{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.11160714285714286, "eval_steps": 500, "global_step": 50, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1990.0, "completions/mean_length": 538.546875, "completions/mean_terminated_length": 514.5873413085938, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.002232142857142857, "frac_reward_zero_std": 0.0, "grad_norm": 5.357117652893066, "learning_rate": 0.0, "loss": -0.0, "num_tokens": 112787.0, "reward": 0.30156248807907104, "reward_std": 0.23618578910827637, "rewards/format_reward/mean": 0.359375, "rewards/format_reward/std": 0.22658175230026245, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 1528.0, "completions/max_terminated_length": 1528.0, "completions/mean_length": 472.5625, "completions/mean_terminated_length": 472.5625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.004464285714285714, "frac_reward_zero_std": 0.0, "grad_norm": 6.3122878074646, "learning_rate": 2e-07, "loss": -0.0, "num_tokens": 229703.0, "reward": 0.2749999761581421, "reward_std": 0.3567870259284973, "rewards/format_reward/mean": 0.25, "rewards/format_reward/std": 0.2519763112068176, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 2 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1485.0, "completions/max_terminated_length": 1485.0, "completions/mean_length": 529.59375, "completions/mean_terminated_length": 529.59375, "completions/min_length": 9.0, "completions/min_terminated_length": 9.0, "epoch": 0.006696428571428571, "frac_reward_zero_std": 0.0, "grad_norm": 4.27625846862793, "learning_rate": 4e-07, "loss": 0.0, "num_tokens": 334789.0, "reward": 0.2554687261581421, "reward_std": 0.40990781784057617, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.28510910272598267, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "step": 3 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1175.0, "completions/mean_length": 357.5625, "completions/mean_terminated_length": 330.73016357421875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.008928571428571428, "frac_reward_zero_std": 0.125, "grad_norm": 23.33127212524414, "learning_rate": 6e-07, "loss": 0.0, "num_tokens": 458113.0, "reward": 0.27265623211860657, "reward_std": 0.284196674823761, "rewards/format_reward/mean": 0.3828125, "rewards/format_reward/std": 0.2780628502368927, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "step": 4 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1768.0, "completions/mean_length": 566.59375, "completions/mean_terminated_length": 518.8064575195312, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.011160714285714286, "frac_reward_zero_std": 0.0, "grad_norm": 6.540904998779297, "learning_rate": 8e-07, "loss": 0.0, "num_tokens": 592039.0, "reward": 0.3968749940395355, "reward_std": 0.4285091459751129, "rewards/format_reward/mean": 0.375, "rewards/format_reward/std": 0.2357022762298584, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 1872.0, "completions/max_terminated_length": 1872.0, "completions/mean_length": 469.515625, "completions/mean_terminated_length": 469.515625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.013392857142857142, "frac_reward_zero_std": 0.0, "grad_norm": 11.142348289489746, "learning_rate": 1e-06, "loss": 0.0, "num_tokens": 708496.0, "reward": 0.36328125, "reward_std": 0.4375847578048706, "rewards/format_reward/mean": 0.3515625, "rewards/format_reward/std": 0.29113471508026123, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 1686.0, "completions/mean_length": 506.09375, "completions/mean_terminated_length": 481.61907958984375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.015625, "frac_reward_zero_std": 0.0, "grad_norm": 7.097582817077637, "learning_rate": 9.99726628670463e-07, "loss": 0.0, "num_tokens": 826550.0, "reward": 0.46562498807907104, "reward_std": 0.361045241355896, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.2597312331199646, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 7 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1335.0, "completions/mean_length": 493.46875, "completions/mean_terminated_length": 468.7936706542969, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.017857142857142856, "frac_reward_zero_std": 0.0, "grad_norm": 10.78849983215332, "learning_rate": 9.989068136093872e-07, "loss": -0.0, "num_tokens": 950020.0, "reward": 0.4062499701976776, "reward_std": 0.43662554025650024, "rewards/format_reward/mean": 0.3125, "rewards/format_reward/std": 0.24397502839565277, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 0.0625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1590.0, "completions/mean_length": 573.0, "completions/mean_terminated_length": 