{ "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": 1684.0, "completions/max_terminated_length": 1684.0, "completions/mean_length": 546.828125, "completions/mean_terminated_length": 546.828125, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.0025, "frac_reward_zero_std": 0.0, "grad_norm": 14.495871543884277, "learning_rate": 0.0, "loss": -0.0, "num_tokens": 128317.0, "reward": 0.3664062023162842, "reward_std": 0.4292623698711395, "rewards/format_reward/mean": 0.3828125, "rewards/format_reward/std": 0.2313210517168045, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 1258.0, "completions/max_terminated_length": 1258.0, "completions/mean_length": 387.03125, "completions/mean_terminated_length": 387.03125, "completions/min_length": 7.0, "completions/min_terminated_length": 7.0, "epoch": 0.005, "frac_reward_zero_std": 0.0, "grad_norm": 3.316119909286499, "learning_rate": 2e-07, "loss": -0.0, "num_tokens": 219231.0, "reward": 0.2359374761581421, "reward_std": 0.2545059621334076, "rewards/format_reward/mean": 0.328125, "rewards/format_reward/std": 0.29839184880256653, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "step": 2 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1645.0, "completions/mean_length": 498.5625, "completions/mean_terminated_length": 473.9682922363281, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0075, "frac_reward_zero_std": 0.0, "grad_norm": 7.069388389587402, "learning_rate": 4e-07, "loss": -0.0, "num_tokens": 346923.0, "reward": 0.20546872913837433, "reward_std": 0.2610474228858948, "rewards/format_reward/mean": 0.3359375, "rewards/format_reward/std": 0.2366211861371994, "rewards/mcq_exact_match_reward/mean": 0.171875, "rewards/mcq_exact_match_reward/std": 0.38025420904159546, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1938.0, "completions/mean_length": 635.796875, "completions/mean_terminated_length": 566.34423828125, "completions/min_length": 56.0, "completions/min_terminated_length": 56.0, "epoch": 0.01, "frac_reward_zero_std": 0.0, "grad_norm": 1.7901220321655273, "learning_rate": 6e-07, "loss": 0.0, "num_tokens": 506542.0, "reward": 0.16249999403953552, "reward_std": 0.18332982063293457, "rewards/format_reward/mean": 0.375, "rewards/format_reward/std": 0.2182178944349289, "rewards/mcq_exact_match_reward/mean": 0.125, "rewards/mcq_exact_match_reward/std": 0.3333333432674408, "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": 0.0, "completions/max_length": 1792.0, "completions/max_terminated_length": 1792.0, "completions/mean_length": 431.40625, "completions/mean_terminated_length": 431.40625, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.0125, "frac_reward_zero_std": 0.0, "grad_norm": 23.413793563842773, "learning_rate": 8e-07, "loss": 0.0, "num_tokens": 624448.0, "reward": 0.13984374701976776, "reward_std": 0.2583358883857727, "rewards/format_reward/mean": 0.3046875, "rewards/format_reward/std": 0.2615155577659607, "rewards/mcq_exact_match_reward/mean": 0.109375, "rewards/mcq_exact_match_reward/std": 0.3145764470100403, "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": 1707.0, "completions/max_terminated_length": 1707.0, "completions/mean_length": 613.921875, "completions/mean_terminated_length": 613.921875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.015, "frac_reward_zero_std": 0.0, "grad_norm": 7.8830718994140625, "learning_rate": 1e-06, "loss": 0.0, "num_tokens": 767579.0, "reward": 0.27421873807907104, "reward_std": 0.32973340153694153, "rewards/format_reward/mean": 0.3984375, "rewards/format_reward/std": 0.20275264978408813, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "step": 6 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1334.0, "completions/mean_length": 510.734375, "completions/mean_terminated_length": 486.3333740234375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0175, "frac_reward_zero_std": 0.0, "grad_norm": 12.138351440429688, "learning_rate": 9.99726628670463e-07, "loss": 0.0, "num_tokens": 897962.0, "reward": 0.17734374105930328, "reward_std": 0.28381556272506714, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.23974503576755524, "rewards/mcq_exact_match_reward/mean": 0.140625, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "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": 1592.0, "completions/max_terminated_length": 1592.0, "completions/mean_length": 552.140625, "completions/mean_terminated_length": 552.140625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.02, "frac_reward_zero_std": 0.0, "grad_norm": 12.481257438659668, "learning_rate": 9.989068136093872e-07, "loss": 0.0, "num_tokens": 1013635.0, "reward": 0.34687501192092896, "reward_std": 0.41548749804496765, "rewards/format_reward/mean": 0.34375, "rewards/format_reward/std": 0.29378482699394226, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 1593.0, "completions/mean_length": 550.703125, "completions/mean_terminated_length": 526.9365234375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0225, "frac_reward_zero_std": 0.0, "grad_norm": 6.228868007659912, "learning_rate": 9.975414512725056e-07, "loss": 0.0, "num_tokens": 1149904.0, "reward": 