{ "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1901.0, "completions/mean_length": 489.890625, "completions/mean_terminated_length": 465.15875244140625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0025, "frac_reward_zero_std": 0.125, "grad_norm": 5.311183452606201, "learning_rate": 0.0, "loss": 0.0, "num_tokens": 130033.0, "reward": 0.24453124403953552, "reward_std": 0.33292192220687866, "rewards/format_reward/mean": 0.4140625, "rewards/format_reward/std": 0.19012710452079773, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "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": 1418.0, "completions/max_terminated_length": 1418.0, "completions/mean_length": 521.546875, "completions/mean_terminated_length": 521.546875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.005, "frac_reward_zero_std": 0.0, "grad_norm": 9.74411392211914, "learning_rate": 2e-07, "loss": -0.0, "num_tokens": 251772.0, "reward": 0.30312496423721313, "reward_std": 0.3246499300003052, "rewards/format_reward/mean": 0.375, "rewards/format_reward/std": 0.2182178944349289, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1361.0, "completions/mean_length": 495.078125, "completions/mean_terminated_length": 444.9838562011719, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.0075, "frac_reward_zero_std": 0.0, "grad_norm": 22.562915802001953, "learning_rate": 4e-07, "loss": -0.0, "num_tokens": 393841.0, "reward": 0.30156248807907104, "reward_std": 0.41690510511398315, "rewards/format_reward/mean": 0.359375, "rewards/format_reward/std": 0.22658175230026245, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 1112.0, "completions/mean_length": 549.78125, "completions/mean_terminated_length": 526.0, "completions/min_length": 60.0, "completions/min_terminated_length": 60.0, "epoch": 0.01, "frac_reward_zero_std": 0.0, "grad_norm": 1.2926867008209229, "learning_rate": 6e-07, "loss": -0.0, "num_tokens": 521459.0, "reward": 0.2828124761581421, "reward_std": 0.36704355478286743, "rewards/format_reward/mean": 0.328125, "rewards/format_reward/std": 0.23935678601264954, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 1620.0, "completions/max_terminated_length": 1620.0, "completions/mean_length": 443.078125, "completions/mean_terminated_length": 443.078125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0125, "frac_reward_zero_std": 0.125, "grad_norm": 6.607117176055908, "learning_rate": 8e-07, "loss": -0.0, "num_tokens": 634128.0, "reward": 0.44609373807907104, "reward_std": 0.3606116771697998, "rewards/format_reward/mean": 0.3984375, "rewards/format_reward/std": 0.31090864539146423, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1636.0, "completions/mean_length": 413.96875, "completions/mean_terminated_length": 361.258056640625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.015, "frac_reward_zero_std": 0.125, "grad_norm": 11.48267936706543, "learning_rate": 1e-06, "loss": 0.0, "num_tokens": 758086.0, "reward": 0.30703121423721313, "reward_std": 0.32097184658050537, "rewards/format_reward/mean": 0.4140625, "rewards/format_reward/std": 0.31565865874290466, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 1745.0, "completions/mean_length": 606.40625, "completions/mean_terminated_length": 583.5238647460938, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0175, "frac_reward_zero_std": 0.0, "grad_norm": 6.747123718261719, "learning_rate": 9.99726628670463e-07, "loss": -0.0, "num_tokens": 907928.0, "reward": 0.3359375, "reward_std": 0.34430092573165894, "rewards/format_reward/mean": 0.390625, "rewards/format_reward/std": 0.22658175230026245, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 0.0, "completions/max_length": 1922.0, "completions/max_terminated_length": 1922.0, "completions/mean_length": 505.65625, "completions/mean_terminated_length": 505.65625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.02, "frac_reward_zero_std": 0.0, "grad_norm": 8.986135482788086, "learning_rate": 9.989068136093872e-07, "loss": -0.0, "num_tokens": 1045098.0, "reward": 0.33828121423721313, "reward_std": 0.3621408939361572, "rewards/format_reward/mean": 0.4140625, "rewards/format_reward/std": 0.19012710452079773, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 1857.0, "completions/mean_length": 598.0625, "completions/mean_terminated_length": 526.7540893554688, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0225, "frac_reward_zero_std": 0.0, "grad_norm": 3.457052707672119, "learning_rate": 9.975414512725056e-07, "loss": 0.0, "num_tokens": 1181022.0, "reward": 0.2710937261581421, "reward_std": 