{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.125, "eval_steps": 500, "global_step": 50, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1998.0, "completions/mean_length": 617.0625, "completions/mean_terminated_length": 570.9031982421875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0025, "frac_reward_zero_std": 0.0, "grad_norm": 6.5873212814331055, "learning_rate": 0.0, "loss": 0.0, "num_tokens": 163788.0, "reward": 0.29218748211860657, "reward_std": 0.3750533163547516, "rewards/format_reward/mean": 0.421875, "rewards/format_reward/std": 0.23935678601264954, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 1 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1681.0, "completions/mean_length": 615.75, "completions/mean_terminated_length": 569.54833984375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.005, "frac_reward_zero_std": 0.0, "grad_norm": 16.994749069213867, "learning_rate": 2e-07, "loss": -0.0, "num_tokens": 291028.0, "reward": 0.24062499403953552, "reward_std": 0.3783254623413086, "rewards/format_reward/mean": 0.375, "rewards/format_reward/std": 0.26726123690605164, "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": 0.0, "completions/max_length": 1544.0, "completions/max_terminated_length": 1544.0, "completions/mean_length": 542.484375, "completions/mean_terminated_length": 542.484375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0075, "frac_reward_zero_std": 0.25, "grad_norm": 5.403246879577637, "learning_rate": 4e-07, "loss": -0.0, "num_tokens": 425931.0, "reward": 0.22968749701976776, "reward_std": 0.24118764698505402, "rewards/format_reward/mean": 0.421875, "rewards/format_reward/std": 0.20351573824882507, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "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": 1915.0, "completions/mean_length": 549.96875, "completions/mean_terminated_length": 526.1904907226562, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.01, "frac_reward_zero_std": 0.125, "grad_norm": 9.253963470458984, "learning_rate": 6e-07, "loss": 0.0, "num_tokens": 557993.0, "reward": 0.13515624403953552, "reward_std": 0.17939597368240356, "rewards/format_reward/mean": 0.4140625, "rewards/format_reward/std": 0.19012710452079773, "rewards/mcq_exact_match_reward/mean": 0.09375, "rewards/mcq_exact_match_reward/std": 0.29378482699394226, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1963.0, "completions/mean_length": 519.921875, "completions/mean_terminated_length": 495.66668701171875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0125, "frac_reward_zero_std": 0.25, "grad_norm": 1.3908439874649048, "learning_rate": 8e-07, "loss": 0.0, "num_tokens": 690364.0, "reward": 0.19218748807907104, "reward_std": 0.29305610060691833, "rewards/format_reward/mean": 0.359375, "rewards/format_reward/std": 0.24346621334552765, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "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": 1678.0, "completions/max_terminated_length": 1678.0, "completions/mean_length": 333.984375, "completions/mean_terminated_length": 333.984375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.015, "frac_reward_zero_std": 0.0, "grad_norm": 6.072598934173584, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 819347.0, "reward": 0.33828121423721313, "reward_std": 0.4058018624782562, "rewards/format_reward/mean": 0.4140625, "rewards/format_reward/std": 0.2280818521976471, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "step": 6 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1771.0, "completions/max_terminated_length": 1771.0, "completions/mean_length": 354.0, "completions/mean_terminated_length": 354.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0175, "frac_reward_zero_std": 0.125, "grad_norm": 16.699325561523438, "learning_rate": 9.99726628670463e-07, "loss": -0.0, "num_tokens": 925811.0, "reward": 0.1914062350988388, "reward_std": 0.3238601088523865, "rewards/format_reward/mean": 0.3515625, "rewards/format_reward/std": 0.24688033759593964, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "step": 7 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1472.0, "completions/mean_length": 520.828125, "completions/mean_terminated_length": 471.56451416015625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.02, "frac_reward_zero_std": 0.0, "grad_norm": 12.57072925567627, "learning_rate": 9.989068136093872e-07, "loss": -0.0, "num_tokens": 1048440.0, "reward": 0.23906247317790985, "reward_std": 0.31088316440582275, "rewards/format_reward/mean": 0.359375, "rewards/format_reward/std": 0.2592533528804779, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "step": 8 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1214.0, "completions/max_terminated_length": 1214.0, "completions/mean_length": 281.671875, "completions/mean_terminated_length": 281.671875, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.0225, "frac_reward_zero_std": 0.125, "grad_norm": 17.275171279907227, "learning_rate": 9.975414512725056e-07, "loss": 0.0, "num_tokens": 1141299.0, "reward": 0.25624996423721313, "reward_std": 0.3120592534542084, "rewards/format_reward/mean": 0.375, "rewards/format_reward/std": 0.28171807527542114, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "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": 1379.0, "completions/mean_length": 302.296875, "completions/mean_terminated_length": 274.5873107910156, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.025, "frac_reward_zero_std": 0.25, "grad_norm": 10.201237678527832, "learning_rate": 9.956320346634875e-07, "loss": -0.0, "num_tokens": 1243270.0, "reward": 0.3882812261581421, "reward_std": 0.37965649366378784, "rewards/format_reward/mean": 0.4453125, "rewards/format_reward/std": 0.2203386276960373, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "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": 865.0, "completions/max_terminated_length": 865.0, "completions/mean_length": 214.390625, "completions/mean_terminated_length": 214.390625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0275, "frac_reward_zero_std": 0.25, "grad_norm": 19.455276489257812, "learning_rate": 9.931806517013612e-07, "loss": 0.0, "num_tokens": 1346943.0, "reward": 0.5625, "reward_std": 0.31731271743774414, "rewards/format_reward/mean": 0.46875, "rewards/format_reward/std": 0.12198751419782639, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "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": 1532.0, "completions/max_terminated_length": 1532.0, "completions/mean_length": 236.640625, "completions/mean_terminated_length": 236.640625, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.03, "frac_reward_zero_std": 0.25, "grad_norm": 56.06517028808594, "learning_rate": 9.901899829374047e-07, "loss": 0.0, "num_tokens": 1470600.0, "reward": 0.24453124403953552, "reward_std": 0.23627185821533203, "rewards/format_reward/mean": 0.4140625, "rewards/format_reward/std": 0.19012710452079773, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "step": 12 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1141.0, "completions/max_terminated_length": 1141.0, "completions/mean_length": 160.859375, "completions/mean_terminated_length": 160.859375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0325, "frac_reward_zero_std": 0.25, "grad_norm": 30.387075424194336, "learning_rate": 9.866632986240029e-07, "loss": -0.0, "num_tokens": 1572047.0, "reward": 0.3179687261581421, "reward_std": 0.29590028524398804, "rewards/format_reward/mean": 0.5234375, "rewards/format_reward/std": 0.22589658200740814, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 1285.0, "completions/mean_length": 340.390625, "completions/mean_terminated_length": 313.2857360839844, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.035, "frac_reward_zero_std": 0.25, "grad_norm": 9.51353931427002, "learning_rate": 9.826044551386742e-07, "loss": 0.0, "num_tokens": 1691744.0, "reward": 0.32890623807907104, "reward_std": 0.1864037811756134, "rewards/format_reward/mean": 0.4765625, "rewards/format_reward/std": 0.24282869696617126, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 14 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 825.0, "completions/max_terminated_length": 825.0, "completions/mean_length": 168.75, "completions/mean_terminated_length": 168.75, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0375, "frac_reward_zero_std": 0.25, "grad_norm": 15.794662475585938, "learning_rate": 9.780178907671788e-07, "loss": -0.0, "num_tokens": 1789216.0, "reward": 0.42656248807907104, "reward_std": 0.1613350808620453, "rewards/format_reward/mean": 0.515625, "rewards/format_reward/std": 0.26679685711860657, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1444.0, "completions/mean_length": 104.59375, "completions/mean_terminated_length": 73.74603271484375, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.04, "frac_reward_zero_std": 0.25, "grad_norm": 21.912553787231445, "learning_rate": 9.729086208503173e-07, "loss": -0.0, "num_tokens": 1873166.0, "reward": 0.66015625, "reward_std": 0.28690463304519653, "rewards/format_reward/mean": 0.5078125, "rewards/format_reward/std": 0.20877929031848907, "rewards/mcq_exact_match_reward/mean": 0.609375, "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": 854.0, "completions/max_terminated_length": 854.0, "completions/mean_length": 56.171875, "completions/mean_terminated_length": 56.171875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0425, "frac_reward_zero_std": 0.75, "grad_norm": 6.873574256896973, "learning_rate": 9.672822322997304e-07, "loss": 0.0, "num_tokens": 1985401.0, "reward": 0.09999999403953552, "reward_std": 0.10215428471565247, "rewards/format_reward/mean": 0.53125, "rewards/format_reward/std": 0.12198751419782639, "rewards/mcq_exact_match_reward/mean": 0.046875, "rewards/mcq_exact_match_reward/std": 0.21304203569889069, "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": 22.0, "completions/max_terminated_length": 22.0, "completions/mean_length": 6.25, "completions/mean_terminated_length": 6.25, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.045, "frac_reward_zero_std": 0.5, "grad_norm": 18.71595573425293, "learning_rate": 9.611448774886923e-07, "loss": -0.0, "num_tokens": 2092457.0, "reward": 0.6585937142372131, "reward_std": 0.22210699319839478, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.609375, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 554.0, "completions/max_terminated_length": 554.0, "completions/mean_length": 48.65625, "completions/mean_terminated_length": 48.65625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0475, "frac_reward_zero_std": 0.625, "grad_norm": 18.441650390625, "learning_rate": 9.545032675245813e-07, "loss": -0.0, "num_tokens": 2192075.0, "reward": 0.11328125, "reward_std": 0.14858190715312958, "rewards/format_reward/mean": 0.5078125, "rewards/format_reward/std": 0.10882443934679031, "rewards/mcq_exact_match_reward/mean": 0.0625, "rewards/mcq_exact_match_reward/std": 0.24397502839565277, "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": 142.0, "completions/max_terminated_length": 142.0, "completions/mean_length": 8.234375, "completions/mean_terminated_length": 8.234375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05, "frac_reward_zero_std": 0.625, "grad_norm": 25.329486846923828, "learning_rate": 9.473646649103817e-07, "loss": -0.0, "num_tokens": 2297002.0, "reward": 0.5507811903953552, "reward_std": 0.17607249319553375, "rewards/format_reward/mean": 0.5078125, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "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": 191.0, "completions/max_terminated_length": 191.0, "completions/mean_length": 10.46875, "completions/mean_terminated_length": 10.46875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0525, "frac_reward_zero_std": 0.75, "grad_norm": 13.107504844665527, "learning_rate": 9.397368756032444e-07, "loss": 0.0, "num_tokens": 2392752.0, "reward": 0.33046871423721313, "reward_std": 0.0887577086687088, "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": 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": 349.0, "completions/max_terminated_length": 349.0, "completions/mean_length": 17.8125, "completions/mean_terminated_length": 17.8125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.055, "frac_reward_zero_std": 0.625, "grad_norm": 26.923320770263672, "learning_rate": 9.316282404787869e-07, "loss": -0.0, "num_tokens": 2481588.0, "reward": 0.5453124642372131, "reward_std": 0.10573774576187134, "rewards/format_reward/mean": 0.609375, "rewards/format_reward/std": 0.2083333432674408, "rewards/mcq_exact_match_reward/mean": 0.484375, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "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": 990.0, "completions/max_terminated_length": 990.0, "completions/mean_length": 56.515625, "completions/mean_terminated_length": 56.515625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0575, "frac_reward_zero_std": 0.25, "grad_norm": 21.975736618041992, "learning_rate": 9.230476262104676e-07, "loss": 0.0, "num_tokens": 2581757.0, "reward": 0.47343748807907104, "reward_std": 0.2496831715106964, "rewards/format_reward/mean": 0.515625, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 345.0, "completions/max_terminated_length": 345.0, "completions/mean_length": 24.53125, "completions/mean_terminated_length": 24.53125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06, "frac_reward_zero_std": 0.5, "grad_norm": 19.507545471191406, "learning_rate": 9.1400441557401e-07, "loss": -0.0, "num_tokens": 2674911.0, "reward": 0.40234375, "reward_std": 0.23218470811843872, "rewards/format_reward/mean": 0.5859375, "rewards/format_reward/std": 0.19012710452079773, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "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": 747.0, "completions/max_terminated_length": 747.0, "completions/mean_length": 49.328125, "completions/mean_terminated_length": 49.328125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0625, "frac_reward_zero_std": 0.5, "grad_norm": 14.060385704040527, "learning_rate": 9.045084971874737e-07, "loss": -0.0, "num_tokens": 2776588.0, "reward": 0.24453124403953552, "reward_std": 0.11687872558832169, "rewards/format_reward/mean": 0.5703125, "rewards/format_reward/std": 0.19654127955436707, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 7.640625, "completions/mean_terminated_length": 7.640625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.065, "frac_reward_zero_std": 0.625, "grad_norm": 24.995521545410156, "learning_rate": 8.945702546981968e-07, "loss": 0.0, "num_tokens": 2853725.0, "reward": 0.125, "reward_std": 0.1462520956993103, "rewards/format_reward/mean": 0.625, "rewards/format_reward/std": 0.2182178944349289, "rewards/mcq_exact_match_reward/mean": 0.0625, "rewards/mcq_exact_match_reward/std": 0.24397502839565277, "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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 6.859375, "completions/mean_terminated_length": 6.859375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0675, "frac_reward_zero_std": 0.875, "grad_norm": 12.034884452819824, "learning_rate": 8.842005554284295e-07, "loss": -0.0, "num_tokens": 2958020.0, "reward": 0.5874999761581421, "reward_std": 0.0578637532889843, "rewards/format_reward/mean": 0.5625, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "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": 432.0, "completions/max_terminated_length": 432.0, "completions/mean_length": 23.96875, "completions/mean_terminated_length": 23.96875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07, "frac_reward_zero_std": 0.625, "grad_norm": 16.252504348754883, "learning_rate": 8.734107384920769e-07, "loss": 0.0, "num_tokens": 3058178.0, "reward": 0.653124988079071, "reward_std": 0.17567718029022217, "rewards/format_reward/mean": 0.59375, "rewards/format_reward/std": 0.21593283116817474, "rewards/mcq_exact_match_reward/mean": 0.59375, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 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": 0.0725, "frac_reward_zero_std": 0.5, "grad_norm": 19.131885528564453, "learning_rate": 8.622126023955445e-07, "loss": -0.0, "num_tokens": 3165074.0, "reward": 0.3937499523162842, "reward_std": 0.24251841008663177, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "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": 327.0, "completions/max_terminated_length": 327.0, "completions/mean_length": 29.59375, "completions/mean_terminated_length": 29.59375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.075, "frac_reward_zero_std": 0.375, "grad_norm": 12.563758850097656, "learning_rate": 8.506183921362442e-07, "loss": -0.0, "num_tokens": 3254080.0, "reward": 0.2601562440395355, "reward_std": 0.2749083638191223, "rewards/format_reward/mean": 0.7265625, "rewards/format_reward/std": 0.250866562128067, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 6.859375, "completions/mean_terminated_length": 6.859375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0775, "frac_reward_zero_std": 0.875, "grad_norm": 7.785823822021484, "learning_rate": 8.386407858128706e-07, "loss": -0.0, "num_tokens": 3320999.0, "reward": 0.4937499761581421, "reward_std": 0.06681530177593231, "rewards/format_reward/mean": 0.5625, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.4375, "rewards/mcq_exact_match_reward/std": 0.5, "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": 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": 0.08, "frac_reward_zero_std": 