59170653907496,
|
|
"grad_norm": 0.4584533151540396,
|
|
"learning_rate": 3.1755705045849465e-05,
|
|
"loss": 0.1897,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17102062702178955,
|
|
"step": 1625,
|
|
"valid_targets_mean": 4546.4,
|
|
"valid_targets_min": 1534
|
|
},
|
|
{
|
|
"epoch": 2.5996810207336525,
|
|
"grad_norm": 0.4353991390783487,
|
|
"learning_rate": 3.1691267911848765e-05,
|
|
"loss": 0.1938,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20528733730316162,
|
|
"step": 1630,
|
|
"valid_targets_mean": 4818.4,
|
|
"valid_targets_min": 2380
|
|
},
|
|
{
|
|
"epoch": 2.6076555023923444,
|
|
"grad_norm": 0.4725264329143068,
|
|
"learning_rate": 3.1626645890751167e-05,
|
|
"loss": 0.1972,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23138327896595,
|
|
"step": 1635,
|
|
"valid_targets_mean": 4452.9,
|
|
"valid_targets_min": 1687
|
|
},
|
|
{
|
|
"epoch": 2.6156299840510364,
|
|
"grad_norm": 0.7190616362034168,
|
|
"learning_rate": 3.156184000449697e-05,
|
|
"loss": 0.1889,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2099694162607193,
|
|
"step": 1640,
|
|
"valid_targets_mean": 4351.5,
|
|
"valid_targets_min": 792
|
|
},
|
|
{
|
|
"epoch": 2.623604465709729,
|
|
"grad_norm": 0.4262432921313368,
|
|
"learning_rate": 3.149685127793415e-05,
|
|
"loss": 0.2084,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2072952389717102,
|
|
"step": 1645,
|
|
"valid_targets_mean": 5261.7,
|
|
"valid_targets_min": 1564
|
|
},
|
|
{
|
|
"epoch": 2.6315789473684212,
|
|
"grad_norm": 0.44663659913615,
|
|
"learning_rate": 3.143168073880214e-05,
|
|
"loss": 0.1839,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20910018682479858,
|
|
"step": 1650,
|
|
"valid_targets_mean": 4689.8,
|
|
"valid_targets_min": 1977
|
|
},
|
|
{
|
|
"epoch": 2.639553429027113,
|
|
"grad_norm": 0.47855265259414465,
|
|
"learning_rate": 3.1366329417715556e-05,
|
|
"loss": 0.1859,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17901211977005005,
|
|
"step": 1655,
|
|
"valid_targets_mean": 3450.9,
|
|
"valid_targets_min": 764
|
|
},
|
|
{
|
|
"epoch": 2.647527910685805,
|
|
"grad_norm": 0.42030685495576364,
|
|
"learning_rate": 3.1300798348147954e-05,
|
|
"loss": 0.187,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17088496685028076,
|
|
"step": 1660,
|
|
"valid_targets_mean": 4424.1,
|
|
"valid_targets_min": 685
|
|
},
|
|
{
|
|
"epoch": 2.6555023923444976,
|
|
"grad_norm": 0.424205274613458,
|
|
"learning_rate": 3.123508856641542e-05,
|
|
"loss": 0.178,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20387335121631622,
|
|
"step": 1665,
|
|
"valid_targets_mean": 4416.8,
|
|
"valid_targets_min": 641
|
|
},
|
|
{
|
|
"epoch": 2.66347687400319,
|
|
"grad_norm": 0.6446111039179884,
|
|
"learning_rate": 3.116920111166025e-05,
|
|
"loss": 0.1893,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18233443796634674,
|
|
"step": 1670,
|
|
"valid_targets_mean": 3812.8,
|
|
"valid_targets_min": 1710
|
|
},
|
|
{
|
|
"epoch": 2.671451355661882,
|
|
"grad_norm": 0.4016792889917327,
|
|
"learning_rate": 3.1103137025834456e-05,
|
|
"loss": 0.1921,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17425809800624847,
|
|
"step": 1675,
|
|
"valid_targets_mean": 5096.6,
|
|
"valid_targets_min": 1249
|
|
},
|
|
{
|
|
"epoch": 2.679425837320574,
|
|
"grad_norm": 0.4651781503641287,
|
|
"learning_rate": 3.103689735368333e-05,
|
|
"loss": 0.2157,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1736563742160797,
|
|
"step": 1680,
|
|
"valid_targets_mean": 4534.1,
|
|
"valid_targets_min": 1866
|
|
},
|
|
{
|
|
"epoch": 2.6874003189792663,
|
|
"grad_norm": 0.5822658951368973,
|
|
"learning_rate": 3.097048314272889e-05,
|
|
"loss": 0.2072,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.25176823139190674,
|
|
"step": 1685,
|
|
"valid_targets_mean": 3692.4,
|
|
"valid_targets_min": 904
|
|
},
|
|
{
|
|
"epoch": 2.6953748006379588,
|
|
"grad_norm": 0.49469625871209677,
|
|
"learning_rate": 3.090389544325335e-05,
|
|
"loss": 0.19,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20558789372444153,
|
|
"step": 1690,
|
|
"valid_targets_mean": 3640.8,
|
|
"valid_targets_min": 683
|
|
},
|
|
{
|
|
"epoch": 2.7033492822966507,
|
|
"grad_norm": 0.6014487435674147,
|
|
"learning_rate": 3.08371353082825e-05,
|
|
"loss": 0.1866,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18729646503925323,
|
|
"step": 1695,
|
|
"valid_targets_mean": 3934.1,
|
|
"valid_targets_min": 1318
|
|
},
|
|
{
|
|
"epoch": 2.7113237639553427,
|
|
"grad_norm": 0.4723809602275162,
|
|
"learning_rate": 3.0770203793568994e-05,
|
|
"loss": 0.1866,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1865779161453247,
|
|
"step": 1700,
|
|
"valid_targets_mean": 4184.6,
|
|
"valid_targets_min": 1924
|
|
},
|
|
{
|
|
"epoch": 2.719298245614035,
|
|
"grad_norm": 0.4575136224294941,
|
|
"learning_rate": 3.0703101957575765e-05,
|
|
"loss": 0.188,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18605884909629822,
|
|
"step": 1705,
|
|
"valid_targets_mean": 4517.9,
|
|
"valid_targets_min": 1272
|
|
},
|
|
{
|
|
"epoch": 2.7272727272727275,
|
|
"grad_norm": 0.44365096011449917,
|
|
"learning_rate": 3.0635830861459204e-05,
|
|
"loss": 0.201,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.211789071559906,
|
|
"step": 1710,
|
|
"valid_targets_mean": 5082.9,
|
|
"valid_targets_min": 667
|
|
},
|
|
{
|
|
"epoch": 2.7352472089314195,
|
|
"grad_norm": 0.46945868579564015,
|
|
"learning_rate": 3.056839156905239e-05,
|
|
"loss": 0.194,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17882491648197174,
|
|
"step": 1715,
|
|
"valid_targets_mean": 4237.8,
|
|
"valid_targets_min": 1611
|
|
},
|
|
{
|
|
"epoch": 2.7432216905901115,
|
|
"grad_norm": 0.44204098954171,
|
|
"learning_rate": 3.0500785146848303e-05,
|
|
"loss": 0.1905,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20405025780200958,
|
|
"step": 1720,
|
|
"valid_targets_mean": 5329.2,
|
|
"valid_targets_min": 1796
|
|
},
|
|
{
|
|
"epoch": 2.751196172248804,
|
|
"grad_norm": 0.46803832657642275,
|
|
"learning_rate": 3.04330126639829e-05,
|
|
"loss": 0.1801,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.188456192612648,
|
|
"step": 1725,
|
|
"valid_targets_mean": 3660.6,
|
|
"valid_targets_min": 1421
|
|
},
|
|
{
|
|
"epoch": 2.7591706539074963,
|
|
"grad_norm": 0.46909903458430224,
|
|
"learning_rate": 3.0365075192218278e-05,
|
|
"loss": 0.1912,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20096969604492188,
|
|
"step": 1730,
|
|
"valid_targets_mean": 3955.4,
|
|
"valid_targets_min": 1782
|
|
},
|
|
{
|
|
"epoch": 2.7671451355661882,
|
|
"grad_norm": 0.4540996711308176,
|
|
"learning_rate": 3.0296973805925663e-05,
|
|
"loss": 0.1817,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15702204406261444,
|
|
"step": 1735,
|
|
"valid_targets_mean": 3424.9,
|
|
"valid_targets_min": 996
|
|
},
|
|
{
|
|
"epoch": 2.77511961722488,
|
|
"grad_norm": 0.43447333950282657,
|
|
"learning_rate": 3.022870958206845e-05,
|
|
"loss": 0.1854,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17967906594276428,
|
|
"step": 1740,
|
|
"valid_targets_mean": 4727.9,
|
|
"valid_targets_min": 1797
|
|
},
|
|
{
|
|
"epoch": 2.7830940988835726,
|
|
"grad_norm": 0.534010470635787,
|
|
"learning_rate": 3.0160283600185168e-05,
|
|
"loss": 0.1955,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16738936305046082,
|
|
"step": 1745,
|
|
"valid_targets_mean": 3008.1,
|
|
"valid_targets_min": 1061
|
|
},
|
|
{
|
|
"epoch": 2.7910685805422646,
|
|
"grad_norm": 0.4657128578979155,
|
|
"learning_rate": 3.0091696942372412e-05,
|
|
"loss": 0.2044,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2012743055820465,
|
|
"step": 1750,
|
|
"valid_targets_mean": 4893.4,
|
|
"valid_targets_min": 486
|
|
},
|
|
{
|
|
"epoch": 2.799043062200957,
|
|
"grad_norm": 0.4392915887871567,
|
|
"learning_rate": 3.002295069326772e-05,
|
|
"loss": 0.1945,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16907723248004913,
|
|
"step": 1755,
|
|
"valid_targets_mean": 4969.5,
|
|
"valid_targets_min": 1327
|
|
},
|
|
{
|
|
"epoch": 2.807017543859649,
|
|
"grad_norm": 0.4982700232598375,
|
|
"learning_rate": 2.9954045940032423e-05,
|
|
"loss": 0.1964,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21579095721244812,
|
|
"step": 1760,
|
|
"valid_targets_mean": 3663.6,
|
|
"valid_targets_min": 1646
|
|
},
|
|
{
|
|
"epoch": 2.8149920255183414,
|
|
"grad_norm": 0.4632009960069745,
|
|
"learning_rate": 2.988498377233446e-05,
|
|
"loss": 0.1865,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17049282789230347,
|
|
"step": 1765,
|
|
"valid_targets_mean": 4540.4,
|
|
"valid_targets_min": 1038
|
|
},
|
|
{
|
|
"epoch": 2.8229665071770333,
|
|
"grad_norm": 0.42410704640339714,
|
|
"learning_rate": 2.981576528233114e-05,
|
|
"loss": 0.1772,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19153660535812378,
|
|
"step": 1770,
|
|
"valid_targets_mean": 4711.1,
|
|
"valid_targets_min": 2124
|
|
},
|
|
{
|
|
"epoch": 2.8309409888357258,
|
|
"grad_norm": 0.45542782962545275,
|
|
"learning_rate": 2.9746391564651867e-05,
|
|
"loss": 0.1772,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16209903359413147,
|
|
"step": 1775,
|
|
"valid_targets_mean": 3815.8,
|
|
"valid_targets_min": 1518
|
|
},
|
|
{
|
|
"epoch": 2.8389154704944177,
|
|
"grad_norm": 0.44181038225703473,
|
|
"learning_rate": 2.9676863716380845e-05,
|
|
"loss": 0.1894,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1918773651123047,
|
|
"step": 1780,
|
|
"valid_targets_mean": 5145.6,
|
|
"valid_targets_min": 2006
|
|
},
|
|
{
|
|
"epoch": 2.84688995215311,
|
|
"grad_norm": 0.4733731410179992,
|
|
"learning_rate": 2.9607182837039697e-05,
|
|
"loss": 0.1862,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17720836400985718,
|
|
"step": 1785,
|
|
"valid_targets_mean": 3921.2,
|
|
"valid_targets_min": 1306
|
|
},
|
|
{
|
|
"epoch": 2.854864433811802,
|
|
"grad_norm": 0.4759505220208452,
|
|
"learning_rate": 2.9537350028570126e-05,
|
|
"loss": 0.1993,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19989919662475586,
|
|
"step": 1790,
|
|
"valid_targets_mean": 4174.9,
|
|
"valid_targets_min": 1135
|
|
},
|
|
{
|
|
"epoch": 2.8628389154704945,
|
|
"grad_norm": 0.42878416628933597,
|
|
"learning_rate": 2.946736639531643e-05,
|
|
"loss": 0.1896,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19443926215171814,
|
|
"step": 1795,
|
|
"valid_targets_mean": 4803.4,
|
|
"valid_targets_min": 2250
|
|
},
|
|
{
|
|
"epoch": 2.8708133971291865,
|
|
"grad_norm": 0.4976297417784473,
|
|
"learning_rate": 2.9397233044008092e-05,
|
|
"loss": 0.1833,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19648616015911102,
|
|
"step": 1800,
|
|
"valid_targets_mean": 3719.8,
|
|
"valid_targets_min": 1844
|
|
},
|
|
{
|
|
"epoch": 2.878787878787879,
|
|
"grad_norm": 0.431683591573211,
|
|
"learning_rate": 2.9326951083742243e-05,
|
|
"loss": 0.188,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21068888902664185,
|
|
"step": 1805,
|
|
"valid_targets_mean": 5563.5,
|
|
"valid_targets_min": 1363
|
|
},
|
|
{
|
|
"epoch": 2.886762360446571,
|
|
"grad_norm": 0.5121916169724076,
|
|
"learning_rate": 2.925652162596613e-05,
|
|
"loss": 0.1816,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2092716246843338,
|
|
"step": 1810,
|
|
"valid_targets_mean": 4329.8,
|
|
"valid_targets_min": 1729
|
|
},
|
|
{
|
|
"epoch": 2.8947368421052633,
|
|
"grad_norm": 0.4272155165121044,
|
|
"learning_rate": 2.9185945784459558e-05,
|
|
"loss": 0.1841,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17961260676383972,
|
|
"step": 1815,
|
|
"valid_targets_mean": 4433.8,
|
|
"valid_targets_min": 1092
|
|
},
|
|
{
|
|
"epoch": 2.9027113237639552,
|
|
"grad_norm": 0.4446934153344442,
|
|
"learning_rate": 2.9115224675317252e-05,
|
|
"loss": 0.1854,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.189774751663208,
|
|
"step": 1820,
|
|
"valid_targets_mean": 4546.6,
|
|
"valid_targets_min": 1965
|
|
},
|
|
{
|
|
"epoch": 2.9106858054226477,
|
|
"grad_norm": 0.4076827298604876,
|
|
"learning_rate": 2.9044359416931206e-05,
|
|
"loss": 0.1812,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1695403754711151,
|
|
"step": 1825,
|
|
"valid_targets_mean": 4729.8,
|
|
"valid_targets_min": 2092
|
|
},
|
|
{
|
|
"epoch": 2.9186602870813396,
|
|
"grad_norm": 0.43032076390253393,
|
|
"learning_rate": 2.897335112997302e-05,
|
|
"loss": 0.1817,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15757311880588531,
|
|
"step": 1830,
|
|
"valid_targets_mean": 3931.1,
|
|
"valid_targets_min": 1880
|
|
},
|
|
{
|
|
"epoch": 2.926634768740032,
|
|
"grad_norm": 0.7730974548275344,
|
|
"learning_rate": 2.8902200937376173e-05,
|
|
"loss": 0.1881,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18478021025657654,
|
|
"step": 1835,
|
|
"valid_targets_mean": 4594.4,
|
|
"valid_targets_min": 880
|
|
},
|
|
{
|
|
"epoch": 2.934609250398724,
|
|
"grad_norm": 0.43970061090571333,
|
|
"learning_rate": 2.8830909964318242e-05,
|
|
"loss": 0.1954,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17417211830615997,
|
|
"step": 1840,
|
|
"valid_targets_mean": 4185.7,
|
|
"valid_targets_min": 1643
|
|
},
|
|
{
|
|
"epoch": 2.9425837320574164,
|
|
"grad_norm": 0.46307277533553975,
|
|
"learning_rate": 2.875947933820312e-05,
|
|
"loss": 0.1941,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1813255250453949,
|
|
"step": 1845,
|
|
"valid_targets_mean": 4267.8,
|
|
"valid_targets_min": 412
|
|
},
|
|
{
|
|
"epoch": 2.9505582137161084,
|
|
"grad_norm": 0.45885144568744957,
|
|
"learning_rate": 2.868791018864321e-05,
|
|
"loss": 0.189,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17331145703792572,
|
|
"step": 1850,
|
|
"valid_targets_mean": 3941.5,
|
|
"valid_targets_min": 1577
|
|
},
|
|
{
|
|
"epoch": 2.958532695374801,
|
|
"grad_norm": 0.4564977890190464,
|
|
"learning_rate": 2.861620364744151e-05,
|
|
"loss": 0.1648,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16592752933502197,
|
|
"step": 1855,
|
|
"valid_targets_mean": 3880.0,
|
|
"valid_targets_min": 1417
|
|
},
|
|
{
|
|
"epoch": 2.9665071770334928,
|
|
"grad_norm": 0.4644093884900869,
|
|
"learning_rate": 2.854436084857379e-05,
|
|
"loss": 0.2011,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15251806378364563,
|
|
"step": 1860,
|
|
"valid_targets_mean": 3713.1,
|
|
"valid_targets_min": 1441
|
|
},
|
|
{
|
|
"epoch": 2.974481658692185,
|
|
"grad_norm": 0.4959719091360015,
|
|
"learning_rate": 2.847238292817057e-05,
|
|
"loss": 0.1849,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19827379286289215,
|
|
"step": 1865,
|
|
"valid_targets_mean": 3770.2,
|
|
"valid_targets_min": 600
|
|
},
|
|
{
|
|
"epoch": 2.982456140350877,
|
|
"grad_norm": 0.5612759105888359,
|
|
"learning_rate": 2.8400271024499212e-05,
|
|
"loss": 0.1842,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1830437183380127,
|
|
"step": 1870,
|
|
"valid_targets_mean": 3588.1,
|
|
"valid_targets_min": 510
|
|
},
|
|
{
|
|
"epoch": 2.990430622009569,
|
|
"grad_norm": 0.4841101151001515,
|
|
"learning_rate": 2.832802627794593e-05,
|
|
"loss": 0.1868,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17299309372901917,
|
|
"step": 1875,
|
|
"valid_targets_mean": 4010.5,
|
|
"valid_targets_min": 1156
|
|
},
|
|
{
|
|
"epoch": 2.9984051036682615,
|
|
"grad_norm": 0.48228383639741784,
|
|
"learning_rate": 2.8255649830997704e-05,
|
|
"loss": 0.1939,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.170659139752388,
|
|
"step": 1880,
|
|
"valid_targets_mean": 3891.1,
|
|
"valid_targets_min": 1303
|
|
},
|
|
{
|
|
"epoch": 3.006379585326954,
|
|
"grad_norm": 0.4187959247951961,
|
|
"learning_rate": 2.8183142828224238e-05,
|
|
"loss": 0.1638,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15005925297737122,
|
|
"step": 1885,
|
|
"valid_targets_mean": 4229.8,
|
|
"valid_targets_min": 726
|
|
},
|
|
{
|
|
"epoch": 3.014354066985646,
|
|
"grad_norm": 0.5518696834811211,
|
|
"learning_rate": 2.811050641625988e-05,
|
|
"loss": 0.1718,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16052091121673584,
|
|
"step": 1890,
|
|
"valid_targets_mean": 3637.3,
|
|
"valid_targets_min": 1523
|
|
},
|
|
{
|
|
"epoch": 3.0223285486443383,
|
|
"grad_norm": 0.5272361647263227,
|
|
"learning_rate": 2.8037741743785458e-05,
|
|
"loss": 0.1712,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17966750264167786,
|
|
"step": 1895,
|
|
"valid_targets_mean": 4879.4,
|
|
"valid_targets_min": 1303
|
|
},
|
|
{
|
|
"epoch": 3.0303030303030303,
|
|
"grad_norm": 0.44821310582608315,
|
|
"learning_rate": 2.796484996151013e-05,
|
|
"loss": 0.1883,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21618738770484924,
|
|
"step": 1900,
|
|
"valid_targets_mean": 5611.4,
|
|
"valid_targets_min": 1867
|
|
},
|
|
{
|
|
"epoch": 3.0382775119617227,
|
|
"grad_norm": 0.4784008624223626,
|
|
"learning_rate": 2.7891832222153188e-05,
|
|
"loss": 0.1802,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16092903912067413,
|
|
"step": 1905,
|
|
"valid_targets_mean": 4007.4,
|
|
"valid_targets_min": 291
|
|
},
|
|
{
|
|
"epoch": 3.0462519936204147,
|
|
"grad_norm": 0.46337869290279277,
|
|
"learning_rate": 2.7818689680425807e-05,
|
|
"loss": 0.173,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18314269185066223,
|
|
"step": 1910,
|
|
"valid_targets_mean": 4119.5,
|
|
"valid_targets_min": 319
|
|
},
|
|
{
|
|
"epoch": 3.054226475279107,
|
|
"grad_norm": 0.45687713151005843,
|
|
"learning_rate": 2.7745423493012827e-05,
|
|
"loss": 0.1813,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2106841802597046,
|
|
"step": 1915,
|
|
"valid_targets_mean": 4950.8,
|
|
"valid_targets_min": 1469
|
|
},
|
|
{
|
|
"epoch": 3.062200956937799,
|
|
"grad_norm": 0.4790710436995416,
|
|
"learning_rate": 2.767203481855441e-05,
|
|
"loss": 0.1738,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17638225853443146,
|
|
"step": 1920,
|
|
"valid_targets_mean": 4521.6,
|
|
"valid_targets_min": 1922
|
|
},
|
|
{
|
|
"epoch": 3.0701754385964914,
|
|
"grad_norm": 0.46515663038891586,
|
|
"learning_rate": 2.7598524817627764e-05,
|
|
"loss": 0.1619,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1621520221233368,
|
|
"step": 1925,
|
|
"valid_targets_mean": 5203.6,
|
|
"valid_targets_min": 2173
|
|
},
|
|
{
|
|
"epoch": 3.0781499202551834,
|
|
"grad_norm": 0.6079774346243542,
|
|
"learning_rate": 2.7524894652728754e-05,
|
|
"loss": 0.1762,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18975286185741425,
|
|
"step": 1930,
|
|
"valid_targets_mean": 3732.1,
|
|
"valid_targets_min": 906
|
|
},
|
|
{
|
|
"epoch": 3.0861244019138754,
|
|
"grad_norm": 0.4548224335500046,
|
|
"learning_rate": 2.7451145488253544e-05,
|
|
"loss": 0.1816,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18545740842819214,
|
|
"step": 1935,
|
|
"valid_targets_mean": 4543.8,
|
|
"valid_targets_min": 763
|
|
},
|
|
{
|
|
"epoch": 3.094098883572568,
|
|
"grad_norm": 0.43076682756967427,
|
|
"learning_rate": 2.7377278490480157e-05,
|
|
"loss": 0.1737,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18288424611091614,
|
|
"step": 1940,
|
|
"valid_targets_mean": 5278.6,
|
|
"valid_targets_min": 2115
|
|
},
|
|
{
|
|
"epoch": 3.1020733652312598,
|
|
"grad_norm": 0.44219495400736336,
|
|
"learning_rate": 2.730329482755006e-05,
|
|
"loss": 0.1636,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17927348613739014,
|
|
"step": 1945,
|
|
"valid_targets_mean": 4926.1,
|
|
"valid_targets_min": 1333
|
|
},
|
|
{
|
|
"epoch": 3.110047846889952,
|
|
"grad_norm": 0.4642793792070131,
|
|
"learning_rate": 2.7229195669449667e-05,
|
|
"loss": 0.1687,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15698301792144775,
|
|
"step": 1950,
|
|
"valid_targets_mean": 5145.6,
|
|
"valid_targets_min": 2758
|
|
},
|
|
{
|
|
"epoch": 3.118022328548644,
|
|
"grad_norm": 0.5081120712069186,
|
|
"learning_rate": 2.7154982187991855e-05,
|
|
"loss": 0.1732,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16398665308952332,
|
|
"step": 1955,
|
|
"valid_targets_mean": 3972.9,
|
|
"valid_targets_min": 1024
|
|
},
|
|
{
|
|
"epoch": 3.1259968102073366,
|
|
"grad_norm": 0.39334994682912094,
|
|
"learning_rate": 2.7080655556797406e-05,
|
|
"loss": 0.1637,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1446482390165329,
|
|
"step": 1960,
|
|
"valid_targets_mean": 5047.1,
|
|
"valid_targets_min": 1067
|
|
},
|
|
{
|
|
"epoch": 3.1339712918660285,
|
|
"grad_norm": 0.49370701263368166,
|
|
"learning_rate": 2.700621695127649e-05,
|
|
"loss": 0.1617,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1659671664237976,
|
|
"step": 1965,
|
|
"valid_targets_mean": 3601.8,
|
|
"valid_targets_min": 809
|
|
},
|
|
{
|
|
"epoch": 3.141945773524721,
|
|
"grad_norm": 0.5236111485377425,
|
|