0.3995788097381592, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.23974503576755524, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1289.0, "completions/mean_length": 541.375, "completions/mean_terminated_length": 517.4603271484375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.025, "frac_reward_zero_std": 0.125, "grad_norm": 2.6495301723480225, "learning_rate": 9.956320346634875e-07, "loss": 0.0, "num_tokens": 1300238.0, "reward": 0.2304687350988388, "reward_std": 0.29889369010925293, "rewards/format_reward/mean": 0.4296875, "rewards/format_reward/std": 0.21578919887542725, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "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": 1661.0, "completions/max_terminated_length": 1661.0, "completions/mean_length": 471.5, "completions/mean_terminated_length": 471.5, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0275, "frac_reward_zero_std": 0.0, "grad_norm": 13.053606033325195, "learning_rate": 9.931806517013612e-07, "loss": -0.0, "num_tokens": 1423542.0, "reward": 0.40546873211860657, "reward_std": 0.42098402976989746, "rewards/format_reward/mean": 0.4609375, "rewards/format_reward/std": 0.28597769141197205, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1664.0, "completions/mean_length": 632.6875, "completions/mean_terminated_length": 610.2222290039062, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.03, "frac_reward_zero_std": 0.125, "grad_norm": 0.9276574850082397, "learning_rate": 9.901899829374047e-07, "loss": 0.0, "num_tokens": 1575978.0, "reward": 0.3312499523162842, "reward_std": 0.34990543127059937, "rewards/format_reward/mean": 0.34375, "rewards/format_reward/std": 0.25, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 0.0, "completions/max_length": 1613.0, "completions/max_terminated_length": 1613.0, "completions/mean_length": 623.453125, "completions/mean_terminated_length": 623.453125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0325, "frac_reward_zero_std": 0.0, "grad_norm": 4.8317461013793945, "learning_rate": 9.866632986240029e-07, "loss": -0.0, "num_tokens": 1695735.0, "reward": 0.35234373807907104, "reward_std": 0.4299824833869934, "rewards/format_reward/mean": 0.3984375, "rewards/format_reward/std": 0.2387082874774933, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1672.0, "completions/mean_length": 461.140625, "completions/mean_terminated_length": 383.09832763671875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.035, "frac_reward_zero_std": 0.0, "grad_norm": 10.147533416748047, "learning_rate": 9.826044551386742e-07, "loss": -0.0, "num_tokens": 1820248.0, "reward": 0.39140623807907104, "reward_std": 0.3290434181690216, "rewards/format_reward/mean": 0.4765625, "rewards/format_reward/std": 0.2735668122768402, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1859.0, "completions/mean_length": 733.859375, "completions/mean_terminated_length": 669.2294921875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0375, "frac_reward_zero_std": 0.125, "grad_norm": 4.119254112243652, "learning_rate": 9.780178907671788e-07, "loss": 0.0, "num_tokens": 1963455.0, "reward": 0.14999999105930328, "reward_std": 0.14054429531097412, "rewards/format_reward/mean": 0.40625, "rewards/format_reward/std": 0.25, "rewards/mcq_exact_match_reward/mean": 0.109375, "rewards/mcq_exact_match_reward/std": 0.3145764470100403, "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": 1593.0, "completions/mean_length": 476.421875, "completions/mean_terminated_length": 451.4762268066406, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.04, "frac_reward_zero_std": 0.0, "grad_norm": 3.720846176147461, "learning_rate": 9.729086208503173e-07, "loss": -0.0, "num_tokens": 2070122.0, "reward": 0.29609373211860657, "reward_std": 0.3335476517677307, "rewards/format_reward/mean": 0.4609375, "rewards/format_reward/std": 0.28597769141197205, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1960.0, "completions/mean_length": 768.640625, "completions/mean_terminated_length": 705.7212524414062, "completions/min_length": 279.0, "completions/min_terminated_length": 279.0, "epoch": 0.0425, "frac_reward_zero_std": 0.0, "grad_norm": 0.7833253145217896, "learning_rate": 9.672822322997304e-07, "loss": -0.0, "num_tokens": 2197947.0, "reward": 0.17890623211860657, "reward_std": 0.2567874789237976, "rewards/format_reward/mean": 0.3828125, "rewards/format_reward/std": 0.2478829026222229, "rewards/mcq_exact_match_reward/mean": 0.140625, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1184.0, "completions/mean_length": 