461 lines
11 KiB
JSON
461 lines
11 KiB
JSON
|
{
|
||
|
|
"best_global_step": 362,
|
||
|
|
"best_metric": 0.15059404075145721,
|
||
|
|
"best_model_checkpoint": "/scratch/gk_sft_model/_trainer/checkpoint-362",
|
||
|
|
"epoch": 3.0,
|
||
|
|
"eval_steps": 500,
|
||
|
|
"global_step": 543,
|
||
|
|
"is_hyper_param_search": false,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 0.055382485289027346,
|
||
|
|
"grad_norm": 1.9609375,
|
||
|
|
"learning_rate": 5.294117647058824e-06,
|
||
|
|
"loss": 0.17566636800765992,
|
||
|
|
"step": 10
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11076497057805469,
|
||
|
|
"grad_norm": 2.578125,
|
||
|
|
"learning_rate": 9.99964328358639e-06,
|
||
|
|
"loss": 0.1814702033996582,
|
||
|
|
"step": 20
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16614745586708204,
|
||
|
|
"grad_norm": 1.7578125,
|
||
|
|
"learning_rate": 9.987163552653802e-06,
|
||
|
|
"loss": 0.17184715270996093,
|
||
|
|
"step": 30
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22152994115610938,
|
||
|
|
"grad_norm": 3.078125,
|
||
|
|
"learning_rate": 9.956898867045e-06,
|
||
|
|
"loss": 0.1755874752998352,
|
||
|
|
"step": 40
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.27691242644513675,
|
||
|
|
"grad_norm": 2.078125,
|
||
|
|
"learning_rate": 9.908957155055522e-06,
|
||
|
|
"loss": 0.17331844568252563,
|
||
|
|
"step": 50
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3322949117341641,
|
||
|
|
"grad_norm": 2.625,
|
||
|
|
"learning_rate": 9.843509383842524e-06,
|
||
|
|
"loss": 0.1772994041442871,
|
||
|
|
"step": 60
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38767739702319143,
|
||
|
|
"grad_norm": 2.453125,
|
||
|
|
"learning_rate": 9.760788949730866e-06,
|
||
|
|
"loss": 0.17646037340164183,
|
||
|
|
"step": 70
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.44305988231221877,
|
||
|
|
"grad_norm": 2.46875,
|
||
|
|
"learning_rate": 9.661090845887693e-06,
|
||
|
|
"loss": 0.17809184789657592,
|
||
|
|
"step": 80
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4984423676012461,
|
||
|
|
"grad_norm": 3.640625,
|
||
|
|
"learning_rate": 9.544770610333656e-06,
|
||
|
|
"loss": 0.16971004009246826,
|
||
|
|
"step": 90
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5538248528902735,
|
||
|
|
"grad_norm": 2.578125,
|
||
|
|
"learning_rate": 9.412243058042316e-06,
|
||
|
|
"loss": 0.1714487314224243,
|
||
|
|
"step": 100
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6092073381793008,
|
||
|
|
"grad_norm": 3.0625,
|
||
|
|
"learning_rate": 9.263980801649286e-06,
|
||
|
|
"loss": 0.164695143699646,
|
||
|
|
"step": 110
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6645898234683282,
|
||
|
|
"grad_norm": 2.234375,
|
||
|
|
"learning_rate": 9.100512566046445e-06,
|
||
|
|
"loss": 0.17420523166656493,
|
||
|
|
"step": 120
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7199723087573555,
|
||
|
|
"grad_norm": 2.875,
|
||
|
|
"learning_rate": 8.922421302871628e-06,
|
||
|
|
"loss": 0.16593801975250244,
|
||
|
|
"step": 130
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7753547940463829,
|
||
|
|
"grad_norm": 2.609375,
|
||
|
|
"learning_rate": 8.73034211161784e-06,
|
||
|
|
"loss": 0.1752484083175659,
|
||
|
|
"step": 140
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8307372793354102,
|
||
|
|
"grad_norm": 1.546875,
|
||
|
|
"learning_rate": 8.524959974775551e-06,
|
||
|
|
"loss": 0.16712336540222167,
|
||
|
|
"step": 150
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8861197646244375,
|
||
|
|
"grad_norm": 3.546875,
|
||
|
|
"learning_rate": 8.307007315084958e-06,
|
||
|
|
"loss": 0.16917212009429933,
|
||
|
|
"step": 160
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9415022499134649,
|
||
|
|
"grad_norm": 2.1875,
|
||
|
|
"learning_rate": 8.077261383609363e-06,
|
||
|
|
"loss": 0.16140577793121338,
|
||
|
|
"step": 170
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9968847352024922,
|
||
|
|
"grad_norm": 3.109375,
|
||
|
|
"learning_rate": 7.836541487944194e-06,
|
||
|
|
"loss": 0.1660953164100647,
|
||
|
|
"step": 180
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0,
|
||
|
|
"eval_val_loss": 0.1515691578388214,
|
||
|
|
"eval_val_runtime": 13.4914,
|
||
|
|
"eval_val_samples_per_second": 23.719,
|
||
|
|
"eval_val_steps_per_second": 23.719,
|
||
|
|
"step": 181
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0,
|
||
|
|
"eval_test_loss": 0.1465539038181305,
|
||
|
|
"eval_test_runtime": 13.6527,
|
||
|
|
"eval_test_samples_per_second": 23.439,
|
||
|
|
"eval_test_steps_per_second": 23.439,
|
||
|
|
"step": 181
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0498442367601246,
|
||
|
|
"grad_norm": 1.7578125,
|
||
|
|
"learning_rate": 7.585706070446288e-06,
|
||
|
|
"loss": 0.15730363130569458,
|
||
|
|
"step": 190
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.105226722049152,
|
||
|
|
"grad_norm": 2.203125,
|
||
|
|
"learning_rate": 7.325649646902887e-06,
|
||
|
|
"loss": 0.16893280744552613,
|
||
|
|
"step": 200
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1606092073381793,
|
||
|
|
"grad_norm": 2.46875,
|
||
|
|
"learning_rate": 7.057299616557494e-06,
|
||
|
|
"loss": 0.1647992253303528,
|
||
|
|
"step": 210
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2159916926272065,
|
||
|
|
"grad_norm": 2.03125,
|
||
|
|
"learning_rate": 6.781612954868538e-06,
|
||
|
|
"loss": 0.1556432366371155,
|
||
|
|
"step": 220
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.271374177916234,
|
||
|
|
"grad_norm": 2.78125,
|
||
|
|
"learning_rate": 6.499572800794911e-06,
|
||
|
|
"loss": 0.16220529079437257,
|
||
|
|
"step": 230
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3267566632052614,
|
||
|
|
"grad_norm": 1.84375,
|
||
|
|
"learning_rate": 6.21218495077864e-06,
|
||
|
|
"loss": 0.16749459505081177,
|
||
|
|
"step": 240
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3821391484942886,
|
||
|
|
"grad_norm": 3.1875,
|
||
|
|
"learning_rate": 5.920474271927668e-06,
|
||
|
|
"loss": 0.15629700422286988,
|
||
|
|
"step": 250
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.437521633783316,
|
||
|
|
"grad_norm": 2.859375,
|
||
|
|
"learning_rate": 5.625481047189835e-06,
|
||
|
|
"loss": 0.1582696795463562,
|
||
|
|
"step": 260
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4929041190723433,
|
||
|
|
"grad_norm": 2.3125,
|
||
|
|
"learning_rate": 5.328257265551742e-06,
|
||
|
|
"loss": 0.16032217741012572,
|
||
|
|
"step": 270
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5482866043613708,
|
||
|
|
"grad_norm": 2.71875,
|
||
|
|
"learning_rate": 5.029862870492142e-06,
|
||
|
|
"loss": 0.15796239376068116,
|
||
|
|
"step": 280
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6036690896503982,
|
||
|
|
"grad_norm": 2.09375,
|
||
|
|
"learning_rate": 4.7313619800684765e-06,
|
||
|
|
"loss": 0.1616731643676758,
|
||
|
|
"step": 290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6590515749394255,
|
||
|
|
"grad_norm": 2.03125,
|
||
|
|
"learning_rate": 4.433819092116223e-06,
|
||
|
|
"loss": 0.16614921092987062,
|
||
|
|
"step": 300
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7144340602284527,
|
||
|
|
"grad_norm": 2.6875,
|
||
|
|
"learning_rate": 4.138295288093935e-06,
|
||
|
|
"loss": 0.16132607460021972,
|
||
|
|
"step": 310
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7698165455174801,
|
||
|
|
"grad_norm": 2.453125,
|
||
|
|
"learning_rate": 3.845844449111535e-06,
|
||
|
|
"loss": 0.14961661100387574,
|
||
|
|
"step": 320
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8251990308065076,
|
||
|
|
"grad_norm": 2.75,
|
||
|
|
"learning_rate": 3.557509497636163e-06,
|
||
|
|
"loss": 0.16481106281280516,
|
||
|
|
"step": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8805815160955348,
|
||
|
|
"grad_norm": 2.9375,
|
||
|
|
"learning_rate": 3.2743186782780855e-06,
|
||
|
|
"loss": 0.16425250768661498,
|
||
|
|
"step": 340
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.935964001384562,
|
||
|
|
"grad_norm": 1.5,
|
||
|
|
"learning_rate": 2.99728189092004e-06,
|
||
|
|
"loss": 0.1606146812438965,
|
||
|
|
"step": 350
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9913464866735895,
|
||
|
|
"grad_norm": 1.9375,
|
||
|
|
"learning_rate": 2.7273870892665257e-06,
|
||
|
|
"loss": 0.1580709457397461,
|
||
|
|
"step": 360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0,
|
||
|
|
"eval_val_loss": 0.15059404075145721,
|
||
|
|
"eval_val_runtime": 13.4279,
|
||
|
|
"eval_val_samples_per_second": 23.831,
|
||
|
|
"eval_val_steps_per_second": 23.831,
|
||
|
|
"step": 362
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0,
|
||
|
|
"eval_test_loss": 0.14516279101371765,
|
||
|
|
"eval_test_runtime": 13.5429,
|
||
|
|
"eval_test_samples_per_second": 23.629,
|
||
|
|
"eval_test_steps_per_second": 23.629,
|
||
|
|
"step": 362
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.044305988231222,
|
||
|
|
"grad_norm": 2.03125,
|
||
|
|
"learning_rate": 2.4655967576564066e-06,
|
||
|
|
"loss": 0.15324945449829103,
|
||
|
|
"step": 370
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0996884735202492,
|
||
|
|
"grad_norm": 2.109375,
|
||
|
|
"learning_rate": 2.212844478702962e-06,
|
||
|
|
"loss": 0.1620323896408081,
|
||
|
|
"step": 380
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1550709588092767,
|
||
|
|
"grad_norm": 2.25,
|
||
|
|
"learning_rate": 1.9700316040017513e-06,
|
||
|
|
"loss": 0.1526092529296875,
|
||
|
|
"step": 390
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.210453444098304,
|
||
|
|
"grad_norm": 2.3125,
|
||
|
|
"learning_rate": 1.7380240397789838e-06,
|
||
|
|
"loss": 0.15873063802719117,
|
||
|
|
"step": 400
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.265835929387331,
|
||
|
|
"grad_norm": 1.9609375,
|
||
|
|
"learning_rate": 1.517649158943263e-06,
|
||
|
|
"loss": 0.15675312280654907,
|
||
|
|
"step": 410
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3212184146763586,
|
||
|
|
"grad_norm": 3.1875,
|
||
|
|
"learning_rate": 1.3096928505527812e-06,
|
||
|
|
"loss": 0.15652062892913818,
|
||
|
|
"step": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.376600899965386,
|
||
|
|
"grad_norm": 2.421875,
|
||
|
|
"learning_rate": 1.1148967172199848e-06,
|
||
|
|
"loss": 0.15628126859664918,
|
||
|
|
"step": 430
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.431983385254413,
|
||
|
|
"grad_norm": 2.515625,
|
||
|
|
"learning_rate": 9.339554304481952e-07,
|
||
|
|
"loss": 0.15665985345840455,
|
||
|
|
"step": 440
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4873658705434405,
|
||
|
|
"grad_norm": 2.234375,
|
||
|
|
"learning_rate": 7.675142533314173e-07,
|
||
|
|
"loss": 0.16646597385406495,
|
||
|
|
"step": 450
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.542748355832468,
|
||
|
|
"grad_norm": 2.21875,
|
||
|
|
"learning_rate": 6.161667394517684e-07,
|
||
|
|
"loss": 0.15952544212341307,
|
||
|
|
"step": 460
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5981308411214954,
|
||
|
|
"grad_norm": 1.6484375,
|
||
|
|
"learning_rate": 4.804526161805834e-07,
|
||
|
|
"loss": 0.1583627700805664,
|
||
|
|
"step": 470
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.653513326410523,
|
||
|
|
"grad_norm": 2.234375,
|
||
|
|
"learning_rate": 3.6085585993167804e-07,
|
||
|
|
"loss": 0.15457649230957032,
|
||
|
|
"step": 480
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7088958116995503,
|
||
|
|
"grad_norm": 2.875,
|
||
|
|
"learning_rate": 2.578029702306506e-07,
|
||
|
|
"loss": 0.15568662881851197,
|
||
|
|
"step": 490
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7642782969885773,
|
||
|
|
"grad_norm": 1.9453125,
|
||
|
|
"learning_rate": 1.7166144875520762e-07,
|
||
|
|
"loss": 0.15728613138198852,
|
||
|
|
"step": 500
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8196607822776047,
|
||
|
|
"grad_norm": 2.125,
|
||
|
|
"learning_rate": 1.0273848877041804e-07,
|
||
|
|
"loss": 0.15580954551696777,
|
||
|
|
"step": 510
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.875043267566632,
|
||
|
|
"grad_norm": 2.1875,
|
||
|
|
"learning_rate": 5.127987963260583e-08,
|
||
|
|
"loss": 0.16028742790222167,
|
||
|
|
"step": 520
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.930425752855659,
|
||
|
|
"grad_norm": 1.8046875,
|
||
|
|
"learning_rate": 1.746913026854924e-08,
|
||
|
|
"loss": 0.15576037168502807,
|
||
|
|
"step": 530
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9858082381446867,
|
||
|
|
"grad_norm": 2.328125,
|
||
|
|
"learning_rate": 1.4268147558088585e-09,
|
||
|
|
"loss": 0.14897778034210205,
|
||
|
|
"step": 540
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0,
|
||
|
|
"eval_val_loss": 0.15068988502025604,
|
||
|
|
"eval_val_runtime": 13.4175,
|
||
|
|
"eval_val_samples_per_second": 23.849,
|
||
|
|
"eval_val_steps_per_second": 23.849,
|
||
|
|
"step": 543
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0,
|
||
|
|
"eval_test_loss": 0.14553283154964447,
|
||
|
|
"eval_test_runtime": 13.5626,
|
||
|
|
"eval_test_samples_per_second": 23.594,
|
||
|
|
"eval_test_steps_per_second": 23.594,
|
||
|
|
"step": 543
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 10,
|
||
|
|
"max_steps": 543,
|
||
|
|
"num_input_tokens_seen": 0,
|
||
|
|
"num_train_epochs": 3,
|
||
|
|
"save_steps": 500,
|
||
|
|
"stateful_callbacks": {
|
||
|
|
"TrainerControl": {
|
||
|
|
"args": {
|
||
|
|
"should_epoch_stop": false,
|
||
|
|
"should_evaluate": false,
|
||
|
|
"should_log": false,
|
||
|
|
"should_save": true,
|
||
|
|
"should_training_stop": true
|
||
|
|
},
|
||
|
|
"attributes": {}
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"total_flos": 1.3613072776965734e+17,
|
||
|
|
"train_batch_size": 1,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|