{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.125, "eval_steps": 500, "global_step": 50, "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": 1346.0, "completions/max_terminated_length": 1346.0, "completions/mean_length": 438.984375, "completions/mean_terminated_length": 438.984375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0025, "frac_reward_zero_std": 0.125, "grad_norm": 5.347014904022217, "learning_rate": 0.0, "loss": -0.0, "num_tokens": 97919.0, "reward": 0.3414062261581421, "reward_std": 0.3232210874557495, "rewards/format_reward/mean": 0.4453125, "rewards/format_reward/std": 0.20152568817138672, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "step": 1 }, { "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": 1862.0, "completions/max_terminated_length": 1862.0, "completions/mean_length": 434.203125, "completions/mean_terminated_length": 434.203125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.005, "frac_reward_zero_std": 0.0, "grad_norm": 20.511768341064453, "learning_rate": 2e-07, "loss": 0.0, "num_tokens": 210508.0, "reward": 0.3695312440395355, "reward_std": 0.41905108094215393, "rewards/format_reward/mean": 0.4140625, "rewards/format_reward/std": 0.28942593932151794, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 2 }, { "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": 1350.0, "completions/max_terminated_length": 1350.0, "completions/mean_length": 351.984375, "completions/mean_terminated_length": 351.984375, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.0075, "frac_reward_zero_std": 0.0, "grad_norm": 41.398521423339844, "learning_rate": 4e-07, "loss": 0.0, "num_tokens": 325963.0, "reward": 0.28984373807907104, "reward_std": 0.3419463634490967, "rewards/format_reward/mean": 0.3984375, "rewards/format_reward/std": 0.3354656398296356, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 3 }, { "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": 1376.0, "completions/max_terminated_length": 1376.0, "completions/mean_length": 495.0, "completions/mean_terminated_length": 495.0, "completions/min_length": 10.0, "completions/min_terminated_length": 10.0, "epoch": 0.01, "frac_reward_zero_std": 0.0, "grad_norm": 8.007179260253906, "learning_rate": 6e-07, "loss": 0.0, "num_tokens": 440395.0, "reward": 0.26484373211860657, "reward_std": 0.24948760867118835, "rewards/format_reward/mean": 0.4609375, "rewards/format_reward/std": 0.24077731370925903, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "step": 4 }, { "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": 0.078125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1161.0, "completions/mean_length": 598.59375, "completions/mean_terminated_length": 475.7627258300781, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0125, "frac_reward_zero_std": 0.0, "grad_norm": 4.848970413208008, "learning_rate": 8e-07, "loss": -0.0, "num_tokens": 555537.0, "reward": 0.2679687440395355, "reward_std": 0.3987445831298828, "rewards/format_reward/mean": 0.3359375, "rewards/format_reward/std": 0.29620200395584106, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "step": 5 }, { "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1073.0, "completions/mean_length": 531.1875, "completions/mean_terminated_length": 507.11114501953125, "completions/min_length": 111.0, "completions/min_terminated_length": 111.0, "epoch": 0.015, "frac_reward_zero_std": 0.0, "grad_norm": 0.9525830149650574, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 666189.0, "reward": 0.3335937261581421, "reward_std": 0.34877651929855347, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.23974503576755524, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "step": 6 }, { "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": 1797.0, "completions/max_terminated_length": 1797.0, "completions/mean_length": 434.421875, "completions/mean_terminated_length": 434.421875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0175, "frac_reward_zero_std": 0.125, "grad_norm": 5.778356552124023, "learning_rate": 9.99726628670463e-07, "loss": -0.0, "num_tokens": 778024.0, "reward": 0.26484376192092896, "reward_std": 0.38700973987579346, "rewards/format_reward/mean": 0.4609375, "rewards/format_reward/std": 0.18483558297157288, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "step": 7 }, { "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": 2048.0, "completions/max_terminated_length": 854.0, "completions/mean_length": 400.171875, "completions/mean_terminated_length": 347.01611328125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.02, "frac_reward_zero_std": 0.0, "grad_norm": 9.09164810180664, "learning_rate": 9.989068136093872e-07, "loss": 0.0, "num_tokens": 876187.0, "reward": 0.17656248807907104, "reward_std": 0.26669788360595703, "rewards/format_reward/mean": 0.515625, "rewards/format_reward/std": 0.29504841566085815, "rewards/mcq_exact_match_reward/mean": 0.125, "rewards/mcq_exact_match_reward/std": 0.3333333432674408, "step": 8 }, { "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1625.0, "completions/mean_length": 734.171875, "completions/mean_terminated_length": 669.5573120117188, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0225, "frac_reward_zero_std": 0.0, "grad_norm": 7.612185001373291, "learning_rate": 9.975414512725056e-07, "loss": -0.0, "num_tokens": 1005886.0, "reward": 0.23046875, "reward_std": 0.3647712469100952, "rewards/format_reward/mean": 0.4296875, "rewards/format_reward/std": 0.23345555365085602, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "step": 9 }, { "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": 1873.0, "completions/max_terminated_length": 1873.0, "completions/mean_length": 379.625, "completions/mean_terminated_length": 379.625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.025, "frac_reward_zero_std": 0.0, "grad_norm": 24.19389533996582, "learning_rate": 9.956320346634875e-07, "loss": -0.0, "num_tokens": 1152838.0, "reward": 0.28515625, "reward_std": 0.36808687448501587, "rewards/format_reward/mean": 0.3515625, "rewards/format_reward/std": 0.2302463799715042, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 10 }, { "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": 2048.0, "completions/max_terminated_length": 1767.0, "completions/mean_length": 680.4375, "completions/mean_terminated_length": 636.3225708007812, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0275, "frac_reward_zero_std": 0.0, "grad_norm": 6.806507110595703, "learning_rate": 9.931806517013612e-07, "loss": 0.0, "num_tokens": 1321834.0, "reward": 0.2460937350988388, "reward_std": 0.32732176780700684, "rewards/format_reward/mean": 0.4296875, "rewards/format_reward/std": 0.19654127955436707, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "step": 11 }, { "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1104.0, "completions/mean_length": 421.5625, "completions/mean_terminated_length": 341.5737609863281, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.03, "frac_reward_zero_std": 0.0, "grad_norm": 17.674768447875977, "learning_rate": 9.901899829374047e-07, "loss": -0.0, "num_tokens": 1437710.0, "reward": 0.2617187201976776, "reward_std": 0.34293755888938904, "rewards/format_reward/mean": 0.4296875, "rewards/format_reward/std": 0.2498759627342224, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "step": 12 }, { "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1319.0, "completions/mean_length": 415.5, "completions/mean_terminated_length": 389.5873107910156, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0325, "frac_reward_zero_std": 0.0, "grad_norm": 13.535994529724121, "learning_rate": 9.866632986240029e-07, "loss": 0.0, "num_tokens": 1529142.0, "reward": 0.3570312559604645, "reward_std": 0.34329360723495483, "rewards/format_reward/mean": 0.4453125, "rewards/format_reward/std": 0.32265254855155945, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "step": 13 }, { "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": 1422.0, "completions/max_terminated_length": 1422.0, "completions/mean_length": 413.984375, "completions/mean_terminated_length": 413.984375, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.035, "frac_reward_zero_std": 0.375, "grad_norm": 21.664846420288086, "learning_rate": 9.826044551386742e-07, "loss": -0.0, "num_tokens": 1651269.0, "reward": 0.33906251192092896, "reward_std": 0.3009297549724579, "rewards/format_reward/mean": 0.421875, "rewards/format_reward/std": 0.2847827076911926, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "step": 14 }, { "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1574.0, "completions/mean_length": 384.921875, "completions/mean_terminated_length": 358.5238342285156, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0375, "frac_reward_zero_std": 0.125, "grad_norm": 34.72894287109375, "learning_rate": 9.780178907671788e-07, "loss": -0.0, "num_tokens": 1759752.0, "reward": 0.31562498211860657, "reward_std": 0.37008920311927795, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.2357022762298584, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "step": 15 }, { "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1989.0, "completions/mean_length": 531.640625, "completions/mean_terminated_length": 507.57147216796875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.04, "frac_reward_zero_std": 0.125, "grad_norm": 11.444356918334961, "learning_rate": 9.729086208503173e-07, "loss": 0.0, "num_tokens": 1894297.0, "reward": 0.25390625, "reward_std": 0.2943572700023651, "rewards/format_reward/mean": 0.5078125, "rewards/format_reward/std": 0.31487196683883667, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "step": 16 }, { "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": 869.0, "completions/max_terminated_length": 869.0, "completions/mean_length": 208.453125, "completions/mean_terminated_length": 208.453125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0425, "frac_reward_zero_std": 0.25, "grad_norm": 12.398904800415039, "learning_rate": 9.672822322997304e-07, "loss": 0.0, "num_tokens": 2022254.0, "reward": 0.5609375238418579, "reward_std": 0.22499367594718933, "rewards/format_reward/mean": 0.453125, "rewards/format_reward/std": 0.2630521357059479, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "step": 17 }, { "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": 2048.0, "completions/max_terminated_length": 1637.0, "completions/mean_length": 386.109375, "completions/mean_terminated_length": 332.5, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.045, "frac_reward_zero_std": 0.25, "grad_norm": 5.189897060394287, "learning_rate": 9.611448774886923e-07, "loss": 0.0, "num_tokens": 2154293.0, "reward": 0.2421874850988388, "reward_std": 0.12163974344730377, "rewards/format_reward/mean": 0.546875, "rewards/format_reward/std": 0.2777281701564789, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "step": 18 }, { "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": 0.078125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1639.0, "completions/mean_length": 353.609375, "completions/mean_terminated_length": 210.01695251464844, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0475, "frac_reward_zero_std": 0.375, "grad_norm": 9.442644119262695, "learning_rate": 9.545032675245813e-07, "loss": 0.0, "num_tokens": 2274284.0, "reward": 0.28984373807907104, "reward_std": 0.1657325029373169, "rewards/format_reward/mean": 0.5546875, "rewards/format_reward/std": 0.32265254855155945, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "step": 19 }, { "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1604.0, "completions/mean_length": 209.765625, "completions/mean_terminated_length": 180.58731079101562, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05, "frac_reward_zero_std": 0.375, "grad_norm": 18.869022369384766, "learning_rate": 9.473646649103817e-07, "loss": 0.0, "num_tokens": 2382253.0, "reward": 0.19843749701976776, "reward_std": 0.22571490705013275, "rewards/format_reward/mean": 0.578125, "rewards/format_reward/std": 0.23935678601264954, "rewards/mcq_exact_match_reward/mean": 0.140625, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "step": 20 }, { "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": 1305.0, "completions/max_terminated_length": 1305.0, "completions/mean_length": 160.828125, "completions/mean_terminated_length": 160.828125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0525, "frac_reward_zero_std": 0.375, "grad_norm": 10.548457145690918, "learning_rate": 9.397368756032444e-07, "loss": 0.0, "num_tokens": 2498002.0, "reward": 0.26953125, "reward_std": 0.25943902134895325, "rewards/format_reward/mean": 0.6640625, "rewards/format_reward/std": 0.2680720090866089, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "step": 21 }, { "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": 628.0, "completions/max_terminated_length": 628.0, "completions/mean_length": 71.828125, "completions/mean_terminated_length": 71.828125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.055, "frac_reward_zero_std": 0.375, "grad_norm": 28.167722702026367, "learning_rate": 9.316282404787869e-07, "loss": 0.0, "num_tokens": 2587023.0, "reward": 0.58984375, "reward_std": 0.24428340792655945, "rewards/format_reward/mean": 0.5859375, "rewards/format_reward/std": 0.209963858127594, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 22 }, { "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": 622.0, "completions/max_terminated_length": 622.0, "completions/mean_length": 49.421875, "completions/mean_terminated_length": 49.421875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0575, "frac_reward_zero_std": 0.125, "grad_norm": 21.865549087524414, "learning_rate": 9.230476262104676e-07, "loss": 0.0, "num_tokens": 2673410.0, "reward": 0.5265624523162842, "reward_std": 0.2945059835910797, "rewards/format_reward/mean": 0.578125, "rewards/format_reward/std": 0.25539806485176086, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 23 }, { "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 609.0, "completions/mean_length": 114.6875, "completions/mean_terminated_length": 84.00000762939453, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06, "frac_reward_zero_std": 0.25, "grad_norm": 17.985027313232422, "learning_rate": 9.1400441557401e-07, "loss": 0.0, "num_tokens": 2774182.0, "reward": 0.682812511920929, "reward_std": 0.2891410291194916, "rewards/format_reward/mean": 0.578125, "rewards/format_reward/std": 0.20351573824882507, "rewards/mcq_exact_match_reward/mean": 0.625, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 24 }, { "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 952.0, "completions/mean_length": 152.890625, "completions/mean_terminated_length": 122.80953216552734, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0625, "frac_reward_zero_std": 0.0, "grad_norm": 27.75064468383789, "learning_rate": 9.045084971874737e-07, "loss": -0.0, "num_tokens": 2894279.0, "reward": 0.5187499523162842, "reward_std": 0.3897084593772888, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.2182178944349289, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 25 }, { "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": 810.0, "completions/max_terminated_length": 810.0, "completions/mean_length": 65.015625, "completions/mean_terminated_length": 65.015625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.065, "frac_reward_zero_std": 0.625, "grad_norm": 10.022387504577637, "learning_rate": 8.945702546981968e-07, "loss": 0.0, "num_tokens": 3004024.0, "reward": 0.4335937201976776, "reward_std": 0.17460909485816956, "rewards/format_reward/mean": 0.5859375, "rewards/format_reward/std": 0.2280818521976471, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 26 }, { "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": 509.0, "completions/max_terminated_length": 509.0, "completions/mean_length": 13.859375, "completions/mean_terminated_length": 13.859375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0675, "frac_reward_zero_std": 0.5, "grad_norm": 20.696189880371094, "learning_rate": 8.842005554284295e-07, "loss": 0.0, "num_tokens": 3095935.0, "reward": 0.36249998211860657, "reward_std": 0.19727616012096405, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "step": 27 }, { "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": 835.0, "completions/max_terminated_length": 835.0, "completions/mean_length": 41.390625, "completions/mean_terminated_length": 41.390625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07, "frac_reward_zero_std": 0.375, "grad_norm": 23.081401824951172, "learning_rate": 8.734107384920769e-07, "loss": 0.0, "num_tokens": 3192760.0, "reward": 0.649218738079071, "reward_std": 0.20237906277179718, "rewards/format_reward/mean": 0.5546875, "rewards/format_reward/std": 0.15728822350502014, "rewards/mcq_exact_match_reward/mean": 0.59375, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 28 }, { "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": 1023.0, "completions/max_terminated_length": 1023.0, "completions/mean_length": 113.1875, "completions/mean_terminated_length": 113.1875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0725, "frac_reward_zero_std": 0.375, "grad_norm": 16.17728042602539, "learning_rate": 8.622126023955445e-07, "loss": -0.0, "num_tokens": 3273764.0, "reward": 0.48750001192092896, "reward_std": 0.2276211977005005, "rewards/format_reward/mean": 0.65625, "rewards/format_reward/std": 0.233588308095932, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 29 }, { "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": 754.0, "completions/max_terminated_length": 754.0, "completions/mean_length": 110.40625, "completions/mean_terminated_length": 110.40625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.075, "frac_reward_zero_std": 0.25, "grad_norm": 13.023463249206543, "learning_rate": 8.506183921362442e-07, "loss": -0.0, "num_tokens": 3356478.0, "reward": 0.3578124940395355, "reward_std": 0.2516772747039795, "rewards/format_reward/mean": 0.609375, "rewards/format_reward/std": 0.2592533528804779, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "step": 30 }, { "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": 1034.0, "completions/max_terminated_length": 1034.0, "completions/mean_length": 65.1875, "completions/mean_terminated_length": 65.1875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0775, "frac_reward_zero_std": 0.375, "grad_norm": 10.686674118041992, "learning_rate": 8.386407858128706e-07, "loss": -0.0, "num_tokens": 3461266.0, "reward": 0.3671875, "reward_std": 0.23324333131313324, "rewards/format_reward/mean": 0.703125, "rewards/format_reward/std": 0.2630521357059479, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "step": 31 }, { "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": 624.0, "completions/max_terminated_length": 624.0, "completions/mean_length": 89.0625, "completions/mean_terminated_length": 89.0625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08, "frac_reward_zero_std": 0.125, "grad_norm": 23.29245948791504, "learning_rate": 8.262928807620843e-07, "loss": 0.0, "num_tokens": 3549766.0, "reward": 0.766406238079071, "reward_std": 0.2440863996744156, "rewards/format_reward/mean": 0.7890625, "rewards/format_reward/std": 0.2643453776836395, "rewards/mcq_exact_match_reward/mean": 0.6875, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "step": 32 }, { "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": 329.0, "completions/max_terminated_length": 329.0, "completions/mean_length": 41.53125, "completions/mean_terminated_length": 41.53125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0825, "frac_reward_zero_std": 0.25, "grad_norm": 32.976783752441406, "learning_rate": 8.135881792367685e-07, "loss": 0.0, "num_tokens": 3646184.0, "reward": 0.53125, "reward_std": 0.22289422154426575, "rewards/format_reward/mean": 0.78125, "rewards/format_reward/std": 0.25, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 33 }, { "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": 421.0, "completions/max_terminated_length": 421.0, "completions/mean_length": 29.8125, "completions/mean_terminated_length": 29.8125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.085, "frac_reward_zero_std": 0.25, "grad_norm": 19.8100643157959, "learning_rate": 8.005405736415125e-07, "loss": 0.0, "num_tokens": 3729868.0, "reward": 0.604687511920929, "reward_std": 0.19311955571174622, "rewards/format_reward/mean": 0.890625, "rewards/format_reward/std": 0.2083333432674408, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "step": 34 }, { "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": 594.0, "completions/max_terminated_length": 594.0, "completions/mean_length": 53.390625, "completions/mean_terminated_length": 53.390625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0875, "frac_reward_zero_std": 0.25, "grad_norm": 11.324256896972656, "learning_rate": 7.871643313414718e-07, "loss": 0.0, "num_tokens": 3810837.0, "reward": 0.5648437738418579, "reward_std": 0.3285239338874817, "rewards/format_reward/mean": 0.8046875, "rewards/format_reward/std": 0.24587368965148926, "rewards/mcq_exact_match_reward/mean": 0.484375, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "step": 35 }, { "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": 531.0, "completions/max_terminated_length": 531.0, "completions/mean_length": 30.546875, "completions/mean_terminated_length": 30.546875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.09, "frac_reward_zero_std": 0.375, "grad_norm": 27.54176139831543, "learning_rate": 7.734740790612136e-07, "loss": -0.0, "num_tokens": 3901168.0, "reward": 0.4476562738418579, "reward_std": 0.1889566034078598, "rewards/format_reward/mean": 0.8828125, "rewards/format_reward/std": 0.2313210517168045, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "step": 36 }, { "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 338.0, "completions/mean_length": 49.328125, "completions/mean_terminated_length": 17.60317611694336, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0925, "frac_reward_zero_std": 0.375, "grad_norm": 10.54279899597168, "learning_rate": 7.594847868906076e-07, "loss": -0.0, "num_tokens": 4011277.0, "reward": 0.4429687559604645, "reward_std": 0.21485351026058197, "rewards/format_reward/mean": 0.8359375, "rewards/format_reward/std": 0.2680720090866089, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "step": 37 }, { "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": 242.0, "completions/max_terminated_length": 242.0, "completions/mean_length": 28.359375, "completions/mean_terminated_length": 28.359375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.095, "frac_reward_zero_std": 0.5, "grad_norm": 9.654556274414062, "learning_rate": 7.452117519152541e-07, "loss": 0.0, "num_tokens": 4093108.0, "reward": 0.515625, "reward_std": 0.23144522309303284, "rewards/format_reward/mean": 0.9375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 38 }, { "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": 375.0, "completions/max_terminated_length": 375.0, "completions/mean_length": 30.09375, "completions/mean_terminated_length": 30.09375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0975, "frac_reward_zero_std": 0.375, "grad_norm": 31.476085662841797, "learning_rate": 7.306705814893439e-07, "loss": 0.0, "num_tokens": 4188306.0, "reward": 0.44921875, "reward_std": 0.2185998558998108, "rewards/format_reward/mean": 0.8984375, "rewards/format_reward/std": 0.22146137058734894, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "step": 39 }, { "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": 200.0, "completions/max_terminated_length": 200.0, "completions/mean_length": 26.671875, "completions/mean_terminated_length": 26.671875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1, "frac_reward_zero_std": 0.75, "grad_norm": 7.092618465423584, "learning_rate": 7.158771761692464e-07, "loss": 0.0, "num_tokens": 4278861.0, "reward": 0.5687500238418579, "reward_std": 0.10888782143592834, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 40 }, { "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": 137.0, "completions/max_terminated_length": 137.0, "completions/mean_length": 25.328125, "completions/mean_terminated_length": 25.328125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1025, "frac_reward_zero_std": 0.5, "grad_norm": 26.821311950683594, "learning_rate": 7.008477123264847e-07, "loss": 0.0, "num_tokens": 4369482.0, "reward": 0.4164062440395355, "reward_std": 0.15971112251281738, "rewards/format_reward/mean": 0.8828125, "rewards/format_reward/std": 0.31800705194473267, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 41 }, { "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": 67.0, "completions/max_terminated_length": 67.0, "completions/mean_length": 14.265625, "completions/mean_terminated_length": 14.265625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.105, "frac_reward_zero_std": 0.5, "grad_norm": 9.240446090698242, "learning_rate": 6.855986244591103e-07, "loss": -0.0, "num_tokens": 4456739.0, "reward": 0.2562500238418579, "reward_std": 0.22201895713806152, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "step": 42 }, { "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": 325.0, "completions/max_terminated_length": 325.0, "completions/mean_length": 21.859375, "completions/mean_terminated_length": 21.859375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1075, "frac_reward_zero_std": 0.5, "grad_norm": 17.005455017089844, "learning_rate": 6.701465872208216e-07, "loss": 0.0, "num_tokens": 4563602.0, "reward": 0.5765625238418579, "reward_std": 0.22431671619415283, "rewards/format_reward/mean": 0.921875, "rewards/format_reward/std": 0.20351573824882507, "rewards/mcq_exact_match_reward/mean": 0.484375, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "step": 43 }, { "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": 207.0, "completions/max_terminated_length": 207.0, "completions/mean_length": 15.0, "completions/mean_terminated_length": 15.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.11, "frac_reward_zero_std": 0.625, "grad_norm": 32.06852722167969, "learning_rate": 6.545084971874736e-07, "loss": 0.0, "num_tokens": 4639778.0, "reward": 0.3023437559604645, "reward_std": 0.12476710975170135, "rewards/format_reward/mean": 0.9921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "step": 44 }, { "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": 50.0, "completions/max_terminated_length": 50.0, "completions/mean_length": 13.453125, "completions/mean_terminated_length": 13.453125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1125, "frac_reward_zero_std": 0.875, "grad_norm": 3.3413596153259277, "learning_rate": 6.387014543809223e-07, "loss": -0.0, "num_tokens": 4725151.0, "reward": 0.3656250238418579, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "step": 45 }, { "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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.03125, "completions/mean_terminated_length": 12.03125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.115, "frac_reward_zero_std": 0.625, "grad_norm": 11.85780143737793, "learning_rate": 6.227427435703995e-07, "loss": 0.0, "num_tokens": 4813545.0, "reward": 0.4125000238418579, "reward_std": 0.1462520956993103, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "step": 46 }, { "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": 19.0, "completions/max_terminated_length": 19.0, "completions/mean_length": 12.109375, "completions/mean_terminated_length": 12.109375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1175, "frac_reward_zero_std": 0.5, "grad_norm": 10.877081871032715, "learning_rate": 6.066498153718734e-07, "loss": 0.0, "num_tokens": 4923528.0, "reward": 0.4437500238418579, "reward_std": 0.2041158527135849, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "step": 47 }, { "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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.015625, "completions/mean_terminated_length": 12.015625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.12, "frac_reward_zero_std": 0.5, "grad_norm": 11.632145881652832, "learning_rate": 5.90440267166055e-07, "loss": 0.0, "num_tokens": 5008537.0, "reward": 0.5828125476837158, "reward_std": 0.21179001033306122, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.484375, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "step": 48 }, { "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": 232.0, "completions/max_terminated_length": 232.0, "completions/mean_length": 21.515625, "completions/mean_terminated_length": 21.515625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1225, "frac_reward_zero_std": 0.5, "grad_norm": 15.242279052734375, "learning_rate": 5.741318238559209e-07, "loss": -0.0, "num_tokens": 5093242.0, "reward": 0.4671875238418579, "reward_std": 0.20908603072166443, "rewards/format_reward/mean": 0.921875, "rewards/format_reward/std": 0.27048972249031067, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 49 }, { "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": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 11.921875, "completions/mean_terminated_length": 11.921875, "completions/min_length": 7.0, "completions/min_terminated_length": 7.0, "epoch": 0.125, "frac_reward_zero_std": 0.5, "grad_norm": 16.313396453857422, "learning_rate": 5.577423184847931e-07, "loss": 0.0, "num_tokens": 5177245.0, "reward": 0.33281251788139343, "reward_std": 0.17800088226795197, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "step": 50 } ], "logging_steps": 1, "max_steps": 100, "num_input_tokens_seen": 5177245, "num_train_epochs": 1, "save_steps": 50, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }