{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.11160714285714286, "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": 1386.0, "completions/mean_length": 515.03125, "completions/mean_terminated_length": 490.6984558105469, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.002232142857142857, "frac_reward_zero_std": 0.0, "grad_norm": 7.372617721557617, "learning_rate": 0.0, "loss": 0.0, "num_tokens": 116874.0, "reward": 0.296875, "reward_std": 0.36100417375564575, "rewards/format_reward/mean": 0.3125, "rewards/format_reward/std": 0.24397502839565277, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 1737.0, "completions/max_terminated_length": 1737.0, "completions/mean_length": 464.234375, "completions/mean_terminated_length": 464.234375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.004464285714285714, "frac_reward_zero_std": 0.0, "grad_norm": 9.42634105682373, "learning_rate": 2e-07, "loss": -0.0, "num_tokens": 230473.0, "reward": 0.33281248807907104, "reward_std": 0.35629093647003174, "rewards/format_reward/mean": 0.359375, "rewards/format_reward/std": 0.28824523091316223, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 1479.0, "completions/max_terminated_length": 1479.0, "completions/mean_length": 375.984375, "completions/mean_terminated_length": 375.984375, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.006696428571428571, "frac_reward_zero_std": 0.0, "grad_norm": 12.815962791442871, "learning_rate": 4e-07, "loss": 0.0, "num_tokens": 346168.0, "reward": 0.21718749403953552, "reward_std": 0.3329676389694214, "rewards/format_reward/mean": 0.296875, "rewards/format_reward/std": 0.2630521357059479, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "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": 0.0625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1428.0, "completions/mean_length": 490.515625, "completions/mean_terminated_length": 386.683349609375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.008928571428571428, "frac_reward_zero_std": 0.0, "grad_norm": 6.687461853027344, "learning_rate": 6e-07, "loss": 0.0, "num_tokens": 486377.0, "reward": 0.2593749761581421, "reward_std": 0.3835652470588684, "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": 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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1506.0, "completions/mean_length": 606.28125, "completions/mean_terminated_length": 535.3770141601562, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.011160714285714286, "frac_reward_zero_std": 0.0, "grad_norm": 16.5729923248291, "learning_rate": 8e-07, "loss": 0.0, "num_tokens": 620971.0, "reward": 0.23984374105930328, "reward_std": 0.3459582030773163, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.23974503576755524, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "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": 1333.0, "completions/mean_length": 531.765625, "completions/mean_terminated_length": 482.8548278808594, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.013392857142857142, "frac_reward_zero_std": 0.0, "grad_norm": 7.315047264099121, "learning_rate": 1e-06, "loss": 0.0, "num_tokens": 739660.0, "reward": 0.28203123807907104, "reward_std": 0.23655115067958832, "rewards/format_reward/mean": 0.3203125, "rewards/format_reward/std": 0.27265870571136475, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 6 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1883.0, "completions/max_terminated_length": 1883.0, "completions/mean_length": 534.65625, "completions/mean_terminated_length": 534.65625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.015625, "frac_reward_zero_std": 0.0, "grad_norm": 10.712610244750977, "learning_rate": 9.99726628670463e-07, "loss": -0.0, "num_tokens": 868742.0, "reward": 0.3554687201976776, "reward_std": 0.20643417537212372, "rewards/format_reward/mean": 0.4296875, "rewards/format_reward/std": 0.23345555365085602, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1883.0, "completions/mean_length": 639.4375, "completions/mean_terminated_length": 570.1638793945312, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.017857142857142856, "frac_reward_zero_std": 0.0, "grad_norm": 10.94524097442627, "learning_rate": 9.989068136093872e-07, "loss": 0.0, "num_tokens": 1012874.0, "reward": 0.20781248807907104, "reward_std": 0.30935126543045044, "rewards/format_reward/mean": 0.359375, "rewards/format_reward/std": 0.2741328477859497, "rewards/mcq_exact_match_reward/mean": 0.171875, "rewards/mcq_exact_match_reward/std": 0.38025420904159546, "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": 0.0, "completions/max_length": 1730.0, "completions/max_terminated_length": 1730.0, "completions/mean_length": 345.8125, "completions/mean_terminated_length": 345.8125, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.020089285714285716, "frac_reward_zero_std": 0.25, "grad_norm": 10.861742973327637, "learning_rate": 9.975414512725056e-07, "loss": -0.0, "num_tokens": 1141086.0, "reward": 0.2593749761581421, "reward_std": 0.2899813652038574, "rewards/format_reward/mean": 0.40625, "rewards/format_reward/std": 0.21593283116817474, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "step": 9 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1430.0, "completions/max_terminated_length": 