{ "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": 0.0, "completions/max_length": 1102.0, "completions/max_terminated_length": 1102.0, "completions/mean_length": 354.328125, "completions/mean_terminated_length": 354.328125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0025, "frac_reward_zero_std": 0.125, "grad_norm": 11.511406898498535, "learning_rate": 0.0, "loss": 0.0, "num_tokens": 116829.0, "reward": 0.2796874940395355, "reward_std": 0.2978697121143341, "rewards/format_reward/mean": 0.296875, "rewards/format_reward/std": 0.2777281701564789, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 1 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1077.0, "completions/max_terminated_length": 1077.0, "completions/mean_length": 404.921875, "completions/mean_terminated_length": 404.921875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.005, "frac_reward_zero_std": 0.0, "grad_norm": 4.572214126586914, "learning_rate": 2e-07, "loss": 0.0, "num_tokens": 206256.0, "reward": 0.35546875, "reward_std": 0.4315900206565857, "rewards/format_reward/mean": 0.4296875, "rewards/format_reward/std": 0.3068941831588745, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 1111.0, "completions/max_terminated_length": 1111.0, "completions/mean_length": 375.625, "completions/mean_terminated_length": 375.625, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.0075, "frac_reward_zero_std": 0.0, "grad_norm": 6.5303473472595215, "learning_rate": 4e-07, "loss": 0.0, "num_tokens": 316488.0, "reward": 0.3179687261581421, "reward_std": 0.41223379969596863, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.23974503576755524, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 3 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1923.0, "completions/mean_length": 591.765625, "completions/mean_terminated_length": 568.6508178710938, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.01, "frac_reward_zero_std": 0.0, "grad_norm": 3.7336020469665527, "learning_rate": 6e-07, "loss": 0.0, "num_tokens": 455041.0, "reward": 0.21796873211860657, "reward_std": 0.2683701813220978, "rewards/format_reward/mean": 0.3046875, "rewards/format_reward/std": 0.24587368965148926, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "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": 1456.0, "completions/mean_length": 520.109375, "completions/mean_terminated_length": 470.82257080078125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0125, "frac_reward_zero_std": 0.0, "grad_norm": 2.97904634475708, "learning_rate": 8e-07, "loss": -0.0, "num_tokens": 592568.0, "reward": 0.11406249552965164, "reward_std": 0.2054193764925003, "rewards/format_reward/mean": 0.359375, "rewards/format_reward/std": 0.33886775374412537, "rewards/mcq_exact_match_reward/mean": 0.078125, "rewards/mcq_exact_match_reward/std": 0.27048972249031067, "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": 1474.0, "completions/max_terminated_length": 1474.0, "completions/mean_length": 500.03125, "completions/mean_terminated_length": 500.03125, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.015, "frac_reward_zero_std": 0.125, "grad_norm": 14.984542846679688, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 708602.0, "reward": 0.3499999940395355, "reward_std": 0.4164764881134033, "rewards/format_reward/mean": 0.375, "rewards/format_reward/std": 0.2357022762298584, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 1562.0, "completions/max_terminated_length": 1562.0, "completions/mean_length": 595.9375, "completions/mean_terminated_length": 595.9375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0175, "frac_reward_zero_std": 0.0, "grad_norm": 8.494792938232422, "learning_rate": 9.99726628670463e-07, "loss": -0.0, "num_tokens": 844854.0, "reward": 0.4609374403953552, "reward_std": 0.4118325710296631, "rewards/format_reward/mean": 0.390625, "rewards/format_reward/std": 0.24346621334552765, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1106.0, "completions/mean_length": 397.875, "completions/mean_terminated_length": 344.6451416015625, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.02, "frac_reward_zero_std": 0.0, "grad_norm": 22.35193634033203, "learning_rate": 9.989068136093872e-07, "loss": 0.0, "num_tokens": 956454.0, "reward": 0.2640624940395355, "reward_std": 0.3728104829788208, "rewards/format_reward/mean": 0.296875, "rewards/format_reward/std": 0.2916666865348816, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "step": 8 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1855.0, "completions/max_terminated_length": 1855.0, "completions/mean_length": 640.5, "completions/mean_terminated_length": 640.5, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0225, "frac_reward_zero_std": 0.0, "grad_norm": 10.466068267822266, "learning_rate": 9.975414512725056e-07, "loss": 0.0, "num_tokens": 1105982.0, "reward": 0.19062499701976776, "reward_std": 0.3181054890155792, "rewards/format_reward/mean": 0.34375, "rewards/format_reward/std": 0.233588308095932, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "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": 0.078125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1440.0, "completions/mean_length": 639.578125, "completions/mean_terminated_length": 520.2203369140625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.025, "frac_reward_zero_std": 0.0, "grad_norm": 9.383023262023926, "learning_rate": 9.956320346634875e-07, "loss": -0.0, "num_tokens": 1239139.0, "reward": 0.3218749761581421, "reward_std": 0.4199468791484833, "rewards/format_reward/mean": 0.40625, "rewards/format_reward/std": 0.265398770570755, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 10 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1201.0, "completions/max_terminated_length": 1201.0, "completions/mean_length": 506.03125, "completions/mean_terminated_length": 506.03125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0275, "frac_reward_zero_std": 0.125, "grad_norm": 7.184911727905273, "learning_rate": 9.931806517013612e-07, "loss": -0.0, "num_tokens": 1348437.0, "reward": 0.3765624761581421, "reward_std": 0.3668774366378784, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 11 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1857.0, "completions/mean_length": 400.734375, "completions/mean_terminated_length": 374.5873107910156, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.03, "frac_reward_zero_std": 0.375, "grad_norm": 7.753147602081299, "learning_rate": 9.901899829374047e-07, "loss": -0.0, "num_tokens": 1483092.0, "reward": 0.31171873211860657, "reward_std": 0.19852636754512787, "rewards/format_reward/mean": 0.4609375, "rewards/format_reward/std": 0.20518454909324646, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "step": 12 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1823.0, "completions/max_terminated_length": 1823.0, "completions/mean_length": 400.671875, "completions/mean_terminated_length": 400.671875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0325, "frac_reward_zero_std": 0.125, "grad_norm": 8.15406322479248, "learning_rate": 9.866632986240029e-07, "loss": 0.0, "num_tokens": 1590639.0, "reward": 0.2968749701976776, "reward_std": 0.3839948773384094, "rewards/format_reward/mean": 0.46875, "rewards/format_reward/std": 0.17536810040473938, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 13 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1152.0, "completions/mean_length": 302.40625, "completions/mean_terminated_length": 274.69842529296875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.035, "frac_reward_zero_std": 0.25, "grad_norm": 7.261423587799072, "learning_rate": 9.826044551386742e-07, "loss": 0.0, "num_tokens": 1709585.0, "reward": 0.27031248807907104, "reward_std": 0.31415098905563354, "rewards/format_reward/mean": 0.515625, "rewards/format_reward/std": 0.1985812783241272, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "step": 14 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 987.0, "completions/mean_length": 274.5625, "completions/mean_terminated_length": 246.4127197265625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0375, "frac_reward_zero_std": 0.0, "grad_norm": 17.917457580566406, "learning_rate": 9.780178907671788e-07, "loss": 0.0, "num_tokens": 1808749.0, "reward": 0.3335937559604645, "reward_std": 0.4015946388244629, "rewards/format_reward/mean": 0.5234375, "rewards/format_reward/std": 0.25865477323532104, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "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": 1725.0, "completions/max_terminated_length": 1725.0, "completions/mean_length": 125.34375, "completions/mean_terminated_length": 125.34375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.04, "frac_reward_zero_std": 0.5, "grad_norm": 10.193765640258789, "learning_rate": 9.729086208503173e-07, "loss": 0.0, "num_tokens": 1910163.0, "reward": 0.5023437142372131, "reward_std": 0.15529169142246246, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 996.0, "completions/max_terminated_length": 996.0, "completions/mean_length": 209.265625, "completions/mean_terminated_length": 209.265625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0425, "frac_reward_zero_std": 