{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.26666666666666666, "eval_steps": 50, "global_step": 50, "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": 331.125, "completions/max_terminated_length": 331.125, "completions/mean_length": 288.96875, "completions/mean_terminated_length": 288.96875, "completions/min_length": 249.125, "completions/min_terminated_length": 249.125, "entropy": 0.14002850465476513, "epoch": 0.005333333333333333, "frac_reward_zero_std": 0.25, "grad_norm": 12.074647903442383, "learning_rate": 1e-06, "loss": -0.105, "num_tokens": 131934.0, "reward": 1.4322916865348816, "reward_std": 0.17777812760323286, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.43229166930541396, "rewards/frames_in_gt_reward/std": 0.17777812760323286, "step": 1, "step_time": 401.7192056638887 }, { "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": 356.5, "completions/max_terminated_length": 356.5, "completions/mean_length": 285.96875, "completions/mean_terminated_length": 285.96875, "completions/min_length": 217.625, "completions/min_terminated_length": 217.625, "entropy": 0.14247188344597816, "epoch": 0.010666666666666666, "frac_reward_zero_std": 0.5, "grad_norm": 8.800053596496582, "learning_rate": 9.973404255319148e-07, "loss": 0.1199, "num_tokens": 264676.0, "reward": 1.2343750149011612, "reward_std": 0.14060229063034058, "rewards/consistency_reward/mean": 0.9375, "rewards/consistency_reward/std": 0.1157275140285492, "rewards/frames_in_gt_reward/mean": 0.2968750027939677, "rewards/frames_in_gt_reward/std": 0.09036078676581383, "step": 2, "step_time": 240.82988019194454 }, { "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": 0.046875, "completions/max_length": 361.375, "completions/max_terminated_length": 340.875, "completions/mean_length": 297.40625, "completions/mean_terminated_length": 287.81562423706055, "completions/min_length": 223.75, "completions/min_terminated_length": 223.75, "entropy": 0.12379888165742159, "epoch": 0.016, "frac_reward_zero_std": 0.25, "grad_norm": 14.743064880371094, "learning_rate": 9.946808510638297e-07, "loss": 0.3748, "num_tokens": 397534.0, "reward": 1.4088541716337204, "reward_std": 0.26582484506070614, "rewards/consistency_reward/mean": 0.890625, "rewards/consistency_reward/std": 0.12387890368700027, "rewards/frames_in_gt_reward/mean": 0.5182291716337204, "rewards/frames_in_gt_reward/std": 0.16439844015985727, "step": 3, "step_time": 235.56323385995347 }, { "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": 0.015625, "completions/max_length": 373.0, "completions/max_terminated_length": 348.375, "completions/mean_length": 280.75, "completions/mean_terminated_length": 276.5870552062988, "completions/min_length": 201.125, "completions/min_terminated_length": 201.125, "entropy": 0.15426983498036861, "epoch": 0.021333333333333333, "frac_reward_zero_std": 0.125, "grad_norm": 14.374115943908691, "learning_rate": 9.920212765957446e-07, "loss": 0.7572, "num_tokens": 529638.0, "reward": 1.2890625, "reward_std": 0.2656492292881012, "rewards/consistency_reward/mean": 0.96875, "rewards/consistency_reward/std": 0.0883883461356163, "rewards/frames_in_gt_reward/mean": 0.3203125, "rewards/frames_in_gt_reward/std": 0.1772608980536461, "step": 4, "step_time": 210.90831525682006 }, { "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": 338.25, "completions/max_terminated_length": 338.25, "completions/mean_length": 280.9375, "completions/mean_terminated_length": 280.9375, "completions/min_length": 224.0, "completions/min_terminated_length": 224.0, "entropy": 0.14072747062891722, "epoch": 0.02666666666666667, "frac_reward_zero_std": 0.375, "grad_norm": 7.985368251800537, "learning_rate": 9.893617021276595e-07, "loss": 0.1124, "num_tokens": 661450.0, "reward": 1.1953125, "reward_std": 0.14211230352520943, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.1953125, "rewards/frames_in_gt_reward/std": 0.14211230352520943, "step": 5, "step_time": 198.13520157209132 }, { "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": 310.0, "completions/max_terminated_length": 310.0, "completions/mean_length": 256.234375, "completions/mean_terminated_length": 256.234375, "completions/min_length": 192.125, "completions/min_terminated_length": 192.125, "entropy": 0.15063040610402822, "epoch": 0.032, "frac_reward_zero_std": 0.5, "grad_norm": 10.914612770080566, "learning_rate": 9.867021276595744e-07, "loss": -0.0601, "num_tokens": 791385.0, "reward": 1.4821428507566452, "reward_std": 0.08940844610333443, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.4821428582072258, "rewards/frames_in_gt_reward/std": 0.08940845355391502, "step": 6, "step_time": 176.5121703321347 }, { "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": 325.0, "completions/max_terminated_length": 325.0, "completions/mean_length": 263.296875, "completions/mean_terminated_length": 263.296875, "completions/min_length": 215.0, "completions/min_terminated_length": 215.0, "entropy": 0.14749469235539436, "epoch": 0.037333333333333336, "frac_reward_zero_std": 0.25, "grad_norm": 11.562383651733398, "learning_rate": 9.840425531914893e-07, "loss": 0.1772, "num_tokens": 921996.0, "reward": 1.2734375, "reward_std": 0.17820549197494984, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.2734375, "rewards/frames_in_gt_reward/std": 0.17820549570024014, "step": 7, "step_time": 332.42883461911697 }, { "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": 308.0, "completions/max_terminated_length": 308.0, "completions/mean_length": 253.921875, "completions/mean_terminated_length": 253.921875, "completions/min_length": 217.125, "completions/min_terminated_length": 217.125, "entropy": 0.1459873951971531, "epoch": 0.042666666666666665, "frac_reward_zero_std": 0.375, "grad_norm": 13.197957038879395, "learning_rate": 9.813829787234042e-07, "loss": -0.5726, "num_tokens": 1052039.0, "reward": 1.2802083492279053, "reward_std": 0.12584944535046816, "rewards/consistency_reward/mean": 0.984375, "rewards/consistency_reward/std": 0.04419417306780815, "rewards/frames_in_gt_reward/mean": 0.2958333343267441, "rewards/frames_in_gt_reward/std": 0.11169478669762611, "step": 8, "step_time": 197.40488314104732 }, { "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": 324.125, "completions/max_terminated_length": 324.125, "completions/mean_length": 271.640625, "completions/mean_terminated_length": 271.640625, "completions/min_length": 201.375, "completions/min_terminated_length": 201.375, "entropy": 0.11871037073433399, "epoch": 0.048, "frac_reward_zero_std": 0.125, "grad_norm": 12.982590675354004, "learning_rate": 9.78723404255319e-07, "loss": 0.1239, "num_tokens": 1183376.0, "reward": 1.2401041835546494, "reward_std": 0.17192464135587215, "rewards/consistency_reward/mean": 0.984375, "rewards/consistency_reward/std": 0.04419417306780815, "rewards/frames_in_gt_reward/mean": 0.255729166790843, "rewards/frames_in_gt_reward/std": 0.1529034459963441, "step": 9, "step_time": 177.05780373478774 }, { "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": 320.375, "completions/max_terminated_length": 320.375, "completions/mean_length": 275.140625, "completions/mean_terminated_length": 275.140625, "completions/min_length": 207.5, "completions/min_terminated_length": 207.5, "entropy": 0.14815808087587357, "epoch": 0.05333333333333334, "frac_reward_zero_std": 0.5, "grad_norm": 12.542384147644043, "learning_rate": 9.76063829787234e-07, "loss": 0.0202, "num_tokens": 1316441.0, "reward": 1.1770833432674408, "reward_std": 0.07513886131346226, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.1770833358168602, "rewards/frames_in_gt_reward/std": 0.07513886503875256, "step": 10, "step_time": 202.85511767759454 }, { "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": 329.25, "completions/max_terminated_length": 329.25, "completions/mean_length": 273.421875, "completions/mean_terminated_length": 273.421875, "completions/min_length": 227.125, "completions/min_terminated_length": 227.125, "entropy": 0.1386905750259757, "epoch": 0.058666666666666666, "frac_reward_zero_std": 0.25, "grad_norm": 33.540565490722656, "learning_rate": 9.73404255319149e-07, "loss": 0.0765, "num_tokens": 1449076.0, "reward": 1.3359375149011612, "reward_std": 0.15955893136560917, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.33593750558793545, "rewards/frames_in_gt_reward/std": 0.15955893881618977, "step": 11, "step_time": 207.8940927479416 }, { "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": 338.875, "completions/max_terminated_length": 338.875, "completions/mean_length": 297.65625, "completions/mean_terminated_length": 297.65625, "completions/min_length": 251.875, "completions/min_terminated_length": 251.875, "entropy": 0.10804447624832392, "epoch": 0.064, "frac_reward_zero_std": 0.375, "grad_norm": 8.172316551208496, "learning_rate": 9.707446808510637e-07, "loss": 0.1203, "num_tokens": 1582126.0, "reward": 1.1523437649011612, "reward_std": 0.11345583852380514, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.15234375186264515, "rewards/frames_in_gt_reward/std": 0.11345584597438574, "step": 12, "step_time": 225.7350684221601 }, { "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": 338.625, "completions/max_terminated_length": 338.625, "completions/mean_length": 273.625, "completions/mean_terminated_length": 273.625, "completions/min_length": 215.375, "completions/min_terminated_length": 215.375, "entropy": 0.1306639350950718, "epoch": 0.06933333333333333, "frac_reward_zero_std": 0.125, "grad_norm": 13.5219144821167, "learning_rate": 9.680851063829786e-07, "loss": 0.0027, "num_tokens": 1714486.0, "reward": 1.2018229216337204, "reward_std": 0.2505111303180456, "rewards/consistency_reward/mean": 0.9375, "rewards/consistency_reward/std": 0.06681530922651291, "rewards/frames_in_gt_reward/mean": 0.2643229179084301, "rewards/frames_in_gt_reward/std": 0.18369582667946815, "step": 13, "step_time": 215.64708924212027 }, { "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": 0.078125, "completions/max_length": 316.0, "completions/max_terminated_length": 294.75, "completions/mean_length": 271.296875, "completions/mean_terminated_length": 253.328125, "completions/min_length": 208.375, "completions/min_terminated_length": 208.375, "entropy": 0.14128621108829975, "epoch": 0.07466666666666667, "frac_reward_zero_std": 0.625, "grad_norm": 6.406787872314453, "learning_rate": 9.654255319148937e-07, "loss": 0.3658, "num_tokens": 1846489.0, "reward": 1.113281261175871, "reward_std": 0.2654474452137947, "rewards/consistency_reward/mean": 0.828125, "rewards/consistency_reward/std": 0.17358147725462914, "rewards/frames_in_gt_reward/mean": 0.2851562537252903, "rewards/frames_in_gt_reward/std": 0.10744580812752247, "step": 14, "step_time": 300.93962029227987 }, { "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": 0.03125, "completions/max_length": 326.875, "completions/max_terminated_length": 313.375, "completions/mean_length": 262.5625, "completions/mean_terminated_length": 257.109375, "completions/min_length": 213.125, "completions/min_terminated_length": 213.125, "entropy": 0.1355037186294794, "epoch": 0.08, "frac_reward_zero_std": 0.625, "grad_norm": 7.533238410949707, "learning_rate": 9.627659574468084e-07, "loss": 0.0361, "num_tokens": 1977213.0, "reward": 1.3450520932674408, "reward_std": 0.17334979306906462, "rewards/consistency_reward/mean": 0.9375, "rewards/consistency_reward/std": 0.1157275140285492, "rewards/frames_in_gt_reward/mean": 0.4075520858168602, "rewards/frames_in_gt_reward/std": 0.07232724782079458, "step": 15, "step_time": 216.9358362380881 }, { "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": 301.625, "completions/max_terminated_length": 301.625, "completions/mean_length": 254.375, "completions/mean_terminated_length": 254.375, "completions/min_length": 216.0, "completions/min_terminated_length": 216.0, "entropy": 0.11875399947166443, "epoch": 0.08533333333333333, "frac_reward_zero_std": 0.875, "grad_norm": 3.654115915298462, "learning_rate": 9.601063829787233e-07, "loss": 0.0342, "num_tokens": 2106493.0, "reward": 1.08984375, "reward_std": 0.007232969626784325, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.08984375, "rewards/frames_in_gt_reward/std": 0.007232969626784325, "step": 16, "step_time": 210.4247310848441 }, { "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": 325.125, "completions/max_terminated_length": 325.125, "completions/mean_length": 273.28125, "completions/mean_terminated_length": 273.28125, "completions/min_length": 211.25, "completions/min_terminated_length": 211.25, "entropy": 0.10801671631634235, "epoch": 0.09066666666666667, "frac_reward_zero_std": 0.375, "grad_norm": 10.859893798828125, "learning_rate": 9.574468085106384e-07, "loss": 0.0553, "num_tokens": 2237527.0, "reward": 1.3191964477300644, "reward_std": 0.11243656650185585, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.3191964281722903, "rewards/frames_in_gt_reward/std": 0.11243656743317842, "step": 17, "step_time": 188.87649188411888 }, { "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": 319.0, "completions/max_terminated_length": 319.0, "completions/mean_length": 263.25, "completions/mean_terminated_length": 263.25, "completions/min_length": 199.125, "completions/min_terminated_length": 199.125, "entropy": 0.12452559638768435, "epoch": 0.096, "frac_reward_zero_std": 0.375, "grad_norm": 11.082453727722168, "learning_rate": 9.54787234042553e-07, "loss": -0.0828, "num_tokens": 2368031.0, "reward": 1.2994791716337204, "reward_std": 0.13045400893315673, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.2994791716337204, "rewards/frames_in_gt_reward/std": 0.13045401126146317, "step": 18, "step_time": 230.66768762213178 }, { "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": 298.375, "completions/max_terminated_length": 298.375, "completions/mean_length": 268.125, "completions/mean_terminated_length": 268.125, "completions/min_length": 245.625, "completions/min_terminated_length": 245.625, "entropy": 0.11268881615251303, "epoch": 0.10133333333333333, "frac_reward_zero_std": 0.625, "grad_norm": 8.177054405212402, "learning_rate": 9.52127659574468e-07, "loss": 0.0458, "num_tokens": 2498807.0, "reward": 1.348437488079071, "reward_std": 0.06134095508605242, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.34843750298023224, "rewards/frames_in_gt_reward/std": 0.061340964399278164, "step": 19, "step_time": 214.88793752901256 }, { "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": 288.75, "completions/max_terminated_length": 288.75, "completions/mean_length": 244.265625, "completions/mean_terminated_length": 244.265625, "completions/min_length": 208.375, "completions/min_terminated_length": 208.375, "entropy": 0.12906462606042624, "epoch": 0.10666666666666667, "frac_reward_zero_std": 0.625, "grad_norm": 12.999517440795898, "learning_rate": 9.49468085106383e-07, "loss": 0.0627, "num_tokens": 2627448.0, "reward": 1.2526041716337204, "reward_std": 0.052546086721122265, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.2526041679084301, "rewards/frames_in_gt_reward/std": 0.052546086721122265, "step": 20, "step_time": 173.11658866901416 }, { "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": 301.75, "completions/max_terminated_length": 301.75, "completions/mean_length": 249.078125, "completions/mean_terminated_length": 249.078125, "completions/min_length": 188.25, "completions/min_terminated_length": 188.25, "entropy": 0.1279576886445284, "epoch": 0.112, "frac_reward_zero_std": 0.5, "grad_norm": 11.39144515991211, "learning_rate": 9.468085106382978e-07, "loss": -0.1775, "num_tokens": 2758277.0, "reward": 1.4029017835855484, "reward_std": 0.09685657173395157, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.4029017873108387, "rewards/frames_in_gt_reward/std": 0.09685657918453217, "step": 21, "step_time": 176.74677460396197 }, { "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": 287.5, "completions/max_terminated_length": 287.5, "completions/mean_length": 238.875, "completions/mean_terminated_length": 238.875, "completions/min_length": 148.25, "completions/min_terminated_length": 148.25, "entropy": 0.11993568483740091, "epoch": 0.11733333333333333, "frac_reward_zero_std": 0.75, "grad_norm": 6.358952522277832, "learning_rate": 9.441489361702127e-07, "loss": 0.1801, "num_tokens": 2886469.0, "reward": 1.1953125, "reward_std": 0.04436885751783848, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.1953125, "rewards/frames_in_gt_reward/std": 0.04436885938048363, "step": 22, "step_time": 200.77307505102362 }, { "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": 310.25, "completions/max_terminated_length": 310.25, "completions/mean_length": 251.25, "completions/mean_terminated_length": 251.25, "completions/min_length": 185.5, "completions/min_terminated_length": 185.5, "entropy": 0.10936911404132843, "epoch": 0.12266666666666666, "frac_reward_zero_std": 0.375, "grad_norm": 10.427143096923828, "learning_rate": 9.414893617021277e-07, "loss": 0.0662, "num_tokens": 3016549.0, "reward": 1.5559895932674408, "reward_std": 0.12049377709627151, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.5559895858168602, "rewards/frames_in_gt_reward/std": 0.12049377895891666, "step": 23, "step_time": 183.04031058377586 }, { "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": 302.75, "completions/max_terminated_length": 302.75, "completions/mean_length": 261.65625, "completions/mean_terminated_length": 261.65625, "completions/min_length": 233.125, "completions/min_terminated_length": 233.125, "entropy": 0.10936919413506985, "epoch": 0.128, "frac_reward_zero_std": 0.75, "grad_norm": 6.140288829803467, "learning_rate": 9.388297872340425e-07, "loss": 0.2902, "num_tokens": 3147463.0, "reward": 1.397656261920929, "reward_std": 0.08078660443425179, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.39765625447034836, "rewards/frames_in_gt_reward/std": 0.08078660443425179, "step": 24, "step_time": 209.9473960272735 }, { "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": 304.875, "completions/max_terminated_length": 304.875, "completions/mean_length": 262.15625, "completions/mean_terminated_length": 262.15625, "completions/min_length": 212.5, "completions/min_terminated_length": 212.5, "entropy": 0.11673647444695234, "epoch": 0.13333333333333333, "frac_reward_zero_std": 0.5, "grad_norm": 7.445509433746338, "learning_rate": 9.361702127659575e-07, "loss": -0.0854, "num_tokens": 3278265.0, "reward": 1.309895858168602, "reward_std": 0.10223261453211308, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.30989583767950535, "rewards/frames_in_gt_reward/std": 0.10223261639475822, "step": 25, "step_time": 163.6448761896463 }, { "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": 284.875, "completions/max_terminated_length": 284.875, "completions/mean_length": 247.59375, "completions/mean_terminated_length": 247.59375, "completions/min_length": 216.875, "completions/min_terminated_length": 216.875, "entropy": 0.08394424710422754, "epoch": 0.13866666666666666, "frac_reward_zero_std": 0.375, "grad_norm": 10.775681495666504, "learning_rate": 9.335106382978722e-07, "loss": -0.3014, "num_tokens": 3408119.0, "reward": 1.2987351268529892, "reward_std": 0.1820310577750206, "rewards/consistency_reward/mean": 0.9375, "rewards/consistency_reward/std": 0.06681530922651291, "rewards/frames_in_gt_reward/mean": 0.3612351231276989, "rewards/frames_in_gt_reward/std": 0.13430584128946066, "step": 26, "step_time": 184.15799631096888 }, { "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": 300.5, "completions/max_terminated_length": 300.5, "completions/mean_length": 261.6875, "completions/mean_terminated_length": 261.6875, "completions/min_length": 224.125, "completions/min_terminated_length": 224.125, "entropy": 0.11061851400882006, "epoch": 0.144, "frac_reward_zero_std": 0.75, "grad_norm": 5.968808650970459, "learning_rate": 9.308510638297871e-07, "loss": 0.0682, "num_tokens": 3538219.0, "reward": 1.3489583432674408, "reward_std": 0.04813904967159033, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.34895833348855376, "rewards/frames_in_gt_reward/std": 0.04813904780894518, "step": 27, "step_time": 195.55628155276645 }, { "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": 296.875, "completions/max_terminated_length": 296.875, "completions/mean_length": 230.59375, "completions/mean_terminated_length": 230.59375, "completions/min_length": 180.625, "completions/min_terminated_length": 180.625, "entropy": 0.1191996717825532, "epoch": 0.14933333333333335, "frac_reward_zero_std": 0.875, "grad_norm": 15.380400657653809, "learning_rate": 9.281914893617021e-07, "loss": -0.1266, "num_tokens": 3666281.0, "reward": 1.3600260466337204, "reward_std": 0.008086705580353737, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.3600260429084301, "rewards/frames_in_gt_reward/std": 0.008086706511676311, "step": 28, "step_time": 209.72345635097008 }, { "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": 272.625, "completions/max_terminated_length": 272.625, "completions/mean_length": 232.65625, "completions/mean_terminated_length": 232.65625, "completions/min_length": 204.125, "completions/min_terminated_length": 204.125, "entropy": 0.10214738175272942, "epoch": 0.15466666666666667, "frac_reward_zero_std": 0.375, "grad_norm": 12.323505401611328, "learning_rate": 9.255319148936169e-07, "loss": 0.1631, "num_tokens": 3794819.0, "reward": 1.3828125149011612, "reward_std": 0.11083479970693588, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.38281250186264515, "rewards/frames_in_gt_reward/std": 0.11083480529487133, "step": 29, "step_time": 161.68485497008078 }, { "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": 306.75, "completions/max_terminated_length": 306.75, "completions/mean_length": 265.734375, "completions/mean_terminated_length": 265.734375, "completions/min_length": 234.25, "completions/min_terminated_length": 234.25, "entropy": 0.10308584477752447, "epoch": 0.16, "frac_reward_zero_std": 0.625, "grad_norm": 6.997777462005615, "learning_rate": 9.228723404255319e-07, "loss": 0.1127, "num_tokens": 3926154.0, "reward": 1.2005208283662796, "reward_std": 0.07312181312590837, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.20052083395421505, "rewards/frames_in_gt_reward/std": 0.07312180753797293, "step": 30, "step_time": 176.4764612171566 }, { "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": 0.046875, "completions/max_length": 314.375, "completions/max_terminated_length": 313.5, "completions/mean_length": 266.953125, "completions/mean_terminated_length": 257.00625228881836, "completions/min_length": 186.75, "completions/min_terminated_length": 186.75, "entropy": 0.10218324046581984, "epoch": 0.16533333333333333, "frac_reward_zero_std": 0.375, "grad_norm": 10.449390411376953, "learning_rate": 9.202127659574468e-07, "loss": 0.1191, "num_tokens": 4057063.0, "reward": 1.1932291723787785, "reward_std": 0.250632893294096, "rewards/consistency_reward/mean": 0.90625, "rewards/consistency_reward/std": 0.12938730418682098, "rewards/frames_in_gt_reward/mean": 0.28697916865348816, "rewards/frames_in_gt_reward/std": 0.12315054796636105, "step": 31, "step_time": 184.1667816147674 }, { "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": 268.375, "completions/max_terminated_length": 268.375, "completions/mean_length": 246.203125, "completions/mean_terminated_length": 246.203125, "completions/min_length": 228.25, "completions/min_terminated_length": 228.25, "entropy": 0.09946651849895716, "epoch": 0.17066666666666666, "frac_reward_zero_std": 0.625, "grad_norm": 8.795154571533203, "learning_rate": 9.175531914893616e-07, "loss": 0.0528, "num_tokens": 4187380.0, "reward": 1.3828125, "reward_std": 0.0942012406885624, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.3828125, "rewards/frames_in_gt_reward/std": 0.09420124441385269, "step": 32, "step_time": 171.93527664185967 }, { "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": 293.875, "completions/max_terminated_length": 293.875, "completions/mean_length": 244.203125, "completions/mean_terminated_length": 244.203125, "completions/min_length": 187.875, "completions/min_terminated_length": 187.875, "entropy": 0.10002793837338686, "epoch": 0.176, "frac_reward_zero_std": 0.25, "grad_norm": 12.763372421264648, "learning_rate": 9.148936170212766e-07, "loss": 0.82, "num_tokens": 4316961.0, "reward": 1.3802083432674408, "reward_std": 0.15811070054769516, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.3802083358168602, "rewards/frames_in_gt_reward/std": 0.15811070799827576, "step": 33, "step_time": 169.94250934314914 }, { "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": 239.5, "completions/max_terminated_length": 239.5, "completions/mean_length": 216.78125, "completions/mean_terminated_length": 216.78125, "completions/min_length": 186.375, "completions/min_terminated_length": 186.375, "entropy": 0.1106764068827033, "epoch": 0.18133333333333335, "frac_reward_zero_std": 0.5, "grad_norm": 11.295879364013672, "learning_rate": 9.122340425531915e-07, "loss": -0.0516, "num_tokens": 4444443.0, "reward": 1.2630208432674408, "reward_std": 0.08294470608234406, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.2630208386108279, "rewards/frames_in_gt_reward/std": 0.08294469863176346, "step": 34, "step_time": 174.51075469097123 }, { "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": 292.5, "completions/max_terminated_length": 292.5, "completions/mean_length": 247.09375, "completions/mean_terminated_length": 247.09375, "completions/min_length": 217.5, "completions/min_terminated_length": 217.5, "entropy": 0.10420085256919265, "epoch": 0.18666666666666668, "frac_reward_zero_std": 0.375, "grad_norm": 10.102306365966797, "learning_rate": 9.095744680851063e-07, "loss": 0.1036, "num_tokens": 4572457.0, "reward": 1.5052083432674408, "reward_std": 0.13959199748933315, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.505208333954215, "rewards/frames_in_gt_reward/std": 0.13959199748933315, "step": 35, "step_time": 211.78892861900385 }, { "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": 287.0, "completions/max_terminated_length": 287.0, "completions/mean_length": 258.96875, "completions/mean_terminated_length": 258.96875, "completions/min_length": 231.875, "completions/min_terminated_length": 231.875, "entropy": 0.08505961950868368, "epoch": 0.192, "frac_reward_zero_std": 0.375, "grad_norm": 14.342461585998535, "learning_rate": 9.069148936170212e-07, "loss": -0.0092, "num_tokens": 4703567.0, "reward": 1.2356770932674408, "reward_std": 0.10872681951150298, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.2356770858168602, "rewards/frames_in_gt_reward/std": 0.10872681951150298, "step": 36, "step_time": 172.780337791075 }, { "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": 250.75, "completions/max_terminated_length": 250.75, "completions/mean_length": 223.609375, "completions/mean_terminated_length": 223.609375, "completions/min_length": 204.75, "completions/min_terminated_length": 204.75, "entropy": 0.10470232181251049, "epoch": 0.19733333333333333, "frac_reward_zero_std": 0.5, "grad_norm": 8.937066078186035, "learning_rate": 9.042553191489361e-07, "loss": 0.1419, "num_tokens": 4831630.0, "reward": 1.346875011920929, "reward_std": 0.08741450682282448, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.34687500447034836, "rewards/frames_in_gt_reward/std": 0.08741450682282448, "step": 37, "step_time": 157.44973521889187 }, { "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": 271.5, "completions/max_terminated_length": 271.5, "completions/mean_length": 230.0625, "completions/mean_terminated_length": 230.0625, "completions/min_length": 175.0, "completions/min_terminated_length": 175.0, "entropy": 0.10912697296589613, "epoch": 0.20266666666666666, "frac_reward_zero_std": 0.625, "grad_norm": 11.766274452209473, "learning_rate": 9.01595744680851e-07, "loss": -0.1302, "num_tokens": 4960234.0, "reward": 1.4153645932674408, "reward_std": 0.044061451219022274, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.4153645858168602, "rewards/frames_in_gt_reward/std": 0.04406144842505455, "step": 38, "step_time": 169.90393041598145 }, { "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": 273.625, "completions/max_terminated_length": 273.625, "completions/mean_length": 231.5, "completions/mean_terminated_length": 231.5, "completions/min_length": 180.75, "completions/min_terminated_length": 180.75, "entropy": 0.11349139083176851, "epoch": 0.208, "frac_reward_zero_std": 0.5, "grad_norm": 10.078503608703613, "learning_rate": 8.989361702127659e-07, "loss": -0.3446, "num_tokens": 5089426.0, "reward": 1.4661458432674408, "reward_std": 0.09397010691463947, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.4661458358168602, "rewards/frames_in_gt_reward/std": 0.09397011063992977, "step": 39, "step_time": 201.72449492895976 }, { "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": 307.75, "completions/max_terminated_length": 307.75, "completions/mean_length": 216.390625, "completions/mean_terminated_length": 216.390625, "completions/min_length": 154.25, "completions/min_terminated_length": 154.25, "entropy": 0.10795157123357058, "epoch": 0.21333333333333335, "frac_reward_zero_std": 0.25, "grad_norm": 17.03178596496582, "learning_rate": 8.962765957446809e-07, "loss": 0.0831, "num_tokens": 5217691.0, "reward": 1.5156250149011612, "reward_std": 0.13668302074074745, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.5156250027939677, "rewards/frames_in_gt_reward/std": 0.1366830226033926, "step": 40, "step_time": 168.2611275301315 }, { "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": 254.625, "completions/max_terminated_length": 254.625, "completions/mean_length": 200.4375, "completions/mean_terminated_length": 200.4375, "completions/min_length": 163.25, "completions/min_terminated_length": 163.25, "entropy": 0.14065338438376784, "epoch": 0.21866666666666668, "frac_reward_zero_std": 0.75, "grad_norm": 8.907140731811523, "learning_rate": 8.936170212765957e-07, "loss": -0.0433, "num_tokens": 5343503.0, "reward": 1.359375, "reward_std": 0.04813904967159033, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.359375, "rewards/frames_in_gt_reward/std": 0.048139046877622604, "step": 41, "step_time": 256.3465889458312 }, { "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": 279.625, "completions/max_terminated_length": 279.625, "completions/mean_length": 241.765625, "completions/mean_terminated_length": 241.765625, "completions/min_length": 201.5, "completions/min_terminated_length": 201.5, "entropy": 0.11048000399023294, "epoch": 0.224, "frac_reward_zero_std": 0.625, "grad_norm": 8.570083618164062, "learning_rate": 8.909574468085106e-07, "loss": 0.1567, "num_tokens": 5472896.0, "reward": 1.3072916865348816, "reward_std": 0.06408787239342928, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.3072916716337204, "rewards/frames_in_gt_reward/std": 0.06408787239342928, "step": 42, "step_time": 178.55916466517374 }, { "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": 279.75, "completions/max_terminated_length": 279.75, "completions/mean_length": 230.390625, "completions/mean_terminated_length": 230.390625, "completions/min_length": 173.5, "completions/min_terminated_length": 173.5, "entropy": 0.12323535978794098, "epoch": 0.22933333333333333, "frac_reward_zero_std": 0.375, "grad_norm": 51.93794250488281, "learning_rate": 8.882978723404255e-07, "loss": -0.3081, "num_tokens": 5600705.0, "reward": 1.4609375149011612, "reward_std": 0.14706461317837238, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.4609375037252903, "rewards/frames_in_gt_reward/std": 0.14706461317837238, "step": 43, "step_time": 164.4088007080136 }, { "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": 265.0, "completions/max_terminated_length": 265.0, "completions/mean_length": 223.84375, "completions/mean_terminated_length": 223.84375, "completions/min_length": 159.375, "completions/min_terminated_length": 159.375, "entropy": 0.09587646834552288, "epoch": 0.23466666666666666, "frac_reward_zero_std": 0.75, "grad_norm": 13.000391006469727, "learning_rate": 8.856382978723403e-07, "loss": -0.3432, "num_tokens": 5728135.0, "reward": 1.34375, "reward_std": 0.02437535859644413, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.34375, "rewards/frames_in_gt_reward/std": 0.024375353008508682, "step": 44, "step_time": 188.58748597512022 }, { "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": 269.0, "completions/max_terminated_length": 269.0, "completions/mean_length": 211.921875, "completions/mean_terminated_length": 211.921875, "completions/min_length": 172.625, "completions/min_terminated_length": 172.625, "entropy": 0.11408914299681783, "epoch": 0.24, "frac_reward_zero_std": 0.5, "grad_norm": 12.961020469665527, "learning_rate": 8.829787234042553e-07, "loss": 0.0283, "num_tokens": 5854594.0, "reward": 1.48828125, "reward_std": 0.09894341975450516, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.48828125, "rewards/frames_in_gt_reward/std": 0.0989434253424406, "step": 45, "step_time": 188.53382495220285 }, { "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": 291.875, "completions/max_terminated_length": 291.875, "completions/mean_length": 244.40625, "completions/mean_terminated_length": 244.40625, "completions/min_length": 201.625, "completions/min_terminated_length": 201.625, "entropy": 0.11054041376337409, "epoch": 0.24533333333333332, "frac_reward_zero_std": 0.75, "grad_norm": 9.556541442871094, "learning_rate": 8.803191489361702e-07, "loss": 0.2408, "num_tokens": 5984508.0, "reward": 1.3046875149011612, "reward_std": 0.0470782145857811, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.3046875037252903, "rewards/frames_in_gt_reward/std": 0.047078218311071396, "step": 46, "step_time": 185.13557544711512 }, { "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": 291.5, "completions/max_terminated_length": 291.5, "completions/mean_length": 203.609375, "completions/mean_terminated_length": 203.609375, "completions/min_length": 158.5, "completions/min_terminated_length": 158.5, "entropy": 0.14272658433765173, "epoch": 0.25066666666666665, "frac_reward_zero_std": 0.5, "grad_norm": 13.52221393585205, "learning_rate": 8.77659574468085e-07, "loss": 0.0824, "num_tokens": 6111195.0, "reward": 1.3554687798023224, "reward_std": 0.10578059032559395, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.35546875558793545, "rewards/frames_in_gt_reward/std": 0.1057805959135294, "step": 47, "step_time": 157.14214496605564 }, { "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": 285.375, "completions/max_terminated_length": 285.375, "completions/mean_length": 197.546875, "completions/mean_terminated_length": 197.546875, "completions/min_length": 137.375, "completions/min_terminated_length": 137.375, "entropy": 0.12425858527421951, "epoch": 0.256, "frac_reward_zero_std": 0.5, "grad_norm": 16.29790687561035, "learning_rate": 8.75e-07, "loss": -0.5537, "num_tokens": 6237238.0, "reward": 1.2808779925107956, "reward_std": 0.10152212623506784, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.2808779771439731, "rewards/frames_in_gt_reward/std": 0.10152213182300329, "step": 48, "step_time": 176.74100091005675 }, { "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": 292.125, "completions/max_terminated_length": 292.125, "completions/mean_length": 234.65625, "completions/mean_terminated_length": 234.65625, "completions/min_length": 169.625, "completions/min_terminated_length": 169.625, "entropy": 0.12429688218981028, "epoch": 0.2613333333333333, "frac_reward_zero_std": 0.5, "grad_norm": 7.397375106811523, "learning_rate": 8.723404255319149e-07, "loss": -0.0678, "num_tokens": 6366344.0, "reward": 1.4071614742279053, "reward_std": 0.08160671405494213, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.4071614630520344, "rewards/frames_in_gt_reward/std": 0.0816067261621356, "step": 49, "step_time": 181.59245612518862 }, { "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": 296.25, "completions/max_terminated_length": 296.25, "completions/mean_length": 234.140625, "completions/mean_terminated_length": 234.140625, "completions/min_length": 161.75, "completions/min_terminated_length": 161.75, "entropy": 0.13882701564580202, "epoch": 0.26666666666666666, "frac_reward_zero_std": 0.625, "grad_norm": 9.337924003601074, "learning_rate": 8.696808510638298e-07, "loss": 0.3199, "num_tokens": 6495169.0, "reward": 1.3098958432674408, "reward_std": 0.09704046696424484, "rewards/consistency_reward/mean": 1.0, "rewards/consistency_reward/std": 0.0, "rewards/frames_in_gt_reward/mean": 0.3098958358168602, "rewards/frames_in_gt_reward/std": 0.09704047814011574, "step": 50, "step_time": 181.65238326694816 }, { "epoch": 0.26666666666666666, "eval_clip_ratio/high_max": 0.0, "eval_clip_ratio/high_mean": 0.0, "eval_clip_ratio/low_mean": 0.0, "eval_clip_ratio/low_min": 0.0, "eval_clip_ratio/region_mean": 0.0, "eval_completions/clipped_ratio": 0.005, "eval_completions/max_length": 308.92, "eval_completions/max_terminated_length": 301.48, "eval_completions/mean_length": 224.795, "eval_completions/mean_terminated_length": 223.3914288330078, "eval_completions/min_length": 130.52, "eval_completions/min_terminated_length": 130.52, "eval_entropy": 0.11699200719594956, "eval_frac_reward_zero_std": 1.0, "eval_loss": 0.0, "eval_num_tokens": 6495169.0, "eval_reward": 1.313458342552185, "eval_reward_std": 0.34629204392433166, "eval_rewards/consistency_reward/mean": 0.99, "eval_rewards/consistency_reward/std": 0.028284270763397217, "eval_rewards/frames_in_gt_reward/mean": 0.32345833599567414, "eval_rewards/frames_in_gt_reward/std": 0.32378916561603543, "eval_runtime": 1088.1998, "eval_samples_per_second": 0.184, "eval_steps_per_second": 0.023, "step": 50 } ], "logging_steps": 1, "max_steps": 376, "num_input_tokens_seen": 6495169, "num_train_epochs": 2, "save_steps": 25, "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": 2, "trial_name": null, "trial_params": null }