1430.0, "completions/mean_length": 466.390625, "completions/mean_terminated_length": 466.390625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.022321428571428572, "frac_reward_zero_std": 0.0, "grad_norm": 10.960936546325684, "learning_rate": 9.956320346634875e-07, "loss": 0.0, "num_tokens": 1260815.0, "reward": 0.27421873807907104, "reward_std": 0.3824688494205475, "rewards/format_reward/mean": 0.3984375, "rewards/format_reward/std": 0.28423789143562317, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "step": 10 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1601.0, "completions/mean_length": 428.53125, "completions/mean_terminated_length": 376.2903137207031, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.024553571428571428, "frac_reward_zero_std": 0.0, "grad_norm": 18.18084716796875, "learning_rate": 9.931806517013612e-07, "loss": 0.0, "num_tokens": 1379769.0, "reward": 0.23281249403953552, "reward_std": 0.2880294919013977, "rewards/format_reward/mean": 0.453125, "rewards/format_reward/std": 0.2630521357059479, "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": 0.0, "completions/max_length": 680.0, "completions/max_terminated_length": 680.0, "completions/mean_length": 135.90625, "completions/mean_terminated_length": 135.90625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.026785714285714284, "frac_reward_zero_std": 0.375, "grad_norm": 17.66448402404785, "learning_rate": 9.901899829374047e-07, "loss": -0.0, "num_tokens": 1470971.0, "reward": 0.4703124761581421, "reward_std": 0.28840553760528564, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.17747680842876434, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 1568.0, "completions/max_terminated_length": 1568.0, "completions/mean_length": 217.578125, "completions/mean_terminated_length": 217.578125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.029017857142857144, "frac_reward_zero_std": 0.625, "grad_norm": 10.263340950012207, "learning_rate": 9.866632986240029e-07, "loss": 0.0, "num_tokens": 1578096.0, "reward": 0.2523437440395355, "reward_std": 0.11321917921304703, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "step": 13 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 842.0, "completions/max_terminated_length": 842.0, "completions/mean_length": 99.265625, "completions/mean_terminated_length": 99.265625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.03125, "frac_reward_zero_std": 0.125, "grad_norm": 18.594722747802734, "learning_rate": 9.826044551386742e-07, "loss": 0.0, "num_tokens": 1670537.0, "reward": 0.42812496423721313, "reward_std": 0.3584539592266083, "rewards/format_reward/mean": 0.53125, "rewards/format_reward/std": 0.17536810040473938, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 0.0, "completions/max_length": 533.0, "completions/max_terminated_length": 533.0, "completions/mean_length": 27.75, "completions/mean_terminated_length": 27.75, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.033482142857142856, "frac_reward_zero_std": 0.75, "grad_norm": 10.696029663085938, "learning_rate": 9.780178907671788e-07, "loss": -0.0, "num_tokens": 1772369.0, "reward": 0.5031249523162842, "reward_std": 0.11100947111845016, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.03571428571428571, "frac_reward_zero_std": 0.75, "grad_norm": 16.516544342041016, "learning_rate": 9.729086208503173e-07, "loss": -0.0, "num_tokens": 1875425.0, "reward": 0.40937498211860657, "reward_std": 0.12255740165710449, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.03794642857142857, "frac_reward_zero_std": 0.375, "grad_norm": 21.267547607421875, "learning_rate": 9.672822322997304e-07, "loss": 0.0, "num_tokens": 1988737.0, "reward": 0.28437498211860657, "reward_std": 0.26196980476379395, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.04017857142857143, "frac_reward_zero_std": 0.625, "grad_norm": 20.922260284423828, "learning_rate": 9.611448774886923e-07, "loss": -0.0, "num_tokens": 2100009.0, "reward": 0.7687499523162842, "reward_std": 0.1872510462999344, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.71875, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "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": 0.0, "completions/max_length": 59.0, "completions/max_terminated_length": 59.0, "completions/mean_length": 6.828125, "completions/mean_terminated_length": 6.828125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.04241071428571429, "frac_reward_zero_std": 0.5, "grad_norm": 22.29154396057129, "learning_rate": 9.545032675245813e-07, "loss": -0.0, "num_tokens": 2228990.0, "reward": 0.39374998211860657, "reward_std": 0.23356686532497406, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "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": 40.0, "completions/max_terminated_length": 40.0, "completions/mean_length": 6.53125, "completions/mean_terminated_length": 6.53125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.044642857142857144, "frac_reward_zero_std": 0.5, "grad_norm": 17.501502990722656, "learning_rate": 9.473646649103817e-07, "loss": 0.0, "num_tokens": 2338600.0, "reward": 0.4093749523162842, "reward_std": 0.19044625759124756, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 19.0, "completions/max_terminated_length": 19.0, "completions/mean_length": 7.3125, "completions/mean_terminated_length": 7.3125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.046875, "frac_reward_zero_std": 0.75, "grad_norm": 14.592903137207031, "learning_rate": 9.397368756032444e-07, "loss": 0.0, "num_tokens": 2423444.0, "reward": 0.19999998807907104, "reward_std": 0.06365012377500534, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.049107142857142856, "frac_reward_zero_std": 0.625, "grad_norm": 15.976240158081055, "learning_rate": 9.316282404787869e-07, "loss": -0.0, "num_tokens": 2520124.0, "reward": 0.31562498211860657, "reward_std": 0.17782476544380188, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 540.0, "completions/max_terminated_length": 540.0, "completions/mean_length": 17.328125, "completions/mean_terminated_length": 17.328125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05133928571428571, "frac_reward_zero_std": 0.625, "grad_norm": 14.27862548828125, "learning_rate": 9.230476262104676e-07, "loss": -0.0, "num_tokens": 2650241.0, "reward": 0.5812499523162842, "reward_std": 0.15520364046096802, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 23 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 11.0, "completions/max_terminated_length": 11.0, "completions/mean_length": 6.078125, "completions/mean_terminated_length": 6.078125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05357142857142857, "frac_reward_zero_std": 0.5, "grad_norm": 15.152713775634766, "learning_rate": 9.1400441557401e-07, "loss": -0.0, "num_tokens": 2770894.0, "reward": 0.19062498211860657, "reward_std": 0.2198972851037979, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.140625, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "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": 211.0, "completions/max_terminated_length": 211.0, "completions/mean_length": 9.203125, "completions/mean_terminated_length": 9.203125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05580357142857143, "frac_reward_zero_std": 0.625, "grad_norm": 29.53569984436035, "learning_rate": 9.045084971874737e-07, "loss": -0.0, "num_tokens": 2841411.0, "reward": 0.4406249523162842, "reward_std": 0.17782476544380188, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05803571428571429, "frac_reward_zero_std": 0.625, "grad_norm": 23.521568298339844, "learning_rate": 8.945702546981968e-07, "loss": -0.0, "num_tokens": 2926323.0, "reward": 0.5031249523162842, "reward_std": 0.15992167592048645, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.060267857142857144, "frac_reward_zero_std": 0.625, "grad_norm": 17.729854583740234, "learning_rate": 8.842005554284295e-07, "loss": -0.0, "num_tokens": 3047355.0, "reward": 0.4093749523162842, "reward_std": 0.15992167592048645, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0625, "frac_reward_zero_std": 0.75, "grad_norm": 10.808194160461426, "learning_rate": 8.734107384920769e-07, "loss": 0.0, "num_tokens": 3145515.0, "reward": 0.29999998211860657, "reward_std": 0.0883883386850357, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06473214285714286, "frac_reward_zero_std": 0.625, "grad_norm": 18.28078269958496, "learning_rate": 8.622126023955445e-07, "loss": -0.0, "num_tokens": 3238355.0, "reward": 0.28437498211860657, "reward_std": 0.16887322068214417, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 650.0, "completions/max_terminated_length": 650.0, "completions/mean_length": 16.0625, "completions/mean_terminated_length": 16.0625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06696428571428571, "frac_reward_zero_std": 0.75, "grad_norm": 12.452601432800293, "learning_rate": 8.506183921362442e-07, "loss": -0.0, "num_tokens": 3350191.0, "reward": 0.3781249523162842, "reward_std": 0.10205792635679245, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06919642857142858, "frac_reward_zero_std": 0.5, "grad_norm": 32.80612564086914, "learning_rate": 8.386407858128706e-07, "loss": -0.0, "num_tokens": 3443559.0, "reward": 0.33124998211860657, "reward_std": 0.21306739747524261, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "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": 183.0, "completions/max_terminated_length": 183.0, "completions/mean_length": 10.84375, "completions/mean_terminated_length": 10.84375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07142857142857142, "frac_reward_zero_std": 0.5, "grad_norm": 21.804208755493164, "learning_rate": 8.262928807620843e-07, "loss": 0.0, "num_tokens": 3553701.0, "reward": 0.40937498211860657, "reward_std": 0.2109457403421402, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07366071428571429, "frac_reward_zero_std": 0.625, "grad_norm": 15.285557746887207, "learning_rate": 8.135881792367685e-07, "loss": 0.0, "num_tokens": 3658485.0, "reward": 0.40937498211860657, "reward_std": 0.1530819833278656, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07589285714285714, "frac_reward_zero_std": 0.75, "grad_norm": 19.146930694580078, "learning_rate": 8.005405736415125e-07, "loss": -0.0, "num_tokens": 3753581.0, "reward": 0.5031249523162842, "reward_std": 0.11100947111845016, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.078125, "frac_reward_zero_std": 0.875, "grad_norm": 12.522613525390625, "learning_rate": 