0.625, "grad_norm": 15.087549209594727, "learning_rate": 8.262928807620843e-07, "loss": -0.0, "num_tokens": 3432231.0, "reward": 0.3624999523162842, "reward_std": 0.1872510462999344, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 342.0, "completions/max_terminated_length": 342.0, "completions/mean_length": 13.96875, "completions/mean_terminated_length": 13.96875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0825, "frac_reward_zero_std": 0.5, "grad_norm": 16.296710968017578, "learning_rate": 8.135881792367685e-07, "loss": 0.0, "num_tokens": 3512573.0, "reward": 0.28125, "reward_std": 0.24511480331420898, "rewards/format_reward/mean": 0.625, "rewards/format_reward/std": 0.2182178944349289, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "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": 100.0, "completions/max_terminated_length": 100.0, "completions/mean_length": 8.21875, "completions/mean_terminated_length": 8.21875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.085, "frac_reward_zero_std": 0.875, "grad_norm": 15.109249114990234, "learning_rate": 8.005405736415125e-07, "loss": -0.0, "num_tokens": 3601043.0, "reward": 0.6187499761581421, "reward_std": 0.06681530177593231, "rewards/format_reward/mean": 0.5625, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.5625, "rewards/mcq_exact_match_reward/std": 0.5, "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": 268.0, "completions/max_terminated_length": 268.0, "completions/mean_length": 17.671875, "completions/mean_terminated_length": 17.671875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0875, "frac_reward_zero_std": 0.625, "grad_norm": 18.687572479248047, "learning_rate": 7.871643313414718e-07, "loss": 0.0, "num_tokens": 3685510.0, "reward": 0.5249999761581421, "reward_std": 0.1462520956993103, "rewards/format_reward/mean": 0.5625, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 35 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 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": 1.0, "grad_norm": 0.0, "learning_rate": 7.734740790612136e-07, "loss": 0.0, "num_tokens": 3793934.0, "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": 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": 21.0, "completions/max_terminated_length": 21.0, "completions/mean_length": 6.234375, "completions/mean_terminated_length": 6.234375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0925, "frac_reward_zero_std": 0.125, "grad_norm": 36.460479736328125, "learning_rate": 7.594847868906076e-07, "loss": -0.0, "num_tokens": 3908741.0, "reward": 0.5648437142372131, "reward_std": 0.38751497864723206, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "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": 881.0, "completions/max_terminated_length": 881.0, "completions/mean_length": 23.515625, "completions/mean_terminated_length": 23.515625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.095, "frac_reward_zero_std": 0.75, "grad_norm": 8.36151123046875, "learning_rate": 7.452117519152541e-07, "loss": -0.0, "num_tokens": 3993790.0, "reward": 0.3499999940395355, "reward_std": 0.10888782143592834, "rewards/format_reward/mean": 0.6875, "rewards/format_reward/std": 0.24397502839565277, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 38 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 6.828125, "completions/mean_terminated_length": 6.828125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0975, "frac_reward_zero_std": 0.5, "grad_norm": 15.606691360473633, "learning_rate": 7.306705814893439e-07, "loss": -0.0, "num_tokens": 4080659.0, "reward": 0.5874999761581421, "reward_std": 0.22201895713806152, "rewards/format_reward/mean": 0.5625, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "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": 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": 0.1, "frac_reward_zero_std": 0.875, "grad_norm": 12.567124366760254, "learning_rate": 7.158771761692464e-07, "loss": -0.0, "num_tokens": 4160227.0, "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": 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": 451.0, "completions/max_terminated_length": 451.0, "completions/mean_length": 12.953125, "completions/mean_terminated_length": 12.953125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1025, "frac_reward_zero_std": 0.5, "grad_norm": 57.7166633605957, "learning_rate": 7.008477123264847e-07, "loss": -0.0, "num_tokens": 4269648.0, "reward": 0.6593749523162842, "reward_std": 0.2109457552433014, "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": 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": 0.875, "grad_norm": 17.067180633544922, "learning_rate": 6.855986244591103e-07, "loss": 0.0, "num_tokens": 4395936.0, "reward": 0.31562498211860657, "reward_std": 0.04419416934251785, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 6.40625, "completions/mean_terminated_length": 6.40625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1075, "frac_reward_zero_std": 0.5, "grad_norm": 30.13214874267578, "learning_rate": 6.701465872208216e-07, "loss": 0.0, "num_tokens": 4467898.0, "reward": 0.45390623807907104, "reward_std": 0.09393882006406784, "rewards/format_reward/mean": 0.4765625, "rewards/format_reward/std": 0.13886408507823944, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 43 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 120.0, "completions/max_terminated_length": 120.0, "completions/mean_length": 8.53125, "completions/mean_terminated_length": 8.53125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.11, "frac_reward_zero_std": 0.625, "grad_norm": 15.19147777557373, "learning_rate": 6.545084971874736e-07, "loss": -0.0, "num_tokens": 4568684.0, "reward": 0.6499999761581421, "reward_std": 0.16675157845020294, "rewards/format_reward/mean": 0.5625, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.59375, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 135.0, "completions/max_terminated_length": 135.0, "completions/mean_length": 8.796875, "completions/mean_terminated_length": 8.796875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1125, "frac_reward_zero_std": 0.625, "grad_norm": 14.918896675109863, "learning_rate": 6.387014543809223e-07, "loss": 0.0, "num_tokens": 4672255.0, "reward": 0.5093749761581421, "reward_std": 0.06911306828260422, "rewards/format_reward/mean": 0.5625, "rewards/format_reward/std": 0.18898223340511322, "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": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 7.515625, "completions/mean_terminated_length": 7.515625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.115, "frac_reward_zero_std": 0.5, "grad_norm": 20.45856475830078, "learning_rate": 6.227427435703995e-07, "loss": 0.0, "num_tokens": 4758496.0, "reward": 0.484375, "reward_std": 0.21778543293476105, "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": 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": 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": 0.1175, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 6.066498153718734e-07, "loss": 0.0, "num_tokens": 4822192.0, "reward": 0.42499998211860657, "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": 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": 0.75, "grad_norm": 12.950883865356445, "learning_rate": 5.90440267166055e-07, "loss": -0.0, "num_tokens": 4919512.0, "reward": 0.7218749523162842, "reward_std": 0.12255740165710449, "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": 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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 6.953125, "completions/mean_terminated_length": 6.953125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1225, "frac_reward_zero_std": 0.5, "grad_norm": 28.649272918701172, "learning_rate": 5.741318238559209e-07, "loss": 0.0, "num_tokens": 4991973.0, "reward": 0.6359374523162842, "reward_std": 0.10647735744714737, "rewards/format_reward/mean": 0.578125, "rewards/format_reward/std": 0.18298126757144928, "rewards/mcq_exact_match_reward/mean": 0.578125, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 83.0, "completions/max_terminated_length": 83.0, "completions/mean_length": 8.984375, "completions/mean_terminated_length": 8.984375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.125, "frac_reward_zero_std": 0.375, "grad_norm": 26.786386489868164, "learning_rate": 5.577423184847931e-07, "loss": 0.0, "num_tokens": 5084700.0, "reward": 0.5796874761581421, "reward_std": 0.19239477813243866, "rewards/format_reward/mean": 0.640625, "rewards/format_reward/std": 0.22658175230026245, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "step": 50 } ], "logging_steps": 1, "max_steps": 100, "num_input_tokens_seen": 5084700, "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 }