{ "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1653.0, "completions/mean_length": 421.875, "completions/mean_terminated_length": 396.0635070800781, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0025, "frac_reward_zero_std": 0.0, "grad_norm": 17.444801330566406, "learning_rate": 0.0, "loss": 0.0, "num_tokens": 111584.0, "reward": 0.2710937261581421, "reward_std": 0.28761613368988037, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.3117053508758545, "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": 1314.0, "completions/max_terminated_length": 1314.0, "completions/mean_length": 433.140625, "completions/mean_terminated_length": 433.140625, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.005, "frac_reward_zero_std": 0.0, "grad_norm": 11.636614799499512, "learning_rate": 2e-07, "loss": 0.0, "num_tokens": 260057.0, "reward": 0.2984374761581421, "reward_std": 0.3332170844078064, "rewards/format_reward/mean": 0.328125, "rewards/format_reward/std": 0.23935678601264954, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "step": 2 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1357.0, "completions/mean_length": 469.609375, "completions/mean_terminated_length": 444.5555725097656, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0075, "frac_reward_zero_std": 0.0, "grad_norm": 11.887650489807129, "learning_rate": 4e-07, "loss": -0.0, "num_tokens": 409712.0, "reward": 0.3164062201976776, "reward_std": 0.3063698410987854, "rewards/format_reward/mean": 0.3515625, "rewards/format_reward/std": 0.24688033759593964, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "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": 1976.0, "completions/max_terminated_length": 1976.0, "completions/mean_length": 521.875, "completions/mean_terminated_length": 521.875, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.01, "frac_reward_zero_std": 0.0, "grad_norm": 9.74310302734375, "learning_rate": 6e-07, "loss": 0.0, "num_tokens": 592296.0, "reward": 0.140625, "reward_std": 0.2123577743768692, "rewards/format_reward/mean": 0.3125, "rewards/format_reward/std": 0.24397502839565277, "rewards/mcq_exact_match_reward/mean": 0.109375, "rewards/mcq_exact_match_reward/std": 0.3145764470100403, "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": 1710.0, "completions/mean_length": 533.3125, "completions/mean_terminated_length": 484.45159912109375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0125, "frac_reward_zero_std": 0.0, "grad_norm": 8.553196907043457, "learning_rate": 8e-07, "loss": -0.0, "num_tokens": 730156.0, "reward": 0.32890623807907104, "reward_std": 0.38638970255851746, "rewards/format_reward/mean": 0.3203125, "rewards/format_reward/std": 0.24180518090724945, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "step": 5 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1274.0, "completions/max_terminated_length": 1274.0, "completions/mean_length": 288.578125, "completions/mean_terminated_length": 288.578125, "completions/min_length": 5.0, "completions/min_terminated_length": 5.0, "epoch": 0.015, "frac_reward_zero_std": 0.0, "grad_norm": 11.435674667358398, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 835745.0, "reward": 0.38984373211860657, "reward_std": 0.27904677391052246, "rewards/format_reward/mean": 0.3046875, "rewards/format_reward/std": 0.29028159379959106, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 1493.0, "completions/max_terminated_length": 1493.0, "completions/mean_length": 391.453125, "completions/mean_terminated_length": 391.453125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0175, "frac_reward_zero_std": 0.125, "grad_norm": 10.314692497253418, "learning_rate": 9.99726628670463e-07, "loss": -0.0, "num_tokens": 963934.0, "reward": 0.3648437261581421, "reward_std": 0.3693491518497467, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.22257846593856812, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 7 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1840.0, "completions/max_terminated_length": 1840.0, "completions/mean_length": 406.203125, "completions/mean_terminated_length": 406.203125, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.02, "frac_reward_zero_std": 0.0, "grad_norm": 9.152073860168457, "learning_rate": 9.989068136093872e-07, "loss": -0.0, "num_tokens": 1067211.0, "reward": 0.4398437440395355, "reward_std": 0.43246692419052124, "rewards/format_reward/mean": 0.3359375, "rewards/format_reward/std": 0.29620200395584106, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 1250.0, "completions/max_terminated_length": 1250.0, "completions/mean_length": 417.140625, "completions/mean_terminated_length": 417.140625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0225, "frac_reward_zero_std": 0.0, "grad_norm": 7.290132999420166, "learning_rate": 9.975414512725056e-07, "loss": -0.0, "num_tokens": 1186452.0, "reward": 0.40312498807907104, "reward_std": 0.39377546310424805, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.22712838649749756, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1970.0, "completions/mean_length": 682.625, "completions/mean_terminated_length": 615.4754028320312, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.025, "frac_reward_zero_std": 0.0, "grad_norm": 6.748217582702637, "learning_rate": 9.956320346634875e-07, "loss": 0.0, "num_tokens": 1336380.0, "reward": 0.26015621423721313, "reward_std": 0.3470621705055237, "rewards/format_reward/mean": 0.4140625, "rewards/format_reward/std": 0.2280818521976471, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "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": 1743.0, "completions/max_terminated_length": 1743.0, "completions/mean_length": 380.15625, "completions/mean_terminated_length": 380.15625, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.0275, "frac_reward_zero_std": 0.0, "grad_norm": 20.983882904052734, "learning_rate": 9.931806517013612e-07, "loss": -0.0, "num_tokens": 1459254.0, "reward": 0.590624988079071, "reward_std": 0.456085205078125, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.3149704039096832, "rewards/mcq_exact_match_reward/mean": 0.546875, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 1615.0, "completions/max_terminated_length": 1615.0, "completions/mean_length": 463.6875, "completions/mean_terminated_length": 463.6875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.03, "frac_reward_zero_std": 0.125, "grad_norm": 7.596439838409424, "learning_rate": 9.901899829374047e-07, "loss": 0.0, "num_tokens": 1582506.0, "reward": 0.265625, "reward_std": 0.325631320476532, "rewards/format_reward/mean": 0.46875, "rewards/format_reward/std": 0.21593283116817474, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "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": 1949.0, "completions/max_terminated_length": 1949.0, "completions/mean_length": 439.6875, "completions/mean_terminated_length": 439.6875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0325, "frac_reward_zero_std": 0.0, "grad_norm": 17.293394088745117, "learning_rate": 9.866632986240029e-07, "loss": -0.0, "num_tokens": 1704590.0, "reward": 0.32734373211860657, "reward_std": 0.42422762513160706, "rewards/format_reward/mean": 0.4609375, "rewards/format_reward/std": 0.28597769141197205, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 13 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1491.0, "completions/max_terminated_length": 1491.0, "completions/mean_length": 370.171875, "completions/mean_terminated_length": 370.171875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.035, "frac_reward_zero_std": 0.125, "grad_norm": 7.3942999839782715, "learning_rate": 9.826044551386742e-07, "loss": -0.0, "num_tokens": 1833185.0, "reward": 0.36406248807907104, "reward_std": 0.291256308555603, "rewards/format_reward/mean": 0.515625, "rewards/format_reward/std": 0.26679685711860657, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 1148.0, "completions/max_terminated_length": 1148.0, "completions/mean_length": 287.953125, "completions/mean_terminated_length": 287.953125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0375, "frac_reward_zero_std": 0.25, "grad_norm": 10.878680229187012, "learning_rate": 9.780178907671788e-07, "loss": 0.0, "num_tokens": 1941766.0, "reward": 0.30937498807907104, "reward_std": 0.30712881684303284, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.24397502839565277, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1383.0, "completions/mean_length": 334.140625, "completions/mean_terminated_length": 278.8548278808594, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.04, "frac_reward_zero_std": 0.125, "grad_norm": 12.395663261413574, "learning_rate": 9.729086208503173e-07, "loss": 0.0, "num_tokens": 2072911.0, "reward": 0.4546874761581421, "reward_std": 0.31789684295654297, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.23517554998397827, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 16 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1632.0, "completions/max_terminated_length": 1632.0, "completions/mean_length": 136.171875, "completions/mean_terminated_length": 136.171875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0425, "frac_reward_zero_std": 0.375, "grad_norm": 11.112771034240723, "learning_rate": 9.672822322997304e-07, "loss": -0.0, "num_tokens": 2171266.0, "reward": 0.33203125, "reward_std": 0.21000537276268005, "rewards/format_reward/mean": 0.5078125, "rewards/format_reward/std": 0.1406387835741043, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "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": 1100.0, "completions/max_terminated_length": 1100.0, "completions/mean_length": 171.21875, "completions/mean_terminated_length": 171.21875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.045, "frac_reward_zero_std": 0.625, "grad_norm": 2.1026268005371094, "learning_rate": 9.611448774886923e-07, "loss": 0.0, "num_tokens": 2287912.0, "reward": 0.20546874403953552, "reward_std": 0.09102033078670502, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.10882443934679031, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "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": 551.0, "completions/max_terminated_length": 551.0, "completions/mean_length": 68.375, "completions/mean_terminated_length": 68.375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0475, "frac_reward_zero_std": 0.375, "grad_norm": 16.021671295166016, "learning_rate": 9.545032675245813e-07, "loss": -0.0, "num_tokens": 2378208.0, "reward": 0.46875, "reward_std": 0.29339390993118286, "rewards/format_reward/mean": 0.625, "rewards/format_reward/std": 0.2357022762298584, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 1008.0, "completions/max_terminated_length": 1008.0, "completions/mean_length": 107.71875, "completions/mean_terminated_length": 107.71875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05, "frac_reward_zero_std": 0.5, "grad_norm": 12.5459623336792, "learning_rate": 9.473646649103817e-07, "loss": 0.0, "num_tokens": 2515638.0, "reward": 0.5984374284744263, "reward_std": 0.21166592836380005, "rewards/format_reward/mean": 0.515625, "rewards/format_reward/std": 0.08768405020236969, "rewards/mcq_exact_match_reward/mean": 0.546875, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 779.0, "completions/max_terminated_length": 779.0, "completions/mean_length": 47.15625, "completions/mean_terminated_length": 47.15625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0525, "frac_reward_zero_std": 0.75, "grad_norm": 30.497756958007812, "learning_rate": 9.397368756032444e-07, "loss": 0.0, "num_tokens": 2589528.0, "reward": 0.5382812023162842, "reward_std": 0.047428861260414124, "rewards/format_reward/mean": 0.5390625, "rewards/format_reward/std": 0.13524486124515533, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1931.0, "completions/mean_length": 173.125, "completions/mean_terminated_length": 143.36508178710938, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.055, "frac_reward_zero_std": 0.375, "grad_norm": 21.298641204833984, "learning_rate": 9.316282404787869e-07, "loss": 0.0, "num_tokens": 2702176.0, "reward": 0.22343751788139343, "reward_std": 0.0986991822719574, "rewards/format_reward/mean": 0.671875, "rewards/format_reward/std": 0.25539806485176086, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "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": 1177.0, "completions/max_terminated_length": 1177.0, "completions/mean_length": 114.453125, "completions/mean_terminated_length": 114.453125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0575, "frac_reward_zero_std": 0.375, "grad_norm": 19.646657943725586, "learning_rate": 9.230476262104676e-07, "loss": -0.0, "num_tokens": 2797717.0, "reward": 0.44609373807907104, "reward_std": 0.2525854706764221, "rewards/format_reward/mean": 0.5546875, "rewards/format_reward/std": 0.20152568817138672, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 705.0, "completions/max_terminated_length": 705.0, "completions/mean_length": 93.1875, "completions/mean_terminated_length": 93.1875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06, "frac_reward_zero_std": 0.0, "grad_norm": 39.375282287597656, "learning_rate": 9.1400441557401e-07, "loss": 0.0, "num_tokens": 2887681.0, "reward": 0.33437496423721313, "reward_std": 0.24721816182136536, "rewards/format_reward/mean": 0.6875, "rewards/format_reward/std": 0.2597312331199646, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 392.0, "completions/max_terminated_length": 392.0, "completions/mean_length": 21.890625, "completions/mean_terminated_length": 21.890625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0625, "frac_reward_zero_std": 0.25, "grad_norm": 44.67606735229492, "learning_rate": 9.045084971874737e-07, "loss": 0.0, "num_tokens": 3012890.0, "reward": 0.4125000238418579, "reward_std": 0.26012107729911804, "rewards/format_reward/mean": 0.6875, "rewards/format_reward/std": 0.24397502839565277, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1383.0, "completions/mean_length": 179.3125, "completions/mean_terminated_length": 119.03225708007812, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.065, "frac_reward_zero_std": 0.5, "grad_norm": 6.391251087188721, "learning_rate": 8.945702546981968e-07, "loss": 0.0, "num_tokens": 3128766.0, "reward": 0.22499999403953552, "reward_std": 0.16906382143497467, "rewards/format_reward/mean": 0.53125, "rewards/format_reward/std": 0.21593283116817474, "rewards/mcq_exact_match_reward/mean": 0.171875, "rewards/mcq_exact_match_reward/std": 0.38025420904159546, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 718.0, "completions/mean_length": 110.390625, "completions/mean_terminated_length": 47.8870964050293, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0675, "frac_reward_zero_std": 0.25, "grad_norm": 20.161277770996094, "learning_rate": 8.842005554284295e-07, "loss": -0.0, "num_tokens": 3233959.0, "reward": 0.44218748807907104, "reward_std": 0.18106591701507568, "rewards/format_reward/mean": 0.828125, "rewards/format_reward/std": 0.27048972249031067, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 1012.0, "completions/max_terminated_length": 1012.0, "completions/mean_length": 51.21875, "completions/mean_terminated_length": 51.21875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07, "frac_reward_zero_std": 0.25, "grad_norm": 16.92597007751465, "learning_rate": 8.734107384920769e-07, "loss": -0.0, "num_tokens": 3331741.0, "reward": 0.4164062738418579, "reward_std": 0.23654812574386597, "rewards/format_reward/mean": 0.8828125, "rewards/format_reward/std": 0.21347814798355103, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 402.0, "completions/max_terminated_length": 402.0, "completions/mean_length": 24.859375, "completions/mean_terminated_length": 24.859375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0725, "frac_reward_zero_std": 0.125, "grad_norm": 55.0001335144043, "learning_rate": 8.622126023955445e-07, "loss": -0.0, "num_tokens": 3443268.0, "reward": 0.3148437440395355, "reward_std": 0.2504793107509613, "rewards/format_reward/mean": 0.8046875, "rewards/format_reward/std": 0.24587368965148926, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 682.0, "completions/mean_length": 71.015625, "completions/mean_terminated_length": 39.63492202758789, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.075, "frac_reward_zero_std": 0.25, "grad_norm": 12.861172676086426, "learning_rate": 8.506183921362442e-07, "loss": 0.0, "num_tokens": 3536253.0, "reward": 0.46406251192092896, "reward_std": 0.2636348605155945, "rewards/format_reward/mean": 0.890625, "rewards/format_reward/std": 0.24346621334552765, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 30 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 410.0, "completions/mean_length": 61.015625, "completions/mean_terminated_length": 29.476192474365234, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0775, "frac_reward_zero_std": 0.25, "grad_norm": 22.783077239990234, "learning_rate": 8.386407858128706e-07, "loss": 0.0, "num_tokens": 3634966.0, "reward": 0.16718751192092896, "reward_std": 0.12469000369310379, "rewards/format_reward/mean": 0.890625, "rewards/format_reward/std": 0.24346621334552765, "rewards/mcq_exact_match_reward/mean": 0.078125, "rewards/mcq_exact_match_reward/std": 0.27048972249031067, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 981.0, "completions/mean_length": 71.484375, "completions/mean_terminated_length": 40.111114501953125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.08, "frac_reward_zero_std": 0.375, "grad_norm": 8.757485389709473, "learning_rate": 8.262928807620843e-07, "loss": 0.0, "num_tokens": 3719477.0, "reward": 0.51953125, "reward_std": 0.19523507356643677, "rewards/format_reward/mean": 0.9765625, "rewards/format_reward/std": 0.13886408507823944, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 32 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 20.0, "completions/max_terminated_length": 20.0, "completions/mean_length": 12.71875, "completions/mean_terminated_length": 12.71875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0825, "frac_reward_zero_std": 0.625, "grad_norm": 7.886214256286621, "learning_rate": 8.135881792367685e-07, "loss": -0.0, "num_tokens": 3833627.0, "reward": 0.6156250238418579, "reward_std": 0.13258251547813416, "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": 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": 169.0, "completions/max_terminated_length": 169.0, "completions/mean_length": 19.78125, "completions/mean_terminated_length": 19.78125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.085, "frac_reward_zero_std": 0.375, "grad_norm": 15.526958465576172, "learning_rate": 8.005405736415125e-07, "loss": 0.0, "num_tokens": 3929165.0, "reward": 0.49687501788139343, "reward_std": 0.21885380148887634, "rewards/format_reward/mean": 0.90625, "rewards/format_reward/std": 0.233588308095932, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 314.0, "completions/max_terminated_length": 314.0, "completions/mean_length": 21.140625, "completions/mean_terminated_length": 21.140625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0875, "frac_reward_zero_std": 0.5, "grad_norm": 13.258960723876953, "learning_rate": 7.871643313414718e-07, "loss": -0.0, "num_tokens": 4035982.0, "reward": 0.2710937559604645, "reward_std": 0.20053395628929138, "rewards/format_reward/mean": 0.9921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.171875, "rewards/mcq_exact_match_reward/std": 0.38025420904159546, "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": 206.0, "completions/max_terminated_length": 206.0, "completions/mean_length": 25.46875, "completions/mean_terminated_length": 25.46875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.09, "frac_reward_zero_std": 0.625, "grad_norm": 6.211149215698242, "learning_rate": 7.734740790612136e-07, "loss": -0.0, "num_tokens": 4121124.0, "reward": 0.39375001192092896, "reward_std": 0.13407263159751892, "rewards/format_reward/mean": 0.96875, "rewards/format_reward/std": 0.1510545015335083, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "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": 83.0, "completions/max_terminated_length": 83.0, "completions/mean_length": 14.0, "completions/mean_terminated_length": 14.0, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0925, "frac_reward_zero_std": 0.375, "grad_norm": 13.185394287109375, "learning_rate": 7.594847868906076e-07, "loss": -0.0, "num_tokens": 4203380.0, "reward": 0.4125000238418579, "reward_std": 0.25513991713523865, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.3125, "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": 255.0, "completions/max_terminated_length": 255.0, "completions/mean_length": 23.484375, "completions/mean_terminated_length": 23.484375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.095, "frac_reward_zero_std": 0.25, "grad_norm": 13.131759643554688, "learning_rate": 7.452117519152541e-07, "loss": 0.0, "num_tokens": 4306307.0, "reward": 0.6390625238418579, "reward_std": 0.30261802673339844, "rewards/format_reward/mean": 0.921875, "rewards/format_reward/std": 0.27048972249031067, "rewards/mcq_exact_match_reward/mean": 0.546875, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 194.0, "completions/max_terminated_length": 194.0, "completions/mean_length": 15.859375, "completions/mean_terminated_length": 15.859375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.0975, "frac_reward_zero_std": 0.125, "grad_norm": 16.97125816345215, "learning_rate": 7.306705814893439e-07, "loss": -0.0, "num_tokens": 4411434.0, "reward": 0.9117187261581421, "reward_std": 0.3241649270057678, "rewards/format_reward/mean": 0.9921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.8125, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "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": 150.0, "completions/max_terminated_length": 150.0, "completions/mean_length": 19.546875, "completions/mean_terminated_length": 19.546875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1, "frac_reward_zero_std": 0.25, "grad_norm": 12.075810432434082, "learning_rate": 7.158771761692464e-07, "loss": -0.0, "num_tokens": 4486165.0, "reward": 0.6156250238418579, "reward_std": 0.3356248140335083, "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": 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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.828125, "completions/mean_terminated_length": 12.828125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1025, "frac_reward_zero_std": 0.875, "grad_norm": 5.789536952972412, "learning_rate": 7.008477123264847e-07, "loss": 0.0, "num_tokens": 4576306.0, "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": 41 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 202.0, "completions/mean_length": 53.375, "completions/mean_terminated_length": 21.71428680419922, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.105, "frac_reward_zero_std": 0.625, "grad_norm": 4.968082427978516, "learning_rate": 6.855986244591103e-07, "loss": 0.0, "num_tokens": 4663722.0, "reward": 0.6921875476837158, "reward_std": 0.11330723762512207, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.59375, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 85.0, "completions/max_terminated_length": 85.0, "completions/mean_length": 14.09375, "completions/mean_terminated_length": 14.09375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1075, "frac_reward_zero_std": 0.75, "grad_norm": 8.742327690124512, "learning_rate": 6.701465872208216e-07, "loss": 0.0, "num_tokens": 4751688.0, "reward": 0.4750000238418579, "reward_std": 0.1157275140285492, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 12.90625, "completions/mean_terminated_length": 12.90625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.11, "frac_reward_zero_std": 0.25, "grad_norm": 13.507596969604492, "learning_rate": 6.545084971874736e-07, "loss": 0.0, "num_tokens": 4867642.0, "reward": 0.5843750238418579, "reward_std": 0.3266732692718506, "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": 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": 175.0, "completions/max_terminated_length": 175.0, "completions/mean_length": 16.328125, "completions/mean_terminated_length": 16.328125, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.1125, "frac_reward_zero_std": 0.625, "grad_norm": 14.993759155273438, "learning_rate": 6.387014543809223e-07, "loss": -0.0, "num_tokens": 4938135.0, "reward": 0.42656251788139343, "reward_std": 0.16290666162967682, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.125, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 45 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 225.0, "completions/mean_length": 51.875, "completions/mean_terminated_length": 20.19047737121582, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.115, "frac_reward_zero_std": 0.25, "grad_norm": 9.617093086242676, "learning_rate": 6.227427435703995e-07, "loss": 0.0, "num_tokens": 5047567.0, "reward": 0.784375011920929, "reward_std": 0.34824109077453613, "rewards/format_reward/mean": 0.96875, "rewards/format_reward/std": 0.17536810040473938, "rewards/mcq_exact_match_reward/mean": 0.6875, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 13.0, "completions/mean_terminated_length": 13.0, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.1175, "frac_reward_zero_std": 0.875, "grad_norm": 3.811880111694336, "learning_rate": 6.066498153718734e-07, "loss": 0.0, "num_tokens": 5135175.0, "reward": 0.3031250238418579, "reward_std": 0.0646936446428299, "rewards/format_reward/mean": 1.0, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "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": 418.0, "completions/max_terminated_length": 418.0, "completions/mean_length": 21.28125, "completions/mean_terminated_length": 21.28125, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.12, "frac_reward_zero_std": 0.5, "grad_norm": 10.27923583984375, "learning_rate": 5.90440267166055e-07, "loss": -0.0, "num_tokens": 5246865.0, "reward": 0.48984378576278687, "reward_std": 0.22760312259197235, "rewards/format_reward/mean": 0.9921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 206.0, "completions/max_terminated_length": 206.0, "completions/mean_length": 16.125, "completions/mean_terminated_length": 16.125, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.1225, "frac_reward_zero_std": 0.375, "grad_norm": 27.7751522064209, "learning_rate": 5.741318238559209e-07, "loss": -0.0, "num_tokens": 5321729.0, "reward": 0.5218750238418579, "reward_std": 0.25726157426834106, "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": 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": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 13.0, "completions/mean_terminated_length": 13.0, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "epoch": 0.125, "frac_reward_zero_std": 0.625, "grad_norm": 7.580718994140625, "learning_rate": 5.577423184847931e-07, "loss": 0.0, "num_tokens": 5438025.0, "reward": 0.3656250238418579, "reward_std": 0.17358146607875824, "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": 50 } ], "logging_steps": 1, "max_steps": 100, "num_input_tokens_seen": 5438025, "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 }