0.125, "grad_norm": 9.754120826721191, "learning_rate": 9.672822322997304e-07, "loss": -0.0, "num_tokens": 1997124.0, "reward": 0.4132812023162842, "reward_std": 0.3032131493091583, "rewards/format_reward/mean": 0.5390625, "rewards/format_reward/std": 0.2567298710346222, "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": 705.0, "completions/max_terminated_length": 705.0, "completions/mean_length": 80.09375, "completions/mean_terminated_length": 80.09375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.045, "frac_reward_zero_std": 0.5, "grad_norm": 19.73709487915039, "learning_rate": 9.611448774886923e-07, "loss": 0.0, "num_tokens": 2076906.0, "reward": 0.38515621423721313, "reward_std": 0.1931585669517517, "rewards/format_reward/mean": 0.5703125, "rewards/format_reward/std": 0.1751912236213684, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 18 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 851.0, "completions/max_terminated_length": 851.0, "completions/mean_length": 70.734375, "completions/mean_terminated_length": 70.734375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0475, "frac_reward_zero_std": 0.375, "grad_norm": 22.699562072753906, "learning_rate": 9.545032675245813e-07, "loss": 0.0, "num_tokens": 2170569.0, "reward": 0.4898437261581421, "reward_std": 0.25680527091026306, "rewards/format_reward/mean": 0.5234375, "rewards/format_reward/std": 0.10652101784944534, "rewards/mcq_exact_match_reward/mean": 0.4375, "rewards/mcq_exact_match_reward/std": 0.5, "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": 681.0, "completions/max_terminated_length": 681.0, "completions/mean_length": 50.921875, "completions/mean_terminated_length": 50.921875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05, "frac_reward_zero_std": 0.5, "grad_norm": 15.743022918701172, "learning_rate": 9.473646649103817e-07, "loss": 0.0, "num_tokens": 2264148.0, "reward": 0.3804687261581421, "reward_std": 0.2108108103275299, "rewards/format_reward/mean": 0.5234375, "rewards/format_reward/std": 0.10652101784944534, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 917.0, "completions/max_terminated_length": 917.0, "completions/mean_length": 70.546875, "completions/mean_terminated_length": 70.546875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0525, "frac_reward_zero_std": 0.375, "grad_norm": 18.87118148803711, "learning_rate": 9.397368756032444e-07, "loss": -0.0, "num_tokens": 2373239.0, "reward": 0.3632812201976776, "reward_std": 0.2548786997795105, "rewards/format_reward/mean": 0.5078125, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "step": 21 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 387.0, "completions/max_terminated_length": 387.0, "completions/mean_length": 17.921875, "completions/mean_terminated_length": 17.921875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.055, "frac_reward_zero_std": 0.375, "grad_norm": 25.482959747314453, "learning_rate": 9.316282404787869e-07, "loss": 0.0, "num_tokens": 2462138.0, "reward": 0.75390625, "reward_std": 0.2730957865715027, "rewards/format_reward/mean": 0.5078125, "rewards/format_reward/std": 0.18881812691688538, "rewards/mcq_exact_match_reward/mean": 0.703125, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "step": 22 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1641.0, "completions/max_terminated_length": 1641.0, "completions/mean_length": 101.4375, "completions/mean_terminated_length": 101.4375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0575, "frac_reward_zero_std": 0.0, "grad_norm": 31.341495513916016, "learning_rate": 9.230476262104676e-07, "loss": -0.0, "num_tokens": 2549926.0, "reward": 0.4296874701976776, "reward_std": 0.38230717182159424, "rewards/format_reward/mean": 0.546875, "rewards/format_reward/std": 0.14689241349697113, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 23 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 550.0, "completions/max_terminated_length": 550.0, "completions/mean_length": 24.375, "completions/mean_terminated_length": 24.375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06, "frac_reward_zero_std": 0.375, "grad_norm": 17.629453659057617, "learning_rate": 9.1400441557401e-07, "loss": -0.0, "num_tokens": 2683486.0, "reward": 0.4874999523162842, "reward_std": 0.24359197914600372, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.4375, "rewards/mcq_exact_match_reward/std": 0.5, "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": 453.0, "completions/max_terminated_length": 453.0, "completions/mean_length": 18.875, "completions/mean_terminated_length": 18.875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0625, "frac_reward_zero_std": 