{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.8064516129032258, "eval_steps": 500, "global_step": 150, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "clip_ratio": 1.5976482927726465e-06, "completions/clipped_ratio": 0.10546875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4069.0, "completions/mean_length": 1215.251953125, "completions/mean_terminated_length": 883.8755493164062, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.005376344086021506, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 687505.0, "register/okay": 0.21875, "register/okay_any": 0.40625, "register/premium": 0.6799999999999999, "register/verify_any": 0.375, "reward": 0.396484375, "reward_std": 0.4896455705165863, "rewards/correct/mean": 0.396484375, "rewards/correct/std": 0.4896455705165863, "rollout/chars": 1732.3125, "rollout/commit": 0.71875, "rollout/reward": 0.46875, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.788275718688965, "sampling/importance_sampling_ratio/mean": 0.9801996946334839, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.8374314308166504, "sampling/sampling_logp_difference/mean": 0.00512130418792367, "step": 1, "step_time": 150.84468165802537 }, { "epoch": 0.005376344086021506, "grad_norm": 0.005536369979381561, "learning_rate": 0.0, "loss": -0.10681209713220596, "step": 1 }, { "clip_ratio": 3.647639914561296e-06, "completions/clipped_ratio": 0.1171875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4028.0, "completions/mean_length": 1264.6640625, "completions/mean_terminated_length": 906.8141479492188, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.010752688172043012, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 1395861.0, "register/okay": 0.25, "register/okay_any": 0.3125, "register/premium": 0.625, "register/verify_any": 0.3125, "reward": 0.4140625, "reward_std": 0.49304109811782837, "rewards/correct/mean": 0.4140625, "rewards/correct/std": 0.49304109811782837, "rollout/chars": 1863.84375, "rollout/commit": 0.6875, "rollout/reward": 0.53125, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.8976492881774902, "sampling/importance_sampling_ratio/mean": 0.9634870290756226, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.6949191093444824, "sampling/sampling_logp_difference/mean": 0.00507094943895936, "step": 2, "step_time": 126.88190104090609 }, { "epoch": 0.010752688172043012, "grad_norm": 0.004960422404110432, "learning_rate": 2.0000000000000003e-06, "loss": -0.06525798887014389, "step": 2 }, { "clip_ratio": 0.0, "completions/clipped_ratio": 0.095703125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4039.0, "completions/mean_length": 1090.421875, "completions/mean_terminated_length": 772.3369140625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.016129032258064516, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 2006541.0, "register/okay": 0.15625, "register/okay_any": 0.28125, "register/premium": 0.7407407407407407, "register/verify_any": 0.28125, "reward": 0.3984375, "reward_std": 0.4900552034378052, "rewards/correct/mean": 0.3984375, "rewards/correct/std": 0.4900552034378052, "rollout/chars": 2198.59375, "rollout/commit": 0.53125, "rollout/reward": 0.375, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.531585693359375, "sampling/importance_sampling_ratio/mean": 0.9748052954673767, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.445171594619751, "sampling/sampling_logp_difference/mean": 0.005298345349729061, "step": 3, "step_time": 133.05447170190746 }, { "epoch": 0.016129032258064516, "grad_norm": 0.006240344140678644, "learning_rate": 4.000000000000001e-06, "loss": -0.1190323755145073, "step": 3 }, { "clip_ratio": 2.7389637580199633e-06, "completions/clipped_ratio": 0.1171875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4058.0, "completions/mean_length": 1281.017578125, "completions/mean_terminated_length": 914.2079467773438, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.021505376344086023, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 2743142.0, "register/okay": 0.3125, "register/okay_any": 0.46875, "register/premium": 0.2727272727272727, "register/verify_any": 0.46875, "reward": 0.3828125, "reward_std": 0.486548513174057, "rewards/correct/mean": 0.3828125, "rewards/correct/std": 0.486548513174057, "rollout/chars": 4298.34375, "rollout/commit": 0.71875, "rollout/reward": 0.3125, "rollout/trunc": 0.25, "sampling/importance_sampling_ratio/max": 2.8071212768554688, "sampling/importance_sampling_ratio/mean": 0.9747036099433899, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.5947754383087158, "sampling/sampling_logp_difference/mean": 0.005121023394167423, "step": 4, "step_time": 288.29422205802985 }, { "epoch": 0.021505376344086023, "grad_norm": 0.004777645226567984, "learning_rate": 6e-06, "loss": -0.08420069515705109, "step": 4 }, { "clip_ratio": 4.7834116173817165e-05, "completions/clipped_ratio": 0.08984375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4077.0, "completions/mean_length": 1166.685546875, "completions/mean_terminated_length": 893.030029296875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.026881720430107527, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 3396405.0, "register/okay": 0.40625, "register/okay_any": 0.65625, "register/premium": 0.27530364372469635, "register/verify_any": 0.6875, "reward": 0.47265625, "reward_std": 0.49974003434181213, "rewards/correct/mean": 0.47265625, "rewards/correct/std": 0.49974003434181213, "rollout/chars": 5199.34375, "rollout/commit": 0.6875, "rollout/reward": 0.375, "rollout/trunc": 0.21875, "sampling/importance_sampling_ratio/max": 2.6029703617095947, "sampling/importance_sampling_ratio/mean": 0.9715023040771484, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.4979090690612793, "sampling/sampling_logp_difference/mean": 0.00514589948579669, "step": 5, "step_time": 118.11216273508035 }, { "epoch": 0.026881720430107527, "grad_norm": 0.006235240027308464, "learning_rate": 8.000000000000001e-06, "loss": -0.06412481516599655, "step": 5 }, { "clip_ratio": 0.0013401048065588839, "completions/clipped_ratio": 0.130859375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4087.0, "completions/mean_length": 1430.537109375, "completions/mean_terminated_length": 1046.298828125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.03225806451612903, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 4183160.0, "register/okay": 0.21875, "register/okay_any": 0.5, "register/premium": 0.4971428571428571, "register/verify_any": 0.53125, "reward": 0.41796875, "reward_std": 0.4937073290348053, "rewards/correct/mean": 0.41796875, "rewards/correct/std": 0.4937073290348053, "rollout/chars": 4344.03125, "rollout/commit": 0.8125, "rollout/reward": 0.46875, "rollout/trunc": 0.125, "sampling/importance_sampling_ratio/max": 2.8645777702331543, "sampling/importance_sampling_ratio/mean": 0.9869682788848877, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.3402305841445923, "sampling/sampling_logp_difference/mean": 0.005047679413110018, "step": 6, "step_time": 126.73521159705706 }, { "epoch": 0.03225806451612903, "grad_norm": 0.00521544786170125, "learning_rate": 1e-05, "loss": -0.09711194038391113, "step": 6 }, { "clip_ratio": 0.004071735407706001, "completions/clipped_ratio": 0.119140625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4087.0, "completions/mean_length": 1213.759765625, "completions/mean_terminated_length": 842.0376586914062, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.03763440860215054, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 4863773.0, "register/okay": 0.25, "register/okay_any": 0.3125, "register/premium": 0.2916666666666667, "register/verify_any": 0.34375, "reward": 0.37890625, "reward_std": 0.4855891764163971, "rewards/correct/mean": 0.37890625, "rewards/correct/std": 0.4855891764163971, "rollout/chars": 3508.96875, "rollout/commit": 0.6875, "rollout/reward": 0.15625, "rollout/trunc": 0.125, "sampling/importance_sampling_ratio/max": 2.994431257247925, "sampling/importance_sampling_ratio/mean": 0.9885159134864807, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.8572158813476562, "sampling/sampling_logp_difference/mean": 0.0053655486553907394, "step": 7, "step_time": 147.0131771651213 }, { "epoch": 0.03763440860215054, "grad_norm": 0.004825656767934561, "learning_rate": 1e-05, "loss": -0.07372352480888367, "step": 7 }, { "clip_ratio": 0.0074993749899476825, "completions/clipped_ratio": 0.125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4058.0, "completions/mean_length": 1266.484375, "completions/mean_terminated_length": 880.3058471679688, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.043010752688172046, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 5574261.0, "register/okay": 0.25, "register/okay_any": 0.40625, "register/premium": 0.5, "register/verify_any": 0.40625, "reward": 0.404296875, "reward_std": 0.4912354052066803, "rewards/correct/mean": 0.404296875, "rewards/correct/std": 0.4912354052066803, "rollout/chars": 3730.25, "rollout/commit": 0.6875, "rollout/reward": 0.25, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.7395002841949463, "sampling/importance_sampling_ratio/mean": 0.9624147415161133, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.5250225067138672, "sampling/sampling_logp_difference/mean": 0.005175733007490635, "step": 8, "step_time": 141.7585549650248 }, { "epoch": 0.043010752688172046, "grad_norm": 0.0049206651747226715, "learning_rate": 1e-05, "loss": -0.09447586536407471, "step": 8 }, { "clip_ratio": 0.010272848696331494, "completions/clipped_ratio": 0.109375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4093.0, "completions/mean_length": 1234.017578125, "completions/mean_terminated_length": 900.2850952148438, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.04838709677419355, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 6267934.0, "register/okay": 0.34375, "register/okay_any": 0.53125, "register/premium": 0.393939393939394, "register/verify_any": 0.625, "reward": 0.462890625, "reward_std": 0.4991086423397064, "rewards/correct/mean": 0.462890625, "rewards/correct/std": 0.4991086423397064, "rollout/chars": 5054.625, "rollout/commit": 0.8125, "rollout/reward": 0.46875, "rollout/trunc": 0.125, "sampling/importance_sampling_ratio/max": 2.8355865478515625, "sampling/importance_sampling_ratio/mean": 0.9936923980712891, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 4.098477363586426, "sampling/sampling_logp_difference/mean": 0.005666417069733143, "step": 9, "step_time": 144.23339352413313 }, { "epoch": 0.04838709677419355, "grad_norm": 0.0060818931087851524, "learning_rate": 1e-05, "loss": -0.12299580872058868, "step": 9 }, { "clip_ratio": 0.01380008689375245, "completions/clipped_ratio": 0.087890625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4091.0, "completions/mean_length": 1173.78125, "completions/mean_terminated_length": 892.197021484375, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.053763440860215055, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 6927438.0, "register/okay": 0.5, "register/okay_any": 0.625, "register/premium": 0.5, "register/verify_any": 0.59375, "reward": 0.49609375, "reward_std": 0.5004737377166748, "rewards/correct/mean": 0.49609375, "rewards/correct/std": 0.5004737377166748, "rollout/chars": 2617.8125, "rollout/commit": 0.78125, "rollout/reward": 0.6875, "rollout/trunc": 0.0625, "sampling/importance_sampling_ratio/max": 2.481905221939087, "sampling/importance_sampling_ratio/mean": 0.969363272190094, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.00347900390625, "sampling/sampling_logp_difference/mean": 0.0053543029353022575, "step": 10, "step_time": 212.75371372816153 }, { "epoch": 0.053763440860215055, "grad_norm": 0.005507849156856537, "learning_rate": 1e-05, "loss": -0.09586161375045776, "step": 10 }, { "clip_ratio": 0.014677733306598384, "completions/clipped_ratio": 0.109375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4023.0, "completions/mean_length": 1157.3125, "completions/mean_terminated_length": 805.37939453125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.05913978494623656, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 7582542.0, "register/okay": 0.0625, "register/okay_any": 0.375, "register/premium": 0.4666666666666667, "register/verify_any": 0.40625, "reward": 0.494140625, "reward_std": 0.5004546642303467, "rewards/correct/mean": 0.494140625, "rewards/correct/std": 0.5004546642303467, "rollout/chars": 2919.6875, "rollout/commit": 0.8125, "rollout/reward": 0.5625, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.9180502891540527, "sampling/importance_sampling_ratio/mean": 0.9749065637588501, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.5184186697006226, "sampling/sampling_logp_difference/mean": 0.005147980526089668, "step": 11, "step_time": 146.4343996279058 }, { "epoch": 0.05913978494623656, "grad_norm": 0.004948989488184452, "learning_rate": 1e-05, "loss": -0.07085411250591278, "step": 11 }, { "clip_ratio": 0.01596076832538529, "completions/clipped_ratio": 0.083984375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4089.0, "completions/mean_length": 1122.380859375, "completions/mean_terminated_length": 866.9168701171875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.06451612903225806, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 8216993.0, "register/okay": 0.28125, "register/okay_any": 0.5, "register/premium": 0.47342995169082125, "register/verify_any": 0.5, "reward": 0.484375, "reward_std": 0.5002445578575134, "rewards/correct/mean": 0.484375, "rewards/correct/std": 0.5002445578575134, "rollout/chars": 5048.125, "rollout/commit": 0.71875, "rollout/reward": 0.4375, "rollout/trunc": 0.15625, "sampling/importance_sampling_ratio/max": 2.8764474391937256, "sampling/importance_sampling_ratio/mean": 0.9590312838554382, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.4287886619567871, "sampling/sampling_logp_difference/mean": 0.004995434079319239, "step": 12, "step_time": 121.18447578209452 }, { "epoch": 0.06451612903225806, "grad_norm": 0.0055179595947265625, "learning_rate": 1e-05, "loss": -0.10851699858903885, "step": 12 }, { "clip_ratio": 0.008574715991926496, "completions/clipped_ratio": 0.162109375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4021.0, "completions/mean_length": 1473.20703125, "completions/mean_terminated_length": 1001.4149169921875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.06989247311827956, "frac_reward_zero_std": 0.125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 9032427.0, "register/okay": 0.34375, "register/okay_any": 0.53125, "register/premium": 0.4805194805194805, "register/verify_any": 0.5625, "reward": 0.4765625, "reward_std": 0.49993884563446045, "rewards/correct/mean": 0.4765625, "rewards/correct/std": 0.49993884563446045, "rollout/chars": 3908.8125, "rollout/commit": 0.75, "rollout/reward": 0.59375, "rollout/trunc": 0.0625, "sampling/importance_sampling_ratio/max": 2.951629161834717, "sampling/importance_sampling_ratio/mean": 0.9317843317985535, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.5227203369140625, "sampling/sampling_logp_difference/mean": 0.005291281268000603, "step": 13, "step_time": 191.56482845492428 }, { "epoch": 0.06989247311827956, "grad_norm": 0.0041601029224693775, "learning_rate": 1e-05, "loss": -0.06557237356901169, "step": 13 }, { "clip_ratio": 0.013084022144539631, "completions/clipped_ratio": 0.109375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4094.0, "completions/mean_length": 1318.9921875, "completions/mean_terminated_length": 994.7105102539062, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.07526881720430108, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 9763207.0, "register/okay": 0.40625, "register/okay_any": 0.75, "register/premium": 0.12145748987854255, "register/verify_any": 0.8125, "reward": 0.513671875, "reward_std": 0.5003018379211426, "rewards/correct/mean": 0.513671875, "rewards/correct/std": 0.5003018379211426, "rollout/chars": 8604.15625, "rollout/commit": 0.5, "rollout/reward": 0.3125, "rollout/trunc": 0.46875, "sampling/importance_sampling_ratio/max": 2.9212090969085693, "sampling/importance_sampling_ratio/mean": 0.9688857197761536, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.9780625104904175, "sampling/sampling_logp_difference/mean": 0.005228726658970118, "step": 14, "step_time": 357.421331763966 }, { "epoch": 0.07526881720430108, "grad_norm": 0.005021767690777779, "learning_rate": 1e-05, "loss": -0.06424752622842789, "step": 14 }, { "clip_ratio": 0.02033586335255677, "completions/clipped_ratio": 0.111328125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4048.0, "completions/mean_length": 1445.634765625, "completions/mean_terminated_length": 1114.4571533203125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.08064516129032258, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 10567756.0, "register/okay": 0.25, "register/okay_any": 0.34375, "register/premium": 0.625, "register/verify_any": 0.4375, "reward": 0.50390625, "reward_std": 0.5004737377166748, "rewards/correct/mean": 0.50390625, "rewards/correct/std": 0.5004737377166748, "rollout/chars": 3988.84375, "rollout/commit": 0.78125, "rollout/reward": 0.40625, "rollout/trunc": 0.09375, "sampling/importance_sampling_ratio/max": 2.932295083999634, "sampling/importance_sampling_ratio/mean": 0.9546321630477905, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.5548479557037354, "sampling/sampling_logp_difference/mean": 0.005285207182168961, "step": 15, "step_time": 166.41254067909904 }, { "epoch": 0.08064516129032258, "grad_norm": 0.004748669918626547, "learning_rate": 1e-05, "loss": -0.07962360978126526, "step": 15 }, { "clip_ratio": 0.021063678299469757, "completions/clipped_ratio": 0.064453125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4073.0, "completions/mean_length": 1114.67578125, "completions/mean_terminated_length": 934.5532836914062, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.08602150537634409, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 11193990.0, "register/okay": 0.375, "register/okay_any": 0.46875, "register/premium": 0.65, "register/verify_any": 0.46875, "reward": 0.537109375, "reward_std": 0.4991086423397064, "rewards/correct/mean": 0.537109375, "rewards/correct/std": 0.4991086423397064, "rollout/chars": 3170.40625, "rollout/commit": 0.8125, "rollout/reward": 0.59375, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.9447994232177734, "sampling/importance_sampling_ratio/mean": 0.9693629741668701, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.8633896112442017, "sampling/sampling_logp_difference/mean": 0.005218314006924629, "step": 16, "step_time": 119.10572393791517 }, { "epoch": 0.08602150537634409, "grad_norm": 0.005955121014267206, "learning_rate": 1e-05, "loss": -0.09967996180057526, "step": 16 }, { "clip_ratio": 0.010124757566700282, "completions/clipped_ratio": 0.166015625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4044.0, "completions/mean_length": 1443.7734375, "completions/mean_terminated_length": 943.2716064453125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.0913978494623656, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 11993010.0, "register/okay": 0.25, "register/okay_any": 0.4375, "register/premium": 0.75, "register/verify_any": 0.5625, "reward": 0.47265625, "reward_std": 0.49974003434181213, "rewards/correct/mean": 0.47265625, "rewards/correct/std": 0.49974003434181213, "rollout/chars": 3285.5, "rollout/commit": 0.8125, "rollout/reward": 0.4375, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.907860517501831, "sampling/importance_sampling_ratio/mean": 0.9712395668029785, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.7675509452819824, "sampling/sampling_logp_difference/mean": 0.005249205976724625, "step": 17, "step_time": 169.88692099996842 }, { "epoch": 0.0913978494623656, "grad_norm": 0.005303401034325361, "learning_rate": 1e-05, "loss": -0.051697950810194016, "step": 17 }, { "clip_ratio": 0.016437006976047996, "completions/clipped_ratio": 0.09375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4040.0, "completions/mean_length": 1151.44140625, "completions/mean_terminated_length": 864.4482421875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.0967741935483871, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 12633684.0, "register/okay": 0.25, "register/okay_any": 0.46875, "register/premium": 0.6666666666666667, "register/verify_any": 0.5, "reward": 0.537109375, "reward_std": 0.4991086423397064, "rewards/correct/mean": 0.537109375, "rewards/correct/std": 0.4991086423397064, "rollout/chars": 1661.1875, "rollout/commit": 0.71875, "rollout/reward": 0.5, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.9537272453308105, "sampling/importance_sampling_ratio/mean": 0.987904965877533, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.7457265853881836, "sampling/sampling_logp_difference/mean": 0.005279700271785259, "step": 18, "step_time": 147.74140067375265 }, { "epoch": 0.0967741935483871, "grad_norm": 0.005066175479441881, "learning_rate": 1e-05, "loss": -0.09199890494346619, "step": 18 }, { "clip_ratio": 0.016431453828772646, "completions/clipped_ratio": 0.111328125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4067.0, "completions/mean_length": 1284.71875, "completions/mean_terminated_length": 955.4945678710938, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.10215053763440861, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 13357524.0, "register/okay": 0.3125, "register/okay_any": 0.5625, "register/premium": 0.6818181818181819, "register/verify_any": 0.5625, "reward": 0.46484375, "reward_std": 0.49925029277801514, "rewards/correct/mean": 0.46484375, "rewards/correct/std": 0.49925029277801514, "rollout/chars": 3002.90625, "rollout/commit": 0.71875, "rollout/reward": 0.53125, "rollout/trunc": 0.0625, "sampling/importance_sampling_ratio/max": 2.83459210395813, "sampling/importance_sampling_ratio/mean": 0.9570035934448242, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.8500914573669434, "sampling/sampling_logp_difference/mean": 0.005208856426179409, "step": 19, "step_time": 134.55601952417055 }, { "epoch": 0.10215053763440861, "grad_norm": 0.0050915940664708614, "learning_rate": 1e-05, "loss": -0.06887590885162354, "step": 19 }, { "clip_ratio": 0.015742540885185008, "completions/clipped_ratio": 0.068359375, "completions/max_length": 4096.0, "completions/max_terminated_length": 3939.0, "completions/mean_length": 1023.6875, "completions/mean_terminated_length": 823.9559326171875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.10752688172043011, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 13931684.0, "register/okay": 0.375, "register/okay_any": 0.5, "register/premium": 0.4, "register/verify_any": 0.5, "reward": 0.533203125, "reward_std": 0.4993842542171478, "rewards/correct/mean": 0.533203125, "rewards/correct/std": 0.4993842542171478, "rollout/chars": 3815.75, "rollout/commit": 0.8125, "rollout/reward": 0.5, "rollout/trunc": 0.09375, "sampling/importance_sampling_ratio/max": 2.7999613285064697, "sampling/importance_sampling_ratio/mean": 1.0000779628753662, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.5036005973815918, "sampling/sampling_logp_difference/mean": 0.005762252025306225, "step": 20, "step_time": 113.17120536096627 }, { "epoch": 0.10752688172043011, "grad_norm": 0.007009309716522694, "learning_rate": 1e-05, "loss": -0.09315481781959534, "step": 20 }, { "clip_ratio": 0.00888046599811787, "completions/clipped_ratio": 0.087890625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4087.0, "completions/mean_length": 1349.1796875, "completions/mean_terminated_length": 1109.8543701171875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.11290322580645161, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 14678448.0, "register/okay": 0.40625, "register/okay_any": 0.5, "register/premium": 0.631578947368421, "register/verify_any": 0.5625, "reward": 0.548828125, "reward_std": 0.498096764087677, "rewards/correct/mean": 0.548828125, "rewards/correct/std": 0.498096764087677, "rollout/chars": 2420.65625, "rollout/commit": 0.8125, "rollout/reward": 0.625, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.826205253601074, "sampling/importance_sampling_ratio/mean": 0.9306512475013733, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.6579908132553101, "sampling/sampling_logp_difference/mean": 0.005072219762951136, "step": 21, "step_time": 136.9637122741551 }, { "epoch": 0.11290322580645161, "grad_norm": 0.004881712142378092, "learning_rate": 1e-05, "loss": -0.07977926731109619, "step": 21 }, { "clip_ratio": 0.006619919394324825, "completions/clipped_ratio": 0.177734375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4033.0, "completions/mean_length": 1557.130859375, "completions/mean_terminated_length": 1008.3491821289062, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.11827956989247312, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 15541459.0, "register/okay": 0.375, "register/okay_any": 0.65625, "register/premium": 0.6, "register/verify_any": 0.625, "reward": 0.486328125, "reward_std": 0.5003018379211426, "rewards/correct/mean": 0.486328125, "rewards/correct/std": 0.5003018379211426, "rollout/chars": 4022.375, "rollout/commit": 0.90625, "rollout/reward": 0.625, "rollout/trunc": 0.09375, "sampling/importance_sampling_ratio/max": 2.895916223526001, "sampling/importance_sampling_ratio/mean": 0.9804240465164185, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.6260337829589844, "sampling/sampling_logp_difference/mean": 0.005503639578819275, "step": 22, "step_time": 166.04075432108948 }, { "epoch": 0.11827956989247312, "grad_norm": 0.004497839603573084, "learning_rate": 1e-05, "loss": -0.03457781299948692, "step": 22 }, { "clip_ratio": 0.008130440383865789, "completions/clipped_ratio": 0.130859375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4082.0, "completions/mean_length": 1388.0390625, "completions/mean_terminated_length": 980.3236083984375, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.12365591397849462, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.5, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 16304375.0, "register/okay": 0.46875, "register/okay_any": 0.71875, "register/premium": -0.019607843137254943, "register/verify_any": 0.84375, "reward": 0.521484375, "reward_std": 0.5000267624855042, "rewards/correct/mean": 0.521484375, "rewards/correct/std": 0.5000267624855042, "rollout/chars": 7497.34375, "rollout/commit": 0.5625, "rollout/reward": 0.34375, "rollout/trunc": 0.4375, "sampling/importance_sampling_ratio/max": 2.9332101345062256, "sampling/importance_sampling_ratio/mean": 0.9471936821937561, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.8108747005462646, "sampling/sampling_logp_difference/mean": 0.0053689200431108475, "step": 23, "step_time": 153.59572070796276 }, { "epoch": 0.12365591397849462, "grad_norm": 0.005164750851690769, "learning_rate": 1e-05, "loss": -0.08419081568717957, "step": 23 }, { "clip_ratio": 0.010075022632008768, "completions/clipped_ratio": 0.10546875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4033.0, "completions/mean_length": 1299.1953125, "completions/mean_terminated_length": 969.4410400390625, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.12903225806451613, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 17027259.0, "register/okay": 0.3125, "register/okay_any": 0.5, "register/premium": 0.5454545454545454, "register/verify_any": 0.53125, "reward": 0.525390625, "reward_std": 0.4998432695865631, "rewards/correct/mean": 0.525390625, "rewards/correct/std": 0.4998432695865631, "rollout/chars": 2898.625, "rollout/commit": 0.84375, "rollout/reward": 0.625, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.7062180042266846, "sampling/importance_sampling_ratio/mean": 0.9389370083808899, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.5482387542724609, "sampling/sampling_logp_difference/mean": 0.005153043195605278, "step": 24, "step_time": 147.44872645987198 }, { "epoch": 0.12903225806451613, "grad_norm": 0.004218562971800566, "learning_rate": 1e-05, "loss": -0.0943070650100708, "step": 24 }, { "clip_ratio": 0.011164478141836298, "completions/clipped_ratio": 0.14453125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4049.0, "completions/mean_length": 1490.28515625, "completions/mean_terminated_length": 1059.365234375, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.13440860215053763, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 17844301.0, "register/okay": 0.625, "register/okay_any": 0.75, "register/premium": 0.6, "register/verify_any": 0.8125, "reward": 0.509765625, "reward_std": 0.5003935098648071, "rewards/correct/mean": 0.509765625, "rewards/correct/std": 0.5003935098648071, "rollout/chars": 5824.28125, "rollout/commit": 0.78125, "rollout/reward": 0.625, "rollout/trunc": 0.09375, "sampling/importance_sampling_ratio/max": 2.9208223819732666, "sampling/importance_sampling_ratio/mean": 0.9381150007247925, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.46186184883117676, "sampling/sampling_logp_difference/mean": 0.005271736532449722, "step": 25, "step_time": 175.4866898949258 }, { "epoch": 0.13440860215053763, "grad_norm": 0.004921833518892527, "learning_rate": 1e-05, "loss": -0.02945581078529358, "step": 25 }, { "clip_ratio": 0.008224556875575217, "completions/clipped_ratio": 0.111328125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4072.0, "completions/mean_length": 1501.216796875, "completions/mean_terminated_length": 1185.1275634765625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.13978494623655913, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 18674172.0, "register/okay": 0.28125, "register/okay_any": 0.5, "register/premium": 0.6521739130434783, "register/verify_any": 0.625, "reward": 0.53125, "reward_std": 0.4995105266571045, "rewards/correct/mean": 0.53125, "rewards/correct/std": 0.4995105266571045, "rollout/chars": 4819.0625, "rollout/commit": 0.71875, "rollout/reward": 0.53125, "rollout/trunc": 0.125, "sampling/importance_sampling_ratio/max": 2.943610429763794, "sampling/importance_sampling_ratio/mean": 0.9341355562210083, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.6876890659332275, "sampling/sampling_logp_difference/mean": 0.005215954966843128, "step": 26, "step_time": 167.32986848492874 }, { "epoch": 0.13978494623655913, "grad_norm": 0.005027454346418381, "learning_rate": 1e-05, "loss": -0.010844079777598381, "step": 26 }, { "clip_ratio": 0.011227854249227676, "completions/clipped_ratio": 0.11328125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4079.0, "completions/mean_length": 1291.7265625, "completions/mean_terminated_length": 951.308349609375, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.14516129032258066, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 19391184.0, "register/okay": 0.46875, "register/okay_any": 0.65625, "register/premium": 0.5215686274509804, "register/verify_any": 0.65625, "reward": 0.560546875, "reward_std": 0.49680593609809875, "rewards/correct/mean": 0.560546875, "rewards/correct/std": 0.49680593609809875, "rollout/chars": 2457.03125, "rollout/commit": 0.875, "rollout/reward": 0.65625, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.8571557998657227, "sampling/importance_sampling_ratio/mean": 0.9634403586387634, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.5841646194458008, "sampling/sampling_logp_difference/mean": 0.005507075227797031, "step": 27, "step_time": 190.42925496015232 }, { "epoch": 0.14516129032258066, "grad_norm": 0.0052156648598611355, "learning_rate": 1e-05, "loss": -0.08258605003356934, "step": 27 }, { "clip_ratio": 0.007539385935615428, "completions/clipped_ratio": 0.142578125, "completions/max_length": 4096.0, "completions/max_terminated_length": 3857.0, "completions/mean_length": 1458.96484375, "completions/mean_terminated_length": 1029.77001953125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.15053763440860216, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 20195006.0, "register/okay": 0.4375, "register/okay_any": 0.625, "register/premium": 0.5, "register/verify_any": 0.625, "reward": 0.51953125, "reward_std": 0.5001069903373718, "rewards/correct/mean": 0.51953125, "rewards/correct/std": 0.5001069903373718, "rollout/chars": 3357.65625, "rollout/commit": 0.875, "rollout/reward": 0.71875, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.9831702709198, "sampling/importance_sampling_ratio/mean": 0.964399516582489, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.4702794551849365, "sampling/sampling_logp_difference/mean": 0.0050812773406505585, "step": 28, "step_time": 169.85662743984722 }, { "epoch": 0.15053763440860216, "grad_norm": 0.005234717857092619, "learning_rate": 1e-05, "loss": -0.03653641790151596, "step": 28 }, { "clip_ratio": 0.00900801509123994, "completions/clipped_ratio": 0.140625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4049.0, "completions/mean_length": 1361.64453125, "completions/mean_terminated_length": 915.3340454101562, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.15591397849462366, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 20957400.0, "register/okay": 0.5, "register/okay_any": 0.65625, "register/premium": 0.4375, "register/verify_any": 0.65625, "reward": 0.509765625, "reward_std": 0.5003935098648071, "rewards/correct/mean": 0.509765625, "rewards/correct/std": 0.5003935098648071, "rollout/chars": 4667.1875, "rollout/commit": 0.8125, "rollout/reward": 0.46875, "rollout/trunc": 0.15625, "sampling/importance_sampling_ratio/max": 2.649383783340454, "sampling/importance_sampling_ratio/mean": 0.925227165222168, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.4926145076751709, "sampling/sampling_logp_difference/mean": 0.005465826950967312, "step": 29, "step_time": 179.66692242206773 }, { "epoch": 0.15591397849462366, "grad_norm": 0.0046501075848937035, "learning_rate": 1e-05, "loss": -0.01778368279337883, "step": 29 }, { "clip_ratio": 0.008490964444717974, "completions/clipped_ratio": 0.12109375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4025.0, "completions/mean_length": 1251.921875, "completions/mean_terminated_length": 860.0711059570312, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.16129032258064516, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 21658480.0, "register/okay": 0.25, "register/okay_any": 0.59375, "register/premium": 0.4166666666666667, "register/verify_any": 0.65625, "reward": 0.537109375, "reward_std": 0.4991086423397064, "rewards/correct/mean": 0.537109375, "rewards/correct/std": 0.4991086423397064, "rollout/chars": 3536.9375, "rollout/commit": 0.78125, "rollout/reward": 0.5625, "rollout/trunc": 0.09375, "sampling/importance_sampling_ratio/max": 2.795497417449951, "sampling/importance_sampling_ratio/mean": 0.9529191255569458, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.5288147926330566, "sampling/sampling_logp_difference/mean": 0.005513270851224661, "step": 30, "step_time": 155.76831682719057 }, { "epoch": 0.16129032258064516, "grad_norm": 0.005014779977500439, "learning_rate": 1e-05, "loss": -0.05131439492106438, "step": 30 }, { "clip_ratio": 0.006336418812679767, "completions/clipped_ratio": 0.125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4083.0, "completions/mean_length": 1301.068359375, "completions/mean_terminated_length": 901.79248046875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.16666666666666666, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 22375811.0, "register/okay": 0.40625, "register/okay_any": 0.59375, "register/premium": 0.5546558704453441, "register/verify_any": 0.59375, "reward": 0.560546875, "reward_std": 0.49680593609809875, "rewards/correct/mean": 0.560546875, "rewards/correct/std": 0.49680593609809875, "rollout/chars": 3711.875, "rollout/commit": 0.84375, "rollout/reward": 0.59375, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.959876775741577, "sampling/importance_sampling_ratio/mean": 0.9501587748527527, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.3710575103759766, "sampling/sampling_logp_difference/mean": 0.005541889928281307, "step": 31, "step_time": 131.88608792208834 }, { "epoch": 0.16666666666666666, "grad_norm": 0.0051728710532188416, "learning_rate": 1e-05, "loss": -0.024730639532208443, "step": 31 }, { "clip_ratio": 0.005722886576222663, "completions/clipped_ratio": 0.13671875, "completions/max_length": 4096.0, "completions/max_terminated_length": 3969.0, "completions/mean_length": 1450.232421875, "completions/mean_terminated_length": 1031.219482421875, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.17204301075268819, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 23181034.0, "register/okay": 0.3125, "register/okay_any": 0.53125, "register/premium": 0.5, "register/verify_any": 0.53125, "reward": 0.525390625, "reward_std": 0.4998432695865631, "rewards/correct/mean": 0.525390625, "rewards/correct/std": 0.4998432695865631, "rollout/chars": 1586.09375, "rollout/commit": 0.875, "rollout/reward": 0.65625, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.913379192352295, "sampling/importance_sampling_ratio/mean": 0.9456106424331665, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.5503673553466797, "sampling/sampling_logp_difference/mean": 0.0051387008279562, "step": 32, "step_time": 166.45166590507142 }, { "epoch": 0.17204301075268819, "grad_norm": 0.004503391217440367, "learning_rate": 1e-05, "loss": -0.05110776051878929, "step": 32 }, { "clip_ratio": 0.00871692816963332, "completions/clipped_ratio": 0.080078125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4093.0, "completions/mean_length": 1226.68359375, "completions/mean_terminated_length": 976.9130249023438, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.1774193548387097, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 23870856.0, "register/okay": 0.40625, "register/okay_any": 0.59375, "register/premium": 0.631578947368421, "register/verify_any": 0.59375, "reward": 0.595703125, "reward_std": 0.4912354052066803, "rewards/correct/mean": 0.595703125, "rewards/correct/std": 0.4912354052066803, "rollout/chars": 2484.125, "rollout/commit": 0.8125, "rollout/reward": 0.625, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.6280722618103027, "sampling/importance_sampling_ratio/mean": 0.948501467704773, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.574876070022583, "sampling/sampling_logp_difference/mean": 0.005526247434318066, "step": 33, "step_time": 150.0948413569713 }, { "epoch": 0.1774193548387097, "grad_norm": 0.005056716967374086, "learning_rate": 1e-05, "loss": -0.08073143661022186, "step": 33 }, { "clip_ratio": 0.011121355913019215, "completions/clipped_ratio": 0.126953125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4096.0, "completions/mean_length": 1472.6640625, "completions/mean_terminated_length": 1109.4608154296875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.1827956989247312, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 24677244.0, "register/okay": 0.375, "register/okay_any": 0.5625, "register/premium": 0.5, "register/verify_any": 0.59375, "reward": 0.552734375, "reward_std": 0.4976975917816162, "rewards/correct/mean": 0.552734375, "rewards/correct/std": 0.4976975917816162, "rollout/chars": 5478.375, "rollout/commit": 0.71875, "rollout/reward": 0.4375, "rollout/trunc": 0.1875, "sampling/importance_sampling_ratio/max": 2.9688801765441895, "sampling/importance_sampling_ratio/mean": 0.9284334778785706, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.5757560729980469, "sampling/sampling_logp_difference/mean": 0.004951587412506342, "step": 34, "step_time": 136.5056155999191 }, { "epoch": 0.1827956989247312, "grad_norm": 0.0041446140967309475, "learning_rate": 1e-05, "loss": -0.07102659344673157, "step": 34 }, { "clip_ratio": 0.010774545253298129, "completions/clipped_ratio": 0.103515625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4085.0, "completions/mean_length": 1302.275390625, "completions/mean_terminated_length": 986.1568603515625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.1881720430107527, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 25401865.0, "register/okay": 0.28125, "register/okay_any": 0.4375, "register/premium": 0.5217391304347826, "register/verify_any": 0.4375, "reward": 0.55078125, "reward_std": 0.497901052236557, "rewards/correct/mean": 0.55078125, "rewards/correct/std": 0.497901052236557, "rollout/chars": 1811.84375, "rollout/commit": 0.96875, "rollout/reward": 0.625, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.8669655323028564, "sampling/importance_sampling_ratio/mean": 0.9753417372703552, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.639234721660614, "sampling/sampling_logp_difference/mean": 0.005423135124146938, "step": 35, "step_time": 143.07610333099728 }, { "epoch": 0.1881720430107527, "grad_norm": 0.005199028179049492, "learning_rate": 1e-05, "loss": -0.10055884718894958, "step": 35 }, { "clip_ratio": 0.008821962311230891, "completions/clipped_ratio": 0.09765625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4078.0, "completions/mean_length": 1438.88671875, "completions/mean_terminated_length": 1151.3203125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.1935483870967742, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 26195471.0, "register/okay": 0.40625, "register/okay_any": 0.59375, "register/premium": 0.35222672064777333, "register/verify_any": 0.625, "reward": 0.603515625, "reward_std": 0.4896455705165863, "rewards/correct/mean": 0.603515625, "rewards/correct/std": 0.4896455705165863, "rollout/chars": 4416.875, "rollout/commit": 0.625, "rollout/reward": 0.40625, "rollout/trunc": 0.25, "sampling/importance_sampling_ratio/max": 2.8039920330047607, "sampling/importance_sampling_ratio/mean": 0.9573048949241638, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.043118953704834, "sampling/sampling_logp_difference/mean": 0.005074475891888142, "step": 36, "step_time": 133.7379490098101 }, { "epoch": 0.1935483870967742, "grad_norm": 0.004406781867146492, "learning_rate": 1e-05, "loss": -0.1069176122546196, "step": 36 }, { "clip_ratio": 0.006849257575595402, "completions/clipped_ratio": 0.115234375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4060.0, "completions/mean_length": 1347.248046875, "completions/mean_terminated_length": 1007.0419311523438, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.1989247311827957, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 26947182.0, "register/okay": 0.40625, "register/okay_any": 0.59375, "register/premium": 0.6356275303643725, "register/verify_any": 0.5625, "reward": 0.556640625, "reward_std": 0.49726733565330505, "rewards/correct/mean": 0.556640625, "rewards/correct/std": 0.49726733565330505, "rollout/chars": 3451.28125, "rollout/commit": 0.84375, "rollout/reward": 0.46875, "rollout/trunc": 0.0625, "sampling/importance_sampling_ratio/max": 2.5512001514434814, "sampling/importance_sampling_ratio/mean": 0.9806995391845703, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.6822056770324707, "sampling/sampling_logp_difference/mean": 0.005445747636258602, "step": 37, "step_time": 152.8698032599641 }, { "epoch": 0.1989247311827957, "grad_norm": 0.005230598617345095, "learning_rate": 1e-05, "loss": -0.025864053517580032, "step": 37 }, { "clip_ratio": 0.006331337699066353, "completions/clipped_ratio": 0.09375, "completions/max_length": 4096.0, "completions/max_terminated_length": 3999.0, "completions/mean_length": 1394.583984375, "completions/mean_terminated_length": 1123.70263671875, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.20430107526881722, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 27718505.0, "register/okay": 0.59375, "register/okay_any": 0.71875, "register/premium": 0.17004048582995945, "register/verify_any": 0.75, "reward": 0.599609375, "reward_std": 0.4904567301273346, "rewards/correct/mean": 0.599609375, "rewards/correct/std": 0.4904567301273346, "rollout/chars": 4813.65625, "rollout/commit": 0.78125, "rollout/reward": 0.5625, "rollout/trunc": 0.25, "sampling/importance_sampling_ratio/max": 2.8853559494018555, "sampling/importance_sampling_ratio/mean": 0.9194067716598511, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.876793622970581, "sampling/sampling_logp_difference/mean": 0.005094734486192465, "step": 38, "step_time": 154.3992670099833 }, { "epoch": 0.20430107526881722, "grad_norm": 0.004638301208615303, "learning_rate": 1e-05, "loss": -0.08557215332984924, "step": 38 }, { "clip_ratio": 0.0064660851248845574, "completions/clipped_ratio": 0.115234375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4018.0, "completions/mean_length": 1461.224609375, "completions/mean_terminated_length": 1136.0728759765625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.20967741935483872, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 28517884.0, "register/okay": 0.375, "register/okay_any": 0.625, "register/premium": 0.3666666666666666, "register/verify_any": 0.65625, "reward": 0.56640625, "reward_std": 0.4960552453994751, "rewards/correct/mean": 0.56640625, "rewards/correct/std": 0.4960552453994751, "rollout/chars": 3073.21875, "rollout/commit": 0.875, "rollout/reward": 0.6875, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.964137077331543, "sampling/importance_sampling_ratio/mean": 0.9699006080627441, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.7239212989807129, "sampling/sampling_logp_difference/mean": 0.00531615037471056, "step": 39, "step_time": 165.53434173611458 }, { "epoch": 0.20967741935483872, "grad_norm": 0.004682127386331558, "learning_rate": 1e-05, "loss": -0.07406766712665558, "step": 39 }, { "clip_ratio": 0.00743856916869845, "completions/clipped_ratio": 0.1875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4065.0, "completions/mean_length": 1560.82421875, "completions/mean_terminated_length": 988.454345703125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.21505376344086022, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 29374962.0, "register/okay": 0.375, "register/okay_any": 0.65625, "register/premium": 0.44999999999999996, "register/verify_any": 0.65625, "reward": 0.5234375, "reward_std": 0.49993884563446045, "rewards/correct/mean": 0.5234375, "rewards/correct/std": 0.49993884563446045, "rollout/chars": 4500.40625, "rollout/commit": 0.84375, "rollout/reward": 0.71875, "rollout/trunc": 0.09375, "sampling/importance_sampling_ratio/max": 2.669318199157715, "sampling/importance_sampling_ratio/mean": 0.9309188723564148, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.1114287376403809, "sampling/sampling_logp_difference/mean": 0.005226319655776024, "step": 40, "step_time": 165.58794789918466 }, { "epoch": 0.21505376344086022, "grad_norm": 0.004323912318795919, "learning_rate": 1e-05, "loss": -0.0437479205429554, "step": 40 }, { "clip_ratio": 0.008895926593140757, "completions/clipped_ratio": 0.068359375, "completions/max_length": 4096.0, "completions/max_terminated_length": 3994.0, "completions/mean_length": 1094.49609375, "completions/mean_terminated_length": 899.8070678710938, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.22043010752688172, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 29994576.0, "register/okay": 0.46875, "register/okay_any": 0.65625, "register/premium": 0.5294117647058824, "register/verify_any": 0.6875, "reward": 0.60546875, "reward_std": 0.4892277717590332, "rewards/correct/mean": 0.60546875, "rewards/correct/std": 0.4892277717590332, "rollout/chars": 4548.75, "rollout/commit": 0.875, "rollout/reward": 0.71875, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.583728313446045, "sampling/importance_sampling_ratio/mean": 0.9422752857208252, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.7799820899963379, "sampling/sampling_logp_difference/mean": 0.00539364293217659, "step": 41, "step_time": 130.81026078393916 }, { "epoch": 0.22043010752688172, "grad_norm": 0.004999271593987942, "learning_rate": 1e-05, "loss": -0.09776099026203156, "step": 41 }, { "clip_ratio": 0.008774890723543649, "completions/clipped_ratio": 0.119140625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4048.0, "completions/mean_length": 1410.099609375, "completions/mean_terminated_length": 1046.818115234375, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.22580645161290322, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 30777635.0, "register/okay": 0.5, "register/okay_any": 0.75, "register/premium": 0.5, "register/verify_any": 0.78125, "reward": 0.572265625, "reward_std": 0.4952339828014374, "rewards/correct/mean": 0.572265625, "rewards/correct/std": 0.4952339828014374, "rollout/chars": 2519.5, "rollout/commit": 0.90625, "rollout/reward": 0.6875, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.90441632270813, "sampling/importance_sampling_ratio/mean": 0.9259432554244995, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.7504162788391113, "sampling/sampling_logp_difference/mean": 0.005266791209578514, "step": 42, "step_time": 182.9241154739866 }, { "epoch": 0.22580645161290322, "grad_norm": 0.004195198882371187, "learning_rate": 1e-05, "loss": -0.08093886822462082, "step": 42 }, { "clip_ratio": 0.003392930002974026, "completions/clipped_ratio": 0.138671875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4073.0, "completions/mean_length": 1570.396484375, "completions/mean_terminated_length": 1182.2109375, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.23118279569892472, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 31636510.0, "register/okay": 0.53125, "register/okay_any": 0.8125, "register/premium": 0.4313725490196078, "register/verify_any": 0.8125, "reward": 0.552734375, "reward_std": 0.4976975917816162, "rewards/correct/mean": 0.552734375, "rewards/correct/std": 0.4976975917816162, "rollout/chars": 7332.375, "rollout/commit": 0.71875, "rollout/reward": 0.5625, "rollout/trunc": 0.25, "sampling/importance_sampling_ratio/max": 2.9183237552642822, "sampling/importance_sampling_ratio/mean": 0.9194202423095703, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.0795272588729858, "sampling/sampling_logp_difference/mean": 0.005144014023244381, "step": 43, "step_time": 154.99469502893044 }, { "epoch": 0.23118279569892472, "grad_norm": 0.004625420551747084, "learning_rate": 1e-05, "loss": -0.04634333774447441, "step": 43 }, { "clip_ratio": 0.005952214763055963, "completions/clipped_ratio": 0.169921875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4095.0, "completions/mean_length": 1643.234375, "completions/mean_terminated_length": 1150.73876953125, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.23655913978494625, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 32532246.0, "register/okay": 0.5, "register/okay_any": 0.75, "register/premium": 0.25, "register/verify_any": 0.75, "reward": 0.591796875, "reward_std": 0.49198177456855774, "rewards/correct/mean": 0.591796875, "rewards/correct/std": 0.49198177456855774, "rollout/chars": 7713.09375, "rollout/commit": 0.5625, "rollout/reward": 0.125, "rollout/trunc": 0.4375, "sampling/importance_sampling_ratio/max": 2.849212646484375, "sampling/importance_sampling_ratio/mean": 0.9619724750518799, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.8619403839111328, "sampling/sampling_logp_difference/mean": 0.0050093987956643105, "step": 44, "step_time": 210.83732629718725 }, { "epoch": 0.23655913978494625, "grad_norm": 0.004370514303445816, "learning_rate": 1e-05, "loss": -0.024845460429787636, "step": 44 }, { "clip_ratio": 0.005458997462028492, "completions/clipped_ratio": 0.0859375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4088.0, "completions/mean_length": 1299.287109375, "completions/mean_terminated_length": 1036.348388671875, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.24193548387096775, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 33250953.0, "register/okay": 0.4375, "register/okay_any": 0.59375, "register/premium": 0.373015873015873, "register/verify_any": 0.59375, "reward": 0.640625, "reward_std": 0.48028653860092163, "rewards/correct/mean": 0.640625, "rewards/correct/std": 0.48028653860092163, "rollout/chars": 3825.34375, "rollout/commit": 0.84375, "rollout/reward": 0.71875, "rollout/trunc": 0.0625, "sampling/importance_sampling_ratio/max": 2.873892068862915, "sampling/importance_sampling_ratio/mean": 0.9793621301651001, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.9061906337738037, "sampling/sampling_logp_difference/mean": 0.00526129174977541, "step": 45, "step_time": 140.2460780791589 }, { "epoch": 0.24193548387096775, "grad_norm": 0.005457329098135233, "learning_rate": 1e-05, "loss": -0.10607008635997772, "step": 45 }, { "clip_ratio": 0.010226659814179584, "completions/clipped_ratio": 0.09375, "completions/max_length": 4096.0, "completions/max_terminated_length": 3943.0, "completions/mean_length": 1321.83984375, "completions/mean_terminated_length": 1034.8577880859375, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.24731182795698925, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 33977671.0, "register/okay": 0.625, "register/okay_any": 0.75, "register/premium": 0.13333333333333336, "register/verify_any": 0.75, "reward": 0.6484375, "reward_std": 0.4779251217842102, "rewards/correct/mean": 0.6484375, "rewards/correct/std": 0.4779251217842102, "rollout/chars": 6811.09375, "rollout/commit": 0.625, "rollout/reward": 0.5, "rollout/trunc": 0.21875, "sampling/importance_sampling_ratio/max": 2.659039258956909, "sampling/importance_sampling_ratio/mean": 0.9590080976486206, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.9100682735443115, "sampling/sampling_logp_difference/mean": 0.005064851604402065, "step": 46, "step_time": 143.94649956107605 }, { "epoch": 0.24731182795698925, "grad_norm": 0.004310651682317257, "learning_rate": 1e-05, "loss": -0.061595045030117035, "step": 46 }, { "clip_ratio": 0.006405189862107363, "completions/clipped_ratio": 0.123046875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4082.0, "completions/mean_length": 1577.265625, "completions/mean_terminated_length": 1229.140380859375, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.25268817204301075, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 34842799.0, "register/okay": 0.4375, "register/okay_any": 0.65625, "register/premium": 0.6190476190476191, "register/verify_any": 0.71875, "reward": 0.595703125, "reward_std": 0.4912354052066803, "rewards/correct/mean": 0.595703125, "rewards/correct/std": 0.4912354052066803, "rollout/chars": 4768.96875, "rollout/commit": 0.625, "rollout/reward": 0.4375, "rollout/trunc": 0.1875, "sampling/importance_sampling_ratio/max": 2.9756243228912354, "sampling/importance_sampling_ratio/mean": 0.9440420269966125, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 0.9075045585632324, "sampling/sampling_logp_difference/mean": 0.005123917944729328, "step": 47, "step_time": 161.0352724850527 }, { "epoch": 0.25268817204301075, "grad_norm": 0.004760539624840021, "learning_rate": 1e-05, "loss": -0.07464903593063354, "step": 47 }, { "clip_ratio": 0.00796015004743822, "completions/clipped_ratio": 0.13671875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4063.0, "completions/mean_length": 1538.916015625, "completions/mean_terminated_length": 1133.947998046875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.25806451612903225, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 35684628.0, "register/okay": 0.5, "register/okay_any": 0.59375, "register/premium": 0.4375, "register/verify_any": 0.65625, "reward": 0.595703125, "reward_std": 0.4912354052066803, "rewards/correct/mean": 0.595703125, "rewards/correct/std": 0.4912354052066803, "rollout/chars": 3902.28125, "rollout/commit": 0.8125, "rollout/reward": 0.71875, "rollout/trunc": 0.0625, "sampling/importance_sampling_ratio/max": 2.9745614528656006, "sampling/importance_sampling_ratio/mean": 0.923351526260376, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.1928386688232422, "sampling/sampling_logp_difference/mean": 0.004994367249310017, "step": 48, "step_time": 152.8169575711945 }, { "epoch": 0.25806451612903225, "grad_norm": 0.004467393737286329, "learning_rate": 1e-05, "loss": -0.037505295127630234, "step": 48 }, { "clip_ratio": 0.008797254737146432, "completions/clipped_ratio": 0.115234375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4058.0, "completions/mean_length": 1352.208984375, "completions/mean_terminated_length": 1037.169921875, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.26344086021505375, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 36437887.0, "register/okay": 0.625, "register/okay_any": 0.75, "register/premium": 0.4166666666666667, "register/verify_any": 0.71875, "reward": 0.6015625, "reward_std": 0.4900552034378052, "rewards/correct/mean": 0.6015625, "rewards/correct/std": 0.4900552034378052, "rollout/chars": 4644.5625, "rollout/commit": 0.78125, "rollout/reward": 0.59375, "rollout/trunc": 0.125, "sampling/importance_sampling_ratio/max": 2.9898746013641357, "sampling/importance_sampling_ratio/mean": 0.9452433586120605, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.5847194194793701, "sampling/sampling_logp_difference/mean": 0.005605063866823912, "step": 49, "step_time": 146.1176283750101 }, { "epoch": 0.26344086021505375, "grad_norm": 0.005190466996282339, "learning_rate": 1e-05, "loss": -0.052763931453228, "step": 49 }, { "clip_ratio": 0.009056178430000728, "completions/clipped_ratio": 0.1171875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4052.0, "completions/mean_length": 1506.935546875, "completions/mean_terminated_length": 1172.281005859375, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.26881720430107525, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 37264702.0, "register/okay": 0.53125, "register/okay_any": 0.625, "register/premium": 0.40784313725490196, "register/verify_any": 0.65625, "reward": 0.59765625, "reward_std": 0.4908501207828522, "rewards/correct/mean": 0.59765625, "rewards/correct/std": 0.4908501207828522, "rollout/chars": 3848.09375, "rollout/commit": 0.875, "rollout/reward": 0.75, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.8477158546447754, "sampling/importance_sampling_ratio/mean": 0.9113643765449524, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.059030532836914, "sampling/sampling_logp_difference/mean": 0.005309573840349913, "step": 50, "step_time": 146.82213812193368 }, { "epoch": 0.26881720430107525, "grad_norm": 0.004830545280128717, "learning_rate": 1e-05, "loss": -0.0666024386882782, "step": 50 }, { "clip_ratio": 0.008195207654352998, "completions/clipped_ratio": 0.150390625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4062.0, "completions/mean_length": 1586.126953125, "completions/mean_terminated_length": 1151.23681640625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.27419354838709675, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 38132543.0, "register/okay": 0.5625, "register/okay_any": 0.6875, "register/premium": 0.34920634920634924, "register/verify_any": 0.6875, "reward": 0.595703125, "reward_std": 0.4912354052066803, "rewards/correct/mean": 0.595703125, "rewards/correct/std": 0.4912354052066803, "rollout/chars": 4520.6875, "rollout/commit": 0.75, "rollout/reward": 0.625, "rollout/trunc": 0.125, "sampling/importance_sampling_ratio/max": 2.7535383701324463, "sampling/importance_sampling_ratio/mean": 0.9479492902755737, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.0449275970458984, "sampling/sampling_logp_difference/mean": 0.005112256854772568, "step": 51, "step_time": 149.37743002310162 }, { "epoch": 0.27419354838709675, "grad_norm": 0.00440118508413434, "learning_rate": 1e-05, "loss": -0.02765597775578499, "step": 51 }, { "clip_ratio": 0.016025727740270668, "completions/clipped_ratio": 0.109375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4052.0, "completions/mean_length": 1429.7421875, "completions/mean_terminated_length": 1102.3070068359375, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.27956989247311825, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 38922571.0, "register/okay": 0.4375, "register/okay_any": 0.625, "register/premium": 0.4444444444444444, "register/verify_any": 0.625, "reward": 0.595703125, "reward_std": 0.4912354052066803, "rewards/correct/mean": 0.595703125, "rewards/correct/std": 0.4912354052066803, "rollout/chars": 2273.375, "rollout/commit": 0.9375, "rollout/reward": 0.75, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.898261070251465, "sampling/importance_sampling_ratio/mean": 0.9588029384613037, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.0916423797607422, "sampling/sampling_logp_difference/mean": 0.005135898012667894, "step": 52, "step_time": 195.76150241907453 }, { "epoch": 0.27956989247311825, "grad_norm": 0.005357522517442703, "learning_rate": 1e-05, "loss": -0.08132000267505646, "step": 52 }, { "clip_ratio": 0.010641372934514948, "completions/clipped_ratio": 0.091796875, "completions/max_length": 4096.0, "completions/max_terminated_length": 3996.0, "completions/mean_length": 1264.96875, "completions/mean_terminated_length": 987.3290405273438, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.2849462365591398, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 39628539.0, "register/okay": 0.46875, "register/okay_any": 0.71875, "register/premium": 0.047058823529411764, "register/verify_any": 0.75, "reward": 0.5703125, "reward_std": 0.4955156147480011, "rewards/correct/mean": 0.5703125, "rewards/correct/std": 0.4955156147480011, "rollout/chars": 5198.84375, "rollout/commit": 0.65625, "rollout/reward": 0.375, "rollout/trunc": 0.25, "sampling/importance_sampling_ratio/max": 2.841999053955078, "sampling/importance_sampling_ratio/mean": 0.9411661624908447, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.083268642425537, "sampling/sampling_logp_difference/mean": 0.005341468378901482, "step": 53, "step_time": 146.18545961700147 }, { "epoch": 0.2849462365591398, "grad_norm": 0.005267903674393892, "learning_rate": 1e-05, "loss": -0.10027610510587692, "step": 53 }, { "clip_ratio": 0.00735647075362067, "completions/clipped_ratio": 0.0703125, "completions/max_length": 4096.0, "completions/max_terminated_length": 3921.0, "completions/mean_length": 1263.234375, "completions/mean_terminated_length": 1048.99169921875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.2903225806451613, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 40329507.0, "register/okay": 0.5625, "register/okay_any": 0.84375, "register/premium": 0.373015873015873, "register/verify_any": 0.78125, "reward": 0.6875, "reward_std": 0.4639657139778137, "rewards/correct/mean": 0.6875, "rewards/correct/std": 0.4639657139778137, "rollout/chars": 2322.53125, "rollout/commit": 0.90625, "rollout/reward": 0.78125, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.850562572479248, "sampling/importance_sampling_ratio/mean": 0.9741051197052002, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.863353967666626, "sampling/sampling_logp_difference/mean": 0.0051050647161901, "step": 54, "step_time": 139.77709489304107 }, { "epoch": 0.2903225806451613, "grad_norm": 0.00515007833018899, "learning_rate": 1e-05, "loss": -0.04062722623348236, "step": 54 }, { "clip_ratio": 0.007077405376548995, "completions/clipped_ratio": 0.111328125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4007.0, "completions/mean_length": 1374.513671875, "completions/mean_terminated_length": 1050.729736328125, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.2956989247311828, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 41087002.0, "register/okay": 0.71875, "register/okay_any": 0.96875, "register/premium": 0.40096618357487923, "register/verify_any": 0.9375, "reward": 0.640625, "reward_std": 0.48028653860092163, "rewards/correct/mean": 0.640625, "rewards/correct/std": 0.48028653860092163, "rollout/chars": 5640.90625, "rollout/commit": 0.84375, "rollout/reward": 0.84375, "rollout/trunc": 0.15625, "sampling/importance_sampling_ratio/max": 2.9735164642333984, "sampling/importance_sampling_ratio/mean": 0.9822587966918945, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.0670156478881836, "sampling/sampling_logp_difference/mean": 0.005598124116659164, "step": 55, "step_time": 192.065424226108 }, { "epoch": 0.2956989247311828, "grad_norm": 0.004856748506426811, "learning_rate": 1e-05, "loss": -0.0492158904671669, "step": 55 }, { "clip_ratio": 0.006070916996122833, "completions/clipped_ratio": 0.123046875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4067.0, "completions/mean_length": 1510.1484375, "completions/mean_terminated_length": 1147.322998046875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.3010752688172043, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 41916454.0, "register/okay": 0.53125, "register/okay_any": 0.6875, "register/premium": 0.3215686274509804, "register/verify_any": 0.75, "reward": 0.64453125, "reward_std": 0.47912323474884033, "rewards/correct/mean": 0.64453125, "rewards/correct/std": 0.47912323474884033, "rollout/chars": 6616.84375, "rollout/commit": 0.65625, "rollout/reward": 0.4375, "rollout/trunc": 0.34375, "sampling/importance_sampling_ratio/max": 2.86000657081604, "sampling/importance_sampling_ratio/mean": 0.9301953315734863, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.1263437271118164, "sampling/sampling_logp_difference/mean": 0.005294946953654289, "step": 56, "step_time": 152.33958495402476 }, { "epoch": 0.3010752688172043, "grad_norm": 0.004720254335552454, "learning_rate": 1e-05, "loss": -0.04862502962350845, "step": 56 }, { "clip_ratio": 0.011137786609651812, "completions/clipped_ratio": 0.138671875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4039.0, "completions/mean_length": 1544.7421875, "completions/mean_terminated_length": 1142.6734619140625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.3064516129032258, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 42764114.0, "register/okay": 0.59375, "register/okay_any": 0.75, "register/premium": 0.5627530364372468, "register/verify_any": 0.75, "reward": 0.64453125, "reward_std": 0.47912323474884033, "rewards/correct/mean": 0.64453125, "rewards/correct/std": 0.47912323474884033, "rollout/chars": 2879.96875, "rollout/commit": 0.90625, "rollout/reward": 0.71875, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.9222769737243652, "sampling/importance_sampling_ratio/mean": 0.9105566740036011, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.594200611114502, "sampling/sampling_logp_difference/mean": 0.005475620273500681, "step": 57, "step_time": 284.45609347720165 }, { "epoch": 0.3064516129032258, "grad_norm": 0.0041741072200238705, "learning_rate": 1e-05, "loss": -0.052645400166511536, "step": 57 }, { "clip_ratio": 0.013457226558784896, "completions/clipped_ratio": 0.1015625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4063.0, "completions/mean_length": 1524.853515625, "completions/mean_terminated_length": 1234.2021484375, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.3118279569892473, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 43600183.0, "register/okay": 0.59375, "register/okay_any": 0.84375, "register/premium": 0.40080971659919024, "register/verify_any": 0.875, "reward": 0.62890625, "reward_std": 0.4835699498653412, "rewards/correct/mean": 0.62890625, "rewards/correct/std": 0.4835699498653412, "rollout/chars": 7962.625, "rollout/commit": 0.5625, "rollout/reward": 0.46875, "rollout/trunc": 0.40625, "sampling/importance_sampling_ratio/max": 2.8918421268463135, "sampling/importance_sampling_ratio/mean": 0.9498208165168762, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.192861557006836, "sampling/sampling_logp_difference/mean": 0.005280203651636839, "step": 58, "step_time": 150.89825956796994 }, { "epoch": 0.3118279569892473, "grad_norm": 0.004743666388094425, "learning_rate": 1e-05, "loss": -0.08027879893779755, "step": 58 }, { "clip_ratio": 0.005621587801670103, "completions/clipped_ratio": 0.056640625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4046.0, "completions/mean_length": 1223.916015625, "completions/mean_terminated_length": 1051.4720458984375, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.3172043010752688, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 44280364.0, "register/okay": 0.46875, "register/okay_any": 0.84375, "register/premium": 0.5294117647058824, "register/verify_any": 0.8125, "reward": 0.728515625, "reward_std": 0.44516023993492126, "rewards/correct/mean": 0.728515625, "rewards/correct/std": 0.44516023993492126, "rollout/chars": 4256.625, "rollout/commit": 0.875, "rollout/reward": 0.71875, "rollout/trunc": 0.0625, "sampling/importance_sampling_ratio/max": 2.8661625385284424, "sampling/importance_sampling_ratio/mean": 0.978600025177002, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.225687026977539, "sampling/sampling_logp_difference/mean": 0.005175590515136719, "step": 59, "step_time": 143.29344098106958 }, { "epoch": 0.3172043010752688, "grad_norm": 0.00537512730807066, "learning_rate": 1e-05, "loss": -0.027813546359539032, "step": 59 }, { "clip_ratio": 0.007545421271970554, "completions/clipped_ratio": 0.103515625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4079.0, "completions/mean_length": 1501.64453125, "completions/mean_terminated_length": 1202.078369140625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.3225806451612903, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 45104086.0, "register/okay": 0.625, "register/okay_any": 0.875, "register/premium": 0.33333333333333337, "register/verify_any": 0.90625, "reward": 0.6484375, "reward_std": 0.4779251217842102, "rewards/correct/mean": 0.6484375, "rewards/correct/std": 0.4779251217842102, "rollout/chars": 3686.5, "rollout/commit": 0.9375, "rollout/reward": 0.875, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.868577003479004, "sampling/importance_sampling_ratio/mean": 0.9491663575172424, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.2289714813232422, "sampling/sampling_logp_difference/mean": 0.0054664998315274715, "step": 60, "step_time": 177.59955923078815 }, { "epoch": 0.3225806451612903, "grad_norm": 0.004134178627282381, "learning_rate": 1e-05, "loss": -0.06762109696865082, "step": 60 }, { "clip_ratio": 0.006728495395691425, "completions/clipped_ratio": 0.11328125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4032.0, "completions/mean_length": 1600.771484375, "completions/mean_terminated_length": 1290.9493408203125, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.3279569892473118, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 45978993.0, "register/okay": 0.46875, "register/okay_any": 0.75, "register/premium": 0.6313725490196078, "register/verify_any": 0.78125, "reward": 0.625, "reward_std": 0.4845963716506958, "rewards/correct/mean": 0.625, "rewards/correct/std": 0.4845963716506958, "rollout/chars": 6318.375, "rollout/commit": 0.75, "rollout/reward": 0.53125, "rollout/trunc": 0.1875, "sampling/importance_sampling_ratio/max": 2.9828646183013916, "sampling/importance_sampling_ratio/mean": 0.941106915473938, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.3497099876403809, "sampling/sampling_logp_difference/mean": 0.004914080258458853, "step": 61, "step_time": 148.84755163395312 }, { "epoch": 0.3279569892473118, "grad_norm": 0.003976790234446526, "learning_rate": 1e-05, "loss": -0.0447138249874115, "step": 61 }, { "clip_ratio": 0.005132095544468029, "completions/clipped_ratio": 0.087890625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4029.0, "completions/mean_length": 1496.93359375, "completions/mean_terminated_length": 1246.4881591796875, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.3333333333333333, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 46808031.0, "register/okay": 0.6875, "register/okay_any": 0.8125, "register/premium": 0.4636363636363636, "register/verify_any": 0.8125, "reward": 0.662109375, "reward_std": 0.4734536409378052, "rewards/correct/mean": 0.662109375, "rewards/correct/std": 0.4734536409378052, "rollout/chars": 4522.3125, "rollout/commit": 0.875, "rollout/reward": 0.71875, "rollout/trunc": 0.09375, "sampling/importance_sampling_ratio/max": 2.840716600418091, "sampling/importance_sampling_ratio/mean": 0.9038251638412476, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.2466607093811035, "sampling/sampling_logp_difference/mean": 0.005272192880511284, "step": 62, "step_time": 156.2710520021501 }, { "epoch": 0.3333333333333333, "grad_norm": 0.004485093057155609, "learning_rate": 1e-05, "loss": -0.05115474760532379, "step": 62 }, { "clip_ratio": 0.004695364639701438, "completions/clipped_ratio": 0.154296875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4061.0, "completions/mean_length": 1540.57421875, "completions/mean_terminated_length": 1083.78515625, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.3387096774193548, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 47667445.0, "register/okay": 0.53125, "register/okay_any": 0.78125, "register/premium": 0.34901960784313724, "register/verify_any": 0.8125, "reward": 0.611328125, "reward_std": 0.4879252314567566, "rewards/correct/mean": 0.611328125, "rewards/correct/std": 0.4879252314567566, "rollout/chars": 2376.65625, "rollout/commit": 0.8125, "rollout/reward": 0.71875, "rollout/trunc": 0.0625, "sampling/importance_sampling_ratio/max": 2.8307008743286133, "sampling/importance_sampling_ratio/mean": 0.9323355555534363, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.351186752319336, "sampling/sampling_logp_difference/mean": 0.005336794536560774, "step": 63, "step_time": 175.78382128389785 }, { "epoch": 0.3387096774193548, "grad_norm": 0.005287384148687124, "learning_rate": 1e-05, "loss": -0.03088352084159851, "step": 63 }, { "clip_ratio": 0.005930900174462295, "completions/clipped_ratio": 0.134765625, "completions/max_length": 4096.0, "completions/max_terminated_length": 3982.0, "completions/mean_length": 1616.599609375, "completions/mean_terminated_length": 1239.6343994140625, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.34408602150537637, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 48554488.0, "register/okay": 0.625, "register/okay_any": 0.90625, "register/premium": 0.30000000000000004, "register/verify_any": 0.96875, "reward": 0.642578125, "reward_std": 0.4797092080116272, "rewards/correct/mean": 0.642578125, "rewards/correct/std": 0.4797092080116272, "rollout/chars": 8004.9375, "rollout/commit": 0.6875, "rollout/reward": 0.4375, "rollout/trunc": 0.375, "sampling/importance_sampling_ratio/max": 2.9949817657470703, "sampling/importance_sampling_ratio/mean": 0.9279459118843079, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.3719978332519531, "sampling/sampling_logp_difference/mean": 0.005151782184839249, "step": 64, "step_time": 212.30247571098153 }, { "epoch": 0.34408602150537637, "grad_norm": 0.004117595963180065, "learning_rate": 1e-05, "loss": -0.04964715987443924, "step": 64 }, { "clip_ratio": 0.003913742443728552, "completions/clipped_ratio": 0.15625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4046.0, "completions/mean_length": 1651.2578125, "completions/mean_terminated_length": 1198.52783203125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.34946236559139787, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 49466316.0, "register/okay": 0.4375, "register/okay_any": 0.65625, "register/premium": 0.15873015873015872, "register/verify_any": 0.6875, "reward": 0.630859375, "reward_std": 0.4830440282821655, "rewards/correct/mean": 0.630859375, "rewards/correct/std": 0.4830440282821655, "rollout/chars": 8608.09375, "rollout/commit": 0.21875, "rollout/reward": 0.125, "rollout/trunc": 0.5625, "sampling/importance_sampling_ratio/max": 2.9574170112609863, "sampling/importance_sampling_ratio/mean": 0.9506480097770691, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.3145928382873535, "sampling/sampling_logp_difference/mean": 0.005280144978314638, "step": 65, "step_time": 184.43613532104064 }, { "epoch": 0.34946236559139787, "grad_norm": 0.0043518198654055595, "learning_rate": 1e-05, "loss": -0.019979700446128845, "step": 65 }, { "clip_ratio": 0.0029513799468077195, "completions/clipped_ratio": 0.099609375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4092.0, "completions/mean_length": 1411.640625, "completions/mean_terminated_length": 1123.22998046875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.3548387096774194, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 50251620.0, "register/okay": 0.65625, "register/okay_any": 0.75, "register/premium": 0.47619047619047616, "register/verify_any": 0.78125, "reward": 0.646484375, "reward_std": 0.47852855920791626, "rewards/correct/mean": 0.646484375, "rewards/correct/std": 0.47852855920791626, "rollout/chars": 7166.9375, "rollout/commit": 0.71875, "rollout/reward": 0.3125, "rollout/trunc": 0.25, "sampling/importance_sampling_ratio/max": 2.8520288467407227, "sampling/importance_sampling_ratio/mean": 0.9455735087394714, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.321157455444336, "sampling/sampling_logp_difference/mean": 0.005345956888049841, "step": 66, "step_time": 157.48279377399012 }, { "epoch": 0.3548387096774194, "grad_norm": 0.004264642018824816, "learning_rate": 1e-05, "loss": -0.07250965386629105, "step": 66 }, { "clip_ratio": 0.005317563226526545, "completions/clipped_ratio": 0.078125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4090.0, "completions/mean_length": 1319.310546875, "completions/mean_terminated_length": 1083.9979248046875, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.3602150537634409, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 50974499.0, "register/okay": 0.65625, "register/okay_any": 0.78125, "register/premium": 0.36363636363636365, "register/verify_any": 0.8125, "reward": 0.734375, "reward_std": 0.44209739565849304, "rewards/correct/mean": 0.734375, "rewards/correct/std": 0.44209739565849304, "rollout/chars": 3158.5, "rollout/commit": 0.9375, "rollout/reward": 0.875, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.9903197288513184, "sampling/importance_sampling_ratio/mean": 0.9033775329589844, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.5022363662719727, "sampling/sampling_logp_difference/mean": 0.0051087504252791405, "step": 67, "step_time": 158.63203663489548 }, { "epoch": 0.3602150537634409, "grad_norm": 0.003832301590591669, "learning_rate": 1e-05, "loss": -0.05876774713397026, "step": 67 }, { "clip_ratio": 0.013483937193996098, "completions/clipped_ratio": 0.04296875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4000.0, "completions/mean_length": 1192.51953125, "completions/mean_terminated_length": 1070.461181640625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.3655913978494624, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 51647901.0, "register/okay": 0.625, "register/okay_any": 0.8125, "register/premium": 0.41666666666666663, "register/verify_any": 0.84375, "reward": 0.732421875, "reward_std": 0.4431293308734894, "rewards/correct/mean": 0.732421875, "rewards/correct/std": 0.4431293308734894, "rollout/chars": 3496.6875, "rollout/commit": 0.9375, "rollout/reward": 0.84375, "rollout/trunc": 0.0625, "sampling/importance_sampling_ratio/max": 2.949833393096924, "sampling/importance_sampling_ratio/mean": 0.9847578406333923, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.3805537223815918, "sampling/sampling_logp_difference/mean": 0.005482224747538567, "step": 68, "step_time": 148.91818148101447 }, { "epoch": 0.3655913978494624, "grad_norm": 0.004574422724545002, "learning_rate": 1e-05, "loss": -0.1127491295337677, "step": 68 }, { "clip_ratio": 0.006227771713383845, "completions/clipped_ratio": 0.142578125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4086.0, "completions/mean_length": 1532.8828125, "completions/mean_terminated_length": 1115.984130859375, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.3709677419354839, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 52487105.0, "register/okay": 0.5625, "register/okay_any": 0.84375, "register/premium": 0.1428571428571429, "register/verify_any": 0.8125, "reward": 0.666015625, "reward_std": 0.47209542989730835, "rewards/correct/mean": 0.666015625, "rewards/correct/std": 0.47209542989730835, "rollout/chars": 1678.4375, "rollout/commit": 0.96875, "rollout/reward": 0.9375, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.929974317550659, "sampling/importance_sampling_ratio/mean": 0.9206405878067017, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 3.0331263542175293, "sampling/sampling_logp_difference/mean": 0.005199924111366272, "step": 69, "step_time": 210.57059248513542 }, { "epoch": 0.3709677419354839, "grad_norm": 0.004228264559060335, "learning_rate": 1e-05, "loss": -0.06214360147714615, "step": 69 }, { "clip_ratio": 0.011256539903115481, "completions/clipped_ratio": 0.09765625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4085.0, "completions/mean_length": 1517.287109375, "completions/mean_terminated_length": 1238.20556640625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.3763440860215054, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 53318116.0, "register/okay": 0.625, "register/okay_any": 0.84375, "register/premium": 0.2, "register/verify_any": 0.84375, "reward": 0.677734375, "reward_std": 0.46780112385749817, "rewards/correct/mean": 0.677734375, "rewards/correct/std": 0.46780112385749817, "rollout/chars": 6570.28125, "rollout/commit": 0.625, "rollout/reward": 0.375, "rollout/trunc": 0.40625, "sampling/importance_sampling_ratio/max": 2.8080251216888428, "sampling/importance_sampling_ratio/mean": 0.9402785897254944, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.4765124320983887, "sampling/sampling_logp_difference/mean": 0.0052244579419493675, "step": 70, "step_time": 146.23567948798882 }, { "epoch": 0.3763440860215054, "grad_norm": 0.004268636927008629, "learning_rate": 1e-05, "loss": -0.07649495452642441, "step": 70 }, { "clip_ratio": 0.002998780707230253, "completions/clipped_ratio": 0.09765625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4082.0, "completions/mean_length": 1574.05078125, "completions/mean_terminated_length": 1301.112548828125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.3817204301075269, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 54185822.0, "register/okay": 0.71875, "register/okay_any": 0.875, "register/premium": 0.048309178743961345, "register/verify_any": 0.875, "reward": 0.72265625, "reward_std": 0.4481254518032074, "rewards/correct/mean": 0.72265625, "rewards/correct/std": 0.4481254518032074, "rollout/chars": 4861.1875, "rollout/commit": 0.96875, "rollout/reward": 0.8125, "rollout/trunc": 0.09375, "sampling/importance_sampling_ratio/max": 2.947408437728882, "sampling/importance_sampling_ratio/mean": 0.9684253931045532, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.4401793479919434, "sampling/sampling_logp_difference/mean": 0.00485216174274683, "step": 71, "step_time": 154.77642090886366 }, { "epoch": 0.3817204301075269, "grad_norm": 0.004496898036450148, "learning_rate": 1e-05, "loss": -0.035110678523778915, "step": 71 }, { "clip_ratio": 0.002711852112724955, "completions/clipped_ratio": 0.119140625, "completions/max_length": 4096.0, "completions/max_terminated_length": 3998.0, "completions/mean_length": 1509.287109375, "completions/mean_terminated_length": 1168.3436279296875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.3870967741935484, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 55015841.0, "register/okay": 0.59375, "register/okay_any": 0.78125, "register/premium": 0.46153846153846156, "register/verify_any": 0.84375, "reward": 0.703125, "reward_std": 0.45732781291007996, "rewards/correct/mean": 0.703125, "rewards/correct/std": 0.45732781291007996, "rollout/chars": 2444.5625, "rollout/commit": 0.90625, "rollout/reward": 0.8125, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.974992513656616, "sampling/importance_sampling_ratio/mean": 0.9554036855697632, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 3.1328630447387695, "sampling/sampling_logp_difference/mean": 0.005356314592063427, "step": 72, "step_time": 176.32808893895708 }, { "epoch": 0.3870967741935484, "grad_norm": 0.004561584908515215, "learning_rate": 1e-05, "loss": -0.027693230658769608, "step": 72 }, { "clip_ratio": 0.00298470003804141, "completions/clipped_ratio": 0.0703125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4087.0, "completions/mean_length": 1426.7578125, "completions/mean_terminated_length": 1224.8824462890625, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.3924731182795699, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 55797973.0, "register/okay": 0.53125, "register/okay_any": 0.78125, "register/premium": 0.5647058823529412, "register/verify_any": 0.75, "reward": 0.74609375, "reward_std": 0.43567025661468506, "rewards/correct/mean": 0.74609375, "rewards/correct/std": 0.43567025661468506, "rollout/chars": 5966.5, "rollout/commit": 0.71875, "rollout/reward": 0.5, "rollout/trunc": 0.1875, "sampling/importance_sampling_ratio/max": 2.9081454277038574, "sampling/importance_sampling_ratio/mean": 0.9251010417938232, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.3671326637268066, "sampling/sampling_logp_difference/mean": 0.005126387346535921, "step": 73, "step_time": 138.39579898305237 }, { "epoch": 0.3924731182795699, "grad_norm": 0.004180382937192917, "learning_rate": 1e-05, "loss": -0.05125332996249199, "step": 73 }, { "clip_ratio": 0.006533934260232854, "completions/clipped_ratio": 0.083984375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4080.0, "completions/mean_length": 1336.974609375, "completions/mean_terminated_length": 1091.7740478515625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.3978494623655914, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 56545432.0, "register/okay": 0.75, "register/okay_any": 0.8125, "register/premium": 0.41666666666666663, "register/verify_any": 0.8125, "reward": 0.712890625, "reward_std": 0.45285552740097046, "rewards/correct/mean": 0.712890625, "rewards/correct/std": 0.45285552740097046, "rollout/chars": 4382.53125, "rollout/commit": 0.875, "rollout/reward": 0.8125, "rollout/trunc": 0.0625, "sampling/importance_sampling_ratio/max": 2.965951442718506, "sampling/importance_sampling_ratio/mean": 0.9607799053192139, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.454908847808838, "sampling/sampling_logp_difference/mean": 0.00535286869853735, "step": 74, "step_time": 137.45202958310256 }, { "epoch": 0.3978494623655914, "grad_norm": 0.004460392985492945, "learning_rate": 1e-05, "loss": -0.06593317538499832, "step": 74 }, { "clip_ratio": 0.005335613825536711, "completions/clipped_ratio": 0.12109375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4055.0, "completions/mean_length": 1464.697265625, "completions/mean_terminated_length": 1102.1622314453125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.4032258064516129, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 57352061.0, "register/okay": 0.5625, "register/okay_any": 0.875, "register/premium": 0.23015873015873012, "register/verify_any": 0.84375, "reward": 0.69140625, "reward_std": 0.4623647928237915, "rewards/correct/mean": 0.69140625, "rewards/correct/std": 0.4623647928237915, "rollout/chars": 2905.4375, "rollout/commit": 0.9375, "rollout/reward": 0.84375, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.9146065711975098, "sampling/importance_sampling_ratio/mean": 0.9481654167175293, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.4810781478881836, "sampling/sampling_logp_difference/mean": 0.005374530330300331, "step": 75, "step_time": 210.9365189120872 }, { "epoch": 0.4032258064516129, "grad_norm": 0.004631032235920429, "learning_rate": 1e-05, "loss": -0.05312083661556244, "step": 75 }, { "clip_ratio": 0.003812480284977937, "completions/clipped_ratio": 0.09765625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4042.0, "completions/mean_length": 1685.5078125, "completions/mean_terminated_length": 1424.632080078125, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.40860215053763443, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 58271025.0, "register/okay": 0.71875, "register/okay_any": 0.84375, "register/premium": 0.1449275362318841, "register/verify_any": 0.84375, "reward": 0.69921875, "reward_std": 0.45904624462127686, "rewards/correct/mean": 0.69921875, "rewards/correct/std": 0.45904624462127686, "rollout/chars": 7379.46875, "rollout/commit": 0.71875, "rollout/reward": 0.4375, "rollout/trunc": 0.3125, "sampling/importance_sampling_ratio/max": 2.794290781021118, "sampling/importance_sampling_ratio/mean": 0.9341869354248047, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.6867361068725586, "sampling/sampling_logp_difference/mean": 0.004922474268823862, "step": 76, "step_time": 165.76109177281614 }, { "epoch": 0.40860215053763443, "grad_norm": 0.0037442913744598627, "learning_rate": 1e-05, "loss": -0.0564141720533371, "step": 76 }, { "clip_ratio": 0.002230702930546613, "completions/clipped_ratio": 0.091796875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4079.0, "completions/mean_length": 1556.19140625, "completions/mean_terminated_length": 1308.268798828125, "completions/min_length": 5.0, "completions/min_terminated_length": 5.0, "epoch": 0.41397849462365593, "frac_reward_zero_std": 0.09375, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 59137411.0, "register/okay": 0.625, "register/okay_any": 0.71875, "register/premium": 0.7, "register/verify_any": 0.71875, "reward": 0.75390625, "reward_std": 0.4311550557613373, "rewards/correct/mean": 0.75390625, "rewards/correct/std": 0.4311550557613373, "rollout/chars": 4749.09375, "rollout/commit": 0.90625, "rollout/reward": 0.6875, "rollout/trunc": 0.0625, "sampling/importance_sampling_ratio/max": 2.8531060218811035, "sampling/importance_sampling_ratio/mean": 0.888579785823822, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.5669324398040771, "sampling/sampling_logp_difference/mean": 0.005278721451759338, "step": 77, "step_time": 180.0787935939734 }, { "epoch": 0.41397849462365593, "grad_norm": 0.003649241290986538, "learning_rate": 1e-05, "loss": -0.04705454409122467, "step": 77 }, { "clip_ratio": 0.0019156661230681493, "completions/clipped_ratio": 0.083984375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4062.0, "completions/mean_length": 1436.734375, "completions/mean_terminated_length": 1192.921142578125, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.41935483870967744, "frac_reward_zero_std": 0.125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 59937195.0, "register/okay": 0.71875, "register/okay_any": 0.90625, "register/premium": 0.4251207729468599, "register/verify_any": 0.90625, "reward": 0.73828125, "reward_std": 0.44000017642974854, "rewards/correct/mean": 0.73828125, "rewards/correct/std": 0.44000017642974854, "rollout/chars": 2967.625, "rollout/commit": 0.9375, "rollout/reward": 0.75, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.985253095626831, "sampling/importance_sampling_ratio/mean": 0.9347658157348633, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.7055559158325195, "sampling/sampling_logp_difference/mean": 0.005423901602625847, "step": 78, "step_time": 150.5979076999938 }, { "epoch": 0.41935483870967744, "grad_norm": 0.004253201652318239, "learning_rate": 1e-05, "loss": -0.0239992868155241, "step": 78 }, { "clip_ratio": 0.0028138360569300858, "completions/clipped_ratio": 0.0703125, "completions/max_length": 4096.0, "completions/max_terminated_length": 3986.0, "completions/mean_length": 1403.796875, "completions/mean_terminated_length": 1208.724853515625, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.42473118279569894, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 60711219.0, "register/okay": 0.6875, "register/okay_any": 0.84375, "register/premium": 0.30000000000000004, "register/verify_any": 0.8125, "reward": 0.75390625, "reward_std": 0.4311550557613373, "rewards/correct/mean": 0.75390625, "rewards/correct/std": 0.4311550557613373, "rollout/chars": 1788.78125, "rollout/commit": 1.0, "rollout/reward": 0.90625, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.9944705963134766, "sampling/importance_sampling_ratio/mean": 0.9171857237815857, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.7286925315856934, "sampling/sampling_logp_difference/mean": 0.00513107143342495, "step": 79, "step_time": 169.8534768060199 }, { "epoch": 0.42473118279569894, "grad_norm": 0.0040318164974451065, "learning_rate": 1e-05, "loss": -0.09426576644182205, "step": 79 }, { "clip_ratio": 0.00250267129013082, "completions/clipped_ratio": 0.111328125, "completions/max_length": 4096.0, "completions/max_terminated_length": 3964.0, "completions/mean_length": 1536.748046875, "completions/mean_terminated_length": 1216.1385498046875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.43010752688172044, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 61562850.0, "register/okay": 0.8125, "register/okay_any": 0.875, "register/premium": 0.32051282051282054, "register/verify_any": 0.875, "reward": 0.7265625, "reward_std": 0.4461594223976135, "rewards/correct/mean": 0.7265625, "rewards/correct/std": 0.4461594223976135, "rollout/chars": 5504.625, "rollout/commit": 0.65625, "rollout/reward": 0.59375, "rollout/trunc": 0.28125, "sampling/importance_sampling_ratio/max": 2.8255505561828613, "sampling/importance_sampling_ratio/mean": 0.9517054557800293, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.867004156112671, "sampling/sampling_logp_difference/mean": 0.005516871809959412, "step": 80, "step_time": 176.51940037903842 }, { "epoch": 0.43010752688172044, "grad_norm": 0.0035058551002293825, "learning_rate": 1e-05, "loss": -0.047525111585855484, "step": 80 }, { "clip_ratio": 0.0045759479553453275, "completions/clipped_ratio": 0.119140625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4088.0, "completions/mean_length": 1629.345703125, "completions/mean_terminated_length": 1304.7672119140625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.43548387096774194, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 62456915.0, "register/okay": 0.65625, "register/okay_any": 0.875, "register/premium": 0.354978354978355, "register/verify_any": 0.84375, "reward": 0.6875, "reward_std": 0.4639657139778137, "rewards/correct/mean": 0.6875, "rewards/correct/std": 0.4639657139778137, "rollout/chars": 4725.3125, "rollout/commit": 0.875, "rollout/reward": 0.6875, "rollout/trunc": 0.09375, "sampling/importance_sampling_ratio/max": 2.874391794204712, "sampling/importance_sampling_ratio/mean": 0.8829848766326904, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.5236120223999023, "sampling/sampling_logp_difference/mean": 0.005041817203164101, "step": 81, "step_time": 193.45293980598217 }, { "epoch": 0.43548387096774194, "grad_norm": 0.003919823095202446, "learning_rate": 1e-05, "loss": -0.02535119280219078, "step": 81 }, { "clip_ratio": 0.0035743965909205144, "completions/clipped_ratio": 0.078125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4096.0, "completions/mean_length": 1554.0234375, "completions/mean_terminated_length": 1338.6016845703125, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.44086021505376344, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 63307231.0, "register/okay": 0.78125, "register/okay_any": 0.9375, "register/premium": 0.5542857142857143, "register/verify_any": 0.9375, "reward": 0.771484375, "reward_std": 0.4202871024608612, "rewards/correct/mean": 0.771484375, "rewards/correct/std": 0.4202871024608612, "rollout/chars": 7963.75, "rollout/commit": 0.84375, "rollout/reward": 0.71875, "rollout/trunc": 0.15625, "sampling/importance_sampling_ratio/max": 2.9019503593444824, "sampling/importance_sampling_ratio/mean": 0.9446427822113037, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.7874631881713867, "sampling/sampling_logp_difference/mean": 0.00514925317838788, "step": 82, "step_time": 163.73912520805607 }, { "epoch": 0.44086021505376344, "grad_norm": 0.004091319162398577, "learning_rate": 1e-05, "loss": -0.03757448121905327, "step": 82 }, { "clip_ratio": 0.0028516641732494463, "completions/clipped_ratio": 0.07421875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4050.0, "completions/mean_length": 1344.78125, "completions/mean_terminated_length": 1124.2193603515625, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.44623655913978494, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 64059695.0, "register/okay": 0.75, "register/okay_any": 0.90625, "register/premium": 0.375, "register/verify_any": 0.90625, "reward": 0.755859375, "reward_std": 0.42999663949012756, "rewards/correct/mean": 0.755859375, "rewards/correct/std": 0.42999663949012756, "rollout/chars": 6170.875, "rollout/commit": 0.96875, "rollout/reward": 0.90625, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.738131523132324, "sampling/importance_sampling_ratio/mean": 0.9077047109603882, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 3.0889058113098145, "sampling/sampling_logp_difference/mean": 0.005292165093123913, "step": 83, "step_time": 145.8009820789448 }, { "epoch": 0.44623655913978494, "grad_norm": 0.004029201343655586, "learning_rate": 1e-05, "loss": -0.05332073196768761, "step": 83 }, { "clip_ratio": 0.002568083282767475, "completions/clipped_ratio": 0.123046875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4027.0, "completions/mean_length": 1656.58203125, "completions/mean_terminated_length": 1314.302978515625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.45161290322580644, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 64973081.0, "register/okay": 0.625, "register/okay_any": 0.75, "register/premium": 0.46666666666666673, "register/verify_any": 0.75, "reward": 0.71484375, "reward_std": 0.45193037390708923, "rewards/correct/mean": 0.71484375, "rewards/correct/std": 0.45193037390708923, "rollout/chars": 4307.28125, "rollout/commit": 0.8125, "rollout/reward": 0.625, "rollout/trunc": 0.0625, "sampling/importance_sampling_ratio/max": 2.981221914291382, "sampling/importance_sampling_ratio/mean": 0.92153000831604, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.6738214492797852, "sampling/sampling_logp_difference/mean": 0.005299369804561138, "step": 84, "step_time": 168.01446572120767 }, { "epoch": 0.45161290322580644, "grad_norm": 0.003875133115798235, "learning_rate": 1e-05, "loss": -0.020854422822594643, "step": 84 }, { "clip_ratio": 0.004443318704261401, "completions/clipped_ratio": 0.037109375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4066.0, "completions/mean_length": 1213.986328125, "completions/mean_terminated_length": 1111.2088623046875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.45698924731182794, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 65644098.0, "register/okay": 0.59375, "register/okay_any": 0.6875, "register/premium": 0.4817813765182186, "register/verify_any": 0.71875, "reward": 0.814453125, "reward_std": 0.38912075757980347, "rewards/correct/mean": 0.814453125, "rewards/correct/std": 0.38912075757980347, "rollout/chars": 3110.40625, "rollout/commit": 0.84375, "rollout/reward": 0.59375, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.9168148040771484, "sampling/importance_sampling_ratio/mean": 0.9617250561714172, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.7552123069763184, "sampling/sampling_logp_difference/mean": 0.005059793591499329, "step": 85, "step_time": 136.79254827008117 }, { "epoch": 0.45698924731182794, "grad_norm": 0.004062873311340809, "learning_rate": 1e-05, "loss": -0.04033565893769264, "step": 85 }, { "clip_ratio": 0.006071408856769267, "completions/clipped_ratio": 0.08984375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4035.0, "completions/mean_length": 1351.279296875, "completions/mean_terminated_length": 1088.5858154296875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.46236559139784944, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 66388289.0, "register/okay": 0.65625, "register/okay_any": 0.875, "register/premium": 0.04329004329004327, "register/verify_any": 0.875, "reward": 0.802734375, "reward_std": 0.3983237147331238, "rewards/correct/mean": 0.802734375, "rewards/correct/std": 0.3983237147331238, "rollout/chars": 2916.28125, "rollout/commit": 1.0, "rollout/reward": 0.9375, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.883626699447632, "sampling/importance_sampling_ratio/mean": 0.9316653609275818, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.9200735092163086, "sampling/sampling_logp_difference/mean": 0.005202242638915777, "step": 86, "step_time": 190.0210038361838 }, { "epoch": 0.46236559139784944, "grad_norm": 0.003920720424503088, "learning_rate": 1e-05, "loss": -0.026152227073907852, "step": 86 }, { "clip_ratio": 0.004060513333001836, "completions/clipped_ratio": 0.10546875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4034.0, "completions/mean_length": 1601.9453125, "completions/mean_terminated_length": 1312.81005859375, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.46774193548387094, "frac_reward_zero_std": 0.09375, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 67274901.0, "register/okay": 0.84375, "register/okay_any": 1.0, "register/premium": 0.2148148148148148, "register/verify_any": 0.90625, "reward": 0.71484375, "reward_std": 0.45193037390708923, "rewards/correct/mean": 0.71484375, "rewards/correct/std": 0.45193037390708923, "rollout/chars": 6292.75, "rollout/commit": 0.84375, "rollout/reward": 0.78125, "rollout/trunc": 0.125, "sampling/importance_sampling_ratio/max": 2.953037977218628, "sampling/importance_sampling_ratio/mean": 0.9629545211791992, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.6056404113769531, "sampling/sampling_logp_difference/mean": 0.005148344207555056, "step": 87, "step_time": 161.60170425794786 }, { "epoch": 0.46774193548387094, "grad_norm": 0.00469123525545001, "learning_rate": 1e-05, "loss": -0.009884790517389774, "step": 87 }, { "clip_ratio": 0.008021032465876488, "completions/clipped_ratio": 0.1171875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4018.0, "completions/mean_length": 1528.216796875, "completions/mean_terminated_length": 1187.389404296875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.4731182795698925, "frac_reward_zero_std": 0.09375, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 68127332.0, "register/okay": 0.625, "register/okay_any": 0.84375, "register/premium": -0.09999999999999998, "register/verify_any": 0.84375, "reward": 0.71484375, "reward_std": 0.45193037390708923, "rewards/correct/mean": 0.71484375, "rewards/correct/std": 0.45193037390708923, "rollout/chars": 5176.375, "rollout/commit": 0.78125, "rollout/reward": 0.6875, "rollout/trunc": 0.1875, "sampling/importance_sampling_ratio/max": 2.8957531452178955, "sampling/importance_sampling_ratio/mean": 0.9481742978096008, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.7427730560302734, "sampling/sampling_logp_difference/mean": 0.005054045002907515, "step": 88, "step_time": 169.949737901974 }, { "epoch": 0.4731182795698925, "grad_norm": 0.004581862594932318, "learning_rate": 1e-05, "loss": -0.0018323007971048355, "step": 88 }, { "clip_ratio": 0.002654670177889784, "completions/clipped_ratio": 0.146484375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4080.0, "completions/mean_length": 1628.328125, "completions/mean_terminated_length": 1223.52392578125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.478494623655914, "frac_reward_zero_std": 0.15625, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 69018108.0, "register/okay": 0.75, "register/okay_any": 0.875, "register/premium": 0.20833333333333337, "register/verify_any": 0.875, "reward": 0.703125, "reward_std": 0.45732781291007996, "rewards/correct/mean": 0.703125, "rewards/correct/std": 0.45732781291007996, "rollout/chars": 6276.6875, "rollout/commit": 0.59375, "rollout/reward": 0.53125, "rollout/trunc": 0.40625, "sampling/importance_sampling_ratio/max": 2.8828799724578857, "sampling/importance_sampling_ratio/mean": 0.9430744051933289, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.265594005584717, "sampling/sampling_logp_difference/mean": 0.005124185234308243, "step": 89, "step_time": 198.9125518440269 }, { "epoch": 0.478494623655914, "grad_norm": 0.0035977123770862818, "learning_rate": 1e-05, "loss": -0.026564961299300194, "step": 89 }, { "clip_ratio": 0.0012757878755564889, "completions/clipped_ratio": 0.0625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4050.0, "completions/mean_length": 1320.9765625, "completions/mean_terminated_length": 1143.92919921875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.4838709677419355, "frac_reward_zero_std": 0.21875, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 69753280.0, "register/okay": 0.625, "register/okay_any": 0.78125, "register/premium": 0.2833333333333333, "register/verify_any": 0.78125, "reward": 0.8203125, "reward_std": 0.38430243730545044, "rewards/correct/mean": 0.8203125, "rewards/correct/std": 0.38430243730545044, "rollout/chars": 2687.15625, "rollout/commit": 0.90625, "rollout/reward": 0.84375, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.881863594055176, "sampling/importance_sampling_ratio/mean": 0.9441045522689819, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.7411904335021973, "sampling/sampling_logp_difference/mean": 0.005257273092865944, "step": 90, "step_time": 126.86479857406812 }, { "epoch": 0.4838709677419355, "grad_norm": 0.004005418624728918, "learning_rate": 1e-05, "loss": -0.04725252091884613, "step": 90 }, { "clip_ratio": 0.002722416330698252, "completions/clipped_ratio": 0.16796875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4039.0, "completions/mean_length": 1679.91015625, "completions/mean_terminated_length": 1192.1549072265625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.489247311827957, "frac_reward_zero_std": 0.15625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 70684130.0, "register/okay": 0.625, "register/okay_any": 0.75, "register/premium": 0.55, "register/verify_any": 0.75, "reward": 0.681640625, "reward_std": 0.46629536151885986, "rewards/correct/mean": 0.681640625, "rewards/correct/std": 0.46629536151885986, "rollout/chars": 3912.28125, "rollout/commit": 0.78125, "rollout/reward": 0.59375, "rollout/trunc": 0.125, "sampling/importance_sampling_ratio/max": 2.8961503505706787, "sampling/importance_sampling_ratio/mean": 0.9452307820320129, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.6795144081115723, "sampling/sampling_logp_difference/mean": 0.005413533188402653, "step": 91, "step_time": 200.4919060600805 }, { "epoch": 0.489247311827957, "grad_norm": 0.004250395577400923, "learning_rate": 1e-05, "loss": 0.010246439836919308, "step": 91 }, { "clip_ratio": 0.004926546272372434, "completions/clipped_ratio": 0.076171875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4069.0, "completions/mean_length": 1475.33203125, "completions/mean_terminated_length": 1259.2515869140625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.4946236559139785, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 71499084.0, "register/okay": 0.8125, "register/okay_any": 0.96875, "register/premium": 0.1282051282051282, "register/verify_any": 0.9375, "reward": 0.771484375, "reward_std": 0.4202871024608612, "rewards/correct/mean": 0.771484375, "rewards/correct/std": 0.4202871024608612, "rollout/chars": 3188.53125, "rollout/commit": 0.96875, "rollout/reward": 0.9375, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.9482409954071045, "sampling/importance_sampling_ratio/mean": 0.9591692090034485, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.9159293174743652, "sampling/sampling_logp_difference/mean": 0.00535973347723484, "step": 92, "step_time": 178.4256179970107 }, { "epoch": 0.4946236559139785, "grad_norm": 0.004815950524061918, "learning_rate": 1e-05, "loss": -0.03477675840258598, "step": 92 }, { "clip_ratio": 0.006788148959230966, "completions/clipped_ratio": 0.125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4042.0, "completions/mean_length": 1540.63671875, "completions/mean_terminated_length": 1193.8125, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.5, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 72341794.0, "register/okay": 0.84375, "register/okay_any": 0.9375, "register/premium": -0.09629629629629632, "register/verify_any": 0.9375, "reward": 0.736328125, "reward_std": 0.4410543739795685, "rewards/correct/mean": 0.736328125, "rewards/correct/std": 0.4410543739795685, "rollout/chars": 5976.1875, "rollout/commit": 0.875, "rollout/reward": 0.71875, "rollout/trunc": 0.21875, "sampling/importance_sampling_ratio/max": 2.954418659210205, "sampling/importance_sampling_ratio/mean": 0.946603536605835, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.7686195373535156, "sampling/sampling_logp_difference/mean": 0.005590211600065231, "step": 93, "step_time": 199.2062720261165 }, { "epoch": 0.5, "grad_norm": 0.004292262252420187, "learning_rate": 1e-05, "loss": -0.02804984711110592, "step": 93 }, { "clip_ratio": 0.0038180052279130905, "completions/clipped_ratio": 0.09765625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4079.0, "completions/mean_length": 1538.888671875, "completions/mean_terminated_length": 1262.14501953125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.5053763440860215, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 73188377.0, "register/okay": 0.78125, "register/okay_any": 0.96875, "register/premium": 0.24571428571428566, "register/verify_any": 0.9375, "reward": 0.744140625, "reward_std": 0.43676990270614624, "rewards/correct/mean": 0.744140625, "rewards/correct/std": 0.43676990270614624, "rollout/chars": 3969.71875, "rollout/commit": 0.9375, "rollout/reward": 0.90625, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.6801586151123047, "sampling/importance_sampling_ratio/mean": 0.9305009841918945, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.8462352752685547, "sampling/sampling_logp_difference/mean": 0.005204683169722557, "step": 94, "step_time": 200.44946134305792 }, { "epoch": 0.5053763440860215, "grad_norm": 0.00365109508857131, "learning_rate": 1e-05, "loss": -0.04738461971282959, "step": 94 }, { "clip_ratio": 0.0021819106390807974, "completions/clipped_ratio": 0.125, "completions/max_length": 4096.0, "completions/max_terminated_length": 3980.0, "completions/mean_length": 1412.916015625, "completions/mean_terminated_length": 1029.618408203125, "completions/min_length": 8.0, "completions/min_terminated_length": 8.0, "epoch": 0.510752688172043, "frac_reward_zero_std": 0.1875, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 73961758.0, "register/okay": 0.71875, "register/okay_any": 0.84375, "register/premium": 0.3381642512077295, "register/verify_any": 0.84375, "reward": 0.724609375, "reward_std": 0.44714778661727905, "rewards/correct/mean": 0.724609375, "rewards/correct/std": 0.44714778661727905, "rollout/chars": 4634.75, "rollout/commit": 0.875, "rollout/reward": 0.6875, "rollout/trunc": 0.125, "sampling/importance_sampling_ratio/max": 2.8489890098571777, "sampling/importance_sampling_ratio/mean": 0.9574590921401978, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.983301877975464, "sampling/sampling_logp_difference/mean": 0.005294046364724636, "step": 95, "step_time": 153.16449551511323 }, { "epoch": 0.510752688172043, "grad_norm": 0.004286632873117924, "learning_rate": 1e-05, "loss": -0.02807465009391308, "step": 95 }, { "clip_ratio": 0.0030943442397983745, "completions/clipped_ratio": 0.08984375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4094.0, "completions/mean_length": 1513.775390625, "completions/mean_terminated_length": 1258.877685546875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.5161290322580645, "frac_reward_zero_std": 0.0, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 74789531.0, "register/okay": 0.71875, "register/okay_any": 0.875, "register/premium": 0.4251207729468599, "register/verify_any": 0.84375, "reward": 0.734375, "reward_std": 0.44209739565849304, "rewards/correct/mean": 0.734375, "rewards/correct/std": 0.44209739565849304, "rollout/chars": 3571.625, "rollout/commit": 0.9375, "rollout/reward": 0.75, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.9837379455566406, "sampling/importance_sampling_ratio/mean": 0.9289745092391968, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.7906036376953125, "sampling/sampling_logp_difference/mean": 0.005226943176239729, "step": 96, "step_time": 182.66966320894426 }, { "epoch": 0.5161290322580645, "grad_norm": 0.00381438247859478, "learning_rate": 1e-05, "loss": -0.03624545782804489, "step": 96 }, { "clip_ratio": 0.0024594998874363228, "completions/clipped_ratio": 0.048828125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4046.0, "completions/mean_length": 1241.30078125, "completions/mean_terminated_length": 1094.7557373046875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.521505376344086, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 75476677.0, "register/okay": 0.84375, "register/okay_any": 0.96875, "register/premium": -0.13333333333333341, "register/verify_any": 0.9375, "reward": 0.822265625, "reward_std": 0.3826628625392914, "rewards/correct/mean": 0.822265625, "rewards/correct/std": 0.3826628625392914, "rollout/chars": 6663.25, "rollout/commit": 0.6875, "rollout/reward": 0.6875, "rollout/trunc": 0.3125, "sampling/importance_sampling_ratio/max": 2.845426321029663, "sampling/importance_sampling_ratio/mean": 0.8784292936325073, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.7628402709960938, "sampling/sampling_logp_difference/mean": 0.0056271725334227085, "step": 97, "step_time": 135.0193039138685 }, { "epoch": 0.521505376344086, "grad_norm": 0.003446342423558235, "learning_rate": 1e-05, "loss": -0.04358059912919998, "step": 97 }, { "clip_ratio": 0.001906061550926097, "completions/clipped_ratio": 0.0625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4082.0, "completions/mean_length": 1534.107421875, "completions/mean_terminated_length": 1363.314697265625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.5268817204301075, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 76314668.0, "register/okay": 0.59375, "register/okay_any": 0.84375, "register/premium": 0.4048582995951417, "register/verify_any": 0.84375, "reward": 0.771484375, "reward_std": 0.4202871024608612, "rewards/correct/mean": 0.771484375, "rewards/correct/std": 0.4202871024608612, "rollout/chars": 5977.65625, "rollout/commit": 0.96875, "rollout/reward": 0.625, "rollout/trunc": 0.0625, "sampling/importance_sampling_ratio/max": 2.9907774925231934, "sampling/importance_sampling_ratio/mean": 0.9597388505935669, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.6243586540222168, "sampling/sampling_logp_difference/mean": 0.005072455853223801, "step": 98, "step_time": 148.87809039419517 }, { "epoch": 0.5268817204301075, "grad_norm": 0.0044438703916966915, "learning_rate": 1e-05, "loss": -0.045310527086257935, "step": 98 }, { "clip_ratio": 0.001449380104759257, "completions/clipped_ratio": 0.103515625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4071.0, "completions/mean_length": 1514.09765625, "completions/mean_terminated_length": 1215.969482421875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.532258064516129, "frac_reward_zero_std": 0.1875, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 77144878.0, "register/okay": 0.8125, "register/okay_any": 0.9375, "register/premium": 0.5512820512820513, "register/verify_any": 0.9375, "reward": 0.748046875, "reward_std": 0.43455907702445984, "rewards/correct/mean": 0.748046875, "rewards/correct/std": 0.43455907702445984, "rollout/chars": 4781.375, "rollout/commit": 0.96875, "rollout/reward": 0.78125, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.7978718280792236, "sampling/importance_sampling_ratio/mean": 0.9005168676376343, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.9524598121643066, "sampling/sampling_logp_difference/mean": 0.00544597115367651, "step": 99, "step_time": 205.2432117689168 }, { "epoch": 0.532258064516129, "grad_norm": 0.0032835553865879774, "learning_rate": 1e-05, "loss": -0.055465392768383026, "step": 99 }, { "clip_ratio": 0.004729641624464875, "completions/clipped_ratio": 0.068359375, "completions/max_length": 4096.0, "completions/max_terminated_length": 3999.0, "completions/mean_length": 1295.177734375, "completions/mean_terminated_length": 1089.6666259765625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.5376344086021505, "frac_reward_zero_std": 0.125, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 77860569.0, "register/okay": 0.875, "register/okay_any": 0.96875, "register/premium": 0.4285714285714286, "register/verify_any": 0.96875, "reward": 0.8046875, "reward_std": 0.3968288004398346, "rewards/correct/mean": 0.8046875, "rewards/correct/std": 0.3968288004398346, "rollout/chars": 6258.25, "rollout/commit": 0.90625, "rollout/reward": 0.875, "rollout/trunc": 0.125, "sampling/importance_sampling_ratio/max": 2.657564640045166, "sampling/importance_sampling_ratio/mean": 0.9256951808929443, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.9996352195739746, "sampling/sampling_logp_difference/mean": 0.005451052449643612, "step": 100, "step_time": 128.70375625789165 }, { "epoch": 0.5376344086021505, "grad_norm": 0.00425604498013854, "learning_rate": 1e-05, "loss": -0.027618352323770523, "step": 100 }, { "clip_ratio": 0.006132574421599202, "completions/clipped_ratio": 0.119140625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4084.0, "completions/mean_length": 1527.84765625, "completions/mean_terminated_length": 1189.5543212890625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.543010752688172, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 78699291.0, "register/okay": 0.84375, "register/okay_any": 0.96875, "register/premium": -0.06666666666666671, "register/verify_any": 0.96875, "reward": 0.73046875, "reward_std": 0.44415023922920227, "rewards/correct/mean": 0.73046875, "rewards/correct/std": 0.44415023922920227, "rollout/chars": 9932.78125, "rollout/commit": 0.5625, "rollout/reward": 0.34375, "rollout/trunc": 0.5625, "sampling/importance_sampling_ratio/max": 2.8872036933898926, "sampling/importance_sampling_ratio/mean": 0.8818156123161316, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.8228774070739746, "sampling/sampling_logp_difference/mean": 0.005235197022557259, "step": 101, "step_time": 208.4062208579271 }, { "epoch": 0.543010752688172, "grad_norm": 0.003870123066008091, "learning_rate": 1e-05, "loss": -0.008510656654834747, "step": 101 }, { "clip_ratio": 0.0023458289954305656, "completions/clipped_ratio": 0.130859375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4068.0, "completions/mean_length": 1561.935546875, "completions/mean_terminated_length": 1184.042724609375, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.5483870967741935, "frac_reward_zero_std": 0.125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 79554762.0, "register/okay": 0.65625, "register/okay_any": 0.9375, "register/premium": -0.025974025974025983, "register/verify_any": 0.9375, "reward": 0.74609375, "reward_std": 0.43567025661468506, "rewards/correct/mean": 0.74609375, "rewards/correct/std": 0.43567025661468506, "rollout/chars": 5361.71875, "rollout/commit": 0.84375, "rollout/reward": 0.4375, "rollout/trunc": 0.25, "sampling/importance_sampling_ratio/max": 2.8685460090637207, "sampling/importance_sampling_ratio/mean": 0.9590466022491455, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.9776654243469238, "sampling/sampling_logp_difference/mean": 0.0051699173636734486, "step": 102, "step_time": 182.73181396716973 }, { "epoch": 0.5483870967741935, "grad_norm": 0.0039959866553545, "learning_rate": 1e-05, "loss": 0.004553236998617649, "step": 102 }, { "clip_ratio": 0.0022633807608940515, "completions/clipped_ratio": 0.0703125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4077.0, "completions/mean_length": 1380.849609375, "completions/mean_terminated_length": 1175.502197265625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.553763440860215, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 80313197.0, "register/okay": 0.6875, "register/okay_any": 0.96875, "register/premium": 0.3090909090909091, "register/verify_any": 0.96875, "reward": 0.783203125, "reward_std": 0.4124660789966583, "rewards/correct/mean": 0.783203125, "rewards/correct/std": 0.4124660789966583, "rollout/chars": 5332.96875, "rollout/commit": 0.96875, "rollout/reward": 0.8125, "rollout/trunc": 0.0625, "sampling/importance_sampling_ratio/max": 2.8667449951171875, "sampling/importance_sampling_ratio/mean": 0.955405592918396, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.9627022743225098, "sampling/sampling_logp_difference/mean": 0.005295293405652046, "step": 103, "step_time": 163.34218965290347 }, { "epoch": 0.553763440860215, "grad_norm": 0.003959255758672953, "learning_rate": 1e-05, "loss": -0.056593164801597595, "step": 103 }, { "clip_ratio": 0.0005578181179544117, "completions/clipped_ratio": 0.09375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4068.0, "completions/mean_length": 1631.8046875, "completions/mean_terminated_length": 1376.887939453125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.5591397849462365, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 81215273.0, "register/okay": 0.84375, "register/okay_any": 0.96875, "register/premium": 0.45185185185185184, "register/verify_any": 0.96875, "reward": 0.759765625, "reward_std": 0.4276435375213623, "rewards/correct/mean": 0.759765625, "rewards/correct/std": 0.4276435375213623, "rollout/chars": 6477.6875, "rollout/commit": 0.96875, "rollout/reward": 0.78125, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.9729526042938232, "sampling/importance_sampling_ratio/mean": 0.9439687132835388, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.9384076595306396, "sampling/sampling_logp_difference/mean": 0.004915799479931593, "step": 104, "step_time": 152.0168167478987 }, { "epoch": 0.5591397849462365, "grad_norm": 0.00464340765029192, "learning_rate": 1e-05, "loss": -0.003207690082490444, "step": 104 }, { "clip_ratio": 0.0029293977775068925, "completions/clipped_ratio": 0.14453125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4096.0, "completions/mean_length": 1825.064453125, "completions/mean_terminated_length": 1441.390380859375, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.5645161290322581, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 82215834.0, "register/okay": 0.71875, "register/okay_any": 0.8125, "register/premium": 0.36231884057971014, "register/verify_any": 0.84375, "reward": 0.705078125, "reward_std": 0.4564536213874817, "rewards/correct/mean": 0.705078125, "rewards/correct/std": 0.4564536213874817, "rollout/chars": 5686.3125, "rollout/commit": 0.75, "rollout/reward": 0.59375, "rollout/trunc": 0.28125, "sampling/importance_sampling_ratio/max": 2.993610382080078, "sampling/importance_sampling_ratio/mean": 0.9261826276779175, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.921788215637207, "sampling/sampling_logp_difference/mean": 0.005259702913463116, "step": 105, "step_time": 194.43338847195264 }, { "epoch": 0.5645161290322581, "grad_norm": 0.0032441350631415844, "learning_rate": 1e-05, "loss": -0.03612882271409035, "step": 105 }, { "clip_ratio": 0.0023796914588274376, "completions/clipped_ratio": 0.03515625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4076.0, "completions/mean_length": 1087.74609375, "completions/mean_terminated_length": 986.2611694335938, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.5698924731182796, "frac_reward_zero_std": 0.15625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 82830712.0, "register/okay": 0.78125, "register/okay_any": 1.0, "register/premium": 0.24571428571428566, "register/verify_any": 1.0, "reward": 0.80078125, "reward_std": 0.39980348944664, "rewards/correct/mean": 0.80078125, "rewards/correct/std": 0.39980348944664, "rollout/chars": 3055.34375, "rollout/commit": 1.0, "rollout/reward": 0.90625, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.878859043121338, "sampling/importance_sampling_ratio/mean": 0.9483377933502197, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.1337199211120605, "sampling/sampling_logp_difference/mean": 0.005619116593152285, "step": 106, "step_time": 150.9896876619896 }, { "epoch": 0.5698924731182796, "grad_norm": 0.0034882868640124798, "learning_rate": 1e-05, "loss": -0.056708455085754395, "step": 106 }, { "clip_ratio": 0.0007898037578115691, "completions/clipped_ratio": 0.123046875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4086.0, "completions/mean_length": 1613.849609375, "completions/mean_terminated_length": 1265.57470703125, "completions/min_length": 5.0, "completions/min_terminated_length": 5.0, "epoch": 0.5752688172043011, "frac_reward_zero_std": 0.1875, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 83716075.0, "register/okay": 0.84375, "register/okay_any": 0.90625, "register/premium": -0.022222222222222254, "register/verify_any": 0.9375, "reward": 0.78125, "reward_std": 0.41380295157432556, "rewards/correct/mean": 0.78125, "rewards/correct/std": 0.41380295157432556, "rollout/chars": 5727.15625, "rollout/commit": 0.84375, "rollout/reward": 0.78125, "rollout/trunc": 0.1875, "sampling/importance_sampling_ratio/max": 2.9724841117858887, "sampling/importance_sampling_ratio/mean": 0.9478975534439087, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.8725066184997559, "sampling/sampling_logp_difference/mean": 0.005161927081644535, "step": 107, "step_time": 321.37006054999074 }, { "epoch": 0.5752688172043011, "grad_norm": 0.003943478688597679, "learning_rate": 1e-05, "loss": 0.02987394481897354, "step": 107 }, { "clip_ratio": 0.001146576663813903, "completions/clipped_ratio": 0.130859375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4044.0, "completions/mean_length": 1655.458984375, "completions/mean_terminated_length": 1288.0067138671875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.5806451612903226, "frac_reward_zero_std": 0.25, "groups/all_right": 0.5, "groups/all_wrong": 0.5, "groups/live_frac": 0.0, "groups/n": 2, "num_tokens": 84616150.0, "register/okay": 0.875, "register/okay_any": 0.96875, "register/premium": -0.5714285714285714, "register/verify_any": 1.0, "reward": 0.7734375, "reward_std": 0.4190165400505066, "rewards/correct/mean": 0.7734375, "rewards/correct/std": 0.4190165400505066, "rollout/chars": 7346.25, "rollout/commit": 0.65625, "rollout/reward": 0.5, "rollout/trunc": 0.4375, "sampling/importance_sampling_ratio/max": 2.622361898422241, "sampling/importance_sampling_ratio/mean": 0.941210150718689, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.9779677391052246, "sampling/sampling_logp_difference/mean": 0.00511105265468359, "step": 108, "step_time": 212.85403366899118 }, { "epoch": 0.5806451612903226, "grad_norm": 0.003243943676352501, "learning_rate": 1e-05, "loss": -0.01560639776289463, "step": 108 }, { "clip_ratio": 0.0017009870779247649, "completions/clipped_ratio": 0.154296875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4074.0, "completions/mean_length": 1643.6875, "completions/mean_terminated_length": 1205.69970703125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.5860215053763441, "frac_reward_zero_std": 0.125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 85517366.0, "register/okay": 0.71875, "register/okay_any": 0.8125, "register/premium": 0.40579710144927533, "register/verify_any": 0.78125, "reward": 0.734375, "reward_std": 0.44209739565849304, "rewards/correct/mean": 0.734375, "rewards/correct/std": 0.44209739565849304, "rollout/chars": 5649.75, "rollout/commit": 0.71875, "rollout/reward": 0.625, "rollout/trunc": 0.25, "sampling/importance_sampling_ratio/max": 2.875641107559204, "sampling/importance_sampling_ratio/mean": 0.9502192735671997, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.9211153984069824, "sampling/sampling_logp_difference/mean": 0.005179607309401035, "step": 109, "step_time": 193.66581050987588 }, { "epoch": 0.5860215053763441, "grad_norm": 0.0039439816027879715, "learning_rate": 1e-05, "loss": -0.010169509798288345, "step": 109 }, { "clip_ratio": 0.0007577786858519175, "completions/clipped_ratio": 0.029296875, "completions/max_length": 4096.0, "completions/max_terminated_length": 3749.0, "completions/mean_length": 1055.728515625, "completions/mean_terminated_length": 963.9697875976562, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.5913978494623656, "frac_reward_zero_std": 0.21875, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 86114763.0, "register/okay": 0.90625, "register/okay_any": 0.9375, "register/premium": 0.5977011494252873, "register/verify_any": 0.9375, "reward": 0.853515625, "reward_std": 0.35393697023391724, "rewards/correct/mean": 0.853515625, "rewards/correct/std": 0.35393697023391724, "rollout/chars": 4833.8125, "rollout/commit": 0.9375, "rollout/reward": 0.875, "rollout/trunc": 0.0625, "sampling/importance_sampling_ratio/max": 2.959141492843628, "sampling/importance_sampling_ratio/mean": 0.9853883981704712, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.07066011428833, "sampling/sampling_logp_difference/mean": 0.005485204979777336, "step": 110, "step_time": 137.919253686734 }, { "epoch": 0.5913978494623656, "grad_norm": 0.005126842763274908, "learning_rate": 1e-05, "loss": -0.03054872527718544, "step": 110 }, { "clip_ratio": 0.002382328377734666, "completions/clipped_ratio": 0.080078125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4071.0, "completions/mean_length": 1337.865234375, "completions/mean_terminated_length": 1106.380126953125, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.5967741935483871, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 86849526.0, "register/okay": 0.875, "register/okay_any": 0.90625, "register/premium": -0.0714285714285714, "register/verify_any": 0.90625, "reward": 0.80859375, "reward_std": 0.3937928080558777, "rewards/correct/mean": 0.80859375, "rewards/correct/std": 0.3937928080558777, "rollout/chars": 4992.4375, "rollout/commit": 0.8125, "rollout/reward": 0.6875, "rollout/trunc": 0.25, "sampling/importance_sampling_ratio/max": 2.9767589569091797, "sampling/importance_sampling_ratio/mean": 0.967754602432251, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.400851249694824, "sampling/sampling_logp_difference/mean": 0.00542665459215641, "step": 111, "step_time": 142.2021625560592 }, { "epoch": 0.5967741935483871, "grad_norm": 0.004243534058332443, "learning_rate": 1e-05, "loss": -0.0039045270532369614, "step": 111 }, { "clip_ratio": 0.0012413713384376024, "completions/clipped_ratio": 0.083984375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4021.0, "completions/mean_length": 1326.625, "completions/mean_terminated_length": 1072.7164306640625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.6021505376344086, "frac_reward_zero_std": 0.09375, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 87585430.0, "register/okay": 0.84375, "register/okay_any": 0.9375, "register/premium": 0.36296296296296293, "register/verify_any": 0.9375, "reward": 0.771484375, "reward_std": 0.4202871024608612, "rewards/correct/mean": 0.771484375, "rewards/correct/std": 0.4202871024608612, "rollout/chars": 2726.75, "rollout/commit": 0.96875, "rollout/reward": 0.90625, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.925628185272217, "sampling/importance_sampling_ratio/mean": 0.942460298538208, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.1665735244750977, "sampling/sampling_logp_difference/mean": 0.005194856319576502, "step": 112, "step_time": 175.13962990115397 }, { "epoch": 0.6021505376344086, "grad_norm": 0.004170512780547142, "learning_rate": 1e-05, "loss": -0.009271319024264812, "step": 112 }, { "clip_ratio": 0.0004819950700039044, "completions/clipped_ratio": 0.12890625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4044.0, "completions/mean_length": 1711.861328125, "completions/mean_terminated_length": 1359.0516357421875, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.6075268817204301, "frac_reward_zero_std": 0.09375, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 88525711.0, "register/okay": 0.8125, "register/okay_any": 0.9375, "register/premium": 0.33333333333333337, "register/verify_any": 0.90625, "reward": 0.759765625, "reward_std": 0.4276435375213623, "rewards/correct/mean": 0.759765625, "rewards/correct/std": 0.4276435375213623, "rollout/chars": 3578.1875, "rollout/commit": 0.96875, "rollout/reward": 0.9375, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.968146562576294, "sampling/importance_sampling_ratio/mean": 0.9759321808815002, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.0367932319641113, "sampling/sampling_logp_difference/mean": 0.004968651570379734, "step": 113, "step_time": 198.56515949894674 }, { "epoch": 0.6075268817204301, "grad_norm": 0.004303612280637026, "learning_rate": 1e-05, "loss": 0.01028444804251194, "step": 113 }, { "clip_ratio": 0.006990602395831047, "completions/clipped_ratio": 0.14453125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4093.0, "completions/mean_length": 1598.763671875, "completions/mean_terminated_length": 1176.8560791015625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.6129032258064516, "frac_reward_zero_std": 0.09375, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 89402598.0, "register/okay": 0.6875, "register/okay_any": 0.90625, "register/premium": 0.13636363636363635, "register/verify_any": 0.9375, "reward": 0.689453125, "reward_std": 0.46317005157470703, "rewards/correct/mean": 0.689453125, "rewards/correct/std": 0.46317005157470703, "rollout/chars": 6123.71875, "rollout/commit": 0.6875, "rollout/reward": 0.59375, "rollout/trunc": 0.34375, "sampling/importance_sampling_ratio/max": 2.986074447631836, "sampling/importance_sampling_ratio/mean": 0.9108353853225708, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.7797436714172363, "sampling/sampling_logp_difference/mean": 0.0051242513582110405, "step": 114, "step_time": 154.62885084090522 }, { "epoch": 0.6129032258064516, "grad_norm": 0.0036691303830593824, "learning_rate": 1e-05, "loss": -0.026565581560134888, "step": 114 }, { "clip_ratio": 0.0020432556212881536, "completions/clipped_ratio": 0.078125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4060.0, "completions/mean_length": 1426.37109375, "completions/mean_terminated_length": 1203.8326416015625, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.6182795698924731, "frac_reward_zero_std": 0.15625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 90185668.0, "register/okay": 0.71875, "register/okay_any": 0.875, "register/premium": 0.2946859903381642, "register/verify_any": 0.9375, "reward": 0.796875, "reward_std": 0.4027182459831238, "rewards/correct/mean": 0.796875, "rewards/correct/std": 0.4027182459831238, "rollout/chars": 4462.09375, "rollout/commit": 0.875, "rollout/reward": 0.65625, "rollout/trunc": 0.125, "sampling/importance_sampling_ratio/max": 2.8912196159362793, "sampling/importance_sampling_ratio/mean": 0.9350206851959229, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.8074588775634766, "sampling/sampling_logp_difference/mean": 0.005165154114365578, "step": 115, "step_time": 144.79554436699254 }, { "epoch": 0.6182795698924731, "grad_norm": 0.0035540899261832237, "learning_rate": 1e-05, "loss": -0.0564810112118721, "step": 115 }, { "clip_ratio": 0.0006885218797378911, "completions/clipped_ratio": 0.10546875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4064.0, "completions/mean_length": 1749.990234375, "completions/mean_terminated_length": 1473.386474609375, "completions/min_length": 9.0, "completions/min_terminated_length": 9.0, "epoch": 0.6236559139784946, "frac_reward_zero_std": 0.09375, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 91148255.0, "register/okay": 0.71875, "register/okay_any": 0.75, "register/premium": 0.8260869565217391, "register/verify_any": 0.75, "reward": 0.76171875, "reward_std": 0.42644867300987244, "rewards/correct/mean": 0.76171875, "rewards/correct/std": 0.42644867300987244, "rollout/chars": 5027.28125, "rollout/commit": 0.875, "rollout/reward": 0.59375, "rollout/trunc": 0.0625, "sampling/importance_sampling_ratio/max": 2.987457036972046, "sampling/importance_sampling_ratio/mean": 0.9786955714225769, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.0797312259674072, "sampling/sampling_logp_difference/mean": 0.0052751884795725346, "step": 116, "step_time": 288.9033424438676 }, { "epoch": 0.6236559139784946, "grad_norm": 0.0038148893509060144, "learning_rate": 1e-05, "loss": 0.007892679423093796, "step": 116 }, { "clip_ratio": 0.00038166562620745026, "completions/clipped_ratio": 0.0390625, "completions/max_length": 4096.0, "completions/max_terminated_length": 3954.0, "completions/mean_length": 1227.751953125, "completions/mean_terminated_length": 1111.156494140625, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.6290322580645161, "frac_reward_zero_std": 0.1875, "groups/all_right": 1.0, "groups/all_wrong": 0.0, "groups/live_frac": 0.0, "groups/n": 2, "num_tokens": 91829968.0, "register/okay": 0.78125, "register/okay_any": 0.875, "register/premium": 0.0, "register/verify_any": 0.875, "reward": 0.873046875, "reward_std": 0.33324605226516724, "rewards/correct/mean": 0.873046875, "rewards/correct/std": 0.33324605226516724, "rollout/chars": 2254.25, "rollout/commit": 1.0, "rollout/reward": 1.0, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.7070865631103516, "sampling/importance_sampling_ratio/mean": 0.9875777363777161, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.9461407661437988, "sampling/sampling_logp_difference/mean": 0.005102950148284435, "step": 117, "step_time": 123.0542707979912 }, { "epoch": 0.6290322580645161, "grad_norm": 0.003495279233902693, "learning_rate": 1e-05, "loss": -0.0375981405377388, "step": 117 }, { "clip_ratio": 0.0009069568764061842, "completions/clipped_ratio": 0.060546875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4091.0, "completions/mean_length": 1587.955078125, "completions/mean_terminated_length": 1426.31396484375, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.6344086021505376, "frac_reward_zero_std": 0.1875, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 92694825.0, "register/okay": 0.78125, "register/okay_any": 0.84375, "register/premium": 0.2857142857142857, "register/verify_any": 0.84375, "reward": 0.802734375, "reward_std": 0.3983237147331238, "rewards/correct/mean": 0.802734375, "rewards/correct/std": 0.3983237147331238, "rollout/chars": 1914.1875, "rollout/commit": 0.96875, "rollout/reward": 0.9375, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.9640774726867676, "sampling/importance_sampling_ratio/mean": 0.91834557056427, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.704322338104248, "sampling/sampling_logp_difference/mean": 0.004920692183077335, "step": 118, "step_time": 161.2203213350149 }, { "epoch": 0.6344086021505376, "grad_norm": 0.004057480022311211, "learning_rate": 1e-05, "loss": -0.012538833543658257, "step": 118 }, { "clip_ratio": 0.004270345917575469, "completions/clipped_ratio": 0.041015625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4094.0, "completions/mean_length": 1387.33984375, "completions/mean_terminated_length": 1271.4908447265625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.6397849462365591, "frac_reward_zero_std": 0.1875, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 93458695.0, "register/okay": 0.8125, "register/okay_any": 0.90625, "register/premium": 0.33333333333333337, "register/verify_any": 0.90625, "reward": 0.849609375, "reward_std": 0.35780346393585205, "rewards/correct/mean": 0.849609375, "rewards/correct/std": 0.35780346393585205, "rollout/chars": 3309.34375, "rollout/commit": 0.96875, "rollout/reward": 0.9375, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.782698154449463, "sampling/importance_sampling_ratio/mean": 0.9493420720100403, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.880380630493164, "sampling/sampling_logp_difference/mean": 0.005209434311836958, "step": 119, "step_time": 121.49903532804456 }, { "epoch": 0.6397849462365591, "grad_norm": 0.0035750337410718203, "learning_rate": 1e-05, "loss": -0.04472940415143967, "step": 119 }, { "clip_ratio": 0.001427118637593594, "completions/clipped_ratio": 0.044921875, "completions/max_length": 4096.0, "completions/max_terminated_length": 3996.0, "completions/mean_length": 1149.73046875, "completions/mean_terminated_length": 1011.1533203125, "completions/min_length": 5.0, "completions/min_terminated_length": 5.0, "epoch": 0.6451612903225806, "frac_reward_zero_std": 0.09375, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 94105325.0, "register/okay": 0.875, "register/okay_any": 0.96875, "register/premium": 0.25, "register/verify_any": 0.96875, "reward": 0.8203125, "reward_std": 0.38430243730545044, "rewards/correct/mean": 0.8203125, "rewards/correct/std": 0.38430243730545044, "rollout/chars": 2050.65625, "rollout/commit": 1.0, "rollout/reward": 0.96875, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.910790205001831, "sampling/importance_sampling_ratio/mean": 0.898614764213562, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.8654298782348633, "sampling/sampling_logp_difference/mean": 0.005589846987277269, "step": 120, "step_time": 133.06771285098512 }, { "epoch": 0.6451612903225806, "grad_norm": 0.004232720006257296, "learning_rate": 1e-05, "loss": -0.03292077034711838, "step": 120 }, { "clip_ratio": 0.00211628355089033, "completions/clipped_ratio": 0.111328125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4085.0, "completions/mean_length": 1629.953125, "completions/mean_terminated_length": 1321.0198974609375, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.6505376344086021, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 94992069.0, "register/okay": 0.84375, "register/okay_any": 0.9375, "register/premium": 0.22962962962962963, "register/verify_any": 0.9375, "reward": 0.748046875, "reward_std": 0.43455907702445984, "rewards/correct/mean": 0.748046875, "rewards/correct/std": 0.43455907702445984, "rollout/chars": 7256.5625, "rollout/commit": 0.71875, "rollout/reward": 0.59375, "rollout/trunc": 0.34375, "sampling/importance_sampling_ratio/max": 2.921647310256958, "sampling/importance_sampling_ratio/mean": 0.9043474197387695, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.4236202239990234, "sampling/sampling_logp_difference/mean": 0.005125326104462147, "step": 121, "step_time": 179.37945296004182 }, { "epoch": 0.6505376344086021, "grad_norm": 0.0042900824919342995, "learning_rate": 1e-05, "loss": -0.01529322937130928, "step": 121 }, { "clip_ratio": 0.0022294048846447367, "completions/clipped_ratio": 0.12109375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4089.0, "completions/mean_length": 1643.615234375, "completions/mean_terminated_length": 1305.731201171875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.6559139784946236, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 95885824.0, "register/okay": 0.84375, "register/okay_any": 0.9375, "register/premium": 0.45185185185185184, "register/verify_any": 0.9375, "reward": 0.75390625, "reward_std": 0.4311550557613373, "rewards/correct/mean": 0.75390625, "rewards/correct/std": 0.4311550557613373, "rollout/chars": 7258.25, "rollout/commit": 0.875, "rollout/reward": 0.78125, "rollout/trunc": 0.125, "sampling/importance_sampling_ratio/max": 2.661055088043213, "sampling/importance_sampling_ratio/mean": 0.9271708130836487, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.05570125579834, "sampling/sampling_logp_difference/mean": 0.005328204948455095, "step": 122, "step_time": 254.4314533189754 }, { "epoch": 0.6559139784946236, "grad_norm": 0.004211827646940947, "learning_rate": 1e-05, "loss": -0.0053047481924295425, "step": 122 }, { "clip_ratio": 0.001567298888573987, "completions/clipped_ratio": 0.091796875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4045.0, "completions/mean_length": 1322.505859375, "completions/mean_terminated_length": 1050.9483642578125, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.6612903225806451, "frac_reward_zero_std": 0.1875, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 96620179.0, "register/okay": 0.875, "register/okay_any": 0.96875, "register/premium": 0.4642857142857143, "register/verify_any": 1.0, "reward": 0.81640625, "reward_std": 0.3875311613082886, "rewards/correct/mean": 0.81640625, "rewards/correct/std": 0.3875311613082886, "rollout/chars": 4045.625, "rollout/commit": 1.0, "rollout/reward": 0.90625, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.723147392272949, "sampling/importance_sampling_ratio/mean": 0.9296847581863403, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.0484890937805176, "sampling/sampling_logp_difference/mean": 0.005254649557173252, "step": 123, "step_time": 174.38490604690742 }, { "epoch": 0.6612903225806451, "grad_norm": 0.004190710838884115, "learning_rate": 1e-05, "loss": -0.006984353065490723, "step": 123 }, { "clip_ratio": 0.005397408350745536, "completions/clipped_ratio": 0.076171875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4061.0, "completions/mean_length": 1509.341796875, "completions/mean_terminated_length": 1296.0655517578125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.6666666666666666, "frac_reward_zero_std": 0.125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 97442290.0, "register/okay": 0.6875, "register/okay_any": 0.9375, "register/premium": 0.25454545454545463, "register/verify_any": 0.9375, "reward": 0.78515625, "reward_std": 0.4111155867576599, "rewards/correct/mean": 0.78515625, "rewards/correct/std": 0.4111155867576599, "rollout/chars": 2709.9375, "rollout/commit": 0.96875, "rollout/reward": 0.875, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.929424524307251, "sampling/importance_sampling_ratio/mean": 0.9954087734222412, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.8879876136779785, "sampling/sampling_logp_difference/mean": 0.005107347387820482, "step": 124, "step_time": 149.38224557612557 }, { "epoch": 0.6666666666666666, "grad_norm": 0.004237816669046879, "learning_rate": 1e-05, "loss": -0.018431853502988815, "step": 124 }, { "clip_ratio": 0.0014971091863458241, "completions/clipped_ratio": 0.080078125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4008.0, "completions/mean_length": 1305.845703125, "completions/mean_terminated_length": 1062.966064453125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.6720430107526881, "frac_reward_zero_std": 0.15625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 98166851.0, "register/okay": 0.71875, "register/okay_any": 0.8125, "register/premium": 0.18357487922705307, "register/verify_any": 0.90625, "reward": 0.8046875, "reward_std": 0.3968288004398346, "rewards/correct/mean": 0.8046875, "rewards/correct/std": 0.3968288004398346, "rollout/chars": 7007.03125, "rollout/commit": 0.8125, "rollout/reward": 0.6875, "rollout/trunc": 0.1875, "sampling/importance_sampling_ratio/max": 2.9069948196411133, "sampling/importance_sampling_ratio/mean": 0.964645504951477, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.852027416229248, "sampling/sampling_logp_difference/mean": 0.005209826864302158, "step": 125, "step_time": 145.16907722194446 }, { "epoch": 0.6720430107526881, "grad_norm": 0.003756337333470583, "learning_rate": 1e-05, "loss": -0.04658762738108635, "step": 125 }, { "clip_ratio": 0.002894880046369508, "completions/clipped_ratio": 0.0546875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4094.0, "completions/mean_length": 1279.169921875, "completions/mean_terminated_length": 1124.656982421875, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.6774193548387096, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 98870666.0, "register/okay": 0.875, "register/okay_any": 0.9375, "register/premium": 0.4285714285714286, "register/verify_any": 0.9375, "reward": 0.8125, "reward_std": 0.39069411158561707, "rewards/correct/mean": 0.8125, "rewards/correct/std": 0.39069411158561707, "rollout/chars": 2280.21875, "rollout/commit": 0.9375, "rollout/reward": 0.875, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.957799196243286, "sampling/importance_sampling_ratio/mean": 0.9181804656982422, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.884596824645996, "sampling/sampling_logp_difference/mean": 0.005212935619056225, "step": 126, "step_time": 138.1352167102159 }, { "epoch": 0.6774193548387096, "grad_norm": 0.003459247527644038, "learning_rate": 1e-05, "loss": -0.015211652964353561, "step": 126 }, { "clip_ratio": 0.004568072673464485, "completions/clipped_ratio": 0.08984375, "completions/max_length": 4096.0, "completions/max_terminated_length": 3947.0, "completions/mean_length": 1523.05859375, "completions/mean_terminated_length": 1269.0772705078125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.6827956989247311, "frac_reward_zero_std": 0.15625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 99708264.0, "register/okay": 0.625, "register/okay_any": 0.78125, "register/premium": 0.3333333333333333, "register/verify_any": 0.8125, "reward": 0.76171875, "reward_std": 0.42644867300987244, "rewards/correct/mean": 0.76171875, "rewards/correct/std": 0.42644867300987244, "rollout/chars": 3477.6875, "rollout/commit": 0.875, "rollout/reward": 0.625, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.9986112117767334, "sampling/importance_sampling_ratio/mean": 0.8933644890785217, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.661247730255127, "sampling/sampling_logp_difference/mean": 0.005158610641956329, "step": 127, "step_time": 158.240861376049 }, { "epoch": 0.6827956989247311, "grad_norm": 0.004215425346046686, "learning_rate": 1e-05, "loss": -0.007604209706187248, "step": 127 }, { "clip_ratio": 0.00033087212602822547, "completions/clipped_ratio": 0.09765625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4078.0, "completions/mean_length": 1635.673828125, "completions/mean_terminated_length": 1369.40478515625, "completions/min_length": 10.0, "completions/min_terminated_length": 10.0, "epoch": 0.6881720430107527, "frac_reward_zero_std": 0.28125, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 100619681.0, "register/okay": 0.84375, "register/okay_any": 1.0, "register/premium": 0.12592592592592589, "register/verify_any": 0.9375, "reward": 0.779296875, "reward_std": 0.4151262938976288, "rewards/correct/mean": 0.779296875, "rewards/correct/std": 0.4151262938976288, "rollout/chars": 4446.75, "rollout/commit": 1.0, "rollout/reward": 0.90625, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.6978135108947754, "sampling/importance_sampling_ratio/mean": 0.9727709293365479, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.7191987037658691, "sampling/sampling_logp_difference/mean": 0.0051291510462760925, "step": 128, "step_time": 161.86852610809729 }, { "epoch": 0.6881720430107527, "grad_norm": 0.0036358756478875875, "learning_rate": 1e-05, "loss": -0.003682537470012903, "step": 128 }, { "clip_ratio": 0.0012130105602068397, "completions/clipped_ratio": 0.0625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4028.0, "completions/mean_length": 1310.6875, "completions/mean_terminated_length": 1125.0, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.6935483870967742, "frac_reward_zero_std": 0.1875, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 101353457.0, "register/okay": 0.6875, "register/okay_any": 0.8125, "register/premium": 0.7, "register/verify_any": 0.8125, "reward": 0.814453125, "reward_std": 0.38912075757980347, "rewards/correct/mean": 0.814453125, "rewards/correct/std": 0.38912075757980347, "rollout/chars": 4165.90625, "rollout/commit": 0.90625, "rollout/reward": 0.78125, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.7910377979278564, "sampling/importance_sampling_ratio/mean": 0.9697359800338745, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.0373849868774414, "sampling/sampling_logp_difference/mean": 0.005320750642567873, "step": 129, "step_time": 145.34099759312812 }, { "epoch": 0.6935483870967742, "grad_norm": 0.003982315305620432, "learning_rate": 1e-05, "loss": -0.006669994909316301, "step": 129 }, { "clip_ratio": 0.0030817392986932646, "completions/clipped_ratio": 0.037109375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4047.0, "completions/mean_length": 1255.916015625, "completions/mean_terminated_length": 1146.46044921875, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.6989247311827957, "frac_reward_zero_std": 0.09375, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 102057366.0, "register/okay": 0.71875, "register/okay_any": 0.875, "register/premium": 0.2898550724637682, "register/verify_any": 0.875, "reward": 0.865234375, "reward_std": 0.3418070077896118, "rewards/correct/mean": 0.865234375, "rewards/correct/std": 0.3418070077896118, "rollout/chars": 1949.3125, "rollout/commit": 0.90625, "rollout/reward": 0.875, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.821599006652832, "sampling/importance_sampling_ratio/mean": 0.9962527751922607, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.7566192150115967, "sampling/sampling_logp_difference/mean": 0.005252516828477383, "step": 130, "step_time": 139.68812240491388 }, { "epoch": 0.6989247311827957, "grad_norm": 0.003503515152260661, "learning_rate": 1e-05, "loss": -0.023015204817056656, "step": 130 }, { "clip_ratio": 0.0009853643232986542, "completions/clipped_ratio": 0.103515625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4018.0, "completions/mean_length": 1503.408203125, "completions/mean_terminated_length": 1204.0457763671875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.7043010752688172, "frac_reward_zero_std": 0.15625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 102882279.0, "register/okay": 0.75, "register/okay_any": 0.84375, "register/premium": 0.5, "register/verify_any": 0.8125, "reward": 0.7890625, "reward_std": 0.4083731174468994, "rewards/correct/mean": 0.7890625, "rewards/correct/std": 0.4083731174468994, "rollout/chars": 3213.8125, "rollout/commit": 0.96875, "rollout/reward": 0.875, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.973461151123047, "sampling/importance_sampling_ratio/mean": 0.9383447170257568, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.067605495452881, "sampling/sampling_logp_difference/mean": 0.00515537615865469, "step": 131, "step_time": 225.02693718491355 }, { "epoch": 0.7043010752688172, "grad_norm": 0.004043659660965204, "learning_rate": 1e-05, "loss": -0.03220412880182266, "step": 131 }, { "clip_ratio": 0.0020985204889711895, "completions/clipped_ratio": 0.068359375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4044.0, "completions/mean_length": 1235.525390625, "completions/mean_terminated_length": 1025.6373291015625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.7096774193548387, "frac_reward_zero_std": 0.125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 103570084.0, "register/okay": 0.75, "register/okay_any": 0.90625, "register/premium": 0.41666666666666663, "register/verify_any": 0.9375, "reward": 0.818359375, "reward_std": 0.38592514395713806, "rewards/correct/mean": 0.818359375, "rewards/correct/std": 0.38592514395713806, "rollout/chars": 3408.875, "rollout/commit": 0.9375, "rollout/reward": 0.8125, "rollout/trunc": 0.0625, "sampling/importance_sampling_ratio/max": 2.8096940517425537, "sampling/importance_sampling_ratio/mean": 0.9707400798797607, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.065889358520508, "sampling/sampling_logp_difference/mean": 0.005303150042891502, "step": 132, "step_time": 210.1624309498584 }, { "epoch": 0.7096774193548387, "grad_norm": 0.0035728602670133114, "learning_rate": 1e-05, "loss": -0.01307845488190651, "step": 132 }, { "clip_ratio": 0.0020416387906152522, "completions/clipped_ratio": 0.080078125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4063.0, "completions/mean_length": 1461.8515625, "completions/mean_terminated_length": 1232.5521240234375, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.7150537634408602, "frac_reward_zero_std": 0.125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 104377672.0, "register/okay": 0.78125, "register/okay_any": 0.84375, "register/premium": 0.6571428571428573, "register/verify_any": 0.84375, "reward": 0.78515625, "reward_std": 0.4111155867576599, "rewards/correct/mean": 0.78515625, "rewards/correct/std": 0.4111155867576599, "rollout/chars": 5205.34375, "rollout/commit": 0.875, "rollout/reward": 0.65625, "rollout/trunc": 0.09375, "sampling/importance_sampling_ratio/max": 2.9373812675476074, "sampling/importance_sampling_ratio/mean": 0.9162903428077698, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.862320899963379, "sampling/sampling_logp_difference/mean": 0.005375402979552746, "step": 133, "step_time": 193.21614323207177 }, { "epoch": 0.7150537634408602, "grad_norm": 0.004236122127622366, "learning_rate": 1e-05, "loss": -0.0358210951089859, "step": 133 }, { "clip_ratio": 0.0012835730135520862, "completions/clipped_ratio": 0.0703125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4020.0, "completions/mean_length": 1467.419921875, "completions/mean_terminated_length": 1268.619873046875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.7204301075268817, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 105188031.0, "register/okay": 0.90625, "register/okay_any": 0.96875, "register/premium": 0.35632183908045983, "register/verify_any": 0.96875, "reward": 0.82421875, "reward_std": 0.3810062110424042, "rewards/correct/mean": 0.82421875, "rewards/correct/std": 0.3810062110424042, "rollout/chars": 7117.9375, "rollout/commit": 0.71875, "rollout/reward": 0.65625, "rollout/trunc": 0.3125, "sampling/importance_sampling_ratio/max": 2.9422264099121094, "sampling/importance_sampling_ratio/mean": 0.9780012965202332, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.8682868480682373, "sampling/sampling_logp_difference/mean": 0.005046343430876732, "step": 134, "step_time": 153.6082283431315 }, { "epoch": 0.7204301075268817, "grad_norm": 0.0037022875621914864, "learning_rate": 1e-05, "loss": -0.041721973568201065, "step": 134 }, { "clip_ratio": 0.0008443065696610574, "completions/clipped_ratio": 0.09375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4071.0, "completions/mean_length": 1528.900390625, "completions/mean_terminated_length": 1263.33837890625, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.7258064516129032, "frac_reward_zero_std": 0.125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 106030268.0, "register/okay": 0.75, "register/okay_any": 0.8125, "register/premium": 0.33333333333333337, "register/verify_any": 0.78125, "reward": 0.779296875, "reward_std": 0.4151262938976288, "rewards/correct/mean": 0.779296875, "rewards/correct/std": 0.4151262938976288, "rollout/chars": 2988.84375, "rollout/commit": 0.9375, "rollout/reward": 0.875, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.916624069213867, "sampling/importance_sampling_ratio/mean": 0.9376813173294067, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.8596224784851074, "sampling/sampling_logp_difference/mean": 0.005318491719663143, "step": 135, "step_time": 173.8136348160333 }, { "epoch": 0.7258064516129032, "grad_norm": 0.003522350685670972, "learning_rate": 1e-05, "loss": -0.03370062634348869, "step": 135 }, { "clip_ratio": 0.0018292587003543304, "completions/clipped_ratio": 0.0546875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4057.0, "completions/mean_length": 1464.861328125, "completions/mean_terminated_length": 1313.4813232421875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.7311827956989247, "frac_reward_zero_std": 0.03125, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 106845461.0, "register/okay": 0.78125, "register/okay_any": 0.96875, "register/premium": 0.10285714285714287, "register/verify_any": 0.96875, "reward": 0.830078125, "reward_std": 0.3759314715862274, "rewards/correct/mean": 0.830078125, "rewards/correct/std": 0.3759314715862274, "rollout/chars": 4768.40625, "rollout/commit": 1.0, "rollout/reward": 0.9375, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.8801794052124023, "sampling/importance_sampling_ratio/mean": 0.9635810852050781, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.9282703399658203, "sampling/sampling_logp_difference/mean": 0.0053610303439199924, "step": 136, "step_time": 177.47986487013986 }, { "epoch": 0.7311827956989247, "grad_norm": 0.004176784306764603, "learning_rate": 1e-05, "loss": -0.030467376112937927, "step": 136 }, { "clip_ratio": 0.0003524359503899177, "completions/clipped_ratio": 0.1328125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4094.0, "completions/mean_length": 1746.12109375, "completions/mean_terminated_length": 1388.4459228515625, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.7365591397849462, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 107811939.0, "register/okay": 0.90625, "register/okay_any": 0.9375, "register/premium": -0.04597701149425282, "register/verify_any": 0.9375, "reward": 0.759765625, "reward_std": 0.4276435375213623, "rewards/correct/mean": 0.759765625, "rewards/correct/std": 0.4276435375213623, "rollout/chars": 6492.1875, "rollout/commit": 0.875, "rollout/reward": 0.625, "rollout/trunc": 0.15625, "sampling/importance_sampling_ratio/max": 2.8439531326293945, "sampling/importance_sampling_ratio/mean": 0.9758110642433167, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.973917007446289, "sampling/sampling_logp_difference/mean": 0.00526106171309948, "step": 137, "step_time": 188.44013428402832 }, { "epoch": 0.7365591397849462, "grad_norm": 0.004476306028664112, "learning_rate": 1e-05, "loss": 0.02768535539507866, "step": 137 }, { "clip_ratio": 0.0006067518021382057, "completions/clipped_ratio": 0.07421875, "completions/max_length": 4096.0, "completions/max_terminated_length": 3990.0, "completions/mean_length": 1413.01171875, "completions/mean_terminated_length": 1197.9197998046875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.7419354838709677, "frac_reward_zero_std": 0.15625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 108590953.0, "register/okay": 0.875, "register/okay_any": 0.96875, "register/premium": -0.1071428571428571, "register/verify_any": 0.96875, "reward": 0.8125, "reward_std": 0.39069411158561707, "rewards/correct/mean": 0.8125, "rewards/correct/std": 0.39069411158561707, "rollout/chars": 3269.875, "rollout/commit": 1.0, "rollout/reward": 0.90625, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.9473440647125244, "sampling/importance_sampling_ratio/mean": 1.0286420583724976, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.9075703620910645, "sampling/sampling_logp_difference/mean": 0.005084763281047344, "step": 138, "step_time": 143.81512518797535 }, { "epoch": 0.7419354838709677, "grad_norm": 0.005723567679524422, "learning_rate": 1e-05, "loss": 0.003882540389895439, "step": 138 }, { "clip_ratio": 0.0006320240884178929, "completions/clipped_ratio": 0.123046875, "completions/max_length": 4096.0, "completions/max_terminated_length": 3997.0, "completions/mean_length": 1649.34375, "completions/mean_terminated_length": 1306.049072265625, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.7473118279569892, "frac_reward_zero_std": 0.09375, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 109493673.0, "register/okay": 0.75, "register/okay_any": 0.875, "register/premium": 0.29166666666666663, "register/verify_any": 0.875, "reward": 0.794921875, "reward_std": 0.4041535556316376, "rewards/correct/mean": 0.794921875, "rewards/correct/std": 0.4041535556316376, "rollout/chars": 2752.75, "rollout/commit": 0.96875, "rollout/reward": 0.84375, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.9239203929901123, "sampling/importance_sampling_ratio/mean": 0.9388471245765686, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.9743530750274658, "sampling/sampling_logp_difference/mean": 0.005093180574476719, "step": 139, "step_time": 196.84098436305067 }, { "epoch": 0.7473118279569892, "grad_norm": 0.003386896103620529, "learning_rate": 1e-05, "loss": -0.020759442821145058, "step": 139 }, { "clip_ratio": 0.0004905687171685713, "completions/clipped_ratio": 0.044921875, "completions/max_length": 4096.0, "completions/max_terminated_length": 3959.0, "completions/mean_length": 1178.193359375, "completions/mean_terminated_length": 1041.6011962890625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.7526881720430108, "frac_reward_zero_std": 0.1875, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 110166060.0, "register/okay": 0.875, "register/okay_any": 0.96875, "register/premium": 0.4642857142857143, "register/verify_any": 0.96875, "reward": 0.822265625, "reward_std": 0.3826628625392914, "rewards/correct/mean": 0.822265625, "rewards/correct/std": 0.3826628625392914, "rollout/chars": 3430.15625, "rollout/commit": 0.96875, "rollout/reward": 0.90625, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.856393575668335, "sampling/importance_sampling_ratio/mean": 0.9868543148040771, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.82421875, "sampling/sampling_logp_difference/mean": 0.00560029549524188, "step": 140, "step_time": 161.10334520693868 }, { "epoch": 0.7526881720430108, "grad_norm": 0.0033938083797693253, "learning_rate": 1e-05, "loss": -0.04584335908293724, "step": 140 }, { "clip_ratio": 0.00023589162890402804, "completions/clipped_ratio": 0.0859375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4088.0, "completions/mean_length": 1509.681640625, "completions/mean_terminated_length": 1266.5235595703125, "completions/min_length": 19.0, "completions/min_terminated_length": 19.0, "epoch": 0.7580645161290323, "frac_reward_zero_std": 0.1875, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 110993097.0, "register/okay": 0.96875, "register/okay_any": 0.96875, "register/premium": -0.032258064516129004, "register/verify_any": 1.0, "reward": 0.798828125, "reward_std": 0.4012683033943176, "rewards/correct/mean": 0.798828125, "rewards/correct/std": 0.4012683033943176, "rollout/chars": 4073.53125, "rollout/commit": 1.0, "rollout/reward": 0.96875, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.6428966522216797, "sampling/importance_sampling_ratio/mean": 0.9553660750389099, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.6986770629882812, "sampling/sampling_logp_difference/mean": 0.004981483798474073, "step": 141, "step_time": 135.95673296507448 }, { "epoch": 0.7580645161290323, "grad_norm": 0.0035385265946388245, "learning_rate": 1e-05, "loss": 0.027638427913188934, "step": 141 }, { "clip_ratio": 0.003148552472907795, "completions/clipped_ratio": 0.064453125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4084.0, "completions/mean_length": 1454.826171875, "completions/mean_terminated_length": 1272.866455078125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.7634408602150538, "frac_reward_zero_std": 0.09375, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 111805232.0, "register/okay": 0.8125, "register/okay_any": 0.90625, "register/premium": 0.34615384615384615, "register/verify_any": 0.9375, "reward": 0.810546875, "reward_std": 0.3922513723373413, "rewards/correct/mean": 0.810546875, "rewards/correct/std": 0.3922513723373413, "rollout/chars": 4182.21875, "rollout/commit": 0.90625, "rollout/reward": 0.78125, "rollout/trunc": 0.0625, "sampling/importance_sampling_ratio/max": 2.9433541297912598, "sampling/importance_sampling_ratio/mean": 0.9924004077911377, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.0392608642578125, "sampling/sampling_logp_difference/mean": 0.005207089241594076, "step": 142, "step_time": 152.6588836440351 }, { "epoch": 0.7634408602150538, "grad_norm": 0.004539084620773792, "learning_rate": 1e-05, "loss": -0.018689453601837158, "step": 142 }, { "clip_ratio": 0.0005103872063045856, "completions/clipped_ratio": 0.056640625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4083.0, "completions/mean_length": 1295.373046875, "completions/mean_terminated_length": 1127.219482421875, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.7688172043010753, "frac_reward_zero_std": 0.21875, "groups/all_right": 0.0, "groups/all_wrong": 0.5, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 112525967.0, "register/okay": 0.90625, "register/okay_any": 1.0, "register/premium": 0.14942528735632188, "register/verify_any": 1.0, "reward": 0.80859375, "reward_std": 0.3937928080558777, "rewards/correct/mean": 0.80859375, "rewards/correct/std": 0.3937928080558777, "rollout/chars": 6926.375, "rollout/commit": 0.46875, "rollout/reward": 0.46875, "rollout/trunc": 0.5, "sampling/importance_sampling_ratio/max": 2.9270646572113037, "sampling/importance_sampling_ratio/mean": 0.9750213623046875, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.8899827003479004, "sampling/sampling_logp_difference/mean": 0.005004197359085083, "step": 143, "step_time": 149.1724165190244 }, { "epoch": 0.7688172043010753, "grad_norm": 0.0031139806378632784, "learning_rate": 1e-05, "loss": -0.04227989912033081, "step": 143 }, { "clip_ratio": 0.0006977089772135514, "completions/clipped_ratio": 0.076171875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4042.0, "completions/mean_length": 1336.3984375, "completions/mean_terminated_length": 1108.862548828125, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.7741935483870968, "frac_reward_zero_std": 0.15625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 113268907.0, "register/okay": 0.875, "register/okay_any": 0.96875, "register/premium": 0.1785714285714286, "register/verify_any": 0.96875, "reward": 0.80078125, "reward_std": 0.39980348944664, "rewards/correct/mean": 0.80078125, "rewards/correct/std": 0.39980348944664, "rollout/chars": 3024.8125, "rollout/commit": 0.96875, "rollout/reward": 0.90625, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.9194934368133545, "sampling/importance_sampling_ratio/mean": 0.9629621505737305, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.100451946258545, "sampling/sampling_logp_difference/mean": 0.005598437041044235, "step": 144, "step_time": 153.35740845493274 }, { "epoch": 0.7741935483870968, "grad_norm": 0.004070339724421501, "learning_rate": 1e-05, "loss": -0.009588368237018585, "step": 144 }, { "clip_ratio": 0.000484032189433492, "completions/clipped_ratio": 0.1015625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4015.0, "completions/mean_length": 1531.9375, "completions/mean_terminated_length": 1242.0869140625, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.7795698924731183, "frac_reward_zero_std": 0.1875, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 114105259.0, "register/okay": 0.8125, "register/okay_any": 0.875, "register/premium": 0.2820512820512821, "register/verify_any": 0.84375, "reward": 0.806640625, "reward_std": 0.39531853795051575, "rewards/correct/mean": 0.806640625, "rewards/correct/std": 0.39531853795051575, "rollout/chars": 6307.0625, "rollout/commit": 0.65625, "rollout/reward": 0.5625, "rollout/trunc": 0.34375, "sampling/importance_sampling_ratio/max": 2.7866883277893066, "sampling/importance_sampling_ratio/mean": 0.9515845775604248, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.9126052856445312, "sampling/sampling_logp_difference/mean": 0.0052135055884718895, "step": 145, "step_time": 196.57410402304959 }, { "epoch": 0.7795698924731183, "grad_norm": 0.004036587662994862, "learning_rate": 1e-05, "loss": -0.02931736409664154, "step": 145 }, { "clip_ratio": 0.0006457828447992142, "completions/clipped_ratio": 0.0859375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4054.0, "completions/mean_length": 1444.646484375, "completions/mean_terminated_length": 1195.3740234375, "completions/min_length": 5.0, "completions/min_terminated_length": 5.0, "epoch": 0.7849462365591398, "frac_reward_zero_std": 0.25, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 114914086.0, "register/okay": 0.90625, "register/okay_any": 0.9375, "register/premium": 0.896551724137931, "register/verify_any": 0.90625, "reward": 0.822265625, "reward_std": 0.3826628625392914, "rewards/correct/mean": 0.822265625, "rewards/correct/std": 0.3826628625392914, "rollout/chars": 4707.0625, "rollout/commit": 0.90625, "rollout/reward": 0.8125, "rollout/trunc": 0.0625, "sampling/importance_sampling_ratio/max": 2.9893362522125244, "sampling/importance_sampling_ratio/mean": 0.9981089234352112, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.9266424179077148, "sampling/sampling_logp_difference/mean": 0.005181650165468454, "step": 146, "step_time": 161.48640546388924 }, { "epoch": 0.7849462365591398, "grad_norm": 0.004142721649259329, "learning_rate": 1e-05, "loss": 0.0031005889177322388, "step": 146 }, { "clip_ratio": 0.0007522905600580998, "completions/clipped_ratio": 0.0703125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4079.0, "completions/mean_length": 1282.04296875, "completions/mean_terminated_length": 1069.2227783203125, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.7903225806451613, "frac_reward_zero_std": 0.1875, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 115624588.0, "register/okay": 0.65625, "register/okay_any": 0.8125, "register/premium": 0.18181818181818177, "register/verify_any": 0.84375, "reward": 0.826171875, "reward_std": 0.3793322443962097, "rewards/correct/mean": 0.826171875, "rewards/correct/std": 0.3793322443962097, "rollout/chars": 2295.53125, "rollout/commit": 0.96875, "rollout/reward": 0.9375, "rollout/trunc": 0.0, "sampling/importance_sampling_ratio/max": 2.8664562702178955, "sampling/importance_sampling_ratio/mean": 0.9619696736335754, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.9203853607177734, "sampling/sampling_logp_difference/mean": 0.005349650979042053, "step": 147, "step_time": 151.0539762271219 }, { "epoch": 0.7903225806451613, "grad_norm": 0.004408734850585461, "learning_rate": 1e-05, "loss": -0.019962511956691742, "step": 147 }, { "clip_ratio": 0.0004631544966287038, "completions/clipped_ratio": 0.115234375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4023.0, "completions/mean_length": 1804.40234375, "completions/mean_terminated_length": 1510.7725830078125, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "epoch": 0.7956989247311828, "frac_reward_zero_std": 0.0625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 116610698.0, "register/okay": 0.78125, "register/okay_any": 0.90625, "register/premium": 0.24571428571428566, "register/verify_any": 0.90625, "reward": 0.775390625, "reward_std": 0.41773295402526855, "rewards/correct/mean": 0.775390625, "rewards/correct/std": 0.41773295402526855, "rollout/chars": 4067.625, "rollout/commit": 0.96875, "rollout/reward": 0.90625, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.9090511798858643, "sampling/importance_sampling_ratio/mean": 0.954256534576416, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.163297176361084, "sampling/sampling_logp_difference/mean": 0.005101749673485756, "step": 148, "step_time": 180.33780744014075 }, { "epoch": 0.7956989247311828, "grad_norm": 0.004125646315515041, "learning_rate": 1e-05, "loss": -0.00448123924434185, "step": 148 }, { "clip_ratio": 0.0012813573852668014, "completions/clipped_ratio": 0.083984375, "completions/max_length": 4096.0, "completions/max_terminated_length": 3976.0, "completions/mean_length": 1568.501953125, "completions/mean_terminated_length": 1336.769775390625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.8010752688172043, "frac_reward_zero_std": 0.15625, "groups/all_right": 0.0, "groups/all_wrong": 0.0, "groups/live_frac": 1.0, "groups/n": 2, "num_tokens": 117465963.0, "register/okay": 0.90625, "register/okay_any": 0.90625, "register/premium": 0.42528735632183906, "register/verify_any": 0.875, "reward": 0.83203125, "reward_std": 0.374204158782959, "rewards/correct/mean": 0.83203125, "rewards/correct/std": 0.374204158782959, "rollout/chars": 7244.9375, "rollout/commit": 0.8125, "rollout/reward": 0.71875, "rollout/trunc": 0.21875, "sampling/importance_sampling_ratio/max": 2.9920871257781982, "sampling/importance_sampling_ratio/mean": 0.9578415155410767, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 2.406337261199951, "sampling/sampling_logp_difference/mean": 0.005244790576398373, "step": 149, "step_time": 185.52631116600242 }, { "epoch": 0.8010752688172043, "grad_norm": 0.004305888433009386, "learning_rate": 1e-05, "loss": 0.025283318012952805, "step": 149 }, { "clip_ratio": 0.0001835316579672508, "completions/clipped_ratio": 0.044921875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4052.0, "completions/mean_length": 1302.625, "completions/mean_terminated_length": 1171.2392578125, "completions/min_length": 9.0, "completions/min_terminated_length": 9.0, "epoch": 0.8064516129032258, "frac_reward_zero_std": 0.15625, "groups/all_right": 0.5, "groups/all_wrong": 0.0, "groups/live_frac": 0.5, "groups/n": 2, "num_tokens": 118183083.0, "register/okay": 0.96875, "register/okay_any": 1.0, "register/premium": -0.032258064516129004, "register/verify_any": 1.0, "reward": 0.857421875, "reward_std": 0.3499840497970581, "rewards/correct/mean": 0.857421875, "rewards/correct/std": 0.3499840497970581, "rollout/chars": 2721.09375, "rollout/commit": 0.96875, "rollout/reward": 0.96875, "rollout/trunc": 0.03125, "sampling/importance_sampling_ratio/max": 2.9949357509613037, "sampling/importance_sampling_ratio/mean": 1.0004043579101562, "sampling/importance_sampling_ratio/min": 0.0, "sampling/sampling_logp_difference/max": 1.9084324836730957, "sampling/sampling_logp_difference/mean": 0.004931967705488205, "step": 150, "step_time": 136.8910830579698 }, { "epoch": 0.8064516129032258, "grad_norm": 0.0036076863761991262, "learning_rate": 1e-05, "loss": -0.031052041798830032, "step": 150 } ], "logging_steps": 1, "max_steps": 300, "num_input_tokens_seen": 118183083, "num_train_epochs": 2, "save_steps": 25, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 2, "trial_name": null, "trial_params": null }