497.671875, "completions/mean_terminated_length": 473.0635070800781, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.045, "frac_reward_zero_std": 0.125, "grad_norm": 7.033801555633545, "learning_rate": 9.611448774886923e-07, "loss": 0.0, "num_tokens": 2340942.0, "reward": 0.47343748807907104, "reward_std": 0.31370896100997925, "rewards/format_reward/mean": 0.515625, "rewards/format_reward/std": 0.26679685711860657, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1587.0, "completions/mean_length": 615.40625, "completions/mean_terminated_length": 592.6666870117188, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0475, "frac_reward_zero_std": 0.0, "grad_norm": 8.892571449279785, "learning_rate": 9.545032675245813e-07, "loss": 0.0, "num_tokens": 2470536.0, "reward": 0.3515625, "reward_std": 0.45494481921195984, "rewards/format_reward/mean": 0.546875, "rewards/format_reward/std": 0.3299681544303894, "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": 0.0625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1861.0, "completions/mean_length": 702.625, "completions/mean_terminated_length": 612.933349609375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05, "frac_reward_zero_std": 0.125, "grad_norm": 7.654572486877441, "learning_rate": 9.473646649103817e-07, "loss": 0.0, "num_tokens": 2613472.0, "reward": 0.34453123807907104, "reward_std": 0.3284533619880676, "rewards/format_reward/mean": 0.4765625, "rewards/format_reward/std": 0.3140833079814911, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1319.0, "completions/mean_length": 539.0, "completions/mean_terminated_length": 490.32257080078125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0525, "frac_reward_zero_std": 0.0, "grad_norm": 11.230612754821777, "learning_rate": 9.397368756032444e-07, "loss": 0.0, "num_tokens": 2761352.0, "reward": 0.20937497913837433, "reward_std": 0.2642296552658081, "rewards/format_reward/mean": 0.375, "rewards/format_reward/std": 0.29546841979026794, "rewards/mcq_exact_match_reward/mean": 0.171875, "rewards/mcq_exact_match_reward/std": 0.38025420904159546, "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": 1265.0, "completions/mean_length": 470.15625, "completions/mean_terminated_length": 445.11114501953125, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.055, "frac_reward_zero_std": 0.0, "grad_norm": 5.04013204574585, "learning_rate": 9.316282404787869e-07, "loss": 0.0, "num_tokens": 2861978.0, "reward": 0.2812499701976776, "reward_std": 0.2999628186225891, "rewards/format_reward/mean": 0.46875, "rewards/format_reward/std": 0.279951810836792, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1235.0, "completions/mean_length": 537.890625, "completions/mean_terminated_length": 513.920654296875, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.0575, "frac_reward_zero_std": 0.0, "grad_norm": 7.080632209777832, "learning_rate": 9.230476262104676e-07, "loss": -0.0, "num_tokens": 3002899.0, "reward": 0.2562499940395355, "reward_std": 0.33555495738983154, "rewards/format_reward/mean": 0.53125, "rewards/format_reward/std": 0.3549536168575287, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1687.0, "completions/mean_length": 675.484375, "completions/mean_terminated_length": 607.9835815429688, "completions/min_length": 105.0, "completions/min_terminated_length": 105.0, "epoch": 0.06, "frac_reward_zero_std": 0.0, "grad_norm": 1.1298067569732666, "learning_rate": 9.1400441557401e-07, "loss": 0.0, "num_tokens": 3147458.0, "reward": 0.36015623807907104, "reward_std": 0.38480520248413086, "rewards/format_reward/mean": 0.4765625, "rewards/format_reward/std": 0.33841001987457275, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 0.0625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1723.0, "completions/mean_length": 725.984375, "completions/mean_terminated_length": 637.8500366210938, "completions/min_length": 19.0, "completions/min_terminated_length": 19.0, "epoch": 0.0625, "frac_reward_zero_std": 0.0, "grad_norm": 2.4508216381073, "learning_rate": 9.045084971874737e-07, "loss": 0.0, "num_tokens": 3306545.0, "reward": 0.3109375238418579, "reward_std": 0.3069794178009033, "rewards/format_reward/mean": 0.453125, "rewards/format_reward/std": 0.30496877431869507, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 0.0625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1914.0, "completions/mean_length": 762.59375, "completions/mean_terminated_length": 676.9000244140625, "completions/min_length": 242.0, "completions/min_terminated_length": 242.0, "epoch": 0.065, "frac_reward_zero_std": 0.0, "grad_norm": 1.101258635520935, "learning_rate": 8.945702546981968e-07, "loss": 0.0, "num_tokens": 3480399.0, "reward": 0.36640626192092896, "reward_std": 0.3461141884326935, "rewards/format_reward/mean": 0.5390625, "rewards/format_reward/std": 0.3485194146633148, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 0.09375, "completions/max_length": 2048.0, "completions/max_terminated_length": 1191.0, "completions/mean_length": 618.4375, "completions/mean_terminated_length": 470.5517272949219, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.0675, "frac_reward_zero_std": 0.0, "grad_norm": 9.459016799926758, "learning_rate": 8.842005554284295e-07, "loss": 0.0, "num_tokens": 3591611.0, "reward": 0.4351562261581421, "reward_std": 0.4266842007637024, "rewards/format_reward/mean": 0.6015625, "rewards/format_reward/std": 0.37984633445739746, "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": 0.0, "completions/max_length": 1179.0, "completions/max_terminated_length": 1179.0, "completions/mean_length": 580.9375, "completions/mean_terminated_length": 580.9375, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.07, "frac_reward_zero_std": 0.0, "grad_norm": 3.6875884532928467, "learning_rate": 8.734107384920769e-07, "loss": -0.0, "num_tokens": 3730847.0, "reward": 0.18828123807907104, "reward_std": 0.2507384717464447, "rewards/format_reward/mean": 0.4765625, "rewards/format_reward/std": 0.3719285726547241, "rewards/mcq_exact_match_reward/mean": 0.140625, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "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": 1266.0, "completions/mean_length": 434.359375, "completions/mean_terminated_length": 408.7460632324219, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.0725, "frac_reward_zero_std": 0.0, "grad_norm": 7.714194297790527, "learning_rate": 8.622126023955445e-07, "loss": -0.0, "num_tokens": 3851966.0, "reward": 0.3515625, "reward_std": 0.4523616135120392, "rewards/format_reward/mean": 0.546875, "rewards/format_reward/std": 0.4338710308074951, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1535.0, "completions/mean_length": 579.703125, "completions/mean_terminated_length": 532.3386840820312, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.075, "frac_reward_zero_std": 0.0, "grad_norm": 7.635128498077393, "learning_rate": 8.506183921362442e-07, "loss": -0.0, "num_tokens": 3982987.0, "reward": 0.3125, "reward_std": 0.3674323558807373, "rewards/format_reward/mean": 0.625, "rewards/format_reward/std": 0.37796446681022644, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1293.0, "completions/mean_length": 409.375, "completions/mean_terminated_length": 383.3651123046875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0775, "frac_reward_zero_std": 0.0, "grad_norm": 8.861705780029297, "learning_rate": 8.386407858128706e-07, "loss": 0.0, "num_tokens": 4122091.0, "reward": 0.18437498807907104, "reward_std": 0.24302847683429718, "rewards/format_reward/mean": 0.59375, "rewards/format_reward/std": 0.3971625566482544, "rewards/mcq_exact_match_reward/mean": 0.125, "rewards/mcq_exact_match_reward/std": 0.3333333432674408, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1802.0, "completions/mean_length": 473.734375, "completions/mean_terminated_length": 448.7460632324219, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.08, "frac_reward_zero_std": 0.0, "grad_norm": 17.855640411376953, "learning_rate": 8.262928807620843e-07, "loss": -0.0, "num_tokens": 4237170.0, "reward": 0.44218748807907104, "reward_std": 0.37794405221939087, "rewards/format_reward/mean": 0.671875, "rewards/format_reward/std": 0.3905506134033203, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1071.0, "completions/mean_length": 304.859375, "completions/mean_terminated_length": 277.19049072265625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0825, "frac_reward_zero_std": 0.125, "grad_norm": 16.608196258544922, "learning_rate": 8.135881792367685e-07, "loss": -0.0, "num_tokens": 4356737.0, "reward": 0.5320312976837158, "reward_std": 0.4117293357849121, "rewards/format_reward/mean": 0.7890625, "rewards/format_reward/std": 0.3060849606990814, "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": 798.0, "completions/max_terminated_length": 798.0, "completions/mean_length": 197.828125, "completions/mean_terminated_length": 197.828125, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.085, "frac_reward_zero_std": 0.125, "grad_norm": 13.8006591796875, "learning_rate": 8.005405736415125e-07, "loss": 0.0, "num_tokens": 4446470.0, "reward": 0.42500001192092896, "reward_std": 0.2939152717590332, "rewards/format_reward/mean": 0.8125, "rewards/format_reward/std": 0.28867512941360474, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1113.0, "completions/mean_length": 414.09375, "completions/mean_terminated_length": 333.7377014160156, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.0875, "frac_reward_zero_std": 0.0, "grad_norm": 6.627899169921875, "learning_rate": 7.871643313414718e-07, "loss": 0.0, "num_tokens": 4582108.0, "reward": 0.4281250238418579, "reward_std": 0.3943573534488678, "rewards/format_reward/mean": 0.84375, "rewards/format_reward/std": 0.3196600377559662, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "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": 1564.0, "completions/max_terminated_length": 1564.0, "completions/mean_length": 207.015625, "completions/mean_terminated_length": 207.015625, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.09, "frac_reward_zero_std": 0.125, "grad_norm": 11.127843856811523, "learning_rate": 7.734740790612136e-07, "loss": 0.0, "num_tokens": 4704157.0, "reward": 0.5867187976837158, "reward_std": 0.299810528755188, "rewards/format_reward/mean": 0.8671875, "rewards/format_reward/std": 0.23974503576755524, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "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": 1068.0, "completions/max_terminated_length": 1068.0, "completions/mean_length": 108.703125, "completions/mean_terminated_length": 108.703125, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.0925, "frac_reward_zero_std": 0.5, "grad_norm": 9.205948829650879, "learning_rate": 7.594847868906076e-07, "loss": 0.0, "num_tokens": 4788218.0, "reward": 0.50390625, "reward_std": 0.22566819190979004, "rewards/format_reward/mean": 0.9765625, "rewards/format_reward/std": 0.10652101784944534, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 632.0, "completions/max_terminated_length": 632.0, "completions/mean_length": 69.4375, "completions/mean_terminated_length": 69.4375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.095, "frac_reward_zero_std": 0.125, "grad_norm": 16.434175491333008, "learning_rate": 7.452117519152541e-07, "loss": 0.0, "num_tokens": 4889774.0, "reward": 0.4398437738418579, "reward_std": 0.35628199577331543, "rewards/format_reward/mean": 0.9609375, "rewards/format_reward/std": 0.16194961965084076, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 703.0, "completions/mean_length": 80.46875, "completions/mean_terminated_length": 49.23809814453125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0975, "frac_reward_zero_std": 0.5, "grad_norm": 6.326141357421875, "learning_rate": 7.306705814893439e-07, "loss": -0.0, "num_tokens": 4971756.0, "reward": 0.581250011920929, "reward_std": 0.12809909880161285, "rewards/format_reward/mean": 0.96875, "rewards/format_reward/std": 0.1510545015335083, "rewards/mcq_exact_match_reward/mean": 0.484375, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "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": 614.0, "completions/max_terminated_length": 614.0, "completions/mean_length": 52.6875, "completions/mean_terminated_length": 52.6875, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.1, "frac_reward_zero_std": 0.375, "grad_norm": 8.57425308227539, "learning_rate": 7.158771761692464e-07, "loss": 0.0, "num_tokens": 5081792.0, "reward": 0.4085937738418579, "reward_std": 0.22734402120113373, "rewards/format_reward/mean": 0.9609375, "rewards/format_reward/std": 0.13524486124515533, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 1080.0, "completions/max_terminated_length": 1080.0, "completions/mean_length": 89.578125, "completions/mean_terminated_length": 89.578125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1025, "frac_reward_zero_std": 0.5, "grad_norm": 8.523602485656738, "learning_rate": 7.008477123264847e-07, "loss": 0.0, "num_tokens": 5180501.0, "reward": 0.4710937440395355, "reward_std": 0.1250883936882019, "rewards/format_reward/mean": 0.9609375, "rewards/format_reward/std": 0.18483558297157288, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 289.0, "completions/max_terminated_length": 289.0, "completions/mean_length": 20.4375, "completions/mean_terminated_length": 20.4375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.105, "frac_reward_zero_std": 0.625, "grad_norm": 6.494924068450928, "learning_rate": 6.855986244591103e-07, "loss": -0.0, "num_tokens": 5254777.0, "reward": 0.5218750238418579, "reward_std": 0.16887325048446655, "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": 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": 402.0, "completions/max_terminated_length": 402.0, "completions/mean_length": 21.40625, "completions/mean_terminated_length": 21.40625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1075, "frac_reward_zero_std": 0.375, "grad_norm": 16.94335174560547, "learning_rate": 6.701465872208216e-07, "loss": 0.0, "num_tokens": 5350915.0, "reward": 0.6156250238418579, "reward_std": 0.28930896520614624, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "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": 483.0, "completions/max_terminated_length": 483.0, "completions/mean_length": 21.125, "completions/mean_terminated_length": 21.125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.11, "frac_reward_zero_std": 0.625, "grad_norm": 5.9126176834106445, "learning_rate": 6.545084971874736e-07, "loss": 0.0, "num_tokens": 5448971.0, "reward": 0.6304687261581421, "reward_std": 0.14677414298057556, "rewards/format_reward/mean": 0.9921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "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": 372.0, "completions/max_terminated_length": 372.0, "completions/mean_length": 25.203125, "completions/mean_terminated_length": 25.203125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1125, "frac_reward_zero_std": 0.75, "grad_norm": 5.639616012573242, "learning_rate": 6.387014543809223e-07, "loss": 0.0, "num_tokens": 5539960.0, "reward": 0.6468750238418579, "reward_std": 0.10205793380737305, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.546875, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 179.0, "completions/max_terminated_length": 179.0, "completions/mean_length": 15.453125, "completions/mean_terminated_length": 15.453125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.115, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 6.227427435703995e-07, "loss": 0.0, "num_tokens": 5657637.0, "reward": 0.2187500149011612, "reward_std": 0.0, "rewards/format_reward/mean": 0.9375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.125, "rewards/mcq_exact_match_reward/std": 0.3333333432674408, "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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.984375, "completions/mean_terminated_length": 12.984375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1175, "frac_reward_zero_std": 0.5, "grad_norm": 8.273690223693848, "learning_rate": 6.066498153718734e-07, "loss": 0.0, "num_tokens": 5751292.0, "reward": 0.1937500238418579, "reward_std": 0.19727617502212524, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.09375, "rewards/mcq_exact_match_reward/std": 0.29378482699394226, "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": 158.0, "completions/max_terminated_length": 158.0, "completions/mean_length": 15.265625, "completions/mean_terminated_length": 15.265625, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.12, "frac_reward_zero_std": 0.25, "grad_norm": 18.4805965423584, "learning_rate": 5.90440267166055e-07, "loss": 0.0, "num_tokens": 5858837.0, "reward": 0.5218750238418579, "reward_std": 0.3403330445289612, "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": 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": 180.0, "completions/max_terminated_length": 180.0, "completions/mean_length": 17.015625, "completions/mean_terminated_length": 17.015625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1225, "frac_reward_zero_std": 0.625, "grad_norm": 7.702163219451904, "learning_rate": 5.741318238559209e-07, "loss": 0.0, "num_tokens": 5942198.0, "reward": 0.23593750596046448, "reward_std": 0.13204257190227509, "rewards/format_reward/mean": 0.953125, "rewards/format_reward/std": 0.14689241349697113, "rewards/mcq_exact_match_reward/mean": 0.140625, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "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": 91.0, "completions/max_terminated_length": 91.0, "completions/mean_length": 14.171875, "completions/mean_terminated_length": 14.171875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.125, "frac_reward_zero_std": 0.375, "grad_norm": 13.468091011047363, "learning_rate": 5.577423184847931e-07, "loss": 0.0, "num_tokens": 6061641.0, "reward": 0.5843750238418579, "reward_std": 0.30721205472946167, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.484375, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "step": 50 } ], "logging_steps": 1, "max_steps": 100, "num_input_tokens_seen": 6061641, "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 }