"learning_rate": 2.693166754861003e-05,
|
|
"loss": 0.1831,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15575477480888367,
|
|
"step": 1970,
|
|
"valid_targets_mean": 3251.2,
|
|
"valid_targets_min": 1446
|
|
},
|
|
{
|
|
"epoch": 3.149920255183413,
|
|
"grad_norm": 0.5184956275914406,
|
|
"learning_rate": 2.685700852773113e-05,
|
|
"loss": 0.1792,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17251455783843994,
|
|
"step": 1975,
|
|
"valid_targets_mean": 5407.9,
|
|
"valid_targets_min": 1637
|
|
},
|
|
{
|
|
"epoch": 3.1578947368421053,
|
|
"grad_norm": 0.567900202509518,
|
|
"learning_rate": 2.6782241069306395e-05,
|
|
"loss": 0.1753,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17170360684394836,
|
|
"step": 1980,
|
|
"valid_targets_mean": 3757.9,
|
|
"valid_targets_min": 1669
|
|
},
|
|
{
|
|
"epoch": 3.1658692185007973,
|
|
"grad_norm": 0.5331605536049749,
|
|
"learning_rate": 2.6707366355717268e-05,
|
|
"loss": 0.1745,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.183954119682312,
|
|
"step": 1985,
|
|
"valid_targets_mean": 3772.0,
|
|
"valid_targets_min": 1645
|
|
},
|
|
{
|
|
"epoch": 3.1738437001594897,
|
|
"grad_norm": 0.5372666317594943,
|
|
"learning_rate": 2.663238557104136e-05,
|
|
"loss": 0.1825,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19271928071975708,
|
|
"step": 1990,
|
|
"valid_targets_mean": 4226.8,
|
|
"valid_targets_min": 462
|
|
},
|
|
{
|
|
"epoch": 3.1818181818181817,
|
|
"grad_norm": 0.4511808563006348,
|
|
"learning_rate": 2.655729990103368e-05,
|
|
"loss": 0.1667,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18819640576839447,
|
|
"step": 1995,
|
|
"valid_targets_mean": 4724.9,
|
|
"valid_targets_min": 1801
|
|
},
|
|
{
|
|
"epoch": 3.189792663476874,
|
|
"grad_norm": 0.515719063530578,
|
|
"learning_rate": 2.648211053310792e-05,
|
|
"loss": 0.1754,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17429378628730774,
|
|
"step": 2000,
|
|
"valid_targets_mean": 3824.9,
|
|
"valid_targets_min": 1608
|
|
},
|
|
{
|
|
"epoch": 3.197767145135566,
|
|
"grad_norm": 0.49018438347190046,
|
|
"learning_rate": 2.6406818656317654e-05,
|
|
"loss": 0.1883,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.23138472437858582,
|
|
"step": 2005,
|
|
"valid_targets_mean": 4367.4,
|
|
"valid_targets_min": 1162
|
|
},
|
|
{
|
|
"epoch": 3.2057416267942584,
|
|
"grad_norm": 0.5140123035534574,
|
|
"learning_rate": 2.6331425461337557e-05,
|
|
"loss": 0.1659,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15981221199035645,
|
|
"step": 2010,
|
|
"valid_targets_mean": 3901.4,
|
|
"valid_targets_min": 1592
|
|
},
|
|
{
|
|
"epoch": 3.2137161084529504,
|
|
"grad_norm": 0.5084601119365288,
|
|
"learning_rate": 2.6255932140444546e-05,
|
|
"loss": 0.1795,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16400453448295593,
|
|
"step": 2015,
|
|
"valid_targets_mean": 3658.4,
|
|
"valid_targets_min": 418
|
|
},
|
|
{
|
|
"epoch": 3.221690590111643,
|
|
"grad_norm": 0.5222629370653686,
|
|
"learning_rate": 2.618033988749895e-05,
|
|
"loss": 0.181,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18251457810401917,
|
|
"step": 2020,
|
|
"valid_targets_mean": 3590.5,
|
|
"valid_targets_min": 404
|
|
},
|
|
{
|
|
"epoch": 3.229665071770335,
|
|
"grad_norm": 0.44929273918603535,
|
|
"learning_rate": 2.6104649897925622e-05,
|
|
"loss": 0.171,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18098227679729462,
|
|
"step": 2025,
|
|
"valid_targets_mean": 4840.8,
|
|
"valid_targets_min": 1725
|
|
},
|
|
{
|
|
"epoch": 3.237639553429027,
|
|
"grad_norm": 0.5128148092418456,
|
|
"learning_rate": 2.602886336869503e-05,
|
|
"loss": 0.1653,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18772810697555542,
|
|
"step": 2030,
|
|
"valid_targets_mean": 4087.9,
|
|
"valid_targets_min": 278
|
|
},
|
|
{
|
|
"epoch": 3.245614035087719,
|
|
"grad_norm": 0.5085322011859369,
|
|
"learning_rate": 2.595298149830433e-05,
|
|
"loss": 0.1695,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16362358629703522,
|
|
"step": 2035,
|
|
"valid_targets_mean": 3683.7,
|
|
"valid_targets_min": 1530
|
|
},
|
|
{
|
|
"epoch": 3.2535885167464116,
|
|
"grad_norm": 0.44741903202469246,
|
|
"learning_rate": 2.5877005486758424e-05,
|
|
"loss": 0.1781,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1595516800880432,
|
|
"step": 2040,
|
|
"valid_targets_mean": 4753.9,
|
|
"valid_targets_min": 2291
|
|
},
|
|
{
|
|
"epoch": 3.2615629984051036,
|
|
"grad_norm": 0.46510640494197375,
|
|
"learning_rate": 2.5800936535550954e-05,
|
|
"loss": 0.1641,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16583102941513062,
|
|
"step": 2045,
|
|
"valid_targets_mean": 4312.9,
|
|
"valid_targets_min": 1997
|
|
},
|
|
{
|
|
"epoch": 3.269537480063796,
|
|
"grad_norm": 0.4531657121086904,
|
|
"learning_rate": 2.5724775847645345e-05,
|
|
"loss": 0.1854,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19412560760974884,
|
|
"step": 2050,
|
|
"valid_targets_mean": 5226.4,
|
|
"valid_targets_min": 1255
|
|
},
|
|
{
|
|
"epoch": 3.277511961722488,
|
|
"grad_norm": 0.46205211802785623,
|
|
"learning_rate": 2.5648524627455738e-05,
|
|
"loss": 0.1733,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17929866909980774,
|
|
"step": 2055,
|
|
"valid_targets_mean": 4219.8,
|
|
"valid_targets_min": 962
|
|
},
|
|
{
|
|
"epoch": 3.2854864433811803,
|
|
"grad_norm": 0.583980913889235,
|
|
"learning_rate": 2.557218408082798e-05,
|
|
"loss": 0.185,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17891603708267212,
|
|
"step": 2060,
|
|
"valid_targets_mean": 5263.7,
|
|
"valid_targets_min": 1495
|
|
},
|
|
{
|
|
"epoch": 3.2934609250398723,
|
|
"grad_norm": 0.528345239667559,
|
|
"learning_rate": 2.5495755415020504e-05,
|
|
"loss": 0.1588,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16911692917346954,
|
|
"step": 2065,
|
|
"valid_targets_mean": 4109.6,
|
|
"valid_targets_min": 710
|
|
},
|
|
{
|
|
"epoch": 3.3014354066985647,
|
|
"grad_norm": 0.5330821422864389,
|
|
"learning_rate": 2.5419239838685325e-05,
|
|
"loss": 0.1817,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19775941967964172,
|
|
"step": 2070,
|
|
"valid_targets_mean": 4235.2,
|
|
"valid_targets_min": 989
|
|
},
|
|
{
|
|
"epoch": 3.3094098883572567,
|
|
"grad_norm": 0.5494101250916248,
|
|
"learning_rate": 2.5342638561848817e-05,
|
|
"loss": 0.1817,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20631219446659088,
|
|
"step": 2075,
|
|
"valid_targets_mean": 3423.0,
|
|
"valid_targets_min": 1477
|
|
},
|
|
{
|
|
"epoch": 3.317384370015949,
|
|
"grad_norm": 0.45567413053215694,
|
|
"learning_rate": 2.526595279589265e-05,
|
|
"loss": 0.1742,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16964775323867798,
|
|
"step": 2080,
|
|
"valid_targets_mean": 4667.2,
|
|
"valid_targets_min": 1610
|
|
},
|
|
{
|
|
"epoch": 3.325358851674641,
|
|
"grad_norm": 0.48291376026654725,
|
|
"learning_rate": 2.5189183753534634e-05,
|
|
"loss": 0.1743,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17720086872577667,
|
|
"step": 2085,
|
|
"valid_targets_mean": 4395.1,
|
|
"valid_targets_min": 1859
|
|
},
|
|
{
|
|
"epoch": 3.3333333333333335,
|
|
"grad_norm": 0.5572381771566919,
|
|
"learning_rate": 2.5112332648809495e-05,
|
|
"loss": 0.1747,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16899794340133667,
|
|
"step": 2090,
|
|
"valid_targets_mean": 4103.3,
|
|
"valid_targets_min": 655
|
|
},
|
|
{
|
|
"epoch": 3.3413078149920254,
|
|
"grad_norm": 0.5124868674052819,
|
|
"learning_rate": 2.503540069704973e-05,
|
|
"loss": 0.1963,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.195106640458107,
|
|
"step": 2095,
|
|
"valid_targets_mean": 4012.4,
|
|
"valid_targets_min": 1067
|
|
},
|
|
{
|
|
"epoch": 3.349282296650718,
|
|
"grad_norm": 0.5470151593968711,
|
|
"learning_rate": 2.4958389114866326e-05,
|
|
"loss": 0.1811,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21948151290416718,
|
|
"step": 2100,
|
|
"valid_targets_mean": 3900.2,
|
|
"valid_targets_min": 515
|
|
},
|
|
{
|
|
"epoch": 3.35725677830941,
|
|
"grad_norm": 0.46712894349683487,
|
|
"learning_rate": 2.488129912012958e-05,
|
|
"loss": 0.1875,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1717987209558487,
|
|
"step": 2105,
|
|
"valid_targets_mean": 4451.4,
|
|
"valid_targets_min": 1283
|
|
},
|
|
{
|
|
"epoch": 3.3652312599681022,
|
|
"grad_norm": 0.5296054451681127,
|
|
"learning_rate": 2.4804131931949823e-05,
|
|
"loss": 0.1743,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16957393288612366,
|
|
"step": 2110,
|
|
"valid_targets_mean": 4148.4,
|
|
"valid_targets_min": 858
|
|
},
|
|
{
|
|
"epoch": 3.373205741626794,
|
|
"grad_norm": 0.44324708492039827,
|
|
"learning_rate": 2.4726888770658103e-05,
|
|
"loss": 0.1779,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19824856519699097,
|
|
"step": 2115,
|
|
"valid_targets_mean": 5021.0,
|
|
"valid_targets_min": 1764
|
|
},
|
|
{
|
|
"epoch": 3.3811802232854866,
|
|
"grad_norm": 0.44782890540327536,
|
|
"learning_rate": 2.4649570857786928e-05,
|
|
"loss": 0.1793,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19559164345264435,
|
|
"step": 2120,
|
|
"valid_targets_mean": 5260.7,
|
|
"valid_targets_min": 1522
|
|
},
|
|
{
|
|
"epoch": 3.3891547049441786,
|
|
"grad_norm": 0.4415358571325815,
|
|
"learning_rate": 2.4572179416050953e-05,
|
|
"loss": 0.1694,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16224873065948486,
|
|
"step": 2125,
|
|
"valid_targets_mean": 4277.9,
|
|
"valid_targets_min": 1156
|
|
},
|
|
{
|
|
"epoch": 3.397129186602871,
|
|
"grad_norm": 0.5354109744414436,
|
|
"learning_rate": 2.449471566932761e-05,
|
|
"loss": 0.1642,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15482710301876068,
|
|
"step": 2130,
|
|
"valid_targets_mean": 3514.0,
|
|
"valid_targets_min": 791
|
|
},
|
|
{
|
|
"epoch": 3.405103668261563,
|
|
"grad_norm": 0.4598141951361569,
|
|
"learning_rate": 2.4417180842637764e-05,
|
|
"loss": 0.1714,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1662302166223526,
|
|
"step": 2135,
|
|
"valid_targets_mean": 5137.8,
|
|
"valid_targets_min": 1509
|
|
},
|
|
{
|
|
"epoch": 3.4130781499202554,
|
|
"grad_norm": 0.5263732946909472,
|
|
"learning_rate": 2.4339576162126362e-05,
|
|
"loss": 0.1936,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1471291333436966,
|
|
"step": 2140,
|
|
"valid_targets_mean": 3313.0,
|
|
"valid_targets_min": 1479
|
|
},
|
|
{
|
|
"epoch": 3.4210526315789473,
|
|
"grad_norm": 0.5328365830297036,
|
|
"learning_rate": 2.4261902855043027e-05,
|
|
"loss": 0.1636,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16807958483695984,
|
|
"step": 2145,
|
|
"valid_targets_mean": 3053.6,
|
|
"valid_targets_min": 1038
|
|
},
|
|
{
|
|
"epoch": 3.4290271132376393,
|
|
"grad_norm": 0.4973411156089335,
|
|
"learning_rate": 2.418416214972265e-05,
|
|
"loss": 0.1688,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1936582326889038,
|
|
"step": 2150,
|
|
"valid_targets_mean": 3973.1,
|
|
"valid_targets_min": 1500
|
|
},
|
|
{
|
|
"epoch": 3.4370015948963317,
|
|
"grad_norm": 0.48444835144260445,
|
|
"learning_rate": 2.4106355275565937e-05,
|
|
"loss": 0.1852,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18759359419345856,
|
|
"step": 2155,
|
|
"valid_targets_mean": 4261.8,
|
|
"valid_targets_min": 400
|
|
},
|
|
{
|
|
"epoch": 3.444976076555024,
|
|
"grad_norm": 0.43370463589864844,
|
|
"learning_rate": 2.4028483463020053e-05,
|
|
"loss": 0.1696,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14529302716255188,
|
|
"step": 2160,
|
|
"valid_targets_mean": 4517.1,
|
|
"valid_targets_min": 961
|
|
},
|
|
{
|
|
"epoch": 3.452950558213716,
|
|
"grad_norm": 0.5083145951827011,
|
|
"learning_rate": 2.3950547943559056e-05,
|
|
"loss": 0.1804,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1669323891401291,
|
|
"step": 2165,
|
|
"valid_targets_mean": 3752.9,
|
|
"valid_targets_min": 286
|
|
},
|
|
{
|
|
"epoch": 3.460925039872408,
|
|
"grad_norm": 0.44649606232597117,
|
|
"learning_rate": 2.3872549949664493e-05,
|
|
"loss": 0.1612,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14849191904067993,
|
|
"step": 2170,
|
|
"valid_targets_mean": 4235.8,
|
|
"valid_targets_min": 1152
|
|
},
|
|
{
|
|
"epoch": 3.4688995215311005,
|
|
"grad_norm": 0.39918099350110425,
|
|
"learning_rate": 2.3794490714805897e-05,
|
|
"loss": 0.1627,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1473017781972885,
|
|
"step": 2175,
|
|
"valid_targets_mean": 4902.6,
|
|
"valid_targets_min": 1706
|
|
},
|
|
{
|
|
"epoch": 3.476874003189793,
|
|
"grad_norm": 0.4752038213110687,
|
|
"learning_rate": 2.3716371473421242e-05,
|
|
"loss": 0.1799,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1683822125196457,
|
|
"step": 2180,
|
|
"valid_targets_mean": 3975.9,
|
|
"valid_targets_min": 819
|
|
},
|
|
{
|
|
"epoch": 3.484848484848485,
|
|
"grad_norm": 0.528684880085982,
|
|
"learning_rate": 2.36381934608975e-05,
|
|
"loss": 0.1679,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19594570994377136,
|
|
"step": 2185,
|
|
"valid_targets_mean": 3924.7,
|
|
"valid_targets_min": 1300
|
|
},
|
|
{
|
|
"epoch": 3.492822966507177,
|
|
"grad_norm": 0.45930586621782155,
|
|
"learning_rate": 2.3559957913551014e-05,
|
|
"loss": 0.1737,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1664031445980072,
|
|
"step": 2190,
|
|
"valid_targets_mean": 4340.3,
|
|
"valid_targets_min": 1633
|
|
},
|
|
{
|
|
"epoch": 3.5007974481658692,
|
|
"grad_norm": 0.5587103197549831,
|
|
"learning_rate": 2.3481666068607994e-05,
|
|
"loss": 0.1729,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16250400245189667,
|
|
"step": 2195,
|
|
"valid_targets_mean": 4534.1,
|
|
"valid_targets_min": 1007
|
|
},
|
|
{
|
|
"epoch": 3.5087719298245617,
|
|
"grad_norm": 0.4849102461514037,
|
|
"learning_rate": 2.3403319164184986e-05,
|
|
"loss": 0.1821,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20459748804569244,
|
|
"step": 2200,
|
|
"valid_targets_mean": 4523.4,
|
|
"valid_targets_min": 1154
|
|
},
|
|
{
|
|
"epoch": 3.5167464114832536,
|
|
"grad_norm": 0.5005441651860468,
|
|
"learning_rate": 2.3324918439269206e-05,
|
|
"loss": 0.1717,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21811363101005554,
|
|
"step": 2205,
|
|
"valid_targets_mean": 4844.4,
|
|
"valid_targets_min": 961
|
|
},
|
|
{
|
|
"epoch": 3.5247208931419456,
|
|
"grad_norm": 0.43346313086997673,
|
|
"learning_rate": 2.3246465133699024e-05,
|
|
"loss": 0.1612,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15447084605693817,
|
|
"step": 2210,
|
|
"valid_targets_mean": 4603.5,
|
|
"valid_targets_min": 1367
|
|
},
|
|
{
|
|
"epoch": 3.532695374800638,
|
|
"grad_norm": 0.47292276496360025,
|
|
"learning_rate": 2.31679604881443e-05,
|
|
"loss": 0.1743,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17732763290405273,
|
|
"step": 2215,
|
|
"valid_targets_mean": 3960.4,
|
|
"valid_targets_min": 1642
|
|
},
|
|
{
|
|
"epoch": 3.5406698564593304,
|
|
"grad_norm": 0.495679757867468,
|
|
"learning_rate": 2.3089405744086823e-05,
|
|
"loss": 0.1865,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21921885013580322,
|
|
"step": 2220,
|
|
"valid_targets_mean": 4405.0,
|
|
"valid_targets_min": 1545
|
|
},
|
|
{
|
|
"epoch": 3.5486443381180224,
|
|
"grad_norm": 0.4778196917275496,
|
|
"learning_rate": 2.3010802143800626e-05,
|
|
"loss": 0.1714,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17508621513843536,
|
|
"step": 2225,
|
|
"valid_targets_mean": 4082.6,
|
|
"valid_targets_min": 733
|
|
},
|
|
{
|
|
"epoch": 3.5566188197767143,
|
|
"grad_norm": 0.4115894050467132,
|
|
"learning_rate": 2.2932150930332363e-05,
|
|
"loss": 0.1825,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18152204155921936,
|
|
"step": 2230,
|
|
"valid_targets_mean": 5820.9,
|
|
"valid_targets_min": 1092
|
|
},
|
|
{
|
|
"epoch": 3.5645933014354068,
|
|
"grad_norm": 0.47451933923829037,
|
|
"learning_rate": 2.2853453347481635e-05,
|
|
"loss": 0.1673,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1449630856513977,
|
|
"step": 2235,
|
|
"valid_targets_mean": 3705.7,
|
|
"valid_targets_min": 1707
|
|
},
|
|
{
|
|
"epoch": 3.5725677830940987,
|
|
"grad_norm": 0.4246625993601654,
|
|
"learning_rate": 2.277471063978137e-05,
|
|
"loss": 0.1679,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16793304681777954,
|
|
"step": 2240,
|
|
"valid_targets_mean": 5003.7,
|
|
"valid_targets_min": 1953
|
|
},
|
|
{
|
|
"epoch": 3.580542264752791,
|
|
"grad_norm": 0.39671974046040576,
|
|
"learning_rate": 2.2695924052478065e-05,
|
|
"loss": 0.1656,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15611745417118073,
|
|
"step": 2245,
|
|
"valid_targets_mean": 6036.0,
|
|
"valid_targets_min": 876
|
|
},
|
|
{
|
|
"epoch": 3.588516746411483,
|
|
"grad_norm": 0.4870061065232744,
|
|
"learning_rate": 2.2617094831512167e-05,
|
|
"loss": 0.1643,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18916553258895874,
|
|
"step": 2250,
|
|
"valid_targets_mean": 4665.1,
|
|
"valid_targets_min": 279
|
|
},
|
|
{
|
|
"epoch": 3.5964912280701755,
|
|
"grad_norm": 0.42205864082695105,
|
|
"learning_rate": 2.253822422349831e-05,
|
|
"loss": 0.1547,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14394357800483704,
|
|
"step": 2255,
|
|
"valid_targets_mean": 4616.1,
|
|
"valid_targets_min": 1506
|
|
},
|
|
{
|
|
"epoch": 3.6044657097288675,
|
|
"grad_norm": 0.5077629191461961,
|
|
"learning_rate": 2.2459313475705645e-05,
|
|
"loss": 0.1727,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.186385378241539,
|
|
"step": 2260,
|
|
"valid_targets_mean": 4445.6,
|
|
"valid_targets_min": 1350
|
|
},
|
|
{
|
|
"epoch": 3.61244019138756,
|
|
"grad_norm": 0.5109160496433331,
|
|
"learning_rate": 2.23803638360381e-05,
|
|
"loss": 0.1676,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1869809627532959,
|
|
"step": 2265,
|
|
"valid_targets_mean": 4321.5,
|
|
"valid_targets_min": 1637
|
|
},
|
|
{
|
|
"epoch": 3.620414673046252,
|
|
"grad_norm": 0.47044498703469384,
|
|
"learning_rate": 2.2301376553014625e-05,
|
|
"loss": 0.1752,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17290177941322327,
|
|
"step": 2270,
|
|
"valid_targets_mean": 4751.1,
|
|
"valid_targets_min": 1507
|
|
},
|
|
{
|
|
"epoch": 3.6283891547049443,
|
|
"grad_norm": 0.4729512095272896,
|
|
"learning_rate": 2.2222352875749493e-05,
|
|
"loss": 0.1816,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16890928149223328,
|
|
"step": 2275,
|
|
"valid_targets_mean": 4695.2,
|
|
"valid_targets_min": 449
|
|
},
|
|
{
|
|
"epoch": 3.6363636363636362,
|
|
"grad_norm": 0.49129927177444355,
|
|
"learning_rate": 2.214329405393249e-05,
|
|
"loss": 0.1758,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.22857588529586792,
|
|
"step": 2280,
|
|
"valid_targets_mean": 5797.1,
|
|
"valid_targets_min": 1477
|
|
},
|
|
{
|
|
"epoch": 3.6443381180223287,
|
|
"grad_norm": 0.5144669886043687,
|
|
"learning_rate": 2.2064201337809212e-05,
|
|
"loss": 0.1665,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17676298320293427,
|
|
"step": 2285,
|
|
"valid_targets_mean": 4023.9,
|
|
"valid_targets_min": 1965
|
|
},
|
|
{
|
|
"epoch": 3.6523125996810206,
|
|
"grad_norm": 0.5205684042429719,
|
|
"learning_rate": 2.1985075978161236e-05,
|
|
"loss": 0.1753,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18539035320281982,
|
|
"step": 2290,
|
|
"valid_targets_mean": 3931.9,
|
|
"valid_targets_min": 1260
|
|
},
|
|
{
|
|
"epoch": 3.660287081339713,
|
|
"grad_norm": 0.45299343870889275,
|
|
"learning_rate": 2.1905919226286385e-05,
|
|
"loss": 0.1612,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1681886911392212,
|
|
"step": 2295,
|
|
"valid_targets_mean": 4429.4,
|
|
"valid_targets_min": 1525
|
|
},
|
|
{
|
|
"epoch": 3.668261562998405,
|
|
"grad_norm": 0.4506347012164917,
|
|
"learning_rate": 2.1826732333978924e-05,
|
|
"loss": 0.1623,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16298139095306396,
|
|
"step": 2300,
|
|
"valid_targets_mean": 5305.1,
|
|
"valid_targets_min": 1115
|
|
},
|
|
{
|
|
"epoch": 3.6762360446570974,
|
|
"grad_norm": 0.49361016290996385,
|
|
"learning_rate": 2.1747516553509755e-05,
|
|
"loss": 0.1669,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15665629506111145,
|
|
"step": 2305,
|
|
"valid_targets_mean": 3544.4,
|
|
"valid_targets_min": 1180
|
|
},
|
|
{
|
|
"epoch": 3.6842105263157894,
|
|
"grad_norm": 0.5778205395291415,
|
|
"learning_rate": 2.166827313760662e-05,
|
|
"loss": 0.1805,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.2410949170589447,
|
|
"step": 2310,
|
|
"valid_targets_mean": 4867.1,
|
|
"valid_targets_min": 949
|
|
},
|
|
{
|
|
"epoch": 3.692185007974482,
|
|
"grad_norm": 0.4707485849572123,
|
|
"learning_rate": 2.1589003339434292e-05,
|
|
"loss": 0.1753,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16910578310489655,
|
|
"step": 2315,
|
|
"valid_targets_mean": 4868.1,
|
|
"valid_targets_min": 1811
|
|
},
|
|
{
|
|
"epoch": 3.7001594896331738,
|
|
"grad_norm": 0.5270779646775441,
|
|
"learning_rate": 2.150970841257476e-05,
|
|
"loss": 0.1793,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17331373691558838,
|
|
"step": 2320,
|
|
"valid_targets_mean": 3425.9,
|
|
"valid_targets_min": 691
|
|
},
|
|
{
|
|
"epoch": 3.708133971291866,
|
|
"grad_norm": 0.4559801428274919,
|
|
"learning_rate": 2.1430389611007393e-05,
|
|
"loss": 0.1606,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16542251408100128,
|
|
"step": 2325,
|
|
"valid_targets_mean": 4824.6,
|
|
"valid_targets_min": 1039
|
|
},
|
|
{
|
|
"epoch": 3.716108452950558,
|
|
"grad_norm": 0.39739881900048674,
|
|
"learning_rate": 2.135104818908913e-05,
|
|
"loss": 0.1749,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14475056529045105,
|
|
"step": 2330,
|
|
"valid_targets_mean": 4815.1,
|
|
"valid_targets_min": 1431
|
|
},
|
|
{
|
|
"epoch": 3.7240829346092506,
|
|
"grad_norm": 0.46555703219656114,
|
|
"learning_rate": 2.1271685401534617e-05,
|
|
"loss": 0.1606,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16371889412403107,
|
|
"step": 2335,
|
|
"valid_targets_mean": 3977.3,
|
|
"valid_targets_min": 1661
|
|
},
|
|
{
|
|
"epoch": 3.7320574162679425,
|
|
"grad_norm": 0.538529815559161,
|
|
"learning_rate": 2.11923025033964e-05,
|
|
"loss": 0.1629,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17419461905956268,
|
|
"step": 2340,
|
|
"valid_targets_mean": 4204.9,
|
|
"valid_targets_min": 1325
|
|
},
|
|
{
|
|
"epoch": 3.740031897926635,
|
|
"grad_norm": 0.4835471565226542,
|
|
"learning_rate": 2.111290075004503e-05,
|
|
"loss": 0.1569,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16316327452659607,
|
|
"step": 2345,
|
|
"valid_targets_mean": 4010.9,
|
|
"valid_targets_min": 621
|
|
},
|
|
{
|
|
"epoch": 3.748006379585327,
|
|
"grad_norm": 0.474398046842084,
|
|
"learning_rate": 2.103348139714925e-05,
|
|
"loss": 0.1645,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15339785814285278,
|
|
"step": 2350,
|
|
"valid_targets_mean": 4455.0,
|
|
"valid_targets_min": 1738
|
|
},
|
|
{
|
|
"epoch": 3.7559808612440193,
|
|
"grad_norm": 0.5090815418316118,
|
|
"learning_rate": 2.0954045700656126e-05,
|
|
"loss": 0.1757,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17221853137016296,
|
|
"step": 2355,
|
|
"valid_targets_mean": 3577.1,
|
|
"valid_targets_min": 1003
|
|
},
|
|
{
|
|
"epoch": 3.7639553429027113,
|
|
"grad_norm": 0.43326727747863436,
|
|
"learning_rate": 2.0874594916771174e-05,
|
|
"loss": 0.162,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14630118012428284,
|
|
"step": 2360,
|
|
"valid_targets_mean": 4644.5,
|
|
"valid_targets_min": 1902
|
|
},
|
|
{
|
|
"epoch": 3.7719298245614032,
|
|
"grad_norm": 0.4759734629916401,
|
|
"learning_rate": 2.079513030193852e-05,
|
|
"loss": 0.1719,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17981478571891785,
|
|
"step": 2365,
|
|
"valid_targets_mean": 4401.8,
|
|
"valid_targets_min": 636
|
|
},
|
|
{
|
|
"epoch": 3.7799043062200957,
|
|
"grad_norm": 0.47047851791101014,
|
|
"learning_rate": 2.071565311282099e-05,
|
|
"loss": 0.1867,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17851416766643524,
|
|
"step": 2370,
|
|
"valid_targets_mean": 4240.1,
|
|
"valid_targets_min": 1589
|
|
},
|
|
{
|
|
"epoch": 3.787878787878788,
|
|
"grad_norm": 0.5464081586672188,
|
|
"learning_rate": 2.063616460628029e-05,
|
|
"loss": 0.1638,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18592001497745514,
|
|
"step": 2375,
|
|
"valid_targets_mean": 4041.9,
|
|
"valid_targets_min": 1807
|
|
},
|
|
{
|
|
"epoch": 3.79585326953748,
|
|
"grad_norm": 0.4464818206087508,
|
|
"learning_rate": 2.0556666039357084e-05,
|
|
"loss": 0.1776,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18792858719825745,
|
|
"step": 2380,
|
|
"valid_targets_mean": 4739.8,
|
|
"valid_targets_min": 1157
|
|
},
|
|
{
|
|
"epoch": 3.803827751196172,
|
|
"grad_norm": 0.5041109661524343,
|
|
"learning_rate": 2.047715866925113e-05,
|
|
"loss": 0.1771,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1874503344297409,
|
|
"step": 2385,
|
|
"valid_targets_mean": 4454.7,
|
|
"valid_targets_min": 1432
|
|
},
|
|
{
|
|
"epoch": 3.8118022328548644,
|
|
"grad_norm": 0.45113237417450835,
|
|
"learning_rate": 2.0397643753301403e-05,
|
|
"loss": 0.1814,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1641354113817215,
|
|
"step": 2390,
|
|
"valid_targets_mean": 5005.2,
|
|
"valid_targets_min": 1886
|
|
},
|
|
{
|
|
"epoch": 3.819776714513557,
|
|
"grad_norm": 0.49530584534417454,
|
|
"learning_rate": 2.0318122548966225e-05,
|
|
"loss": 0.1705,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1944631189107895,
|
|
"step": 2395,
|
|
"valid_targets_mean": 4375.2,
|
|
"valid_targets_min": 1665
|
|
},
|
|
{
|
|
"epoch": 3.827751196172249,
|
|
"grad_norm": 0.49543993336649245,
|
|
"learning_rate": 2.0238596313803337e-05,
|
|
"loss": 0.1678,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1980539858341217,
|
|
"step": 2400,
|
|
"valid_targets_mean": 4541.7,
|
|
"valid_targets_min": 627
|
|
},
|
|
{
|
|
"epoch": 3.8357256778309408,
|
|
"grad_norm": 0.4254188193829258,
|
|
"learning_rate": 2.015906630545005e-05,
|
|
"loss": 0.1766,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15166829526424408,
|
|
"step": 2405,
|
|
"valid_targets_mean": 4715.5,
|
|
"valid_targets_min": 2173
|
|
},
|
|
{
|
|
"epoch": 3.843700159489633,
|
|
"grad_norm": 0.4882032464005021,
|
|
"learning_rate": 2.0079533781603352e-05,
|
|
"loss": 0.1846,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19435849785804749,
|
|
"step": 2410,
|
|
"valid_targets_mean": 4388.9,
|
|
"valid_targets_min": 531
|
|
},
|
|
{
|
|
"epoch": 3.8516746411483256,
|
|
"grad_norm": 0.5265806814763052,
|
|
"learning_rate": 2e-05,
|
|
"loss": 0.1647,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18883755803108215,
|
|
"step": 2415,
|
|
"valid_targets_mean": 3900.7,
|
|
"valid_targets_min": 1194
|
|
},
|
|
{
|
|
"epoch": 3.8596491228070176,
|
|
"grad_norm": 0.4970672078467313,
|
|
"learning_rate": 1.9920466218396655e-05,
|
|
"loss": 0.1665,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14973965287208557,
|
|
"step": 2420,
|
|
"valid_targets_mean": 3951.0,
|
|
"valid_targets_min": 1906
|
|
},
|
|
{
|
|
"epoch": 3.8676236044657095,
|
|
"grad_norm": 0.4485320461327396,
|
|
"learning_rate": 1.9840933694549956e-05,
|
|
"loss": 0.1725,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1715766191482544,
|
|
"step": 2425,
|
|
"valid_targets_mean": 4628.1,
|
|
"valid_targets_min": 2325
|
|
},
|
|
{
|
|
"epoch": 3.875598086124402,
|
|
"grad_norm": 0.46144828607434146,
|
|
"learning_rate": 1.976140368619667e-05,
|
|
"loss": 0.1849,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1804232895374298,
|
|
"step": 2430,
|
|
"valid_targets_mean": 4036.7,
|
|
"valid_targets_min": 1579
|
|
},
|
|
{
|
|
"epoch": 3.8835725677830943,
|
|
"grad_norm": 0.40935989183035376,
|
|
"learning_rate": 1.9681877451033778e-05,
|
|
"loss": 0.1678,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13349954783916473,
|
|
"step": 2435,
|
|
"valid_targets_mean": 4621.5,
|
|
"valid_targets_min": 2387
|
|
},
|
|
{
|
|
"epoch": 3.8915470494417863,
|
|
"grad_norm": 0.5822879432451648,
|
|
"learning_rate": 1.9602356246698597e-05,
|
|
"loss": 0.1796,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17016389966011047,
|
|
"step": 2440,
|
|
"valid_targets_mean": 4279.2,
|
|
"valid_targets_min": 1442
|
|
},
|
|
{
|
|
"epoch": 3.8995215311004783,
|
|
"grad_norm": 0.45662436879070906,
|
|
"learning_rate": 1.9522841330748877e-05,
|
|
"loss": 0.1743,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16033823788166046,
|
|
"step": 2445,
|
|
"valid_targets_mean": 4220.1,
|
|
"valid_targets_min": 1598
|
|
},
|
|
{
|
|
"epoch": 3.9074960127591707,
|
|
"grad_norm": 0.48077089254158484,
|
|
"learning_rate": 1.944333396064292e-05,
|
|
"loss": 0.1569,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14935660362243652,
|
|
"step": 2450,
|
|
"valid_targets_mean": 3772.9,
|
|
"valid_targets_min": 1647
|
|
},
|
|
{
|
|
"epoch": 3.915470494417863,
|
|
"grad_norm": 0.4244926540119392,
|
|
"learning_rate": 1.936383539371971e-05,
|
|
"loss": 0.1662,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14506709575653076,
|
|
"step": 2455,
|
|
"valid_targets_mean": 4454.8,
|
|
"valid_targets_min": 1710
|
|
},
|
|
{
|
|
"epoch": 3.923444976076555,
|
|
"grad_norm": 0.4427100557851415,
|
|
"learning_rate": 1.9284346887179016e-05,
|
|
"loss": 0.1653,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15511883795261383,
|
|
"step": 2460,
|
|
"valid_targets_mean": 4552.0,
|
|
"valid_targets_min": 1761
|
|
},
|
|
{
|
|
"epoch": 3.931419457735247,
|
|
"grad_norm": 0.47909923376130753,
|
|
"learning_rate": 1.9204869698061493e-05,
|
|
"loss": 0.1718,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15708695352077484,
|
|
"step": 2465,
|
|
"valid_targets_mean": 3728.4,
|
|
"valid_targets_min": 1978
|
|
},
|
|
{
|
|
"epoch": 3.9393939393939394,
|
|
"grad_norm": 0.595539634060728,
|
|
"learning_rate": 1.9125405083228833e-05,
|
|
"loss": 0.1578,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16844023764133453,
|
|
"step": 2470,
|
|
"valid_targets_mean": 3325.6,
|
|
"valid_targets_min": 1501
|
|
},
|
|
{
|
|
"epoch": 3.9473684210526314,
|
|
"grad_norm": 0.4533444186226723,
|
|
"learning_rate": 1.9045954299343884e-05,
|
|
"loss": 0.1677,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17140570282936096,
|
|
"step": 2475,
|
|
"valid_targets_mean": 4405.3,
|
|
"valid_targets_min": 2264
|
|
},
|
|
{
|
|
"epoch": 3.955342902711324,
|
|
"grad_norm": 0.4595885145274142,
|
|
"learning_rate": 1.896651860285076e-05,
|
|
"loss": 0.1826,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16533583402633667,
|
|
"step": 2480,
|
|
"valid_targets_mean": 4279.9,
|
|
"valid_targets_min": 1464
|
|
},
|
|
{
|
|
"epoch": 3.963317384370016,
|
|
"grad_norm": 0.45222684065372687,
|
|
"learning_rate": 1.8887099249954976e-05,
|
|
"loss": 0.1701,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1614869087934494,
|
|
"step": 2485,
|
|
"valid_targets_mean": 4647.2,
|
|
"valid_targets_min": 1192
|
|
},
|
|
{
|
|
"epoch": 3.971291866028708,
|
|
"grad_norm": 0.4725644857876905,
|
|
"learning_rate": 1.8807697496603604e-05,
|
|
"loss": 0.1643,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17880479991436005,
|
|
"step": 2490,
|
|
"valid_targets_mean": 4414.1,
|
|
"valid_targets_min": 2303
|
|
},
|
|
{
|
|
"epoch": 3.9792663476874,
|
|
"grad_norm": 0.45652270717722077,
|
|
"learning_rate": 1.8728314598465386e-05,
|
|
"loss": 0.1797,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1617339849472046,
|
|
"step": 2495,
|
|
"valid_targets_mean": 4411.3,
|
|
"valid_targets_min": 1827
|
|
},
|
|
{
|
|
"epoch": 3.9872408293460926,
|
|
"grad_norm": 0.5151368519656011,
|
|
"learning_rate": 1.8648951810910878e-05,
|
|
"loss": 0.1743,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15443146228790283,
|
|
"step": 2500,
|
|
"valid_targets_mean": 3280.9,
|
|
"valid_targets_min": 1115
|
|
},
|
|
{
|
|
"epoch": 3.9952153110047846,
|
|
"grad_norm": 0.47470285857166405,
|
|
"learning_rate": 1.856961038899261e-05,
|
|
"loss": 0.1606,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15671250224113464,
|
|
"step": 2505,
|
|
"valid_targets_mean": 4404.2,
|
|
"valid_targets_min": 1362
|
|
},
|
|
{
|
|
"epoch": 4.003189792663477,
|
|
"grad_norm": 0.4899880225039297,
|
|
"learning_rate": 1.849029158742525e-05,
|
|
"loss": 0.1654,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15422114729881287,
|
|
"step": 2510,
|
|
"valid_targets_mean": 3694.6,
|
|
"valid_targets_min": 1506
|
|
},
|
|
{
|
|
"epoch": 4.011164274322169,
|
|
"grad_norm": 0.4634790318267732,
|
|
"learning_rate": 1.8410996660565714e-05,
|
|
"loss": 0.1608,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1345663070678711,
|
|
"step": 2515,
|
|
"valid_targets_mean": 4565.0,
|
|
"valid_targets_min": 2064
|
|
},
|
|
{
|
|
"epoch": 4.019138755980861,
|
|
"grad_norm": 0.5042108705454043,
|
|
"learning_rate": 1.8331726862393385e-05,
|
|
"loss": 0.1545,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16057871282100677,
|
|
"step": 2520,
|
|
"valid_targets_mean": 4519.1,
|
|
"valid_targets_min": 1717
|
|
},
|
|
{
|
|
"epoch": 4.027113237639553,
|
|
"grad_norm": 0.4844796153115308,
|
|
"learning_rate": 1.8252483446490245e-05,
|
|
"loss": 0.1659,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17984318733215332,
|
|
"step": 2525,
|
|
"valid_targets_mean": 4351.6,
|
|
"valid_targets_min": 1530
|
|
},
|
|
{
|
|
"epoch": 4.035087719298246,
|
|
"grad_norm": 0.45265367637894693,
|
|
"learning_rate": 1.8173267666021075e-05,
|
|
"loss": 0.1462,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16487132012844086,
|
|
"step": 2530,
|
|
"valid_targets_mean": 5113.8,
|
|
"valid_targets_min": 1797
|
|
},
|
|
{
|
|
"epoch": 4.043062200956938,
|
|
"grad_norm": 0.5341917502431222,
|
|
"learning_rate": 1.8094080773713615e-05,
|
|
"loss": 0.152,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1509234458208084,
|
|
"step": 2535,
|
|
"valid_targets_mean": 3534.0,
|
|
"valid_targets_min": 1379
|
|
},
|
|
{
|
|
"epoch": 4.05103668261563,
|
|
"grad_norm": 0.5250909339903316,
|
|
"learning_rate": 1.8014924021838774e-05,
|
|
"loss": 0.1664,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1545257717370987,
|
|
"step": 2540,
|
|
"valid_targets_mean": 3596.7,
|
|
"valid_targets_min": 524
|
|
},
|
|
{
|
|
"epoch": 4.059011164274322,
|
|
"grad_norm": 0.4771773683854749,
|
|
"learning_rate": 1.7935798662190798e-05,
|
|
"loss": 0.1594,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14937272667884827,
|
|
"step": 2545,
|
|
"valid_targets_mean": 4365.2,
|
|
"valid_targets_min": 1184
|
|
},
|
|
{
|
|
"epoch": 4.0669856459330145,
|
|
"grad_norm": 0.4378658709624504,
|
|
"learning_rate": 1.785670594606752e-05,
|
|
"loss": 0.1631,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16334344446659088,
|
|
"step": 2550,
|
|
"valid_targets_mean": 4933.5,
|
|
"valid_targets_min": 1487
|
|
},
|
|
{
|
|
"epoch": 4.074960127591707,
|
|
"grad_norm": 0.5408025302725352,
|
|
"learning_rate": 1.777764712425052e-05,
|
|
"loss": 0.1589,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1732405722141266,
|
|
"step": 2555,
|
|
"valid_targets_mean": 3627.1,
|
|
"valid_targets_min": 317
|
|
},
|
|
{
|
|
"epoch": 4.082934609250398,
|
|
"grad_norm": 0.507827216712315,
|
|
"learning_rate": 1.7698623446985378e-05,
|
|
"loss": 0.1578,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17435893416404724,
|
|
"step": 2560,
|
|
"valid_targets_mean": 4402.0,
|
|
"valid_targets_min": 1466
|
|
},
|
|
{
|
|
"epoch": 4.090909090909091,
|
|
"grad_norm": 0.4951560011181849,
|
|
"learning_rate": 1.7619636163961907e-05,
|
|
"loss": 0.1556,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14189505577087402,
|
|
"step": 2565,
|
|
"valid_targets_mean": 4492.1,
|
|
"valid_targets_min": 1893
|
|
},
|
|
{
|
|
"epoch": 4.098883572567783,
|
|
"grad_norm": 0.48077419738077953,
|
|
"learning_rate": 1.754068652429436e-05,
|
|
"loss": 0.1465,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14747288823127747,
|
|
"step": 2570,
|
|
"valid_targets_mean": 4816.2,
|
|
"valid_targets_min": 1338
|
|
},
|
|
{
|
|
"epoch": 4.106858054226476,
|
|
"grad_norm": 0.48118894084351693,
|
|
"learning_rate": 1.74617757765017e-05,
|
|
"loss": 0.1472,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15636608004570007,
|
|
"step": 2575,
|
|
"valid_targets_mean": 3937.2,
|
|
"valid_targets_min": 385
|
|
},
|
|
{
|
|
"epoch": 4.114832535885167,
|
|
"grad_norm": 0.46253282549841945,
|
|
"learning_rate": 1.738290516848784e-05,
|
|
"loss": 0.1558,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13956312835216522,
|
|
"step": 2580,
|
|
"valid_targets_mean": 5142.4,
|
|
"valid_targets_min": 1541
|
|
},
|
|
{
|
|
"epoch": 4.12280701754386,
|
|
"grad_norm": 0.4840910750690081,
|
|
"learning_rate": 1.730407594752194e-05,
|
|
"loss": 0.1616,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18448464572429657,
|
|
"step": 2585,
|
|
"valid_targets_mean": 4904.5,
|
|
"valid_targets_min": 610
|
|
},
|
|
{
|
|
"epoch": 4.130781499202552,
|
|
"grad_norm": 0.46098830201420854,
|
|
"learning_rate": 1.722528936021864e-05,
|
|
"loss": 0.1606,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16083885729312897,
|
|
"step": 2590,
|
|
"valid_targets_mean": 4776.6,
|
|
"valid_targets_min": 1693
|
|
},
|
|
{
|
|
"epoch": 4.138755980861244,
|
|
"grad_norm": 0.4509630939011778,
|
|
"learning_rate": 1.714654665251837e-05,
|
|
"loss": 0.1528,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1386050283908844,
|
|
"step": 2595,
|
|
"valid_targets_mean": 4520.9,
|
|
"valid_targets_min": 430
|
|
},
|
|
{
|
|
"epoch": 4.146730462519936,
|
|
"grad_norm": 0.5078283848796444,
|
|
"learning_rate": 1.7067849069667644e-05,
|
|
"loss": 0.1543,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14574451744556427,
|
|
"step": 2600,
|
|
"valid_targets_mean": 3890.8,
|
|
"valid_targets_min": 1673
|
|
},
|
|
{
|
|
"epoch": 4.154704944178628,
|
|
"grad_norm": 0.5425342658424712,
|
|
"learning_rate": 1.6989197856199377e-05,
|
|
"loss": 0.165,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14644119143486023,
|
|
"step": 2605,
|
|
"valid_targets_mean": 4598.1,
|
|
"valid_targets_min": 599
|
|
},
|
|
{
|
|
"epoch": 4.162679425837321,
|
|
"grad_norm": 0.6814195331715653,
|
|
"learning_rate": 1.6910594255913177e-05,
|
|
"loss": 0.1587,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16274163126945496,
|
|
"step": 2610,
|
|
"valid_targets_mean": 3990.4,
|
|
"valid_targets_min": 1780
|
|
},
|
|
{
|
|
"epoch": 4.170653907496013,
|
|
"grad_norm": 0.47901972456824515,
|
|
"learning_rate": 1.6832039511855702e-05,
|
|
"loss": 0.1599,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1412867307662964,
|
|
"step": 2615,
|
|
"valid_targets_mean": 4189.2,
|
|
"valid_targets_min": 1917
|
|
},
|
|
{
|
|
"epoch": 4.178628389154705,
|
|
"grad_norm": 0.47380491777234185,
|
|
"learning_rate": 1.6753534866300983e-05,
|
|
"loss": 0.1575,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16598361730575562,
|
|
"step": 2620,
|
|
"valid_targets_mean": 4522.2,
|
|
"valid_targets_min": 2215
|
|
},
|
|
{
|
|
"epoch": 4.186602870813397,
|
|
"grad_norm": 0.4623200429689346,
|
|
"learning_rate": 1.6675081560730804e-05,
|
|
"loss": 0.1655,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1583814024925232,
|
|
"step": 2625,
|
|
"valid_targets_mean": 4811.9,
|
|
"valid_targets_min": 484
|
|
},
|
|
{
|
|
"epoch": 4.1945773524720895,
|
|
"grad_norm": 0.4639372494986275,
|
|
"learning_rate": 1.6596680835815024e-05,
|
|
"loss": 0.1469,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1427120864391327,
|
|
"step": 2630,
|
|
"valid_targets_mean": 4565.5,
|
|
"valid_targets_min": 1893
|
|
},
|
|
{
|
|
"epoch": 4.202551834130782,
|
|
"grad_norm": 0.5202843864845481,
|
|
"learning_rate": 1.651833393139201e-05,
|
|
"loss": 0.1565,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16338421404361725,
|
|
"step": 2635,
|
|
"valid_targets_mean": 4019.4,
|
|
"valid_targets_min": 1725
|
|
},
|
|
{
|
|
"epoch": 4.2105263157894735,
|
|
"grad_norm": 0.5235868701035974,
|
|
"learning_rate": 1.6440042086449e-05,
|
|
"loss": 0.1482,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15927179157733917,
|
|
"step": 2640,
|
|
"valid_targets_mean": 4495.2,
|
|
"valid_targets_min": 951
|
|
},
|
|
{
|
|
"epoch": 4.218500797448166,
|
|
"grad_norm": 0.44531838825532105,
|
|
"learning_rate": 1.6361806539102508e-05,
|
|
"loss": 0.1597,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1629904806613922,
|
|
"step": 2645,
|
|
"valid_targets_mean": 5343.2,
|
|
"valid_targets_min": 867
|
|
},
|
|
{
|
|
"epoch": 4.226475279106858,
|
|
"grad_norm": 0.4251988292309876,
|
|
"learning_rate": 1.628362852657876e-05,
|
|
"loss": 0.1467,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13012798130512238,
|
|
"step": 2650,
|
|
"valid_targets_mean": 5094.8,
|
|
"valid_targets_min": 339
|
|
},
|
|
{
|
|
"epoch": 4.23444976076555,
|
|
"grad_norm": 0.46887168314695904,
|
|
"learning_rate": 1.620550928519411e-05,
|
|
"loss": 0.1574,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15908288955688477,
|
|
"step": 2655,
|
|
"valid_targets_mean": 4939.9,
|
|
"valid_targets_min": 2028
|
|
},
|
|
{
|
|
"epoch": 4.242424242424242,
|
|
"grad_norm": 0.5098973831313105,
|
|
"learning_rate": 1.612745005033551e-05,
|
|
"loss": 0.1407,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1334664523601532,
|
|
"step": 2660,
|
|
"valid_targets_mean": 4081.6,
|
|
"valid_targets_min": 1556
|
|
},
|
|
{
|
|
"epoch": 4.250398724082935,
|
|
"grad_norm": 0.46808481227323445,
|
|
"learning_rate": 1.6049452056440948e-05,
|
|
"loss": 0.15,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15534178912639618,
|
|
"step": 2665,
|
|
"valid_targets_mean": 4708.3,
|
|
"valid_targets_min": 2017
|
|
},
|
|
{
|
|
"epoch": 4.258373205741627,
|
|
"grad_norm": 0.4875790771425733,
|
|
"learning_rate": 1.597151653697995e-05,
|
|
"loss": 0.16,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12777553498744965,
|
|
"step": 2670,
|
|
"valid_targets_mean": 4391.9,
|
|
"valid_targets_min": 278
|
|
},
|
|
{
|
|
"epoch": 4.266347687400319,
|
|
"grad_norm": 0.47842605994377785,
|
|
"learning_rate": 1.5893644724434066e-05,
|
|
"loss": 0.1671,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15351736545562744,
|
|
"step": 2675,
|
|
"valid_targets_mean": 4784.4,
|
|
"valid_targets_min": 440
|
|
},
|
|
{
|
|
"epoch": 4.274322169059011,
|
|
"grad_norm": 0.4798020507410421,
|
|
"learning_rate": 1.581583785027736e-05,
|
|
"loss": 0.1435,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12768876552581787,
|
|
"step": 2680,
|
|
"valid_targets_mean": 4058.9,
|
|
"valid_targets_min": 1533
|
|
},
|
|
{
|
|
"epoch": 4.282296650717703,
|
|
"grad_norm": 0.4890971139697186,
|
|
"learning_rate": 1.5738097144956976e-05,
|
|
"loss": 0.1576,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1449412852525711,
|
|
"step": 2685,
|
|
"valid_targets_mean": 4657.4,
|
|
"valid_targets_min": 2325
|
|
},
|
|
{
|
|
"epoch": 4.290271132376396,
|
|
"grad_norm": 0.5461562109497161,
|
|
"learning_rate": 1.566042383787364e-05,
|
|
"loss": 0.1562,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1455443799495697,
|
|
"step": 2690,
|
|
"valid_targets_mean": 3337.1,
|
|
"valid_targets_min": 1529
|
|
},
|
|
{
|
|
"epoch": 4.298245614035087,
|
|
"grad_norm": 0.46410280159872597,
|
|
"learning_rate": 1.558281915736224e-05,
|
|
"loss": 0.1546,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17412641644477844,
|
|
"step": 2695,
|
|
"valid_targets_mean": 4995.4,
|
|
"valid_targets_min": 2412
|
|
},
|
|
{
|
|
"epoch": 4.30622009569378,
|
|
"grad_norm": 0.5330477782573131,
|
|
"learning_rate": 1.550528433067239e-05,
|
|
"loss": 0.1501,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17384281754493713,
|
|
"step": 2700,
|
|
"valid_targets_mean": 4280.9,
|
|
"valid_targets_min": 396
|
|
},
|
|
{
|
|
"epoch": 4.314194577352472,
|
|
"grad_norm": 0.4993801267582135,
|
|
"learning_rate": 1.5427820583949054e-05,
|
|
"loss": 0.1555,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1459612250328064,
|
|
"step": 2705,
|
|
"valid_targets_mean": 3986.2,
|
|
"valid_targets_min": 1182
|
|
},
|
|
{
|
|
"epoch": 4.3221690590111645,
|
|
"grad_norm": 0.5020099422312442,
|
|
"learning_rate": 1.5350429142213075e-05,
|
|
"loss": 0.1648,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.24648582935333252,
|
|
"step": 2710,
|
|
"valid_targets_mean": 6519.9,
|
|
"valid_targets_min": 2514
|
|
},
|
|
{
|
|
"epoch": 4.330143540669856,
|
|
"grad_norm": 0.544133437246787,
|
|
"learning_rate": 1.5273111229341907e-05,
|
|
"loss": 0.1447,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13865166902542114,
|
|
"step": 2715,
|
|
"valid_targets_mean": 4138.2,
|
|
"valid_targets_min": 1620
|
|
},
|
|
{
|
|
"epoch": 4.3381180223285485,
|
|
"grad_norm": 0.5733519120938113,
|
|
"learning_rate": 1.5195868068050185e-05,
|
|
"loss": 0.1579,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16256359219551086,
|
|
"step": 2720,
|
|
"valid_targets_mean": 3474.5,
|
|
"valid_targets_min": 1836
|
|
},
|
|
{
|
|
"epoch": 4.346092503987241,
|
|
"grad_norm": 0.4716089878439407,
|
|
"learning_rate": 1.5118700879870426e-05,
|
|
"loss": 0.1533,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17399930953979492,
|
|
"step": 2725,
|
|
"valid_targets_mean": 5075.4,
|
|
"valid_targets_min": 1690
|
|
},
|
|
{
|
|
"epoch": 4.354066985645933,
|
|
"grad_norm": 0.4209893499367196,
|
|
"learning_rate": 1.5041610885133681e-05,
|
|
"loss": 0.1638,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12980689108371735,
|
|
"step": 2730,
|
|
"valid_targets_mean": 5178.8,
|
|
"valid_targets_min": 1476
|
|
},
|
|
{
|
|
"epoch": 4.362041467304625,
|
|
"grad_norm": 0.5515449457928758,
|
|
"learning_rate": 1.496459930295028e-05,
|
|
"loss": 0.1531,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16415995359420776,
|
|
"step": 2735,
|
|
"valid_targets_mean": 3561.8,
|
|
"valid_targets_min": 1387
|
|
},
|
|
{
|
|
"epoch": 4.370015948963317,
|
|
"grad_norm": 0.5109525433836578,
|
|
"learning_rate": 1.4887667351190508e-05,
|
|
"loss": 0.1669,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18121805787086487,
|
|
"step": 2740,
|
|
"valid_targets_mean": 4864.2,
|
|
"valid_targets_min": 1895
|
|
},
|
|
{
|
|
"epoch": 4.37799043062201,
|
|
"grad_norm": 0.5037396991640567,
|
|
"learning_rate": 1.4810816246465374e-05,
|
|
"loss": 0.152,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1513875126838684,
|
|
"step": 2745,
|
|
"valid_targets_mean": 4453.0,
|
|
"valid_targets_min": 907
|
|
},
|
|
{
|
|
"epoch": 4.385964912280702,
|
|
"grad_norm": 0.4934516813748904,
|
|
"learning_rate": 1.4734047204107358e-05,
|
|
"loss": 0.1558,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1657409369945526,
|
|
"step": 2750,
|
|
"valid_targets_mean": 4435.2,
|
|
"valid_targets_min": 1760
|
|
},
|
|
{
|
|
"epoch": 4.393939393939394,
|
|
"grad_norm": 0.4417129235128675,
|
|
"learning_rate": 1.4657361438151192e-05,
|
|
"loss": 0.1572,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13217982649803162,
|
|
"step": 2755,
|
|
"valid_targets_mean": 4826.6,
|
|
"valid_targets_min": 1233
|
|
},
|
|
{
|
|
"epoch": 4.401913875598086,
|
|
"grad_norm": 0.572380140644963,
|
|
"learning_rate": 1.458076016131468e-05,
|
|
"loss": 0.1689,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1532914787530899,
|
|
"step": 2760,
|
|
"valid_targets_mean": 3163.1,
|
|
"valid_targets_min": 1432
|
|
},
|
|
{
|
|
"epoch": 4.409888357256778,
|
|
"grad_norm": 0.5850789870287324,
|
|
"learning_rate": 1.4504244584979494e-05,
|
|
"loss": 0.1568,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17484025657176971,
|
|
"step": 2765,
|
|
"valid_targets_mean": 3527.3,
|
|
"valid_targets_min": 759
|
|
},
|
|
{
|
|
"epoch": 4.417862838915471,
|
|
"grad_norm": 0.5345615763784187,
|
|
"learning_rate": 1.442781591917203e-05,
|
|
"loss": 0.1611,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17847514152526855,
|
|
"step": 2770,
|
|
"valid_targets_mean": 3803.9,
|
|
"valid_targets_min": 288
|
|
},
|
|
{
|
|
"epoch": 4.425837320574162,
|
|
"grad_norm": 0.49877751842401186,
|
|
"learning_rate": 1.4351475372544262e-05,
|
|
"loss": 0.1466,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14222899079322815,
|
|
"step": 2775,
|
|
"valid_targets_mean": 4224.1,
|
|
"valid_targets_min": 2213
|
|
},
|
|
{
|
|
"epoch": 4.433811802232855,
|
|
"grad_norm": 0.5127528633812499,
|
|
"learning_rate": 1.4275224152354658e-05,
|
|
"loss": 0.1398,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14961972832679749,
|
|
"step": 2780,
|
|
"valid_targets_mean": 3905.4,
|
|
"valid_targets_min": 1318
|
|
},
|
|
{
|
|
"epoch": 4.441786283891547,
|
|
"grad_norm": 0.48555317932260317,
|
|
"learning_rate": 1.4199063464449047e-05,
|
|
"loss": 0.1626,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17323818802833557,
|
|
"step": 2785,
|
|
"valid_targets_mean": 4936.8,
|
|
"valid_targets_min": 1178
|
|
},
|
|
{
|
|
"epoch": 4.44976076555024,
|
|
"grad_norm": 0.46902570337103017,
|
|
"learning_rate": 1.4122994513241587e-05,
|
|
"loss": 0.1639,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14024415612220764,
|
|
"step": 2790,
|
|
"valid_targets_mean": 4709.8,
|
|
"valid_targets_min": 1527
|
|
},
|
|
{
|
|
"epoch": 4.457735247208931,
|
|
"grad_norm": 0.5286358404792354,
|
|
"learning_rate": 1.4047018501695677e-05,
|
|
"loss": 0.1688,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14404083788394928,
|
|
"step": 2795,
|
|
"valid_targets_mean": 4177.2,
|
|
"valid_targets_min": 949
|
|
},
|
|
{
|
|
"epoch": 4.4657097288676235,
|
|
"grad_norm": 0.5835959583862976,
|
|
"learning_rate": 1.3971136631304978e-05,
|
|
"loss": 0.1534,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16053429245948792,
|
|
"step": 2800,
|
|
"valid_targets_mean": 4404.4,
|
|
"valid_targets_min": 635
|
|
},
|
|
{
|
|
"epoch": 4.473684210526316,
|
|
"grad_norm": 0.5087390219672198,
|
|
"learning_rate": 1.3895350102074386e-05,
|
|
"loss": 0.1662,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.154604971408844,
|
|
"step": 2805,
|
|
"valid_targets_mean": 4275.4,
|
|
"valid_targets_min": 311
|
|
},
|
|
{
|
|
"epoch": 4.481658692185008,
|
|
"grad_norm": 0.4701979924293451,
|
|
"learning_rate": 1.3819660112501054e-05,
|
|
"loss": 0.1476,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16549868881702423,
|
|
"step": 2810,
|
|
"valid_targets_mean": 4737.1,
|
|
"valid_targets_min": 860
|
|
},
|
|
{
|
|
"epoch": 4.4896331738437,
|
|
"grad_norm": 0.4801457713814075,
|
|
"learning_rate": 1.3744067859555461e-05,
|
|
"loss": 0.1571,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14757005870342255,
|
|
"step": 2815,
|
|
"valid_targets_mean": 4083.9,
|
|
"valid_targets_min": 1011
|
|
},
|
|
{
|
|
"epoch": 4.497607655502392,
|
|
"grad_norm": 0.4327428514767497,
|
|
"learning_rate": 1.3668574538662451e-05,
|
|
"loss": 0.1694,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16689707338809967,
|
|
"step": 2820,
|
|
"valid_targets_mean": 5532.1,
|
|
"valid_targets_min": 1640
|
|
},
|
|
{
|
|
"epoch": 4.505582137161085,
|
|
"grad_norm": 0.48232554647060877,
|
|
"learning_rate": 1.3593181343682353e-05,
|
|
"loss": 0.1579,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1623343527317047,
|
|
"step": 2825,
|
|
"valid_targets_mean": 4389.2,
|
|
"valid_targets_min": 815
|
|
},
|
|
{
|
|
"epoch": 4.513556618819777,
|
|
"grad_norm": 0.46842062004881196,
|
|
"learning_rate": 1.3517889466892085e-05,
|
|
"loss": 0.1583,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16388767957687378,
|
|
"step": 2830,
|
|
"valid_targets_mean": 4568.9,
|
|
"valid_targets_min": 1589
|
|
},
|
|
{
|
|
"epoch": 4.521531100478469,
|
|
"grad_norm": 0.5316043989251976,
|
|
"learning_rate": 1.3442700098966326e-05,
|
|
"loss": 0.1532,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13952653110027313,
|
|
"step": 2835,
|
|
"valid_targets_mean": 4084.7,
|
|
"valid_targets_min": 1625
|
|
},
|
|
{
|
|
"epoch": 4.529505582137161,
|
|
"grad_norm": 0.6131005374748497,
|
|
"learning_rate": 1.3367614428958646e-05,
|
|
"loss": 0.1448,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1712581217288971,
|
|
"step": 2840,
|
|
"valid_targets_mean": 4815.7,
|
|
"valid_targets_min": 738
|
|
},
|
|
{
|
|
"epoch": 4.5374800637958534,
|
|
"grad_norm": 0.511066316660687,
|
|
"learning_rate": 1.3292633644282735e-05,
|
|
"loss": 0.1449,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14669950306415558,
|
|
"step": 2845,
|
|
"valid_targets_mean": 3795.1,
|
|
"valid_targets_min": 811
|
|
},
|
|
{
|
|
"epoch": 4.545454545454545,
|
|
"grad_norm": 0.44495333907690426,
|
|
"learning_rate": 1.3217758930693608e-05,
|
|
"loss": 0.1628,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14287234842777252,
|
|
"step": 2850,
|
|
"valid_targets_mean": 4819.1,
|
|
"valid_targets_min": 1363
|
|
},
|
|
{
|
|
"epoch": 4.553429027113237,
|
|
"grad_norm": 0.4803993591600635,
|
|
"learning_rate": 1.314299147226887e-05,
|
|
"loss": 0.1582,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1529885232448578,
|
|
"step": 2855,
|
|
"valid_targets_mean": 4219.8,
|
|
"valid_targets_min": 1637
|
|
},
|
|
{
|
|
"epoch": 4.56140350877193,
|
|
"grad_norm": 0.534456680714764,
|
|
"learning_rate": 1.3068332451389969e-05,
|
|
"loss": 0.1644,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18377503752708435,
|
|
"step": 2860,
|
|
"valid_targets_mean": 4056.7,
|
|
"valid_targets_min": 1252
|
|
},
|
|
{
|
|
"epoch": 4.569377990430622,
|
|
"grad_norm": 0.5715320414560933,
|
|
"learning_rate": 1.2993783048723515e-05,
|
|
"loss": 0.1493,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14145058393478394,
|
|
"step": 2865,
|
|
"valid_targets_mean": 2752.7,
|
|
"valid_targets_min": 634
|
|
},
|
|
{
|
|
"epoch": 4.577352472089315,
|
|
"grad_norm": 0.490531412083574,
|
|
"learning_rate": 1.2919344443202602e-05,
|
|
"loss": 0.1584,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13521042466163635,
|
|
"step": 2870,
|
|
"valid_targets_mean": 3873.3,
|
|
"valid_targets_min": 1432
|
|
},
|
|
{
|
|
"epoch": 4.585326953748006,
|
|
"grad_norm": 0.4960382986414196,
|
|
"learning_rate": 1.2845017812008158e-05,
|
|
"loss": 0.1466,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13930058479309082,
|
|
"step": 2875,
|
|
"valid_targets_mean": 4009.3,
|
|
"valid_targets_min": 1653
|
|
},
|
|
{
|
|
"epoch": 4.5933014354066986,
|
|
"grad_norm": 0.45490882957573536,
|
|
"learning_rate": 1.277080433055034e-05,
|
|
"loss": 0.1659,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16347578167915344,
|
|
"step": 2880,
|
|
"valid_targets_mean": 4968.5,
|
|
"valid_targets_min": 2414
|
|
},
|
|
{
|
|
"epoch": 4.601275917065391,
|
|
"grad_norm": 0.49053829703055934,
|
|
"learning_rate": 1.2696705172449944e-05,
|
|
"loss": 0.159,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17954492568969727,
|
|
"step": 2885,
|
|
"valid_targets_mean": 4561.9,
|
|
"valid_targets_min": 937
|
|
},
|
|
{
|
|
"epoch": 4.6092503987240825,
|
|
"grad_norm": 0.49085289338663335,
|
|
"learning_rate": 1.2622721509519846e-05,
|
|
"loss": 0.1508,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14600206911563873,
|
|
"step": 2890,
|
|
"valid_targets_mean": 4638.7,
|
|
"valid_targets_min": 1362
|
|
},
|
|
{
|
|
"epoch": 4.617224880382775,
|
|
"grad_norm": 0.4440863342043673,
|
|
"learning_rate": 1.2548854511746461e-05,
|
|
"loss": 0.1605,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1403721123933792,
|
|
"step": 2895,
|
|
"valid_targets_mean": 4791.0,
|
|
"valid_targets_min": 1727
|
|
},
|
|
{
|
|
"epoch": 4.625199362041467,
|
|
"grad_norm": 0.4290742285274184,
|
|
"learning_rate": 1.247510534727125e-05,
|
|
"loss": 0.1656,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14262665808200836,
|
|
"step": 2900,
|
|
"valid_targets_mean": 4883.1,
|
|
"valid_targets_min": 1800
|
|
},
|
|
{
|
|
"epoch": 4.63317384370016,
|
|
"grad_norm": 0.48233893104018916,
|
|
"learning_rate": 1.240147518237224e-05,
|
|
"loss": 0.1614,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1487930715084076,
|
|
"step": 2905,
|
|
"valid_targets_mean": 4299.9,
|
|
"valid_targets_min": 720
|
|
},
|
|
{
|
|
"epoch": 4.641148325358852,
|
|
"grad_norm": 0.49119003845245157,
|
|
"learning_rate": 1.2327965181445593e-05,
|
|
"loss": 0.1546,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15125395357608795,
|
|
"step": 2910,
|
|
"valid_targets_mean": 4226.4,
|
|
"valid_targets_min": 412
|
|
},
|
|
{
|
|
"epoch": 4.649122807017544,
|
|
"grad_norm": 0.5514755092195026,
|
|
"learning_rate": 1.2254576506987182e-05,
|
|
"loss": 0.1714,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.20615872740745544,
|
|
"step": 2915,
|
|
"valid_targets_mean": 3670.6,
|
|
"valid_targets_min": 287
|
|
},
|
|
{
|
|
"epoch": 4.657097288676236,
|
|
"grad_norm": 0.4836449422046179,
|
|
"learning_rate": 1.21813103195742e-05,
|
|
"loss": 0.1547,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1694192886352539,
|
|
"step": 2920,
|
|
"valid_targets_mean": 4252.4,
|
|
"valid_targets_min": 1091
|
|
},
|
|
{
|
|
"epoch": 4.6650717703349285,
|
|
"grad_norm": 0.5227248323901297,
|
|
"learning_rate": 1.2108167777846815e-05,
|
|
"loss": 0.1571,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1699051558971405,
|
|
"step": 2925,
|
|
"valid_targets_mean": 4134.8,
|
|
"valid_targets_min": 716
|
|
},
|
|
{
|
|
"epoch": 4.67304625199362,
|
|
"grad_norm": 0.4660658722544293,
|
|
"learning_rate": 1.203515003848987e-05,
|
|
"loss": 0.163,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15797314047813416,
|
|
"step": 2930,
|
|
"valid_targets_mean": 4928.8,
|
|
"valid_targets_min": 1582
|
|
},
|
|
{
|
|
"epoch": 4.681020733652312,
|
|
"grad_norm": 0.6376936646502744,
|
|
"learning_rate": 1.1962258256214545e-05,
|
|
"loss": 0.1426,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14403367042541504,
|
|
"step": 2935,
|
|
"valid_targets_mean": 3905.9,
|
|
"valid_targets_min": 766
|
|
},
|
|
{
|
|
"epoch": 4.688995215311005,
|
|
"grad_norm": 0.4353897798410051,
|
|
"learning_rate": 1.1889493583740124e-05,
|
|
"loss": 0.1651,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17408692836761475,
|
|
"step": 2940,
|
|
"valid_targets_mean": 5558.9,
|
|
"valid_targets_min": 1941
|
|
},
|
|
{
|
|
"epoch": 4.696969696969697,
|
|
"grad_norm": 0.5138974967310053,
|
|
"learning_rate": 1.1816857171775767e-05,
|
|
"loss": 0.1566,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18357430398464203,
|
|
"step": 2945,
|
|
"valid_targets_mean": 4750.4,
|
|
"valid_targets_min": 2115
|
|
},
|
|
{
|
|
"epoch": 4.70494417862839,
|
|
"grad_norm": 0.5058909214087609,
|
|
"learning_rate": 1.1744350169002308e-05,
|
|
"loss": 0.151,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1348288655281067,
|
|
"step": 2950,
|
|
"valid_targets_mean": 3484.4,
|
|
"valid_targets_min": 1397
|
|
},
|
|
{
|
|
"epoch": 4.712918660287081,
|
|
"grad_norm": 0.49111490538221414,
|
|
"learning_rate": 1.1671973722054077e-05,
|
|
"loss": 0.1488,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1464148610830307,
|
|
"step": 2955,
|
|
"valid_targets_mean": 4043.7,
|
|
"valid_targets_min": 1224
|
|
},
|
|
{
|
|
"epoch": 4.720893141945774,
|
|
"grad_norm": 0.551693194914327,
|
|
"learning_rate": 1.159972897550079e-05,
|
|
"loss": 0.1649,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18348699808120728,
|
|
"step": 2960,
|
|
"valid_targets_mean": 3622.8,
|
|
"valid_targets_min": 1035
|
|
},
|
|
{
|
|
"epoch": 4.728867623604466,
|
|
"grad_norm": 0.48537829529174753,
|
|
"learning_rate": 1.1527617071829447e-05,
|
|
"loss": 0.1663,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12741652131080627,
|
|
"step": 2965,
|
|
"valid_targets_mean": 4625.4,
|
|
"valid_targets_min": 1135
|
|
},
|
|
{
|
|
"epoch": 4.7368421052631575,
|
|
"grad_norm": 0.5167280393109993,
|
|
"learning_rate": 1.145563915142622e-05,
|
|
"loss": 0.1608,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1738428771495819,
|
|
"step": 2970,
|
|
"valid_targets_mean": 4045.1,
|
|
"valid_targets_min": 1934
|
|
},
|
|
{
|
|
"epoch": 4.74481658692185,
|
|
"grad_norm": 0.46590567459493326,
|
|
"learning_rate": 1.1383796352558489e-05,
|
|
"loss": 0.1589,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1390235275030136,
|
|
"step": 2975,
|
|
"valid_targets_mean": 4557.5,
|
|
"valid_targets_min": 377
|
|
},
|
|
{
|
|
"epoch": 4.752791068580542,
|
|
"grad_norm": 0.45132284661428174,
|
|
"learning_rate": 1.1312089811356803e-05,
|
|
"loss": 0.163,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14673513174057007,
|
|
"step": 2980,
|
|
"valid_targets_mean": 4593.8,
|
|
"valid_targets_min": 1424
|
|
},
|
|
{
|
|
"epoch": 4.760765550239235,
|
|
"grad_norm": 0.5190466302755331,
|
|
"learning_rate": 1.1240520661796882e-05,
|
|
"loss": 0.1515,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15376582741737366,
|
|
"step": 2985,
|
|
"valid_targets_mean": 3806.1,
|
|
"valid_targets_min": 858
|
|
},
|
|
{
|
|
"epoch": 4.768740031897926,
|
|
"grad_norm": 0.48812938806701267,
|
|
"learning_rate": 1.1169090035681772e-05,
|
|
"loss": 0.157,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14771802723407745,
|
|
"step": 2990,
|
|
"valid_targets_mean": 4181.2,
|
|
"valid_targets_min": 1647
|
|
},
|
|
{
|
|
"epoch": 4.776714513556619,
|
|
"grad_norm": 0.5183768837688895,
|
|
"learning_rate": 1.109779906262383e-05,
|
|
"loss": 0.1589,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17195749282836914,
|
|
"step": 2995,
|
|
"valid_targets_mean": 4079.7,
|
|
"valid_targets_min": 1639
|
|
},
|
|
{
|
|
"epoch": 4.784688995215311,
|
|
"grad_norm": 0.7151241582202109,
|
|
"learning_rate": 1.102664887002698e-05,
|
|
"loss": 0.1602,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1477375328540802,
|
|
"step": 3000,
|
|
"valid_targets_mean": 3195.2,
|
|
"valid_targets_min": 1061
|
|
},
|
|
{
|
|
"epoch": 4.7926634768740035,
|
|
"grad_norm": 0.5340265757646905,
|
|
"learning_rate": 1.0955640583068802e-05,
|
|
"loss": 0.1511,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15154235064983368,
|
|
"step": 3005,
|
|
"valid_targets_mean": 3735.8,
|
|
"valid_targets_min": 1180
|
|
},
|
|
{
|
|
"epoch": 4.800637958532695,
|
|
"grad_norm": 0.5191731924116039,
|
|
"learning_rate": 1.0884775324682755e-05,
|
|
"loss": 0.1539,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15678870677947998,
|
|
"step": 3010,
|
|
"valid_targets_mean": 4505.0,
|
|
"valid_targets_min": 1436
|
|
},
|
|
{
|
|
"epoch": 4.8086124401913874,
|
|
"grad_norm": 0.577257395101735,
|
|
"learning_rate": 1.081405421554044e-05,
|
|
"loss": 0.1529,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15045231580734253,
|
|
"step": 3015,
|
|
"valid_targets_mean": 3325.9,
|
|
"valid_targets_min": 349
|
|
},
|
|
{
|
|
"epoch": 4.81658692185008,
|
|
"grad_norm": 0.47952301871619524,
|
|
"learning_rate": 1.074347837403387e-05,
|
|
"loss": 0.1703,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15675462782382965,
|
|
"step": 3020,
|
|
"valid_targets_mean": 4365.0,
|
|
"valid_targets_min": 2000
|
|
},
|
|
{
|
|
"epoch": 4.824561403508772,
|
|
"grad_norm": 0.5171197959497484,
|
|
"learning_rate": 1.067304891625776e-05,
|
|
"loss": 0.1716,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.21107935905456543,
|
|
"step": 3025,
|
|
"valid_targets_mean": 4033.0,
|
|
"valid_targets_min": 1431
|
|
},
|
|
{
|
|
"epoch": 4.832535885167464,
|
|
"grad_norm": 0.457639324980029,
|
|
"learning_rate": 1.0602766955991913e-05,
|
|
"loss": 0.1591,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1720237135887146,
|
|
"step": 3030,
|
|
"valid_targets_mean": 4731.8,
|
|
"valid_targets_min": 1708
|
|
},
|
|
{
|
|
"epoch": 4.840510366826156,
|
|
"grad_norm": 0.48700471070095636,
|
|
"learning_rate": 1.053263360468358e-05,
|
|
"loss": 0.1595,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14990770816802979,
|
|
"step": 3035,
|
|
"valid_targets_mean": 4321.4,
|
|
"valid_targets_min": 271
|
|
},
|
|
{
|
|
"epoch": 4.848484848484849,
|
|
"grad_norm": 0.4333197612254627,
|
|
"learning_rate": 1.0462649971429884e-05,
|
|
"loss": 0.1595,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14838045835494995,
|
|
"step": 3040,
|
|
"valid_targets_mean": 5209.8,
|
|
"valid_targets_min": 1739
|
|
},
|
|
{
|
|
"epoch": 4.856459330143541,
|
|
"grad_norm": 0.4874500871691921,
|
|
"learning_rate": 1.0392817162960304e-05,
|
|
"loss": 0.1412,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13294383883476257,
|
|
"step": 3045,
|
|
"valid_targets_mean": 4401.8,
|
|
"valid_targets_min": 907
|
|
},
|
|
{
|
|
"epoch": 4.8644338118022326,
|
|
"grad_norm": 0.5503368290230514,
|
|
"learning_rate": 1.0323136283619167e-05,
|
|
"loss": 0.1604,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18634894490242004,
|
|
"step": 3050,
|
|
"valid_targets_mean": 3853.9,
|
|
"valid_targets_min": 2069
|
|
},
|
|
{
|
|
"epoch": 4.872408293460925,
|
|
"grad_norm": 0.5438858738387928,
|
|
"learning_rate": 1.0253608435348136e-05,
|
|
"loss": 0.1571,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17999961972236633,
|
|
"step": 3055,
|
|
"valid_targets_mean": 4998.1,
|
|
"valid_targets_min": 1754
|
|
},
|
|
{
|
|
"epoch": 4.880382775119617,
|
|
"grad_norm": 0.484007746641252,
|
|
"learning_rate": 1.0184234717668867e-05,
|
|
"loss": 0.1584,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14526879787445068,
|
|
"step": 3060,
|
|
"valid_targets_mean": 4370.1,
|
|
"valid_targets_min": 2135
|
|
},
|
|
{
|
|
"epoch": 4.88835725677831,
|
|
"grad_norm": 0.4798298193869541,
|
|
"learning_rate": 1.0115016227665544e-05,
|
|
"loss": 0.1515,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17290471494197845,
|
|
"step": 3065,
|
|
"valid_targets_mean": 4554.9,
|
|
"valid_targets_min": 690
|
|
},
|
|
{
|
|
"epoch": 4.896331738437001,
|
|
"grad_norm": 0.4847848102732291,
|
|
"learning_rate": 1.0045954059967577e-05,
|
|
"loss": 0.1445,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1566402018070221,
|
|
"step": 3070,
|
|
"valid_targets_mean": 4250.6,
|
|
"valid_targets_min": 1662
|
|
},
|
|
{
|
|
"epoch": 4.904306220095694,
|
|
"grad_norm": 0.5079181491087512,
|
|
"learning_rate": 9.977049306732287e-06,
|
|
"loss": 0.1538,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15510109066963196,
|
|
"step": 3075,
|
|
"valid_targets_mean": 4287.0,
|
|
"valid_targets_min": 1349
|
|
},
|
|
{
|
|
"epoch": 4.912280701754386,
|
|
"grad_norm": 0.49369228412711796,
|
|
"learning_rate": 9.908303057627591e-06,
|
|
"loss": 0.175,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1516781747341156,
|
|
"step": 3080,
|
|
"valid_targets_mean": 4807.1,
|
|
"valid_targets_min": 1399
|
|
},
|
|
{
|
|
"epoch": 4.920255183413078,
|
|
"grad_norm": 0.46206631563701656,
|
|
"learning_rate": 9.83971639981484e-06,
|
|
"loss": 0.1608,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15620173513889313,
|
|
"step": 3085,
|
|
"valid_targets_mean": 4509.9,
|
|
"valid_targets_min": 1755
|
|
},
|
|
{
|
|
"epoch": 4.92822966507177,
|
|
"grad_norm": 0.4764073964647667,
|
|
"learning_rate": 9.771290417931559e-06,
|
|
"loss": 0.1432,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14277634024620056,
|
|
"step": 3090,
|
|
"valid_targets_mean": 4633.9,
|
|
"valid_targets_min": 540
|
|
},
|
|
{
|
|
"epoch": 4.9362041467304625,
|
|
"grad_norm": 0.499728819601026,
|
|
"learning_rate": 9.703026194074342e-06,
|
|
"loss": 0.1608,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14103826880455017,
|
|
"step": 3095,
|
|
"valid_targets_mean": 4783.8,
|
|
"valid_targets_min": 1889
|
|
},
|
|
{
|
|
"epoch": 4.944178628389155,
|
|
"grad_norm": 0.5030379720704173,
|
|
"learning_rate": 9.634924807781729e-06,
|
|
"loss": 0.1576,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18518777191638947,
|
|
"step": 3100,
|
|
"valid_targets_mean": 4487.1,
|
|
"valid_targets_min": 1345
|
|
},
|
|
{
|
|
"epoch": 4.952153110047847,
|
|
"grad_norm": 0.5312971046827829,
|
|
"learning_rate": 9.566987336017102e-06,
|
|
"loss": 0.1532,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13996370136737823,
|
|
"step": 3105,
|
|
"valid_targets_mean": 3201.3,
|
|
"valid_targets_min": 1367
|
|
},
|
|
{
|
|
"epoch": 4.960127591706539,
|
|
"grad_norm": 0.47987309157405533,
|
|
"learning_rate": 9.499214853151699e-06,
|
|
"loss": 0.1494,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.142959326505661,
|
|
"step": 3110,
|
|
"valid_targets_mean": 4422.2,
|
|
"valid_targets_min": 1032
|
|
},
|
|
{
|
|
"epoch": 4.968102073365231,
|
|
"grad_norm": 0.5664816415312744,
|
|
"learning_rate": 9.431608430947619e-06,
|
|
"loss": 0.1663,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.176031231880188,
|
|
"step": 3115,
|
|
"valid_targets_mean": 4770.7,
|
|
"valid_targets_min": 891
|
|
},
|
|
{
|
|
"epoch": 4.976076555023924,
|
|
"grad_norm": 0.5518137212422529,
|
|
"learning_rate": 9.364169138540805e-06,
|
|
"loss": 0.1511,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15234243869781494,
|
|
"step": 3120,
|
|
"valid_targets_mean": 4083.6,
|
|
"valid_targets_min": 1519
|
|
},
|
|
{
|
|
"epoch": 4.984051036682615,
|
|
"grad_norm": 0.4682479995624355,
|
|
"learning_rate": 9.296898042424237e-06,
|
|
"loss": 0.1653,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16865205764770508,
|
|
"step": 3125,
|
|
"valid_targets_mean": 4818.3,
|
|
"valid_targets_min": 1142
|
|
},
|
|
{
|
|
"epoch": 4.992025518341308,
|
|
"grad_norm": 0.5040455659035196,
|
|
"learning_rate": 9.229796206431015e-06,
|
|
"loss": 0.1596,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14097434282302856,
|
|
"step": 3130,
|
|
"valid_targets_mean": 4541.0,
|
|
"valid_targets_min": 1290
|
|
},
|
|
{
|
|
"epoch": 5.0,
|
|
"grad_norm": 0.47913131975592566,
|
|
"learning_rate": 9.162864691717513e-06,
|
|
"loss": 0.1626,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1694181263446808,
|
|
"step": 3135,
|
|
"valid_targets_mean": 4800.7,
|
|
"valid_targets_min": 1983
|
|
},
|
|
{
|
|
"epoch": 5.007974481658692,
|
|
"grad_norm": 0.4631690140133372,
|
|
"learning_rate": 9.096104556746654e-06,
|
|
"loss": 0.1409,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1334855705499649,
|
|
"step": 3140,
|
|
"valid_targets_mean": 4212.7,
|
|
"valid_targets_min": 1470
|
|
},
|
|
{
|
|
"epoch": 5.015948963317384,
|
|
"grad_norm": 0.4738719091728486,
|
|
"learning_rate": 9.029516857271115e-06,
|
|
"loss": 0.1505,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15238399803638458,
|
|
"step": 3145,
|
|
"valid_targets_mean": 4365.6,
|
|
"valid_targets_min": 861
|
|
},
|
|
{
|
|
"epoch": 5.023923444976076,
|
|
"grad_norm": 0.5490118882632362,
|
|
"learning_rate": 8.963102646316677e-06,
|
|
"loss": 0.1425,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1590614765882492,
|
|
"step": 3150,
|
|
"valid_targets_mean": 3971.9,
|
|
"valid_targets_min": 1032
|
|
},
|
|
{
|
|
"epoch": 5.031897926634769,
|
|
"grad_norm": 0.5723827642772381,
|
|
"learning_rate": 8.896862974165553e-06,
|
|
"loss": 0.1451,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1540474146604538,
|
|
"step": 3155,
|
|
"valid_targets_mean": 3627.2,
|
|
"valid_targets_min": 1479
|
|
},
|
|
{
|
|
"epoch": 5.039872408293461,
|
|
"grad_norm": 0.47302389032511105,
|
|
"learning_rate": 8.830798888339756e-06,
|
|
"loss": 0.1445,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14463520050048828,
|
|
"step": 3160,
|
|
"valid_targets_mean": 4906.4,
|
|
"valid_targets_min": 1707
|
|
},
|
|
{
|
|
"epoch": 5.047846889952153,
|
|
"grad_norm": 0.5469416912728198,
|
|
"learning_rate": 8.764911433584581e-06,
|
|
"loss": 0.1423,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1377723217010498,
|
|
"step": 3165,
|
|
"valid_targets_mean": 3464.6,
|
|
"valid_targets_min": 578
|
|
},
|
|
{
|
|
"epoch": 5.055821371610845,
|
|
"grad_norm": 0.5582713551464368,
|
|
"learning_rate": 8.699201651852056e-06,
|
|
"loss": 0.1426,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14744654297828674,
|
|
"step": 3170,
|
|
"valid_targets_mean": 4384.8,
|
|
"valid_targets_min": 455
|
|
},
|
|
{
|
|
"epoch": 5.0637958532695375,
|
|
"grad_norm": 0.4525826677728015,
|
|
"learning_rate": 8.633670582284446e-06,
|
|
"loss": 0.1552,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12246271222829819,
|
|
"step": 3175,
|
|
"valid_targets_mean": 5568.0,
|
|
"valid_targets_min": 1686
|
|
},
|
|
{
|
|
"epoch": 5.07177033492823,
|
|
"grad_norm": 0.4802895756565889,
|
|
"learning_rate": 8.56831926119787e-06,
|
|
"loss": 0.1481,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13796019554138184,
|
|
"step": 3180,
|
|
"valid_targets_mean": 4717.9,
|
|
"valid_targets_min": 883
|
|
},
|
|
{
|
|
"epoch": 5.0797448165869215,
|
|
"grad_norm": 0.47933543045230076,
|
|
"learning_rate": 8.503148722065851e-06,
|
|
"loss": 0.1485,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1501651108264923,
|
|
"step": 3185,
|
|
"valid_targets_mean": 4850.2,
|
|
"valid_targets_min": 792
|
|
},
|
|
{
|
|
"epoch": 5.087719298245614,
|
|
"grad_norm": 0.46167193488495856,
|
|
"learning_rate": 8.43815999550303e-06,
|
|
"loss": 0.142,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17034366726875305,
|
|
"step": 3190,
|
|
"valid_targets_mean": 5809.8,
|
|
"valid_targets_min": 2272
|
|
},
|
|
{
|
|
"epoch": 5.095693779904306,
|
|
"grad_norm": 0.5147068478030644,
|
|
"learning_rate": 8.373354109248842e-06,
|
|
"loss": 0.1362,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13025915622711182,
|
|
"step": 3195,
|
|
"valid_targets_mean": 3514.9,
|
|
"valid_targets_min": 1532
|
|
},
|
|
{
|
|
"epoch": 5.103668261562999,
|
|
"grad_norm": 0.4918615867478781,
|
|
"learning_rate": 8.308732088151245e-06,
|
|
"loss": 0.1464,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16296672821044922,
|
|
"step": 3200,
|
|
"valid_targets_mean": 5018.5,
|
|
"valid_targets_min": 1439
|
|
},
|
|
{
|
|
"epoch": 5.11164274322169,
|
|
"grad_norm": 0.5788923126986796,
|
|
"learning_rate": 8.24429495415054e-06,
|
|
"loss": 0.1431,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1460484266281128,
|
|
"step": 3205,
|
|
"valid_targets_mean": 4154.6,
|
|
"valid_targets_min": 670
|
|
},
|
|
{
|
|
"epoch": 5.119617224880383,
|
|
"grad_norm": 0.4553069456977529,
|
|
"learning_rate": 8.180043726263216e-06,
|
|
"loss": 0.1408,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12702570855617523,
|
|
"step": 3210,
|
|
"valid_targets_mean": 4684.9,
|
|
"valid_targets_min": 2310
|
|
},
|
|
{
|
|
"epoch": 5.127591706539075,
|
|
"grad_norm": 0.5218901188519967,
|
|
"learning_rate": 8.115979420565794e-06,
|
|
"loss": 0.1567,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14124715328216553,
|
|
"step": 3215,
|
|
"valid_targets_mean": 3825.7,
|
|
"valid_targets_min": 1939
|
|
},
|
|
{
|
|
"epoch": 5.1355661881977674,
|
|
"grad_norm": 0.455006451357129,
|
|
"learning_rate": 8.052103050178806e-06,
|
|
"loss": 0.1368,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13781721889972687,
|
|
"step": 3220,
|
|
"valid_targets_mean": 4437.4,
|
|
"valid_targets_min": 1007
|
|
},
|
|
{
|
|
"epoch": 5.143540669856459,
|
|
"grad_norm": 0.5047251677784329,
|
|
"learning_rate": 7.988415625250755e-06,
|
|
"loss": 0.131,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14169715344905853,
|
|
"step": 3225,
|
|
"valid_targets_mean": 3893.2,
|
|
"valid_targets_min": 1632
|
|
},
|
|
{
|
|
"epoch": 5.151515151515151,
|
|
"grad_norm": 0.48563253677866414,
|
|
"learning_rate": 7.924918152942117e-06,
|
|
"loss": 0.1393,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13769680261611938,
|
|
"step": 3230,
|
|
"valid_targets_mean": 4708.1,
|
|
"valid_targets_min": 2084
|
|
},
|
|
{
|
|
"epoch": 5.159489633173844,
|
|
"grad_norm": 0.49252720325100796,
|
|
"learning_rate": 7.861611637409462e-06,
|
|
"loss": 0.1427,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1714155673980713,
|
|
"step": 3235,
|
|
"valid_targets_mean": 4988.3,
|
|
"valid_targets_min": 1499
|
|
},
|
|
{
|
|
"epoch": 5.167464114832536,
|
|
"grad_norm": 0.564418000148086,
|
|
"learning_rate": 7.798497079789513e-06,
|
|
"loss": 0.143,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14852869510650635,
|
|
"step": 3240,
|
|
"valid_targets_mean": 3508.2,
|
|
"valid_targets_min": 1539
|
|
},
|
|
{
|
|
"epoch": 5.175438596491228,
|
|
"grad_norm": 0.5346686667466659,
|
|
"learning_rate": 7.735575478183381e-06,
|
|
"loss": 0.1493,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15654098987579346,
|
|
"step": 3245,
|
|
"valid_targets_mean": 4420.9,
|
|
"valid_targets_min": 1403
|
|
},
|
|
{
|
|
"epoch": 5.18341307814992,
|
|
"grad_norm": 0.5129242426435193,
|
|
"learning_rate": 7.672847827640735e-06,
|
|
"loss": 0.1582,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1257268786430359,
|
|
"step": 3250,
|
|
"valid_targets_mean": 3798.2,
|
|
"valid_targets_min": 632
|
|
},
|
|
{
|
|
"epoch": 5.1913875598086126,
|
|
"grad_norm": 0.5278462460463896,
|
|
"learning_rate": 7.610315120144067e-06,
|
|
"loss": 0.1417,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1487414836883545,
|
|
"step": 3255,
|
|
"valid_targets_mean": 3692.8,
|
|
"valid_targets_min": 1770
|
|
},
|
|
{
|
|
"epoch": 5.199362041467305,
|
|
"grad_norm": 0.5164937797343533,
|
|
"learning_rate": 7.5479783445930414e-06,
|
|
"loss": 0.1448,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14907556772232056,
|
|
"step": 3260,
|
|
"valid_targets_mean": 4389.5,
|
|
"valid_targets_min": 1326
|
|
},
|
|
{
|
|
"epoch": 5.2073365231259965,
|
|
"grad_norm": 0.5605235918702897,
|
|
"learning_rate": 7.485838486788803e-06,
|
|
"loss": 0.1368,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15365715324878693,
|
|
"step": 3265,
|
|
"valid_targets_mean": 3910.3,
|
|
"valid_targets_min": 510
|
|
},
|
|
{
|
|
"epoch": 5.215311004784689,
|
|
"grad_norm": 0.4804446704770563,
|
|
"learning_rate": 7.4238965294184374e-06,
|
|
"loss": 0.1386,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11510901153087616,
|
|
"step": 3270,
|
|
"valid_targets_mean": 4616.4,
|
|
"valid_targets_min": 2092
|
|
},
|
|
{
|
|
"epoch": 5.223285486443381,
|
|
"grad_norm": 0.48788340937166225,
|
|
"learning_rate": 7.362153452039409e-06,
|
|
"loss": 0.1532,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12575878202915192,
|
|
"step": 3275,
|
|
"valid_targets_mean": 4244.5,
|
|
"valid_targets_min": 2169
|
|
},
|
|
{
|
|
"epoch": 5.231259968102074,
|
|
"grad_norm": 0.4998963625738302,
|
|
"learning_rate": 7.300610231064056e-06,
|
|
"loss": 0.153,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1369197964668274,
|
|
"step": 3280,
|
|
"valid_targets_mean": 4486.5,
|
|
"valid_targets_min": 1157
|
|
},
|
|
{
|
|
"epoch": 5.239234449760765,
|
|
"grad_norm": 0.5597438659865254,
|
|
"learning_rate": 7.239267839744166e-06,
|
|
"loss": 0.144,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13235260546207428,
|
|
"step": 3285,
|
|
"valid_targets_mean": 3675.8,
|
|
"valid_targets_min": 1767
|
|
},
|
|
{
|
|
"epoch": 5.247208931419458,
|
|
"grad_norm": 0.4637157957912727,
|
|
"learning_rate": 7.178127248155604e-06,
|
|
"loss": 0.14,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12913677096366882,
|
|
"step": 3290,
|
|
"valid_targets_mean": 5069.9,
|
|
"valid_targets_min": 2045
|
|
},
|
|
{
|
|
"epoch": 5.25518341307815,
|
|
"grad_norm": 0.5234421904422631,
|
|
"learning_rate": 7.117189423182917e-06,
|
|
"loss": 0.1421,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1609647274017334,
|
|
"step": 3295,
|
|
"valid_targets_mean": 4219.6,
|
|
"valid_targets_min": 1568
|
|
},
|
|
{
|
|
"epoch": 5.2631578947368425,
|
|
"grad_norm": 0.8939060446670861,
|
|
"learning_rate": 7.056455328504104e-06,
|
|
"loss": 0.1501,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17754587531089783,
|
|
"step": 3300,
|
|
"valid_targets_mean": 4179.9,
|
|
"valid_targets_min": 1021
|
|
},
|
|
{
|
|
"epoch": 5.271132376395534,
|
|
"grad_norm": 0.5613035580302348,
|
|
"learning_rate": 6.995925924575342e-06,
|
|
"loss": 0.1352,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14372298121452332,
|
|
"step": 3305,
|
|
"valid_targets_mean": 3731.8,
|
|
"valid_targets_min": 1233
|
|
},
|
|
{
|
|
"epoch": 5.279106858054226,
|
|
"grad_norm": 0.4154591658982006,
|
|
"learning_rate": 6.935602168615792e-06,
|
|
"loss": 0.147,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12290593981742859,
|
|
"step": 3310,
|
|
"valid_targets_mean": 5637.1,
|
|
"valid_targets_min": 1673
|
|
},
|
|
{
|
|
"epoch": 5.287081339712919,
|
|
"grad_norm": 0.6416800447593414,
|
|
"learning_rate": 6.875485014592493e-06,
|
|
"loss": 0.1553,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1722438931465149,
|
|
"step": 3315,
|
|
"valid_targets_mean": 4116.6,
|
|
"valid_targets_min": 1508
|
|
},
|
|
{
|
|
"epoch": 5.295055821371611,
|
|
"grad_norm": 0.5871320560825972,
|
|
"learning_rate": 6.815575413205235e-06,
|
|
"loss": 0.1466,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13305939733982086,
|
|
"step": 3320,
|
|
"valid_targets_mean": 3895.7,
|
|
"valid_targets_min": 416
|
|
},
|
|
{
|
|
"epoch": 5.303030303030303,
|
|
"grad_norm": 0.5007068898056246,
|
|
"learning_rate": 6.755874311871562e-06,
|
|
"loss": 0.1451,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12426762282848358,
|
|
"step": 3325,
|
|
"valid_targets_mean": 3973.1,
|
|
"valid_targets_min": 1233
|
|
},
|
|
{
|
|
"epoch": 5.311004784688995,
|
|
"grad_norm": 0.5044473336922294,
|
|
"learning_rate": 6.696382654711777e-06,
|
|
"loss": 0.1306,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13455146551132202,
|
|
"step": 3330,
|
|
"valid_targets_mean": 3863.1,
|
|
"valid_targets_min": 1300
|
|
},
|
|
{
|
|
"epoch": 5.318979266347688,
|
|
"grad_norm": 0.4845451776935124,
|
|
"learning_rate": 6.637101382533986e-06,
|
|
"loss": 0.1522,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15252287685871124,
|
|
"step": 3335,
|
|
"valid_targets_mean": 4714.7,
|
|
"valid_targets_min": 1551
|
|
},
|
|
{
|
|
"epoch": 5.32695374800638,
|
|
"grad_norm": 0.49995222258967786,
|
|
"learning_rate": 6.578031432819263e-06,
|
|
"loss": 0.1412,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13115063309669495,
|
|
"step": 3340,
|
|
"valid_targets_mean": 3895.4,
|
|
"valid_targets_min": 342
|
|
},
|
|
{
|
|
"epoch": 5.3349282296650715,
|
|
"grad_norm": 0.5147718024489065,
|
|
"learning_rate": 6.5191737397068015e-06,
|
|
"loss": 0.146,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1277836561203003,
|
|
"step": 3345,
|
|
"valid_targets_mean": 4424.9,
|
|
"valid_targets_min": 1402
|
|
},
|
|
{
|
|
"epoch": 5.342902711323764,
|
|
"grad_norm": 0.581498600134185,
|
|
"learning_rate": 6.460529233979127e-06,
|
|
"loss": 0.1432,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13812121748924255,
|
|
"step": 3350,
|
|
"valid_targets_mean": 3206.0,
|
|
"valid_targets_min": 1569
|
|
},
|
|
{
|
|
"epoch": 5.350877192982456,
|
|
"grad_norm": 0.44412427640952046,
|
|
"learning_rate": 6.402098843047417e-06,
|
|
"loss": 0.1408,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14585547149181366,
|
|
"step": 3355,
|
|
"valid_targets_mean": 5636.0,
|
|
"valid_targets_min": 2343
|
|
},
|
|
{
|
|
"epoch": 5.358851674641148,
|
|
"grad_norm": 0.46631543622836114,
|
|
"learning_rate": 6.343883490936791e-06,
|
|
"loss": 0.1419,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1294684112071991,
|
|
"step": 3360,
|
|
"valid_targets_mean": 5182.2,
|
|
"valid_targets_min": 1761
|
|
},
|
|
{
|
|
"epoch": 5.36682615629984,
|
|
"grad_norm": 0.7967138290866428,
|
|
"learning_rate": 6.285884098271739e-06,
|
|
"loss": 0.1386,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13547652959823608,
|
|
"step": 3365,
|
|
"valid_targets_mean": 4249.9,
|
|
"valid_targets_min": 738
|
|
},
|
|
{
|
|
"epoch": 5.374800637958533,
|
|
"grad_norm": 0.5636272491598279,
|
|
"learning_rate": 6.228101582261532e-06,
|
|
"loss": 0.1496,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13094601035118103,
|
|
"step": 3370,
|
|
"valid_targets_mean": 4631.6,
|
|
"valid_targets_min": 692
|
|
},
|
|
{
|
|
"epoch": 5.382775119617225,
|
|
"grad_norm": 0.5018163908163288,
|
|
"learning_rate": 6.170536856685716e-06,
|
|
"loss": 0.1574,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17899195849895477,
|
|
"step": 3375,
|
|
"valid_targets_mean": 5286.1,
|
|
"valid_targets_min": 2007
|
|
},
|
|
{
|
|
"epoch": 5.3907496012759175,
|
|
"grad_norm": 0.5181322415866998,
|
|
"learning_rate": 6.113190831879698e-06,
|
|
"loss": 0.1396,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13492557406425476,
|
|
"step": 3380,
|
|
"valid_targets_mean": 4542.4,
|
|
"valid_targets_min": 1544
|
|
},
|
|
{
|
|
"epoch": 5.398724082934609,
|
|
"grad_norm": 0.49897907969004646,
|
|
"learning_rate": 6.056064414720317e-06,
|
|
"loss": 0.1492,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1465134173631668,
|
|
"step": 3385,
|
|
"valid_targets_mean": 4784.2,
|
|
"valid_targets_min": 1590
|
|
},
|
|
{
|
|
"epoch": 5.4066985645933014,
|
|
"grad_norm": 0.511553521072539,
|
|
"learning_rate": 5.999158508611496e-06,
|
|
"loss": 0.1387,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.133215069770813,
|
|
"step": 3390,
|
|
"valid_targets_mean": 3946.7,
|
|
"valid_targets_min": 1646
|
|
},
|
|
{
|
|
"epoch": 5.414673046251994,
|
|
"grad_norm": 0.5330358447457354,
|
|
"learning_rate": 5.942474013469983e-06,
|
|
"loss": 0.143,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16982445120811462,
|
|
"step": 3395,
|
|
"valid_targets_mean": 4178.5,
|
|
"valid_targets_min": 1664
|
|
},
|
|
{
|
|
"epoch": 5.422647527910685,
|
|
"grad_norm": 0.5597901604211492,
|
|
"learning_rate": 5.886011825711117e-06,
|
|
"loss": 0.1713,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17069560289382935,
|
|
"step": 3400,
|
|
"valid_targets_mean": 4582.6,
|
|
"valid_targets_min": 1345
|
|
},
|
|
{
|
|
"epoch": 5.430622009569378,
|
|
"grad_norm": 0.5472872828033137,
|
|
"learning_rate": 5.829772838234615e-06,
|
|
"loss": 0.1459,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15048807859420776,
|
|
"step": 3405,
|
|
"valid_targets_mean": 4030.6,
|
|
"valid_targets_min": 1675
|
|
},
|
|
{
|
|
"epoch": 5.43859649122807,
|
|
"grad_norm": 0.5792848668345226,
|
|
"learning_rate": 5.773757940410503e-06,
|
|
"loss": 0.143,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1506005823612213,
|
|
"step": 3410,
|
|
"valid_targets_mean": 4001.0,
|
|
"valid_targets_min": 652
|
|
},
|
|
{
|
|
"epoch": 5.446570972886763,
|
|
"grad_norm": 0.5741421209243341,
|
|
"learning_rate": 5.7179680180650055e-06,
|
|
"loss": 0.1466,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1489039659500122,
|
|
"step": 3415,
|
|
"valid_targets_mean": 3604.6,
|
|
"valid_targets_min": 1131
|
|
},
|
|
{
|
|
"epoch": 5.454545454545454,
|
|
"grad_norm": 0.4976492316442447,
|
|
"learning_rate": 5.6624039534665775e-06,
|
|
"loss": 0.1443,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1596824824810028,
|
|
"step": 3420,
|
|
"valid_targets_mean": 5098.1,
|
|
"valid_targets_min": 1663
|
|
},
|
|
{
|
|
"epoch": 5.4625199362041466,
|
|
"grad_norm": 0.5192694981391627,
|
|
"learning_rate": 5.607066625311925e-06,
|
|
"loss": 0.1461,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18752288818359375,
|
|
"step": 3425,
|
|
"valid_targets_mean": 4772.7,
|
|
"valid_targets_min": 1762
|
|
},
|
|
{
|
|
"epoch": 5.470494417862839,
|
|
"grad_norm": 0.4898189694458072,
|
|
"learning_rate": 5.55195690871211e-06,
|
|
"loss": 0.1435,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.136124387383461,
|
|
"step": 3430,
|
|
"valid_targets_mean": 4446.1,
|
|
"valid_targets_min": 1368
|
|
},
|
|
{
|
|
"epoch": 5.478468899521531,
|
|
"grad_norm": 0.5134289481443551,
|
|
"learning_rate": 5.497075675178727e-06,
|
|
"loss": 0.1492,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12175813317298889,
|
|
"step": 3435,
|
|
"valid_targets_mean": 3663.2,
|
|
"valid_targets_min": 1077
|
|
},
|
|
{
|
|
"epoch": 5.486443381180223,
|
|
"grad_norm": 0.4923964128209256,
|
|
"learning_rate": 5.442423792610118e-06,
|
|
"loss": 0.1435,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.158582404255867,
|
|
"step": 3440,
|
|
"valid_targets_mean": 4724.8,
|
|
"valid_targets_min": 1120
|
|
},
|
|
{
|
|
"epoch": 5.494417862838915,
|
|
"grad_norm": 0.5728414484181402,
|
|
"learning_rate": 5.388002125277627e-06,
|
|
"loss": 0.1555,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15127599239349365,
|
|
"step": 3445,
|
|
"valid_targets_mean": 3708.5,
|
|
"valid_targets_min": 1555
|
|
},
|
|
{
|
|
"epoch": 5.502392344497608,
|
|
"grad_norm": 0.5316153344749247,
|
|
"learning_rate": 5.333811533811945e-06,
|
|
"loss": 0.1448,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1468171775341034,
|
|
"step": 3450,
|
|
"valid_targets_mean": 4261.4,
|
|
"valid_targets_min": 1179
|
|
},
|
|
{
|
|
"epoch": 5.5103668261563,
|
|
"grad_norm": 0.5059201934559844,
|
|
"learning_rate": 5.2798528751895265e-06,
|
|
"loss": 0.149,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16418465971946716,
|
|
"step": 3455,
|
|
"valid_targets_mean": 4627.7,
|
|
"valid_targets_min": 659
|
|
},
|
|
{
|
|
"epoch": 5.518341307814992,
|
|
"grad_norm": 0.5138855686834195,
|
|
"learning_rate": 5.226127002718984e-06,
|
|
"loss": 0.1334,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14713597297668457,
|
|
"step": 3460,
|
|
"valid_targets_mean": 4193.9,
|
|
"valid_targets_min": 465
|
|
},
|
|
{
|
|
"epoch": 5.526315789473684,
|
|
"grad_norm": 0.8327422979910104,
|
|
"learning_rate": 5.1726347660276424e-06,
|
|
"loss": 0.1448,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12202882766723633,
|
|
"step": 3465,
|
|
"valid_targets_mean": 3889.8,
|
|
"valid_targets_min": 2307
|
|
},
|
|
{
|
|
"epoch": 5.5342902711323765,
|
|
"grad_norm": 0.5417866845035682,
|
|
"learning_rate": 5.119377011048066e-06,
|
|
"loss": 0.149,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15193571150302887,
|
|
"step": 3470,
|
|
"valid_targets_mean": 3883.6,
|
|
"valid_targets_min": 1027
|
|
},
|
|
{
|
|
"epoch": 5.542264752791069,
|
|
"grad_norm": 0.4869683717405225,
|
|
"learning_rate": 5.066354580004713e-06,
|
|
"loss": 0.1427,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12537409365177155,
|
|
"step": 3475,
|
|
"valid_targets_mean": 4069.8,
|
|
"valid_targets_min": 1800
|
|
},
|
|
{
|
|
"epoch": 5.55023923444976,
|
|
"grad_norm": 0.4657169670509649,
|
|
"learning_rate": 5.013568311400599e-06,
|
|
"loss": 0.1381,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1281999796628952,
|
|
"step": 3480,
|
|
"valid_targets_mean": 4824.8,
|
|
"valid_targets_min": 1840
|
|
},
|
|
{
|
|
"epoch": 5.558213716108453,
|
|
"grad_norm": 0.48908460901251527,
|
|
"learning_rate": 4.96101904000402e-06,
|
|
"loss": 0.1416,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14244484901428223,
|
|
"step": 3485,
|
|
"valid_targets_mean": 4372.4,
|
|
"valid_targets_min": 588
|
|
},
|
|
{
|
|
"epoch": 5.566188197767145,
|
|
"grad_norm": 0.5080720826305037,
|
|
"learning_rate": 4.908707596835396e-06,
|
|
"loss": 0.1417,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13216745853424072,
|
|
"step": 3490,
|
|
"valid_targets_mean": 4219.7,
|
|
"valid_targets_min": 1831
|
|
},
|
|
{
|
|
"epoch": 5.574162679425838,
|
|
"grad_norm": 0.49012390976455616,
|
|
"learning_rate": 4.856634809154093e-06,
|
|
"loss": 0.1405,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1266259402036667,
|
|
"step": 3495,
|
|
"valid_targets_mean": 4140.4,
|
|
"valid_targets_min": 1462
|
|
},
|
|
{
|
|
"epoch": 5.582137161084529,
|
|
"grad_norm": 0.4655698992214377,
|
|
"learning_rate": 4.804801500445338e-06,
|
|
"loss": 0.1474,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1438761204481125,
|
|
"step": 3500,
|
|
"valid_targets_mean": 5119.6,
|
|
"valid_targets_min": 1727
|
|
},
|
|
{
|
|
"epoch": 5.590111642743222,
|
|
"grad_norm": 0.47702768130088696,
|
|
"learning_rate": 4.753208490407233e-06,
|
|
"loss": 0.1502,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13481447100639343,
|
|
"step": 3505,
|
|
"valid_targets_mean": 4350.1,
|
|
"valid_targets_min": 1564
|
|
},
|
|
{
|
|
"epoch": 5.598086124401914,
|
|
"grad_norm": 0.5869307333733439,
|
|
"learning_rate": 4.701856594937744e-06,
|
|
"loss": 0.1579,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13290971517562866,
|
|
"step": 3510,
|
|
"valid_targets_mean": 3775.4,
|
|
"valid_targets_min": 1844
|
|
},
|
|
{
|
|
"epoch": 5.606060606060606,
|
|
"grad_norm": 0.5164996482721174,
|
|
"learning_rate": 4.650746626121838e-06,
|
|
"loss": 0.1558,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18071457743644714,
|
|
"step": 3515,
|
|
"valid_targets_mean": 4425.1,
|
|
"valid_targets_min": 1501
|
|
},
|
|
{
|
|
"epoch": 5.614035087719298,
|
|
"grad_norm": 0.5064382852224343,
|
|
"learning_rate": 4.5998793922186315e-06,
|
|
"loss": 0.1562,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13357748091220856,
|
|
"step": 3520,
|
|
"valid_targets_mean": 4476.5,
|
|
"valid_targets_min": 311
|
|
},
|
|
{
|
|
"epoch": 5.62200956937799,
|
|
"grad_norm": 0.4774092956312079,
|
|
"learning_rate": 4.549255697648576e-06,
|
|
"loss": 0.1437,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13738349080085754,
|
|
"step": 3525,
|
|
"valid_targets_mean": 4918.1,
|
|
"valid_targets_min": 1763
|
|
},
|
|
{
|
|
"epoch": 5.629984051036683,
|
|
"grad_norm": 0.4869106482742091,
|
|
"learning_rate": 4.498876342980796e-06,
|
|
"loss": 0.149,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13851019740104675,
|
|
"step": 3530,
|
|
"valid_targets_mean": 4753.4,
|
|
"valid_targets_min": 2359
|
|
},
|
|
{
|
|
"epoch": 5.637958532695375,
|
|
"grad_norm": 0.5444025100400355,
|
|
"learning_rate": 4.448742124920368e-06,
|
|
"loss": 0.1594,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15897592902183533,
|
|
"step": 3535,
|
|
"valid_targets_mean": 4504.8,
|
|
"valid_targets_min": 1086
|
|
},
|
|
{
|
|
"epoch": 5.645933014354067,
|
|
"grad_norm": 0.5202794879141485,
|
|
"learning_rate": 4.39885383629576e-06,
|
|
"loss": 0.1458,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12069641053676605,
|
|
"step": 3540,
|
|
"valid_targets_mean": 3759.1,
|
|
"valid_targets_min": 840
|
|
},
|
|
{
|
|
"epoch": 5.653907496012759,
|
|
"grad_norm": 0.4955637184896322,
|
|
"learning_rate": 4.349212266046285e-06,
|
|
"loss": 0.1414,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15988540649414062,
|
|
"step": 3545,
|
|
"valid_targets_mean": 4708.2,
|
|
"valid_targets_min": 2546
|
|
},
|
|
{
|
|
"epoch": 5.6618819776714515,
|
|
"grad_norm": 0.5349829325713732,
|
|
"learning_rate": 4.299818199209629e-06,
|
|
"loss": 0.16,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15441840887069702,
|
|
"step": 3550,
|
|
"valid_targets_mean": 3981.0,
|
|
"valid_targets_min": 1162
|
|
},
|
|
{
|
|
"epoch": 5.669856459330144,
|
|
"grad_norm": 0.5752960820375173,
|
|
"learning_rate": 4.250672416909407e-06,
|
|
"loss": 0.1602,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1708446890115738,
|
|
"step": 3555,
|
|
"valid_targets_mean": 4396.9,
|
|
"valid_targets_min": 1424
|
|
},
|
|
{
|
|
"epoch": 5.6778309409888355,
|
|
"grad_norm": 0.5751417483583992,
|
|
"learning_rate": 4.201775696342862e-06,
|
|
"loss": 0.1537,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18202140927314758,
|
|
"step": 3560,
|
|
"valid_targets_mean": 4092.4,
|
|
"valid_targets_min": 630
|
|
},
|
|
{
|
|
"epoch": 5.685805422647528,
|
|
"grad_norm": 1.1827861103154333,
|
|
"learning_rate": 4.153128810768517e-06,
|
|
"loss": 0.1373,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14385861158370972,
|
|
"step": 3565,
|
|
"valid_targets_mean": 4737.3,
|
|
"valid_targets_min": 1622
|
|
},
|
|
{
|
|
"epoch": 5.69377990430622,
|
|
"grad_norm": 0.5422374910484713,
|
|
"learning_rate": 4.104732529493991e-06,
|
|
"loss": 0.1487,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.139515221118927,
|
|
"step": 3570,
|
|
"valid_targets_mean": 4079.3,
|
|
"valid_targets_min": 1344
|
|
},
|
|
{
|
|
"epoch": 5.701754385964913,
|
|
"grad_norm": 0.5363029556528964,
|
|
"learning_rate": 4.056587617863825e-06,
|
|
"loss": 0.1549,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1622910350561142,
|
|
"step": 3575,
|
|
"valid_targets_mean": 4282.9,
|
|
"valid_targets_min": 316
|
|
},
|
|
{
|
|
"epoch": 5.709728867623604,
|
|
"grad_norm": 0.4597016933287751,
|
|
"learning_rate": 4.008694837247345e-06,
|
|
"loss": 0.1503,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13833516836166382,
|
|
"step": 3580,
|
|
"valid_targets_mean": 5463.1,
|
|
"valid_targets_min": 1688
|
|
},
|
|
{
|
|
"epoch": 5.717703349282297,
|
|
"grad_norm": 0.5063083170748712,
|
|
"learning_rate": 3.961054945026674e-06,
|
|
"loss": 0.1539,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15499377250671387,
|
|
"step": 3585,
|
|
"valid_targets_mean": 4217.2,
|
|
"valid_targets_min": 1999
|
|
},
|
|
{
|
|
"epoch": 5.725677830940989,
|
|
"grad_norm": 0.5248353026554393,
|
|
"learning_rate": 3.913668694584705e-06,
|
|
"loss": 0.1387,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15698176622390747,
|
|
"step": 3590,
|
|
"valid_targets_mean": 4311.0,
|
|
"valid_targets_min": 1237
|
|
},
|
|
{
|
|
"epoch": 5.733652312599681,
|
|
"grad_norm": 0.5039937842772115,
|
|
"learning_rate": 3.866536835293227e-06,
|
|
"loss": 0.1494,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1316099911928177,
|
|
"step": 3595,
|
|
"valid_targets_mean": 4095.2,
|
|
"valid_targets_min": 1637
|
|
},
|
|
{
|
|
"epoch": 5.741626794258373,
|
|
"grad_norm": 0.5325358744438344,
|
|
"learning_rate": 3.819660112501053e-06,
|
|
"loss": 0.1459,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14324098825454712,
|
|
"step": 3600,
|
|
"valid_targets_mean": 4207.9,
|
|
"valid_targets_min": 876
|
|
},
|
|
{
|
|
"epoch": 5.749601275917065,
|
|
"grad_norm": 0.47962766801791007,
|
|
"learning_rate": 3.773039267522227e-06,
|
|
"loss": 0.1462,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13331088423728943,
|
|
"step": 3605,
|
|
"valid_targets_mean": 4672.8,
|
|
"valid_targets_min": 414
|
|
},
|
|
{
|
|
"epoch": 5.757575757575758,
|
|
"grad_norm": 0.513570541506682,
|
|
"learning_rate": 3.72667503762433e-06,
|
|
"loss": 0.1427,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11546094715595245,
|
|
"step": 3610,
|
|
"valid_targets_mean": 4141.8,
|
|
"valid_targets_min": 1381
|
|
},
|
|
{
|
|
"epoch": 5.76555023923445,
|
|
"grad_norm": 0.7010946630275743,
|
|
"learning_rate": 3.680568156016786e-06,
|
|
"loss": 0.13,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14983190596103668,
|
|
"step": 3615,
|
|
"valid_targets_mean": 4744.1,
|
|
"valid_targets_min": 302
|
|
},
|
|
{
|
|
"epoch": 5.773524720893142,
|
|
"grad_norm": 0.5750246801245077,
|
|
"learning_rate": 3.6347193518392776e-06,
|
|
"loss": 0.145,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15094132721424103,
|
|
"step": 3620,
|
|
"valid_targets_mean": 3853.7,
|
|
"valid_targets_min": 1087
|
|
},
|
|
{
|
|
"epoch": 5.781499202551834,
|
|
"grad_norm": 0.4450147099326662,
|
|
"learning_rate": 3.58912935015024e-06,
|
|
"loss": 0.1349,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1421867311000824,
|
|
"step": 3625,
|
|
"valid_targets_mean": 5715.1,
|
|
"valid_targets_min": 2194
|
|
},
|
|
{
|
|
"epoch": 5.7894736842105265,
|
|
"grad_norm": 0.539335801690844,
|
|
"learning_rate": 3.543798871915367e-06,
|
|
"loss": 0.144,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14866751432418823,
|
|
"step": 3630,
|
|
"valid_targets_mean": 3839.4,
|
|
"valid_targets_min": 1946
|
|
},
|
|
{
|
|
"epoch": 5.797448165869218,
|
|
"grad_norm": 0.5828493833087499,
|
|
"learning_rate": 3.498728633996209e-06,
|
|
"loss": 0.132,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1669713258743286,
|
|
"step": 3635,
|
|
"valid_targets_mean": 5967.7,
|
|
"valid_targets_min": 1818
|
|
},
|
|
{
|
|
"epoch": 5.8054226475279105,
|
|
"grad_norm": 0.534453291890428,
|
|
"learning_rate": 3.453919349138859e-06,
|
|
"loss": 0.1436,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12102772295475006,
|
|
"step": 3640,
|
|
"valid_targets_mean": 3891.5,
|
|
"valid_targets_min": 1323
|
|
},
|
|
{
|
|
"epoch": 5.813397129186603,
|
|
"grad_norm": 0.48171779799429404,
|
|
"learning_rate": 3.4093717259626514e-06,
|
|
"loss": 0.1522,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14127930998802185,
|
|
"step": 3645,
|
|
"valid_targets_mean": 4453.7,
|
|
"valid_targets_min": 736
|
|
},
|
|
{
|
|
"epoch": 5.821371610845295,
|
|
"grad_norm": 0.49861046151709937,
|
|
"learning_rate": 3.365086468948988e-06,
|
|
"loss": 0.1534,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14490726590156555,
|
|
"step": 3650,
|
|
"valid_targets_mean": 4365.1,
|
|
"valid_targets_min": 1254
|
|
},
|
|
{
|
|
"epoch": 5.829346092503988,
|
|
"grad_norm": 0.5855078575877344,
|
|
"learning_rate": 3.321064278430175e-06,
|
|
"loss": 0.1534,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.18041715025901794,
|
|
"step": 3655,
|
|
"valid_targets_mean": 3792.9,
|
|
"valid_targets_min": 1380
|
|
},
|
|
{
|
|
"epoch": 5.837320574162679,
|
|
"grad_norm": 0.6060290822284615,
|
|
"learning_rate": 3.277305850578345e-06,
|
|
"loss": 0.1351,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14750179648399353,
|
|
"step": 3660,
|
|
"valid_targets_mean": 4823.8,
|
|
"valid_targets_min": 1604
|
|
},
|
|
{
|
|
"epoch": 5.845295055821372,
|
|
"grad_norm": 0.49296159557823577,
|
|
"learning_rate": 3.2338118773944684e-06,
|
|
"loss": 0.1446,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16297249495983124,
|
|
"step": 3665,
|
|
"valid_targets_mean": 4690.5,
|
|
"valid_targets_min": 1229
|
|
},
|
|
{
|
|
"epoch": 5.853269537480064,
|
|
"grad_norm": 0.5335159615396101,
|
|
"learning_rate": 3.1905830466973975e-06,
|
|
"loss": 0.144,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13889795541763306,
|
|
"step": 3670,
|
|
"valid_targets_mean": 4028.1,
|
|
"valid_targets_min": 1224
|
|
},
|
|
{
|
|
"epoch": 5.861244019138756,
|
|
"grad_norm": 0.48653483581031726,
|
|
"learning_rate": 3.14762004211298e-06,
|
|
"loss": 0.1435,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14789119362831116,
|
|
"step": 3675,
|
|
"valid_targets_mean": 4518.0,
|
|
"valid_targets_min": 1242
|
|
},
|
|
{
|
|
"epoch": 5.869218500797448,
|
|
"grad_norm": 0.44980623299065636,
|
|
"learning_rate": 3.1049235430632696e-06,
|
|
"loss": 0.1475,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14088287949562073,
|
|
"step": 3680,
|
|
"valid_targets_mean": 5030.7,
|
|
"valid_targets_min": 2268
|
|
},
|
|
{
|
|
"epoch": 5.87719298245614,
|
|
"grad_norm": 0.5454087979799263,
|
|
"learning_rate": 3.062494224755759e-06,
|
|
"loss": 0.1468,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14683279395103455,
|
|
"step": 3685,
|
|
"valid_targets_mean": 3636.9,
|
|
"valid_targets_min": 1406
|
|
},
|
|
{
|
|
"epoch": 5.885167464114833,
|
|
"grad_norm": 0.5284650556492287,
|
|
"learning_rate": 3.0203327581727195e-06,
|
|
"loss": 0.144,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14028066396713257,
|
|
"step": 3690,
|
|
"valid_targets_mean": 3733.6,
|
|
"valid_targets_min": 1271
|
|
},
|
|
{
|
|
"epoch": 5.893141945773524,
|
|
"grad_norm": 1.3422902233127916,
|
|
"learning_rate": 2.9784398100605937e-06,
|
|
"loss": 0.1375,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14661850035190582,
|
|
"step": 3695,
|
|
"valid_targets_mean": 3551.7,
|
|
"valid_targets_min": 1107
|
|
},
|
|
{
|
|
"epoch": 5.901116427432217,
|
|
"grad_norm": 0.5040044272291349,
|
|
"learning_rate": 2.9368160429194127e-06,
|
|
"loss": 0.1495,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13759127259254456,
|
|
"step": 3700,
|
|
"valid_targets_mean": 4007.2,
|
|
"valid_targets_min": 458
|
|
},
|
|
{
|
|
"epoch": 5.909090909090909,
|
