{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.125, "eval_steps": 500, "global_step": 50, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1467.0, "completions/max_terminated_length": 1467.0, "completions/mean_length": 340.46875, "completions/mean_terminated_length": 340.46875, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.0025, "frac_reward_zero_std": 0.0, "grad_norm": 24.121065139770508, "learning_rate": 0.0, "loss": -0.0, "num_tokens": 104782.0, "reward": 0.2320312261581421, "reward_std": 0.2813003361225128, "rewards/format_reward/mean": 0.2890625, "rewards/format_reward/std": 0.2643453776836395, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "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": 1311.0, "completions/max_terminated_length": 1311.0, "completions/mean_length": 349.15625, "completions/mean_terminated_length": 349.15625, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.005, "frac_reward_zero_std": 0.0, "grad_norm": 6.631368637084961, "learning_rate": 2e-07, "loss": -0.0, "num_tokens": 199528.0, "reward": 0.34218746423721313, "reward_std": 0.4091617465019226, "rewards/format_reward/mean": 0.296875, "rewards/format_reward/std": 0.2630521357059479, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "step": 2 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1933.0, "completions/max_terminated_length": 1933.0, "completions/mean_length": 378.484375, "completions/mean_terminated_length": 378.484375, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.0075, "frac_reward_zero_std": 0.0, "grad_norm": 22.637144088745117, "learning_rate": 4e-07, "loss": 0.0, "num_tokens": 342431.0, "reward": 0.43593746423721313, "reward_std": 0.44063660502433777, "rewards/format_reward/mean": 0.296875, "rewards/format_reward/std": 0.2777281701564789, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 0.0, "completions/max_length": 1120.0, "completions/max_terminated_length": 1120.0, "completions/mean_length": 296.15625, "completions/mean_terminated_length": 296.15625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.01, "frac_reward_zero_std": 0.125, "grad_norm": 5.168887615203857, "learning_rate": 6e-07, "loss": 0.0, "num_tokens": 430457.0, "reward": 0.2671875059604645, "reward_std": 0.261038213968277, "rewards/format_reward/mean": 0.328125, "rewards/format_reward/std": 0.2847827076911926, "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": 0.0, "completions/max_length": 1408.0, "completions/max_terminated_length": 1408.0, "completions/mean_length": 373.171875, "completions/mean_terminated_length": 373.171875, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.0125, "frac_reward_zero_std": 0.0, "grad_norm": 22.746238708496094, "learning_rate": 8e-07, "loss": -0.0, "num_tokens": 552004.0, "reward": 0.4624999761581421, "reward_std": 0.34055960178375244, "rewards/format_reward/mean": 0.40625, "rewards/format_reward/std": 0.233588308095932, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 2036.0, "completions/mean_length": 423.859375, "completions/mean_terminated_length": 398.0793762207031, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.015, "frac_reward_zero_std": 0.125, "grad_norm": 5.042137145996094, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 679451.0, "reward": 0.4781249761581421, "reward_std": 0.29430097341537476, "rewards/format_reward/mean": 0.40625, "rewards/format_reward/std": 0.233588308095932, "rewards/mcq_exact_match_reward/mean": 0.4375, "rewards/mcq_exact_match_reward/std": 0.5, "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": 0.0625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1730.0, "completions/mean_length": 719.296875, "completions/mean_terminated_length": 630.7166748046875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0175, "frac_reward_zero_std": 0.0, "grad_norm": 2.657170295715332, "learning_rate": 9.99726628670463e-07, "loss": -0.0, "num_tokens": 842902.0, "reward": 0.25312498211860657, "reward_std": 0.3533875346183777, "rewards/format_reward/mean": 0.34375, "rewards/format_reward/std": 0.233588308095932, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "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": 1097.0, "completions/mean_length": 528.5625, "completions/mean_terminated_length": 504.4444885253906, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.02, "frac_reward_zero_std": 0.0, "grad_norm": 7.923519611358643, "learning_rate": 9.989068136093872e-07, "loss": 0.0, "num_tokens": 967962.0, "reward": 0.24453124403953552, "reward_std": 0.24999795854091644, "rewards/format_reward/mean": 0.4140625, "rewards/format_reward/std": 0.2605654299259186, "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": 1769.0, "completions/max_terminated_length": 1769.0, "completions/mean_length": 430.078125, "completions/mean_terminated_length": 430.078125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0225, "frac_reward_zero_std": 0.0, "grad_norm": 9.529799461364746, "learning_rate": 9.975414512725056e-07, "loss": 0.0, "num_tokens": 1090047.0, "reward": 0.24296873807907104, "reward_std": 0.24107582867145538, "rewards/format_reward/mean": 0.3984375, "rewards/format_reward/std": 0.2387082874774933, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "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": 1396.0, "completions/max_terminated_length": 1396.0, "completions/mean_length": 507.59375, "completions/mean_terminated_length": 507.59375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.025, "frac_reward_zero_std": 0.125, "grad_norm": 3.837048053741455, "learning_rate": 9.956320346634875e-07, "loss": -0.0, "num_tokens": 1221797.0, "reward": 0.5843749642372131, "reward_std": 0.3462211489677429, "rewards/format_reward/mean": 0.375, "rewards/format_reward/std": 0.2357022762298584, "rewards/mcq_exact_match_reward/mean": 0.546875, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 0.078125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1354.0, "completions/mean_length": 652.03125, "completions/mean_terminated_length": 533.7288208007812, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0275, "frac_reward_zero_std": 0.0, "grad_norm": 3.759780168533325, "learning_rate": 9.931806517013612e-07, "loss": -0.0, "num_tokens": 1362423.0, "reward": 0.4781249463558197, "reward_std": 0.4550410211086273, "rewards/format_reward/mean": 0.40625, "rewards/format_reward/std": 0.25, "rewards/mcq_exact_match_reward/mean": 0.4375, "rewards/mcq_exact_match_reward/std": 0.5, "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": 1516.0, "completions/max_terminated_length": 1516.0, "completions/mean_length": 391.5625, "completions/mean_terminated_length": 391.5625, "completions/min_length": 5.0, "completions/min_terminated_length": 5.0, "epoch": 0.03, "frac_reward_zero_std": 0.0, "grad_norm": 7.199992656707764, "learning_rate": 9.901899829374047e-07, "loss": -0.0, "num_tokens": 1478563.0, "reward": 0.2851562201976776, "reward_std": 0.37806522846221924, "rewards/format_reward/mean": 0.3515625, "rewards/format_reward/std": 0.26246222853660583, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 1248.0, "completions/max_terminated_length": 1248.0, "completions/mean_length": 402.9375, "completions/mean_terminated_length": 402.9375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0325, "frac_reward_zero_std": 0.0, "grad_norm": 8.057401657104492, "learning_rate": 9.866632986240029e-07, "loss": 0.0, "num_tokens": 1587399.0, "reward": 0.3820312023162842, "reward_std": 0.36425960063934326, "rewards/format_reward/mean": 0.3828125, "rewards/format_reward/std": 0.2313210517168045, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "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": 0.0625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1439.0, "completions/mean_length": 581.5, "completions/mean_terminated_length": 483.7333679199219, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.035, "frac_reward_zero_std": 0.0, "grad_norm": 1.4237709045410156, "learning_rate": 9.826044551386742e-07, "loss": -0.0, "num_tokens": 1717319.0, "reward": 0.4156249761581421, "reward_std": 0.37006694078445435, "rewards/format_reward/mean": 0.40625, "rewards/format_reward/std": 0.19669894874095917, "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": 0.0, "completions/max_length": 1978.0, "completions/max_terminated_length": 1978.0, "completions/mean_length": 451.609375, "completions/mean_terminated_length": 451.609375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0375, "frac_reward_zero_std": 0.0, "grad_norm": 5.386332035064697, "learning_rate": 9.780178907671788e-07, "loss": -0.0, "num_tokens": 1822350.0, "reward": 0.23671874403953552, "reward_std": 0.3205496072769165, "rewards/format_reward/mean": 0.3359375, "rewards/format_reward/std": 0.2824873626232147, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "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": 1391.0, "completions/mean_length": 554.8125, "completions/mean_terminated_length": 531.1111450195312, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.04, "frac_reward_zero_std": 0.0, "grad_norm": 6.627169609069824, "learning_rate": 9.729086208503173e-07, "loss": 0.0, "num_tokens": 1938426.0, "reward": 0.43437498807907104, "reward_std": 0.36538147926330566, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.1666666716337204, "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": 1754.0, "completions/max_terminated_length": 1754.0, "completions/mean_length": 504.859375, "completions/mean_terminated_length": 504.859375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0425, "frac_reward_zero_std": 0.125, "grad_norm": 10.752079963684082, "learning_rate": 9.672822322997304e-07, "loss": 0.0, "num_tokens": 2077889.0, "reward": 0.3359375, "reward_std": 0.29728609323501587, "rewards/format_reward/mean": 0.390625, "rewards/format_reward/std": 0.2083333432674408, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1745.0, "completions/mean_length": 477.875, "completions/mean_terminated_length": 452.952392578125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.045, "frac_reward_zero_std": 0.125, "grad_norm": 15.702384948730469, "learning_rate": 9.611448774886923e-07, "loss": 0.0, "num_tokens": 2201481.0, "reward": 0.39296871423721313, "reward_std": 0.25133559107780457, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.22699186205863953, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "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": 1560.0, "completions/mean_length": 479.328125, "completions/mean_terminated_length": 454.4285888671875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0475, "frac_reward_zero_std": 0.125, "grad_norm": 8.670381546020508, "learning_rate": 9.545032675245813e-07, "loss": 0.0, "num_tokens": 2328654.0, "reward": 0.26484373211860657, "reward_std": 0.2534114122390747, "rewards/format_reward/mean": 0.4609375, "rewards/format_reward/std": 0.18483558297157288, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "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": 1680.0, "completions/max_terminated_length": 1680.0, "completions/mean_length": 339.71875, "completions/mean_terminated_length": 339.71875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05, "frac_reward_zero_std": 0.125, "grad_norm": 7.443161487579346, "learning_rate": 9.473646649103817e-07, "loss": 0.0, "num_tokens": 2444524.0, "reward": 0.5249999761581421, "reward_std": 0.30251359939575195, "rewards/format_reward/mean": 0.5625, "rewards/format_reward/std": 0.22712838649749756, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1120.0, "completions/mean_length": 241.296875, "completions/mean_terminated_length": 212.6190643310547, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0525, "frac_reward_zero_std": 0.5, "grad_norm": 11.614701271057129, "learning_rate": 9.397368756032444e-07, "loss": -0.0, "num_tokens": 2531023.0, "reward": 0.53125, "reward_std": 0.2136937975883484, "rewards/format_reward/mean": 0.46875, "rewards/format_reward/std": 0.17536810040473938, "rewards/mcq_exact_match_reward/mean": 0.484375, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "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": 1872.0, "completions/max_terminated_length": 1872.0, "completions/mean_length": 150.1875, "completions/mean_terminated_length": 150.1875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.055, "frac_reward_zero_std": 0.25, "grad_norm": 14.320113182067871, "learning_rate": 9.316282404787869e-07, "loss": 0.0, "num_tokens": 2643883.0, "reward": 0.53125, "reward_std": 0.3013129234313965, "rewards/format_reward/mean": 0.46875, "rewards/format_reward/std": 0.17536810040473938, "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": 1496.0, "completions/max_terminated_length": 1496.0, "completions/mean_length": 89.8125, "completions/mean_terminated_length": 89.8125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0575, "frac_reward_zero_std": 0.5, "grad_norm": 15.463613510131836, "learning_rate": 9.230476262104676e-07, "loss": -0.0, "num_tokens": 2744167.0, "reward": 0.4718749523162842, "reward_std": 0.22673699259757996, "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": 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": 477.0, "completions/max_terminated_length": 477.0, "completions/mean_length": 32.15625, "completions/mean_terminated_length": 32.15625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06, "frac_reward_zero_std": 0.625, "grad_norm": 17.539522171020508, "learning_rate": 9.1400441557401e-07, "loss": 0.0, "num_tokens": 2856569.0, "reward": 0.39296871423721313, "reward_std": 0.13159699738025665, "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": 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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 881.0, "completions/mean_length": 150.96875, "completions/mean_terminated_length": 120.85714721679688, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0625, "frac_reward_zero_std": 0.125, "grad_norm": 22.964664459228516, "learning_rate": 9.045084971874737e-07, "loss": -0.0, "num_tokens": 2956095.0, "reward": 0.40703120827674866, "reward_std": 0.2988312840461731, "rewards/format_reward/mean": 0.4765625, "rewards/format_reward/std": 0.13886408507823944, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 327.0, "completions/max_terminated_length": 327.0, "completions/mean_length": 18.0625, "completions/mean_terminated_length": 18.0625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.065, "frac_reward_zero_std": 0.5, "grad_norm": 21.533161163330078, "learning_rate": 8.945702546981968e-07, "loss": -0.0, "num_tokens": 3050171.0, "reward": 0.34687498211860657, "reward_std": 0.21778541803359985, "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": 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": 1794.0, "completions/max_terminated_length": 1794.0, "completions/mean_length": 96.390625, "completions/mean_terminated_length": 96.390625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0675, "frac_reward_zero_std": 0.625, "grad_norm": 19.354248046875, "learning_rate": 8.842005554284295e-07, "loss": -0.0, "num_tokens": 3161332.0, "reward": 0.42499998211860657, "reward_std": 0.16675156354904175, "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": 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": 522.0, "completions/max_terminated_length": 522.0, "completions/mean_length": 36.171875, "completions/mean_terminated_length": 36.171875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07, "frac_reward_zero_std": 0.75, "grad_norm": 4.9072980880737305, "learning_rate": 8.734107384920769e-07, "loss": 0.0, "num_tokens": 3233679.0, "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": 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": 438.0, "completions/max_terminated_length": 438.0, "completions/mean_length": 16.421875, "completions/mean_terminated_length": 16.421875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0725, "frac_reward_zero_std": 0.625, "grad_norm": 12.05967903137207, "learning_rate": 8.622126023955445e-07, "loss": -0.0, "num_tokens": 3352234.0, "reward": 0.25312498211860657, "reward_std": 0.15992167592048645, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "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": 957.0, "completions/max_terminated_length": 957.0, "completions/mean_length": 91.296875, "completions/mean_terminated_length": 91.296875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.075, "frac_reward_zero_std": 0.625, "grad_norm": 17.8170108795166, "learning_rate": 8.506183921362442e-07, "loss": -0.0, "num_tokens": 3459293.0, "reward": 0.5960937142372131, "reward_std": 0.1534513533115387, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.546875, "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": 1977.0, "completions/max_terminated_length": 1977.0, "completions/mean_length": 114.9375, "completions/mean_terminated_length": 114.9375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0775, "frac_reward_zero_std": 0.625, "grad_norm": 11.964653015136719, "learning_rate": 8.386407858128706e-07, "loss": -0.0, "num_tokens": 3566153.0, "reward": 0.4570312201976776, "reward_std": 0.09059804677963257, "rewards/format_reward/mean": 0.5078125, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 712.0, "completions/max_terminated_length": 712.0, "completions/mean_length": 69.9375, "completions/mean_terminated_length": 69.9375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08, "frac_reward_zero_std": 0.625, "grad_norm": 13.64649772644043, "learning_rate": 8.262928807620843e-07, "loss": 0.0, "num_tokens": 3642269.0, "reward": 0.5523437261581421, "reward_std": 0.09162302315235138, "rewards/format_reward/mean": 0.5234375, "rewards/format_reward/std": 0.10652101784944534, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "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": 593.0, "completions/max_terminated_length": 593.0, "completions/mean_length": 40.515625, "completions/mean_terminated_length": 40.515625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0825, "frac_reward_zero_std": 0.625, "grad_norm": 14.757148742675781, "learning_rate": 8.135881792367685e-07, "loss": -0.0, "num_tokens": 3736686.0, "reward": 0.4101562201976776, "reward_std": 0.10426762700080872, "rewards/format_reward/mean": 0.5078125, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 775.0, "completions/mean_length": 134.0, "completions/mean_terminated_length": 103.61905670166016, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.085, "frac_reward_zero_std": 0.5, "grad_norm": 12.825127601623535, "learning_rate": 8.005405736415125e-07, "loss": -0.0, "num_tokens": 3833502.0, "reward": 0.6906249523162842, "reward_std": 0.19939783215522766, "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": 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": 609.0, "completions/max_terminated_length": 609.0, "completions/mean_length": 83.71875, "completions/mean_terminated_length": 83.71875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0875, "frac_reward_zero_std": 0.75, "grad_norm": 14.550989151000977, "learning_rate": 7.871643313414718e-07, "loss": -0.0, "num_tokens": 3934252.0, "reward": 0.6281249523162842, "reward_std": 0.10205792635679245, "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": 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": 411.0, "completions/max_terminated_length": 411.0, "completions/mean_length": 52.0, "completions/mean_terminated_length": 52.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.09, "frac_reward_zero_std": 0.625, "grad_norm": 21.93817901611328, "learning_rate": 7.734740790612136e-07, "loss": 0.0, "num_tokens": 4037148.0, "reward": 0.5695312023162842, "reward_std": 0.13474363088607788, "rewards/format_reward/mean": 0.5390625, "rewards/format_reward/std": 0.13524486124515533, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "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": 748.0, "completions/max_terminated_length": 748.0, "completions/mean_length": 105.453125, "completions/mean_terminated_length": 105.453125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0925, "frac_reward_zero_std": 0.5, "grad_norm": 13.75621509552002, "learning_rate": 7.594847868906076e-07, "loss": -0.0, "num_tokens": 4133001.0, "reward": 0.73828125, "reward_std": 0.23312518000602722, "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": 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": 319.0, "completions/max_terminated_length": 319.0, "completions/mean_length": 39.96875, "completions/mean_terminated_length": 39.96875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.095, "frac_reward_zero_std": 0.5, "grad_norm": 10.660466194152832, "learning_rate": 7.452117519152541e-07, "loss": -0.0, "num_tokens": 4225415.0, "reward": 0.5929687023162842, "reward_std": 0.20632557570934296, "rewards/format_reward/mean": 0.6171875, "rewards/format_reward/std": 0.21347814798355103, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "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": 801.0, "completions/max_terminated_length": 801.0, "completions/mean_length": 112.671875, "completions/mean_terminated_length": 112.671875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0975, "frac_reward_zero_std": 0.75, "grad_norm": 11.935098648071289, "learning_rate": 7.306705814893439e-07, "loss": -0.0, "num_tokens": 4319626.0, "reward": 0.39296871423721313, "reward_std": 0.12520109117031097, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.24384792149066925, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "step": 39 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 667.0, "completions/mean_length": 128.875, "completions/mean_terminated_length": 98.41270446777344, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1, "frac_reward_zero_std": 0.625, "grad_norm": 12.141583442687988, "learning_rate": 7.158771761692464e-07, "loss": 0.0, "num_tokens": 4425530.0, "reward": 0.4468749761581421, "reward_std": 0.13258251547813416, "rewards/format_reward/mean": 0.5625, "rewards/format_reward/std": 0.1666666716337204, "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": 618.0, "completions/max_terminated_length": 618.0, "completions/mean_length": 115.109375, "completions/mean_terminated_length": 115.109375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1025, "frac_reward_zero_std": 0.5, "grad_norm": 13.006111145019531, "learning_rate": 7.008477123264847e-07, "loss": 0.0, "num_tokens": 4515321.0, "reward": 0.37343746423721313, "reward_std": 0.16964474320411682, "rewards/format_reward/mean": 0.609375, "rewards/format_reward/std": 0.24346621334552765, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 485.0, "completions/max_terminated_length": 485.0, "completions/mean_length": 78.375, "completions/mean_terminated_length": 78.375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.105, "frac_reward_zero_std": 0.375, "grad_norm": 32.34177017211914, "learning_rate": 6.855986244591103e-07, "loss": -0.0, "num_tokens": 4604545.0, "reward": 0.62890625, "reward_std": 0.23583920300006866, "rewards/format_reward/mean": 0.6640625, "rewards/format_reward/std": 0.2366211861371994, "rewards/mcq_exact_match_reward/mean": 0.5625, "rewards/mcq_exact_match_reward/std": 0.5, "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": 621.0, "completions/max_terminated_length": 621.0, "completions/mean_length": 