{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.25, "eval_steps": 500, "global_step": 100, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1148.0, "completions/max_terminated_length": 1148.0, "completions/mean_length": 365.59375, "completions/mean_terminated_length": 365.59375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0025, "frac_reward_zero_std": 0.0, "grad_norm": 3.4189677238464355, "learning_rate": 0.0, "loss": -0.0, "num_tokens": 109126.0, "reward": 0.26484373211860657, "reward_std": 0.28110986948013306, "rewards/format_reward/mean": 0.3046875, "rewards/format_reward/std": 0.3036443889141083, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 1977.0, "completions/max_terminated_length": 1977.0, "completions/mean_length": 338.890625, "completions/mean_terminated_length": 338.890625, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.005, "frac_reward_zero_std": 0.0, "grad_norm": 15.90227222442627, "learning_rate": 2e-07, "loss": 0.0, "num_tokens": 230415.0, "reward": 0.328125, "reward_std": 0.3611314296722412, "rewards/format_reward/mean": 0.3125, "rewards/format_reward/std": 0.2745848298072815, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1597.0, "completions/mean_length": 456.71875, "completions/mean_terminated_length": 405.3870849609375, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.0075, "frac_reward_zero_std": 0.0, "grad_norm": 8.803783416748047, "learning_rate": 4e-07, "loss": 0.0, "num_tokens": 337765.0, "reward": 0.16953124105930328, "reward_std": 0.30735519528388977, "rewards/format_reward/mean": 0.2890625, "rewards/format_reward/std": 0.2789533734321594, "rewards/mcq_exact_match_reward/mean": 0.140625, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "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": 1510.0, "completions/max_terminated_length": 1510.0, "completions/mean_length": 528.90625, "completions/mean_terminated_length": 528.90625, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.01, "frac_reward_zero_std": 0.0, "grad_norm": 47.8989372253418, "learning_rate": 6e-07, "loss": 0.0, "num_tokens": 444183.0, "reward": 0.15859374403953552, "reward_std": 0.24041227996349335, "rewards/format_reward/mean": 0.3359375, "rewards/format_reward/std": 0.25283610820770264, "rewards/mcq_exact_match_reward/mean": 0.125, "rewards/mcq_exact_match_reward/std": 0.3333333432674408, "step": 4 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1486.0, "completions/mean_length": 557.3125, "completions/mean_terminated_length": 509.2257995605469, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.0125, "frac_reward_zero_std": 0.125, "grad_norm": 13.940407752990723, "learning_rate": 8e-07, "loss": -0.0, "num_tokens": 567939.0, "reward": 0.15937498211860657, "reward_std": 0.27569344639778137, "rewards/format_reward/mean": 0.34375, "rewards/format_reward/std": 0.25, "rewards/mcq_exact_match_reward/mean": 0.125, "rewards/mcq_exact_match_reward/std": 0.3333333432674408, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1699.0, "completions/mean_length": 590.765625, "completions/mean_terminated_length": 519.0983276367188, "completions/min_length": 5.0, "completions/min_terminated_length": 5.0, "epoch": 0.015, "frac_reward_zero_std": 0.125, "grad_norm": 8.428192138671875, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 718588.0, "reward": 0.20078125596046448, "reward_std": 0.32229453325271606, "rewards/format_reward/mean": 0.2890625, "rewards/format_reward/std": 0.2789533734321594, "rewards/mcq_exact_match_reward/mean": 0.171875, "rewards/mcq_exact_match_reward/std": 0.38025420904159546, "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": 1950.0, "completions/max_terminated_length": 1950.0, "completions/mean_length": 447.453125, "completions/mean_terminated_length": 447.453125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0175, "frac_reward_zero_std": 0.0, "grad_norm": 7.821189880371094, "learning_rate": 9.99726628670463e-07, "loss": 0.0, "num_tokens": 835953.0, "reward": 0.12031249701976776, "reward_std": 0.21482786536216736, "rewards/format_reward/mean": 0.265625, "rewards/format_reward/std": 0.2514837086200714, "rewards/mcq_exact_match_reward/mean": 0.09375, "rewards/mcq_exact_match_reward/std": 0.29378482699394226, "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": 1311.0, "completions/mean_length": 465.65625, "completions/mean_terminated_length": 440.5397033691406, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.02, "frac_reward_zero_std": 0.0, "grad_norm": 13.07833480834961, "learning_rate": 9.989068136093872e-07, "loss": 0.0, "num_tokens": 938315.0, "reward": 0.26875001192092896, "reward_std": 0.3125661611557007, "rewards/format_reward/mean": 0.34375, "rewards/format_reward/std": 0.29378482699394226, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 1398.0, "completions/max_terminated_length": 1398.0, "completions/mean_length": 434.421875, "completions/mean_terminated_length": 434.421875, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.0225, "frac_reward_zero_std": 0.0, "grad_norm": 10.79680061340332, "learning_rate": 9.975414512725056e-07, "loss": -0.0, "num_tokens": 1055974.0, "reward": 0.20078124105930328, "reward_std": 0.2931196689605713, "rewards/format_reward/mean": 0.2890625, "rewards/format_reward/std": 0.2928335666656494, "rewards/mcq_exact_match_reward/mean": 0.171875, "rewards/mcq_exact_match_reward/std": 0.38025420904159546, "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": 1588.0, "completions/max_terminated_length": 1588.0, "completions/mean_length": 395.328125, "completions/mean_terminated_length": 395.328125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.025, "frac_reward_zero_std": 0.125, "grad_norm": 5.580694675445557, "learning_rate": 9.956320346634875e-07, "loss": -0.0, "num_tokens": 1170515.0, "reward": 0.23046875, "reward_std": 0.26288744807243347, "rewards/format_reward/mean": 0.4296875, "rewards/format_reward/std": 0.29367929697036743, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "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": 1624.0, "completions/max_terminated_length": 1624.0, "completions/mean_length": 416.84375, "completions/mean_terminated_length": 416.84375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0275, "frac_reward_zero_std": 0.0, "grad_norm": 13.259758949279785, "learning_rate": 9.931806517013612e-07, "loss": -0.0, "num_tokens": 1283785.0, "reward": 0.30781248211860657, "reward_std": 0.3837190568447113, "rewards/format_reward/mean": 0.421875, "rewards/format_reward/std": 0.27048972249031067, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "step": 11 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1758.0, "completions/mean_length": 505.28125, "completions/mean_terminated_length": 455.51611328125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.03, "frac_reward_zero_std": 0.0, "grad_norm": 7.822777271270752, "learning_rate": 9.901899829374047e-07, "loss": -0.0, "num_tokens": 1397251.0, "reward": 0.3765624761581421, "reward_std": 0.35432469844818115, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.26679685711860657, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 1199.0, "completions/max_terminated_length": 1199.0, "completions/mean_length": 307.9375, "completions/mean_terminated_length": 307.9375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0325, "frac_reward_zero_std": 0.0, "grad_norm": 11.520330429077148, "learning_rate": 9.866632986240029e-07, "loss": -0.0, "num_tokens": 1503791.0, "reward": 0.30781248211860657, "reward_std": 0.44272083044052124, "rewards/format_reward/mean": 0.421875, "rewards/format_reward/std": 0.2221602201461792, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1612.0, "completions/mean_length": 509.734375, "completions/mean_terminated_length": 434.0819396972656, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.035, "frac_reward_zero_std": 0.25, "grad_norm": 11.170822143554688, "learning_rate": 9.826044551386742e-07, "loss": 0.0, "num_tokens": 1634846.0, "reward": 0.26249998807907104, "reward_std": 0.17784461379051208, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.208927720785141, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "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": 1530.0, "completions/max_terminated_length": 1530.0, "completions/mean_length": 310.625, "completions/mean_terminated_length": 310.625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0375, "frac_reward_zero_std": 0.125, "grad_norm": 12.4185152053833, "learning_rate": 9.780178907671788e-07, "loss": 0.0, "num_tokens": 1752902.0, "reward": 0.2874999940395355, "reward_std": 0.31605303287506104, "rewards/format_reward/mean": 0.53125, "rewards/format_reward/std": 0.279951810836792, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 1315.0, "completions/max_terminated_length": 1315.0, "completions/mean_length": 255.484375, "completions/mean_terminated_length": 255.484375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.04, "frac_reward_zero_std": 0.125, "grad_norm": 15.137930870056152, "learning_rate": 9.729086208503173e-07, "loss": 0.0, "num_tokens": 1839637.0, "reward": 0.2828124761581421, "reward_std": 0.25739848613739014, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.1985812783241272, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 1279.0, "completions/max_terminated_length": 1279.0, "completions/mean_length": 158.46875, "completions/mean_terminated_length": 158.46875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0425, "frac_reward_zero_std": 0.25, "grad_norm": 16.657760620117188, "learning_rate": 9.672822322997304e-07, "loss": -0.0, "num_tokens": 1950499.0, "reward": 0.3648437261581421, "reward_std": 0.28681617975234985, "rewards/format_reward/mean": 0.5234375, "rewards/format_reward/std": 0.1875000149011612, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 819.0, "completions/max_terminated_length": 819.0, "completions/mean_length": 74.84375, "completions/mean_terminated_length": 74.84375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.045, "frac_reward_zero_std": 0.375, "grad_norm": 13.936868667602539, "learning_rate": 9.611448774886923e-07, "loss": -0.0, "num_tokens": 2038145.0, "reward": 0.43359375, "reward_std": 0.2583305537700653, "rewards/format_reward/mean": 0.5859375, "rewards/format_reward/std": 0.209963858127594, "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": 730.0, "completions/max_terminated_length": 730.0, "completions/mean_length": 171.96875, "completions/mean_terminated_length": 171.96875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0475, "frac_reward_zero_std": 0.25, "grad_norm": 13.823277473449707, "learning_rate": 9.545032675245813e-07, "loss": 0.0, "num_tokens": 2135143.0, "reward": 0.5234375, "reward_std": 0.3237568140029907, "rewards/format_reward/mean": 0.546875, "rewards/format_reward/std": 0.1717960685491562, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "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": 1072.0, "completions/max_terminated_length": 1072.0, "completions/mean_length": 178.25, "completions/mean_terminated_length": 178.25, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05, "frac_reward_zero_std": 0.375, "grad_norm": 48.28337860107422, "learning_rate": 9.473646649103817e-07, "loss": 0.0, "num_tokens": 2228263.0, "reward": 0.453125, "reward_std": 0.29175418615341187, "rewards/format_reward/mean": 0.625, "rewards/format_reward/std": 0.2182178944349289, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 508.0, "completions/max_terminated_length": 508.0, "completions/mean_length": 21.0, "completions/mean_terminated_length": 21.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0525, "frac_reward_zero_std": 0.5, "grad_norm": 23.907447814941406, "learning_rate": 9.397368756032444e-07, "loss": 0.0, "num_tokens": 2334327.0, "reward": 0.2734375, "reward_std": 0.16964475810527802, "rewards/format_reward/mean": 0.546875, "rewards/format_reward/std": 0.14689241349697113, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "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": 629.0, "completions/max_terminated_length": 629.0, "completions/mean_length": 52.109375, "completions/mean_terminated_length": 52.109375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.055, "frac_reward_zero_std": 0.5, "grad_norm": 29.481897354125977, "learning_rate": 9.316282404787869e-07, "loss": -0.0, "num_tokens": 2414950.0, "reward": 0.5421874523162842, "reward_std": 0.13995961844921112, "rewards/format_reward/mean": 0.578125, "rewards/format_reward/std": 0.20351573824882507, "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": 647.0, "completions/max_terminated_length": 647.0, "completions/mean_length": 43.3125, "completions/mean_terminated_length": 43.3125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0575, "frac_reward_zero_std": 0.375, "grad_norm": 23.262107849121094, "learning_rate": 9.230476262104676e-07, "loss": 0.0, "num_tokens": 2509898.0, "reward": 0.5578124523162842, "reward_std": 0.30553871393203735, "rewards/format_reward/mean": 0.578125, "rewards/format_reward/std": 0.20351573824882507, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 7.703125, "completions/mean_terminated_length": 7.703125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06, "frac_reward_zero_std": 0.5, "grad_norm": 33.23601531982422, "learning_rate": 9.1400441557401e-07, "loss": 0.0, "num_tokens": 2591999.0, "reward": 0.3125, "reward_std": 0.1903783231973648, "rewards/format_reward/mean": 0.625, "rewards/format_reward/std": 0.2182178944349289, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 209.0, "completions/max_terminated_length": 209.0, "completions/mean_length": 18.640625, "completions/mean_terminated_length": 18.640625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0625, "frac_reward_zero_std": 0.375, "grad_norm": 37.055049896240234, "learning_rate": 9.045084971874737e-07, "loss": -0.0, "num_tokens": 2685136.0, "reward": 0.5539062023162842, "reward_std": 0.22436748445034027, "rewards/format_reward/mean": 0.6953125, "rewards/format_reward/std": 0.24587368965148926, "rewards/mcq_exact_match_reward/mean": 0.484375, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "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": 186.0, "completions/max_terminated_length": 186.0, "completions/mean_length": 9.75, "completions/mean_terminated_length": 9.75, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.065, "frac_reward_zero_std": 0.25, "grad_norm": 28.075769424438477, "learning_rate": 8.945702546981968e-07, "loss": -0.0, "num_tokens": 2764240.0, "reward": 0.41718748211860657, "reward_std": 0.28056418895721436, "rewards/format_reward/mean": 0.578125, "rewards/format_reward/std": 0.18298126757144928, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 616.0, "completions/max_terminated_length": 616.0, "completions/mean_length": 50.515625, "completions/mean_terminated_length": 50.515625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0675, "frac_reward_zero_std": 0.5, "grad_norm": 18.022727966308594, "learning_rate": 8.842005554284295e-07, "loss": -0.0, "num_tokens": 2851385.0, "reward": 0.4429687261581421, "reward_std": 0.23292513191699982, "rewards/format_reward/mean": 0.6796875, "rewards/format_reward/std": 0.2576941251754761, "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": 344.0, "completions/max_terminated_length": 344.0, "completions/mean_length": 20.140625, "completions/mean_terminated_length": 20.140625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07, "frac_reward_zero_std": 0.5, "grad_norm": 33.50341033935547, "learning_rate": 8.734107384920769e-07, "loss": 0.0, "num_tokens": 2974386.0, "reward": 0.5390625, "reward_std": 0.15597516298294067, "rewards/format_reward/mean": 0.546875, "rewards/format_reward/std": 0.14689241349697113, "rewards/mcq_exact_match_reward/mean": 0.484375, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "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": 307.0, "completions/max_terminated_length": 307.0, "completions/mean_length": 16.96875, "completions/mean_terminated_length": 16.96875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0725, "frac_reward_zero_std": 0.5, "grad_norm": 11.394798278808594, "learning_rate": 8.622126023955445e-07, "loss": -0.0, "num_tokens": 3065512.0, "reward": 0.32421875, "reward_std": 0.1689612865447998, "rewards/format_reward/mean": 0.7421875, "rewards/format_reward/std": 0.25185325741767883, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 342.0, "completions/max_terminated_length": 342.0, "completions/mean_length": 14.109375, "completions/mean_terminated_length": 14.109375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.075, "frac_reward_zero_std": 0.5, "grad_norm": 24.388032913208008, "learning_rate": 8.506183921362442e-07, "loss": 0.0, "num_tokens": 3141575.0, "reward": 0.6109374761581421, "reward_std": 0.21881303191184998, "rewards/format_reward/mean": 0.640625, "rewards/format_reward/std": 0.22658175230026245, "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": 407.0, "completions/max_terminated_length": 407.0, "completions/mean_length": 17.890625, "completions/mean_terminated_length": 17.890625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0775, "frac_reward_zero_std": 0.375, "grad_norm": 25.361494064331055, "learning_rate": 8.386407858128706e-07, "loss": -0.0, "num_tokens": 3220200.0, "reward": 0.5257812738418579, "reward_std": 0.3186529874801636, "rewards/format_reward/mean": 0.7265625, "rewards/format_reward/std": 0.250866562128067, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 178.0, "completions/max_terminated_length": 178.0, "completions/mean_length": 17.515625, "completions/mean_terminated_length": 17.515625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08, "frac_reward_zero_std": 0.25, "grad_norm": 18.510908126831055, "learning_rate": 8.262928807620843e-07, "loss": 0.0, "num_tokens": 3292849.0, "reward": 0.3929687738418579, "reward_std": 0.2731408476829529, "rewards/format_reward/mean": 0.8046875, "rewards/format_reward/std": 0.2615155577659607, "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": 185.0, "completions/max_terminated_length": 185.0, "completions/mean_length": 12.53125, "completions/mean_terminated_length": 12.53125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0825, "frac_reward_zero_std": 0.25, "grad_norm": 44.879940032958984, "learning_rate": 8.135881792367685e-07, "loss": -0.0, "num_tokens": 3398835.0, "reward": 0.47734373807907104, "reward_std": 0.22218023240566254, "rewards/format_reward/mean": 0.7109375, "rewards/format_reward/std": 0.2643453776836395, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 156.0, "completions/max_terminated_length": 156.0, "completions/mean_length": 16.71875, "completions/mean_terminated_length": 16.71875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.085, "frac_reward_zero_std": 0.625, "grad_norm": 7.9521484375, "learning_rate": 8.005405736415125e-07, "loss": 0.0, "num_tokens": 3464233.0, "reward": 0.5406250357627869, "reward_std": 0.1530819982290268, "rewards/format_reward/mean": 0.875, "rewards/format_reward/std": 0.2182178944349289, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1218.0, "completions/mean_length": 86.609375, "completions/mean_terminated_length": 55.476192474365234, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0875, "frac_reward_zero_std": 0.375, "grad_norm": 22.898433685302734, "learning_rate": 7.871643313414718e-07, "loss": -0.0, "num_tokens": 3562752.0, "reward": 0.5648437738418579, "reward_std": 0.17209085822105408, "rewards/format_reward/mean": 0.8046875, "rewards/format_reward/std": 0.29028159379959106, "rewards/mcq_exact_match_reward/mean": 0.484375, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "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": 244.0, "completions/max_terminated_length": 244.0, "completions/mean_length": 14.875, "completions/mean_terminated_length": 14.875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.09, "frac_reward_zero_std": 0.25, "grad_norm": 26.0075740814209, "learning_rate": 7.734740790612136e-07, "loss": 0.0, "num_tokens": 3634192.0, "reward": 0.7484375238418579, "reward_std": 0.2853722870349884, "rewards/format_reward/mean": 0.921875, "rewards/format_reward/std": 0.18298126757144928, "rewards/mcq_exact_match_reward/mean": 0.65625, "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": 380.0, "completions/max_terminated_length": 380.0, "completions/mean_length": 24.859375, "completions/mean_terminated_length": 24.859375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0925, "frac_reward_zero_std": 0.5, "grad_norm": 21.14771842956543, "learning_rate": 7.594847868906076e-07, "loss": 0.0, "num_tokens": 3724279.0, "reward": 0.4531250298023224, "reward_std": 0.24258562922477722, "rewards/format_reward/mean": 0.9375, "rewards/format_reward/std": 0.18898223340511322, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 241.0, "completions/max_terminated_length": 241.0, "completions/mean_length": 21.03125, "completions/mean_terminated_length": 21.03125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.095, "frac_reward_zero_std": 0.5, "grad_norm": 25.86863899230957, "learning_rate": 7.452117519152541e-07, "loss": 0.0, "num_tokens": 3802521.0, "reward": 0.34921878576278687, "reward_std": 0.18475252389907837, "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": 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": 246.0, "completions/max_terminated_length": 246.0, "completions/mean_length": 18.703125, "completions/mean_terminated_length": 18.703125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0975, "frac_reward_zero_std": 0.625, "grad_norm": 10.359986305236816, "learning_rate": 7.306705814893439e-07, "loss": 0.0, "num_tokens": 3890246.0, "reward": 0.578125, "reward_std": 0.15992169082164764, "rewards/format_reward/mean": 0.9375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.484375, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "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": 94.0, "completions/max_terminated_length": 94.0, "completions/mean_length": 14.609375, "completions/mean_terminated_length": 14.609375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1, "frac_reward_zero_std": 0.875, "grad_norm": 4.429067611694336, "learning_rate": 7.158771761692464e-07, "loss": 0.0, "num_tokens": 3978645.0, "reward": 0.7406250238418579, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.640625, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 1148.0, "completions/max_terminated_length": 1148.0, "completions/mean_length": 30.90625, "completions/mean_terminated_length": 30.90625, "completions/min_length": 8.0, "completions/min_terminated_length": 8.0, "epoch": 0.1025, "frac_reward_zero_std": 0.75, "grad_norm": 12.020353317260742, "learning_rate": 7.008477123264847e-07, "loss": -0.0, "num_tokens": 4085879.0, "reward": 0.5875000357627869, "reward_std": 0.13363061845302582, "rewards/format_reward/mean": 0.875, "rewards/format_reward/std": 0.3333333432674408, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "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": 1190.0, "completions/max_terminated_length": 1190.0, "completions/mean_length": 33.4375, "completions/mean_terminated_length": 33.4375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.105, "frac_reward_zero_std": 0.75, "grad_norm": 7.975914001464844, "learning_rate": 6.855986244591103e-07, "loss": -0.0, "num_tokens": 4186219.0, "reward": 0.31796878576278687, "reward_std": 0.1099943220615387, "rewards/format_reward/mean": 0.9921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "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": 228.0, "completions/max_terminated_length": 228.0, "completions/mean_length": 20.796875, "completions/mean_terminated_length": 20.796875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1075, "frac_reward_zero_std": 0.625, "grad_norm": 28.40608787536621, "learning_rate": 6.701465872208216e-07, "loss": 0.0, "num_tokens": 4292846.0, "reward": 0.5687500238418579, "reward_std": 0.1735912710428238, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 43 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 17.0, "completions/max_terminated_length": 17.0, "completions/mean_length": 12.109375, "completions/mean_terminated_length": 12.109375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.11, "frac_reward_zero_std": 0.625, "grad_norm": 19.773279190063477, "learning_rate": 6.545084971874736e-07, "loss": 0.0, "num_tokens": 4378973.0, "reward": 0.5218750238418579, "reward_std": 0.1804211586713791, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "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": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1125, "frac_reward_zero_std": 0.5, "grad_norm": 10.283831596374512, "learning_rate": 6.387014543809223e-07, "loss": 0.0, "num_tokens": 4476549.0, "reward": 0.6312500238418579, "reward_std": 0.19727617502212524, "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": 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": 175.0, "completions/max_terminated_length": 175.0, "completions/mean_length": 14.546875, "completions/mean_terminated_length": 14.546875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.115, "frac_reward_zero_std": 0.5, "grad_norm": 28.187740325927734, "learning_rate": 6.227427435703995e-07, "loss": 0.0, "num_tokens": 4572880.0, "reward": 0.7718750238418579, "reward_std": 0.2109457552433014, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.671875, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 285.0, "completions/max_terminated_length": 285.0, "completions/mean_length": 18.453125, "completions/mean_terminated_length": 18.453125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1175, "frac_reward_zero_std": 0.5, "grad_norm": 17.003896713256836, "learning_rate": 6.066498153718734e-07, "loss": 0.0, "num_tokens": 4669397.0, "reward": 0.3656250238418579, "reward_std": 0.19044628739356995, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 385.0, "completions/max_terminated_length": 385.0, "completions/mean_length": 28.28125, "completions/mean_terminated_length": 28.28125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.12, "frac_reward_zero_std": 0.625, "grad_norm": 11.66069221496582, "learning_rate": 5.90440267166055e-07, "loss": 0.0, "num_tokens": 4779695.0, "reward": 0.6156250238418579, "reward_std": 0.189372718334198, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "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": 18.0, "completions/max_terminated_length": 18.0, "completions/mean_length": 12.09375, "completions/mean_terminated_length": 12.09375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1225, "frac_reward_zero_std": 0.875, "grad_norm": 6.9331135749816895, "learning_rate": 5.741318238559209e-07, "loss": 0.0, "num_tokens": 4883989.0, "reward": 0.6312500238418579, "reward_std": 0.0578637570142746, "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": 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": 15.0, "completions/max_terminated_length": 15.0, "completions/mean_length": 12.109375, "completions/mean_terminated_length": 12.109375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.125, "frac_reward_zero_std": 0.75, "grad_norm": 8.471827507019043, "learning_rate": 5.577423184847931e-07, "loss": 0.0, "num_tokens": 4966268.0, "reward": 0.3968750238418579, "reward_std": 0.10205793380737305, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "step": 50 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1275, "frac_reward_zero_std": 0.625, "grad_norm": 12.442874908447266, "learning_rate": 5.412896727361662e-07, "loss": -0.0, "num_tokens": 5053284.0, "reward": 0.3656250238418579, "reward_std": 0.13258251547813416, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "step": 51 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 68.0, "completions/max_terminated_length": 68.0, "completions/mean_length": 13.421875, "completions/mean_terminated_length": 13.421875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.13, "frac_reward_zero_std": 0.5, "grad_norm": 28.07390785217285, "learning_rate": 5.247918773366111e-07, "loss": 0.0, "num_tokens": 5150279.0, "reward": 0.7703125476837158, "reward_std": 0.23586463928222656, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.671875, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 52 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 16.0, "completions/max_terminated_length": 16.0, "completions/mean_length": 12.296875, "completions/mean_terminated_length": 12.296875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1325, "frac_reward_zero_std": 0.75, "grad_norm": 10.012133598327637, "learning_rate": 5.082669723831793e-07, "loss": 0.0, "num_tokens": 5261018.0, "reward": 0.6234375238418579, "reward_std": 0.09188060462474823, "rewards/format_reward/mean": 0.921875, "rewards/format_reward/std": 0.27048972249031067, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 53 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 25.0, "completions/max_terminated_length": 25.0, "completions/mean_length": 12.28125, "completions/mean_terminated_length": 12.28125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.135, "frac_reward_zero_std": 0.5, "grad_norm": 9.309820175170898, "learning_rate": 4.917330276168208e-07, "loss": 0.0, "num_tokens": 5357212.0, "reward": 0.8015625476837158, "reward_std": 0.15750141441822052, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.703125, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "step": 54 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 16.0, "completions/max_terminated_length": 16.0, "completions/mean_length": 12.0625, "completions/mean_terminated_length": 12.0625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1375, "frac_reward_zero_std": 0.875, "grad_norm": 6.933794021606445, "learning_rate": 4.752081226633888e-07, "loss": 0.0, "num_tokens": 5440000.0, "reward": 0.9593750238418579, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.859375, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "step": 55 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 51.0, "completions/max_terminated_length": 51.0, "completions/mean_length": 12.90625, "completions/mean_terminated_length": 12.90625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.14, "frac_reward_zero_std": 0.625, "grad_norm": 9.150296211242676, "learning_rate": 4.5871032726383385e-07, "loss": 0.0, "num_tokens": 5520882.0, "reward": 0.5531250238418579, "reward_std": 0.189372718334198, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 56 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1425, "frac_reward_zero_std": 0.375, "grad_norm": 14.611623764038086, "learning_rate": 4.4225768151520694e-07, "loss": 0.0, "num_tokens": 5596346.0, "reward": 0.6468750238418579, "reward_std": 0.2414703369140625, "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": 57 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 77.0, "completions/max_terminated_length": 77.0, "completions/mean_length": 13.96875, "completions/mean_terminated_length": 13.96875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.145, "frac_reward_zero_std": 0.875, "grad_norm": 4.403719425201416, "learning_rate": 4.258681761440789e-07, "loss": 0.0, "num_tokens": 5668120.0, "reward": 0.8343750238418579, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.734375, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "step": 58 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1466.0, "completions/mean_length": 104.265625, "completions/mean_terminated_length": 73.41270446777344, "completions/min_length": 11.0, "completions/min_terminated_length": 11.0, "epoch": 