| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 5.0, |
| "eval_steps": 500, |
| "global_step": 780, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0322061191626409, |
| "grad_norm": 12.898349962866439, |
| "learning_rate": 2.0512820512820513e-06, |
| "loss": 0.8621, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.11310747265815735, |
| "step": 5, |
| "valid_targets_mean": 552.5, |
| "valid_targets_min": 297 |
| }, |
| { |
| "epoch": 0.0644122383252818, |
| "grad_norm": 6.85206740052359, |
| "learning_rate": 4.615384615384616e-06, |
| "loss": 0.872, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.11378927528858185, |
| "step": 10, |
| "valid_targets_mean": 1438.5, |
| "valid_targets_min": 332 |
| }, |
| { |
| "epoch": 0.0966183574879227, |
| "grad_norm": 3.405492909651102, |
| "learning_rate": 7.17948717948718e-06, |
| "loss": 0.7405, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06183416768908501, |
| "step": 15, |
| "valid_targets_mean": 580.8, |
| "valid_targets_min": 260 |
| }, |
| { |
| "epoch": 0.1288244766505636, |
| "grad_norm": 1.903925034713442, |
| "learning_rate": 9.743589743589744e-06, |
| "loss": 0.7208, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.08290648460388184, |
| "step": 20, |
| "valid_targets_mean": 922.6, |
| "valid_targets_min": 337 |
| }, |
| { |
| "epoch": 0.1610305958132045, |
| "grad_norm": 1.3698491699971576, |
| "learning_rate": 1.230769230769231e-05, |
| "loss": 0.6653, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.08934478461742401, |
| "step": 25, |
| "valid_targets_mean": 716.9, |
| "valid_targets_min": 355 |
| }, |
| { |
| "epoch": 0.1932367149758454, |
| "grad_norm": 1.0986695170310006, |
| "learning_rate": 1.4871794871794874e-05, |
| "loss": 0.6276, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.07862719893455505, |
| "step": 30, |
| "valid_targets_mean": 893.6, |
| "valid_targets_min": 342 |
| }, |
| { |
| "epoch": 0.22544283413848631, |
| "grad_norm": 0.8812267490478244, |
| "learning_rate": 1.7435897435897438e-05, |
| "loss": 0.615, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.12738855183124542, |
| "step": 35, |
| "valid_targets_mean": 1057.4, |
| "valid_targets_min": 363 |
| }, |
| { |
| "epoch": 0.2576489533011272, |
| "grad_norm": 0.8401277977595764, |
| "learning_rate": 2e-05, |
| "loss": 0.5848, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.0551215335726738, |
| "step": 40, |
| "valid_targets_mean": 723.9, |
| "valid_targets_min": 292 |
| }, |
| { |
| "epoch": 0.2898550724637681, |
| "grad_norm": 0.6745283792399535, |
| "learning_rate": 2.2564102564102566e-05, |
| "loss": 0.5279, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05893056094646454, |
| "step": 45, |
| "valid_targets_mean": 787.1, |
| "valid_targets_min": 285 |
| }, |
| { |
| "epoch": 0.322061191626409, |
| "grad_norm": 0.6559777036049393, |
| "learning_rate": 2.512820512820513e-05, |
| "loss": 0.5298, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06752099096775055, |
| "step": 50, |
| "valid_targets_mean": 897.5, |
| "valid_targets_min": 221 |
| }, |
| { |
| "epoch": 0.35426731078904994, |
| "grad_norm": 0.5512846197250572, |
| "learning_rate": 2.7692307692307694e-05, |
| "loss": 0.4835, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.10814902186393738, |
| "step": 55, |
| "valid_targets_mean": 1876.0, |
| "valid_targets_min": 240 |
| }, |
| { |
| "epoch": 0.3864734299516908, |
| "grad_norm": 0.4993982742101459, |
| "learning_rate": 3.0256410256410257e-05, |
| "loss": 0.4953, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06639184057712555, |
| "step": 60, |
| "valid_targets_mean": 1391.9, |
| "valid_targets_min": 344 |
| }, |
| { |
| "epoch": 0.41867954911433175, |
| "grad_norm": 0.6383132839926805, |
| "learning_rate": 3.282051282051282e-05, |
| "loss": 0.4583, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.07518483698368073, |
| "step": 65, |
| "valid_targets_mean": 898.4, |
| "valid_targets_min": 327 |
| }, |
| { |
| "epoch": 0.45088566827697263, |
| "grad_norm": 0.5621105173468922, |
| "learning_rate": 3.538461538461539e-05, |
| "loss": 0.469, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.028787091374397278, |
| "step": 70, |
| "valid_targets_mean": 518.4, |
| "valid_targets_min": 304 |
| }, |
| { |
| "epoch": 0.4830917874396135, |
| "grad_norm": 0.5544434814590523, |
| "learning_rate": 3.794871794871795e-05, |
| "loss": 0.4697, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04727187752723694, |
| "step": 75, |
| "valid_targets_mean": 674.9, |
| "valid_targets_min": 298 |
| }, |
| { |
| "epoch": 0.5152979066022544, |
| "grad_norm": 0.4882906127614797, |
| "learning_rate": 3.999979972589652e-05, |
| "loss": 0.4368, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04346778243780136, |
| "step": 80, |
| "valid_targets_mean": 946.8, |
| "valid_targets_min": 300 |
| }, |
| { |
| "epoch": 0.5475040257648953, |
| "grad_norm": 0.9150926943645863, |
| "learning_rate": 3.999279055341771e-05, |
| "loss": 0.4631, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.09739387035369873, |
| "step": 85, |
| "valid_targets_mean": 1282.0, |
| "valid_targets_min": 408 |
| }, |
| { |
| "epoch": 0.5797101449275363, |
| "grad_norm": 0.6709571649722427, |
| "learning_rate": 3.9975771686375434e-05, |
| "loss": 0.4388, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03802517428994179, |
| "step": 90, |
| "valid_targets_mean": 515.5, |
| "valid_targets_min": 329 |
| }, |
| { |
| "epoch": 0.6119162640901772, |
| "grad_norm": 0.8539827107022786, |
| "learning_rate": 3.9948751645524235e-05, |
| "loss": 0.4325, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.062230631709098816, |
| "step": 95, |
| "valid_targets_mean": 573.1, |
| "valid_targets_min": 335 |
| }, |
| { |
| "epoch": 0.644122383252818, |
| "grad_norm": 0.6365846252008591, |
| "learning_rate": 3.9911743958858376e-05, |
| "loss": 0.4566, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04755745083093643, |
| "step": 100, |
| "valid_targets_mean": 709.1, |
| "valid_targets_min": 322 |
| }, |
| { |
| "epoch": 0.6763285024154589, |
| "grad_norm": 0.4964193415485056, |
| "learning_rate": 3.9864767154838864e-05, |
| "loss": 0.4385, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05032774806022644, |
| "step": 105, |
| "valid_targets_mean": 1279.0, |
| "valid_targets_min": 295 |
| }, |
| { |
| "epoch": 0.7085346215780999, |
| "grad_norm": 0.5582500327494776, |
| "learning_rate": 3.9807844753116864e-05, |
| "loss": 0.4292, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.13067539036273956, |
| "step": 110, |
| "valid_targets_mean": 2418.5, |
| "valid_targets_min": 276 |
| }, |
| { |
| "epoch": 0.7407407407407407, |
| "grad_norm": 0.5605455694371961, |
| "learning_rate": 3.9741005252758255e-05, |
| "loss": 0.4141, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03115726262331009, |
| "step": 115, |
| "valid_targets_mean": 681.2, |
| "valid_targets_min": 283 |
| }, |
| { |
| "epoch": 0.7729468599033816, |
| "grad_norm": 0.4737117451016544, |
| "learning_rate": 3.966428211797514e-05, |
| "loss": 0.441, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06667988002300262, |
| "step": 120, |
| "valid_targets_mean": 1534.2, |
| "valid_targets_min": 305 |
| }, |
| { |
| "epoch": 0.8051529790660226, |
| "grad_norm": 0.43538248655180545, |
| "learning_rate": 3.957771376137144e-05, |
| "loss": 0.4257, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04947737231850624, |
| "step": 125, |
| "valid_targets_mean": 1130.1, |
| "valid_targets_min": 308 |
| }, |
| { |
| "epoch": 0.8373590982286635, |
| "grad_norm": 0.48578578549548673, |
| "learning_rate": 3.9481343524711095e-05, |
| "loss": 0.4285, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.036311663687229156, |
| "step": 130, |
| "valid_targets_mean": 1009.8, |
| "valid_targets_min": 333 |
| }, |
| { |
| "epoch": 0.8695652173913043, |
| "grad_norm": 0.4857684316354408, |
| "learning_rate": 3.937521965721831e-05, |
| "loss": 0.4334, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.058429233729839325, |
| "step": 135, |
| "valid_targets_mean": 1255.6, |
| "valid_targets_min": 379 |
| }, |
| { |
| "epoch": 0.9017713365539453, |
| "grad_norm": 0.536215147216915, |
| "learning_rate": 3.9259395291420864e-05, |
| "loss": 0.4324, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.02372780255973339, |
| "step": 140, |
| "valid_targets_mean": 394.0, |
| "valid_targets_min": 283 |
| }, |
| { |
| "epoch": 0.9339774557165862, |
| "grad_norm": 0.4911344506126531, |
| "learning_rate": 3.913392841654851e-05, |
| "loss": 0.4102, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.021754739806056023, |
| "step": 145, |
| "valid_targets_mean": 400.6, |
| "valid_targets_min": 290 |
| }, |
| { |
| "epoch": 0.966183574879227, |
| "grad_norm": 0.5895368205854837, |
| "learning_rate": 3.8998881849499756e-05, |
| "loss": 0.4187, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.10034637898206711, |
| "step": 150, |
| "valid_targets_mean": 1143.1, |
| "valid_targets_min": 321 |
| }, |
| { |
| "epoch": 0.998389694041868, |
| "grad_norm": 0.4934508890573737, |
| "learning_rate": 3.885432320339167e-05, |
| "loss": 0.4174, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.08337146043777466, |
| "step": 155, |
| "valid_targets_mean": 2523.6, |
| "valid_targets_min": 281 |
| }, |
| { |
| "epoch": 1.0257648953301126, |
| "grad_norm": 0.6397313297428109, |
| "learning_rate": 3.8700324853708304e-05, |
| "loss": 0.4417, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.059281207621097565, |
| "step": 160, |
| "valid_targets_mean": 753.4, |
| "valid_targets_min": 309 |
| }, |
| { |
| "epoch": 1.0579710144927537, |
| "grad_norm": 0.46415104272287283, |
| "learning_rate": 3.853696390206484e-05, |
| "loss": 0.3984, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.028974976390600204, |
| "step": 165, |
| "valid_targets_mean": 518.6, |
| "valid_targets_min": 336 |
| }, |
| { |
| "epoch": 1.0901771336553945, |
| "grad_norm": 0.5352714190562025, |
| "learning_rate": 3.8364322137605484e-05, |
| "loss": 0.3912, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.054242536425590515, |
| "step": 170, |
| "valid_targets_mean": 1146.2, |
| "valid_targets_min": 307 |
| }, |
| { |
| "epoch": 1.1223832528180355, |
| "grad_norm": 0.5623597503666744, |
| "learning_rate": 3.818248599605448e-05, |
| "loss": 0.3861, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.0729055255651474, |
| "step": 175, |
| "valid_targets_mean": 1848.2, |
| "valid_targets_min": 325 |
| }, |
| { |
| "epoch": 1.1545893719806763, |
| "grad_norm": 0.5742605838509625, |
| "learning_rate": 3.799154651644075e-05, |
| "loss": 0.397, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05786103010177612, |
| "step": 180, |
| "valid_targets_mean": 768.2, |
| "valid_targets_min": 255 |
| }, |
| { |
| "epoch": 1.1867954911433172, |
| "grad_norm": 0.48599772051318924, |
| "learning_rate": 3.7791599295517825e-05, |
| "loss": 0.3868, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.044465214014053345, |
| "step": 185, |
| "valid_targets_mean": 953.5, |
| "valid_targets_min": 305 |
| }, |
| { |
| "epoch": 1.2190016103059582, |
| "grad_norm": 0.4427239963256647, |
| "learning_rate": 3.75827444399019e-05, |
| "loss": 0.4163, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03635489195585251, |
| "step": 190, |
| "valid_targets_mean": 787.0, |
| "valid_targets_min": 287 |
| }, |
| { |
| "epoch": 1.251207729468599, |
| "grad_norm": 0.46071305423252523, |
| "learning_rate": 3.736508651595188e-05, |
| "loss": 0.3759, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.1078839898109436, |
| "step": 195, |
| "valid_targets_mean": 1857.1, |
| "valid_targets_min": 310 |
| }, |
| { |
| "epoch": 1.2834138486312399, |
| "grad_norm": 0.45110054102426617, |
| "learning_rate": 3.713873449741669e-05, |
| "loss": 0.4067, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.040066909044981, |
| "step": 200, |
| "valid_targets_mean": 1158.4, |
| "valid_targets_min": 337 |
| }, |
| { |
| "epoch": 1.315619967793881, |
| "grad_norm": 0.4952144859012548, |
| "learning_rate": 3.69038017108759e-05, |
| "loss": 0.3931, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.02214495651423931, |
| "step": 205, |
| "valid_targets_mean": 509.6, |
| "valid_targets_min": 257 |
| }, |
| { |
| "epoch": 1.3478260869565217, |
| "grad_norm": 0.4303160555876925, |
| "learning_rate": 3.6660405779001036e-05, |
| "loss": 0.4067, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.10445412993431091, |
| "step": 210, |
| "valid_targets_mean": 3890.0, |
| "valid_targets_min": 318 |
| }, |
| { |
| "epoch": 1.3800322061191626, |
| "grad_norm": 0.4365789322635305, |
| "learning_rate": 3.640866856166601e-05, |
| "loss": 0.3827, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.0636882334947586, |
| "step": 215, |
| "valid_targets_mean": 1477.5, |
| "valid_targets_min": 390 |
| }, |
| { |
| "epoch": 1.4122383252818036, |
| "grad_norm": 0.4837420644157074, |
| "learning_rate": 3.614871609493614e-05, |
| "loss": 0.3846, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05718361958861351, |
| "step": 220, |
| "valid_targets_mean": 1256.9, |
| "valid_targets_min": 353 |
| }, |
| { |
| "epoch": 1.4444444444444444, |
| "grad_norm": 0.47344211330731195, |
| "learning_rate": 3.5880678527966224e-05, |
| "loss": 0.3729, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.07174763828516006, |
| "step": 225, |
| "valid_targets_mean": 1642.0, |
| "valid_targets_min": 260 |
| }, |
| { |
| "epoch": 1.4766505636070852, |
| "grad_norm": 0.5547472032191869, |
| "learning_rate": 3.5604690057839456e-05, |
| "loss": 0.3808, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04665002226829529, |
| "step": 230, |
| "valid_targets_mean": 620.6, |
| "valid_targets_min": 358 |
| }, |
| { |
| "epoch": 1.5088566827697263, |
| "grad_norm": 0.5106632095827158, |
| "learning_rate": 3.532088886237956e-05, |
| "loss": 0.3796, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.060500893741846085, |
| "step": 235, |
| "valid_targets_mean": 1519.4, |
| "valid_targets_min": 288 |
| }, |
| { |
| "epoch": 1.541062801932367, |
| "grad_norm": 0.4498419929162088, |
| "learning_rate": 3.5029417030970005e-05, |
| "loss": 0.4032, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.027759376913309097, |
| "step": 240, |
| "valid_targets_mean": 679.4, |
| "valid_targets_min": 294 |
| }, |
| { |
| "epoch": 1.573268921095008, |
| "grad_norm": 0.4631774519072297, |
| "learning_rate": 3.473042049341474e-05, |
| "loss": 0.3899, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04057846963405609, |
| "step": 245, |
| "valid_targets_mean": 1025.0, |
| "valid_targets_min": 218 |
| }, |
| { |
| "epoch": 1.605475040257649, |
| "grad_norm": 0.47311421989321817, |
| "learning_rate": 3.442404894687629e-05, |
| "loss": 0.395, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.027594510465860367, |
| "step": 250, |
| "valid_targets_mean": 574.0, |
| "valid_targets_min": 269 |
| }, |
| { |
| "epoch": 1.6376811594202898, |
| "grad_norm": 0.6220706207123248, |
| "learning_rate": 3.411045578092754e-05, |
| "loss": 0.4178, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.02938500978052616, |
| "step": 255, |
| "valid_targets_mean": 496.4, |
| "valid_targets_min": 325 |
| }, |
| { |
| "epoch": 1.6698872785829306, |
| "grad_norm": 0.5013668028351838, |
| "learning_rate": 3.3789798000754924e-05, |
| "loss": 0.3814, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.027927782386541367, |
| "step": 260, |
| "valid_targets_mean": 767.4, |
| "valid_targets_min": 263 |
| }, |
| { |
| "epoch": 1.7020933977455717, |
| "grad_norm": 0.5397613733002328, |
| "learning_rate": 3.34622361485514e-05, |
| "loss": 0.3818, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03136827051639557, |
| "step": 265, |
| "valid_targets_mean": 570.8, |
| "valid_targets_min": 341 |
| }, |
| { |
| "epoch": 1.7342995169082127, |
| "grad_norm": 0.5549058173437013, |
| "learning_rate": 3.31279342231386e-05, |
| "loss": 0.3931, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05748458951711655, |
| "step": 270, |
| "valid_targets_mean": 809.4, |
| "valid_targets_min": 263 |
| }, |
| { |
| "epoch": 1.7665056360708533, |
| "grad_norm": 0.5121030877701617, |
| "learning_rate": 3.278705959785821e-05, |
| "loss": 0.4033, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03404642269015312, |
| "step": 275, |
| "valid_targets_mean": 479.6, |
| "valid_targets_min": 275 |
| }, |
| { |
| "epoch": 1.7987117552334944, |
| "grad_norm": 0.4382008464023779, |
| "learning_rate": 3.243978293677408e-05, |
| "loss": 0.3693, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03826048597693443, |
| "step": 280, |
| "valid_targets_mean": 1542.5, |
| "valid_targets_min": 317 |
| }, |
| { |
| "epoch": 1.8309178743961354, |
| "grad_norm": 0.6050902730329194, |
| "learning_rate": 3.208627810922665e-05, |
| "loss": 0.3794, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.058419372886419296, |
| "step": 285, |
| "valid_targets_mean": 855.5, |
| "valid_targets_min": 303 |
| }, |
| { |
| "epoch": 1.863123993558776, |
| "grad_norm": 0.5038584214895873, |
| "learning_rate": 3.172672210278254e-05, |
| "loss": 0.3898, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05743902176618576, |
| "step": 290, |
| "valid_targets_mean": 1938.6, |
| "valid_targets_min": 333 |
| }, |
| { |
| "epoch": 1.895330112721417, |
| "grad_norm": 0.564278397643887, |
| "learning_rate": 3.136129493462312e-05, |
| "loss": 0.4063, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04776623472571373, |
| "step": 295, |
| "valid_targets_mean": 636.6, |
| "valid_targets_min": 333 |
| }, |
| { |
| "epoch": 1.927536231884058, |
| "grad_norm": 0.5937603264774977, |
| "learning_rate": 3.0990179561416124e-05, |
| "loss": 0.3934, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06801211833953857, |
| "step": 300, |
| "valid_targets_mean": 944.6, |
| "valid_targets_min": 340 |
| }, |
| { |
| "epoch": 1.959742351046699, |
| "grad_norm": 0.4964999875942384, |
| "learning_rate": 3.061356178771564e-05, |
| "loss": 0.3916, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03988070785999298, |
| "step": 305, |
| "valid_targets_mean": 664.6, |
| "valid_targets_min": 292 |
| }, |
| { |
| "epoch": 1.9919484702093397, |
| "grad_norm": 0.5661676600187325, |
| "learning_rate": 3.0231630172936205e-05, |
| "loss": 0.397, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.041811276227235794, |
| "step": 310, |
| "valid_targets_mean": 593.0, |
| "valid_targets_min": 265 |
| }, |
| { |
| "epoch": 2.0193236714975846, |
| "grad_norm": 0.5035992902460946, |
| "learning_rate": 2.98445759369477e-05, |
| "loss": 0.3887, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.024304592981934547, |
| "step": 315, |
| "valid_targets_mean": 545.0, |
| "valid_targets_min": 335 |
| }, |
| { |
| "epoch": 2.0515297906602252, |
| "grad_norm": 0.4746849218727487, |
| "learning_rate": 2.945259286433819e-05, |
| "loss": 0.3821, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.035836152732372284, |
| "step": 320, |
| "valid_targets_mean": 911.9, |
| "valid_targets_min": 302 |
| }, |
| { |
| "epoch": 2.0837359098228663, |
| "grad_norm": 0.6069458438713761, |
| "learning_rate": 2.9055877207392752e-05, |
| "loss": 0.3795, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04563501104712486, |
| "step": 325, |
| "valid_targets_mean": 767.9, |
| "valid_targets_min": 316 |
| }, |
| { |
| "epoch": 2.1159420289855073, |
| "grad_norm": 0.464114101694879, |
| "learning_rate": 2.8654627587836724e-05, |
| "loss": 0.3588, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.0893658697605133, |
| "step": 330, |
| "valid_targets_mean": 2450.0, |
| "valid_targets_min": 416 |
| }, |
| { |
| "epoch": 2.148148148148148, |
| "grad_norm": 0.4979196420820879, |
| "learning_rate": 2.8249044897392814e-05, |
| "loss": 0.3649, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.051502641290426254, |
| "step": 335, |
| "valid_targets_mean": 1717.1, |
| "valid_targets_min": 244 |
| }, |
| { |
| "epoch": 2.180354267310789, |
| "grad_norm": 0.5588899086430609, |
| "learning_rate": 2.7839332197201505e-05, |
| "loss": 0.3799, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.045202601701021194, |
| "step": 340, |
| "valid_targets_mean": 951.6, |
| "valid_targets_min": 360 |
| }, |
| { |
| "epoch": 2.21256038647343, |
| "grad_norm": 0.4764396199562437, |
| "learning_rate": 2.7425694616155474e-05, |
| "loss": 0.3623, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06349411606788635, |
| "step": 345, |
| "valid_targets_mean": 1610.1, |
| "valid_targets_min": 298 |
| }, |
| { |
| "epoch": 2.244766505636071, |
| "grad_norm": 0.4543103599849946, |
| "learning_rate": 2.700833924819868e-05, |
| "loss": 0.3587, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.042356789112091064, |
| "step": 350, |
| "valid_targets_mean": 1044.9, |
| "valid_targets_min": 348 |
| }, |
| { |
| "epoch": 2.2769726247987117, |
| "grad_norm": 0.4714348506524609, |
| "learning_rate": 2.6587475048641596e-05, |
| "loss": 0.3681, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.040921151638031006, |
| "step": 355, |
| "valid_targets_mean": 958.1, |
| "valid_targets_min": 252 |
| }, |
| { |
| "epoch": 2.3091787439613527, |
| "grad_norm": 0.45052091460255717, |
| "learning_rate": 2.6163312729544616e-05, |
| "loss": 0.3651, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.0605086013674736, |
| "step": 360, |
| "valid_targets_mean": 1398.6, |
| "valid_targets_min": 407 |
| }, |
| { |
| "epoch": 2.3413848631239937, |
| "grad_norm": 0.8153337300604954, |
| "learning_rate": 2.5736064654221808e-05, |
| "loss": 0.3775, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.035735368728637695, |
| "step": 365, |
| "valid_targets_mean": 977.4, |
| "valid_targets_min": 308 |
| }, |
| { |
| "epoch": 2.3735909822866343, |
| "grad_norm": 0.469458881165414, |
| "learning_rate": 2.5305944730918017e-05, |
| "loss": 0.3979, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.031210483983159065, |
| "step": 370, |
| "valid_targets_mean": 778.0, |
| "valid_targets_min": 251 |
| }, |
| { |
| "epoch": 2.4057971014492754, |
| "grad_norm": 0.5033946973514036, |
| "learning_rate": 2.487316830571244e-05, |
| "loss": 0.344, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.0375363752245903, |
| "step": 375, |
| "valid_targets_mean": 912.9, |
| "valid_targets_min": 282 |
| }, |
| { |
| "epoch": 2.4380032206119164, |
| "grad_norm": 0.5816567624854461, |
| "learning_rate": 2.4437952054702354e-05, |
| "loss": 0.3515, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.029283087700605392, |
| "step": 380, |
| "valid_targets_mean": 378.5, |
| "valid_targets_min": 253 |
| }, |
| { |
| "epoch": 2.470209339774557, |
| "grad_norm": 0.43089304741489437, |
| "learning_rate": 2.4000513875520892e-05, |
| "loss": 0.349, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05593574792146683, |
| "step": 385, |
| "valid_targets_mean": 2434.5, |
| "valid_targets_min": 326 |
| }, |
| { |
| "epoch": 2.502415458937198, |
| "grad_norm": 0.45716732308341723, |
| "learning_rate": 2.3561072778243335e-05, |
| "loss": 0.3504, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.033897001296281815, |
| "step": 390, |
| "valid_targets_mean": 945.1, |
| "valid_targets_min": 355 |
| }, |
| { |
| "epoch": 2.534621578099839, |
| "grad_norm": 0.5807110440832902, |
| "learning_rate": 2.311984877573636e-05, |
| "loss": 0.3755, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.050817061215639114, |
| "step": 395, |
| "valid_targets_mean": 629.4, |
| "valid_targets_min": 297 |
| }, |
| { |
| "epoch": 2.5668276972624797, |
| "grad_norm": 0.48048954818776785, |
| "learning_rate": 2.2677062773505235e-05, |
| "loss": 0.3542, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.026756778359413147, |
| "step": 400, |
| "valid_targets_mean": 713.8, |
| "valid_targets_min": 343 |
| }, |
| { |
| "epoch": 2.5990338164251208, |
| "grad_norm": 0.571474727405292, |
| "learning_rate": 2.2232936459094158e-05, |
| "loss": 0.3498, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03278893977403641, |
| "step": 405, |
| "valid_targets_mean": 543.1, |
| "valid_targets_min": 348 |
| }, |
| { |
| "epoch": 2.631239935587762, |
| "grad_norm": 0.46213821153329887, |
| "learning_rate": 2.1787692191094986e-05, |
| "loss": 0.3661, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.02265809290111065, |
| "step": 410, |
| "valid_targets_mean": 691.4, |
| "valid_targets_min": 264 |
| }, |
| { |
| "epoch": 2.6634460547504024, |
| "grad_norm": 0.4201064077434344, |
| "learning_rate": 2.1341552887820048e-05, |
| "loss": 0.3667, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04259542003273964, |
| "step": 415, |
| "valid_targets_mean": 1292.5, |
| "valid_targets_min": 356 |
| }, |
| { |
| "epoch": 2.6956521739130435, |
| "grad_norm": 0.5548360913759484, |
| "learning_rate": 2.0894741915694685e-05, |
| "loss": 0.3912, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04088819772005081, |
| "step": 420, |
| "valid_targets_mean": 663.2, |
| "valid_targets_min": 347 |
| }, |
| { |
| "epoch": 2.7278582930756845, |
| "grad_norm": 0.5304481578098202, |
| "learning_rate": 2.0447482977425465e-05, |
| "loss": 0.3596, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04688024893403053, |
| "step": 425, |
| "valid_targets_mean": 819.1, |
| "valid_targets_min": 324 |
| }, |
| { |
| "epoch": 2.760064412238325, |
| "grad_norm": 0.5165835397260369, |
| "learning_rate": 2e-05, |
| "loss": 0.3846, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.028544938191771507, |
| "step": 430, |
| "valid_targets_mean": 738.5, |
| "valid_targets_min": 337 |
| }, |
| { |
| "epoch": 2.792270531400966, |
| "grad_norm": 0.4492097572019024, |
| "learning_rate": 1.9552517022574542e-05, |
| "loss": 0.3614, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03934304788708687, |
| "step": 435, |
| "valid_targets_mean": 950.9, |
| "valid_targets_min": 317 |
| }, |
| { |
| "epoch": 2.824476650563607, |
| "grad_norm": 0.5323764358603231, |
| "learning_rate": 1.910525808430532e-05, |
| "loss": 0.3961, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.041988566517829895, |
| "step": 440, |
| "valid_targets_mean": 720.8, |
| "valid_targets_min": 321 |
| }, |
| { |
| "epoch": 2.8566827697262482, |
| "grad_norm": 0.461336682212016, |
| "learning_rate": 1.8658447112179952e-05, |
| "loss": 0.3608, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.039135366678237915, |
| "step": 445, |
| "valid_targets_mean": 938.9, |
| "valid_targets_min": 258 |
| }, |
| { |
| "epoch": 2.888888888888889, |
| "grad_norm": 0.3990312478235952, |
| "learning_rate": 1.821230780890502e-05, |
| "loss": 0.3463, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.0437677726149559, |
| "step": 450, |
| "valid_targets_mean": 1247.5, |
| "valid_targets_min": 344 |
| }, |
| { |
| "epoch": 2.92109500805153, |
| "grad_norm": 0.549057571424804, |
| "learning_rate": 1.776706354090585e-05, |
| "loss": 0.3843, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06449249386787415, |
| "step": 455, |
| "valid_targets_mean": 1010.9, |
| "valid_targets_min": 296 |
| }, |
| { |
| "epoch": 2.9533011272141705, |
| "grad_norm": 0.5774462386833141, |
| "learning_rate": 1.7322937226494775e-05, |
| "loss": 0.3701, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.038682036101818085, |
| "step": 460, |
| "valid_targets_mean": 597.2, |
| "valid_targets_min": 275 |
| }, |
| { |
| "epoch": 2.9855072463768115, |
| "grad_norm": 0.45901296564550076, |
| "learning_rate": 1.6880151224263646e-05, |
| "loss": 0.3596, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04841262102127075, |
| "step": 465, |
| "valid_targets_mean": 1967.9, |
| "valid_targets_min": 300 |
| }, |
| { |
| "epoch": 3.0128824476650564, |
| "grad_norm": 0.5272617877932585, |
| "learning_rate": 1.643892722175667e-05, |
| "loss": 0.3841, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.034066297113895416, |
| "step": 470, |
| "valid_targets_mean": 499.1, |
| "valid_targets_min": 291 |
| }, |
| { |
| "epoch": 3.0450885668276975, |
| "grad_norm": 0.5673083616980614, |
| "learning_rate": 1.5999486124479115e-05, |
| "loss": 0.3609, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.039988331496715546, |
| "step": 475, |
| "valid_targets_mean": 589.9, |
| "valid_targets_min": 285 |
| }, |
| { |
| "epoch": 3.077294685990338, |
| "grad_norm": 0.6094441969597235, |
| "learning_rate": 1.5562047945297652e-05, |
| "loss": 0.3685, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.040959179401397705, |
| "step": 480, |
| "valid_targets_mean": 667.4, |
| "valid_targets_min": 304 |
| }, |
| { |
| "epoch": 3.109500805152979, |
| "grad_norm": 0.4240021705550922, |
| "learning_rate": 1.5126831694287564e-05, |
| "loss": 0.357, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.026873812079429626, |
| "step": 485, |
| "valid_targets_mean": 763.5, |
| "valid_targets_min": 303 |
| }, |
| { |
| "epoch": 3.14170692431562, |
| "grad_norm": 0.46197206380035316, |
| "learning_rate": 1.4694055269081987e-05, |
| "loss": 0.3592, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.028340086340904236, |
| "step": 490, |
| "valid_targets_mean": 740.5, |
| "valid_targets_min": 305 |
| }, |
| { |
| "epoch": 3.1739130434782608, |
| "grad_norm": 0.443735264461422, |
| "learning_rate": 1.4263935345778202e-05, |
| "loss": 0.3604, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06339392066001892, |
| "step": 495, |
| "valid_targets_mean": 1977.4, |
| "valid_targets_min": 302 |
| }, |
| { |
| "epoch": 3.206119162640902, |
| "grad_norm": 0.44481778304862124, |
| "learning_rate": 1.3836687270455388e-05, |
| "loss": 0.3501, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.027158450335264206, |
| "step": 500, |
| "valid_targets_mean": 723.2, |
| "valid_targets_min": 253 |
| }, |
| { |
| "epoch": 3.238325281803543, |
| "grad_norm": 0.4835785420746217, |
| "learning_rate": 1.341252495135841e-05, |
| "loss": 0.3512, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06052570044994354, |
| "step": 505, |
| "valid_targets_mean": 1464.1, |
| "valid_targets_min": 244 |
| }, |
| { |
| "epoch": 3.2705314009661834, |
| "grad_norm": 0.565332271009312, |
| "learning_rate": 1.299166075180133e-05, |
| "loss": 0.3572, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.040864065289497375, |
| "step": 510, |
| "valid_targets_mean": 729.4, |
| "valid_targets_min": 287 |
| }, |
| { |
| "epoch": 3.3027375201288245, |
| "grad_norm": 0.4361579841431306, |
| "learning_rate": 1.2574305383844528e-05, |
| "loss": 0.3674, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05954238772392273, |
| "step": 515, |
| "valid_targets_mean": 2146.2, |
| "valid_targets_min": 278 |
| }, |
| { |
| "epoch": 3.3349436392914655, |
| "grad_norm": 0.6338988451640348, |
| "learning_rate": 1.2160667802798503e-05, |
| "loss": 0.3485, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.09376896917819977, |
| "step": 520, |
| "valid_targets_mean": 1447.6, |
| "valid_targets_min": 249 |
| }, |
| { |
| "epoch": 3.367149758454106, |
| "grad_norm": 0.4967736013730134, |
| "learning_rate": 1.1750955102607193e-05, |
| "loss": 0.346, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04483143985271454, |
| "step": 525, |
| "valid_targets_mean": 1566.4, |
| "valid_targets_min": 313 |
| }, |
| { |
| "epoch": 3.399355877616747, |
| "grad_norm": 0.3662991770351534, |
| "learning_rate": 1.1345372412163276e-05, |
| "loss": 0.3406, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.02428291365504265, |
| "step": 530, |
| "valid_targets_mean": 1073.8, |
| "valid_targets_min": 314 |
| }, |
| { |
| "epoch": 3.4315619967793882, |
| "grad_norm": 0.5695193572731221, |
| "learning_rate": 1.094412279260726e-05, |
| "loss": 0.349, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03791878744959831, |
| "step": 535, |
| "valid_targets_mean": 812.0, |
| "valid_targets_min": 293 |
| }, |
| { |
| "epoch": 3.463768115942029, |
| "grad_norm": 0.456955054097705, |
| "learning_rate": 1.0547407135661816e-05, |
| "loss": 0.3597, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.037714216858148575, |
| "step": 540, |
| "valid_targets_mean": 1002.9, |
| "valid_targets_min": 312 |
| }, |
| { |
| "epoch": 3.49597423510467, |
| "grad_norm": 0.5084684377157883, |
| "learning_rate": 1.0155424063052306e-05, |
| "loss": 0.3581, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.044180020689964294, |
| "step": 545, |
| "valid_targets_mean": 1112.9, |
| "valid_targets_min": 284 |
| }, |
| { |
| "epoch": 3.528180354267311, |
| "grad_norm": 0.5865778091291848, |
| "learning_rate": 9.768369827063803e-06, |
| "loss": 0.375, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04223884642124176, |
| "step": 550, |
| "valid_targets_mean": 846.5, |
| "valid_targets_min": 231 |
| }, |
| { |
| "epoch": 3.5603864734299515, |
| "grad_norm": 0.484120177701455, |
| "learning_rate": 9.386438212284372e-06, |
| "loss": 0.3564, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.0398784764111042, |
| "step": 555, |
| "valid_targets_mean": 1083.4, |
| "valid_targets_min": 397 |
| }, |
| { |
| "epoch": 3.5925925925925926, |
| "grad_norm": 0.5654014733466529, |
| "learning_rate": 9.009820438583881e-06, |
| "loss": 0.3629, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06269536912441254, |
| "step": 560, |
| "valid_targets_mean": 954.4, |
| "valid_targets_min": 297 |
| }, |
| { |
| "epoch": 3.6247987117552336, |
| "grad_norm": 0.42473927669630696, |
| "learning_rate": 8.638705065376887e-06, |
| "loss": 0.3464, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.021862631663680077, |
| "step": 565, |
| "valid_targets_mean": 506.1, |
| "valid_targets_min": 266 |
| }, |
| { |
| "epoch": 3.6570048309178746, |
| "grad_norm": 0.3862733964555012, |
| "learning_rate": 8.27327789721746e-06, |
| "loss": 0.3585, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04623720049858093, |
| "step": 570, |
| "valid_targets_mean": 1697.0, |
| "valid_targets_min": 298 |
| }, |
| { |
| "epoch": 3.6892109500805152, |
| "grad_norm": 0.4125076115623328, |
| "learning_rate": 7.913721890773354e-06, |
| "loss": 0.3258, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.0626937747001648, |
| "step": 575, |
| "valid_targets_mean": 2128.5, |
| "valid_targets_min": 338 |
| }, |
| { |
| "epoch": 3.7214170692431563, |
| "grad_norm": 0.5688902989190885, |
| "learning_rate": 7.560217063225923e-06, |
| "loss": 0.3681, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05661293864250183, |
| "step": 580, |
| "valid_targets_mean": 855.2, |
| "valid_targets_min": 322 |
| }, |
| { |
| "epoch": 3.753623188405797, |
| "grad_norm": 0.47371336118949064, |
| "learning_rate": 7.212940402141808e-06, |
| "loss": 0.3543, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.032223887741565704, |
| "step": 585, |
| "valid_targets_mean": 570.1, |
| "valid_targets_min": 331 |
| }, |
| { |
| "epoch": 3.785829307568438, |
| "grad_norm": 0.4522966191465321, |
| "learning_rate": 6.8720657768614115e-06, |
| "loss": 0.3299, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.035974081605672836, |
| "step": 590, |
| "valid_targets_mean": 813.8, |
| "valid_targets_min": 331 |
| }, |
| { |
| "epoch": 3.818035426731079, |
| "grad_norm": 0.4652139120483709, |
| "learning_rate": 6.537763851448593e-06, |
| "loss": 0.3628, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04356799274682999, |
| "step": 595, |
| "valid_targets_mean": 1215.0, |
| "valid_targets_min": 288 |
| }, |
| { |
| "epoch": 3.85024154589372, |
| "grad_norm": 0.46874463875570954, |
| "learning_rate": 6.210201999245087e-06, |
| "loss": 0.3436, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.029137909412384033, |
| "step": 600, |
| "valid_targets_mean": 621.4, |
| "valid_targets_min": 278 |
| }, |
| { |
| "epoch": 3.8824476650563606, |
| "grad_norm": 0.41711178274476524, |
| "learning_rate": 5.889544219072465e-06, |
| "loss": 0.3406, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.045179564505815506, |
| "step": 605, |
| "valid_targets_mean": 1414.2, |
| "valid_targets_min": 342 |
| }, |
| { |
| "epoch": 3.9146537842190017, |
| "grad_norm": 0.36858876028850934, |
| "learning_rate": 5.5759510531237116e-06, |
| "loss": 0.3473, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03258288651704788, |
| "step": 610, |
| "valid_targets_mean": 1184.9, |
| "valid_targets_min": 350 |
| }, |
| { |
| "epoch": 3.9468599033816423, |
| "grad_norm": 0.3655458437589055, |
| "learning_rate": 5.269579506585259e-06, |
| "loss": 0.3383, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.024550924077630043, |
| "step": 615, |
| "valid_targets_mean": 761.0, |
| "valid_targets_min": 367 |
| }, |
| { |
| "epoch": 3.9790660225442833, |
| "grad_norm": 0.5867554039598152, |
| "learning_rate": 4.97058296903e-06, |
| "loss": 0.3549, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04499742016196251, |
| "step": 620, |
| "valid_targets_mean": 819.9, |
| "valid_targets_min": 312 |
| }, |
| { |
| "epoch": 4.006441223832528, |
| "grad_norm": 0.5337919310243177, |
| "learning_rate": 4.679111137620442e-06, |
| "loss": 0.3519, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04486808925867081, |
| "step": 625, |
| "valid_targets_mean": 827.9, |
| "valid_targets_min": 261 |
| }, |
| { |
| "epoch": 4.038647342995169, |
| "grad_norm": 0.5000808046641099, |
| "learning_rate": 4.395309942160551e-06, |
| "loss": 0.3509, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03933977708220482, |
| "step": 630, |
| "valid_targets_mean": 830.9, |
| "valid_targets_min": 299 |
| }, |
| { |
| "epoch": 4.07085346215781, |
| "grad_norm": 0.4557917329913038, |
| "learning_rate": 4.119321472033779e-06, |
| "loss": 0.3471, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.054948341101408005, |
| "step": 635, |
| "valid_targets_mean": 1991.6, |
| "valid_targets_min": 267 |
| }, |
| { |
| "epoch": 4.1030595813204505, |
| "grad_norm": 0.4721247455404434, |
| "learning_rate": 3.851283905063863e-06, |
| "loss": 0.3463, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04883628711104393, |
| "step": 640, |
| "valid_targets_mean": 1025.4, |
| "valid_targets_min": 344 |
| }, |
| { |
| "epoch": 4.1352657004830915, |
| "grad_norm": 0.44811926966002247, |
| "learning_rate": 3.5913314383339937e-06, |
| "loss": 0.3443, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.02158988080918789, |
| "step": 645, |
| "valid_targets_mean": 505.0, |
| "valid_targets_min": 200 |
| }, |
| { |
| "epoch": 4.1674718196457325, |
| "grad_norm": 0.4830380662288485, |
| "learning_rate": 3.3395942209989694e-06, |
| "loss": 0.3395, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03771327808499336, |
| "step": 650, |
| "valid_targets_mean": 780.9, |
| "valid_targets_min": 342 |
| }, |
| { |
| "epoch": 4.199677938808374, |
| "grad_norm": 0.4975062818136211, |
| "learning_rate": 3.0961982891241083e-06, |
| "loss": 0.338, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04027272388339043, |
| "step": 655, |
| "valid_targets_mean": 965.4, |
| "valid_targets_min": 331 |
| }, |
| { |
| "epoch": 4.231884057971015, |
| "grad_norm": 0.5186580520002718, |
| "learning_rate": 2.8612655025833148e-06, |
| "loss": 0.3311, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.0684180036187172, |
| "step": 660, |
| "valid_targets_mean": 1254.0, |
| "valid_targets_min": 253 |
| }, |
| { |
| "epoch": 4.264090177133656, |
| "grad_norm": 0.5925008943137404, |
| "learning_rate": 2.6349134840481294e-06, |
| "loss": 0.3621, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05215730518102646, |
| "step": 665, |
| "valid_targets_mean": 731.0, |
| "valid_targets_min": 272 |
| }, |
| { |
| "epoch": 4.296296296296296, |
| "grad_norm": 0.41226419968489375, |
| "learning_rate": 2.417255560098111e-06, |
| "loss": 0.341, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03634045645594597, |
| "step": 670, |
| "valid_targets_mean": 1005.0, |
| "valid_targets_min": 357 |
| }, |
| { |
| "epoch": 4.328502415458937, |
| "grad_norm": 0.4129907136798245, |
| "learning_rate": 2.2084007044821764e-06, |
| "loss": 0.3291, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03767111897468567, |
| "step": 675, |
| "valid_targets_mean": 1171.0, |
| "valid_targets_min": 289 |
| }, |
| { |
| "epoch": 4.360708534621578, |
| "grad_norm": 0.5348812254442832, |
| "learning_rate": 2.0084534835592584e-06, |
| "loss": 0.3387, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04408322647213936, |
| "step": 680, |
| "valid_targets_mean": 861.4, |
| "valid_targets_min": 346 |
| }, |
| { |
| "epoch": 4.392914653784219, |
| "grad_norm": 0.4983477080423943, |
| "learning_rate": 1.817514003945524e-06, |
| "loss": 0.3446, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03348889201879501, |
| "step": 685, |
| "valid_targets_mean": 840.8, |
| "valid_targets_min": 301 |
| }, |
| { |
| "epoch": 4.42512077294686, |
| "grad_norm": 0.5216601301709456, |
| "learning_rate": 1.6356778623945223e-06, |
| "loss": 0.3368, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.034096863120794296, |
| "step": 690, |
| "valid_targets_mean": 715.5, |
| "valid_targets_min": 244 |
| }, |
| { |
| "epoch": 4.457326892109501, |
| "grad_norm": 0.4655020726002043, |
| "learning_rate": 1.4630360979351644e-06, |
| "loss": 0.3668, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.07300788164138794, |
| "step": 695, |
| "valid_targets_mean": 2403.8, |
| "valid_targets_min": 313 |
| }, |
| { |
| "epoch": 4.489533011272142, |
| "grad_norm": 0.5310053686779203, |
| "learning_rate": 1.2996751462917057e-06, |
| "loss": 0.3544, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.037691399455070496, |
| "step": 700, |
| "valid_targets_mean": 850.4, |
| "valid_targets_min": 306 |
| }, |
| { |
| "epoch": 4.521739130434782, |
| "grad_norm": 0.4310213806844335, |
| "learning_rate": 1.1456767966083393e-06, |
| "loss": 0.3441, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03169054538011551, |
| "step": 705, |
| "valid_targets_mean": 730.5, |
| "valid_targets_min": 316 |
| }, |
| { |
| "epoch": 4.553945249597423, |
| "grad_norm": 0.40942357788440725, |
| "learning_rate": 1.0011181505002488e-06, |
| "loss": 0.3616, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.02835988998413086, |
| "step": 710, |
| "valid_targets_mean": 966.9, |
| "valid_targets_min": 248 |
| }, |
| { |
| "epoch": 4.586151368760064, |
| "grad_norm": 0.4064126072802411, |
| "learning_rate": 8.660715834514977e-07, |
| "loss": 0.3394, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.019462428987026215, |
| "step": 715, |
| "valid_targets_mean": 683.4, |
| "valid_targets_min": 303 |
| }, |
| { |
| "epoch": 4.618357487922705, |
| "grad_norm": 0.4659035655355515, |
| "learning_rate": 7.406047085791379e-07, |
| "loss": 0.3367, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04081908240914345, |
| "step": 720, |
| "valid_targets_mean": 1303.8, |
| "valid_targets_min": 284 |
| }, |
| { |
| "epoch": 4.650563607085346, |
| "grad_norm": 0.5553298528236867, |
| "learning_rate": 6.247803427816945e-07, |
| "loss": 0.3613, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.0552799291908741, |
| "step": 725, |
| "valid_targets_mean": 973.4, |
| "valid_targets_min": 357 |
| }, |
| { |
| "epoch": 4.6827697262479875, |
| "grad_norm": 0.5045394202661283, |
| "learning_rate": 5.186564752889078e-07, |
| "loss": 0.3547, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.08811699599027634, |
| "step": 730, |
| "valid_targets_mean": 1683.6, |
| "valid_targets_min": 439 |
| }, |
| { |
| "epoch": 4.714975845410628, |
| "grad_norm": 0.41108260312594874, |
| "learning_rate": 4.22286238628562e-07, |
| "loss": 0.3564, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04261046648025513, |
| "step": 735, |
| "valid_targets_mean": 1162.9, |
| "valid_targets_min": 225 |
| }, |
| { |
| "epoch": 4.747181964573269, |
| "grad_norm": 0.37863905329842407, |
| "learning_rate": 3.357178820248641e-07, |
| "loss": 0.3402, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06985923647880554, |
| "step": 740, |
| "valid_targets_mean": 3784.5, |
| "valid_targets_min": 348 |
| }, |
| { |
| "epoch": 4.77938808373591, |
| "grad_norm": 0.38057216907053537, |
| "learning_rate": 2.5899474724174313e-07, |
| "loss": 0.3272, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.024681851267814636, |
| "step": 745, |
| "valid_targets_mean": 856.8, |
| "valid_targets_min": 320 |
| }, |
| { |
| "epoch": 4.811594202898551, |
| "grad_norm": 0.4980702323011905, |
| "learning_rate": 1.9215524688313893e-07, |
| "loss": 0.3422, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03393610566854477, |
| "step": 750, |
| "valid_targets_mean": 697.4, |
| "valid_targets_min": 304 |
| }, |
| { |
| "epoch": 4.843800322061192, |
| "grad_norm": 0.48534143807405106, |
| "learning_rate": 1.3523284516113955e-07, |
| "loss": 0.3442, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04657585173845291, |
| "step": 755, |
| "valid_targets_mean": 1234.6, |
| "valid_targets_min": 294 |
| }, |
| { |
| "epoch": 4.876006441223833, |
| "grad_norm": 0.4327518473258517, |
| "learning_rate": 8.825604114162556e-08, |
| "loss": 0.3246, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03470573574304581, |
| "step": 760, |
| "valid_targets_mean": 690.1, |
| "valid_targets_min": 276 |
| }, |
| { |
| "epoch": 4.908212560386474, |
| "grad_norm": 0.5175651819741588, |
| "learning_rate": 5.1248354475768034e-08, |
| "loss": 0.3555, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05181993171572685, |
| "step": 765, |
| "valid_targets_mean": 1085.9, |
| "valid_targets_min": 303 |
| }, |
| { |
| "epoch": 4.940418679549114, |
| "grad_norm": 0.5505526615345873, |
| "learning_rate": 2.4228313624568013e-08, |
| "loss": 0.3575, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.03321858495473862, |
| "step": 770, |
| "valid_targets_mean": 728.8, |
| "valid_targets_min": 168 |
| }, |
| { |
| "epoch": 4.972624798711755, |
| "grad_norm": 0.5754568951885141, |
| "learning_rate": 7.209446582292501e-09, |
| "loss": 0.3547, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06288649886846542, |
| "step": 775, |
| "valid_targets_mean": 1244.1, |
| "valid_targets_min": 286 |
| }, |
| { |
| "epoch": 5.0, |
| "grad_norm": 1.1816753894303154, |
| "learning_rate": 2.002741034812239e-10, |
| "loss": 0.3613, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.23872976005077362, |
| "step": 780, |
| "valid_targets_mean": 1017.1, |
| "valid_targets_min": 327 |
| }, |
| { |
| "epoch": 5.0, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.23872976005077362, |
| "step": 780, |
| "total_flos": 4.410167938263286e+17, |
| "train_loss": 0.3955344903163421, |
| "train_runtime": 13041.82, |
| "train_samples_per_second": 3.807, |
| "train_steps_per_second": 0.06, |
| "valid_targets_mean": 1017.1, |
| "valid_targets_min": 327 |
| } |
| ], |
| "logging_steps": 5, |
| "max_steps": 780, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 5, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": false, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 4.410167938263286e+17, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
