{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.125, "eval_steps": 500, "global_step": 50, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1225.0, "completions/mean_length": 431.109375, "completions/mean_terminated_length": 405.4444580078125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0025, "frac_reward_zero_std": 0.125, "grad_norm": 5.259467124938965, "learning_rate": 0.0, "loss": -0.0, "num_tokens": 113231.0, "reward": 0.4351562261581421, "reward_std": 0.399507611989975, "rewards/format_reward/mean": 0.4453125, "rewards/format_reward/std": 0.180765300989151, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1913.0, "completions/mean_length": 676.65625, "completions/mean_terminated_length": 609.2130737304688, "completions/min_length": 9.0, "completions/min_terminated_length": 9.0, "epoch": 0.005, "frac_reward_zero_std": 0.0, "grad_norm": 10.034707069396973, "learning_rate": 2e-07, "loss": -0.0, "num_tokens": 264049.0, "reward": 0.4648436903953552, "reward_std": 0.30636128783226013, "rewards/format_reward/mean": 0.4296875, "rewards/format_reward/std": 0.23345555365085602, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 1345.0, "completions/max_terminated_length": 1345.0, "completions/mean_length": 452.875, "completions/mean_terminated_length": 452.875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0075, "frac_reward_zero_std": 0.0, "grad_norm": 3.9717142581939697, "learning_rate": 4e-07, "loss": -0.0, "num_tokens": 366009.0, "reward": 0.41874998807907104, "reward_std": 0.38659223914146423, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.3021090030670166, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 1307.0, "completions/max_terminated_length": 1307.0, "completions/mean_length": 520.125, "completions/mean_terminated_length": 520.125, "completions/min_length": 143.0, "completions/min_terminated_length": 143.0, "epoch": 0.01, "frac_reward_zero_std": 0.125, "grad_norm": 0.8136993646621704, "learning_rate": 6e-07, "loss": 0.0, "num_tokens": 486657.0, "reward": 0.2914062440395355, "reward_std": 0.3199944496154785, "rewards/format_reward/mean": 0.4140625, "rewards/format_reward/std": 0.2280818521976471, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1689.0, "completions/mean_length": 465.375, "completions/mean_terminated_length": 414.32257080078125, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.0125, "frac_reward_zero_std": 0.0, "grad_norm": 48.27979278564453, "learning_rate": 8e-07, "loss": -0.0, "num_tokens": 596601.0, "reward": 0.27031248807907104, "reward_std": 0.3643844723701477, "rewards/format_reward/mean": 0.359375, "rewards/format_reward/std": 0.301698237657547, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "step": 5 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 866.0, "completions/max_terminated_length": 866.0, "completions/mean_length": 270.53125, "completions/mean_terminated_length": 270.53125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.015, "frac_reward_zero_std": 0.0, "grad_norm": 16.251083374023438, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 704523.0, "reward": 0.28749996423721313, "reward_std": 0.31347018480300903, "rewards/format_reward/mean": 0.375, "rewards/format_reward/std": 0.2357022762298584, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 6 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1120.0, "completions/mean_length": 367.234375, "completions/mean_terminated_length": 340.5555725097656, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.0175, "frac_reward_zero_std": 0.0, "grad_norm": 11.4781494140625, "learning_rate": 9.99726628670463e-07, "loss": -0.0, "num_tokens": 806458.0, "reward": 0.3179687261581421, "reward_std": 0.4396277666091919, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.2987033724784851, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "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": 2000.0, "completions/max_terminated_length": 2000.0, "completions/mean_length": 483.640625, "completions/mean_terminated_length": 483.640625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.02, "frac_reward_zero_std": 0.0, "grad_norm": 4.160367965698242, "learning_rate": 9.989068136093872e-07, "loss": -0.0, "num_tokens": 920035.0, "reward": 0.32265621423721313, "reward_std": 0.39968013763427734, "rewards/format_reward/mean": 0.4140625, "rewards/format_reward/std": 0.19012710452079773, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "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": 2035.0, "completions/mean_length": 687.9375, "completions/mean_terminated_length": 666.3492431640625, "completions/min_length": 129.0, "completions/min_terminated_length": 129.0, "epoch": 0.0225, "frac_reward_zero_std": 0.25, "grad_norm": 1.0734870433807373, "learning_rate": 9.975414512725056e-07, "loss": 0.0, "num_tokens": 1043311.0, "reward": 0.1328125, "reward_std": 0.18026423454284668, "rewards/format_reward/mean": 0.390625, "rewards/format_reward/std": 0.22658175230026245, "rewards/mcq_exact_match_reward/mean": 0.09375, "rewards/mcq_exact_match_reward/std": 0.29378482699394226, "step": 9 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1260.0, "completions/mean_length": 388.09375, "completions/mean_terminated_length": 361.7460632324219, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.025, "frac_reward_zero_std": 0.125, "grad_norm": 16.148521423339844, "learning_rate": 9.956320346634875e-07, "loss": 0.0, "num_tokens": 1165533.0, "reward": 0.46953123807907104, "reward_std": 0.38408195972442627, "rewards/format_reward/mean": 0.4765625, "rewards/format_reward/std": 0.28770697116851807, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1282.0, "completions/mean_length": 580.203125, "completions/mean_terminated_length": 508.016357421875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0275, "frac_reward_zero_std": 0.0, "grad_norm": 3.1740071773529053, "learning_rate": 9.931806517013612e-07, "loss": -0.0, "num_tokens": 1278858.0, "reward": 0.3960937261581421, "reward_std": 0.3790547847747803, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.2708333432674408, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 1479.0, "completions/max_terminated_length": 1479.0, "completions/mean_length": 523.4375, "completions/mean_terminated_length": 523.4375, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.03, "frac_reward_zero_std": 0.0, "grad_norm": 5.912919521331787, "learning_rate": 9.901899829374047e-07, "loss": -0.0, "num_tokens": 1411390.0, "reward": 0.28359371423721313, "reward_std": 0.39816731214523315, "rewards/format_reward/mean": 0.3359375, "rewards/format_reward/std": 0.2366211861371994, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 1946.0, "completions/mean_length": 643.5, "completions/mean_terminated_length": 621.2063598632812, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0325, "frac_reward_zero_std": 0.25, "grad_norm": 4.178334712982178, "learning_rate": 9.866632986240029e-07, "loss": 0.0, "num_tokens": 1551454.0, "reward": 0.2757812440395355, "reward_std": 0.34187954664230347, "rewards/format_reward/mean": 0.4140625, "rewards/format_reward/std": 0.209963858127594, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 1380.0, "completions/mean_length": 597.53125, "completions/mean_terminated_length": 550.741943359375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.035, "frac_reward_zero_std": 0.0, "grad_norm": 4.846618175506592, "learning_rate": 9.826044551386742e-07, "loss": 0.0, "num_tokens": 1661736.0, "reward": 0.4468749761581421, "reward_std": 0.4076019525527954, "rewards/format_reward/mean": 0.40625, "rewards/format_reward/std": 0.3069951832294464, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 1224.0, "completions/max_terminated_length": 1224.0, "completions/mean_length": 402.3125, "completions/mean_terminated_length": 402.3125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0375, "frac_reward_zero_std": 0.0, "grad_norm": 21.241313934326172, "learning_rate": 9.780178907671788e-07, "loss": -0.0, "num_tokens": 1781228.0, "reward": 0.5023436546325684, "reward_std": 0.4687495827674866, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.22699186205863953, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 15 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1578.0, "completions/max_terminated_length": 1578.0, "completions/mean_length": 511.65625, "completions/mean_terminated_length": 511.65625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.04, "frac_reward_zero_std": 0.0, "grad_norm": 11.121533393859863, "learning_rate": 9.729086208503173e-07, "loss": -0.0, "num_tokens": 1890062.0, "reward": 0.31953123211860657, "reward_std": 0.40660345554351807, "rewards/format_reward/mean": 0.5390625, "rewards/format_reward/std": 0.3125, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "step": 16 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1753.0, "completions/max_terminated_length": 1753.0, "completions/mean_length": 501.59375, "completions/mean_terminated_length": 501.59375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0425, "frac_reward_zero_std": 0.125, "grad_norm": 8.208738327026367, "learning_rate": 9.672822322997304e-07, "loss": -0.0, "num_tokens": 2016348.0, "reward": 0.40468746423721313, "reward_std": 0.3911667466163635, "rewards/format_reward/mean": 0.453125, "rewards/format_reward/std": 0.19352105259895325, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "step": 17 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1847.0, "completions/max_terminated_length": 1847.0, "completions/mean_length": 533.9375, "completions/mean_terminated_length": 533.9375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.045, "frac_reward_zero_std": 0.0, "grad_norm": 8.234257698059082, "learning_rate": 9.611448774886923e-07, "loss": 0.0, "num_tokens": 2161912.0, "reward": 0.17265625298023224, "reward_std": 0.24125564098358154, "rewards/format_reward/mean": 0.4765625, "rewards/format_reward/std": 0.28770697116851807, "rewards/mcq_exact_match_reward/mean": 0.125, "rewards/mcq_exact_match_reward/std": 0.3333333432674408, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1279.0, "completions/mean_length": 354.703125, "completions/mean_terminated_length": 271.42620849609375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0475, "frac_reward_zero_std": 0.25, "grad_norm": 12.127586364746094, "learning_rate": 9.545032675245813e-07, "loss": -0.0, "num_tokens": 2279485.0, "reward": 0.37421873211860657, "reward_std": 0.34257689118385315, "rewards/format_reward/mean": 0.4609375, "rewards/format_reward/std": 0.2236899733543396, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 1580.0, "completions/max_terminated_length": 1580.0, "completions/mean_length": 431.359375, "completions/mean_terminated_length": 431.359375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05, "frac_reward_zero_std": 0.0, "grad_norm": 11.236443519592285, "learning_rate": 9.473646649103817e-07, "loss": -0.0, "num_tokens": 2425436.0, "reward": 0.3140624761581421, "reward_std": 0.4075997769832611, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.30820462107658386, "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": 0.0625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1699.0, "completions/mean_length": 542.546875, "completions/mean_terminated_length": 442.183349609375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0525, "frac_reward_zero_std": 0.25, "grad_norm": 11.123575210571289, "learning_rate": 9.397368756032444e-07, "loss": 0.0, "num_tokens": 2568383.0, "reward": 0.13984374701976776, "reward_std": 0.2072017937898636, "rewards/format_reward/mean": 0.4609375, "rewards/format_reward/std": 0.2236899733543396, "rewards/mcq_exact_match_reward/mean": 0.09375, "rewards/mcq_exact_match_reward/std": 0.29378482699394226, "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": 1141.0, "completions/mean_length": 346.5625, "completions/mean_terminated_length": 319.5555725097656, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.055, "frac_reward_zero_std": 0.125, "grad_norm": 20.075658798217773, "learning_rate": 9.316282404787869e-07, "loss": -0.0, "num_tokens": 2661115.0, "reward": 0.4859375059604645, "reward_std": 0.398270845413208, "rewards/format_reward/mean": 0.640625, "rewards/format_reward/std": 0.28824523091316223, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 22 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1276.0, "completions/mean_length": 287.71875, "completions/mean_terminated_length": 259.7778015136719, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0575, "frac_reward_zero_std": 0.375, "grad_norm": 10.893749237060547, "learning_rate": 9.230476262104676e-07, "loss": -0.0, "num_tokens": 2753361.0, "reward": 0.2796874940395355, "reward_std": 0.21289312839508057, "rewards/format_reward/mean": 0.609375, "rewards/format_reward/std": 0.22658175230026245, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "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": 1407.0, "completions/mean_length": 290.921875, "completions/mean_terminated_length": 263.0317687988281, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06, "frac_reward_zero_std": 0.125, "grad_norm": 13.405001640319824, "learning_rate": 9.1400441557401e-07, "loss": 0.0, "num_tokens": 2877572.0, "reward": 0.4351562559604645, "reward_std": 0.2841757535934448, "rewards/format_reward/mean": 0.6015625, "rewards/format_reward/std": 0.26991602778434753, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 24 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1321.0, "completions/max_terminated_length": 1321.0, "completions/mean_length": 274.375, "completions/mean_terminated_length": 274.375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0625, "frac_reward_zero_std": 0.25, "grad_norm": 17.836549758911133, "learning_rate": 9.045084971874737e-07, "loss": -0.0, "num_tokens": 3004036.0, "reward": 0.32265621423721313, "reward_std": 0.23537221550941467, "rewards/format_reward/mean": 0.5703125, "rewards/format_reward/std": 0.2498759627342224, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 1289.0, "completions/max_terminated_length": 1289.0, "completions/mean_length": 141.171875, "completions/mean_terminated_length": 141.171875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.065, "frac_reward_zero_std": 0.375, "grad_norm": 15.09114933013916, "learning_rate": 8.945702546981968e-07, "loss": 0.0, "num_tokens": 3094383.0, "reward": 0.4078125059604645, "reward_std": 0.2618940472602844, "rewards/format_reward/mean": 0.796875, "rewards/format_reward/std": 0.24750742316246033, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 797.0, "completions/mean_length": 100.546875, "completions/mean_terminated_length": 69.63492584228516, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0675, "frac_reward_zero_std": 0.125, "grad_norm": 31.52334976196289, "learning_rate": 8.842005554284295e-07, "loss": -0.0, "num_tokens": 3187418.0, "reward": 0.6171875, "reward_std": 0.3499646484851837, "rewards/format_reward/mean": 0.703125, "rewards/format_reward/std": 0.30496877431869507, "rewards/mcq_exact_match_reward/mean": 0.546875, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1209.0, "completions/mean_length": 216.953125, "completions/mean_terminated_length": 157.8870849609375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07, "frac_reward_zero_std": 0.25, "grad_norm": 38.22136688232422, "learning_rate": 8.734107384920769e-07, "loss": -0.0, "num_tokens": 3301759.0, "reward": 0.5718749761581421, "reward_std": 0.17830196022987366, "rewards/format_reward/mean": 0.71875, "rewards/format_reward/std": 0.3069951832294464, "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": 1263.0, "completions/max_terminated_length": 1263.0, "completions/mean_length": 101.828125, "completions/mean_terminated_length": 101.828125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0725, "frac_reward_zero_std": 0.25, "grad_norm": 10.861287117004395, "learning_rate": 8.622126023955445e-07, "loss": 0.0, "num_tokens": 3381796.0, "reward": 0.5921875238418579, "reward_std": 0.3222452402114868, "rewards/format_reward/mean": 0.921875, "rewards/format_reward/std": 0.2221602201461792, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "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": 488.0, "completions/max_terminated_length": 488.0, "completions/mean_length": 28.640625, "completions/mean_terminated_length": 28.640625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.075, "frac_reward_zero_std": 0.375, "grad_norm": 10.657682418823242, "learning_rate": 8.506183921362442e-07, "loss": -0.0, "num_tokens": 3477045.0, "reward": 0.34687501192092896, "reward_std": 0.15013158321380615, "rewards/format_reward/mean": 0.96875, "rewards/format_reward/std": 0.12198751419782639, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 2007.0, "completions/max_terminated_length": 2007.0, "completions/mean_length": 72.5625, "completions/mean_terminated_length": 72.5625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0775, "frac_reward_zero_std": 0.625, "grad_norm": 24.812374114990234, "learning_rate": 8.386407858128706e-07, "loss": -0.0, "num_tokens": 3572769.0, "reward": 0.515625, "reward_std": 0.10445605218410492, "rewards/format_reward/mean": 0.9375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 48.0, "completions/max_terminated_length": 48.0, "completions/mean_length": 12.734375, "completions/mean_terminated_length": 12.734375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08, "frac_reward_zero_std": 0.875, "grad_norm": 22.429168701171875, "learning_rate": 8.262928807620843e-07, "loss": 0.0, "num_tokens": 3655176.0, "reward": 0.7234375476837158, "reward_std": 0.0028931896667927504, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.08768405020236969, "rewards/mcq_exact_match_reward/mean": 0.625, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 133.0, "completions/max_terminated_length": 133.0, "completions/mean_length": 17.609375, "completions/mean_terminated_length": 17.609375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0825, "frac_reward_zero_std": 0.625, "grad_norm": 9.004563331604004, "learning_rate": 8.135881792367685e-07, "loss": -0.0, "num_tokens": 3759983.0, "reward": 0.6304687261581421, "reward_std": 0.15494245290756226, "rewards/format_reward/mean": 0.9921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "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": 422.0, "completions/max_terminated_length": 422.0, "completions/mean_length": 19.421875, "completions/mean_terminated_length": 19.421875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.085, "frac_reward_zero_std": 0.375, "grad_norm": 14.119917869567871, "learning_rate": 8.005405736415125e-07, "loss": 0.0, "num_tokens": 3852194.0, "reward": 0.6468750238418579, "reward_std": 0.28930896520614624, "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": 34 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.109375, "completions/mean_terminated_length": 12.109375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0875, "frac_reward_zero_std": 0.625, "grad_norm": 8.968405723571777, "learning_rate": 7.871643313414718e-07, "loss": 0.0, "num_tokens": 3935377.0, "reward": 0.5687500238418579, "reward_std": 0.16675157845020294, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 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": 254.0, "completions/max_terminated_length": 254.0, "completions/mean_length": 19.09375, "completions/mean_terminated_length": 19.09375, "completions/min_length": 8.0, "completions/min_terminated_length": 8.0, "epoch": 0.09, "frac_reward_zero_std": 0.5, "grad_norm": 15.726703643798828, "learning_rate": 7.734740790612136e-07, "loss": 0.0, "num_tokens": 4048999.0, "reward": 0.37968751788139343, "reward_std": 0.23583455383777618, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "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": 798.0, "completions/max_terminated_length": 798.0, "completions/mean_length": 23.765625, "completions/mean_terminated_length": 23.765625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0925, "frac_reward_zero_std": 0.625, "grad_norm": 8.501358985900879, "learning_rate": 7.594847868906076e-07, "loss": -0.0, "num_tokens": 4172896.0, "reward": 0.4203125238418579, "reward_std": 0.15526627004146576, "rewards/format_reward/mean": 0.921875, "rewards/format_reward/std": 0.2221602201461792, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 22.0, "completions/max_terminated_length": 22.0, "completions/mean_length": 12.328125, "completions/mean_terminated_length": 12.328125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.095, "frac_reward_zero_std": 0.75, "grad_norm": 8.62055778503418, "learning_rate": 7.452117519152541e-07, "loss": -0.0, "num_tokens": 4270837.0, "reward": 0.5515625476837158, "reward_std": 0.06629126518964767, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.08768405020236969, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 235.0, "completions/max_terminated_length": 235.0, "completions/mean_length": 15.671875, "completions/mean_terminated_length": 15.671875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0975, "frac_reward_zero_std": 0.75, "grad_norm": 10.415192604064941, "learning_rate": 7.306705814893439e-07, "loss": 0.0, "num_tokens": 4359496.0, "reward": 0.8187500238418579, "reward_std": 0.0883883461356163, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.71875, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 39 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.046875, "completions/mean_terminated_length": 12.046875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1, "frac_reward_zero_std": 0.75, "grad_norm": 5.868995189666748, "learning_rate": 7.158771761692464e-07, "loss": 0.0, "num_tokens": 4456883.0, "reward": 0.5062500238418579, "reward_std": 0.1293872892856598, "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": 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": 85.0, "completions/max_terminated_length": 85.0, "completions/mean_length": 14.4375, "completions/mean_terminated_length": 14.4375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1025, "frac_reward_zero_std": 0.5, "grad_norm": 13.44471263885498, "learning_rate": 7.008477123264847e-07, "loss": 0.0, "num_tokens": 4552767.0, "reward": 0.547656238079071, "reward_std": 0.16052383184432983, "rewards/format_reward/mean": 0.9453125, "rewards/format_reward/std": 0.2203386276960373, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 41 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.046875, "completions/mean_terminated_length": 12.046875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.105, "frac_reward_zero_std": 0.5, "grad_norm": 14.134708404541016, "learning_rate": 6.855986244591103e-07, "loss": 0.0, "num_tokens": 4625970.0, "reward": 0.8500000238418579, "reward_std": 0.249358132481575, "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": 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": 763.0, "completions/max_terminated_length": 763.0, "completions/mean_length": 23.75, "completions/mean_terminated_length": 23.75, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1075, "frac_reward_zero_std": 0.5, "grad_norm": 14.159423828125, "learning_rate": 6.701465872208216e-07, "loss": 0.0, "num_tokens": 4735562.0, "reward": 0.6937500238418579, "reward_std": 0.2177756428718567, "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": 43 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.015625, "completions/mean_terminated_length": 12.015625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.11, "frac_reward_zero_std": 0.75, "grad_norm": 6.4924702644348145, "learning_rate": 6.545084971874736e-07, "loss": -0.0, "num_tokens": 4817275.0, "reward": 0.2875000238418579, "reward_std": 0.10888782143592834, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "step": 44 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.015625, "completions/mean_terminated_length": 12.015625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1125, "frac_reward_zero_std": 0.75, "grad_norm": 13.048378944396973, "learning_rate": 6.387014543809223e-07, "loss": 0.0, "num_tokens": 4910468.0, "reward": 0.6156250238418579, "reward_std": 0.1315089464187622, "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": 45 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.115, "frac_reward_zero_std": 0.75, "grad_norm": 10.74089241027832, "learning_rate": 6.227427435703995e-07, "loss": 0.0, "num_tokens": 5013692.0, "reward": 0.5531250238418579, "reward_std": 0.10205793380737305, "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": 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": 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": 0.1175, "frac_reward_zero_std": 0.75, "grad_norm": 13.447330474853516, "learning_rate": 6.066498153718734e-07, "loss": 0.0, "num_tokens": 5116044.0, "reward": 0.6781250238418579, "reward_std": 0.10205793380737305, "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": 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": 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": 0.12, "frac_reward_zero_std": 0.625, "grad_norm": 16.091989517211914, "learning_rate": 5.90440267166055e-07, "loss": 0.0, "num_tokens": 5219452.0, "reward": 0.6000000238418579, "reward_std": 0.17570313811302185, "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": 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": 710.0, "completions/max_terminated_length": 710.0, "completions/mean_length": 46.421875, "completions/mean_terminated_length": 46.421875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1225, "frac_reward_zero_std": 0.75, "grad_norm": 15.343292236328125, "learning_rate": 5.741318238559209e-07, "loss": 0.0, "num_tokens": 5318119.0, "reward": 0.3500000238418579, "reward_std": 0.1293872892856598, "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": 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": 0.125, "frac_reward_zero_std": 0.75, "grad_norm": 11.404319763183594, "learning_rate": 5.577423184847931e-07, "loss": 0.0, "num_tokens": 5389239.0, "reward": 0.14687500894069672, "reward_std": 0.10205793380737305, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.046875, "rewards/mcq_exact_match_reward/std": 0.21304203569889069, "step": 50 } ], "logging_steps": 1, "max_steps": 100, "num_input_tokens_seen": 5389239, "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 }