{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.25, "eval_steps": 500, "global_step": 100, "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": 1639.0, "completions/mean_length": 657.09375, "completions/mean_terminated_length": 635.0159301757812, "completions/min_length": 23.0, "completions/min_terminated_length": 23.0, "epoch": 0.0025, "frac_reward_zero_std": 0.0, "grad_norm": 1.382034182548523, "learning_rate": 0.0, "loss": 0.0, "num_tokens": 119230.0, "reward": 0.22890622913837433, "reward_std": 0.32265496253967285, "rewards/format_reward/mean": 0.4140625, "rewards/format_reward/std": 0.2280818521976471, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1431.0, "completions/mean_length": 530.390625, "completions/mean_terminated_length": 481.4354553222656, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.005, "frac_reward_zero_std": 0.0, "grad_norm": 4.274089336395264, "learning_rate": 2e-07, "loss": -0.0, "num_tokens": 236031.0, "reward": 0.2906249761581421, "reward_std": 0.41764312982559204, "rewards/format_reward/mean": 0.40625, "rewards/format_reward/std": 0.279951810836792, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 1717.0, "completions/max_terminated_length": 1717.0, "completions/mean_length": 447.28125, "completions/mean_terminated_length": 447.28125, "completions/min_length": 90.0, "completions/min_terminated_length": 90.0, "epoch": 0.0075, "frac_reward_zero_std": 0.0, "grad_norm": 1.3015592098236084, "learning_rate": 4e-07, "loss": 0.0, "num_tokens": 344849.0, "reward": 0.5367187261581421, "reward_std": 0.3515271842479706, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.2987033724784851, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "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": 0.0, "completions/max_length": 1284.0, "completions/max_terminated_length": 1284.0, "completions/mean_length": 449.421875, "completions/mean_terminated_length": 449.421875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.01, "frac_reward_zero_std": 0.125, "grad_norm": 4.185198783874512, "learning_rate": 6e-07, "loss": -0.0, "num_tokens": 473748.0, "reward": 0.1289062350988388, "reward_std": 0.24294739961624146, "rewards/format_reward/mean": 0.3515625, "rewards/format_reward/std": 0.2302463799715042, "rewards/mcq_exact_match_reward/mean": 0.09375, "rewards/mcq_exact_match_reward/std": 0.29378482699394226, "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": 873.0, "completions/mean_length": 474.328125, "completions/mean_terminated_length": 449.3492431640625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0125, "frac_reward_zero_std": 0.0, "grad_norm": 8.912360191345215, "learning_rate": 8e-07, "loss": -0.0, "num_tokens": 592969.0, "reward": 0.3828124701976776, "reward_std": 0.38378074765205383, "rewards/format_reward/mean": 0.390625, "rewards/format_reward/std": 0.28824523091316223, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "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": 0.078125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1662.0, "completions/mean_length": 646.96875, "completions/mean_terminated_length": 528.2373046875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.015, "frac_reward_zero_std": 0.0, "grad_norm": 9.784747123718262, "learning_rate": 1e-06, "loss": 0.0, "num_tokens": 748743.0, "reward": 0.2554687261581421, "reward_std": 0.293965220451355, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.23974503576755524, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "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": 1555.0, "completions/max_terminated_length": 1555.0, "completions/mean_length": 436.375, "completions/mean_terminated_length": 436.375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0175, "frac_reward_zero_std": 0.0, "grad_norm": 9.773689270019531, "learning_rate": 9.99726628670463e-07, "loss": -0.0, "num_tokens": 878191.0, "reward": 0.3531249761581421, "reward_std": 0.3999304175376892, "rewards/format_reward/mean": 0.40625, "rewards/format_reward/std": 0.19669894874095917, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1206.0, "completions/mean_length": 464.90625, "completions/mean_terminated_length": 439.7778015136719, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.02, "frac_reward_zero_std": 0.0, "grad_norm": 11.763710975646973, "learning_rate": 9.989068136093872e-07, "loss": -0.0, "num_tokens": 986217.0, "reward": 0.33281248807907104, "reward_std": 0.437870591878891, "rewards/format_reward/mean": 0.359375, "rewards/format_reward/std": 0.301698237657547, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "step": 8 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1965.0, "completions/mean_length": 579.140625, "completions/mean_terminated_length": 555.825439453125, "completions/min_length": 46.0, "completions/min_terminated_length": 46.0, "epoch": 0.0225, "frac_reward_zero_std": 0.125, "grad_norm": 1.4943907260894775, "learning_rate": 9.975414512725056e-07, "loss": 0.0, "num_tokens": 1118698.0, "reward": 0.2757812440395355, "reward_std": 0.30947786569595337, "rewards/format_reward/mean": 0.4140625, "rewards/format_reward/std": 0.244862899184227, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1914.0, "completions/mean_length": 574.078125, "completions/mean_terminated_length": 501.59014892578125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.025, "frac_reward_zero_std": 0.0, "grad_norm": 1.6521011590957642, "learning_rate": 9.956320346634875e-07, "loss": -0.0, "num_tokens": 1274359.0, "reward": 0.29765623807907104, "reward_std": 0.40693485736846924, "rewards/format_reward/mean": 0.3203125, "rewards/format_reward/std": 0.24180518090724945, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 1337.0, "completions/mean_length": 543.453125, "completions/mean_terminated_length": 494.9193420410156, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0275, "frac_reward_zero_std": 0.0, "grad_norm": 15.775018692016602, "learning_rate": 9.931806517013612e-07, "loss": -0.0, "num_tokens": 1401548.0, "reward": 0.3531250059604645, "reward_std": 0.37828850746154785, "rewards/format_reward/mean": 0.40625, "rewards/format_reward/std": 0.3069951832294464, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 1339.0, "completions/max_terminated_length": 1339.0, "completions/mean_length": 487.6875, "completions/mean_terminated_length": 487.6875, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.03, "frac_reward_zero_std": 0.0, "grad_norm": 3.059255838394165, "learning_rate": 9.901899829374047e-07, "loss": 0.0, "num_tokens": 1510656.0, "reward": 0.43828123807907104, "reward_std": 0.3373667001724243, "rewards/format_reward/mean": 0.4765625, "rewards/format_reward/std": 0.25865477323532104, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 2024.0, "completions/mean_length": 498.578125, "completions/mean_terminated_length": 473.9841613769531, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0325, "frac_reward_zero_std": 0.0, "grad_norm": 4.521096229553223, "learning_rate": 9.866632986240029e-07, "loss": -0.0, "num_tokens": 1624901.0, "reward": 0.23984375596046448, "reward_std": 0.3577025532722473, "rewards/format_reward/mean": 0.5234375, "rewards/format_reward/std": 0.36110159754753113, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "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": 0.0625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1732.0, "completions/mean_length": 534.8125, "completions/mean_terminated_length": 433.933349609375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.035, "frac_reward_zero_std": 0.0, "grad_norm": 6.450885772705078, "learning_rate": 9.826044551386742e-07, "loss": -0.0, "num_tokens": 1789881.0, "reward": 0.3296874761581421, "reward_std": 0.3826944828033447, "rewards/format_reward/mean": 0.328125, "rewards/format_reward/std": 0.2847827076911926, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 1480.0, "completions/max_terminated_length": 1480.0, "completions/mean_length": 416.71875, "completions/mean_terminated_length": 416.71875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0375, "frac_reward_zero_std": 0.0, "grad_norm": 4.970447540283203, "learning_rate": 9.780178907671788e-07, "loss": 0.0, "num_tokens": 1897015.0, "reward": 0.28203126788139343, "reward_std": 0.2876996695995331, "rewards/format_reward/mean": 0.4765625, "rewards/format_reward/std": 0.42191633582115173, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 1281.0, "completions/max_terminated_length": 1281.0, "completions/mean_length": 372.8125, "completions/mean_terminated_length": 372.8125, "completions/min_length": 7.0, "completions/min_terminated_length": 7.0, "epoch": 0.04, "frac_reward_zero_std": 0.125, "grad_norm": 5.324711322784424, "learning_rate": 9.729086208503173e-07, "loss": 0.0, "num_tokens": 2014515.0, "reward": 0.45703125, "reward_std": 0.3959631323814392, "rewards/format_reward/mean": 0.3515625, "rewards/format_reward/std": 0.3952063024044037, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 1216.0, "completions/mean_length": 343.078125, "completions/mean_terminated_length": 316.0158996582031, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.0425, "frac_reward_zero_std": 0.0, "grad_norm": 9.116941452026367, "learning_rate": 9.672822322997304e-07, "loss": 0.0, "num_tokens": 2109992.0, "reward": 0.24296875298023224, "reward_std": 0.30575716495513916, "rewards/format_reward/mean": 0.5546875, "rewards/format_reward/std": 0.4189668893814087, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "step": 17 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1191.0, "completions/mean_length": 470.703125, "completions/mean_terminated_length": 445.66668701171875, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.045, "frac_reward_zero_std": 0.0, "grad_norm": 18.01677131652832, "learning_rate": 9.611448774886923e-07, "loss": -0.0, "num_tokens": 2247021.0, "reward": 0.31718748807907104, "reward_std": 0.2856425940990448, "rewards/format_reward/mean": 0.359375, "rewards/format_reward/std": 0.3503824472427368, "rewards/mcq_exact_match_reward/mean": 0.28125, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1253.0, "completions/mean_length": 443.984375, "completions/mean_terminated_length": 392.2419128417969, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.0475, "frac_reward_zero_std": 0.0, "grad_norm": 6.768802165985107, "learning_rate": 9.545032675245813e-07, "loss": -0.0, "num_tokens": 2359900.0, "reward": 0.41328123211860657, "reward_std": 0.4769665002822876, "rewards/format_reward/mean": 0.5390625, "rewards/format_reward/std": 0.37059250473976135, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1511.0, "completions/mean_length": 465.90625, "completions/mean_terminated_length": 440.7936706542969, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.05, "frac_reward_zero_std": 0.0, "grad_norm": 7.05373477935791, "learning_rate": 9.473646649103817e-07, "loss": 0.0, "num_tokens": 2487934.0, "reward": 0.31953126192092896, "reward_std": 0.44596385955810547, "rewards/format_reward/mean": 0.5390625, "rewards/format_reward/std": 0.35972529649734497, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 1650.0, "completions/max_terminated_length": 1650.0, "completions/mean_length": 358.578125, "completions/mean_terminated_length": 358.578125, "completions/min_length": 11.0, "completions/min_terminated_length": 11.0, "epoch": 0.0525, "frac_reward_zero_std": 0.0, "grad_norm": 17.915525436401367, "learning_rate": 9.397368756032444e-07, "loss": -0.0, "num_tokens": 2609083.0, "reward": 0.22968751192092896, "reward_std": 0.3376360535621643, "rewards/format_reward/mean": 0.578125, "rewards/format_reward/std": 0.419927716255188, "rewards/mcq_exact_match_reward/mean": 0.171875, "rewards/mcq_exact_match_reward/std": 0.38025420904159546, "step": 21 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1009.0, "completions/mean_length": 268.859375, "completions/mean_terminated_length": 240.6190643310547, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.055, "frac_reward_zero_std": 0.0, "grad_norm": 9.154450416564941, "learning_rate": 9.316282404787869e-07, "loss": 0.0, "num_tokens": 2710506.0, "reward": 0.40234375, "reward_std": 0.4602089822292328, "rewards/format_reward/mean": 0.5859375, "rewards/format_reward/std": 0.4038965702056885, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1738.0, "completions/mean_length": 407.109375, "completions/mean_terminated_length": 326.4098205566406, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.0575, "frac_reward_zero_std": 0.0, "grad_norm": 13.803871154785156, "learning_rate": 9.230476262104676e-07, "loss": 0.0, "num_tokens": 2812017.0, "reward": 0.27265626192092896, "reward_std": 0.325811505317688, "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": 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": 1871.0, "completions/mean_length": 323.765625, "completions/mean_terminated_length": 296.3968505859375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.06, "frac_reward_zero_std": 0.0, "grad_norm": 18.435890197753906, "learning_rate": 9.1400441557401e-07, "loss": 0.0, "num_tokens": 2927170.0, "reward": 0.2796875238418579, "reward_std": 0.2613222002983093, "rewards/format_reward/mean": 0.609375, "rewards/format_reward/std": 0.40305137634277344, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1604.0, "completions/mean_length": 364.0, "completions/mean_terminated_length": 337.2698669433594, "completions/min_length": 11.0, "completions/min_terminated_length": 11.0, "epoch": 0.0625, "frac_reward_zero_std": 0.0, "grad_norm": 19.538883209228516, "learning_rate": 9.045084971874737e-07, "loss": -0.0, "num_tokens": 3073786.0, "reward": 0.36328125, "reward_std": 0.26463496685028076, "rewards/format_reward/mean": 0.6640625, "rewards/format_reward/std": 0.38888004422187805, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 990.0, "completions/mean_length": 143.96875, "completions/mean_terminated_length": 113.74604034423828, "completions/min_length": 9.0, "completions/min_terminated_length": 9.0, "epoch": 0.065, "frac_reward_zero_std": 0.125, "grad_norm": 16.721471786499023, "learning_rate": 8.945702546981968e-07, "loss": 0.0, "num_tokens": 3157032.0, "reward": 0.530468761920929, "reward_std": 0.3712921142578125, "rewards/format_reward/mean": 0.7734375, "rewards/format_reward/std": 0.38760238885879517, "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": 1373.0, "completions/max_terminated_length": 1373.0, "completions/mean_length": 84.53125, "completions/mean_terminated_length": 84.53125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0675, "frac_reward_zero_std": 0.375, "grad_norm": 9.660776138305664, "learning_rate": 8.842005554284295e-07, "loss": -0.0, "num_tokens": 3234450.0, "reward": 0.38750001788139343, "reward_std": 0.11329139769077301, "rewards/format_reward/mean": 0.90625, "rewards/format_reward/std": 0.25, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 842.0, "completions/mean_length": 159.28125, "completions/mean_terminated_length": 66.39344024658203, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07, "frac_reward_zero_std": 0.125, "grad_norm": 14.32498836517334, "learning_rate": 8.734107384920769e-07, "loss": -0.0, "num_tokens": 3332324.0, "reward": 0.5789062976837158, "reward_std": 0.31568628549575806, "rewards/format_reward/mean": 0.7890625, "rewards/format_reward/std": 0.34277912974357605, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "step": 28 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 628.0, "completions/max_terminated_length": 628.0, "completions/mean_length": 48.984375, "completions/mean_terminated_length": 48.984375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0725, "frac_reward_zero_std": 0.625, "grad_norm": 7.845927715301514, "learning_rate": 8.622126023955445e-07, "loss": -0.0, "num_tokens": 3434587.0, "reward": 0.44687503576278687, "reward_std": 0.17732149362564087, "rewards/format_reward/mean": 0.875, "rewards/format_reward/std": 0.30860671401023865, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 788.0, "completions/max_terminated_length": 788.0, "completions/mean_length": 58.84375, "completions/mean_terminated_length": 58.84375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.075, "frac_reward_zero_std": 0.125, "grad_norm": 14.274752616882324, "learning_rate": 8.506183921362442e-07, "loss": -0.0, "num_tokens": 3535785.0, "reward": 0.5140625238418579, "reward_std": 0.34334200620651245, "rewards/format_reward/mean": 0.921875, "rewards/format_reward/std": 0.25539806485176086, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 69.0, "completions/mean_length": 77.375, "completions/mean_terminated_length": 13.806450843811035, "completions/min_length": 11.0, "completions/min_terminated_length": 11.0, "epoch": 0.0775, "frac_reward_zero_std": 0.375, "grad_norm": 12.343521118164062, "learning_rate": 8.386407858128706e-07, "loss": 0.0, "num_tokens": 3635465.0, "reward": 0.1875, "reward_std": 0.2048131227493286, "rewards/format_reward/mean": 0.9375, "rewards/format_reward/std": 0.24397502839565277, "rewards/mcq_exact_match_reward/mean": 0.09375, "rewards/mcq_exact_match_reward/std": 0.29378482699394226, "step": 31 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 326.0, "completions/mean_length": 68.0, "completions/mean_terminated_length": 36.57143020629883, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.08, "frac_reward_zero_std": 0.375, "grad_norm": 14.074780464172363, "learning_rate": 8.262928807620843e-07, "loss": 0.0, "num_tokens": 3720145.0, "reward": 0.702343761920929, "reward_std": 0.2966356575489044, "rewards/format_reward/mean": 0.9296875, "rewards/format_reward/std": 0.2498759627342224, "rewards/mcq_exact_match_reward/mean": 0.609375, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 99.0, "completions/max_terminated_length": 99.0, "completions/mean_length": 15.3125, "completions/mean_terminated_length": 15.3125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0825, "frac_reward_zero_std": 0.5, "grad_norm": 15.079535484313965, "learning_rate": 8.135881792367685e-07, "loss": -0.0, "num_tokens": 3814877.0, "reward": 0.42656251788139343, "reward_std": 0.12697681784629822, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.08768405020236969, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 1271.0, "completions/max_terminated_length": 1271.0, "completions/mean_length": 44.9375, "completions/mean_terminated_length": 44.9375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.085, "frac_reward_zero_std": 0.625, "grad_norm": 5.545784950256348, "learning_rate": 8.005405736415125e-07, "loss": 0.0, "num_tokens": 3916945.0, "reward": 0.67578125, "reward_std": 0.13921165466308594, "rewards/format_reward/mean": 0.9765625, "rewards/format_reward/std": 0.13886408507823944, "rewards/mcq_exact_match_reward/mean": 0.578125, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 374.0, "completions/max_terminated_length": 374.0, "completions/mean_length": 18.515625, "completions/mean_terminated_length": 18.515625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0875, "frac_reward_zero_std": 0.125, "grad_norm": 28.097858428955078, "learning_rate": 7.871643313414718e-07, "loss": 0.0, "num_tokens": 4004202.0, "reward": 0.31640625, "reward_std": 0.3092179298400879, "rewards/format_reward/mean": 0.9765625, "rewards/format_reward/std": 0.13886408507823944, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "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": 1473.0, "completions/max_terminated_length": 1473.0, "completions/mean_length": 101.15625, "completions/mean_terminated_length": 101.15625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.09, "frac_reward_zero_std": 0.375, "grad_norm": 13.161815643310547, "learning_rate": 7.734740790612136e-07, "loss": -0.0, "num_tokens": 4118812.0, "reward": 0.4203125238418579, "reward_std": 0.2513866424560547, "rewards/format_reward/mean": 0.921875, "rewards/format_reward/std": 0.25539806485176086, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 353.0, "completions/max_terminated_length": 353.0, "completions/mean_length": 21.125, "completions/mean_terminated_length": 21.125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0925, "frac_reward_zero_std": 0.5, "grad_norm": 10.958316802978516, "learning_rate": 7.594847868906076e-07, "loss": 0.0, "num_tokens": 4234556.0, "reward": 0.4726562798023224, "reward_std": 0.22422221302986145, "rewards/format_reward/mean": 0.9765625, "rewards/format_reward/std": 0.10652101784944534, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 37 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.65625, "completions/mean_terminated_length": 12.65625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.095, "frac_reward_zero_std": 0.25, "grad_norm": 22.143489837646484, "learning_rate": 7.452117519152541e-07, "loss": 0.0, "num_tokens": 4313686.0, "reward": 0.5218750238418579, "reward_std": 0.3356248140335083, "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": 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": 124.0, "completions/max_terminated_length": 124.0, "completions/mean_length": 14.25, "completions/mean_terminated_length": 14.25, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0975, "frac_reward_zero_std": 0.875, "grad_norm": 3.8069100379943848, "learning_rate": 7.306705814893439e-07, "loss": -0.0, "num_tokens": 4407614.0, "reward": 0.3656250238418579, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 12.734375, "completions/mean_terminated_length": 12.734375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1, "frac_reward_zero_std": 0.5, "grad_norm": 18.4069766998291, "learning_rate": 7.158771761692464e-07, "loss": 0.0, "num_tokens": 4506413.0, "reward": 0.5375000238418579, "reward_std": 0.22461533546447754, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.4375, "rewards/mcq_exact_match_reward/std": 0.5, "step": 40 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 36.0, "completions/max_terminated_length": 36.0, "completions/mean_length": 12.765625, "completions/mean_terminated_length": 12.765625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1025, "frac_reward_zero_std": 0.75, "grad_norm": 9.79914665222168, "learning_rate": 7.008477123264847e-07, "loss": 0.0, "num_tokens": 4589486.0, "reward": 0.4281250238418579, "reward_std": 0.11100947856903076, "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": 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": 90.0, "completions/max_terminated_length": 90.0, "completions/mean_length": 13.953125, "completions/mean_terminated_length": 13.953125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.105, "frac_reward_zero_std": 0.5, "grad_norm": 12.262887001037598, "learning_rate": 6.855986244591103e-07, "loss": -0.0, "num_tokens": 4710843.0, "reward": 0.5531250238418579, "reward_std": 0.19939783215522766, "rewards/format_reward/mean": 1.0, "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": 135.0, "completions/max_terminated_length": 135.0, "completions/mean_length": 15.859375, "completions/mean_terminated_length": 15.859375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1075, "frac_reward_zero_std": 0.5, "grad_norm": 14.82951545715332, "learning_rate": 6.701465872208216e-07, "loss": 0.0, "num_tokens": 4803378.0, "reward": 0.45468753576278687, "reward_std": 0.1663910448551178, "rewards/format_reward/mean": 0.953125, "rewards/format_reward/std": 0.21304203569889069, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 17.0, "completions/max_terminated_length": 17.0, "completions/mean_length": 12.4375, "completions/mean_terminated_length": 12.4375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.11, "frac_reward_zero_std": 0.75, "grad_norm": 17.398311614990234, "learning_rate": 6.545084971874736e-07, "loss": 0.0, "num_tokens": 4884430.0, "reward": 0.16250000894069672, "reward_std": 0.10888782143592834, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.0625, "rewards/mcq_exact_match_reward/std": 0.24397502839565277, "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": 110.0, "completions/max_terminated_length": 110.0, "completions/mean_length": 14.4375, "completions/mean_terminated_length": 14.4375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1125, "frac_reward_zero_std": 0.375, "grad_norm": 19.101451873779297, "learning_rate": 6.387014543809223e-07, "loss": 0.0, "num_tokens": 5003522.0, "reward": 0.608593761920929, "reward_std": 0.24867939949035645, "rewards/format_reward/mean": 0.9296875, "rewards/format_reward/std": 0.19654127955436707, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "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": 142.0, "completions/max_terminated_length": 142.0, "completions/mean_length": 14.21875, "completions/mean_terminated_length": 14.21875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.115, "frac_reward_zero_std": 0.625, "grad_norm": 8.656336784362793, "learning_rate": 6.227427435703995e-07, "loss": 0.0, "num_tokens": 5085776.0, "reward": 0.6625000238418579, "reward_std": 0.1962026059627533, "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": 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": 130.0, "completions/max_terminated_length": 130.0, "completions/mean_length": 14.1875, "completions/mean_terminated_length": 14.1875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1175, "frac_reward_zero_std": 0.625, "grad_norm": 11.477785110473633, "learning_rate": 6.066498153718734e-07, "loss": 0.0, "num_tokens": 5180652.0, "reward": 0.7718750238418579, "reward_std": 0.1530819982290268, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.671875, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.09375, "completions/mean_terminated_length": 12.09375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.12, "frac_reward_zero_std": 0.75, "grad_norm": 7.277567386627197, "learning_rate": 5.90440267166055e-07, "loss": -0.0, "num_tokens": 5280978.0, "reward": 0.4750000238418579, "reward_std": 0.0883883461356163, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 11.6875, "completions/mean_terminated_length": 11.6875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1225, "frac_reward_zero_std": 0.25, "grad_norm": 28.435686111450195, "learning_rate": 5.741318238559209e-07, "loss": 0.0, "num_tokens": 5408774.0, "reward": 0.5492187738418579, "reward_std": 0.3656456470489502, "rewards/format_reward/mean": 0.9609375, "rewards/format_reward/std": 0.13524486124515533, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 12.140625, "completions/mean_terminated_length": 12.140625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.125, "frac_reward_zero_std": 0.75, "grad_norm": 9.53724193572998, "learning_rate": 5.577423184847931e-07, "loss": 0.0, "num_tokens": 5488895.0, "reward": 0.5062500238418579, "reward_std": 0.10888782143592834, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 50 }, { "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": 12.15625, "completions/mean_terminated_length": 12.15625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1275, "frac_reward_zero_std": 0.75, "grad_norm": 6.96045446395874, "learning_rate": 5.412896727361662e-07, "loss": 0.0, "num_tokens": 5593945.0, "reward": 0.5062500238418579, "reward_std": 0.10888782143592834, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 51 }, { "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": 36.0, "completions/max_terminated_length": 36.0, "completions/mean_length": 12.625, "completions/mean_terminated_length": 12.625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.13, "frac_reward_zero_std": 0.5, "grad_norm": 12.546669006347656, "learning_rate": 5.247918773366111e-07, "loss": 0.0, "num_tokens": 5707569.0, "reward": 0.6453125476837158, "reward_std": 0.23586463928222656, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.546875, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 52 }, { "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": 12.09375, "completions/mean_terminated_length": 12.09375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1325, "frac_reward_zero_std": 0.625, "grad_norm": 15.857230186462402, "learning_rate": 5.082669723831793e-07, "loss": 0.0, "num_tokens": 5776607.0, "reward": 0.5531250238418579, "reward_std": 0.15992169082164764, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 53 }, { "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": 203.0, "completions/max_terminated_length": 203.0, "completions/mean_length": 15.25, "completions/mean_terminated_length": 15.25, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.135, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 4.917330276168208e-07, "loss": 0.0, "num_tokens": 5869567.0, "reward": 0.4750000238418579, "reward_std": 0.0, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 54 }, { "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": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1375, "frac_reward_zero_std": 0.5, "grad_norm": 16.801105499267578, "learning_rate": 4.752081226633888e-07, "loss": 0.0, "num_tokens": 5946503.0, "reward": 0.6617187857627869, "reward_std": 0.22400324046611786, "rewards/format_reward/mean": 0.9921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.5625, "rewards/mcq_exact_match_reward/std": 0.5, "step": 55 }, { "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": 93.0, "completions/max_terminated_length": 93.0, "completions/mean_length": 13.546875, "completions/mean_terminated_length": 13.546875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.14, "frac_reward_zero_std": 0.625, "grad_norm": 15.91934871673584, "learning_rate": 4.5871032726383385e-07, "loss": 0.0, "num_tokens": 6056042.0, "reward": 0.5359375476837158, "reward_std": 0.11330723762512207, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.4375, "rewards/mcq_exact_match_reward/std": 0.5, "step": 56 }, { "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": 227.0, "completions/max_terminated_length": 227.0, "completions/mean_length": 15.609375, "completions/mean_terminated_length": 15.609375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1425, "frac_reward_zero_std": 0.625, "grad_norm": 16.60999298095703, "learning_rate": 4.4225768151520694e-07, "loss": 0.0, "num_tokens": 6149761.0, "reward": 0.4000000059604645, "reward_std": 0.1462520956993103, "rewards/format_reward/mean": 0.875, "rewards/format_reward/std": 0.3333333432674408, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "step": 57 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.046875, "completions/mean_terminated_length": 12.046875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.145, "frac_reward_zero_std": 0.625, "grad_norm": 32.35861587524414, "learning_rate": 4.258681761440789e-07, "loss": 0.0, "num_tokens": 6241516.0, "reward": 0.6937500238418579, "reward_std": 0.19149437546730042, "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": 58 }, { "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": 12.09375, "completions/mean_terminated_length": 12.09375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1475, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 4.095597328339452e-07, "loss": 0.0, "num_tokens": 6330114.0, "reward": 0.6000000238418579, "reward_std": 0.0, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "step": 59 