7.871643313414718e-07, "loss": -0.0, "num_tokens": 3851317.0, "reward": 0.4562499523162842, "reward_std": 0.0578637532889843, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 29.0, "completions/max_terminated_length": 29.0, "completions/mean_length": 6.578125, "completions/mean_terminated_length": 6.578125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08035714285714286, "frac_reward_zero_std": 0.375, "grad_norm": 38.662662506103516, "learning_rate": 7.734740790612136e-07, "loss": 0.0, "num_tokens": 3935226.0, "reward": 0.34609371423721313, "reward_std": 0.20160752534866333, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.296875, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08258928571428571, "frac_reward_zero_std": 0.75, "grad_norm": 32.49333572387695, "learning_rate": 7.594847868906076e-07, "loss": -0.0, "num_tokens": 4054722.0, "reward": 0.5968749523162842, "reward_std": 0.11100947856903076, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.546875, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08482142857142858, "frac_reward_zero_std": 0.625, "grad_norm": 38.19132614135742, "learning_rate": 7.452117519152541e-07, "loss": 0.0, "num_tokens": 4151506.0, "reward": 0.7164062261581421, "reward_std": 0.15529169142246246, "rewards/format_reward/mean": 0.4453125, "rewards/format_reward/std": 0.15728822350502014, "rewards/mcq_exact_match_reward/mean": 0.671875, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08705357142857142, "frac_reward_zero_std": 0.5, "grad_norm": 30.554147720336914, "learning_rate": 7.306705814893439e-07, "loss": 0.0, "num_tokens": 4230210.0, "reward": 0.3937499523162842, "reward_std": 0.19727614521980286, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08928571428571429, "frac_reward_zero_std": 0.625, "grad_norm": 23.661088943481445, "learning_rate": 7.158771761692464e-07, "loss": -0.0, "num_tokens": 4335138.0, "reward": 0.34609371423721313, "reward_std": 0.15529169142246246, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.09151785714285714, "frac_reward_zero_std": 0.625, "grad_norm": 32.43943786621094, "learning_rate": 7.008477123264847e-07, "loss": 0.0, "num_tokens": 4439562.0, "reward": 0.23749998211860657, "reward_std": 0.16675156354904175, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.09375, "frac_reward_zero_std": 0.625, "grad_norm": 13.679961204528809, "learning_rate": 6.855986244591103e-07, "loss": -0.0, "num_tokens": 4555042.0, "reward": 0.5031249523162842, "reward_std": 0.13258251547813416, "rewards/format_reward/mean": 0.5, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.09598214285714286, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 6.701465872208216e-07, "loss": 0.0, "num_tokens": 4661210.0, "reward": 0.4249999523162842, "reward_std": 0.0, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 125.0, "completions/max_terminated_length": 125.0, "completions/mean_length": 7.859375, "completions/mean_terminated_length": 7.859375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.09821428571428571, "frac_reward_zero_std": 0.875, "grad_norm": 6.7772626876831055, "learning_rate": 6.545084971874736e-07, "loss": -0.0, "num_tokens": 4763577.0, "reward": 0.4093749523162842, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.10044642857142858, "frac_reward_zero_std": 0.625, "grad_norm": 29.50361442565918, "learning_rate": 6.387014543809223e-07, "loss": 0.0, "num_tokens": 4850985.0, "reward": 0.5499999523162842, "reward_std": 0.16675156354904175, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.10267857142857142, "frac_reward_zero_std": 0.875, "grad_norm": 14.329254150390625, "learning_rate": 6.227427435703995e-07, "loss": -0.0, "num_tokens": 4946729.0, "reward": 0.5187499523162842, "reward_std": 0.0578637532889843, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.10491071428571429, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 6.066498153718734e-07, "loss": 0.0, "num_tokens": 5027633.0, "reward": 0.6749999523162842, "reward_std": 0.0, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.625, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.10714285714285714, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 5.90440267166055e-07, "loss": 0.0, "num_tokens": 5123361.0, "reward": 0.5499999523162842, "reward_std": 0.0, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.109375, "frac_reward_zero_std": 0.875, "grad_norm": 15.217447280883789, "learning_rate": 5.741318238559209e-07, "loss": -0.0, "num_tokens": 5215425.0, "reward": 0.5187499523162842, "reward_std": 0.0578637532889843, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "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": 418.0, "completions/max_terminated_length": 418.0, "completions/mean_length": 12.4375, "completions/mean_terminated_length": 12.4375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.11160714285714286, "frac_reward_zero_std": 0.625, "grad_norm": 29.34973907470703, "learning_rate": 5.577423184847931e-07, "loss": 0.0, "num_tokens": 5331149.0, "reward": 0.26874998211860657, "reward_std": 0.1462520956993103, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "step": 50 } ], "logging_steps": 1, "max_steps": 100, "num_input_tokens_seen": 5331149, "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 }