0.5, "grad_norm": 13.41634750366211, "learning_rate": 9.045084971874737e-07, "loss": -0.0, "num_tokens": 2776526.0, "reward": 0.20781248807907104, "reward_std": 0.1462520956993103, "rewards/format_reward/mean": 0.515625, "rewards/format_reward/std": 0.08768405020236969, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "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": 706.0, "completions/max_terminated_length": 706.0, "completions/mean_length": 36.578125, "completions/mean_terminated_length": 36.578125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.065, "frac_reward_zero_std": 0.5, "grad_norm": 33.40193176269531, "learning_rate": 8.945702546981968e-07, "loss": -0.0, "num_tokens": 2891163.0, "reward": 0.22499997913837433, "reward_std": 0.1078442931175232, "rewards/format_reward/mean": 0.53125, "rewards/format_reward/std": 0.12198751419782639, "rewards/mcq_exact_match_reward/mean": 0.171875, "rewards/mcq_exact_match_reward/std": 0.38025420904159546, "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": 608.0, "completions/max_terminated_length": 608.0, "completions/mean_length": 16.25, "completions/mean_terminated_length": 16.25, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0675, "frac_reward_zero_std": 0.375, "grad_norm": 43.69699478149414, "learning_rate": 8.842005554284295e-07, "loss": -0.0, "num_tokens": 2994803.0, "reward": 0.43203121423721313, "reward_std": 0.182278573513031, "rewards/format_reward/mean": 0.5703125, "rewards/format_reward/std": 0.1751912236213684, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 27 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 202.0, "completions/max_terminated_length": 202.0, "completions/mean_length": 10.03125, "completions/mean_terminated_length": 10.03125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07, "frac_reward_zero_std": 0.5, "grad_norm": 37.87979507446289, "learning_rate": 8.734107384920769e-07, "loss": 0.0, "num_tokens": 3080637.0, "reward": 0.4476562440395355, "reward_std": 0.13726258277893066, "rewards/format_reward/mean": 0.5703125, "rewards/format_reward/std": 0.19654127955436707, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 934.0, "completions/max_terminated_length": 934.0, "completions/mean_length": 23.96875, "completions/mean_terminated_length": 23.96875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0725, "frac_reward_zero_std": 0.625, "grad_norm": 23.51087188720703, "learning_rate": 8.622126023955445e-07, "loss": 0.0, "num_tokens": 3172627.0, "reward": 0.4624999761581421, "reward_std": 0.06228317692875862, "rewards/format_reward/mean": 0.5625, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 789.0, "completions/max_terminated_length": 789.0, "completions/mean_length": 20.390625, "completions/mean_terminated_length": 20.390625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.075, "frac_reward_zero_std": 0.375, "grad_norm": 15.329917907714844, "learning_rate": 8.506183921362442e-07, "loss": 0.0, "num_tokens": 3261748.0, "reward": 0.3812500238418579, "reward_std": 0.2709311544895172, "rewards/format_reward/mean": 0.6875, "rewards/format_reward/std": 0.24397502839565277, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 516.0, "completions/max_terminated_length": 516.0, "completions/mean_length": 35.953125, "completions/mean_terminated_length": 35.953125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0775, "frac_reward_zero_std": 0.375, "grad_norm": 21.827795028686523, "learning_rate": 8.386407858128706e-07, "loss": -0.0, "num_tokens": 3355569.0, "reward": 0.5796874761581421, "reward_std": 0.15852637588977814, "rewards/format_reward/mean": 0.640625, "rewards/format_reward/std": 0.22658175230026245, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "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": 76.0, "completions/max_terminated_length": 76.0, "completions/mean_length": 10.796875, "completions/mean_terminated_length": 10.796875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08, "frac_reward_zero_std": 0.25, "grad_norm": 21.881502151489258, "learning_rate": 8.262928807620843e-07, "loss": 0.0, "num_tokens": 3427308.0, "reward": 0.69921875, "reward_std": 0.2810230255126953, "rewards/format_reward/mean": 0.7421875, "rewards/format_reward/std": 0.25185325741767883, "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": 120.0, "completions/max_terminated_length": 120.0, "completions/mean_length": 13.28125, "completions/mean_terminated_length": 13.28125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0825, "frac_reward_zero_std": 0.125, "grad_norm": 84.99024963378906, "learning_rate": 8.135881792367685e-07, "loss": 0.0, "num_tokens": 3513686.0, "reward": 0.5703125, "reward_std": 0.1591934710741043, "rewards/format_reward/mean": 0.703125, "rewards/format_reward/std": 0.24750742316246033, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "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": 544.0, "completions/max_terminated_length": 544.0, "completions/mean_length": 63.5625, "completions/mean_terminated_length": 63.5625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.085, "frac_reward_zero_std": 0.375, "grad_norm": 15.079115867614746, "learning_rate": 8.005405736415125e-07, "loss": -0.0, "num_tokens": 3602858.0, "reward": 0.54296875, "reward_std": 0.28624227643013, "rewards/format_reward/mean": 0.8984375, "rewards/format_reward/std": 0.20275264978408813, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 34 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 502.0, "completions/max_terminated_length": 502.0, "completions/mean_length": 31.3125, "completions/mean_terminated_length": 31.3125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0875, "frac_reward_zero_std": 0.375, "grad_norm": 13.162272453308105, "learning_rate": 7.871643313414718e-07, "loss": -0.0, "num_tokens": 3694886.0, "reward": 0.6062500476837158, "reward_std": 0.25862938165664673, "rewards/format_reward/mean": 0.90625, "rewards/format_reward/std": 0.19669894874095917, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "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": 859.0, "completions/max_terminated_length": 859.0, "completions/mean_length": 52.9375, "completions/mean_terminated_length": 52.9375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.09, "frac_reward_zero_std": 0.125, "grad_norm": 26.653566360473633, "learning_rate": 7.734740790612136e-07, "loss": 0.0, "num_tokens": 3803314.0, "reward": 0.2906250059604645, "reward_std": 0.3420897126197815, "rewards/format_reward/mean": 0.875, "rewards/format_reward/std": 0.2357022762298584, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "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": 107.0, "completions/max_terminated_length": 107.0, "completions/mean_length": 18.28125, "completions/mean_terminated_length": 18.28125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0925, "frac_reward_zero_std": 0.75, "grad_norm": 18.36152458190918, "learning_rate": 7.594847868906076e-07, "loss": 0.0, "num_tokens": 3892388.0, "reward": 0.44218751788139343, "reward_std": 0.09280776977539062, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "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": 939.0, "completions/max_terminated_length": 939.0, "completions/mean_length": 52.578125, "completions/mean_terminated_length": 52.578125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.095, "frac_reward_zero_std": 0.375, "grad_norm": 18.953285217285156, "learning_rate": 7.452117519152541e-07, "loss": 0.0, "num_tokens": 3969345.0, "reward": 0.737500011920929, "reward_std": 0.2959960699081421, "rewards/format_reward/mean": 0.96875, "rewards/format_reward/std": 0.17536810040473938, "rewards/mcq_exact_match_reward/mean": 0.640625, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 11.625, "completions/mean_terminated_length": 11.625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0975, "frac_reward_zero_std": 0.75, "grad_norm": 26.365705490112305, "learning_rate": 7.306705814893439e-07, "loss": 0.0, "num_tokens": 4080665.0, "reward": 0.48750001192092896, "reward_std": 0.047534942626953125, "rewards/format_reward/mean": 0.96875, "rewards/format_reward/std": 0.12198751419782639, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1007.0, "completions/mean_length": 99.390625, "completions/mean_terminated_length": 68.46031951904297, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 7.158771761692464e-07, "loss": 0.0, "num_tokens": 4202850.0, "reward": 0.8375000357627869, "reward_std": 0.0, "rewards/format_reward/mean": 0.875, "rewards/format_reward/std": 0.3333333432674408, "rewards/mcq_exact_match_reward/mean": 0.75, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 114.0, "completions/max_terminated_length": 114.0, "completions/mean_length": 14.84375, "completions/mean_terminated_length": 14.84375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1025, "frac_reward_zero_std": 0.875, "grad_norm": 6.025423049926758, "learning_rate": 7.008477123264847e-07, "loss": 0.0, "num_tokens": 4289952.0, "reward": 0.3968750238418579, "reward_std": 0.0646936446428299, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 126.0, "completions/max_terminated_length": 126.0, "completions/mean_length": 15.890625, "completions/mean_terminated_length": 15.890625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.105, "frac_reward_zero_std": 0.75, "grad_norm": 15.665976524353027, "learning_rate": 6.855986244591103e-07, "loss": -0.0, "num_tokens": 4369649.0, "reward": 0.25, "reward_std": 0.0618792288005352, "rewards/format_reward/mean": 0.9375, "rewards/format_reward/std": 0.18898223340511322, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "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": 430.0, "completions/max_terminated_length": 430.0, "completions/mean_length": 28.90625, "completions/mean_terminated_length": 28.90625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1075, "frac_reward_zero_std": 0.625, "grad_norm": 9.897542953491211, "learning_rate": 6.701465872208216e-07, "loss": 0.0, "num_tokens": 4477763.0, "reward": 0.3968750238418579, "reward_std": 0.1804211586713791, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 869.0, "completions/max_terminated_length": 869.0, "completions/mean_length": 35.90625, "completions/mean_terminated_length": 35.90625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.11, "frac_reward_zero_std": 0.375, "grad_norm": 18.9465274810791, "learning_rate": 6.545084971874736e-07, "loss": 0.0, "num_tokens": 4574165.0, "reward": 0.6468750238418579, "reward_std": 0.29826053977012634, "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": 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": 22.0, "completions/max_terminated_length": 22.0, "completions/mean_length": 12.46875, "completions/mean_terminated_length": 12.46875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1125, "frac_reward_zero_std": 0.875, "grad_norm": 3.1505491733551025, "learning_rate": 6.387014543809223e-07, "loss": -0.0, "num_tokens": 4676571.0, "reward": 0.6156250238418579, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "step": 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": 302.0, "completions/max_terminated_length": 302.0, "completions/mean_length": 16.53125, "completions/mean_terminated_length": 16.53125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.115, "frac_reward_zero_std": 0.375, "grad_norm": 17.579618453979492, "learning_rate": 6.227427435703995e-07, "loss": 0.0, "num_tokens": 4805829.0, "reward": 0.40156251192092896, "reward_std": 0.2469378411769867, "rewards/format_reward/mean": 0.890625, "rewards/format_reward/std": 0.3145764470100403, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 116.0, "completions/max_terminated_length": 116.0, "completions/mean_length": 13.625, "completions/mean_terminated_length": 13.625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1175, "frac_reward_zero_std": 0.5, "grad_norm": 14.595392227172852, "learning_rate": 6.066498153718734e-07, "loss": 0.0, "num_tokens": 4919605.0, "reward": 0.44218751788139343, "reward_std": 0.21748682856559753, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "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": 11.4375, "completions/mean_terminated_length": 11.4375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.12, "frac_reward_zero_std": 0.5, "grad_norm": 16.478628158569336, "learning_rate": 5.90440267166055e-07, "loss": 0.0, "num_tokens": 5002329.0, "reward": 0.6421874761581421, "reward_std": 0.21040581166744232, "rewards/format_reward/mean": 0.953125, "rewards/format_reward/std": 0.14689241349697113, "rewards/mcq_exact_match_reward/mean": 0.546875, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 402.0, "completions/max_terminated_length": 402.0, "completions/mean_length": 18.09375, "completions/mean_terminated_length": 18.09375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1225, "frac_reward_zero_std": 0.5, "grad_norm": 18.771331787109375, "learning_rate": 5.741318238559209e-07, "loss": 0.0, "num_tokens": 5105351.0, "reward": 0.7718750238418579, "reward_std": 0.19044628739356995, "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": 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": 1091.0, "completions/max_terminated_length": 1091.0, "completions/mean_length": 55.734375, "completions/mean_terminated_length": 55.734375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.125, "frac_reward_zero_std": 0.375, "grad_norm": 23.37836265563965, "learning_rate": 5.577423184847931e-07, "loss": 0.0, "num_tokens": 5191422.0, "reward": 0.7250000238418579, "reward_std": 0.3093337416648865, "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": 50 } ], "logging_steps": 1, "max_steps": 100, "num_input_tokens_seen": 5191422, "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 }