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.046875, "completions/mean_terminated_length": 12.046875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.15, "frac_reward_zero_std": 0.75, "grad_norm": 15.280664443969727, "learning_rate": 3.9335018462812664e-07, "loss": -0.0, "num_tokens": 6410589.0, "reward": 0.6781250238418579, "reward_std": 0.11100947856903076, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.578125, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 60 }, { "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": 12.125, "completions/mean_terminated_length": 12.125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1525, "frac_reward_zero_std": 0.875, "grad_norm": 8.745597839355469, "learning_rate": 3.772572564296004e-07, "loss": 0.0, "num_tokens": 6513493.0, "reward": 0.5531250238418579, "reward_std": 0.0646936446428299, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 61 }, { "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": 1288.0, "completions/mean_length": 133.421875, "completions/mean_terminated_length": 103.03175354003906, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.155, "frac_reward_zero_std": 0.5, "grad_norm": 25.65398597717285, "learning_rate": 3.612985456190778e-07, "loss": 0.0, "num_tokens": 6627480.0, "reward": 0.717968761920929, "reward_std": 0.1689612865447998, "rewards/format_reward/mean": 0.9296875, "rewards/format_reward/std": 0.19654127955436707, "rewards/mcq_exact_match_reward/mean": 0.625, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 62 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.03125, "completions/mean_terminated_length": 12.03125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1575, "frac_reward_zero_std": 0.875, "grad_norm": 6.467506408691406, "learning_rate": 3.454915028125263e-07, "loss": -0.0, "num_tokens": 6712658.0, "reward": 0.7406250238418579, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.640625, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "step": 63 }, { "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": 100.0, "completions/max_terminated_length": 100.0, "completions/mean_length": 13.40625, "completions/mean_terminated_length": 13.40625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.16, "frac_reward_zero_std": 0.5, "grad_norm": 19.417722702026367, "learning_rate": 3.2985341277917846e-07, "loss": 0.0, "num_tokens": 6835196.0, "reward": 0.6609375476837158, "reward_std": 0.21920469403266907, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.5625, "rewards/mcq_exact_match_reward/std": 0.5, "step": 64 }, { "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": 11.921875, "completions/mean_terminated_length": 11.921875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1625, "frac_reward_zero_std": 0.75, "grad_norm": 54.821903228759766, "learning_rate": 3.1440137554088953e-07, "loss": -0.0, "num_tokens": 6921519.0, "reward": 0.6773437857627869, "reward_std": 0.06690335273742676, "rewards/format_reward/mean": 0.9921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.578125, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 65 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.03125, "completions/mean_terminated_length": 12.03125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.165, "frac_reward_zero_std": 0.625, "grad_norm": 9.333979606628418, "learning_rate": 2.9915228767351535e-07, "loss": 0.0, "num_tokens": 7047585.0, "reward": 0.7875000238418579, "reward_std": 0.16675157845020294, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.6875, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "step": 66 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.03125, "completions/mean_terminated_length": 12.03125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1675, "frac_reward_zero_std": 0.75, "grad_norm": 8.179891586303711, "learning_rate": 2.841228238307536e-07, "loss": -0.0, "num_tokens": 7144067.0, "reward": 0.6312500238418579, "reward_std": 0.0883883461356163, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 67 }, { "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": 128.0, "completions/max_terminated_length": 128.0, "completions/mean_length": 13.84375, "completions/mean_terminated_length": 13.84375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.17, "frac_reward_zero_std": 0.875, "grad_norm": 10.27669906616211, "learning_rate": 2.6932941851065615e-07, "loss": 0.0, "num_tokens": 7232393.0, "reward": 0.5062500238418579, "reward_std": 0.0578637570142746, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 68 }, { "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": 12.125, "completions/mean_terminated_length": 12.125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1725, "frac_reward_zero_std": 0.625, "grad_norm": 9.707029342651367, "learning_rate": 2.547882480847461e-07, "loss": 0.0, "num_tokens": 7340721.0, "reward": 0.47343751788139343, "reward_std": 0.09280776232481003, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 69 }, { "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": 12.078125, "completions/mean_terminated_length": 12.078125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.175, "frac_reward_zero_std": 0.625, "grad_norm": 20.616575241088867, "learning_rate": 2.4051521310939254e-07, "loss": -0.0, "num_tokens": 7432398.0, "reward": 0.6468750238418579, "reward_std": 0.1530819982290268, "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": 70 }, { "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": 12.515625, "completions/mean_terminated_length": 12.515625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1775, "frac_reward_zero_std": 0.5, "grad_norm": 18.274648666381836, "learning_rate": 2.2652592093878665e-07, "loss": 0.0, "num_tokens": 7531343.0, "reward": 0.49531251192092896, "reward_std": 0.22903475165367126, "rewards/format_reward/mean": 0.890625, "rewards/format_reward/std": 0.3145764470100403, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 71 }, { "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": 12.078125, "completions/mean_terminated_length": 12.078125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.18, "frac_reward_zero_std": 0.875, "grad_norm": 8.250680923461914, "learning_rate": 2.128356686585282e-07, "loss": 0.0, "num_tokens": 7611988.0, "reward": 0.5843750238418579, "reward_std": 0.04419417306780815, "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": 72 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.0625, "completions/mean_terminated_length": 12.0625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1825, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 1.9945942635848745e-07, "loss": 0.0, "num_tokens": 7687112.0, "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": 73 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.03125, "completions/mean_terminated_length": 12.03125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.185, "frac_reward_zero_std": 0.75, "grad_norm": 19.1580810546875, "learning_rate": 1.8641182076323148e-07, "loss": 0.0, "num_tokens": 7761962.0, "reward": 0.4437500238418579, "reward_std": 0.10888782143592834, "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": 74 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.0625, "completions/mean_terminated_length": 12.0625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1875, "frac_reward_zero_std": 0.875, "grad_norm": 7.743622303009033, "learning_rate": 1.7370711923791564e-07, "loss": 0.0, "num_tokens": 7849702.0, "reward": 0.5531250238418579, "reward_std": 0.0646936446428299, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 75 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.046875, "completions/mean_terminated_length": 12.046875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.19, "frac_reward_zero_std": 0.875, "grad_norm": 6.07141637802124, "learning_rate": 1.6135921418712955e-07, "loss": 0.0, "num_tokens": 7920697.0, "reward": 0.3343750238418579, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "step": 76 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.046875, "completions/mean_terminated_length": 12.046875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1925, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 1.493816078637557e-07, "loss": 0.0, "num_tokens": 8022396.0, "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": 77 }, { "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": 0.195, "frac_reward_zero_std": 0.75, "grad_norm": 15.382882118225098, "learning_rate": 1.3778739760445552e-07, "loss": 0.0, "num_tokens": 8098229.0, "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": 78 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.03125, "completions/mean_terminated_length": 12.03125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1975, "frac_reward_zero_std": 0.5, "grad_norm": 16.139266967773438, "learning_rate": 1.2658926150792322e-07, "loss": 0.0, "num_tokens": 8212863.0, "reward": 0.7250000238418579, "reward_std": 0.1767766922712326, "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": 79 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.046875, "completions/mean_terminated_length": 12.046875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2, "frac_reward_zero_std": 0.875, "grad_norm": 15.221053123474121, "learning_rate": 1.1579944457157059e-07, "loss": 0.0, "num_tokens": 8304634.0, "reward": 0.7562500238418579, "reward_std": 0.0578637570142746, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.65625, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "step": 80 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.046875, "completions/mean_terminated_length": 12.046875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2025, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 1.0542974530180327e-07, "loss": 0.0, "num_tokens": 8402525.0, "reward": 0.8500000238418579, "reward_std": 0.0, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.75, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 81 }, { "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": 0.205, "frac_reward_zero_std": 0.25, "grad_norm": 37.38897705078125, "learning_rate": 9.549150281252632e-08, "loss": 0.0, "num_tokens": 8476261.0, "reward": 0.3500000238418579, "reward_std": 0.3014557659626007, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 82 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 21.0, "completions/max_terminated_length": 21.0, "completions/mean_length": 12.34375, "completions/mean_terminated_length": 12.34375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2075, "frac_reward_zero_std": 0.875, "grad_norm": 4.418399810791016, "learning_rate": 8.599558442598998e-08, "loss": 0.0, "num_tokens": 8572003.0, "reward": 0.48750001192092896, "reward_std": 0.045806266367435455, "rewards/format_reward/mean": 0.96875, "rewards/format_reward/std": 0.17536810040473938, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "step": 83 }, { "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": 12.140625, "completions/mean_terminated_length": 12.140625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.21, "frac_reward_zero_std": 0.625, "grad_norm": 22.378751754760742, "learning_rate": 7.695237378953224e-08, "loss": 0.0, "num_tokens": 8677996.0, "reward": 0.6312500238418579, "reward_std": 0.17570313811302185, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 84 }, { "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": 12.078125, "completions/mean_terminated_length": 12.078125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2125, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 6.837175952121304e-08, "loss": 0.0, "num_tokens": 8747201.0, "reward": 0.6000000238418579, "reward_std": 0.0, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "step": 85 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.0625, "completions/mean_terminated_length": 12.0625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.215, "frac_reward_zero_std": 0.625, "grad_norm": 16.450885772705078, "learning_rate": 6.026312439675551e-08, "loss": -0.0, "num_tokens": 8836277.0, "reward": 0.6468750238418579, "reward_std": 0.1530819982290268, "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": 86 }, { "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": 0.2175, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 5.263533508961826e-08, "loss": 0.0, "num_tokens": 8924174.0, "reward": 0.6000000238418579, "reward_std": 0.0, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "step": 87 }, { "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": 12.09375, "completions/mean_terminated_length": 12.09375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.22, "frac_reward_zero_std": 0.625, "grad_norm": 24.914026260375977, "learning_rate": 4.549673247541874e-08, "loss": 0.0, "num_tokens": 9019764.0, "reward": 0.5531250238418579, "reward_std": 0.15992169082164764, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 88 }, { "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": 72.0, "completions/max_terminated_length": 72.0, "completions/mean_length": 19.65625, "completions/mean_terminated_length": 19.65625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2225, "frac_reward_zero_std": 0.75, "grad_norm": 12.63296127319336, "learning_rate": 3.8855122511307626e-08, "loss": 0.0, "num_tokens": 9131878.0, "reward": 0.4625000059604645, "reward_std": 0.0883883461356163, "rewards/format_reward/mean": 0.875, "rewards/format_reward/std": 0.3333333432674408, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 89 }, { "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": 68.0, "completions/max_terminated_length": 68.0, "completions/mean_length": 13.734375, "completions/mean_terminated_length": 13.734375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.225, "frac_reward_zero_std": 0.875, "grad_norm": 4.911560535430908, "learning_rate": 3.271776770026963e-08, "loss": 0.0, "num_tokens": 9211605.0, "reward": 0.6312500238418579, "reward_std": 0.0578637570142746, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 90 }, { "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": 0.2275, "frac_reward_zero_std": 0.75, "grad_norm": 15.331509590148926, "learning_rate": 2.7091379149682682e-08, "loss": 0.0, "num_tokens": 9303022.0, "reward": 0.7093750238418579, "reward_std": 0.12255740165710449, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.609375, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "step": 91 }, { "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": 0.23, "frac_reward_zero_std": 0.75, "grad_norm": 16.129894256591797, "learning_rate": 2.1982109232821176e-08, "loss": 0.0, "num_tokens": 9389302.0, "reward": 0.9593750238418579, "reward_std": 0.1315089464187622, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.859375, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "step": 92 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.046875, "completions/mean_terminated_length": 12.046875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2325, "frac_reward_zero_std": 0.875, "grad_norm": 5.276142120361328, "learning_rate": 1.7395544861325718e-08, "loss": 0.0, "num_tokens": 9459145.0, "reward": 0.6937500238418579, "reward_std": 0.0578637570142746, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.59375, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 93 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.03125, "completions/mean_terminated_length": 12.03125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.235, "frac_reward_zero_std": 0.75, "grad_norm": 19.702011108398438, "learning_rate": 1.3336701375997127e-08, "loss": 0.0, "num_tokens": 9565043.0, "reward": 0.5531250238418579, "reward_std": 0.11100947856903076, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 94 }, { "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": 15.0, "completions/max_terminated_length": 15.0, "completions/mean_length": 12.046875, "completions/mean_terminated_length": 12.046875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2375, "frac_reward_zero_std": 0.75, "grad_norm": 11.032377243041992, "learning_rate": 9.810017062595321e-09, "loss": 0.0, "num_tokens": 9666214.0, "reward": 0.6937500238418579, "reward_std": 0.10888782143592834, "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": 95 }, { "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": 178.0, "completions/max_terminated_length": 178.0, "completions/mean_length": 20.3125, "completions/mean_terminated_length": 20.3125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.24, "frac_reward_zero_std": 0.875, "grad_norm": 13.167874336242676, "learning_rate": 6.819348298638839e-09, "loss": 0.0, "num_tokens": 9774682.0, "reward": 0.7718750238418579, "reward_std": 0.0646936446428299, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.671875, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 96 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.046875, "completions/mean_terminated_length": 12.046875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2425, "frac_reward_zero_std": 0.625, "grad_norm": 18.6101016998291, "learning_rate": 4.367965336512403e-09, "loss": 0.0, "num_tokens": 9877533.0, "reward": 0.4750000238418579, "reward_std": 0.1462521106004715, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 97 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.046875, "completions/mean_terminated_length": 12.046875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.245, "frac_reward_zero_std": 0.75, "grad_norm": 19.820520401000977, "learning_rate": 2.458548727494292e-09, "loss": 0.0, "num_tokens": 9972744.0, "reward": 0.3656250238418579, "reward_std": 0.12255740165710449, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "step": 98 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.046875, "completions/mean_terminated_length": 12.046875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2475, "frac_reward_zero_std": 0.875, "grad_norm": 4.595024108886719, "learning_rate": 1.0931863906127325e-09, "loss": 0.0, "num_tokens": 10059371.0, "reward": 0.7093750238418579, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.609375, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "step": 99 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.046875, "completions/mean_terminated_length": 12.046875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.25, "frac_reward_zero_std": 0.5, "grad_norm": 18.83749771118164, "learning_rate": 2.733713295369755e-10, "loss": 0.0, "num_tokens": 10163190.0, "reward": 0.6468750238418579, "reward_std": 0.2109457552433014, "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": 100 } ], "logging_steps": 1, "max_steps": 100, "num_input_tokens_seen": 10163190, "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": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }