{ "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": 1764.0, "completions/mean_length": 685.28125, "completions/mean_terminated_length": 618.2622680664062, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.002232142857142857, "frac_reward_zero_std": 0.125, "grad_norm": 3.4717795848846436, "learning_rate": 0.0, "loss": -0.0, "num_tokens": 156770.0, "reward": 0.2343749850988388, "reward_std": 0.273389607667923, "rewards/format_reward/mean": 0.46875, "rewards/format_reward/std": 0.25, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "step": 1 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1645.0, "completions/mean_length": 531.640625, "completions/mean_terminated_length": 507.57147216796875, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.004464285714285714, "frac_reward_zero_std": 0.0, "grad_norm": 5.85898494720459, "learning_rate": 2e-07, "loss": -0.0, "num_tokens": 287467.0, "reward": 0.15390624105930328, "reward_std": 0.22925737500190735, "rewards/format_reward/mean": 0.2890625, "rewards/format_reward/std": 0.2643453776836395, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1352.0, "completions/mean_length": 592.78125, "completions/mean_terminated_length": 545.8386840820312, "completions/min_length": 9.0, "completions/min_terminated_length": 9.0, "epoch": 0.006696428571428571, "frac_reward_zero_std": 0.0, "grad_norm": 5.323749542236328, "learning_rate": 4e-07, "loss": 0.0, "num_tokens": 416525.0, "reward": 0.11484375596046448, "reward_std": 0.14524583518505096, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.2987033724784851, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1899.0, "completions/mean_length": 576.625, "completions/mean_terminated_length": 529.1612548828125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.008928571428571428, "frac_reward_zero_std": 0.125, "grad_norm": 5.891542434692383, "learning_rate": 6e-07, "loss": 0.0, "num_tokens": 544053.0, "reward": 0.2671874761581421, "reward_std": 0.34134700894355774, "rewards/format_reward/mean": 0.328125, "rewards/format_reward/std": 0.2847827076911926, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 1201.0, "completions/mean_length": 457.953125, "completions/mean_terminated_length": 432.7143249511719, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.011160714285714286, "frac_reward_zero_std": 0.125, "grad_norm": 16.81123161315918, "learning_rate": 8e-07, "loss": 0.0, "num_tokens": 660802.0, "reward": 0.3984375, "reward_std": 0.2901628017425537, "rewards/format_reward/mean": 0.390625, "rewards/format_reward/std": 0.2741328477859497, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "step": 5 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1577.0, "completions/mean_length": 612.15625, "completions/mean_terminated_length": 589.3651123046875, "completions/min_length": 8.0, "completions/min_terminated_length": 8.0, "epoch": 0.013392857142857142, "frac_reward_zero_std": 0.0, "grad_norm": 8.97691535949707, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 782508.0, "reward": 0.4132812023162842, "reward_std": 0.31584274768829346, "rewards/format_reward/mean": 0.3828125, "rewards/format_reward/std": 0.31800705194473267, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 1857.0, "completions/max_terminated_length": 1857.0, "completions/mean_length": 465.375, "completions/mean_terminated_length": 465.375, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.015625, "frac_reward_zero_std": 0.125, "grad_norm": 6.818016052246094, "learning_rate": 9.99726628670463e-07, "loss": 0.0, "num_tokens": 899100.0, "reward": 0.11953124403953552, "reward_std": 0.2439277023077011, "rewards/format_reward/mean": 0.2578125, "rewards/format_reward/std": 0.25185325741767883, "rewards/mcq_exact_match_reward/mean": 0.09375, "rewards/mcq_exact_match_reward/std": 0.29378482699394226, "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": 1812.0, "completions/max_terminated_length": 1812.0, "completions/mean_length": 626.390625, "completions/mean_terminated_length": 626.390625, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.017857142857142856, "frac_reward_zero_std": 0.0, "grad_norm": 17.15423011779785, "learning_rate": 9.989068136093872e-07, "loss": -0.0, "num_tokens": 1045413.0, "reward": 0.42890623211860657, "reward_std": 0.2928798794746399, "rewards/format_reward/mean": 0.3828125, "rewards/format_reward/std": 0.21347814798355103, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 1152.0, "completions/mean_length": 554.1875, "completions/mean_terminated_length": 530.4761962890625, "completions/min_length": 139.0, "completions/min_terminated_length": 139.0, "epoch": 0.020089285714285716, "frac_reward_zero_std": 0.0, "grad_norm": 1.0637177228927612, "learning_rate": 9.975414512725056e-07, "loss": 0.0, "num_tokens": 1163073.0, "reward": 0.2734375, "reward_std": 0.3395315408706665, "rewards/format_reward/mean": 0.390625, "rewards/format_reward/std": 0.2592533528804779, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "step": 9 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1679.0, "completions/mean_length": 509.59375, "completions/mean_terminated_length": 459.9677429199219, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.022321428571428572, "frac_reward_zero_std": 0.0, "grad_norm": 14.886730194091797, "learning_rate": 9.956320346634875e-07, "loss": -0.0, "num_tokens": 1294495.0, "reward": 0.38124996423721313, "reward_std": 0.2799506187438965, "rewards/format_reward/mean": 0.375, "rewards/format_reward/std": 0.26726123690605164, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "step": 10 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1826.0, "completions/max_terminated_length": 1826.0, "completions/mean_length": 414.34375, "completions/mean_terminated_length": 414.34375, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.024553571428571428, "frac_reward_zero_std": 0.125, "grad_norm": 8.528685569763184, "learning_rate": 9.931806517013612e-07, "loss": -0.0, "num_tokens": 1398101.0, "reward": 0.31171873211860657, "reward_std": 0.3370498716831207, "rewards/format_reward/mean": 0.3046875, "rewards/format_reward/std": 0.32874447107315063, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 11 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1712.0, "completions/max_terminated_length": 1712.0, "completions/mean_length": 528.40625, "completions/mean_terminated_length": 528.40625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.026785714285714284, "frac_reward_zero_std": 0.0, "grad_norm": 5.6365132331848145, "learning_rate": 9.901899829374047e-07, "loss": -0.0, "num_tokens": 1525559.0, "reward": 0.32734373211860657, "reward_std": 0.4506267011165619, "rewards/format_reward/mean": 0.4609375, "rewards/format_reward/std": 0.28597769141197205, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 12 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1688.0, "completions/mean_length": 539.84375, "completions/mean_terminated_length": 465.672119140625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.029017857142857144, "frac_reward_zero_std": 0.125, "grad_norm": 7.917036056518555, "learning_rate": 9.866632986240029e-07, "loss": -0.0, "num_tokens": 1658021.0, "reward": 0.2914062440395355, "reward_std": 0.30685722827911377, "rewards/format_reward/mean": 0.4140625, "rewards/format_reward/std": 0.19012710452079773, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1840.0, "completions/mean_length": 649.96875, "completions/mean_terminated_length": 604.8709716796875, "completions/min_length": 127.0, "completions/min_terminated_length": 127.0, "epoch": 0.03125, "frac_reward_zero_std": 0.125, "grad_norm": 1.2111611366271973, "learning_rate": 9.826044551386742e-07, "loss": -0.0, "num_tokens": 1786931.0, "reward": 0.12031249701976776, "reward_std": 0.20228664577007294, "rewards/format_reward/mean": 0.421875, "rewards/format_reward/std": 0.25539806485176086, "rewards/mcq_exact_match_reward/mean": 0.078125, "rewards/mcq_exact_match_reward/std": 0.27048972249031067, "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": 1927.0, "completions/mean_length": 543.21875, "completions/mean_terminated_length": 519.3333740234375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.033482142857142856, "frac_reward_zero_std": 0.0, "grad_norm": 7.481484889984131, "learning_rate": 9.780178907671788e-07, "loss": -0.0, "num_tokens": 1909233.0, "reward": 0.2867187261581421, "reward_std": 0.3402034342288971, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.22257846593856812, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 1413.0, "completions/mean_length": 566.890625, "completions/mean_terminated_length": 543.3809814453125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.03571428571428571, "frac_reward_zero_std": 0.0, "grad_norm": 3.754464626312256, "learning_rate": 9.729086208503173e-07, "loss": -0.0, "num_tokens": 2034146.0, "reward": 0.23984374105930328, "reward_std": 0.30626606941223145, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.2708333432674408, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "step": 16 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1942.0, "completions/mean_length": 623.46875, "completions/mean_terminated_length": 600.857177734375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.03794642857142857, "frac_reward_zero_std": 0.0, "grad_norm": 3.560889720916748, "learning_rate": 9.672822322997304e-07, "loss": 0.0, "num_tokens": 2180672.0, "reward": 0.2421874850988388, "reward_std": 0.3310280442237854, "rewards/format_reward/mean": 0.390625, "rewards/format_reward/std": 0.24346621334552765, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "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": 1735.0, "completions/mean_length": 696.171875, "completions/mean_terminated_length": 652.5645141601562, "completions/min_length": 220.0, "completions/min_terminated_length": 220.0, "epoch": 0.04017857142857143, "frac_reward_zero_std": 0.0, "grad_norm": 0.6656014323234558, "learning_rate": 9.611448774886923e-07, "loss": 0.0, "num_tokens": 2338299.0, "reward": 0.20468749105930328, "reward_std": 0.3216380476951599, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.2514837086200714, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "step": 18 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1630.0, "completions/mean_length": 486.171875, "completions/mean_terminated_length": 461.3809814453125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.04241071428571429, "frac_reward_zero_std": 0.25, "grad_norm": 9.904446601867676, "learning_rate": 9.545032675245813e-07, "loss": 0.0, "num_tokens": 2452854.0, "reward": 0.46015626192092896, "reward_std": 0.34995272755622864, "rewards/format_reward/mean": 0.5390625, "rewards/format_reward/std": 0.27174752950668335, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 1333.0, "completions/max_terminated_length": 1333.0, "completions/mean_length": 354.046875, "completions/mean_terminated_length": 354.046875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.044642857142857144, "frac_reward_zero_std": 0.0, "grad_norm": 14.993556022644043, "learning_rate": 9.473646649103817e-07, "loss": -0.0, "num_tokens": 2576777.0, "reward": 0.4492187201976776, "reward_std": 0.4537230432033539, "rewards/format_reward/mean": 0.5859375, "rewards/format_reward/std": 0.3398726284503937, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 1101.0, "completions/max_terminated_length": 1101.0, "completions/mean_length": 538.484375, "completions/mean_terminated_length": 538.484375, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.046875, "frac_reward_zero_std": 0.0, "grad_norm": 2.769683361053467, "learning_rate": 9.397368756032444e-07, "loss": -0.0, "num_tokens": 2721288.0, "reward": 0.37031248211860657, "reward_std": 0.47065046429634094, "rewards/format_reward/mean": 0.578125, "rewards/format_reward/std": 0.2847827076911926, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1826.0, "completions/mean_length": 603.53125, "completions/mean_terminated_length": 532.4917602539062, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.049107142857142856, "frac_reward_zero_std": 0.125, "grad_norm": 4.0598554611206055, "learning_rate": 9.316282404787869e-07, "loss": -0.0, "num_tokens": 2850274.0, "reward": 0.39140623807907104, "reward_std": 0.25593301653862, "rewards/format_reward/mean": 0.6328125, "rewards/format_reward/std": 0.3590351641178131, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 0.0625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1540.0, "completions/mean_length": 451.734375, "completions/mean_terminated_length": 345.3166809082031, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.05133928571428571, "frac_reward_zero_std": 0.0, "grad_norm": 9.803508758544922, "learning_rate": 9.230476262104676e-07, "loss": 0.0, "num_tokens": 2995873.0, "reward": 0.2632812261581421, "reward_std": 0.2542516589164734, "rewards/format_reward/mean": 0.6015625, "rewards/format_reward/std": 0.3692515790462494, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "step": 23 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1625.0, "completions/max_terminated_length": 1625.0, "completions/mean_length": 294.265625, "completions/mean_terminated_length": 294.265625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05357142857142857, "frac_reward_zero_std": 0.125, "grad_norm": 24.852619171142578, "learning_rate": 9.1400441557401e-07, "loss": -0.0, "num_tokens": 3103778.0, "reward": 0.27265626192092896, "reward_std": 0.20474214851856232, "rewards/format_reward/mean": 0.6953125, "rewards/format_reward/std": 0.394578218460083, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "step": 24 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1746.0, "completions/mean_length": 419.0625, "completions/mean_terminated_length": 338.9508056640625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05580357142857143, "frac_reward_zero_std": 0.125, "grad_norm": 7.306559085845947, "learning_rate": 9.045084971874737e-07, "loss": -0.0, "num_tokens": 3231798.0, "reward": 0.5171874761581421, "reward_std": 0.35653990507125854, "rewards/format_reward/mean": 0.640625, "rewards/format_reward/std": 0.301698237657547, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 1307.0, "completions/max_terminated_length": 1307.0, "completions/mean_length": 209.140625, "completions/mean_terminated_length": 209.140625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05803571428571429, "frac_reward_zero_std": 0.125, "grad_norm": 41.05641555786133, "learning_rate": 8.945702546981968e-07, "loss": 0.0, "num_tokens": 3351783.0, "reward": 0.28437501192092896, "reward_std": 0.2017095386981964, "rewards/format_reward/mean": 0.8125, "rewards/format_reward/std": 0.3021090030670166, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "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": 1823.0, "completions/max_terminated_length": 1823.0, "completions/mean_length": 361.203125, "completions/mean_terminated_length": 361.203125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.060267857142857144, "frac_reward_zero_std": 0.125, "grad_norm": 19.25481605529785, "learning_rate": 8.842005554284295e-07, "loss": -0.0, "num_tokens": 3483740.0, "reward": 0.27031248807907104, "reward_std": 0.34534239768981934, "rewards/format_reward/mean": 0.671875, "rewards/format_reward/std": 0.38025420904159546, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "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": 1101.0, "completions/max_terminated_length": 1101.0, "completions/mean_length": 193.515625, "completions/mean_terminated_length": 193.515625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0625, "frac_reward_zero_std": 0.25, "grad_norm": 15.254220008850098, "learning_rate": 8.734107384920769e-07, "loss": 0.0, "num_tokens": 3610165.0, "reward": 0.43281251192092896, "reward_std": 0.22094213962554932, "rewards/format_reward/mean": 0.734375, "rewards/format_reward/std": 0.3669776916503906, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 1832.0, "completions/max_terminated_length": 1832.0, "completions/mean_length": 237.296875, "completions/mean_terminated_length": 237.296875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.06473214285714286, "frac_reward_zero_std": 0.125, "grad_norm": 21.24951171875, "learning_rate": 8.622126023955445e-07, "loss": 0.0, "num_tokens": 3724232.0, "reward": 0.5414062738418579, "reward_std": 0.31315740942955017, "rewards/format_reward/mean": 0.8828125, "rewards/format_reward/std": 0.21347814798355103, "rewards/mcq_exact_match_reward/mean": 0.453125, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 733.0, "completions/mean_length": 109.9375, "completions/mean_terminated_length": 79.17460632324219, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.06696428571428571, "frac_reward_zero_std": 0.5, "grad_norm": 7.420022964477539, "learning_rate": 8.506183921362442e-07, "loss": -0.0, "num_tokens": 3802340.0, "reward": 0.737500011920929, "reward_std": 0.22764386236667633, "rewards/format_reward/mean": 0.96875, "rewards/format_reward/std": 0.1510545015335083, "rewards/mcq_exact_match_reward/mean": 0.640625, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 236.0, "completions/max_terminated_length": 236.0, "completions/mean_length": 28.8125, "completions/mean_terminated_length": 28.8125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.06919642857142858, "frac_reward_zero_std": 0.5, "grad_norm": 17.47530746459961, "learning_rate": 8.386407858128706e-07, "loss": -0.0, "num_tokens": 3880016.0, "reward": 0.4890625476837158, "reward_std": 0.12697681784629822, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.08768405020236969, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 268.0, "completions/max_terminated_length": 268.0, "completions/mean_length": 34.15625, "completions/mean_terminated_length": 34.15625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.07142857142857142, "frac_reward_zero_std": 0.5, "grad_norm": 10.235955238342285, "learning_rate": 8.262928807620843e-07, "loss": 0.0, "num_tokens": 3959194.0, "reward": 0.5203125476837158, "reward_std": 0.2278592735528946, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 32 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1299.0, "completions/mean_length": 82.21875, "completions/mean_terminated_length": 51.01587677001953, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.07366071428571429, "frac_reward_zero_std": 0.125, "grad_norm": 14.980599403381348, "learning_rate": 8.135881792367685e-07, "loss": 0.0, "num_tokens": 4056352.0, "reward": 0.6765625476837158, "reward_std": 0.3137786388397217, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.578125, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 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": 353.0, "completions/max_terminated_length": 353.0, "completions/mean_length": 27.109375, "completions/mean_terminated_length": 27.109375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.07589285714285714, "frac_reward_zero_std": 0.875, "grad_norm": 2.6166975498199463, "learning_rate": 8.005405736415125e-07, "loss": -0.0, "num_tokens": 4158159.0, "reward": 0.4906250238418579, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 274.0, "completions/max_terminated_length": 274.0, "completions/mean_length": 21.484375, "completions/mean_terminated_length": 21.484375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.078125, "frac_reward_zero_std": 0.5, "grad_norm": 10.179203987121582, "learning_rate": 7.871643313414718e-07, "loss": 0.0, "num_tokens": 4261702.0, "reward": 0.5375000238418579, "reward_std": 0.19727616012096405, "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": 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": 14.0, "completions/max_terminated_length": 14.0, "completions/mean_length": 12.109375, "completions/mean_terminated_length": 12.109375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.08035714285714286, "frac_reward_zero_std": 0.5, "grad_norm": 13.746647834777832, "learning_rate": 7.734740790612136e-07, "loss": 0.0, "num_tokens": 4355221.0, "reward": 0.5960937738418579, "reward_std": 0.17209626734256744, "rewards/format_reward/mean": 0.9609375, "rewards/format_reward/std": 0.18483558297157288, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "step": 36 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1133.0, "completions/max_terminated_length": 1133.0, "completions/mean_length": 31.921875, "completions/mean_terminated_length": 31.921875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.08258928571428571, "frac_reward_zero_std": 0.625, "grad_norm": 13.752304077148438, "learning_rate": 7.594847868906076e-07, "loss": -0.0, "num_tokens": 4430720.0, "reward": 0.3031250238418579, "reward_std": 0.189372718334198, "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": 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": 597.0, "completions/max_terminated_length": 597.0, "completions/mean_length": 30.390625, "completions/mean_terminated_length": 30.390625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.08482142857142858, "frac_reward_zero_std": 0.5, "grad_norm": 13.630165100097656, "learning_rate": 7.452117519152541e-07, "loss": 0.0, "num_tokens": 4521617.0, "reward": 0.6156250238418579, "reward_std": 0.23144522309303284, "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": 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": 353.0, "completions/max_terminated_length": 353.0, "completions/mean_length": 22.125, "completions/mean_terminated_length": 22.125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.08705357142857142, "frac_reward_zero_std": 0.625, "grad_norm": 11.657012939453125, "learning_rate": 7.306705814893439e-07, "loss": 0.0, "num_tokens": 4637849.0, "reward": 0.5843750238418579, "reward_std": 0.13258251547813416, "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": 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": 17.0, "completions/max_terminated_length": 17.0, "completions/mean_length": 12.078125, "completions/mean_terminated_length": 12.078125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.08928571428571429, "frac_reward_zero_std": 0.625, "grad_norm": 6.2598443031311035, "learning_rate": 7.158771761692464e-07, "loss": 0.0, "num_tokens": 4730630.0, "reward": 0.3812500238418579, "reward_std": 0.1462521106004715, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "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": 19.0, "completions/max_terminated_length": 19.0, "completions/mean_length": 12.3125, "completions/mean_terminated_length": 12.3125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.09151785714285714, "frac_reward_zero_std": 0.5, "grad_norm": 9.939923286437988, "learning_rate": 7.008477123264847e-07, "loss": 0.0, "num_tokens": 4822458.0, "reward": 0.6148437857627869, "reward_std": 0.19000458717346191, "rewards/format_reward/mean": 0.9921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "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": 185.0, "completions/max_terminated_length": 185.0, "completions/mean_length": 15.671875, "completions/mean_terminated_length": 15.671875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.09375, "frac_reward_zero_std": 0.625, "grad_norm": 7.561997890472412, "learning_rate": 6.855986244591103e-07, "loss": -0.0, "num_tokens": 4909189.0, "reward": 0.6937500238418579, "reward_std": 0.1462521106004715, "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": 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": 12.015625, "completions/mean_terminated_length": 12.015625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.09598214285714286, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 6.701465872208216e-07, "loss": 0.0, "num_tokens": 4985798.0, "reward": 0.10000000149011612, "reward_std": 0.0, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.0, "rewards/mcq_exact_match_reward/std": 0.0, "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": 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": 0.09821428571428571, "frac_reward_zero_std": 0.5, "grad_norm": 15.431774139404297, "learning_rate": 6.545084971874736e-07, "loss": -0.0, "num_tokens": 5065070.0, "reward": 0.43437501788139343, "reward_std": 0.15203848481178284, "rewards/format_reward/mean": 0.90625, "rewards/format_reward/std": 0.29378482699394226, "rewards/mcq_exact_match_reward/mean": 0.34375, "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": 231.0, "completions/max_terminated_length": 231.0, "completions/mean_length": 17.328125, "completions/mean_terminated_length": 17.328125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.10044642857142858, "frac_reward_zero_std": 0.75, "grad_norm": 9.880815505981445, "learning_rate": 6.387014543809223e-07, "loss": 0.0, "num_tokens": 5162771.0, "reward": 0.4281250238418579, "reward_std": 0.12255740165710449, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 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": 803.0, "completions/max_terminated_length": 803.0, "completions/mean_length": 25.25, "completions/mean_terminated_length": 25.25, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.10267857142857142, "frac_reward_zero_std": 0.75, "grad_norm": 17.39849281311035, "learning_rate": 6.227427435703995e-07, "loss": 0.0, "num_tokens": 5279363.0, "reward": 0.4593750238418579, "reward_std": 0.1315089464187622, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 12.078125, "completions/mean_terminated_length": 12.078125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.10491071428571429, "frac_reward_zero_std": 0.75, "grad_norm": 11.483086585998535, "learning_rate": 6.066498153718734e-07, "loss": 0.0, "num_tokens": 5370520.0, "reward": 0.5218750238418579, "reward_std": 0.11100947856903076, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 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": 157.0, "completions/max_terminated_length": 157.0, "completions/mean_length": 16.734375, "completions/mean_terminated_length": 16.734375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.10714285714285714, "frac_reward_zero_std": 0.75, "grad_norm": 13.122138977050781, "learning_rate": 5.90440267166055e-07, "loss": -0.0, "num_tokens": 5456783.0, "reward": 0.3812500238418579, "reward_std": 0.0883883461356163, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "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": 0.109375, "frac_reward_zero_std": 0.625, "grad_norm": 9.711867332458496, "learning_rate": 5.741318238559209e-07, "loss": 0.0, "num_tokens": 5566751.0, "reward": 0.5843750238418579, "reward_std": 0.15992169082164764, "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": 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": 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": 0.11160714285714286, "frac_reward_zero_std": 0.875, "grad_norm": 3.2205755710601807, "learning_rate": 5.577423184847931e-07, "loss": -0.0, "num_tokens": 5656215.0, "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": 50 } ], "logging_steps": 1, "max_steps": 100, "num_input_tokens_seen": 5656215, "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 }