474.66668701171875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.020089285714285716, "frac_reward_zero_std": 0.0, "grad_norm": 15.594640731811523, "learning_rate": 9.975414512725056e-07, "loss": -0.0, "num_tokens": 1076804.0, "reward": 0.5062499642372131, "reward_std": 0.4312324523925781, "rewards/format_reward/mean": 0.375, "rewards/format_reward/std": 0.2357022762298584, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 9 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1214.0, "completions/mean_length": 438.484375, "completions/mean_terminated_length": 412.9365234375, "completions/min_length": 5.0, "completions/min_terminated_length": 5.0, "epoch": 0.022321428571428572, "frac_reward_zero_std": 0.125, "grad_norm": 13.5233736038208, "learning_rate": 9.956320346634875e-07, "loss": -0.0, "num_tokens": 1191307.0, "reward": 0.3671874701976776, "reward_std": 0.3545979857444763, "rewards/format_reward/mean": 0.390625, "rewards/format_reward/std": 0.2083333432674408, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 10 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1973.0, "completions/max_terminated_length": 1973.0, "completions/mean_length": 430.59375, "completions/mean_terminated_length": 430.59375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.024553571428571428, "frac_reward_zero_std": 0.0, "grad_norm": 6.149971008300781, "learning_rate": 9.931806517013612e-07, "loss": 0.0, "num_tokens": 1319913.0, "reward": 0.28203123807907104, "reward_std": 0.3983161449432373, "rewards/format_reward/mean": 0.4765625, "rewards/format_reward/std": 0.25865477323532104, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "step": 11 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1246.0, "completions/max_terminated_length": 1246.0, "completions/mean_length": 423.96875, "completions/mean_terminated_length": 423.96875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.026785714285714284, "frac_reward_zero_std": 0.0, "grad_norm": 14.51206111907959, "learning_rate": 9.901899829374047e-07, "loss": -0.0, "num_tokens": 1450607.0, "reward": 0.2164062261581421, "reward_std": 0.32999470829963684, "rewards/format_reward/mean": 0.4453125, "rewards/format_reward/std": 0.20152568817138672, "rewards/mcq_exact_match_reward/mean": 0.171875, "rewards/mcq_exact_match_reward/std": 0.38025420904159546, "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": 0.0625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1615.0, "completions/mean_length": 610.75, "completions/mean_terminated_length": 514.933349609375, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.029017857142857144, "frac_reward_zero_std": 0.0, "grad_norm": 15.69182014465332, "learning_rate": 9.866632986240029e-07, "loss": -0.0, "num_tokens": 1610687.0, "reward": 0.21015623211860657, "reward_std": 0.2368287742137909, "rewards/format_reward/mean": 0.3828125, "rewards/format_reward/std": 0.2634054720401764, "rewards/mcq_exact_match_reward/mean": 0.171875, "rewards/mcq_exact_match_reward/std": 0.38025420904159546, "step": 13 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1177.0, "completions/mean_length": 271.390625, "completions/mean_terminated_length": 243.19049072265625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.03125, "frac_reward_zero_std": 0.125, "grad_norm": 17.6346492767334, "learning_rate": 9.826044551386742e-07, "loss": 0.0, "num_tokens": 1715944.0, "reward": 0.3101562261581421, "reward_std": 0.3299850821495056, "rewards/format_reward/mean": 0.4453125, "rewards/format_reward/std": 0.26899561285972595, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 1551.0, "completions/mean_length": 314.53125, "completions/mean_terminated_length": 287.0158996582031, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.033482142857142856, "frac_reward_zero_std": 0.375, "grad_norm": 9.082239151000977, "learning_rate": 9.780178907671788e-07, "loss": -0.0, "num_tokens": 1837930.0, "reward": 0.3453124761581421, "reward_std": 0.20426981151103973, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.17747680842876434, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 749.0, "completions/max_terminated_length": 749.0, "completions/mean_length": 70.03125, "completions/mean_terminated_length": 70.03125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.03571428571428571, "frac_reward_zero_std": 0.375, "grad_norm": 15.018674850463867, "learning_rate": 9.729086208503173e-07, "loss": -0.0, "num_tokens": 1933756.0, "reward": 0.6445311903953552, "reward_std": 0.27968671917915344, "rewards/format_reward/mean": 0.5078125, "rewards/format_reward/std": 0.1406387835741043, "rewards/mcq_exact_match_reward/mean": 0.59375, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 