413 lines
14 KiB
JSON
413 lines
14 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 3.0,
|
|
"eval_steps": 500,
|
|
"global_step": 144,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completion_length": 100.0,
|
|
"completions/clipped_ratio": 1.0,
|
|
"completions/max_length": 100.0,
|
|
"completions/max_terminated_length": 0.0,
|
|
"completions/mean_length": 100.0,
|
|
"completions/mean_terminated_length": 0.0,
|
|
"completions/min_length": 100.0,
|
|
"completions/min_terminated_length": 0.0,
|
|
"epoch": 0.20833333333333334,
|
|
"frac_reward_zero_std": 0.275,
|
|
"grad_norm": 0.252999871969223,
|
|
"kl": 4.623506138159428e-05,
|
|
"learning_rate": 3e-05,
|
|
"loss": 7.562339305877686e-08,
|
|
"num_tokens": 30976.0,
|
|
"reward": -1.63874990940094,
|
|
"reward_std": 0.6745969772338867,
|
|
"rewards/reward_fn/mean": -1.63874990940094,
|
|
"rewards/reward_fn/std": 0.8239436686038971,
|
|
"step": 10
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completion_length": 100.0,
|
|
"completions/clipped_ratio": 1.0,
|
|
"completions/max_length": 100.0,
|
|
"completions/max_terminated_length": 0.0,
|
|
"completions/mean_length": 100.0,
|
|
"completions/mean_terminated_length": 0.0,
|
|
"completions/min_length": 100.0,
|
|
"completions/min_terminated_length": 0.0,
|
|
"epoch": 0.4166666666666667,
|
|
"frac_reward_zero_std": 0.175,
|
|
"grad_norm": 0.2783929705619812,
|
|
"kl": 0.001219680597569095,
|
|
"learning_rate": 4.8449612403100775e-05,
|
|
"loss": 1.2345612049102782e-06,
|
|
"num_tokens": 62268.0,
|
|
"reward": -1.4524999380111694,
|
|
"reward_std": 0.7632413059473038,
|
|
"rewards/reward_fn/mean": -1.4524999260902405,
|
|
"rewards/reward_fn/std": 0.8776885092258453,
|
|
"step": 20
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completion_length": 100.0,
|
|
"completions/clipped_ratio": 1.0,
|
|
"completions/max_length": 100.0,
|
|
"completions/max_terminated_length": 0.0,
|
|
"completions/mean_length": 100.0,
|
|
"completions/mean_terminated_length": 0.0,
|
|
"completions/min_length": 100.0,
|
|
"completions/min_terminated_length": 0.0,
|
|
"epoch": 0.625,
|
|
"frac_reward_zero_std": 0.075,
|
|
"grad_norm": 0.310812383890152,
|
|
"kl": 0.012675344076706096,
|
|
"learning_rate": 4.4573643410852715e-05,
|
|
"loss": 1.2670829892158509e-05,
|
|
"num_tokens": 93408.0,
|
|
"reward": -0.9787499606609344,
|
|
"reward_std": 0.8382474899291992,
|
|
"rewards/reward_fn/mean": -0.9787499666213989,
|
|
"rewards/reward_fn/std": 0.874516224861145,
|
|
"step": 30
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completion_length": 100.0,
|
|
"completions/clipped_ratio": 1.0,
|
|
"completions/max_length": 100.0,
|
|
"completions/max_terminated_length": 0.0,
|
|
"completions/mean_length": 100.0,
|
|
"completions/mean_terminated_length": 0.0,
|
|
"completions/min_length": 100.0,
|
|
"completions/min_terminated_length": 0.0,
|
|
"epoch": 0.8333333333333334,
|
|
"frac_reward_zero_std": 0.075,
|
|
"grad_norm": 0.3383573889732361,
|
|
"kl": 0.03188293152488768,
|
|
"learning_rate": 4.0697674418604655e-05,
|
|
"loss": 3.1878240406513214e-05,
|
|
"num_tokens": 124444.0,
|
|
"reward": -0.7437499940395356,
|
|
"reward_std": 0.7179304540157319,
|
|
"rewards/reward_fn/mean": -0.7437499821186065,
|
|
"rewards/reward_fn/std": 0.7846309900283813,
|
|
"step": 40
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completion_length": 100.0,
|
|
"completions/clipped_ratio": 1.0,
|
|
"completions/max_length": 100.0,
|
|
"completions/max_terminated_length": 0.0,
|
|
"completions/mean_length": 100.0,
|
|
"completions/mean_terminated_length": 0.0,
|
|
"completions/min_length": 100.0,
|
|
"completions/min_terminated_length": 0.0,
|
|
"epoch": 1.0416666666666667,
|
|
"frac_reward_zero_std": 0.225,
|
|
"grad_norm": 0.35372963547706604,
|
|
"kl": 0.04432902028784156,
|
|
"learning_rate": 3.682170542635659e-05,
|
|
"loss": 4.432052373886109e-05,
|
|
"num_tokens": 155440.0,
|
|
"reward": -0.6174999922513962,
|
|
"reward_std": 0.5982875227928162,
|
|
"rewards/reward_fn/mean": -0.6174999952316285,
|
|
"rewards/reward_fn/std": 0.698276162147522,
|
|
"step": 50
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completion_length": 100.0,
|
|
"completions/clipped_ratio": 1.0,
|
|
"completions/max_length": 100.0,
|
|
"completions/max_terminated_length": 0.0,
|
|
"completions/mean_length": 100.0,
|
|
"completions/mean_terminated_length": 0.0,
|
|
"completions/min_length": 100.0,
|
|
"completions/min_terminated_length": 0.0,
|
|
"epoch": 1.25,
|
|
"frac_reward_zero_std": 0.475,
|
|
"grad_norm": 0.19890302419662476,
|
|
"kl": 0.06396519551053644,
|
|
"learning_rate": 3.294573643410852e-05,
|
|
"loss": 6.395205855369568e-05,
|
|
"num_tokens": 186420.0,
|
|
"reward": -0.4387499988079071,
|
|
"reward_std": 0.3215150013566017,
|
|
"rewards/reward_fn/mean": -0.4387499988079071,
|
|
"rewards/reward_fn/std": 0.48022162914276123,
|
|
"step": 60
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completion_length": 100.0,
|
|
"completions/clipped_ratio": 1.0,
|
|
"completions/max_length": 100.0,
|
|
"completions/max_terminated_length": 0.0,
|
|
"completions/mean_length": 100.0,
|
|
"completions/mean_terminated_length": 0.0,
|
|
"completions/min_length": 100.0,
|
|
"completions/min_terminated_length": 0.0,
|
|
"epoch": 1.4583333333333333,
|
|
"frac_reward_zero_std": 0.5,
|
|
"grad_norm": 0.3041526675224304,
|
|
"kl": 0.08288784278556705,
|
|
"learning_rate": 2.9069767441860467e-05,
|
|
"loss": 8.288174867630005e-05,
|
|
"num_tokens": 217644.0,
|
|
"reward": -0.3737500041723251,
|
|
"reward_std": 0.2771410092711449,
|
|
"rewards/reward_fn/mean": -0.3737500011920929,
|
|
"rewards/reward_fn/std": 0.4215263336896896,
|
|
"step": 70
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completion_length": 100.0,
|
|
"completions/clipped_ratio": 1.0,
|
|
"completions/max_length": 100.0,
|
|
"completions/max_terminated_length": 0.0,
|
|
"completions/mean_length": 100.0,
|
|
"completions/mean_terminated_length": 0.0,
|
|
"completions/min_length": 100.0,
|
|
"completions/min_terminated_length": 0.0,
|
|
"epoch": 1.6666666666666665,
|
|
"frac_reward_zero_std": 0.725,
|
|
"grad_norm": 0.3231094479560852,
|
|
"kl": 0.10231811739504337,
|
|
"learning_rate": 2.5193798449612404e-05,
|
|
"loss": 0.00010231174528598785,
|
|
"num_tokens": 248780.0,
|
|
"reward": -0.34500000476837156,
|
|
"reward_std": 0.17121320217847824,
|
|
"rewards/reward_fn/mean": -0.34500000476837156,
|
|
"rewards/reward_fn/std": 0.26369245946407316,
|
|
"step": 80
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completion_length": 100.0,
|
|
"completions/clipped_ratio": 1.0,
|
|
"completions/max_length": 100.0,
|
|
"completions/max_terminated_length": 0.0,
|
|
"completions/mean_length": 100.0,
|
|
"completions/mean_terminated_length": 0.0,
|
|
"completions/min_length": 100.0,
|
|
"completions/min_terminated_length": 0.0,
|
|
"epoch": 1.875,
|
|
"frac_reward_zero_std": 0.675,
|
|
"grad_norm": 0.3435279428958893,
|
|
"kl": 0.123704750277102,
|
|
"learning_rate": 2.131782945736434e-05,
|
|
"loss": 0.0001236904412508011,
|
|
"num_tokens": 279836.0,
|
|
"reward": -0.3937500029802322,
|
|
"reward_std": 0.18749999403953552,
|
|
"rewards/reward_fn/mean": -0.3937500029802322,
|
|
"rewards/reward_fn/std": 0.29925094842910765,
|
|
"step": 90
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completion_length": 100.0,
|
|
"completions/clipped_ratio": 1.0,
|
|
"completions/max_length": 100.0,
|
|
"completions/max_terminated_length": 0.0,
|
|
"completions/mean_length": 100.0,
|
|
"completions/mean_terminated_length": 0.0,
|
|
"completions/min_length": 100.0,
|
|
"completions/min_terminated_length": 0.0,
|
|
"epoch": 2.0833333333333335,
|
|
"frac_reward_zero_std": 0.725,
|
|
"grad_norm": 0.0014862708048895001,
|
|
"kl": 0.1360227341763675,
|
|
"learning_rate": 1.744186046511628e-05,
|
|
"loss": 0.0001360199647024274,
|
|
"num_tokens": 310908.0,
|
|
"reward": -0.33250000327825546,
|
|
"reward_std": 0.20633567869663239,
|
|
"rewards/reward_fn/mean": -0.3325000062584877,
|
|
"rewards/reward_fn/std": 0.3232262939214706,
|
|
"step": 100
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completion_length": 100.0,
|
|
"completions/clipped_ratio": 1.0,
|
|
"completions/max_length": 100.0,
|
|
"completions/max_terminated_length": 0.0,
|
|
"completions/mean_length": 100.0,
|
|
"completions/mean_terminated_length": 0.0,
|
|
"completions/min_length": 100.0,
|
|
"completions/min_terminated_length": 0.0,
|
|
"epoch": 2.2916666666666665,
|
|
"frac_reward_zero_std": 0.725,
|
|
"grad_norm": 0.2269871085882187,
|
|
"kl": 0.11381789557635784,
|
|
"learning_rate": 1.3565891472868217e-05,
|
|
"loss": 0.00011381134390830994,
|
|
"num_tokens": 341928.0,
|
|
"reward": -0.3806250035762787,
|
|
"reward_std": 0.16293191239237786,
|
|
"rewards/reward_fn/mean": -0.38062500655651094,
|
|
"rewards/reward_fn/std": 0.2841194272041321,
|
|
"step": 110
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completion_length": 100.0,
|
|
"completions/clipped_ratio": 1.0,
|
|
"completions/max_length": 100.0,
|
|
"completions/max_terminated_length": 0.0,
|
|
"completions/mean_length": 100.0,
|
|
"completions/mean_terminated_length": 0.0,
|
|
"completions/min_length": 100.0,
|
|
"completions/min_terminated_length": 0.0,
|
|
"epoch": 2.5,
|
|
"frac_reward_zero_std": 0.9,
|
|
"grad_norm": 0.22058050334453583,
|
|
"kl": 0.12228449042886495,
|
|
"learning_rate": 9.689922480620156e-06,
|
|
"loss": 0.0001222832128405571,
|
|
"num_tokens": 373136.0,
|
|
"reward": -0.318125008046627,
|
|
"reward_std": 0.06874999850988388,
|
|
"rewards/reward_fn/mean": -0.318125008046627,
|
|
"rewards/reward_fn/std": 0.1375,
|
|
"step": 120
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completion_length": 100.0,
|
|
"completions/clipped_ratio": 1.0,
|
|
"completions/max_length": 100.0,
|
|
"completions/max_terminated_length": 0.0,
|
|
"completions/mean_length": 100.0,
|
|
"completions/mean_terminated_length": 0.0,
|
|
"completions/min_length": 100.0,
|
|
"completions/min_terminated_length": 0.0,
|
|
"epoch": 2.7083333333333335,
|
|
"frac_reward_zero_std": 0.85,
|
|
"grad_norm": 0.003290389198809862,
|
|
"kl": 0.12402632609009742,
|
|
"learning_rate": 5.8139534883720935e-06,
|
|
"loss": 0.00012402099091559648,
|
|
"num_tokens": 404156.0,
|
|
"reward": -0.3487500101327896,
|
|
"reward_std": 0.09749999642372131,
|
|
"rewards/reward_fn/mean": -0.3487500071525574,
|
|
"rewards/reward_fn/std": 0.14059183597564698,
|
|
"step": 130
|
|
},
|
|
{
|
|
"clip_ratio/high_max": 0.0,
|
|
"clip_ratio/high_mean": 0.0,
|
|
"clip_ratio/low_mean": 0.0,
|
|
"clip_ratio/low_min": 0.0,
|
|
"clip_ratio/region_mean": 0.0,
|
|
"completion_length": 100.0,
|
|
"completions/clipped_ratio": 1.0,
|
|
"completions/max_length": 100.0,
|
|
"completions/max_terminated_length": 0.0,
|
|
"completions/mean_length": 100.0,
|
|
"completions/mean_terminated_length": 0.0,
|
|
"completions/min_length": 100.0,
|
|
"completions/min_terminated_length": 0.0,
|
|
"epoch": 2.9166666666666665,
|
|
"frac_reward_zero_std": 0.75,
|
|
"grad_norm": 0.45249366760253906,
|
|
"kl": 0.12512001674622297,
|
|
"learning_rate": 1.937984496124031e-06,
|
|
"loss": 0.00012511350214481354,
|
|
"num_tokens": 435268.0,
|
|
"reward": -0.3462500095367432,
|
|
"reward_std": 0.1474999986588955,
|
|
"rewards/reward_fn/mean": -0.34625000655651095,
|
|
"rewards/reward_fn/std": 0.25199120342731474,
|
|
"step": 140
|
|
}
|
|
],
|
|
"logging_steps": 10,
|
|
"max_steps": 144,
|
|
"num_input_tokens_seen": 447744,
|
|
"num_train_epochs": 3,
|
|
"save_steps": 50,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": true
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 0.0,
|
|
"train_batch_size": 4,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|