{ "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1869.0, "completions/mean_length": 485.078125, "completions/mean_terminated_length": 408.2131042480469, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.0025, "frac_reward_zero_std": 0.125, "grad_norm": 11.975702285766602, "learning_rate": 0.0, "loss": 0.0, "num_tokens": 140029.0, "reward": 0.203125, "reward_std": 0.2905668318271637, "rewards/format_reward/mean": 0.3125, "rewards/format_reward/std": 0.24397502839565277, "rewards/mcq_exact_match_reward/mean": 0.171875, "rewards/mcq_exact_match_reward/std": 0.38025420904159546, "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": 1776.0, "completions/max_terminated_length": 1776.0, "completions/mean_length": 396.0, "completions/mean_terminated_length": 396.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.005, "frac_reward_zero_std": 0.0, "grad_norm": 3.944349765777588, "learning_rate": 2e-07, "loss": 0.0, "num_tokens": 224253.0, "reward": 0.17031249403953552, "reward_std": 0.30059775710105896, "rewards/format_reward/mean": 0.453125, "rewards/format_reward/std": 0.3750000298023224, "rewards/mcq_exact_match_reward/mean": 0.125, "rewards/mcq_exact_match_reward/std": 0.3333333432674408, "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": 1881.0, "completions/max_terminated_length": 1881.0, "completions/mean_length": 379.359375, "completions/mean_terminated_length": 379.359375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0075, "frac_reward_zero_std": 0.0, "grad_norm": 7.817681312561035, "learning_rate": 4e-07, "loss": 0.0, "num_tokens": 352036.0, "reward": 0.18984374403953552, "reward_std": 0.2685413360595703, "rewards/format_reward/mean": 0.3359375, "rewards/format_reward/std": 0.2366211861371994, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "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": 1588.0, "completions/mean_length": 596.4375, "completions/mean_terminated_length": 573.3968505859375, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.01, "frac_reward_zero_std": 0.0, "grad_norm": 6.446213722229004, "learning_rate": 6e-07, "loss": 0.0, "num_tokens": 497552.0, "reward": 0.18984374403953552, "reward_std": 0.2477949857711792, "rewards/format_reward/mean": 0.3359375, "rewards/format_reward/std": 0.2366211861371994, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1618.0, "completions/mean_length": 478.484375, "completions/mean_terminated_length": 453.5714416503906, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0125, "frac_reward_zero_std": 0.0, "grad_norm": 14.41591739654541, "learning_rate": 8e-07, "loss": 0.0, "num_tokens": 618759.0, "reward": 0.36953121423721313, "reward_std": 0.2460281401872635, "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": 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": 1360.0, "completions/max_terminated_length": 1360.0, "completions/mean_length": 448.875, "completions/mean_terminated_length": 448.875, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.015, "frac_reward_zero_std": 0.125, "grad_norm": 2.243814706802368, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 750335.0, "reward": 0.30312496423721313, "reward_std": 0.3401804566383362, "rewards/format_reward/mean": 0.375, "rewards/format_reward/std": 0.2357022762298584, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1968.0, "completions/mean_length": 600.21875, "completions/mean_terminated_length": 553.51611328125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0175, "frac_reward_zero_std": 0.0, "grad_norm": 11.396003723144531, "learning_rate": 9.99726628670463e-07, "loss": -0.0, "num_tokens": 896389.0, "reward": 0.3515625, "reward_std": 0.4192287027835846, "rewards/format_reward/mean": 0.390625, "rewards/format_reward/std": 0.2741328477859497, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 1808.0, "completions/max_terminated_length": 1808.0, "completions/mean_length": 503.78125, "completions/mean_terminated_length": 503.78125, "completions/min_length": 7.0, "completions/min_terminated_length": 7.0, "epoch": 0.02, "frac_reward_zero_std": 0.0, "grad_norm": 1.7651492357254028, "learning_rate": 9.989068136093872e-07, "loss": -0.0, "num_tokens": 1013231.0, "reward": 0.34218746423721313, "reward_std": 0.4568067193031311, "rewards/format_reward/mean": 0.296875, "rewards/format_reward/std": 0.2630521357059479, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 1956.0, "completions/mean_length": 712.46875, "completions/mean_terminated_length": 691.2698974609375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0225, "frac_reward_zero_std": 0.25, "grad_norm": 2.015846014022827, "learning_rate": 9.975414512725056e-07, "loss": -0.0, "num_tokens": 1164093.0, "reward": 0.16953124105930328, "reward_std": 0.2928316295146942, "rewards/format_reward/mean": 0.4453125, "rewards/format_reward/std": 0.180765300989151, "rewards/mcq_exact_match_reward/mean": 0.125, "rewards/mcq_exact_match_reward/std": 0.3333333432674408, "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": 1623.0, "completions/mean_length": 621.421875, "completions/mean_terminated_length": 