1400.0, "completions/max_terminated_length": 1400.0, "completions/mean_length": 109.015625, "completions/mean_terminated_length": 109.015625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.03794642857142857, "frac_reward_zero_std": 0.625, "grad_norm": 11.6762056350708, "learning_rate": 9.672822322997304e-07, "loss": 0.0, "num_tokens": 2044469.0, "reward": 0.14296874403953552, "reward_std": 0.11109752207994461, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.09375, "rewards/mcq_exact_match_reward/std": 0.29378482699394226, "step": 17 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 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": 0.04017857142857143, "frac_reward_zero_std": 0.625, "grad_norm": 14.160598754882812, "learning_rate": 9.611448774886923e-07, "loss": -0.0, "num_tokens": 2139605.0, "reward": 0.4249999523162842, "reward_std": 0.17570310831069946, "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": 18 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 721.0, "completions/max_terminated_length": 721.0, "completions/mean_length": 27.453125, "completions/mean_terminated_length": 27.453125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.04241071428571429, "frac_reward_zero_std": 0.375, "grad_norm": 23.57636833190918, "learning_rate": 9.545032675245813e-07, "loss": 0.0, "num_tokens": 2234258.0, "reward": 0.5679687261581421, "reward_std": 0.23467989265918732, "rewards/format_reward/mean": 0.5234375, "rewards/format_reward/std": 0.10652101784944534, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "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": 579.0, "completions/max_terminated_length": 579.0, "completions/mean_length": 33.765625, "completions/mean_terminated_length": 33.765625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.044642857142857144, "frac_reward_zero_std": 0.5, "grad_norm": 20.334789276123047, "learning_rate": 9.473646649103817e-07, "loss": 0.0, "num_tokens": 2350395.0, "reward": 0.28749996423721313, "reward_std": 0.17692223191261292, "rewards/format_reward/mean": 0.53125, "rewards/format_reward/std": 0.12198751419782639, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 977.0, "completions/max_terminated_length": 977.0, "completions/mean_length": 23.625, "completions/mean_terminated_length": 23.625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.046875, "frac_reward_zero_std": 0.625, "grad_norm": 10.189783096313477, "learning_rate": 9.397368756032444e-07, "loss": 0.0, "num_tokens": 2440331.0, "reward": 0.4874999523162842, "reward_std": 0.1462520956993103, "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": 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": 448.0, "completions/max_terminated_length": 448.0, "completions/mean_length": 21.625, "completions/mean_terminated_length": 21.625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.049107142857142856, "frac_reward_zero_std": 0.625, "grad_norm": 10.06070327758789, "learning_rate": 9.316282404787869e-07, "loss": -0.0, "num_tokens": 2555947.0, "reward": 0.3304687440395355, "reward_std": 0.18777307868003845, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "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": 550.0, "completions/max_terminated_length": 550.0, "completions/mean_length": 40.859375, "completions/mean_terminated_length": 40.859375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05133928571428571, "frac_reward_zero_std": 0.625, "grad_norm": 9.419411659240723, "learning_rate": 9.230476262104676e-07, "loss": -0.0, "num_tokens": 2655266.0, "reward": 0.34687498211860657, "reward_std": 0.1530819833278656, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 115.0, "completions/max_terminated_length": 115.0, "completions/mean_length": 7.703125, "completions/mean_terminated_length": 7.703125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05357142857142857, "frac_reward_zero_std": 0.375, "grad_norm": 23.631738662719727, "learning_rate": 9.1400441557401e-07, "loss": -0.0, "num_tokens": 2753767.0, "reward": 0.5187499523162842, "reward_std": 0.2961388826370239, "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": 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": 331.0, "completions/max_terminated_length": 331.0, "completions/mean_length": 21.5625, "completions/mean_terminated_length": 21.5625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05580357142857143, "frac_reward_zero_std": 0.625, "grad_norm": 32.62712478637695, "learning_rate": 9.045084971874737e-07, "loss": -0.0, "num_tokens": 2854067.0, "reward": 0.7687499523162842, "reward_std": 0.16675157845020294, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.71875, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "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": 749.0, "completions/max_terminated_length": 749.0, "completions/mean_length": 46.765625, "completions/mean_terminated_length": 46.765625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05803571428571429, "frac_reward_zero_std": 0.625, "grad_norm": 14.9358491897583, "learning_rate": 8.945702546981968e-07, "loss": -0.0, "num_tokens": 2941572.0, "reward": 0.4273437261581421, "reward_std": 0.16922587156295776, "rewards/format_reward/mean": 0.5234375, "rewards/format_reward/std": 0.13886408507823944, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 26 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 473.0, "completions/mean_length": 55.65625, "completions/mean_terminated_length": 24.031747817993164, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.060267857142857144, "frac_reward_zero_std": 0.125, "grad_norm": 17.296993255615234, "learning_rate": 8.842005554284295e-07, "loss": -0.0, "num_tokens": 3036318.0, "reward": 0.45546871423721313, "reward_std": 0.3812890648841858, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 118.0, "completions/max_terminated_length": 118.0, "completions/mean_length": 9.265625, "completions/mean_terminated_length": 9.265625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0625, "frac_reward_zero_std": 0.375, "grad_norm": 22.46074676513672, "learning_rate": 8.734107384920769e-07, "loss": 0.0, "num_tokens": 3126295.0, "reward": 0.7382811903953552, "reward_std": 0.240484818816185, "rewards/format_reward/mean": 0.5078125, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.6875, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 919.0, "completions/max_terminated_length": 919.0, "completions/mean_length": 34.875, "completions/mean_terminated_length": 34.875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06473214285714286, "frac_reward_zero_std": 0.625, "grad_norm": 26.231388092041016, "learning_rate": 8.622126023955445e-07, "loss": 0.0, "num_tokens": 3218455.0, "reward": 0.524218738079071, "reward_std": 0.06296665221452713, "rewards/format_reward/mean": 0.5546875, "rewards/format_reward/std": 0.15728822350502014, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "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": 267.0, "completions/max_terminated_length": 267.0, "completions/mean_length": 27.46875, "completions/mean_terminated_length": 27.46875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06696428571428571, "frac_reward_zero_std": 0.625, "grad_norm": 12.698409080505371, "learning_rate": 8.506183921362442e-07, "loss": -0.0, "num_tokens": 3300133.0, "reward": 0.42578125, "reward_std": 0.09059804677963257, "rewards/format_reward/mean": 0.5078125, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 295.0, "completions/max_terminated_length": 295.0, "completions/mean_length": 20.296875, "completions/mean_terminated_length": 20.296875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06919642857142858, "frac_reward_zero_std": 0.5, "grad_norm": 25.770362854003906, "learning_rate": 8.386407858128706e-07, "loss": -0.0, "num_tokens": 3392480.0, "reward": 0.6601561903953552, "reward_std": 0.2199951410293579, "rewards/format_reward/mean": 0.5078125, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.609375, "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": 58.0, "completions/max_terminated_length": 58.0, "completions/mean_length": 9.203125, "completions/mean_terminated_length": 9.203125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07142857142857142, "frac_reward_zero_std": 0.125, "grad_norm": 25.56088638305664, "learning_rate": 8.262928807620843e-07, "loss": -0.0, "num_tokens": 3487341.0, "reward": 0.4843749701976776, "reward_std": 0.35943299531936646, "rewards/format_reward/mean": 0.625, "rewards/format_reward/std": 0.2182178944349289, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 32 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1187.0, "completions/max_terminated_length": 1187.0, "completions/mean_length": 74.703125, "completions/mean_terminated_length": 74.703125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07366071428571429, "frac_reward_zero_std": 0.375, "grad_norm": 16.140928268432617, "learning_rate": 8.135881792367685e-07, "loss": -0.0, "num_tokens": 3597818.0, "reward": 0.6210936903953552, "reward_std": 0.15935926139354706, "rewards/format_reward/mean": 0.5859375, "rewards/format_reward/std": 0.19012710452079773, "rewards/mcq_exact_match_reward/mean": 0.5625, "rewards/mcq_exact_match_reward/std": 0.5, "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": 521.0, "completions/max_terminated_length": 521.0, "completions/mean_length": 78.921875, "completions/mean_terminated_length": 78.921875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07589285714285714, "frac_reward_zero_std": 0.5, "grad_norm": 36.14888381958008, "learning_rate": 8.005405736415125e-07, "loss": 0.0, "num_tokens": 3699197.0, "reward": 1.0531249046325684, "reward_std": 0.052852895110845566, "rewards/format_reward/mean": 0.6875, "rewards/format_reward/std": 0.24397502839565277, "rewards/mcq_exact_match_reward/mean": 0.984375, "rewards/mcq_exact_match_reward/std": 0.125, "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": 701.0, "completions/max_terminated_length": 701.0, "completions/mean_length": 129.09375, "completions/mean_terminated_length": 129.09375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.078125, "frac_reward_zero_std": 0.0, "grad_norm": 38.96165084838867, "learning_rate": 7.871643313414718e-07, "loss": 0.0, "num_tokens": 3785907.0, "reward": 0.581250011920929, "reward_std": 0.19127598404884338, "rewards/format_reward/mean": 0.8125, "rewards/format_reward/std": 0.24397502839565277, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "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": 1201.0, "completions/max_terminated_length": 1201.0, "completions/mean_length": 78.640625, "completions/mean_terminated_length": 78.640625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08035714285714286, "frac_reward_zero_std": 0.25, "grad_norm": 23.0406436920166, "learning_rate": 7.734740790612136e-07, "loss": 0.0, "num_tokens": 3888292.0, "reward": 0.43828126788139343, "reward_std": 0.23249360918998718, "rewards/format_reward/mean": 0.9453125, "rewards/format_reward/std": 0.180765300989151, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "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": 1375.0, "completions/max_terminated_length": 1375.0, "completions/mean_length": 107.90625, "completions/mean_terminated_length": 107.90625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08258928571428571, "frac_reward_zero_std": 0.0, "grad_norm": 29.743480682373047, "learning_rate": 7.594847868906076e-07, "loss": 0.0, "num_tokens": 3997214.0, "reward": 0.43437501788139343, "reward_std": 0.3827434182167053, "rewards/format_reward/mean": 0.90625, "rewards/format_reward/std": 0.19669894874095917, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "step": 37 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 594.0, "completions/max_terminated_length": 594.0, "completions/mean_length": 65.421875, "completions/mean_terminated_length": 65.421875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08482142857142858, "frac_reward_zero_std": 0.25, "grad_norm": 22.222890853881836, "learning_rate": 7.452117519152541e-07, "loss": -0.0, "num_tokens": 4083289.0, "reward": 0.8304687738418579, "reward_std": 0.27762293815612793, "rewards/format_reward/mean": 0.9609375, "rewards/format_reward/std": 0.16194961965084076, "rewards/mcq_exact_match_reward/mean": 0.734375, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 612.0, "completions/max_terminated_length": 612.0, "completions/mean_length": 104.859375, "completions/mean_terminated_length": 104.859375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08705357142857142, "frac_reward_zero_std": 0.25, "grad_norm": 15.55250358581543, "learning_rate": 7.306705814893439e-07, "loss": 0.0, "num_tokens": 4180168.0, "reward": 0.4671874940395355, "reward_std": 0.22475478053092957, "rewards/format_reward/mean": 0.921875, "rewards/format_reward/std": 0.20351573824882507, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 968.0, "completions/max_terminated_length": 968.0, "completions/mean_length": 113.1875, "completions/mean_terminated_length": 113.1875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08928571428571429, "frac_reward_zero_std": 0.375, "grad_norm": 5.488610744476318, "learning_rate": 7.158771761692464e-07, "loss": 0.0, "num_tokens": 4310372.0, "reward": 0.48359376192092896, "reward_std": 0.19265598058700562, "rewards/format_reward/mean": 0.9296875, "rewards/format_reward/std": 0.1751912236213684, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 1017.0, "completions/max_terminated_length": 1017.0, "completions/mean_length": 122.84375, "completions/mean_terminated_length": 122.84375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.09151785714285714, "frac_reward_zero_std": 0.375, "grad_norm": 7.906015396118164, "learning_rate": 7.008477123264847e-07, "loss": -0.0, "num_tokens": 4400802.0, "reward": 0.34921878576278687, "reward_std": 0.17898640036582947, "rewards/format_reward/mean": 0.9921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 104.0, "completions/mean_length": 47.65625, "completions/mean_terminated_length": 15.904763221740723, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.09375, "frac_reward_zero_std": 0.5, "grad_norm": 18.783111572265625, "learning_rate": 6.855986244591103e-07, "loss": 0.0, "num_tokens": 4490844.0, "reward": 0.8484375476837158, "reward_std": 0.15067152678966522, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.75, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 42 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 386.0, "completions/max_terminated_length": 386.0, "completions/mean_length": 44.71875, "completions/mean_terminated_length": 44.71875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.09598214285714286, "frac_reward_zero_std": 0.75, "grad_norm": 10.034747123718262, "learning_rate": 6.701465872208216e-07, "loss": -0.0, "num_tokens": 4581682.0, "reward": 0.6312500238418579, "reward_std": 0.0883883461356163, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 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": 169.0, "completions/max_terminated_length": 169.0, "completions/mean_length": 28.828125, "completions/mean_terminated_length": 28.828125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.09821428571428571, "frac_reward_zero_std": 0.625, "grad_norm": 14.294143676757812, "learning_rate": 6.545084971874736e-07, "loss": -0.0, "num_tokens": 4687551.0, "reward": 0.4593750238418579, "reward_std": 0.1530819982290268, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 138.0, "completions/max_terminated_length": 138.0, "completions/mean_length": 20.34375, "completions/mean_terminated_length": 20.34375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.10044642857142858, "frac_reward_zero_std": 0.5, "grad_norm": 19.66748046875, "learning_rate": 6.387014543809223e-07, "loss": 0.0, "num_tokens": 4780245.0, "reward": 0.5515625476837158, "reward_std": 0.21536517143249512, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.453125, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 288.0, "completions/mean_length": 63.03125, "completions/mean_terminated_length": 31.52381134033203, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.10267857142857142, "frac_reward_zero_std": 0.75, "grad_norm": 8.040264129638672, "learning_rate": 6.227427435703995e-07, "loss": 0.0, "num_tokens": 4885047.0, "reward": 0.7250000238418579, "reward_std": 0.0883883461356163, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.625, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 560.0, "completions/mean_length": 54.53125, "completions/mean_terminated_length": 22.88888931274414, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.10491071428571429, "frac_reward_zero_std": 0.75, "grad_norm": 4.303212642669678, "learning_rate": 6.066498153718734e-07, "loss": 0.0, "num_tokens": 4981921.0, "reward": 0.6140625476837158, "reward_std": 0.04861358925700188, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "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": 423.0, "completions/max_terminated_length": 423.0, "completions/mean_length": 36.078125, "completions/mean_terminated_length": 36.078125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.10714285714285714, "frac_reward_zero_std": 0.5, "grad_norm": 14.959633827209473, "learning_rate": 5.90440267166055e-07, "loss": -0.0, "num_tokens": 5097710.0, "reward": 0.7867187261581421, "reward_std": 0.14846181869506836, "rewards/format_reward/mean": 0.9921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.6875, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 1036.0, "completions/max_terminated_length": 1036.0, "completions/mean_length": 39.046875, "completions/mean_terminated_length": 39.046875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.109375, "frac_reward_zero_std": 0.625, "grad_norm": 13.03551197052002, "learning_rate": 5.741318238559209e-07, "loss": 0.0, "num_tokens": 5197737.0, "reward": 0.6468750238418579, "reward_std": 0.15992169082164764, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.546875, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 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": 240.0, "completions/max_terminated_length": 240.0, "completions/mean_length": 22.109375, "completions/mean_terminated_length": 22.109375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.11160714285714286, "frac_reward_zero_std": 0.625, "grad_norm": 7.772871017456055, "learning_rate": 5.577423184847931e-07, "loss": 0.0, "num_tokens": 5273552.0, "reward": 0.5062500238418579, "reward_std": 0.1462521106004715, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 50 } ], "logging_steps": 1, "max_steps": 100, "num_input_tokens_seen": 5273552, "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 }