| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 1.3888888888888888, |
| "eval_steps": 500, |
| "global_step": 500, |
| "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": 0.3453125, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2711.7, |
| "completions/mean_length": 3878.6890625, |
| "completions/mean_terminated_length": 1619.041455078125, |
| "completions/min_length": 609.9, |
| "completions/min_terminated_length": 609.9, |
| "entropy": 0.5342199512757361, |
| "epoch": 0.027777777777777776, |
| "frac_reward_zero_std": 0.275, |
| "grad_norm": 0.2434006631374359, |
| "learning_rate": 3e-07, |
| "loss": -0.0252, |
| "num_tokens": 2530705.0, |
| "reward": 0.3828125, |
| "reward_std": 0.4778271645307541, |
| "rewards/reward_correct/mean": 0.3828125, |
| "rewards/reward_correct/std": 0.47782717645168304, |
| "sampling/importance_sampling_ratio/max": 2.229939651489258, |
| "sampling/importance_sampling_ratio/mean": 0.28947882503271105, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 20.731588554382324, |
| "sampling/sampling_logp_difference/mean": 0.042482714727520944, |
| "step": 10, |
| "step_time": 97.8341444697231 |
| }, |
| { |
| "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": 0.296875, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2827.0, |
| "completions/mean_length": 3604.1640625, |
| "completions/mean_terminated_length": 1672.880029296875, |
| "completions/min_length": 578.1, |
| "completions/min_terminated_length": 578.1, |
| "entropy": 0.5231569674331695, |
| "epoch": 0.05555555555555555, |
| "frac_reward_zero_std": 0.175, |
| "grad_norm": 0.1266385316848755, |
| "learning_rate": 6.333333333333332e-07, |
| "loss": -0.0326, |
| "num_tokens": 4884674.0, |
| "reward": 0.4109375, |
| "reward_std": 0.47415409684181214, |
| "rewards/reward_correct/mean": 0.4109375, |
| "rewards/reward_correct/std": 0.4741541087627411, |
| "sampling/importance_sampling_ratio/max": 2.4683921217918394, |
| "sampling/importance_sampling_ratio/mean": 0.3206924065947533, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 19.872397041320802, |
| "sampling/sampling_logp_difference/mean": 0.04122307561337948, |
| "step": 20, |
| "step_time": 91.14647748004646 |
| }, |
| { |
| "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": 0.284375, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2853.0, |
| "completions/mean_length": 3462.8375, |
| "completions/mean_terminated_length": 1606.2551391601562, |
| "completions/min_length": 553.3, |
| "completions/min_terminated_length": 553.3, |
| "entropy": 0.489674887387082, |
| "epoch": 0.08333333333333333, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.10141309350728989, |
| "learning_rate": 9.666666666666666e-07, |
| "loss": -0.0178, |
| "num_tokens": 7152210.0, |
| "reward": 0.45, |
| "reward_std": 0.4725195735692978, |
| "rewards/reward_correct/mean": 0.45, |
| "rewards/reward_correct/std": 0.47251958549022677, |
| "sampling/importance_sampling_ratio/max": 2.1169467091560366, |
| "sampling/importance_sampling_ratio/mean": 0.3329242721199989, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 22.969456291198732, |
| "sampling/sampling_logp_difference/mean": 0.03882257957011461, |
| "step": 30, |
| "step_time": 89.79702748470008 |
| }, |
| { |
| "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": 0.4125, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2707.2, |
| "completions/mean_length": 4290.6703125, |
| "completions/mean_terminated_length": 1551.9008666992188, |
| "completions/min_length": 537.6, |
| "completions/min_terminated_length": 537.6, |
| "entropy": 0.5838452530559153, |
| "epoch": 0.1111111111111111, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.11795759201049805, |
| "learning_rate": 9.997876019358083e-07, |
| "loss": -0.0185, |
| "num_tokens": 9945431.0, |
| "reward": 0.3640625, |
| "reward_std": 0.4712675392627716, |
| "rewards/reward_correct/mean": 0.3640625, |
| "rewards/reward_correct/std": 0.471267557144165, |
| "sampling/importance_sampling_ratio/max": 2.536797857284546, |
| "sampling/importance_sampling_ratio/mean": 0.29656108319759367, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 23.390139389038087, |
| "sampling/sampling_logp_difference/mean": 0.04604039713740349, |
| "step": 40, |
| "step_time": 104.82241893839091 |
| }, |
| { |
| "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": 0.3453125, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2767.5, |
| "completions/mean_length": 3832.1796875, |
| "completions/mean_terminated_length": 1559.965576171875, |
| "completions/min_length": 624.6, |
| "completions/min_terminated_length": 624.6, |
| "entropy": 0.5342930512968451, |
| "epoch": 0.1388888888888889, |
| "frac_reward_zero_std": 0.3125, |
| "grad_norm": 1.948204517364502, |
| "learning_rate": 9.990536180750723e-07, |
| "loss": -0.0307, |
| "num_tokens": 12443042.0, |
| "reward": 0.403125, |
| "reward_std": 0.4718716889619827, |
| "rewards/reward_correct/mean": 0.403125, |
| "rewards/reward_correct/std": 0.4718716889619827, |
| "sampling/importance_sampling_ratio/max": 2.631464684009552, |
| "sampling/importance_sampling_ratio/mean": 0.3282940939068794, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 23.956416893005372, |
| "sampling/sampling_logp_difference/mean": 0.045470394939184186, |
| "step": 50, |
| "step_time": 96.40531954467296 |
| }, |
| { |
| "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": 0.353125, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2789.8, |
| "completions/mean_length": 3885.1921875, |
| "completions/mean_terminated_length": 1538.73271484375, |
| "completions/min_length": 554.3, |
| "completions/min_terminated_length": 554.3, |
| "entropy": 0.5527781342156232, |
| "epoch": 0.16666666666666666, |
| "frac_reward_zero_std": 0.2125, |
| "grad_norm": 0.2698710858821869, |
| "learning_rate": 9.97796195883804e-07, |
| "loss": -0.0329, |
| "num_tokens": 14976877.0, |
| "reward": 0.4359375, |
| "reward_std": 0.4925714761018753, |
| "rewards/reward_correct/mean": 0.4359375, |
| "rewards/reward_correct/std": 0.49257148802280426, |
| "sampling/importance_sampling_ratio/max": 2.5233256936073305, |
| "sampling/importance_sampling_ratio/mean": 0.32925176024436953, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 21.101856422424316, |
| "sampling/sampling_logp_difference/mean": 0.04616186544299126, |
| "step": 60, |
| "step_time": 97.68703336380422 |
| }, |
| { |
| "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": 0.346875, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2801.9, |
| "completions/mean_length": 3842.059375, |
