{ "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": 1424.0, "completions/max_terminated_length": 1424.0, "completions/mean_length": 516.875, "completions/mean_terminated_length": 516.875, "completions/min_length": 15.0, "completions/min_terminated_length": 15.0, "epoch": 0.0025, "frac_reward_zero_std": 0.125, "grad_norm": 8.525991439819336, "learning_rate": 0.0, "loss": 0.0, "num_tokens": 115704.0, "reward": 0.1851562261581421, "reward_std": 0.29114237427711487, "rewards/format_reward/mean": 0.2890625, "rewards/format_reward/std": 0.3060849606990814, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 2038.0, "completions/mean_length": 548.1875, "completions/mean_terminated_length": 524.3809814453125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.005, "frac_reward_zero_std": 0.0, "grad_norm": 9.42932415008545, "learning_rate": 2e-07, "loss": 0.0, "num_tokens": 237500.0, "reward": 0.3624999523162842, "reward_std": 0.37413233518600464, "rewards/format_reward/mean": 0.34375, "rewards/format_reward/std": 0.279951810836792, "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": 1824.0, "completions/max_terminated_length": 1824.0, "completions/mean_length": 614.359375, "completions/mean_terminated_length": 614.359375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0075, "frac_reward_zero_std": 0.0, "grad_norm": 2.7854650020599365, "learning_rate": 4e-07, "loss": -0.0, "num_tokens": 383059.0, "reward": 0.41874998807907104, "reward_std": 0.48726803064346313, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1311.0, "completions/mean_length": 519.125, "completions/mean_terminated_length": 494.857177734375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.01, "frac_reward_zero_std": 0.0, "grad_norm": 12.556537628173828, "learning_rate": 6e-07, "loss": -0.0, "num_tokens": 518635.0, "reward": 0.36953121423721313, "reward_std": 0.44310489296913147, "rewards/format_reward/mean": 0.4140625, "rewards/format_reward/std": 0.2280818521976471, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 0.0, "completions/max_length": 1315.0, "completions/max_terminated_length": 1315.0, "completions/mean_length": 452.46875, "completions/mean_terminated_length": 452.46875, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.0125, "frac_reward_zero_std": 0.0, "grad_norm": 14.461836814880371, "learning_rate": 8e-07, "loss": -0.0, "num_tokens": 633593.0, "reward": 0.39921873807907104, "reward_std": 0.43056797981262207, "rewards/format_reward/mean": 0.3984375, "rewards/format_reward/std": 0.26991602778434753, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 0.0, "completions/max_length": 1837.0, "completions/max_terminated_length": 1837.0, "completions/mean_length": 508.78125, "completions/mean_terminated_length": 508.78125, "completions/min_length": 51.0, "completions/min_terminated_length": 51.0, "epoch": 0.015, "frac_reward_zero_std": 0.0, "grad_norm": 1.5115914344787598, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 744331.0, "reward": 0.27265623211860657, "reward_std": 0.3936590552330017, "rewards/format_reward/mean": 0.3828125, "rewards/format_reward/std": 0.2780628502368927, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1754.0, "completions/mean_length": 510.1875, "completions/mean_terminated_length": 485.7778015136719, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.0175, "frac_reward_zero_std": 0.0, "grad_norm": 8.98042106628418, "learning_rate": 9.99726628670463e-07, "loss": -0.0, "num_tokens": 858703.0, "reward": 0.19843748211860657, "reward_std": 0.34161263704299927, "rewards/format_reward/mean": 0.265625, "rewards/format_reward/std": 0.2812775671482086, "rewards/mcq_exact_match_reward/mean": 0.171875, "rewards/mcq_exact_match_reward/std": 0.38025420904159546, "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": 1470.0, "completions/mean_length": 633.484375, "completions/mean_terminated_length": 587.8547973632812, "completions/min_length": 10.0, "completions/min_terminated_length": 10.0, "epoch": 0.02, "frac_reward_zero_std": 0.0, "grad_norm": 8.4592924118042, "learning_rate": 9.989068136093872e-07, "loss": 0.0, "num_tokens": 996710.0, "reward": 0.1929687261581421, "reward_std": 0.28227367997169495, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.23974503576755524, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1708.0, "completions/mean_length": 504.515625, "completions/mean_terminated_length": 480.0158996582031, "completions/min_length": 19.0, "completions/min_terminated_length": 19.0, "epoch": 0.0225, "frac_reward_zero_std": 0.125, "grad_norm": 2.371509313583374, "learning_rate": 9.975414512725056e-07, "loss": -0.0, "num_tokens": 1114319.0, "reward": 0.22421875596046448, "reward_std": 0.21774323284626007, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.34780704975128174, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1534.0, "completions/mean_length": 667.671875, "completions/mean_terminated_length": 623.1451416015625, "completions/min_length": 19.0, "completions/min_terminated_length": 19.0, "epoch": 0.025, "frac_reward_zero_std": 0.0, "grad_norm": 3.2990942001342773, "learning_rate": 9.956320346634875e-07, "loss": -0.0, "num_tokens": 1258938.0, "reward": 0.36953121423721313, "reward_std": 0.39852094650268555, "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": 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": 0.0, "completions/max_length": 1851.0, "completions/max_terminated_length": 1851.0, "completions/mean_length": 634.984375, "completions/mean_terminated_length": 634.984375, "completions/min_length": 96.0, "completions/min_terminated_length": 96.0, "epoch": 0.0275, "frac_reward_zero_std": 0.0, "grad_norm": 0.9042181968688965, "learning_rate": 9.931806517013612e-07, "loss": 0.0, "num_tokens": 1395537.0, "reward": 0.32343748211860657, "reward_std": 0.358438640832901, "rewards/format_reward/mean": 0.421875, "rewards/format_reward/std": 0.23935678601264954, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "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": 0.0, "completions/max_length": 1947.0, "completions/max_terminated_length": 1947.0, "completions/mean_length": 606.46875, "completions/mean_terminated_length": 606.46875, "completions/min_length": 32.0, "completions/min_terminated_length": 32.0, "epoch": 0.03, "frac_reward_zero_std": 0.0, "grad_norm": 1.4510782957077026, "learning_rate": 9.901899829374047e-07, "loss": -0.0, "num_tokens": 1540367.0, "reward": 0.3195312023162842, "reward_std": 0.4172312021255493, "rewards/format_reward/mean": 0.3828125, "rewards/format_reward/std": 0.2634054720401764, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "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": 1893.0, "completions/mean_length": 720.796875, "completions/mean_terminated_length": 699.730224609375, "completions/min_length": 98.0, "completions/min_terminated_length": 98.0, "epoch": 0.0325, "frac_reward_zero_std": 0.0, "grad_norm": 1.063207745552063, "learning_rate": 9.866632986240029e-07, "loss": -0.0, "num_tokens": 1703218.0, "reward": 0.2593749761581421, "reward_std": 0.3210762143135071, "rewards/format_reward/mean": 0.40625, "rewards/format_reward/std": 0.25, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "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": 0.0625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1556.0, "completions/mean_length": 739.9375, "completions/mean_terminated_length": 652.7333374023438, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.035, "frac_reward_zero_std": 0.0, "grad_norm": 6.66965913772583, "learning_rate": 9.826044551386742e-07, "loss": 0.0, "num_tokens": 1852670.0, "reward": 0.18046875298023224, "reward_std": 0.20850324630737305, "rewards/format_reward/mean": 0.3984375, "rewards/format_reward/std": 0.25479042530059814, "rewards/mcq_exact_match_reward/mean": 0.140625, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1591.0, "completions/mean_length": 529.828125, "completions/mean_terminated_length": 480.8548278808594, "completions/min_length": 19.0, "completions/min_terminated_length": 19.0, "epoch": 0.0375, "frac_reward_zero_std": 0.25, "grad_norm": 3.0365588665008545, "learning_rate": 9.780178907671788e-07, "loss": -0.0, "num_tokens": 1956155.0, "reward": 0.6171875, "reward_std": 0.3435981273651123, "rewards/format_reward/mean": 0.546875, "rewards/format_reward/std": 0.3299681544303894, "rewards/mcq_exact_match_reward/mean": 0.5625, "rewards/mcq_exact_match_reward/std": 0.5, "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": 1282.0, "completions/mean_length": 476.328125, "completions/mean_terminated_length": 451.3809814453125, "completions/min_length": 20.0, "completions/min_terminated_length": 20.0, "epoch": 0.04, "frac_reward_zero_std": 0.0, "grad_norm": 2.078115940093994, "learning_rate": 9.729086208503173e-07, "loss": 0.0, "num_tokens": 2063920.0, "reward": 0.4468750059604645, "reward_std": 0.22694434225559235, "rewards/format_reward/mean": 0.5625, "rewards/format_reward/std": 0.3021090030670166, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1920.0, "completions/mean_length": 544.25, "completions/mean_terminated_length": 520.3809814453125, "completions/min_length": 19.0, "completions/min_terminated_length": 19.0, "epoch": 0.0425, "frac_reward_zero_std": 0.0, "grad_norm": 5.524308681488037, "learning_rate": 9.672822322997304e-07, "loss": 0.0, "num_tokens": 2209224.0, "reward": 0.2109375, "reward_std": 0.23218126595020294, "rewards/format_reward/mean": 0.546875, "rewards/format_reward/std": 0.3177144229412079, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "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": 0.0, "completions/max_length": 2044.0, "completions/max_terminated_length": 2044.0, "completions/mean_length": 467.859375, "completions/mean_terminated_length": 467.859375, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.045, "frac_reward_zero_std": 0.125, "grad_norm": 4.5600175857543945, "learning_rate": 9.611448774886923e-07, "loss": 0.0, "num_tokens": 2330519.0, "reward": 0.21015623211860657, "reward_std": 0.2244272232055664, "rewards/format_reward/mean": 0.5390625, "rewards/format_reward/std": 0.37059250473976135, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1852.0, "completions/mean_length": 474.671875, "completions/mean_terminated_length": 423.9193420410156, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0475, "frac_reward_zero_std": 0.125, "grad_norm": 9.626194953918457, "learning_rate": 9.545032675245813e-07, "loss": 0.0, "num_tokens": 2448962.0, "reward": 0.35078123211860657, "reward_std": 0.3747270107269287, "rewards/format_reward/mean": 0.5390625, "rewards/format_reward/std": 0.3249503970146179, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 0.078125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1599.0, "completions/mean_length": 508.1875, "completions/mean_terminated_length": 377.6949157714844, "completions/min_length": 17.0, "completions/min_terminated_length": 17.0, "epoch": 0.05, "frac_reward_zero_std": 0.125, "grad_norm": 5.258202075958252, "learning_rate": 9.473646649103817e-07, "loss": -0.0, "num_tokens": 2561582.0, "reward": 0.20234374701976776, "reward_std": 0.24121138453483582, "rewards/format_reward/mean": 0.6171875, "rewards/format_reward/std": 0.3752479553222656, "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": 0.109375, "completions/max_length": 2048.0, "completions/max_terminated_length": 2034.0, "completions/mean_length": 630.953125, "completions/mean_terminated_length": 456.9298400878906, "completions/min_length": 15.0, "completions/min_terminated_length": 15.0, "epoch": 0.0525, "frac_reward_zero_std": 0.0, "grad_norm": 7.231940269470215, "learning_rate": 9.397368756032444e-07, "loss": -0.0, "num_tokens": 2708651.0, "reward": 0.2953124940395355, "reward_std": 0.31727102398872375, "rewards/format_reward/mean": 0.609375, "rewards/format_reward/std": 0.38285425305366516, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1786.0, "completions/mean_length": 422.484375, "completions/mean_terminated_length": 396.68255615234375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.055, "frac_reward_zero_std": 0.0, "grad_norm": 8.500676155090332, "learning_rate": 9.316282404787869e-07, "loss": -0.0, "num_tokens": 2846146.0, "reward": 0.4664062559604645, "reward_std": 0.29066649079322815, "rewards/format_reward/mean": 0.6015625, "rewards/format_reward/std": 0.3354656398296356, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 1610.0, "completions/max_terminated_length": 1610.0, "completions/mean_length": 166.46875, "completions/mean_terminated_length": 166.46875, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.0575, "frac_reward_zero_std": 0.0, "grad_norm": 20.38620376586914, "learning_rate": 9.230476262104676e-07, "loss": 0.0, "num_tokens": 2938992.0, "reward": 0.4750000238418579, "reward_std": 0.37396490573883057, "rewards/format_reward/mean": 0.6875, "rewards/format_reward/std": 0.4225771427154541, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 0.078125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1886.0, "completions/mean_length": 471.84375, "completions/mean_terminated_length": 338.27117919921875, "completions/min_length": 16.0, "completions/min_terminated_length": 16.0, "epoch": 0.06, "frac_reward_zero_std": 0.25, "grad_norm": 12.495400428771973, "learning_rate": 9.1400441557401e-07, "loss": 0.0, "num_tokens": 3084518.0, "reward": 0.23906248807907104, "reward_std": 0.2132878452539444, "rewards/format_reward/mean": 0.671875, "rewards/format_reward/std": 0.4103690981864929, "rewards/mcq_exact_match_reward/mean": 0.171875, "rewards/mcq_exact_match_reward/std": 0.38025420904159546, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 806.0, "completions/mean_length": 178.953125, "completions/mean_terminated_length": 87.03278350830078, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0625, "frac_reward_zero_std": 0.125, "grad_norm": 11.406326293945312, "learning_rate": 9.045084971874737e-07, "loss": 0.0, "num_tokens": 3165979.0, "reward": 0.33125001192092896, "reward_std": 0.22604334354400635, "rewards/format_reward/mean": 0.8125, "rewards/format_reward/std": 0.36187344789505005, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 734.0, "completions/max_terminated_length": 734.0, "completions/mean_length": 56.59375, "completions/mean_terminated_length": 56.59375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.065, "frac_reward_zero_std": 0.125, "grad_norm": 14.60300064086914, "learning_rate": 8.945702546981968e-07, "loss": 0.0, "num_tokens": 3266065.0, "reward": 0.29999998211860657, "reward_std": 0.28624725341796875, "rewards/format_reward/mean": 0.8125, "rewards/format_reward/std": 0.3021090030670166, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "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": 674.0, "completions/max_terminated_length": 674.0, "completions/mean_length": 50.96875, "completions/mean_terminated_length": 50.96875, "completions/min_length": 16.0, "completions/min_terminated_length": 16.0, "epoch": 0.0675, "frac_reward_zero_std": 0.375, "grad_norm": 7.735217094421387, "learning_rate": 8.842005554284295e-07, "loss": 0.0, "num_tokens": 3365199.0, "reward": 0.4750000238418579, "reward_std": 0.2619796395301819, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 0.0625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1436.0, "completions/mean_length": 221.71875, "completions/mean_terminated_length": 99.9666748046875, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.07, "frac_reward_zero_std": 0.625, "grad_norm": 12.549527168273926, "learning_rate": 8.734107384920769e-07, "loss": 0.0, "num_tokens": 3470861.0, "reward": 0.21406251192092896, "reward_std": 0.13916602730751038, "rewards/format_reward/mean": 0.890625, "rewards/format_reward/std": 0.28824523091316223, "rewards/mcq_exact_match_reward/mean": 0.125, "rewards/mcq_exact_match_reward/std": 0.3333333432674408, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 680.0, "completions/mean_length": 91.875, "completions/mean_terminated_length": 60.825401306152344, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.0725, "frac_reward_zero_std": 0.375, "grad_norm": 12.2145357131958, "learning_rate": 8.622126023955445e-07, "loss": 0.0, "num_tokens": 3576205.0, "reward": 0.608593761920929, "reward_std": 0.18384604156017303, "rewards/format_reward/mean": 0.9296875, "rewards/format_reward/std": 0.21578919887542725, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "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": 584.0, "completions/max_terminated_length": 584.0, "completions/mean_length": 25.734375, "completions/mean_terminated_length": 25.734375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.075, "frac_reward_zero_std": 0.375, "grad_norm": 15.464653968811035, "learning_rate": 8.506183921362442e-07, "loss": 0.0, "num_tokens": 3660676.0, "reward": 0.6078125238418579, "reward_std": 0.2516558766365051, "rewards/format_reward/mean": 0.921875, "rewards/format_reward/std": 0.20351573824882507, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "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": 257.0, "completions/max_terminated_length": 257.0, "completions/mean_length": 28.265625, "completions/mean_terminated_length": 28.265625, "completions/min_length": 14.0, "completions/min_terminated_length": 14.0, "epoch": 0.0775, "frac_reward_zero_std": 0.375, "grad_norm": 13.753981590270996, "learning_rate": 8.386407858128706e-07, "loss": 0.0, "num_tokens": 3741293.0, "reward": 0.5328125357627869, "reward_std": 0.2937379479408264, "rewards/format_reward/mean": 0.953125, "rewards/format_reward/std": 0.21304203569889069, "rewards/mcq_exact_match_reward/mean": 0.4375, "rewards/mcq_exact_match_reward/std": 0.5, "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": 504.0, "completions/max_terminated_length": 504.0, "completions/mean_length": 32.15625, "completions/mean_terminated_length": 32.15625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.08, "frac_reward_zero_std": 0.375, "grad_norm": 13.3984375, "learning_rate": 8.262928807620843e-07, "loss": -0.0, "num_tokens": 3850007.0, "reward": 0.5179687738418579, "reward_std": 0.21698468923568726, "rewards/format_reward/mean": 0.9609375, "rewards/format_reward/std": 0.18483558297157288, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 313.0, "completions/max_terminated_length": 313.0, "completions/mean_length": 21.21875, "completions/mean_terminated_length": 21.21875, "completions/min_length": 11.0, "completions/min_terminated_length": 11.0, "epoch": 0.0825, "frac_reward_zero_std": 0.5, "grad_norm": 15.135663986206055, "learning_rate": 8.135881792367685e-07, "loss": -0.0, "num_tokens": 3921093.0, "reward": 0.41093751788139343, "reward_std": 0.2397041618824005, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 20.0, "completions/max_terminated_length": 20.0, "completions/mean_length": 16.65625, "completions/mean_terminated_length": 16.65625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.085, "frac_reward_zero_std": 0.5, "grad_norm": 15.264263153076172, "learning_rate": 8.005405736415125e-07, "loss": 0.0, "num_tokens": 3986967.0, "reward": 0.6765625476837158, "reward_std": 0.23586463928222656, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.578125, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 19.0, "completions/max_terminated_length": 19.0, "completions/mean_length": 16.546875, "completions/mean_terminated_length": 16.546875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0875, "frac_reward_zero_std": 0.5, "grad_norm": 7.794801235198975, "learning_rate": 7.871643313414718e-07, "loss": 0.0, "num_tokens": 4090786.0, "reward": 0.4281250238418579, "reward_std": 0.22673699259757996, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 19.0, "completions/max_terminated_length": 19.0, "completions/mean_length": 16.421875, "completions/mean_terminated_length": 16.421875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.09, "frac_reward_zero_std": 0.5, "grad_norm": 9.687009811401367, "learning_rate": 7.734740790612136e-07, "loss": 0.0, "num_tokens": 4167205.0, "reward": 0.7250000238418579, "reward_std": 0.2130674123764038, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.625, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 0.0, "completions/max_length": 313.0, "completions/max_terminated_length": 313.0, "completions/mean_length": 22.328125, "completions/mean_terminated_length": 22.328125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0925, "frac_reward_zero_std": 0.5, "grad_norm": 14.43371295928955, "learning_rate": 7.594847868906076e-07, "loss": -0.0, "num_tokens": 4276314.0, "reward": 0.5531250238418579, "reward_std": 0.19939783215522766, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 454.0, "completions/max_terminated_length": 454.0, "completions/mean_length": 21.859375, "completions/mean_terminated_length": 21.859375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.095, "frac_reward_zero_std": 0.625, "grad_norm": 11.441950798034668, "learning_rate": 7.452117519152541e-07, "loss": 0.0, "num_tokens": 4381377.0, "reward": 0.5375000238418579, "reward_std": 0.1825428307056427, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.4375, "rewards/mcq_exact_match_reward/std": 0.5, "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": 62.0, "completions/max_terminated_length": 62.0, "completions/mean_length": 17.140625, "completions/mean_terminated_length": 17.140625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0975, "frac_reward_zero_std": 0.5, "grad_norm": 9.27878475189209, "learning_rate": 7.306705814893439e-07, "loss": 0.0, "num_tokens": 4482722.0, "reward": 0.5218750238418579, "reward_std": 0.19939783215522766, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 18.0, "completions/max_terminated_length": 18.0, "completions/mean_length": 15.78125, "completions/mean_terminated_length": 15.78125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1, "frac_reward_zero_std": 0.5, "grad_norm": 13.436245918273926, "learning_rate": 7.158771761692464e-07, "loss": 0.0, "num_tokens": 4560580.0, "reward": 0.7562500238418579, "reward_std": 0.2540663480758667, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.65625, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "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": 18.0, "completions/max_terminated_length": 18.0, "completions/mean_length": 14.9375, "completions/mean_terminated_length": 14.9375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1025, "frac_reward_zero_std": 0.75, "grad_norm": 7.61528205871582, "learning_rate": 7.008477123264847e-07, "loss": 0.0, "num_tokens": 4640080.0, "reward": 0.34375, "reward_std": 0.0883883461356163, "rewards/format_reward/mean": 0.9375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 16.0, "completions/max_terminated_length": 16.0, "completions/mean_length": 15.703125, "completions/mean_terminated_length": 15.703125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.105, "frac_reward_zero_std": 0.625, "grad_norm": 9.347208976745605, "learning_rate": 6.855986244591103e-07, "loss": 0.0, "num_tokens": 4740893.0, "reward": 0.5531250238418579, "reward_std": 0.16887323558330536, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 65.0, "completions/max_terminated_length": 65.0, "completions/mean_length": 17.609375, "completions/mean_terminated_length": 17.609375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1075, "frac_reward_zero_std": 0.625, "grad_norm": 9.331953048706055, "learning_rate": 6.701465872208216e-07, "loss": 0.0, "num_tokens": 4832028.0, "reward": 0.6140625476837158, "reward_std": 0.18340612947940826, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.515625, "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": 18.0, "completions/max_terminated_length": 18.0, "completions/mean_length": 14.953125, "completions/mean_terminated_length": 14.953125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.11, "frac_reward_zero_std": 0.75, "grad_norm": 15.550042152404785, "learning_rate": 6.545084971874736e-07, "loss": 0.0, "num_tokens": 4905585.0, "reward": 0.7250000238418579, "reward_std": 0.0883883461356163, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.625, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 18.0, "completions/max_terminated_length": 18.0, "completions/mean_length": 12.96875, "completions/mean_terminated_length": 12.96875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1125, "frac_reward_zero_std": 0.875, "grad_norm": 2.6361594200134277, "learning_rate": 6.387014543809223e-07, "loss": 0.0, "num_tokens": 5024887.0, "reward": 0.3031250238418579, "reward_std": 0.0646936446428299, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "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": 115.0, "completions/max_terminated_length": 115.0, "completions/mean_length": 16.109375, "completions/mean_terminated_length": 16.109375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.115, "frac_reward_zero_std": 0.75, "grad_norm": 7.038691997528076, "learning_rate": 6.227427435703995e-07, "loss": 0.0, "num_tokens": 5116830.0, "reward": 0.359375, "reward_std": 0.0508757047355175, "rewards/format_reward/mean": 0.9375, "rewards/format_reward/std": 0.24397502839565277, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 22.0, "completions/max_terminated_length": 22.0, "completions/mean_length": 13.03125, "completions/mean_terminated_length": 13.03125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1175, "frac_reward_zero_std": 0.5, "grad_norm": 19.56507110595703, "learning_rate": 6.066498153718734e-07, "loss": -0.0, "num_tokens": 5210408.0, "reward": 0.453125, "reward_std": 0.19939783215522766, "rewards/format_reward/mean": 0.9375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 16.0, "completions/max_terminated_length": 16.0, "completions/mean_length": 13.1875, "completions/mean_terminated_length": 13.1875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.12, "frac_reward_zero_std": 0.625, "grad_norm": 28.4608211517334, "learning_rate": 5.90440267166055e-07, "loss": -0.0, "num_tokens": 5309796.0, "reward": 0.6312500238418579, "reward_std": 0.16675157845020294, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "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": 19.0, "completions/max_terminated_length": 19.0, "completions/mean_length": 13.75, "completions/mean_terminated_length": 13.75, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1225, "frac_reward_zero_std": 0.875, "grad_norm": 10.606098175048828, "learning_rate": 5.741318238559209e-07, "loss": 0.0, "num_tokens": 5409708.0, "reward": 0.7875000238418579, "reward_std": 0.06681530922651291, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.6875, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 18.0, "completions/max_terminated_length": 18.0, "completions/mean_length": 13.953125, "completions/mean_terminated_length": 13.953125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.125, "frac_reward_zero_std": 0.75, "grad_norm": 15.378028869628906, "learning_rate": 5.577423184847931e-07, "loss": 0.0, "num_tokens": 5493121.0, "reward": 0.9421875476837158, "reward_std": 0.09280776977539062, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.84375, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "step": 50 } ], "logging_steps": 1, "max_steps": 100, "num_input_tokens_seen": 5493121, "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 }