{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.390625, "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": 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 } ], "logging_steps": 5, "max_steps": 200, "num_input_tokens_seen": 293245, "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": false }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }