Files
counsel-env-qwen3-0.6b-grpo/checkpoint-200/trainer_state.json
ModelHub XC 2ea3d793ee 初始化项目,由ModelHub XC社区提供模型
Model: heavycoderhh/counsel-env-qwen3-0.6b-grpo
Source: Original Platform
2026-06-16 07:21:17 +08:00

1235 lines
44 KiB
JSON

{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.78125,
"eval_steps": 500,
"global_step": 200,
"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": 152.4,
"completions/max_terminated_length": 152.4,
"completions/mean_length": 117.8,
"completions/mean_terminated_length": 117.8,
"completions/min_length": 93.8,
"completions/min_terminated_length": 93.8,
"entropy": 0.20053473562002183,
"epoch": 0.01953125,
"frac_reward_zero_std": 0.0,
"grad_norm": 20.259794235229492,
"kl": 0.1606330933049321,
"learning_rate": 9.800000000000001e-06,
"loss": 0.020250317454338074,
"num_tokens": 11544.0,
"reward": 0.4544999921694398,
"reward_std": 0.2574530947953463,
"rewards/reward_func/mean": 0.4544999921694398,
"rewards/reward_func/std": 0.2574530977755785,
"step": 5,
"step_time": 8.736884885399922,
"tools/call_frequency": 3.1,
"tools/failure_frequency": 0.0
},
{
"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": 205.8,
"completions/max_terminated_length": 205.8,
"completions/mean_length": 174.4,
"completions/mean_terminated_length": 174.4,
"completions/min_length": 140.6,
"completions/min_terminated_length": 140.6,
"entropy": 0.45406929701566695,
"epoch": 0.0390625,
"frac_reward_zero_std": 0.0,
"grad_norm": 20.727792739868164,
"kl": 0.5858453318476677,
"learning_rate": 9.55e-06,
"loss": 0.00982179045677185,
"num_tokens": 24254.0,
"reward": 0.08549999967217445,
"reward_std": 0.14715017192065716,
"rewards/reward_func/mean": 0.08549999967217445,
"rewards/reward_func/std": 0.14715017192065716,
"step": 10,
"step_time": 11.54933958799811,
"tools/call_frequency": 3.45,
"tools/failure_frequency": 0.0
},
{
"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": 0.35,
"completions/max_length": 320.8,
"completions/max_terminated_length": 233.6,
"completions/mean_length": 232.35,
"completions/mean_terminated_length": 189.8,
"completions/min_length": 141.0,
"completions/min_terminated_length": 141.0,
"entropy": 0.5534275218844413,
"epoch": 0.05859375,
"frac_reward_zero_std": 0.2,
"grad_norm": 6.85633659362793,
"kl": 0.6838902086019516,
"learning_rate": 9.3e-06,
"loss": 0.026860833168029785,
"num_tokens": 38189.0,
"reward": 0.10483333505690098,
"reward_std": 0.16720814146101476,
"rewards/reward_func/mean": 0.10483333505690098,
"rewards/reward_func/std": 0.1672081384807825,
"step": 15,
"step_time": 18.40421949800075,
"tools/call_frequency": 4.35,
"tools/failure_frequency": 0.0
},
{
"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": 0.4,
"completions/max_length": 413.0,
"completions/max_terminated_length": 351.2,
"completions/mean_length": 340.5,
"completions/mean_terminated_length": 306.28333435058596,
"completions/min_length": 263.0,
"completions/min_terminated_length": 263.0,
"entropy": 0.5062903374433517,
"epoch": 0.078125,
"frac_reward_zero_std": 0.4,
"grad_norm": 6.486283302307129,
"kl": 0.5253867790102958,
"learning_rate": 9.050000000000001e-06,
"loss": -0.045569175481796266,
"num_tokens": 54196.0,
"reward": 0.09649999756366015,
"reward_std": 0.10897674113512039,
"rewards/reward_func/mean": 0.09649999756366015,
"rewards/reward_func/std": 0.10897674113512039,
"step": 20,
"step_time": 29.913843032000294,
"tools/call_frequency": 3.95,
"tools/failure_frequency": 0.22069264352321624
},
{
"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": 0.65,
"completions/max_length": 480.4,
"completions/max_terminated_length": 241.0,
"completions/mean_length": 361.25,
"completions/mean_terminated_length": 225.56666870117186,
"completions/min_length": 247.4,
"completions/min_terminated_length": 202.8,
"entropy": 0.4256091395393014,
"epoch": 0.09765625,
"frac_reward_zero_std": 0.2,
"grad_norm": 4.082395553588867,
"kl": 0.43983293175697324,
"learning_rate": 8.8e-06,
"loss": 0.02868407070636749,
"num_tokens": 70641.0,
"reward": 0.0399999987334013,
"reward_std": 0.060056403279304504,
"rewards/reward_func/mean": 0.0399999987334013,
"rewards/reward_func/std": 0.060056403279304504,
"step": 25,
"step_time": 35.14591444559992,
"tools/call_frequency": 3.7,
"tools/failure_frequency": 0.18648459613323212
},
{
"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": 0.4,
"completions/max_length": 342.0,
"completions/max_terminated_length": 286.6,
"completions/mean_length": 267.45,
"completions/mean_terminated_length": 208.86666870117188,
"completions/min_length": 93.2,
"completions/min_terminated_length": 93.2,
"entropy": 0.27065443359315394,
"epoch": 0.1171875,
"frac_reward_zero_std": 0.0,
"grad_norm": 5.178186893463135,
"kl": 0.6691923886537552,
"learning_rate": 8.550000000000001e-06,
"loss": -0.30545821189880373,
"num_tokens": 85232.0,
"reward": 0.029999999329447746,
"reward_std": 0.019999999552965164,
"rewards/reward_func/mean": 0.029999999329447746,
"rewards/reward_func/std": 0.019999999552965164,
"step": 30,
"step_time": 21.171669380999084,
"tools/call_frequency": 5.3,
"tools/failure_frequency": 0.30696970224380493
},
{
"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": 0.4,
"completions/max_length": 314.6,
"completions/max_terminated_length": 300.6,
"completions/mean_length": 275.95,
"completions/mean_terminated_length": 258.2,
"completions/min_length": 191.6,
"completions/min_terminated_length": 193.2,
"entropy": 0.12053632475435734,
"epoch": 0.13671875,
"frac_reward_zero_std": 0.0,
"grad_norm": 3.0749428272247314,
"kl": 0.6381830915808677,
"learning_rate": 8.3e-06,
"loss": -0.07461150288581848,
"num_tokens": 100010.0,
"reward": 0.023999999463558196,
"reward_std": 0.021856406703591347,
"rewards/reward_func/mean": 0.023999999463558196,
"rewards/reward_func/std": 0.021856406703591347,
"step": 35,
"step_time": 18.075483031600744,
"tools/call_frequency": 7.2,
"tools/failure_frequency": 0.3397361934185028
},
{
"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": 0.55,
"completions/max_length": 406.2,
"completions/max_terminated_length": 194.6,
"completions/mean_length": 318.55,
"completions/mean_terminated_length": 139.6666717529297,
"completions/min_length": 207.2,
"completions/min_terminated_length": 81.6,
"entropy": 0.25037811435759066,
"epoch": 0.15625,
"frac_reward_zero_std": 0.2,
"grad_norm": 5.45590877532959,
"kl": 0.632529079169035,
"learning_rate": 8.050000000000001e-06,
"loss": -0.08673273324966431,
"num_tokens": 115597.0,
"reward": 0.013999999687075614,
"reward_std": 0.016618802025914193,
"rewards/reward_func/mean": 0.013999999687075614,
"rewards/reward_func/std": 0.016618802025914193,
"step": 40,
"step_time": 25.49324105279957,
"tools/call_frequency": 4.8,
"tools/failure_frequency": 0.2961344659328461
},
{
"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": 0.7,
"completions/max_length": 488.8,
"completions/max_terminated_length": 167.4,
"completions/mean_length": 351.65,
"completions/mean_terminated_length": 126.56666870117188,
"completions/min_length": 218.6,
"completions/min_terminated_length": 102.6,
"entropy": 0.30095756258815526,
"epoch": 0.17578125,
"frac_reward_zero_std": 0.2,
"grad_norm": 2.2991526126861572,
"kl": 0.549846900999546,
"learning_rate": 7.800000000000002e-06,
"loss": 0.01944226920604706,
"num_tokens": 131856.0,
"reward": 0.017999999597668646,
"reward_std": 0.016618802025914193,
"rewards/reward_func/mean": 0.017999999597668646,
"rewards/reward_func/std": 0.016618802025914193,
"step": 45,
"step_time": 35.21552088999961,
"tools/call_frequency": 3.3,
"tools/failure_frequency": 0.2740601599216461
},
{
"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": 0.6,
"completions/max_length": 381.0,
"completions/max_terminated_length": 284.0,
"completions/mean_length": 294.35,
"completions/mean_terminated_length": 227.53333435058593,
"completions/min_length": 160.8,
"completions/min_terminated_length": 160.8,
"entropy": 0.23832304682582617,
"epoch": 0.1953125,
"frac_reward_zero_std": 0.2,
"grad_norm": 0.2268025130033493,
"kl": 0.6058375500142574,
"learning_rate": 7.5500000000000006e-06,
"loss": -0.08301904201507568,
"num_tokens": 146914.0,
"reward": 0.023999999463558196,
"reward_std": 0.018475209176540375,
"rewards/reward_func/mean": 0.023999999463558196,
"rewards/reward_func/std": 0.018475209176540375,
"step": 50,
"step_time": 25.430086624599788,
"tools/call_frequency": 5.5,
"tools/failure_frequency": 0.30512219071388247
},
{
"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": 0.5,
"completions/max_length": 339.2,
"completions/max_terminated_length": 290.2,
"completions/mean_length": 287.3,
"completions/mean_terminated_length": 284.8333374023438,
"completions/min_length": 262.0,
"completions/min_terminated_length": 279.2,
"entropy": 0.05971384542062878,
"epoch": 0.21484375,
"frac_reward_zero_std": 0.4,
"grad_norm": 0.22013917565345764,
"kl": 0.5083232939243316,
"learning_rate": 7.3e-06,
"loss": 0.0854921281337738,
"num_tokens": 161945.0,
"reward": 0.031999999284744264,
"reward_std": 0.012618802115321159,
"rewards/reward_func/mean": 0.031999999284744264,
"rewards/reward_func/std": 0.012618802115321159,
"step": 55,
"step_time": 21.9785563156016,
"tools/call_frequency": 8.3,
"tools/failure_frequency": 0.3500013709068298
},
{
"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": 0.6,
"completions/max_length": 264.6,
"completions/max_terminated_length": 219.6,
"completions/mean_length": 249.7,
"completions/mean_terminated_length": 217.26666870117188,
"completions/min_length": 213.4,
"completions/min_terminated_length": 215.6,
"entropy": 0.030067690974101426,
"epoch": 0.234375,
"frac_reward_zero_std": 0.2,
"grad_norm": 2.1762969493865967,
"kl": 0.5388822212815285,
"learning_rate": 7.05e-06,
"loss": -0.06041957139968872,
"num_tokens": 176143.0,
"reward": 0.029999999329447746,
"reward_std": 0.016618802025914193,
"rewards/reward_func/mean": 0.029999999329447746,
"rewards/reward_func/std": 0.016618802025914193,
"step": 60,
"step_time": 16.232493841598625,
"tools/call_frequency": 8.85,
"tools/failure_frequency": 0.3545756280422211
},
{
"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": 0.4,
"completions/max_length": 288.2,
"completions/max_terminated_length": 148.2,
"completions/mean_length": 213.95,
"completions/mean_terminated_length": 137.4000030517578,
"completions/min_length": 181.4,
"completions/min_terminated_length": 129.6,
"entropy": 0.07268587870057672,
"epoch": 0.25390625,
"frac_reward_zero_std": 0.4,
"grad_norm": 0.1308048814535141,
"kl": 0.8497638419270516,
"learning_rate": 6.800000000000001e-06,
"loss": -0.2692496538162231,
"num_tokens": 189629.0,
"reward": 0.03799999877810478,
"reward_std": 0.010309400968253613,
"rewards/reward_func/mean": 0.03799999877810478,
"rewards/reward_func/std": 0.010309400968253613,
"step": 65,
"step_time": 17.90531996159916,
"tools/call_frequency": 5.85,
"tools/failure_frequency": 0.3754902005195618
},
{
"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": 0.3,
"completions/max_length": 274.6,
"completions/max_terminated_length": 256.6,
"completions/mean_length": 248.25,
"completions/mean_terminated_length": 241.73333740234375,
"completions/min_length": 226.0,
"completions/min_terminated_length": 226.0,
"entropy": 0.05142406928353012,
"epoch": 0.2734375,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.09222911298274994,
"kl": 0.57062017172575,
"learning_rate": 6.550000000000001e-06,
"loss": 0.02273084670305252,
"num_tokens": 203795.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 70,
"step_time": 16.734460433397906,
"tools/call_frequency": 9.3,
"tools/failure_frequency": 0.1542351394891739
},
{
"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": 0.35,
"completions/max_length": 303.4,
"completions/max_terminated_length": 242.2,
"completions/mean_length": 244.05,
"completions/mean_terminated_length": 231.43333435058594,
"completions/min_length": 215.6,
"completions/min_terminated_length": 222.2,
"entropy": 0.03808064609766006,
"epoch": 0.29296875,
"frac_reward_zero_std": 0.8,
"grad_norm": 0.2428537756204605,
"kl": 0.49708728566765786,
"learning_rate": 6.300000000000001e-06,
"loss": 0.08911536335945129,
"num_tokens": 217958.0,
"reward": 0.0569999985396862,
"reward_std": 0.006000000238418579,
"rewards/reward_func/mean": 0.0569999985396862,
"rewards/reward_func/std": 0.005999999865889549,
"step": 75,
"step_time": 18.7449420207995,
"tools/call_frequency": 9.25,
"tools/failure_frequency": 0.09236167445778846
},
{
"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": 0.4,
"completions/max_length": 298.6,
"completions/max_terminated_length": 238.2,
"completions/mean_length": 243.95,
"completions/mean_terminated_length": 229.53333435058593,
"completions/min_length": 217.8,
"completions/min_terminated_length": 221.8,
"entropy": 0.03462462190072983,
"epoch": 0.3125,
"frac_reward_zero_std": 0.6,
"grad_norm": 3.802891969680786,
"kl": 0.48679155968129634,
"learning_rate": 6.0500000000000005e-06,
"loss": 0.0853162169456482,
"num_tokens": 232107.0,
"reward": 0.05399999842047691,
"reward_std": 0.012000000476837159,
"rewards/reward_func/mean": 0.05399999842047691,
"rewards/reward_func/std": 0.011999999731779098,
"step": 80,
"step_time": 18.624738503398838,
"tools/call_frequency": 9.4,
"tools/failure_frequency": 0.1012024775147438
},
{
"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": 0.55,
"completions/max_length": 315.8,
"completions/max_terminated_length": 245.4,
"completions/mean_length": 287.0,
"completions/mean_terminated_length": 223.70000305175782,
"completions/min_length": 235.4,
"completions/min_terminated_length": 188.8,
"entropy": 0.018990062386728825,
"epoch": 0.33203125,
"frac_reward_zero_std": 0.8,
"grad_norm": 0.24913819134235382,
"kl": 0.36854752600193025,
"learning_rate": 5.8e-06,
"loss": -0.06027485728263855,
"num_tokens": 247068.0,
"reward": 0.0569999985396862,
"reward_std": 0.006000000238418579,
"rewards/reward_func/mean": 0.0569999985396862,
"rewards/reward_func/std": 0.005999999865889549,
"step": 85,
"step_time": 19.23757492739969,
"tools/call_frequency": 7.25,
"tools/failure_frequency": 0.38577277064323423
},
{
"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": 0.6,
"completions/max_length": 336.4,
"completions/max_terminated_length": 313.2,
"completions/mean_length": 315.8,
"completions/mean_terminated_length": 307.6,
"completions/min_length": 291.8,
"completions/min_terminated_length": 302.0,
"entropy": 0.005942635895917192,
"epoch": 0.3515625,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.03795464709401131,
"kl": 0.27816674262285235,
"learning_rate": 5.550000000000001e-06,
"loss": 0.011118948459625244,
"num_tokens": 262655.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 90,
"step_time": 20.623258255800465,
"tools/call_frequency": 7.0,
"tools/failure_frequency": 0.378443706035614
},
{
"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": 0.3,
"completions/max_length": 326.2,
"completions/max_terminated_length": 326.2,
"completions/mean_length": 305.55,
"completions/mean_terminated_length": 311.4333435058594,
"completions/min_length": 286.4,
"completions/min_terminated_length": 300.0,
"entropy": 0.00408092990401201,
"epoch": 0.37109375,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.03232214227318764,
"kl": 0.2662045076489449,
"learning_rate": 5.300000000000001e-06,
"loss": 0.010630997270345688,
"num_tokens": 277935.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 95,
"step_time": 19.27180065259963,
"tools/call_frequency": 7.5,
"tools/failure_frequency": 0.40017797946929934
},
{
"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": 0.4,
"completions/max_length": 370.4,
"completions/max_terminated_length": 320.0,
"completions/mean_length": 305.75,
"completions/mean_terminated_length": 274.26666870117185,
"completions/min_length": 213.2,
"completions/min_terminated_length": 213.8,
"entropy": 0.020784856198588386,
"epoch": 0.390625,
"frac_reward_zero_std": 0.6,
"grad_norm": 2.253676414489746,
"kl": 0.30799318477511406,
"learning_rate": 5.050000000000001e-06,
"loss": -0.07253098487854004,
"num_tokens": 293245.0,
"reward": 0.050999998673796655,
"reward_std": 0.012928203493356705,
"rewards/reward_func/mean": 0.050999998673796655,
"rewards/reward_func/std": 0.012928203120827675,
"step": 100,
"step_time": 22.40414413039907,
"tools/call_frequency": 5.7,
"tools/failure_frequency": 0.3851664662361145
},
{
"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": 0.65,
"completions/max_length": 337.8,
"completions/max_terminated_length": 273.4,
"completions/mean_length": 313.45,
"completions/mean_terminated_length": 271.1,
"completions/min_length": 262.0,
"completions/min_terminated_length": 268.8,
"entropy": 0.04189231124473736,
"epoch": 0.41015625,
"frac_reward_zero_std": 0.8,
"grad_norm": 0.11439066380262375,
"kl": 0.3796409673988819,
"learning_rate": 4.800000000000001e-06,
"loss": -0.05525785088539124,
"num_tokens": 308812.0,
"reward": 0.0569999985396862,
"reward_std": 0.006000000238418579,
"rewards/reward_func/mean": 0.0569999985396862,
"rewards/reward_func/std": 0.005999999865889549,
"step": 105,
"step_time": 20.08870074040169,
"tools/call_frequency": 6.3,
"tools/failure_frequency": 0.3948905646800995
},
{
"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": 0.65,
"completions/max_length": 342.4,
"completions/max_terminated_length": 329.0,
"completions/mean_length": 324.9,
"completions/mean_terminated_length": 314.3,
"completions/min_length": 294.4,
"completions/min_terminated_length": 299.6,
"entropy": 0.027399512368720024,
"epoch": 0.4296875,
"frac_reward_zero_std": 0.6,
"grad_norm": 0.04786663129925728,
"kl": 0.35288467034697535,
"learning_rate": 4.5500000000000005e-06,
"loss": -0.010480480641126633,
"num_tokens": 324626.0,
"reward": 0.05399999842047691,
"reward_std": 0.012000000476837159,
"rewards/reward_func/mean": 0.05399999842047691,
"rewards/reward_func/std": 0.011999999731779098,
"step": 110,
"step_time": 20.751387215401337,
"tools/call_frequency": 6.05,
"tools/failure_frequency": 0.3629230856895447
},
{
"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": 0.65,
"completions/max_length": 335.0,
"completions/max_terminated_length": 250.6,
"completions/mean_length": 324.05,
"completions/mean_terminated_length": 247.85,
"completions/min_length": 306.4,
"completions/min_terminated_length": 244.4,
"entropy": 0.011146409685898107,
"epoch": 0.44921875,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.02133430540561676,
"kl": 0.31119330078363416,
"learning_rate": 4.3e-06,
"loss": 0.012449412047863007,
"num_tokens": 340312.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 115,
"step_time": 19.889900923999086,
"tools/call_frequency": 6.7,
"tools/failure_frequency": 0.364724200963974
},
{
"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": 0.7,
"completions/max_length": 338.4,
"completions/max_terminated_length": 254.8,
"completions/mean_length": 326.0,
"completions/mean_terminated_length": 253.13333740234376,
"completions/min_length": 310.8,
"completions/min_terminated_length": 250.0,
"entropy": 0.0049434174259658905,
"epoch": 0.46875,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.07528958469629288,
"kl": 0.2956920139491558,
"learning_rate": 4.05e-06,
"loss": 0.011839952319860458,
"num_tokens": 356100.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 120,
"step_time": 20.338967355799106,
"tools/call_frequency": 6.6,
"tools/failure_frequency": 0.37093899846076966
},
{
"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": 0.45,
"completions/max_length": 328.4,
"completions/max_terminated_length": 266.6,
"completions/mean_length": 292.65,
"completions/mean_terminated_length": 243.46666870117187,
"completions/min_length": 212.6,
"completions/min_terminated_length": 220.8,
"entropy": 0.010821938829030842,
"epoch": 0.48828125,
"frac_reward_zero_std": 0.6,
"grad_norm": 0.02426336519420147,
"kl": 0.3538802661001682,
"learning_rate": 3.8000000000000005e-06,
"loss": -0.12166147232055664,
"num_tokens": 371141.0,
"reward": 0.05399999842047691,
"reward_std": 0.012000000476837159,
"rewards/reward_func/mean": 0.05399999842047691,
"rewards/reward_func/std": 0.011999999731779098,
"step": 125,
"step_time": 19.236361916999158,
"tools/call_frequency": 6.5,
"tools/failure_frequency": 0.3801395773887634
},
{
"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": 0.15,
"completions/max_length": 323.4,
"completions/max_terminated_length": 316.6,
"completions/mean_length": 311.1,
"completions/mean_terminated_length": 309.3333374023438,
"completions/min_length": 301.6,
"completions/min_terminated_length": 302.8,
"entropy": 0.003383969687274657,
"epoch": 0.5078125,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.025379447266459465,
"kl": 0.28744339048862455,
"learning_rate": 3.5500000000000003e-06,
"loss": 0.011490081995725631,
"num_tokens": 386449.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 130,
"step_time": 17.496410710996862,
"tools/call_frequency": 7.45,
"tools/failure_frequency": 0.3971812605857849
},
{
"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": 0.5,
"completions/max_length": 336.2,
"completions/max_terminated_length": 320.8,
"completions/mean_length": 319.85,
"completions/mean_terminated_length": 315.1333374023437,
"completions/min_length": 303.8,
"completions/min_terminated_length": 310.4,
"entropy": 0.0029613528924528508,
"epoch": 0.52734375,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.029231328517198563,
"kl": 0.28883620277047156,
"learning_rate": 3.3000000000000006e-06,
"loss": 0.011552013456821442,
"num_tokens": 402092.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 135,
"step_time": 19.767896748798375,
"tools/call_frequency": 6.95,
"tools/failure_frequency": 0.3816740334033966
},
{
"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": 0.5,
"completions/max_length": 336.8,
"completions/max_terminated_length": 303.4,
"completions/mean_length": 314.65,
"completions/mean_terminated_length": 297.0666748046875,
"completions/min_length": 291.8,
"completions/min_terminated_length": 291.8,
"entropy": 0.0022313889967335854,
"epoch": 0.546875,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.006076246965676546,
"kl": 0.2769774109125137,
"learning_rate": 3.05e-06,
"loss": 0.011077215522527694,
"num_tokens": 417523.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 140,
"step_time": 19.793050527399465,
"tools/call_frequency": 7.05,
"tools/failure_frequency": 0.36877211928367615
},
{
"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": 0.35,
"completions/max_length": 334.8,
"completions/max_terminated_length": 321.0,
"completions/mean_length": 317.05,
"completions/mean_terminated_length": 311.78333740234376,
"completions/min_length": 299.8,
"completions/min_terminated_length": 302.2,
"entropy": 0.0023406802676618097,
"epoch": 0.56640625,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.016859112307429314,
"kl": 0.27094974666833876,
"learning_rate": 2.8000000000000003e-06,
"loss": 0.010821019113063813,
"num_tokens": 433028.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 145,
"step_time": 19.179231669600995,
"tools/call_frequency": 7.1,
"tools/failure_frequency": 0.3943915367126465
},
{
"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": 0.4,
"completions/max_length": 333.4,
"completions/max_terminated_length": 324.2,
"completions/mean_length": 321.25,
"completions/mean_terminated_length": 315.7,
"completions/min_length": 299.6,
"completions/min_terminated_length": 304.6,
"entropy": 0.002267455725814216,
"epoch": 0.5859375,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.026256700977683067,
"kl": 0.27512960955500604,
"learning_rate": 2.55e-06,
"loss": 0.010997948795557022,
"num_tokens": 448676.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 150,
"step_time": 18.90324738700001,
"tools/call_frequency": 6.9,
"tools/failure_frequency": 0.39089863300323485
},
{
"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": 0.55,
"completions/max_length": 336.4,
"completions/max_terminated_length": 320.8,
"completions/mean_length": 322.25,
"completions/mean_terminated_length": 316.3666687011719,
"completions/min_length": 304.2,
"completions/min_terminated_length": 311.8,
"entropy": 0.001911457245296333,
"epoch": 0.60546875,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.03210851177573204,
"kl": 0.2622846975922585,
"learning_rate": 2.3000000000000004e-06,
"loss": 0.010500120371580124,
"num_tokens": 464367.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 155,
"step_time": 19.15542162460188,
"tools/call_frequency": 6.8,
"tools/failure_frequency": 0.38988603949546813
},
{
"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": 0.6,
"completions/max_length": 333.0,
"completions/max_terminated_length": 253.4,
"completions/mean_length": 318.85,
"completions/mean_terminated_length": 250.6,
"completions/min_length": 307.4,
"completions/min_terminated_length": 247.0,
"entropy": 0.003048306703567505,
"epoch": 0.625,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.03355943784117699,
"kl": 0.2786222040653229,
"learning_rate": 2.05e-06,
"loss": 0.011140207946300506,
"num_tokens": 479978.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 160,
"step_time": 19.03406370099983,
"tools/call_frequency": 6.85,
"tools/failure_frequency": 0.4012160897254944
},
{
"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": 0.4,
"completions/max_length": 337.4,
"completions/max_terminated_length": 318.8,
"completions/mean_length": 320.15,
"completions/mean_terminated_length": 311.73333740234375,
"completions/min_length": 303.2,
"completions/min_terminated_length": 304.6,
"entropy": 0.0024299834709381684,
"epoch": 0.64453125,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.00971182156354189,
"kl": 0.26761081293225286,
"learning_rate": 1.8000000000000001e-06,
"loss": 0.010710306465625763,
"num_tokens": 495637.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 165,
"step_time": 19.779936843597532,
"tools/call_frequency": 6.9,
"tools/failure_frequency": 0.39147052764892576
},
{
"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": 0.45,
"completions/max_length": 335.0,
"completions/max_terminated_length": 317.4,
"completions/mean_length": 320.7,
"completions/mean_terminated_length": 313.7166687011719,
"completions/min_length": 303.6,
"completions/min_terminated_length": 308.6,
"entropy": 0.0015991570777259766,
"epoch": 0.6640625,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.0209937933832407,
"kl": 0.2613979808986187,
"learning_rate": 1.5500000000000002e-06,
"loss": 0.010462091118097306,
"num_tokens": 511254.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 170,
"step_time": 18.80054657560031,
"tools/call_frequency": 7.0,
"tools/failure_frequency": 0.3783483743667603
},
{
"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": 0.6,
"completions/max_length": 336.0,
"completions/max_terminated_length": 325.4,
"completions/mean_length": 321.4,
"completions/mean_terminated_length": 317.1,
"completions/min_length": 298.4,
"completions/min_terminated_length": 308.8,
"entropy": 0.0019770772167248653,
"epoch": 0.68359375,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.030799515545368195,
"kl": 0.2681220337748528,
"learning_rate": 1.3e-06,
"loss": 0.01071011796593666,
"num_tokens": 526902.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 175,
"step_time": 20.021315099197817,
"tools/call_frequency": 6.75,
"tools/failure_frequency": 0.370084285736084
},
{
"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": 0.4,
"completions/max_length": 332.4,
"completions/max_terminated_length": 318.2,
"completions/mean_length": 317.5,
"completions/mean_terminated_length": 311.3333374023438,
"completions/min_length": 299.8,
"completions/min_terminated_length": 304.4,
"entropy": 0.0017659664415987208,
"epoch": 0.703125,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.012540716677904129,
"kl": 0.26256203353405,
"learning_rate": 1.0500000000000001e-06,
"loss": 0.010497495532035828,
"num_tokens": 542478.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 180,
"step_time": 19.730400246601494,
"tools/call_frequency": 6.95,
"tools/failure_frequency": 0.38820105195045473
},
{
"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": 0.55,
"completions/max_length": 337.2,
"completions/max_terminated_length": 259.8,
"completions/mean_length": 323.55,
"completions/mean_terminated_length": 254.73333740234375,
"completions/min_length": 306.2,
"completions/min_terminated_length": 249.4,
"entropy": 0.001894339967839187,
"epoch": 0.72265625,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.019277706742286682,
"kl": 0.25229193195700644,
"learning_rate": 8.000000000000001e-07,
"loss": 0.010086478292942047,
"num_tokens": 558253.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 185,
"step_time": 19.658330725000997,
"tools/call_frequency": 6.7,
"tools/failure_frequency": 0.3805657267570496
},
{
"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": 0.55,
"completions/max_length": 336.0,
"completions/max_terminated_length": 317.4,
"completions/mean_length": 322.9,
"completions/mean_terminated_length": 313.75,
"completions/min_length": 300.0,
"completions/min_terminated_length": 309.2,
"entropy": 0.0026780343818245456,
"epoch": 0.7421875,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.01523889135569334,
"kl": 0.26018148586153983,
"learning_rate": 5.5e-07,
"loss": 0.010390447080135345,
"num_tokens": 573937.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 190,
"step_time": 19.327826142799314,
"tools/call_frequency": 6.7,
"tools/failure_frequency": 0.3721375405788422
},
{
"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": 0.55,
"completions/max_length": 336.2,
"completions/max_terminated_length": 315.6,
"completions/mean_length": 320.45,
"completions/mean_terminated_length": 308.2,
"completions/min_length": 297.2,
"completions/min_terminated_length": 300.6,
"entropy": 0.0018196408695075662,
"epoch": 0.76171875,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.010020147077739239,
"kl": 0.28274188563227654,
"learning_rate": 3.0000000000000004e-07,
"loss": 0.011302116513252258,
"num_tokens": 589569.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 195,
"step_time": 19.777142194598856,
"tools/call_frequency": 6.85,
"tools/failure_frequency": 0.3721937298774719
},
{
"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": 0.55,
"completions/max_length": 336.2,
"completions/max_terminated_length": 319.8,
"completions/mean_length": 320.25,
"completions/mean_terminated_length": 315.95,
"completions/min_length": 299.6,
"completions/min_terminated_length": 313.0,
"entropy": 0.0022158807900268585,
"epoch": 0.78125,
"frac_reward_zero_std": 1.0,
"grad_norm": 0.022795701399445534,
"kl": 0.2784128800034523,
"learning_rate": 5.0000000000000004e-08,
"loss": 0.011128613352775573,
"num_tokens": 605249.0,
"reward": 0.05999999865889549,
"reward_std": 0.0,
"rewards/reward_func/mean": 0.05999999865889549,
"rewards/reward_func/std": 0.0,
"step": 200,
"step_time": 19.714544834000115,
"tools/call_frequency": 6.95,
"tools/failure_frequency": 0.38814000487327577
}
],
"logging_steps": 5,
"max_steps": 200,
"num_input_tokens_seen": 605249,
"num_train_epochs": 1,
"save_steps": 100,
"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": 1,
"trial_name": null,
"trial_params": null
}