|
"grad_norm": 0.6963247570508141,
|
|
"learning_rate": 2.895462114992371e-06,
|
|
"loss": 0.1389,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1433190256357193,
|
|
"step": 3705,
|
|
"valid_targets_mean": 3502.1,
|
|
"valid_targets_min": 1281
|
|
},
|
|
{
|
|
"epoch": 5.917065390749602,
|
|
"grad_norm": 0.5251079681315951,
|
|
"learning_rate": 2.8543786802553943e-06,
|
|
"loss": 0.1503,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1811850666999817,
|
|
"step": 3710,
|
|
"valid_targets_mean": 5324.9,
|
|
"valid_targets_min": 882
|
|
},
|
|
{
|
|
"epoch": 5.925039872408293,
|
|
"grad_norm": 0.5241665849437455,
|
|
"learning_rate": 2.813566388406781e-06,
|
|
"loss": 0.1516,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1358180046081543,
|
|
"step": 3715,
|
|
"valid_targets_mean": 3920.8,
|
|
"valid_targets_min": 599
|
|
},
|
|
{
|
|
"epoch": 5.9330143540669855,
|
|
"grad_norm": 0.49869493382972957,
|
|
"learning_rate": 2.773025884856957e-06,
|
|
"loss": 0.1448,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1408381462097168,
|
|
"step": 3720,
|
|
"valid_targets_mean": 4674.2,
|
|
"valid_targets_min": 1306
|
|
},
|
|
{
|
|
"epoch": 5.940988835725678,
|
|
"grad_norm": 0.5711342385096281,
|
|
"learning_rate": 2.7327578107182585e-06,
|
|
"loss": 0.14,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1314079761505127,
|
|
"step": 3725,
|
|
"valid_targets_mean": 3162.1,
|
|
"valid_targets_min": 892
|
|
},
|
|
{
|
|
"epoch": 5.94896331738437,
|
|
"grad_norm": 0.6108334839280424,
|
|
"learning_rate": 2.692762802794775e-06,
|
|
"loss": 0.1317,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12222002446651459,
|
|
"step": 3730,
|
|
"valid_targets_mean": 3391.4,
|
|
"valid_targets_min": 961
|
|
},
|
|
{
|
|
"epoch": 5.956937799043062,
|
|
"grad_norm": 0.5027197439334232,
|
|
"learning_rate": 2.6530414935723104e-06,
|
|
"loss": 0.142,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13072869181632996,
|
|
"step": 3735,
|
|
"valid_targets_mean": 3994.5,
|
|
"valid_targets_min": 1639
|
|
},
|
|
{
|
|
"epoch": 5.964912280701754,
|
|
"grad_norm": 0.5075537096649687,
|
|
"learning_rate": 2.6135945112083506e-06,
|
|
"loss": 0.1456,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12960845232009888,
|
|
"step": 3740,
|
|
"valid_targets_mean": 4494.9,
|
|
"valid_targets_min": 597
|
|
},
|
|
{
|
|
"epoch": 5.972886762360447,
|
|
"grad_norm": 0.5380248417532908,
|
|
"learning_rate": 2.574422479522156e-06,
|
|
"loss": 0.1516,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1744580864906311,
|
|
"step": 3745,
|
|
"valid_targets_mean": 4334.4,
|
|
"valid_targets_min": 1703
|
|
},
|
|
{
|
|
"epoch": 5.980861244019139,
|
|
"grad_norm": 0.5515990373712637,
|
|
"learning_rate": 2.535526017984884e-06,
|
|
"loss": 0.1456,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14526067674160004,
|
|
"step": 3750,
|
|
"valid_targets_mean": 3798.1,
|
|
"valid_targets_min": 353
|
|
},
|
|
{
|
|
"epoch": 5.988835725677831,
|
|
"grad_norm": 0.49011701337432784,
|
|
"learning_rate": 2.4969057417097807e-06,
|
|
"loss": 0.1574,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15856046974658966,
|
|
"step": 3755,
|
|
"valid_targets_mean": 5457.1,
|
|
"valid_targets_min": 893
|
|
},
|
|
{
|
|
"epoch": 5.996810207336523,
|
|
"grad_norm": 0.5383249113038507,
|
|
"learning_rate": 2.458562261442483e-06,
|
|
"loss": 0.1552,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14745956659317017,
|
|
"step": 3760,
|
|
"valid_targets_mean": 4042.0,
|
|
"valid_targets_min": 900
|
|
},
|
|
{
|
|
"epoch": 6.0047846889952154,
|
|
"grad_norm": 0.49430702596220477,
|
|
"learning_rate": 2.4204961835513263e-06,
|
|
"loss": 0.14,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14110133051872253,
|
|
"step": 3765,
|
|
"valid_targets_mean": 4544.2,
|
|
"valid_targets_min": 1362
|
|
},
|
|
{
|
|
"epoch": 6.012759170653908,
|
|
"grad_norm": 0.5009175563097514,
|
|
"learning_rate": 2.3827081100177797e-06,
|
|
"loss": 0.142,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1476556956768036,
|
|
"step": 3770,
|
|
"valid_targets_mean": 5110.4,
|
|
"valid_targets_min": 1185
|
|
},
|
|
{
|
|
"epoch": 6.020733652312599,
|
|
"grad_norm": 0.49313677516660054,
|
|
"learning_rate": 2.3451986384269266e-06,
|
|
"loss": 0.1309,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13937491178512573,
|
|
"step": 3775,
|
|
"valid_targets_mean": 4621.9,
|
|
"valid_targets_min": 1624
|
|
},
|
|
{
|
|
"epoch": 6.028708133971292,
|
|
"grad_norm": 0.6860362912088609,
|
|
"learning_rate": 2.307968361957993e-06,
|
|
"loss": 0.154,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.187550351023674,
|
|
"step": 3780,
|
|
"valid_targets_mean": 4856.3,
|
|
"valid_targets_min": 1571
|
|
},
|
|
{
|
|
"epoch": 6.036682615629984,
|
|
"grad_norm": 0.5773716723883122,
|
|
"learning_rate": 2.2710178693749805e-06,
|
|
"loss": 0.1387,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1661595106124878,
|
|
"step": 3785,
|
|
"valid_targets_mean": 3802.4,
|
|
"valid_targets_min": 1601
|
|
},
|
|
{
|
|
"epoch": 6.044657097288677,
|
|
"grad_norm": 0.497541583677315,
|
|
"learning_rate": 2.2343477450173665e-06,
|
|
"loss": 0.131,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11052682250738144,
|
|
"step": 3790,
|
|
"valid_targets_mean": 3915.4,
|
|
"valid_targets_min": 1388
|
|
},
|
|
{
|
|
"epoch": 6.052631578947368,
|
|
"grad_norm": 0.4960366052870153,
|
|
"learning_rate": 2.197958568790839e-06,
|
|
"loss": 0.1416,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14265236258506775,
|
|
"step": 3795,
|
|
"valid_targets_mean": 4880.4,
|
|
"valid_targets_min": 2259
|
|
},
|
|
{
|
|
"epoch": 6.0606060606060606,
|
|
"grad_norm": 0.4925604930180625,
|
|
"learning_rate": 2.161850916158148e-06,
|
|
"loss": 0.1414,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13926450908184052,
|
|
"step": 3800,
|
|
"valid_targets_mean": 4805.2,
|
|
"valid_targets_min": 1930
|
|
},
|
|
{
|
|
"epoch": 6.068580542264753,
|
|
"grad_norm": 0.49478746503871057,
|
|
"learning_rate": 2.1260253581299996e-06,
|
|
"loss": 0.1383,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1338876485824585,
|
|
"step": 3805,
|
|
"valid_targets_mean": 4109.5,
|
|
"valid_targets_min": 414
|
|
},
|
|
{
|
|
"epoch": 6.076555023923445,
|
|
"grad_norm": 0.5064549765633942,
|
|
"learning_rate": 2.0904824612560046e-06,
|
|
"loss": 0.1445,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1764906346797943,
|
|
"step": 3810,
|
|
"valid_targets_mean": 4899.3,
|
|
"valid_targets_min": 1593
|
|
},
|
|
{
|
|
"epoch": 6.084529505582137,
|
|
"grad_norm": 0.5180065700775449,
|
|
"learning_rate": 2.0552227876157536e-06,
|
|
"loss": 0.1252,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1284601241350174,
|
|
"step": 3815,
|
|
"valid_targets_mean": 3687.7,
|
|
"valid_targets_min": 732
|
|
},
|
|
{
|
|
"epoch": 6.092503987240829,
|
|
"grad_norm": 0.5259373294338707,
|
|
"learning_rate": 2.020246894809912e-06,
|
|
"loss": 0.1394,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13603927195072174,
|
|
"step": 3820,
|
|
"valid_targets_mean": 4060.6,
|
|
"valid_targets_min": 308
|
|
},
|
|
{
|
|
"epoch": 6.100478468899522,
|
|
"grad_norm": 0.5235335460898032,
|
|
"learning_rate": 1.9855553359513836e-06,
|
|
"loss": 0.1447,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11099904775619507,
|
|
"step": 3825,
|
|
"valid_targets_mean": 4016.0,
|
|
"valid_targets_min": 1483
|
|
},
|
|
{
|
|
"epoch": 6.108452950558214,
|
|
"grad_norm": 0.6341054094850744,
|
|
"learning_rate": 1.9511486596566054e-06,
|
|
"loss": 0.1576,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17271441221237183,
|
|
"step": 3830,
|
|
"valid_targets_mean": 3144.9,
|
|
"valid_targets_min": 1179
|
|
},
|
|
{
|
|
"epoch": 6.116427432216906,
|
|
"grad_norm": 0.5684654706262051,
|
|
"learning_rate": 1.917027410036825e-06,
|
|
"loss": 0.1416,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15632422268390656,
|
|
"step": 3835,
|
|
"valid_targets_mean": 3489.4,
|
|
"valid_targets_min": 440
|
|
},
|
|
{
|
|
"epoch": 6.124401913875598,
|
|
"grad_norm": 0.47186770873479894,
|
|
"learning_rate": 1.8831921266895348e-06,
|
|
"loss": 0.1317,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12489638477563858,
|
|
"step": 3840,
|
|
"valid_targets_mean": 4751.5,
|
|
"valid_targets_min": 564
|
|
},
|
|
{
|
|
"epoch": 6.1323763955342905,
|
|
"grad_norm": 0.5812261927451817,
|
|
"learning_rate": 1.8496433446899197e-06,
|
|
"loss": 0.1332,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11479625850915909,
|
|
"step": 3845,
|
|
"valid_targets_mean": 3060.4,
|
|
"valid_targets_min": 1069
|
|
},
|
|
{
|
|
"epoch": 6.140350877192983,
|
|
"grad_norm": 0.5402386536039824,
|
|
"learning_rate": 1.8163815945823881e-06,
|
|
"loss": 0.1354,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1345522552728653,
|
|
"step": 3850,
|
|
"valid_targets_mean": 4273.4,
|
|
"valid_targets_min": 1083
|
|
},
|
|
{
|
|
"epoch": 6.148325358851674,
|
|
"grad_norm": 0.496965598633163,
|
|
"learning_rate": 1.7834074023722082e-06,
|
|
"loss": 0.1277,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12266291677951813,
|
|
"step": 3855,
|
|
"valid_targets_mean": 4934.3,
|
|
"valid_targets_min": 1487
|
|
},
|
|
{
|
|
"epoch": 6.156299840510367,
|
|
"grad_norm": 0.5519860673495093,
|
|
"learning_rate": 1.7507212895171632e-06,
|
|
"loss": 0.1439,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1321459710597992,
|
|
"step": 3860,
|
|
"valid_targets_mean": 3861.4,
|
|
"valid_targets_min": 686
|
|
},
|
|
{
|
|
"epoch": 6.164274322169059,
|
|
"grad_norm": 0.49076514558644085,
|
|
"learning_rate": 1.7183237729193081e-06,
|
|
"loss": 0.152,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15181314945220947,
|
|
"step": 3865,
|
|
"valid_targets_mean": 5045.6,
|
|
"valid_targets_min": 1188
|
|
},
|
|
{
|
|
"epoch": 6.172248803827751,
|
|
"grad_norm": 0.5372730767695284,
|
|
"learning_rate": 1.6862153649168211e-06,
|
|
"loss": 0.134,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12448375672101974,
|
|
"step": 3870,
|
|
"valid_targets_mean": 3875.8,
|
|
"valid_targets_min": 1162
|
|
},
|
|
{
|
|
"epoch": 6.180223285486443,
|
|
"grad_norm": 0.49684499279704325,
|
|
"learning_rate": 1.6543965732758737e-06,
|
|
"loss": 0.1427,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1295873075723648,
|
|
"step": 3875,
|
|
"valid_targets_mean": 4275.8,
|
|
"valid_targets_min": 1810
|
|
},
|
|
{
|
|
"epoch": 6.188197767145136,
|
|
"grad_norm": 0.5504960661897019,
|
|
"learning_rate": 1.6228679011826032e-06,
|
|
"loss": 0.1391,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1351630687713623,
|
|
"step": 3880,
|
|
"valid_targets_mean": 3684.4,
|
|
"valid_targets_min": 1036
|
|
},
|
|
{
|
|
"epoch": 6.196172248803828,
|
|
"grad_norm": 0.4995587060471986,
|
|
"learning_rate": 1.591629847235172e-06,
|
|
"loss": 0.1318,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12718336284160614,
|
|
"step": 3885,
|
|
"valid_targets_mean": 4705.2,
|
|
"valid_targets_min": 1038
|
|
},
|
|
{
|
|
"epoch": 6.2041467304625195,
|
|
"grad_norm": 0.4959141881965582,
|
|
"learning_rate": 1.5606829054358686e-06,
|
|
"loss": 0.1428,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1419508159160614,
|
|
"step": 3890,
|
|
"valid_targets_mean": 4684.0,
|
|
"valid_targets_min": 2001
|
|
},
|
|
{
|
|
"epoch": 6.212121212121212,
|
|
"grad_norm": 0.623950185968585,
|
|
"learning_rate": 1.5300275651832963e-06,
|
|
"loss": 0.136,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13223516941070557,
|
|
"step": 3895,
|
|
"valid_targets_mean": 5110.7,
|
|
"valid_targets_min": 1969
|
|
},
|
|
{
|
|
"epoch": 6.220095693779904,
|
|
"grad_norm": 0.6346640130343832,
|
|
"learning_rate": 1.499664311264648e-06,
|
|
"loss": 0.1488,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16367286443710327,
|
|
"step": 3900,
|
|
"valid_targets_mean": 3819.8,
|
|
"valid_targets_min": 860
|
|
},
|
|
{
|
|
"epoch": 6.228070175438597,
|
|
"grad_norm": 0.6122246042123756,
|
|
"learning_rate": 1.4695936238480135e-06,
|
|
"loss": 0.1426,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12655913829803467,
|
|
"step": 3905,
|
|
"valid_targets_mean": 3684.9,
|
|
"valid_targets_min": 736
|
|
},
|
|
{
|
|
"epoch": 6.236044657097288,
|
|
"grad_norm": 0.5744548130501511,
|
|
"learning_rate": 1.4398159784748144e-06,
|
|
"loss": 0.1265,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16123470664024353,
|
|
"step": 3910,
|
|
"valid_targets_mean": 3574.8,
|
|
"valid_targets_min": 1859
|
|
},
|
|
{
|
|
"epoch": 6.244019138755981,
|
|
"grad_norm": 0.5996106329357315,
|
|
"learning_rate": 1.4103318460522598e-06,
|
|
"loss": 0.1316,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1403590589761734,
|
|
"step": 3915,
|
|
"valid_targets_mean": 3877.1,
|
|
"valid_targets_min": 625
|
|
},
|
|
{
|
|
"epoch": 6.251993620414673,
|
|
"grad_norm": 0.5132749502243364,
|
|
"learning_rate": 1.3811416928459177e-06,
|
|
"loss": 0.137,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14636051654815674,
|
|
"step": 3920,
|
|
"valid_targets_mean": 4128.2,
|
|
"valid_targets_min": 465
|
|
},
|
|
{
|
|
"epoch": 6.2599681020733655,
|
|
"grad_norm": 0.5650057439215522,
|
|
"learning_rate": 1.3522459804723353e-06,
|
|
"loss": 0.1245,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1392631232738495,
|
|
"step": 3925,
|
|
"valid_targets_mean": 4433.2,
|
|
"valid_targets_min": 916
|
|
},
|
|
{
|
|
"epoch": 6.267942583732057,
|
|
"grad_norm": 0.5185553837899172,
|
|
"learning_rate": 1.3236451658917293e-06,
|
|
"loss": 0.1493,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13320839405059814,
|
|
"step": 3930,
|
|
"valid_targets_mean": 4020.2,
|
|
"valid_targets_min": 1580
|
|
},
|
|
{
|
|
"epoch": 6.2759170653907494,
|
|
"grad_norm": 0.5525213441195974,
|
|
"learning_rate": 1.2953397014007728e-06,
|
|
"loss": 0.1611,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17850124835968018,
|
|
"step": 3935,
|
|
"valid_targets_mean": 4348.1,
|
|
"valid_targets_min": 1432
|
|
},
|
|
{
|
|
"epoch": 6.283891547049442,
|
|
"grad_norm": 0.488391409608166,
|
|
"learning_rate": 1.2673300346254447e-06,
|
|
"loss": 0.1341,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15462128818035126,
|
|
"step": 3940,
|
|
"valid_targets_mean": 4733.1,
|
|
"valid_targets_min": 1935
|
|
},
|
|
{
|
|
"epoch": 6.291866028708134,
|
|
"grad_norm": 0.5108981894562138,
|
|
"learning_rate": 1.239616608513925e-06,
|
|
"loss": 0.1409,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1496082842350006,
|
|
"step": 3945,
|
|
"valid_targets_mean": 4025.8,
|
|
"valid_targets_min": 535
|
|
},
|
|
{
|
|
"epoch": 6.299840510366826,
|
|
"grad_norm": 0.6093838052506947,
|
|
"learning_rate": 1.2121998613296259e-06,
|
|
"loss": 0.1214,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10867314040660858,
|
|
"step": 3950,
|
|
"valid_targets_mean": 3494.8,
|
|
"valid_targets_min": 670
|
|
},
|
|
{
|
|
"epoch": 6.307814992025518,
|
|
"grad_norm": 0.5119509074281289,
|
|
"learning_rate": 1.1850802266442396e-06,
|
|
"loss": 0.1597,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16429255902767181,
|
|
"step": 3955,
|
|
"valid_targets_mean": 4612.4,
|
|
"valid_targets_min": 759
|
|
},
|
|
{
|
|
"epoch": 6.315789473684211,
|
|
"grad_norm": 0.5347696760359358,
|
|
"learning_rate": 1.1582581333308784e-06,
|
|
"loss": 0.1558,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1720159351825714,
|
|
"step": 3960,
|
|
"valid_targets_mean": 4205.4,
|
|
"valid_targets_min": 515
|
|
},
|
|
{
|
|
"epoch": 6.323763955342903,
|
|
"grad_norm": 0.4852207515725228,
|
|
"learning_rate": 1.1317340055573122e-06,
|
|
"loss": 0.1381,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1445663422346115,
|
|
"step": 3965,
|
|
"valid_targets_mean": 5052.6,
|
|
"valid_targets_min": 1283
|
|
},
|
|
{
|
|
"epoch": 6.3317384370015946,
|
|
"grad_norm": 0.507436285150948,
|
|
"learning_rate": 1.1055082627792357e-06,
|
|
"loss": 0.1329,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12436027824878693,
|
|
"step": 3970,
|
|
"valid_targets_mean": 4297.2,
|
|
"valid_targets_min": 1663
|
|
},
|
|
{
|
|
"epoch": 6.339712918660287,
|
|
"grad_norm": 0.5368792578229027,
|
|
"learning_rate": 1.0795813197336602e-06,
|
|
"loss": 0.1349,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11996762454509735,
|
|
"step": 3975,
|
|
"valid_targets_mean": 3953.6,
|
|
"valid_targets_min": 1257
|
|
},
|
|
{
|
|
"epoch": 6.347687400318979,
|
|
"grad_norm": 0.48044682571789116,
|
|
"learning_rate": 1.0539535864323391e-06,
|
|
"loss": 0.1308,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1088440790772438,
|
|
"step": 3980,
|
|
"valid_targets_mean": 4750.7,
|
|
"valid_targets_min": 1344
|
|
},
|
|
{
|
|
"epoch": 6.355661881977672,
|
|
"grad_norm": 0.5329144232637667,
|
|
"learning_rate": 1.0286254681552777e-06,
|
|
"loss": 0.1351,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1612853854894638,
|
|
"step": 3985,
|
|
"valid_targets_mean": 4179.6,
|
|
"valid_targets_min": 1229
|
|
},
|
|
{
|
|
"epoch": 6.363636363636363,
|
|
"grad_norm": 0.5748925946370642,
|
|
"learning_rate": 1.0035973654443466e-06,
|
|
"loss": 0.1342,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12145799398422241,
|
|
"step": 3990,
|
|
"valid_targets_mean": 4831.9,
|
|
"valid_targets_min": 861
|
|
},
|
|
{
|
|
"epoch": 6.371610845295056,
|
|
"grad_norm": 0.5705983078076705,
|
|
"learning_rate": 9.788696740969295e-07,
|
|
"loss": 0.1337,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14661964774131775,
|
|
"step": 3995,
|
|
"valid_targets_mean": 3695.1,
|
|
"valid_targets_min": 550
|
|
},
|
|
{
|
|
"epoch": 6.379585326953748,
|
|
"grad_norm": 0.5645513513745306,
|
|
"learning_rate": 9.544427851596661e-07,
|
|
"loss": 0.1306,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13105720281600952,
|
|
"step": 4000,
|
|
"valid_targets_mean": 3997.6,
|
|
"valid_targets_min": 287
|
|
},
|
|
{
|
|
"epoch": 6.3875598086124405,
|
|
"grad_norm": 0.5275976634530215,
|
|
"learning_rate": 9.303170849222764e-07,
|
|
"loss": 0.1486,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15079137682914734,
|
|
"step": 4005,
|
|
"valid_targets_mean": 4694.5,
|
|
"valid_targets_min": 359
|
|
},
|
|
{
|
|
"epoch": 6.395534290271132,
|
|
"grad_norm": 0.5285990530880269,
|
|
"learning_rate": 9.064929549114421e-07,
|
|
"loss": 0.1469,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.10869532823562622,
|
|
"step": 4010,
|
|
"valid_targets_mean": 3649.6,
|
|
"valid_targets_min": 1504
|
|
},
|
|
{
|
|
"epoch": 6.4035087719298245,
|
|
"grad_norm": 0.7090154333622751,
|
|
"learning_rate": 8.829707718847835e-07,
|
|
"loss": 0.1411,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14588609337806702,
|
|
"step": 4015,
|
|
"valid_targets_mean": 3740.7,
|
|
"valid_targets_min": 1132
|
|
},
|
|
{
|
|
"epoch": 6.411483253588517,
|
|
"grad_norm": 0.5137768138489031,
|
|
"learning_rate": 8.597509078248923e-07,
|
|
"loss": 0.1456,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1488763988018036,
|
|
"step": 4020,
|
|
"valid_targets_mean": 4146.1,
|
|
"valid_targets_min": 531
|
|
},
|
|
{
|
|
"epoch": 6.419457735247209,
|
|
"grad_norm": 0.48049275831230703,
|
|
"learning_rate": 8.368337299334461e-07,
|
|
"loss": 0.13,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11516915261745453,
|
|
"step": 4025,
|
|
"valid_targets_mean": 4326.6,
|
|
"valid_targets_min": 1295
|
|
},
|
|
{
|
|
"epoch": 6.427432216905901,
|
|
"grad_norm": 0.5161570168262671,
|
|
"learning_rate": 8.142196006254144e-07,
|
|
"loss": 0.1328,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12275591492652893,
|
|
"step": 4030,
|
|
"valid_targets_mean": 4094.8,
|
|
"valid_targets_min": 1898
|
|
},
|
|
{
|
|
"epoch": 6.435406698564593,
|
|
"grad_norm": 0.5362563187603275,
|
|
"learning_rate": 7.919088775233264e-07,
|
|
"loss": 0.1382,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13758490979671478,
|
|
"step": 4035,
|
|
"valid_targets_mean": 3916.8,
|
|
"valid_targets_min": 1555
|
|
},
|
|
{
|
|
"epoch": 6.443381180223286,
|
|
"grad_norm": 0.49875755343739087,
|
|
"learning_rate": 7.699019134515917e-07,
|
|
"loss": 0.1365,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1160312294960022,
|
|
"step": 4040,
|
|
"valid_targets_mean": 4468.5,
|
|
"valid_targets_min": 2508
|
|
},
|
|
{
|
|
"epoch": 6.451355661881978,
|
|
"grad_norm": 0.5262408222186133,
|
|
"learning_rate": 7.48199056430956e-07,
|
|
"loss": 0.1446,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1481764316558838,
|
|
"step": 4045,
|
|
"valid_targets_mean": 4616.4,
|
|
"valid_targets_min": 458
|
|
},
|
|
{
|
|
"epoch": 6.45933014354067,
|
|
"grad_norm": 0.507178503336427,
|
|
"learning_rate": 7.268006496729762e-07,
|
|
"loss": 0.1328,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13112816214561462,
|
|
"step": 4050,
|
|
"valid_targets_mean": 3990.1,
|
|
"valid_targets_min": 641
|
|
},
|
|
{
|
|
"epoch": 6.467304625199362,
|
|
"grad_norm": 0.5669657663732001,
|
|
"learning_rate": 7.057070315745851e-07,
|
|
"loss": 0.1344,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11124549061059952,
|
|
"step": 4055,
|
|
"valid_targets_mean": 3979.8,
|
|
"valid_targets_min": 1798
|
|
},
|
|
{
|
|
"epoch": 6.475279106858054,
|
|
"grad_norm": 0.5377654564783906,
|
|
"learning_rate": 6.849185357127686e-07,
|
|
"loss": 0.135,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13395267724990845,
|
|
"step": 4060,
|
|
"valid_targets_mean": 4332.1,
|
|
"valid_targets_min": 1825
|
|
},
|
|
{
|
|
"epoch": 6.483253588516747,
|
|
"grad_norm": 0.6589422038099677,
|
|
"learning_rate": 6.64435490839257e-07,
|
|
"loss": 0.1385,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12837636470794678,
|
|
"step": 4065,
|
|
"valid_targets_mean": 3748.6,
|
|
"valid_targets_min": 291
|
|
},
|
|
{
|
|
"epoch": 6.491228070175438,
|
|
"grad_norm": 0.46549715916839196,
|
|
"learning_rate": 6.442582208753578e-07,
|
|
"loss": 0.1658,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16415579617023468,
|
|
"step": 4070,
|
|
"valid_targets_mean": 5833.8,
|
|
"valid_targets_min": 1597
|
|
},
|
|
{
|
|
"epoch": 6.499202551834131,
|
|
"grad_norm": 0.6622687035422515,
|
|
"learning_rate": 6.243870449068068e-07,
|
|
"loss": 0.1536,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1493580937385559,
|
|
"step": 4075,
|
|
"valid_targets_mean": 4838.5,
|
|
"valid_targets_min": 867
|
|
},
|
|
{
|
|
"epoch": 6.507177033492823,
|
|
"grad_norm": 0.5198345078048688,
|
|
"learning_rate": 6.048222771787382e-07,
|
|
"loss": 0.1366,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1303584724664688,
|
|
"step": 4080,
|
|
"valid_targets_mean": 4566.9,
|
|
"valid_targets_min": 1867
|
|
},
|
|
{
|
|
"epoch": 6.515151515151516,
|
|
"grad_norm": 0.48509162868152,
|
|
"learning_rate": 5.85564227090707e-07,
|
|
"loss": 0.1545,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16224688291549683,
|
|
"step": 4085,
|
|
"valid_targets_mean": 5195.4,
|
|
"valid_targets_min": 1471
|
|
},
|
|
{
|
|
"epoch": 6.523125996810207,
|
|
"grad_norm": 0.5873719255870822,
|
|
"learning_rate": 5.666131991917989e-07,
|
|
"loss": 0.1348,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13083861768245697,
|
|
"step": 4090,
|
|
"valid_targets_mean": 3929.3,
|
|
"valid_targets_min": 376
|
|
},
|
|
{
|
|
"epoch": 6.5311004784688995,
|
|
"grad_norm": 0.5977724356369237,
|
|
"learning_rate": 5.479694931758194e-07,
|
|
"loss": 0.129,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13004256784915924,
|
|
"step": 4095,
|
|
"valid_targets_mean": 3322.4,
|
|
"valid_targets_min": 854
|
|
},
|
|
{
|
|
"epoch": 6.539074960127592,
|
|
"grad_norm": 0.5240351703130457,
|
|
"learning_rate": 5.296334038765483e-07,
|
|
"loss": 0.1402,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13967791199684143,
|
|
"step": 4100,
|
|
"valid_targets_mean": 3777.9,
|
|
"valid_targets_min": 1111
|
|
},
|
|
{
|
|
"epoch": 6.5470494417862835,
|
|
"grad_norm": 0.5440153780671622,
|
|
"learning_rate": 5.116052212630696e-07,
|
|
"loss": 0.1347,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11996950209140778,
|
|
"step": 4105,
|
|
"valid_targets_mean": 3631.4,
|
|
"valid_targets_min": 1837
|
|
},
|
|
{
|
|
"epoch": 6.555023923444976,
|
|
"grad_norm": 0.48914053795353435,
|
|
"learning_rate": 4.938852304352026e-07,
|
|
"loss": 0.1245,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11675041913986206,
|
|
"step": 4110,
|
|
"valid_targets_mean": 4209.9,
|
|
"valid_targets_min": 1039
|
|
},
|
|
{
|
|
"epoch": 6.562998405103668,
|
|
"grad_norm": 0.5196936360126563,
|
|
"learning_rate": 4.7647371161898547e-07,
|
|
"loss": 0.144,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14445306360721588,
|
|
"step": 4115,
|
|
"valid_targets_mean": 4266.6,
|
|
"valid_targets_min": 1671
|
|
},
|
|
{
|
|
"epoch": 6.570972886762361,
|
|
"grad_norm": 0.6270985256379639,
|
|
"learning_rate": 4.593709401622359e-07,
|
|
"loss": 0.1344,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11855871975421906,
|
|
"step": 4120,
|
|
"valid_targets_mean": 4256.4,
|
|
"valid_targets_min": 1403
|
|
},
|
|
{
|
|
"epoch": 6.578947368421053,
|
|
"grad_norm": 0.5513191005216137,
|
|
"learning_rate": 4.425771865302153e-07,
|
|
"loss": 0.1387,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16223138570785522,
|
|
"step": 4125,
|
|
"valid_targets_mean": 5632.7,
|
|
"valid_targets_min": 2783
|
|
},
|
|
{
|
|
"epoch": 6.586921850079745,
|
|
"grad_norm": 0.47622414847630423,
|
|
"learning_rate": 4.2609271630133174e-07,
|
|
"loss": 0.1431,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15721635520458221,
|
|
"step": 4130,
|
|
"valid_targets_mean": 4821.3,
|
|
"valid_targets_min": 1248
|
|
},
|
|
{
|
|
"epoch": 6.594896331738437,
|
|
"grad_norm": 0.5867580468822079,
|
|
"learning_rate": 4.099177901629525e-07,
|
|
"loss": 0.1376,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1569528430700302,
|
|
"step": 4135,
|
|
"valid_targets_mean": 4191.9,
|
|
"valid_targets_min": 351
|
|
},
|
|
{
|
|
"epoch": 6.6028708133971294,
|
|
"grad_norm": 0.5051611881659711,
|
|
"learning_rate": 3.9405266390727836e-07,
|
|
"loss": 0.1387,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13498680293560028,
|
|
"step": 4140,
|
|
"valid_targets_mean": 4330.7,
|
|
"valid_targets_min": 338
|
|
},
|
|
{
|
|
"epoch": 6.610845295055821,
|
|
"grad_norm": 0.47853079889882155,
|
|
"learning_rate": 3.7849758842729344e-07,
|
|
"loss": 0.1418,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12443745136260986,
|
|
"step": 4145,
|
|
"valid_targets_mean": 4602.2,
|
|
"valid_targets_min": 1840
|
|
},
|
|
{
|
|
"epoch": 6.618819776714513,
|
|
"grad_norm": 0.5338550959823287,
|
|
"learning_rate": 3.632528097128085e-07,
|
|
"loss": 0.1234,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12975472211837769,
|
|
"step": 4150,
|
|
"valid_targets_mean": 4234.5,
|
|
"valid_targets_min": 943
|
|
},
|
|
{
|
|
"epoch": 6.626794258373206,
|
|
"grad_norm": 0.49688890655045326,
|
|
"learning_rate": 3.48318568846564e-07,
|
|
"loss": 0.1297,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13185477256774902,
|
|
"step": 4155,
|
|
"valid_targets_mean": 4626.2,
|
|
"valid_targets_min": 481
|
|
},
|
|
{
|
|
"epoch": 6.634768740031898,
|
|
"grad_norm": 0.5763693407472297,
|
|
"learning_rate": 3.336951020004087e-07,
|
|
"loss": 0.1561,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.19649842381477356,
|
|
"step": 4160,
|
|
"valid_targets_mean": 4184.9,
|
|
"valid_targets_min": 1350
|
|
},
|
|
{
|
|
"epoch": 6.64274322169059,
|
|
"grad_norm": 0.5531016551635043,
|
|
"learning_rate": 3.1938264043158694e-07,
|
|
"loss": 0.1434,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17289824783802032,
|
|
"step": 4165,
|
|
"valid_targets_mean": 4395.6,
|
|
"valid_targets_min": 1557
|
|
},
|
|
{
|
|
"epoch": 6.650717703349282,
|
|
"grad_norm": 0.47177742265711486,
|
|
"learning_rate": 3.05381410479062e-07,
|
|
"loss": 0.1439,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17544347047805786,
|
|
"step": 4170,
|
|
"valid_targets_mean": 5182.7,
|
|
"valid_targets_min": 1911
|
|
},
|
|
{
|
|
"epoch": 6.6586921850079746,
|
|
"grad_norm": 0.5216521849278853,
|
|
"learning_rate": 2.916916335599407e-07,
|
|
"loss": 0.1289,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.11967335641384125,
|
|
"step": 4175,
|
|
"valid_targets_mean": 3633.3,
|
|
"valid_targets_min": 1290
|
|
},
|
|
{
|
|
"epoch": 6.666666666666667,
|
|
"grad_norm": 0.5244652547649694,
|
|
"learning_rate": 2.783135261659831e-07,
|
|
"loss": 0.1503,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14524878561496735,
|
|
"step": 4180,
|
|
"valid_targets_mean": 5046.4,
|
|
"valid_targets_min": 1755
|
|
},
|
|
{
|
|
"epoch": 6.6746411483253585,
|
|
"grad_norm": 0.5242548984299902,
|
|
"learning_rate": 2.6524729986016293e-07,
|
|
"loss": 0.142,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1346084475517273,
|
|
"step": 4185,
|
|
"valid_targets_mean": 4556.6,
|
|
"valid_targets_min": 1156
|
|
},
|
|
{
|
|
"epoch": 6.682615629984051,
|
|
"grad_norm": 0.48588135941150645,
|
|
"learning_rate": 2.524931612733328e-07,
|
|
"loss": 0.1315,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12049822509288788,
|
|
"step": 4190,
|
|
"valid_targets_mean": 4792.8,
|
|
"valid_targets_min": 583
|
|
},
|
|
{
|
|
"epoch": 6.690590111642743,
|
|
"grad_norm": 0.5157831592987673,
|
|
"learning_rate": 2.400513121009529e-07,
|
|
"loss": 0.1399,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12056033313274384,
|
|
"step": 4195,
|
|
"valid_targets_mean": 3902.8,
|
|
"valid_targets_min": 900
|
|
},
|
|
{
|
|
"epoch": 6.698564593301436,
|
|
"grad_norm": 0.6181941397201824,
|
|
"learning_rate": 2.279219490998985e-07,
|
|
"loss": 0.1555,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12805788218975067,
|
|
"step": 4200,
|
|
"valid_targets_mean": 4067.9,
|
|
"valid_targets_min": 448
|
|
},
|
|
{
|
|
"epoch": 6.706539074960127,
|
|
"grad_norm": 0.5411702652574677,
|
|
"learning_rate": 2.161052640853578e-07,
|
|
"loss": 0.1597,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13948936760425568,
|
|
"step": 4205,
|
|
"valid_targets_mean": 4612.0,
|
|
"valid_targets_min": 2172
|
|
},
|
|
{
|
|
"epoch": 6.71451355661882,
|
|
"grad_norm": 0.5203767410756568,
|
|
"learning_rate": 2.0460144392778768e-07,
|
|
"loss": 0.1352,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14677637815475464,
|
|
"step": 4210,
|
|
"valid_targets_mean": 4351.0,
|
|
"valid_targets_min": 1795
|
|
},
|
|
{
|
|
"epoch": 6.722488038277512,
|
|
"grad_norm": 0.48076115868735375,
|
|
"learning_rate": 1.9341067054996277e-07,
|
|
"loss": 0.1355,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12446603178977966,
|
|
"step": 4215,
|
|
"valid_targets_mean": 4783.4,
|
|
"valid_targets_min": 1169
|
|
},
|
|
{
|
|
"epoch": 6.7304625199362045,
|
|
"grad_norm": 0.4938032695536993,
|
|
"learning_rate": 1.8253312092409992e-07,
|
|
"loss": 0.1464,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14762309193611145,
|
|
"step": 4220,
|
|
"valid_targets_mean": 5270.6,
|
|
"valid_targets_min": 2308
|
|
},
|
|
{
|
|
"epoch": 6.738437001594896,
|
|
"grad_norm": 0.5073690650738787,
|
|
"learning_rate": 1.7196896706906273e-07,
|
|
"loss": 0.1476,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13876202702522278,
|
|
"step": 4225,
|
|
"valid_targets_mean": 4535.9,
|
|
"valid_targets_min": 2133
|
|
},
|
|
{
|
|
"epoch": 6.746411483253588,
|
|
"grad_norm": 0.6239349122520454,
|
|
"learning_rate": 1.6171837604762597e-07,
|
|
"loss": 0.1394,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14759038388729095,
|
|
"step": 4230,
|
|
"valid_targets_mean": 3153.2,
|
|
"valid_targets_min": 712
|
|
},
|
|
{
|
|
"epoch": 6.754385964912281,
|
|
"grad_norm": 0.5294503911396128,
|
|
"learning_rate": 1.5178150996385755e-07,
|
|
"loss": 0.1298,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14934271574020386,
|
|
"step": 4235,
|
|
"valid_targets_mean": 4826.8,
|
|
"valid_targets_min": 2045
|
|
},
|
|
{
|
|
"epoch": 6.762360446570973,
|
|
"grad_norm": 0.5474589095143887,
|
|
"learning_rate": 1.421585259605318e-07,
|
|
"loss": 0.1569,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17565397918224335,
|
|
"step": 4240,
|
|
"valid_targets_mean": 3994.0,
|
|
"valid_targets_min": 811
|
|
},
|
|
{
|
|
"epoch": 6.770334928229665,
|
|
"grad_norm": 0.4968111197576825,
|
|
"learning_rate": 1.3284957621666039e-07,
|
|
"loss": 0.1449,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15295177698135376,
|
|
"step": 4245,
|
|
"valid_targets_mean": 5280.7,
|
|
"valid_targets_min": 2153
|
|
},
|
|
{
|
|
"epoch": 6.778309409888357,
|
|
"grad_norm": 0.5712193249343777,
|
|
"learning_rate": 1.2385480794507853e-07,
|
|
"loss": 0.1565,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15290597081184387,
|
|
"step": 4250,
|
|
"valid_targets_mean": 4323.8,
|
|
"valid_targets_min": 1528
|
|
},
|
|
{
|
|
"epoch": 6.78628389154705,
|
|
"grad_norm": 0.5387472877772308,
|
|
"learning_rate": 1.1517436339011589e-07,
|
|
"loss": 0.1368,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16408362984657288,
|
|
"step": 4255,
|
|
"valid_targets_mean": 4533.3,
|
|
"valid_targets_min": 1733
|
|
},
|
|
{
|
|
"epoch": 6.794258373205742,
|
|
"grad_norm": 0.5262055497554436,
|
|
"learning_rate": 1.0680837982535607e-07,
|
|
"loss": 0.1491,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17279741168022156,
|
|
"step": 4260,
|
|
"valid_targets_mean": 4844.5,
|
|
"valid_targets_min": 1115
|
|
},
|
|
{
|
|
"epoch": 6.8022328548644335,
|
|
"grad_norm": 0.5393251228907873,
|
|
"learning_rate": 9.875698955145174e-08,
|
|
"loss": 0.1503,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.17203345894813538,
|
|
"step": 4265,
|
|
"valid_targets_mean": 4560.6,
|
|
"valid_targets_min": 835
|
|
},
|
|
{
|
|
"epoch": 6.810207336523126,
|
|
"grad_norm": 0.4950986090823543,
|
|
"learning_rate": 9.102031989404403e-08,
|
|
"loss": 0.132,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1278517246246338,
|
|
"step": 4270,
|
|
"valid_targets_mean": 4280.2,
|
|
"valid_targets_min": 271
|
|
},
|
|
{
|
|
"epoch": 6.818181818181818,
|
|
"grad_norm": 0.5322683037430497,
|
|
"learning_rate": 8.359849320174196e-08,
|
|
"loss": 0.1405,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14759370684623718,
|
|
"step": 4275,
|
|
"valid_targets_mean": 4334.2,
|
|
"valid_targets_min": 1141
|
|
},
|
|
{
|
|
"epoch": 6.826156299840511,
|
|
"grad_norm": 0.5106524077335042,
|
|
"learning_rate": 7.649162684419731e-08,
|
|
"loss": 0.1324,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13630405068397522,
|
|
"step": 4280,
|
|
"valid_targets_mean": 5122.1,
|
|
"valid_targets_min": 1651
|
|
},
|
|
{
|
|
"epoch": 6.834130781499202,
|
|
"grad_norm": 0.48090627611402476,
|
|
"learning_rate": 6.969983321023499e-08,
|
|
"loss": 0.1369,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14416742324829102,
|
|
"step": 4285,
|
|
"valid_targets_mean": 4816.6,
|
|
"valid_targets_min": 1821
|
|
},
|
|
{
|
|
"epoch": 6.842105263157895,
|
|
"grad_norm": 0.48778544545536157,
|
|
"learning_rate": 6.322321970608337e-08,
|
|
"loss": 0.1506,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1339418888092041,
|
|
"step": 4290,
|
|
"valid_targets_mean": 4488.4,
|
|
"valid_targets_min": 1616
|
|
},
|
|
{
|
|
"epoch": 6.850079744816587,
|
|
"grad_norm": 0.46499159243236704,
|
|
"learning_rate": 5.7061888753677796e-08,
|
|
"loss": 0.1413,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14237193763256073,
|
|
"step": 4295,
|
|
"valid_targets_mean": 4920.9,
|
|
"valid_targets_min": 1271
|
|
},
|
|
{
|
|
"epoch": 6.858054226475279,
|
|
"grad_norm": 0.4818822507608203,
|
|
"learning_rate": 5.121593778903755e-08,
|
|
"loss": 0.1341,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13319122791290283,
|
|
"step": 4300,
|
|
"valid_targets_mean": 4507.4,
|
|
"valid_targets_min": 2343
|
|
},
|
|
{
|
|
"epoch": 6.866028708133971,
|
|
"grad_norm": 0.5030961917930569,
|
|
"learning_rate": 4.5685459260722544e-08,
|
|
"loss": 0.1348,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15166345238685608,
|
|
"step": 4305,
|
|
"valid_targets_mean": 4678.6,
|
|
"valid_targets_min": 1154
|
|
},
|
|
{
|
|
"epoch": 6.8740031897926634,
|
|
"grad_norm": 0.5753815923974399,
|
|
"learning_rate": 4.047054062837452e-08,
|
|
"loss": 0.1378,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.16344240307807922,
|
|
"step": 4310,
|
|
"valid_targets_mean": 4398.2,
|
|
"valid_targets_min": 1732
|
|
},
|
|
{
|
|
"epoch": 6.881977671451356,
|
|
"grad_norm": 0.5355466593328079,
|
|
"learning_rate": 3.55712643613404e-08,
|
|
"loss": 0.1419,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14519166946411133,
|
|
"step": 4315,
|
|
"valid_targets_mean": 4389.8,
|
|
"valid_targets_min": 1213
|
|
},
|
|
{
|
|
"epoch": 6.889952153110048,
|
|
"grad_norm": 0.5278125498932654,
|
|
"learning_rate": 3.0987707937351066e-08,
|
|
"loss": 0.1428,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13044758141040802,
|
|
"step": 4320,
|
|
"valid_targets_mean": 4127.8,
|
|
"valid_targets_min": 1647
|
|
},
|
|
{
|
|
"epoch": 6.89792663476874,
|
|
"grad_norm": 0.4856712767135322,
|
|
"learning_rate": 2.6719943841311268e-08,
|
|
"loss": 0.1372,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13070060312747955,
|
|
"step": 4325,
|
|
"valid_targets_mean": 4700.8,
|
|
"valid_targets_min": 608
|
|
},
|
|
{
|
|
"epoch": 6.905901116427432,
|
|
"grad_norm": 0.6415911642154425,
|
|
"learning_rate": 2.2768039564151635e-08,
|
|
"loss": 0.1403,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1640297770500183,
|
|
"step": 4330,
|
|
"valid_targets_mean": 3901.8,
|
|
"valid_targets_min": 655
|
|
},
|
|
{
|
|
"epoch": 6.913875598086125,
|
|
"grad_norm": 0.6626092253360519,
|
|
"learning_rate": 1.913205760175174e-08,
|
|
"loss": 0.1442,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1504662036895752,
|
|
"step": 4335,
|
|
"valid_targets_mean": 5199.6,
|
|
"valid_targets_min": 1387
|
|
},
|
|
{
|
|
"epoch": 6.921850079744816,
|
|
"grad_norm": 0.5982491019172199,
|
|
"learning_rate": 1.5812055453963136e-08,
|
|
"loss": 0.1383,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14008523523807526,
|
|
"step": 4340,
|
|
"valid_targets_mean": 4319.2,
|
|
"valid_targets_min": 1729
|
|
},
|
|
{
|
|
"epoch": 6.9298245614035086,
|
|
"grad_norm": 0.5104174835877258,
|
|
"learning_rate": 1.280808562369229e-08,
|
|
"loss": 0.14,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.133617103099823,
|
|
"step": 4345,
|
|
"valid_targets_mean": 4219.4,
|
|
"valid_targets_min": 1451
|
|
},
|
|
{
|
|
"epoch": 6.937799043062201,
|
|
"grad_norm": 0.4877409631231481,
|
|
"learning_rate": 1.01201956160768e-08,
|
|
"loss": 0.144,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.174456387758255,
|
|
"step": 4350,
|
|
"valid_targets_mean": 4994.6,
|
|
"valid_targets_min": 1946
|
|
},
|
|
{
|
|
"epoch": 6.945773524720893,
|
|
"grad_norm": 0.5184348308045106,
|
|
"learning_rate": 7.74842793772601e-09,
|
|
"loss": 0.1363,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.1534990817308426,
|
|
"step": 4355,
|
|
"valid_targets_mean": 4945.9,
|
|
"valid_targets_min": 1800
|
|
},
|
|
{
|
|
"epoch": 6.953748006379586,
|
|
"grad_norm": 0.5492476418437654,
|
|
"learning_rate": 5.692820096054874e-09,
|
|
"loss": 0.1349,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.14275865256786346,
|
|
"step": 4360,
|
|
"valid_targets_mean": 4028.4,
|
|
"valid_targets_min": 1148
|
|
},
|
|
{
|
|
"epoch": 6.961722488038277,
|
|
"grad_norm": 0.5057721726447473,
|
|
"learning_rate": 3.9534045986888706e-09,
|
|
"loss": 0.1375,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.13061657547950745,
|
|
"step": 4365,
|
|
"valid_targets_mean": 4621.9,
|
|
"valid_targets_min": 1483
|
|
},
|
|
{
|
|
"epoch": 6.96969696969697,
|
|
"grad_norm": 0.5165832319676158,
|
|
"learning_rate": 2.530208952953306e-09,
|
|
"loss": 0.1369,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.15254126489162445,
|
|
"step": 4370,
|
|
"valid_targets_mean": 4441.8,
|
|
"valid_targets_min": 1501
|
|
},
|
|
{
|
|
"epoch": 6.977671451355662,
|
|
"grad_norm": 0.4938874383244835,
|
|
"learning_rate": 1.4232556654314445e-09,
|
|
"loss": 0.1412,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12354663759469986,
|
|
"step": 4375,
|
|
"valid_targets_mean": 4462.6,
|
|
"valid_targets_min": 396
|
|
},
|
|
{
|
|
"epoch": 6.985645933014354,
|
|
"grad_norm": 0.5172931781731442,
|
|
"learning_rate": 6.325622416136767e-10,
|
|
"loss": 0.1396,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12898609042167664,
|
|
"step": 4380,
|
|
"valid_targets_mean": 4070.1,
|
|
"valid_targets_min": 1645
|
|
},
|
|
{
|
|
"epoch": 6.993620414673046,
|
|
"grad_norm": 0.5813118538526274,
|
|
"learning_rate": 1.581411856199644e-10,
|
|
"loss": 0.1526,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12935660779476166,
|
|
"step": 4385,
|
|
"valid_targets_mean": 3541.5,
|
|
"valid_targets_min": 1587
|
|
},
|
|
{
|
|
"epoch": 7.0,
|
|
"loss_nan_ranks": 0,
|
|
"loss_rank_avg": 0.12793053686618805,
|
|
"step": 4389,
|
|
"total_flos": 1406471795245056.0,
|
|
"train_loss": 0.18805862067907703,
|
|
"train_runtime": 23263.3401,
|
|
"train_samples_per_second": 3.014,
|
|
"train_steps_per_second": 0.189,
|
|
"valid_targets_mean": 3981.4,
|
|
"valid_targets_min": 1669
|
|
}
|
|
],
|
|
"logging_steps": 5,
|
|
"max_steps": 4389,
|
|
"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": 1406471795245056.0,
|
|
"train_batch_size": 1,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|