{ "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1797.0, "completions/mean_length": 640.1875, "completions/mean_terminated_length": 570.9508056640625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.002232142857142857, "frac_reward_zero_std": 0.0, "grad_norm": 2.6361892223358154, "learning_rate": 0.0, "loss": -0.0, "num_tokens": 146636.0, "reward": 0.39921873807907104, "reward_std": 0.3636443316936493, "rewards/format_reward/mean": 0.3984375, "rewards/format_reward/std": 0.20275264978408813, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 1288.0, "completions/max_terminated_length": 1288.0, "completions/mean_length": 445.984375, "completions/mean_terminated_length": 445.984375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.004464285714285714, "frac_reward_zero_std": 0.0, "grad_norm": 6.810795307159424, "learning_rate": 2e-07, "loss": -0.0, "num_tokens": 261267.0, "reward": 0.30546873807907104, "reward_std": 0.39963752031326294, "rewards/format_reward/mean": 0.3984375, "rewards/format_reward/std": 0.25479042530059814, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1379.0, "completions/mean_length": 625.875, "completions/mean_terminated_length": 603.3016357421875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.006696428571428571, "frac_reward_zero_std": 0.0, "grad_norm": 6.051039695739746, "learning_rate": 4e-07, "loss": 0.0, "num_tokens": 417747.0, "reward": 0.11874999105930328, "reward_std": 0.17684173583984375, "rewards/format_reward/mean": 0.40625, "rewards/format_reward/std": 0.233588308095932, "rewards/mcq_exact_match_reward/mean": 0.078125, "rewards/mcq_exact_match_reward/std": 0.27048972249031067, "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": 1456.0, "completions/mean_length": 572.15625, "completions/mean_terminated_length": 548.7301635742188, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.008928571428571428, "frac_reward_zero_std": 0.125, "grad_norm": 2.4216501712799072, "learning_rate": 6e-07, "loss": -0.0, "num_tokens": 553933.0, "reward": 0.33515623211860657, "reward_std": 0.2986903786659241, "rewards/format_reward/mean": 0.3828125, "rewards/format_reward/std": 0.21347814798355103, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 1309.0, "completions/mean_length": 588.78125, "completions/mean_terminated_length": 517.016357421875, "completions/min_length": 82.0, "completions/min_terminated_length": 82.0, "epoch": 0.011160714285714286, "frac_reward_zero_std": 0.0, "grad_norm": 1.31868314743042, "learning_rate": 8e-07, "loss": -0.0, "num_tokens": 685175.0, "reward": 0.33671873807907104, "reward_std": 0.37096071243286133, "rewards/format_reward/mean": 0.3984375, "rewards/format_reward/std": 0.26991602778434753, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "step": 5 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1901.0, "completions/mean_length": 505.390625, "completions/mean_terminated_length": 480.90478515625, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.013392857142857142, "frac_reward_zero_std": 0.0, "grad_norm": 11.582802772521973, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 826336.0, "reward": 0.27031248807907104, "reward_std": 0.20724275708198547, "rewards/format_reward/mean": 0.359375, "rewards/format_reward/std": 0.2741328477859497, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "step": 6 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1880.0, "completions/mean_length": 533.4375, "completions/mean_terminated_length": 509.3968505859375, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.015625, "frac_reward_zero_std": 0.0, "grad_norm": 31.721981048583984, "learning_rate": 9.99726628670463e-07, "loss": 0.0, "num_tokens": 969636.0, "reward": 0.25078123807907104, "reward_std": 0.36516356468200684, "rewards/format_reward/mean": 0.3203125, "rewards/format_reward/std": 0.24180518090724945, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "step": 7 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1544.0, "completions/mean_length": 435.796875, "completions/mean_terminated_length": 410.20635986328125, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.017857142857142856, "frac_reward_zero_std": 0.0, "grad_norm": 39.52054977416992, "learning_rate": 9.989068136093872e-07, "loss": -0.0, "num_tokens": 1089463.0, "reward": 0.29374998807907104, "reward_std": 0.4106873869895935, "rewards/format_reward/mean": 0.28125, "rewards/format_reward/std": 0.265398770570755, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 1330.0, "completions/max_terminated_length": 1330.0, "completions/mean_length": 472.859375, "completions/mean_terminated_length": 472.859375, "completions/min_length": 10.0, "completions/min_terminated_length": 10.0, "epoch": 0.020089285714285716, "frac_reward_zero_std": 0.0, "grad_norm": 9.149499893188477, "learning_rate": 9.975414512725056e-07, "loss": 0.0, "num_tokens": 1184590.0, "reward": 0.234375, "reward_std": 0.29499173164367676, "rewards/format_reward/mean": 0.46875, "rewards/format_reward/std": 0.33184191584587097, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "step": 9 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 921.0, "completions/max_terminated_length": 921.0, "completions/mean_length": 283.4375, "completions/mean_terminated_length": 283.4375, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.022321428571428572, "frac_reward_zero_std": 0.25, "grad_norm": 24.136234283447266, "learning_rate": 9.956320346634875e-07, "loss": 0.0, "num_tokens": 1303474.0, "reward": 0.18515625596046448, "reward_std": 0.23798231780529022, "rewards/format_reward/mean": 0.4453125, "rewards/format_reward/std": 0.2203386276960373, "rewards/mcq_exact_match_reward/mean": 0.140625, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "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": 0.0625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1860.0, "completions/mean_length": 680.984375, "completions/mean_terminated_length": 589.8500366210938, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.024553571428571428, "frac_reward_zero_std": 0.125, "grad_norm": 6.937000274658203, "learning_rate": 9.931806517013612e-07, "loss": -0.0, "num_tokens": 1454961.0, "reward": 0.20859374105930328, "reward_std": 0.33535653352737427, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.2708333432674408, "rewards/mcq_exact_match_reward/mean": 0.171875, "rewards/mcq_exact_match_reward/std": 0.38025420904159546, "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": 1922.0, "completions/mean_length": 455.734375, "completions/mean_terminated_length": 430.4603271484375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.026785714285714284, "frac_reward_zero_std": 0.125, "grad_norm": 5.525171756744385, "learning_rate": 9.901899829374047e-07, "loss": -0.0, "num_tokens": 1587416.0, "reward": 0.16718748211860657, "reward_std": 0.26241543889045715, "rewards/format_reward/mean": 0.421875, "rewards/format_reward/std": 0.25539806485176086, "rewards/mcq_exact_match_reward/mean": 0.125, "rewards/mcq_exact_match_reward/std": 0.3333333432674408, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1391.0, "completions/mean_length": 383.84375, "completions/mean_terminated_length": 330.1612854003906, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.029017857142857144, "frac_reward_zero_std": 0.0, "grad_norm": 10.084939956665039, "learning_rate": 9.866632986240029e-07, "loss": 0.0, "num_tokens": 1694382.0, "reward": 0.47968748211860657, "reward_std": 0.4627240300178528, "rewards/format_reward/mean": 0.421875, "rewards/format_reward/std": 0.27048972249031067, "rewards/mcq_exact_match_reward/mean": 0.4375, "rewards/mcq_exact_match_reward/std": 0.5, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1491.0, "completions/mean_length": 407.65625, "completions/mean_terminated_length": 381.61907958984375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.03125, "frac_reward_zero_std": 0.0, "grad_norm": 19.48894691467285, "learning_rate": 9.826044551386742e-07, "loss": -0.0, "num_tokens": 1819624.0, "reward": 0.42265623807907104, "reward_std": 0.38675984740257263, "rewards/format_reward/mean": 0.4765625, "rewards/format_reward/std": 0.25865477323532104, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1608.0, "completions/mean_length": 379.59375, "completions/mean_terminated_length": 353.11114501953125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.033482142857142856, "frac_reward_zero_std": 0.5, "grad_norm": 6.431363105773926, "learning_rate": 9.780178907671788e-07, "loss": 0.0, "num_tokens": 1921990.0, "reward": 0.47343748807907104, "reward_std": 0.1926751434803009, "rewards/format_reward/mean": 0.515625, "rewards/format_reward/std": 0.1985812783241272, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 1698.0, "completions/mean_length": 472.28125, "completions/mean_terminated_length": 447.2698669433594, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.03571428571428571, "frac_reward_zero_std": 0.0, "grad_norm": 7.701583385467529, "learning_rate": 9.729086208503173e-07, "loss": 0.0, "num_tokens": 2046472.0, "reward": 0.18359375, "reward_std": 0.24951808154582977, "rewards/format_reward/mean": 0.4296875, "rewards/format_reward/std": 0.3195630609989166, "rewards/mcq_exact_match_reward/mean": 0.140625, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "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": 1771.0, "completions/max_terminated_length": 1771.0, "completions/mean_length": 373.984375, "completions/mean_terminated_length": 373.984375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.03794642857142857, "frac_reward_zero_std": 0.125, "grad_norm": 18.922931671142578, "learning_rate": 9.672822322997304e-07, "loss": 0.0, "num_tokens": 2166279.0, "reward": 0.27031248807907104, "reward_std": 0.18254360556602478, "rewards/format_reward/mean": 0.515625, "rewards/format_reward/std": 0.2812775671482086, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "step": 17 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1453.0, "completions/mean_length": 328.4375, "completions/mean_terminated_length": 272.9677429199219, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.04017857142857143, "frac_reward_zero_std": 0.25, "grad_norm": 13.542136192321777, "learning_rate": 9.611448774886923e-07, "loss": -0.0, "num_tokens": 2286787.0, "reward": 0.5539062023162842, "reward_std": 0.20302043855190277, "rewards/format_reward/mean": 0.5390625, "rewards/format_reward/std": 0.2567298710346222, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "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": 809.0, "completions/mean_length": 225.578125, "completions/mean_terminated_length": 196.6508026123047, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.04241071428571429, "frac_reward_zero_std": 0.125, "grad_norm": 23.45460319519043, "learning_rate": 9.545032675245813e-07, "loss": -0.0, "num_tokens": 2375832.0, "reward": 0.6328125, "reward_std": 0.43864309787750244, "rewards/format_reward/mean": 0.546875, "rewards/format_reward/std": 0.24750742316246033, "rewards/mcq_exact_match_reward/mean": 0.578125, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 1126.0, "completions/max_terminated_length": 1126.0, "completions/mean_length": 229.171875, "completions/mean_terminated_length": 229.171875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.044642857142857144, "frac_reward_zero_std": 0.125, "grad_norm": 9.905097961425781, "learning_rate": 9.473646649103817e-07, "loss": -0.0, "num_tokens": 2489147.0, "reward": 0.3414062559604645, "reward_std": 0.3022947907447815, "rewards/format_reward/mean": 0.6015625, "rewards/format_reward/std": 0.28423789143562317, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "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": 1054.0, "completions/max_terminated_length": 1054.0, "completions/mean_length": 198.84375, "completions/mean_terminated_length": 198.84375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.046875, "frac_reward_zero_std": 0.125, "grad_norm": 13.886911392211914, "learning_rate": 9.397368756032444e-07, "loss": -0.0, "num_tokens": 2593489.0, "reward": 0.5960937142372131, "reward_std": 0.40328171849250793, "rewards/format_reward/mean": 0.6484375, "rewards/format_reward/std": 0.2302463799715042, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "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": 988.0, "completions/mean_length": 262.265625, "completions/mean_terminated_length": 233.920654296875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.049107142857142856, "frac_reward_zero_std": 0.125, "grad_norm": 11.873527526855469, "learning_rate": 9.316282404787869e-07, "loss": 0.0, "num_tokens": 2692234.0, "reward": 0.42109373211860657, "reward_std": 0.33247458934783936, "rewards/format_reward/mean": 0.6171875, "rewards/format_reward/std": 0.31800705194473267, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 1099.0, "completions/mean_length": 188.046875, "completions/mean_terminated_length": 158.52381896972656, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05133928571428571, "frac_reward_zero_std": 0.125, "grad_norm": 20.855073928833008, "learning_rate": 9.230476262104676e-07, "loss": -0.0, "num_tokens": 2800709.0, "reward": 0.41874998807907104, "reward_std": 0.39405784010887146, "rewards/format_reward/mean": 0.59375, "rewards/format_reward/std": 0.3069951832294464, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "step": 23 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1004.0, "completions/mean_length": 169.65625, "completions/mean_terminated_length": 139.84127807617188, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05357142857142857, "frac_reward_zero_std": 0.25, "grad_norm": 19.31253433227539, "learning_rate": 9.1400441557401e-07, "loss": 0.0, "num_tokens": 2907087.0, "reward": 0.45156246423721313, "reward_std": 0.2916262149810791, "rewards/format_reward/mean": 0.609375, "rewards/format_reward/std": 0.2741328477859497, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 654.0, "completions/max_terminated_length": 654.0, "completions/mean_length": 120.1875, "completions/mean_terminated_length": 120.1875, "completions/min_length": 5.0, "completions/min_terminated_length": 5.0, "epoch": 0.05580357142857143, "frac_reward_zero_std": 0.125, "grad_norm": 22.144765853881836, "learning_rate": 9.045084971874737e-07, "loss": 0.0, "num_tokens": 3018971.0, "reward": 0.43359375, "reward_std": 0.2608272433280945, "rewards/format_reward/mean": 0.5859375, "rewards/format_reward/std": 0.28942593932151794, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 1558.0, "completions/max_terminated_length": 1558.0, "completions/mean_length": 42.453125, "completions/mean_terminated_length": 42.453125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05803571428571429, "frac_reward_zero_std": 0.375, "grad_norm": 17.29846954345703, "learning_rate": 8.945702546981968e-07, "loss": -0.0, "num_tokens": 3117608.0, "reward": 0.30156248807907104, "reward_std": 0.21868138015270233, "rewards/format_reward/mean": 0.671875, "rewards/format_reward/std": 0.25539806485176086, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 461.0, "completions/max_terminated_length": 461.0, "completions/mean_length": 49.5, "completions/mean_terminated_length": 49.5, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.060267857142857144, "frac_reward_zero_std": 0.25, "grad_norm": 24.37734603881836, "learning_rate": 8.842005554284295e-07, "loss": -0.0, "num_tokens": 3206816.0, "reward": 0.39531251788139343, "reward_std": 0.2395390272140503, "rewards/format_reward/mean": 0.828125, "rewards/format_reward/std": 0.25539806485176086, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "step": 27 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1220.0, "completions/max_terminated_length": 1220.0, "completions/mean_length": 109.078125, "completions/mean_terminated_length": 109.078125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0625, "frac_reward_zero_std": 0.25, "grad_norm": 22.970813751220703, "learning_rate": 8.734107384920769e-07, "loss": -0.0, "num_tokens": 3304517.0, "reward": 0.75390625, "reward_std": 0.2941003143787384, "rewards/format_reward/mean": 0.6640625, "rewards/format_reward/std": 0.321882963180542, "rewards/mcq_exact_match_reward/mean": 0.6875, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 734.0, "completions/mean_length": 74.125, "completions/mean_terminated_length": 42.79365158081055, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06473214285714286, "frac_reward_zero_std": 0.0, "grad_norm": 22.676082611083984, "learning_rate": 8.622126023955445e-07, "loss": -0.0, "num_tokens": 3406469.0, "reward": 0.6265624761581421, "reward_std": 0.3221682608127594, "rewards/format_reward/mean": 0.796875, "rewards/format_reward/std": 0.30496877431869507, "rewards/mcq_exact_match_reward/mean": 0.546875, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 715.0, "completions/max_terminated_length": 715.0, "completions/mean_length": 45.828125, "completions/mean_terminated_length": 45.828125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06696428571428571, "frac_reward_zero_std": 0.125, "grad_norm": 31.81020164489746, "learning_rate": 8.506183921362442e-07, "loss": 0.0, "num_tokens": 3490378.0, "reward": 0.5710937976837158, "reward_std": 0.28889110684394836, "rewards/format_reward/mean": 0.8671875, "rewards/format_reward/std": 0.22257846593856812, "rewards/mcq_exact_match_reward/mean": 0.484375, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "step": 30 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 400.0, "completions/mean_length": 51.859375, "completions/mean_terminated_length": 20.174604415893555, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06919642857142858, "frac_reward_zero_std": 0.5, "grad_norm": 15.087133407592773, "learning_rate": 8.386407858128706e-07, "loss": 0.0, "num_tokens": 3584129.0, "reward": 0.5234375, "reward_std": 0.17117097973823547, "rewards/format_reward/mean": 0.859375, "rewards/format_reward/std": 0.24346621334552765, "rewards/mcq_exact_match_reward/mean": 0.4375, "rewards/mcq_exact_match_reward/std": 0.5, "step": 31 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 160.0, "completions/max_terminated_length": 160.0, "completions/mean_length": 15.78125, "completions/mean_terminated_length": 15.78125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07142857142857142, "frac_reward_zero_std": 0.375, "grad_norm": 17.448902130126953, "learning_rate": 8.262928807620843e-07, "loss": -0.0, "num_tokens": 3701563.0, "reward": 0.5453125238418579, "reward_std": 0.18484057486057281, "rewards/format_reward/mean": 0.921875, "rewards/format_reward/std": 0.18298126757144928, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 267.0, "completions/max_terminated_length": 267.0, "completions/mean_length": 17.015625, "completions/mean_terminated_length": 17.015625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07366071428571429, "frac_reward_zero_std": 0.375, "grad_norm": 23.23429298400879, "learning_rate": 8.135881792367685e-07, "loss": -0.0, "num_tokens": 3794492.0, "reward": 0.686718761920929, "reward_std": 0.2268250435590744, "rewards/format_reward/mean": 0.9296875, "rewards/format_reward/std": 0.1751912236213684, "rewards/mcq_exact_match_reward/mean": 0.59375, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 961.0, "completions/max_terminated_length": 961.0, "completions/mean_length": 59.78125, "completions/mean_terminated_length": 59.78125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.07589285714285714, "frac_reward_zero_std": 0.5, "grad_norm": 10.30454158782959, "learning_rate": 8.005405736415125e-07, "loss": -0.0, "num_tokens": 3903710.0, "reward": 0.4125000238418579, "reward_std": 0.2041158676147461, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "step": 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": 775.0, "completions/max_terminated_length": 775.0, "completions/mean_length": 41.796875, "completions/mean_terminated_length": 41.796875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.078125, "frac_reward_zero_std": 0.125, "grad_norm": 22.40190887451172, "learning_rate": 7.871643313414718e-07, "loss": 0.0, "num_tokens": 4010041.0, "reward": 0.453125, "reward_std": 0.42556095123291016, "rewards/format_reward/mean": 0.9375, "rewards/format_reward/std": 0.22712838649749756, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 1687.0, "completions/max_terminated_length": 1687.0, "completions/mean_length": 75.359375, "completions/mean_terminated_length": 75.359375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.08035714285714286, "frac_reward_zero_std": 0.25, "grad_norm": 15.008723258972168, "learning_rate": 7.734740790612136e-07, "loss": 0.0, "num_tokens": 4086320.0, "reward": 0.23828125, "reward_std": 0.3008594512939453, "rewards/format_reward/mean": 0.9765625, "rewards/format_reward/std": 0.13886408507823944, "rewards/mcq_exact_match_reward/mean": 0.140625, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "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": 463.0, "completions/max_terminated_length": 463.0, "completions/mean_length": 30.984375, "completions/mean_terminated_length": 30.984375, "completions/min_length": 11.0, "completions/min_terminated_length": 11.0, "epoch": 0.08258928571428571, "frac_reward_zero_std": 0.125, "grad_norm": 31.971960067749023, "learning_rate": 7.594847868906076e-07, "loss": 0.0, "num_tokens": 4168255.0, "reward": 0.28046876192092896, "reward_std": 0.28750497102737427, "rewards/format_reward/mean": 0.9296875, "rewards/format_reward/std": 0.2498759627342224, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 318.0, "completions/mean_length": 50.171875, "completions/mean_terminated_length": 18.460317611694336, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.08482142857142858, "frac_reward_zero_std": 0.0, "grad_norm": 24.996009826660156, "learning_rate": 7.452117519152541e-07, "loss": 0.0, "num_tokens": 4282330.0, "reward": 0.5249999761581421, "reward_std": 0.39607352018356323, "rewards/format_reward/mean": 0.875, "rewards/format_reward/std": 0.3333333432674408, "rewards/mcq_exact_match_reward/mean": 0.4375, "rewards/mcq_exact_match_reward/std": 0.5, "step": 38 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 402.0, "completions/max_terminated_length": 402.0, "completions/mean_length": 18.828125, "completions/mean_terminated_length": 18.828125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08705357142857142, "frac_reward_zero_std": 0.625, "grad_norm": 13.527844429016113, "learning_rate": 7.306705814893439e-07, "loss": 0.0, "num_tokens": 4376327.0, "reward": 0.23749999701976776, "reward_std": 0.13143110275268555, "rewards/format_reward/mean": 0.96875, "rewards/format_reward/std": 0.12198751419782639, "rewards/mcq_exact_match_reward/mean": 0.140625, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "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": 342.0, "completions/max_terminated_length": 342.0, "completions/mean_length": 24.328125, "completions/mean_terminated_length": 24.328125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.08928571428571429, "frac_reward_zero_std": 0.375, "grad_norm": 23.073726654052734, "learning_rate": 7.158771761692464e-07, "loss": 0.0, "num_tokens": 4438556.0, "reward": 0.8031250238418579, "reward_std": 0.24831002950668335, "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": 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": 286.0, "completions/max_terminated_length": 286.0, "completions/mean_length": 21.703125, "completions/mean_terminated_length": 21.703125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.09151785714285714, "frac_reward_zero_std": 0.875, "grad_norm": 13.344606399536133, "learning_rate": 7.008477123264847e-07, "loss": 0.0, "num_tokens": 4531281.0, "reward": 0.2562500238418579, "reward_std": 0.0578637570142746, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "step": 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": 12.015625, "completions/mean_terminated_length": 12.015625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.09375, "frac_reward_zero_std": 0.5, "grad_norm": 10.37992000579834, "learning_rate": 6.855986244591103e-07, "loss": 0.0, "num_tokens": 4600082.0, "reward": 0.5687500238418579, "reward_std": 0.1767766922712326, "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": 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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 11.359375, "completions/mean_terminated_length": 11.359375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.09598214285714286, "frac_reward_zero_std": 0.5, "grad_norm": 14.126476287841797, "learning_rate": 6.701465872208216e-07, "loss": 0.0, "num_tokens": 4670465.0, "reward": 0.48515626788139343, "reward_std": 0.19081565737724304, "rewards/format_reward/mean": 0.9453125, "rewards/format_reward/std": 0.15728822350502014, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 445.0, "completions/max_terminated_length": 445.0, "completions/mean_length": 19.046875, "completions/mean_terminated_length": 19.046875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.09821428571428571, "frac_reward_zero_std": 0.75, "grad_norm": 9.034393310546875, "learning_rate": 6.545084971874736e-07, "loss": 0.0, "num_tokens": 4748140.0, "reward": 0.6625000238418579, "reward_std": 0.10888782143592834, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.5625, "rewards/mcq_exact_match_reward/std": 0.5, "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": 12.015625, "completions/mean_terminated_length": 12.015625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.10044642857142858, "frac_reward_zero_std": 0.625, "grad_norm": 15.553948402404785, "learning_rate": 6.387014543809223e-07, "loss": 0.0, "num_tokens": 4829981.0, "reward": 0.9125000238418579, "reward_std": 0.1462521106004715, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.8125, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "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": 20.0, "completions/max_terminated_length": 20.0, "completions/mean_length": 12.125, "completions/mean_terminated_length": 12.125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.10267857142857142, "frac_reward_zero_std": 0.625, "grad_norm": 10.595111846923828, "learning_rate": 6.227427435703995e-07, "loss": 0.0, "num_tokens": 4930573.0, "reward": 0.4437500238418579, "reward_std": 0.1462521106004715, "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": 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": 24.0, "completions/max_terminated_length": 24.0, "completions/mean_length": 12.46875, "completions/mean_terminated_length": 12.46875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.10491071428571429, "frac_reward_zero_std": 0.875, "grad_norm": 9.08613395690918, "learning_rate": 6.066498153718734e-07, "loss": 0.0, "num_tokens": 5006731.0, "reward": 0.6468750238418579, "reward_std": 0.0646936446428299, "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": 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": 19.0, "completions/max_terminated_length": 19.0, "completions/mean_length": 12.203125, "completions/mean_terminated_length": 12.203125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.10714285714285714, "frac_reward_zero_std": 0.75, "grad_norm": 8.32162094116211, "learning_rate": 5.90440267166055e-07, "loss": -0.0, "num_tokens": 5081592.0, "reward": 0.2562500238418579, "reward_std": 0.0883883461356163, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "step": 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": 18.0, "completions/max_terminated_length": 18.0, "completions/mean_length": 12.09375, "completions/mean_terminated_length": 12.09375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.109375, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 5.741318238559209e-07, "loss": 0.0, "num_tokens": 5182278.0, "reward": 0.7250000238418579, "reward_std": 0.0, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.625, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 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": 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": 0.11160714285714286, "frac_reward_zero_std": 0.875, "grad_norm": 4.81080961227417, "learning_rate": 5.577423184847931e-07, "loss": 0.0, "num_tokens": 5263119.0, "reward": 0.3031250238418579, "reward_std": 0.0646936446428299, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "step": 50 } ], "logging_steps": 1, "max_steps": 100, "num_input_tokens_seen": 5263119, "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 }