{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.25, "eval_steps": 500, "global_step": 100, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "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 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 110.0, "completions/max_terminated_length": 110.0, "completions/mean_length": 9.65625, "completions/mean_terminated_length": 9.65625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1275, "frac_reward_zero_std": 0.625, "grad_norm": 35.150230407714844, "learning_rate": 5.412896727361662e-07, "loss": -0.0, "num_tokens": 5463781.0, "reward": 0.6124999523162842, "reward_std": 0.1962025910615921, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.5625, "rewards/mcq_exact_match_reward/std": 0.5, "step": 51 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 25.0, "completions/max_terminated_length": 25.0, "completions/mean_length": 6.296875, "completions/mean_terminated_length": 6.296875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.13, "frac_reward_zero_std": 0.375, "grad_norm": 31.79669952392578, "learning_rate": 5.247918773366111e-07, "loss": -0.0, "num_tokens": 5547904.0, "reward": 0.5179687142372131, "reward_std": 0.29211172461509705, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 52 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 638.0, "completions/max_terminated_length": 638.0, "completions/mean_length": 15.875, "completions/mean_terminated_length": 15.875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1325, "frac_reward_zero_std": 0.75, "grad_norm": 22.680667877197266, "learning_rate": 5.082669723831793e-07, "loss": 0.0, "num_tokens": 5624936.0, "reward": 0.6273437142372131, "reward_std": 0.06690335273742676, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.578125, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 53 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 375.0, "completions/max_terminated_length": 375.0, "completions/mean_length": 18.765625, "completions/mean_terminated_length": 18.765625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.135, "frac_reward_zero_std": 0.875, "grad_norm": 11.646509170532227, "learning_rate": 4.917330276168208e-07, "loss": -0.0, "num_tokens": 5729593.0, "reward": 0.9093749523162842, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.859375, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "step": 54 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 98.0, "completions/max_terminated_length": 98.0, "completions/mean_length": 7.4375, "completions/mean_terminated_length": 7.4375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1375, "frac_reward_zero_std": 0.5, "grad_norm": 40.399471282958984, "learning_rate": 4.752081226633888e-07, "loss": -0.0, "num_tokens": 5803605.0, "reward": 0.8312499523162842, "reward_std": 0.22461533546447754, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.78125, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "step": 55 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 358.0, "completions/max_terminated_length": 358.0, "completions/mean_length": 12.8125, "completions/mean_terminated_length": 12.8125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.14, "frac_reward_zero_std": 0.625, "grad_norm": 22.172245025634766, "learning_rate": 4.5871032726383385e-07, "loss": -0.0, "num_tokens": 5901849.0, "reward": 0.6851562261581421, "reward_std": 0.10426763445138931, "rewards/format_reward/mean": 0.4453125, "rewards/format_reward/std": 0.15728822350502014, "rewards/mcq_exact_match_reward/mean": 0.640625, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "step": 56 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 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": 0.1425, "frac_reward_zero_std": 0.75, "grad_norm": 30.754718780517578, "learning_rate": 4.4225768151520694e-07, "loss": -0.0, "num_tokens": 5985785.0, "reward": 0.37812498211860657, "reward_std": 0.10205792635679245, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 57 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1088.0, "completions/max_terminated_length": 1088.0, "completions/mean_length": 34.625, "completions/mean_terminated_length": 34.625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.145, "frac_reward_zero_std": 0.5, "grad_norm": 25.583332061767578, "learning_rate": 4.258681761440789e-07, "loss": -0.0, "num_tokens": 6085025.0, "reward": 0.4562499523162842, "reward_std": 0.22461533546447754, "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": 58 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 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": 0.1475, "frac_reward_zero_std": 0.75, "grad_norm": 79.62020874023438, "learning_rate": 4.095597328339452e-07, "loss": 0.0, "num_tokens": 6179769.0, "reward": 0.6124999523162842, "reward_std": 0.10888781398534775, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.5625, "rewards/mcq_exact_match_reward/std": 0.5, "step": 59 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 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": 0.15, "frac_reward_zero_std": 0.75, "grad_norm": 25.738685607910156, "learning_rate": 