0.1475, "frac_reward_zero_std": 0.625, "grad_norm": 13.01597785949707, "learning_rate": 4.095597328339452e-07, "loss": 0.0, "num_tokens": 5763281.0, "reward": 0.6382812857627869, "reward_std": 0.16143880784511566, "rewards/format_reward/mean": 0.9140625, "rewards/format_reward/std": 0.244862899184227, "rewards/mcq_exact_match_reward/mean": 0.546875, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 59 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 19.0, "completions/max_terminated_length": 19.0, "completions/mean_length": 12.21875, "completions/mean_terminated_length": 12.21875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.15, "frac_reward_zero_std": 0.875, "grad_norm": 4.214562892913818, "learning_rate": 3.9335018462812664e-07, "loss": 0.0, "num_tokens": 5854455.0, "reward": 0.5843750238418579, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.484375, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "step": 60 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1525, "frac_reward_zero_std": 0.875, "grad_norm": 10.081838607788086, "learning_rate": 3.772572564296004e-07, "loss": 0.0, "num_tokens": 5938351.0, "reward": 0.6781250238418579, "reward_std": 0.0646936446428299, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.578125, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 61 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 55.0, "completions/max_terminated_length": 55.0, "completions/mean_length": 12.6875, "completions/mean_terminated_length": 12.6875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.155, "frac_reward_zero_std": 0.5, "grad_norm": 12.646625518798828, "learning_rate": 3.612985456190778e-07, "loss": -0.0, "num_tokens": 6012259.0, "reward": 0.8812500238418579, "reward_std": 0.2041158527135849, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.78125, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "step": 62 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1575, "frac_reward_zero_std": 0.75, "grad_norm": 12.317092895507812, "learning_rate": 3.454915028125263e-07, "loss": 0.0, "num_tokens": 6104523.0, "reward": 0.5062500238418579, "reward_std": 0.1246790662407875, "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": 63 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 130.0, "completions/max_terminated_length": 130.0, "completions/mean_length": 13.859375, "completions/mean_terminated_length": 13.859375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.16, "frac_reward_zero_std": 0.75, "grad_norm": 15.709736824035645, "learning_rate": 3.2985341277917846e-07, "loss": 0.0, "num_tokens": 6168714.0, "reward": 0.8656250238418579, "reward_std": 0.1315089464187622, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.765625, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "step": 64 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 11.25, "completions/mean_terminated_length": 11.25, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1625, "frac_reward_zero_std": 0.75, "grad_norm": 7.530425071716309, "learning_rate": 3.1440137554088953e-07, "loss": 0.0, "num_tokens": 6244338.0, "reward": 0.515625, "reward_std": 0.10205793380737305, "rewards/format_reward/mean": 0.9375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 65 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 122.0, "completions/max_terminated_length": 122.0, "completions/mean_length": 13.71875, "completions/mean_terminated_length": 13.71875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.165, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 2.9915228767351535e-07, "loss": 0.0, "num_tokens": 6329944.0, "reward": 0.6000000238418579, "reward_std": 0.0, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "step": 66 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1675, "frac_reward_zero_std": 0.75, "grad_norm": 8.945871353149414, "learning_rate": 2.841228238307536e-07, "loss": 0.0, "num_tokens": 6411056.0, "reward": 0.5062500238418579, "reward_std": 0.10888782143592834, "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": 67 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 584.0, "completions/max_terminated_length": 584.0, "completions/mean_length": 21.40625, "completions/mean_terminated_length": 21.40625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.17, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 2.6932941851065615e-07, "loss": 0.0, "num_tokens": 6510226.0, "reward": 0.6000000238418579, "reward_std": 0.0, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "step": 68 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 17.0, "completions/max_terminated_length": 17.0, "completions/mean_length": 12.09375, "completions/mean_terminated_length": 12.09375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1725, "frac_reward_zero_std": 0.625, "grad_norm": 29.34805679321289, "learning_rate": 2.547882480847461e-07, "loss": 0.0, "num_tokens": 6598168.0, "reward": 0.8187500238418579, "reward_std": 0.1462521106004715, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.71875, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 69 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 15.0, "completions/max_terminated_length": 15.0, "completions/mean_length": 12.046875, "completions/mean_terminated_length": 12.046875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.175, "frac_reward_zero_std": 0.625, "grad_norm": 20.082977294921875, "learning_rate": 2.4051521310939254e-07, "loss": 0.0, "num_tokens": 6687171.0, "reward": 0.9906250238418579, "reward_std": 0.1804211586713791, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.890625, "rewards/mcq_exact_match_reward/std": 0.3145764470100403, "step": 70 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1775, "frac_reward_zero_std": 0.75, "grad_norm": 11.251187324523926, "learning_rate": 2.2652592093878665e-07, "loss": 0.0, "num_tokens": 6773563.0, "reward": 0.5531250238418579, "reward_std": 0.10205793380737305, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 71 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 47.0, "completions/max_terminated_length": 47.0, "completions/mean_length": 12.953125, "completions/mean_terminated_length": 12.953125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.18, "frac_reward_zero_std": 0.75, "grad_norm": 10.648567199707031, "learning_rate": 2.128356686585282e-07, "loss": 0.0, "num_tokens": 6877512.0, "reward": 0.5531250238418579, "reward_std": 0.12255740165710449, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 72 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 11.96875, "completions/mean_terminated_length": 11.96875, "completions/min_length": 10.0, "completions/min_terminated_length": 10.0, "epoch": 0.1825, "frac_reward_zero_std": 0.875, "grad_norm": 6.908886909484863, "learning_rate": 1.9945942635848745e-07, "loss": 0.0, "num_tokens": 6997470.0, "reward": 0.3812500238418579, "reward_std": 0.0578637570142746, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 73 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.185, "frac_reward_zero_std": 0.5, "grad_norm": 18.639074325561523, "learning_rate": 1.8641182076323148e-07, "loss": -0.0, "num_tokens": 7062070.0, "reward": 0.5687500238418579, "reward_std": 0.19727617502212524, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 74 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1875, "frac_reward_zero_std": 0.75, "grad_norm": 15.986225128173828, "learning_rate": 1.7370711923791564e-07, "loss": 0.0, "num_tokens": 7171262.0, "reward": 0.5515625476837158, "reward_std": 0.06911306083202362, "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": 75 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 78.0, "completions/max_terminated_length": 78.0, "completions/mean_length": 13.421875, "completions/mean_terminated_length": 13.421875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.19, "frac_reward_zero_std": 0.75, "grad_norm": 6.071596622467041, "learning_rate": 1.6135921418712955e-07, "loss": -0.0, "num_tokens": 7266889.0, "reward": 0.3500000238418579, "reward_std": 0.0883883461356163, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 76 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1925, "frac_reward_zero_std": 0.75, "grad_norm": 12.055264472961426, "learning_rate": 1.493816078637557e-07, "loss": 0.0, "num_tokens": 7349513.0, "reward": 0.8187500238418579, "reward_std": 0.10888782143592834, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.71875, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 77 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.195, "frac_reward_zero_std": 0.75, "grad_norm": 11.782408714294434, "learning_rate": 1.3778739760445552e-07, "loss": 0.0, "num_tokens": 7462945.0, "reward": 0.6937500238418579, "reward_std": 0.10888782143592834, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.59375, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 78 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 169.0, "completions/max_terminated_length": 169.0, "completions/mean_length": 14.453125, "completions/mean_terminated_length": 14.453125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1975, "frac_reward_zero_std": 0.75, "grad_norm": 8.004434585571289, "learning_rate": 1.2658926150792322e-07, "loss": 0.0, "num_tokens": 7530590.0, "reward": 0.729687511920929, "reward_std": 0.04861358925700188, "rewards/format_reward/mean": 0.890625, "rewards/format_reward/std": 0.3145764470100403, "rewards/mcq_exact_match_reward/mean": 0.640625, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "step": 79 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 51.0, "completions/max_terminated_length": 51.0, "completions/mean_length": 13.3125, "completions/mean_terminated_length": 13.3125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2, "frac_reward_zero_std": 0.5, "grad_norm": 16.81325340270996, "learning_rate": 1.1579944457157059e-07, "loss": 0.0, "num_tokens": 7635010.0, "reward": 0.6312500238418579, "reward_std": 0.2130674123764038, "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": 80 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 20.0, "completions/max_terminated_length": 20.0, "completions/mean_length": 12.671875, "completions/mean_terminated_length": 12.671875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2025, "frac_reward_zero_std": 0.75, "grad_norm": 12.965169906616211, "learning_rate": 1.0542974530180327e-07, "loss": 0.0, "num_tokens": 7717781.0, "reward": 0.6468750238418579, "reward_std": 0.11100947856903076, "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": 81 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 37.0, "completions/max_terminated_length": 37.0, "completions/mean_length": 12.421875, "completions/mean_terminated_length": 12.421875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.205, "frac_reward_zero_std": 0.75, "grad_norm": 7.267765998840332, "learning_rate": 9.549150281252632e-08, "loss": 0.0, "num_tokens": 7822384.0, "reward": 0.6609375476837158, "reward_std": 0.07123472541570663, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.5625, "rewards/mcq_exact_match_reward/std": 0.5, "step": 82 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2075, "frac_reward_zero_std": 0.875, "grad_norm": 8.191621780395508, "learning_rate": 8.599558442598998e-08, "loss": -0.0, "num_tokens": 7899424.0, "reward": 0.8656250238418579, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.765625, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "step": 83 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.21, "frac_reward_zero_std": 0.75, "grad_norm": 8.849154472351074, "learning_rate": 7.695237378953224e-08, "loss": 0.0, "num_tokens": 8005656.0, "reward": 0.6000000238418579, "reward_std": 0.0883883461356163, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "step": 84 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2125, "frac_reward_zero_std": 0.75, "grad_norm": 12.004926681518555, "learning_rate": 6.837175952121304e-08, "loss": -0.0, "num_tokens": 8092296.0, "reward": 0.4281250238418579, "reward_std": 0.11100947856903076, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 85 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 16.0, "completions/max_terminated_length": 16.0, "completions/mean_length": 12.0625, "completions/mean_terminated_length": 12.0625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.215, "frac_reward_zero_std": 0.625, "grad_norm": 18.872772216796875, "learning_rate": 6.026312439675551e-08, "loss": 0.0, "num_tokens": 8180828.0, "reward": 0.6781250238418579, "reward_std": 0.1804211586713791, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.578125, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 86 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2175, "frac_reward_zero_std": 0.875, "grad_norm": 10.038103103637695, "learning_rate": 5.263533508961826e-08, "loss": 0.0, "num_tokens": 8254508.0, "reward": 1.021875023841858, "reward_std": 0.0646936446428299, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.921875, "rewards/mcq_exact_match_reward/std": 0.27048972249031067, "step": 87 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.22, "frac_reward_zero_std": 0.875, "grad_norm": 11.348250389099121, "learning_rate": 4.549673247541874e-08, "loss": 0.0, "num_tokens": 8321012.0, "reward": 0.7562500238418579, "reward_std": 0.0578637570142746, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.65625, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "step": 88 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2225, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 3.8855122511307626e-08, "loss": 0.0, "num_tokens": 8415084.0, "reward": 0.6000000238418579, "reward_std": 0.0, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "step": 89 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.225, "frac_reward_zero_std": 0.875, "grad_norm": 9.320042610168457, "learning_rate": 3.271776770026963e-08, "loss": 0.0, "num_tokens": 8500924.0, "reward": 0.5218750238418579, "reward_std": 0.0646936446428299, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 90 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2275, "frac_reward_zero_std": 0.625, "grad_norm": 14.410379409790039, "learning_rate": 2.7091379149682682e-08, "loss": 0.0, "num_tokens": 8586316.0, "reward": 0.6781250238418579, "reward_std": 0.1530819833278656, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.578125, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 91 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.23, "frac_reward_zero_std": 0.875, "grad_norm": 19.93406105041504, "learning_rate": 2.1982109232821176e-08, "loss": 0.0, "num_tokens": 8666996.0, "reward": 0.4906250238418579, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "step": 92 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2325, "frac_reward_zero_std": 0.75, "grad_norm": 10.138134002685547, "learning_rate": 1.7395544861325718e-08, "loss": 0.0, "num_tokens": 8747804.0, "reward": 0.6781250238418579, "reward_std": 0.10205793380737305, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.578125, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 93 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 35.0, "completions/max_terminated_length": 35.0, "completions/mean_length": 13.109375, "completions/mean_terminated_length": 13.109375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.235, "frac_reward_zero_std": 0.875, "grad_norm": 13.264481544494629, "learning_rate": 1.3336701375997127e-08, "loss": 0.0, "num_tokens": 8824003.0, "reward": 0.5062500238418579, "reward_std": 0.0578637570142746, "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": 94 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 128.0, "completions/max_terminated_length": 128.0, "completions/mean_length": 14.84375, "completions/mean_terminated_length": 14.84375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2375, "frac_reward_zero_std": 0.75, "grad_norm": 12.212809562683105, "learning_rate": 9.810017062595321e-09, "loss": -0.0, "num_tokens": 8903281.0, "reward": 0.8031250238418579, "reward_std": 0.10205793380737305, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.703125, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "step": 95 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 11.8125, "completions/mean_terminated_length": 11.8125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.24, "frac_reward_zero_std": 0.875, "grad_norm": 19.784290313720703, "learning_rate": 6.819348298638839e-09, "loss": 0.0, "num_tokens": 9002229.0, "reward": 0.848437488079071, "reward_std": 0.0028931896667927504, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.08768405020236969, "rewards/mcq_exact_match_reward/mean": 0.75, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 96 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.015625, "completions/mean_terminated_length": 12.015625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2425, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 4.367965336512403e-09, "loss": 0.0, "num_tokens": 9076830.0, "reward": 0.8500000238418579, "reward_std": 0.0, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.75, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 97 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 12.0, "completions/max_terminated_length": 12.0, "completions/mean_length": 12.0, "completions/mean_terminated_length": 12.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.245, "frac_reward_zero_std": 0.75, "grad_norm": 6.8440117835998535, "learning_rate": 2.458548727494292e-09, "loss": -0.0, "num_tokens": 9194030.0, "reward": 0.7093750238418579, "reward_std": 0.10205793380737305, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.609375, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "step": 98 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 126.0, "completions/max_terminated_length": 126.0, "completions/mean_length": 14.015625, "completions/mean_terminated_length": 14.015625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.2475, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 1.0931863906127325e-09, "loss": 0.0, "num_tokens": 9278191.0, "reward": 0.6000000238418579, "reward_std": 0.0, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "step": 99 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 98.0, "completions/max_terminated_length": 98.0, "completions/mean_length": 14.375, "completions/mean_terminated_length": 14.375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.25, "frac_reward_zero_std": 0.625, "grad_norm": 12.697029113769531, "learning_rate": 2.733713295369755e-10, "loss": 0.0, "num_tokens": 9361111.0, "reward": 0.6421875357627869, "reward_std": 0.1085273027420044, "rewards/format_reward/mean": 0.953125, "rewards/format_reward/std": 0.21304203569889069, "rewards/mcq_exact_match_reward/mean": 0.546875, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 100 } ], "logging_steps": 1, "max_steps": 100, "num_input_tokens_seen": 9361111, "num_train_epochs": 1, "save_steps": 50, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }