615 lines
21 KiB
JSON
615 lines
21 KiB
JSON
|
{
|
||
|
|
"best_global_step": null,
|
||
|
|
"best_metric": null,
|
||
|
|
"best_model_checkpoint": null,
|
||
|
|
"epoch": 0.4,
|
||
|
|
"eval_steps": 500,
|
||
|
|
"global_step": 100,
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.0,
|
||
|
|
"completions/max_length": 188.4,
|
||
|
|
"completions/max_terminated_length": 188.4,
|
||
|
|
"completions/mean_length": 121.775,
|
||
|
|
"completions/mean_terminated_length": 121.775,
|
||
|
|
"completions/min_length": 78.2,
|
||
|
|
"completions/min_terminated_length": 78.2,
|
||
|
|
"entropy": 0.24368538679555057,
|
||
|
|
"epoch": 0.02,
|
||
|
|
"frac_reward_zero_std": 0.5,
|
||
|
|
"grad_norm": 5.132767200469971,
|
||
|
|
"learning_rate": 9.84e-07,
|
||
|
|
"loss": 0.021476665139198305,
|
||
|
|
"num_tokens": 29677.0,
|
||
|
|
"reward": 0.5543750047683715,
|
||
|
|
"reward_std": 0.14613429307937623,
|
||
|
|
"rewards/reward_func/mean": 0.5543750047683715,
|
||
|
|
"rewards/reward_func/std": 0.14613429754972457,
|
||
|
|
"step": 5,
|
||
|
|
"step_time": 14.864119119599128,
|
||
|
|
"tools/call_frequency": 3.875,
|
||
|
|
"tools/failure_frequency": 0.8958312749862671
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.0,
|
||
|
|
"completions/max_length": 168.6,
|
||
|
|
"completions/max_terminated_length": 168.6,
|
||
|
|
"completions/mean_length": 130.725,
|
||
|
|
"completions/mean_terminated_length": 130.725,
|
||
|
|
"completions/min_length": 104.0,
|
||
|
|
"completions/min_terminated_length": 104.0,
|
||
|
|
"entropy": 0.24438673248514534,
|
||
|
|
"epoch": 0.04,
|
||
|
|
"frac_reward_zero_std": 0.5,
|
||
|
|
"grad_norm": 3.591146469116211,
|
||
|
|
"learning_rate": 9.64e-07,
|
||
|
|
"loss": 0.029430291056632994,
|
||
|
|
"num_tokens": 59716.0,
|
||
|
|
"reward": 0.5546875,
|
||
|
|
"reward_std": 0.14634020775556564,
|
||
|
|
"rewards/reward_func/mean": 0.5546875,
|
||
|
|
"rewards/reward_func/std": 0.14634020924568175,
|
||
|
|
"step": 10,
|
||
|
|
"step_time": 14.065522156999942,
|
||
|
|
"tools/call_frequency": 4.025,
|
||
|
|
"tools/failure_frequency": 0.9068181872367859
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.025,
|
||
|
|
"completions/max_length": 163.4,
|
||
|
|
"completions/max_terminated_length": 163.4,
|
||
|
|
"completions/mean_length": 128.0,
|
||
|
|
"completions/mean_terminated_length": 128.1500015258789,
|
||
|
|
"completions/min_length": 105.8,
|
||
|
|
"completions/min_terminated_length": 105.8,
|
||
|
|
"entropy": 0.2112921039108187,
|
||
|
|
"epoch": 0.06,
|
||
|
|
"frac_reward_zero_std": 0.4,
|
||
|
|
"grad_norm": 4.043612480163574,
|
||
|
|
"learning_rate": 9.439999999999999e-07,
|
||
|
|
"loss": 0.010420092940330505,
|
||
|
|
"num_tokens": 89639.0,
|
||
|
|
"reward": 0.5423749923706055,
|
||
|
|
"reward_std": 0.1744689464569092,
|
||
|
|
"rewards/reward_func/mean": 0.5423749923706055,
|
||
|
|
"rewards/reward_func/std": 0.17446894943714142,
|
||
|
|
"step": 15,
|
||
|
|
"step_time": 11.504540603999704,
|
||
|
|
"tools/call_frequency": 3.95,
|
||
|
|
"tools/failure_frequency": 0.8987920165061951
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.025,
|
||
|
|
"completions/max_length": 209.8,
|
||
|
|
"completions/max_terminated_length": 178.0,
|
||
|
|
"completions/mean_length": 138.075,
|
||
|
|
"completions/mean_terminated_length": 132.8,
|
||
|
|
"completions/min_length": 105.4,
|
||
|
|
"completions/min_terminated_length": 105.4,
|
||
|
|
"entropy": 0.22869818531908095,
|
||
|
|
"epoch": 0.08,
|
||
|
|
"frac_reward_zero_std": 0.5,
|
||
|
|
"grad_norm": 3.174783229827881,
|
||
|
|
"learning_rate": 9.24e-07,
|
||
|
|
"loss": 0.030705225467681885,
|
||
|
|
"num_tokens": 119967.0,
|
||
|
|
"reward": 0.5546875,
|
||
|
|
"reward_std": 0.15467961132526398,
|
||
|
|
"rewards/reward_func/mean": 0.5546875,
|
||
|
|
"rewards/reward_func/std": 0.15467961132526398,
|
||
|
|
"step": 20,
|
||
|
|
"step_time": 15.2172645851997,
|
||
|
|
"tools/call_frequency": 4.225,
|
||
|
|
"tools/failure_frequency": 0.9109591126441956
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.0,
|
||
|
|
"completions/max_length": 179.2,
|
||
|
|
"completions/max_terminated_length": 179.2,
|
||
|
|
"completions/mean_length": 133.6,
|
||
|
|
"completions/mean_terminated_length": 133.6,
|
||
|
|
"completions/min_length": 108.6,
|
||
|
|
"completions/min_terminated_length": 108.6,
|
||
|
|
"entropy": 0.23596344091929494,
|
||
|
|
"epoch": 0.1,
|
||
|
|
"frac_reward_zero_std": 0.5,
|
||
|
|
"grad_norm": 2.9762465953826904,
|
||
|
|
"learning_rate": 9.039999999999999e-07,
|
||
|
|
"loss": 0.03595758378505707,
|
||
|
|
"num_tokens": 150118.0,
|
||
|
|
"reward": 0.5421875,
|
||
|
|
"reward_std": 0.1739386200904846,
|
||
|
|
"rewards/reward_func/mean": 0.5421875,
|
||
|
|
"rewards/reward_func/std": 0.1739386200904846,
|
||
|
|
"step": 25,
|
||
|
|
"step_time": 11.905974677799714,
|
||
|
|
"tools/call_frequency": 4.05,
|
||
|
|
"tools/failure_frequency": 0.9073529362678527
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.0,
|
||
|
|
"completions/max_length": 174.6,
|
||
|
|
"completions/max_terminated_length": 174.6,
|
||
|
|
"completions/mean_length": 131.3,
|
||
|
|
"completions/mean_terminated_length": 131.3,
|
||
|
|
"completions/min_length": 106.4,
|
||
|
|
"completions/min_terminated_length": 106.4,
|
||
|
|
"entropy": 0.24171450519934295,
|
||
|
|
"epoch": 0.12,
|
||
|
|
"frac_reward_zero_std": 0.5,
|
||
|
|
"grad_norm": 2.964181900024414,
|
||
|
|
"learning_rate": 8.839999999999999e-07,
|
||
|
|
"loss": 0.03578909039497376,
|
||
|
|
"num_tokens": 180183.0,
|
||
|
|
"reward": 0.5546875,
|
||
|
|
"reward_std": 0.15467961132526398,
|
||
|
|
"rewards/reward_func/mean": 0.5546875,
|
||
|
|
"rewards/reward_func/std": 0.15467961132526398,
|
||
|
|
"step": 30,
|
||
|
|
"step_time": 11.773073928199846,
|
||
|
|
"tools/call_frequency": 4.025,
|
||
|
|
"tools/failure_frequency": 0.9068181872367859
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.0,
|
||
|
|
"completions/max_length": 164.0,
|
||
|
|
"completions/max_terminated_length": 164.0,
|
||
|
|
"completions/mean_length": 133.5,
|
||
|
|
"completions/mean_terminated_length": 133.5,
|
||
|
|
"completions/min_length": 106.4,
|
||
|
|
"completions/min_terminated_length": 106.4,
|
||
|
|
"entropy": 0.25739177097566424,
|
||
|
|
"epoch": 0.14,
|
||
|
|
"frac_reward_zero_std": 0.5,
|
||
|
|
"grad_norm": 3.43412184715271,
|
||
|
|
"learning_rate": 8.639999999999999e-07,
|
||
|
|
"loss": 0.037665295600891116,
|
||
|
|
"num_tokens": 210329.0,
|
||
|
|
"reward": 0.5546875,
|
||
|
|
"reward_std": 0.15467961132526398,
|
||
|
|
"rewards/reward_func/mean": 0.5546875,
|
||
|
|
"rewards/reward_func/std": 0.15467961132526398,
|
||
|
|
"step": 35,
|
||
|
|
"step_time": 11.653867780799919,
|
||
|
|
"tools/call_frequency": 4.1,
|
||
|
|
"tools/failure_frequency": 0.9084893107414246
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.0,
|
||
|
|
"completions/max_length": 163.0,
|
||
|
|
"completions/max_terminated_length": 163.0,
|
||
|
|
"completions/mean_length": 130.35,
|
||
|
|
"completions/mean_terminated_length": 130.35,
|
||
|
|
"completions/min_length": 105.2,
|
||
|
|
"completions/min_terminated_length": 105.2,
|
||
|
|
"entropy": 0.24240239977370948,
|
||
|
|
"epoch": 0.16,
|
||
|
|
"frac_reward_zero_std": 0.5,
|
||
|
|
"grad_norm": 1.9415134191513062,
|
||
|
|
"learning_rate": 8.439999999999999e-07,
|
||
|
|
"loss": 0.038230502605438234,
|
||
|
|
"num_tokens": 240351.0,
|
||
|
|
"reward": 0.5546875,
|
||
|
|
"reward_std": 0.15467961132526398,
|
||
|
|
"rewards/reward_func/mean": 0.5546875,
|
||
|
|
"rewards/reward_func/std": 0.15467961132526398,
|
||
|
|
"step": 40,
|
||
|
|
"step_time": 11.493494538199593,
|
||
|
|
"tools/call_frequency": 4.1,
|
||
|
|
"tools/failure_frequency": 0.9085227489471436
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.0,
|
||
|
|
"completions/max_length": 189.2,
|
||
|
|
"completions/max_terminated_length": 189.2,
|
||
|
|
"completions/mean_length": 144.325,
|
||
|
|
"completions/mean_terminated_length": 144.325,
|
||
|
|
"completions/min_length": 103.4,
|
||
|
|
"completions/min_terminated_length": 103.4,
|
||
|
|
"entropy": 0.2922910688444972,
|
||
|
|
"epoch": 0.18,
|
||
|
|
"frac_reward_zero_std": 0.5,
|
||
|
|
"grad_norm": 2.625551223754883,
|
||
|
|
"learning_rate": 8.24e-07,
|
||
|
|
"loss": 0.029571545124053956,
|
||
|
|
"num_tokens": 270934.0,
|
||
|
|
"reward": 0.5421875,
|
||
|
|
"reward_std": 0.1739386200904846,
|
||
|
|
"rewards/reward_func/mean": 0.5421875,
|
||
|
|
"rewards/reward_func/std": 0.1739386200904846,
|
||
|
|
"step": 45,
|
||
|
|
"step_time": 13.904022978000103,
|
||
|
|
"tools/call_frequency": 4.3,
|
||
|
|
"tools/failure_frequency": 0.9124376058578492
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.0,
|
||
|
|
"completions/max_length": 173.4,
|
||
|
|
"completions/max_terminated_length": 173.4,
|
||
|
|
"completions/mean_length": 136.9,
|
||
|
|
"completions/mean_terminated_length": 136.9,
|
||
|
|
"completions/min_length": 102.0,
|
||
|
|
"completions/min_terminated_length": 102.0,
|
||
|
|
"entropy": 0.30844923059921714,
|
||
|
|
"epoch": 0.2,
|
||
|
|
"frac_reward_zero_std": 0.5,
|
||
|
|
"grad_norm": 2.4793224334716797,
|
||
|
|
"learning_rate": 8.04e-07,
|
||
|
|
"loss": 0.05630779266357422,
|
||
|
|
"num_tokens": 301212.0,
|
||
|
|
"reward": 0.5546875,
|
||
|
|
"reward_std": 0.15467961132526398,
|
||
|
|
"rewards/reward_func/mean": 0.5546875,
|
||
|
|
"rewards/reward_func/std": 0.15467961132526398,
|
||
|
|
"step": 50,
|
||
|
|
"step_time": 12.183540330400138,
|
||
|
|
"tools/call_frequency": 4.075,
|
||
|
|
"tools/failure_frequency": 0.9079211235046387
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.0,
|
||
|
|
"completions/max_length": 169.2,
|
||
|
|
"completions/max_terminated_length": 169.2,
|
||
|
|
"completions/mean_length": 133.675,
|
||
|
|
"completions/mean_terminated_length": 133.675,
|
||
|
|
"completions/min_length": 104.0,
|
||
|
|
"completions/min_terminated_length": 104.0,
|
||
|
|
"entropy": 0.2755190956639126,
|
||
|
|
"epoch": 0.22,
|
||
|
|
"frac_reward_zero_std": 0.5,
|
||
|
|
"grad_norm": 2.5465078353881836,
|
||
|
|
"learning_rate": 7.84e-07,
|
||
|
|
"loss": 0.044404512643814086,
|
||
|
|
"num_tokens": 331366.0,
|
||
|
|
"reward": 0.5546875,
|
||
|
|
"reward_std": 0.15467961132526398,
|
||
|
|
"rewards/reward_func/mean": 0.5546875,
|
||
|
|
"rewards/reward_func/std": 0.15467961132526398,
|
||
|
|
"step": 55,
|
||
|
|
"step_time": 12.093883183000253,
|
||
|
|
"tools/call_frequency": 4.05,
|
||
|
|
"tools/failure_frequency": 0.9073863744735717
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.0,
|
||
|
|
"completions/max_length": 169.4,
|
||
|
|
"completions/max_terminated_length": 169.4,
|
||
|
|
"completions/mean_length": 131.625,
|
||
|
|
"completions/mean_terminated_length": 131.625,
|
||
|
|
"completions/min_length": 99.6,
|
||
|
|
"completions/min_terminated_length": 99.6,
|
||
|
|
"entropy": 0.27003198163583875,
|
||
|
|
"epoch": 0.24,
|
||
|
|
"frac_reward_zero_std": 0.4,
|
||
|
|
"grad_norm": 2.265756845474243,
|
||
|
|
"learning_rate": 7.64e-07,
|
||
|
|
"loss": 0.023787906765937804,
|
||
|
|
"num_tokens": 361436.0,
|
||
|
|
"reward": 0.5421875,
|
||
|
|
"reward_std": 0.1739386200904846,
|
||
|
|
"rewards/reward_func/mean": 0.5421875,
|
||
|
|
"rewards/reward_func/std": 0.1739386200904846,
|
||
|
|
"step": 60,
|
||
|
|
"step_time": 11.345259791600256,
|
||
|
|
"tools/call_frequency": 3.975,
|
||
|
|
"tools/failure_frequency": 0.9049531102180481
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.0,
|
||
|
|
"completions/max_length": 161.0,
|
||
|
|
"completions/max_terminated_length": 161.0,
|
||
|
|
"completions/mean_length": 134.475,
|
||
|
|
"completions/mean_terminated_length": 134.475,
|
||
|
|
"completions/min_length": 102.2,
|
||
|
|
"completions/min_terminated_length": 102.2,
|
||
|
|
"entropy": 0.31135775299044327,
|
||
|
|
"epoch": 0.26,
|
||
|
|
"frac_reward_zero_std": 0.5,
|
||
|
|
"grad_norm": 2.0467045307159424,
|
||
|
|
"learning_rate": 7.44e-07,
|
||
|
|
"loss": 0.03975436091423035,
|
||
|
|
"num_tokens": 391623.0,
|
||
|
|
"reward": 0.5421875,
|
||
|
|
"reward_std": 0.1739386200904846,
|
||
|
|
"rewards/reward_func/mean": 0.5421875,
|
||
|
|
"rewards/reward_func/std": 0.1739386200904846,
|
||
|
|
"step": 65,
|
||
|
|
"step_time": 11.19052229420031,
|
||
|
|
"tools/call_frequency": 3.9,
|
||
|
|
"tools/failure_frequency": 0.9035714268684387
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.0,
|
||
|
|
"completions/max_length": 187.2,
|
||
|
|
"completions/max_terminated_length": 187.2,
|
||
|
|
"completions/mean_length": 138.75,
|
||
|
|
"completions/mean_terminated_length": 138.75,
|
||
|
|
"completions/min_length": 109.8,
|
||
|
|
"completions/min_terminated_length": 109.8,
|
||
|
|
"entropy": 0.30972914764424786,
|
||
|
|
"epoch": 0.28,
|
||
|
|
"frac_reward_zero_std": 0.5,
|
||
|
|
"grad_norm": 1.9595667123794556,
|
||
|
|
"learning_rate": 7.24e-07,
|
||
|
|
"loss": 0.04363401532173157,
|
||
|
|
"num_tokens": 421981.0,
|
||
|
|
"reward": 0.5546875,
|
||
|
|
"reward_std": 0.15467961132526398,
|
||
|
|
"rewards/reward_func/mean": 0.5546875,
|
||
|
|
"rewards/reward_func/std": 0.15467961132526398,
|
||
|
|
"step": 70,
|
||
|
|
"step_time": 12.639552449400071,
|
||
|
|
"tools/call_frequency": 4.0,
|
||
|
|
"tools/failure_frequency": 0.90625
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.0,
|
||
|
|
"completions/max_length": 229.0,
|
||
|
|
"completions/max_terminated_length": 229.0,
|
||
|
|
"completions/mean_length": 151.35,
|
||
|
|
"completions/mean_terminated_length": 151.35,
|
||
|
|
"completions/min_length": 108.4,
|
||
|
|
"completions/min_terminated_length": 108.4,
|
||
|
|
"entropy": 0.3386134950676933,
|
||
|
|
"epoch": 0.3,
|
||
|
|
"frac_reward_zero_std": 0.5,
|
||
|
|
"grad_norm": 2.162438154220581,
|
||
|
|
"learning_rate": 7.04e-07,
|
||
|
|
"loss": 0.10028836727142335,
|
||
|
|
"num_tokens": 452843.0,
|
||
|
|
"reward": 0.5421875,
|
||
|
|
"reward_std": 0.1739386200904846,
|
||
|
|
"rewards/reward_func/mean": 0.5421875,
|
||
|
|
"rewards/reward_func/std": 0.1739386200904846,
|
||
|
|
"step": 75,
|
||
|
|
"step_time": 15.725879665600042,
|
||
|
|
"tools/call_frequency": 3.975,
|
||
|
|
"tools/failure_frequency": 0.9052759885787964
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.0,
|
||
|
|
"completions/max_length": 177.8,
|
||
|
|
"completions/max_terminated_length": 177.8,
|
||
|
|
"completions/mean_length": 142.1,
|
||
|
|
"completions/mean_terminated_length": 142.1,
|
||
|
|
"completions/min_length": 108.2,
|
||
|
|
"completions/min_terminated_length": 108.2,
|
||
|
|
"entropy": 0.3440137348137796,
|
||
|
|
"epoch": 0.32,
|
||
|
|
"frac_reward_zero_std": 0.5,
|
||
|
|
"grad_norm": 1.5661996603012085,
|
||
|
|
"learning_rate": 6.84e-07,
|
||
|
|
"loss": 0.04704100489616394,
|
||
|
|
"num_tokens": 483332.0,
|
||
|
|
"reward": 0.5546875,
|
||
|
|
"reward_std": 0.15467961132526398,
|
||
|
|
"rewards/reward_func/mean": 0.5546875,
|
||
|
|
"rewards/reward_func/std": 0.15467961132526398,
|
||
|
|
"step": 80,
|
||
|
|
"step_time": 12.160148031599965,
|
||
|
|
"tools/call_frequency": 4.025,
|
||
|
|
"tools/failure_frequency": 0.9068181872367859
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.025,
|
||
|
|
"completions/max_length": 201.8,
|
||
|
|
"completions/max_terminated_length": 201.8,
|
||
|
|
"completions/mean_length": 153.2,
|
||
|
|
"completions/mean_terminated_length": 152.66428833007814,
|
||
|
|
"completions/min_length": 112.0,
|
||
|
|
"completions/min_terminated_length": 112.0,
|
||
|
|
"entropy": 0.4032384747406468,
|
||
|
|
"epoch": 0.34,
|
||
|
|
"frac_reward_zero_std": 0.5,
|
||
|
|
"grad_norm": 4.639192581176758,
|
||
|
|
"learning_rate": 6.64e-07,
|
||
|
|
"loss": 0.01302967369556427,
|
||
|
|
"num_tokens": 514266.0,
|
||
|
|
"reward": 0.5546875,
|
||
|
|
"reward_std": 0.15467961132526398,
|
||
|
|
"rewards/reward_func/mean": 0.5546875,
|
||
|
|
"rewards/reward_func/std": 0.15467961132526398,
|
||
|
|
"step": 85,
|
||
|
|
"step_time": 14.85692119880041,
|
||
|
|
"tools/call_frequency": 4.275,
|
||
|
|
"tools/failure_frequency": 0.9116071462631226
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.0,
|
||
|
|
"completions/max_length": 217.0,
|
||
|
|
"completions/max_terminated_length": 217.0,
|
||
|
|
"completions/mean_length": 155.25,
|
||
|
|
"completions/mean_terminated_length": 155.25,
|
||
|
|
"completions/min_length": 107.2,
|
||
|
|
"completions/min_terminated_length": 107.2,
|
||
|
|
"entropy": 0.3891718311235309,
|
||
|
|
"epoch": 0.36,
|
||
|
|
"frac_reward_zero_std": 0.5,
|
||
|
|
"grad_norm": 2.346120834350586,
|
||
|
|
"learning_rate": 6.44e-07,
|
||
|
|
"loss": 0.07838824391365051,
|
||
|
|
"num_tokens": 545274.0,
|
||
|
|
"reward": 0.5546875,
|
||
|
|
"reward_std": 0.15467961132526398,
|
||
|
|
"rewards/reward_func/mean": 0.5546875,
|
||
|
|
"rewards/reward_func/std": 0.15467961132526398,
|
||
|
|
"step": 90,
|
||
|
|
"step_time": 14.502108011600649,
|
||
|
|
"tools/call_frequency": 4.05,
|
||
|
|
"tools/failure_frequency": 0.9073529362678527
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.025,
|
||
|
|
"completions/max_length": 277.2,
|
||
|
|
"completions/max_terminated_length": 277.2,
|
||
|
|
"completions/mean_length": 173.75,
|
||
|
|
"completions/mean_terminated_length": 174.2357147216797,
|
||
|
|
"completions/min_length": 115.4,
|
||
|
|
"completions/min_terminated_length": 115.4,
|
||
|
|
"entropy": 0.3660696781706065,
|
||
|
|
"epoch": 0.38,
|
||
|
|
"frac_reward_zero_std": 0.5,
|
||
|
|
"grad_norm": 2.0744283199310303,
|
||
|
|
"learning_rate": 6.24e-07,
|
||
|
|
"loss": 0.04411016702651978,
|
||
|
|
"num_tokens": 577035.0,
|
||
|
|
"reward": 0.5546875,
|
||
|
|
"reward_std": 0.15467961132526398,
|
||
|
|
"rewards/reward_func/mean": 0.5546875,
|
||
|
|
"rewards/reward_func/std": 0.15467961132526398,
|
||
|
|
"step": 95,
|
||
|
|
"step_time": 17.604619891600485,
|
||
|
|
"tools/call_frequency": 5.275,
|
||
|
|
"tools/failure_frequency": 0.927957046031952
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"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,
|
||
|
|
"completions/clipped_ratio": 0.025,
|
||
|
|
"completions/max_length": 262.6,
|
||
|
|
"completions/max_terminated_length": 249.4,
|
||
|
|
"completions/mean_length": 185.4,
|
||
|
|
"completions/mean_terminated_length": 181.58214416503907,
|
||
|
|
"completions/min_length": 113.4,
|
||
|
|
"completions/min_terminated_length": 113.4,
|
||
|
|
"entropy": 0.3639457892393693,
|
||
|
|
"epoch": 0.4,
|
||
|
|
"frac_reward_zero_std": 0.5,
|
||
|
|
"grad_norm": 2.8498334884643555,
|
||
|
|
"learning_rate": 6.04e-07,
|
||
|
|
"loss": 0.06711583137512207,
|
||
|
|
"num_tokens": 609256.0,
|
||
|
|
"reward": 0.5546875,
|
||
|
|
"reward_std": 0.15467961132526398,
|
||
|
|
"rewards/reward_func/mean": 0.5546875,
|
||
|
|
"rewards/reward_func/std": 0.15467961132526398,
|
||
|
|
"step": 100,
|
||
|
|
"step_time": 18.920671705199673,
|
||
|
|
"tools/call_frequency": 5.6,
|
||
|
|
"tools/failure_frequency": 0.9292140007019043
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"logging_steps": 5,
|
||
|
|
"max_steps": 250,
|
||
|
|
"num_input_tokens_seen": 609256,
|
||
|
|
"num_train_epochs": 1,
|
||
|
|
"save_steps": 50,
|
||
|
|
"stateful_callbacks": {
|
||
|
|
"TrainerControl": {
|
||
|
|
"args": {
|
||
|
|
"should_epoch_stop": false,
|
||
|
|
"should_evaluate": false,
|
||
|
|
"should_log": false,
|
||
|
|
"should_save": true,
|
||
|
|
"should_training_stop": false
|
||
|
|
},
|
||
|
|
"attributes": {}
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"total_flos": 0.0,
|
||
|
|
"train_batch_size": 1,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|