{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 450, "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": 571.8, "completions/max_terminated_length": 571.8, "completions/mean_length": 210.3375, "completions/mean_terminated_length": 210.3375, "completions/min_length": 25.9, "completions/min_terminated_length": 25.9, "entropy": 0.5260405464097857, "epoch": 0.022222222222222223, "frac_reward_zero_std": 0.6, "grad_norm": 0.08984375, "kl": 0.004743382518677208, "learning_rate": 9.990133642141357e-07, "loss": -0.1532900333404541, "num_tokens": 28079.0, "reward": 1.5, "reward_std": 0.4325482964515686, "rewards/accuracy_reward_func/mean": 0.525, "rewards/accuracy_reward_func/std": 0.3935012608766556, "rewards/format_reward_func/mean": 0.975, "rewards/format_reward_func/std": 0.07071067690849304, "step": 10, "step_time": 22.153983098920435 }, { "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": 0.025, "completions/max_length": 781.7, "completions/max_terminated_length": 520.5, "completions/mean_length": 244.575, "completions/mean_terminated_length": 199.82678833007813, "completions/min_length": 11.8, "completions/min_terminated_length": 11.8, "entropy": 0.34257757353479973, "epoch": 0.044444444444444446, "frac_reward_zero_std": 0.65, "grad_norm": 0.765625, "kl": 0.006423782935695499, "learning_rate": 9.956077701257707e-07, "loss": 0.05708537697792053, "num_tokens": 58605.0, "reward": 1.725, "reward_std": 0.4088952898979187, "rewards/accuracy_reward_func/mean": 0.775, "rewards/accuracy_reward_func/std": 0.32741602063179015, "rewards/format_reward_func/mean": 0.95, "rewards/format_reward_func/std": 0.09258201122283935, "step": 20, "step_time": 29.545378133747725 }, { "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": 0.025, "completions/max_length": 941.8, "completions/max_terminated_length": 809.2, "completions/mean_length": 266.275, "completions/mean_terminated_length": 228.87679138183594, "completions/min_length": 19.2, "completions/min_terminated_length": 19.2, "entropy": 0.4463253241032362, "epoch": 0.06666666666666667, "frac_reward_zero_std": 0.5, "grad_norm": 0.83984375, "kl": 0.004194882212846096, "learning_rate": 9.89787624799672e-07, "loss": -0.03688110411167145, "num_tokens": 90843.0, "reward": 1.55, "reward_std": 0.46539856791496276, "rewards/accuracy_reward_func/mean": 0.575, "rewards/accuracy_reward_func/std": 0.4237646222114563, "rewards/format_reward_func/mean": 0.975, "rewards/format_reward_func/std": 0.07071067690849304, "step": 30, "step_time": 35.055862438119945 }, { "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": 623.7, "completions/max_terminated_length": 623.7, "completions/mean_length": 174.3375, "completions/mean_terminated_length": 174.3375, "completions/min_length": 12.5, "completions/min_terminated_length": 12.5, "entropy": 0.4108163947239518, "epoch": 0.08888888888888889, "frac_reward_zero_std": 0.7, "grad_norm": 0.86328125, "kl": 0.004012970752546608, "learning_rate": 9.81581283398829e-07, "loss": -0.041548117995262146, "num_tokens": 115550.0, "reward": 1.45, "reward_std": 0.3992423236370087, "rewards/accuracy_reward_func/mean": 0.475, "rewards/accuracy_reward_func/std": 0.36994043588638303, "rewards/format_reward_func/mean": 0.975, "rewards/format_reward_func/std": 0.046291005611419675, "step": 40, "step_time": 23.89164825808257 }, { "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": 0.0375, "completions/max_length": 1086.7, "completions/max_terminated_length": 770.1, "completions/mean_length": 295.6625, "completions/mean_terminated_length": 225.37857208251953, "completions/min_length": 19.0, "completions/min_terminated_length": 19.0, "entropy": 0.3411364992149174, "epoch": 0.1111111111111111, "frac_reward_zero_std": 0.45, "grad_norm": 0.1416015625, "kl": 0.0024333317864744684, "learning_rate": 9.710287263936483e-07, "loss": 0.01952042579650879, "num_tokens": 150471.0, "reward": 1.5875, "reward_std": 0.5131408303976059, "rewards/accuracy_reward_func/mean": 0.6375, "rewards/accuracy_reward_func/std": 0.43809494078159333, "rewards/format_reward_func/mean": 0.95, "rewards/format_reward_func/std": 0.09258201122283935, "step": 50, "step_time": 40.592786005605014 }, { "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": 416.1, "completions/max_terminated_length": 416.1, "completions/mean_length": 132.075, "completions/mean_terminated_length": 132.075, "completions/min_length": 11.1, "completions/min_terminated_length": 11.1, "entropy": 0.4589052753522992, "epoch": 0.13333333333333333, "frac_reward_zero_std": 0.75, "grad_norm": 10.5, "kl": 0.0038458615549870957, "learning_rate": 9.581813647811197e-07, "loss": 0.013749442994594574, "num_tokens": 171793.0, "reward": 1.775, "reward_std": 0.2839782267808914, "rewards/accuracy_reward_func/mean": 0.8, "rewards/accuracy_reward_func/std": 0.24407235085964202, "rewards/format_reward_func/mean": 0.975, "rewards/format_reward_func/std": 0.07071067690849304, "step": 60, "step_time": 16.592803716100754 }, { "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": 0.05, "completions/max_length": 838.2, "completions/max_terminated_length": 672.2, "completions/mean_length": 310.8625, "completions/mean_terminated_length": 223.80107421875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "entropy": 0.39748903196305035, "epoch": 0.15555555555555556, "frac_reward_zero_std": 0.6, "grad_norm": 3.203125, "kl": 0.009320074055262273, "learning_rate": 9.431017896156073e-07, "loss": 0.04533530175685883, "num_tokens": 208122.0, "reward": 1.525, "reward_std": 0.44547489285469055, "rewards/accuracy_reward_func/mean": 0.5875, "rewards/accuracy_reward_func/std": 0.34551292955875396, "rewards/format_reward_func/mean": 0.9375, "rewards/format_reward_func/std": 0.12246559858322144, "step": 70, "step_time": 31.752976473141462 }, { "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": 0.025, "completions/max_length": 666.2, "completions/max_terminated_length": 512.1, "completions/mean_length": 155.05, "completions/mean_terminated_length": 111.56607513427734, "completions/min_length": 11.2, "completions/min_terminated_length": 11.2, "entropy": 0.4051839893683791, "epoch": 0.17777777777777778, "frac_reward_zero_std": 0.6, "grad_norm": 6.03125, "kl": 0.006855625934451837, "learning_rate": 9.258634670715237e-07, "loss": 0.1526792049407959, "num_tokens": 231274.0, "reward": 1.625, "reward_std": 0.36304850578308107, "rewards/accuracy_reward_func/mean": 0.65, "rewards/accuracy_reward_func/std": 0.30385519564151764, "rewards/format_reward_func/mean": 0.975, "rewards/format_reward_func/std": 0.07071067690849304, "step": 80, "step_time": 25.53765450278297 }, { "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": 0.0375, "completions/max_length": 641.0, "completions/max_terminated_length": 541.7, "completions/mean_length": 252.0, "completions/mean_terminated_length": 197.65595703125, "completions/min_length": 11.9, "completions/min_terminated_length": 11.9, "entropy": 0.3982577932416461, "epoch": 0.2, "frac_reward_zero_std": 0.7, "grad_norm": 18.5, "kl": 0.004655605388043149, "learning_rate": 9.065503805235137e-07, "loss": 0.05322921872138977, "num_tokens": 262278.0, "reward": 1.6, "reward_std": 0.3093269467353821, "rewards/accuracy_reward_func/mean": 0.6375, "rewards/accuracy_reward_func/std": 0.22768060266971588, "rewards/format_reward_func/mean": 0.9625, "rewards/format_reward_func/std": 0.0816463440656662, "step": 90, "step_time": 24.73598325951025 }, { "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": 501.6, "completions/max_terminated_length": 501.6, "completions/mean_length": 117.4625, "completions/mean_terminated_length": 117.4625, "completions/min_length": 11.9, "completions/min_terminated_length": 11.9, "entropy": 0.4523235411616042, "epoch": 0.2222222222222222, "frac_reward_zero_std": 0.8, "grad_norm": 0.2412109375, "kl": 0.0038861612925586543, "learning_rate": 8.852566213878946e-07, "loss": -0.17166446447372435, "num_tokens": 282819.0, "reward": 1.9, "reward_std": 0.1796922653913498, "rewards/accuracy_reward_func/mean": 0.9, "rewards/accuracy_reward_func/std": 0.17969227135181426, "rewards/format_reward_func/mean": 1.0, "rewards/format_reward_func/std": 0.0, "step": 100, "step_time": 19.58473698506132 }, { "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": 364.9, "completions/max_terminated_length": 364.9, "completions/mean_length": 93.0125, "completions/mean_terminated_length": 93.0125, "completions/min_length": 10.7, "completions/min_terminated_length": 10.7, "entropy": 0.3852502123452723, "epoch": 0.24444444444444444, "frac_reward_zero_std": 0.7, "grad_norm": 38.5, "kl": 0.0029849309575197937, "learning_rate": 8.620859307187338e-07, "loss": -0.06970181465148925, "num_tokens": 301756.0, "reward": 1.675, "reward_std": 0.3110164314508438, "rewards/accuracy_reward_func/mean": 0.6875, "rewards/accuracy_reward_func/std": 0.2929195284843445, "rewards/format_reward_func/mean": 0.9875, "rewards/format_reward_func/std": 0.03535533845424652, "step": 110, "step_time": 14.820625300519168 }, { "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": 0.0125, "completions/max_length": 782.8, "completions/max_terminated_length": 633.1, "completions/mean_length": 221.725, "completions/mean_terminated_length": 