598.77783203125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.025, "frac_reward_zero_std": 0.0, "grad_norm": 12.08847427368164, "learning_rate": 9.956320346634875e-07, "loss": -0.0, "num_tokens": 1311568.0, "reward": 0.4335936903953552, "reward_std": 0.3458997309207916, "rewards/format_reward/mean": 0.4296875, "rewards/format_reward/std": 0.19654127955436707, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1560.0, "completions/mean_length": 535.15625, "completions/mean_terminated_length": 511.14288330078125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0275, "frac_reward_zero_std": 0.25, "grad_norm": 3.640256881713867, "learning_rate": 9.931806517013612e-07, "loss": 0.0, "num_tokens": 1423130.0, "reward": 0.23203124105930328, "reward_std": 0.20354680716991425, "rewards/format_reward/mean": 0.4453125, "rewards/format_reward/std": 0.15728822350502014, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "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": 1656.0, "completions/mean_length": 532.796875, "completions/mean_terminated_length": 508.7460632324219, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.03, "frac_reward_zero_std": 0.125, "grad_norm": 6.753490924835205, "learning_rate": 9.901899829374047e-07, "loss": -0.0, "num_tokens": 1569437.0, "reward": 0.17421874403953552, "reward_std": 0.2824912667274475, "rewards/format_reward/mean": 0.3359375, "rewards/format_reward/std": 0.29620200395584106, "rewards/mcq_exact_match_reward/mean": 0.140625, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "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": 1096.0, "completions/max_terminated_length": 1096.0, "completions/mean_length": 468.734375, "completions/mean_terminated_length": 468.734375, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.0325, "frac_reward_zero_std": 0.125, "grad_norm": 8.269699096679688, "learning_rate": 9.866632986240029e-07, "loss": -0.0, "num_tokens": 1678564.0, "reward": 0.37890625, "reward_std": 0.38244131207466125, "rewards/format_reward/mean": 0.5078125, "rewards/format_reward/std": 0.2885677218437195, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 1457.0, "completions/mean_length": 459.875, "completions/mean_terminated_length": 381.7704772949219, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.035, "frac_reward_zero_std": 0.125, "grad_norm": 7.154005527496338, "learning_rate": 9.826044551386742e-07, "loss": -0.0, "num_tokens": 1803660.0, "reward": 0.37812498211860657, "reward_std": 0.4340607523918152, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.28171807527542114, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 2039.0, "completions/mean_length": 576.921875, "completions/mean_terminated_length": 504.57373046875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0375, "frac_reward_zero_std": 0.0, "grad_norm": 6.8013691902160645, "learning_rate": 9.780178907671788e-07, "loss": -0.0, "num_tokens": 1923263.0, "reward": 0.3499999940395355, "reward_std": 0.34579959511756897, "rewards/format_reward/mean": 0.53125, "rewards/format_reward/std": 0.3549536168575287, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 1148.0, "completions/mean_length": 412.03125, "completions/mean_terminated_length": 386.0635070800781, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.04, "frac_reward_zero_std": 0.125, "grad_norm": 11.156859397888184, "learning_rate": 9.729086208503173e-07, "loss": 0.0, "num_tokens": 2045377.0, "reward": 0.2953124940395355, "reward_std": 0.27578848600387573, "rewards/format_reward/mean": 0.609375, "rewards/format_reward/std": 0.3503824472427368, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 1787.0, "completions/max_terminated_length": 1787.0, "completions/mean_length": 383.125, "completions/mean_terminated_length": 383.125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0425, "frac_reward_zero_std": 0.125, "grad_norm": 13.340229034423828, "learning_rate": 9.672822322997304e-07, "loss": -0.0, "num_tokens": 2137337.0, "reward": 0.2562499940395355, "reward_std": 0.2679210901260376, "rewards/format_reward/mean": 0.6875, "rewards/format_reward/std": 0.35073620080947876, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "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": 1353.0, "completions/mean_length": 423.21875, "completions/mean_terminated_length": 397.4285888671875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.045, "frac_reward_zero_std": 0.0, "grad_norm": 14.747516632080078, "learning_rate": 9.611448774886923e-07, "loss": 0.0, "num_tokens": 2252199.0, "reward": 0.45390623807907104, "reward_std": 0.3495957851409912, "rewards/format_reward/mean": 0.6328125, "rewards/format_reward/std": 0.3699226677417755, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 1226.0, "completions/mean_length": 411.09375, "completions/mean_terminated_length": 385.11114501953125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0475, "frac_reward_zero_std": 0.0, "grad_norm": 9.82708740234375, "learning_rate": 9.545032675245813e-07, "loss": -0.0, "num_tokens": 2386589.0, "reward": 0.3140624761581421, "reward_std": 0.4299202859401703, "rewards/format_reward/mean": 0.640625, "rewards/format_reward/std": 0.3503824472427368, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 0.0, "completions/max_length": 1669.0, "completions/max_terminated_length": 1669.0, "completions/mean_length": 399.3125, "completions/mean_terminated_length": 399.3125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05, "frac_reward_zero_std": 0.0, "grad_norm": 18.517269134521484, "learning_rate": 9.473646649103817e-07, "loss": 0.0, "num_tokens": 2498441.0, "reward": 0.52734375, "reward_std": 0.41716527938842773, "rewards/format_reward/mean": 0.7421875, "rewards/format_reward/std": 0.30850622057914734, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 1284.0, "completions/max_terminated_length": 1284.0, "completions/mean_length": 281.59375, "completions/mean_terminated_length": 281.59375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0525, "frac_reward_zero_std": 0.0, "grad_norm": 16.47442626953125, "learning_rate": 9.397368756032444e-07, "loss": 0.0, "num_tokens": 2607039.0, "reward": 0.40546876192092896, "reward_std": 0.29514557123184204, "rewards/format_reward/mean": 0.7734375, "rewards/format_reward/std": 0.33249524235725403, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 644.0, "completions/max_terminated_length": 644.0, "completions/mean_length": 107.21875, "completions/mean_terminated_length": 107.21875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.055, "frac_reward_zero_std": 0.25, "grad_norm": 20.29006576538086, "learning_rate": 9.316282404787869e-07, "loss": 0.0, "num_tokens": 2718141.0, "reward": 0.50390625, "reward_std": 0.23895862698554993, "rewards/format_reward/mean": 0.8203125, "rewards/format_reward/std": 0.31329265236854553, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 893.0, "completions/max_terminated_length": 893.0, "completions/mean_length": 95.46875, "completions/mean_terminated_length": 95.46875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0575, "frac_reward_zero_std": 0.125, "grad_norm": 15.167083740234375, "learning_rate": 9.230476262104676e-07, "loss": 0.0, "num_tokens": 2797115.0, "reward": 0.46953126788139343, "reward_std": 0.3944826126098633, "rewards/format_reward/mean": 0.9453125, "rewards/format_reward/std": 0.180765300989151, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 0.0, "completions/max_length": 1358.0, "completions/max_terminated_length": 1358.0, "completions/mean_length": 63.4375, "completions/mean_terminated_length": 63.4375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06, "frac_reward_zero_std": 0.375, "grad_norm": 12.069833755493164, "learning_rate": 9.1400441557401e-07, "loss": 0.0, "num_tokens": 2916479.0, "reward": 0.14531250298023224, "reward_std": 0.15578331053256989, "rewards/format_reward/mean": 0.828125, "rewards/format_reward/std": 0.29839184880256653, "rewards/mcq_exact_match_reward/mean": 0.0625, "rewards/mcq_exact_match_reward/std": 0.24397502839565277, "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": 0.0, "completions/max_length": 723.0, "completions/max_terminated_length": 723.0, "completions/mean_length": 43.515625, "completions/mean_terminated_length": 43.515625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0625, "frac_reward_zero_std": 0.375, "grad_norm": 13.992871284484863, "learning_rate": 9.045084971874737e-07, "loss": 0.0, "num_tokens": 3010432.0, "reward": 0.41015625, "reward_std": 0.21074499189853668, "rewards/format_reward/mean": 0.9765625, "rewards/format_reward/std": 0.13886408507823944, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 198.0, "completions/max_terminated_length": 198.0, "completions/mean_length": 27.4375, "completions/mean_terminated_length": 27.4375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.065, "frac_reward_zero_std": 0.625, "grad_norm": 6.05722713470459, "learning_rate": 8.945702546981968e-07, "loss": 0.0, "num_tokens": 3099780.0, "reward": 0.3187500238418579, "reward_std": 0.1872510462999344, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "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": 913.0, "completions/max_terminated_length": 913.0, "completions/mean_length": 52.015625, "completions/mean_terminated_length": 52.015625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0675, "frac_reward_zero_std": 0.375, "grad_norm": 10.7685546875, "learning_rate": 8.842005554284295e-07, "loss": 0.0, "num_tokens": 3197813.0, "reward": 0.28437501192092896, "reward_std": 0.28456324338912964, "rewards/format_reward/mean": 0.96875, "rewards/format_reward/std": 0.1510545015335083, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "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": 184.0, "completions/max_terminated_length": 184.0, "completions/mean_length": 21.03125, "completions/mean_terminated_length": 21.03125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.07, "frac_reward_zero_std": 0.625, "grad_norm": 11.545337677001953, "learning_rate": 8.734107384920769e-07, "loss": 0.0, "num_tokens": 3286063.0, "reward": 0.59765625, "reward_std": 0.09102033078670502, "rewards/format_reward/mean": 0.9765625, "rewards/format_reward/std": 0.10652101784944534, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "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": 189.0, "completions/max_terminated_length": 189.0, "completions/mean_length": 19.71875, "completions/mean_terminated_length": 19.71875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0725, "frac_reward_zero_std": 0.25, "grad_norm": 15.760375022888184, "learning_rate": 8.622126023955445e-07, "loss": 0.0, "num_tokens": 3380445.0, "reward": 0.6156250238418579, "reward_std": 0.2993341088294983, "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": 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": 17.0, "completions/max_terminated_length": 17.0, "completions/mean_length": 12.28125, "completions/mean_terminated_length": 12.28125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.075, "frac_reward_zero_std": 0.625, "grad_norm": 9.981792449951172, "learning_rate": 8.506183921362442e-07, "loss": 0.0, "num_tokens": 3464063.0, "reward": 0.25468751788139343, "reward_std": 0.11330723762512207, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "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": 17.0, "completions/max_terminated_length": 17.0, "completions/mean_length": 12.1875, "completions/mean_terminated_length": 12.1875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0775, "frac_reward_zero_std": 0.5, "grad_norm": 12.168708801269531, "learning_rate": 8.386407858128706e-07, "loss": 0.0, "num_tokens": 3551139.0, "reward": 0.5843750238418579, "reward_std": 0.2109457552433014, "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": 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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.171875, "completions/mean_terminated_length": 12.171875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.08, "frac_reward_zero_std": 0.5, "grad_norm": 12.254404067993164, "learning_rate": 8.262928807620843e-07, "loss": -0.0, "num_tokens": 3633270.0, "reward": 0.4437500238418579, "reward_std": 0.24251843988895416, "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": 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": 252.0, "completions/max_terminated_length": 252.0, "completions/mean_length": 18.71875, "completions/mean_terminated_length": 18.71875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0825, "frac_reward_zero_std": 0.75, "grad_norm": 6.7919206619262695, "learning_rate": 8.135881792367685e-07, "loss": 0.0, "num_tokens": 3730156.0, "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": 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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.046875, "completions/mean_terminated_length": 12.046875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.085, "frac_reward_zero_std": 0.625, "grad_norm": 8.68002986907959, "learning_rate": 8.005405736415125e-07, "loss": 0.0, "num_tokens": 3836479.0, "reward": 0.4437500238418579, "reward_std": 0.1552036553621292, "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": 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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.171875, "completions/mean_terminated_length": 12.171875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0875, "frac_reward_zero_std": 0.625, "grad_norm": 7.427948474884033, "learning_rate": 7.871643313414718e-07, "loss": 0.0, "num_tokens": 3936578.0, "reward": 0.5687500238418579, "reward_std": 0.1735912710428238, "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": 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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.109375, "completions/mean_terminated_length": 12.109375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.09, "frac_reward_zero_std": 0.875, "grad_norm": 8.654311180114746, "learning_rate": 7.734740790612136e-07, "loss": 0.0, "num_tokens": 4036369.0, "reward": 0.8031250238418579, "reward_std": 0.0646936446428299, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.703125, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 45.0, "completions/max_terminated_length": 45.0, "completions/mean_length": 12.640625, "completions/mean_terminated_length": 12.640625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0925, "frac_reward_zero_std": 0.75, "grad_norm": 10.818239212036133, "learning_rate": 7.594847868906076e-07, "loss": 0.0, "num_tokens": 4120738.0, "reward": 0.7406250238418579, "reward_std": 0.12255740165710449, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.640625, "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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.0625, "completions/mean_terminated_length": 12.0625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.095, "frac_reward_zero_std": 0.625, "grad_norm": 11.268946647644043, "learning_rate": 7.452117519152541e-07, "loss": 0.0, "num_tokens": 4216870.0, "reward": 0.3500000238418579, "reward_std": 0.16675157845020294, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 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": 0.0975, "frac_reward_zero_std": 0.625, "grad_norm": 9.37126350402832, "learning_rate": 7.306705814893439e-07, "loss": 0.0, "num_tokens": 4302000.0, "reward": 0.6781250238418579, "reward_std": 0.1530819833278656, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.578125, "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": 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": 0.1, "frac_reward_zero_std": 0.875, "grad_norm": 5.179771423339844, "learning_rate": 7.158771761692464e-07, "loss": 0.0, "num_tokens": 4397090.0, "reward": 0.5375000238418579, "reward_std": 0.06681530922651291, "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": 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": 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": 0.1025, "frac_reward_zero_std": 0.625, "grad_norm": 14.984646797180176, "learning_rate": 7.008477123264847e-07, "loss": 0.0, "num_tokens": 4500211.0, "reward": 0.5687500238418579, "reward_std": 0.16675157845020294, "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": 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": 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": 0.105, "frac_reward_zero_std": 0.875, "grad_norm": 3.1286957263946533, "learning_rate": 6.855986244591103e-07, "loss": -0.0, "num_tokens": 4578861.0, "reward": 0.6156250238418579, "reward_std": 0.04419417306780815, "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": 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": 1113.0, "completions/max_terminated_length": 1113.0, "completions/mean_length": 29.203125, "completions/mean_terminated_length": 29.203125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1075, "frac_reward_zero_std": 0.75, "grad_norm": 10.211320877075195, "learning_rate": 6.701465872208216e-07, "loss": 0.0, "num_tokens": 4693106.0, "reward": 0.5062500238418579, "reward_std": 0.10888782143592834, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 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": 0.11, "frac_reward_zero_std": 0.375, "grad_norm": 21.76201629638672, "learning_rate": 6.545084971874736e-07, "loss": -0.0, "num_tokens": 4767099.0, "reward": 0.7562500238418579, "reward_std": 0.30509042739868164, "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": 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": 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": 0.1125, "frac_reward_zero_std": 0.875, "grad_norm": 7.693737030029297, "learning_rate": 6.387014543809223e-07, "loss": 0.0, "num_tokens": 4843029.0, "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": 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": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.115, "frac_reward_zero_std": 0.75, "grad_norm": 7.3927836418151855, "learning_rate": 6.227427435703995e-07, "loss": 0.0, "num_tokens": 4941237.0, "reward": 0.6468750238418579, "reward_std": 0.11100947856903076, "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": 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": 17.0, "completions/max_terminated_length": 17.0, "completions/mean_length": 12.234375, "completions/mean_terminated_length": 12.234375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1175, "frac_reward_zero_std": 0.75, "grad_norm": 6.7654948234558105, "learning_rate": 6.066498153718734e-07, "loss": 0.0, "num_tokens": 5045612.0, "reward": 0.5843750238418579, "reward_std": 0.12255740165710449, "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": 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": 35.0, "completions/max_terminated_length": 35.0, "completions/mean_length": 12.71875, "completions/mean_terminated_length": 12.71875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.12, "frac_reward_zero_std": 0.625, "grad_norm": 14.427403450012207, "learning_rate": 5.90440267166055e-07, "loss": 0.0, "num_tokens": 5170154.0, "reward": 0.659375011920929, "reward_std": 0.1894281953573227, "rewards/format_reward/mean": 0.96875, "rewards/format_reward/std": 0.17536810040473938, "rewards/mcq_exact_match_reward/mean": 0.5625, "rewards/mcq_exact_match_reward/std": 0.5, "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": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1225, "frac_reward_zero_std": 0.875, "grad_norm": 5.07783317565918, "learning_rate": 5.741318238559209e-07, "loss": -0.0, "num_tokens": 5261834.0, "reward": 0.6937500238418579, "reward_std": 0.0578637570142746, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.59375, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 21.0, "completions/max_terminated_length": 21.0, "completions/mean_length": 12.140625, "completions/mean_terminated_length": 12.140625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.125, "frac_reward_zero_std": 0.5, "grad_norm": 15.970564842224121, "learning_rate": 5.577423184847931e-07, "loss": 0.0, "num_tokens": 5351211.0, "reward": 0.6296875476837158, "reward_std": 0.26010218262672424, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 50 } ], "logging_steps": 1, "max_steps": 100, "num_input_tokens_seen": 5351211, "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 }