3.9335018462812664e-07, "loss": -0.0, "num_tokens": 6273489.0, "reward": 0.5656249523162842, "reward_std": 0.10205792635679245, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "step": 60 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 315.0, "completions/max_terminated_length": 315.0, "completions/mean_length": 10.828125, "completions/mean_terminated_length": 10.828125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1525, "frac_reward_zero_std": 0.75, "grad_norm": 20.475622177124023, "learning_rate": 3.772572564296004e-07, "loss": 0.0, "num_tokens": 6385854.0, "reward": 0.43984371423721313, "reward_std": 0.046403877437114716, "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": 61 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 250.0, "completions/max_terminated_length": 250.0, "completions/mean_length": 11.421875, "completions/mean_terminated_length": 11.421875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.155, "frac_reward_zero_std": 0.625, "grad_norm": 31.395235061645508, "learning_rate": 3.612985456190778e-07, "loss": -0.0, "num_tokens": 6484497.0, "reward": 0.29999998211860657, "reward_std": 0.17570312321186066, "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": 62 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 11.0, "completions/max_terminated_length": 11.0, "completions/mean_length": 6.078125, "completions/mean_terminated_length": 6.078125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1575, "frac_reward_zero_std": 0.375, "grad_norm": 29.636781692504883, "learning_rate": 3.454915028125263e-07, "loss": -0.0, "num_tokens": 6571750.0, "reward": 0.33124998211860657, "reward_std": 0.2709311246871948, "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": 63 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 132.0, "completions/max_terminated_length": 132.0, "completions/mean_length": 8.40625, "completions/mean_terminated_length": 8.40625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.16, "frac_reward_zero_std": 0.875, "grad_norm": 9.249646186828613, "learning_rate": 3.2985341277917846e-07, "loss": -0.0, "num_tokens": 6675304.0, "reward": 0.27812498807907104, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "step": 64 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 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": 0.1625, "frac_reward_zero_std": 0.625, "grad_norm": 22.266313552856445, "learning_rate": 3.1440137554088953e-07, "loss": -0.0, "num_tokens": 6734528.0, "reward": 0.5968749523162842, "reward_std": 0.1893727034330368, "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": 65 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 197.0, "completions/max_terminated_length": 197.0, "completions/mean_length": 14.578125, "completions/mean_terminated_length": 14.578125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.165, "frac_reward_zero_std": 0.75, "grad_norm": 23.659265518188477, "learning_rate": 2.9915228767351535e-07, "loss": 0.0, "num_tokens": 6822317.0, "reward": 0.6593749523162842, "reward_std": 0.10205791890621185, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.609375, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "step": 66 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 278.0, "completions/max_terminated_length": 278.0, "completions/mean_length": 10.25, "completions/mean_terminated_length": 10.25, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1675, "frac_reward_zero_std": 0.875, "grad_norm": 22.198257446289062, "learning_rate": 2.841228238307536e-07, "loss": -0.0, "num_tokens": 6896733.0, "reward": 0.4718749523162842, "reward_std": 0.0646936446428299, "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": 67 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 802.0, "completions/max_terminated_length": 802.0, "completions/mean_length": 28.03125, "completions/mean_terminated_length": 28.03125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.17, "frac_reward_zero_std": 0.75, "grad_norm": 32.631858825683594, "learning_rate": 2.6932941851065615e-07, "loss": -0.0, "num_tokens": 6981895.0, "reward": 0.41874998807907104, "reward_std": 0.1157275065779686, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 68 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 268.0, "completions/max_terminated_length": 268.0, "completions/mean_length": 15.015625, "completions/mean_terminated_length": 15.015625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1725, "frac_reward_zero_std": 0.875, "grad_norm": 11.709912300109863, "learning_rate": 2.547882480847461e-07, "loss": -0.0, "num_tokens": 7074120.0, "reward": 0.4093749523162842, "reward_std": 0.04419417306780815, "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": 69 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 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": 0.175, "frac_reward_zero_std": 0.75, "grad_norm": 44.08403015136719, "learning_rate": 2.4051521310939254e-07, "loss": -0.0, "num_tokens": 7155736.0, "reward": 0.4406249523162842, "reward_std": 0.12255740165710449, "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": 70 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 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": 0.1775, "frac_reward_zero_std": 0.5, "grad_norm": 33.12507247924805, "learning_rate": 2.2652592093878665e-07, "loss": -0.0, "num_tokens": 7237824.0, "reward": 0.5187499523162842, "reward_std": 0.2540663480758667, "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": 71 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 90.0, "completions/max_terminated_length": 90.0, "completions/mean_length": 7.3125, "completions/mean_terminated_length": 7.3125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.18, "frac_reward_zero_std": 0.75, "grad_norm": 15.500550270080566, "learning_rate": 2.128356686585282e-07, "loss": 0.0, "num_tokens": 7322692.0, "reward": 0.5343749523162842, "reward_std": 0.10205791890621185, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.484375, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "step": 72 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 363.0, "completions/max_terminated_length": 363.0, "completions/mean_length": 11.578125, "completions/mean_terminated_length": 11.578125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1825, "frac_reward_zero_std": 0.875, "grad_norm": 10.39971923828125, "learning_rate": 1.9945942635848745e-07, "loss": -0.0, "num_tokens": 7398641.0, "reward": 0.4562499523162842, "reward_std": 0.0578637532889843, "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": 73 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 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": 0.185, "frac_reward_zero_std": 0.75, "grad_norm": 44.666664123535156, "learning_rate": 1.8641182076323148e-07, "loss": -0.0, "num_tokens": 7484937.0, "reward": 0.6906249523162842, "reward_std": 0.10205792635679245, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.640625, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "step": 74 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 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": 0.1875, "frac_reward_zero_std": 0.75, "grad_norm": 22.949417114257812, "learning_rate": 1.7370711923791564e-07, "loss": -0.0, "num_tokens": 7564001.0, "reward": 0.5492187142372131, "reward_std": 0.11624953895807266, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "step": 75 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 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": 0.19, "frac_reward_zero_std": 0.75, "grad_norm": 17.887191772460938, "learning_rate": 1.6135921418712955e-07, "loss": -0.0, "num_tokens": 7658593.0, "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": 76 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 939.0, "completions/max_terminated_length": 939.0, "completions/mean_length": 20.578125, "completions/mean_terminated_length": 20.578125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1925, "frac_reward_zero_std": 0.75, "grad_norm": 23.46506118774414, "learning_rate": 1.493816078637557e-07, "loss": -0.0, "num_tokens": 7774854.0, "reward": 0.6281249523162842, "reward_std": 0.11100947111845016, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.578125, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 77 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 10.0, "completions/max_terminated_length": 10.0, "completions/mean_length": 6.0625, "completions/mean_terminated_length": 6.0625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.195, "frac_reward_zero_std": 0.625, "grad_norm": 31.387968063354492, "learning_rate": 1.3778739760445552e-07, "loss": -0.0, "num_tokens": 7854682.0, "reward": 0.7210937142372131, "reward_std": 0.1990053355693817, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.671875, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 78 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 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": 0.1975, "frac_reward_zero_std": 0.875, "grad_norm": 44.366573333740234, "learning_rate": 1.2658926150792322e-07, "loss": -0.0, "num_tokens": 7938162.0, "reward": 0.7843749523162842, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.734375, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "step": 79 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 411.0, "completions/max_terminated_length": 411.0, "completions/mean_length": 12.625, "completions/mean_terminated_length": 12.625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.2, "frac_reward_zero_std": 0.5, "grad_norm": 23.749942779541016, "learning_rate": 1.1579944457157059e-07, "loss": -0.0, "num_tokens": 8028026.0, "reward": 0.46718746423721313, "reward_std": 0.15597516298294067, "rewards/format_reward/mean": 0.453125, "rewards/format_reward/std": 0.14689241349697113, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 80 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1036.0, "completions/max_terminated_length": 1036.0, "completions/mean_length": 30.390625, "completions/mean_terminated_length": 30.390625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.2025, "frac_reward_zero_std": 0.625, "grad_norm": 43.081417083740234, "learning_rate": 1.0542974530180327e-07, "loss": -0.0, "num_tokens": 8128419.0, "reward": 0.6062499284744263, "reward_std": 0.1825428009033203, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.5625, "rewards/mcq_exact_match_reward/std": 0.5, "step": 81 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 163.0, "completions/max_terminated_length": 163.0, "completions/mean_length": 8.453125, "completions/mean_terminated_length": 8.453125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.205, "frac_reward_zero_std": 0.75, "grad_norm": 32.80108642578125, "learning_rate": 9.549150281252632e-08, "loss": -0.0, "num_tokens": 8231240.0, "reward": 0.4718749523162842, "reward_std": 0.11100947856903076, "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": 82 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 311.0, "completions/max_terminated_length": 311.0, "completions/mean_length": 16.125, "completions/mean_terminated_length": 16.125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.2075, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 8.599558442598998e-08, "loss": 0.0, "num_tokens": 8325712.0, "reward": 0.4249999523162842, "reward_std": 0.0, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 83 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 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": 0.21, "frac_reward_zero_std": 0.625, "grad_norm": 35.28303527832031, "learning_rate": 7.695237378953224e-08, "loss": 0.0, "num_tokens": 8397312.0, "reward": 0.7062499523162842, "reward_std": 0.16675156354904175, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.65625, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "step": 84 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 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": 0.2125, "frac_reward_zero_std": 0.625, "grad_norm": 31.763187408447266, "learning_rate": 6.837175952121304e-08, "loss": -0.0, "num_tokens": 8476328.0, "reward": 0.4562499523162842, "reward_std": 0.1462520956993103, "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": 85 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1151.0, "completions/max_terminated_length": 1151.0, "completions/mean_length": 44.578125, "completions/mean_terminated_length": 44.578125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.215, "frac_reward_zero_std": 0.625, "grad_norm": 27.318849563598633, "learning_rate": 6.026312439675551e-08, "loss": -0.0, "num_tokens": 8578765.0, "reward": 0.6593749523162842, "reward_std": 0.13258251547813416, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.609375, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "step": 86 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 371.0, "completions/max_terminated_length": 371.0, "completions/mean_length": 11.703125, "completions/mean_terminated_length": 11.703125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.2175, "frac_reward_zero_std": 0.75, "grad_norm": 36.50062561035156, "learning_rate": 5.263533508961826e-08, "loss": -0.0, "num_tokens": 8684858.0, "reward": 0.7999999523162842, "reward_std": 0.1157275065779686, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.75, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 87 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 284.0, "completions/max_terminated_length": 284.0, "completions/mean_length": 10.34375, "completions/mean_terminated_length": 10.34375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.22, "frac_reward_zero_std": 0.875, "grad_norm": 21.864126205444336, "learning_rate": 4.549673247541874e-08, "loss": -0.0, "num_tokens": 8764608.0, "reward": 0.9093749523162842, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.859375, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "step": 88 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 126.0, "completions/max_terminated_length": 126.0, "completions/mean_length": 8.90625, "completions/mean_terminated_length": 8.90625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.2225, "frac_reward_zero_std": 0.875, "grad_norm": 10.64477252960205, "learning_rate": 3.8855122511307626e-08, "loss": -0.0, "num_tokens": 8871890.0, "reward": 0.8624999523162842, "reward_std": 0.06681530177593231, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.8125, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "step": 89 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 67.0, "completions/max_terminated_length": 67.0, "completions/mean_length": 6.953125, "completions/mean_terminated_length": 6.953125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.225, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 3.271776770026963e-08, "loss": 0.0, "num_tokens": 8962527.0, "reward": 0.5437499284744263, "reward_std": 0.0, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "step": 90 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 595.0, "completions/max_terminated_length": 595.0, "completions/mean_length": 18.53125, "completions/mean_terminated_length": 18.53125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.2275, "frac_reward_zero_std": 0.75, "grad_norm": 11.680058479309082, "learning_rate": 2.7091379149682682e-08, "loss": 0.0, "num_tokens": 9051889.0, "reward": 0.7062499523162842, "reward_std": 0.0883883386850357, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.65625, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "step": 91 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 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": 0.23, "frac_reward_zero_std": 0.75, "grad_norm": 31.518062591552734, "learning_rate": 2.1982109232821176e-08, "loss": -0.0, "num_tokens": 9144145.0, "reward": 0.6437499523162842, "reward_std": 0.12467905879020691, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.59375, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 92 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 315.0, "completions/max_terminated_length": 315.0, "completions/mean_length": 10.828125, "completions/mean_terminated_length": 10.828125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.2325, "frac_reward_zero_std": 0.875, "grad_norm": 11.914877891540527, "learning_rate": 1.7395544861325718e-08, "loss": -0.0, "num_tokens": 9232974.0, "reward": 0.6593749523162842, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.609375, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "step": 93 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 495.0, "completions/max_terminated_length": 495.0, "completions/mean_length": 13.640625, "completions/mean_terminated_length": 13.640625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.235, "frac_reward_zero_std": 0.625, "grad_norm": 32.350364685058594, "learning_rate": 1.3336701375997127e-08, "loss": -0.0, "num_tokens": 9326127.0, "reward": 0.6749999523162842, "reward_std": 0.16675156354904175, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.625, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 94 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 24.0, "completions/max_terminated_length": 24.0, "completions/mean_length": 6.28125, "completions/mean_terminated_length": 6.28125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.2375, "frac_reward_zero_std": 0.75, "grad_norm": 18.34619903564453, "learning_rate": 9.810017062595321e-09, "loss": -0.0, "num_tokens": 9416065.0, "reward": 0.4398437440395355, "reward_std": 0.12402750551700592, "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": 95 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 164.0, "completions/max_terminated_length": 164.0, "completions/mean_length": 8.46875, "completions/mean_terminated_length": 8.46875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.24, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 6.819348298638839e-09, "loss": 0.0, "num_tokens": 9499807.0, "reward": 0.7999999523162842, "reward_std": 0.0, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.75, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 96 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 771.0, "completions/max_terminated_length": 771.0, "completions/mean_length": 20.0, "completions/mean_terminated_length": 20.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.2425, "frac_reward_zero_std": 0.875, "grad_norm": 7.09723424911499, "learning_rate": 4.367965336512403e-09, "loss": -0.0, "num_tokens": 9618511.0, "reward": 0.4874999523162842, "reward_std": 0.06681530177593231, "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": 97 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 304.0, "completions/max_terminated_length": 304.0, "completions/mean_length": 13.0625, "completions/mean_terminated_length": 13.0625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.245, "frac_reward_zero_std": 0.75, "grad_norm": 18.552555084228516, "learning_rate": 2.458548727494292e-09, "loss": 0.0, "num_tokens": 9718563.0, "reward": 0.4507812261581421, "reward_std": 0.06007346138358116, "rewards/format_reward/mean": 0.4453125, "rewards/format_reward/std": 0.15728822350502014, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 98 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1017.0, "completions/max_terminated_length": 1017.0, "completions/mean_length": 22.328125, "completions/mean_terminated_length": 22.328125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.2475, "frac_reward_zero_std": 0.875, "grad_norm": 25.63117027282715, "learning_rate": 1.0931863906127325e-09, "loss": -0.0, "num_tokens": 9828168.0, "reward": 0.6281249523162842, "reward_std": 0.0646936446428299, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.578125, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 99 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 299.0, "completions/max_terminated_length": 299.0, "completions/mean_length": 10.578125, "completions/mean_terminated_length": 10.578125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.25, "frac_reward_zero_std": 0.875, "grad_norm": 15.285822868347168, "learning_rate": 2.733713295369755e-10, "loss": -0.0, "num_tokens": 9929653.0, "reward": 0.5812499523162842, "reward_std": 0.0578637532889843, "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": 100 } ], "logging_steps": 1, "max_steps": 100, "num_input_tokens_seen": 9929653, "num_train_epochs": 1, "save_steps": 50, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }