| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.00068, |
| "eval_steps": 500, |
| "global_step": 68, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 309.0, |
| "completions/max_terminated_length": 309.0, |
| "completions/mean_length": 160.90625, |
| "completions/mean_terminated_length": 160.90625, |
| "completions/min_length": 6.0, |
| "completions/min_terminated_length": 6.0, |
| "entropy": 8.38103711605072, |
| "epoch": 1e-05, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.0028616045601665974, |
| "kl": 0.0, |
| "learning_rate": 0.0, |
| "loss": -0.0002, |
| "num_tokens": 18085.0, |
| "reward": 0.03125002980232239, |
| "reward_std": 0.6862481236457825, |
| "rewards/rollout_reward_func/mean": 0.03125002980232239, |
| "rewards/rollout_reward_func/std": 1.011366844177246, |
| "sampling/importance_sampling_ratio/max": 0.010557955130934715, |
| "sampling/importance_sampling_ratio/mean": 0.003285687882453203, |
| "sampling/importance_sampling_ratio/min": 7.365059625542847e-13, |
| "sampling/sampling_logp_difference/max": 10.706700325012207, |
| "sampling/sampling_logp_difference/mean": 1.4828054904937744, |
| "step": 1, |
| "step_time": 5.417499241000769 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.38103711605072, |
| "epoch": 2e-05, |
| "grad_norm": 0.002821348374709487, |
| "kl": 0.0, |
| "learning_rate": 2.8571428571428575e-07, |
| "loss": -0.0002, |
| "step": 2, |
| "step_time": 2.1234856260016386 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 297.0, |
| "completions/max_terminated_length": 297.0, |
| "completions/mean_length": 80.875, |
| "completions/mean_terminated_length": 80.875, |
| "completions/min_length": 2.0, |
| "completions/min_terminated_length": 2.0, |
| "entropy": 7.941936016082764, |
| "epoch": 3e-05, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.041607797145843506, |
| "kl": 0.0004363941061455989, |
| "learning_rate": 5.714285714285715e-07, |
| "loss": 0.0023, |
| "num_tokens": 33737.0, |
| "reward": -0.13124999403953552, |
| "reward_std": 0.7598094344139099, |
| "rewards/rollout_reward_func/mean": -0.13124999403953552, |
| "rewards/rollout_reward_func/std": 1.0014303922653198, |
| "sampling/importance_sampling_ratio/max": 0.11680073291063309, |
| "sampling/importance_sampling_ratio/mean": 0.033671747893095016, |
| "sampling/importance_sampling_ratio/min": 1.0370337122367346e-06, |
| "sampling/sampling_logp_difference/max": 4.6255574226379395, |
| "sampling/sampling_logp_difference/mean": 1.339035987854004, |
| "step": 3, |
| "step_time": 3.9910208220007917 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 7.934386253356934, |
| "epoch": 4e-05, |
| "grad_norm": 0.04307129234075546, |
| "kl": 0.0004773353211930953, |
| "learning_rate": 8.571428571428572e-07, |
| "loss": 0.0023, |
| "step": 4, |
| "step_time": 2.6708831029991416 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 297.0, |
| "completions/max_terminated_length": 297.0, |
| "completions/mean_length": 129.0625, |
| "completions/mean_terminated_length": 129.0625, |
| "completions/min_length": 2.0, |
| "completions/min_terminated_length": 2.0, |
| "entropy": 8.268575668334961, |
| "epoch": 5e-05, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.03974386304616928, |
| "kl": 0.0007551452181360219, |
| "learning_rate": 1.142857142857143e-06, |
| "loss": 0.0022, |
| "num_tokens": 51251.0, |
| "reward": -0.53125, |
| "reward_std": 0.5484436750411987, |
| "rewards/rollout_reward_func/mean": -0.53125, |
| "rewards/rollout_reward_func/std": 0.8789427280426025, |
| "sampling/importance_sampling_ratio/max": 0.11807744950056076, |
| "sampling/importance_sampling_ratio/mean": 0.024541109800338745, |
| "sampling/importance_sampling_ratio/min": 1.4743216955430405e-13, |
| "sampling/sampling_logp_difference/max": 10.85576057434082, |
| "sampling/sampling_logp_difference/mean": 1.503469467163086, |
| "step": 5, |
| "step_time": 4.814693420999902 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.273618221282959, |
| "epoch": 6e-05, |
| "grad_norm": 0.04039419814944267, |
| "kl": 0.0003924804532289272, |
| "learning_rate": 1.4285714285714286e-06, |
| "loss": 0.0022, |
| "step": 6, |
| "step_time": 2.0722021359997598 |
| }, |
| { |
| "clip_ratio/high_max": 0.010416666977107525, |
| "clip_ratio/high_mean": 0.0052083334885537624, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0052083334885537624, |
| "completions/clipped_ratio": 0.03125, |
| "completions/max_length": 369.0, |
| "completions/max_terminated_length": 369.0, |
| "completions/mean_length": 159.3125, |
| "completions/mean_terminated_length": 157.77418518066406, |
| "completions/min_length": 2.0, |
| "completions/min_terminated_length": 2.0, |
| "entropy": 8.400971293449402, |
| "epoch": 7e-05, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.02198818139731884, |
| "kl": 0.0007719468412688002, |
| "learning_rate": 1.7142857142857145e-06, |
| "loss": 0.0013, |
| "num_tokens": 69453.0, |
| "reward": -0.6624999642372131, |
| "reward_std": 0.48922622203826904, |
| "rewards/rollout_reward_func/mean": -0.6624999642372131, |
| "rewards/rollout_reward_func/std": 0.8071255087852478, |
| "sampling/importance_sampling_ratio/max": 0.10809381306171417, |
| "sampling/importance_sampling_ratio/mean": 0.011609362438321114, |
| "sampling/importance_sampling_ratio/min": 1.4364835228825295e-22, |
| "sampling/sampling_logp_difference/max": 11.3631591796875, |
| "sampling/sampling_logp_difference/mean": 1.7411997318267822, |
| "step": 7, |
| "step_time": 4.814280139999937 |
| }, |
| { |
| "clip_ratio/high_max": 0.014583333861082792, |
| "clip_ratio/high_mean": 0.007291666930541396, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.007291666930541396, |
| "entropy": 8.400816917419434, |
| "epoch": 8e-05, |
| "grad_norm": 0.021912436932325363, |
| "kl": 0.000808713368314784, |
| "learning_rate": 2.0000000000000003e-06, |
| "loss": 0.0013, |
| "step": 8, |
| "step_time": 2.084449717998723 |
| }, |
| { |
| "clip_ratio/high_max": 0.0052083334885537624, |
| "clip_ratio/high_mean": 0.0026041667442768812, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0026041667442768812, |
| "completions/clipped_ratio": 0.0625, |
| "completions/max_length": 394.0, |
| "completions/max_terminated_length": 394.0, |
| "completions/mean_length": 160.40625, |
| "completions/mean_terminated_length": 156.1666717529297, |
| "completions/min_length": 2.0, |
| "completions/min_terminated_length": 2.0, |
| "entropy": 8.220785737037659, |
| "epoch": 9e-05, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.00841094646602869, |
| "kl": 0.0007819603888492566, |
| "learning_rate": 2.285714285714286e-06, |
| "loss": -0.0006, |
| "num_tokens": 88018.0, |
| "reward": -0.6656249761581421, |
| "reward_std": 0.30112773180007935, |
| "rewards/rollout_reward_func/mean": -0.6656249761581421, |
| "rewards/rollout_reward_func/std": 0.8090652823448181, |
| "sampling/importance_sampling_ratio/max": 0.10453330725431442, |
| "sampling/importance_sampling_ratio/mean": 0.009016389958560467, |
| "sampling/importance_sampling_ratio/min": 3.77039369235492e-14, |
| "sampling/sampling_logp_difference/max": 11.177996635437012, |
| "sampling/sampling_logp_difference/mean": 1.46906578540802, |
| "step": 9, |
| "step_time": 5.767028449001373 |
| }, |
| { |
| "clip_ratio/high_max": 0.010416666977107525, |
| "clip_ratio/high_mean": 0.0078125, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0078125, |
| "entropy": 8.221423089504242, |
| "epoch": 0.0001, |
| "grad_norm": 0.00875311903655529, |
| "kl": 0.0009726146636239719, |
| "learning_rate": 2.571428571428571e-06, |
| "loss": -0.0006, |
| "step": 10, |
| "step_time": 3.8331103400005304 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 476.0, |
| "completions/max_terminated_length": 476.0, |
| "completions/mean_length": 146.875, |
| "completions/mean_terminated_length": 146.875, |
| "completions/min_length": 2.0, |
| "completions/min_terminated_length": 2.0, |
| "entropy": 8.224214434623718, |
| "epoch": 0.00011, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.005308263469487429, |
| "kl": 0.0006370486844389234, |
| "learning_rate": 2.8571428571428573e-06, |
| "loss": -0.0005, |
| "num_tokens": 105118.0, |
| "reward": -0.6531249284744263, |
| "reward_std": 0.27538806200027466, |
| "rewards/rollout_reward_func/mean": -0.6531249284744263, |
| "rewards/rollout_reward_func/std": 0.7935158014297485, |
| "sampling/importance_sampling_ratio/max": 0.09377396106719971, |
| "sampling/importance_sampling_ratio/mean": 0.007022843696177006, |
| "sampling/importance_sampling_ratio/min": 1.1420680792263728e-27, |
| "sampling/sampling_logp_difference/max": 4.9811320304870605, |
| "sampling/sampling_logp_difference/mean": 1.3924936056137085, |
| "step": 11, |
| "step_time": 5.839474645000337 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.226899027824402, |
| "epoch": 0.00012, |
| "grad_norm": 0.005312995053827763, |
| "kl": 0.0006140958357718773, |
| "learning_rate": 3.142857142857143e-06, |
| "loss": -0.0005, |
| "step": 12, |
| "step_time": 2.6753236439999455 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 294.0, |
| "completions/max_terminated_length": 294.0, |
| "completions/mean_length": 188.25, |
| "completions/mean_terminated_length": 188.25, |
| "completions/min_length": 87.0, |
| "completions/min_terminated_length": 87.0, |
| "entropy": 8.331215858459473, |
| "epoch": 0.00013, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.0016523015219718218, |
| "kl": 0.0006200866155268159, |
| "learning_rate": 3.428571428571429e-06, |
| "loss": -0.0001, |
| "num_tokens": 124622.0, |
| "reward": -1.009374976158142, |
| "reward_std": 0.0265165027230978, |
| "rewards/rollout_reward_func/mean": -1.009374976158142, |
| "rewards/rollout_reward_func/std": 0.03901509940624237, |
| "sampling/importance_sampling_ratio/max": 0.0077222432009875774, |
| "sampling/importance_sampling_ratio/mean": 0.0027147922664880753, |
| "sampling/importance_sampling_ratio/min": 3.714021278022894e-14, |
| "sampling/sampling_logp_difference/max": 11.583379745483398, |
| "sampling/sampling_logp_difference/mean": 1.4247815608978271, |
| "step": 13, |
| "step_time": 4.811962196999048 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.332964897155762, |
| "epoch": 0.00014, |
| "grad_norm": 0.0017943360144272447, |
| "kl": 0.0005818934332637582, |
| "learning_rate": 3.7142857142857146e-06, |
| "loss": -0.0001, |
| "step": 14, |
| "step_time": 2.0559995119983796 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 314.0, |
| "completions/max_terminated_length": 314.0, |
| "completions/mean_length": 134.34375, |
| "completions/mean_terminated_length": 134.34375, |
| "completions/min_length": 2.0, |
| "completions/min_terminated_length": 2.0, |
| "entropy": 8.289089798927307, |
| "epoch": 0.00015, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.006275709718465805, |
| "kl": 0.0006801459057896864, |
| "learning_rate": 4.000000000000001e-06, |
| "loss": -0.0006, |
| "num_tokens": 142137.0, |
| "reward": -1.0187499523162842, |
| "reward_std": 0.04082316905260086, |
| "rewards/rollout_reward_func/mean": -1.0187499523162842, |
| "rewards/rollout_reward_func/std": 0.04709291458129883, |
| "sampling/importance_sampling_ratio/max": 0.1032625362277031, |
| "sampling/importance_sampling_ratio/mean": 0.010007976554334164, |
| "sampling/importance_sampling_ratio/min": 1.6740179376029118e-07, |
| "sampling/sampling_logp_difference/max": 4.918362617492676, |
| "sampling/sampling_logp_difference/mean": 1.347560167312622, |
| "step": 15, |
| "step_time": 4.26574305899976 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.295971155166626, |
| "epoch": 0.00016, |
| "grad_norm": 0.006583907175809145, |
| "kl": 0.0010946946458716411, |
| "learning_rate": 4.2857142857142855e-06, |
| "loss": -0.0006, |
| "step": 16, |
| "step_time": 3.0452185289996123 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 342.0, |
| "completions/max_terminated_length": 342.0, |
| "completions/mean_length": 176.59375, |
| "completions/mean_terminated_length": 176.59375, |
| "completions/min_length": 2.0, |
| "completions/min_terminated_length": 2.0, |
| "entropy": 8.386258959770203, |
| "epoch": 0.00017, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.019049938768148422, |
| "kl": 0.0009286534441343974, |
| "learning_rate": 4.571428571428572e-06, |
| "loss": 0.0008, |
| "num_tokens": 161764.0, |
| "reward": 0.34687501192092896, |
| "reward_std": 0.3751429617404938, |
| "rewards/rollout_reward_func/mean": 0.34687501192092896, |
| "rewards/rollout_reward_func/std": 0.9510552287101746, |
| "sampling/importance_sampling_ratio/max": 0.10295701771974564, |
| "sampling/importance_sampling_ratio/mean": 0.00530514121055603, |
| "sampling/importance_sampling_ratio/min": 1.1150266265858022e-09, |
| "sampling/sampling_logp_difference/max": 8.94926643371582, |
| "sampling/sampling_logp_difference/mean": 1.3967115879058838, |
| "step": 17, |
| "step_time": 4.39325438600099 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.376275658607483, |
| "epoch": 0.00018, |
| "grad_norm": 0.017463266849517822, |
| "kl": 0.0013417988302535377, |
| "learning_rate": 4.857142857142858e-06, |
| "loss": 0.0007, |
| "step": 18, |
| "step_time": 2.0553056610006024 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 405.0, |
| "completions/max_terminated_length": 405.0, |
| "completions/mean_length": 163.53125, |
| "completions/mean_terminated_length": 163.53125, |
| "completions/min_length": 62.0, |
| "completions/min_terminated_length": 62.0, |
| "entropy": 8.383512496948242, |
| "epoch": 0.00019, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.0029027618002146482, |
| "kl": 0.0011069331085309386, |
| "learning_rate": 5.142857142857142e-06, |
| "loss": -0.0001, |
| "num_tokens": 180341.0, |
| "reward": -0.840624988079071, |
| "reward_std": 0.2795426845550537, |
| "rewards/rollout_reward_func/mean": -0.840624988079071, |
| "rewards/rollout_reward_func/std": 0.570715069770813, |
| "sampling/importance_sampling_ratio/max": 0.009223885834217072, |
| "sampling/importance_sampling_ratio/mean": 0.002876629587262869, |
| "sampling/importance_sampling_ratio/min": 3.4141774769962514e-21, |
| "sampling/sampling_logp_difference/max": 11.647392272949219, |
| "sampling/sampling_logp_difference/mean": 1.74857759475708, |
| "step": 19, |
| "step_time": 5.028274022000915 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.370469331741333, |
| "epoch": 0.0002, |
| "grad_norm": 0.002040296094492078, |
| "kl": 0.0010489341875654645, |
| "learning_rate": 5.428571428571429e-06, |
| "loss": -0.0001, |
| "step": 20, |
| "step_time": 2.1169578549997823 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 373.0, |
| "completions/max_terminated_length": 373.0, |
| "completions/mean_length": 155.15625, |
| "completions/mean_terminated_length": 155.15625, |
| "completions/min_length": 2.0, |
| "completions/min_terminated_length": 2.0, |
| "entropy": 8.354791045188904, |
| "epoch": 0.00021, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.014582234434783459, |
| "kl": 0.0016839846794027835, |
| "learning_rate": 5.7142857142857145e-06, |
| "loss": 0.0006, |
| "num_tokens": 198770.0, |
| "reward": 0.46875, |
| "reward_std": 0.6740255355834961, |
| "rewards/rollout_reward_func/mean": 0.46875, |
| "rewards/rollout_reward_func/std": 0.9006941318511963, |
| "sampling/importance_sampling_ratio/max": 0.08340345323085785, |
| "sampling/importance_sampling_ratio/mean": 0.006231832783669233, |
| "sampling/importance_sampling_ratio/min": 2.269645449359814e-08, |
| "sampling/sampling_logp_difference/max": 2.1503682136535645, |
| "sampling/sampling_logp_difference/mean": 1.321890115737915, |
| "step": 21, |
| "step_time": 4.95798639899931 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.337073802947998, |
| "epoch": 0.00022, |
| "grad_norm": 0.014086912386119366, |
| "kl": 0.0020258916774764657, |
| "learning_rate": 6e-06, |
| "loss": 0.0006, |
| "step": 22, |
| "step_time": 2.598393530000976 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0020833334419876337, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0020833334419876337, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 417.0, |
| "completions/max_terminated_length": 417.0, |
| "completions/mean_length": 152.59375, |
| "completions/mean_terminated_length": 152.59375, |
| "completions/min_length": 2.0, |
| "completions/min_terminated_length": 2.0, |
| "entropy": 8.294317483901978, |
| "epoch": 0.00023, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.019008534029126167, |
| "kl": 0.004680573721998371, |
| "learning_rate": 6.285714285714286e-06, |
| "loss": 0.0008, |
| "num_tokens": 217253.0, |
| "reward": -0.53125, |
| "reward_std": 0.3699263334274292, |
| "rewards/rollout_reward_func/mean": -0.53125, |
| "rewards/rollout_reward_func/std": 0.8785756230354309, |
| "sampling/importance_sampling_ratio/max": 0.09325665980577469, |
| "sampling/importance_sampling_ratio/mean": 0.009642375633120537, |
| "sampling/importance_sampling_ratio/min": 1.0861621381728576e-20, |
| "sampling/sampling_logp_difference/max": 10.062246322631836, |
| "sampling/sampling_logp_difference/mean": 1.4383571147918701, |
| "step": 23, |
| "step_time": 4.488339367000663 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0020833334419876337, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0020833334419876337, |
| "entropy": 8.280492901802063, |
| "epoch": 0.00024, |
| "grad_norm": 0.01694045588374138, |
| "kl": 0.007116928434697911, |
| "learning_rate": 6.571428571428572e-06, |
| "loss": 0.0007, |
| "step": 24, |
| "step_time": 2.144616393000433 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 262.0, |
| "completions/max_terminated_length": 262.0, |
| "completions/mean_length": 141.1875, |
| "completions/mean_terminated_length": 141.1875, |
| "completions/min_length": 2.0, |
| "completions/min_terminated_length": 2.0, |
| "entropy": 8.207048416137695, |
| "epoch": 0.00025, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.008826951496303082, |
| "kl": 0.011807448376202956, |
| "learning_rate": 6.857142857142858e-06, |
| "loss": -0.0003, |
| "num_tokens": 235259.0, |
| "reward": -0.7625000476837158, |
| "reward_std": 0.2917833626270294, |
| "rewards/rollout_reward_func/mean": -0.7625000476837158, |
| "rewards/rollout_reward_func/std": 0.6776382327079773, |
| "sampling/importance_sampling_ratio/max": 0.06867893040180206, |
| "sampling/importance_sampling_ratio/mean": 0.007195095531642437, |
| "sampling/importance_sampling_ratio/min": 8.830933421045788e-15, |
| "sampling/sampling_logp_difference/max": 4.469038486480713, |
| "sampling/sampling_logp_difference/mean": 1.387784719467163, |
| "step": 25, |
| "step_time": 5.051108140999531 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.201621413230896, |
| "epoch": 0.00026, |
| "grad_norm": 0.009136058390140533, |
| "kl": 0.014795138500630856, |
| "learning_rate": 7.1428571428571436e-06, |
| "loss": -0.0003, |
| "step": 26, |
| "step_time": 2.5399822110020978 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 290.0, |
| "completions/max_terminated_length": 290.0, |
| "completions/mean_length": 136.9375, |
| "completions/mean_terminated_length": 136.9375, |
| "completions/min_length": 87.0, |
| "completions/min_terminated_length": 87.0, |
| "entropy": 8.213225603103638, |
| "epoch": 0.00027, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.002163404831662774, |
| "kl": 0.0031390516232931986, |
| "learning_rate": 7.428571428571429e-06, |
| "loss": -0.0001, |
| "num_tokens": 252577.0, |
| "reward": -0.3937499523162842, |
| "reward_std": 0.2508378326892853, |
| "rewards/rollout_reward_func/mean": -0.3937499523162842, |
| "rewards/rollout_reward_func/std": 0.9557761549949646, |
| "sampling/importance_sampling_ratio/max": 0.010245459154248238, |
| "sampling/importance_sampling_ratio/mean": 0.004281196743249893, |
| "sampling/importance_sampling_ratio/min": 9.54090864979662e-07, |
| "sampling/sampling_logp_difference/max": 3.967635154724121, |
| "sampling/sampling_logp_difference/mean": 1.3100483417510986, |
| "step": 27, |
| "step_time": 6.014637065999523 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.203812718391418, |
| "epoch": 0.00028, |
| "grad_norm": 0.002143328543752432, |
| "kl": 0.003412064048461616, |
| "learning_rate": 7.714285714285716e-06, |
| "loss": -0.0001, |
| "step": 28, |
| "step_time": 2.5550204580013087 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 402.0, |
| "completions/max_terminated_length": 402.0, |
| "completions/mean_length": 146.15625, |
| "completions/mean_terminated_length": 146.15625, |
| "completions/min_length": 2.0, |
| "completions/min_terminated_length": 2.0, |
| "entropy": 8.294090986251831, |
| "epoch": 0.00029, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.014697319827973843, |
| "kl": 0.025721593061462045, |
| "learning_rate": 8.000000000000001e-06, |
| "loss": -0.0002, |
| "num_tokens": 270222.0, |
| "reward": -0.703125, |
| "reward_std": 0.2893909811973572, |
| "rewards/rollout_reward_func/mean": -0.703125, |
| "rewards/rollout_reward_func/std": 0.7459500432014465, |
| "sampling/importance_sampling_ratio/max": 0.05455424264073372, |
| "sampling/importance_sampling_ratio/mean": 0.006639046128839254, |
| "sampling/importance_sampling_ratio/min": 5.018679106327676e-15, |
| "sampling/sampling_logp_difference/max": 9.214568138122559, |
| "sampling/sampling_logp_difference/mean": 1.469120979309082, |
| "step": 29, |
| "step_time": 5.175338321999334 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.281907558441162, |
| "epoch": 0.0003, |
| "grad_norm": 0.01356051117181778, |
| "kl": 0.02538611611817032, |
| "learning_rate": 8.285714285714287e-06, |
| "loss": -0.0002, |
| "step": 30, |
| "step_time": 2.570307294000486 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 217.0, |
| "completions/max_terminated_length": 217.0, |
| "completions/mean_length": 122.40625, |
| "completions/mean_terminated_length": 122.40625, |
| "completions/min_length": 2.0, |
| "completions/min_terminated_length": 2.0, |
| "entropy": 7.991218149662018, |
| "epoch": 0.00031, |
| "frac_reward_zero_std": 1.0, |
| "grad_norm": 0.0034507170785218477, |
| "kl": 0.012121076317271218, |
| "learning_rate": 8.571428571428571e-06, |
| "loss": 0.0, |
| "num_tokens": 287003.0, |
| "reward": -1.0, |
| "reward_std": 0.0, |
| "rewards/rollout_reward_func/mean": -1.0, |
| "rewards/rollout_reward_func/std": 0.0, |
| "sampling/importance_sampling_ratio/max": 0.06423835456371307, |
| "sampling/importance_sampling_ratio/mean": 0.006418607663363218, |
| "sampling/importance_sampling_ratio/min": 2.267779519726787e-09, |
| "sampling/sampling_logp_difference/max": 7.289473533630371, |
| "sampling/sampling_logp_difference/mean": 1.3377501964569092, |
| "step": 31, |
| "step_time": 4.991982824997649 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 7.985842347145081, |
| "epoch": 0.00032, |
| "grad_norm": 0.0031677871011197567, |
| "kl": 0.01152854437532369, |
| "learning_rate": 8.857142857142858e-06, |
| "loss": 0.0, |
| "step": 32, |
| "step_time": 2.551680210001905 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 290.0, |
| "completions/max_terminated_length": 290.0, |
| "completions/mean_length": 141.4375, |
| "completions/mean_terminated_length": 141.4375, |
| "completions/min_length": 2.0, |
| "completions/min_terminated_length": 2.0, |
| "entropy": 8.251873970031738, |
| "epoch": 0.00033, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.011402878910303116, |
| "kl": 0.015730057610198855, |
| "learning_rate": 9.142857142857144e-06, |
| "loss": -0.0004, |
| "num_tokens": 305185.0, |
| "reward": -0.3999999761581421, |
| "reward_std": 0.5431233644485474, |
| "rewards/rollout_reward_func/mean": -0.3999999761581421, |
| "rewards/rollout_reward_func/std": 0.9466408491134644, |
| "sampling/importance_sampling_ratio/max": 0.05344817042350769, |
| "sampling/importance_sampling_ratio/mean": 0.005628373473882675, |
| "sampling/importance_sampling_ratio/min": 1.8508519615559533e-23, |
| "sampling/sampling_logp_difference/max": 13.587614059448242, |
| "sampling/sampling_logp_difference/mean": 1.4650746583938599, |
| "step": 33, |
| "step_time": 6.090412677998756 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.246023058891296, |
| "epoch": 0.00034, |
| "grad_norm": 0.008096246048808098, |
| "kl": 0.014423061889829114, |
| "learning_rate": 9.42857142857143e-06, |
| "loss": -0.0004, |
| "step": 34, |
| "step_time": 2.5453261089996886 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 282.0, |
| "completions/max_terminated_length": 282.0, |
| "completions/mean_length": 132.625, |
| "completions/mean_terminated_length": 132.625, |
| "completions/min_length": 2.0, |
| "completions/min_terminated_length": 2.0, |
| "entropy": 8.162899851799011, |
| "epoch": 0.00035, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.004175754263997078, |
| "kl": 0.018283673271071166, |
| "learning_rate": 9.714285714285715e-06, |
| "loss": -0.0001, |
| "num_tokens": 323325.0, |
| "reward": -0.5718749761581421, |
| "reward_std": 0.18343815207481384, |
| "rewards/rollout_reward_func/mean": -0.5718749761581421, |
| "rewards/rollout_reward_func/std": 0.8301706910133362, |
| "sampling/importance_sampling_ratio/max": 0.062451351433992386, |
| "sampling/importance_sampling_ratio/mean": 0.007372735999524593, |
| "sampling/importance_sampling_ratio/min": 1.2278145562505762e-20, |
| "sampling/sampling_logp_difference/max": 9.404362678527832, |
| "sampling/sampling_logp_difference/mean": 1.5296763181686401, |
| "step": 35, |
| "step_time": 5.175936893000653 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.150159418582916, |
| "epoch": 0.00036, |
| "grad_norm": 0.004381407983601093, |
| "kl": 0.017532640282297507, |
| "learning_rate": 1e-05, |
| "loss": -0.0001, |
| "step": 36, |
| "step_time": 2.551337270998374 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 350.0, |
| "completions/max_terminated_length": 350.0, |
| "completions/mean_length": 193.5625, |
| "completions/mean_terminated_length": 193.5625, |
| "completions/min_length": 2.0, |
| "completions/min_terminated_length": 2.0, |
| "entropy": 8.269986033439636, |
| "epoch": 0.00037, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.008302879519760609, |
| "kl": 0.009193298814352602, |
| "learning_rate": 9.9999999995372e-06, |
| "loss": 0.0005, |
| "num_tokens": 342711.0, |
| "reward": -0.03749999403953552, |
| "reward_std": 0.7515531778335571, |
| "rewards/rollout_reward_func/mean": -0.03749999403953552, |
| "rewards/rollout_reward_func/std": 1.0247737169265747, |
| "sampling/importance_sampling_ratio/max": 0.06419403851032257, |
| "sampling/importance_sampling_ratio/mean": 0.003890752326697111, |
| "sampling/importance_sampling_ratio/min": 1.164774526829504e-10, |
| "sampling/sampling_logp_difference/max": 3.794468402862549, |
| "sampling/sampling_logp_difference/mean": 1.2929356098175049, |
| "step": 37, |
| "step_time": 4.535253317000752 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.270979046821594, |
| "epoch": 0.00038, |
| "grad_norm": 0.00880998931825161, |
| "kl": 0.007889840024290606, |
| "learning_rate": 9.999999998148802e-06, |
| "loss": 0.0005, |
| "step": 38, |
| "step_time": 2.492936824000026 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.03125, |
| "completions/max_length": 347.0, |
| "completions/max_terminated_length": 347.0, |
| "completions/mean_length": 171.09375, |
| "completions/mean_terminated_length": 172.93548583984375, |
| "completions/min_length": 89.0, |
| "completions/min_terminated_length": 89.0, |
| "entropy": 8.237572193145752, |
| "epoch": 0.00039, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.003999819979071617, |
| "kl": 0.003984437556937337, |
| "learning_rate": 9.999999995834804e-06, |
| "loss": -0.0003, |
| "num_tokens": 361754.0, |
| "reward": -0.2750000059604645, |
| "reward_std": 0.29003918170928955, |
| "rewards/rollout_reward_func/mean": -0.2750000059604645, |
| "rewards/rollout_reward_func/std": 0.9721940755844116, |
| "sampling/importance_sampling_ratio/max": 0.00961074884980917, |
| "sampling/importance_sampling_ratio/mean": 0.0027457564137876034, |
| "sampling/importance_sampling_ratio/min": 1.3412947045145382e-17, |
| "sampling/sampling_logp_difference/max": 10.32085132598877, |
| "sampling/sampling_logp_difference/mean": 1.3388471603393555, |
| "step": 39, |
| "step_time": 4.8369908450004 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.241283535957336, |
| "epoch": 0.0004, |
| "grad_norm": 0.003973928280174732, |
| "kl": 0.004160504468018189, |
| "learning_rate": 9.999999992595207e-06, |
| "loss": -0.0003, |
| "step": 40, |
| "step_time": 2.064312154999243 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.03125, |
| "completions/max_length": 408.0, |
| "completions/max_terminated_length": 408.0, |
| "completions/mean_length": 149.6875, |
| "completions/mean_terminated_length": 150.7096710205078, |
| "completions/min_length": 87.0, |
| "completions/min_terminated_length": 87.0, |
| "entropy": 8.331173419952393, |
| "epoch": 0.00041, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.0036302730441093445, |
| "kl": 0.002889991897973232, |
| "learning_rate": 9.999999988430008e-06, |
| "loss": -0.0, |
| "num_tokens": 380000.0, |
| "reward": -0.40312498807907104, |
| "reward_std": 0.5359131097793579, |
| "rewards/rollout_reward_func/mean": -0.40312498807907104, |
| "rewards/rollout_reward_func/std": 0.9361845850944519, |
| "sampling/importance_sampling_ratio/max": 0.009129444137215614, |
| "sampling/importance_sampling_ratio/mean": 0.003908202983438969, |
| "sampling/importance_sampling_ratio/min": 3.258544557229945e-14, |
| "sampling/sampling_logp_difference/max": 4.673151016235352, |
| "sampling/sampling_logp_difference/mean": 1.426315426826477, |
| "step": 41, |
| "step_time": 4.397045022999009 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.317348182201385, |
| "epoch": 0.00042, |
| "grad_norm": 0.002291926182806492, |
| "kl": 0.002586090617114678, |
| "learning_rate": 9.999999983339212e-06, |
| "loss": -0.0, |
| "step": 42, |
| "step_time": 2.1006949159991564 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 408.0, |
| "completions/max_terminated_length": 408.0, |
| "completions/mean_length": 168.1875, |
| "completions/mean_terminated_length": 168.1875, |
| "completions/min_length": 89.0, |
| "completions/min_terminated_length": 89.0, |
| "entropy": 8.245356917381287, |
| "epoch": 0.00043, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.004080729093402624, |
| "kl": 0.003670690639410168, |
| "learning_rate": 9.999999977322818e-06, |
| "loss": -0.0004, |
| "num_tokens": 398190.0, |
| "reward": -0.5843750238418579, |
| "reward_std": 0.22265997529029846, |
| "rewards/rollout_reward_func/mean": -0.5843750238418579, |
| "rewards/rollout_reward_func/std": 0.8462857007980347, |
| "sampling/importance_sampling_ratio/max": 0.010589290410280228, |
| "sampling/importance_sampling_ratio/mean": 0.003948138561099768, |
| "sampling/importance_sampling_ratio/min": 1.0654254561925924e-10, |
| "sampling/sampling_logp_difference/max": 9.166659355163574, |
| "sampling/sampling_logp_difference/mean": 1.3605971336364746, |
| "step": 43, |
| "step_time": 4.460575121997863 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.227819621562958, |
| "epoch": 0.00044, |
| "grad_norm": 0.004100060556083918, |
| "kl": 0.003359506925335154, |
| "learning_rate": 9.999999970380822e-06, |
| "loss": -0.0004, |
| "step": 44, |
| "step_time": 3.0909940020010254 |
| }, |
| { |
| "clip_ratio/high_max": 0.01785714365541935, |
| "clip_ratio/high_mean": 0.008928571827709675, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.008928571827709675, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 297.0, |
| "completions/max_terminated_length": 297.0, |
| "completions/mean_length": 160.40625, |
| "completions/mean_terminated_length": 160.40625, |
| "completions/min_length": 87.0, |
| "completions/min_terminated_length": 87.0, |
| "entropy": 8.268202066421509, |
| "epoch": 0.00045, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.0023337905295193195, |
| "kl": 0.0028031190449837595, |
| "learning_rate": 9.999999962513228e-06, |
| "loss": -0.0001, |
| "num_tokens": 416179.0, |
| "reward": -0.5843750238418579, |
| "reward_std": 0.20319493114948273, |
| "rewards/rollout_reward_func/mean": -0.5843750238418579, |
| "rewards/rollout_reward_func/std": 0.830510675907135, |
| "sampling/importance_sampling_ratio/max": 0.010615027509629726, |
| "sampling/importance_sampling_ratio/mean": 0.003981100395321846, |
| "sampling/importance_sampling_ratio/min": 1.0911494689562484e-13, |
| "sampling/sampling_logp_difference/max": 11.518604278564453, |
| "sampling/sampling_logp_difference/mean": 1.4530669450759888, |
| "step": 45, |
| "step_time": 4.283326912999655 |
| }, |
| { |
| "clip_ratio/high_max": 0.01785714365541935, |
| "clip_ratio/high_mean": 0.008928571827709675, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.008928571827709675, |
| "entropy": 8.239543914794922, |
| "epoch": 0.00046, |
| "grad_norm": 0.0035347214434295893, |
| "kl": 0.0033496549731353298, |
| "learning_rate": 9.999999953720035e-06, |
| "loss": -0.0001, |
| "step": 46, |
| "step_time": 2.0328167930001655 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 346.0, |
| "completions/max_terminated_length": 346.0, |
| "completions/mean_length": 154.96875, |
| "completions/mean_terminated_length": 154.96875, |
| "completions/min_length": 6.0, |
| "completions/min_terminated_length": 6.0, |
| "entropy": 8.14276933670044, |
| "epoch": 0.00047, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.003259505843743682, |
| "kl": 0.005610214080661535, |
| "learning_rate": 9.99999994400124e-06, |
| "loss": -0.0004, |
| "num_tokens": 434762.0, |
| "reward": -1.0218749046325684, |
| "reward_std": 0.04966200143098831, |
| "rewards/rollout_reward_func/mean": -1.0218749046325684, |
| "rewards/rollout_reward_func/std": 0.0490843690931797, |
| "sampling/importance_sampling_ratio/max": 0.011610783636569977, |
| "sampling/importance_sampling_ratio/mean": 0.003400737652555108, |
| "sampling/importance_sampling_ratio/min": 9.674043361674659e-17, |
| "sampling/sampling_logp_difference/max": 10.514336585998535, |
| "sampling/sampling_logp_difference/mean": 1.5065617561340332, |
| "step": 47, |
| "step_time": 4.387635872000828 |
| }, |
| { |
| "clip_ratio/high_max": 0.004807692486792803, |
| "clip_ratio/high_mean": 0.0024038462433964014, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0024038462433964014, |
| "entropy": 8.125671744346619, |
| "epoch": 0.00048, |
| "grad_norm": 0.0028330760542303324, |
| "kl": 0.005603832833003253, |
| "learning_rate": 9.999999933356848e-06, |
| "loss": -0.0004, |
| "step": 48, |
| "step_time": 2.031070074998752 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.03125, |
| "completions/max_length": 265.0, |
| "completions/max_terminated_length": 265.0, |
| "completions/mean_length": 153.46875, |
| "completions/mean_terminated_length": 154.61289978027344, |
| "completions/min_length": 2.0, |
| "completions/min_terminated_length": 2.0, |
| "entropy": 8.048729181289673, |
| "epoch": 0.00049, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.005494742188602686, |
| "kl": 0.03256297600455582, |
| "learning_rate": 9.999999921786855e-06, |
| "loss": -0.0001, |
| "num_tokens": 453105.0, |
| "reward": -0.078125, |
| "reward_std": 0.5431658029556274, |
| "rewards/rollout_reward_func/mean": -0.078125, |
| "rewards/rollout_reward_func/std": 1.0044207572937012, |
| "sampling/importance_sampling_ratio/max": 0.05582950636744499, |
| "sampling/importance_sampling_ratio/mean": 0.004962150938808918, |
| "sampling/importance_sampling_ratio/min": 2.203801160657881e-12, |
| "sampling/sampling_logp_difference/max": 9.083192825317383, |
| "sampling/sampling_logp_difference/mean": 1.358087182044983, |
| "step": 49, |
| "step_time": 4.150341610999931 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.026359498500824, |
| "epoch": 0.0005, |
| "grad_norm": 0.005025926977396011, |
| "kl": 0.02888420899398625, |
| "learning_rate": 9.999999909291265e-06, |
| "loss": -0.0001, |
| "step": 50, |
| "step_time": 2.9815445680005723 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 290.0, |
| "completions/max_terminated_length": 290.0, |
| "completions/mean_length": 143.65625, |
| "completions/mean_terminated_length": 143.65625, |
| "completions/min_length": 2.0, |
| "completions/min_terminated_length": 2.0, |
| "entropy": 8.10213428735733, |
| "epoch": 0.00051, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.001717552193440497, |
| "kl": 0.011197627696674317, |
| "learning_rate": 9.999999895870075e-06, |
| "loss": -0.0002, |
| "num_tokens": 470670.0, |
| "reward": -1.015625, |
| "reward_std": 0.036278266459703445, |
| "rewards/rollout_reward_func/mean": -1.015625, |
| "rewards/rollout_reward_func/std": 0.05148990824818611, |
| "sampling/importance_sampling_ratio/max": 0.06585416942834854, |
| "sampling/importance_sampling_ratio/mean": 0.006765150930732489, |
| "sampling/importance_sampling_ratio/min": 1.496036702519632e-06, |
| "sampling/sampling_logp_difference/max": 4.3300042152404785, |
| "sampling/sampling_logp_difference/mean": 1.3094828128814697, |
| "step": 51, |
| "step_time": 4.126140448999649 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.08252465724945, |
| "epoch": 0.00052, |
| "grad_norm": 0.0015695166075602174, |
| "kl": 0.0106147377518937, |
| "learning_rate": 9.999999881523285e-06, |
| "loss": -0.0002, |
| "step": 52, |
| "step_time": 2.0502745620005953 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.03125, |
| "completions/max_length": 314.0, |
| "completions/max_terminated_length": 314.0, |
| "completions/mean_length": 173.5, |
| "completions/mean_terminated_length": 175.29031372070312, |
| "completions/min_length": 91.0, |
| "completions/min_terminated_length": 91.0, |
| "entropy": 7.971049547195435, |
| "epoch": 0.00053, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.002233212348073721, |
| "kl": 0.004628648399375379, |
| "learning_rate": 9.999999866250896e-06, |
| "loss": -0.0001, |
| "num_tokens": 489622.0, |
| "reward": 0.04375000298023224, |
| "reward_std": 0.6950876712799072, |
| "rewards/rollout_reward_func/mean": 0.04375000298023224, |
| "rewards/rollout_reward_func/std": 1.0162986516952515, |
| "sampling/importance_sampling_ratio/max": 0.014256482943892479, |
| "sampling/importance_sampling_ratio/mean": 0.0037497361190617085, |
| "sampling/importance_sampling_ratio/min": 3.703341886623912e-13, |
| "sampling/sampling_logp_difference/max": 10.441887855529785, |
| "sampling/sampling_logp_difference/mean": 1.3701958656311035, |
| "step": 53, |
| "step_time": 4.342945265999333 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 7.953936517238617, |
| "epoch": 0.00054, |
| "grad_norm": 0.0023060261737555265, |
| "kl": 0.004763354663737118, |
| "learning_rate": 9.999999850052909e-06, |
| "loss": -0.0001, |
| "step": 54, |
| "step_time": 2.074805829001889 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.03125, |
| "completions/max_length": 469.0, |
| "completions/max_terminated_length": 469.0, |
| "completions/mean_length": 133.125, |
| "completions/mean_terminated_length": 133.61289978027344, |
| "completions/min_length": 2.0, |
| "completions/min_terminated_length": 2.0, |
| "entropy": 7.858851313591003, |
| "epoch": 0.00055, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.010865299962460995, |
| "kl": 0.05283547495491803, |
| "learning_rate": 9.99999983292932e-06, |
| "loss": 0.0005, |
| "num_tokens": 507130.0, |
| "reward": -0.08750000596046448, |
| "reward_std": 0.2218937873840332, |
| "rewards/rollout_reward_func/mean": -0.08750000596046448, |
| "rewards/rollout_reward_func/std": 1.0082721710205078, |
| "sampling/importance_sampling_ratio/max": 0.0794573649764061, |
| "sampling/importance_sampling_ratio/mean": 0.01186932623386383, |
| "sampling/importance_sampling_ratio/min": 2.8194972302655857e-18, |
| "sampling/sampling_logp_difference/max": 11.174360275268555, |
| "sampling/sampling_logp_difference/mean": 1.4186177253723145, |
| "step": 55, |
| "step_time": 4.7503355889994054 |
| }, |
| { |
| "clip_ratio/high_max": 0.005681818351149559, |
| "clip_ratio/high_mean": 0.0028409091755747795, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0028409091755747795, |
| "entropy": 7.84703141450882, |
| "epoch": 0.00056, |
| "grad_norm": 0.009876980446279049, |
| "kl": 0.04835269978502765, |
| "learning_rate": 9.999999814880132e-06, |
| "loss": 0.0005, |
| "step": 56, |
| "step_time": 2.697479030000977 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 309.0, |
| "completions/max_terminated_length": 309.0, |
| "completions/mean_length": 161.03125, |
| "completions/mean_terminated_length": 161.03125, |
| "completions/min_length": 87.0, |
| "completions/min_terminated_length": 87.0, |
| "entropy": 7.946120798587799, |
| "epoch": 0.00057, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.00416550925001502, |
| "kl": 0.005225592787610367, |
| "learning_rate": 9.999999795905347e-06, |
| "loss": 0.0003, |
| "num_tokens": 524811.0, |
| "reward": -0.328125, |
| "reward_std": 0.7972557544708252, |
| "rewards/rollout_reward_func/mean": -0.328125, |
| "rewards/rollout_reward_func/std": 0.9642762541770935, |
| "sampling/importance_sampling_ratio/max": 0.011825657449662685, |
| "sampling/importance_sampling_ratio/mean": 0.004940683953464031, |
| "sampling/importance_sampling_ratio/min": 7.442185295759504e-14, |
| "sampling/sampling_logp_difference/max": 10.764989852905273, |
| "sampling/sampling_logp_difference/mean": 1.3635454177856445, |
| "step": 57, |
| "step_time": 4.399483632999363 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 7.954219579696655, |
| "epoch": 0.00058, |
| "grad_norm": 0.004001974128186703, |
| "kl": 0.005270991328870878, |
| "learning_rate": 9.999999776004962e-06, |
| "loss": 0.0003, |
| "step": 58, |
| "step_time": 2.068970861998423 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 305.0, |
| "completions/max_terminated_length": 305.0, |
| "completions/mean_length": 137.78125, |
| "completions/mean_terminated_length": 137.78125, |
| "completions/min_length": 100.0, |
| "completions/min_terminated_length": 100.0, |
| "entropy": 7.7305819392204285, |
| "epoch": 0.00059, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.002240613801404834, |
| "kl": 0.007570188608951867, |
| "learning_rate": 9.999999755178978e-06, |
| "loss": -0.0002, |
| "num_tokens": 542324.0, |
| "reward": -0.565625011920929, |
| "reward_std": 0.18561552464962006, |
| "rewards/rollout_reward_func/mean": -0.565625011920929, |
| "rewards/rollout_reward_func/std": 0.841890811920166, |
| "sampling/importance_sampling_ratio/max": 0.014670869335532188, |
| "sampling/importance_sampling_ratio/mean": 0.006734498776495457, |
| "sampling/importance_sampling_ratio/min": 9.527041698945138e-13, |
| "sampling/sampling_logp_difference/max": 10.319395065307617, |
| "sampling/sampling_logp_difference/mean": 1.346876621246338, |
| "step": 59, |
| "step_time": 4.076921171999857 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 7.7493292689323425, |
| "epoch": 0.0006, |
| "grad_norm": 0.0021323147229850292, |
| "kl": 0.007373731641564518, |
| "learning_rate": 9.999999733427394e-06, |
| "loss": -0.0002, |
| "step": 60, |
| "step_time": 2.065178036998077 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0625, |
| "completions/max_length": 305.0, |
| "completions/max_terminated_length": 305.0, |
| "completions/mean_length": 131.15625, |
| "completions/mean_terminated_length": 132.53334045410156, |
| "completions/min_length": 100.0, |
| "completions/min_terminated_length": 100.0, |
| "entropy": 7.80867725610733, |
| "epoch": 0.00061, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.002022077329456806, |
| "kl": 0.004840011039050296, |
| "learning_rate": 9.99999971075021e-06, |
| "loss": -0.0003, |
| "num_tokens": 559385.0, |
| "reward": -0.25312501192092896, |
| "reward_std": 0.2761000692844391, |
| "rewards/rollout_reward_func/mean": -0.25312501192092896, |
| "rewards/rollout_reward_func/std": 0.9863533973693848, |
| "sampling/importance_sampling_ratio/max": 0.011284389533102512, |
| "sampling/importance_sampling_ratio/mean": 0.006653377786278725, |
| "sampling/importance_sampling_ratio/min": 2.8632414147966578e-11, |
| "sampling/sampling_logp_difference/max": 4.169203758239746, |
| "sampling/sampling_logp_difference/mean": 1.2035651206970215, |
| "step": 61, |
| "step_time": 4.5231009249973795 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 7.829549431800842, |
| "epoch": 0.00062, |
| "grad_norm": 0.0023180257994681597, |
| "kl": 0.004549442324787378, |
| "learning_rate": 9.999999687147426e-06, |
| "loss": -0.0003, |
| "step": 62, |
| "step_time": 2.04733120399942 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 212.0, |
| "completions/max_terminated_length": 212.0, |
| "completions/mean_length": 123.1875, |
| "completions/mean_terminated_length": 123.1875, |
| "completions/min_length": 6.0, |
| "completions/min_terminated_length": 6.0, |
| "entropy": 8.036745607852936, |
| "epoch": 0.00063, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.0024225530214607716, |
| "kl": 0.007575233932584524, |
| "learning_rate": 9.999999662619046e-06, |
| "loss": -0.0004, |
| "num_tokens": 577175.0, |
| "reward": -0.6437500715255737, |
| "reward_std": 0.2552982568740845, |
| "rewards/rollout_reward_func/mean": -0.6437500715255737, |
| "rewards/rollout_reward_func/std": 0.7873751521110535, |
| "sampling/importance_sampling_ratio/max": 0.010909978300333023, |
| "sampling/importance_sampling_ratio/mean": 0.005395432468503714, |
| "sampling/importance_sampling_ratio/min": 2.1033281019655625e-11, |
| "sampling/sampling_logp_difference/max": 10.216733932495117, |
| "sampling/sampling_logp_difference/mean": 1.502685546875, |
| "step": 63, |
| "step_time": 3.9012105610008803 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.038543045520782, |
| "epoch": 0.00064, |
| "grad_norm": 0.0024051358923316, |
| "kl": 0.007686805154662579, |
| "learning_rate": 9.999999637165062e-06, |
| "loss": -0.0004, |
| "step": 64, |
| "step_time": 2.0093538759983858 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 282.0, |
| "completions/max_terminated_length": 282.0, |
| "completions/mean_length": 141.65625, |
| "completions/mean_terminated_length": 141.65625, |
| "completions/min_length": 62.0, |
| "completions/min_terminated_length": 62.0, |
| "entropy": 8.191402852535248, |
| "epoch": 0.00065, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.003754268866032362, |
| "kl": 0.005535919044632465, |
| "learning_rate": 9.999999610785483e-06, |
| "loss": -0.0, |
| "num_tokens": 594644.0, |
| "reward": -0.33125001192092896, |
| "reward_std": 0.4657542407512665, |
| "rewards/rollout_reward_func/mean": -0.33125001192092896, |
| "rewards/rollout_reward_func/std": 0.9663491249084473, |
| "sampling/importance_sampling_ratio/max": 0.010239495895802975, |
| "sampling/importance_sampling_ratio/mean": 0.004091148264706135, |
| "sampling/importance_sampling_ratio/min": 1.6428119986328787e-13, |
| "sampling/sampling_logp_difference/max": 11.840818405151367, |
| "sampling/sampling_logp_difference/mean": 1.458545446395874, |
| "step": 65, |
| "step_time": 4.098211152998374 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "entropy": 8.207817077636719, |
| "epoch": 0.00066, |
| "grad_norm": 0.0028874659910798073, |
| "kl": 0.005369087448343635, |
| "learning_rate": 9.999999583480304e-06, |
| "loss": -0.0, |
| "step": 66, |
| "step_time": 2.03046784199978 |
| }, |
| { |
| "clip_ratio/high_max": 0.005434782709926367, |
| "clip_ratio/high_mean": 0.0027173913549631834, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0027173913549631834, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 286.0, |
| "completions/max_terminated_length": 286.0, |
| "completions/mean_length": 131.0625, |
| "completions/mean_terminated_length": 131.0625, |
| "completions/min_length": 2.0, |
| "completions/min_terminated_length": 2.0, |
| "entropy": 8.138030529022217, |
| "epoch": 0.00067, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.008762937970459461, |
| "kl": 0.03389387877541594, |
| "learning_rate": 9.999999555249524e-06, |
| "loss": 0.0004, |
| "num_tokens": 612302.0, |
| "reward": -0.32500001788139343, |
| "reward_std": 0.44983184337615967, |
| "rewards/rollout_reward_func/mean": -0.32500001788139343, |
| "rewards/rollout_reward_func/std": 0.9615175724029541, |
| "sampling/importance_sampling_ratio/max": 0.06631788611412048, |
| "sampling/importance_sampling_ratio/mean": 0.009149353951215744, |
| "sampling/importance_sampling_ratio/min": 2.947741995564272e-15, |
| "sampling/sampling_logp_difference/max": 4.271122455596924, |
| "sampling/sampling_logp_difference/mean": 1.307502269744873, |
| "step": 67, |
| "step_time": 4.921273821997602 |
| }, |
| { |
| "clip_ratio/high_max": 0.005434782709926367, |
| "clip_ratio/high_mean": 0.0027173913549631834, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0027173913549631834, |
| "entropy": 8.132037341594696, |
| "epoch": 0.00068, |
| "grad_norm": 0.009227960370481014, |
| "kl": 0.03418682742631063, |
| "learning_rate": 9.999999526093148e-06, |
| "loss": 0.0004, |
| "step": 68, |
| "step_time": 2.032772711999314 |
| } |
| ], |
| "logging_steps": 1.0, |
| "max_steps": 200000, |
| "num_input_tokens_seen": 612302, |
| "num_train_epochs": 2, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 0.0, |
| "train_batch_size": 2, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