| "completions/mean_terminated_length": 1533.8720336914062, |
| "completions/min_length": 606.3, |
| "completions/min_terminated_length": 606.3, |
| "entropy": 0.5389381644781679, |
| "epoch": 0.19444444444444445, |
| "frac_reward_zero_std": 0.275, |
| "grad_norm": 0.1625521332025528, |
| "learning_rate": 9.960166542242428e-07, |
| "loss": -0.0281, |
| "num_tokens": 17480451.0, |
| "reward": 0.4265625, |
| "reward_std": 0.4911981523036957, |
| "rewards/reward_correct/mean": 0.4265625, |
| "rewards/reward_correct/std": 0.49119815826416013, |
| "sampling/importance_sampling_ratio/max": 2.4373825907707216, |
| "sampling/importance_sampling_ratio/mean": 0.32550778836011884, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 24.449700927734376, |
| "sampling/sampling_logp_difference/mean": 0.04671826101839542, |
| "step": 70, |
| "step_time": 96.85919274576008 |
| }, |
| { |
| "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": 0.3375, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2702.0, |
| "completions/mean_length": 3792.390625, |
| "completions/mean_terminated_length": 1554.6982788085938, |
| "completions/min_length": 523.0, |
| "completions/min_terminated_length": 523.0, |
| "entropy": 0.5755266681313514, |
| "epoch": 0.2222222222222222, |
| "frac_reward_zero_std": 0.275, |
| "grad_norm": 0.2741508483886719, |
| "learning_rate": 9.937168595898508e-07, |
| "loss": -0.0216, |
| "num_tokens": 19956309.0, |
| "reward": 0.396875, |
| "reward_std": 0.48693000376224516, |
| "rewards/reward_correct/mean": 0.396875, |
| "rewards/reward_correct/std": 0.48693000674247744, |
| "sampling/importance_sampling_ratio/max": 2.2106072545051574, |
| "sampling/importance_sampling_ratio/mean": 0.30533475875854493, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 23.59271354675293, |
| "sampling/sampling_logp_difference/mean": 0.04881819523870945, |
| "step": 80, |
| "step_time": 95.62034683078528 |
| }, |
| { |
| "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": 0.3796875, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2839.8, |
| "completions/mean_length": 4084.5625, |
| "completions/mean_terminated_length": 1582.0239868164062, |
| "completions/min_length": 598.2, |
| "completions/min_terminated_length": 598.2, |
| "entropy": 0.5755758906248957, |
| "epoch": 0.25, |
| "frac_reward_zero_std": 0.2375, |
| "grad_norm": 0.11705684661865234, |
| "learning_rate": 9.908992241476186e-07, |
| "loss": -0.0328, |
| "num_tokens": 22618117.0, |
| "reward": 0.3671875, |
| "reward_std": 0.46752808690071107, |
| "rewards/reward_correct/mean": 0.3671875, |
| "rewards/reward_correct/std": 0.4675280928611755, |
| "sampling/importance_sampling_ratio/max": 2.1488356232643127, |
| "sampling/importance_sampling_ratio/mean": 0.282461379468441, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 21.50088768005371, |
| "sampling/sampling_logp_difference/mean": 0.04865880124270916, |
| "step": 90, |
| "step_time": 101.01973776686937 |
| }, |
| { |
| "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": 0.4203125, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2645.3, |
| "completions/mean_length": 4314.771875, |
| "completions/mean_terminated_length": 1509.4629516601562, |
| "completions/min_length": 569.4, |
| "completions/min_terminated_length": 569.4, |
| "entropy": 0.5954360837582499, |
| "epoch": 0.2777777777777778, |
| "frac_reward_zero_std": 0.275, |
| "grad_norm": 0.09120581299066544, |
| "learning_rate": 9.875667032080352e-07, |
| "loss": -0.0301, |
| "num_tokens": 25430539.0, |
| "reward": 0.353125, |
| "reward_std": 0.4764838844537735, |
| "rewards/reward_correct/mean": 0.353125, |
| "rewards/reward_correct/std": 0.4764839023351669, |
| "sampling/importance_sampling_ratio/max": 2.3067823886871337, |
| "sampling/importance_sampling_ratio/mean": 0.28874450623989106, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 23.806869316101075, |
| "sampling/sampling_logp_difference/mean": 0.050743330270051956, |
| "step": 100, |
| "step_time": 106.27121401969343 |
| }, |
| { |
| "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": 0.2765625, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2885.7, |
| "completions/mean_length": 3313.5390625, |
| "completions/mean_terminated_length": 1451.3173950195312, |
| "completions/min_length": 524.1, |
| "completions/min_terminated_length": 524.1, |
| "entropy": 0.5162243159022182, |
| "epoch": 0.3055555555555556, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.20105834305286407, |
| "learning_rate": 9.837227921253745e-07, |
| "loss": -0.0222, |
| "num_tokens": 27594124.0, |
| "reward": 0.4359375, |
| "reward_std": 0.48942558765411376, |
| "rewards/reward_correct/mean": 0.4359375, |
| "rewards/reward_correct/std": 0.4894255995750427, |
| "sampling/importance_sampling_ratio/max": 2.4590456604957582, |
| "sampling/importance_sampling_ratio/mean": 0.36281110197305677, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 21.5187047958374, |
| "sampling/sampling_logp_difference/mean": 0.04329596869647503, |
| "step": 110, |
| "step_time": 88.54633442051708 |
| }, |
| { |
| "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": 0.3703125, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2659.6, |
| "completions/mean_length": 3929.0671875, |
| "completions/mean_terminated_length": 1439.959716796875, |
| "completions/min_length": 555.3, |
| "completions/min_terminated_length": 555.3, |
| "entropy": 0.5905495826154947, |
| "epoch": 0.3333333333333333, |
| "frac_reward_zero_std": 0.2375, |
| "grad_norm": 0.08803282678127289, |
| "learning_rate": 9.793715226315528e-07, |
| "loss": -0.0261, |
| "num_tokens": 30159399.0, |
| "reward": 0.421875, |
| "reward_std": 0.47909272611141207, |
| "rewards/reward_correct/mean": 0.421875, |
| "rewards/reward_correct/std": 0.47909273505210875, |
| "sampling/importance_sampling_ratio/max": 2.126016139984131, |
| "sampling/importance_sampling_ratio/mean": 0.27617028504610064, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 23.618224716186525, |
| "sampling/sampling_logp_difference/mean": 0.05163493566215038, |
| "step": 120, |
| "step_time": 99.54315063767135 |
| }, |
| { |
| "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": 0.378125, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2735.7, |
| "completions/mean_length": 4040.6890625, |
| "completions/mean_terminated_length": 1536.42490234375, |
| "completions/min_length": 645.0, |
| "completions/min_terminated_length": 645.0, |
| "entropy": 0.5655143281910568, |
| "epoch": 0.3611111111111111, |
| "frac_reward_zero_std": 0.2125, |
| "grad_norm": 0.1966542899608612, |
| "learning_rate": 9.745174586073982e-07, |
| "loss": -0.0311, |
| "num_tokens": 32795344.0, |
| "reward": 0.384375, |
| "reward_std": 0.47214375138282777, |
| "rewards/reward_correct/mean": 0.384375, |
| "rewards/reward_correct/std": 0.4721437692642212, |
| "sampling/importance_sampling_ratio/max": 2.5283960938453673, |
| "sampling/importance_sampling_ratio/mean": 0.3151987254619598, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 22.23869686126709, |
| "sampling/sampling_logp_difference/mean": 0.049242925643920896, |
| "step": 130, |
| "step_time": 100.30694740768521 |
| }, |
| { |
| "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": 0.2765625, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2714.3, |
| "completions/mean_length": 3301.6859375, |
| "completions/mean_terminated_length": 1435.6876953125, |
| "completions/min_length": 568.2, |
| "completions/min_terminated_length": 568.2, |
| "entropy": 0.5224350173491985, |
| "epoch": 0.3888888888888889, |
| "frac_reward_zero_std": 0.225, |
| "grad_norm": 0.266464501619339, |
| "learning_rate": 9.691656912957684e-07, |
| "loss": -0.0196, |
| "num_tokens": 34952999.0, |
| "reward": 0.478125, |
| "reward_std": 0.5006843686103821, |
| "rewards/reward_correct/mean": 0.478125, |
| "rewards/reward_correct/std": 0.500684380531311, |
| "sampling/importance_sampling_ratio/max": 2.5264899134635925, |
| "sampling/importance_sampling_ratio/mean": 0.4062116891145706, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 24.331268119812012, |
| "sampling/sampling_logp_difference/mean": 0.0450795441865921, |
| "step": 140, |
| "step_time": 88.51336960084737 |
| }, |
| { |
| "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": 0.321875, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2729.3, |
| "completions/mean_length": 3610.175, |
| "completions/mean_terminated_length": 1446.8697387695313, |
| "completions/min_length": 536.4, |
| "completions/min_terminated_length": 536.4, |
| "entropy": 0.5646064455155283, |
| "epoch": 0.4166666666666667, |
| "frac_reward_zero_std": 0.225, |
| "grad_norm": 0.2582959830760956, |
| "learning_rate": 9.633218339615432e-07, |
| "loss": -0.0341, |
| "num_tokens": 37314103.0, |
| "reward": 0.4046875, |
| "reward_std": 0.47060964107513426, |
| "rewards/reward_correct/mean": 0.4046875, |
| "rewards/reward_correct/std": 0.470609650015831, |
| "sampling/importance_sampling_ratio/max": 2.4045764088630674, |
| "sampling/importance_sampling_ratio/mean": 0.3732295721769333, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 21.39520835876465, |
| "sampling/sampling_logp_difference/mean": 0.04807058311998844, |
| "step": 150, |
| "step_time": 93.85687965862454 |
| }, |
| { |
| "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": 0.3375, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2705.6, |
| "completions/mean_length": 3746.8171875, |
| "completions/mean_terminated_length": 1490.0490234375, |
| "completions/min_length": 519.2, |
| "completions/min_terminated_length": 519.2, |
| "entropy": 0.5504673863761127, |
| "epoch": 0.4444444444444444, |
| "frac_reward_zero_std": 0.1625, |
| "grad_norm": 0.20146092772483826, |
| "learning_rate": 9.569920160040814e-07, |
| "loss": -0.0387, |
| "num_tokens": 39761090.0, |
| "reward": 0.4515625, |
| "reward_std": 0.48955525159835817, |
| "rewards/reward_correct/mean": 0.4515625, |
| "rewards/reward_correct/std": 0.48955525159835817, |
| "sampling/importance_sampling_ratio/max": 2.39810996055603, |
| "sampling/importance_sampling_ratio/mean": 0.35599239766597746, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 23.339636421203615, |
| "sampling/sampling_logp_difference/mean": 0.04856264889240265, |
| "step": 160, |
| "step_time": 95.43752448409796 |
| }, |
| { |
| "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": 0.3578125, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2787.9, |
| "completions/mean_length": 3883.9484375, |
| "completions/mean_terminated_length": 1488.986962890625, |
| "completions/min_length": 600.0, |
| "completions/min_terminated_length": 600.0, |
| "entropy": 0.5517090468667447, |
| "epoch": 0.4722222222222222, |
| "frac_reward_zero_std": 0.3625, |
| "grad_norm": 0.23341695964336395, |
| "learning_rate": 9.501828765283294e-07, |
| "loss": -0.0221, |
| "num_tokens": 42296545.0, |
| "reward": 0.3953125, |
| "reward_std": 0.47710200548172, |
| "rewards/reward_correct/mean": 0.3953125, |
| "rewards/reward_correct/std": 0.4771020084619522, |
| "sampling/importance_sampling_ratio/max": 2.4312403798103333, |
| "sampling/importance_sampling_ratio/mean": 0.3166383683681488, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 23.572778701782227, |
| "sampling/sampling_logp_difference/mean": 0.04768793135881424, |
| "step": 170, |
| "step_time": 98.4165175858885 |
| }, |
| { |
| "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": 0.340625, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2983.3, |
| "completions/mean_length": 3747.296875, |
| "completions/mean_terminated_length": 1446.30537109375, |
| "completions/min_length": 559.2, |
| "completions/min_terminated_length": 559.2, |
| "entropy": 0.5527338188607246, |
| "epoch": 0.5, |
| "frac_reward_zero_std": 0.2375, |
| "grad_norm": 0.13981030881404877, |
| "learning_rate": 9.429015573813162e-07, |
| "loss": -0.0258, |
| "num_tokens": 44746079.0, |
| "reward": 0.390625, |
| "reward_std": 0.48312163949012754, |
| "rewards/reward_correct/mean": 0.390625, |
| "rewards/reward_correct/std": 0.4831216514110565, |
| "sampling/importance_sampling_ratio/max": 2.236375367641449, |
| "sampling/importance_sampling_ratio/mean": 0.35382367074489596, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 22.376666259765624, |
| "sampling/sampling_logp_difference/mean": 0.04669986329972744, |
| "step": 180, |
| "step_time": 95.65556847061961 |
| }, |
| { |
| "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": 0.31875, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2660.9, |
| "completions/mean_length": 3635.1734375, |
| "completions/mean_terminated_length": 1505.3306030273438, |
| "completions/min_length": 580.8, |
| "completions/min_terminated_length": 580.8, |
| "entropy": 0.536462911311537, |
| "epoch": 0.5277777777777778, |
| "frac_reward_zero_std": 0.1875, |
| "grad_norm": 0.16308102011680603, |
| "learning_rate": 9.351556956613422e-07, |
| "loss": -0.018, |
| "num_tokens": 47120230.0, |
| "reward": 0.41875, |
| "reward_std": 0.48690163493156435, |
| "rewards/reward_correct/mean": 0.41875, |
| "rewards/reward_correct/std": 0.48690164685249326, |
| "sampling/importance_sampling_ratio/max": 2.3551482439041136, |
| "sampling/importance_sampling_ratio/mean": 0.3526595965027809, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 22.034694099426268, |
| "sampling/sampling_logp_difference/mean": 0.04380165822803974, |
| "step": 190, |
| "step_time": 92.7884238248691 |
| }, |
| { |
| "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": 0.3421875, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2682.7, |
| "completions/mean_length": 3784.634375, |
| "completions/mean_terminated_length": 1501.1128051757812, |
| "completions/min_length": 506.2, |
| "completions/min_terminated_length": 506.2, |
| "entropy": 0.5288248831871897, |
| "epoch": 0.5555555555555556, |
| "frac_reward_zero_std": 0.225, |
| "grad_norm": 0.3080485165119171, |
| "learning_rate": 9.269534157077176e-07, |
| "loss": -0.0415, |
| "num_tokens": 49589284.0, |
| "reward": 0.4328125, |
| "reward_std": 0.49167430102825166, |
| "rewards/reward_correct/mean": 0.4328125, |
| "rewards/reward_correct/std": 0.4916743189096451, |
| "sampling/importance_sampling_ratio/max": 2.2188847064971924, |
| "sampling/importance_sampling_ratio/mean": 0.3379298195242882, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 22.151332664489747, |
| "sampling/sampling_logp_difference/mean": 0.04422352984547615, |
| "step": 200, |
| "step_time": 96.42145730126649 |
| }, |
| { |
| "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": 0.3015625, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2692.1, |
| "completions/mean_length": 3486.7875, |
| "completions/mean_terminated_length": 1456.8675537109375, |
| "completions/min_length": 424.7, |
| "completions/min_terminated_length": 424.7, |
| "entropy": 0.5138872295152396, |
| "epoch": 0.5833333333333334, |
| "frac_reward_zero_std": 0.2375, |
| "grad_norm": 0.14368104934692383, |
| "learning_rate": 9.183033205794524e-07, |
| "loss": -0.0271, |
| "num_tokens": 51866052.0, |
| "reward": 0.4359375, |
| "reward_std": 0.4865183919668198, |
| "rewards/reward_correct/mean": 0.4359375, |
| "rewards/reward_correct/std": 0.48651840090751647, |
| "sampling/importance_sampling_ratio/max": 2.529734969139099, |
| "sampling/importance_sampling_ratio/mean": 0.3310353755950928, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 21.605073738098145, |
| "sampling/sampling_logp_difference/mean": 0.04353258796036243, |
| "step": 210, |
| "step_time": 91.41761814020575 |
| }, |
| { |
| "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": 0.3109375, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2680.2, |
| "completions/mean_length": 3544.5046875, |
| "completions/mean_terminated_length": 1457.0344970703125, |
| "completions/min_length": 569.1, |
| "completions/min_terminated_length": 569.1, |
| "entropy": 0.5135231144260615, |
| "epoch": 0.6111111111111112, |
| "frac_reward_zero_std": 0.225, |
| "grad_norm": 0.17278236150741577, |
| "learning_rate": 9.092144830318357e-07, |
| "loss": -0.0254, |
| "num_tokens": 54182551.0, |
| "reward": 0.43125, |
| "reward_std": 0.4832146823406219, |
| "rewards/reward_correct/mean": 0.43125, |
| "rewards/reward_correct/std": 0.4832146853208542, |
| "sampling/importance_sampling_ratio/max": 2.305103623867035, |
| "sampling/importance_sampling_ratio/mean": 0.35843545794487, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 21.62167282104492, |
| "sampling/sampling_logp_difference/mean": 0.04431740902364254, |
| "step": 220, |
| "step_time": 92.90139632392675 |
| }, |
| { |
| "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": 0.3234375, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2705.8, |
| "completions/mean_length": 3632.15, |
| "completions/mean_terminated_length": 1470.5746459960938, |
| "completions/min_length": 552.3, |
| "completions/min_terminated_length": 552.3, |
| "entropy": 0.5150754484347999, |
| "epoch": 0.6388888888888888, |
| "frac_reward_zero_std": 0.1875, |
| "grad_norm": 0.4685378968715668, |
| "learning_rate": 8.99696436000368e-07, |
| "loss": -0.0234, |
| "num_tokens": 56552135.0, |
| "reward": 0.4265625, |
| "reward_std": 0.48878895938396455, |
| "rewards/reward_correct/mean": 0.4265625, |
| "rewards/reward_correct/std": 0.48878897428512574, |
| "sampling/importance_sampling_ratio/max": 2.552997076511383, |
| "sampling/importance_sampling_ratio/mean": 0.3578918635845184, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 23.531307792663576, |
| "sampling/sampling_logp_difference/mean": 0.04508708380162716, |
| "step": 230, |
| "step_time": 93.72880325838923 |
| }, |
| { |
| "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": 0.3234375, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 3232.5, |
| "completions/mean_length": 3705.865625, |
| "completions/mean_terminated_length": 1559.2656982421875, |
| "completions/min_length": 649.0, |
| "completions/min_terminated_length": 649.0, |
| "entropy": 0.5137905349489301, |
| "epoch": 0.6666666666666666, |
| "frac_reward_zero_std": 0.2875, |
| "grad_norm": 0.1393118053674698, |
| "learning_rate": 8.897591626020284e-07, |
| "loss": -0.0271, |
| "num_tokens": 58972081.0, |
| "reward": 0.4296875, |
| "reward_std": 0.4902578145265579, |
| "rewards/reward_correct/mean": 0.4296875, |
| "rewards/reward_correct/std": 0.49025782346725466, |
| "sampling/importance_sampling_ratio/max": 2.492326486110687, |
| "sampling/importance_sampling_ratio/mean": 0.33982470631599426, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 19.714303398132323, |
| "sampling/sampling_logp_difference/mean": 0.04368722774088383, |
| "step": 240, |
| "step_time": 94.55036701764911 |
| }, |
| { |
| "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": 0.3140625, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2707.3, |
| "completions/mean_length": 3529.0140625, |
| "completions/mean_terminated_length": 1392.2738159179687, |
| "completions/min_length": 495.1, |
| "completions/min_terminated_length": 495.1, |
| "entropy": 0.5081628792919218, |
| "epoch": 0.6944444444444444, |
| "frac_reward_zero_std": 0.175, |
| "grad_norm": 0.08988256007432938, |
| "learning_rate": 8.794130856643633e-07, |
| "loss": -0.0172, |
| "num_tokens": 61275938.0, |
| "reward": 0.4484375, |
| "reward_std": 0.49697442054748536, |
| "rewards/reward_correct/mean": 0.4484375, |
| "rewards/reward_correct/std": 0.4969744235277176, |
| "sampling/importance_sampling_ratio/max": 2.635455274581909, |
| "sampling/importance_sampling_ratio/mean": 0.35613458752632143, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 22.6952974319458, |
| "sampling/sampling_logp_difference/mean": 0.044385163486003874, |
| "step": 250, |
| "step_time": 92.77678105384112 |
| }, |
| { |
| "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": 0.275, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 3083.6, |
| "completions/mean_length": 3301.2484375, |
| "completions/mean_terminated_length": 1451.84169921875, |
| "completions/min_length": 537.7, |
| "completions/min_terminated_length": 537.7, |
| "entropy": 0.48980009178631007, |
| "epoch": 0.7222222222222222, |
| "frac_reward_zero_std": 0.175, |
| "grad_norm": 0.13209204375743866, |
| "learning_rate": 8.686690567933801e-07, |
| "loss": -0.0239, |
| "num_tokens": 63433217.0, |
| "reward": 0.4625, |
| "reward_std": 0.5000784158706665, |
| "rewards/reward_correct/mean": 0.4625, |
| "rewards/reward_correct/std": 0.5000784218311309, |
| "sampling/importance_sampling_ratio/max": 2.4186819911003115, |
| "sampling/importance_sampling_ratio/mean": 0.3675165086984634, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 22.942532539367676, |
| "sampling/sampling_logp_difference/mean": 0.04302049204707146, |
| "step": 260, |
| "step_time": 88.44859280735254 |
| }, |
| { |
| "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": 0.3078125, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2721.3, |
| "completions/mean_length": 3572.4546875, |
| "completions/mean_terminated_length": 1534.3489501953125, |
| "completions/min_length": 564.8, |
| "completions/min_terminated_length": 564.8, |
| "entropy": 0.5065572742838412, |
| "epoch": 0.75, |
| "frac_reward_zero_std": 0.2625, |
| "grad_norm": 0.3504384756088257, |
| "learning_rate": 8.575383449917102e-07, |
| "loss": -0.0168, |
| "num_tokens": 65767596.0, |
| "reward": 0.453125, |
| "reward_std": 0.4748325616121292, |
| "rewards/reward_correct/mean": 0.453125, |
| "rewards/reward_correct/std": 0.4748325675725937, |
| "sampling/importance_sampling_ratio/max": 2.2907237648963927, |
| "sampling/importance_sampling_ratio/mean": 0.3160614877939224, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 25.577795219421386, |
| "sampling/sampling_logp_difference/mean": 0.043240266665816304, |
| "step": 270, |
| "step_time": 92.66369731239975 |
| }, |
| { |
| "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": 0.2734375, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2857.0, |
| "completions/mean_length": 3278.6984375, |
| "completions/mean_terminated_length": 1444.8076782226562, |
| "completions/min_length": 553.0, |
| "completions/min_terminated_length": 553.0, |
| "entropy": 0.4928085923660547, |
| "epoch": 0.7777777777777778, |
| "frac_reward_zero_std": 0.2, |
| "grad_norm": 0.11743981391191483, |
| "learning_rate": 8.460326248389824e-07, |
| "loss": -0.0328, |
| "num_tokens": 67909899.0, |
| "reward": 0.4796875, |
| "reward_std": 0.4822675257921219, |
| "rewards/reward_correct/mean": 0.4796875, |
| "rewards/reward_correct/std": 0.4822675287723541, |
| "sampling/importance_sampling_ratio/max": 2.5242951631546022, |
| "sampling/importance_sampling_ratio/mean": 0.38842698484659194, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 24.473637580871582, |
| "sampling/sampling_logp_difference/mean": 0.04052827805280686, |
| "step": 280, |
| "step_time": 87.84348612818867 |
| }, |
| { |
| "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": 0.2953125, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2763.8, |
| "completions/mean_length": 3433.0734375, |
| "completions/mean_terminated_length": 1448.4840942382812, |
| "completions/min_length": 560.3, |
| "completions/min_terminated_length": 560.3, |
| "entropy": 0.5100364525802433, |
| "epoch": 0.8055555555555556, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.2548004686832428, |
| "learning_rate": 8.341639642468001e-07, |
| "loss": -0.0231, |
| "num_tokens": 70154122.0, |
| "reward": 0.39375, |
| "reward_std": 0.4714914709329605, |
| "rewards/reward_correct/mean": 0.39375, |
| "rewards/reward_correct/std": 0.4714914858341217, |
| "sampling/importance_sampling_ratio/max": 2.33950617313385, |
| "sampling/importance_sampling_ratio/mean": 0.3633014544844627, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 22.3909854888916, |
| "sampling/sampling_logp_difference/mean": 0.04575198851525784, |
| "step": 290, |
| "step_time": 90.32039845213293 |
| }, |
| { |
| "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": 0.265625, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2642.8, |
| "completions/mean_length": 3212.525, |
| "completions/mean_terminated_length": 1411.774169921875, |
| "completions/min_length": 572.2, |
| "completions/min_terminated_length": 572.2, |
| "entropy": 0.47054290105588736, |
| "epoch": 0.8333333333333334, |
| "frac_reward_zero_std": 0.2125, |
| "grad_norm": 0.27349135279655457, |
| "learning_rate": 8.219448118011687e-07, |
| "loss": -0.0334, |
| "num_tokens": 72256530.0, |
| "reward": 0.421875, |
| "reward_std": 0.4876495271921158, |
| "rewards/reward_correct/mean": 0.421875, |
| "rewards/reward_correct/std": 0.48764953911304476, |
| "sampling/importance_sampling_ratio/max": 2.2759945511817934, |
| "sampling/importance_sampling_ratio/mean": 0.3820392429828644, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 22.031887817382813, |
| "sampling/sampling_logp_difference/mean": 0.041293776035308837, |
| "step": 300, |
| "step_time": 86.86856737490743 |
| }, |
| { |
| "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": 0.2796875, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2730.5, |
| "completions/mean_length": 3313.04375, |
| "completions/mean_terminated_length": 1431.5855224609375, |
| "completions/min_length": 443.0, |
| "completions/min_terminated_length": 443.0, |
| "entropy": 0.500835029873997, |
| "epoch": 0.8611111111111112, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.460864394903183, |
| "learning_rate": 8.093879837056485e-07, |
| "loss": -0.0167, |
| "num_tokens": 74422630.0, |
| "reward": 0.4078125, |
| "reward_std": 0.4829194873571396, |
| "rewards/reward_correct/mean": 0.4078125, |
| "rewards/reward_correct/std": 0.4829194962978363, |
| "sampling/importance_sampling_ratio/max": 2.56204674243927, |
| "sampling/importance_sampling_ratio/mean": 0.3848576098680496, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 22.089742851257324, |
| "sampling/sampling_logp_difference/mean": 0.04200447015464306, |
| "step": 310, |
| "step_time": 88.47893992438912 |
| }, |
| { |
| "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": 0.303125, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2956.5, |
| "completions/mean_length": 3513.2265625, |
| "completions/mean_terminated_length": 1483.2241455078124, |
| "completions/min_length": 610.5, |
| "completions/min_terminated_length": 610.5, |
| "entropy": 0.514991466421634, |
| "epoch": 0.8888888888888888, |
| "frac_reward_zero_std": 0.2625, |
| "grad_norm": 0.179782897233963, |
| "learning_rate": 7.965066503389264e-07, |
| "loss": -0.0309, |
| "num_tokens": 76716127.0, |
| "reward": 0.365625, |
| "reward_std": 0.47816264927387236, |
| "rewards/reward_correct/mean": 0.365625, |
| "rewards/reward_correct/std": 0.4781626582145691, |
| "sampling/importance_sampling_ratio/max": 2.519623303413391, |
| "sampling/importance_sampling_ratio/mean": 0.3532961577177048, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 24.083291053771973, |
| "sampling/sampling_logp_difference/mean": 0.04559468738734722, |
| "step": 320, |
| "step_time": 91.2618560783565 |
| }, |
| { |
| "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": 0.253125, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2747.5, |
| "completions/mean_length": 3135.4203125, |
| "completions/mean_terminated_length": 1434.197314453125, |
| "completions/min_length": 520.0, |
| "completions/min_terminated_length": 520.0, |
| "entropy": 0.4833090817555785, |
| "epoch": 0.9166666666666666, |
| "frac_reward_zero_std": 0.1875, |
| "grad_norm": 0.13914607465267181, |
| "learning_rate": 7.833143224409075e-07, |
| "loss": -0.0312, |
| "num_tokens": 78771436.0, |
| "reward": 0.4390625, |
| "reward_std": 0.4904475510120392, |
| "rewards/reward_correct/mean": 0.4390625, |
| "rewards/reward_correct/std": 0.49044756293296815, |
| "sampling/importance_sampling_ratio/max": 2.455558753013611, |
| "sampling/importance_sampling_ratio/mean": 0.38001332730054854, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 24.64868927001953, |
| "sampling/sampling_logp_difference/mean": 0.04199961461126804, |
| "step": 330, |
| "step_time": 85.40838945042341 |
| }, |
| { |
| "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": 0.290625, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2742.6, |
| "completions/mean_length": 3391.753125, |
| "completions/mean_terminated_length": 1421.10107421875, |
| "completions/min_length": 449.9, |
| "completions/min_terminated_length": 449.9, |
| "entropy": 0.4844976777676493, |
| "epoch": 0.9444444444444444, |
| "frac_reward_zero_std": 0.225, |
| "grad_norm": 0.16815045475959778, |
| "learning_rate": 7.698248369418146e-07, |
| "loss": -0.0224, |
| "num_tokens": 80986918.0, |
| "reward": 0.4578125, |
| "reward_std": 0.49601399302482607, |
| "rewards/reward_correct/mean": 0.4578125, |
| "rewards/reward_correct/std": 0.496014004945755, |
| "sampling/importance_sampling_ratio/max": 2.3630632042884825, |
| "sampling/importance_sampling_ratio/mean": 0.3652906835079193, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 24.45225372314453, |
| "sampling/sampling_logp_difference/mean": 0.0402510404586792, |
| "step": 340, |
| "step_time": 89.63123638182878 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.3, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2764.0, |
| "completions/mean_length": 3510.5140625, |
| "completions/mean_terminated_length": 1515.0855346679687, |
| "completions/min_length": 589.2, |
| "completions/min_terminated_length": 589.2, |
| "entropy": 0.48650123639963566, |
| "epoch": 0.9722222222222222, |
| "frac_reward_zero_std": 0.2125, |
| "grad_norm": 0.34350690245628357, |
| "learning_rate": 7.560523424491594e-07, |
| "loss": -0.0327, |
| "num_tokens": 83279343.0, |
| "reward": 0.40625, |
| "reward_std": 0.48524410724639894, |
| "rewards/reward_correct/mean": 0.40625, |
| "rewards/reward_correct/std": 0.4852441310882568, |
| "sampling/importance_sampling_ratio/max": 2.621637427806854, |
| "sampling/importance_sampling_ratio/mean": 0.3800614088773727, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 24.8350399017334, |
| "sampling/sampling_logp_difference/mean": 0.04253811165690422, |
| "step": 350, |
| "step_time": 90.70822816956789 |
| }, |
| { |
| "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": 0.315625, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2796.9, |
| "completions/mean_length": 3580.165625, |
| "completions/mean_terminated_length": 1469.3232055664062, |
| "completions/min_length": 481.0, |
| "completions/min_terminated_length": 481.0, |
| "entropy": 0.5141544749494642, |
| "epoch": 1.0, |
| "frac_reward_zero_std": 0.2375, |
| "grad_norm": 0.09957528114318848, |
| "learning_rate": 7.420112844078065e-07, |
| "loss": -0.0264, |
| "num_tokens": 85620353.0, |
| "reward": 0.45, |
| "reward_std": 0.48938204646110534, |
| "rewards/reward_correct/mean": 0.45, |
| "rewards/reward_correct/std": 0.48938204944133756, |
| "sampling/importance_sampling_ratio/max": 2.5152801275253296, |
| "sampling/importance_sampling_ratio/mean": 0.3644652560353279, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 23.44343662261963, |
| "sampling/sampling_logp_difference/mean": 0.043626537173986436, |
| "step": 360, |
| "step_time": 92.83569850176573 |
| }, |
| { |
| "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": 0.2875, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2694.5, |
| "completions/mean_length": 3396.1671875, |
| "completions/mean_terminated_length": 1472.170263671875, |
| "completions/min_length": 472.1, |
| "completions/min_terminated_length": 472.1, |
| "entropy": 0.5021241253707558, |
| "epoch": 1.0277777777777777, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.20708823204040527, |
| "learning_rate": 7.277163899486974e-07, |
| "loss": -0.0084, |
| "num_tokens": 87839972.0, |
| "reward": 0.4609375, |
| "reward_std": 0.4824483096599579, |
| "rewards/reward_correct/mean": 0.4609375, |
| "rewards/reward_correct/std": 0.4824483096599579, |
| "sampling/importance_sampling_ratio/max": 2.517948567867279, |
| "sampling/importance_sampling_ratio/mean": 0.4134147673845291, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 22.941598892211914, |
| "sampling/sampling_logp_difference/mean": 0.04388279486447573, |
| "step": 370, |
| "step_time": 89.34413652997463 |
| }, |
| { |
| "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": 0.3171875, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2708.5, |
| "completions/mean_length": 3588.9296875, |
| "completions/mean_terminated_length": 1454.8683471679688, |
| "completions/min_length": 524.1, |
| "completions/min_terminated_length": 524.1, |
| "entropy": 0.502693377295509, |
| "epoch": 1.0555555555555556, |
| "frac_reward_zero_std": 0.275, |
| "grad_norm": 0.1858949214220047, |
| "learning_rate": 7.131826524421229e-07, |
| "loss": -0.0318, |
| "num_tokens": 90182887.0, |
| "reward": 0.440625, |
| "reward_std": 0.4875584036111832, |
| "rewards/reward_correct/mean": 0.440625, |
| "rewards/reward_correct/std": 0.4875584065914154, |
| "sampling/importance_sampling_ratio/max": 2.4215506672859193, |
| "sampling/importance_sampling_ratio/mean": 0.36463034301996233, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 22.958362197875978, |
| "sampling/sampling_logp_difference/mean": 0.04441062249243259, |
| "step": 380, |
| "step_time": 92.67752550859004 |
| }, |
| { |
| "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": 0.3890625, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2740.1, |
| "completions/mean_length": 4121.328125, |
| "completions/mean_terminated_length": 1532.1543334960938, |
| "completions/min_length": 603.0, |
| "completions/min_terminated_length": 603.0, |
| "entropy": 0.5237304116133601, |
| "epoch": 1.0833333333333333, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.09304118901491165, |
| "learning_rate": 6.984253157717485e-07, |
| "loss": -0.018, |
| "num_tokens": 92870865.0, |
| "reward": 0.4078125, |
| "reward_std": 0.49269229769706724, |
| "rewards/reward_correct/mean": 0.4078125, |
| "rewards/reward_correct/std": 0.492692306637764, |
| "sampling/importance_sampling_ratio/max": 2.14275484085083, |
| "sampling/importance_sampling_ratio/mean": 0.29739712178707123, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 23.217792129516603, |
| "sampling/sampling_logp_difference/mean": 0.04599142037332058, |
| "step": 390, |
| "step_time": 102.05327154491097 |
| }, |
| { |
| "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": 0.3640625, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2618.8, |
| "completions/mean_length": 3917.4375, |
| "completions/mean_terminated_length": 1481.8446899414062, |
| "completions/min_length": 622.0, |
| "completions/min_terminated_length": 622.0, |
| "entropy": 0.5371272402349859, |
| "epoch": 1.1111111111111112, |
| "frac_reward_zero_std": 0.2125, |
| "grad_norm": 0.14799313247203827, |
| "learning_rate": 6.834598583458861e-07, |
| "loss": -0.0227, |
| "num_tokens": 95421425.0, |
| "reward": 0.428125, |
| "reward_std": 0.482084384560585, |
| "rewards/reward_correct/mean": 0.428125, |
| "rewards/reward_correct/std": 0.4820843905210495, |
| "sampling/importance_sampling_ratio/max": 2.424322748184204, |
| "sampling/importance_sampling_ratio/mean": 0.32391665428876876, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 22.81523914337158, |
| "sampling/sampling_logp_difference/mean": 0.04645326770842075, |
| "step": 400, |
| "step_time": 99.04672075342387 |
| }, |
| { |
| "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": 0.371875, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2735.0, |
| "completions/mean_length": 3984.6125, |
| "completions/mean_terminated_length": 1514.8505249023438, |
| "completions/min_length": 551.9, |
| "completions/min_terminated_length": 551.9, |
| "entropy": 0.5294327767100185, |
| "epoch": 1.1388888888888888, |
| "frac_reward_zero_std": 0.2, |
| "grad_norm": 0.2487545609474182, |
| "learning_rate": 6.683019768627794e-07, |
| "loss": -0.0202, |
| "num_tokens": 98020353.0, |
| "reward": 0.4203125, |
| "reward_std": 0.4853229820728302, |
| "rewards/reward_correct/mean": 0.4203125, |
| "rewards/reward_correct/std": 0.48532298803329466, |
| "sampling/importance_sampling_ratio/max": 2.425787329673767, |
| "sampling/importance_sampling_ratio/mean": 0.34082319438457487, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 22.107694625854492, |
| "sampling/sampling_logp_difference/mean": 0.04505102261900902, |
| "step": 410, |
| "step_time": 100.24638866372406 |
| }, |
| { |
| "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": 0.4046875, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 3061.0, |
| "completions/mean_length": 4253.1953125, |
| "completions/mean_terminated_length": 1604.5595947265624, |
| "completions/min_length": 520.9, |
| "completions/min_terminated_length": 520.9, |
| "entropy": 0.5594399727415293, |
| "epoch": 1.1666666666666667, |
| "frac_reward_zero_std": 0.3125, |
| "grad_norm": 0.09589551389217377, |
| "learning_rate": 6.529675698469369e-07, |
| "loss": -0.0194, |
| "num_tokens": 100791518.0, |
| "reward": 0.340625, |
| "reward_std": 0.4659861117601395, |
| "rewards/reward_correct/mean": 0.340625, |
| "rewards/reward_correct/std": 0.4659861266613007, |
| "sampling/importance_sampling_ratio/max": 2.2143738508224486, |
| "sampling/importance_sampling_ratio/mean": 0.2883808270096779, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 24.680903816223143, |
| "sampling/sampling_logp_difference/mean": 0.045639732107520106, |
| "step": 420, |
| "step_time": 103.95541290417313 |
| }, |
| { |
| "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": 0.3640625, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2721.9, |
| "completions/mean_length": 3928.7765625, |
| "completions/mean_terminated_length": 1515.8645141601562, |
| "completions/min_length": 658.3, |
| "completions/min_terminated_length": 658.3, |
| "entropy": 0.5309726831037551, |
| "epoch": 1.1944444444444444, |
| "frac_reward_zero_std": 0.2125, |
| "grad_norm": 0.11377198994159698, |
| "learning_rate": 6.374727209737742e-07, |
| "loss": -0.0375, |
| "num_tokens": 103356455.0, |
| "reward": 0.3875, |
| "reward_std": 0.4833220809698105, |
| "rewards/reward_correct/mean": 0.3875, |
| "rewards/reward_correct/std": 0.48332209289073946, |
| "sampling/importance_sampling_ratio/max": 2.697994363307953, |
| "sampling/importance_sampling_ratio/mean": 0.3253849983215332, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 23.634774589538573, |
| "sampling/sampling_logp_difference/mean": 0.04580756314098835, |
| "step": 430, |
| "step_time": 98.64194328989834 |
| }, |
| { |
| "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": 0.3828125, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2776.0, |
| "completions/mean_length": 4030.6875, |
| "completions/mean_terminated_length": 1456.2660522460938, |
| "completions/min_length": 534.9, |
| "completions/min_terminated_length": 534.9, |
| "entropy": 0.5253222627099603, |
| "epoch": 1.2222222222222223, |
| "frac_reward_zero_std": 0.2875, |
| "grad_norm": 0.1834605634212494, |
| "learning_rate": 6.218336822000597e-07, |
| "loss": -0.0311, |
| "num_tokens": 105985199.0, |
| "reward": 0.371875, |
| "reward_std": 0.4680081695318222, |
| "rewards/reward_correct/mean": 0.371875, |
| "rewards/reward_correct/std": 0.46800817251205445, |
| "sampling/importance_sampling_ratio/max": 2.4830429673194887, |
| "sampling/importance_sampling_ratio/mean": 0.3344820261001587, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 25.512220191955567, |
| "sampling/sampling_logp_difference/mean": 0.04608744233846664, |
| "step": 440, |
| "step_time": 101.21402945108711 |
| }, |
| { |
| "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": 0.4046875, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2800.7, |
| "completions/mean_length": 4262.6609375, |
| "completions/mean_terminated_length": 1580.7835083007812, |
| "completions/min_length": 585.1, |
| "completions/min_terminated_length": 585.1, |
| "entropy": 0.5569739128928631, |
| "epoch": 1.25, |
| "frac_reward_zero_std": 0.175, |
| "grad_norm": 0.12380044907331467, |
| "learning_rate": 6.060668567178559e-07, |
| "loss": -0.0332, |
| "num_tokens": 108762526.0, |
| "reward": 0.4375, |
| "reward_std": 0.4926897346973419, |
| "rewards/reward_correct/mean": 0.4375, |
| "rewards/reward_correct/std": 0.4926897376775742, |
| "sampling/importance_sampling_ratio/max": 2.0959500551223753, |
| "sampling/importance_sampling_ratio/mean": 0.28630392253398895, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 23.07707347869873, |
| "sampling/sampling_logp_difference/mean": 0.048703621700406076, |
| "step": 450, |
| "step_time": 104.36413167230785 |
| }, |
| { |
| "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": 0.3203125, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2636.1, |
| "completions/mean_length": 3645.103125, |
| "completions/mean_terminated_length": 1502.7676513671875, |
| "completions/min_length": 578.3, |
| "completions/min_terminated_length": 578.3, |
| "entropy": 0.5052359893452376, |
| "epoch": 1.2777777777777777, |
| "frac_reward_zero_std": 0.1375, |
| "grad_norm": 0.3007439970970154, |
| "learning_rate": 5.901887817498367e-07, |
| "loss": -0.0368, |
| "num_tokens": 111138808.0, |
| "reward": 0.41875, |
| "reward_std": 0.49026038646698, |
| "rewards/reward_correct/mean": 0.41875, |
| "rewards/reward_correct/std": 0.49026039242744446, |
| "sampling/importance_sampling_ratio/max": 2.2202050924301147, |
| "sampling/importance_sampling_ratio/mean": 0.3308444514870644, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 23.17190055847168, |
| "sampling/sampling_logp_difference/mean": 0.04389919862151146, |
| "step": 460, |
| "step_time": 93.38349273055792 |
| }, |
| { |
| "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": 0.3265625, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2857.9, |
| "completions/mean_length": 3689.3828125, |
| "completions/mean_terminated_length": 1499.214501953125, |
| "completions/min_length": 502.0, |
| "completions/min_terminated_length": 502.0, |
| "entropy": 0.5224037424195558, |
| "epoch": 1.3055555555555556, |
| "frac_reward_zero_std": 0.2, |
| "grad_norm": 0.2663036286830902, |
| "learning_rate": 5.742161112040236e-07, |
| "loss": -0.0214, |
| "num_tokens": 113549213.0, |
| "reward": 0.375, |
| "reward_std": 0.4806476831436157, |
| "rewards/reward_correct/mean": 0.375, |
| "rewards/reward_correct/std": 0.48064770102500914, |
| "sampling/importance_sampling_ratio/max": 2.4689416170120237, |
| "sampling/importance_sampling_ratio/mean": 0.3618432849645615, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 20.738582611083984, |
| "sampling/sampling_logp_difference/mean": 0.046205943822860716, |
| "step": 470, |
| "step_time": 94.04668763801456 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.35, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 3171.3, |
| "completions/mean_length": 3906.228125, |
| "completions/mean_terminated_length": 1615.4124145507812, |
| "completions/min_length": 649.1, |
| "completions/min_terminated_length": 649.1, |
| "entropy": 0.5468584059271961, |
| "epoch": 1.3333333333333333, |
| "frac_reward_zero_std": 0.2, |
| "grad_norm": 0.1444406360387802, |
| "learning_rate": 5.581655982061366e-07, |
| "loss": -0.0381, |
| "num_tokens": 116095783.0, |
| "reward": 0.4109375, |
| "reward_std": 0.48344514071941375, |
| "rewards/reward_correct/mean": 0.4109375, |
| "rewards/reward_correct/std": 0.48344514071941375, |
| "sampling/importance_sampling_ratio/max": 2.212104618549347, |
| "sampling/importance_sampling_ratio/mean": 0.3308627545833588, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 22.928424453735353, |
| "sampling/sampling_logp_difference/mean": 0.04624480046331882, |
| "step": 480, |
| "step_time": 97.89495770912617 |
| }, |
| { |
| "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": 0.3640625, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2847.0, |
| "completions/mean_length": 3933.9484375, |
| "completions/mean_terminated_length": 1534.0626586914063, |
| "completions/min_length": 625.0, |
| "completions/min_terminated_length": 625.0, |
| "entropy": 0.5143518089316785, |
| "epoch": 1.3611111111111112, |
| "frac_reward_zero_std": 0.1625, |
| "grad_norm": 0.14443330466747284, |
| "learning_rate": 5.420540775278788e-07, |
| "loss": -0.0338, |
| "num_tokens": 118660662.0, |
| "reward": 0.453125, |
| "reward_std": 0.490170481801033, |
| "rewards/reward_correct/mean": 0.453125, |
| "rewards/reward_correct/std": 0.49017049074172975, |
| "sampling/importance_sampling_ratio/max": 2.4240808367729185, |
| "sampling/importance_sampling_ratio/mean": 0.32956405729055405, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 25.39155864715576, |
| "sampling/sampling_logp_difference/mean": 0.044797389209270476, |
| "step": 490, |
| "step_time": 99.29671751111746 |
| }, |
| { |
| "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": 0.31875, |
| "completions/max_length": 8192.0, |
| "completions/max_terminated_length": 2790.6, |
| "completions/mean_length": 3592.6796875, |
| "completions/mean_terminated_length": 1455.6937377929687, |
| "completions/min_length": 574.1, |
| "completions/min_terminated_length": 574.1, |
| "entropy": 0.5269695220049471, |
| "epoch": 1.3888888888888888, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.15532448887825012, |
| "learning_rate": 5.258984479295852e-07, |
| "loss": -0.028, |
| "num_tokens": 121006129.0, |
| "reward": 0.4578125, |
| "reward_std": 0.4924586594104767, |
| "rewards/reward_correct/mean": 0.4578125, |
| "rewards/reward_correct/std": 0.49245867431163787, |
| "sampling/importance_sampling_ratio/max": 2.5489956021308897, |
| "sampling/importance_sampling_ratio/mean": 0.37822050750255587, |
| "sampling/importance_sampling_ratio/min": 0.0, |
| "sampling/sampling_logp_difference/max": 21.463522529602052, |
| "sampling/sampling_logp_difference/mean": 0.046779343485832216, |
| "step": 500, |
| "step_time": 92.61194766778499 |
| } |
| ], |
| "logging_steps": 10, |
| "max_steps": 1000, |
| "num_input_tokens_seen": 121006129, |
| "num_train_epochs": 3, |
| "save_steps": 100, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 0.0, |
| "train_batch_size": 2, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