0.23124998807907104, "reward_std": 0.27496111392974854, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.22712838649749756, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "step": 9 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1936.0, "completions/max_terminated_length": 1936.0, "completions/mean_length": 530.953125, "completions/mean_terminated_length": 530.953125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.025, "frac_reward_zero_std": 0.125, "grad_norm": 11.466665267944336, "learning_rate": 9.956320346634875e-07, "loss": -0.0, "num_tokens": 1268701.0, "reward": 0.46015623211860657, "reward_std": 0.44074398279190063, "rewards/format_reward/mean": 0.3828125, "rewards/format_reward/std": 0.21347814798355103, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1606.0, "completions/mean_length": 573.9375, "completions/mean_terminated_length": 550.5397338867188, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0275, "frac_reward_zero_std": 0.0, "grad_norm": 4.589084148406982, "learning_rate": 9.931806517013612e-07, "loss": -0.0, "num_tokens": 1399529.0, "reward": 0.3062499761581421, "reward_std": 0.3655306100845337, "rewards/format_reward/mean": 0.40625, "rewards/format_reward/std": 0.21593283116817474, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1703.0, "completions/mean_length": 576.796875, "completions/mean_terminated_length": 504.4425964355469, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.03, "frac_reward_zero_std": 0.25, "grad_norm": 3.7078964710235596, "learning_rate": 9.901899829374047e-07, "loss": 0.0, "num_tokens": 1543364.0, "reward": 0.23281249403953552, "reward_std": 0.2547384798526764, "rewards/format_reward/mean": 0.453125, "rewards/format_reward/std": 0.19352105259895325, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "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": 1525.0, "completions/mean_length": 628.265625, "completions/mean_terminated_length": 605.7301635742188, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0325, "frac_reward_zero_std": 0.0, "grad_norm": 5.075721263885498, "learning_rate": 9.866632986240029e-07, "loss": 0.0, "num_tokens": 1671269.0, "reward": 0.24765624105930328, "reward_std": 0.3335675597190857, "rewards/format_reward/mean": 0.4453125, "rewards/format_reward/std": 0.2203386276960373, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "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": 0.0, "completions/max_length": 1191.0, "completions/max_terminated_length": 1191.0, "completions/mean_length": 350.703125, "completions/mean_terminated_length": 350.703125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.035, "frac_reward_zero_std": 0.125, "grad_norm": 16.176406860351562, "learning_rate": 9.826044551386742e-07, "loss": -0.0, "num_tokens": 1792610.0, "reward": 0.42031246423721313, "reward_std": 0.31208381056785583, "rewards/format_reward/mean": 0.453125, "rewards/format_reward/std": 0.14689241349697113, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 14 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1870.0, "completions/mean_length": 516.71875, "completions/mean_terminated_length": 492.4127197265625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0375, "frac_reward_zero_std": 0.25, "grad_norm": 1.744857668876648, "learning_rate": 9.780178907671788e-07, "loss": 0.0, "num_tokens": 1907352.0, "reward": 0.32734373211860657, "reward_std": 0.2742621898651123, "rewards/format_reward/mean": 0.4609375, "rewards/format_reward/std": 0.16194961965084076, "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": 1174.0, "completions/max_terminated_length": 1174.0, "completions/mean_length": 323.96875, "completions/mean_terminated_length": 323.96875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.04, "frac_reward_zero_std": 0.0, "grad_norm": 18.771692276000977, "learning_rate": 9.729086208503173e-07, "loss": -0.0, "num_tokens": 2031870.0, "reward": 0.4375, "reward_std": 0.4681442677974701, "rewards/format_reward/mean": 0.46875, "rewards/format_reward/std": 0.12198751419782639, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 1318.0, "completions/max_terminated_length": 1318.0, "completions/mean_length": 414.171875, "completions/mean_terminated_length": 414.171875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0425, "frac_reward_zero_std": 0.0, "grad_norm": 15.930991172790527, "learning_rate": 9.672822322997304e-07, "loss": 0.0, "num_tokens": 2114057.0, "reward": 0.2671874761581421, "reward_std": 0.3394935727119446, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.17747680842876434, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "step": 17 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1849.0, "completions/mean_length": 490.859375, "completions/mean_terminated_length": 440.6290283203125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.045, "frac_reward_zero_std": 0.25, "grad_norm": 20.626331329345703, "learning_rate": 9.611448774886923e-07, "loss": -0.0, "num_tokens": 2228400.0, "reward": 0.2968749701976776, "reward_std": 0.23979520797729492, "rewards/format_reward/mean": 0.46875, "rewards/format_reward/std": 0.1510545015335083, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 18 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1590.0, "completions/mean_length": 391.890625, "completions/mean_terminated_length": 365.60321044921875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0475, "frac_reward_zero_std": 0.25, "grad_norm": 11.943554878234863, "learning_rate": 9.545032675245813e-07, "loss": 0.0, "num_tokens": 2357113.0, "reward": 0.4039062261581421, "reward_std": 0.22949570417404175, "rewards/format_reward/mean": 0.4453125, "rewards/format_reward/std": 0.180765300989151, "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": 0.0, "completions/max_length": 1165.0, "completions/max_terminated_length": 1165.0, "completions/mean_length": 95.828125, "completions/mean_terminated_length": 95.828125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05, "frac_reward_zero_std": 0.625, "grad_norm": 18.862329483032227, "learning_rate": 9.473646649103817e-07, "loss": -0.0, "num_tokens": 2453430.0, "reward": 0.19140625, "reward_std": 0.1594799906015396, "rewards/format_reward/mean": 0.5078125, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.140625, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "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": 1178.0, "completions/max_terminated_length": 1178.0, "completions/mean_length": 241.328125, "completions/mean_terminated_length": 241.328125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0525, "frac_reward_zero_std": 0.25, "grad_norm": 12.630736351013184, "learning_rate": 9.397368756032444e-07, "loss": -0.0, "num_tokens": 2584963.0, "reward": 0.3460937440395355, "reward_std": 0.33498305082321167, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 854.0, "completions/max_terminated_length": 854.0, "completions/mean_length": 87.859375, "completions/mean_terminated_length": 87.859375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.055, "frac_reward_zero_std": 0.25, "grad_norm": 18.5394229888916, "learning_rate": 9.316282404787869e-07, "loss": -0.0, "num_tokens": 2677362.0, "reward": 0.3312499523162842, "reward_std": 0.36295413970947266, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "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": 1750.0, "completions/max_terminated_length": 1750.0, "completions/mean_length": 398.90625, "completions/mean_terminated_length": 398.90625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0575, "frac_reward_zero_std": 0.125, "grad_norm": 16.368852615356445, "learning_rate": 9.230476262104676e-07, "loss": -0.0, "num_tokens": 2790796.0, "reward": 0.4234374761581421, "reward_std": 0.3852473795413971, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.08768405020236969, "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": 1432.0, "completions/max_terminated_length": 1432.0, "completions/mean_length": 131.0, "completions/mean_terminated_length": 131.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06, "frac_reward_zero_std": 0.75, "grad_norm": 5.738178730010986, "learning_rate": 9.1400441557401e-07, "loss": -0.0, "num_tokens": 2897452.0, "reward": 0.2835937440395355, "reward_std": 0.12366390228271484, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 811.0, "completions/max_terminated_length": 811.0, "completions/mean_length": 115.46875, "completions/mean_terminated_length": 115.46875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0625, "frac_reward_zero_std": 0.375, "grad_norm": 27.89461898803711, "learning_rate": 9.045084971874737e-07, "loss": -0.0, "num_tokens": 2970330.0, "reward": 0.36171871423721313, "reward_std": 0.3036596477031708, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 834.0, "completions/max_terminated_length": 834.0, "completions/mean_length": 119.828125, "completions/mean_terminated_length": 119.828125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.065, "frac_reward_zero_std": 0.375, "grad_norm": 20.274412155151367, "learning_rate": 8.945702546981968e-07, "loss": -0.0, "num_tokens": 3067951.0, "reward": 0.4562499523162842, "reward_std": 0.2798826992511749, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 2015.0, "completions/max_terminated_length": 2015.0, "completions/mean_length": 128.5625, "completions/mean_terminated_length": 128.5625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0675, "frac_reward_zero_std": 0.25, "grad_norm": 19.104936599731445, "learning_rate": 8.842005554284295e-07, "loss": 0.0, "num_tokens": 3181171.0, "reward": 0.3140624761581421, "reward_std": 0.2512032091617584, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.08768405020236969, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 508.0, "completions/max_terminated_length": 508.0, "completions/mean_length": 25.09375, "completions/mean_terminated_length": 25.09375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07, "frac_reward_zero_std": 0.75, "grad_norm": 12.246910095214844, "learning_rate": 8.734107384920769e-07, "loss": -0.0, "num_tokens": 3273865.0, "reward": 0.7531249523162842, "reward_std": 0.10205792635679245, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.703125, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 651.0, "completions/max_terminated_length": 651.0, "completions/mean_length": 62.15625, "completions/mean_terminated_length": 62.15625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0725, "frac_reward_zero_std": 0.625, "grad_norm": 24.603952407836914, "learning_rate": 8.622126023955445e-07, "loss": -0.0, "num_tokens": 3368763.0, "reward": 0.3929687440395355, "reward_std": 0.1555730104446411, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "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": 1078.0, "completions/max_terminated_length": 1078.0, "completions/mean_length": 195.828125, "completions/mean_terminated_length": 195.828125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.075, "frac_reward_zero_std": 0.375, "grad_norm": 25.076967239379883, "learning_rate": 8.506183921362442e-07, "loss": -0.0, "num_tokens": 3468920.0, "reward": 0.5031249523162842, "reward_std": 0.3098084330558777, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 830.0, "completions/max_terminated_length": 830.0, "completions/mean_length": 108.453125, "completions/mean_terminated_length": 108.453125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0775, "frac_reward_zero_std": 0.5, "grad_norm": 13.948243141174316, "learning_rate": 8.386407858128706e-07, "loss": 0.0, "num_tokens": 3554133.0, "reward": 0.43984371423721313, "reward_std": 0.19158241152763367, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "step": 31 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 953.0, "completions/max_terminated_length": 953.0, "completions/mean_length": 80.96875, "completions/mean_terminated_length": 80.96875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08, "frac_reward_zero_std": 0.5, "grad_norm": 24.682003021240234, "learning_rate": 8.262928807620843e-07, "loss": -0.0, "num_tokens": 3643467.0, "reward": 0.4562499523162842, "reward_std": 0.23356688022613525, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 32 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1012.0, "completions/mean_length": 95.578125, "completions/mean_terminated_length": 64.5873031616211, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0825, "frac_reward_zero_std": 0.5, "grad_norm": 14.139244079589844, "learning_rate": 8.135881792367685e-07, "loss": 0.0, "num_tokens": 3732528.0, "reward": 0.8773437142372131, "reward_std": 0.13479222357273102, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.828125, "rewards/mcq_exact_match_reward/std": 0.38025420904159546, "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": 262.0, "completions/max_terminated_length": 262.0, "completions/mean_length": 23.96875, "completions/mean_terminated_length": 23.96875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.085, "frac_reward_zero_std": 0.625, "grad_norm": 18.356752395629883, "learning_rate": 8.005405736415125e-07, "loss": -0.0, "num_tokens": 3835070.0, "reward": 0.5187499523162842, "reward_std": 0.19149436056613922, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "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": 479.0, "completions/max_terminated_length": 479.0, "completions/mean_length": 22.1875, "completions/mean_terminated_length": 22.1875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0875, "frac_reward_zero_std": 0.375, "grad_norm": 23.347909927368164, "learning_rate": 7.871643313414718e-07, "loss": 0.0, "num_tokens": 3940994.0, "reward": 0.4406249523162842, "reward_std": 0.24831001460552216, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.09, "frac_reward_zero_std": 0.75, "grad_norm": 15.116031646728516, "learning_rate": 7.734740790612136e-07, "loss": 0.0, "num_tokens": 4044306.0, "reward": 0.40937498211860657, "reward_std": 0.10205791890621185, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 337.0, "completions/max_terminated_length": 337.0, "completions/mean_length": 12.203125, "completions/mean_terminated_length": 12.203125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0925, "frac_reward_zero_std": 0.75, "grad_norm": 16.849199295043945, "learning_rate": 7.594847868906076e-07, "loss": 0.0, "num_tokens": 4155711.0, "reward": 0.5187499523162842, "reward_std": 0.10888781398534775, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "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": 663.0, "completions/max_terminated_length": 663.0, "completions/mean_length": 22.421875, "completions/mean_terminated_length": 22.421875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.095, "frac_reward_zero_std": 0.5, "grad_norm": 17.486907958984375, "learning_rate": 7.452117519152541e-07, "loss": 0.0, "num_tokens": 4242322.0, "reward": 0.29999998211860657, "reward_std": 0.21306739747524261, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 308.0, "completions/max_terminated_length": 308.0, "completions/mean_length": 12.421875, "completions/mean_terminated_length": 12.421875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0975, "frac_reward_zero_std": 0.5, "grad_norm": 20.448984146118164, "learning_rate": 7.306705814893439e-07, "loss": -0.0, "num_tokens": 4342661.0, "reward": 0.33124998211860657, "reward_std": 0.22461532056331635, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.28125, "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": 638.0, "completions/max_terminated_length": 638.0, "completions/mean_length": 26.59375, "completions/mean_terminated_length": 26.59375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1, "frac_reward_zero_std": 0.5, "grad_norm": 18.941287994384766, "learning_rate": 7.158771761692464e-07, "loss": 0.0, "num_tokens": 4438755.0, "reward": 0.4874999523162842, "reward_std": 0.19727616012096405, "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": 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": 460.0, "completions/max_terminated_length": 460.0, "completions/mean_length": 29.109375, "completions/mean_terminated_length": 29.109375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1025, "frac_reward_zero_std": 0.5, "grad_norm": 28.462017059326172, "learning_rate": 7.008477123264847e-07, "loss": -0.0, "num_tokens": 4525850.0, "reward": 0.4874999523162842, "reward_std": 0.22461532056331635, "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": 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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.105, "frac_reward_zero_std": 0.75, "grad_norm": 20.120864868164062, "learning_rate": 6.855986244591103e-07, "loss": 0.0, "num_tokens": 4611154.0, "reward": 0.4718749523162842, "reward_std": 0.10205791890621185, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1075, "frac_reward_zero_std": 0.875, "grad_norm": 17.602794647216797, "learning_rate": 6.701465872208216e-07, "loss": -0.0, "num_tokens": 4714530.0, "reward": 0.5187499523162842, "reward_std": 0.0578637532889843, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "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": 338.0, "completions/max_terminated_length": 338.0, "completions/mean_length": 11.1875, "completions/mean_terminated_length": 11.1875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.11, "frac_reward_zero_std": 0.5, "grad_norm": 32.34278106689453, "learning_rate": 6.545084971874736e-07, "loss": -0.0, "num_tokens": 4782310.0, "reward": 0.8156249523162842, "reward_std": 0.2109457403421402, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.765625, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 308.0, "completions/max_terminated_length": 308.0, "completions/mean_length": 17.59375, "completions/mean_terminated_length": 17.59375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1125, "frac_reward_zero_std": 0.875, "grad_norm": 18.601213455200195, "learning_rate": 6.387014543809223e-07, "loss": -0.0, "num_tokens": 4845348.0, "reward": 0.5968749523162842, "reward_std": 0.0646936446428299, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.546875, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 603.0, "completions/max_terminated_length": 603.0, "completions/mean_length": 32.75, "completions/mean_terminated_length": 32.75, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.115, "frac_reward_zero_std": 0.625, "grad_norm": 23.18716049194336, "learning_rate": 6.227427435703995e-07, "loss": -0.0, "num_tokens": 4962700.0, "reward": 0.4562499523162842, "reward_std": 0.1735912561416626, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 729.0, "completions/max_terminated_length": 729.0, "completions/mean_length": 22.359375, "completions/mean_terminated_length": 22.359375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1175, "frac_reward_zero_std": 0.75, "grad_norm": 17.54852294921875, "learning_rate": 6.066498153718734e-07, "loss": -0.0, "num_tokens": 5056715.0, "reward": 0.7218749523162842, "reward_std": 0.10205791890621185, "rewards/format_reward/mean": 0.5, "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": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.12, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 5.90440267166055e-07, "loss": 0.0, "num_tokens": 5176635.0, "reward": 0.29999998211860657, "reward_std": 0.0, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 139.0, "completions/max_terminated_length": 139.0, "completions/mean_length": 12.703125, "completions/mean_terminated_length": 12.703125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1225, "frac_reward_zero_std": 0.625, "grad_norm": 23.021032333374023, "learning_rate": 5.741318238559209e-07, "loss": -0.0, "num_tokens": 5299304.0, "reward": 0.5812499523162842, "reward_std": 0.1735912561416626, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "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": 1053.0, "completions/max_terminated_length": 1053.0, "completions/mean_length": 51.296875, "completions/mean_terminated_length": 51.296875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.125, "frac_reward_zero_std": 0.75, "grad_norm": 23.403974533081055, "learning_rate": 5.577423184847931e-07, "loss": -0.0, "num_tokens": 5386643.0, "reward": 0.32499998807907104, "reward_std": 0.10888782143592834, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 50 } ], "logging_steps": 1, "max_steps": 100, "num_input_tokens_seen": 5386643, "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 }