113.390625, "completions/mean_terminated_length": 113.390625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1075, "frac_reward_zero_std": 0.75, "grad_norm": 13.46899127960205, "learning_rate": 6.701465872208216e-07, "loss": -0.0, "num_tokens": 4684994.0, "reward": 0.5203124284744263, "reward_std": 0.10834787786006927, "rewards/format_reward/mean": 0.515625, "rewards/format_reward/std": 0.08768405020236969, "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": 1031.0, "completions/max_terminated_length": 1031.0, "completions/mean_length": 148.828125, "completions/mean_terminated_length": 148.828125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.11, "frac_reward_zero_std": 0.75, "grad_norm": 16.012969970703125, "learning_rate": 6.545084971874736e-07, "loss": 0.0, "num_tokens": 4803671.0, "reward": 0.53125, "reward_std": 0.06154358386993408, "rewards/format_reward/mean": 0.625, "rewards/format_reward/std": 0.2357022762298584, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "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": 973.0, "completions/max_terminated_length": 973.0, "completions/mean_length": 105.0625, "completions/mean_terminated_length": 105.0625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1125, "frac_reward_zero_std": 0.125, "grad_norm": 24.89419937133789, "learning_rate": 6.387014543809223e-07, "loss": 0.0, "num_tokens": 4899051.0, "reward": 0.6148437261581421, "reward_std": 0.3771878480911255, "rewards/format_reward/mean": 0.6796875, "rewards/format_reward/std": 0.24180518090724945, "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": 689.0, "completions/max_terminated_length": 689.0, "completions/mean_length": 112.140625, "completions/mean_terminated_length": 112.140625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.115, "frac_reward_zero_std": 0.25, "grad_norm": 53.93398666381836, "learning_rate": 6.227427435703995e-07, "loss": 0.0, "num_tokens": 4996884.0, "reward": 0.6117187738418579, "reward_std": 0.2741013467311859, "rewards/format_reward/mean": 0.6484375, "rewards/format_reward/std": 0.24688033759593964, "rewards/mcq_exact_match_reward/mean": 0.546875, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 499.0, "completions/max_terminated_length": 499.0, "completions/mean_length": 41.1875, "completions/mean_terminated_length": 41.1875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1175, "frac_reward_zero_std": 0.5, "grad_norm": 22.220945358276367, "learning_rate": 6.066498153718734e-07, "loss": -0.0, "num_tokens": 5085840.0, "reward": 0.421875, "reward_std": 0.24641379714012146, "rewards/format_reward/mean": 0.78125, "rewards/format_reward/std": 0.25, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "step": 47 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 815.0, "completions/mean_length": 232.984375, "completions/mean_terminated_length": 204.17462158203125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.12, "frac_reward_zero_std": 0.125, "grad_norm": 11.711411476135254, "learning_rate": 5.90440267166055e-07, "loss": 0.0, "num_tokens": 5208495.0, "reward": 0.34765625, "reward_std": 0.24685458838939667, "rewards/format_reward/mean": 0.6640625, "rewards/format_reward/std": 0.2680720090866089, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "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": 545.0, "completions/max_terminated_length": 545.0, "completions/mean_length": 65.984375, "completions/mean_terminated_length": 65.984375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1225, "frac_reward_zero_std": 0.5, "grad_norm": 21.655317306518555, "learning_rate": 5.741318238559209e-07, "loss": 0.0, "num_tokens": 5306174.0, "reward": 0.62890625, "reward_std": 0.11573586612939835, "rewards/format_reward/mean": 0.6640625, "rewards/format_reward/std": 0.2366211861371994, "rewards/mcq_exact_match_reward/mean": 0.5625, "rewards/mcq_exact_match_reward/std": 0.5, "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": 737.0, "completions/max_terminated_length": 737.0, "completions/mean_length": 79.53125, "completions/mean_terminated_length": 79.53125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.125, "frac_reward_zero_std": 0.375, "grad_norm": 14.4029541015625, "learning_rate": 5.577423184847931e-07, "loss": -0.0, "num_tokens": 5413376.0, "reward": 0.6429687142372131, "reward_std": 0.19145923852920532, "rewards/format_reward/mean": 0.8046875, "rewards/format_reward/std": 0.24587368965148926, "rewards/mcq_exact_match_reward/mean": 0.5625, "rewards/mcq_exact_match_reward/std": 0.5, "step": 50 } ], "logging_steps": 1, "max_steps": 100, "num_input_tokens_seen": 5413376, "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 }