198.3125015258789, "completions/min_length": 12.1, "completions/min_terminated_length": 12.1, "entropy": 0.40864241449162364, "epoch": 0.26666666666666666, "frac_reward_zero_std": 0.75, "grad_norm": 4.0625, "kl": 0.003238640889708222, "learning_rate": 8.371511937918617e-07, "loss": 0.06257223486900329, "num_tokens": 331842.0, "reward": 1.65, "reward_std": 0.36446867883205414, "rewards/accuracy_reward_func/mean": 0.6625, "rewards/accuracy_reward_func/std": 0.3291133463382721, "rewards/format_reward_func/mean": 0.9875, "rewards/format_reward_func/std": 0.03535533845424652, "step": 120, "step_time": 29.600185937993228 }, { "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": 0.0125, "completions/max_length": 615.2, "completions/max_terminated_length": 424.7, "completions/mean_length": 178.8875, "completions/mean_terminated_length": 153.88571434020997, "completions/min_length": 11.4, "completions/min_terminated_length": 11.4, "entropy": 0.36546113309450445, "epoch": 0.28888888888888886, "frac_reward_zero_std": 0.55, "grad_norm": 4.03125, "kl": 0.043059189406889686, "learning_rate": 8.105738901391551e-07, "loss": 0.0829508662223816, "num_tokens": 357617.0, "reward": 1.6625, "reward_std": 0.3564689576625824, "rewards/accuracy_reward_func/mean": 0.6875, "rewards/accuracy_reward_func/std": 0.3185574531555176, "rewards/format_reward_func/mean": 0.975, "rewards/format_reward_func/std": 0.07071067690849304, "step": 130, "step_time": 23.732911340519784 }, { "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": 0.0375, "completions/max_length": 794.2, "completions/max_terminated_length": 587.0, "completions/mean_length": 254.4625, "completions/mean_terminated_length": 190.2952407836914, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "entropy": 0.4260031552053988, "epoch": 0.3111111111111111, "frac_reward_zero_std": 0.7, "grad_norm": 2.40625, "kl": 0.004566994193298512, "learning_rate": 7.82483501712469e-07, "loss": 0.1280433177947998, "num_tokens": 389998.0, "reward": 1.5875, "reward_std": 0.31729506254196166, "rewards/accuracy_reward_func/mean": 0.6375, "rewards/accuracy_reward_func/std": 0.23484184443950654, "rewards/format_reward_func/mean": 0.95, "rewards/format_reward_func/std": 0.11700168251991272, "step": 140, "step_time": 29.948555017076433 }, { "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": 0.025, "completions/max_length": 481.9, "completions/max_terminated_length": 379.7, "completions/mean_length": 156.2375, "completions/mean_terminated_length": 113.08333435058594, "completions/min_length": 11.0, "completions/min_terminated_length": 11.0, "entropy": 0.35120144123211505, "epoch": 0.3333333333333333, "frac_reward_zero_std": 0.75, "grad_norm": 0.1015625, "kl": 0.0030827762759145115, "learning_rate": 7.530168820605818e-07, "loss": -0.0070336699485778805, "num_tokens": 413489.0, "reward": 1.5375, "reward_std": 0.41186747550964353, "rewards/accuracy_reward_func/mean": 0.575, "rewards/accuracy_reward_func/std": 0.3472102552652359, "rewards/format_reward_func/mean": 0.9625, "rewards/format_reward_func/std": 0.0816463440656662, "step": 150, "step_time": 18.941299455426634 }, { "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": 0.0125, "completions/max_length": 822.9, "completions/max_terminated_length": 656.7, "completions/mean_length": 179.2125, "completions/mean_terminated_length": 154.84642868041993, "completions/min_length": 11.8, "completions/min_terminated_length": 11.8, "entropy": 0.40060292216949167, "epoch": 0.35555555555555557, "frac_reward_zero_std": 0.6, "grad_norm": 3.8125, "kl": 0.009431924868613351, "learning_rate": 7.223175895924637e-07, "loss": -0.025944623351097106, "num_tokens": 438754.0, "reward": 1.675, "reward_std": 0.41999801993370056, "rewards/accuracy_reward_func/mean": 0.6875, "rewards/accuracy_reward_func/std": 0.4010422766208649, "rewards/format_reward_func/mean": 0.9875, "rewards/format_reward_func/std": 0.03535533845424652, "step": 160, "step_time": 31.001335567049683 }, { "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": 371.1, "completions/max_terminated_length": 371.1, "completions/mean_length": 124.725, "completions/mean_terminated_length": 124.725, "completions/min_length": 12.1, "completions/min_terminated_length": 12.1, "entropy": 0.39380953572690486, "epoch": 0.37777777777777777, "frac_reward_zero_std": 0.75, "grad_norm": 0.78515625, "kl": 0.006049546646372051, "learning_rate": 6.905351881751371e-07, "loss": -0.037034496665000916, "num_tokens": 458932.0, "reward": 1.75, "reward_std": 0.3400490075349808, "rewards/accuracy_reward_func/mean": 0.75, "rewards/accuracy_reward_func/std": 0.34004901349544525, "rewards/format_reward_func/mean": 1.0, "rewards/format_reward_func/std": 0.0, "step": 170, "step_time": 14.985138600133359 }, { "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": 692.3, "completions/max_terminated_length": 692.3, "completions/mean_length": 206.1875, "completions/mean_terminated_length": 206.1875, "completions/min_length": 13.7, "completions/min_terminated_length": 13.7, "entropy": 0.3043558836914599, "epoch": 0.4, "frac_reward_zero_std": 0.5, "grad_norm": 0.0791015625, "kl": 0.0056259966624109484, "learning_rate": 6.578245184735512e-07, "loss": 0.18330377340316772, "num_tokens": 486087.0, "reward": 1.7, "reward_std": 0.34984527826309203, "rewards/accuracy_reward_func/mean": 0.7125, "rewards/accuracy_reward_func/std": 0.3320492923259735, "rewards/format_reward_func/mean": 0.9875, "rewards/format_reward_func/std": 0.03535533845424652, "step": 180, "step_time": 26.418801236059515 }, { "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": 484.5, "completions/max_terminated_length": 484.5, "completions/mean_length": 115.0125, "completions/mean_terminated_length": 115.0125, "completions/min_length": 11.9, "completions/min_terminated_length": 11.9, "entropy": 0.34211956479121, "epoch": 0.4222222222222222, "frac_reward_zero_std": 0.55, "grad_norm": 95.0, "kl": 0.006216998511573602, "learning_rate": 6.243449435824276e-07, "loss": -0.023849168419837953, "num_tokens": 506532.0, "reward": 1.5875, "reward_std": 0.4254619300365448, "rewards/accuracy_reward_func/mean": 0.5875, "rewards/accuracy_reward_func/std": 0.42546194791793823, "rewards/format_reward_func/mean": 1.0, "rewards/format_reward_func/std": 0.0, "step": 190, "step_time": 19.032863819692285 }, { "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": 0.0125, "completions/max_length": 742.6, "completions/max_terminated_length": 740.7, "completions/mean_length": 240.3625, "completions/mean_terminated_length": 221.21607666015626, "completions/min_length": 10.3, "completions/min_terminated_length": 10.3, "entropy": 0.35800585690885783, "epoch": 0.4444444444444444, "frac_reward_zero_std": 0.75, "grad_norm": 0.1171875, "kl": 0.004671698746193442, "learning_rate": 5.9025957262528e-07, "loss": 0.02241768389940262, "num_tokens": 536945.0, "reward": 1.7, "reward_std": 0.37737098932266233, "rewards/accuracy_reward_func/mean": 0.725, "rewards/accuracy_reward_func/std": 0.3345772624015808, "rewards/format_reward_func/mean": 0.975, "rewards/format_reward_func/std": 0.07071067690849304, "step": 200, "step_time": 28.164707153476776 }, { "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": 554.7, "completions/max_terminated_length": 554.7, "completions/mean_length": 129.0125, "completions/mean_terminated_length": 129.0125, "completions/min_length": 10.7, "completions/min_terminated_length": 10.7, "entropy": 0.35905973901972177, "epoch": 0.4666666666666667, "frac_reward_zero_std": 0.5, "grad_norm": 0.50390625, "kl": 0.0016817269792369417, "learning_rate": 5.557344661031627e-07, "loss": -0.027326536178588868, "num_tokens": 557854.0, "reward": 1.6625, "reward_std": 0.4926281929016113, "rewards/accuracy_reward_func/mean": 0.6875, "rewards/accuracy_reward_func/std": 0.4364054501056671, "rewards/format_reward_func/mean": 0.975, "rewards/format_reward_func/std": 0.07071067690849304, "step": 210, "step_time": 21.530417766235768 }, { "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": 600.1, "completions/max_terminated_length": 600.1, "completions/mean_length": 168.9875, "completions/mean_terminated_length": 168.9875, "completions/min_length": 12.3, "completions/min_terminated_length": 12.3, "entropy": 0.4736802303697914, "epoch": 0.4888888888888889, "frac_reward_zero_std": 0.75, "grad_norm": 0.126953125, "kl": 0.001979434840226446, "learning_rate": 5.209378268645997e-07, "loss": -0.03603552579879761, "num_tokens": 582233.0, "reward": 1.725, "reward_std": 0.3101575791835785, "rewards/accuracy_reward_func/mean": 0.725, "rewards/accuracy_reward_func/std": 0.31015759110450747, "rewards/format_reward_func/mean": 1.0, "rewards/format_reward_func/std": 0.0, "step": 220, "step_time": 23.071198928635567 }, { "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": 0.025, "completions/max_length": 723.2, "completions/max_terminated_length": 645.5, "completions/mean_length": 215.75, "completions/mean_terminated_length": 174.425, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "entropy": 0.4890560990199447, "epoch": 0.5111111111111111, "frac_reward_zero_std": 0.6, "grad_norm": 5.09375, "kl": 0.003485281078062741, "learning_rate": 4.860391806382156e-07, "loss": 0.11066329479217529, "num_tokens": 610901.0, "reward": 1.7375, "reward_std": 0.32810779809951784, "rewards/accuracy_reward_func/mean": 0.7625, "rewards/accuracy_reward_func/std": 0.2929195284843445, "rewards/format_reward_func/mean": 0.975, "rewards/format_reward_func/std": 0.046291005611419675, "step": 230, "step_time": 27.49671770753339 }, { "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": 0.0125, "completions/max_length": 521.2, "completions/max_terminated_length": 394.2, "completions/mean_length": 108.225, "completions/mean_terminated_length": 85.31785736083984, "completions/min_length": 11.1, "completions/min_terminated_length": 11.1, "entropy": 0.3478309240192175, "epoch": 0.5333333333333333, "frac_reward_zero_std": 0.5, "grad_norm": 0.94140625, "kl": 0.0021312192547497943, "learning_rate": 4.512085501204253e-07, "loss": -0.029661089181900024, "num_tokens": 629719.0, "reward": 1.6375, "reward_std": 0.41708244681358336, "rewards/accuracy_reward_func/mean": 0.6625, "rewards/accuracy_reward_func/std": 0.37917094230651854, "rewards/format_reward_func/mean": 0.975, "rewards/format_reward_func/std": 0.07071067690849304, "step": 240, "step_time": 20.277328117284924 }, { "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": 0.0125, "completions/max_length": 895.5, "completions/max_terminated_length": 734.6, "completions/mean_length": 298.4, "completions/mean_terminated_length": 275.63392944335936, "completions/min_length": 26.8, "completions/min_terminated_length": 26.8, "entropy": 0.4690596304833889, "epoch": 0.5555555555555556, "frac_reward_zero_std": 0.45, "grad_norm": 12.625, "kl": 0.0021533174171267434, "learning_rate": 4.166156266419489e-07, "loss": -0.16938039064407348, "num_tokens": 664259.0, "reward": 1.6125, "reward_std": 0.5434043407440186, "rewards/accuracy_reward_func/mean": 0.65, "rewards/accuracy_reward_func/std": 0.4843781113624573, "rewards/format_reward_func/mean": 0.9625, "rewards/format_reward_func/std": 0.0816463440656662, "step": 250, "step_time": 33.734226391091944 }, { "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": 0.0125, "completions/max_length": 741.6, "completions/max_terminated_length": 679.4, "completions/mean_length": 193.675, "completions/mean_terminated_length": 173.20714416503907, "completions/min_length": 11.8, "completions/min_terminated_length": 11.8, "entropy": 0.6225330086424947, "epoch": 0.5777777777777777, "frac_reward_zero_std": 0.6, "grad_norm": 3.890625, "kl": 0.017426777686920046, "learning_rate": 3.8242894344870495e-07, "loss": 0.038351666927337644, "num_tokens": 690625.0, "reward": 1.5375, "reward_std": 0.35561010539531707, "rewards/accuracy_reward_func/mean": 0.55, "rewards/accuracy_reward_func/std": 0.3202547788619995, "rewards/format_reward_func/mean": 0.9875, "rewards/format_reward_func/std": 0.03535533845424652, "step": 260, "step_time": 28.100541374552996 }, { "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": 528.1, "completions/max_terminated_length": 528.1, "completions/mean_length": 151.1375, "completions/mean_terminated_length": 151.1375, "completions/min_length": 10.9, "completions/min_terminated_length": 10.9, "entropy": 0.41103129191324117, "epoch": 0.6, "frac_reward_zero_std": 0.65, "grad_norm": 19.125, "kl": 0.0025848565255216726, "learning_rate": 3.488150546247778e-07, "loss": -0.0187175914645195, "num_tokens": 713576.0, "reward": 1.7, "reward_std": 0.4127540558576584, "rewards/accuracy_reward_func/mean": 0.7125, "rewards/accuracy_reward_func/std": 0.384642693400383, "rewards/format_reward_func/mean": 0.9875, "rewards/format_reward_func/std": 0.03535533845424652, "step": 270, "step_time": 20.493615136761218 }, { "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": 432.0, "completions/max_terminated_length": 432.0, "completions/mean_length": 123.0625, "completions/mean_terminated_length": 123.0625, "completions/min_length": 11.7, "completions/min_terminated_length": 11.7, "entropy": 0.336948611587286, "epoch": 0.6222222222222222, "frac_reward_zero_std": 0.5, "grad_norm": 0.20703125, "kl": 0.008485621147895018, "learning_rate": 3.15937723657661e-07, "loss": 0.037566670775413515, "num_tokens": 734029.0, "reward": 1.775, "reward_std": 0.3610740154981613, "rewards/accuracy_reward_func/mean": 0.775, "rewards/accuracy_reward_func/std": 0.36107403337955474, "rewards/format_reward_func/mean": 1.0, "rewards/format_reward_func/std": 0.0, "step": 280, "step_time": 17.11939077787101 }, { "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": 0.025, "completions/max_length": 720.1, "completions/max_terminated_length": 317.9, "completions/mean_length": 155.725, "completions/mean_terminated_length": 104.97142877578736, "completions/min_length": 11.8, "completions/min_terminated_length": 11.8, "entropy": 0.4226549585349858, "epoch": 0.6444444444444445, "frac_reward_zero_std": 0.65, "grad_norm": 9.625, "kl": 0.00767935578212473, "learning_rate": 2.8395712559900874e-07, "loss": 0.25207085609436036, "num_tokens": 757547.0, "reward": 1.575, "reward_std": 0.5132734388113022, "rewards/accuracy_reward_func/mean": 0.6125, "rewards/accuracy_reward_func/std": 0.445256182551384, "rewards/format_reward_func/mean": 0.9625, "rewards/format_reward_func/std": 0.10606601536273956, "step": 290, "step_time": 27.44823620794341 }, { "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": 551.9, "completions/max_terminated_length": 551.9, "completions/mean_length": 178.525, "completions/mean_terminated_length": 178.525, "completions/min_length": 14.1, "completions/min_terminated_length": 14.1, "entropy": 0.3967325270175934, "epoch": 0.6666666666666666, "frac_reward_zero_std": 0.6, "grad_norm": 0.5859375, "kl": 0.0035352140832515034, "learning_rate": 2.530290667078846e-07, "loss": -0.0781548261642456, "num_tokens": 782485.0, "reward": 1.7375, "reward_std": 0.3946571469306946, "rewards/accuracy_reward_func/mean": 0.75, "rewards/accuracy_reward_func/std": 0.3665457844734192, "rewards/format_reward_func/mean": 0.9875, "rewards/format_reward_func/std": 0.03535533845424652, "step": 300, "step_time": 21.438914876710623 }, { "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": 0.025, "completions/max_length": 940.3, "completions/max_terminated_length": 594.3, "completions/mean_length": 217.0875, "completions/mean_terminated_length": 169.35892868041992, "completions/min_length": 12.9, "completions/min_terminated_length": 12.9, "entropy": 0.41476303837262096, "epoch": 0.6888888888888889, "frac_reward_zero_std": 0.65, "grad_norm": 0.3515625, "kl": 0.003116685965650845, "learning_rate": 2.23304225378328e-07, "loss": 0.08113812804222106, "num_tokens": 810456.0, "reward": 1.65, "reward_std": 0.45866142213344574, "rewards/accuracy_reward_func/mean": 0.6875, "rewards/accuracy_reward_func/std": 0.384642693400383, "rewards/format_reward_func/mean": 0.9625, "rewards/format_reward_func/std": 0.10606601536273956, "step": 310, "step_time": 35.4386817724444 }, { "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": 0.0375, "completions/max_length": 1016.5, "completions/max_terminated_length": 661.5, "completions/mean_length": 332.925, "completions/mean_terminated_length": 269.355362033844, "completions/min_length": 11.1, "completions/min_terminated_length": 11.1, "entropy": 0.43934575584717095, "epoch": 0.7111111111111111, "frac_reward_zero_std": 0.45, "grad_norm": 19.875, "kl": 0.00665290072620337, "learning_rate": 1.9492741804936618e-07, "loss": 0.13840909004211427, "num_tokens": 848386.0, "reward": 1.575, "reward_std": 0.5897251844406128, "rewards/accuracy_reward_func/mean": 0.6375, "rewards/accuracy_reward_func/std": 0.4717529535293579, "rewards/format_reward_func/mean": 0.9375, "rewards/format_reward_func/std": 0.1767766922712326, "step": 320, "step_time": 38.41773249600082 }, { "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": 0.0125, "completions/max_length": 589.8, "completions/max_terminated_length": 415.4, "completions/mean_length": 143.8125, "completions/mean_terminated_length": 119.04821472167968, "completions/min_length": 11.9, "completions/min_terminated_length": 11.9, "entropy": 0.5242078335955739, "epoch": 0.7333333333333333, "frac_reward_zero_std": 0.55, "grad_norm": 7.90625, "kl": 0.006010404983805983, "learning_rate": 1.6803689367387918e-07, "loss": -0.01796160191297531, "num_tokens": 871079.0, "reward": 1.625, "reward_std": 0.47253691256046293, "rewards/accuracy_reward_func/mean": 0.6375, "rewards/accuracy_reward_func/std": 0.45158677995204927, "rewards/format_reward_func/mean": 0.9875, "rewards/format_reward_func/std": 0.03535533845424652, "step": 330, "step_time": 22.92032507658005 }, { "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": 0.025, "completions/max_length": 621.5, "completions/max_terminated_length": 326.6, "completions/mean_length": 146.0125, "completions/mean_terminated_length": 99.35535984039306, "completions/min_length": 13.7, "completions/min_terminated_length": 13.7, "entropy": 0.4421461643651128, "epoch": 0.7555555555555555, "frac_reward_zero_std": 0.5, "grad_norm": 39.75, "kl": 0.0067155171589301345, "learning_rate": 1.4276366018359842e-07, "loss": 0.22347230911254884, "num_tokens": 892896.0, "reward": 1.75, "reward_std": 0.3668150842189789, "rewards/accuracy_reward_func/mean": 0.7875, "rewards/accuracy_reward_func/std": 0.2722664475440979, "rewards/format_reward_func/mean": 0.9625, "rewards/format_reward_func/std": 0.10606601536273956, "step": 340, "step_time": 23.91549188606441 }, { "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": 0.0125, "completions/max_length": 680.1, "completions/max_terminated_length": 477.5, "completions/mean_length": 148.5625, "completions/mean_terminated_length": 123.15714292526245, "completions/min_length": 11.6, "completions/min_terminated_length": 11.6, "entropy": 0.39340304215438665, "epoch": 0.7777777777777778, "frac_reward_zero_std": 0.7, "grad_norm": 0.052734375, "kl": 0.008201900802868067, "learning_rate": 1.192308462316317e-07, "loss": 0.015751588344573974, "num_tokens": 916161.0, "reward": 1.4875, "reward_std": 0.22220884561538695, "rewards/accuracy_reward_func/mean": 0.5, "rewards/accuracy_reward_func/std": 0.20411194264888763, "rewards/format_reward_func/mean": 0.9875, "rewards/format_reward_func/std": 0.03535533845424652, "step": 350, "step_time": 26.149945951718838 }, { "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": 0.0125, "completions/max_length": 812.1, "completions/max_terminated_length": 633.3, "completions/mean_length": 248.6125, "completions/mean_terminated_length": 224.17857208251954, "completions/min_length": 30.9, "completions/min_terminated_length": 30.9, "entropy": 0.42323864391073585, "epoch": 0.8, "frac_reward_zero_std": 0.75, "grad_norm": 0.294921875, "kl": 0.003600113800894178, "learning_rate": 9.755310132204297e-08, "loss": 0.09096987247467041, "num_tokens": 946826.0, "reward": 1.7, "reward_std": 0.33229822516441343, "rewards/accuracy_reward_func/mean": 0.7125, "rewards/accuracy_reward_func/std": 0.3084602653980255, "rewards/format_reward_func/mean": 0.9875, "rewards/format_reward_func/std": 0.03535533845424652, "step": 360, "step_time": 30.848398898541927 }, { "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": 0.0125, "completions/max_length": 507.7, "completions/max_terminated_length": 331.7, "completions/mean_length": 167.85, "completions/mean_terminated_length": 144.32142944335936, "completions/min_length": 37.6, "completions/min_terminated_length": 37.6, "entropy": 0.5191565339453519, "epoch": 0.8222222222222222, "frac_reward_zero_std": 0.5, "grad_norm": 42.25, "kl": 0.0029767929230956726, "learning_rate": 7.783603724899257e-08, "loss": 0.07828059196472167, "num_tokens": 971274.0, "reward": 1.6, "reward_std": 0.43631487488746645, "rewards/accuracy_reward_func/mean": 0.6125, "rewards/accuracy_reward_func/std": 0.4082035183906555, "rewards/format_reward_func/mean": 0.9875, "rewards/format_reward_func/std": 0.03535533845424652, "step": 370, "step_time": 20.00939663061872 }, { "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": 210.9, "completions/max_terminated_length": 210.9, "completions/mean_length": 87.075, "completions/mean_terminated_length": 87.075, "completions/min_length": 14.3, "completions/min_terminated_length": 14.3, "entropy": 0.3288654647767544, "epoch": 0.8444444444444444, "frac_reward_zero_std": 0.8, "grad_norm": 0.71875, "kl": 0.0030142821157852495, "learning_rate": 6.017571356669182e-08, "loss": 0.022350619733333587, "num_tokens": 988584.0, "reward": 1.8625, "reward_std": 0.15149816870689392, "rewards/accuracy_reward_func/mean": 0.8625, "rewards/accuracy_reward_func/std": 0.1514981746673584, "rewards/format_reward_func/mean": 1.0, "rewards/format_reward_func/std": 0.0, "step": 380, "step_time": 9.370467808935791 }, { "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": 0.0125, "completions/max_length": 485.1, "completions/max_terminated_length": 418.8, "completions/mean_length": 151.8125, "completions/mean_terminated_length": 130.91071472167968, "completions/min_length": 10.9, "completions/min_terminated_length": 10.9, "entropy": 0.41746671898290516, "epoch": 0.8666666666666667, "frac_reward_zero_std": 0.65, "grad_norm": 14.75, "kl": 0.006796541407470613, "learning_rate": 4.465816959691149e-08, "loss": -0.13778045177459716, "num_tokens": 1011545.0, "reward": 1.6125, "reward_std": 0.3564248263835907, "rewards/accuracy_reward_func/mean": 0.6375, "rewards/accuracy_reward_func/std": 0.3147908627986908, "rewards/format_reward_func/mean": 0.975, "rewards/format_reward_func/std": 0.07071067690849304, "step": 390, "step_time": 19.206426683813334 }, { "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": 433.1, "completions/max_terminated_length": 433.1, "completions/mean_length": 130.625, "completions/mean_terminated_length": 130.625, "completions/min_length": 11.6, "completions/min_terminated_length": 11.6, "entropy": 0.46799973752349616, "epoch": 0.8888888888888888, "frac_reward_zero_std": 0.75, "grad_norm": 0.76171875, "kl": 0.0035236999536151004, "learning_rate": 3.135900525405427e-08, "loss": -0.03830570280551911, "num_tokens": 1032867.0, "reward": 1.5875, "reward_std": 0.345512917637825, "rewards/accuracy_reward_func/mean": 0.5875, "rewards/accuracy_reward_func/std": 0.34551292955875396, "rewards/format_reward_func/mean": 1.0, "rewards/format_reward_func/std": 0.0, "step": 400, "step_time": 17.119698034226893 }, { "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": 0.0125, "completions/max_length": 839.7, "completions/max_terminated_length": 758.7, "completions/mean_length": 268.6875, "completions/mean_terminated_length": 247.98035888671876, "completions/min_length": 21.7, "completions/min_terminated_length": 21.7, "entropy": 0.42415032302960753, "epoch": 0.9111111111111111, "frac_reward_zero_std": 0.55, "grad_norm": 0.0274658203125, "kl": 0.0035217325345570315, "learning_rate": 2.0343012729971243e-08, "loss": -0.016824692487716675, "num_tokens": 1065190.0, "reward": 1.5375, "reward_std": 0.4409481525421143, "rewards/accuracy_reward_func/mean": 0.5625, "rewards/accuracy_reward_func/std": 0.4010422766208649, "rewards/format_reward_func/mean": 0.975, "rewards/format_reward_func/std": 0.07071067690849304, "step": 410, "step_time": 31.805962641444058 }, { "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": 423.3, "completions/max_terminated_length": 423.3, "completions/mean_length": 105.2, "completions/mean_terminated_length": 105.2, "completions/min_length": 13.9, "completions/min_terminated_length": 13.9, "entropy": 0.4501968648750335, "epoch": 0.9333333333333333, "frac_reward_zero_std": 0.6, "grad_norm": 0.77734375, "kl": 0.0039933641563425935, "learning_rate": 1.166386083291604e-08, "loss": -0.00834755301475525, "num_tokens": 1085050.0, "reward": 1.6375, "reward_std": 0.3918039232492447, "rewards/accuracy_reward_func/mean": 0.6375, "rewards/accuracy_reward_func/std": 0.39180393517017365, "rewards/format_reward_func/mean": 1.0, "rewards/format_reward_func/std": 0.0, "step": 420, "step_time": 16.97432085731998 }, { "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": 0.0125, "completions/max_length": 580.5, "completions/max_terminated_length": 377.9, "completions/mean_length": 117.6625, "completions/mean_terminated_length": 92.25357151031494, "completions/min_length": 11.6, "completions/min_terminated_length": 11.6, "entropy": 0.4956971381790936, "epoch": 0.9555555555555556, "frac_reward_zero_std": 0.55, "grad_norm": 0.060302734375, "kl": 0.0021189929543652397, "learning_rate": 5.3638335185058335e-09, "loss": 0.16330193281173705, "num_tokens": 1106215.0, "reward": 1.65, "reward_std": 0.43717372715473174, "rewards/accuracy_reward_func/mean": 0.675, "rewards/accuracy_reward_func/std": 0.3808682680130005, "rewards/format_reward_func/mean": 0.975, "rewards/format_reward_func/std": 0.07071067690849304, "step": 430, "step_time": 22.55132986428216 }, { "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": 0.0125, "completions/max_length": 603.6, "completions/max_terminated_length": 467.1, "completions/mean_length": 169.625, "completions/mean_terminated_length": 148.05535888671875, "completions/min_length": 12.9, "completions/min_terminated_length": 12.9, "entropy": 0.4719172019511461, "epoch": 0.9777777777777777, "frac_reward_zero_std": 0.6, "grad_norm": 2.375, "kl": 0.003301422532410925, "learning_rate": 1.4736238865398765e-09, "loss": 0.14782559871673584, "num_tokens": 1131041.0, "reward": 1.5125, "reward_std": 0.5642522245645523, "rewards/accuracy_reward_func/mean": 0.55, "rewards/accuracy_reward_func/std": 0.4898498624563217, "rewards/format_reward_func/mean": 0.9625, "rewards/format_reward_func/std": 0.0816463440656662, "step": 440, "step_time": 23.233798024896533 }, { "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": 364.2, "completions/max_terminated_length": 364.2, "completions/mean_length": 111.775, "completions/mean_terminated_length": 111.775, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "entropy": 0.40269730454310776, "epoch": 1.0, "frac_reward_zero_std": 0.55, "grad_norm": 2.265625, "kl": 0.006289730004789362, "learning_rate": 1.2184647302626582e-11, "loss": 0.004454012960195541, "num_tokens": 1151755.0, "reward": 1.65, "reward_std": 0.43639759719371796, "rewards/accuracy_reward_func/mean": 0.65, "rewards/accuracy_reward_func/std": 0.4363976150751114, "rewards/format_reward_func/mean": 1.0, "rewards/format_reward_func/std": 0.0, "step": 450, "step_time": 14.730636461358518 } ], "logging_steps": 10, "max_steps": 450, "num_input_tokens_seen": 1151755, "num_train_epochs": 1, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }