| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.1251500085719184, |
| "eval_steps": 500, |
| "global_step": 730, |
| "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": 0.009895833674818277, |
| "completions/max_length": 1024.0, |
| "completions/max_terminated_length": 921.2, |
| "completions/mean_length": 747.1151245117187, |
| "completions/mean_terminated_length": 744.3689392089843, |
| "completions/min_length": 457.8, |
| "completions/min_terminated_length": 457.8, |
| "entropy": 0.7938701828320821, |
| "epoch": 0.0017143836790673753, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.294921875, |
| "learning_rate": 1e-06, |
| "loss": 0.0024, |
| "num_tokens": 4482493.0, |
| "reward": -11.88613681793213, |
| "reward_std": 9.975114345550537, |
| "rewards/ADERawReward/mean": -11.914521789550781, |
| "rewards/ADERawReward/std": 9.975466585159301, |
| "rewards/StrictFormatReward/mean": 0.28385417386889455, |
| "rewards/StrictFormatReward/std": 2.689612889289856, |
| "sampling/importance_sampling_ratio/max": 2.7710394859313965, |
| "sampling/importance_sampling_ratio/mean": 0.38329996466636657, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.348051142692566, |
| "sampling/sampling_logp_difference/mean": 0.02634742669761181, |
| "step": 10, |
| "step_time": 31.851699317700696 |
| }, |
| { |
| "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": 0.014062500465661288, |
| "completions/max_length": 1024.0, |
| "completions/max_terminated_length": 901.0, |
| "completions/mean_length": 746.5093994140625, |
| "completions/mean_terminated_length": 742.5637084960938, |
| "completions/min_length": 319.8, |
| "completions/min_terminated_length": 319.8, |
| "entropy": 0.783385141690572, |
| "epoch": 0.0034287673581347507, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.341796875, |
| "learning_rate": 1e-06, |
| "loss": -0.0167, |
| "num_tokens": 8963695.0, |
| "reward": -10.739660549163819, |
| "reward_std": 7.596912527084351, |
| "rewards/ADERawReward/mean": -10.76804552078247, |
| "rewards/ADERawReward/std": 7.586498594284057, |
| "rewards/StrictFormatReward/mean": 0.2838541708886623, |
| "rewards/StrictFormatReward/std": 2.69962215423584, |
| "sampling/importance_sampling_ratio/max": 2.778630328178406, |
| "sampling/importance_sampling_ratio/mean": 0.38422776758670807, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.3024064064025875, |
| "sampling/sampling_logp_difference/mean": 0.026079285889863968, |
| "step": 20, |
| "step_time": 31.08811371029442 |
| }, |
| { |
| "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": 0.01302083358168602, |
| "completions/max_length": 1024.0, |
| "completions/max_terminated_length": 916.5, |
| "completions/mean_length": 745.9713745117188, |
| "completions/mean_terminated_length": 742.3107482910157, |
| "completions/min_length": 344.2, |
| "completions/min_terminated_length": 344.2, |
| "entropy": 0.777154250939687, |
| "epoch": 0.005143151037202126, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.240234375, |
| "learning_rate": 1e-06, |
| "loss": -0.0033, |
| "num_tokens": 13443080.0, |
| "reward": -10.726643276214599, |
| "reward_std": 6.941620349884033, |
| "rewards/ADERawReward/mean": -10.759612274169921, |
| "rewards/ADERawReward/std": 6.939747667312622, |
| "rewards/StrictFormatReward/mean": 0.32968750670552255, |
| "rewards/StrictFormatReward/std": 2.6174988746643066, |
| "sampling/importance_sampling_ratio/max": 2.820967411994934, |
| "sampling/importance_sampling_ratio/mean": 0.3970964789390564, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.230282878875732, |
| "sampling/sampling_logp_difference/mean": 0.025640238262712955, |
| "step": 30, |
| "step_time": 31.281337887697738 |
| }, |
| { |
| "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": 0.006250000186264515, |
| "completions/max_length": 1006.1, |
| "completions/max_terminated_length": 952.6, |
| "completions/mean_length": 746.8578247070312, |
| "completions/mean_terminated_length": 745.1074523925781, |
| "completions/min_length": 574.6, |
| "completions/min_terminated_length": 574.6, |
| "entropy": 0.7786477545897166, |
| "epoch": 0.006857534716269501, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.353515625, |
| "learning_rate": 1e-06, |
| "loss": 0.0052, |
| "num_tokens": 17925079.0, |
| "reward": -10.400494289398193, |
| "reward_std": 8.363166570663452, |
| "rewards/ADERawReward/mean": -10.453514766693115, |
| "rewards/ADERawReward/std": 8.35603289604187, |
| "rewards/StrictFormatReward/mean": 0.5302083313465118, |
| "rewards/StrictFormatReward/std": 2.1997122406959533, |
| "sampling/importance_sampling_ratio/max": 2.7520124673843385, |
| "sampling/importance_sampling_ratio/mean": 0.4182109236717224, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.882514882087707, |
| "sampling/sampling_logp_difference/mean": 0.025688859820365905, |
| "step": 40, |
| "step_time": 30.94830545460136 |
| }, |
| { |
| "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": 0.010416666883975268, |
| "completions/max_length": 1008.7, |
| "completions/max_terminated_length": 908.0, |
| "completions/mean_length": 744.6708557128907, |
| "completions/mean_terminated_length": 741.7320251464844, |
| "completions/min_length": 318.0, |
| "completions/min_terminated_length": 318.0, |
| "entropy": 0.7797636012236278, |
| "epoch": 0.008571918395336876, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.296875, |
| "learning_rate": 1e-06, |
| "loss": 0.0059, |
| "num_tokens": 22402927.0, |
| "reward": -11.912008953094482, |
| "reward_std": 25.465690279006957, |
| "rewards/ADERawReward/mean": -11.948414325714111, |
| "rewards/ADERawReward/std": 25.452851724624633, |
| "rewards/StrictFormatReward/mean": 0.3640625074505806, |
| "rewards/StrictFormatReward/std": 2.5515830755233764, |
| "sampling/importance_sampling_ratio/max": 2.7972999811172485, |
| "sampling/importance_sampling_ratio/mean": 0.41742126941680907, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.1138537406921385, |
| "sampling/sampling_logp_difference/mean": 0.025952593609690666, |
| "step": 50, |
| "step_time": 30.99103705089656 |
| }, |
| { |
| "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": 0.00572916679084301, |
| "completions/max_length": 982.4, |
| "completions/max_terminated_length": 894.4, |
| "completions/mean_length": 743.1708618164063, |
| "completions/mean_terminated_length": 741.5553588867188, |
| "completions/min_length": 389.5, |
| "completions/min_terminated_length": 389.5, |
| "entropy": 0.7705018659432729, |
| "epoch": 0.010286302074404252, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.294921875, |
| "learning_rate": 1e-06, |
| "loss": -0.001, |
| "num_tokens": 26877415.0, |
| "reward": -10.354321956634521, |
| "reward_std": 7.085966682434082, |
| "rewards/ADERawReward/mean": -10.3998948097229, |
| "rewards/ADERawReward/std": 7.078661155700684, |
| "rewards/StrictFormatReward/mean": 0.45572916492819787, |
| "rewards/StrictFormatReward/std": 2.357165718078613, |
| "sampling/importance_sampling_ratio/max": 2.8156795501708984, |
| "sampling/importance_sampling_ratio/mean": 0.4263548344373703, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.213912606239319, |
| "sampling/sampling_logp_difference/mean": 0.025748718157410623, |
| "step": 60, |
| "step_time": 30.629806772600567 |
| }, |
| { |
| "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": 0.007812500139698387, |
| "completions/max_length": 997.5, |
| "completions/max_terminated_length": 901.5, |
| "completions/mean_length": 745.1635559082031, |
| "completions/mean_terminated_length": 742.9772766113281, |
| "completions/min_length": 448.0, |
| "completions/min_terminated_length": 448.0, |
| "entropy": 0.7808759689331055, |
| "epoch": 0.012000685753471628, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.416015625, |
| "learning_rate": 1e-06, |
| "loss": -0.0022, |
| "num_tokens": 31355793.0, |
| "reward": -10.355865859985352, |
| "reward_std": 7.629749822616577, |
| "rewards/ADERawReward/mean": -10.399147129058838, |
| "rewards/ADERawReward/std": 7.623885774612427, |
| "rewards/StrictFormatReward/mean": 0.43281250447034836, |
| "rewards/StrictFormatReward/std": 2.4088792324066164, |
| "sampling/importance_sampling_ratio/max": 2.7858126163482666, |
| "sampling/importance_sampling_ratio/mean": 0.4104765444993973, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.654694700241089, |
| "sampling/sampling_logp_difference/mean": 0.02596179526299238, |
| "step": 70, |
| "step_time": 30.90157012869895 |
| }, |
| { |
| "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": 0.006250000139698386, |
| "completions/max_length": 997.0, |
| "completions/max_terminated_length": 911.7, |
| "completions/mean_length": 743.9010620117188, |
| "completions/mean_terminated_length": 742.1458801269531, |
| "completions/min_length": 506.5, |
| "completions/min_terminated_length": 506.5, |
| "entropy": 0.7690683722496032, |
| "epoch": 0.013715069432539003, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.259765625, |
| "learning_rate": 1e-06, |
| "loss": -0.0147, |
| "num_tokens": 35831699.0, |
| "reward": -9.603075504302979, |
| "reward_std": 7.397923803329467, |
| "rewards/ADERawReward/mean": -9.64406452178955, |
| "rewards/ADERawReward/std": 7.394052314758301, |
| "rewards/StrictFormatReward/mean": 0.40989583656191825, |
| "rewards/StrictFormatReward/std": 2.446663808822632, |
| "sampling/importance_sampling_ratio/max": 2.8509432315826415, |
| "sampling/importance_sampling_ratio/mean": 0.42068196535110475, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 3.871529459953308, |
| "sampling/sampling_logp_difference/mean": 0.025504560954868794, |
| "step": 80, |
| "step_time": 30.775000178898335 |
| }, |
| { |
| "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": 0.006770833488553762, |
| "completions/max_length": 1018.6, |
| "completions/max_terminated_length": 921.4, |
| "completions/mean_length": 743.0687744140625, |
| "completions/mean_terminated_length": 741.1454040527344, |
| "completions/min_length": 413.2, |
| "completions/min_terminated_length": 413.2, |
| "entropy": 0.7730992416540782, |
| "epoch": 0.015429453111606378, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.330078125, |
| "learning_rate": 1e-06, |
| "loss": -0.0209, |
| "num_tokens": 40306439.0, |
| "reward": -9.999512577056885, |
| "reward_std": 6.795107650756836, |
| "rewards/ADERawReward/mean": -10.04222068786621, |
| "rewards/ADERawReward/std": 6.785987043380738, |
| "rewards/StrictFormatReward/mean": 0.4270833358168602, |
| "rewards/StrictFormatReward/std": 2.427087187767029, |
| "sampling/importance_sampling_ratio/max": 2.806507611274719, |
| "sampling/importance_sampling_ratio/mean": 0.41283826231956483, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.893343424797058, |
| "sampling/sampling_logp_difference/mean": 0.02597447969019413, |
| "step": 90, |
| "step_time": 31.03075063330034 |
| }, |
| { |
| "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": 0.006250000139698386, |
| "completions/max_length": 989.1, |
| "completions/max_terminated_length": 911.9, |
| "completions/mean_length": 744.7958557128907, |
| "completions/mean_terminated_length": 743.0538269042969, |
| "completions/min_length": 479.0, |
| "completions/min_terminated_length": 479.0, |
| "entropy": 0.7617896695931753, |
| "epoch": 0.01714383679067375, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.26953125, |
| "learning_rate": 1e-06, |
| "loss": 0.0056, |
| "num_tokens": 44785151.0, |
| "reward": -10.150122165679932, |
| "reward_std": 6.947455310821534, |
| "rewards/ADERawReward/mean": -10.1985595703125, |
| "rewards/ADERawReward/std": 6.9321370124816895, |
| "rewards/StrictFormatReward/mean": 0.48437499925494193, |
| "rewards/StrictFormatReward/std": 2.2780819296836854, |
| "sampling/importance_sampling_ratio/max": 2.8471882343292236, |
| "sampling/importance_sampling_ratio/mean": 0.4186240643262863, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 5.192489242553711, |
| "sampling/sampling_logp_difference/mean": 0.025915385968983174, |
| "step": 100, |
| "step_time": 30.777787425796852 |
| }, |
| { |
| "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": 0.008854166837409139, |
| "completions/max_length": 1016.1, |
| "completions/max_terminated_length": 927.6, |
| "completions/mean_length": 747.430224609375, |
| "completions/mean_terminated_length": 744.9568664550782, |
| "completions/min_length": 542.5, |
| "completions/min_terminated_length": 542.5, |
| "entropy": 0.7839122792085012, |
| "epoch": 0.018858220469741126, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.369140625, |
| "learning_rate": 1e-06, |
| "loss": 0.0095, |
| "num_tokens": 49267497.0, |
| "reward": -10.34236192703247, |
| "reward_std": 6.522932863235473, |
| "rewards/ADERawReward/mean": -10.377049255371094, |
| "rewards/ADERawReward/std": 6.512392950057984, |
| "rewards/StrictFormatReward/mean": 0.3468750029802322, |
| "rewards/StrictFormatReward/std": 2.5984415769577027, |
| "sampling/importance_sampling_ratio/max": 2.919663596153259, |
| "sampling/importance_sampling_ratio/mean": 0.42760642170906066, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.582067346572876, |
| "sampling/sampling_logp_difference/mean": 0.025711694732308388, |
| "step": 110, |
| "step_time": 31.13533423260669 |
| }, |
| { |
| "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": 0.006250000139698386, |
| "completions/max_length": 981.8, |
| "completions/max_terminated_length": 905.6, |
| "completions/mean_length": 741.3906433105469, |
| "completions/mean_terminated_length": 739.6170837402344, |
| "completions/min_length": 322.0, |
| "completions/min_terminated_length": 322.0, |
| "entropy": 0.7655617932478587, |
| "epoch": 0.020572604148808505, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.2255859375, |
| "learning_rate": 1e-06, |
| "loss": -0.0003, |
| "num_tokens": 53739095.0, |
| "reward": -9.396285820007325, |
| "reward_std": 6.019436550140381, |
| "rewards/ADERawReward/mean": -9.43211908340454, |
| "rewards/ADERawReward/std": 5.99921350479126, |
| "rewards/StrictFormatReward/mean": 0.35833333134651185, |
| "rewards/StrictFormatReward/std": 2.565337634086609, |
| "sampling/importance_sampling_ratio/max": 2.7736236810684205, |
| "sampling/importance_sampling_ratio/mean": 0.4329004347324371, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 5.103514838218689, |
| "sampling/sampling_logp_difference/mean": 0.02581946086138487, |
| "step": 120, |
| "step_time": 30.675611695800036 |
| }, |
| { |
| "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": 0.006250000139698386, |
| "completions/max_length": 993.0, |
| "completions/max_terminated_length": 917.3, |
| "completions/mean_length": 743.2172119140625, |
| "completions/mean_terminated_length": 741.4531127929688, |
| "completions/min_length": 448.6, |
| "completions/min_terminated_length": 448.6, |
| "entropy": 0.7609198371569316, |
| "epoch": 0.02228698782787588, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.306640625, |
| "learning_rate": 1e-06, |
| "loss": -0.0186, |
| "num_tokens": 58213992.0, |
| "reward": -10.035307788848877, |
| "reward_std": 7.6791582107543945, |
| "rewards/ADERawReward/mean": -10.080307579040527, |
| "rewards/ADERawReward/std": 7.661759281158448, |
| "rewards/StrictFormatReward/mean": 0.45, |
| "rewards/StrictFormatReward/std": 2.383046817779541, |
| "sampling/importance_sampling_ratio/max": 2.7497065782547, |
| "sampling/importance_sampling_ratio/mean": 0.38673609495162964, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.264748442173004, |
| "sampling/sampling_logp_difference/mean": 0.02555535715073347, |
| "step": 130, |
| "step_time": 30.85125565490016 |
| }, |
| { |
| "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": 0.008333333535119891, |
| "completions/max_length": 995.9, |
| "completions/max_terminated_length": 936.1, |
| "completions/mean_length": 747.1015808105469, |
| "completions/mean_terminated_length": 744.7781860351563, |
| "completions/min_length": 504.1, |
| "completions/min_terminated_length": 504.1, |
| "entropy": 0.7680625875790914, |
| "epoch": 0.024001371506943255, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.283203125, |
| "learning_rate": 1e-06, |
| "loss": -0.014, |
| "num_tokens": 62696459.0, |
| "reward": -9.6462965965271, |
| "reward_std": 6.933882856369019, |
| "rewards/ADERawReward/mean": -9.687859153747558, |
| "rewards/ADERawReward/std": 6.932486724853516, |
| "rewards/StrictFormatReward/mean": 0.41562500298023225, |
| "rewards/StrictFormatReward/std": 2.435987985134125, |
| "sampling/importance_sampling_ratio/max": 2.7624801874160765, |
| "sampling/importance_sampling_ratio/mean": 0.39736475646495817, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.469653701782226, |
| "sampling/sampling_logp_difference/mean": 0.025803259573876858, |
| "step": 140, |
| "step_time": 30.751169363802184 |
| }, |
| { |
| "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": 0.0067708335351198915, |
| "completions/max_length": 996.8, |
| "completions/max_terminated_length": 927.2, |
| "completions/mean_length": 743.0797119140625, |
| "completions/mean_terminated_length": 741.1633605957031, |
| "completions/min_length": 364.4, |
| "completions/min_terminated_length": 364.4, |
| "entropy": 0.771180788675944, |
| "epoch": 0.02571575518601063, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.322265625, |
| "learning_rate": 1e-06, |
| "loss": -0.0135, |
| "num_tokens": 67171268.0, |
| "reward": -9.683474731445312, |
| "reward_std": 6.344115209579468, |
| "rewards/ADERawReward/mean": -9.721026802062989, |
| "rewards/ADERawReward/std": 6.331903743743896, |
| "rewards/StrictFormatReward/mean": 0.3755208313465118, |
| "rewards/StrictFormatReward/std": 2.536852979660034, |
| "sampling/importance_sampling_ratio/max": 2.6906844854354857, |
| "sampling/importance_sampling_ratio/mean": 0.3939074516296387, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.591675233840943, |
| "sampling/sampling_logp_difference/mean": 0.026031140238046646, |
| "step": 150, |
| "step_time": 30.759532438102177 |
| }, |
| { |
| "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": 0.009375000186264516, |
| "completions/max_length": 1001.1, |
| "completions/max_terminated_length": 928.5, |
| "completions/mean_length": 743.9906616210938, |
| "completions/mean_terminated_length": 741.3461547851563, |
| "completions/min_length": 396.0, |
| "completions/min_terminated_length": 396.0, |
| "entropy": 0.772610588868459, |
| "epoch": 0.027430138865078006, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.291015625, |
| "learning_rate": 1e-06, |
| "loss": -0.006, |
| "num_tokens": 71648274.0, |
| "reward": -9.378812217712403, |
| "reward_std": 6.656644868850708, |
| "rewards/ADERawReward/mean": -9.419801712036133, |
| "rewards/ADERawReward/std": 6.638844442367554, |
| "rewards/StrictFormatReward/mean": 0.4098958343267441, |
| "rewards/StrictFormatReward/std": 2.457722854614258, |
| "sampling/importance_sampling_ratio/max": 2.8564789295196533, |
| "sampling/importance_sampling_ratio/mean": 0.3868180692195892, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 5.245336532592773, |
| "sampling/sampling_logp_difference/mean": 0.026049494743347168, |
| "step": 160, |
| "step_time": 31.00511598830053 |
| }, |
| { |
| "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": 0.00572916679084301, |
| "completions/max_length": 990.5, |
| "completions/max_terminated_length": 943.8, |
| "completions/mean_length": 742.833349609375, |
| "completions/mean_terminated_length": 741.2216247558594, |
| "completions/min_length": 507.2, |
| "completions/min_terminated_length": 507.2, |
| "entropy": 0.7669690032800038, |
| "epoch": 0.02914452254414538, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.2421875, |
| "learning_rate": 1e-06, |
| "loss": 0.0046, |
| "num_tokens": 76122162.0, |
| "reward": -9.605478668212891, |
| "reward_std": 5.803140306472779, |
| "rewards/ADERawReward/mean": -9.645895195007324, |
| "rewards/ADERawReward/std": 5.7928056716918945, |
| "rewards/StrictFormatReward/mean": 0.4041666634380817, |
| "rewards/StrictFormatReward/std": 2.4726577758789063, |
| "sampling/importance_sampling_ratio/max": 2.71953125, |
| "sampling/importance_sampling_ratio/mean": 0.4068588227033615, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.668109703063965, |
| "sampling/sampling_logp_difference/mean": 0.025866417028009892, |
| "step": 170, |
| "step_time": 30.801190392104036 |
| }, |
| { |
| "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": 0.0067708335351198915, |
| "completions/max_length": 992.8, |
| "completions/max_terminated_length": 902.9, |
| "completions/mean_length": 745.7635559082031, |
| "completions/mean_terminated_length": 743.8595642089844, |
| "completions/min_length": 539.9, |
| "completions/min_terminated_length": 539.9, |
| "entropy": 0.7681126793225607, |
| "epoch": 0.030858906223212756, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.330078125, |
| "learning_rate": 1e-06, |
| "loss": -0.0009, |
| "num_tokens": 80601532.0, |
| "reward": -10.118904399871827, |
| "reward_std": 6.3121555805206295, |
| "rewards/ADERawReward/mean": -10.168487930297852, |
| "rewards/ADERawReward/std": 6.31596007347107, |
| "rewards/StrictFormatReward/mean": 0.49583332538604735, |
| "rewards/StrictFormatReward/std": 2.2922281503677366, |
| "sampling/importance_sampling_ratio/max": 2.808049750328064, |
| "sampling/importance_sampling_ratio/mean": 0.40117439031600954, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 5.4242959260940555, |
| "sampling/sampling_logp_difference/mean": 0.026261823065578938, |
| "step": 180, |
| "step_time": 30.95573297930241 |
| }, |
| { |
| "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": 0.009375000186264516, |
| "completions/max_length": 1006.3, |
| "completions/max_terminated_length": 940.1, |
| "completions/mean_length": 745.2932495117187, |
| "completions/mean_terminated_length": 742.6480163574219, |
| "completions/min_length": 493.3, |
| "completions/min_terminated_length": 493.3, |
| "entropy": 0.7715065141518911, |
| "epoch": 0.03257328990228013, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.349609375, |
| "learning_rate": 1e-06, |
| "loss": -0.0167, |
| "num_tokens": 85080239.0, |
| "reward": -9.032559633255005, |
| "reward_std": 5.6337450504302975, |
| "rewards/ADERawReward/mean": -9.070111560821534, |
| "rewards/ADERawReward/std": 5.615639925003052, |
| "rewards/StrictFormatReward/mean": 0.3755208343267441, |
| "rewards/StrictFormatReward/std": 2.532916522026062, |
| "sampling/importance_sampling_ratio/max": 2.906028723716736, |
| "sampling/importance_sampling_ratio/mean": 0.43047145903110506, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.560648036003113, |
| "sampling/sampling_logp_difference/mean": 0.02589159458875656, |
| "step": 190, |
| "step_time": 31.04905029379588 |
| }, |
| { |
| "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": 0.004687500139698386, |
| "completions/max_length": 967.5, |
| "completions/max_terminated_length": 901.5, |
| "completions/mean_length": 743.9588745117187, |
| "completions/mean_terminated_length": 742.6530639648438, |
| "completions/min_length": 482.7, |
| "completions/min_terminated_length": 482.7, |
| "entropy": 0.7573511083920796, |
| "epoch": 0.0342876735813475, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.373046875, |
| "learning_rate": 1e-06, |
| "loss": -0.0071, |
| "num_tokens": 89554992.0, |
| "reward": -9.594873905181885, |
| "reward_std": 7.4608996391296385, |
| "rewards/ADERawReward/mean": -9.645030212402343, |
| "rewards/ADERawReward/std": 7.4621100425720215, |
| "rewards/StrictFormatReward/mean": 0.5015625014901162, |
| "rewards/StrictFormatReward/std": 2.2230212569236754, |
| "sampling/importance_sampling_ratio/max": 2.762194800376892, |
| "sampling/importance_sampling_ratio/mean": 0.3942416876554489, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.715522122383118, |
| "sampling/sampling_logp_difference/mean": 0.02548809293657541, |
| "step": 200, |
| "step_time": 30.452962850301994 |
| }, |
| { |
| "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": 0.008854166883975268, |
| "completions/max_length": 1002.2, |
| "completions/max_terminated_length": 908.9, |
| "completions/mean_length": 749.1671997070313, |
| "completions/mean_terminated_length": 746.7061828613281, |
| "completions/min_length": 479.3, |
| "completions/min_terminated_length": 479.3, |
| "entropy": 0.7769242246945699, |
| "epoch": 0.03600205726041488, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.349609375, |
| "learning_rate": 1e-06, |
| "loss": -0.0264, |
| "num_tokens": 94041409.0, |
| "reward": -9.6312180519104, |
| "reward_std": 5.488570928573608, |
| "rewards/ADERawReward/mean": -9.677363443374634, |
| "rewards/ADERawReward/std": 5.482590246200561, |
| "rewards/StrictFormatReward/mean": 0.46145834028720856, |
| "rewards/StrictFormatReward/std": 2.369260883331299, |
| "sampling/importance_sampling_ratio/max": 2.827503228187561, |
| "sampling/importance_sampling_ratio/mean": 0.39889043271541597, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.615737795829773, |
| "sampling/sampling_logp_difference/mean": 0.02642081268131733, |
| "step": 210, |
| "step_time": 31.002005700898007 |
| }, |
| { |
| "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": 0.007291666883975268, |
| "completions/max_length": 1011.2, |
| "completions/max_terminated_length": 917.8, |
| "completions/mean_length": 744.7901306152344, |
| "completions/mean_terminated_length": 742.7363525390625, |
| "completions/min_length": 326.2, |
| "completions/min_terminated_length": 326.2, |
| "entropy": 0.7611714641253153, |
| "epoch": 0.03771644093948225, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.2470703125, |
| "learning_rate": 1e-06, |
| "loss": -0.0072, |
| "num_tokens": 98520158.0, |
| "reward": -9.730458450317382, |
| "reward_std": 6.581094980239868, |
| "rewards/ADERawReward/mean": -9.778323078155518, |
| "rewards/ADERawReward/std": 6.570281600952148, |
| "rewards/StrictFormatReward/mean": 0.478645833581686, |
| "rewards/StrictFormatReward/std": 2.319968819618225, |
| "sampling/importance_sampling_ratio/max": 2.9045305013656617, |
| "sampling/importance_sampling_ratio/mean": 0.4044792056083679, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.901871252059936, |
| "sampling/sampling_logp_difference/mean": 0.026320008002221583, |
| "step": 220, |
| "step_time": 31.057802741799968 |
| }, |
| { |
| "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": 0.00833333358168602, |
| "completions/max_length": 1006.6, |
| "completions/max_terminated_length": 907.8, |
| "completions/mean_length": 744.116162109375, |
| "completions/mean_terminated_length": 741.7729553222656, |
| "completions/min_length": 364.6, |
| "completions/min_terminated_length": 364.6, |
| "entropy": 0.7634056508541107, |
| "epoch": 0.03943082461854963, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.35546875, |
| "learning_rate": 1e-06, |
| "loss": 0.0058, |
| "num_tokens": 102995469.0, |
| "reward": -9.594787311553954, |
| "reward_std": 6.4190247535705565, |
| "rewards/ADERawReward/mean": -9.632339668273925, |
| "rewards/ADERawReward/std": 6.41018123626709, |
| "rewards/StrictFormatReward/mean": 0.37552083283662796, |
| "rewards/StrictFormatReward/std": 2.4954795360565187, |
| "sampling/importance_sampling_ratio/max": 2.7927833795547485, |
| "sampling/importance_sampling_ratio/mean": 0.38037638664245604, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 5.303525161743164, |
| "sampling/sampling_logp_difference/mean": 0.026079374738037585, |
| "step": 230, |
| "step_time": 30.925862592204066 |
| }, |
| { |
| "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": 0.015625000232830644, |
| "completions/max_length": 1024.0, |
| "completions/max_terminated_length": 922.7, |
| "completions/mean_length": 747.7349060058593, |
| "completions/mean_terminated_length": 743.3495300292968, |
| "completions/min_length": 425.3, |
| "completions/min_terminated_length": 425.3, |
| "entropy": 0.7614292343457539, |
| "epoch": 0.04114520829761701, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.3984375, |
| "learning_rate": 1e-06, |
| "loss": 0.0014, |
| "num_tokens": 107478240.0, |
| "reward": -9.159432983398437, |
| "reward_std": 5.858185243606568, |
| "rewards/ADERawReward/mean": -9.192401695251466, |
| "rewards/ADERawReward/std": 5.8467125415802, |
| "rewards/StrictFormatReward/mean": 0.3296875016763806, |
| "rewards/StrictFormatReward/std": 2.6086880922317506, |
| "sampling/importance_sampling_ratio/max": 2.8600486755371093, |
| "sampling/importance_sampling_ratio/mean": 0.38983087837696073, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 5.5773406505584715, |
| "sampling/sampling_logp_difference/mean": 0.026277855411171912, |
| "step": 240, |
| "step_time": 31.37704919550306 |
| }, |
| { |
| "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": 0.011458333488553763, |
| "completions/max_length": 1012.8, |
| "completions/max_terminated_length": 905.7, |
| "completions/mean_length": 743.3661682128907, |
| "completions/mean_terminated_length": 740.1103088378907, |
| "completions/min_length": 399.4, |
| "completions/min_terminated_length": 399.4, |
| "entropy": 0.753006378809611, |
| "epoch": 0.04285959197668438, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.35546875, |
| "learning_rate": 1e-06, |
| "loss": -0.0194, |
| "num_tokens": 111952815.0, |
| "reward": -9.348959684371948, |
| "reward_std": 5.709445333480835, |
| "rewards/ADERawReward/mean": -9.387657499313354, |
| "rewards/ADERawReward/std": 5.683103704452515, |
| "rewards/StrictFormatReward/mean": 0.38697916716337205, |
| "rewards/StrictFormatReward/std": 2.5156437873840334, |
| "sampling/importance_sampling_ratio/max": 2.7942948818206785, |
| "sampling/importance_sampling_ratio/mean": 0.3898850232362747, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.9041135787963865, |
| "sampling/sampling_logp_difference/mean": 0.026003769040107726, |
| "step": 250, |
| "step_time": 31.078463642198766 |
| }, |
| { |
| "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": 0.016145833814516664, |
| "completions/max_length": 1017.2, |
| "completions/max_terminated_length": 912.4, |
| "completions/mean_length": 749.7567993164063, |
| "completions/mean_terminated_length": 745.2748840332031, |
| "completions/min_length": 587.3, |
| "completions/min_terminated_length": 587.3, |
| "entropy": 0.7615062018235524, |
| "epoch": 0.04457397565575176, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.3984375, |
| "learning_rate": 1e-06, |
| "loss": -0.0051, |
| "num_tokens": 116440044.0, |
| "reward": -9.92240343093872, |
| "reward_std": 5.772085857391358, |
| "rewards/ADERawReward/mean": -9.958809566497802, |
| "rewards/ADERawReward/std": 5.756062030792236, |
| "rewards/StrictFormatReward/mean": 0.36406250670552254, |
| "rewards/StrictFormatReward/std": 2.54167058467865, |
| "sampling/importance_sampling_ratio/max": 2.7225013256072996, |
| "sampling/importance_sampling_ratio/mean": 0.3842493683099747, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.781629824638367, |
| "sampling/sampling_logp_difference/mean": 0.02624006476253271, |
| "step": 260, |
| "step_time": 31.103271031705663 |
| }, |
| { |
| "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": 0.008333333535119891, |
| "completions/max_length": 998.2, |
| "completions/max_terminated_length": 911.1, |
| "completions/mean_length": 743.7968994140625, |
| "completions/mean_terminated_length": 741.4540466308594, |
| "completions/min_length": 510.3, |
| "completions/min_terminated_length": 510.3, |
| "entropy": 0.7562108953793844, |
| "epoch": 0.04628835933481913, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.1953125, |
| "learning_rate": 1e-06, |
| "loss": -0.0247, |
| "num_tokens": 120916390.0, |
| "reward": -9.01723370552063, |
| "reward_std": 5.40621428489685, |
| "rewards/ADERawReward/mean": -9.065098094940186, |
| "rewards/ADERawReward/std": 5.384741115570068, |
| "rewards/StrictFormatReward/mean": 0.4786458298563957, |
| "rewards/StrictFormatReward/std": 2.2954741954803466, |
| "sampling/importance_sampling_ratio/max": 2.789908194541931, |
| "sampling/importance_sampling_ratio/mean": 0.40749328434467313, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.034530830383301, |
| "sampling/sampling_logp_difference/mean": 0.026305202022194864, |
| "step": 270, |
| "step_time": 30.820387327201026 |
| }, |
| { |
| "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": 0.009375000139698386, |
| "completions/max_length": 1011.8, |
| "completions/max_terminated_length": 914.4, |
| "completions/mean_length": 745.3156433105469, |
| "completions/mean_terminated_length": 742.68203125, |
| "completions/min_length": 492.1, |
| "completions/min_terminated_length": 492.1, |
| "entropy": 0.757685911655426, |
| "epoch": 0.04800274301388651, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.349609375, |
| "learning_rate": 1e-06, |
| "loss": 0.002, |
| "num_tokens": 125395556.0, |
| "reward": -9.0065682888031, |
| "reward_std": 5.3750749111175535, |
| "rewards/ADERawReward/mean": -9.06875557899475, |
| "rewards/ADERawReward/std": 5.365862655639648, |
| "rewards/StrictFormatReward/mean": 0.6218749970197678, |
| "rewards/StrictFormatReward/std": 1.9849600195884705, |
| "sampling/importance_sampling_ratio/max": 2.791543960571289, |
| "sampling/importance_sampling_ratio/mean": 0.36947061419487, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 5.032503914833069, |
| "sampling/sampling_logp_difference/mean": 0.02653008736670017, |
| "step": 280, |
| "step_time": 31.023602066392776 |
| }, |
| { |
| "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": 0.009375000139698386, |
| "completions/max_length": 1011.5, |
| "completions/max_terminated_length": 906.5, |
| "completions/mean_length": 745.1588745117188, |
| "completions/mean_terminated_length": 742.5252624511719, |
| "completions/min_length": 384.9, |
| "completions/min_terminated_length": 384.9, |
| "entropy": 0.7569859484831493, |
| "epoch": 0.04971712669295388, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.275390625, |
| "learning_rate": 1e-06, |
| "loss": -0.0025, |
| "num_tokens": 129874581.0, |
| "reward": -9.8903018951416, |
| "reward_std": 5.79907751083374, |
| "rewards/ADERawReward/mean": -9.936447715759277, |
| "rewards/ADERawReward/std": 5.782765436172485, |
| "rewards/StrictFormatReward/mean": 0.4614583358168602, |
| "rewards/StrictFormatReward/std": 2.3571563005447387, |
| "sampling/importance_sampling_ratio/max": 2.780865716934204, |
| "sampling/importance_sampling_ratio/mean": 0.3752612203359604, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 5.010831832885742, |
| "sampling/sampling_logp_difference/mean": 0.026486290618777275, |
| "step": 290, |
| "step_time": 31.036961890700333 |
| }, |
| { |
| "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": 0.010937500186264515, |
| "completions/max_length": 1024.0, |
| "completions/max_terminated_length": 909.3, |
| "completions/mean_length": 743.6302307128906, |
| "completions/mean_terminated_length": 740.5345153808594, |
| "completions/min_length": 545.1, |
| "completions/min_terminated_length": 545.1, |
| "entropy": 0.7479812840620677, |
| "epoch": 0.05143151037202126, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.296875, |
| "learning_rate": 1e-06, |
| "loss": 0.0075, |
| "num_tokens": 134349103.0, |
| "reward": -8.726549053192139, |
| "reward_std": 5.281997394561768, |
| "rewards/ADERawReward/mean": -8.780142164230346, |
| "rewards/ADERawReward/std": 5.273246812820434, |
| "rewards/StrictFormatReward/mean": 0.5359374985098839, |
| "rewards/StrictFormatReward/std": 2.176564943790436, |
| "sampling/importance_sampling_ratio/max": 2.9104627132415772, |
| "sampling/importance_sampling_ratio/mean": 0.40789560675621034, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.580249309539795, |
| "sampling/sampling_logp_difference/mean": 0.025899984315037726, |
| "step": 300, |
| "step_time": 31.32498294780089 |
| }, |
| { |
| "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": 0.007812500186264514, |
| "completions/max_length": 998.0, |
| "completions/max_terminated_length": 917.8, |
| "completions/mean_length": 745.6083557128907, |
| "completions/mean_terminated_length": 743.4208374023438, |
| "completions/min_length": 550.1, |
| "completions/min_terminated_length": 550.1, |
| "entropy": 0.7465774556001027, |
| "epoch": 0.05314589405108863, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.353515625, |
| "learning_rate": 1e-06, |
| "loss": -0.0123, |
| "num_tokens": 138828591.0, |
| "reward": -9.077164840698241, |
| "reward_std": 5.992494773864746, |
| "rewards/ADERawReward/mean": -9.131331491470338, |
| "rewards/ADERawReward/std": 5.984670972824096, |
| "rewards/StrictFormatReward/mean": 0.5416666656732559, |
| "rewards/StrictFormatReward/std": 2.169144892692566, |
| "sampling/importance_sampling_ratio/max": 2.749341917037964, |
| "sampling/importance_sampling_ratio/mean": 0.3683441460132599, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 5.0134562969207765, |
| "sampling/sampling_logp_difference/mean": 0.026201115176081657, |
| "step": 310, |
| "step_time": 30.797605072900478 |
| }, |
| { |
| "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": 0.016145833814516664, |
| "completions/max_length": 1024.0, |
| "completions/max_terminated_length": 899.7, |
| "completions/mean_length": 747.6823181152344, |
| "completions/mean_terminated_length": 743.1548828125, |
| "completions/min_length": 458.0, |
| "completions/min_terminated_length": 458.0, |
| "entropy": 0.7556955059369405, |
| "epoch": 0.05486027773015601, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.3671875, |
| "learning_rate": 1e-06, |
| "loss": 0.006, |
| "num_tokens": 143312285.0, |
| "reward": -9.226322269439697, |
| "reward_std": 6.171311950683593, |
| "rewards/ADERawReward/mean": -9.265593338012696, |
| "rewards/ADERawReward/std": 6.160300636291504, |
| "rewards/StrictFormatReward/mean": 0.39270834550261496, |
| "rewards/StrictFormatReward/std": 2.4976075172424315, |
| "sampling/importance_sampling_ratio/max": 2.7495847940444946, |
| "sampling/importance_sampling_ratio/mean": 0.384751632809639, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.004838466644287, |
| "sampling/sampling_logp_difference/mean": 0.026612747460603714, |
| "step": 320, |
| "step_time": 31.281772803403147 |
| }, |
| { |
| "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": 0.007812500093132257, |
| "completions/max_length": 1001.5, |
| "completions/max_terminated_length": 901.0, |
| "completions/mean_length": 745.114599609375, |
| "completions/mean_terminated_length": 742.9226379394531, |
| "completions/min_length": 415.5, |
| "completions/min_terminated_length": 415.5, |
| "entropy": 0.7487649619579315, |
| "epoch": 0.05657466140922338, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.2099609375, |
| "learning_rate": 1e-06, |
| "loss": 0.0105, |
| "num_tokens": 147791241.0, |
| "reward": -8.852140951156617, |
| "reward_std": 6.015402030944824, |
| "rewards/ADERawReward/mean": -8.904588794708252, |
| "rewards/ADERawReward/std": 6.00461745262146, |
| "rewards/StrictFormatReward/mean": 0.5244791731238365, |
| "rewards/StrictFormatReward/std": 2.198140048980713, |
| "sampling/importance_sampling_ratio/max": 2.8983518362045286, |
| "sampling/importance_sampling_ratio/mean": 0.38104148209095, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.405174565315247, |
| "sampling/sampling_logp_difference/mean": 0.026203346997499467, |
| "step": 330, |
| "step_time": 30.943252457803464 |
| }, |
| { |
| "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": 0.013541666883975267, |
| "completions/max_length": 1024.0, |
| "completions/max_terminated_length": 910.4, |
| "completions/mean_length": 746.2218994140625, |
| "completions/mean_terminated_length": 742.4164428710938, |
| "completions/min_length": 451.8, |
| "completions/min_terminated_length": 451.8, |
| "entropy": 0.7482379992802938, |
| "epoch": 0.05828904508829076, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.2490234375, |
| "learning_rate": 1e-06, |
| "loss": -0.0044, |
| "num_tokens": 152271555.0, |
| "reward": -9.297206592559814, |
| "reward_std": 5.522917318344116, |
| "rewards/ADERawReward/mean": -9.338768911361694, |
| "rewards/ADERawReward/std": 5.507352113723755, |
| "rewards/StrictFormatReward/mean": 0.4156250014901161, |
| "rewards/StrictFormatReward/std": 2.437586045265198, |
| "sampling/importance_sampling_ratio/max": 2.782149076461792, |
| "sampling/importance_sampling_ratio/mean": 0.37727462947368623, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.424329519271851, |
| "sampling/sampling_logp_difference/mean": 0.0264679953455925, |
| "step": 340, |
| "step_time": 31.394037494297663 |
| }, |
| { |
| "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": 0.012500000232830644, |
| "completions/max_length": 1024.0, |
| "completions/max_terminated_length": 929.2, |
| "completions/mean_length": 747.1989868164062, |
| "completions/mean_terminated_length": 743.6892578125, |
| "completions/min_length": 461.7, |
| "completions/min_terminated_length": 461.7, |
| "entropy": 0.7526971201101939, |
| "epoch": 0.06000342876735813, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.373046875, |
| "learning_rate": 1e-06, |
| "loss": -0.0279, |
| "num_tokens": 156753761.0, |
| "reward": -8.87293882369995, |
| "reward_std": 5.407143402099609, |
| "rewards/ADERawReward/mean": -8.912782192230225, |
| "rewards/ADERawReward/std": 5.394438552856445, |
| "rewards/StrictFormatReward/mean": 0.3984375029802322, |
| "rewards/StrictFormatReward/std": 2.4990755558013915, |
| "sampling/importance_sampling_ratio/max": 2.7867547035217286, |
| "sampling/importance_sampling_ratio/mean": 0.3730757534503937, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.122068333625793, |
| "sampling/sampling_logp_difference/mean": 0.02654006313532591, |
| "step": 350, |
| "step_time": 31.515721038298217 |
| }, |
| { |
| "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": 0.012500000325962901, |
| "completions/max_length": 1007.6, |
| "completions/max_terminated_length": 894.1, |
| "completions/mean_length": 747.2932495117187, |
| "completions/mean_terminated_length": 743.7852661132813, |
| "completions/min_length": 591.8, |
| "completions/min_terminated_length": 591.8, |
| "entropy": 0.7486960550149282, |
| "epoch": 0.06171781244642551, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.2373046875, |
| "learning_rate": 1e-06, |
| "loss": -0.0224, |
| "num_tokens": 161235908.0, |
| "reward": -9.346630477905274, |
| "reward_std": 5.943923234939575, |
| "rewards/ADERawReward/mean": -9.397359657287598, |
| "rewards/ADERawReward/std": 5.933114051818848, |
| "rewards/StrictFormatReward/mean": 0.5072916746139526, |
| "rewards/StrictFormatReward/std": 2.258626651763916, |
| "sampling/importance_sampling_ratio/max": 2.783932328224182, |
| "sampling/importance_sampling_ratio/mean": 0.3899604082107544, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.8976672172546385, |
| "sampling/sampling_logp_difference/mean": 0.02663377095013857, |
| "step": 360, |
| "step_time": 31.072972166599357 |
| }, |
| { |
| "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": 0.010937500232830643, |
| "completions/max_length": 1024.0, |
| "completions/max_terminated_length": 907.6, |
| "completions/mean_length": 743.9771057128906, |
| "completions/mean_terminated_length": 740.8768371582031, |
| "completions/min_length": 373.6, |
| "completions/min_terminated_length": 373.6, |
| "entropy": 0.7463085273901622, |
| "epoch": 0.06343219612549289, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.43359375, |
| "learning_rate": 1e-06, |
| "loss": -0.0195, |
| "num_tokens": 165711768.0, |
| "reward": -9.58823890686035, |
| "reward_std": 7.584380769729615, |
| "rewards/ADERawReward/mean": -9.62750973701477, |
| "rewards/ADERawReward/std": 7.566325521469116, |
| "rewards/StrictFormatReward/mean": 0.3927083402872086, |
| "rewards/StrictFormatReward/std": 2.502521538734436, |
| "sampling/importance_sampling_ratio/max": 2.824156880378723, |
| "sampling/importance_sampling_ratio/mean": 0.38374905586242675, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.386010980606079, |
| "sampling/sampling_logp_difference/mean": 0.02665374930948019, |
| "step": 370, |
| "step_time": 31.397882584496983 |
| }, |
| { |
| "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": 0.009895833674818277, |
| "completions/max_length": 1000.8, |
| "completions/max_terminated_length": 913.6, |
| "completions/mean_length": 746.1416870117188, |
| "completions/mean_terminated_length": 743.3394592285156, |
| "completions/min_length": 440.4, |
| "completions/min_terminated_length": 440.4, |
| "entropy": 0.7471979439258576, |
| "epoch": 0.06514657980456026, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.33203125, |
| "learning_rate": 1e-06, |
| "loss": 0.0032, |
| "num_tokens": 170192072.0, |
| "reward": -9.293725442886352, |
| "reward_std": 5.743723678588867, |
| "rewards/ADERawReward/mean": -9.341016817092896, |
| "rewards/ADERawReward/std": 5.729398393630982, |
| "rewards/StrictFormatReward/mean": 0.4729166701436043, |
| "rewards/StrictFormatReward/std": 2.320163094997406, |
| "sampling/importance_sampling_ratio/max": 2.869658589363098, |
| "sampling/importance_sampling_ratio/mean": 0.3773952782154083, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.138100743293762, |
| "sampling/sampling_logp_difference/mean": 0.026742291823029517, |
| "step": 380, |
| "step_time": 30.96988229679846 |
| }, |
| { |
| "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": 0.012500000325962901, |
| "completions/max_length": 1016.4, |
| "completions/max_terminated_length": 914.7, |
| "completions/mean_length": 748.4885559082031, |
| "completions/mean_terminated_length": 744.993017578125, |
| "completions/min_length": 477.5, |
| "completions/min_terminated_length": 477.5, |
| "entropy": 0.7559548437595367, |
| "epoch": 0.06686096348362763, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.24609375, |
| "learning_rate": 1e-06, |
| "loss": -0.0181, |
| "num_tokens": 174677202.0, |
| "reward": -9.28088207244873, |
| "reward_std": 5.592328834533691, |
| "rewards/ADERawReward/mean": -9.324163150787353, |
| "rewards/ADERawReward/std": 5.577477645874024, |
| "rewards/StrictFormatReward/mean": 0.43281249795109034, |
| "rewards/StrictFormatReward/std": 2.4112179040908814, |
| "sampling/importance_sampling_ratio/max": 2.825889301300049, |
| "sampling/importance_sampling_ratio/mean": 0.3906464457511902, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 3.902947115898132, |
| "sampling/sampling_logp_difference/mean": 0.026762423664331438, |
| "step": 390, |
| "step_time": 31.20840191419411 |
| }, |
| { |
| "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": 0.007812500186264514, |
| "completions/max_length": 992.4, |
| "completions/max_terminated_length": 934.3, |
| "completions/mean_length": 745.6031433105469, |
| "completions/mean_terminated_length": 743.4128295898438, |
| "completions/min_length": 437.2, |
| "completions/min_terminated_length": 437.2, |
| "entropy": 0.7531329174836476, |
| "epoch": 0.068575347162695, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.376953125, |
| "learning_rate": 1e-06, |
| "loss": 0.008, |
| "num_tokens": 179156696.0, |
| "reward": -9.488264322280884, |
| "reward_std": 5.980498790740967, |
| "rewards/ADERawReward/mean": -9.5326913356781, |
| "rewards/ADERawReward/std": 5.96862416267395, |
| "rewards/StrictFormatReward/mean": 0.44427084028720853, |
| "rewards/StrictFormatReward/std": 2.4021025896072388, |
| "sampling/importance_sampling_ratio/max": 2.796902632713318, |
| "sampling/importance_sampling_ratio/mean": 0.38265294432640073, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.037104201316834, |
| "sampling/sampling_logp_difference/mean": 0.026842619478702544, |
| "step": 400, |
| "step_time": 30.975298491402647 |
| }, |
| { |
| "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": 0.008854166977107525, |
| "completions/max_length": 1020.0, |
| "completions/max_terminated_length": 935.6, |
| "completions/mean_length": 746.836474609375, |
| "completions/mean_terminated_length": 744.3670227050782, |
| "completions/min_length": 503.7, |
| "completions/min_terminated_length": 503.7, |
| "entropy": 0.7571933090686798, |
| "epoch": 0.07028973084176239, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.30859375, |
| "learning_rate": 1e-06, |
| "loss": -0.0061, |
| "num_tokens": 183639214.0, |
| "reward": -8.974328899383545, |
| "reward_std": 6.1143780708312985, |
| "rewards/ADERawReward/mean": -9.019901895523072, |
| "rewards/ADERawReward/std": 6.102041292190552, |
| "rewards/StrictFormatReward/mean": 0.45572916865348817, |
| "rewards/StrictFormatReward/std": 2.3774968147277833, |
| "sampling/importance_sampling_ratio/max": 2.655148983001709, |
| "sampling/importance_sampling_ratio/mean": 0.36126451194286346, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 3.974502754211426, |
| "sampling/sampling_logp_difference/mean": 0.027187051996588708, |
| "step": 410, |
| "step_time": 31.259745221505 |
| }, |
| { |
| "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": 0.010416666930541395, |
| "completions/max_length": 1017.3, |
| "completions/max_terminated_length": 918.2, |
| "completions/mean_length": 747.1531433105469, |
| "completions/mean_terminated_length": 744.2319213867188, |
| "completions/min_length": 565.9, |
| "completions/min_terminated_length": 565.9, |
| "entropy": 0.7566624025503794, |
| "epoch": 0.07200411452082976, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.3359375, |
| "learning_rate": 1e-06, |
| "loss": -0.0087, |
| "num_tokens": 188121876.0, |
| "reward": -8.923426342010497, |
| "reward_std": 6.084955930709839, |
| "rewards/ADERawReward/mean": -8.9724365234375, |
| "rewards/ADERawReward/std": 6.07550630569458, |
| "rewards/StrictFormatReward/mean": 0.4901041708886623, |
| "rewards/StrictFormatReward/std": 2.2838862776756286, |
| "sampling/importance_sampling_ratio/max": 2.812648606300354, |
| "sampling/importance_sampling_ratio/mean": 0.3750308662652969, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.534145450592041, |
| "sampling/sampling_logp_difference/mean": 0.026898518949747086, |
| "step": 420, |
| "step_time": 31.253269567198004 |
| }, |
| { |
| "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": 0.010416666837409138, |
| "completions/max_length": 1000.9, |
| "completions/max_terminated_length": 899.6, |
| "completions/mean_length": 744.9640930175781, |
| "completions/mean_terminated_length": 742.0363525390625, |
| "completions/min_length": 414.5, |
| "completions/min_terminated_length": 414.5, |
| "entropy": 0.7489292581876119, |
| "epoch": 0.07371849819989713, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.58203125, |
| "learning_rate": 1e-06, |
| "loss": -0.0073, |
| "num_tokens": 192599695.0, |
| "reward": -9.504335880279541, |
| "reward_std": 5.772246503829956, |
| "rewards/ADERawReward/mean": -9.548762798309326, |
| "rewards/ADERawReward/std": 5.758863925933838, |
| "rewards/StrictFormatReward/mean": 0.4442708283662796, |
| "rewards/StrictFormatReward/std": 2.37346408367157, |
| "sampling/importance_sampling_ratio/max": 2.820726418495178, |
| "sampling/importance_sampling_ratio/mean": 0.36917952001094817, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.74177827835083, |
| "sampling/sampling_logp_difference/mean": 0.026925336755812167, |
| "step": 430, |
| "step_time": 31.05574051309668 |
| }, |
| { |
| "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": 0.013541666977107524, |
| "completions/max_length": 1003.0, |
| "completions/max_terminated_length": 914.5, |
| "completions/mean_length": 746.8135620117188, |
| "completions/mean_terminated_length": 743.0150817871094, |
| "completions/min_length": 349.1, |
| "completions/min_terminated_length": 349.1, |
| "entropy": 0.7525838236014049, |
| "epoch": 0.0754328818789645, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.3359375, |
| "learning_rate": 1e-06, |
| "loss": -0.0216, |
| "num_tokens": 197081145.0, |
| "reward": -8.942778491973877, |
| "reward_std": 5.134439134597779, |
| "rewards/ADERawReward/mean": -8.988924121856689, |
| "rewards/ADERawReward/std": 5.119956731796265, |
| "rewards/StrictFormatReward/mean": 0.4614583387970924, |
| "rewards/StrictFormatReward/std": 2.3441874384880066, |
| "sampling/importance_sampling_ratio/max": 2.7575156688690186, |
| "sampling/importance_sampling_ratio/mean": 0.3831501841545105, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 6.117226243019104, |
| "sampling/sampling_logp_difference/mean": 0.02673897407948971, |
| "step": 440, |
| "step_time": 31.136345591007558 |
| }, |
| { |
| "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": 0.00989583358168602, |
| "completions/max_length": 1024.0, |
| "completions/max_terminated_length": 962.1, |
| "completions/mean_length": 746.3265747070312, |
| "completions/mean_terminated_length": 743.5626037597656, |
| "completions/min_length": 502.5, |
| "completions/min_terminated_length": 502.5, |
| "entropy": 0.7494863112767537, |
| "epoch": 0.07714726555803189, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.396484375, |
| "learning_rate": 1e-06, |
| "loss": 0.0015, |
| "num_tokens": 201562060.0, |
| "reward": -8.833820581436157, |
| "reward_std": 6.585151100158692, |
| "rewards/ADERawReward/mean": -8.883976650238036, |
| "rewards/ADERawReward/std": 6.576332855224609, |
| "rewards/StrictFormatReward/mean": 0.5015625029802322, |
| "rewards/StrictFormatReward/std": 2.246139848232269, |
| "sampling/importance_sampling_ratio/max": 2.858155703544617, |
| "sampling/importance_sampling_ratio/mean": 0.38457230627536776, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.555598568916321, |
| "sampling/sampling_logp_difference/mean": 0.026770622283220292, |
| "step": 450, |
| "step_time": 31.314683586404136 |
| }, |
| { |
| "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": 0.010937500325962902, |
| "completions/max_length": 1012.7, |
| "completions/max_terminated_length": 908.7, |
| "completions/mean_length": 747.1239807128907, |
| "completions/mean_terminated_length": 744.033935546875, |
| "completions/min_length": 452.2, |
| "completions/min_terminated_length": 452.2, |
| "entropy": 0.7497527619202932, |
| "epoch": 0.07886164923709926, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.2392578125, |
| "learning_rate": 1e-06, |
| "loss": -0.0312, |
| "num_tokens": 206044186.0, |
| "reward": -8.526823663711548, |
| "reward_std": 4.98410964012146, |
| "rewards/ADERawReward/mean": -8.568385744094849, |
| "rewards/ADERawReward/std": 4.961003923416138, |
| "rewards/StrictFormatReward/mean": 0.415625, |
| "rewards/StrictFormatReward/std": 2.4348717212677, |
| "sampling/importance_sampling_ratio/max": 2.7420751094818114, |
| "sampling/importance_sampling_ratio/mean": 0.3720589101314545, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.188339471817017, |
| "sampling/sampling_logp_difference/mean": 0.026370251737535, |
| "step": 460, |
| "step_time": 31.211517236700455 |
| }, |
| { |
| "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": 0.010416666837409138, |
| "completions/max_length": 1013.2, |
| "completions/max_terminated_length": 913.8, |
| "completions/mean_length": 747.784912109375, |
| "completions/mean_terminated_length": 744.8851257324219, |
| "completions/min_length": 416.8, |
| "completions/min_terminated_length": 416.8, |
| "entropy": 0.7524498959382375, |
| "epoch": 0.08057603291616663, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.349609375, |
| "learning_rate": 1e-06, |
| "loss": -0.0088, |
| "num_tokens": 210527581.0, |
| "reward": -9.329790830612183, |
| "reward_std": 5.533093023300171, |
| "rewards/ADERawReward/mean": -9.378228092193604, |
| "rewards/ADERawReward/std": 5.5270590543746945, |
| "rewards/StrictFormatReward/mean": 0.4843750029802322, |
| "rewards/StrictFormatReward/std": 2.30391206741333, |
| "sampling/importance_sampling_ratio/max": 2.7844461679458616, |
| "sampling/importance_sampling_ratio/mean": 0.36996753215789796, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.251289129257202, |
| "sampling/sampling_logp_difference/mean": 0.026920023374259472, |
| "step": 470, |
| "step_time": 31.291990611601797 |
| }, |
| { |
| "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": 0.008333333535119891, |
| "completions/max_length": 1006.7, |
| "completions/max_terminated_length": 924.5, |
| "completions/mean_length": 744.3693054199218, |
| "completions/mean_terminated_length": 742.0288696289062, |
| "completions/min_length": 345.2, |
| "completions/min_terminated_length": 345.2, |
| "entropy": 0.7420653978983561, |
| "epoch": 0.08229041659523402, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.25390625, |
| "learning_rate": 1e-06, |
| "loss": -0.0135, |
| "num_tokens": 215005026.0, |
| "reward": -8.811880826950073, |
| "reward_std": 5.741937494277954, |
| "rewards/ADERawReward/mean": -8.863182735443115, |
| "rewards/ADERawReward/std": 5.721714019775391, |
| "rewards/StrictFormatReward/mean": 0.5130208387970925, |
| "rewards/StrictFormatReward/std": 2.2228100419044496, |
| "sampling/importance_sampling_ratio/max": 2.887178134918213, |
| "sampling/importance_sampling_ratio/mean": 0.3799948424100876, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.423047304153442, |
| "sampling/sampling_logp_difference/mean": 0.026753966324031354, |
| "step": 480, |
| "step_time": 31.235763939598108 |
| }, |
| { |
| "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": 0.007812500232830644, |
| "completions/max_length": 1007.4, |
| "completions/max_terminated_length": 921.0, |
| "completions/mean_length": 745.5041809082031, |
| "completions/mean_terminated_length": 743.31552734375, |
| "completions/min_length": 503.1, |
| "completions/min_terminated_length": 503.1, |
| "entropy": 0.7548379202683767, |
| "epoch": 0.08400480027430139, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.302734375, |
| "learning_rate": 1e-06, |
| "loss": 0.0051, |
| "num_tokens": 219484362.0, |
| "reward": -8.726992988586426, |
| "reward_std": 5.2785539627075195, |
| "rewards/ADERawReward/mean": -8.783451509475707, |
| "rewards/ADERawReward/std": 5.263514995574951, |
| "rewards/StrictFormatReward/mean": 0.5645833373069763, |
| "rewards/StrictFormatReward/std": 2.121866428852081, |
| "sampling/importance_sampling_ratio/max": 2.8357564210891724, |
| "sampling/importance_sampling_ratio/mean": 0.36869050562381744, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.648310852050781, |
| "sampling/sampling_logp_difference/mean": 0.026755044236779213, |
| "step": 490, |
| "step_time": 31.2714247025011 |
| }, |
| { |
| "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": 0.010416666930541395, |
| "completions/max_length": 1011.9, |
| "completions/max_terminated_length": 899.3, |
| "completions/mean_length": 745.266162109375, |
| "completions/mean_terminated_length": 742.3389892578125, |
| "completions/min_length": 504.1, |
| "completions/min_terminated_length": 504.1, |
| "entropy": 0.7439370552698771, |
| "epoch": 0.08571918395336876, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.337890625, |
| "learning_rate": 1e-06, |
| "loss": -0.0111, |
| "num_tokens": 223962569.0, |
| "reward": -8.893217849731446, |
| "reward_std": 6.001188564300537, |
| "rewards/ADERawReward/mean": -8.93993616104126, |
| "rewards/ADERawReward/std": 5.9909703731536865, |
| "rewards/StrictFormatReward/mean": 0.46718750298023226, |
| "rewards/StrictFormatReward/std": 2.34447557926178, |
| "sampling/importance_sampling_ratio/max": 2.847609305381775, |
| "sampling/importance_sampling_ratio/mean": 0.4031797587871552, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.063726282119751, |
| "sampling/sampling_logp_difference/mean": 0.02628465835005045, |
| "step": 500, |
| "step_time": 31.151053833100015 |
| }, |
| { |
| "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": 0.008854166837409139, |
| "completions/max_length": 1009.8, |
| "completions/max_terminated_length": 904.5, |
| "completions/mean_length": 745.569287109375, |
| "completions/mean_terminated_length": 743.0883544921875, |
| "completions/min_length": 499.8, |
| "completions/min_terminated_length": 499.8, |
| "entropy": 0.7393844624360403, |
| "epoch": 0.08743356763243613, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.345703125, |
| "learning_rate": 1e-06, |
| "loss": -0.003, |
| "num_tokens": 228442430.0, |
| "reward": -8.746051263809203, |
| "reward_std": 5.200268793106079, |
| "rewards/ADERawReward/mean": -8.796780586242676, |
| "rewards/ADERawReward/std": 5.189048337936401, |
| "rewards/StrictFormatReward/mean": 0.5072916716337204, |
| "rewards/StrictFormatReward/std": 2.2304036617279053, |
| "sampling/importance_sampling_ratio/max": 2.8343311309814454, |
| "sampling/importance_sampling_ratio/mean": 0.3753476530313492, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 3.841508960723877, |
| "sampling/sampling_logp_difference/mean": 0.026619693823158742, |
| "step": 510, |
| "step_time": 31.27263153729873 |
| }, |
| { |
| "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": 0.007291666837409139, |
| "completions/max_length": 987.4, |
| "completions/max_terminated_length": 894.7, |
| "completions/mean_length": 743.181787109375, |
| "completions/mean_terminated_length": 741.12255859375, |
| "completions/min_length": 500.0, |
| "completions/min_terminated_length": 500.0, |
| "entropy": 0.7444122533003489, |
| "epoch": 0.08914795131150352, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.361328125, |
| "learning_rate": 1e-06, |
| "loss": 0.0089, |
| "num_tokens": 232916635.0, |
| "reward": -8.511340045928955, |
| "reward_std": 4.909872698783874, |
| "rewards/ADERawReward/mean": -8.566652154922485, |
| "rewards/ADERawReward/std": 4.893595862388611, |
| "rewards/StrictFormatReward/mean": 0.5531250022351741, |
| "rewards/StrictFormatReward/std": 2.0853070855140685, |
| "sampling/importance_sampling_ratio/max": 2.862236833572388, |
| "sampling/importance_sampling_ratio/mean": 0.38686116933822634, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.322235846519471, |
| "sampling/sampling_logp_difference/mean": 0.026448269747197627, |
| "step": 520, |
| "step_time": 30.78332566640311 |
| }, |
| { |
| "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": 0.00677083358168602, |
| "completions/max_length": 991.6, |
| "completions/max_terminated_length": 919.4, |
| "completions/mean_length": 747.3963806152344, |
| "completions/mean_terminated_length": 745.5401245117188, |
| "completions/min_length": 380.7, |
| "completions/min_terminated_length": 380.7, |
| "entropy": 0.7540235598882039, |
| "epoch": 0.09086233499057089, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.322265625, |
| "learning_rate": 1e-06, |
| "loss": -0.0113, |
| "num_tokens": 237399348.0, |
| "reward": -8.865944004058838, |
| "reward_std": 5.564558506011963, |
| "rewards/ADERawReward/mean": -8.92240219116211, |
| "rewards/ADERawReward/std": 5.551964378356933, |
| "rewards/StrictFormatReward/mean": 0.5645833313465118, |
| "rewards/StrictFormatReward/std": 2.112958538532257, |
| "sampling/importance_sampling_ratio/max": 2.7960949182510375, |
| "sampling/importance_sampling_ratio/mean": 0.36649490892887115, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 3.830978512763977, |
| "sampling/sampling_logp_difference/mean": 0.02695600502192974, |
| "step": 530, |
| "step_time": 30.918909230499413 |
| }, |
| { |
| "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": 0.006250000186264515, |
| "completions/max_length": 1024.0, |
| "completions/max_terminated_length": 911.9, |
| "completions/mean_length": 744.5010681152344, |
| "completions/mean_terminated_length": 742.7408508300781, |
| "completions/min_length": 418.2, |
| "completions/min_terminated_length": 418.2, |
| "entropy": 0.7437257766723633, |
| "epoch": 0.09257671866963826, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.28515625, |
| "learning_rate": 1e-06, |
| "loss": -0.0034, |
| "num_tokens": 241877222.0, |
| "reward": -8.705044603347778, |
| "reward_std": 4.9693234920501705, |
| "rewards/ADERawReward/mean": -8.75405502319336, |
| "rewards/ADERawReward/std": 4.961550354957581, |
| "rewards/StrictFormatReward/mean": 0.490104167163372, |
| "rewards/StrictFormatReward/std": 2.2909578323364257, |
| "sampling/importance_sampling_ratio/max": 2.806812071800232, |
| "sampling/importance_sampling_ratio/mean": 0.38454858362674715, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.238021278381348, |
| "sampling/sampling_logp_difference/mean": 0.026717523485422133, |
| "step": 540, |
| "step_time": 31.30855839920405 |
| }, |
| { |
| "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": 0.01250000037252903, |
| "completions/max_length": 1024.0, |
| "completions/max_terminated_length": 931.3, |
| "completions/mean_length": 749.8541870117188, |
| "completions/mean_terminated_length": 746.3853088378906, |
| "completions/min_length": 535.8, |
| "completions/min_terminated_length": 535.8, |
| "entropy": 0.7400518854459127, |
| "epoch": 0.09429110234870564, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.4609375, |
| "learning_rate": 1e-06, |
| "loss": 0.0011, |
| "num_tokens": 246365646.0, |
| "reward": -8.35037841796875, |
| "reward_std": 5.094221115112305, |
| "rewards/ADERawReward/mean": -8.400534439086915, |
| "rewards/ADERawReward/std": 5.078604912757873, |
| "rewards/StrictFormatReward/mean": 0.5015625029802322, |
| "rewards/StrictFormatReward/std": 2.2548654079437256, |
| "sampling/importance_sampling_ratio/max": 2.729209136962891, |
| "sampling/importance_sampling_ratio/mean": 0.3943050533533096, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.494288897514343, |
| "sampling/sampling_logp_difference/mean": 0.026609367132186888, |
| "step": 550, |
| "step_time": 31.5098136007 |
| }, |
| { |
| "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": 0.011979166930541397, |
| "completions/max_length": 1011.1, |
| "completions/max_terminated_length": 901.4, |
| "completions/mean_length": 746.1073059082031, |
| "completions/mean_terminated_length": 742.7442687988281, |
| "completions/min_length": 385.7, |
| "completions/min_terminated_length": 385.7, |
| "entropy": 0.7511671622594197, |
| "epoch": 0.09600548602777302, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.421875, |
| "learning_rate": 1e-06, |
| "loss": -0.0039, |
| "num_tokens": 250846300.0, |
| "reward": -8.354753160476685, |
| "reward_std": 4.998448300361633, |
| "rewards/ADERawReward/mean": -8.403190279006958, |
| "rewards/ADERawReward/std": 4.983683443069458, |
| "rewards/StrictFormatReward/mean": 0.484375, |
| "rewards/StrictFormatReward/std": 2.294073450565338, |
| "sampling/importance_sampling_ratio/max": 2.7630903482437135, |
| "sampling/importance_sampling_ratio/mean": 0.371540492773056, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.54563274383545, |
| "sampling/sampling_logp_difference/mean": 0.026747452840209008, |
| "step": 560, |
| "step_time": 31.339443079804187 |
| }, |
| { |
| "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": 0.006770833395421505, |
| "completions/max_length": 1006.3, |
| "completions/max_terminated_length": 919.6, |
| "completions/mean_length": 746.6364807128906, |
| "completions/mean_terminated_length": 744.7229919433594, |
| "completions/min_length": 485.7, |
| "completions/min_terminated_length": 485.7, |
| "entropy": 0.7559505820274353, |
| "epoch": 0.09771986970684039, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.298828125, |
| "learning_rate": 1e-06, |
| "loss": -0.022, |
| "num_tokens": 255327154.0, |
| "reward": -9.671938514709472, |
| "reward_std": 6.359064817428589, |
| "rewards/ADERawReward/mean": -9.72152156829834, |
| "rewards/ADERawReward/std": 6.352941226959229, |
| "rewards/StrictFormatReward/mean": 0.49583333283662795, |
| "rewards/StrictFormatReward/std": 2.2696222186088564, |
| "sampling/importance_sampling_ratio/max": 2.8152915716171263, |
| "sampling/importance_sampling_ratio/mean": 0.37120532989501953, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 3.6147093534469605, |
| "sampling/sampling_logp_difference/mean": 0.027104974165558814, |
| "step": 570, |
| "step_time": 31.2074279195047 |
| }, |
| { |
| "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": 0.009375000093132257, |
| "completions/max_length": 989.5, |
| "completions/max_terminated_length": 893.7, |
| "completions/mean_length": 747.6479309082031, |
| "completions/mean_terminated_length": 745.0316833496094, |
| "completions/min_length": 484.2, |
| "completions/min_terminated_length": 484.2, |
| "entropy": 0.7506202618281047, |
| "epoch": 0.09943425338590776, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.34765625, |
| "learning_rate": 1e-06, |
| "loss": -0.0002, |
| "num_tokens": 259810670.0, |
| "reward": -8.481088829040527, |
| "reward_std": 5.200103354454041, |
| "rewards/ADERawReward/mean": -8.532390689849853, |
| "rewards/ADERawReward/std": 5.18415892124176, |
| "rewards/StrictFormatReward/mean": 0.5130208402872085, |
| "rewards/StrictFormatReward/std": 2.2208899974823, |
| "sampling/importance_sampling_ratio/max": 2.803302502632141, |
| "sampling/importance_sampling_ratio/mean": 0.4020031362771988, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 5.206298041343689, |
| "sampling/sampling_logp_difference/mean": 0.02699448149651289, |
| "step": 580, |
| "step_time": 31.03782037871715 |
| }, |
| { |
| "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": 0.007812500093132257, |
| "completions/max_length": 998.4, |
| "completions/max_terminated_length": 910.7, |
| "completions/mean_length": 747.5250183105469, |
| "completions/mean_terminated_length": 745.33720703125, |
| "completions/min_length": 399.1, |
| "completions/min_terminated_length": 399.1, |
| "entropy": 0.7567769944667816, |
| "epoch": 0.10114863706497514, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.28125, |
| "learning_rate": 1e-06, |
| "loss": -0.0121, |
| "num_tokens": 264294318.0, |
| "reward": -8.613714408874511, |
| "reward_std": 5.104084444046021, |
| "rewards/ADERawReward/mean": -8.672464275360108, |
| "rewards/ADERawReward/std": 5.086045837402343, |
| "rewards/StrictFormatReward/mean": 0.5874999970197677, |
| "rewards/StrictFormatReward/std": 2.0732890486717226, |
| "sampling/importance_sampling_ratio/max": 2.7467212677001953, |
| "sampling/importance_sampling_ratio/mean": 0.36461472511291504, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.379195928573608, |
| "sampling/sampling_logp_difference/mean": 0.027049308083951472, |
| "step": 590, |
| "step_time": 31.140472374291857 |
| }, |
| { |
| "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": 0.0062500000931322575, |
| "completions/max_length": 992.1, |
| "completions/max_terminated_length": 916.2, |
| "completions/mean_length": 745.9307434082032, |
| "completions/mean_terminated_length": 744.1726928710938, |
| "completions/min_length": 395.7, |
| "completions/min_terminated_length": 395.7, |
| "entropy": 0.7494464596112569, |
| "epoch": 0.10286302074404252, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.283203125, |
| "learning_rate": 1e-06, |
| "loss": 0.0102, |
| "num_tokens": 268774665.0, |
| "reward": -8.773472213745118, |
| "reward_std": 5.6750462532043455, |
| "rewards/ADERawReward/mean": -8.832221984863281, |
| "rewards/ADERawReward/std": 5.654839897155762, |
| "rewards/StrictFormatReward/mean": 0.5875000059604645, |
| "rewards/StrictFormatReward/std": 2.0246779561042785, |
| "sampling/importance_sampling_ratio/max": 2.8275955438613893, |
| "sampling/importance_sampling_ratio/mean": 0.3867086052894592, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.084816575050354, |
| "sampling/sampling_logp_difference/mean": 0.027156472206115723, |
| "step": 600, |
| "step_time": 31.06699350780109 |
| }, |
| { |
| "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": 0.0057291668374091385, |
| "completions/max_length": 1017.1, |
| "completions/max_terminated_length": 910.3, |
| "completions/mean_length": 747.0932495117188, |
| "completions/mean_terminated_length": 745.496240234375, |
| "completions/min_length": 485.1, |
| "completions/min_terminated_length": 485.1, |
| "entropy": 0.7526243805885315, |
| "epoch": 0.1045774044231099, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.396484375, |
| "learning_rate": 1e-06, |
| "loss": -0.0151, |
| "num_tokens": 273256972.0, |
| "reward": -8.731144094467163, |
| "reward_std": 5.71333794593811, |
| "rewards/ADERawReward/mean": -8.782446098327636, |
| "rewards/ADERawReward/std": 5.699142122268677, |
| "rewards/StrictFormatReward/mean": 0.5130208373069763, |
| "rewards/StrictFormatReward/std": 2.253889012336731, |
| "sampling/importance_sampling_ratio/max": 2.8028954029083253, |
| "sampling/importance_sampling_ratio/mean": 0.3589755445718765, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.905529975891113, |
| "sampling/sampling_logp_difference/mean": 0.02720630671828985, |
| "step": 610, |
| "step_time": 31.30127082870167 |
| }, |
| { |
| "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": 0.007291666837409139, |
| "completions/max_length": 986.2, |
| "completions/max_terminated_length": 908.3, |
| "completions/mean_length": 746.2526245117188, |
| "completions/mean_terminated_length": 744.2300537109375, |
| "completions/min_length": 523.4, |
| "completions/min_terminated_length": 523.4, |
| "entropy": 0.7530896683533986, |
| "epoch": 0.10629178810217726, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.46875, |
| "learning_rate": 1e-06, |
| "loss": -0.0115, |
| "num_tokens": 277737025.0, |
| "reward": -8.564061069488526, |
| "reward_std": 4.777447652816773, |
| "rewards/ADERawReward/mean": -8.621665000915527, |
| "rewards/ADERawReward/std": 4.770789432525635, |
| "rewards/StrictFormatReward/mean": 0.5760416656732559, |
| "rewards/StrictFormatReward/std": 2.0861072897911073, |
| "sampling/importance_sampling_ratio/max": 2.810882544517517, |
| "sampling/importance_sampling_ratio/mean": 0.39588436484336853, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.1026554822921755, |
| "sampling/sampling_logp_difference/mean": 0.02695068046450615, |
| "step": 620, |
| "step_time": 30.90586839320604 |
| }, |
| { |
| "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": 0.007291666837409139, |
| "completions/max_length": 988.7, |
| "completions/max_terminated_length": 898.0, |
| "completions/mean_length": 748.3562683105469, |
| "completions/mean_terminated_length": 746.3339233398438, |
| "completions/min_length": 629.4, |
| "completions/min_terminated_length": 629.4, |
| "entropy": 0.7435857594013214, |
| "epoch": 0.10800617178124464, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.25390625, |
| "learning_rate": 1e-06, |
| "loss": -0.0193, |
| "num_tokens": 282222253.0, |
| "reward": -9.023520565032959, |
| "reward_std": 5.119320058822632, |
| "rewards/ADERawReward/mean": -9.07825984954834, |
| "rewards/ADERawReward/std": 5.104773378372192, |
| "rewards/StrictFormatReward/mean": 0.5473958343267441, |
| "rewards/StrictFormatReward/std": 2.160345029830933, |
| "sampling/importance_sampling_ratio/max": 2.8361830711364746, |
| "sampling/importance_sampling_ratio/mean": 0.37177495658397675, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.693781447410584, |
| "sampling/sampling_logp_difference/mean": 0.027231728471815587, |
| "step": 630, |
| "step_time": 30.870413991407258 |
| }, |
| { |
| "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": 0.004687500139698386, |
| "completions/max_length": 971.6, |
| "completions/max_terminated_length": 889.4, |
| "completions/mean_length": 747.0187744140625, |
| "completions/mean_terminated_length": 745.7044677734375, |
| "completions/min_length": 558.4, |
| "completions/min_terminated_length": 558.4, |
| "entropy": 0.7533182700475057, |
| "epoch": 0.10972055546031202, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.34375, |
| "learning_rate": 1e-06, |
| "loss": 0.0057, |
| "num_tokens": 286704369.0, |
| "reward": -9.217678833007813, |
| "reward_std": 5.054318857192993, |
| "rewards/ADERawReward/mean": -9.277001667022706, |
| "rewards/ADERawReward/std": 5.038924503326416, |
| "rewards/StrictFormatReward/mean": 0.5932291626930237, |
| "rewards/StrictFormatReward/std": 2.0508111119270325, |
| "sampling/importance_sampling_ratio/max": 2.866812252998352, |
| "sampling/importance_sampling_ratio/mean": 0.3646788477897644, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 3.960925984382629, |
| "sampling/sampling_logp_difference/mean": 0.027474281564354896, |
| "step": 640, |
| "step_time": 30.600575475511143 |
| }, |
| { |
| "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": 0.003645833395421505, |
| "completions/max_length": 960.3, |
| "completions/max_terminated_length": 891.2, |
| "completions/mean_length": 742.1609497070312, |
| "completions/mean_terminated_length": 741.1292663574219, |
| "completions/min_length": 366.8, |
| "completions/min_terminated_length": 366.8, |
| "entropy": 0.7416558722654979, |
| "epoch": 0.1114349391393794, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.31640625, |
| "learning_rate": 1e-06, |
| "loss": -0.0118, |
| "num_tokens": 291177734.0, |
| "reward": -8.407093048095703, |
| "reward_std": 5.326395010948181, |
| "rewards/ADERawReward/mean": -8.467561674118041, |
| "rewards/ADERawReward/std": 5.315518450737, |
| "rewards/StrictFormatReward/mean": 0.6046874970197678, |
| "rewards/StrictFormatReward/std": 2.026141095161438, |
| "sampling/importance_sampling_ratio/max": 2.862113666534424, |
| "sampling/importance_sampling_ratio/mean": 0.3777022361755371, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.170439219474792, |
| "sampling/sampling_logp_difference/mean": 0.027010084502398966, |
| "step": 650, |
| "step_time": 30.520393895095914 |
| }, |
| { |
| "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": 0.010937500279396772, |
| "completions/max_length": 1009.0, |
| "completions/max_terminated_length": 908.3, |
| "completions/mean_length": 746.7343872070312, |
| "completions/mean_terminated_length": 743.6657897949219, |
| "completions/min_length": 455.1, |
| "completions/min_terminated_length": 455.1, |
| "entropy": 0.748453684647878, |
| "epoch": 0.11314932281844677, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.220703125, |
| "learning_rate": 1e-06, |
| "loss": -0.0029, |
| "num_tokens": 295659304.0, |
| "reward": -8.47295446395874, |
| "reward_std": 5.315748786926269, |
| "rewards/ADERawReward/mean": -8.525974988937378, |
| "rewards/ADERawReward/std": 5.30073721408844, |
| "rewards/StrictFormatReward/mean": 0.5302083283662796, |
| "rewards/StrictFormatReward/std": 2.2150806427001952, |
| "sampling/importance_sampling_ratio/max": 2.8308970451354982, |
| "sampling/importance_sampling_ratio/mean": 0.36592633128166197, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 3.81339545249939, |
| "sampling/sampling_logp_difference/mean": 0.027280481532216072, |
| "step": 660, |
| "step_time": 31.34898143170285 |
| }, |
| { |
| "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": 0.0031250000931322573, |
| "completions/max_length": 961.7, |
| "completions/max_terminated_length": 898.4, |
| "completions/mean_length": 745.1661743164062, |
| "completions/mean_terminated_length": 744.2929931640625, |
| "completions/min_length": 464.4, |
| "completions/min_terminated_length": 464.4, |
| "entropy": 0.7453262567520141, |
| "epoch": 0.11486370649751414, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.228515625, |
| "learning_rate": 1e-06, |
| "loss": 0.0075, |
| "num_tokens": 300137783.0, |
| "reward": -8.66427845954895, |
| "reward_std": 4.92835762500763, |
| "rewards/ADERawReward/mean": -8.727611589431763, |
| "rewards/ADERawReward/std": 4.9187664270401, |
| "rewards/StrictFormatReward/mean": 0.6333333373069763, |
| "rewards/StrictFormatReward/std": 1.9716094613075257, |
| "sampling/importance_sampling_ratio/max": 2.7208317041397097, |
| "sampling/importance_sampling_ratio/mean": 0.3644146382808685, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.467776584625244, |
| "sampling/sampling_logp_difference/mean": 0.027055931463837623, |
| "step": 670, |
| "step_time": 30.402021049396716 |
| }, |
| { |
| "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": 0.007812500139698387, |
| "completions/max_length": 1009.0, |
| "completions/max_terminated_length": 900.7, |
| "completions/mean_length": 744.5047119140625, |
| "completions/mean_terminated_length": 742.3053039550781, |
| "completions/min_length": 501.0, |
| "completions/min_terminated_length": 501.0, |
| "entropy": 0.7389722168445587, |
| "epoch": 0.11657809017658152, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.310546875, |
| "learning_rate": 1e-06, |
| "loss": 0.0105, |
| "num_tokens": 304615648.0, |
| "reward": -8.14780044555664, |
| "reward_std": 5.765257406234741, |
| "rewards/ADERawReward/mean": -8.198529577255249, |
| "rewards/ADERawReward/std": 5.752817440032959, |
| "rewards/StrictFormatReward/mean": 0.507291667163372, |
| "rewards/StrictFormatReward/std": 2.2412220239639282, |
| "sampling/importance_sampling_ratio/max": 2.838437080383301, |
| "sampling/importance_sampling_ratio/mean": 0.3927165180444717, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 3.8545892953872682, |
| "sampling/sampling_logp_difference/mean": 0.026737906597554685, |
| "step": 680, |
| "step_time": 31.265542278197245 |
| }, |
| { |
| "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": 0.007291666837409139, |
| "completions/max_length": 1011.0, |
| "completions/max_terminated_length": 918.1, |
| "completions/mean_length": 745.039599609375, |
| "completions/mean_terminated_length": 743.0080322265625, |
| "completions/min_length": 483.0, |
| "completions/min_terminated_length": 483.0, |
| "entropy": 0.7438246866067251, |
| "epoch": 0.1182924738556489, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.326171875, |
| "learning_rate": 1e-06, |
| "loss": -0.0139, |
| "num_tokens": 309093692.0, |
| "reward": -8.50984206199646, |
| "reward_std": 4.543536067008972, |
| "rewards/ADERawReward/mean": -8.562862634658813, |
| "rewards/ADERawReward/std": 4.5306751251220705, |
| "rewards/StrictFormatReward/mean": 0.5302083313465118, |
| "rewards/StrictFormatReward/std": 2.1870830774307253, |
| "sampling/importance_sampling_ratio/max": 2.8570920705795286, |
| "sampling/importance_sampling_ratio/mean": 0.3624374896287918, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.160294127464295, |
| "sampling/sampling_logp_difference/mean": 0.02724489979445934, |
| "step": 690, |
| "step_time": 31.26219962689502 |
| }, |
| { |
| "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": 0.005208333441987634, |
| "completions/max_length": 996.0, |
| "completions/max_terminated_length": 924.7, |
| "completions/mean_length": 745.0484558105469, |
| "completions/mean_terminated_length": 743.5991821289062, |
| "completions/min_length": 539.6, |
| "completions/min_terminated_length": 539.6, |
| "entropy": 0.7445198019345601, |
| "epoch": 0.12000685753471627, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.390625, |
| "learning_rate": 1e-06, |
| "loss": -0.0064, |
| "num_tokens": 313571945.0, |
| "reward": -8.856493520736695, |
| "reward_std": 4.863119053840637, |
| "rewards/ADERawReward/mean": -8.914670515060426, |
| "rewards/ADERawReward/std": 4.857007074356079, |
| "rewards/StrictFormatReward/mean": 0.5817708313465119, |
| "rewards/StrictFormatReward/std": 2.0837946057319643, |
| "sampling/importance_sampling_ratio/max": 2.8482714176177977, |
| "sampling/importance_sampling_ratio/mean": 0.3727655470371246, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.315770363807678, |
| "sampling/sampling_logp_difference/mean": 0.02694130353629589, |
| "step": 700, |
| "step_time": 31.00740701830364 |
| }, |
| { |
| "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": 0.00885416679084301, |
| "completions/max_length": 980.0, |
| "completions/max_terminated_length": 904.4, |
| "completions/mean_length": 745.5870056152344, |
| "completions/mean_terminated_length": 743.1242919921875, |
| "completions/min_length": 423.9, |
| "completions/min_terminated_length": 423.9, |
| "entropy": 0.7471017857392629, |
| "epoch": 0.12172124121378365, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.3125, |
| "learning_rate": 1e-06, |
| "loss": 0.0164, |
| "num_tokens": 318051584.0, |
| "reward": -8.626063299179076, |
| "reward_std": 5.039473915100098, |
| "rewards/ADERawReward/mean": -8.679083728790284, |
| "rewards/ADERawReward/std": 5.022432565689087, |
| "rewards/StrictFormatReward/mean": 0.5302083313465118, |
| "rewards/StrictFormatReward/std": 2.2059712767601014, |
| "sampling/importance_sampling_ratio/max": 2.6881606578826904, |
| "sampling/importance_sampling_ratio/mean": 0.35196583569049833, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 3.9412872314453127, |
| "sampling/sampling_logp_difference/mean": 0.02722361944615841, |
| "step": 710, |
| "step_time": 30.79243855099194 |
| }, |
| { |
| "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": 0.007812500093132257, |
| "completions/max_length": 1011.8, |
| "completions/max_terminated_length": 889.3, |
| "completions/mean_length": 745.9838806152344, |
| "completions/mean_terminated_length": 743.8016540527344, |
| "completions/min_length": 430.3, |
| "completions/min_terminated_length": 430.3, |
| "entropy": 0.737281721830368, |
| "epoch": 0.12343562489285102, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.333984375, |
| "learning_rate": 1e-06, |
| "loss": -0.0019, |
| "num_tokens": 322532273.0, |
| "reward": -8.676169490814209, |
| "reward_std": 6.3611784934997555, |
| "rewards/ADERawReward/mean": -8.736637830734253, |
| "rewards/ADERawReward/std": 6.346572685241699, |
| "rewards/StrictFormatReward/mean": 0.6046874940395355, |
| "rewards/StrictFormatReward/std": 2.027357840538025, |
| "sampling/importance_sampling_ratio/max": 2.7995285272598265, |
| "sampling/importance_sampling_ratio/mean": 0.35478300750255587, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.93809871673584, |
| "sampling/sampling_logp_difference/mean": 0.027037655375897883, |
| "step": 720, |
| "step_time": 31.32133076491009 |
| }, |
| { |
| "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": 0.006250000139698386, |
| "completions/max_length": 1008.3, |
| "completions/max_terminated_length": 902.0, |
| "completions/mean_length": 744.9453369140625, |
| "completions/mean_terminated_length": 743.1854309082031, |
| "completions/min_length": 398.3, |
| "completions/min_terminated_length": 398.3, |
| "entropy": 0.7388640781243642, |
| "epoch": 0.1251500085719184, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.388671875, |
| "learning_rate": 1e-06, |
| "loss": -0.0171, |
| "num_tokens": 327010056.0, |
| "reward": -8.387274026870728, |
| "reward_std": 4.668637943267822, |
| "rewards/ADERawReward/mean": -8.443159294128417, |
| "rewards/ADERawReward/std": 4.659743785858154, |
| "rewards/StrictFormatReward/mean": 0.5588541686534881, |
| "rewards/StrictFormatReward/std": 2.136683487892151, |
| "sampling/importance_sampling_ratio/max": 2.7782902002334593, |
| "sampling/importance_sampling_ratio/mean": 0.3791107714176178, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 4.789157485961914, |
| "sampling/sampling_logp_difference/mean": 0.02694005910307169, |
| "step": 730, |
| "step_time": 31.208155655500015 |
| } |
| ], |
| "logging_steps": 10, |
| "max_steps": 730, |
| "num_input_tokens_seen": 327010056, |
| "num_train_epochs": 1, |
| "save_steps": 500, |
| "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": 16, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
