| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.8191126279863481, |
| "eval_steps": 60, |
| "global_step": 120, |
| "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": 0.0416666679084301, |
| "completions/max_length": 4126.0, |
| "completions/max_terminated_length": 3978.0, |
| "completions/mean_length": 2078.7084350585938, |
| "completions/mean_terminated_length": 1958.6099243164062, |
| "completions/min_length": 962.5, |
| "completions/min_terminated_length": 962.5, |
| "entropy": 0.19557882845401764, |
| "epoch": 0.006825938566552901, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.03708447515964508, |
| "learning_rate": 0.0, |
| "loss": 0.07332071661949158, |
| "num_tokens": 86135.0, |
| "reward": -0.4166666865348816, |
| "reward_std": 0.8109081089496613, |
| "rewards/correctness_reward/mean": -0.4166666567325592, |
| "rewards/correctness_reward/std": 0.8109080791473389, |
| "sampling/importance_sampling_ratio/max": 1.7469300627708435, |
| "sampling/importance_sampling_ratio/mean": 0.6238087564706802, |
| "sampling/importance_sampling_ratio/min": 0.02465523686259985, |
| "sampling/sampling_logp_difference/max": 0.5807536840438843, |
| "sampling/sampling_logp_difference/mean": 0.009548636619001627, |
| "step": 1, |
| "step_time": 143.7903330239933 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3109.5, |
| "completions/max_terminated_length": 3109.5, |
| "completions/mean_length": 1972.4584350585938, |
| "completions/mean_terminated_length": 1972.4584350585938, |
| "completions/min_length": 942.0, |
| "completions/min_terminated_length": 942.0, |
| "entropy": 0.1991811003535986, |
| "epoch": 0.013651877133105802, |
| "frac_reward_zero_std": 1.0, |
| "grad_norm": 0.0, |
| "learning_rate": 2e-08, |
| "loss": 0.0, |
| "num_tokens": 164872.0, |
| "reward": -1.0, |
| "reward_std": 0.0, |
| "rewards/correctness_reward/mean": -1.0, |
| "rewards/correctness_reward/std": 0.0, |
| "sampling/importance_sampling_ratio/max": 2.464595675468445, |
| "sampling/importance_sampling_ratio/mean": 0.7223336100578308, |
| "sampling/importance_sampling_ratio/min": 0.04568341374397278, |
| "sampling/sampling_logp_difference/max": 1.6412757635116577, |
| "sampling/sampling_logp_difference/mean": 0.009283150546252728, |
| "step": 2, |
| "step_time": 103.18303876400023 |
| }, |
| { |
| "clip_ratio/high_max": 0.00012525788770290092, |
| "clip_ratio/high_mean": 0.00012525788770290092, |
| "clip_ratio/low_mean": 0.000619696089415811, |
| "clip_ratio/low_min": 0.000619696089415811, |
| "clip_ratio/region_mean": 0.0007449539843946695, |
| "completions/clipped_ratio": 0.125, |
| "completions/max_length": 4050.5, |
| "completions/max_terminated_length": 3202.0, |
| "completions/mean_length": 2460.9583740234375, |
| "completions/mean_terminated_length": 2179.125, |
| "completions/min_length": 1478.5, |
| "completions/min_terminated_length": 1478.5, |
| "entropy": 0.2095671109855175, |
| "epoch": 0.020477815699658702, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.12216407060623169, |
| "learning_rate": 4e-08, |
| "loss": -0.18665893375873566, |
| "num_tokens": 263763.0, |
| "reward": -0.6666666865348816, |
| "reward_std": 0.7409421801567078, |
| "rewards/correctness_reward/mean": -0.6666666567325592, |
| "rewards/correctness_reward/std": 0.7409421503543854, |
| "sampling/importance_sampling_ratio/max": 2.2802764177322388, |
| "sampling/importance_sampling_ratio/mean": 0.6775670647621155, |
| "sampling/importance_sampling_ratio/min": 0.009539297316223383, |
| "sampling/sampling_logp_difference/max": 0.8306642770767212, |
| "sampling/sampling_logp_difference/mean": 0.00935443863272667, |
| "step": 3, |
| "step_time": 135.05255382400355 |
| }, |
| { |
| "clip_ratio/high_max": 0.00012446496657503303, |
| "clip_ratio/high_mean": 0.00012446496657503303, |
| "clip_ratio/low_mean": 0.00017372676666127518, |
| "clip_ratio/low_min": 0.00017372676666127518, |
| "clip_ratio/region_mean": 0.0002981917332363082, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 3878.0, |
| "completions/max_terminated_length": 3628.0, |
| "completions/mean_length": 2135.791748046875, |
| "completions/mean_terminated_length": 2011.3409423828125, |
| "completions/min_length": 860.5, |
| "completions/min_terminated_length": 860.5, |
| "entropy": 0.1768759936094284, |
| "epoch": 0.027303754266211604, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.04616802930831909, |
| "learning_rate": 6e-08, |
| "loss": 0.047608934342861176, |
| "num_tokens": 354190.0, |
| "reward": -0.3333333358168602, |
| "reward_std": 0.9671956598758698, |
| "rewards/correctness_reward/mean": -0.3333333358168602, |
| "rewards/correctness_reward/std": 0.9671956598758698, |
| "sampling/importance_sampling_ratio/max": 1.2191235721111298, |
| "sampling/importance_sampling_ratio/mean": 0.41029319167137146, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 1.1395794749259949, |
| "sampling/sampling_logp_difference/mean": 0.008533329702913761, |
| "step": 4, |
| "step_time": 129.13586523798585 |
| }, |
| { |
| "clip_ratio/high_max": 0.00011122030264232308, |
| "clip_ratio/high_mean": 0.00011122030264232308, |
| "clip_ratio/low_mean": 0.0007960342863952974, |
| "clip_ratio/low_min": 0.0007960342863952974, |
| "clip_ratio/region_mean": 0.0009072545872186311, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 4103.0, |
| "completions/max_terminated_length": 4103.0, |
| "completions/mean_length": 2328.8333740234375, |
| "completions/mean_terminated_length": 2328.8333740234375, |
| "completions/min_length": 1169.5, |
| "completions/min_terminated_length": 1169.5, |
| "entropy": 0.18773097731173038, |
| "epoch": 0.034129692832764506, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.05336736515164375, |
| "learning_rate": 8e-08, |
| "loss": -0.271327942609787, |
| "num_tokens": 444492.0, |
| "reward": -0.5, |
| "reward_std": 0.9045340418815613, |
| "rewards/correctness_reward/mean": -0.5, |
| "rewards/correctness_reward/std": 0.9045340418815613, |
| "sampling/importance_sampling_ratio/max": 2.003176987171173, |
| "sampling/importance_sampling_ratio/mean": 0.4816400110721588, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.6100131273269653, |
| "sampling/sampling_logp_difference/mean": 0.008833470288664103, |
| "step": 5, |
| "step_time": 133.57865831398522 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 1452.0, |
| "completions/max_terminated_length": 1452.0, |
| "completions/mean_length": 986.5833740234375, |
| "completions/mean_terminated_length": 986.5833740234375, |
| "completions/min_length": 662.5, |
| "completions/min_terminated_length": 662.5, |
| "entropy": 0.1458668615669012, |
| "epoch": 0.040955631399317405, |
| "frac_reward_zero_std": 1.0, |
| "grad_norm": 0.0, |
| "learning_rate": 1e-07, |
| "loss": 0.0, |
| "num_tokens": 503954.0, |
| "reward": -1.0, |
| "reward_std": 0.0, |
| "rewards/correctness_reward/mean": -1.0, |
| "rewards/correctness_reward/std": 0.0, |
| "sampling/importance_sampling_ratio/max": 1.5658745169639587, |
| "sampling/importance_sampling_ratio/mean": 0.6347561478614807, |
| "sampling/importance_sampling_ratio/min": 0.18051937222480774, |
| "sampling/sampling_logp_difference/max": 0.647778332233429, |
| "sampling/sampling_logp_difference/mean": 0.008760722819715738, |
| "step": 6, |
| "step_time": 57.31431425399933 |
| }, |
| { |
| "clip_ratio/high_max": 4.898119004792534e-05, |
| "clip_ratio/high_mean": 4.898119004792534e-05, |
| "clip_ratio/low_mean": 0.0005076292945886962, |
| "clip_ratio/low_min": 0.0005076292945886962, |
| "clip_ratio/region_mean": 0.0005566104846366215, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3297.5, |
| "completions/max_terminated_length": 3297.5, |
| "completions/mean_length": 1440.4166870117188, |
| "completions/mean_terminated_length": 1440.4166870117188, |
| "completions/min_length": 855.0, |
| "completions/min_terminated_length": 855.0, |
| "entropy": 0.1631603855639696, |
| "epoch": 0.04778156996587031, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.10335278511047363, |
| "learning_rate": 1.2e-07, |
| "loss": -0.6942333579063416, |
| "num_tokens": 574770.0, |
| "reward": -0.9166666865348816, |
| "reward_std": 0.28867512941360474, |
| "rewards/correctness_reward/mean": -0.9166666567325592, |
| "rewards/correctness_reward/std": 0.28867512941360474, |
| "sampling/importance_sampling_ratio/max": 1.8089405596256256, |
| "sampling/importance_sampling_ratio/mean": 0.5232424139976501, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.5933345556259155, |
| "sampling/sampling_logp_difference/mean": 0.009010098408907652, |
| "step": 7, |
| "step_time": 115.51082664800197 |
| }, |
| { |
| "clip_ratio/high_max": 2.7079722713097e-05, |
| "clip_ratio/high_mean": 2.7079722713097e-05, |
| "clip_ratio/low_mean": 0.00011799410276580602, |
| "clip_ratio/low_min": 0.00011799410276580602, |
| "clip_ratio/region_mean": 0.00014507382547890302, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 3793.5, |
| "completions/max_terminated_length": 3278.5, |
| "completions/mean_length": 1910.75, |
| "completions/mean_terminated_length": 1647.050048828125, |
| "completions/min_length": 912.5, |
| "completions/min_terminated_length": 912.5, |
| "entropy": 0.18195712938904762, |
| "epoch": 0.05460750853242321, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.003391283331438899, |
| "learning_rate": 1.3999999999999998e-07, |
| "loss": 0.019645797088742256, |
| "num_tokens": 657492.0, |
| "reward": -0.8333333432674408, |
| "reward_std": 0.3892494738101959, |
| "rewards/correctness_reward/mean": -0.8333333432674408, |
| "rewards/correctness_reward/std": 0.3892495036125183, |
| "sampling/importance_sampling_ratio/max": 1.332252025604248, |
| "sampling/importance_sampling_ratio/mean": 0.46315374970436096, |
| "sampling/importance_sampling_ratio/min": 0.03832871373742819, |
| "sampling/sampling_logp_difference/max": 0.6777635812759399, |
| "sampling/sampling_logp_difference/mean": 0.00894957734271884, |
| "step": 8, |
| "step_time": 124.02444752900192 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.00037109364347998053, |
| "clip_ratio/low_min": 0.00037109364347998053, |
| "clip_ratio/region_mean": 0.00037109364347998053, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 3602.0, |
| "completions/max_terminated_length": 2851.5, |
| "completions/mean_length": 1949.0, |
| "completions/mean_terminated_length": 1830.397705078125, |
| "completions/min_length": 1087.0, |
| "completions/min_terminated_length": 1087.0, |
| "entropy": 0.2037021517753601, |
| "epoch": 0.06143344709897611, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.029773741960525513, |
| "learning_rate": 1.6e-07, |
| "loss": -0.05949871987104416, |
| "num_tokens": 749352.0, |
| "reward": -0.9166666865348816, |
| "reward_std": 0.2886751592159271, |
| "rewards/correctness_reward/mean": -0.9166666567325592, |
| "rewards/correctness_reward/std": 0.28867512941360474, |
| "sampling/importance_sampling_ratio/max": 2.4376940727233887, |
| "sampling/importance_sampling_ratio/mean": 0.5821068584918976, |
| "sampling/importance_sampling_ratio/min": 0.005494819954037666, |
| "sampling/sampling_logp_difference/max": 0.8441054224967957, |
| "sampling/sampling_logp_difference/mean": 0.00960251223295927, |
| "step": 9, |
| "step_time": 122.19422086099075 |
| }, |
| { |
| "clip_ratio/high_max": 0.0002809852703649085, |
| "clip_ratio/high_mean": 0.0002809852703649085, |
| "clip_ratio/low_mean": 0.0008867749929777347, |
| "clip_ratio/low_min": 0.0008867749929777347, |
| "clip_ratio/region_mean": 0.0011677602742565796, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3277.0, |
| "completions/max_terminated_length": 3277.0, |
| "completions/mean_length": 1764.25, |
| "completions/mean_terminated_length": 1764.25, |
| "completions/min_length": 786.5, |
| "completions/min_terminated_length": 786.5, |
| "entropy": 0.18435939773917198, |
| "epoch": 0.06825938566552901, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.032720599323511124, |
| "learning_rate": 1.8e-07, |
| "loss": 0.12537115812301636, |
| "num_tokens": 820326.0, |
| "reward": -0.583333358168602, |
| "reward_std": 0.7810410857200623, |
| "rewards/correctness_reward/mean": -0.5833333283662796, |
| "rewards/correctness_reward/std": 0.7810411155223846, |
| "sampling/importance_sampling_ratio/max": 1.6888806819915771, |
| "sampling/importance_sampling_ratio/mean": 0.5177919268608093, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.6338849067687988, |
| "sampling/sampling_logp_difference/mean": 0.00877432944253087, |
| "step": 10, |
| "step_time": 105.65176495799096 |
| }, |
| { |
| "clip_ratio/high_max": 0.00015119444287847728, |
| "clip_ratio/high_mean": 0.00015119444287847728, |
| "clip_ratio/low_mean": 0.0004296297502150992, |
| "clip_ratio/low_min": 0.0004296297502150992, |
| "clip_ratio/region_mean": 0.00058082418217964, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 3852.5, |
| "completions/max_terminated_length": 3593.0, |
| "completions/mean_length": 2337.9166870117188, |
| "completions/mean_terminated_length": 2242.7689819335938, |
| "completions/min_length": 1315.5, |
| "completions/min_terminated_length": 1315.5, |
| "entropy": 0.18981880322098732, |
| "epoch": 0.07508532423208192, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.016524910926818848, |
| "learning_rate": 2e-07, |
| "loss": 0.023730065673589706, |
| "num_tokens": 916900.0, |
| "reward": -0.1666666641831398, |
| "reward_std": 0.9671956598758698, |
| "rewards/correctness_reward/mean": -0.1666666641831398, |
| "rewards/correctness_reward/std": 0.9671956598758698, |
| "sampling/importance_sampling_ratio/max": 2.6923059225082397, |
| "sampling/importance_sampling_ratio/mean": 0.6565994620323181, |
| "sampling/importance_sampling_ratio/min": 0.0022264549043029547, |
| "sampling/sampling_logp_difference/max": 0.5726124048233032, |
| "sampling/sampling_logp_difference/mean": 0.009137257002294064, |
| "step": 11, |
| "step_time": 129.848471055986 |
| }, |
| { |
| "clip_ratio/high_max": 9.70824057731079e-05, |
| "clip_ratio/high_mean": 9.70824057731079e-05, |
| "clip_ratio/low_mean": 0.0009453010425204411, |
| "clip_ratio/low_min": 0.0009453010425204411, |
| "clip_ratio/region_mean": 0.0010423834464745596, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 4029.5, |
| "completions/max_terminated_length": 4029.5, |
| "completions/mean_length": 2146.25, |
| "completions/mean_terminated_length": 2146.25, |
| "completions/min_length": 590.5, |
| "completions/min_terminated_length": 590.5, |
| "entropy": 0.20860109478235245, |
| "epoch": 0.08191126279863481, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.06532567739486694, |
| "learning_rate": 2.1999999999999998e-07, |
| "loss": -0.1271137297153473, |
| "num_tokens": 1000354.0, |
| "reward": -0.5833333432674408, |
| "reward_std": 0.841516524553299, |
| "rewards/correctness_reward/mean": -0.5833333432674408, |
| "rewards/correctness_reward/std": 0.841516524553299, |
| "sampling/importance_sampling_ratio/max": 1.4567700028419495, |
| "sampling/importance_sampling_ratio/mean": 0.5064111053943634, |
| "sampling/importance_sampling_ratio/min": 0.006421342492103577, |
| "sampling/sampling_logp_difference/max": 0.7267114520072937, |
| "sampling/sampling_logp_difference/mean": 0.009367370512336493, |
| "step": 12, |
| "step_time": 130.4215078309935 |
| }, |
| { |
| "clip_ratio/high_max": 3.9846989238867536e-05, |
| "clip_ratio/high_mean": 3.9846989238867536e-05, |
| "clip_ratio/low_mean": 0.0003410607132536825, |
| "clip_ratio/low_min": 0.0003410607132536825, |
| "clip_ratio/region_mean": 0.00038090770249255, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 2148.0, |
| "completions/max_terminated_length": 2148.0, |
| "completions/mean_length": 1350.7916870117188, |
| "completions/mean_terminated_length": 1350.7916870117188, |
| "completions/min_length": 779.5, |
| "completions/min_terminated_length": 779.5, |
| "entropy": 0.14979218784719706, |
| "epoch": 0.08873720136518772, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.03845459595322609, |
| "learning_rate": 2.4e-07, |
| "loss": -0.04120948165655136, |
| "num_tokens": 1064921.0, |
| "reward": -0.9166666865348816, |
| "reward_std": 0.2886751592159271, |
| "rewards/correctness_reward/mean": -0.9166666567325592, |
| "rewards/correctness_reward/std": 0.28867512941360474, |
| "sampling/importance_sampling_ratio/max": 2.167421817779541, |
| "sampling/importance_sampling_ratio/mean": 0.7132795751094818, |
| "sampling/importance_sampling_ratio/min": 0.04248122125864029, |
| "sampling/sampling_logp_difference/max": 0.6888407468795776, |
| "sampling/sampling_logp_difference/mean": 0.008195365080609918, |
| "step": 13, |
| "step_time": 75.68014991000382 |
| }, |
| { |
| "clip_ratio/high_max": 0.0002453310153214261, |
| "clip_ratio/high_mean": 0.0002453310153214261, |
| "clip_ratio/low_mean": 0.00010895619925577193, |
| "clip_ratio/low_min": 0.00010895619925577193, |
| "clip_ratio/region_mean": 0.00035428721457719803, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 3994.5, |
| "completions/max_terminated_length": 3507.5, |
| "completions/mean_length": 1930.166748046875, |
| "completions/mean_terminated_length": 1651.4500122070312, |
| "completions/min_length": 822.5, |
| "completions/min_terminated_length": 822.5, |
| "entropy": 0.19265830144286156, |
| "epoch": 0.09556313993174062, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.03448331356048584, |
| "learning_rate": 2.6e-07, |
| "loss": -0.12286396324634552, |
| "num_tokens": 1147695.0, |
| "reward": -0.5833333358168602, |
| "reward_std": 0.5149286389350891, |
| "rewards/correctness_reward/mean": -0.5833333358168602, |
| "rewards/correctness_reward/std": 0.5149286389350891, |
| "sampling/importance_sampling_ratio/max": 2.6719439029693604, |
| "sampling/importance_sampling_ratio/mean": 0.6675399243831635, |
| "sampling/importance_sampling_ratio/min": 0.0005254613934084773, |
| "sampling/sampling_logp_difference/max": 0.6318219900131226, |
| "sampling/sampling_logp_difference/mean": 0.009593375492841005, |
| "step": 14, |
| "step_time": 127.86682878998545 |
| }, |
| { |
| "clip_ratio/high_max": 0.00015420510499097873, |
| "clip_ratio/high_mean": 0.00015420510499097873, |
| "clip_ratio/low_mean": 0.000891726536792703, |
| "clip_ratio/low_min": 0.000891726536792703, |
| "clip_ratio/region_mean": 0.0010459316545166075, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 3989.0, |
| "completions/max_terminated_length": 3116.0, |
| "completions/mean_length": 2372.041748046875, |
| "completions/mean_terminated_length": 2264.75, |
| "completions/min_length": 1013.0, |
| "completions/min_terminated_length": 1013.0, |
| "entropy": 0.2017787005752325, |
| "epoch": 0.10238907849829351, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.07834500819444656, |
| "learning_rate": 2.7999999999999997e-07, |
| "loss": -0.19363535940647125, |
| "num_tokens": 1237786.0, |
| "reward": -0.5000000149011612, |
| "reward_std": 0.8816154301166534, |
| "rewards/correctness_reward/mean": -0.5000000149011612, |
| "rewards/correctness_reward/std": 0.8816154599189758, |
| "sampling/importance_sampling_ratio/max": 1.4357083439826965, |
| "sampling/importance_sampling_ratio/mean": 0.4262894541025162, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.798983097076416, |
| "sampling/sampling_logp_difference/mean": 0.009441147092729807, |
| "step": 15, |
| "step_time": 131.49785820600664 |
| }, |
| { |
| "clip_ratio/high_max": 0.00020652647253882606, |
| "clip_ratio/high_mean": 0.00020652647253882606, |
| "clip_ratio/low_mean": 0.00024393484818574507, |
| "clip_ratio/low_min": 0.00024393484818574507, |
| "clip_ratio/region_mean": 0.0004504613189055817, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 3986.0, |
| "completions/max_terminated_length": 3296.0, |
| "completions/mean_length": 1959.8333740234375, |
| "completions/mean_terminated_length": 1826.901611328125, |
| "completions/min_length": 1002.0, |
| "completions/min_terminated_length": 1002.0, |
| "entropy": 0.17979588359594345, |
| "epoch": 0.10921501706484642, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.06618808954954147, |
| "learning_rate": 3e-07, |
| "loss": -0.1892964094877243, |
| "num_tokens": 1315212.0, |
| "reward": -0.8333333432674408, |
| "reward_std": 0.3892494738101959, |
| "rewards/correctness_reward/mean": -0.8333333432674408, |
| "rewards/correctness_reward/std": 0.3892495036125183, |
| "sampling/importance_sampling_ratio/max": 1.7599316835403442, |
| "sampling/importance_sampling_ratio/mean": 0.5774809271097183, |
| "sampling/importance_sampling_ratio/min": 0.07288849353790283, |
| "sampling/sampling_logp_difference/max": 0.7448713779449463, |
| "sampling/sampling_logp_difference/mean": 0.008799537550657988, |
| "step": 16, |
| "step_time": 125.46948220199556 |
| }, |
| { |
| "clip_ratio/high_max": 9.079917072085664e-05, |
| "clip_ratio/high_mean": 9.079917072085664e-05, |
| "clip_ratio/low_mean": 7.84781514084898e-05, |
| "clip_ratio/low_min": 7.84781514084898e-05, |
| "clip_ratio/region_mean": 0.00016927732212934643, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 4361.0, |
| "completions/max_terminated_length": 4361.0, |
| "completions/mean_length": 2464.291748046875, |
| "completions/mean_terminated_length": 2464.291748046875, |
| "completions/min_length": 1042.5, |
| "completions/min_terminated_length": 1042.5, |
| "entropy": 0.23505781777203083, |
| "epoch": 0.11604095563139932, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.052851345390081406, |
| "learning_rate": 2.999904907117206e-07, |
| "loss": 0.24631723761558533, |
| "num_tokens": 1409605.0, |
| "reward": -0.5833333358168602, |
| "reward_std": 0.5149286389350891, |
| "rewards/correctness_reward/mean": -0.5833333358168602, |
| "rewards/correctness_reward/std": 0.5149286389350891, |
| "sampling/importance_sampling_ratio/max": 1.705892026424408, |
| "sampling/importance_sampling_ratio/mean": 0.5010480284690857, |
| "sampling/importance_sampling_ratio/min": 0.06038624048233032, |
| "sampling/sampling_logp_difference/max": 0.6204233765602112, |
| "sampling/sampling_logp_difference/mean": 0.01033627800643444, |
| "step": 17, |
| "step_time": 141.54468511500454 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 4423.5, |
| "completions/max_terminated_length": 4423.5, |
| "completions/mean_length": 2031.4166870117188, |
| "completions/mean_terminated_length": 2031.4166870117188, |
| "completions/min_length": 844.5, |
| "completions/min_terminated_length": 844.5, |
| "entropy": 0.20310819149017334, |
| "epoch": 0.12286689419795221, |
| "frac_reward_zero_std": 1.0, |
| "grad_norm": 0.0, |
| "learning_rate": 2.9996196405256996e-07, |
| "loss": 0.0, |
| "num_tokens": 1491989.0, |
| "reward": -0.5, |
| "reward_std": 0.5222329497337341, |
| "rewards/correctness_reward/mean": -0.5, |
| "rewards/correctness_reward/std": 0.5222329497337341, |
| "sampling/importance_sampling_ratio/max": 1.893693208694458, |
| "sampling/importance_sampling_ratio/mean": 0.5588868856430054, |
| "sampling/importance_sampling_ratio/min": 0.013609365560114384, |
| "sampling/sampling_logp_difference/max": 0.5993608236312866, |
| "sampling/sampling_logp_difference/mean": 0.009505682159215212, |
| "step": 18, |
| "step_time": 138.96269857199513 |
| }, |
| { |
| "clip_ratio/high_max": 3.729117088369094e-05, |
| "clip_ratio/high_mean": 3.729117088369094e-05, |
| "clip_ratio/low_mean": 0.0003078869340242818, |
| "clip_ratio/low_min": 0.0003078869340242818, |
| "clip_ratio/region_mean": 0.00034517810490797274, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 3841.5, |
| "completions/max_terminated_length": 3414.5, |
| "completions/mean_length": 2340.541748046875, |
| "completions/mean_terminated_length": 2127.9500732421875, |
| "completions/min_length": 1127.0, |
| "completions/min_terminated_length": 1127.0, |
| "entropy": 0.20708073489367962, |
| "epoch": 0.1296928327645051, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.006704013328999281, |
| "learning_rate": 2.9991442363945767e-07, |
| "loss": 0.05776930972933769, |
| "num_tokens": 1581882.0, |
| "reward": -0.9166666865348816, |
| "reward_std": 0.2886751592159271, |
| "rewards/correctness_reward/mean": -0.9166666567325592, |
| "rewards/correctness_reward/std": 0.28867512941360474, |
| "sampling/importance_sampling_ratio/max": 1.3786442875862122, |
| "sampling/importance_sampling_ratio/mean": 0.4760366827249527, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.523478627204895, |
| "sampling/sampling_logp_difference/mean": 0.009245706256479025, |
| "step": 19, |
| "step_time": 127.39751178900042 |
| }, |
| { |
| "clip_ratio/high_max": 8.210180385503918e-05, |
| "clip_ratio/high_mean": 8.210180385503918e-05, |
| "clip_ratio/low_mean": 0.0006445428680308396, |
| "clip_ratio/low_min": 0.0006445428680308396, |
| "clip_ratio/region_mean": 0.0007266446718858788, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 4055.0, |
| "completions/max_terminated_length": 4055.0, |
| "completions/mean_length": 1693.2083740234375, |
| "completions/mean_terminated_length": 1693.2083740234375, |
| "completions/min_length": 865.5, |
| "completions/min_terminated_length": 865.5, |
| "entropy": 0.16844954527914524, |
| "epoch": 0.13651877133105803, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.07481791824102402, |
| "learning_rate": 2.998478755000571e-07, |
| "loss": -0.2300712913274765, |
| "num_tokens": 1653749.0, |
| "reward": -0.7500000298023224, |
| "reward_std": 0.6779246032238007, |
| "rewards/correctness_reward/mean": -0.75, |
| "rewards/correctness_reward/std": 0.6779246032238007, |
| "sampling/importance_sampling_ratio/max": 1.5487863421440125, |
| "sampling/importance_sampling_ratio/mean": 0.6358051598072052, |
| "sampling/importance_sampling_ratio/min": 0.04564919415861368, |
| "sampling/sampling_logp_difference/max": 0.6578903198242188, |
| "sampling/sampling_logp_difference/mean": 0.008329455275088549, |
| "step": 20, |
| "step_time": 125.59296355299011 |
| }, |
| { |
| "clip_ratio/high_max": 0.0003053993041248759, |
| "clip_ratio/high_mean": 0.0003053993041248759, |
| "clip_ratio/low_mean": 0.0005765805062765139, |
| "clip_ratio/low_min": 0.0005765805062765139, |
| "clip_ratio/region_mean": 0.0008819798167678528, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 4515.5, |
| "completions/max_terminated_length": 4067.5, |
| "completions/mean_length": 2374.875, |
| "completions/mean_terminated_length": 2117.5750732421875, |
| "completions/min_length": 941.0, |
| "completions/min_terminated_length": 941.0, |
| "entropy": 0.18689643032848835, |
| "epoch": 0.14334470989761092, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.03002380020916462, |
| "learning_rate": 2.997623280720407e-07, |
| "loss": -0.08155997842550278, |
| "num_tokens": 1746350.0, |
| "reward": -0.4166666716337204, |
| "reward_std": 0.9446330070495605, |
| "rewards/correctness_reward/mean": -0.4166666716337204, |
| "rewards/correctness_reward/std": 0.9446330070495605, |
| "sampling/importance_sampling_ratio/max": 1.3880415558815002, |
| "sampling/importance_sampling_ratio/mean": 0.5663973093032837, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.7681431770324707, |
| "sampling/sampling_logp_difference/mean": 0.008763373363763094, |
| "step": 21, |
| "step_time": 147.82532634199015 |
| }, |
| { |
| "clip_ratio/high_max": 0.0001005809517664602, |
| "clip_ratio/high_mean": 0.0001005809517664602, |
| "clip_ratio/low_mean": 0.00033674173755571246, |
| "clip_ratio/low_min": 0.00033674173755571246, |
| "clip_ratio/region_mean": 0.00043732268932217266, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3619.0, |
| "completions/max_terminated_length": 3619.0, |
| "completions/mean_length": 1774.0833740234375, |
| "completions/mean_terminated_length": 1774.0833740234375, |
| "completions/min_length": 923.5, |
| "completions/min_terminated_length": 923.5, |
| "entropy": 0.1736101247370243, |
| "epoch": 0.15017064846416384, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.03494279459118843, |
| "learning_rate": 2.996577922020106e-07, |
| "loss": 0.08088157325983047, |
| "num_tokens": 1820956.0, |
| "reward": -0.6666666865348816, |
| "reward_std": 0.7409421503543854, |
| "rewards/correctness_reward/mean": -0.6666666567325592, |
| "rewards/correctness_reward/std": 0.7409421503543854, |
| "sampling/importance_sampling_ratio/max": 2.3706082105636597, |
| "sampling/importance_sampling_ratio/mean": 0.6451850235462189, |
| "sampling/importance_sampling_ratio/min": 0.02104969136416912, |
| "sampling/sampling_logp_difference/max": 0.6389624774456024, |
| "sampling/sampling_logp_difference/mean": 0.008609656244516373, |
| "step": 22, |
| "step_time": 116.07473812201351 |
| }, |
| { |
| "clip_ratio/high_max": 0.00021113662296556868, |
| "clip_ratio/high_mean": 0.00021113662296556868, |
| "clip_ratio/low_mean": 0.00011071744665969163, |
| "clip_ratio/low_min": 0.00011071744665969163, |
| "clip_ratio/region_mean": 0.0003218540696252603, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 4188.0, |
| "completions/max_terminated_length": 3987.0, |
| "completions/mean_length": 2107.0, |
| "completions/mean_terminated_length": 1845.0, |
| "completions/min_length": 903.5, |
| "completions/min_terminated_length": 903.5, |
| "entropy": 0.17278275545686483, |
| "epoch": 0.15699658703071673, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.040954675525426865, |
| "learning_rate": 2.9953428114412303e-07, |
| "loss": 0.018079154193401337, |
| "num_tokens": 1911322.0, |
| "reward": -0.6666666716337204, |
| "reward_std": 0.4923659861087799, |
| "rewards/correctness_reward/mean": -0.6666666716337204, |
| "rewards/correctness_reward/std": 0.4923659861087799, |
| "sampling/importance_sampling_ratio/max": 1.6411811113357544, |
| "sampling/importance_sampling_ratio/mean": 0.565078467130661, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.6274092197418213, |
| "sampling/sampling_logp_difference/mean": 0.00899354089051485, |
| "step": 23, |
| "step_time": 135.8156637239881 |
| }, |
| { |
| "clip_ratio/high_max": 0.00011917552455997793, |
| "clip_ratio/high_mean": 0.00011917552455997793, |
| "clip_ratio/low_mean": 0.00062593947222922, |
| "clip_ratio/low_min": 0.00062593947222922, |
| "clip_ratio/region_mean": 0.0007451150013366714, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 3819.0, |
| "completions/max_terminated_length": 3004.5, |
| "completions/mean_length": 2192.166748046875, |
| "completions/mean_terminated_length": 1937.783447265625, |
| "completions/min_length": 772.5, |
| "completions/min_terminated_length": 772.5, |
| "entropy": 0.19274314120411873, |
| "epoch": 0.16382252559726962, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.04705621302127838, |
| "learning_rate": 2.993918105584082e-07, |
| "loss": -0.0761164128780365, |
| "num_tokens": 1998752.0, |
| "reward": -0.7500000298023224, |
| "reward_std": 0.677924633026123, |
| "rewards/correctness_reward/mean": -0.75, |
| "rewards/correctness_reward/std": 0.677924633026123, |
| "sampling/importance_sampling_ratio/max": 2.2044373750686646, |
| "sampling/importance_sampling_ratio/mean": 0.5416325330734253, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.7398251295089722, |
| "sampling/sampling_logp_difference/mean": 0.009043822530657053, |
| "step": 24, |
| "step_time": 125.81809424500534 |
| }, |
| { |
| "clip_ratio/high_max": 0.00017605540779186413, |
| "clip_ratio/high_mean": 0.00017605540779186413, |
| "clip_ratio/low_mean": 0.00029181257104937686, |
| "clip_ratio/low_min": 0.00029181257104937686, |
| "clip_ratio/region_mean": 0.0004678679906646721, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 5248.0, |
| "completions/max_terminated_length": 4321.0, |
| "completions/mean_length": 2522.541748046875, |
| "completions/mean_terminated_length": 2274.772705078125, |
| "completions/min_length": 658.0, |
| "completions/min_terminated_length": 658.0, |
| "entropy": 0.22828248515725136, |
| "epoch": 0.17064846416382254, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.02641363814473152, |
| "learning_rate": 2.9923039850878424e-07, |
| "loss": -0.0734773501753807, |
| "num_tokens": 2098737.0, |
| "reward": -0.5, |
| "reward_std": 0.5222329497337341, |
| "rewards/correctness_reward/mean": -0.5, |
| "rewards/correctness_reward/std": 0.5222329497337341, |
| "sampling/importance_sampling_ratio/max": 2.215272545814514, |
| "sampling/importance_sampling_ratio/mean": 0.5476150363683701, |
| "sampling/importance_sampling_ratio/min": 0.08045727759599686, |
| "sampling/sampling_logp_difference/max": 0.6184170246124268, |
| "sampling/sampling_logp_difference/mean": 0.010321022942662239, |
| "step": 25, |
| "step_time": 170.05788060501072 |
| }, |
| { |
| "clip_ratio/high_max": 0.00031355412284028716, |
| "clip_ratio/high_mean": 0.00031355412284028716, |
| "clip_ratio/low_mean": 0.000488045678139315, |
| "clip_ratio/low_min": 0.000488045678139315, |
| "clip_ratio/region_mean": 0.0008015997955226339, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 3655.0, |
| "completions/max_terminated_length": 3551.5, |
| "completions/mean_length": 2256.416748046875, |
| "completions/mean_terminated_length": 2062.0250244140625, |
| "completions/min_length": 962.5, |
| "completions/min_terminated_length": 962.5, |
| "entropy": 0.19479426182806492, |
| "epoch": 0.17747440273037543, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.03679096698760986, |
| "learning_rate": 2.9905006546076744e-07, |
| "loss": 0.04023721441626549, |
| "num_tokens": 2185249.0, |
| "reward": -0.1666666641831398, |
| "reward_std": 0.9671956598758698, |
| "rewards/correctness_reward/mean": -0.1666666641831398, |
| "rewards/correctness_reward/std": 0.9671956598758698, |
| "sampling/importance_sampling_ratio/max": 2.40914648771286, |
| "sampling/importance_sampling_ratio/mean": 0.6711709499359131, |
| "sampling/importance_sampling_ratio/min": 0.08630403692950495, |
| "sampling/sampling_logp_difference/max": 0.9873652458190918, |
| "sampling/sampling_logp_difference/mean": 0.009495267178863287, |
| "step": 26, |
| "step_time": 121.49674113400397 |
| }, |
| { |
| "clip_ratio/high_max": 0.0003286330756964162, |
| "clip_ratio/high_mean": 0.0003286330756964162, |
| "clip_ratio/low_mean": 0.0003672421880764887, |
| "clip_ratio/low_min": 0.0003672421880764887, |
| "clip_ratio/region_mean": 0.0006958752710488625, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3712.0, |
| "completions/max_terminated_length": 3712.0, |
| "completions/mean_length": 1608.9166870117188, |
| "completions/mean_terminated_length": 1608.9166870117188, |
| "completions/min_length": 881.0, |
| "completions/min_terminated_length": 881.0, |
| "entropy": 0.19669272750616074, |
| "epoch": 0.18430034129692832, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.048227936029434204, |
| "learning_rate": 2.988508342788768e-07, |
| "loss": 0.15509483218193054, |
| "num_tokens": 2252771.0, |
| "reward": -0.5, |
| "reward_std": 0.5222329497337341, |
| "rewards/correctness_reward/mean": -0.5, |
| "rewards/correctness_reward/std": 0.5222329497337341, |
| "sampling/importance_sampling_ratio/max": 1.7127284407615662, |
| "sampling/importance_sampling_ratio/mean": 0.5389788150787354, |
| "sampling/importance_sampling_ratio/min": 0.028196483850479126, |
| "sampling/sampling_logp_difference/max": 0.8035677671432495, |
| "sampling/sampling_logp_difference/mean": 0.009410608559846878, |
| "step": 27, |
| "step_time": 116.56851660798566 |
| }, |
| { |
| "clip_ratio/high_max": 2.1826435840921476e-05, |
| "clip_ratio/high_mean": 2.1826435840921476e-05, |
| "clip_ratio/low_mean": 0.00039388730510836467, |
| "clip_ratio/low_min": 0.00039388730510836467, |
| "clip_ratio/region_mean": 0.00041571374094928615, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 2754.0, |
| "completions/max_terminated_length": 2754.0, |
| "completions/mean_length": 1630.9166870117188, |
| "completions/mean_terminated_length": 1630.9166870117188, |
| "completions/min_length": 925.0, |
| "completions/min_terminated_length": 925.0, |
| "entropy": 0.16313775815069675, |
| "epoch": 0.19112627986348124, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.006842807866632938, |
| "learning_rate": 2.9863273022373584e-07, |
| "loss": 0.07145501673221588, |
| "num_tokens": 2321181.0, |
| "reward": -0.9166666865348816, |
| "reward_std": 0.2886751592159271, |
| "rewards/correctness_reward/mean": -0.9166666567325592, |
| "rewards/correctness_reward/std": 0.28867512941360474, |
| "sampling/importance_sampling_ratio/max": 2.4500311613082886, |
| "sampling/importance_sampling_ratio/mean": 0.8633750975131989, |
| "sampling/importance_sampling_ratio/min": 0.02915837988257408, |
| "sampling/sampling_logp_difference/max": 0.6632177233695984, |
| "sampling/sampling_logp_difference/mean": 0.008668553549796343, |
| "step": 28, |
| "step_time": 91.4905828859919 |
| }, |
| { |
| "clip_ratio/high_max": 0.0002642529752847622, |
| "clip_ratio/high_mean": 0.0002642529752847622, |
| "clip_ratio/low_mean": 0.0005223052976361942, |
| "clip_ratio/low_min": 0.0005223052976361942, |
| "clip_ratio/region_mean": 0.0007865582774684299, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 4784.5, |
| "completions/max_terminated_length": 4765.0, |
| "completions/mean_length": 2501.666748046875, |
| "completions/mean_terminated_length": 2392.621337890625, |
| "completions/min_length": 888.0, |
| "completions/min_terminated_length": 888.0, |
| "entropy": 0.22441293857991695, |
| "epoch": 0.19795221843003413, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.05232253298163414, |
| "learning_rate": 2.9839578094886877e-07, |
| "loss": 0.29416829347610474, |
| "num_tokens": 2419909.0, |
| "reward": -0.3333333507180214, |
| "reward_std": 0.8036037981510162, |
| "rewards/correctness_reward/mean": -0.333333320915699, |
| "rewards/correctness_reward/std": 0.8036037683486938, |
| "sampling/importance_sampling_ratio/max": 1.3574910461902618, |
| "sampling/importance_sampling_ratio/mean": 0.46175582706928253, |
| "sampling/importance_sampling_ratio/min": 0.023056588135659695, |
| "sampling/sampling_logp_difference/max": 0.7084431648254395, |
| "sampling/sampling_logp_difference/mean": 0.009676191490143538, |
| "step": 29, |
| "step_time": 154.80130539400125 |
| }, |
| { |
| "clip_ratio/high_max": 0.00027174162278242875, |
| "clip_ratio/high_mean": 0.00027174162278242875, |
| "clip_ratio/low_mean": 0.0007876567906350829, |
| "clip_ratio/low_min": 0.0007876567906350829, |
| "clip_ratio/region_mean": 0.0010593984261504374, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 3716.0, |
| "completions/max_terminated_length": 2443.0, |
| "completions/mean_length": 1590.0833740234375, |
| "completions/mean_terminated_length": 1435.5568237304688, |
| "completions/min_length": 855.5, |
| "completions/min_terminated_length": 855.5, |
| "entropy": 0.16478881239891052, |
| "epoch": 0.20477815699658702, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.08924312144517899, |
| "learning_rate": 2.981400164971952e-07, |
| "loss": -0.43429163098335266, |
| "num_tokens": 2485449.0, |
| "reward": -0.4166666716337204, |
| "reward_std": 0.9446329772472382, |
| "rewards/correctness_reward/mean": -0.4166666716337204, |
| "rewards/correctness_reward/std": 0.9446330070495605, |
| "sampling/importance_sampling_ratio/max": 1.8077207803726196, |
| "sampling/importance_sampling_ratio/mean": 0.641688883304596, |
| "sampling/importance_sampling_ratio/min": 0.07364495098590851, |
| "sampling/sampling_logp_difference/max": 0.6166839599609375, |
| "sampling/sampling_logp_difference/mean": 0.00862132990732789, |
| "step": 30, |
| "step_time": 115.32614517699403 |
| }, |
| { |
| "clip_ratio/high_max": 0.00017621346887608524, |
| "clip_ratio/high_mean": 0.00017621346887608524, |
| "clip_ratio/low_mean": 0.0005795853066956624, |
| "clip_ratio/low_min": 0.0005795853066956624, |
| "clip_ratio/region_mean": 0.0007557987719337689, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 3910.5, |
| "completions/max_terminated_length": 3090.0, |
| "completions/mean_length": 2015.8333740234375, |
| "completions/mean_terminated_length": 1889.776611328125, |
| "completions/min_length": 889.0, |
| "completions/min_terminated_length": 889.0, |
| "entropy": 0.2223886251449585, |
| "epoch": 0.21160409556313994, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.06942225992679596, |
| "learning_rate": 2.9786546929722056e-07, |
| "loss": 0.3086526393890381, |
| "num_tokens": 2564933.0, |
| "reward": -0.1666666865348816, |
| "reward_std": 0.7409421503543854, |
| "rewards/correctness_reward/mean": -0.1666666567325592, |
| "rewards/correctness_reward/std": 0.7409421503543854, |
| "sampling/importance_sampling_ratio/max": 2.386189043521881, |
| "sampling/importance_sampling_ratio/mean": 0.6213788092136383, |
| "sampling/importance_sampling_ratio/min": 0.015281436033546925, |
| "sampling/sampling_logp_difference/max": 0.7943930625915527, |
| "sampling/sampling_logp_difference/mean": 0.00985921872779727, |
| "step": 31, |
| "step_time": 125.07035979099601 |
| }, |
| { |
| "clip_ratio/high_max": 0.0002656124333952903, |
| "clip_ratio/high_mean": 0.0002656124333952903, |
| "clip_ratio/low_mean": 0.00017724649842421059, |
| "clip_ratio/low_min": 0.00017724649842421059, |
| "clip_ratio/region_mean": 0.0004428589454619214, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 3942.0, |
| "completions/max_terminated_length": 3820.5, |
| "completions/mean_length": 1951.7500610351562, |
| "completions/mean_terminated_length": 1836.6439819335938, |
| "completions/min_length": 996.5, |
| "completions/min_terminated_length": 996.5, |
| "entropy": 0.17561497911810875, |
| "epoch": 0.21843003412969283, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.01926896534860134, |
| "learning_rate": 2.9757217415892433e-07, |
| "loss": 0.052177101373672485, |
| "num_tokens": 2646665.0, |
| "reward": -0.5, |
| "reward_std": 0.5222329497337341, |
| "rewards/correctness_reward/mean": -0.5, |
| "rewards/correctness_reward/std": 0.5222329497337341, |
| "sampling/importance_sampling_ratio/max": 2.242147386074066, |
| "sampling/importance_sampling_ratio/mean": 0.6626814007759094, |
| "sampling/importance_sampling_ratio/min": 0.031560447067022324, |
| "sampling/sampling_logp_difference/max": 0.6647169589996338, |
| "sampling/sampling_logp_difference/mean": 0.00904133077710867, |
| "step": 32, |
| "step_time": 127.47206454698608 |
| }, |
| { |
| "clip_ratio/high_max": 0.00018125995120499283, |
| "clip_ratio/high_mean": 0.00018125995120499283, |
| "clip_ratio/low_mean": 0.0005849039262102451, |
| "clip_ratio/low_min": 0.0005849039262102451, |
| "clip_ratio/region_mean": 0.000766163877415238, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 3630.5, |
| "completions/max_terminated_length": 3218.5, |
| "completions/mean_length": 1927.5416870117188, |
| "completions/mean_terminated_length": 1798.4393920898438, |
| "completions/min_length": 866.0, |
| "completions/min_terminated_length": 866.0, |
| "entropy": 0.17528906650841236, |
| "epoch": 0.22525597269624573, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.03713307902216911, |
| "learning_rate": 2.9726016826934687e-07, |
| "loss": -0.05819889158010483, |
| "num_tokens": 2732742.0, |
| "reward": -0.5, |
| "reward_std": 0.9045340418815613, |
| "rewards/correctness_reward/mean": -0.5, |
| "rewards/correctness_reward/std": 0.9045340418815613, |
| "sampling/importance_sampling_ratio/max": 1.9077945947647095, |
| "sampling/importance_sampling_ratio/mean": 0.5301178991794586, |
| "sampling/importance_sampling_ratio/min": 0.037661814130842686, |
| "sampling/sampling_logp_difference/max": 0.8656803965568542, |
| "sampling/sampling_logp_difference/mean": 0.008524105418473482, |
| "step": 33, |
| "step_time": 124.27680982700258 |
| }, |
| { |
| "clip_ratio/high_max": 0.0005821565191581612, |
| "clip_ratio/high_mean": 0.0005821565191581612, |
| "clip_ratio/low_mean": 0.0001287642153329216, |
| "clip_ratio/low_min": 0.0001287642153329216, |
| "clip_ratio/region_mean": 0.0007109207344910828, |
| "completions/clipped_ratio": 0.1250000037252903, |
| "completions/max_length": 5248.0, |
| "completions/max_terminated_length": 3240.5, |
| "completions/mean_length": 1932.5, |
| "completions/mean_terminated_length": 1463.3091430664062, |
| "completions/min_length": 811.5, |
| "completions/min_terminated_length": 811.5, |
| "entropy": 0.16712841391563416, |
| "epoch": 0.23208191126279865, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.05192067101597786, |
| "learning_rate": 2.9692949118787414e-07, |
| "loss": 0.12746219336986542, |
| "num_tokens": 2813184.0, |
| "reward": -0.3333333358168602, |
| "reward_std": 0.9671956598758698, |
| "rewards/correctness_reward/mean": -0.3333333358168602, |
| "rewards/correctness_reward/std": 0.9671956598758698, |
| "sampling/importance_sampling_ratio/max": 1.7298161387443542, |
| "sampling/importance_sampling_ratio/mean": 0.6548172235488892, |
| "sampling/importance_sampling_ratio/min": 0.0035946578718721867, |
| "sampling/sampling_logp_difference/max": 0.6443392038345337, |
| "sampling/sampling_logp_difference/mean": 0.008729535154998302, |
| "step": 34, |
| "step_time": 159.94666257500648 |
| }, |
| { |
| "clip_ratio/high_max": 9.326366489403881e-05, |
| "clip_ratio/high_mean": 9.326366489403881e-05, |
| "clip_ratio/low_mean": 0.0005161476583452895, |
| "clip_ratio/low_min": 0.0005161476583452895, |
| "clip_ratio/region_mean": 0.0006094113232393283, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 4532.0, |
| "completions/max_terminated_length": 3245.0, |
| "completions/mean_length": 1874.7500610351562, |
| "completions/mean_terminated_length": 1559.3416748046875, |
| "completions/min_length": 875.0, |
| "completions/min_terminated_length": 875.0, |
| "entropy": 0.1692411694675684, |
| "epoch": 0.23890784982935154, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.04213385283946991, |
| "learning_rate": 2.965801848412221e-07, |
| "loss": 0.07331521809101105, |
| "num_tokens": 2888670.0, |
| "reward": -0.6666666865348816, |
| "reward_std": 0.7409421503543854, |
| "rewards/correctness_reward/mean": -0.6666666567325592, |
| "rewards/correctness_reward/std": 0.7409421503543854, |
| "sampling/importance_sampling_ratio/max": 2.3725807666778564, |
| "sampling/importance_sampling_ratio/mean": 0.8309193253517151, |
| "sampling/importance_sampling_ratio/min": 0.0595044381916523, |
| "sampling/sampling_logp_difference/max": 0.5672608017921448, |
| "sampling/sampling_logp_difference/mean": 0.008656369056552649, |
| "step": 35, |
| "step_time": 139.26266696700623 |
| }, |
| { |
| "clip_ratio/high_max": 0.00012051208250341006, |
| "clip_ratio/high_mean": 0.00012051208250341006, |
| "clip_ratio/low_mean": 0.0003169747687934432, |
| "clip_ratio/low_min": 0.0003169747687934432, |
| "clip_ratio/region_mean": 0.0004374868512968533, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 3681.5, |
| "completions/max_terminated_length": 3509.5, |
| "completions/mean_length": 2362.6250610351562, |
| "completions/mean_terminated_length": 2181.7666625976562, |
| "completions/min_length": 1057.0, |
| "completions/min_terminated_length": 1057.0, |
| "entropy": 0.17967960238456726, |
| "epoch": 0.24573378839590443, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.008928037248551846, |
| "learning_rate": 2.9621229351812067e-07, |
| "loss": 0.05752938240766525, |
| "num_tokens": 2981865.0, |
| "reward": -0.25, |
| "reward_std": 0.9744999706745148, |
| "rewards/correctness_reward/mean": -0.25, |
| "rewards/correctness_reward/std": 0.9744999706745148, |
| "sampling/importance_sampling_ratio/max": 2.1388869285583496, |
| "sampling/importance_sampling_ratio/mean": 0.5870790481567383, |
| "sampling/importance_sampling_ratio/min": 0.001988642383366823, |
| "sampling/sampling_logp_difference/max": 0.9233172535896301, |
| "sampling/sampling_logp_difference/mean": 0.009207553695887327, |
| "step": 36, |
| "step_time": 125.00127266600612 |
| }, |
| { |
| "clip_ratio/high_max": 0.00010507144907023758, |
| "clip_ratio/high_mean": 0.00010507144907023758, |
| "clip_ratio/low_mean": 0.00033121939486591145, |
| "clip_ratio/low_min": 0.00033121939486591145, |
| "clip_ratio/region_mean": 0.00043629084393614903, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 2232.0, |
| "completions/max_terminated_length": 2232.0, |
| "completions/mean_length": 1288.7500610351562, |
| "completions/mean_terminated_length": 1288.7500610351562, |
| "completions/min_length": 793.0, |
| "completions/min_terminated_length": 793.0, |
| "entropy": 0.1518058469519019, |
| "epoch": 0.2525597269624573, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.11457372456789017, |
| "learning_rate": 2.958258638636986e-07, |
| "loss": -0.41082119941711426, |
| "num_tokens": 3040839.0, |
| "reward": -0.9166666865348816, |
| "reward_std": 0.28867512941360474, |
| "rewards/correctness_reward/mean": -0.9166666567325592, |
| "rewards/correctness_reward/std": 0.28867512941360474, |
| "sampling/importance_sampling_ratio/max": 2.3129096031188965, |
| "sampling/importance_sampling_ratio/mean": 0.7068502902984619, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.5720200538635254, |
| "sampling/sampling_logp_difference/mean": 0.008375945966690779, |
| "step": 37, |
| "step_time": 74.90589488900878 |
| }, |
| { |
| "clip_ratio/high_max": 4.4680284190690145e-05, |
| "clip_ratio/high_mean": 4.4680284190690145e-05, |
| "clip_ratio/low_mean": 0.0001608589809620753, |
| "clip_ratio/low_min": 0.0001608589809620753, |
| "clip_ratio/region_mean": 0.00020553926515276544, |
| "completions/clipped_ratio": 0.125, |
| "completions/max_length": 3663.5, |
| "completions/max_terminated_length": 3535.0, |
| "completions/mean_length": 2338.7083740234375, |
| "completions/mean_terminated_length": 2058.736083984375, |
| "completions/min_length": 748.5, |
| "completions/min_terminated_length": 748.5, |
| "entropy": 0.2067808359861374, |
| "epoch": 0.2593856655290102, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.019327981397509575, |
| "learning_rate": 2.9542094487356887e-07, |
| "loss": 0.0506199486553669, |
| "num_tokens": 3132146.0, |
| "reward": -0.5, |
| "reward_std": 0.5222329497337341, |
| "rewards/correctness_reward/mean": -0.5, |
| "rewards/correctness_reward/std": 0.5222329497337341, |
| "sampling/importance_sampling_ratio/max": 1.9883459210395813, |
| "sampling/importance_sampling_ratio/mean": 0.6568911969661713, |
| "sampling/importance_sampling_ratio/min": 0.0012566318036988378, |
| "sampling/sampling_logp_difference/max": 0.622053861618042, |
| "sampling/sampling_logp_difference/mean": 0.009283999912440777, |
| "step": 38, |
| "step_time": 124.15157590300805 |
| }, |
| { |
| "clip_ratio/high_max": 0.0002599538038339233, |
| "clip_ratio/high_mean": 0.0002599538038339233, |
| "clip_ratio/low_mean": 0.0009598340366210323, |
| "clip_ratio/low_min": 0.0009598340366210323, |
| "clip_ratio/region_mean": 0.0012197878386359662, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 4883.0, |
| "completions/max_terminated_length": 4883.0, |
| "completions/mean_length": 2541.3333740234375, |
| "completions/mean_terminated_length": 2541.3333740234375, |
| "completions/min_length": 1086.0, |
| "completions/min_terminated_length": 1086.0, |
| "entropy": 0.20533941686153412, |
| "epoch": 0.26621160409556316, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.03861364349722862, |
| "learning_rate": 2.949975878876169e-07, |
| "loss": 0.37017422914505005, |
| "num_tokens": 3228928.0, |
| "reward": -0.3333333358168602, |
| "reward_std": 0.9671956598758698, |
| "rewards/correctness_reward/mean": -0.3333333358168602, |
| "rewards/correctness_reward/std": 0.9671956598758698, |
| "sampling/importance_sampling_ratio/max": 1.7460360527038574, |
| "sampling/importance_sampling_ratio/mean": 0.4507356733083725, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.7737934589385986, |
| "sampling/sampling_logp_difference/mean": 0.009104832075536251, |
| "step": 39, |
| "step_time": 157.81988949599327 |
| }, |
| { |
| "clip_ratio/high_max": 0.00011135237582493573, |
| "clip_ratio/high_mean": 0.00011135237582493573, |
| "clip_ratio/low_mean": 0.0005540200945688412, |
| "clip_ratio/low_min": 0.0005540200945688412, |
| "clip_ratio/region_mean": 0.0006653724703937769, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 4377.0, |
| "completions/max_terminated_length": 4269.0, |
| "completions/mean_length": 2209.0416870117188, |
| "completions/mean_terminated_length": 1929.1917114257812, |
| "completions/min_length": 875.5, |
| "completions/min_terminated_length": 875.5, |
| "entropy": 0.1815284639596939, |
| "epoch": 0.27303754266211605, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.06392155587673187, |
| "learning_rate": 2.945558465834911e-07, |
| "loss": -0.38024067878723145, |
| "num_tokens": 3317567.0, |
| "reward": -0.6666666865348816, |
| "reward_std": 0.7409421503543854, |
| "rewards/correctness_reward/mean": -0.6666666567325592, |
| "rewards/correctness_reward/std": 0.7409421503543854, |
| "sampling/importance_sampling_ratio/max": 2.0949626564979553, |
| "sampling/importance_sampling_ratio/mean": 0.5830715447664261, |
| "sampling/importance_sampling_ratio/min": 0.007053101202473044, |
| "sampling/sampling_logp_difference/max": 0.4775382876396179, |
| "sampling/sampling_logp_difference/mean": 0.008772795088589191, |
| "step": 40, |
| "step_time": 140.69885612401413 |
| }, |
| { |
| "clip_ratio/high_max": 8.099789556581527e-05, |
| "clip_ratio/high_mean": 8.099789556581527e-05, |
| "clip_ratio/low_mean": 0.0003992624915554188, |
| "clip_ratio/low_min": 0.0003992624915554188, |
| "clip_ratio/region_mean": 0.0004802603871212341, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3610.0, |
| "completions/max_terminated_length": 3610.0, |
| "completions/mean_length": 1637.6250610351562, |
| "completions/mean_terminated_length": 1637.6250610351562, |
| "completions/min_length": 712.5, |
| "completions/min_terminated_length": 712.5, |
| "entropy": 0.18336657993495464, |
| "epoch": 0.27986348122866894, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.04287489131093025, |
| "learning_rate": 2.9409577696979687e-07, |
| "loss": -0.1049296110868454, |
| "num_tokens": 3391028.0, |
| "reward": -0.9166666865348816, |
| "reward_std": 0.28867512941360474, |
| "rewards/correctness_reward/mean": -0.9166666567325592, |
| "rewards/correctness_reward/std": 0.28867512941360474, |
| "sampling/importance_sampling_ratio/max": 1.7601441740989685, |
| "sampling/importance_sampling_ratio/mean": 0.5774812400341034, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.6337925791740417, |
| "sampling/sampling_logp_difference/mean": 0.008997163735330105, |
| "step": 41, |
| "step_time": 116.36959670999931 |
| }, |
| { |
| "clip_ratio/high_max": 0.0003851416477118619, |
| "clip_ratio/high_mean": 0.0003851416477118619, |
| "clip_ratio/low_mean": 0.0006896271115692798, |
| "clip_ratio/low_min": 0.0006896271115692798, |
| "clip_ratio/region_mean": 0.0010747687556431629, |
| "completions/clipped_ratio": 0.1250000037252903, |
| "completions/max_length": 5248.0, |
| "completions/max_terminated_length": 4576.5, |
| "completions/mean_length": 2773.1251220703125, |
| "completions/mean_terminated_length": 2403.9500732421875, |
| "completions/min_length": 867.0, |
| "completions/min_terminated_length": 867.0, |
| "entropy": 0.2158426120877266, |
| "epoch": 0.28668941979522183, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.054242201149463654, |
| "learning_rate": 2.9361743737899534e-07, |
| "loss": 0.15749189257621765, |
| "num_tokens": 3488897.0, |
| "reward": 0.0, |
| "reward_std": 1.0444658994674683, |
| "rewards/correctness_reward/mean": 0.0, |
| "rewards/correctness_reward/std": 1.0444658994674683, |
| "sampling/importance_sampling_ratio/max": 2.1604199409484863, |
| "sampling/importance_sampling_ratio/mean": 0.6884078681468964, |
| "sampling/importance_sampling_ratio/min": 0.034998440532945096, |
| "sampling/sampling_logp_difference/max": 0.7063878774642944, |
| "sampling/sampling_logp_difference/mean": 0.009720813482999802, |
| "step": 42, |
| "step_time": 165.69028340099612 |
| }, |
| { |
| "clip_ratio/high_max": 2.61752702499507e-05, |
| "clip_ratio/high_mean": 2.61752702499507e-05, |
| "clip_ratio/low_mean": 0.00038996605871943757, |
| "clip_ratio/low_min": 0.00038996605871943757, |
| "clip_ratio/region_mean": 0.00041614132896938827, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 4010.5, |
| "completions/max_terminated_length": 4010.5, |
| "completions/mean_length": 1930.5000610351562, |
| "completions/mean_terminated_length": 1930.5000610351562, |
| "completions/min_length": 913.0, |
| "completions/min_terminated_length": 913.0, |
| "entropy": 0.18189280107617378, |
| "epoch": 0.2935153583617747, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.016736026853322983, |
| "learning_rate": 2.931208884600073e-07, |
| "loss": 0.052612289786338806, |
| "num_tokens": 3567983.0, |
| "reward": -0.75, |
| "reward_std": 0.45226702094078064, |
| "rewards/correctness_reward/mean": -0.75, |
| "rewards/correctness_reward/std": 0.45226702094078064, |
| "sampling/importance_sampling_ratio/max": 1.9738622903823853, |
| "sampling/importance_sampling_ratio/mean": 0.554217055439949, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.6788408756256104, |
| "sampling/sampling_logp_difference/mean": 0.008132760412991047, |
| "step": 43, |
| "step_time": 127.23670873800438 |
| }, |
| { |
| "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": 0.125, |
| "completions/max_length": 4686.5, |
| "completions/max_terminated_length": 4259.0, |
| "completions/mean_length": 2431.666748046875, |
| "completions/mean_terminated_length": 2103.15283203125, |
| "completions/min_length": 1169.5, |
| "completions/min_terminated_length": 1169.5, |
| "entropy": 0.1827030796557665, |
| "epoch": 0.3003412969283277, |
| "frac_reward_zero_std": 1.0, |
| "grad_norm": 0.0, |
| "learning_rate": 2.926061931705237e-07, |
| "loss": 0.0, |
| "num_tokens": 3667377.0, |
| "reward": -1.0, |
| "reward_std": 0.0, |
| "rewards/correctness_reward/mean": -1.0, |
| "rewards/correctness_reward/std": 0.0, |
| "sampling/importance_sampling_ratio/max": 1.0915364623069763, |
| "sampling/importance_sampling_ratio/mean": 0.481503888964653, |
| "sampling/importance_sampling_ratio/min": 0.01448526605963707, |
| "sampling/sampling_logp_difference/max": 0.601245641708374, |
| "sampling/sampling_logp_difference/mean": 0.008691030088812113, |
| "step": 44, |
| "step_time": 151.4489465440056 |
| }, |
| { |
| "clip_ratio/high_max": 4.348203037807252e-05, |
| "clip_ratio/high_mean": 4.348203037807252e-05, |
| "clip_ratio/low_mean": 0.00013300800128490664, |
| "clip_ratio/low_min": 0.00013300800128490664, |
| "clip_ratio/region_mean": 0.00017649003166297916, |
| "completions/clipped_ratio": 0.125, |
| "completions/max_length": 3959.5, |
| "completions/max_terminated_length": 3033.0, |
| "completions/mean_length": 2206.7500610351562, |
| "completions/mean_terminated_length": 1852.5695190429688, |
| "completions/min_length": 1081.0, |
| "completions/min_terminated_length": 1081.0, |
| "entropy": 0.17536703869700432, |
| "epoch": 0.30716723549488056, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.019536586478352547, |
| "learning_rate": 2.92073416769023e-07, |
| "loss": -0.05542347952723503, |
| "num_tokens": 3756813.0, |
| "reward": -0.8333333432674408, |
| "reward_std": 0.38924944400787354, |
| "rewards/correctness_reward/mean": -0.8333333432674408, |
| "rewards/correctness_reward/std": 0.3892495036125183, |
| "sampling/importance_sampling_ratio/max": 1.5530624985694885, |
| "sampling/importance_sampling_ratio/mean": 0.5366989821195602, |
| "sampling/importance_sampling_ratio/min": 0.04268880607560277, |
| "sampling/sampling_logp_difference/max": 0.5965178608894348, |
| "sampling/sampling_logp_difference/mean": 0.008516309317201376, |
| "step": 45, |
| "step_time": 131.10619422599848 |
| }, |
| { |
| "clip_ratio/high_max": 0.00021378630663093645, |
| "clip_ratio/high_mean": 0.00021378630663093645, |
| "clip_ratio/low_mean": 0.0004329298644734081, |
| "clip_ratio/low_min": 0.0004329298644734081, |
| "clip_ratio/region_mean": 0.0006467161674663657, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 2647.0, |
| "completions/max_terminated_length": 2647.0, |
| "completions/mean_length": 1851.291748046875, |
| "completions/mean_terminated_length": 1851.291748046875, |
| "completions/min_length": 1014.5, |
| "completions/min_terminated_length": 1014.5, |
| "entropy": 0.17232570238411427, |
| "epoch": 0.31399317406143346, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.04990917444229126, |
| "learning_rate": 2.91522626806497e-07, |
| "loss": -0.06765934079885483, |
| "num_tokens": 3830830.0, |
| "reward": -0.4166666716337204, |
| "reward_std": 0.9446329772472382, |
| "rewards/correctness_reward/mean": -0.4166666716337204, |
| "rewards/correctness_reward/std": 0.9446330070495605, |
| "sampling/importance_sampling_ratio/max": 2.2500372529029846, |
| "sampling/importance_sampling_ratio/mean": 0.6319310963153839, |
| "sampling/importance_sampling_ratio/min": 0.02974645048379898, |
| "sampling/sampling_logp_difference/max": 0.6080000102519989, |
| "sampling/sampling_logp_difference/mean": 0.008410447742789984, |
| "step": 46, |
| "step_time": 90.71643152000615 |
| }, |
| { |
| "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": 0.0416666679084301, |
| "completions/max_length": 4392.5, |
| "completions/max_terminated_length": 3531.5, |
| "completions/mean_length": 1753.2083740234375, |
| "completions/mean_terminated_length": 1598.6931762695312, |
| "completions/min_length": 749.5, |
| "completions/min_terminated_length": 749.5, |
| "entropy": 0.18032599054276943, |
| "epoch": 0.32081911262798635, |
| "frac_reward_zero_std": 1.0, |
| "grad_norm": 0.0, |
| "learning_rate": 2.9095389311788627e-07, |
| "loss": 0.0, |
| "num_tokens": 3907401.0, |
| "reward": -1.0, |
| "reward_std": 0.0, |
| "rewards/correctness_reward/mean": -1.0, |
| "rewards/correctness_reward/std": 0.0, |
| "sampling/importance_sampling_ratio/max": 2.022447347640991, |
| "sampling/importance_sampling_ratio/mean": 0.5649960935115814, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.661316990852356, |
| "sampling/sampling_logp_difference/mean": 0.008854995481669903, |
| "step": 47, |
| "step_time": 136.0928902579908 |
| }, |
| { |
| "clip_ratio/high_max": 0.0002303535029568593, |
| "clip_ratio/high_mean": 0.0002303535029568593, |
| "clip_ratio/low_mean": 0.0008515775716659846, |
| "clip_ratio/low_min": 0.0008515775716659846, |
| "clip_ratio/region_mean": 0.0010819310809893068, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 4282.0, |
| "completions/max_terminated_length": 4095.0, |
| "completions/mean_length": 2293.2083740234375, |
| "completions/mean_terminated_length": 2041.425048828125, |
| "completions/min_length": 985.5, |
| "completions/min_terminated_length": 985.5, |
| "entropy": 0.16349471919238567, |
| "epoch": 0.32764505119453924, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.029812565073370934, |
| "learning_rate": 2.9036728781322535e-07, |
| "loss": 0.2133912295103073, |
| "num_tokens": 4000670.0, |
| "reward": -0.6666666865348816, |
| "reward_std": 0.7409421503543854, |
| "rewards/correctness_reward/mean": -0.6666666567325592, |
| "rewards/correctness_reward/std": 0.7409421503543854, |
| "sampling/importance_sampling_ratio/max": 1.9735003113746643, |
| "sampling/importance_sampling_ratio/mean": 0.5840648710727692, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.8899977207183838, |
| "sampling/sampling_logp_difference/mean": 0.008015737868845463, |
| "step": 48, |
| "step_time": 139.79937548300222 |
| }, |
| { |
| "clip_ratio/high_max": 0.0002933151881734375, |
| "clip_ratio/high_mean": 0.0002933151881734375, |
| "clip_ratio/low_mean": 0.0008847473072819412, |
| "clip_ratio/low_min": 0.0008847473072819412, |
| "clip_ratio/region_mean": 0.001178062506369315, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5097.5, |
| "completions/max_terminated_length": 5097.5, |
| "completions/mean_length": 2149.375, |
| "completions/mean_terminated_length": 2149.375, |
| "completions/min_length": 979.0, |
| "completions/min_terminated_length": 979.0, |
| "entropy": 0.17316051665693521, |
| "epoch": 0.33447098976109213, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.06560072302818298, |
| "learning_rate": 2.8976288526850024e-07, |
| "loss": -0.19419705867767334, |
| "num_tokens": 4081457.0, |
| "reward": -0.0833333432674408, |
| "reward_std": 0.8415164947509766, |
| "rewards/correctness_reward/mean": -0.0833333432674408, |
| "rewards/correctness_reward/std": 0.8415164947509766, |
| "sampling/importance_sampling_ratio/max": 2.216562867164612, |
| "sampling/importance_sampling_ratio/mean": 0.6668431162834167, |
| "sampling/importance_sampling_ratio/min": 0.0018459694692865014, |
| "sampling/sampling_logp_difference/max": 1.8778375387191772, |
| "sampling/sampling_logp_difference/mean": 0.008765012491494417, |
| "step": 49, |
| "step_time": 156.69416434800223 |
| }, |
| { |
| "clip_ratio/high_max": 3.402286165510304e-05, |
| "clip_ratio/high_mean": 3.402286165510304e-05, |
| "clip_ratio/low_mean": 0.0005146802941453643, |
| "clip_ratio/low_min": 0.0005146802941453643, |
| "clip_ratio/region_mean": 0.0005487031558004674, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 3714.0, |
| "completions/max_terminated_length": 3508.5, |
| "completions/mean_length": 1895.25, |
| "completions/mean_terminated_length": 1630.2249755859375, |
| "completions/min_length": 710.5, |
| "completions/min_terminated_length": 710.5, |
| "entropy": 0.17622262984514236, |
| "epoch": 0.3412969283276451, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.027309048920869827, |
| "learning_rate": 2.8914076211621813e-07, |
| "loss": -0.07097186893224716, |
| "num_tokens": 4160171.0, |
| "reward": -0.8333333432674408, |
| "reward_std": 0.38924944400787354, |
| "rewards/correctness_reward/mean": -0.8333333432674408, |
| "rewards/correctness_reward/std": 0.3892495036125183, |
| "sampling/importance_sampling_ratio/max": 1.059992790222168, |
| "sampling/importance_sampling_ratio/mean": 0.4521847665309906, |
| "sampling/importance_sampling_ratio/min": 0.01352127268910408, |
| "sampling/sampling_logp_difference/max": 0.6476984620094299, |
| "sampling/sampling_logp_difference/mean": 0.008724214509129524, |
| "step": 50, |
| "step_time": 121.86180935701123 |
| }, |
| { |
| "clip_ratio/high_max": 0.0002972058136947453, |
| "clip_ratio/high_mean": 0.0002972058136947453, |
| "clip_ratio/low_mean": 0.000536448962520808, |
| "clip_ratio/low_min": 0.000536448962520808, |
| "clip_ratio/region_mean": 0.0008336547798535321, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 4687.0, |
| "completions/max_terminated_length": 4029.5, |
| "completions/mean_length": 2090.8333740234375, |
| "completions/mean_terminated_length": 1952.7614135742188, |
| "completions/min_length": 814.5, |
| "completions/min_terminated_length": 814.5, |
| "entropy": 0.19055325351655483, |
| "epoch": 0.34812286689419797, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.02999180555343628, |
| "learning_rate": 2.8850099723569103e-07, |
| "loss": -0.10701270401477814, |
| "num_tokens": 4253803.0, |
| "reward": -0.4166666865348816, |
| "reward_std": 0.8109080791473389, |
| "rewards/correctness_reward/mean": -0.4166666567325592, |
| "rewards/correctness_reward/std": 0.8109080791473389, |
| "sampling/importance_sampling_ratio/max": 1.5244572162628174, |
| "sampling/importance_sampling_ratio/mean": 0.48259805142879486, |
| "sampling/importance_sampling_ratio/min": 0.005122458096593618, |
| "sampling/sampling_logp_difference/max": 0.5854331254959106, |
| "sampling/sampling_logp_difference/mean": 0.009422262199223042, |
| "step": 51, |
| "step_time": 151.26289811401512 |
| }, |
| { |
| "clip_ratio/high_max": 0.00013200865942053497, |
| "clip_ratio/high_mean": 0.00013200865942053497, |
| "clip_ratio/low_mean": 0.0007060257175908191, |
| "clip_ratio/low_min": 0.0007060257175908191, |
| "clip_ratio/region_mean": 0.0008380343770113541, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 4348.5, |
| "completions/max_terminated_length": 3834.5, |
| "completions/mean_length": 2119.6251220703125, |
| "completions/mean_terminated_length": 2002.0606689453125, |
| "completions/min_length": 1073.5, |
| "completions/min_terminated_length": 1073.5, |
| "entropy": 0.16607994306832552, |
| "epoch": 0.35494880546075086, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.04837702214717865, |
| "learning_rate": 2.878436717430346e-07, |
| "loss": -0.04106302559375763, |
| "num_tokens": 4342324.0, |
| "reward": -0.7500000298023224, |
| "reward_std": 0.677924633026123, |
| "rewards/correctness_reward/mean": -0.75, |
| "rewards/correctness_reward/std": 0.677924633026123, |
| "sampling/importance_sampling_ratio/max": 1.6572447419166565, |
| "sampling/importance_sampling_ratio/mean": 0.5404750406742096, |
| "sampling/importance_sampling_ratio/min": 0.019457601010799408, |
| "sampling/sampling_logp_difference/max": 0.7372019290924072, |
| "sampling/sampling_logp_difference/mean": 0.007984871277585626, |
| "step": 52, |
| "step_time": 140.17531206600688 |
| }, |
| { |
| "clip_ratio/high_max": 0.0001694170387054328, |
| "clip_ratio/high_mean": 0.0001694170387054328, |
| "clip_ratio/low_mean": 0.0002777307599899359, |
| "clip_ratio/low_min": 0.0002777307599899359, |
| "clip_ratio/region_mean": 0.0004471477986953687, |
| "completions/clipped_ratio": 0.25, |
| "completions/max_length": 5248.0, |
| "completions/max_terminated_length": 3680.0, |
| "completions/mean_length": 2798.2501220703125, |
| "completions/mean_terminated_length": 1981.6666870117188, |
| "completions/min_length": 879.0, |
| "completions/min_terminated_length": 879.0, |
| "entropy": 0.19165829755365849, |
| "epoch": 0.36177474402730375, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.01472468115389347, |
| "learning_rate": 2.8716886898088355e-07, |
| "loss": -0.03431159257888794, |
| "num_tokens": 4451944.0, |
| "reward": -0.5833333358168602, |
| "reward_std": 0.5149286389350891, |
| "rewards/correctness_reward/mean": -0.5833333358168602, |
| "rewards/correctness_reward/std": 0.5149286389350891, |
| "sampling/importance_sampling_ratio/max": 1.4794809222221375, |
| "sampling/importance_sampling_ratio/mean": 0.6550164222717285, |
| "sampling/importance_sampling_ratio/min": 0.03906719759106636, |
| "sampling/sampling_logp_difference/max": 0.6392964124679565, |
| "sampling/sampling_logp_difference/mean": 0.009628844913095236, |
| "step": 53, |
| "step_time": 170.42672173799656 |
| }, |
| { |
| "clip_ratio/high_max": 0.00013047829088463914, |
| "clip_ratio/high_mean": 0.00013047829088463914, |
| "clip_ratio/low_mean": 0.0011631374727585353, |
| "clip_ratio/low_min": 0.0011631374727585353, |
| "clip_ratio/region_mean": 0.0012936157727381214, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 5248.0, |
| "completions/max_terminated_length": 4837.5, |
| "completions/mean_length": 2533.08349609375, |
| "completions/mean_terminated_length": 2286.2727661132812, |
| "completions/min_length": 865.0, |
| "completions/min_terminated_length": 865.0, |
| "entropy": 0.206689840182662, |
| "epoch": 0.36860068259385664, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.034136850386857986, |
| "learning_rate": 2.864766745078246e-07, |
| "loss": 0.03963364660739899, |
| "num_tokens": 4551342.0, |
| "reward": -0.6666666865348816, |
| "reward_std": 0.7784989774227142, |
| "rewards/correctness_reward/mean": -0.6666666865348816, |
| "rewards/correctness_reward/std": 0.7784989774227142, |
| "sampling/importance_sampling_ratio/max": 2.0958030223846436, |
| "sampling/importance_sampling_ratio/mean": 0.6207669377326965, |
| "sampling/importance_sampling_ratio/min": 0.010971732437610626, |
| "sampling/sampling_logp_difference/max": 0.5716677308082581, |
| "sampling/sampling_logp_difference/mean": 0.009138945490121841, |
| "step": 54, |
| "step_time": 169.02359899500152 |
| }, |
| { |
| "clip_ratio/high_max": 0.00012305460040806793, |
| "clip_ratio/high_mean": 0.00012305460040806793, |
| "clip_ratio/low_mean": 0.0003682131282403134, |
| "clip_ratio/low_min": 0.0003682131282403134, |
| "clip_ratio/region_mean": 0.0004912677286483813, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 3839.0, |
| "completions/max_terminated_length": 3335.5, |
| "completions/mean_length": 1927.2083740234375, |
| "completions/mean_terminated_length": 1801.401611328125, |
| "completions/min_length": 784.0, |
| "completions/min_terminated_length": 784.0, |
| "entropy": 0.19140255451202393, |
| "epoch": 0.37542662116040953, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.0412936732172966, |
| "learning_rate": 2.857671760875482e-07, |
| "loss": 0.2316821962594986, |
| "num_tokens": 4634279.0, |
| "reward": -0.8333333432674408, |
| "reward_std": 0.38924944400787354, |
| "rewards/correctness_reward/mean": -0.8333333432674408, |
| "rewards/correctness_reward/std": 0.3892495036125183, |
| "sampling/importance_sampling_ratio/max": 1.9013534784317017, |
| "sampling/importance_sampling_ratio/mean": 0.7683794945478439, |
| "sampling/importance_sampling_ratio/min": 0.052524975035339594, |
| "sampling/sampling_logp_difference/max": 0.5893399715423584, |
| "sampling/sampling_logp_difference/mean": 0.009064876940101385, |
| "step": 55, |
| "step_time": 126.5851471220085 |
| }, |
| { |
| "clip_ratio/high_max": 6.827863762737252e-05, |
| "clip_ratio/high_mean": 6.827863762737252e-05, |
| "clip_ratio/low_mean": 0.00023662939202040434, |
| "clip_ratio/low_min": 0.00023662939202040434, |
| "clip_ratio/region_mean": 0.00030490803328575566, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 5248.0, |
| "completions/max_terminated_length": 4025.0, |
| "completions/mean_length": 2566.3751220703125, |
| "completions/mean_terminated_length": 2322.5909423828125, |
| "completions/min_length": 634.0, |
| "completions/min_terminated_length": 634.0, |
| "entropy": 0.22426036186516285, |
| "epoch": 0.3822525597269625, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.017772767692804337, |
| "learning_rate": 2.850404636777212e-07, |
| "loss": 0.10956238210201263, |
| "num_tokens": 4733990.0, |
| "reward": -0.8333333432674408, |
| "reward_std": 0.3892495036125183, |
| "rewards/correctness_reward/mean": -0.8333333432674408, |
| "rewards/correctness_reward/std": 0.3892495036125183, |
| "sampling/importance_sampling_ratio/max": 1.8248947858810425, |
| "sampling/importance_sampling_ratio/mean": 0.4554801285266876, |
| "sampling/importance_sampling_ratio/min": 0.000969524378888309, |
| "sampling/sampling_logp_difference/max": 0.7188509702682495, |
| "sampling/sampling_logp_difference/mean": 0.009816299192607403, |
| "step": 56, |
| "step_time": 167.3950643390126 |
| }, |
| { |
| "clip_ratio/high_max": 0.00019467264428385533, |
| "clip_ratio/high_mean": 0.00019467264428385533, |
| "clip_ratio/low_mean": 0.0008429501613136381, |
| "clip_ratio/low_min": 0.0008429501613136381, |
| "clip_ratio/region_mean": 0.0010376228055974934, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 4506.5, |
| "completions/max_terminated_length": 3469.5, |
| "completions/mean_length": 2122.791748046875, |
| "completions/mean_terminated_length": 1973.5531616210938, |
| "completions/min_length": 829.0, |
| "completions/min_terminated_length": 829.0, |
| "entropy": 0.2076687552034855, |
| "epoch": 0.3890784982935154, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.056109074503183365, |
| "learning_rate": 2.842966294185809e-07, |
| "loss": -0.13481320440769196, |
| "num_tokens": 4833657.0, |
| "reward": -0.583333358168602, |
| "reward_std": 0.781041145324707, |
| "rewards/correctness_reward/mean": -0.5833333283662796, |
| "rewards/correctness_reward/std": 0.7810411155223846, |
| "sampling/importance_sampling_ratio/max": 1.5594306886196136, |
| "sampling/importance_sampling_ratio/mean": 0.45633988082408905, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.6542015075683594, |
| "sampling/sampling_logp_difference/mean": 0.009512615855783224, |
| "step": 57, |
| "step_time": 149.88567502400838 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3770.5, |
| "completions/max_terminated_length": 3770.5, |
| "completions/mean_length": 1786.1250610351562, |
| "completions/mean_terminated_length": 1786.1250610351562, |
| "completions/min_length": 765.5, |
| "completions/min_terminated_length": 765.5, |
| "entropy": 0.18106183130294085, |
| "epoch": 0.39590443686006827, |
| "frac_reward_zero_std": 1.0, |
| "grad_norm": 0.0, |
| "learning_rate": 2.835357676212527e-07, |
| "loss": 0.0, |
| "num_tokens": 4909302.0, |
| "reward": -1.0, |
| "reward_std": 0.0, |
| "rewards/correctness_reward/mean": -1.0, |
| "rewards/correctness_reward/std": 0.0, |
| "sampling/importance_sampling_ratio/max": 1.9227526783943176, |
| "sampling/importance_sampling_ratio/mean": 0.3771863132715225, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.8851069211959839, |
| "sampling/sampling_logp_difference/mean": 0.009029736276715994, |
| "step": 58, |
| "step_time": 120.19218669401016 |
| }, |
| { |
| "clip_ratio/high_max": 0.00031629430486646015, |
| "clip_ratio/high_mean": 0.00031629430486646015, |
| "clip_ratio/low_mean": 0.0007711920734436717, |
| "clip_ratio/low_min": 0.0007711920734436717, |
| "clip_ratio/region_mean": 0.0010874863874050789, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3482.5, |
| "completions/max_terminated_length": 3482.5, |
| "completions/mean_length": 1716.0000610351562, |
| "completions/mean_terminated_length": 1716.0000610351562, |
| "completions/min_length": 797.0, |
| "completions/min_terminated_length": 797.0, |
| "entropy": 0.1720494432374835, |
| "epoch": 0.40273037542662116, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.04583508521318436, |
| "learning_rate": 2.827579747557923e-07, |
| "loss": 0.1589364856481552, |
| "num_tokens": 4978320.0, |
| "reward": -0.5, |
| "reward_std": 0.9045339822769165, |
| "rewards/correctness_reward/mean": -0.5, |
| "rewards/correctness_reward/std": 0.9045340418815613, |
| "sampling/importance_sampling_ratio/max": 1.5915037989616394, |
| "sampling/importance_sampling_ratio/mean": 0.5548273622989655, |
| "sampling/importance_sampling_ratio/min": 0.009591552894562483, |
| "sampling/sampling_logp_difference/max": 0.9339792132377625, |
| "sampling/sampling_logp_difference/mean": 0.00888055469840765, |
| "step": 59, |
| "step_time": 110.96668717899593 |
| }, |
| { |
| "clip_ratio/high_max": 0.0003037179012608249, |
| "clip_ratio/high_mean": 0.0003037179012608249, |
| "clip_ratio/low_mean": 0.0006711918067594524, |
| "clip_ratio/low_min": 0.0006711918067594524, |
| "clip_ratio/region_mean": 0.0009749097152962349, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 4913.0, |
| "completions/max_terminated_length": 4913.0, |
| "completions/mean_length": 1996.2083740234375, |
| "completions/mean_terminated_length": 1996.2083740234375, |
| "completions/min_length": 917.5, |
| "completions/min_terminated_length": 917.5, |
| "entropy": 0.1703935917466879, |
| "epoch": 0.40955631399317405, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.029283197596669197, |
| "learning_rate": 2.81963349438954e-07, |
| "loss": -0.008943863213062286, |
| "num_tokens": 5062631.0, |
| "reward": -0.583333358168602, |
| "reward_std": 0.7810411155223846, |
| "rewards/correctness_reward/mean": -0.5833333283662796, |
| "rewards/correctness_reward/std": 0.7810411155223846, |
| "sampling/importance_sampling_ratio/max": 1.8522260189056396, |
| "sampling/importance_sampling_ratio/mean": 0.4740467965602875, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 1.0869746208190918, |
| "sampling/sampling_logp_difference/mean": 0.007830763701349497, |
| "step": 60, |
| "step_time": 154.14188899399596 |
| }, |
| { |
| "clip_ratio/high_max": 5.544539453694597e-05, |
| "clip_ratio/high_mean": 5.544539453694597e-05, |
| "clip_ratio/low_mean": 0.00029268264552229084, |
| "clip_ratio/low_min": 0.00029268264552229084, |
| "clip_ratio/region_mean": 0.0003481280436972156, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3764.5, |
| "completions/max_terminated_length": 3764.5, |
| "completions/mean_length": 1855.75, |
| "completions/mean_terminated_length": 1855.75, |
| "completions/min_length": 918.5, |
| "completions/min_terminated_length": 918.5, |
| "entropy": 0.16261542588472366, |
| "epoch": 0.41638225255972694, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.01945802941918373, |
| "learning_rate": 2.811519924216873e-07, |
| "loss": 0.05384628474712372, |
| "num_tokens": 5140289.0, |
| "reward": -0.3333333432674408, |
| "reward_std": 0.91148242354393, |
| "rewards/correctness_reward/mean": -0.3333333432674408, |
| "rewards/correctness_reward/std": 0.91148242354393, |
| "sampling/importance_sampling_ratio/max": 1.528772234916687, |
| "sampling/importance_sampling_ratio/mean": 0.4670892059803009, |
| "sampling/importance_sampling_ratio/min": 0.05911402031779289, |
| "sampling/sampling_logp_difference/max": 0.6520828008651733, |
| "sampling/sampling_logp_difference/mean": 0.007892156019806862, |
| "step": 61, |
| "step_time": 121.0787803270141 |
| }, |
| { |
| "clip_ratio/high_max": 4.087170600541867e-05, |
| "clip_ratio/high_mean": 4.087170600541867e-05, |
| "clip_ratio/low_mean": 0.0002310327981831506, |
| "clip_ratio/low_min": 0.0002310327981831506, |
| "clip_ratio/region_mean": 0.00027190449873160105, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3292.5, |
| "completions/max_terminated_length": 3292.5, |
| "completions/mean_length": 1349.1666870117188, |
| "completions/mean_terminated_length": 1349.1666870117188, |
| "completions/min_length": 692.5, |
| "completions/min_terminated_length": 692.5, |
| "entropy": 0.16260085813701153, |
| "epoch": 0.4232081911262799, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.03200114518404007, |
| "learning_rate": 2.8032400657636255e-07, |
| "loss": -0.13627246022224426, |
| "num_tokens": 5202531.0, |
| "reward": -0.8333333432674408, |
| "reward_std": 0.38924944400787354, |
| "rewards/correctness_reward/mean": -0.8333333432674408, |
| "rewards/correctness_reward/std": 0.3892495036125183, |
| "sampling/importance_sampling_ratio/max": 0.8998820185661316, |
| "sampling/importance_sampling_ratio/mean": 0.38338422775268555, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.6040194034576416, |
| "sampling/sampling_logp_difference/mean": 0.008680101484060287, |
| "step": 62, |
| "step_time": 103.78884234600264 |
| }, |
| { |
| "clip_ratio/high_max": 8.822122617857531e-05, |
| "clip_ratio/high_mean": 8.822122617857531e-05, |
| "clip_ratio/low_mean": 2.095557465509046e-05, |
| "clip_ratio/low_min": 2.095557465509046e-05, |
| "clip_ratio/region_mean": 0.00010917680083366577, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3387.0, |
| "completions/max_terminated_length": 3387.0, |
| "completions/mean_length": 1841.0833740234375, |
| "completions/mean_terminated_length": 1841.0833740234375, |
| "completions/min_length": 961.5, |
| "completions/min_terminated_length": 961.5, |
| "entropy": 0.18213766440749168, |
| "epoch": 0.4300341296928328, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.012850549072027206, |
| "learning_rate": 2.7947949688372767e-07, |
| "loss": 0.007055250927805901, |
| "num_tokens": 5286647.0, |
| "reward": -0.6666666716337204, |
| "reward_std": 0.4923659861087799, |
| "rewards/correctness_reward/mean": -0.6666666716337204, |
| "rewards/correctness_reward/std": 0.4923659861087799, |
| "sampling/importance_sampling_ratio/max": 1.8162547945976257, |
| "sampling/importance_sampling_ratio/mean": 0.5006676018238068, |
| "sampling/importance_sampling_ratio/min": 0.015822945162653923, |
| "sampling/sampling_logp_difference/max": 0.787606954574585, |
| "sampling/sampling_logp_difference/mean": 0.008724508807063103, |
| "step": 63, |
| "step_time": 115.6498547349911 |
| }, |
| { |
| "clip_ratio/high_max": 8.47601259010844e-05, |
| "clip_ratio/high_mean": 8.47601259010844e-05, |
| "clip_ratio/low_mean": 0.0001956572632479947, |
| "clip_ratio/low_min": 0.0001956572632479947, |
| "clip_ratio/region_mean": 0.0002804173818731215, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 5248.0, |
| "completions/max_terminated_length": 3345.0, |
| "completions/mean_length": 2251.666748046875, |
| "completions/mean_terminated_length": 1979.2727661132812, |
| "completions/min_length": 803.0, |
| "completions/min_terminated_length": 803.0, |
| "entropy": 0.19157697446644306, |
| "epoch": 0.43686006825938567, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.022400015965104103, |
| "learning_rate": 2.786185704195976e-07, |
| "loss": 0.15593445301055908, |
| "num_tokens": 5381205.0, |
| "reward": -0.9166666865348816, |
| "reward_std": 0.2886751592159271, |
| "rewards/correctness_reward/mean": -0.9166666567325592, |
| "rewards/correctness_reward/std": 0.28867512941360474, |
| "sampling/importance_sampling_ratio/max": 1.956946074962616, |
| "sampling/importance_sampling_ratio/mean": 0.652755469083786, |
| "sampling/importance_sampling_ratio/min": 0.0161164328455925, |
| "sampling/sampling_logp_difference/max": 0.4603617191314697, |
| "sampling/sampling_logp_difference/mean": 0.009298257529735565, |
| "step": 64, |
| "step_time": 166.16724363300455 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3447.5, |
| "completions/max_terminated_length": 3447.5, |
| "completions/mean_length": 1786.75, |
| "completions/mean_terminated_length": 1786.75, |
| "completions/min_length": 725.5, |
| "completions/min_terminated_length": 725.5, |
| "entropy": 0.18305392190814018, |
| "epoch": 0.44368600682593856, |
| "frac_reward_zero_std": 1.0, |
| "grad_norm": 0.0, |
| "learning_rate": 2.7774133634127823e-07, |
| "loss": 0.0, |
| "num_tokens": 5453997.0, |
| "reward": -1.0, |
| "reward_std": 0.0, |
| "rewards/correctness_reward/mean": -1.0, |
| "rewards/correctness_reward/std": 0.0, |
| "sampling/importance_sampling_ratio/max": 1.5538569688796997, |
| "sampling/importance_sampling_ratio/mean": 0.5863878726959229, |
| "sampling/importance_sampling_ratio/min": 0.018695535138249397, |
| "sampling/sampling_logp_difference/max": 0.466518759727478, |
| "sampling/sampling_logp_difference/mean": 0.00866021541878581, |
| "step": 65, |
| "step_time": 110.4639795779949 |
| }, |
| { |
| "clip_ratio/high_max": 0.0001480766040913295, |
| "clip_ratio/high_mean": 0.0001480766040913295, |
| "clip_ratio/low_mean": 0.0002666572872840334, |
| "clip_ratio/low_min": 0.0002666572872840334, |
| "clip_ratio/region_mean": 0.0004147338913753629, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 4059.0, |
| "completions/max_terminated_length": 4059.0, |
| "completions/mean_length": 2139.2501220703125, |
| "completions/mean_terminated_length": 2139.2501220703125, |
| "completions/min_length": 799.0, |
| "completions/min_terminated_length": 799.0, |
| "entropy": 0.18904976546764374, |
| "epoch": 0.45051194539249145, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.04415519908070564, |
| "learning_rate": 2.7684790587372594e-07, |
| "loss": -0.35014355182647705, |
| "num_tokens": 5532975.0, |
| "reward": -0.4166666716337204, |
| "reward_std": 0.9446330070495605, |
| "rewards/correctness_reward/mean": -0.4166666716337204, |
| "rewards/correctness_reward/std": 0.9446330070495605, |
| "sampling/importance_sampling_ratio/max": 1.7579538822174072, |
| "sampling/importance_sampling_ratio/mean": 0.5573211014270782, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.5224248170852661, |
| "sampling/sampling_logp_difference/mean": 0.00864137476310134, |
| "step": 66, |
| "step_time": 128.31755204999354 |
| }, |
| { |
| "clip_ratio/high_max": 0.0003504481737763854, |
| "clip_ratio/high_mean": 0.0003504481737763854, |
| "clip_ratio/low_mean": 0.00035990733886137605, |
| "clip_ratio/low_min": 0.00035990733886137605, |
| "clip_ratio/region_mean": 0.0007103555126377614, |
| "completions/clipped_ratio": 0.125, |
| "completions/max_length": 4231.5, |
| "completions/max_terminated_length": 4021.5, |
| "completions/mean_length": 2462.5416870117188, |
| "completions/mean_terminated_length": 2111.1666870117188, |
| "completions/min_length": 1038.0, |
| "completions/min_terminated_length": 1038.0, |
| "entropy": 0.17603121791034937, |
| "epoch": 0.45733788395904434, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.043682895600795746, |
| "learning_rate": 2.7593839229544584e-07, |
| "loss": -0.15456409752368927, |
| "num_tokens": 5631004.0, |
| "reward": -0.5833333432674408, |
| "reward_std": 0.8415164649486542, |
| "rewards/correctness_reward/mean": -0.5833333432674408, |
| "rewards/correctness_reward/std": 0.841516524553299, |
| "sampling/importance_sampling_ratio/max": 2.3461341857910156, |
| "sampling/importance_sampling_ratio/mean": 0.5733579099178314, |
| "sampling/importance_sampling_ratio/min": 0.04074254631996155, |
| "sampling/sampling_logp_difference/max": 0.6068271398544312, |
| "sampling/sampling_logp_difference/mean": 0.008984747808426619, |
| "step": 67, |
| "step_time": 141.14267089701025 |
| }, |
| { |
| "clip_ratio/high_max": 0.0001881080479506636, |
| "clip_ratio/high_mean": 0.0001881080479506636, |
| "clip_ratio/low_mean": 0.00042036333616124466, |
| "clip_ratio/low_min": 0.00042036333616124466, |
| "clip_ratio/region_mean": 0.0006084713841119083, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 4071.5, |
| "completions/max_terminated_length": 2543.5, |
| "completions/mean_length": 1812.375, |
| "completions/mean_terminated_length": 1657.6931762695312, |
| "completions/min_length": 895.0, |
| "completions/min_terminated_length": 895.0, |
| "entropy": 0.16869750246405602, |
| "epoch": 0.4641638225255973, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.06864888221025467, |
| "learning_rate": 2.750129109241286e-07, |
| "loss": -0.16236957907676697, |
| "num_tokens": 5704339.0, |
| "reward": -0.4166666716337204, |
| "reward_std": 0.9446329772472382, |
| "rewards/correctness_reward/mean": -0.4166666716337204, |
| "rewards/correctness_reward/std": 0.9446330070495605, |
| "sampling/importance_sampling_ratio/max": 2.2440454363822937, |
| "sampling/importance_sampling_ratio/mean": 0.7120912075042725, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.5913791656494141, |
| "sampling/sampling_logp_difference/mean": 0.008479172829538584, |
| "step": 68, |
| "step_time": 127.05132082500495 |
| }, |
| { |
| "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": 0.0416666679084301, |
| "completions/max_length": 4289.0, |
| "completions/max_terminated_length": 3739.0, |
| "completions/mean_length": 2125.7500610351562, |
| "completions/mean_terminated_length": 1998.4622192382812, |
| "completions/min_length": 785.5, |
| "completions/min_terminated_length": 785.5, |
| "entropy": 0.2187929842621088, |
| "epoch": 0.4709897610921502, |
| "frac_reward_zero_std": 1.0, |
| "grad_norm": 0.0, |
| "learning_rate": 2.740715791020298e-07, |
| "loss": 0.0, |
| "num_tokens": 5796139.0, |
| "reward": -1.0, |
| "reward_std": 0.0, |
| "rewards/correctness_reward/mean": -1.0, |
| "rewards/correctness_reward/std": 0.0, |
| "sampling/importance_sampling_ratio/max": 2.015613377094269, |
| "sampling/importance_sampling_ratio/mean": 0.6376230865716934, |
| "sampling/importance_sampling_ratio/min": 0.016034055734053254, |
| "sampling/sampling_logp_difference/max": 0.5144100189208984, |
| "sampling/sampling_logp_difference/mean": 0.009602179750800133, |
| "step": 69, |
| "step_time": 140.69199190499057 |
| }, |
| { |
| "clip_ratio/high_max": 7.757430648780428e-05, |
| "clip_ratio/high_mean": 7.757430648780428e-05, |
| "clip_ratio/low_mean": 0.0004446139955689432, |
| "clip_ratio/low_min": 0.0004446139955689432, |
| "clip_ratio/region_mean": 0.0005221882984187687, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 4236.5, |
| "completions/max_terminated_length": 3391.5, |
| "completions/mean_length": 2219.5000610351562, |
| "completions/mean_terminated_length": 2107.5530395507812, |
| "completions/min_length": 1279.5, |
| "completions/min_terminated_length": 1279.5, |
| "entropy": 0.1923739816993475, |
| "epoch": 0.4778156996587031, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.0228436142206192, |
| "learning_rate": 2.731145161810915e-07, |
| "loss": -0.00633201003074646, |
| "num_tokens": 5885617.0, |
| "reward": -0.7500000298023224, |
| "reward_std": 0.677924633026123, |
| "rewards/correctness_reward/mean": -0.75, |
| "rewards/correctness_reward/std": 0.6779246032238007, |
| "sampling/importance_sampling_ratio/max": 1.157751977443695, |
| "sampling/importance_sampling_ratio/mean": 0.45236852765083313, |
| "sampling/importance_sampling_ratio/min": 0.06210612959694117, |
| "sampling/sampling_logp_difference/max": 1.0103888511657715, |
| "sampling/sampling_logp_difference/mean": 0.009076605550944805, |
| "step": 70, |
| "step_time": 136.6656262119941 |
| }, |
| { |
| "clip_ratio/high_max": 0.0002730607448029332, |
| "clip_ratio/high_mean": 0.0002730607448029332, |
| "clip_ratio/low_mean": 0.0010750940673460718, |
| "clip_ratio/low_min": 0.0010750940673460718, |
| "clip_ratio/region_mean": 0.0013481548157869838, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 3523.0, |
| "completions/max_terminated_length": 3138.0, |
| "completions/mean_length": 2051.291748046875, |
| "completions/mean_terminated_length": 1816.800048828125, |
| "completions/min_length": 989.5, |
| "completions/min_terminated_length": 989.5, |
| "entropy": 0.19321701675653458, |
| "epoch": 0.48464163822525597, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.07435967028141022, |
| "learning_rate": 2.7214184350780987e-07, |
| "loss": -0.22458279132843018, |
| "num_tokens": 5969330.0, |
| "reward": -0.1666666716337204, |
| "reward_std": 1.0298572778701782, |
| "rewards/correctness_reward/mean": -0.1666666716337204, |
| "rewards/correctness_reward/std": 1.0298572778701782, |
| "sampling/importance_sampling_ratio/max": 2.1131256818771362, |
| "sampling/importance_sampling_ratio/mean": 0.6783826649188995, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.5828158259391785, |
| "sampling/sampling_logp_difference/mean": 0.00917751993983984, |
| "step": 71, |
| "step_time": 117.55696539898781 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.00040589617856312543, |
| "clip_ratio/low_min": 0.00040589617856312543, |
| "clip_ratio/region_mean": 0.00040589617856312543, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3371.5, |
| "completions/max_terminated_length": 3371.5, |
| "completions/mean_length": 1673.375, |
| "completions/mean_terminated_length": 1673.375, |
| "completions/min_length": 821.5, |
| "completions/min_terminated_length": 821.5, |
| "entropy": 0.18138444982469082, |
| "epoch": 0.49146757679180886, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.07799665629863739, |
| "learning_rate": 2.711536844078497e-07, |
| "loss": -0.00800640881061554, |
| "num_tokens": 6043151.0, |
| "reward": -0.9166666865348816, |
| "reward_std": 0.28867512941360474, |
| "rewards/correctness_reward/mean": -0.9166666567325592, |
| "rewards/correctness_reward/std": 0.28867512941360474, |
| "sampling/importance_sampling_ratio/max": 1.7861913442611694, |
| "sampling/importance_sampling_ratio/mean": 0.6217601448297501, |
| "sampling/importance_sampling_ratio/min": 0.031388456001877785, |
| "sampling/sampling_logp_difference/max": 0.6879630088806152, |
| "sampling/sampling_logp_difference/mean": 0.008580193389207125, |
| "step": 72, |
| "step_time": 109.68210411000473 |
| }, |
| { |
| "clip_ratio/high_max": 0.00011866039858432487, |
| "clip_ratio/high_mean": 0.00011866039858432487, |
| "clip_ratio/low_mean": 0.0003503719690343132, |
| "clip_ratio/low_min": 0.0003503719690343132, |
| "clip_ratio/region_mean": 0.0004690323676186381, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 5204.5, |
| "completions/max_terminated_length": 4409.5, |
| "completions/mean_length": 2428.5001220703125, |
| "completions/mean_terminated_length": 2314.700927734375, |
| "completions/min_length": 777.5, |
| "completions/min_terminated_length": 777.5, |
| "entropy": 0.22688594833016396, |
| "epoch": 0.49829351535836175, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.028718864545226097, |
| "learning_rate": 2.7015016417040766e-07, |
| "loss": 0.02665596641600132, |
| "num_tokens": 6135209.0, |
| "reward": -0.7500000298023224, |
| "reward_std": 0.6779246032238007, |
| "rewards/correctness_reward/mean": -0.75, |
| "rewards/correctness_reward/std": 0.6779246032238007, |
| "sampling/importance_sampling_ratio/max": 0.949919581413269, |
| "sampling/importance_sampling_ratio/mean": 0.30701562762260437, |
| "sampling/importance_sampling_ratio/min": 0.004413304850459099, |
| "sampling/sampling_logp_difference/max": 0.6100250482559204, |
| "sampling/sampling_logp_difference/mean": 0.009902372024953365, |
| "step": 73, |
| "step_time": 162.6344251230039 |
| }, |
| { |
| "clip_ratio/high_max": 0.0002700084096431965, |
| "clip_ratio/high_mean": 0.0002700084096431965, |
| "clip_ratio/low_mean": 0.0005516534420166863, |
| "clip_ratio/low_min": 0.0005516534420166863, |
| "clip_ratio/region_mean": 0.0008216618552978616, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3201.0, |
| "completions/max_terminated_length": 3201.0, |
| "completions/mean_length": 1547.125, |
| "completions/mean_terminated_length": 1547.125, |
| "completions/min_length": 702.0, |
| "completions/min_terminated_length": 702.0, |
| "entropy": 0.1647388245910406, |
| "epoch": 0.5051194539249146, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.05796932429075241, |
| "learning_rate": 2.691314100323268e-07, |
| "loss": -0.05011383071541786, |
| "num_tokens": 6201476.0, |
| "reward": -0.4166666716337204, |
| "reward_std": 0.9446330070495605, |
| "rewards/correctness_reward/mean": -0.4166666716337204, |
| "rewards/correctness_reward/std": 0.9446330070495605, |
| "sampling/importance_sampling_ratio/max": 2.1914119720458984, |
| "sampling/importance_sampling_ratio/mean": 0.5255703777074814, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.827777624130249, |
| "sampling/sampling_logp_difference/mean": 0.008731815032660961, |
| "step": 74, |
| "step_time": 102.39683082199917 |
| }, |
| { |
| "clip_ratio/high_max": 0.0002842457906808704, |
| "clip_ratio/high_mean": 0.0002842457906808704, |
| "clip_ratio/low_mean": 0.0002800585643853992, |
| "clip_ratio/low_min": 0.0002800585643853992, |
| "clip_ratio/region_mean": 0.0005643043550662696, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 4044.5, |
| "completions/max_terminated_length": 4044.5, |
| "completions/mean_length": 1764.3750610351562, |
| "completions/mean_terminated_length": 1764.3750610351562, |
| "completions/min_length": 800.0, |
| "completions/min_terminated_length": 800.0, |
| "entropy": 0.19630672596395016, |
| "epoch": 0.5119453924914675, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.04505336284637451, |
| "learning_rate": 2.680975511619642e-07, |
| "loss": -0.05365559831261635, |
| "num_tokens": 6276863.0, |
| "reward": -0.5000000149011612, |
| "reward_std": 0.8816154599189758, |
| "rewards/correctness_reward/mean": -0.5000000149011612, |
| "rewards/correctness_reward/std": 0.8816154599189758, |
| "sampling/importance_sampling_ratio/max": 2.4316436648368835, |
| "sampling/importance_sampling_ratio/mean": 0.6821621358394623, |
| "sampling/importance_sampling_ratio/min": 0.060118645429611206, |
| "sampling/sampling_logp_difference/max": 0.9917401075363159, |
| "sampling/sampling_logp_difference/mean": 0.009168979246169329, |
| "step": 75, |
| "step_time": 128.50791397599096 |
| }, |
| { |
| "clip_ratio/high_max": 8.07665528554935e-05, |
| "clip_ratio/high_mean": 8.07665528554935e-05, |
| "clip_ratio/low_mean": 0.0001678744811215438, |
| "clip_ratio/low_min": 0.0001678744811215438, |
| "clip_ratio/region_mean": 0.0002486410339770373, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 4977.0, |
| "completions/max_terminated_length": 3904.5, |
| "completions/mean_length": 2161.7501220703125, |
| "completions/mean_terminated_length": 1859.7833862304688, |
| "completions/min_length": 882.0, |
| "completions/min_terminated_length": 882.0, |
| "entropy": 0.16465575248003006, |
| "epoch": 0.5187713310580204, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.0189689751714468, |
| "learning_rate": 2.670487186428138e-07, |
| "loss": 0.09658992290496826, |
| "num_tokens": 6362699.0, |
| "reward": -0.75, |
| "reward_std": 0.45226702094078064, |
| "rewards/correctness_reward/mean": -0.75, |
| "rewards/correctness_reward/std": 0.45226702094078064, |
| "sampling/importance_sampling_ratio/max": 1.0848638117313385, |
| "sampling/importance_sampling_ratio/mean": 0.5145403891801834, |
| "sampling/importance_sampling_ratio/min": 0.04004676640033722, |
| "sampling/sampling_logp_difference/max": 0.589957058429718, |
| "sampling/sampling_logp_difference/mean": 0.007860165322199464, |
| "step": 76, |
| "step_time": 154.08570145599515 |
| }, |
| { |
| "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": 0.2083333432674408, |
| "completions/max_length": 4109.0, |
| "completions/max_terminated_length": 3585.5, |
| "completions/mean_length": 2682.041748046875, |
| "completions/mean_terminated_length": 2152.9049072265625, |
| "completions/min_length": 1140.0, |
| "completions/min_terminated_length": 1140.0, |
| "entropy": 0.18603627104312181, |
| "epoch": 0.5255972696245734, |
| "frac_reward_zero_std": 1.0, |
| "grad_norm": 0.0, |
| "learning_rate": 2.659850454568859e-07, |
| "loss": 0.0, |
| "num_tokens": 6463632.0, |
| "reward": -1.0, |
| "reward_std": 0.0, |
| "rewards/correctness_reward/mean": -1.0, |
| "rewards/correctness_reward/std": 0.0, |
| "sampling/importance_sampling_ratio/max": 2.0550328493118286, |
| "sampling/importance_sampling_ratio/mean": 0.7448717653751373, |
| "sampling/importance_sampling_ratio/min": 0.004171021748334169, |
| "sampling/sampling_logp_difference/max": 0.5743434429168701, |
| "sampling/sampling_logp_difference/mean": 0.008833530824631453, |
| "step": 77, |
| "step_time": 138.94735664299515 |
| }, |
| { |
| "clip_ratio/high_max": 0.00019208645971957594, |
| "clip_ratio/high_mean": 0.00019208645971957594, |
| "clip_ratio/low_mean": 0.0003050619870919036, |
| "clip_ratio/low_min": 0.0003050619870919036, |
| "clip_ratio/region_mean": 0.0004971484468114795, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3626.0, |
| "completions/max_terminated_length": 3626.0, |
| "completions/mean_length": 1688.7500610351562, |
| "completions/mean_terminated_length": 1688.7500610351562, |
| "completions/min_length": 717.0, |
| "completions/min_terminated_length": 717.0, |
| "entropy": 0.19578155875205994, |
| "epoch": 0.5324232081911263, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.08791548758745193, |
| "learning_rate": 2.6490666646784667e-07, |
| "loss": -0.6725838780403137, |
| "num_tokens": 6537786.0, |
| "reward": -0.416666679084301, |
| "reward_std": 0.904178112745285, |
| "rewards/correctness_reward/mean": -0.416666679084301, |
| "rewards/correctness_reward/std": 0.9041781425476074, |
| "sampling/importance_sampling_ratio/max": 1.8620446026325226, |
| "sampling/importance_sampling_ratio/mean": 0.5930303931236267, |
| "sampling/importance_sampling_ratio/min": 0.021932126954197884, |
| "sampling/sampling_logp_difference/max": 0.49617767333984375, |
| "sampling/sampling_logp_difference/mean": 0.00903140613809228, |
| "step": 78, |
| "step_time": 116.53287122999609 |
| }, |
| { |
| "clip_ratio/high_max": 0.00013660525837622117, |
| "clip_ratio/high_mean": 0.00013660525837622117, |
| "clip_ratio/low_mean": 0.00034289806717424653, |
| "clip_ratio/low_min": 0.00034289806717424653, |
| "clip_ratio/region_mean": 0.0004795033182745101, |
| "completions/clipped_ratio": 0.1250000037252903, |
| "completions/max_length": 5248.0, |
| "completions/max_terminated_length": 4759.0, |
| "completions/mean_length": 2415.4583740234375, |
| "completions/mean_terminated_length": 2004.5682373046875, |
| "completions/min_length": 896.5, |
| "completions/min_terminated_length": 896.5, |
| "entropy": 0.19903319235891104, |
| "epoch": 0.5392491467576792, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.012782029807567596, |
| "learning_rate": 2.6381371840391864e-07, |
| "loss": 0.03598427399992943, |
| "num_tokens": 6634505.0, |
| "reward": -0.583333358168602, |
| "reward_std": 0.7810410857200623, |
| "rewards/correctness_reward/mean": -0.5833333283662796, |
| "rewards/correctness_reward/std": 0.7810411155223846, |
| "sampling/importance_sampling_ratio/max": 1.7369353771209717, |
| "sampling/importance_sampling_ratio/mean": 0.5799808204174042, |
| "sampling/importance_sampling_ratio/min": 0.027315656188875437, |
| "sampling/sampling_logp_difference/max": 0.6335808634757996, |
| "sampling/sampling_logp_difference/mean": 0.009694334119558334, |
| "step": 79, |
| "step_time": 165.85728075699444 |
| }, |
| { |
| "clip_ratio/high_max": 0.00034807420161087066, |
| "clip_ratio/high_mean": 0.00034807420161087066, |
| "clip_ratio/low_mean": 0.001164372471976094, |
| "clip_ratio/low_min": 0.001164372471976094, |
| "clip_ratio/region_mean": 0.0015124466590350494, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 2498.5, |
| "completions/max_terminated_length": 2498.5, |
| "completions/mean_length": 1373.4583740234375, |
| "completions/mean_terminated_length": 1373.4583740234375, |
| "completions/min_length": 766.5, |
| "completions/min_terminated_length": 766.5, |
| "entropy": 0.16406875103712082, |
| "epoch": 0.5460750853242321, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.09411022067070007, |
| "learning_rate": 2.627063398405445e-07, |
| "loss": -0.2403782606124878, |
| "num_tokens": 6699142.0, |
| "reward": -0.5833333432674408, |
| "reward_std": 0.841516524553299, |
| "rewards/correctness_reward/mean": -0.5833333432674408, |
| "rewards/correctness_reward/std": 0.841516524553299, |
| "sampling/importance_sampling_ratio/max": 1.8231101036071777, |
| "sampling/importance_sampling_ratio/mean": 0.6702441573143005, |
| "sampling/importance_sampling_ratio/min": 0.01660044677555561, |
| "sampling/sampling_logp_difference/max": 0.6353662014007568, |
| "sampling/sampling_logp_difference/mean": 0.0088970260694623, |
| "step": 80, |
| "step_time": 84.66325323101773 |
| }, |
| { |
| "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": 0.291666679084301, |
| "completions/max_length": 5248.0, |
| "completions/max_terminated_length": 2788.5, |
| "completions/mean_length": 2848.166748046875, |
| "completions/mean_terminated_length": 1871.357177734375, |
| "completions/min_length": 853.0, |
| "completions/min_terminated_length": 853.0, |
| "entropy": 0.14244998339563608, |
| "epoch": 0.552901023890785, |
| "frac_reward_zero_std": 1.0, |
| "grad_norm": 0.0, |
| "learning_rate": 2.6158467118281766e-07, |
| "loss": 0.0, |
| "num_tokens": 6799250.0, |
| "reward": -0.5, |
| "reward_std": 0.5222329497337341, |
| "rewards/correctness_reward/mean": -0.5, |
| "rewards/correctness_reward/std": 0.5222329497337341, |
| "sampling/importance_sampling_ratio/max": 1.8679000735282898, |
| "sampling/importance_sampling_ratio/mean": 0.7701037526130676, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.7022671699523926, |
| "sampling/sampling_logp_difference/mean": 0.008580916561186314, |
| "step": 81, |
| "step_time": 165.6021267400065 |
| }, |
| { |
| "clip_ratio/high_max": 1.6419282474089414e-05, |
| "clip_ratio/high_mean": 1.6419282474089414e-05, |
| "clip_ratio/low_mean": 0.00016853575289133005, |
| "clip_ratio/low_min": 0.00016853575289133005, |
| "clip_ratio/region_mean": 0.00018495503536541946, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 2679.5, |
| "completions/max_terminated_length": 2679.5, |
| "completions/mean_length": 1440.25, |
| "completions/mean_terminated_length": 1440.25, |
| "completions/min_length": 905.5, |
| "completions/min_terminated_length": 905.5, |
| "entropy": 0.15635963436216116, |
| "epoch": 0.5597269624573379, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.009361106902360916, |
| "learning_rate": 2.6044885464767957e-07, |
| "loss": 0.08625303953886032, |
| "num_tokens": 6865784.0, |
| "reward": -0.4166666641831398, |
| "reward_std": 0.5149286389350891, |
| "rewards/correctness_reward/mean": -0.4166666641831398, |
| "rewards/correctness_reward/std": 0.5149286389350891, |
| "sampling/importance_sampling_ratio/max": 1.5225063860416412, |
| "sampling/importance_sampling_ratio/mean": 0.482978492975235, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 1.0508806705474854, |
| "sampling/sampling_logp_difference/mean": 0.008569143246859312, |
| "step": 82, |
| "step_time": 88.88227481501235 |
| }, |
| { |
| "clip_ratio/high_max": 0.00012391141262924066, |
| "clip_ratio/high_mean": 0.00012391141262924066, |
| "clip_ratio/low_mean": 2.3360120394499972e-05, |
| "clip_ratio/low_min": 2.3360120394499972e-05, |
| "clip_ratio/region_mean": 0.00014727153302374063, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 4927.0, |
| "completions/max_terminated_length": 4851.0, |
| "completions/mean_length": 2604.541748046875, |
| "completions/mean_terminated_length": 2506.9129638671875, |
| "completions/min_length": 1156.0, |
| "completions/min_terminated_length": 1156.0, |
| "entropy": 0.222346480935812, |
| "epoch": 0.5665529010238908, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.009380086325109005, |
| "learning_rate": 2.5929903424588857e-07, |
| "loss": -0.015079451724886894, |
| "num_tokens": 6957447.0, |
| "reward": -0.6666666716337204, |
| "reward_std": 0.4923659563064575, |
| "rewards/correctness_reward/mean": -0.6666666716337204, |
| "rewards/correctness_reward/std": 0.4923659861087799, |
| "sampling/importance_sampling_ratio/max": 1.3343388438224792, |
| "sampling/importance_sampling_ratio/mean": 0.4296713322401047, |
| "sampling/importance_sampling_ratio/min": 0.014223325997591019, |
| "sampling/sampling_logp_difference/max": 0.621342658996582, |
| "sampling/sampling_logp_difference/mean": 0.009445840958505869, |
| "step": 83, |
| "step_time": 154.10503394300758 |
| }, |
| { |
| "clip_ratio/high_max": 3.4239539672853425e-05, |
| "clip_ratio/high_mean": 3.4239539672853425e-05, |
| "clip_ratio/low_mean": 0.0003129054239252582, |
| "clip_ratio/low_min": 0.0003129054239252582, |
| "clip_ratio/region_mean": 0.00034714496359811164, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 3495.0, |
| "completions/max_terminated_length": 2862.0, |
| "completions/mean_length": 1840.041748046875, |
| "completions/mean_terminated_length": 1710.3182373046875, |
| "completions/min_length": 768.0, |
| "completions/min_terminated_length": 768.0, |
| "entropy": 0.20816332660615444, |
| "epoch": 0.5733788395904437, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.06777017563581467, |
| "learning_rate": 2.581353557637602e-07, |
| "loss": -0.5427741408348083, |
| "num_tokens": 7043392.0, |
| "reward": -0.8333333432674408, |
| "reward_std": 0.3892494738101959, |
| "rewards/correctness_reward/mean": -0.8333333432674408, |
| "rewards/correctness_reward/std": 0.3892495036125183, |
| "sampling/importance_sampling_ratio/max": 2.5436028242111206, |
| "sampling/importance_sampling_ratio/mean": 0.7241137623786926, |
| "sampling/importance_sampling_ratio/min": 0.014458982273936272, |
| "sampling/sampling_logp_difference/max": 0.6590601205825806, |
| "sampling/sampling_logp_difference/mean": 0.00995600363239646, |
| "step": 84, |
| "step_time": 117.45463798900164 |
| }, |
| { |
| "clip_ratio/high_max": 0.00032514029953745194, |
| "clip_ratio/high_mean": 0.00032514029953745194, |
| "clip_ratio/low_mean": 5.560498175327666e-05, |
| "clip_ratio/low_min": 5.560498175327666e-05, |
| "clip_ratio/region_mean": 0.0003807452812907286, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 5248.0, |
| "completions/max_terminated_length": 3557.5, |
| "completions/mean_length": 2173.166748046875, |
| "completions/mean_terminated_length": 1893.6364135742188, |
| "completions/min_length": 872.0, |
| "completions/min_terminated_length": 872.0, |
| "entropy": 0.21564749628305435, |
| "epoch": 0.5802047781569966, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.009651628322899342, |
| "learning_rate": 2.56957966744683e-07, |
| "loss": 0.011977184563875198, |
| "num_tokens": 7132724.0, |
| "reward": -0.5833333358168602, |
| "reward_std": 0.5149286389350891, |
| "rewards/correctness_reward/mean": -0.5833333358168602, |
| "rewards/correctness_reward/std": 0.5149286389350891, |
| "sampling/importance_sampling_ratio/max": 1.0573405623435974, |
| "sampling/importance_sampling_ratio/mean": 0.36841432750225067, |
| "sampling/importance_sampling_ratio/min": 0.052347431890666485, |
| "sampling/sampling_logp_difference/max": 0.855263888835907, |
| "sampling/sampling_logp_difference/mean": 0.009965576231479645, |
| "step": 85, |
| "step_time": 162.9061557780151 |
| }, |
| { |
| "clip_ratio/high_max": 4.3267566070426255e-05, |
| "clip_ratio/high_mean": 4.3267566070426255e-05, |
| "clip_ratio/low_mean": 0.00032204676244873554, |
| "clip_ratio/low_min": 0.00032204676244873554, |
| "clip_ratio/region_mean": 0.0003653143357951194, |
| "completions/clipped_ratio": 0.125, |
| "completions/max_length": 3607.0, |
| "completions/max_terminated_length": 3038.5, |
| "completions/mean_length": 2108.3334350585938, |
| "completions/mean_terminated_length": 1769.8472290039062, |
| "completions/min_length": 1082.0, |
| "completions/min_terminated_length": 1082.0, |
| "entropy": 0.19981952756643295, |
| "epoch": 0.5870307167235495, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.04601443558931351, |
| "learning_rate": 2.557670164704119e-07, |
| "loss": 0.07306678593158722, |
| "num_tokens": 7226698.0, |
| "reward": -0.4166666641831398, |
| "reward_std": 0.5149286389350891, |
| "rewards/correctness_reward/mean": -0.4166666641831398, |
| "rewards/correctness_reward/std": 0.5149286389350891, |
| "sampling/importance_sampling_ratio/max": 1.9394845962524414, |
| "sampling/importance_sampling_ratio/mean": 0.650184690952301, |
| "sampling/importance_sampling_ratio/min": 0.03433588892221451, |
| "sampling/sampling_logp_difference/max": 0.633596658706665, |
| "sampling/sampling_logp_difference/mean": 0.009664933197200298, |
| "step": 86, |
| "step_time": 123.40111500900093 |
| }, |
| { |
| "clip_ratio/high_max": 0.00013916854732087813, |
| "clip_ratio/high_mean": 0.00013916854732087813, |
| "clip_ratio/low_mean": 0.0005129784458404174, |
| "clip_ratio/low_min": 0.0005129784458404174, |
| "clip_ratio/region_mean": 0.0006521469986182638, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 2699.0, |
| "completions/max_terminated_length": 2699.0, |
| "completions/mean_length": 1503.5000610351562, |
| "completions/mean_terminated_length": 1503.5000610351562, |
| "completions/min_length": 725.0, |
| "completions/min_terminated_length": 725.0, |
| "entropy": 0.17778978310525417, |
| "epoch": 0.5938566552901023, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.06781456619501114, |
| "learning_rate": 2.5456265594213984e-07, |
| "loss": -0.1386936753988266, |
| "num_tokens": 7291168.0, |
| "reward": -0.5833333432674408, |
| "reward_std": 0.8415164947509766, |
| "rewards/correctness_reward/mean": -0.5833333432674408, |
| "rewards/correctness_reward/std": 0.8415164947509766, |
| "sampling/importance_sampling_ratio/max": 1.674412488937378, |
| "sampling/importance_sampling_ratio/mean": 0.516161248087883, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.7146804332733154, |
| "sampling/sampling_logp_difference/mean": 0.00882220035418868, |
| "step": 87, |
| "step_time": 89.02695259900065 |
| }, |
| { |
| "clip_ratio/high_max": 4.265726238372736e-05, |
| "clip_ratio/high_mean": 4.265726238372736e-05, |
| "clip_ratio/low_mean": 0.0002642589679453522, |
| "clip_ratio/low_min": 0.0002642589679453522, |
| "clip_ratio/region_mean": 0.00030691623032907955, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 3686.5, |
| "completions/max_terminated_length": 3596.0, |
| "completions/mean_length": 1874.291748046875, |
| "completions/mean_terminated_length": 1750.0682373046875, |
| "completions/min_length": 1003.5, |
| "completions/min_terminated_length": 1003.5, |
| "entropy": 0.1829511933028698, |
| "epoch": 0.6006825938566553, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.012510834261775017, |
| "learning_rate": 2.5334503786135296e-07, |
| "loss": 0.06788667291402817, |
| "num_tokens": 7379951.0, |
| "reward": -0.4166666865348816, |
| "reward_std": 0.8109080791473389, |
| "rewards/correctness_reward/mean": -0.4166666567325592, |
| "rewards/correctness_reward/std": 0.8109080791473389, |
| "sampling/importance_sampling_ratio/max": 1.515598714351654, |
| "sampling/importance_sampling_ratio/mean": 0.47678041458129883, |
| "sampling/importance_sampling_ratio/min": 0.011095719411969185, |
| "sampling/sampling_logp_difference/max": 0.580707848072052, |
| "sampling/sampling_logp_difference/mean": 0.00906964810565114, |
| "step": 88, |
| "step_time": 123.02009241499036 |
| }, |
| { |
| "clip_ratio/high_max": 9.884079736366402e-05, |
| "clip_ratio/high_mean": 9.884079736366402e-05, |
| "clip_ratio/low_mean": 0.00012933203470311128, |
| "clip_ratio/low_min": 0.00012933203470311128, |
| "clip_ratio/region_mean": 0.0002281728320667753, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3009.5, |
| "completions/max_terminated_length": 3009.5, |
| "completions/mean_length": 1605.9166870117188, |
| "completions/mean_terminated_length": 1605.9166870117188, |
| "completions/min_length": 824.5, |
| "completions/min_terminated_length": 824.5, |
| "entropy": 0.16808602958917618, |
| "epoch": 0.6075085324232082, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.03892357274889946, |
| "learning_rate": 2.521143166104692e-07, |
| "loss": -0.304046094417572, |
| "num_tokens": 7451169.0, |
| "reward": -0.75, |
| "reward_std": 0.45226702094078064, |
| "rewards/correctness_reward/mean": -0.75, |
| "rewards/correctness_reward/std": 0.45226702094078064, |
| "sampling/importance_sampling_ratio/max": 1.7958517670631409, |
| "sampling/importance_sampling_ratio/mean": 0.7084388136863708, |
| "sampling/importance_sampling_ratio/min": 0.08083060570061207, |
| "sampling/sampling_logp_difference/max": 0.5716913342475891, |
| "sampling/sampling_logp_difference/mean": 0.008326051756739616, |
| "step": 89, |
| "step_time": 100.39591565699811 |
| }, |
| { |
| "clip_ratio/high_max": 0.0003322830416436773, |
| "clip_ratio/high_mean": 0.0003322830416436773, |
| "clip_ratio/low_mean": 0.0004085483597009443, |
| "clip_ratio/low_min": 0.0004085483597009443, |
| "clip_ratio/region_mean": 0.0007408313977066427, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 4037.5, |
| "completions/max_terminated_length": 2841.0, |
| "completions/mean_length": 1738.7500610351562, |
| "completions/mean_terminated_length": 1585.0606689453125, |
| "completions/min_length": 826.5, |
| "completions/min_terminated_length": 826.5, |
| "entropy": 0.1912026945501566, |
| "epoch": 0.6143344709897611, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.02759329229593277, |
| "learning_rate": 2.50870648233264e-07, |
| "loss": 0.1086440235376358, |
| "num_tokens": 7526013.0, |
| "reward": -0.5000000149011612, |
| "reward_std": 0.8816154599189758, |
| "rewards/correctness_reward/mean": -0.5000000149011612, |
| "rewards/correctness_reward/std": 0.8816154301166534, |
| "sampling/importance_sampling_ratio/max": 2.0824206471443176, |
| "sampling/importance_sampling_ratio/mean": 0.7261651158332825, |
| "sampling/importance_sampling_ratio/min": 0.018040256574749947, |
| "sampling/sampling_logp_difference/max": 0.7164096236228943, |
| "sampling/sampling_logp_difference/mean": 0.009469267446547747, |
| "step": 90, |
| "step_time": 128.0492773010119 |
| }, |
| { |
| "clip_ratio/high_max": 0.0001939719804795459, |
| "clip_ratio/high_mean": 0.0001939719804795459, |
| "clip_ratio/low_mean": 0.0007475265156244859, |
| "clip_ratio/low_min": 0.0007475265156244859, |
| "clip_ratio/region_mean": 0.0009414984961040318, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3000.0, |
| "completions/max_terminated_length": 3000.0, |
| "completions/mean_length": 1316.6250610351562, |
| "completions/mean_terminated_length": 1316.6250610351562, |
| "completions/min_length": 819.5, |
| "completions/min_terminated_length": 819.5, |
| "entropy": 0.1598931821063161, |
| "epoch": 0.621160409556314, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.08526404947042465, |
| "learning_rate": 2.496141904150859e-07, |
| "loss": -0.23110927641391754, |
| "num_tokens": 7589220.0, |
| "reward": -0.8333333432674408, |
| "reward_std": 0.3892494738101959, |
| "rewards/correctness_reward/mean": -0.8333333432674408, |
| "rewards/correctness_reward/std": 0.3892495036125183, |
| "sampling/importance_sampling_ratio/max": 1.29092675447464, |
| "sampling/importance_sampling_ratio/mean": 0.47361135482788086, |
| "sampling/importance_sampling_ratio/min": 0.030142774805426598, |
| "sampling/sampling_logp_difference/max": 0.7877267003059387, |
| "sampling/sampling_logp_difference/mean": 0.008897948078811169, |
| "step": 91, |
| "step_time": 97.83714397101721 |
| }, |
| { |
| "clip_ratio/high_max": 0.0002154113790311385, |
| "clip_ratio/high_mean": 0.0002154113790311385, |
| "clip_ratio/low_mean": 0.0007493217526644003, |
| "clip_ratio/low_min": 0.0007493217526644003, |
| "clip_ratio/region_mean": 0.000964733137152507, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 4933.5, |
| "completions/max_terminated_length": 3842.5, |
| "completions/mean_length": 1734.9583740234375, |
| "completions/mean_terminated_length": 1582.9356689453125, |
| "completions/min_length": 715.5, |
| "completions/min_terminated_length": 715.5, |
| "entropy": 0.18329482711851597, |
| "epoch": 0.6279863481228669, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.10777463763952255, |
| "learning_rate": 2.4834510246286263e-07, |
| "loss": -0.3767480254173279, |
| "num_tokens": 7665203.0, |
| "reward": -0.4166666865348816, |
| "reward_std": 0.8109081089496613, |
| "rewards/correctness_reward/mean": -0.4166666567325592, |
| "rewards/correctness_reward/std": 0.8109080791473389, |
| "sampling/importance_sampling_ratio/max": 2.762686014175415, |
| "sampling/importance_sampling_ratio/mean": 0.737210601568222, |
| "sampling/importance_sampling_ratio/min": 0.0013398262672126293, |
| "sampling/sampling_logp_difference/max": 0.8068320751190186, |
| "sampling/sampling_logp_difference/mean": 0.009431012906134129, |
| "step": 92, |
| "step_time": 158.21941642699676 |
| }, |
| { |
| "clip_ratio/high_max": 9.0035306129721e-05, |
| "clip_ratio/high_mean": 9.0035306129721e-05, |
| "clip_ratio/low_mean": 7.182532135630026e-05, |
| "clip_ratio/low_min": 7.182532135630026e-05, |
| "clip_ratio/region_mean": 0.00016186062748602126, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 4526.0, |
| "completions/max_terminated_length": 3674.5, |
| "completions/mean_length": 1940.5833740234375, |
| "completions/mean_terminated_length": 1800.1136474609375, |
| "completions/min_length": 866.0, |
| "completions/min_terminated_length": 866.0, |
| "entropy": 0.2036778088659048, |
| "epoch": 0.6348122866894198, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.019959157332777977, |
| "learning_rate": 2.4706354528490365e-07, |
| "loss": -0.13522621989250183, |
| "num_tokens": 7746607.0, |
| "reward": -0.5833333358168602, |
| "reward_std": 0.5149286389350891, |
| "rewards/correctness_reward/mean": -0.5833333358168602, |
| "rewards/correctness_reward/std": 0.5149286389350891, |
| "sampling/importance_sampling_ratio/max": 1.7632864117622375, |
| "sampling/importance_sampling_ratio/mean": 0.6962140649557114, |
| "sampling/importance_sampling_ratio/min": 0.03561893478035927, |
| "sampling/sampling_logp_difference/max": 0.5904067754745483, |
| "sampling/sampling_logp_difference/mean": 0.009093993343412876, |
| "step": 93, |
| "step_time": 143.2192148319955 |
| }, |
| { |
| "clip_ratio/high_max": 0.0002615741323097609, |
| "clip_ratio/high_mean": 0.0002615741323097609, |
| "clip_ratio/low_mean": 0.00030253883960540406, |
| "clip_ratio/low_min": 0.00030253883960540406, |
| "clip_ratio/region_mean": 0.0005641129682771862, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 2530.5, |
| "completions/max_terminated_length": 2530.5, |
| "completions/mean_length": 1570.5, |
| "completions/mean_terminated_length": 1570.5, |
| "completions/min_length": 689.0, |
| "completions/min_terminated_length": 689.0, |
| "entropy": 0.18163765780627728, |
| "epoch": 0.6416382252559727, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.03428397327661514, |
| "learning_rate": 2.457696813704975e-07, |
| "loss": 0.15175065398216248, |
| "num_tokens": 7813495.0, |
| "reward": -0.5833333358168602, |
| "reward_std": 0.5149286389350891, |
| "rewards/correctness_reward/mean": -0.5833333358168602, |
| "rewards/correctness_reward/std": 0.5149286389350891, |
| "sampling/importance_sampling_ratio/max": 1.7397316694259644, |
| "sampling/importance_sampling_ratio/mean": 0.5945777893066406, |
| "sampling/importance_sampling_ratio/min": 0.020148619078099728, |
| "sampling/sampling_logp_difference/max": 0.7906639575958252, |
| "sampling/sampling_logp_difference/mean": 0.009107701480388641, |
| "step": 94, |
| "step_time": 85.37202533299569 |
| }, |
| { |
| "clip_ratio/high_max": 0.0002010115931625478, |
| "clip_ratio/high_mean": 0.0002010115931625478, |
| "clip_ratio/low_mean": 0.00027532511012395844, |
| "clip_ratio/low_min": 0.00027532511012395844, |
| "clip_ratio/region_mean": 0.00047633669964852743, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3140.5, |
| "completions/max_terminated_length": 3140.5, |
| "completions/mean_length": 1883.5, |
| "completions/mean_terminated_length": 1883.5, |
| "completions/min_length": 952.0, |
| "completions/min_terminated_length": 952.0, |
| "entropy": 0.19250520505011082, |
| "epoch": 0.6484641638225256, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.04572583734989166, |
| "learning_rate": 2.4446367476931036e-07, |
| "loss": 0.1977643370628357, |
| "num_tokens": 7889431.0, |
| "reward": -0.4166666716337204, |
| "reward_std": 0.9446330070495605, |
| "rewards/correctness_reward/mean": -0.4166666716337204, |
| "rewards/correctness_reward/std": 0.9446330070495605, |
| "sampling/importance_sampling_ratio/max": 2.768944263458252, |
| "sampling/importance_sampling_ratio/mean": 0.7922740876674652, |
| "sampling/importance_sampling_ratio/min": 0.029824020341038704, |
| "sampling/sampling_logp_difference/max": 0.5023660659790039, |
| "sampling/sampling_logp_difference/mean": 0.008869106881320477, |
| "step": 95, |
| "step_time": 103.88041676799912 |
| }, |
| { |
| "clip_ratio/high_max": 3.110226316493936e-05, |
| "clip_ratio/high_mean": 3.110226316493936e-05, |
| "clip_ratio/low_mean": 0.00014555339657817967, |
| "clip_ratio/low_min": 0.00014555339657817967, |
| "clip_ratio/region_mean": 0.00017665565610514022, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 4723.0, |
| "completions/max_terminated_length": 4356.0, |
| "completions/mean_length": 2388.125, |
| "completions/mean_terminated_length": 2270.897705078125, |
| "completions/min_length": 867.5, |
| "completions/min_terminated_length": 867.5, |
| "entropy": 0.2005954124033451, |
| "epoch": 0.6552901023890785, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.009689918719232082, |
| "learning_rate": 2.4314569107058573e-07, |
| "loss": 0.007732723373919725, |
| "num_tokens": 7979722.0, |
| "reward": -0.8333333432674408, |
| "reward_std": 0.38924944400787354, |
| "rewards/correctness_reward/mean": -0.8333333432674408, |
| "rewards/correctness_reward/std": 0.3892495036125183, |
| "sampling/importance_sampling_ratio/max": 1.914838969707489, |
| "sampling/importance_sampling_ratio/mean": 0.43378978967666626, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.5895647406578064, |
| "sampling/sampling_logp_difference/mean": 0.008874210994690657, |
| "step": 96, |
| "step_time": 149.69903900801 |
| }, |
| { |
| "clip_ratio/high_max": 8.717645869182888e-05, |
| "clip_ratio/high_mean": 8.717645869182888e-05, |
| "clip_ratio/low_mean": 0.00030598924422520213, |
| "clip_ratio/low_min": 0.00030598924422520213, |
| "clip_ratio/region_mean": 0.0003931656974600628, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 4195.5, |
| "completions/max_terminated_length": 2847.5, |
| "completions/mean_length": 1947.416748046875, |
| "completions/mean_terminated_length": 1795.2197875976562, |
| "completions/min_length": 882.5, |
| "completions/min_terminated_length": 882.5, |
| "entropy": 0.15587482787668705, |
| "epoch": 0.6621160409556314, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.022968977689743042, |
| "learning_rate": 2.418158973821494e-07, |
| "loss": -0.006974846124649048, |
| "num_tokens": 8060606.0, |
| "reward": -0.6666666716337204, |
| "reward_std": 0.4923659861087799, |
| "rewards/correctness_reward/mean": -0.6666666716337204, |
| "rewards/correctness_reward/std": 0.4923659563064575, |
| "sampling/importance_sampling_ratio/max": 1.4343634843826294, |
| "sampling/importance_sampling_ratio/mean": 0.5058686882257462, |
| "sampling/importance_sampling_ratio/min": 0.018607817590236664, |
| "sampling/sampling_logp_difference/max": 0.613651305437088, |
| "sampling/sampling_logp_difference/mean": 0.007983496179804206, |
| "step": 97, |
| "step_time": 133.11254283601738 |
| }, |
| { |
| "clip_ratio/high_max": 5.8293175243306905e-05, |
| "clip_ratio/high_mean": 5.8293175243306905e-05, |
| "clip_ratio/low_mean": 0.00016594905901001766, |
| "clip_ratio/low_min": 0.00016594905901001766, |
| "clip_ratio/region_mean": 0.00022424223425332457, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 4004.0, |
| "completions/max_terminated_length": 4002.5, |
| "completions/mean_length": 2277.5834350585938, |
| "completions/mean_terminated_length": 2169.5416870117188, |
| "completions/min_length": 847.0, |
| "completions/min_terminated_length": 847.0, |
| "entropy": 0.16203098744153976, |
| "epoch": 0.6689419795221843, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.013039804995059967, |
| "learning_rate": 2.4047446230922195e-07, |
| "loss": 0.12956568598747253, |
| "num_tokens": 8148760.0, |
| "reward": -0.9166666865348816, |
| "reward_std": 0.2886751592159271, |
| "rewards/correctness_reward/mean": -0.9166666567325592, |
| "rewards/correctness_reward/std": 0.28867512941360474, |
| "sampling/importance_sampling_ratio/max": 1.7119688391685486, |
| "sampling/importance_sampling_ratio/mean": 0.514645516872406, |
| "sampling/importance_sampling_ratio/min": 0.017148795071989298, |
| "sampling/sampling_logp_difference/max": 0.690927267074585, |
| "sampling/sampling_logp_difference/mean": 0.007868173997849226, |
| "step": 98, |
| "step_time": 130.74494820101245 |
| }, |
| { |
| "clip_ratio/high_max": 7.856693991925567e-05, |
| "clip_ratio/high_mean": 7.856693991925567e-05, |
| "clip_ratio/low_mean": 0.0005291322886478156, |
| "clip_ratio/low_min": 0.0005291322886478156, |
| "clip_ratio/region_mean": 0.0006076992212911136, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 1717.5, |
| "completions/max_terminated_length": 1717.5, |
| "completions/mean_length": 1047.8333740234375, |
| "completions/mean_terminated_length": 1047.8333740234375, |
| "completions/min_length": 759.0, |
| "completions/min_terminated_length": 759.0, |
| "entropy": 0.14885542448610067, |
| "epoch": 0.6757679180887372, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.09916958212852478, |
| "learning_rate": 2.391215559330409e-07, |
| "loss": -0.1368696540594101, |
| "num_tokens": 8202576.0, |
| "reward": -0.75, |
| "reward_std": 0.452267050743103, |
| "rewards/correctness_reward/mean": -0.75, |
| "rewards/correctness_reward/std": 0.45226702094078064, |
| "sampling/importance_sampling_ratio/max": 2.7344236373901367, |
| "sampling/importance_sampling_ratio/mean": 0.7819750905036926, |
| "sampling/importance_sampling_ratio/min": 0.033862002193927765, |
| "sampling/sampling_logp_difference/max": 0.9169849455356598, |
| "sampling/sampling_logp_difference/mean": 0.009436240419745445, |
| "step": 99, |
| "step_time": 60.29736274699826 |
| }, |
| { |
| "clip_ratio/high_max": 6.98723870300455e-05, |
| "clip_ratio/high_mean": 6.98723870300455e-05, |
| "clip_ratio/low_mean": 0.0011807848823082168, |
| "clip_ratio/low_min": 0.0011807848823082168, |
| "clip_ratio/region_mean": 0.0012506572711572517, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3254.0, |
| "completions/max_terminated_length": 3254.0, |
| "completions/mean_length": 1745.5416870117188, |
| "completions/mean_terminated_length": 1745.5416870117188, |
| "completions/min_length": 971.0, |
| "completions/min_terminated_length": 971.0, |
| "entropy": 0.19362994097173214, |
| "epoch": 0.6825938566552902, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.0453786700963974, |
| "learning_rate": 2.3775734978929614e-07, |
| "loss": 0.11239898949861526, |
| "num_tokens": 8280883.0, |
| "reward": -0.3333333507180214, |
| "reward_std": 0.8036037981510162, |
| "rewards/correctness_reward/mean": -0.333333320915699, |
| "rewards/correctness_reward/std": 0.8036037683486938, |
| "sampling/importance_sampling_ratio/max": 1.9463191032409668, |
| "sampling/importance_sampling_ratio/mean": 0.6716760843992233, |
| "sampling/importance_sampling_ratio/min": 0.09884109068661928, |
| "sampling/sampling_logp_difference/max": 0.6764633655548096, |
| "sampling/sampling_logp_difference/mean": 0.009771426673978567, |
| "step": 100, |
| "step_time": 108.78550151999661 |
| }, |
| { |
| "clip_ratio/high_max": 4.9233836762141436e-05, |
| "clip_ratio/high_mean": 4.9233836762141436e-05, |
| "clip_ratio/low_mean": 0.0006928291040821932, |
| "clip_ratio/low_min": 0.0006928291040821932, |
| "clip_ratio/region_mean": 0.0007420629408443347, |
| "completions/clipped_ratio": 0.125, |
| "completions/max_length": 5040.5, |
| "completions/max_terminated_length": 4454.0, |
| "completions/mean_length": 2518.7501220703125, |
| "completions/mean_terminated_length": 2111.8612060546875, |
| "completions/min_length": 920.5, |
| "completions/min_terminated_length": 920.5, |
| "entropy": 0.1930021233856678, |
| "epoch": 0.689419795221843, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.03779743239283562, |
| "learning_rate": 2.3638201684638098e-07, |
| "loss": -0.3188192844390869, |
| "num_tokens": 8373763.0, |
| "reward": -0.7500000298023224, |
| "reward_std": 0.6779245734214783, |
| "rewards/correctness_reward/mean": -0.75, |
| "rewards/correctness_reward/std": 0.677924633026123, |
| "sampling/importance_sampling_ratio/max": 1.8769155144691467, |
| "sampling/importance_sampling_ratio/mean": 0.6328006386756897, |
| "sampling/importance_sampling_ratio/min": 0.0016347093041986227, |
| "sampling/sampling_logp_difference/max": 1.0652188658714294, |
| "sampling/sampling_logp_difference/mean": 0.008982821833342314, |
| "step": 101, |
| "step_time": 157.72760972299875 |
| }, |
| { |
| "clip_ratio/high_max": 0.0002456316269672243, |
| "clip_ratio/high_mean": 0.0002456316269672243, |
| "clip_ratio/low_mean": 0.0006088513764552772, |
| "clip_ratio/low_min": 0.0006088513764552772, |
| "clip_ratio/region_mean": 0.0008544830034225015, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 3448.0, |
| "completions/max_terminated_length": 3441.5, |
| "completions/mean_length": 1830.3333740234375, |
| "completions/mean_terminated_length": 1571.3250732421875, |
| "completions/min_length": 923.0, |
| "completions/min_terminated_length": 923.0, |
| "entropy": 0.17076147999614477, |
| "epoch": 0.6962457337883959, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.041363637894392014, |
| "learning_rate": 2.3499573148346115e-07, |
| "loss": 0.0010967161506414413, |
| "num_tokens": 8450487.0, |
| "reward": -0.6666666865348816, |
| "reward_std": 0.7409421503543854, |
| "rewards/correctness_reward/mean": -0.6666666567325592, |
| "rewards/correctness_reward/std": 0.7409421503543854, |
| "sampling/importance_sampling_ratio/max": 1.3017855882644653, |
| "sampling/importance_sampling_ratio/mean": 0.47573068737983704, |
| "sampling/importance_sampling_ratio/min": 0.011353067122399807, |
| "sampling/sampling_logp_difference/max": 0.5613093376159668, |
| "sampling/sampling_logp_difference/mean": 0.008331953082233667, |
| "step": 102, |
| "step_time": 113.68557855300605 |
| }, |
| { |
| "clip_ratio/high_max": 0.00023342900021816604, |
| "clip_ratio/high_mean": 0.00023342900021816604, |
| "clip_ratio/low_mean": 0.0005807926718262024, |
| "clip_ratio/low_min": 0.0005807926718262024, |
| "clip_ratio/region_mean": 0.0008142216684063897, |
| "completions/clipped_ratio": 0.1250000037252903, |
| "completions/max_length": 5248.0, |
| "completions/max_terminated_length": 3850.5, |
| "completions/mean_length": 2237.1666870117188, |
| "completions/mean_terminated_length": 1812.9091186523438, |
| "completions/min_length": 826.0, |
| "completions/min_terminated_length": 826.0, |
| "entropy": 0.20056550204753876, |
| "epoch": 0.7030716723549488, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.04666848108172417, |
| "learning_rate": 2.3359866946836543e-07, |
| "loss": -0.10311692208051682, |
| "num_tokens": 8540965.0, |
| "reward": -0.583333358168602, |
| "reward_std": 0.7810411155223846, |
| "rewards/correctness_reward/mean": -0.5833333283662796, |
| "rewards/correctness_reward/std": 0.7810411155223846, |
| "sampling/importance_sampling_ratio/max": 1.8114920854568481, |
| "sampling/importance_sampling_ratio/mean": 0.6123473793268204, |
| "sampling/importance_sampling_ratio/min": 0.04999285563826561, |
| "sampling/sampling_logp_difference/max": 0.6254404187202454, |
| "sampling/sampling_logp_difference/mean": 0.009581032674759626, |
| "step": 103, |
| "step_time": 162.55260506199556 |
| }, |
| { |
| "clip_ratio/high_max": 0.00010359214502386749, |
| "clip_ratio/high_mean": 0.00010359214502386749, |
| "clip_ratio/low_mean": 8.616077320766635e-05, |
| "clip_ratio/low_min": 8.616077320766635e-05, |
| "clip_ratio/region_mean": 0.00018975292186951265, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 5049.5, |
| "completions/max_terminated_length": 4768.0, |
| "completions/mean_length": 2366.416748046875, |
| "completions/mean_terminated_length": 2242.166748046875, |
| "completions/min_length": 899.0, |
| "completions/min_terminated_length": 899.0, |
| "entropy": 0.19171350076794624, |
| "epoch": 0.7098976109215017, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.032151710242033005, |
| "learning_rate": 2.3219100793529975e-07, |
| "loss": 0.07015686482191086, |
| "num_tokens": 8629127.0, |
| "reward": -0.25, |
| "reward_std": 0.9744999706745148, |
| "rewards/correctness_reward/mean": -0.25, |
| "rewards/correctness_reward/std": 0.9744999706745148, |
| "sampling/importance_sampling_ratio/max": 2.378260016441345, |
| "sampling/importance_sampling_ratio/mean": 0.6469814479351044, |
| "sampling/importance_sampling_ratio/min": 0.00819753110408783, |
| "sampling/sampling_logp_difference/max": 0.730820894241333, |
| "sampling/sampling_logp_difference/mean": 0.008754108101129532, |
| "step": 104, |
| "step_time": 156.84601822699187 |
| }, |
| { |
| "clip_ratio/high_max": 2.9079910746077076e-05, |
| "clip_ratio/high_mean": 2.9079910746077076e-05, |
| "clip_ratio/low_mean": 0.00017557437968207523, |
| "clip_ratio/low_min": 0.00017557437968207523, |
| "clip_ratio/region_mean": 0.0002046542867901735, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 2998.0, |
| "completions/max_terminated_length": 2998.0, |
| "completions/mean_length": 1943.7916870117188, |
| "completions/mean_terminated_length": 1943.7916870117188, |
| "completions/min_length": 969.5, |
| "completions/min_terminated_length": 969.5, |
| "entropy": 0.1940737683326006, |
| "epoch": 0.7167235494880546, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.02384648658335209, |
| "learning_rate": 2.3077292536238835e-07, |
| "loss": -0.05765508860349655, |
| "num_tokens": 8708640.0, |
| "reward": -0.9166666865348816, |
| "reward_std": 0.28867512941360474, |
| "rewards/correctness_reward/mean": -0.9166666567325592, |
| "rewards/correctness_reward/std": 0.28867512941360474, |
| "sampling/importance_sampling_ratio/max": 1.6162993907928467, |
| "sampling/importance_sampling_ratio/mean": 0.42033903300762177, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 3.9300663471221924, |
| "sampling/sampling_logp_difference/mean": 0.008964573498815298, |
| "step": 105, |
| "step_time": 101.75178025999776 |
| }, |
| { |
| "clip_ratio/high_max": 0.00018588699094834737, |
| "clip_ratio/high_mean": 0.00018588699094834737, |
| "clip_ratio/low_mean": 0.0007062332515488379, |
| "clip_ratio/low_min": 0.0007062332515488379, |
| "clip_ratio/region_mean": 0.000892120246135164, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3964.0, |
| "completions/max_terminated_length": 3964.0, |
| "completions/mean_length": 2023.166748046875, |
| "completions/mean_terminated_length": 2023.166748046875, |
| "completions/min_length": 870.0, |
| "completions/min_terminated_length": 870.0, |
| "entropy": 0.19571753405034542, |
| "epoch": 0.7235494880546075, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.04298916086554527, |
| "learning_rate": 2.2934460154904435e-07, |
| "loss": -0.20362494885921478, |
| "num_tokens": 8785792.0, |
| "reward": -0.5000000223517418, |
| "reward_std": 0.8036037981510162, |
| "rewards/correctness_reward/mean": -0.4999999925494194, |
| "rewards/correctness_reward/std": 0.8036037683486938, |
| "sampling/importance_sampling_ratio/max": 1.5308769345283508, |
| "sampling/importance_sampling_ratio/mean": 0.4027136564254761, |
| "sampling/importance_sampling_ratio/min": 0.05640080850571394, |
| "sampling/sampling_logp_difference/max": 0.7667943835258484, |
| "sampling/sampling_logp_difference/mean": 0.00927647016942501, |
| "step": 106, |
| "step_time": 125.02470587198331 |
| }, |
| { |
| "clip_ratio/high_max": 0.00026201896253041923, |
| "clip_ratio/high_mean": 0.00026201896253041923, |
| "clip_ratio/low_mean": 0.0005470625947054941, |
| "clip_ratio/low_min": 0.0005470625947054941, |
| "clip_ratio/region_mean": 0.0008090815663308604, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 4992.5, |
| "completions/max_terminated_length": 3967.5, |
| "completions/mean_length": 1849.5, |
| "completions/mean_terminated_length": 1689.727294921875, |
| "completions/min_length": 721.0, |
| "completions/min_terminated_length": 721.0, |
| "entropy": 0.17433863878250122, |
| "epoch": 0.7303754266211604, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.06949550658464432, |
| "learning_rate": 2.2790621759317312e-07, |
| "loss": -0.28901875019073486, |
| "num_tokens": 8857900.0, |
| "reward": -0.2500000074505806, |
| "reward_std": 1.0072945952415466, |
| "rewards/correctness_reward/mean": -0.2500000074505806, |
| "rewards/correctness_reward/std": 1.007294625043869, |
| "sampling/importance_sampling_ratio/max": 1.8129876852035522, |
| "sampling/importance_sampling_ratio/mean": 0.8787103593349457, |
| "sampling/importance_sampling_ratio/min": 0.06362062506377697, |
| "sampling/sampling_logp_difference/max": 0.5592628717422485, |
| "sampling/sampling_logp_difference/mean": 0.008585838600993156, |
| "step": 107, |
| "step_time": 149.98139648498181 |
| }, |
| { |
| "clip_ratio/high_max": 0.0001897049278341001, |
| "clip_ratio/high_mean": 0.0001897049278341001, |
| "clip_ratio/low_mean": 0.0005273691203910857, |
| "clip_ratio/low_min": 0.0005273691203910857, |
| "clip_ratio/region_mean": 0.0007170740500441752, |
| "completions/clipped_ratio": 0.2500000074505806, |
| "completions/max_length": 5248.0, |
| "completions/max_terminated_length": 3779.5, |
| "completions/mean_length": 2909.416748046875, |
| "completions/mean_terminated_length": 2244.6500244140625, |
| "completions/min_length": 1466.5, |
| "completions/min_terminated_length": 1466.5, |
| "entropy": 0.2094256915152073, |
| "epoch": 0.7372013651877133, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.041358582675457, |
| "learning_rate": 2.264579558682106e-07, |
| "loss": -0.10254944115877151, |
| "num_tokens": 8968760.0, |
| "reward": -0.5833333432674408, |
| "reward_std": 0.8415164947509766, |
| "rewards/correctness_reward/mean": -0.5833333432674408, |
| "rewards/correctness_reward/std": 0.8415164947509766, |
| "sampling/importance_sampling_ratio/max": 1.5969599485397339, |
| "sampling/importance_sampling_ratio/mean": 0.7714826762676239, |
| "sampling/importance_sampling_ratio/min": 0.02449956108466722, |
| "sampling/sampling_logp_difference/max": 0.5438909530639648, |
| "sampling/sampling_logp_difference/mean": 0.009290697518736124, |
| "step": 108, |
| "step_time": 169.33149299799697 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0002111037993017817, |
| "clip_ratio/low_min": 0.0002111037993017817, |
| "clip_ratio/region_mean": 0.0002111037993017817, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 4061.5, |
| "completions/max_terminated_length": 2890.5, |
| "completions/mean_length": 1793.4583740234375, |
| "completions/mean_terminated_length": 1470.6000366210938, |
| "completions/min_length": 679.0, |
| "completions/min_terminated_length": 679.0, |
| "entropy": 0.15238594356924295, |
| "epoch": 0.7440273037542662, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.004517465829849243, |
| "learning_rate": 2.25e-07, |
| "loss": 0.024714305996894836, |
| "num_tokens": 9052189.0, |
| "reward": -0.9166666865348816, |
| "reward_std": 0.2886751592159271, |
| "rewards/correctness_reward/mean": -0.9166666567325592, |
| "rewards/correctness_reward/std": 0.28867512941360474, |
| "sampling/importance_sampling_ratio/max": 1.6460945010185242, |
| "sampling/importance_sampling_ratio/mean": 0.6706594079732895, |
| "sampling/importance_sampling_ratio/min": 0.05606750026345253, |
| "sampling/sampling_logp_difference/max": 0.6999292373657227, |
| "sampling/sampling_logp_difference/mean": 0.008071616757661104, |
| "step": 109, |
| "step_time": 130.72493847899023 |
| }, |
| { |
| "clip_ratio/high_max": 6.980315811233595e-05, |
| "clip_ratio/high_mean": 6.980315811233595e-05, |
| "clip_ratio/low_mean": 0.0002183696378779132, |
| "clip_ratio/low_min": 0.0002183696378779132, |
| "clip_ratio/region_mean": 0.00028817279235227033, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3514.5, |
| "completions/max_terminated_length": 3514.5, |
| "completions/mean_length": 1805.0000610351562, |
| "completions/mean_terminated_length": 1805.0000610351562, |
| "completions/min_length": 947.0, |
| "completions/min_terminated_length": 947.0, |
| "entropy": 0.16883638687431812, |
| "epoch": 0.7508532423208191, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.01729542203247547, |
| "learning_rate": 2.2353253484351e-07, |
| "loss": -0.04131258651614189, |
| "num_tokens": 9124381.0, |
| "reward": -0.8333333432674408, |
| "reward_std": 0.3892494738101959, |
| "rewards/correctness_reward/mean": -0.8333333432674408, |
| "rewards/correctness_reward/std": 0.3892495036125183, |
| "sampling/importance_sampling_ratio/max": 2.4297763109207153, |
| "sampling/importance_sampling_ratio/mean": 0.6028274744749069, |
| "sampling/importance_sampling_ratio/min": 0.022295957431197166, |
| "sampling/sampling_logp_difference/max": 0.6852223873138428, |
| "sampling/sampling_logp_difference/mean": 0.008053431054577231, |
| "step": 110, |
| "step_time": 112.09678150700347 |
| }, |
| { |
| "clip_ratio/high_max": 2.134927490260452e-05, |
| "clip_ratio/high_mean": 2.134927490260452e-05, |
| "clip_ratio/low_mean": 0.000559045140107628, |
| "clip_ratio/low_min": 0.000559045140107628, |
| "clip_ratio/region_mean": 0.0005803944150102325, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 4755.0, |
| "completions/max_terminated_length": 3513.0, |
| "completions/mean_length": 1957.75, |
| "completions/mean_terminated_length": 1804.0, |
| "completions/min_length": 946.5, |
| "completions/min_terminated_length": 946.5, |
| "entropy": 0.16786883305758238, |
| "epoch": 0.757679180887372, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.021987807005643845, |
| "learning_rate": 2.220557464593968e-07, |
| "loss": 0.15698328614234924, |
| "num_tokens": 9208027.0, |
| "reward": -0.8333333730697632, |
| "reward_std": 0.5773502588272095, |
| "rewards/correctness_reward/mean": -0.8333333134651184, |
| "rewards/correctness_reward/std": 0.5773502588272095, |
| "sampling/importance_sampling_ratio/max": 2.5532615184783936, |
| "sampling/importance_sampling_ratio/mean": 0.5662012547254562, |
| "sampling/importance_sampling_ratio/min": 0.01069966983050108, |
| "sampling/sampling_logp_difference/max": 0.802309513092041, |
| "sampling/sampling_logp_difference/mean": 0.008713751565665007, |
| "step": 111, |
| "step_time": 149.50963626700832 |
| }, |
| { |
| "clip_ratio/high_max": 9.413296356797218e-05, |
| "clip_ratio/high_mean": 9.413296356797218e-05, |
| "clip_ratio/low_mean": 0.00085310242138803, |
| "clip_ratio/low_min": 0.00085310242138803, |
| "clip_ratio/region_mean": 0.0009472353849560022, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3690.0, |
| "completions/max_terminated_length": 3690.0, |
| "completions/mean_length": 1725.5000610351562, |
| "completions/mean_terminated_length": 1725.5000610351562, |
| "completions/min_length": 748.0, |
| "completions/min_terminated_length": 748.0, |
| "entropy": 0.19234910048544407, |
| "epoch": 0.764505119453925, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.08581659942865372, |
| "learning_rate": 2.2056982209041334e-07, |
| "loss": -0.2922076880931854, |
| "num_tokens": 9286429.0, |
| "reward": -0.5000000149011612, |
| "reward_std": 0.8816154599189758, |
| "rewards/correctness_reward/mean": -0.5000000149011612, |
| "rewards/correctness_reward/std": 0.8816154599189758, |
| "sampling/importance_sampling_ratio/max": 2.0683520436286926, |
| "sampling/importance_sampling_ratio/mean": 0.5630929172039032, |
| "sampling/importance_sampling_ratio/min": 0.025849847123026848, |
| "sampling/sampling_logp_difference/max": 0.6896476745605469, |
| "sampling/sampling_logp_difference/mean": 0.009252662304788828, |
| "step": 112, |
| "step_time": 122.5797243149791 |
| }, |
| { |
| "clip_ratio/high_max": 5.361355215427466e-05, |
| "clip_ratio/high_mean": 5.361355215427466e-05, |
| "clip_ratio/low_mean": 0.0003404093768040184, |
| "clip_ratio/low_min": 0.0003404093768040184, |
| "clip_ratio/region_mean": 0.00039402292895829305, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 4042.0, |
| "completions/max_terminated_length": 3283.5, |
| "completions/mean_length": 1776.7083740234375, |
| "completions/mean_terminated_length": 1627.439453125, |
| "completions/min_length": 730.5, |
| "completions/min_terminated_length": 730.5, |
| "entropy": 0.1821483038365841, |
| "epoch": 0.7713310580204779, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.023123860359191895, |
| "learning_rate": 2.1907495013766874e-07, |
| "loss": -0.03904452547430992, |
| "num_tokens": 9365520.0, |
| "reward": -0.9166666865348816, |
| "reward_std": 0.28867512941360474, |
| "rewards/correctness_reward/mean": -0.9166666567325592, |
| "rewards/correctness_reward/std": 0.28867512941360474, |
| "sampling/importance_sampling_ratio/max": 0.9768067002296448, |
| "sampling/importance_sampling_ratio/mean": 0.32466551661491394, |
| "sampling/importance_sampling_ratio/min": 0.0039460924454033375, |
| "sampling/sampling_logp_difference/max": 0.5433354377746582, |
| "sampling/sampling_logp_difference/mean": 0.009037391748279333, |
| "step": 113, |
| "step_time": 131.63761309701658 |
| }, |
| { |
| "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": 0.0833333358168602, |
| "completions/max_length": 5248.0, |
| "completions/max_terminated_length": 2787.5, |
| "completions/mean_length": 2246.4583740234375, |
| "completions/mean_terminated_length": 1973.5909423828125, |
| "completions/min_length": 1119.0, |
| "completions/min_terminated_length": 1119.0, |
| "entropy": 0.19245805777609348, |
| "epoch": 0.7781569965870307, |
| "frac_reward_zero_std": 1.0, |
| "grad_norm": 0.0, |
| "learning_rate": 2.1757132013674093e-07, |
| "loss": 0.0, |
| "num_tokens": 9457013.0, |
| "reward": -0.5, |
| "reward_std": 0.5222329497337341, |
| "rewards/correctness_reward/mean": -0.5, |
| "rewards/correctness_reward/std": 0.5222329497337341, |
| "sampling/importance_sampling_ratio/max": 2.2650750875473022, |
| "sampling/importance_sampling_ratio/mean": 0.6504296064376831, |
| "sampling/importance_sampling_ratio/min": 0.011082793585956097, |
| "sampling/sampling_logp_difference/max": 0.6627502739429474, |
| "sampling/sampling_logp_difference/mean": 0.009212277363985777, |
| "step": 114, |
| "step_time": 164.27334200701443 |
| }, |
| { |
| "clip_ratio/high_max": 0.00012562813935801387, |
| "clip_ratio/high_mean": 0.00012562813935801387, |
| "clip_ratio/low_mean": 0.0004480607240111567, |
| "clip_ratio/low_min": 0.0004480607240111567, |
| "clip_ratio/region_mean": 0.0005736888633691706, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 2268.0, |
| "completions/max_terminated_length": 2268.0, |
| "completions/mean_length": 1129.8333740234375, |
| "completions/mean_terminated_length": 1129.8333740234375, |
| "completions/min_length": 767.5, |
| "completions/min_terminated_length": 767.5, |
| "entropy": 0.1530211679637432, |
| "epoch": 0.7849829351535836, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.17116226255893707, |
| "learning_rate": 2.1605912273364515e-07, |
| "loss": -0.6148641705513, |
| "num_tokens": 9514183.0, |
| "reward": -0.9166666865348816, |
| "reward_std": 0.2886751592159271, |
| "rewards/correctness_reward/mean": -0.9166666567325592, |
| "rewards/correctness_reward/std": 0.28867512941360474, |
| "sampling/importance_sampling_ratio/max": 2.204539656639099, |
| "sampling/importance_sampling_ratio/mean": 0.6949987411499023, |
| "sampling/importance_sampling_ratio/min": 0.050576210021972656, |
| "sampling/sampling_logp_difference/max": 0.9088044762611389, |
| "sampling/sampling_logp_difference/mean": 0.008726058527827263, |
| "step": 115, |
| "step_time": 75.91544955001154 |
| }, |
| { |
| "clip_ratio/high_max": 0.0002909997583628865, |
| "clip_ratio/high_mean": 0.0002909997583628865, |
| "clip_ratio/low_mean": 0.0008542009891243652, |
| "clip_ratio/low_min": 0.0008542009891243652, |
| "clip_ratio/region_mean": 0.0011452007420302834, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 2597.5, |
| "completions/max_terminated_length": 2597.5, |
| "completions/mean_length": 1472.3333740234375, |
| "completions/mean_terminated_length": 1472.3333740234375, |
| "completions/min_length": 778.5, |
| "completions/min_terminated_length": 778.5, |
| "entropy": 0.15164860244840384, |
| "epoch": 0.7918088737201365, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.052857693284749985, |
| "learning_rate": 2.145385496606619e-07, |
| "loss": 0.2539215087890625, |
| "num_tokens": 9578811.0, |
| "reward": -0.5833333432674408, |
| "reward_std": 0.8415164947509766, |
| "rewards/correctness_reward/mean": -0.5833333432674408, |
| "rewards/correctness_reward/std": 0.8415164947509766, |
| "sampling/importance_sampling_ratio/max": 2.0988221168518066, |
| "sampling/importance_sampling_ratio/mean": 0.6562928557395935, |
| "sampling/importance_sampling_ratio/min": 0.020620742812752724, |
| "sampling/sampling_logp_difference/max": 0.5115723609924316, |
| "sampling/sampling_logp_difference/mean": 0.008505056146532297, |
| "step": 116, |
| "step_time": 86.72707642504247 |
| }, |
| { |
| "clip_ratio/high_max": 0.0003095987376582343, |
| "clip_ratio/high_mean": 0.0003095987376582343, |
| "clip_ratio/low_mean": 0.0005687712473445572, |
| "clip_ratio/low_min": 0.0005687712473445572, |
| "clip_ratio/region_mean": 0.0008783699813648127, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 4099.0, |
| "completions/max_terminated_length": 4099.0, |
| "completions/mean_length": 2552.791748046875, |
| "completions/mean_terminated_length": 2552.791748046875, |
| "completions/min_length": 804.5, |
| "completions/min_terminated_length": 804.5, |
| "entropy": 0.23589052446186543, |
| "epoch": 0.7986348122866894, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.03518350422382355, |
| "learning_rate": 2.1300979371202724e-07, |
| "loss": 0.08621601015329361, |
| "num_tokens": 9672556.0, |
| "reward": -0.4166666716337204, |
| "reward_std": 0.9446329772472382, |
| "rewards/correctness_reward/mean": -0.4166666716337204, |
| "rewards/correctness_reward/std": 0.9446330070495605, |
| "sampling/importance_sampling_ratio/max": 1.4925788640975952, |
| "sampling/importance_sampling_ratio/mean": 0.4850645363330841, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.696826696395874, |
| "sampling/sampling_logp_difference/mean": 0.009785029105842113, |
| "step": 117, |
| "step_time": 134.44173820799915 |
| }, |
| { |
| "clip_ratio/high_max": 0.00024063924229267286, |
| "clip_ratio/high_mean": 0.00024063924229267286, |
| "clip_ratio/low_mean": 0.0004523810584942112, |
| "clip_ratio/low_min": 0.0004523810584942112, |
| "clip_ratio/region_mean": 0.0006930202853254741, |
| "completions/clipped_ratio": 0.0833333358168602, |
| "completions/max_length": 5248.0, |
| "completions/max_terminated_length": 4326.0, |
| "completions/mean_length": 3003.916748046875, |
| "completions/mean_terminated_length": 2799.9091796875, |
| "completions/min_length": 1060.0, |
| "completions/min_terminated_length": 1060.0, |
| "entropy": 0.22496706247329712, |
| "epoch": 0.8054607508532423, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.04817913472652435, |
| "learning_rate": 2.1147304871948812e-07, |
| "loss": -0.20130401849746704, |
| "num_tokens": 9779246.0, |
| "reward": -0.3333333358168602, |
| "reward_std": 0.9671956598758698, |
| "rewards/correctness_reward/mean": -0.3333333358168602, |
| "rewards/correctness_reward/std": 0.9671956598758698, |
| "sampling/importance_sampling_ratio/max": 2.08801132440567, |
| "sampling/importance_sampling_ratio/mean": 0.5505248308181763, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 0.7242177128791809, |
| "sampling/sampling_logp_difference/mean": 0.009443401359021664, |
| "step": 118, |
| "step_time": 168.65139187700697 |
| }, |
| { |
| "clip_ratio/high_max": 5.621346281259321e-05, |
| "clip_ratio/high_mean": 5.621346281259321e-05, |
| "clip_ratio/low_mean": 0.0005924530651100213, |
| "clip_ratio/low_min": 0.0005924530651100213, |
| "clip_ratio/region_mean": 0.0006486665315605933, |
| "completions/clipped_ratio": 0.0416666679084301, |
| "completions/max_length": 3854.5, |
| "completions/max_terminated_length": 3206.5, |
| "completions/mean_length": 1865.7500610351562, |
| "completions/mean_terminated_length": 1732.3485107421875, |
| "completions/min_length": 898.0, |
| "completions/min_terminated_length": 898.0, |
| "entropy": 0.18109257612377405, |
| "epoch": 0.8122866894197952, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.02632739767432213, |
| "learning_rate": 2.0992850952772656e-07, |
| "loss": -0.015607806853950024, |
| "num_tokens": 9859526.0, |
| "reward": -0.8333333730697632, |
| "reward_std": 0.5773503184318542, |
| "rewards/correctness_reward/mean": -0.8333333134651184, |
| "rewards/correctness_reward/std": 0.5773502588272095, |
| "sampling/importance_sampling_ratio/max": 1.622683048248291, |
| "sampling/importance_sampling_ratio/mean": 0.5376181304454803, |
| "sampling/importance_sampling_ratio/min": 0.06714884378015995, |
| "sampling/sampling_logp_difference/max": 0.8372238874435425, |
| "sampling/sampling_logp_difference/mean": 0.008855800610035658, |
| "step": 119, |
| "step_time": 125.9446228260058 |
| }, |
| { |
| "clip_ratio/high_max": 0.00019927212451875675, |
| "clip_ratio/high_mean": 0.00019927212451875675, |
| "clip_ratio/low_mean": 0.0003339088070788421, |
| "clip_ratio/low_min": 0.0003339088070788421, |
| "clip_ratio/region_mean": 0.0005331809225026518, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 3373.0, |
| "completions/max_terminated_length": 3373.0, |
| "completions/mean_length": 1848.75, |
| "completions/mean_terminated_length": 1848.75, |
| "completions/min_length": 1006.0, |
| "completions/min_terminated_length": 1006.0, |
| "entropy": 0.185127517208457, |
| "epoch": 0.8191126279863481, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.03174300491809845, |
| "learning_rate": 2.083763719696549e-07, |
| "loss": 0.12570519745349884, |
| "num_tokens": 9934550.0, |
| "reward": -0.5, |
| "reward_std": 0.5222329497337341, |
| "rewards/correctness_reward/mean": -0.5, |
| "rewards/correctness_reward/std": 0.5222329497337341, |
| "sampling/importance_sampling_ratio/max": 1.1787127256393433, |
| "sampling/importance_sampling_ratio/mean": 0.4706980586051941, |
| "sampling/importance_sampling_ratio/min": 0.03853647783398628, |
| "sampling/sampling_logp_difference/max": 0.7632927894592285, |
| "sampling/sampling_logp_difference/mean": 0.00860992493107915, |
| "step": 120, |
| "step_time": 109.8505892399844 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 294, |
| "num_input_tokens_seen": 9934550, |
| "num_train_epochs": 2, |
| "save_steps": 60, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 0.0, |
| "train_batch_size": 3, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
