{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.25, "eval_steps": 500, "global_step": 441, "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": 1433.0, "completions/max_terminated_length": 1433.0, "completions/mean_length": 787.265625, "completions/mean_terminated_length": 787.265625, "completions/min_length": 415.0, "completions/min_terminated_length": 415.0, "entropy": 0.7912661358714104, "epoch": 0.0005668934240362812, "frac_reward_zero_std": 0.75, "grad_norm": 0.3359375, "kl": 0.0005471410513564479, "learning_rate": 1e-06, "loss": -0.0145, "num_tokens": 58849.0, "reward": 0.09375, "reward_std": 0.426084041595459, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 1, "step_time": 55.261071532033384 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1452.0, "completions/max_terminated_length": 1452.0, "completions/mean_length": 712.390625, "completions/mean_terminated_length": 712.390625, "completions/min_length": 277.0, "completions/min_terminated_length": 277.0, "entropy": 0.8053555078804493, "epoch": 0.0011337868480725624, "frac_reward_zero_std": 0.375, "grad_norm": 0.6015625, "kl": 0.0005988259563309839, "learning_rate": 9.994331065759637e-07, "loss": -0.008, "num_tokens": 116154.0, "reward": 0.1875, "reward_std": 0.5875696539878845, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 2, "step_time": 56.21459399070591 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1385.0, "completions/max_terminated_length": 1385.0, "completions/mean_length": 705.8125, "completions/mean_terminated_length": 705.8125, "completions/min_length": 272.0, "completions/min_terminated_length": 272.0, "entropy": 0.9432813562452793, "epoch": 0.0017006802721088435, "frac_reward_zero_std": 0.625, "grad_norm": 0.40625, "kl": 0.0006910295905981911, "learning_rate": 9.988662131519275e-07, "loss": -0.0029, "num_tokens": 170830.0, "reward": 0.09375, "reward_std": 0.426084041595459, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 3, "step_time": 53.85885674227029 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1959.0, "completions/max_terminated_length": 1959.0, "completions/mean_length": 756.921875, "completions/mean_terminated_length": 756.921875, "completions/min_length": 337.0, "completions/min_terminated_length": 337.0, "entropy": 1.0130716506391764, "epoch": 0.0022675736961451248, "frac_reward_zero_std": 0.875, "grad_norm": 0.2158203125, "kl": 0.0007245079577842262, "learning_rate": 9.982993197278912e-07, "loss": -0.0027, "num_tokens": 229481.0, "reward": 0.0625, "reward_std": 0.35073620080947876, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 4, "step_time": 76.4772049030289 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1817.0, "completions/max_terminated_length": 1817.0, "completions/mean_length": 806.5, "completions/mean_terminated_length": 806.5, "completions/min_length": 342.0, "completions/min_terminated_length": 342.0, "entropy": 0.9002897161990404, "epoch": 0.002834467120181406, "frac_reward_zero_std": 1.0, "grad_norm": 0.000316619873046875, "kl": 0.0006696486689179437, "learning_rate": 9.977324263038547e-07, "loss": 0.0, "num_tokens": 290833.0, "reward": 0.0, "reward_std": 0.0, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 5, "step_time": 70.04715261328965 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1433.0, "completions/max_terminated_length": 1433.0, "completions/mean_length": 781.296875, "completions/mean_terminated_length": 781.296875, "completions/min_length": 100.0, "completions/min_terminated_length": 100.0, "entropy": 0.6070616748183966, "epoch": 0.003401360544217687, "frac_reward_zero_std": 0.875, "grad_norm": 0.265625, "kl": 0.000568444460441242, "learning_rate": 9.971655328798185e-07, "loss": -0.0185, "num_tokens": 351660.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 6, "step_time": 55.606025882065296 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1308.0, "completions/max_terminated_length": 1308.0, "completions/mean_length": 745.3125, "completions/mean_terminated_length": 745.3125, "completions/min_length": 459.0, "completions/min_terminated_length": 459.0, "entropy": 0.8628565222024918, "epoch": 0.003968253968253968, "frac_reward_zero_std": 0.75, "grad_norm": 0.353515625, "kl": 0.0006363487227645237, "learning_rate": 9.965986394557822e-07, "loss": 0.0025, "num_tokens": 412128.0, "reward": 0.09375, "reward_std": 0.426084041595459, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 7, "step_time": 51.90570120885968 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1151.0, "completions/max_terminated_length": 1151.0, "completions/mean_length": 666.1875, "completions/mean_terminated_length": 666.1875, "completions/min_length": 238.0, "completions/min_terminated_length": 238.0, "entropy": 0.782693337649107, "epoch": 0.0045351473922902496, "frac_reward_zero_std": 0.875, "grad_norm": 0.39453125, "kl": 0.0006672325453109806, "learning_rate": 9.96031746031746e-07, "loss": -0.0049, "num_tokens": 466628.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 8, "step_time": 44.9562089079991 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1280.0, "completions/max_terminated_length": 1280.0, "completions/mean_length": 789.53125, "completions/mean_terminated_length": 789.53125, "completions/min_length": 453.0, "completions/min_terminated_length": 453.0, "entropy": 0.7709263563156128, "epoch": 0.00510204081632653, "frac_reward_zero_std": 0.875, "grad_norm": 0.20703125, "kl": 0.000592712800425943, "learning_rate": 9.954648526077097e-07, "loss": 0.0029, "num_tokens": 528558.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 9, "step_time": 50.013049360364676 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1361.0, "completions/max_terminated_length": 1361.0, "completions/mean_length": 834.765625, "completions/mean_terminated_length": 834.765625, "completions/min_length": 501.0, "completions/min_terminated_length": 501.0, "entropy": 0.8164468556642532, "epoch": 0.005668934240362812, "frac_reward_zero_std": 0.75, "grad_norm": 0.328125, "kl": 0.0006466719878517324, "learning_rate": 9.948979591836735e-07, "loss": 0.0143, "num_tokens": 591359.0, "reward": 0.0625, "reward_std": 0.35073620080947876, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 10, "step_time": 53.13434873986989 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1307.0, "completions/max_terminated_length": 1307.0, "completions/mean_length": 819.125, "completions/mean_terminated_length": 819.125, "completions/min_length": 415.0, "completions/min_terminated_length": 415.0, "entropy": 0.8207502197474241, "epoch": 0.006235827664399093, "frac_reward_zero_std": 0.75, "grad_norm": 0.35546875, "kl": 0.0006157456555229146, "learning_rate": 9.94331065759637e-07, "loss": -0.0009, "num_tokens": 653527.0, "reward": 0.0625, "reward_std": 0.35073620080947876, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 11, "step_time": 51.11214052326977 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1524.0, "completions/max_terminated_length": 1524.0, "completions/mean_length": 818.0625, "completions/mean_terminated_length": 818.0625, "completions/min_length": 302.0, "completions/min_terminated_length": 302.0, "entropy": 0.7241049911826849, "epoch": 0.006802721088435374, "frac_reward_zero_std": 0.625, "grad_norm": 0.41015625, "kl": 0.0005700781130144605, "learning_rate": 9.937641723356008e-07, "loss": -0.0101, "num_tokens": 715875.0, "reward": 0.125, "reward_std": 0.48795002698898315, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 12, "step_time": 59.11056109983474 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1607.0, "completions/max_terminated_length": 1607.0, "completions/mean_length": 912.9375, "completions/mean_terminated_length": 912.9375, "completions/min_length": 265.0, "completions/min_terminated_length": 265.0, "entropy": 0.7451053559780121, "epoch": 0.007369614512471655, "frac_reward_zero_std": 0.875, "grad_norm": 0.1845703125, "kl": 0.000535154516910552, "learning_rate": 9.931972789115645e-07, "loss": -0.0058, "num_tokens": 785487.0, "reward": 0.0625, "reward_std": 0.35073620080947876, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 13, "step_time": 61.979948569089174 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1222.0, "completions/max_terminated_length": 1222.0, "completions/mean_length": 743.3125, "completions/mean_terminated_length": 743.3125, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "entropy": 0.8540959488600492, "epoch": 0.007936507936507936, "frac_reward_zero_std": 0.75, "grad_norm": 0.333984375, "kl": 0.0006348256429191679, "learning_rate": 9.926303854875283e-07, "loss": 0.0172, "num_tokens": 844387.0, "reward": 0.0625, "reward_std": 0.35073620080947876, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 14, "step_time": 48.2400392210111 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2454.0, "completions/max_terminated_length": 2454.0, "completions/mean_length": 789.453125, "completions/mean_terminated_length": 789.453125, "completions/min_length": 254.0, "completions/min_terminated_length": 254.0, "entropy": 0.6644097231328487, "epoch": 0.008503401360544218, "frac_reward_zero_std": 0.75, "grad_norm": 0.3359375, "kl": 0.0006174067239044234, "learning_rate": 9.92063492063492e-07, "loss": -0.0117, "num_tokens": 905664.0, "reward": 0.09375, "reward_std": 0.426084041595459, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 15, "step_time": 97.98640740290284 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1459.0, "completions/max_terminated_length": 1459.0, "completions/mean_length": 778.0625, "completions/mean_terminated_length": 778.0625, "completions/min_length": 400.0, "completions/min_terminated_length": 400.0, "entropy": 0.8444338124245405, "epoch": 0.009070294784580499, "frac_reward_zero_std": 0.625, "grad_norm": 0.4453125, "kl": 0.0007286928284884198, "learning_rate": 9.914965986394558e-07, "loss": 0.0031, "num_tokens": 965404.0, "reward": 0.125, "reward_std": 0.48795002698898315, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 16, "step_time": 56.34489590395242 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1568.0, "completions/max_terminated_length": 1568.0, "completions/mean_length": 801.015625, "completions/mean_terminated_length": 801.015625, "completions/min_length": 224.0, "completions/min_terminated_length": 224.0, "entropy": 0.967936534434557, "epoch": 0.00963718820861678, "frac_reward_zero_std": 0.875, "grad_norm": 0.48046875, "kl": 0.0007015223854978103, "learning_rate": 9.909297052154193e-07, "loss": -0.0194, "num_tokens": 1027645.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 17, "step_time": 61.04807883966714 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2414.0, "completions/max_terminated_length": 2414.0, "completions/mean_length": 824.53125, "completions/mean_terminated_length": 824.53125, "completions/min_length": 306.0, "completions/min_terminated_length": 306.0, "entropy": 0.6849733348935843, "epoch": 0.01020408163265306, "frac_reward_zero_std": 0.75, "grad_norm": 0.306640625, "kl": 0.0005724504480895121, "learning_rate": 9.90362811791383e-07, "loss": -0.0081, "num_tokens": 1090127.0, "reward": 0.125, "reward_std": 0.48795002698898315, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 18, "step_time": 95.73333193920553 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1571.0, "completions/max_terminated_length": 1571.0, "completions/mean_length": 725.53125, "completions/mean_terminated_length": 725.53125, "completions/min_length": 402.0, "completions/min_terminated_length": 402.0, "entropy": 1.071934748440981, "epoch": 0.010770975056689343, "frac_reward_zero_std": 0.75, "grad_norm": 0.380859375, "kl": 0.0007020830926194321, "learning_rate": 9.897959183673468e-07, "loss": -0.0168, "num_tokens": 1146105.0, "reward": 0.09375, "reward_std": 0.426084041595459, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 19, "step_time": 61.04049335233867 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1878.0, "completions/max_terminated_length": 1878.0, "completions/mean_length": 790.359375, "completions/mean_terminated_length": 790.359375, "completions/min_length": 315.0, "completions/min_terminated_length": 315.0, "entropy": 0.7894083205610514, "epoch": 0.011337868480725623, "frac_reward_zero_std": 0.625, "grad_norm": 0.365234375, "kl": 0.000633090885457932, "learning_rate": 9.892290249433106e-07, "loss": 0.0443, "num_tokens": 1206824.0, "reward": 0.15625, "reward_std": 0.5409793853759766, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 20, "step_time": 73.91199010517448 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1718.0, "completions/max_terminated_length": 1718.0, "completions/mean_length": 854.046875, "completions/mean_terminated_length": 854.046875, "completions/min_length": 296.0, "completions/min_terminated_length": 296.0, "entropy": 0.7625202462077141, "epoch": 0.011904761904761904, "frac_reward_zero_std": 0.875, "grad_norm": 0.201171875, "kl": 0.0006028276457072934, "learning_rate": 9.886621315192743e-07, "loss": 0.0006, "num_tokens": 1274483.0, "reward": 0.0625, "reward_std": 0.35073620080947876, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 21, "step_time": 68.21598854195327 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1261.0, "completions/max_terminated_length": 1261.0, "completions/mean_length": 692.296875, "completions/mean_terminated_length": 692.296875, "completions/min_length": 148.0, "completions/min_terminated_length": 148.0, "entropy": 0.8724542260169983, "epoch": 0.012471655328798186, "frac_reward_zero_std": 0.625, "grad_norm": 0.458984375, "kl": 0.0008029471355257556, "learning_rate": 9.88095238095238e-07, "loss": -0.0082, "num_tokens": 1337598.0, "reward": 0.1953125, "reward_std": 0.5883605480194092, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 22, "step_time": 52.92388445697725 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 4096.0, "completions/max_terminated_length": 1279.0, "completions/mean_length": 820.921875, "completions/mean_terminated_length": 768.9365234375, "completions/min_length": 206.0, "completions/min_terminated_length": 206.0, "entropy": 0.7689987067133188, "epoch": 0.013038548752834467, "frac_reward_zero_std": 0.875, "grad_norm": 0.2060546875, "kl": 0.0006323507077468093, "learning_rate": 9.875283446712018e-07, "loss": 0.007, "num_tokens": 1401201.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 23, "step_time": 181.87214892730117 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1296.0, "completions/max_terminated_length": 1296.0, "completions/mean_length": 679.171875, "completions/mean_terminated_length": 679.171875, "completions/min_length": 321.0, "completions/min_terminated_length": 321.0, "entropy": 1.0151823163032532, "epoch": 0.013605442176870748, "frac_reward_zero_std": 0.875, "grad_norm": 0.34765625, "kl": 0.0006834342384536285, "learning_rate": 9.869614512471656e-07, "loss": -0.0117, "num_tokens": 1455220.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 24, "step_time": 50.659633063711226 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1289.0, "completions/max_terminated_length": 1289.0, "completions/mean_length": 779.484375, "completions/mean_terminated_length": 779.484375, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "entropy": 0.8662557099014521, "epoch": 0.01417233560090703, "frac_reward_zero_std": 1.0, "grad_norm": 0.000492095947265625, "kl": 0.0006726052361045731, "learning_rate": 9.86394557823129e-07, "loss": 0.0, "num_tokens": 1515651.0, "reward": 0.0, "reward_std": 0.0, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 25, "step_time": 50.44343617465347 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1112.0, "completions/max_terminated_length": 1112.0, "completions/mean_length": 746.453125, "completions/mean_terminated_length": 746.453125, "completions/min_length": 393.0, "completions/min_terminated_length": 393.0, "entropy": 0.8149989545345306, "epoch": 0.01473922902494331, "frac_reward_zero_std": 0.75, "grad_norm": 0.3125, "kl": 0.0006371798408508766, "learning_rate": 9.858276643990928e-07, "loss": 0.0002, "num_tokens": 1572696.0, "reward": 0.0390625, "reward_std": 0.2567298710346222, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 26, "step_time": 44.474265119060874 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1443.0, "completions/max_terminated_length": 1443.0, "completions/mean_length": 715.890625, "completions/mean_terminated_length": 715.890625, "completions/min_length": 298.0, "completions/min_terminated_length": 298.0, "entropy": 1.095861166715622, "epoch": 0.015306122448979591, "frac_reward_zero_std": 1.0, "grad_norm": 0.00019359588623046875, "kl": 0.0007180045668064849, "learning_rate": 9.852607709750566e-07, "loss": 0.0, "num_tokens": 1629089.0, "reward": 0.0, "reward_std": 0.0, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 27, "step_time": 57.39974437747151 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1605.0, "completions/max_terminated_length": 1605.0, "completions/mean_length": 718.109375, "completions/mean_terminated_length": 718.109375, "completions/min_length": 193.0, "completions/min_terminated_length": 193.0, "entropy": 0.826047021895647, "epoch": 0.015873015873015872, "frac_reward_zero_std": 0.75, "grad_norm": 0.50390625, "kl": 0.0006650087561865803, "learning_rate": 9.846938775510203e-07, "loss": -0.0281, "num_tokens": 1684944.0, "reward": 0.125, "reward_std": 0.48795002698898315, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 28, "step_time": 64.38864888530225 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1537.0, "completions/max_terminated_length": 1537.0, "completions/mean_length": 826.625, "completions/mean_terminated_length": 826.625, "completions/min_length": 23.0, "completions/min_terminated_length": 23.0, "entropy": 0.8258813861757517, "epoch": 0.016439909297052153, "frac_reward_zero_std": 0.875, "grad_norm": 0.263671875, "kl": 0.0006447574651247123, "learning_rate": 9.84126984126984e-07, "loss": -0.0015, "num_tokens": 1750912.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 29, "step_time": 60.81136659439653 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1433.0, "completions/max_terminated_length": 1433.0, "completions/mean_length": 775.40625, "completions/mean_terminated_length": 775.40625, "completions/min_length": 329.0, "completions/min_terminated_length": 329.0, "entropy": 0.8379686512053013, "epoch": 0.017006802721088437, "frac_reward_zero_std": 0.875, "grad_norm": 0.24609375, "kl": 0.0006821296829002677, "learning_rate": 9.835600907029478e-07, "loss": -0.0105, "num_tokens": 1812026.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 30, "step_time": 57.01410628668964 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2075.0, "completions/max_terminated_length": 2075.0, "completions/mean_length": 785.953125, "completions/mean_terminated_length": 785.953125, "completions/min_length": 367.0, "completions/min_terminated_length": 367.0, "entropy": 0.915758527815342, "epoch": 0.017573696145124718, "frac_reward_zero_std": 0.875, "grad_norm": 0.29296875, "kl": 0.0006477457645814866, "learning_rate": 9.829931972789116e-07, "loss": 0.0075, "num_tokens": 1872023.0, "reward": 0.0078125, "reward_std": 0.0625, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 31, "step_time": 81.80097121186554 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1282.0, "completions/max_terminated_length": 1282.0, "completions/mean_length": 792.578125, "completions/mean_terminated_length": 792.578125, "completions/min_length": 275.0, "completions/min_terminated_length": 275.0, "entropy": 0.9764898531138897, "epoch": 0.018140589569160998, "frac_reward_zero_std": 0.875, "grad_norm": 0.330078125, "kl": 0.000687534624375985, "learning_rate": 9.824263038548753e-07, "loss": 0.0029, "num_tokens": 1935372.0, "reward": 0.015625, "reward_std": 0.08768405020236969, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 32, "step_time": 51.07413536589593 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1377.0, "completions/max_terminated_length": 1377.0, "completions/mean_length": 720.671875, "completions/mean_terminated_length": 720.671875, "completions/min_length": 319.0, "completions/min_terminated_length": 319.0, "entropy": 0.8088947497308254, "epoch": 0.01870748299319728, "frac_reward_zero_std": 0.5, "grad_norm": 0.51953125, "kl": 0.0006506219633592991, "learning_rate": 9.818594104308389e-07, "loss": 0.008, "num_tokens": 1990167.0, "reward": 0.21875, "reward_std": 0.6291528940200806, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 33, "step_time": 53.4132538670674 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1718.0, "completions/max_terminated_length": 1718.0, "completions/mean_length": 785.0625, "completions/mean_terminated_length": 785.0625, "completions/min_length": 341.0, "completions/min_terminated_length": 341.0, "entropy": 0.6793034747242928, "epoch": 0.01927437641723356, "frac_reward_zero_std": 0.75, "grad_norm": 0.314453125, "kl": 0.0005896855145692825, "learning_rate": 9.812925170068026e-07, "loss": -0.0068, "num_tokens": 2050579.0, "reward": 0.09375, "reward_std": 0.426084041595459, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 34, "step_time": 66.1939322706312 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1336.0, "completions/max_terminated_length": 1336.0, "completions/mean_length": 786.03125, "completions/mean_terminated_length": 786.03125, "completions/min_length": 328.0, "completions/min_terminated_length": 328.0, "entropy": 0.7928113117814064, "epoch": 0.01984126984126984, "frac_reward_zero_std": 0.625, "grad_norm": 0.494140625, "kl": 0.0006361865598591976, "learning_rate": 9.807256235827664e-07, "loss": -0.0042, "num_tokens": 2112461.0, "reward": 0.078125, "reward_std": 0.3587760329246521, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 35, "step_time": 52.14811750501394 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1769.0, "completions/max_terminated_length": 1769.0, "completions/mean_length": 745.390625, "completions/mean_terminated_length": 745.390625, "completions/min_length": 210.0, "completions/min_terminated_length": 210.0, "entropy": 0.9053811207413673, "epoch": 0.02040816326530612, "frac_reward_zero_std": 1.0, "grad_norm": 0.00020885467529296875, "kl": 0.0006936117097211536, "learning_rate": 9.801587301587301e-07, "loss": 0.0, "num_tokens": 2171030.0, "reward": 0.0, "reward_std": 0.0, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 36, "step_time": 68.92622819449753 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1255.0, "completions/max_terminated_length": 1255.0, "completions/mean_length": 770.296875, "completions/mean_terminated_length": 770.296875, "completions/min_length": 420.0, "completions/min_terminated_length": 420.0, "entropy": 0.6799955926835537, "epoch": 0.020975056689342405, "frac_reward_zero_std": 0.5, "grad_norm": 0.44921875, "kl": 0.0005374719130486483, "learning_rate": 9.795918367346939e-07, "loss": -0.0226, "num_tokens": 2231729.0, "reward": 0.375, "reward_std": 0.7867957949638367, "rewards/accuracy_reward_func/mean": 0.375, "rewards/accuracy_reward_func/std": 0.7867957949638367, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 37, "step_time": 49.94026922620833 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1254.0, "completions/max_terminated_length": 1254.0, "completions/mean_length": 778.9375, "completions/mean_terminated_length": 778.9375, "completions/min_length": 341.0, "completions/min_terminated_length": 341.0, "entropy": 0.7798807322978973, "epoch": 0.021541950113378686, "frac_reward_zero_std": 0.875, "grad_norm": 0.24609375, "kl": 0.0005999129698466277, "learning_rate": 9.790249433106576e-07, "loss": -0.0042, "num_tokens": 2291117.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 38, "step_time": 49.6709918724373 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1569.0, "completions/max_terminated_length": 1569.0, "completions/mean_length": 762.765625, "completions/mean_terminated_length": 762.765625, "completions/min_length": 436.0, "completions/min_terminated_length": 436.0, "entropy": 1.0847668685019016, "epoch": 0.022108843537414966, "frac_reward_zero_std": 0.875, "grad_norm": 0.26953125, "kl": 0.0007197252125479281, "learning_rate": 9.784580498866214e-07, "loss": 0.0039, "num_tokens": 2350870.0, "reward": 0.0625, "reward_std": 0.35073620080947876, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 39, "step_time": 61.75452910736203 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1609.0, "completions/max_terminated_length": 1609.0, "completions/mean_length": 710.671875, "completions/mean_terminated_length": 710.671875, "completions/min_length": 216.0, "completions/min_terminated_length": 216.0, "entropy": 1.0219030529260635, "epoch": 0.022675736961451247, "frac_reward_zero_std": 1.0, "grad_norm": 0.00030517578125, "kl": 0.0007319577052840032, "learning_rate": 9.778911564625851e-07, "loss": 0.0, "num_tokens": 2411049.0, "reward": 0.0, "reward_std": 0.0, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 40, "step_time": 64.2758515747264 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1166.0, "completions/max_terminated_length": 1166.0, "completions/mean_length": 692.0, "completions/mean_terminated_length": 692.0, "completions/min_length": 23.0, "completions/min_terminated_length": 23.0, "entropy": 0.9765904024243355, "epoch": 0.023242630385487528, "frac_reward_zero_std": 0.75, "grad_norm": 0.447265625, "kl": 0.0007533810530730989, "learning_rate": 9.773242630385487e-07, "loss": -0.018, "num_tokens": 2465801.0, "reward": 0.125, "reward_std": 0.48795002698898315, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 41, "step_time": 45.983314889483154 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1518.0, "completions/max_terminated_length": 1518.0, "completions/mean_length": 939.328125, "completions/mean_terminated_length": 939.328125, "completions/min_length": 475.0, "completions/min_terminated_length": 475.0, "entropy": 0.8025281708687544, "epoch": 0.023809523809523808, "frac_reward_zero_std": 0.75, "grad_norm": 0.298828125, "kl": 0.0006058815342839807, "learning_rate": 9.767573696145124e-07, "loss": 0.0009, "num_tokens": 2537286.0, "reward": 0.0625, "reward_std": 0.35073620080947876, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 42, "step_time": 61.44131015893072 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1457.0, "completions/max_terminated_length": 1457.0, "completions/mean_length": 734.328125, "completions/mean_terminated_length": 734.328125, "completions/min_length": 313.0, "completions/min_terminated_length": 313.0, "entropy": 0.739720705896616, "epoch": 0.024376417233560092, "frac_reward_zero_std": 0.625, "grad_norm": 0.42578125, "kl": 0.0006070303206797689, "learning_rate": 9.761904761904762e-07, "loss": 0.0228, "num_tokens": 2594475.0, "reward": 0.21875, "reward_std": 0.6291528940200806, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 43, "step_time": 57.56407857872546 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1369.0, "completions/max_terminated_length": 1369.0, "completions/mean_length": 799.140625, "completions/mean_terminated_length": 799.140625, "completions/min_length": 77.0, "completions/min_terminated_length": 77.0, "entropy": 0.8235787898302078, "epoch": 0.024943310657596373, "frac_reward_zero_std": 0.75, "grad_norm": 0.3671875, "kl": 0.0006626893518841825, "learning_rate": 9.7562358276644e-07, "loss": 0.0062, "num_tokens": 2655684.0, "reward": 0.0625, "reward_std": 0.35073620080947876, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 44, "step_time": 53.67061932384968 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1431.0, "completions/max_terminated_length": 1431.0, "completions/mean_length": 768.5, "completions/mean_terminated_length": 768.5, "completions/min_length": 312.0, "completions/min_terminated_length": 312.0, "entropy": 1.153789233416319, "epoch": 0.025510204081632654, "frac_reward_zero_std": 1.0, "grad_norm": 0.000347137451171875, "kl": 0.0007488102091883775, "learning_rate": 9.750566893424036e-07, "loss": 0.0, "num_tokens": 2716276.0, "reward": 0.0, "reward_std": 0.0, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 45, "step_time": 56.18650230206549 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1416.0, "completions/max_terminated_length": 1416.0, "completions/mean_length": 771.671875, "completions/mean_terminated_length": 771.671875, "completions/min_length": 156.0, "completions/min_terminated_length": 156.0, "entropy": 0.7739651259034872, "epoch": 0.026077097505668934, "frac_reward_zero_std": 1.0, "grad_norm": 0.0002288818359375, "kl": 0.0006726973533659475, "learning_rate": 9.744897959183674e-07, "loss": 0.0, "num_tokens": 2776359.0, "reward": 0.0, "reward_std": 0.0, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 46, "step_time": 55.43411583080888 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1265.0, "completions/max_terminated_length": 1265.0, "completions/mean_length": 753.828125, "completions/mean_terminated_length": 753.828125, "completions/min_length": 384.0, "completions/min_terminated_length": 384.0, "entropy": 0.7858696226030588, "epoch": 0.026643990929705215, "frac_reward_zero_std": 0.875, "grad_norm": 0.15625, "kl": 0.0006061669355403865, "learning_rate": 9.739229024943311e-07, "loss": 0.0046, "num_tokens": 2837924.0, "reward": 0.1015625, "reward_std": 0.42891231179237366, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 47, "step_time": 50.96035030670464 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1856.0, "completions/max_terminated_length": 1856.0, "completions/mean_length": 832.484375, "completions/mean_terminated_length": 832.484375, "completions/min_length": 113.0, "completions/min_terminated_length": 113.0, "entropy": 0.6921264715492725, "epoch": 0.027210884353741496, "frac_reward_zero_std": 0.75, "grad_norm": 0.314453125, "kl": 0.0005807866637042025, "learning_rate": 9.733560090702947e-07, "loss": 0.0138, "num_tokens": 2900523.0, "reward": 0.09375, "reward_std": 0.426084041595459, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 48, "step_time": 72.53957574628294 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1209.0, "completions/max_terminated_length": 1209.0, "completions/mean_length": 690.640625, "completions/mean_terminated_length": 690.640625, "completions/min_length": 362.0, "completions/min_terminated_length": 362.0, "entropy": 0.9992184825241566, "epoch": 0.027777777777777776, "frac_reward_zero_std": 0.625, "grad_norm": 0.53515625, "kl": 0.0007191828335635364, "learning_rate": 9.727891156462584e-07, "loss": 0.0092, "num_tokens": 2954092.0, "reward": 0.125, "reward_std": 0.48795002698898315, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 49, "step_time": 47.970756710506976 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1233.0, "completions/max_terminated_length": 1233.0, "completions/mean_length": 763.9375, "completions/mean_terminated_length": 763.9375, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "entropy": 0.8453336097300053, "epoch": 0.02834467120181406, "frac_reward_zero_std": 0.875, "grad_norm": 0.27734375, "kl": 0.0006714449973515002, "learning_rate": 9.722222222222222e-07, "loss": -0.006, "num_tokens": 3013360.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 50, "step_time": 49.39462559018284 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1286.0, "completions/max_terminated_length": 1286.0, "completions/mean_length": 784.921875, "completions/mean_terminated_length": 784.921875, "completions/min_length": 355.0, "completions/min_terminated_length": 355.0, "entropy": 0.7711440324783325, "epoch": 0.02891156462585034, "frac_reward_zero_std": 0.625, "grad_norm": 0.431640625, "kl": 0.0006841422764409799, "learning_rate": 9.71655328798186e-07, "loss": -0.0161, "num_tokens": 3077283.0, "reward": 0.125, "reward_std": 0.48795002698898315, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 51, "step_time": 54.76244755461812 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2038.0, "completions/max_terminated_length": 2038.0, "completions/mean_length": 824.328125, "completions/mean_terminated_length": 824.328125, "completions/min_length": 262.0, "completions/min_terminated_length": 262.0, "entropy": 0.8741871435195208, "epoch": 0.02947845804988662, "frac_reward_zero_std": 1.0, "grad_norm": 0.00022220611572265625, "kl": 0.0006717530995956622, "learning_rate": 9.710884353741497e-07, "loss": 0.0, "num_tokens": 3139384.0, "reward": 0.0, "reward_std": 0.0, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 52, "step_time": 81.03633813653141 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1312.0, "completions/max_terminated_length": 1312.0, "completions/mean_length": 685.78125, "completions/mean_terminated_length": 685.78125, "completions/min_length": 385.0, "completions/min_terminated_length": 385.0, "entropy": 1.008056415244937, "epoch": 0.030045351473922902, "frac_reward_zero_std": 0.875, "grad_norm": 0.2099609375, "kl": 0.0008027284893614706, "learning_rate": 9.705215419501134e-07, "loss": 0.0029, "num_tokens": 3193314.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 53, "step_time": 51.80523566156626 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1396.0, "completions/max_terminated_length": 1396.0, "completions/mean_length": 847.921875, "completions/mean_terminated_length": 847.921875, "completions/min_length": 383.0, "completions/min_terminated_length": 383.0, "entropy": 0.8684836961328983, "epoch": 0.030612244897959183, "frac_reward_zero_std": 0.875, "grad_norm": 0.1533203125, "kl": 0.0006654198623436969, "learning_rate": 9.69954648526077e-07, "loss": 0.0056, "num_tokens": 3258621.0, "reward": 0.0078125, "reward_std": 0.0625, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 54, "step_time": 55.535200772807 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1297.0, "completions/max_terminated_length": 1297.0, "completions/mean_length": 714.484375, "completions/mean_terminated_length": 714.484375, "completions/min_length": 184.0, "completions/min_terminated_length": 184.0, "entropy": 0.9482756368815899, "epoch": 0.031179138321995464, "frac_reward_zero_std": 0.875, "grad_norm": 0.16796875, "kl": 0.0007149345219659153, "learning_rate": 9.693877551020407e-07, "loss": 0.0063, "num_tokens": 3315644.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 55, "step_time": 51.57503576390445 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1504.0, "completions/max_terminated_length": 1504.0, "completions/mean_length": 716.9375, "completions/mean_terminated_length": 716.9375, "completions/min_length": 242.0, "completions/min_terminated_length": 242.0, "entropy": 0.92583417147398, "epoch": 0.031746031746031744, "frac_reward_zero_std": 0.75, "grad_norm": 0.42578125, "kl": 0.0007983353825693484, "learning_rate": 9.688208616780045e-07, "loss": -0.0053, "num_tokens": 3373208.0, "reward": 0.09375, "reward_std": 0.426084041595459, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 56, "step_time": 59.63824711274356 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1363.0, "completions/max_terminated_length": 1363.0, "completions/mean_length": 726.875, "completions/mean_terminated_length": 726.875, "completions/min_length": 370.0, "completions/min_terminated_length": 370.0, "entropy": 0.7292650081217289, "epoch": 0.03231292517006803, "frac_reward_zero_std": 1.0, "grad_norm": 0.00052642822265625, "kl": 0.0007186582824942889, "learning_rate": 9.682539682539682e-07, "loss": 0.0, "num_tokens": 3432760.0, "reward": 0.0, "reward_std": 0.0, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 57, "step_time": 55.06257279776037 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1301.0, "completions/max_terminated_length": 1301.0, "completions/mean_length": 712.390625, "completions/mean_terminated_length": 712.390625, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "entropy": 0.7277219910174608, "epoch": 0.032879818594104306, "frac_reward_zero_std": 0.375, "grad_norm": 0.6328125, "kl": 0.0006634861638303846, "learning_rate": 9.67687074829932e-07, "loss": 0.0349, "num_tokens": 3487625.0, "reward": 0.2890625, "reward_std": 0.7003666162490845, "rewards/accuracy_reward_func/mean": 0.28125, "rewards/accuracy_reward_func/std": 0.7007648944854736, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 58, "step_time": 51.896855810657144 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1332.0, "completions/max_terminated_length": 1332.0, "completions/mean_length": 672.078125, "completions/mean_terminated_length": 672.078125, "completions/min_length": 327.0, "completions/min_terminated_length": 327.0, "entropy": 0.9556263759732246, "epoch": 0.03344671201814059, "frac_reward_zero_std": 0.5, "grad_norm": 0.609375, "kl": 0.0008063717141340021, "learning_rate": 9.671201814058957e-07, "loss": -0.0271, "num_tokens": 3541342.0, "reward": 0.140625, "reward_std": 0.4917473793029785, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 59, "step_time": 53.21447585616261 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 4096.0, "completions/max_terminated_length": 1470.0, "completions/mean_length": 816.28125, "completions/mean_terminated_length": 764.2222900390625, "completions/min_length": 314.0, "completions/min_terminated_length": 314.0, "entropy": 0.9738425239920616, "epoch": 0.034013605442176874, "frac_reward_zero_std": 0.875, "grad_norm": 0.25, "kl": 0.0007435131665260997, "learning_rate": 9.665532879818595e-07, "loss": -0.0248, "num_tokens": 3604608.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 60, "step_time": 182.24652972444892 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2197.0, "completions/max_terminated_length": 2197.0, "completions/mean_length": 811.84375, "completions/mean_terminated_length": 811.84375, "completions/min_length": 342.0, "completions/min_terminated_length": 342.0, "entropy": 0.8543225675821304, "epoch": 0.03458049886621315, "frac_reward_zero_std": 0.875, "grad_norm": 0.21484375, "kl": 0.0007339417061302811, "learning_rate": 9.65986394557823e-07, "loss": 0.0047, "num_tokens": 3667670.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 61, "step_time": 86.90420265123248 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1365.0, "completions/max_terminated_length": 1365.0, "completions/mean_length": 659.34375, "completions/mean_terminated_length": 659.34375, "completions/min_length": 273.0, "completions/min_terminated_length": 273.0, "entropy": 0.8573559112846851, "epoch": 0.035147392290249435, "frac_reward_zero_std": 0.75, "grad_norm": 0.36328125, "kl": 0.0006897826133354101, "learning_rate": 9.654195011337868e-07, "loss": -0.0002, "num_tokens": 3718516.0, "reward": 0.0625, "reward_std": 0.35073620080947876, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 62, "step_time": 53.312431974336505 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1478.0, "completions/max_terminated_length": 1478.0, "completions/mean_length": 730.34375, "completions/mean_terminated_length": 730.34375, "completions/min_length": 258.0, "completions/min_terminated_length": 258.0, "entropy": 0.9046593047678471, "epoch": 0.03571428571428571, "frac_reward_zero_std": 0.75, "grad_norm": 0.3359375, "kl": 0.0007334729616559343, "learning_rate": 9.648526077097505e-07, "loss": -0.0, "num_tokens": 3775250.0, "reward": 0.25, "reward_std": 0.6666666865348816, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 63, "step_time": 56.92630753200501 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1827.0, "completions/max_terminated_length": 1827.0, "completions/mean_length": 793.21875, "completions/mean_terminated_length": 793.21875, "completions/min_length": 18.0, "completions/min_terminated_length": 18.0, "entropy": 0.7975286822766066, "epoch": 0.036281179138321996, "frac_reward_zero_std": 0.75, "grad_norm": 0.298828125, "kl": 0.0006675975000689505, "learning_rate": 9.642857142857142e-07, "loss": -0.0224, "num_tokens": 3838456.0, "reward": 0.15625, "reward_std": 0.5409793853759766, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 64, "step_time": 72.07502091303468 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1546.0, "completions/max_terminated_length": 1546.0, "completions/mean_length": 746.296875, "completions/mean_terminated_length": 746.296875, "completions/min_length": 248.0, "completions/min_terminated_length": 248.0, "entropy": 0.9710622020065784, "epoch": 0.036848072562358274, "frac_reward_zero_std": 0.75, "grad_norm": 0.318359375, "kl": 0.0007247600697155576, "learning_rate": 9.63718820861678e-07, "loss": 0.0081, "num_tokens": 3897779.0, "reward": 0.0390625, "reward_std": 0.2567298710346222, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 65, "step_time": 61.013609156012535 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1482.0, "completions/max_terminated_length": 1482.0, "completions/mean_length": 872.578125, "completions/mean_terminated_length": 872.578125, "completions/min_length": 312.0, "completions/min_terminated_length": 312.0, "entropy": 0.6040001548826694, "epoch": 0.03741496598639456, "frac_reward_zero_std": 0.75, "grad_norm": 0.251953125, "kl": 0.0006071323223295622, "learning_rate": 9.631519274376417e-07, "loss": 0.0031, "num_tokens": 3966128.0, "reward": 0.09375, "reward_std": 0.426084041595459, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 66, "step_time": 59.130399737507105 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1829.0, "completions/max_terminated_length": 1829.0, "completions/mean_length": 741.125, "completions/mean_terminated_length": 741.125, "completions/min_length": 82.0, "completions/min_terminated_length": 82.0, "entropy": 0.7512783892452717, "epoch": 0.03798185941043084, "frac_reward_zero_std": 0.75, "grad_norm": 0.41015625, "kl": 0.0006990785877860617, "learning_rate": 9.625850340136053e-07, "loss": -0.0309, "num_tokens": 4024048.0, "reward": 0.0625, "reward_std": 0.35073620080947876, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 67, "step_time": 71.48969585448503 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1737.0, "completions/max_terminated_length": 1737.0, "completions/mean_length": 935.09375, "completions/mean_terminated_length": 935.09375, "completions/min_length": 388.0, "completions/min_terminated_length": 388.0, "entropy": 0.6651214510202408, "epoch": 0.03854875283446712, "frac_reward_zero_std": 0.625, "grad_norm": 0.37109375, "kl": 0.000569740826904308, "learning_rate": 9.62018140589569e-07, "loss": -0.0019, "num_tokens": 4094222.0, "reward": 0.125, "reward_std": 0.48795002698898315, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 68, "step_time": 68.3381945444271 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 4096.0, "completions/max_terminated_length": 1524.0, "completions/mean_length": 856.453125, "completions/mean_terminated_length": 805.0317993164062, "completions/min_length": 315.0, "completions/min_terminated_length": 315.0, "entropy": 0.7249712217599154, "epoch": 0.0391156462585034, "frac_reward_zero_std": 0.875, "grad_norm": 0.27734375, "kl": 0.0006563397655554581, "learning_rate": 9.614512471655328e-07, "loss": -0.0054, "num_tokens": 4160611.0, "reward": 0.0078125, "reward_std": 0.0625, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 69, "step_time": 181.34099934529513 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1827.0, "completions/max_terminated_length": 1827.0, "completions/mean_length": 763.640625, "completions/mean_terminated_length": 763.640625, "completions/min_length": 250.0, "completions/min_terminated_length": 250.0, "entropy": 0.9422737546265125, "epoch": 0.03968253968253968, "frac_reward_zero_std": 0.5, "grad_norm": 0.52734375, "kl": 0.0008529020524292719, "learning_rate": 9.608843537414965e-07, "loss": 0.0356, "num_tokens": 4219740.0, "reward": 0.1640625, "reward_std": 0.5422958731651306, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 70, "step_time": 71.48918853607029 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 629.34375, "completions/mean_terminated_length": 629.34375, "completions/min_length": 295.0, "completions/min_terminated_length": 295.0, "entropy": 0.7320157140493393, "epoch": 0.040249433106575964, "frac_reward_zero_std": 0.875, "grad_norm": 0.291015625, "kl": 0.0007003796126809902, "learning_rate": 9.603174603174603e-07, "loss": 0.0033, "num_tokens": 4269650.0, "reward": 0.0625, "reward_std": 0.35073620080947876, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 71, "step_time": 46.33684809971601 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1225.0, "completions/max_terminated_length": 1225.0, "completions/mean_length": 607.6875, "completions/mean_terminated_length": 607.6875, "completions/min_length": 250.0, "completions/min_terminated_length": 250.0, "entropy": 0.8365393560379744, "epoch": 0.04081632653061224, "frac_reward_zero_std": 0.625, "grad_norm": 0.482421875, "kl": 0.0007384430427919142, "learning_rate": 9.59750566893424e-07, "loss": 0.0073, "num_tokens": 4316630.0, "reward": 0.125, "reward_std": 0.48795002698898315, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 72, "step_time": 47.365977444685996 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1606.0, "completions/max_terminated_length": 1606.0, "completions/mean_length": 766.859375, "completions/mean_terminated_length": 766.859375, "completions/min_length": 334.0, "completions/min_terminated_length": 334.0, "entropy": 0.7943666074424982, "epoch": 0.041383219954648526, "frac_reward_zero_std": 0.625, "grad_norm": 0.404296875, "kl": 0.0006592419031221652, "learning_rate": 9.591836734693876e-07, "loss": 0.0108, "num_tokens": 4383725.0, "reward": 0.1875, "reward_std": 0.5875696539878845, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 73, "step_time": 70.14359752461314 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1080.0, "completions/max_terminated_length": 1080.0, "completions/mean_length": 679.75, "completions/mean_terminated_length": 679.75, "completions/min_length": 261.0, "completions/min_terminated_length": 261.0, "entropy": 1.0987933464348316, "epoch": 0.04195011337868481, "frac_reward_zero_std": 0.625, "grad_norm": 0.56640625, "kl": 0.0008692541960044764, "learning_rate": 9.586167800453513e-07, "loss": -0.0024, "num_tokens": 4436613.0, "reward": 0.1328125, "reward_std": 0.489915668964386, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 74, "step_time": 44.69452762231231 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1387.0, "completions/max_terminated_length": 1387.0, "completions/mean_length": 829.5625, "completions/mean_terminated_length": 829.5625, "completions/min_length": 411.0, "completions/min_terminated_length": 411.0, "entropy": 0.6923223361372948, "epoch": 0.04251700680272109, "frac_reward_zero_std": 0.875, "grad_norm": 0.22265625, "kl": 0.0006142441598058213, "learning_rate": 9.58049886621315e-07, "loss": -0.0138, "num_tokens": 4499945.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 75, "step_time": 56.41232482995838 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1258.0, "completions/max_terminated_length": 1258.0, "completions/mean_length": 701.484375, "completions/mean_terminated_length": 701.484375, "completions/min_length": 264.0, "completions/min_terminated_length": 264.0, "entropy": 0.8171689808368683, "epoch": 0.04308390022675737, "frac_reward_zero_std": 0.75, "grad_norm": 0.30859375, "kl": 0.000773871553974459, "learning_rate": 9.574829931972788e-07, "loss": 0.0193, "num_tokens": 4553632.0, "reward": 0.078125, "reward_std": 0.3587760329246521, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 76, "step_time": 48.506453486159444 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1288.0, "completions/max_terminated_length": 1288.0, "completions/mean_length": 779.9375, "completions/mean_terminated_length": 779.9375, "completions/min_length": 399.0, "completions/min_terminated_length": 399.0, "entropy": 0.9224153310060501, "epoch": 0.04365079365079365, "frac_reward_zero_std": 0.375, "grad_norm": 0.59375, "kl": 0.0008126598149829078, "learning_rate": 9.569160997732426e-07, "loss": 0.0238, "num_tokens": 4613460.0, "reward": 0.15625, "reward_std": 0.49501484632492065, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 77, "step_time": 49.962997142225504 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1602.0, "completions/max_terminated_length": 1602.0, "completions/mean_length": 822.953125, "completions/mean_terminated_length": 822.953125, "completions/min_length": 299.0, "completions/min_terminated_length": 299.0, "entropy": 0.8187730945646763, "epoch": 0.04421768707482993, "frac_reward_zero_std": 0.625, "grad_norm": 0.390625, "kl": 0.0007310177807084983, "learning_rate": 9.563492063492063e-07, "loss": 0.0191, "num_tokens": 4677641.0, "reward": 0.0703125, "reward_std": 0.35486623644828796, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 78, "step_time": 62.56507764104754 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1495.0, "completions/max_terminated_length": 1495.0, "completions/mean_length": 792.046875, "completions/mean_terminated_length": 792.046875, "completions/min_length": 406.0, "completions/min_terminated_length": 406.0, "entropy": 0.772476501762867, "epoch": 0.04478458049886622, "frac_reward_zero_std": 0.75, "grad_norm": 0.373046875, "kl": 0.0007511713683925336, "learning_rate": 9.5578231292517e-07, "loss": 0.0328, "num_tokens": 4738996.0, "reward": 0.125, "reward_std": 0.48795002698898315, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 79, "step_time": 57.763637254945934 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1732.0, "completions/max_terminated_length": 1732.0, "completions/mean_length": 801.203125, "completions/mean_terminated_length": 801.203125, "completions/min_length": 182.0, "completions/min_terminated_length": 182.0, "entropy": 0.8073384631425142, "epoch": 0.045351473922902494, "frac_reward_zero_std": 0.75, "grad_norm": 0.328125, "kl": 0.0006839092584414175, "learning_rate": 9.552154195011338e-07, "loss": -0.0029, "num_tokens": 4800513.0, "reward": 0.125, "reward_std": 0.48795002698898315, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 80, "step_time": 67.18388975318521 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1291.0, "completions/max_terminated_length": 1291.0, "completions/mean_length": 736.703125, "completions/mean_terminated_length": 736.703125, "completions/min_length": 344.0, "completions/min_terminated_length": 344.0, "entropy": 0.8549042195081711, "epoch": 0.04591836734693878, "frac_reward_zero_std": 0.875, "grad_norm": 0.302734375, "kl": 0.0007158639236877207, "learning_rate": 9.546485260770974e-07, "loss": -0.0042, "num_tokens": 4857286.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 81, "step_time": 50.64162682276219 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1231.0, "completions/max_terminated_length": 1231.0, "completions/mean_length": 701.421875, "completions/mean_terminated_length": 701.421875, "completions/min_length": 329.0, "completions/min_terminated_length": 329.0, "entropy": 0.9150978699326515, "epoch": 0.046485260770975055, "frac_reward_zero_std": 1.0, "grad_norm": 0.0002899169921875, "kl": 0.0007863493265176658, "learning_rate": 9.54081632653061e-07, "loss": 0.0, "num_tokens": 4916737.0, "reward": 0.0, "reward_std": 0.0, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 82, "step_time": 51.29425719939172 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1905.0, "completions/max_terminated_length": 1905.0, "completions/mean_length": 811.9375, "completions/mean_terminated_length": 811.9375, "completions/min_length": 34.0, "completions/min_terminated_length": 34.0, "entropy": 0.6648439597338438, "epoch": 0.04705215419501134, "frac_reward_zero_std": 0.5, "grad_norm": 0.44140625, "kl": 0.0006040604894224089, "learning_rate": 9.53514739229025e-07, "loss": 0.0176, "num_tokens": 4977661.0, "reward": 0.125, "reward_std": 0.48795002698898315, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 83, "step_time": 75.30042485240847 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1192.0, "completions/max_terminated_length": 1192.0, "completions/mean_length": 784.046875, "completions/mean_terminated_length": 784.046875, "completions/min_length": 382.0, "completions/min_terminated_length": 382.0, "entropy": 1.0095579735934734, "epoch": 0.047619047619047616, "frac_reward_zero_std": 0.75, "grad_norm": 0.38671875, "kl": 0.000779534788307501, "learning_rate": 9.529478458049887e-07, "loss": 0.001, "num_tokens": 5038568.0, "reward": 0.0390625, "reward_std": 0.2567298710346222, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 84, "step_time": 47.187259843572974 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1261.0, "completions/max_terminated_length": 1261.0, "completions/mean_length": 925.625, "completions/mean_terminated_length": 925.625, "completions/min_length": 563.0, "completions/min_terminated_length": 563.0, "entropy": 0.6331640388816595, "epoch": 0.0481859410430839, "frac_reward_zero_std": 0.75, "grad_norm": 0.380859375, "kl": 0.000579886636842275, "learning_rate": 9.523809523809522e-07, "loss": -0.0118, "num_tokens": 5109528.0, "reward": 0.0625, "reward_std": 0.35073620080947876, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 85, "step_time": 50.76587761286646 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1555.0, "completions/max_terminated_length": 1555.0, "completions/mean_length": 810.25, "completions/mean_terminated_length": 810.25, "completions/min_length": 421.0, "completions/min_terminated_length": 421.0, "entropy": 0.9099563546478748, "epoch": 0.048752834467120185, "frac_reward_zero_std": 0.875, "grad_norm": 0.255859375, "kl": 0.0008195096816052683, "learning_rate": 9.51814058956916e-07, "loss": -0.0075, "num_tokens": 5172280.0, "reward": 0.0078125, "reward_std": 0.0625, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 86, "step_time": 61.36391172464937 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1628.0, "completions/max_terminated_length": 1628.0, "completions/mean_length": 822.515625, "completions/mean_terminated_length": 822.515625, "completions/min_length": 401.0, "completions/min_terminated_length": 401.0, "entropy": 0.9575734511017799, "epoch": 0.04931972789115646, "frac_reward_zero_std": 0.875, "grad_norm": 0.28515625, "kl": 0.0007745505245111417, "learning_rate": 9.512471655328797e-07, "loss": 0.005, "num_tokens": 5237673.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 87, "step_time": 63.86270225327462 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1300.0, "completions/max_terminated_length": 1300.0, "completions/mean_length": 801.8125, "completions/mean_terminated_length": 801.8125, "completions/min_length": 445.0, "completions/min_terminated_length": 445.0, "entropy": 0.9237263891845942, "epoch": 0.049886621315192746, "frac_reward_zero_std": 0.875, "grad_norm": 0.2451171875, "kl": 0.0007447822299582185, "learning_rate": 9.506802721088435e-07, "loss": 0.0031, "num_tokens": 5308589.0, "reward": 0.09375, "reward_std": 0.426084041595459, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 88, "step_time": 57.35689403023571 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1207.0, "completions/max_terminated_length": 1207.0, "completions/mean_length": 754.203125, "completions/mean_terminated_length": 754.203125, "completions/min_length": 349.0, "completions/min_terminated_length": 349.0, "entropy": 0.8471732772886753, "epoch": 0.05045351473922902, "frac_reward_zero_std": 1.0, "grad_norm": 0.000293731689453125, "kl": 0.0007413623989123153, "learning_rate": 9.501133786848072e-07, "loss": 0.0, "num_tokens": 5367594.0, "reward": 0.0, "reward_std": 0.0, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 89, "step_time": 47.73160739801824 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1388.0, "completions/max_terminated_length": 1388.0, "completions/mean_length": 787.71875, "completions/mean_terminated_length": 787.71875, "completions/min_length": 289.0, "completions/min_terminated_length": 289.0, "entropy": 0.8667267635464668, "epoch": 0.05102040816326531, "frac_reward_zero_std": 0.875, "grad_norm": 0.25, "kl": 0.000806073730927892, "learning_rate": 9.49546485260771e-07, "loss": 0.0017, "num_tokens": 5429440.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 90, "step_time": 54.461288248188794 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1508.0, "completions/max_terminated_length": 1508.0, "completions/mean_length": 801.96875, "completions/mean_terminated_length": 801.96875, "completions/min_length": 421.0, "completions/min_terminated_length": 421.0, "entropy": 0.8943926393985748, "epoch": 0.051587301587301584, "frac_reward_zero_std": 0.625, "grad_norm": 0.39453125, "kl": 0.0007918762676126789, "learning_rate": 9.489795918367347e-07, "loss": 0.0153, "num_tokens": 5493286.0, "reward": 0.0703125, "reward_std": 0.35486623644828796, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 91, "step_time": 59.711739867925644 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1538.0, "completions/max_terminated_length": 1538.0, "completions/mean_length": 811.359375, "completions/mean_terminated_length": 811.359375, "completions/min_length": 291.0, "completions/min_terminated_length": 291.0, "entropy": 0.6979294549673796, "epoch": 0.05215419501133787, "frac_reward_zero_std": 0.75, "grad_norm": 0.34765625, "kl": 0.000719607574865222, "learning_rate": 9.484126984126984e-07, "loss": 0.0023, "num_tokens": 5555325.0, "reward": 0.125, "reward_std": 0.48795002698898315, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 92, "step_time": 59.873963047750294 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1534.0, "completions/max_terminated_length": 1534.0, "completions/mean_length": 763.109375, "completions/mean_terminated_length": 763.109375, "completions/min_length": 233.0, "completions/min_terminated_length": 233.0, "entropy": 0.933561509475112, "epoch": 0.05272108843537415, "frac_reward_zero_std": 0.75, "grad_norm": 0.318359375, "kl": 0.0007509427705372218, "learning_rate": 9.47845804988662e-07, "loss": 0.0134, "num_tokens": 5616124.0, "reward": 0.09375, "reward_std": 0.426084041595459, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 93, "step_time": 60.57129462156445 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1352.0, "completions/max_terminated_length": 1352.0, "completions/mean_length": 685.421875, "completions/mean_terminated_length": 685.421875, "completions/min_length": 320.0, "completions/min_terminated_length": 320.0, "entropy": 1.1630984470248222, "epoch": 0.05328798185941043, "frac_reward_zero_std": 0.75, "grad_norm": 0.43359375, "kl": 0.0009033987516886555, "learning_rate": 9.472789115646258e-07, "loss": -0.0213, "num_tokens": 5671479.0, "reward": 0.0390625, "reward_std": 0.2567298710346222, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 94, "step_time": 53.713384873233736 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1834.0, "completions/max_terminated_length": 1834.0, "completions/mean_length": 712.3125, "completions/mean_terminated_length": 712.3125, "completions/min_length": 314.0, "completions/min_terminated_length": 314.0, "entropy": 0.9153989348560572, "epoch": 0.053854875283446714, "frac_reward_zero_std": 0.625, "grad_norm": 0.4609375, "kl": 0.0008535892484360375, "learning_rate": 9.467120181405895e-07, "loss": -0.0601, "num_tokens": 5727011.0, "reward": 0.125, "reward_std": 0.48795002698898315, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 95, "step_time": 71.87342350929976 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1255.0, "completions/max_terminated_length": 1255.0, "completions/mean_length": 716.84375, "completions/mean_terminated_length": 716.84375, "completions/min_length": 82.0, "completions/min_terminated_length": 82.0, "entropy": 0.7735940665006638, "epoch": 0.05442176870748299, "frac_reward_zero_std": 0.375, "grad_norm": 0.5625, "kl": 0.0007864913895900827, "learning_rate": 9.461451247165533e-07, "loss": -0.0109, "num_tokens": 5783737.0, "reward": 0.1640625, "reward_std": 0.5422958731651306, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 96, "step_time": 49.97596748266369 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1406.0, "completions/max_terminated_length": 1406.0, "completions/mean_length": 749.875, "completions/mean_terminated_length": 749.875, "completions/min_length": 264.0, "completions/min_terminated_length": 264.0, "entropy": 0.921129334717989, "epoch": 0.054988662131519275, "frac_reward_zero_std": 0.75, "grad_norm": 0.330078125, "kl": 0.0008132914081215858, "learning_rate": 9.45578231292517e-07, "loss": -0.0044, "num_tokens": 5842865.0, "reward": 0.0625, "reward_std": 0.35073620080947876, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 97, "step_time": 55.25691071804613 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1780.0, "completions/max_terminated_length": 1780.0, "completions/mean_length": 916.84375, "completions/mean_terminated_length": 916.84375, "completions/min_length": 480.0, "completions/min_terminated_length": 480.0, "entropy": 0.8317632135003805, "epoch": 0.05555555555555555, "frac_reward_zero_std": 0.875, "grad_norm": 0.2431640625, "kl": 0.0007306409952434478, "learning_rate": 9.450113378684807e-07, "loss": 0.0047, "num_tokens": 5912511.0, "reward": 0.0078125, "reward_std": 0.0625, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 98, "step_time": 70.03312500286847 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1460.0, "completions/max_terminated_length": 1460.0, "completions/mean_length": 824.359375, "completions/mean_terminated_length": 824.359375, "completions/min_length": 245.0, "completions/min_terminated_length": 245.0, "entropy": 0.7674974091351032, "epoch": 0.05612244897959184, "frac_reward_zero_std": 0.875, "grad_norm": 0.185546875, "kl": 0.0006871784426039085, "learning_rate": 9.444444444444444e-07, "loss": 0.0055, "num_tokens": 5978486.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 99, "step_time": 60.27627642452717 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1437.0, "completions/max_terminated_length": 1437.0, "completions/mean_length": 852.078125, "completions/mean_terminated_length": 852.078125, "completions/min_length": 358.0, "completions/min_terminated_length": 358.0, "entropy": 0.7974452953785658, "epoch": 0.05668934240362812, "frac_reward_zero_std": 0.625, "grad_norm": 0.431640625, "kl": 0.0007822227889846545, "learning_rate": 9.438775510204082e-07, "loss": 0.0245, "num_tokens": 6043827.0, "reward": 0.125, "reward_std": 0.48795002698898315, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 100, "step_time": 57.44859805330634 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1418.0, "completions/max_terminated_length": 1418.0, "completions/mean_length": 756.625, "completions/mean_terminated_length": 756.625, "completions/min_length": 341.0, "completions/min_terminated_length": 341.0, "entropy": 0.7994147948920727, "epoch": 0.0572562358276644, "frac_reward_zero_std": 0.625, "grad_norm": 0.41015625, "kl": 0.0007528209171141498, "learning_rate": 9.433106575963718e-07, "loss": 0.026, "num_tokens": 6101947.0, "reward": 0.1875, "reward_std": 0.5875696539878845, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 101, "step_time": 55.89661477226764 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1311.0, "completions/max_terminated_length": 1311.0, "completions/mean_length": 785.234375, "completions/mean_terminated_length": 785.234375, "completions/min_length": 382.0, "completions/min_terminated_length": 382.0, "entropy": 0.6613637581467628, "epoch": 0.05782312925170068, "frac_reward_zero_std": 0.875, "grad_norm": 0.259765625, "kl": 0.0007186354232544545, "learning_rate": 9.427437641723356e-07, "loss": 0.0114, "num_tokens": 6162258.0, "reward": 0.0078125, "reward_std": 0.0625, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 102, "step_time": 51.73925363551825 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1594.0, "completions/max_terminated_length": 1594.0, "completions/mean_length": 845.875, "completions/mean_terminated_length": 845.875, "completions/min_length": 283.0, "completions/min_terminated_length": 283.0, "entropy": 0.8546389937400818, "epoch": 0.05839002267573696, "frac_reward_zero_std": 0.625, "grad_norm": 0.349609375, "kl": 0.0007117596451280406, "learning_rate": 9.421768707482993e-07, "loss": 0.0217, "num_tokens": 6226426.0, "reward": 0.1015625, "reward_std": 0.42891231179237366, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 103, "step_time": 62.46534308884293 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1694.0, "completions/max_terminated_length": 1694.0, "completions/mean_length": 844.265625, "completions/mean_terminated_length": 844.265625, "completions/min_length": 277.0, "completions/min_terminated_length": 277.0, "entropy": 0.6533396132290363, "epoch": 0.05895691609977324, "frac_reward_zero_std": 0.5, "grad_norm": 0.435546875, "kl": 0.0008017468153411755, "learning_rate": 9.41609977324263e-07, "loss": 0.02, "num_tokens": 6293019.0, "reward": 0.171875, "reward_std": 0.5434950590133667, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 104, "step_time": 66.67652920912951 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1389.0, "completions/max_terminated_length": 1389.0, "completions/mean_length": 734.3125, "completions/mean_terminated_length": 734.3125, "completions/min_length": 253.0, "completions/min_terminated_length": 253.0, "entropy": 0.8464022967964411, "epoch": 0.05952380952380952, "frac_reward_zero_std": 0.625, "grad_norm": 0.427734375, "kl": 0.0008064001049206126, "learning_rate": 9.410430839002267e-07, "loss": 0.0259, "num_tokens": 6349567.0, "reward": 0.21875, "reward_std": 0.6291528940200806, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 105, "step_time": 54.00935931224376 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1550.0, "completions/max_terminated_length": 1550.0, "completions/mean_length": 770.546875, "completions/mean_terminated_length": 770.546875, "completions/min_length": 303.0, "completions/min_terminated_length": 303.0, "entropy": 0.8186229839920998, "epoch": 0.060090702947845805, "frac_reward_zero_std": 0.5, "grad_norm": 0.54296875, "kl": 0.0007566980402771151, "learning_rate": 9.404761904761904e-07, "loss": -0.0066, "num_tokens": 6408538.0, "reward": 0.1953125, "reward_std": 0.5883605480194092, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 106, "step_time": 60.694702457636595 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1288.0, "completions/max_terminated_length": 1288.0, "completions/mean_length": 758.34375, "completions/mean_terminated_length": 758.34375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "entropy": 0.8053422253578901, "epoch": 0.06065759637188209, "frac_reward_zero_std": 0.625, "grad_norm": 0.451171875, "kl": 0.0007592442962049972, "learning_rate": 9.399092970521542e-07, "loss": 0.0093, "num_tokens": 6469112.0, "reward": 0.09375, "reward_std": 0.426084041595459, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 107, "step_time": 50.79260412603617 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1458.0, "completions/max_terminated_length": 1458.0, "completions/mean_length": 797.53125, "completions/mean_terminated_length": 797.53125, "completions/min_length": 352.0, "completions/min_terminated_length": 352.0, "entropy": 0.6849835515022278, "epoch": 0.061224489795918366, "frac_reward_zero_std": 0.625, "grad_norm": 0.345703125, "kl": 0.0007152012603910407, "learning_rate": 9.393424036281179e-07, "loss": -0.0011, "num_tokens": 6529466.0, "reward": 0.1015625, "reward_std": 0.42891231179237366, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 108, "step_time": 57.19498935434967 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1308.0, "completions/max_terminated_length": 1308.0, "completions/mean_length": 659.4375, "completions/mean_terminated_length": 659.4375, "completions/min_length": 215.0, "completions/min_terminated_length": 215.0, "entropy": 1.0229039713740349, "epoch": 0.06179138321995465, "frac_reward_zero_std": 0.875, "grad_norm": 0.384765625, "kl": 0.0009435774336452596, "learning_rate": 9.387755102040816e-07, "loss": 0.0059, "num_tokens": 6580846.0, "reward": 0.0625, "reward_std": 0.35073620080947876, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 109, "step_time": 51.30871072318405 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1400.0, "completions/max_terminated_length": 1400.0, "completions/mean_length": 726.390625, "completions/mean_terminated_length": 726.390625, "completions/min_length": 297.0, "completions/min_terminated_length": 297.0, "entropy": 0.7591048255562782, "epoch": 0.06235827664399093, "frac_reward_zero_std": 0.75, "grad_norm": 0.26171875, "kl": 0.0007802735126460902, "learning_rate": 9.382086167800452e-07, "loss": -0.0063, "num_tokens": 6637271.0, "reward": 0.09375, "reward_std": 0.426084041595459, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 110, "step_time": 55.055217486806214 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1309.0, "completions/max_terminated_length": 1309.0, "completions/mean_length": 742.71875, "completions/mean_terminated_length": 742.71875, "completions/min_length": 320.0, "completions/min_terminated_length": 320.0, "entropy": 0.9038644768297672, "epoch": 0.06292517006802721, "frac_reward_zero_std": 0.5, "grad_norm": 0.53125, "kl": 0.0007633761742908973, "learning_rate": 9.37641723356009e-07, "loss": 0.0132, "num_tokens": 6696757.0, "reward": 0.21875, "reward_std": 0.6291528940200806, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 111, "step_time": 51.84569195844233 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1390.0, "completions/max_terminated_length": 1390.0, "completions/mean_length": 762.453125, "completions/mean_terminated_length": 762.453125, "completions/min_length": 254.0, "completions/min_terminated_length": 254.0, "entropy": 0.7126942295581102, "epoch": 0.06349206349206349, "frac_reward_zero_std": 1.0, "grad_norm": 0.000408172607421875, "kl": 0.0008155866780725773, "learning_rate": 9.370748299319727e-07, "loss": 0.0, "num_tokens": 6755994.0, "reward": 0.0, "reward_std": 0.0, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 112, "step_time": 54.81356189586222 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1438.0, "completions/max_terminated_length": 1438.0, "completions/mean_length": 777.125, "completions/mean_terminated_length": 777.125, "completions/min_length": 295.0, "completions/min_terminated_length": 295.0, "entropy": 0.8607089705765247, "epoch": 0.06405895691609978, "frac_reward_zero_std": 0.75, "grad_norm": 0.294921875, "kl": 0.0008256015426013619, "learning_rate": 9.365079365079365e-07, "loss": -0.0086, "num_tokens": 6817346.0, "reward": 0.0625, "reward_std": 0.35073620080947876, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 113, "step_time": 57.42120492178947 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1348.0, "completions/max_terminated_length": 1348.0, "completions/mean_length": 744.3125, "completions/mean_terminated_length": 744.3125, "completions/min_length": 222.0, "completions/min_terminated_length": 222.0, "entropy": 0.9622498042881489, "epoch": 0.06462585034013606, "frac_reward_zero_std": 0.875, "grad_norm": 0.333984375, "kl": 0.0008918357489164919, "learning_rate": 9.359410430839002e-07, "loss": 0.015, "num_tokens": 6877462.0, "reward": 0.09375, "reward_std": 0.426084041595459, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 114, "step_time": 54.03743090759963 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1034.0, "completions/max_terminated_length": 1034.0, "completions/mean_length": 652.59375, "completions/mean_terminated_length": 652.59375, "completions/min_length": 297.0, "completions/min_terminated_length": 297.0, "entropy": 1.0135034248232841, "epoch": 0.06519274376417233, "frac_reward_zero_std": 0.375, "grad_norm": 0.69140625, "kl": 0.0010756870760815218, "learning_rate": 9.35374149659864e-07, "loss": -0.0047, "num_tokens": 6929452.0, "reward": 0.234375, "reward_std": 0.6297439336776733, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 115, "step_time": 41.20539268106222 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1209.0, "completions/max_terminated_length": 1209.0, "completions/mean_length": 697.765625, "completions/mean_terminated_length": 697.765625, "completions/min_length": 325.0, "completions/min_terminated_length": 325.0, "entropy": 0.8065849877893925, "epoch": 0.06575963718820861, "frac_reward_zero_std": 0.375, "grad_norm": 0.6640625, "kl": 0.0009536606376059353, "learning_rate": 9.348072562358276e-07, "loss": -0.0416, "num_tokens": 6990093.0, "reward": 0.1875, "reward_std": 0.5875696539878845, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 116, "step_time": 51.148926474153996 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1358.0, "completions/max_terminated_length": 1358.0, "completions/mean_length": 774.703125, "completions/mean_terminated_length": 774.703125, "completions/min_length": 364.0, "completions/min_terminated_length": 364.0, "entropy": 0.8654811605811119, "epoch": 0.0663265306122449, "frac_reward_zero_std": 0.75, "grad_norm": 0.345703125, "kl": 0.0008410576774622314, "learning_rate": 9.342403628117913e-07, "loss": 0.0157, "num_tokens": 7053066.0, "reward": 0.0390625, "reward_std": 0.2567298710346222, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 117, "step_time": 55.61251427512616 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1278.0, "completions/max_terminated_length": 1278.0, "completions/mean_length": 691.921875, "completions/mean_terminated_length": 691.921875, "completions/min_length": 322.0, "completions/min_terminated_length": 322.0, "entropy": 0.9328459855169058, "epoch": 0.06689342403628118, "frac_reward_zero_std": 0.625, "grad_norm": 0.46875, "kl": 0.0009202329711115453, "learning_rate": 9.33673469387755e-07, "loss": 0.0169, "num_tokens": 7109197.0, "reward": 0.09375, "reward_std": 0.426084041595459, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 118, "step_time": 51.24461770057678 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1603.0, "completions/max_terminated_length": 1603.0, "completions/mean_length": 726.078125, "completions/mean_terminated_length": 726.078125, "completions/min_length": 259.0, "completions/min_terminated_length": 259.0, "entropy": 0.8497186675667763, "epoch": 0.06746031746031746, "frac_reward_zero_std": 0.625, "grad_norm": 0.4765625, "kl": 0.0008916753904486541, "learning_rate": 9.331065759637188e-07, "loss": -0.0135, "num_tokens": 7164370.0, "reward": 0.125, "reward_std": 0.48795002698898315, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 119, "step_time": 62.421043317765 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1325.0, "completions/max_terminated_length": 1325.0, "completions/mean_length": 864.21875, "completions/mean_terminated_length": 864.21875, "completions/min_length": 526.0, "completions/min_terminated_length": 526.0, "entropy": 0.8159789443016052, "epoch": 0.06802721088435375, "frac_reward_zero_std": 0.5, "grad_norm": 0.470703125, "kl": 0.0007719660407019546, "learning_rate": 9.325396825396825e-07, "loss": 0.0131, "num_tokens": 7230192.0, "reward": 0.2578125, "reward_std": 0.666620135307312, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 120, "step_time": 51.65067602414638 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2271.0, "completions/max_terminated_length": 2271.0, "completions/mean_length": 818.578125, "completions/mean_terminated_length": 818.578125, "completions/min_length": 387.0, "completions/min_terminated_length": 387.0, "entropy": 0.6372475307434797, "epoch": 0.06859410430839002, "frac_reward_zero_std": 0.75, "grad_norm": 0.40234375, "kl": 0.0007827264707884751, "learning_rate": 9.319727891156463e-07, "loss": 0.0045, "num_tokens": 7294253.0, "reward": 0.0546875, "reward_std": 0.26899558305740356, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 121, "step_time": 89.85847285855561 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1541.0, "completions/max_terminated_length": 1541.0, "completions/mean_length": 827.15625, "completions/mean_terminated_length": 827.15625, "completions/min_length": 7.0, "completions/min_terminated_length": 7.0, "entropy": 0.666001146659255, "epoch": 0.0691609977324263, "frac_reward_zero_std": 0.5, "grad_norm": 0.498046875, "kl": 0.0007001044541539159, "learning_rate": 9.3140589569161e-07, "loss": 0.0055, "num_tokens": 7358327.0, "reward": 0.28125, "reward_std": 0.7007648944854736, "rewards/accuracy_reward_func/mean": 0.28125, "rewards/accuracy_reward_func/std": 0.7007648944854736, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 122, "step_time": 60.366386415436864 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1220.0, "completions/max_terminated_length": 1220.0, "completions/mean_length": 598.796875, "completions/mean_terminated_length": 598.796875, "completions/min_length": 195.0, "completions/min_terminated_length": 195.0, "entropy": 1.0049929097294807, "epoch": 0.06972789115646258, "frac_reward_zero_std": 0.875, "grad_norm": 0.30078125, "kl": 0.0010934763122349977, "learning_rate": 9.308390022675737e-07, "loss": -0.0009, "num_tokens": 7410298.0, "reward": 0.125, "reward_std": 0.48795002698898315, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 123, "step_time": 48.925475688651204 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1826.0, "completions/max_terminated_length": 1826.0, "completions/mean_length": 602.09375, "completions/mean_terminated_length": 602.09375, "completions/min_length": 231.0, "completions/min_terminated_length": 231.0, "entropy": 0.9215579964220524, "epoch": 0.07029478458049887, "frac_reward_zero_std": 0.625, "grad_norm": 0.69140625, "kl": 0.0012636464562092442, "learning_rate": 9.302721088435374e-07, "loss": -0.0067, "num_tokens": 7458544.0, "reward": 0.0703125, "reward_std": 0.35486623644828796, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 124, "step_time": 70.9872962711379 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1610.0, "completions/max_terminated_length": 1610.0, "completions/mean_length": 830.53125, "completions/mean_terminated_length": 830.53125, "completions/min_length": 351.0, "completions/min_terminated_length": 351.0, "entropy": 0.8522497117519379, "epoch": 0.07086167800453515, "frac_reward_zero_std": 0.75, "grad_norm": 0.29296875, "kl": 0.0009019470344355796, "learning_rate": 9.29705215419501e-07, "loss": 0.0057, "num_tokens": 7524058.0, "reward": 0.0390625, "reward_std": 0.2567298710346222, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 125, "step_time": 63.3850915748626 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1282.0, "completions/max_terminated_length": 1282.0, "completions/mean_length": 806.25, "completions/mean_terminated_length": 806.25, "completions/min_length": 246.0, "completions/min_terminated_length": 246.0, "entropy": 0.7999003082513809, "epoch": 0.07142857142857142, "frac_reward_zero_std": 0.625, "grad_norm": 0.47265625, "kl": 0.0008742077352508204, "learning_rate": 9.291383219954648e-07, "loss": -0.0328, "num_tokens": 7586338.0, "reward": 0.1015625, "reward_std": 0.42891231179237366, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 126, "step_time": 50.78821122366935 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1572.0, "completions/max_terminated_length": 1572.0, "completions/mean_length": 790.1875, "completions/mean_terminated_length": 790.1875, "completions/min_length": 421.0, "completions/min_terminated_length": 421.0, "entropy": 0.8215701729059219, "epoch": 0.07199546485260772, "frac_reward_zero_std": 0.75, "grad_norm": 0.380859375, "kl": 0.0009467004892940167, "learning_rate": 9.285714285714285e-07, "loss": 0.0209, "num_tokens": 7648062.0, "reward": 0.0390625, "reward_std": 0.2567298710346222, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 127, "step_time": 61.55685986764729 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1501.0, "completions/max_terminated_length": 1501.0, "completions/mean_length": 718.5, "completions/mean_terminated_length": 718.5, "completions/min_length": 317.0, "completions/min_terminated_length": 317.0, "entropy": 0.8467830922454596, "epoch": 0.07256235827664399, "frac_reward_zero_std": 0.875, "grad_norm": 0.2890625, "kl": 0.0010289369311067276, "learning_rate": 9.280045351473923e-07, "loss": 0.0067, "num_tokens": 7704590.0, "reward": 0.015625, "reward_std": 0.08768405020236969, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 128, "step_time": 58.87193254288286 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1444.0, "completions/max_terminated_length": 1444.0, "completions/mean_length": 789.75, "completions/mean_terminated_length": 789.75, "completions/min_length": 377.0, "completions/min_terminated_length": 377.0, "entropy": 0.6771079618483782, "epoch": 0.07312925170068027, "frac_reward_zero_std": 0.5, "grad_norm": 0.51171875, "kl": 0.0008786784237599932, "learning_rate": 9.274376417233559e-07, "loss": 0.0014, "num_tokens": 7764822.0, "reward": 0.171875, "reward_std": 0.5434950590133667, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 129, "step_time": 56.4890464367345 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 4096.0, "completions/max_terminated_length": 2020.0, "completions/mean_length": 925.453125, "completions/mean_terminated_length": 875.1270141601562, "completions/min_length": 389.0, "completions/min_terminated_length": 389.0, "entropy": 0.6869203355163336, "epoch": 0.07369614512471655, "frac_reward_zero_std": 0.75, "grad_norm": 0.283203125, "kl": 0.0009084341945708729, "learning_rate": 9.268707482993197e-07, "loss": -0.0242, "num_tokens": 7834419.0, "reward": 0.0625, "reward_std": 0.2745848298072815, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 130, "step_time": 181.55315329879522 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1208.0, "completions/max_terminated_length": 1208.0, "completions/mean_length": 697.703125, "completions/mean_terminated_length": 697.703125, "completions/min_length": 85.0, "completions/min_terminated_length": 85.0, "entropy": 1.0916701704263687, "epoch": 0.07426303854875284, "frac_reward_zero_std": 0.75, "grad_norm": 0.32421875, "kl": 0.0010625670292938594, "learning_rate": 9.263038548752834e-07, "loss": -0.0137, "num_tokens": 7889696.0, "reward": 0.09375, "reward_std": 0.426084041595459, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 131, "step_time": 47.96899611968547 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1471.0, "completions/max_terminated_length": 1471.0, "completions/mean_length": 795.75, "completions/mean_terminated_length": 795.75, "completions/min_length": 293.0, "completions/min_terminated_length": 293.0, "entropy": 0.7799660600721836, "epoch": 0.07482993197278912, "frac_reward_zero_std": 0.625, "grad_norm": 0.3828125, "kl": 0.0009379767434438691, "learning_rate": 9.257369614512472e-07, "loss": 0.0217, "num_tokens": 7951104.0, "reward": 0.125, "reward_std": 0.48795002698898315, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 132, "step_time": 58.739000890403986 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1454.0, "completions/max_terminated_length": 1454.0, "completions/mean_length": 665.484375, "completions/mean_terminated_length": 665.484375, "completions/min_length": 240.0, "completions/min_terminated_length": 240.0, "entropy": 0.8104948624968529, "epoch": 0.07539682539682539, "frac_reward_zero_std": 0.5, "grad_norm": 0.56640625, "kl": 0.001190572242194321, "learning_rate": 9.251700680272108e-07, "loss": 0.0302, "num_tokens": 8004183.0, "reward": 0.2265625, "reward_std": 0.5904644727706909, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 133, "step_time": 57.126070658676326 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1956.0, "completions/max_terminated_length": 1956.0, "completions/mean_length": 827.75, "completions/mean_terminated_length": 827.75, "completions/min_length": 308.0, "completions/min_terminated_length": 308.0, "entropy": 0.7314121834933758, "epoch": 0.07596371882086168, "frac_reward_zero_std": 0.625, "grad_norm": 0.4140625, "kl": 0.000903994710824918, "learning_rate": 9.246031746031746e-07, "loss": -0.0049, "num_tokens": 8068271.0, "reward": 0.1953125, "reward_std": 0.5883605480194092, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 134, "step_time": 77.03828375507146 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1391.0, "completions/max_terminated_length": 1391.0, "completions/mean_length": 697.5, "completions/mean_terminated_length": 697.5, "completions/min_length": 307.0, "completions/min_terminated_length": 307.0, "entropy": 0.7819268852472305, "epoch": 0.07653061224489796, "frac_reward_zero_std": 0.375, "grad_norm": 0.609375, "kl": 0.0012548193262773566, "learning_rate": 9.240362811791382e-07, "loss": -0.0229, "num_tokens": 8122103.0, "reward": 0.28125, "reward_std": 0.71200031042099, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 135, "step_time": 53.75756964739412 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1664.0, "completions/max_terminated_length": 1664.0, "completions/mean_length": 855.53125, "completions/mean_terminated_length": 855.53125, "completions/min_length": 284.0, "completions/min_terminated_length": 284.0, "entropy": 0.7436029799282551, "epoch": 0.07709750566893424, "frac_reward_zero_std": 0.625, "grad_norm": 0.4296875, "kl": 0.001049405378580559, "learning_rate": 9.23469387755102e-07, "loss": 0.0069, "num_tokens": 8186665.0, "reward": 0.171875, "reward_std": 0.49776285886764526, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 136, "step_time": 64.77417926304042 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1187.0, "completions/max_terminated_length": 1187.0, "completions/mean_length": 681.0625, "completions/mean_terminated_length": 681.0625, "completions/min_length": 254.0, "completions/min_terminated_length": 254.0, "entropy": 0.9490941390395164, "epoch": 0.07766439909297052, "frac_reward_zero_std": 0.625, "grad_norm": 0.4296875, "kl": 0.0011216318016522564, "learning_rate": 9.229024943310657e-07, "loss": -0.0011, "num_tokens": 8240965.0, "reward": 0.1015625, "reward_std": 0.42891231179237366, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 137, "step_time": 47.23937799781561 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1421.0, "completions/max_terminated_length": 1421.0, "completions/mean_length": 748.65625, "completions/mean_terminated_length": 748.65625, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "entropy": 0.9417592454701662, "epoch": 0.0782312925170068, "frac_reward_zero_std": 0.75, "grad_norm": 0.369140625, "kl": 0.0010869996040128171, "learning_rate": 9.223356009070295e-07, "loss": -0.0135, "num_tokens": 8302119.0, "reward": 0.1875, "reward_std": 0.5875696539878845, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 138, "step_time": 56.91432747989893 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1174.0, "completions/max_terminated_length": 1174.0, "completions/mean_length": 727.1875, "completions/mean_terminated_length": 727.1875, "completions/min_length": 354.0, "completions/min_terminated_length": 354.0, "entropy": 0.8250185586512089, "epoch": 0.07879818594104308, "frac_reward_zero_std": 0.875, "grad_norm": 0.2275390625, "kl": 0.0011217291685170494, "learning_rate": 9.217687074829932e-07, "loss": 0.0089, "num_tokens": 8360459.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 139, "step_time": 46.688301916234195 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1552.0, "completions/max_terminated_length": 1552.0, "completions/mean_length": 830.15625, "completions/mean_terminated_length": 830.15625, "completions/min_length": 378.0, "completions/min_terminated_length": 378.0, "entropy": 0.6795956548303366, "epoch": 0.07936507936507936, "frac_reward_zero_std": 0.75, "grad_norm": 0.39453125, "kl": 0.0011381846597942058, "learning_rate": 9.21201814058957e-07, "loss": -0.0006, "num_tokens": 8422341.0, "reward": 0.0625, "reward_std": 0.2745848298072815, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 140, "step_time": 60.8731788918376 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 707.375, "completions/mean_terminated_length": 707.375, "completions/min_length": 266.0, "completions/min_terminated_length": 266.0, "entropy": 0.8753399010747671, "epoch": 0.07993197278911565, "frac_reward_zero_std": 0.5, "grad_norm": 0.578125, "kl": 0.0014630026489612646, "learning_rate": 9.206349206349205e-07, "loss": 0.0028, "num_tokens": 8477165.0, "reward": 0.125, "reward_std": 0.48795002698898315, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 141, "step_time": 64.22281729802489 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1319.0, "completions/max_terminated_length": 1319.0, "completions/mean_length": 809.390625, "completions/mean_terminated_length": 809.390625, "completions/min_length": 357.0, "completions/min_terminated_length": 357.0, "entropy": 0.9984244965016842, "epoch": 0.08049886621315193, "frac_reward_zero_std": 0.875, "grad_norm": 0.2021484375, "kl": 0.0011269914175500162, "learning_rate": 9.200680272108843e-07, "loss": 0.0006, "num_tokens": 8539014.0, "reward": 0.0078125, "reward_std": 0.0625, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 142, "step_time": 51.55915431212634 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1253.0, "completions/max_terminated_length": 1253.0, "completions/mean_length": 814.359375, "completions/mean_terminated_length": 814.359375, "completions/min_length": 466.0, "completions/min_terminated_length": 466.0, "entropy": 0.9719417598098516, "epoch": 0.0810657596371882, "frac_reward_zero_std": 0.5, "grad_norm": 0.4765625, "kl": 0.0009653525630710647, "learning_rate": 9.19501133786848e-07, "loss": 0.0104, "num_tokens": 8601453.0, "reward": 0.1640625, "reward_std": 0.5422958731651306, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 143, "step_time": 50.030846861191094 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1326.0, "completions/max_terminated_length": 1326.0, "completions/mean_length": 650.203125, "completions/mean_terminated_length": 650.203125, "completions/min_length": 47.0, "completions/min_terminated_length": 47.0, "entropy": 0.9308004006743431, "epoch": 0.08163265306122448, "frac_reward_zero_std": 0.75, "grad_norm": 0.4375, "kl": 0.002023561413807329, "learning_rate": 9.189342403628118e-07, "loss": 0.0036, "num_tokens": 8653642.0, "reward": 0.0546875, "reward_std": 0.26899558305740356, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 144, "step_time": 51.93653193488717 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1297.0, "completions/max_terminated_length": 1297.0, "completions/mean_length": 787.859375, "completions/mean_terminated_length": 787.859375, "completions/min_length": 265.0, "completions/min_terminated_length": 265.0, "entropy": 0.8656461779028177, "epoch": 0.08219954648526077, "frac_reward_zero_std": 0.75, "grad_norm": 0.390625, "kl": 0.0011374048954166938, "learning_rate": 9.183673469387755e-07, "loss": -0.0089, "num_tokens": 8715849.0, "reward": 0.078125, "reward_std": 0.3587760329246521, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 145, "step_time": 51.37489194236696 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 4096.0, "completions/max_terminated_length": 1240.0, "completions/mean_length": 791.65625, "completions/mean_terminated_length": 739.2064208984375, "completions/min_length": 353.0, "completions/min_terminated_length": 353.0, "entropy": 0.8582950960844755, "epoch": 0.08276643990929705, "frac_reward_zero_std": 0.375, "grad_norm": 0.5859375, "kl": 0.0014971747732488438, "learning_rate": 9.178004535147393e-07, "loss": -0.0031, "num_tokens": 8779851.0, "reward": 0.2265625, "reward_std": 0.6294977068901062, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 146, "step_time": 184.58719532191753 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1267.0, "completions/max_terminated_length": 1267.0, "completions/mean_length": 861.484375, "completions/mean_terminated_length": 861.484375, "completions/min_length": 475.0, "completions/min_terminated_length": 475.0, "entropy": 0.8132954798638821, "epoch": 0.08333333333333333, "frac_reward_zero_std": 0.75, "grad_norm": 0.345703125, "kl": 0.001158217251941096, "learning_rate": 9.172335600907029e-07, "loss": 0.0055, "num_tokens": 8845346.0, "reward": 0.09375, "reward_std": 0.36596253514289856, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 147, "step_time": 49.048096066340804 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1306.0, "completions/max_terminated_length": 1306.0, "completions/mean_length": 756.921875, "completions/mean_terminated_length": 756.921875, "completions/min_length": 284.0, "completions/min_terminated_length": 284.0, "entropy": 0.8934296295046806, "epoch": 0.08390022675736962, "frac_reward_zero_std": 0.75, "grad_norm": 0.400390625, "kl": 0.0012354206082818564, "learning_rate": 9.166666666666665e-07, "loss": 0.0021, "num_tokens": 8903637.0, "reward": 0.109375, "reward_std": 0.4315783977508545, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 148, "step_time": 50.68960814829916 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1395.0, "completions/max_terminated_length": 1395.0, "completions/mean_length": 807.578125, "completions/mean_terminated_length": 807.578125, "completions/min_length": 309.0, "completions/min_terminated_length": 309.0, "entropy": 0.740367965772748, "epoch": 0.0844671201814059, "frac_reward_zero_std": 0.75, "grad_norm": 0.30859375, "kl": 0.0012059269793098792, "learning_rate": 9.160997732426303e-07, "loss": 0.0085, "num_tokens": 8965810.0, "reward": 0.09375, "reward_std": 0.426084041595459, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 149, "step_time": 54.18200596701354 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1610.0, "completions/max_terminated_length": 1610.0, "completions/mean_length": 812.3125, "completions/mean_terminated_length": 812.3125, "completions/min_length": 345.0, "completions/min_terminated_length": 345.0, "entropy": 0.5668262634426355, "epoch": 0.08503401360544217, "frac_reward_zero_std": 0.5, "grad_norm": 0.4609375, "kl": 0.0013014152864343487, "learning_rate": 9.15532879818594e-07, "loss": 0.0183, "num_tokens": 9027926.0, "reward": 0.296875, "reward_std": 0.69987952709198, "rewards/accuracy_reward_func/mean": 0.28125, "rewards/accuracy_reward_func/std": 0.7007648944854736, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 150, "step_time": 62.40979080926627 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1775.0, "completions/max_terminated_length": 1775.0, "completions/mean_length": 835.859375, "completions/mean_terminated_length": 835.859375, "completions/min_length": 210.0, "completions/min_terminated_length": 210.0, "entropy": 0.9155545383691788, "epoch": 0.08560090702947845, "frac_reward_zero_std": 0.75, "grad_norm": 0.384765625, "kl": 0.0012780445722455624, "learning_rate": 9.149659863945578e-07, "loss": 0.0195, "num_tokens": 9091717.0, "reward": 0.0546875, "reward_std": 0.26899558305740356, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 151, "step_time": 69.37227410264313 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1328.0, "completions/max_terminated_length": 1328.0, "completions/mean_length": 765.390625, "completions/mean_terminated_length": 765.390625, "completions/min_length": 339.0, "completions/min_terminated_length": 339.0, "entropy": 0.8690036237239838, "epoch": 0.08616780045351474, "frac_reward_zero_std": 0.75, "grad_norm": 0.404296875, "kl": 0.001788135792594403, "learning_rate": 9.143990929705215e-07, "loss": -0.0196, "num_tokens": 9151678.0, "reward": 0.0625, "reward_std": 0.2745848298072815, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 152, "step_time": 52.06068628747016 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 4096.0, "completions/max_terminated_length": 1262.0, "completions/mean_length": 783.9375, "completions/mean_terminated_length": 731.3651123046875, "completions/min_length": 306.0, "completions/min_terminated_length": 306.0, "entropy": 0.9219446405768394, "epoch": 0.08673469387755102, "frac_reward_zero_std": 0.75, "grad_norm": 0.376953125, "kl": 0.0015317276192945428, "learning_rate": 9.138321995464853e-07, "loss": -0.0009, "num_tokens": 9212578.0, "reward": 0.1015625, "reward_std": 0.3692515790462494, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 153, "step_time": 180.3162548020482 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1297.0, "completions/max_terminated_length": 1297.0, "completions/mean_length": 828.484375, "completions/mean_terminated_length": 828.484375, "completions/min_length": 484.0, "completions/min_terminated_length": 484.0, "entropy": 0.867816049605608, "epoch": 0.0873015873015873, "frac_reward_zero_std": 0.625, "grad_norm": 0.482421875, "kl": 0.0013349929031392094, "learning_rate": 9.132653061224489e-07, "loss": 0.0106, "num_tokens": 9275665.0, "reward": 0.0859375, "reward_std": 0.36247265338897705, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 154, "step_time": 50.62536738999188 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1414.0, "completions/max_terminated_length": 1414.0, "completions/mean_length": 749.203125, "completions/mean_terminated_length": 749.203125, "completions/min_length": 325.0, "completions/min_terminated_length": 325.0, "entropy": 0.8404502682387829, "epoch": 0.08786848072562359, "frac_reward_zero_std": 0.875, "grad_norm": 0.25, "kl": 0.0014105687332630623, "learning_rate": 9.126984126984127e-07, "loss": 0.0045, "num_tokens": 9338358.0, "reward": 0.03125, "reward_std": 0.12198750674724579, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 155, "step_time": 56.59618319571018 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1363.0, "completions/max_terminated_length": 1363.0, "completions/mean_length": 815.71875, "completions/mean_terminated_length": 815.71875, "completions/min_length": 410.0, "completions/min_terminated_length": 410.0, "entropy": 0.7059785295277834, "epoch": 0.08843537414965986, "frac_reward_zero_std": 0.5, "grad_norm": 0.484375, "kl": 0.0014341858404804952, "learning_rate": 9.121315192743763e-07, "loss": -0.0028, "num_tokens": 9401748.0, "reward": 0.1875, "reward_std": 0.5875696539878845, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 156, "step_time": 53.83780967630446 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1761.0, "completions/max_terminated_length": 1761.0, "completions/mean_length": 754.5625, "completions/mean_terminated_length": 754.5625, "completions/min_length": 142.0, "completions/min_terminated_length": 142.0, "entropy": 0.9523781202733517, "epoch": 0.08900226757369614, "frac_reward_zero_std": 0.5, "grad_norm": 0.5, "kl": 0.0024719537650526036, "learning_rate": 9.115646258503401e-07, "loss": -0.0176, "num_tokens": 9460232.0, "reward": 0.3046875, "reward_std": 0.6878606677055359, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 157, "step_time": 69.29571347311139 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1437.0, "completions/max_terminated_length": 1437.0, "completions/mean_length": 736.546875, "completions/mean_terminated_length": 736.546875, "completions/min_length": 339.0, "completions/min_terminated_length": 339.0, "entropy": 0.8924895077943802, "epoch": 0.08956916099773243, "frac_reward_zero_std": 0.625, "grad_norm": 0.45703125, "kl": 0.0014391718214028515, "learning_rate": 9.109977324263038e-07, "loss": -0.0143, "num_tokens": 9516203.0, "reward": 0.09375, "reward_std": 0.426084041595459, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 158, "step_time": 56.50125625450164 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2517.0, "completions/max_terminated_length": 2517.0, "completions/mean_length": 848.515625, "completions/mean_terminated_length": 848.515625, "completions/min_length": 340.0, "completions/min_terminated_length": 340.0, "entropy": 0.7691221535205841, "epoch": 0.09013605442176871, "frac_reward_zero_std": 0.625, "grad_norm": 0.37890625, "kl": 0.0012151955052104313, "learning_rate": 9.104308390022676e-07, "loss": 0.0268, "num_tokens": 9581780.0, "reward": 0.1875, "reward_std": 0.5875696539878845, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 159, "step_time": 100.30379501078278 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 3178.0, "completions/max_terminated_length": 3178.0, "completions/mean_length": 747.359375, "completions/mean_terminated_length": 747.359375, "completions/min_length": 376.0, "completions/min_terminated_length": 376.0, "entropy": 0.7731830514967442, "epoch": 0.09070294784580499, "frac_reward_zero_std": 0.75, "grad_norm": 0.287109375, "kl": 0.0014131366970104864, "learning_rate": 9.098639455782312e-07, "loss": -0.0093, "num_tokens": 9638755.0, "reward": 0.0625, "reward_std": 0.2745848298072815, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 160, "step_time": 129.41354488953948 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1599.0, "completions/max_terminated_length": 1599.0, "completions/mean_length": 847.859375, "completions/mean_terminated_length": 847.859375, "completions/min_length": 485.0, "completions/min_terminated_length": 485.0, "entropy": 0.879808509722352, "epoch": 0.09126984126984126, "frac_reward_zero_std": 0.25, "grad_norm": 0.58203125, "kl": 0.0015618964862369467, "learning_rate": 9.09297052154195e-07, "loss": -0.004, "num_tokens": 9704434.0, "reward": 0.2109375, "reward_std": 0.5896238088607788, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 161, "step_time": 61.810291890986264 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1938.0, "completions/max_terminated_length": 1938.0, "completions/mean_length": 793.609375, "completions/mean_terminated_length": 793.609375, "completions/min_length": 279.0, "completions/min_terminated_length": 279.0, "entropy": 0.5231372248381376, "epoch": 0.09183673469387756, "frac_reward_zero_std": 0.375, "grad_norm": 0.478515625, "kl": 0.0012791927074431442, "learning_rate": 9.087301587301587e-07, "loss": 0.052, "num_tokens": 9764505.0, "reward": 0.21875, "reward_std": 0.6291528940200806, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 162, "step_time": 75.29955735895783 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1246.0, "completions/max_terminated_length": 1246.0, "completions/mean_length": 740.59375, "completions/mean_terminated_length": 740.59375, "completions/min_length": 410.0, "completions/min_terminated_length": 410.0, "entropy": 0.957040335983038, "epoch": 0.09240362811791383, "frac_reward_zero_std": 0.625, "grad_norm": 0.4765625, "kl": 0.0019501148199196905, "learning_rate": 9.081632653061225e-07, "loss": 0.0302, "num_tokens": 9823023.0, "reward": 0.125, "reward_std": 0.4364357888698578, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 163, "step_time": 48.82911225967109 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1744.0, "completions/max_terminated_length": 1744.0, "completions/mean_length": 821.328125, "completions/mean_terminated_length": 821.328125, "completions/min_length": 464.0, "completions/min_terminated_length": 464.0, "entropy": 0.9202660918235779, "epoch": 0.09297052154195011, "frac_reward_zero_std": 0.75, "grad_norm": 0.3828125, "kl": 0.0016514871385879815, "learning_rate": 9.075963718820861e-07, "loss": -0.0187, "num_tokens": 9887660.0, "reward": 0.046875, "reward_std": 0.2630521357059479, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 164, "step_time": 67.61680617462844 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1266.0, "completions/max_terminated_length": 1266.0, "completions/mean_length": 723.953125, "completions/mean_terminated_length": 723.953125, "completions/min_length": 184.0, "completions/min_terminated_length": 184.0, "entropy": 0.6882897149771452, "epoch": 0.0935374149659864, "frac_reward_zero_std": 0.75, "grad_norm": 0.396484375, "kl": 0.0016417923288827296, "learning_rate": 9.070294784580499e-07, "loss": 0.0162, "num_tokens": 9942961.0, "reward": 0.1015625, "reward_std": 0.3692515790462494, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 165, "step_time": 49.54639142379165 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1351.0, "completions/max_terminated_length": 1351.0, "completions/mean_length": 721.015625, "completions/mean_terminated_length": 721.015625, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "entropy": 0.8900751657783985, "epoch": 0.09410430839002268, "frac_reward_zero_std": 0.875, "grad_norm": 0.291015625, "kl": 0.0020163542503723875, "learning_rate": 9.064625850340135e-07, "loss": -0.0039, "num_tokens": 10000034.0, "reward": 0.046875, "reward_std": 0.14689241349697113, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 166, "step_time": 54.27309240680188 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1461.0, "completions/max_terminated_length": 1461.0, "completions/mean_length": 768.78125, "completions/mean_terminated_length": 768.78125, "completions/min_length": 249.0, "completions/min_terminated_length": 249.0, "entropy": 0.6647769100964069, "epoch": 0.09467120181405896, "frac_reward_zero_std": 0.375, "grad_norm": 0.482421875, "kl": 0.001808562345104292, "learning_rate": 9.058956916099772e-07, "loss": 0.0153, "num_tokens": 10059292.0, "reward": 0.296875, "reward_std": 0.7331100702285767, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 167, "step_time": 57.41250906791538 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1128.0, "completions/max_terminated_length": 1128.0, "completions/mean_length": 678.5625, "completions/mean_terminated_length": 678.5625, "completions/min_length": 299.0, "completions/min_terminated_length": 299.0, "entropy": 0.8922697510570288, "epoch": 0.09523809523809523, "frac_reward_zero_std": 0.75, "grad_norm": 0.4140625, "kl": 0.0014247586659621447, "learning_rate": 9.05328798185941e-07, "loss": 0.0131, "num_tokens": 10113944.0, "reward": 0.078125, "reward_std": 0.3587760329246521, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 168, "step_time": 45.19113690778613 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1348.0, "completions/max_terminated_length": 1348.0, "completions/mean_length": 818.234375, "completions/mean_terminated_length": 818.234375, "completions/min_length": 481.0, "completions/min_terminated_length": 481.0, "entropy": 0.9066624715924263, "epoch": 0.09580498866213152, "frac_reward_zero_std": 0.75, "grad_norm": 0.361328125, "kl": 0.0017615591932553798, "learning_rate": 9.047619047619047e-07, "loss": -0.0157, "num_tokens": 10177631.0, "reward": 0.0703125, "reward_std": 0.35486623644828796, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 169, "step_time": 53.81898599676788 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1307.0, "completions/max_terminated_length": 1307.0, "completions/mean_length": 701.015625, "completions/mean_terminated_length": 701.015625, "completions/min_length": 14.0, "completions/min_terminated_length": 14.0, "entropy": 0.9111842215061188, "epoch": 0.0963718820861678, "frac_reward_zero_std": 0.75, "grad_norm": 0.447265625, "kl": 0.0025995390242314897, "learning_rate": 9.041950113378685e-07, "loss": 0.0198, "num_tokens": 10236248.0, "reward": 0.1328125, "reward_std": 0.4386323094367981, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 170, "step_time": 52.9567281100899 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1325.0, "completions/max_terminated_length": 1325.0, "completions/mean_length": 701.6875, "completions/mean_terminated_length": 701.6875, "completions/min_length": 362.0, "completions/min_terminated_length": 362.0, "entropy": 0.8338793143630028, "epoch": 0.09693877551020408, "frac_reward_zero_std": 0.5, "grad_norm": 0.546875, "kl": 0.0017993416768149473, "learning_rate": 9.036281179138322e-07, "loss": -0.0165, "num_tokens": 10290100.0, "reward": 0.2265625, "reward_std": 0.6294977068901062, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 171, "step_time": 52.467592406086624 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1762.0, "completions/max_terminated_length": 1762.0, "completions/mean_length": 759.390625, "completions/mean_terminated_length": 759.390625, "completions/min_length": 333.0, "completions/min_terminated_length": 333.0, "entropy": 0.9258961863815784, "epoch": 0.09750566893424037, "frac_reward_zero_std": 0.875, "grad_norm": 0.263671875, "kl": 0.002015009449678473, "learning_rate": 9.030612244897958e-07, "loss": -0.004, "num_tokens": 10348733.0, "reward": 0.03125, "reward_std": 0.12198750674724579, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 172, "step_time": 68.8363786675036 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1413.0, "completions/max_terminated_length": 1413.0, "completions/mean_length": 765.84375, "completions/mean_terminated_length": 765.84375, "completions/min_length": 407.0, "completions/min_terminated_length": 407.0, "entropy": 0.9442645497620106, "epoch": 0.09807256235827665, "frac_reward_zero_std": 0.375, "grad_norm": 0.58203125, "kl": 0.0017031964925990906, "learning_rate": 9.024943310657595e-07, "loss": 0.011, "num_tokens": 10407603.0, "reward": 0.234375, "reward_std": 0.6170124411582947, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 173, "step_time": 54.379828153178096 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1700.0, "completions/max_terminated_length": 1700.0, "completions/mean_length": 864.640625, "completions/mean_terminated_length": 864.640625, "completions/min_length": 283.0, "completions/min_terminated_length": 283.0, "entropy": 0.9839955344796181, "epoch": 0.09863945578231292, "frac_reward_zero_std": 0.875, "grad_norm": 0.234375, "kl": 0.0015942358804750256, "learning_rate": 9.019274376417233e-07, "loss": 0.0186, "num_tokens": 10473524.0, "reward": 0.015625, "reward_std": 0.08768405020236969, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 174, "step_time": 66.9548751115799 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1157.0, "completions/max_terminated_length": 1157.0, "completions/mean_length": 778.703125, "completions/mean_terminated_length": 778.703125, "completions/min_length": 413.0, "completions/min_terminated_length": 413.0, "entropy": 0.8145867474377155, "epoch": 0.0992063492063492, "frac_reward_zero_std": 0.75, "grad_norm": 0.31640625, "kl": 0.001623749420105014, "learning_rate": 9.01360544217687e-07, "loss": 0.0009, "num_tokens": 10536297.0, "reward": 0.1015625, "reward_std": 0.40999117493629456, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 175, "step_time": 47.00463814288378 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1492.0, "completions/max_terminated_length": 1492.0, "completions/mean_length": 772.640625, "completions/mean_terminated_length": 772.640625, "completions/min_length": 322.0, "completions/min_terminated_length": 322.0, "entropy": 0.8795701172202826, "epoch": 0.09977324263038549, "frac_reward_zero_std": 0.5, "grad_norm": 0.5703125, "kl": 0.0028222336040926166, "learning_rate": 9.007936507936508e-07, "loss": -0.003, "num_tokens": 10595850.0, "reward": 0.1640625, "reward_std": 0.5274578928947449, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 176, "step_time": 58.35854543372989 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1640.0, "completions/max_terminated_length": 1640.0, "completions/mean_length": 749.1875, "completions/mean_terminated_length": 749.1875, "completions/min_length": 358.0, "completions/min_terminated_length": 358.0, "entropy": 0.6966068465262651, "epoch": 0.10034013605442177, "frac_reward_zero_std": 0.75, "grad_norm": 0.34375, "kl": 0.0030260694547905587, "learning_rate": 9.002267573696145e-07, "loss": 0.0027, "num_tokens": 10653542.0, "reward": 0.1328125, "reward_std": 0.4386323094367981, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 177, "step_time": 64.24640219751745 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1382.0, "completions/max_terminated_length": 1382.0, "completions/mean_length": 713.21875, "completions/mean_terminated_length": 713.21875, "completions/min_length": 243.0, "completions/min_terminated_length": 243.0, "entropy": 0.997425977140665, "epoch": 0.10090702947845805, "frac_reward_zero_std": 0.625, "grad_norm": 0.412109375, "kl": 0.002245221330667846, "learning_rate": 8.996598639455783e-07, "loss": 0.0142, "num_tokens": 10708476.0, "reward": 0.09375, "reward_std": 0.36596253514289856, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 178, "step_time": 53.79487491585314 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1301.0, "completions/max_terminated_length": 1301.0, "completions/mean_length": 785.6875, "completions/mean_terminated_length": 785.6875, "completions/min_length": 360.0, "completions/min_terminated_length": 360.0, "entropy": 0.6753324847668409, "epoch": 0.10147392290249434, "frac_reward_zero_std": 0.75, "grad_norm": 0.326171875, "kl": 0.0016486272761540022, "learning_rate": 8.990929705215419e-07, "loss": 0.0045, "num_tokens": 10770408.0, "reward": 0.125, "reward_std": 0.37796446681022644, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 179, "step_time": 52.98196454253048 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1220.0, "completions/max_terminated_length": 1220.0, "completions/mean_length": 753.8125, "completions/mean_terminated_length": 753.8125, "completions/min_length": 299.0, "completions/min_terminated_length": 299.0, "entropy": 0.7863912023603916, "epoch": 0.10204081632653061, "frac_reward_zero_std": 0.625, "grad_norm": 0.400390625, "kl": 0.0020044324155605864, "learning_rate": 8.985260770975056e-07, "loss": 0.0256, "num_tokens": 10831492.0, "reward": 0.1015625, "reward_std": 0.42891231179237366, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 180, "step_time": 49.65730780083686 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1241.0, "completions/max_terminated_length": 1241.0, "completions/mean_length": 763.765625, "completions/mean_terminated_length": 763.765625, "completions/min_length": 290.0, "completions/min_terminated_length": 290.0, "entropy": 0.8663644455373287, "epoch": 0.10260770975056689, "frac_reward_zero_std": 0.625, "grad_norm": 0.4609375, "kl": 0.0022015888171154074, "learning_rate": 8.979591836734693e-07, "loss": 0.0119, "num_tokens": 10891677.0, "reward": 0.1875, "reward_std": 0.53079754114151, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 181, "step_time": 50.27756413444877 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1282.0, "completions/max_terminated_length": 1282.0, "completions/mean_length": 672.859375, "completions/mean_terminated_length": 672.859375, "completions/min_length": 186.0, "completions/min_terminated_length": 186.0, "entropy": 0.8014075439423323, "epoch": 0.10317460317460317, "frac_reward_zero_std": 0.375, "grad_norm": 0.5859375, "kl": 0.0026391869432700332, "learning_rate": 8.973922902494331e-07, "loss": -0.0063, "num_tokens": 10944524.0, "reward": 0.3125, "reward_std": 0.6871842741966248, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 182, "step_time": 49.701088985428214 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1633.0, "completions/max_terminated_length": 1633.0, "completions/mean_length": 701.78125, "completions/mean_terminated_length": 701.78125, "completions/min_length": 140.0, "completions/min_terminated_length": 140.0, "entropy": 0.9052042663097382, "epoch": 0.10374149659863946, "frac_reward_zero_std": 0.625, "grad_norm": 0.58984375, "kl": 0.0019432959816185758, "learning_rate": 8.968253968253968e-07, "loss": -0.0522, "num_tokens": 10999318.0, "reward": 0.109375, "reward_std": 0.4315783977508545, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 183, "step_time": 63.13254692032933 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1192.0, "completions/max_terminated_length": 1192.0, "completions/mean_length": 774.4375, "completions/mean_terminated_length": 774.4375, "completions/min_length": 309.0, "completions/min_terminated_length": 309.0, "entropy": 0.7909810151904821, "epoch": 0.10430839002267574, "frac_reward_zero_std": 0.375, "grad_norm": 0.61328125, "kl": 0.0019899419603461865, "learning_rate": 8.962585034013606e-07, "loss": 0.0044, "num_tokens": 11058786.0, "reward": 0.234375, "reward_std": 0.5907269716262817, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 184, "step_time": 46.25606250949204 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1666.0, "completions/max_terminated_length": 1666.0, "completions/mean_length": 810.3125, "completions/mean_terminated_length": 810.3125, "completions/min_length": 277.0, "completions/min_terminated_length": 277.0, "entropy": 0.8799561206251383, "epoch": 0.10487528344671201, "frac_reward_zero_std": 0.5, "grad_norm": 0.52734375, "kl": 0.002295402762683807, "learning_rate": 8.956916099773242e-07, "loss": 0.0031, "num_tokens": 11120150.0, "reward": 0.3203125, "reward_std": 0.6978833675384521, "rewards/accuracy_reward_func/mean": 0.28125, "rewards/accuracy_reward_func/std": 0.7007648944854736, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 185, "step_time": 65.20812787115574 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1911.0, "completions/max_terminated_length": 1911.0, "completions/mean_length": 819.484375, "completions/mean_terminated_length": 819.484375, "completions/min_length": 309.0, "completions/min_terminated_length": 309.0, "entropy": 0.6690396033227444, "epoch": 0.1054421768707483, "frac_reward_zero_std": 0.625, "grad_norm": 0.443359375, "kl": 0.0018117764229828026, "learning_rate": 8.95124716553288e-07, "loss": -0.0058, "num_tokens": 11184109.0, "reward": 0.1015625, "reward_std": 0.42891231179237366, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 186, "step_time": 75.52050837129354 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1695.0, "completions/max_terminated_length": 1695.0, "completions/mean_length": 776.59375, "completions/mean_terminated_length": 776.59375, "completions/min_length": 323.0, "completions/min_terminated_length": 323.0, "entropy": 0.5842441543936729, "epoch": 0.10600907029478458, "frac_reward_zero_std": 0.625, "grad_norm": 0.421875, "kl": 0.0021664260639226995, "learning_rate": 8.945578231292517e-07, "loss": 0.014, "num_tokens": 11243443.0, "reward": 0.1328125, "reward_std": 0.4386323094367981, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 187, "step_time": 65.45243280660361 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1191.0, "completions/max_terminated_length": 1191.0, "completions/mean_length": 705.875, "completions/mean_terminated_length": 705.875, "completions/min_length": 225.0, "completions/min_terminated_length": 225.0, "entropy": 0.6692053359001875, "epoch": 0.10657596371882086, "frac_reward_zero_std": 0.375, "grad_norm": 0.53125, "kl": 0.0031446133070858195, "learning_rate": 8.939909297052153e-07, "loss": -0.004, "num_tokens": 11298731.0, "reward": 0.265625, "reward_std": 0.6297439336776733, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 188, "step_time": 46.79448201600462 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1344.0, "completions/max_terminated_length": 1344.0, "completions/mean_length": 754.90625, "completions/mean_terminated_length": 754.90625, "completions/min_length": 286.0, "completions/min_terminated_length": 286.0, "entropy": 1.0256855934858322, "epoch": 0.10714285714285714, "frac_reward_zero_std": 0.75, "grad_norm": 0.361328125, "kl": 0.0024256880205939524, "learning_rate": 8.934240362811791e-07, "loss": 0.0306, "num_tokens": 11358741.0, "reward": 0.0703125, "reward_std": 0.27984103560447693, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 189, "step_time": 52.464261604472995 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1880.0, "completions/max_terminated_length": 1880.0, "completions/mean_length": 763.03125, "completions/mean_terminated_length": 763.03125, "completions/min_length": 268.0, "completions/min_terminated_length": 268.0, "entropy": 0.7950188089162111, "epoch": 0.10770975056689343, "frac_reward_zero_std": 0.875, "grad_norm": 0.29296875, "kl": 0.0021833389000676107, "learning_rate": 8.928571428571428e-07, "loss": 0.0295, "num_tokens": 11417503.0, "reward": 0.0234375, "reward_std": 0.10652101039886475, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 190, "step_time": 73.67372031044215 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1137.0, "completions/max_terminated_length": 1137.0, "completions/mean_length": 695.25, "completions/mean_terminated_length": 695.25, "completions/min_length": 360.0, "completions/min_terminated_length": 360.0, "entropy": 0.7993376590311527, "epoch": 0.1082766439909297, "frac_reward_zero_std": 0.375, "grad_norm": 0.58984375, "kl": 0.002104887673340272, "learning_rate": 8.922902494331065e-07, "loss": 0.0093, "num_tokens": 11474415.0, "reward": 0.203125, "reward_std": 0.5471335649490356, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 191, "step_time": 44.51445526443422 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1473.0, "completions/max_terminated_length": 1473.0, "completions/mean_length": 791.46875, "completions/mean_terminated_length": 791.46875, "completions/min_length": 399.0, "completions/min_terminated_length": 399.0, "entropy": 0.7037545405328274, "epoch": 0.10884353741496598, "frac_reward_zero_std": 0.875, "grad_norm": 0.2080078125, "kl": 0.00409951618348714, "learning_rate": 8.917233560090702e-07, "loss": -0.0098, "num_tokens": 11535149.0, "reward": 0.0703125, "reward_std": 0.33174845576286316, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 192, "step_time": 56.72403950151056 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1284.0, "completions/max_terminated_length": 1284.0, "completions/mean_length": 755.765625, "completions/mean_terminated_length": 755.765625, "completions/min_length": 194.0, "completions/min_terminated_length": 194.0, "entropy": 0.7793229520320892, "epoch": 0.10941043083900227, "frac_reward_zero_std": 0.75, "grad_norm": 0.291015625, "kl": 0.002601281084935181, "learning_rate": 8.91156462585034e-07, "loss": 0.0239, "num_tokens": 11593254.0, "reward": 0.0625, "reward_std": 0.35073620080947876, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 193, "step_time": 50.4504238832742 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1446.0, "completions/max_terminated_length": 1446.0, "completions/mean_length": 849.140625, "completions/mean_terminated_length": 849.140625, "completions/min_length": 495.0, "completions/min_terminated_length": 495.0, "entropy": 0.6648522336035967, "epoch": 0.10997732426303855, "frac_reward_zero_std": 0.5, "grad_norm": 0.4296875, "kl": 0.002077874007227365, "learning_rate": 8.905895691609977e-07, "loss": 0.0106, "num_tokens": 11656559.0, "reward": 0.1328125, "reward_std": 0.489915668964386, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 194, "step_time": 56.28423589468002 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1553.0, "completions/max_terminated_length": 1553.0, "completions/mean_length": 688.515625, "completions/mean_terminated_length": 688.515625, "completions/min_length": 214.0, "completions/min_terminated_length": 214.0, "entropy": 0.8349446803331375, "epoch": 0.11054421768707483, "frac_reward_zero_std": 0.625, "grad_norm": 0.50390625, "kl": 0.003143731963064056, "learning_rate": 8.900226757369615e-07, "loss": 0.0065, "num_tokens": 11709888.0, "reward": 0.125, "reward_std": 0.4714045226573944, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 195, "step_time": 60.17381243593991 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1791.0, "completions/max_terminated_length": 1791.0, "completions/mean_length": 797.109375, "completions/mean_terminated_length": 797.109375, "completions/min_length": 329.0, "completions/min_terminated_length": 329.0, "entropy": 0.9731405824422836, "epoch": 0.1111111111111111, "frac_reward_zero_std": 0.375, "grad_norm": 0.6875, "kl": 0.002387014923442621, "learning_rate": 8.894557823129251e-07, "loss": -0.0437, "num_tokens": 11775207.0, "reward": 0.2109375, "reward_std": 0.6412084698677063, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 196, "step_time": 72.48106692917645 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1246.0, "completions/max_terminated_length": 1246.0, "completions/mean_length": 653.96875, "completions/mean_terminated_length": 653.96875, "completions/min_length": 144.0, "completions/min_terminated_length": 144.0, "entropy": 0.8588339984416962, "epoch": 0.1116780045351474, "frac_reward_zero_std": 0.625, "grad_norm": 0.5390625, "kl": 0.0036632285846280865, "learning_rate": 8.888888888888888e-07, "loss": -0.0134, "num_tokens": 11829389.0, "reward": 0.2109375, "reward_std": 0.5896238088607788, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 197, "step_time": 49.20649313181639 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1651.0, "completions/max_terminated_length": 1651.0, "completions/mean_length": 700.578125, "completions/mean_terminated_length": 700.578125, "completions/min_length": 203.0, "completions/min_terminated_length": 203.0, "entropy": 0.8202382121235132, "epoch": 0.11224489795918367, "frac_reward_zero_std": 0.625, "grad_norm": 0.423828125, "kl": 0.002230228237749543, "learning_rate": 8.883219954648525e-07, "loss": -0.0085, "num_tokens": 11885482.0, "reward": 0.1796875, "reward_std": 0.49894481897354126, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 198, "step_time": 64.34397663455456 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 4096.0, "completions/max_terminated_length": 1470.0, "completions/mean_length": 768.25, "completions/mean_terminated_length": 715.4285888671875, "completions/min_length": 240.0, "completions/min_terminated_length": 240.0, "entropy": 0.7309875469654799, "epoch": 0.11281179138321995, "frac_reward_zero_std": 0.75, "grad_norm": 0.337890625, "kl": 0.00237447653489653, "learning_rate": 8.877551020408163e-07, "loss": 0.0032, "num_tokens": 11943818.0, "reward": 0.0859375, "reward_std": 0.4038965702056885, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 199, "step_time": 179.59818591643125 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1305.0, "completions/max_terminated_length": 1305.0, "completions/mean_length": 728.984375, "completions/mean_terminated_length": 728.984375, "completions/min_length": 319.0, "completions/min_terminated_length": 319.0, "entropy": 0.963052537292242, "epoch": 0.11337868480725624, "frac_reward_zero_std": 0.625, "grad_norm": 0.44921875, "kl": 0.002730268613959197, "learning_rate": 8.8718820861678e-07, "loss": 0.0071, "num_tokens": 12000401.0, "reward": 0.109375, "reward_std": 0.37234508991241455, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 200, "step_time": 51.364328576251864 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1272.0, "completions/max_terminated_length": 1272.0, "completions/mean_length": 699.046875, "completions/mean_terminated_length": 699.046875, "completions/min_length": 400.0, "completions/min_terminated_length": 400.0, "entropy": 0.7668270245194435, "epoch": 0.11394557823129252, "frac_reward_zero_std": 0.5, "grad_norm": 0.51953125, "kl": 0.0026685933116823435, "learning_rate": 8.866213151927438e-07, "loss": 0.0066, "num_tokens": 12055452.0, "reward": 0.25, "reward_std": 0.6299408078193665, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 201, "step_time": 50.06663680355996 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1189.0, "completions/max_terminated_length": 1189.0, "completions/mean_length": 669.21875, "completions/mean_terminated_length": 669.21875, "completions/min_length": 11.0, "completions/min_terminated_length": 11.0, "entropy": 0.8293571621179581, "epoch": 0.1145124716553288, "frac_reward_zero_std": 0.625, "grad_norm": 0.52734375, "kl": 0.002398704025836196, "learning_rate": 8.860544217687075e-07, "loss": -0.0032, "num_tokens": 12110410.0, "reward": 0.1484375, "reward_std": 0.49344661831855774, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 202, "step_time": 47.57376485038549 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1171.0, "completions/max_terminated_length": 1171.0, "completions/mean_length": 698.9375, "completions/mean_terminated_length": 698.9375, "completions/min_length": 314.0, "completions/min_terminated_length": 314.0, "entropy": 1.000132780522108, "epoch": 0.11507936507936507, "frac_reward_zero_std": 0.875, "grad_norm": 0.314453125, "kl": 0.002165637895814143, "learning_rate": 8.854875283446712e-07, "loss": 0.0056, "num_tokens": 12164742.0, "reward": 0.046875, "reward_std": 0.14689241349697113, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 203, "step_time": 46.32271307427436 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1144.0, "completions/max_terminated_length": 1144.0, "completions/mean_length": 746.25, "completions/mean_terminated_length": 746.25, "completions/min_length": 297.0, "completions/min_terminated_length": 297.0, "entropy": 0.9319121092557907, "epoch": 0.11564625850340136, "frac_reward_zero_std": 0.75, "grad_norm": 0.365234375, "kl": 0.002491012099198997, "learning_rate": 8.849206349206348e-07, "loss": -0.009, "num_tokens": 12222470.0, "reward": 0.0546875, "reward_std": 0.26899558305740356, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 204, "step_time": 45.10238230600953 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1662.0, "completions/max_terminated_length": 1662.0, "completions/mean_length": 806.640625, "completions/mean_terminated_length": 806.640625, "completions/min_length": 204.0, "completions/min_terminated_length": 204.0, "entropy": 0.6651069391518831, "epoch": 0.11621315192743764, "frac_reward_zero_std": 0.625, "grad_norm": 0.41796875, "kl": 0.003674557614431251, "learning_rate": 8.843537414965985e-07, "loss": 0.0148, "num_tokens": 12283039.0, "reward": 0.2421875, "reward_std": 0.6423677802085876, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 205, "step_time": 64.43609751947224 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1808.0, "completions/max_terminated_length": 1808.0, "completions/mean_length": 818.53125, "completions/mean_terminated_length": 818.53125, "completions/min_length": 319.0, "completions/min_terminated_length": 319.0, "entropy": 0.7762941531836987, "epoch": 0.11678004535147392, "frac_reward_zero_std": 0.25, "grad_norm": 0.625, "kl": 0.003014502683072351, "learning_rate": 8.837868480725623e-07, "loss": -0.024, "num_tokens": 12347177.0, "reward": 0.28125, "reward_std": 0.7339553236961365, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 206, "step_time": 71.33777440246195 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1298.0, "completions/max_terminated_length": 1298.0, "completions/mean_length": 698.671875, "completions/mean_terminated_length": 698.671875, "completions/min_length": 377.0, "completions/min_terminated_length": 377.0, "entropy": 0.6892206575721502, "epoch": 0.11734693877551021, "frac_reward_zero_std": 0.625, "grad_norm": 0.470703125, "kl": 0.004469722611247562, "learning_rate": 8.83219954648526e-07, "loss": -0.0029, "num_tokens": 12403292.0, "reward": 0.21875, "reward_std": 0.6416434645652771, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 207, "step_time": 51.56914662569761 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1361.0, "completions/max_terminated_length": 1361.0, "completions/mean_length": 708.09375, "completions/mean_terminated_length": 708.09375, "completions/min_length": 370.0, "completions/min_terminated_length": 370.0, "entropy": 0.8025732468813658, "epoch": 0.11791383219954649, "frac_reward_zero_std": 0.75, "grad_norm": 0.3984375, "kl": 0.002947663400846068, "learning_rate": 8.826530612244898e-07, "loss": -0.0204, "num_tokens": 12458490.0, "reward": 0.078125, "reward_std": 0.2847827076911926, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 208, "step_time": 52.94201644603163 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1789.0, "completions/max_terminated_length": 1789.0, "completions/mean_length": 809.0, "completions/mean_terminated_length": 809.0, "completions/min_length": 250.0, "completions/min_terminated_length": 250.0, "entropy": 0.7231426276266575, "epoch": 0.11848072562358276, "frac_reward_zero_std": 0.75, "grad_norm": 0.388671875, "kl": 0.0021686839136236813, "learning_rate": 8.820861678004535e-07, "loss": 0.0086, "num_tokens": 12519674.0, "reward": 0.046875, "reward_std": 0.14689241349697113, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 209, "step_time": 68.73250697460026 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1746.0, "completions/max_terminated_length": 1746.0, "completions/mean_length": 700.84375, "completions/mean_terminated_length": 700.84375, "completions/min_length": 292.0, "completions/min_terminated_length": 292.0, "entropy": 0.7687572091817856, "epoch": 0.11904761904761904, "frac_reward_zero_std": 0.625, "grad_norm": 0.482421875, "kl": 0.003985284827649593, "learning_rate": 8.815192743764172e-07, "loss": 0.0077, "num_tokens": 12574656.0, "reward": 0.1171875, "reward_std": 0.43408530950546265, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 210, "step_time": 67.19688565097749 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1376.0, "completions/max_terminated_length": 1376.0, "completions/mean_length": 775.703125, "completions/mean_terminated_length": 775.703125, "completions/min_length": 273.0, "completions/min_terminated_length": 273.0, "entropy": 0.7454958781599998, "epoch": 0.11961451247165533, "frac_reward_zero_std": 0.625, "grad_norm": 0.478515625, "kl": 0.0025180759803333785, "learning_rate": 8.809523809523809e-07, "loss": 0.0144, "num_tokens": 12634093.0, "reward": 0.1328125, "reward_std": 0.4386323094367981, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 211, "step_time": 54.07804360054433 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1331.0, "completions/max_terminated_length": 1331.0, "completions/mean_length": 844.921875, "completions/mean_terminated_length": 844.921875, "completions/min_length": 367.0, "completions/min_terminated_length": 367.0, "entropy": 0.7186903860419989, "epoch": 0.12018140589569161, "frac_reward_zero_std": 0.75, "grad_norm": 0.376953125, "kl": 0.0023014823818812147, "learning_rate": 8.803854875283446e-07, "loss": -0.0052, "num_tokens": 12698072.0, "reward": 0.078125, "reward_std": 0.3587760329246521, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 212, "step_time": 52.086543631739914 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1639.0, "completions/max_terminated_length": 1639.0, "completions/mean_length": 802.984375, "completions/mean_terminated_length": 802.984375, "completions/min_length": 191.0, "completions/min_terminated_length": 191.0, "entropy": 0.7671890594065189, "epoch": 0.12074829931972789, "frac_reward_zero_std": 0.75, "grad_norm": 0.2734375, "kl": 0.004307562980102375, "learning_rate": 8.798185941043083e-07, "loss": 0.0089, "num_tokens": 12761055.0, "reward": 0.1796875, "reward_std": 0.5866720080375671, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 213, "step_time": 64.068588770926 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1307.0, "completions/max_terminated_length": 1307.0, "completions/mean_length": 792.921875, "completions/mean_terminated_length": 792.921875, "completions/min_length": 388.0, "completions/min_terminated_length": 388.0, "entropy": 0.6857465747743845, "epoch": 0.12131519274376418, "frac_reward_zero_std": 0.5, "grad_norm": 0.4453125, "kl": 0.0037605646284646355, "learning_rate": 8.792517006802721e-07, "loss": 0.0003, "num_tokens": 12822554.0, "reward": 0.1796875, "reward_std": 0.5298036932945251, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 214, "step_time": 53.09345087129623 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1427.0, "completions/max_terminated_length": 1427.0, "completions/mean_length": 748.765625, "completions/mean_terminated_length": 748.765625, "completions/min_length": 301.0, "completions/min_terminated_length": 301.0, "entropy": 0.7146262265741825, "epoch": 0.12188208616780045, "frac_reward_zero_std": 0.75, "grad_norm": 0.35546875, "kl": 0.0030410429681069218, "learning_rate": 8.786848072562358e-07, "loss": -0.0116, "num_tokens": 12880883.0, "reward": 0.1328125, "reward_std": 0.4734388589859009, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 215, "step_time": 58.4134808331728 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1176.0, "completions/max_terminated_length": 1176.0, "completions/mean_length": 782.625, "completions/mean_terminated_length": 782.625, "completions/min_length": 418.0, "completions/min_terminated_length": 418.0, "entropy": 1.0164642632007599, "epoch": 0.12244897959183673, "frac_reward_zero_std": 0.625, "grad_norm": 0.484375, "kl": 0.004625954588846071, "learning_rate": 8.781179138321995e-07, "loss": 0.0318, "num_tokens": 12945563.0, "reward": 0.109375, "reward_std": 0.4315783977508545, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 216, "step_time": 47.95185865368694 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1429.0, "completions/max_terminated_length": 1429.0, "completions/mean_length": 827.921875, "completions/mean_terminated_length": 827.921875, "completions/min_length": 429.0, "completions/min_terminated_length": 429.0, "entropy": 0.765678521245718, "epoch": 0.12301587301587301, "frac_reward_zero_std": 0.5, "grad_norm": 0.53125, "kl": 0.002401860627287533, "learning_rate": 8.775510204081632e-07, "loss": 0.0101, "num_tokens": 13008662.0, "reward": 0.109375, "reward_std": 0.4315783977508545, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 217, "step_time": 56.30959102790803 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2061.0, "completions/max_terminated_length": 2061.0, "completions/mean_length": 891.515625, "completions/mean_terminated_length": 891.515625, "completions/min_length": 443.0, "completions/min_terminated_length": 443.0, "entropy": 0.7714218124747276, "epoch": 0.1235827664399093, "frac_reward_zero_std": 0.5, "grad_norm": 0.474609375, "kl": 0.003679653353174217, "learning_rate": 8.76984126984127e-07, "loss": -0.021, "num_tokens": 13076471.0, "reward": 0.171875, "reward_std": 0.5434950590133667, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 218, "step_time": 80.66285381931812 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1234.0, "completions/max_terminated_length": 1234.0, "completions/mean_length": 693.625, "completions/mean_terminated_length": 693.625, "completions/min_length": 180.0, "completions/min_terminated_length": 180.0, "entropy": 0.789083743467927, "epoch": 0.12414965986394558, "frac_reward_zero_std": 0.375, "grad_norm": 0.66796875, "kl": 0.004488422848226037, "learning_rate": 8.764172335600907e-07, "loss": 0.0159, "num_tokens": 13130831.0, "reward": 0.234375, "reward_std": 0.6297439336776733, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 219, "step_time": 48.07829122338444 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1337.0, "completions/max_terminated_length": 1337.0, "completions/mean_length": 709.78125, "completions/mean_terminated_length": 709.78125, "completions/min_length": 257.0, "completions/min_terminated_length": 257.0, "entropy": 1.0191886126995087, "epoch": 0.12471655328798185, "frac_reward_zero_std": 0.875, "grad_norm": 0.2734375, "kl": 0.0039866922234068625, "learning_rate": 8.758503401360544e-07, "loss": -0.0108, "num_tokens": 13186977.0, "reward": 0.0234375, "reward_std": 0.10652101039886475, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 220, "step_time": 51.81976623181254 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1464.0, "completions/max_terminated_length": 1464.0, "completions/mean_length": 823.171875, "completions/mean_terminated_length": 823.171875, "completions/min_length": 475.0, "completions/min_terminated_length": 475.0, "entropy": 0.7741623353213072, "epoch": 0.12528344671201813, "frac_reward_zero_std": 0.625, "grad_norm": 0.408203125, "kl": 0.004347501919255592, "learning_rate": 8.752834467120181e-07, "loss": -0.0022, "num_tokens": 13250300.0, "reward": 0.1171875, "reward_std": 0.43408530950546265, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 221, "step_time": 56.70674694888294 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1369.0, "completions/max_terminated_length": 1369.0, "completions/mean_length": 766.46875, "completions/mean_terminated_length": 766.46875, "completions/min_length": 255.0, "completions/min_terminated_length": 255.0, "entropy": 0.862474599853158, "epoch": 0.12585034013605442, "frac_reward_zero_std": 0.5, "grad_norm": 0.53515625, "kl": 0.0032451720435346942, "learning_rate": 8.747165532879818e-07, "loss": -0.0089, "num_tokens": 13311818.0, "reward": 0.265625, "reward_std": 0.6544642448425293, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 222, "step_time": 54.62590129673481 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1355.0, "completions/max_terminated_length": 1355.0, "completions/mean_length": 647.515625, "completions/mean_terminated_length": 647.515625, "completions/min_length": 284.0, "completions/min_terminated_length": 284.0, "entropy": 0.9779625628143549, "epoch": 0.1264172335600907, "frac_reward_zero_std": 0.625, "grad_norm": 0.5390625, "kl": 0.004331238793383818, "learning_rate": 8.741496598639455e-07, "loss": 0.0183, "num_tokens": 13362787.0, "reward": 0.1875, "reward_std": 0.5, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 223, "step_time": 52.470416551455855 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1475.0, "completions/max_terminated_length": 1475.0, "completions/mean_length": 722.671875, "completions/mean_terminated_length": 722.671875, "completions/min_length": 358.0, "completions/min_terminated_length": 358.0, "entropy": 0.7243257965892553, "epoch": 0.12698412698412698, "frac_reward_zero_std": 0.5, "grad_norm": 0.48828125, "kl": 0.004345040906628128, "learning_rate": 8.735827664399093e-07, "loss": 0.0078, "num_tokens": 13418782.0, "reward": 0.328125, "reward_std": 0.7622999548912048, "rewards/accuracy_reward_func/mean": 0.28125, "rewards/accuracy_reward_func/std": 0.7007648944854736, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 224, "step_time": 57.42830611113459 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1300.0, "completions/max_terminated_length": 1300.0, "completions/mean_length": 712.6875, "completions/mean_terminated_length": 712.6875, "completions/min_length": 235.0, "completions/min_terminated_length": 235.0, "entropy": 0.9633335061371326, "epoch": 0.12755102040816327, "frac_reward_zero_std": 0.75, "grad_norm": 0.41796875, "kl": 0.003034610257600434, "learning_rate": 8.73015873015873e-07, "loss": -0.007, "num_tokens": 13480690.0, "reward": 0.2109375, "reward_std": 0.653468668460846, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 225, "step_time": 54.13289635069668 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2195.0, "completions/max_terminated_length": 2195.0, "completions/mean_length": 703.703125, "completions/mean_terminated_length": 703.703125, "completions/min_length": 323.0, "completions/min_terminated_length": 323.0, "entropy": 0.6303536836057901, "epoch": 0.12811791383219956, "frac_reward_zero_std": 0.625, "grad_norm": 0.46875, "kl": 0.004575064493110403, "learning_rate": 8.724489795918368e-07, "loss": 0.016, "num_tokens": 13535311.0, "reward": 0.1484375, "reward_std": 0.44257262349128723, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 226, "step_time": 85.67797674704343 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1494.0, "completions/max_terminated_length": 1494.0, "completions/mean_length": 750.703125, "completions/mean_terminated_length": 750.703125, "completions/min_length": 371.0, "completions/min_terminated_length": 371.0, "entropy": 0.8935067839920521, "epoch": 0.12868480725623582, "frac_reward_zero_std": 0.625, "grad_norm": 0.482421875, "kl": 0.005571341884206049, "learning_rate": 8.718820861678005e-07, "loss": -0.0032, "num_tokens": 13593484.0, "reward": 0.125, "reward_std": 0.4364357888698578, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 227, "step_time": 58.77007927093655 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1265.0, "completions/max_terminated_length": 1265.0, "completions/mean_length": 708.484375, "completions/mean_terminated_length": 708.484375, "completions/min_length": 276.0, "completions/min_terminated_length": 276.0, "entropy": 0.8600424602627754, "epoch": 0.1292517006802721, "frac_reward_zero_std": 0.5, "grad_norm": 0.578125, "kl": 0.004000057917437516, "learning_rate": 8.71315192743764e-07, "loss": -0.0138, "num_tokens": 13648171.0, "reward": 0.2734375, "reward_std": 0.6294977068901062, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 228, "step_time": 48.869283626787364 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1423.0, "completions/max_terminated_length": 1423.0, "completions/mean_length": 818.390625, "completions/mean_terminated_length": 818.390625, "completions/min_length": 308.0, "completions/min_terminated_length": 308.0, "entropy": 0.6593409366905689, "epoch": 0.12981859410430838, "frac_reward_zero_std": 0.625, "grad_norm": 0.3671875, "kl": 0.0035780292018898763, "learning_rate": 8.707482993197278e-07, "loss": 0.0011, "num_tokens": 13710476.0, "reward": 0.234375, "reward_std": 0.6782841682434082, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 229, "step_time": 55.18314048089087 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 649.5625, "completions/mean_terminated_length": 649.5625, "completions/min_length": 270.0, "completions/min_terminated_length": 270.0, "entropy": 1.0032381936907768, "epoch": 0.13038548752834467, "frac_reward_zero_std": 0.625, "grad_norm": 0.53125, "kl": 0.004240339570969809, "learning_rate": 8.701814058956915e-07, "loss": 0.0016, "num_tokens": 13762320.0, "reward": 0.1171875, "reward_std": 0.43408530950546265, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 230, "step_time": 60.95228326693177 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1380.0, "completions/max_terminated_length": 1380.0, "completions/mean_length": 783.96875, "completions/mean_terminated_length": 783.96875, "completions/min_length": 242.0, "completions/min_terminated_length": 242.0, "entropy": 0.7472527362406254, "epoch": 0.13095238095238096, "frac_reward_zero_std": 0.5, "grad_norm": 0.46484375, "kl": 0.005484887624334078, "learning_rate": 8.696145124716553e-07, "loss": 0.0353, "num_tokens": 13821494.0, "reward": 0.1640625, "reward_std": 0.496453195810318, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 231, "step_time": 54.40403426066041 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1246.0, "completions/max_terminated_length": 1246.0, "completions/mean_length": 776.765625, "completions/mean_terminated_length": 776.765625, "completions/min_length": 381.0, "completions/min_terminated_length": 381.0, "entropy": 0.7287231758236885, "epoch": 0.13151927437641722, "frac_reward_zero_std": 0.75, "grad_norm": 0.3828125, "kl": 0.003164727291732561, "learning_rate": 8.69047619047619e-07, "loss": 0.0172, "num_tokens": 13882103.0, "reward": 0.109375, "reward_std": 0.4315783977508545, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 232, "step_time": 49.35764399077743 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1446.0, "completions/max_terminated_length": 1446.0, "completions/mean_length": 740.796875, "completions/mean_terminated_length": 740.796875, "completions/min_length": 266.0, "completions/min_terminated_length": 266.0, "entropy": 0.8617698810994625, "epoch": 0.1320861678004535, "frac_reward_zero_std": 0.75, "grad_norm": 0.404296875, "kl": 0.0035138000966981053, "learning_rate": 8.684807256235828e-07, "loss": 0.0197, "num_tokens": 13939714.0, "reward": 0.0625, "reward_std": 0.2745848298072815, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 233, "step_time": 55.777379545383155 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1309.0, "completions/max_terminated_length": 1309.0, "completions/mean_length": 686.359375, "completions/mean_terminated_length": 686.359375, "completions/min_length": 204.0, "completions/min_terminated_length": 204.0, "entropy": 1.0610783658921719, "epoch": 0.1326530612244898, "frac_reward_zero_std": 0.875, "grad_norm": 0.3515625, "kl": 0.006404620158718899, "learning_rate": 8.679138321995464e-07, "loss": 0.002, "num_tokens": 13994833.0, "reward": 0.015625, "reward_std": 0.08768405020236969, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 234, "step_time": 51.19552284292877 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 4096.0, "completions/max_terminated_length": 1355.0, "completions/mean_length": 816.0, "completions/mean_terminated_length": 763.9365234375, "completions/min_length": 382.0, "completions/min_terminated_length": 382.0, "entropy": 0.9454163573682308, "epoch": 0.13321995464852607, "frac_reward_zero_std": 0.75, "grad_norm": 0.328125, "kl": 0.004431686924363021, "learning_rate": 8.673469387755102e-07, "loss": -0.0152, "num_tokens": 14061857.0, "reward": 0.1796875, "reward_std": 0.5729843378067017, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 235, "step_time": 184.50176417734474 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1570.0, "completions/max_terminated_length": 1570.0, "completions/mean_length": 843.78125, "completions/mean_terminated_length": 843.78125, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "entropy": 0.7127547487616539, "epoch": 0.13378684807256236, "frac_reward_zero_std": 0.75, "grad_norm": 0.337890625, "kl": 0.007691268219787162, "learning_rate": 8.667800453514738e-07, "loss": -0.0137, "num_tokens": 14129947.0, "reward": 0.0703125, "reward_std": 0.27984103560447693, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 236, "step_time": 61.732176676392555 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1457.0, "completions/max_terminated_length": 1457.0, "completions/mean_length": 693.15625, "completions/mean_terminated_length": 693.15625, "completions/min_length": 305.0, "completions/min_terminated_length": 305.0, "entropy": 0.8555677160620689, "epoch": 0.13435374149659865, "frac_reward_zero_std": 0.75, "grad_norm": 0.361328125, "kl": 0.010487258521607146, "learning_rate": 8.662131519274376e-07, "loss": 0.0315, "num_tokens": 14184837.0, "reward": 0.0546875, "reward_std": 0.26899558305740356, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 237, "step_time": 56.15231809951365 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 4086.0, "completions/max_terminated_length": 4086.0, "completions/mean_length": 884.59375, "completions/mean_terminated_length": 884.59375, "completions/min_length": 380.0, "completions/min_terminated_length": 380.0, "entropy": 0.6391396187245846, "epoch": 0.1349206349206349, "frac_reward_zero_std": 0.625, "grad_norm": 0.33984375, "kl": 0.003827203137916513, "learning_rate": 8.656462585034013e-07, "loss": 0.0065, "num_tokens": 14251563.0, "reward": 0.140625, "reward_std": 0.4753340482711792, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 238, "step_time": 179.5403899597004 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2424.0, "completions/max_terminated_length": 2424.0, "completions/mean_length": 756.859375, "completions/mean_terminated_length": 756.859375, "completions/min_length": 14.0, "completions/min_terminated_length": 14.0, "entropy": 0.9389139264822006, "epoch": 0.1354875283446712, "frac_reward_zero_std": 0.625, "grad_norm": 0.455078125, "kl": 0.0035379417677177116, "learning_rate": 8.650793650793651e-07, "loss": 0.0103, "num_tokens": 14309914.0, "reward": 0.1640625, "reward_std": 0.496453195810318, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 239, "step_time": 94.99134722538292 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 785.65625, "completions/mean_terminated_length": 785.65625, "completions/min_length": 292.0, "completions/min_terminated_length": 292.0, "entropy": 0.8213918693363667, "epoch": 0.1360544217687075, "frac_reward_zero_std": 0.75, "grad_norm": 0.396484375, "kl": 0.004000949513283558, "learning_rate": 8.645124716553288e-07, "loss": 0.0167, "num_tokens": 14369980.0, "reward": 0.0703125, "reward_std": 0.35486623644828796, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 240, "step_time": 58.402712346054614 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2243.0, "completions/max_terminated_length": 2243.0, "completions/mean_length": 775.125, "completions/mean_terminated_length": 775.125, "completions/min_length": 250.0, "completions/min_terminated_length": 250.0, "entropy": 0.6468172837048769, "epoch": 0.13662131519274376, "frac_reward_zero_std": 0.75, "grad_norm": 0.337890625, "kl": 0.009005385334603488, "learning_rate": 8.639455782312925e-07, "loss": 0.0185, "num_tokens": 14429644.0, "reward": 0.1640625, "reward_std": 0.496453195810318, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 241, "step_time": 89.81586427427828 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1471.0, "completions/max_terminated_length": 1471.0, "completions/mean_length": 817.703125, "completions/mean_terminated_length": 817.703125, "completions/min_length": 298.0, "completions/min_terminated_length": 298.0, "entropy": 0.7675363086163998, "epoch": 0.13718820861678005, "frac_reward_zero_std": 0.625, "grad_norm": 0.40234375, "kl": 0.003092183542321436, "learning_rate": 8.633786848072562e-07, "loss": -0.0005, "num_tokens": 14491881.0, "reward": 0.1484375, "reward_std": 0.44257262349128723, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 242, "step_time": 57.815207845531404 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1283.0, "completions/max_terminated_length": 1283.0, "completions/mean_length": 643.5625, "completions/mean_terminated_length": 643.5625, "completions/min_length": 206.0, "completions/min_terminated_length": 206.0, "entropy": 0.854773510247469, "epoch": 0.1377551020408163, "frac_reward_zero_std": 0.375, "grad_norm": 0.66015625, "kl": 0.004636547317204531, "learning_rate": 8.6281179138322e-07, "loss": -0.0201, "num_tokens": 14544165.0, "reward": 0.203125, "reward_std": 0.5471335649490356, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 243, "step_time": 51.11892264150083 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1606.0, "completions/max_terminated_length": 1606.0, "completions/mean_length": 752.578125, "completions/mean_terminated_length": 752.578125, "completions/min_length": 278.0, "completions/min_terminated_length": 278.0, "entropy": 0.7487936839461327, "epoch": 0.1383219954648526, "frac_reward_zero_std": 0.375, "grad_norm": 0.5703125, "kl": 0.0038994086789898574, "learning_rate": 8.622448979591836e-07, "loss": -0.0097, "num_tokens": 14602258.0, "reward": 0.265625, "reward_std": 0.6782841682434082, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 244, "step_time": 62.81886448804289 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1134.0, "completions/max_terminated_length": 1134.0, "completions/mean_length": 730.46875, "completions/mean_terminated_length": 730.46875, "completions/min_length": 387.0, "completions/min_terminated_length": 387.0, "entropy": 0.8416277430951595, "epoch": 0.1388888888888889, "frac_reward_zero_std": 0.75, "grad_norm": 0.40625, "kl": 0.004320491309044883, "learning_rate": 8.616780045351474e-07, "loss": -0.0115, "num_tokens": 14662184.0, "reward": 0.1015625, "reward_std": 0.3692515790462494, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 245, "step_time": 46.775943524204195 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1520.0, "completions/max_terminated_length": 1520.0, "completions/mean_length": 813.6875, "completions/mean_terminated_length": 813.6875, "completions/min_length": 332.0, "completions/min_terminated_length": 332.0, "entropy": 0.8341892696917057, "epoch": 0.13945578231292516, "frac_reward_zero_std": 0.75, "grad_norm": 0.330078125, "kl": 0.005075214154203422, "learning_rate": 8.611111111111111e-07, "loss": 0.0178, "num_tokens": 14726332.0, "reward": 0.0859375, "reward_std": 0.36247265338897705, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 246, "step_time": 59.55118718650192 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1229.0, "completions/max_terminated_length": 1229.0, "completions/mean_length": 768.390625, "completions/mean_terminated_length": 768.390625, "completions/min_length": 323.0, "completions/min_terminated_length": 323.0, "entropy": 0.8386390097439289, "epoch": 0.14002267573696145, "frac_reward_zero_std": 0.875, "grad_norm": 0.25390625, "kl": 0.004386021581012756, "learning_rate": 8.605442176870747e-07, "loss": -0.002, "num_tokens": 14790253.0, "reward": 0.0546875, "reward_std": 0.26899558305740356, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 247, "step_time": 50.37566922232509 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1222.0, "completions/max_terminated_length": 1222.0, "completions/mean_length": 671.953125, "completions/mean_terminated_length": 671.953125, "completions/min_length": 220.0, "completions/min_terminated_length": 220.0, "entropy": 0.7530262172222137, "epoch": 0.14058956916099774, "frac_reward_zero_std": 0.25, "grad_norm": 0.63671875, "kl": 0.003375259351741988, "learning_rate": 8.599773242630385e-07, "loss": 0.0037, "num_tokens": 14842218.0, "reward": 0.296875, "reward_std": 0.6649904847145081, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 248, "step_time": 47.82787063345313 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1987.0, "completions/max_terminated_length": 1987.0, "completions/mean_length": 831.53125, "completions/mean_terminated_length": 831.53125, "completions/min_length": 401.0, "completions/min_terminated_length": 401.0, "entropy": 0.7898271754384041, "epoch": 0.141156462585034, "frac_reward_zero_std": 0.625, "grad_norm": 0.46875, "kl": 0.003935490451112855, "learning_rate": 8.594104308390022e-07, "loss": -0.0057, "num_tokens": 14907324.0, "reward": 0.15625, "reward_std": 0.49501484632492065, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 249, "step_time": 78.51457949448377 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1259.0, "completions/max_terminated_length": 1259.0, "completions/mean_length": 710.53125, "completions/mean_terminated_length": 710.53125, "completions/min_length": 383.0, "completions/min_terminated_length": 383.0, "entropy": 0.8600624240934849, "epoch": 0.1417233560090703, "frac_reward_zero_std": 0.625, "grad_norm": 0.55859375, "kl": 0.0037620200600940734, "learning_rate": 8.58843537414966e-07, "loss": -0.0108, "num_tokens": 14962966.0, "reward": 0.109375, "reward_std": 0.37234508991241455, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 250, "step_time": 49.08725009299815 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 4096.0, "completions/max_terminated_length": 1145.0, "completions/mean_length": 699.953125, "completions/mean_terminated_length": 646.0476684570312, "completions/min_length": 324.0, "completions/min_terminated_length": 324.0, "entropy": 1.1695469804108143, "epoch": 0.14229024943310659, "frac_reward_zero_std": 0.625, "grad_norm": 0.50390625, "kl": 0.007454679012880661, "learning_rate": 8.582766439909297e-07, "loss": 0.0073, "num_tokens": 15018755.0, "reward": 0.1484375, "reward_std": 0.49344661831855774, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 251, "step_time": 180.536724999547 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1471.0, "completions/max_terminated_length": 1471.0, "completions/mean_length": 689.578125, "completions/mean_terminated_length": 689.578125, "completions/min_length": 14.0, "completions/min_terminated_length": 14.0, "entropy": 0.8454550709575415, "epoch": 0.14285714285714285, "frac_reward_zero_std": 0.5, "grad_norm": 0.59375, "kl": 0.015838683204492554, "learning_rate": 8.577097505668934e-07, "loss": -0.0188, "num_tokens": 15073768.0, "reward": 0.234375, "reward_std": 0.6170124411582947, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 252, "step_time": 56.604044997133315 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1138.0, "completions/max_terminated_length": 1138.0, "completions/mean_length": 725.8125, "completions/mean_terminated_length": 725.8125, "completions/min_length": 428.0, "completions/min_terminated_length": 428.0, "entropy": 0.9270643517374992, "epoch": 0.14342403628117914, "frac_reward_zero_std": 0.375, "grad_norm": 0.64453125, "kl": 0.006181689605000429, "learning_rate": 8.57142857142857e-07, "loss": -0.0016, "num_tokens": 15129764.0, "reward": 0.296875, "reward_std": 0.6649904847145081, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 253, "step_time": 43.90463090222329 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1675.0, "completions/max_terminated_length": 1675.0, "completions/mean_length": 745.328125, "completions/mean_terminated_length": 745.328125, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "entropy": 0.749716816470027, "epoch": 0.14399092970521543, "frac_reward_zero_std": 0.875, "grad_norm": 0.298828125, "kl": 0.007256982797116507, "learning_rate": 8.565759637188208e-07, "loss": -0.015, "num_tokens": 15188457.0, "reward": 0.0390625, "reward_std": 0.13524484634399414, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 254, "step_time": 66.19093087594956 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1119.0, "completions/max_terminated_length": 1119.0, "completions/mean_length": 724.421875, "completions/mean_terminated_length": 724.421875, "completions/min_length": 320.0, "completions/min_terminated_length": 320.0, "entropy": 0.8893743939697742, "epoch": 0.1445578231292517, "frac_reward_zero_std": 0.75, "grad_norm": 0.37890625, "kl": 0.007954366330523044, "learning_rate": 8.560090702947845e-07, "loss": -0.0089, "num_tokens": 15244892.0, "reward": 0.0546875, "reward_std": 0.26899558305740356, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 255, "step_time": 44.45121990144253 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1564.0, "completions/max_terminated_length": 1564.0, "completions/mean_length": 738.25, "completions/mean_terminated_length": 738.25, "completions/min_length": 302.0, "completions/min_terminated_length": 302.0, "entropy": 1.016543436795473, "epoch": 0.14512471655328799, "frac_reward_zero_std": 0.75, "grad_norm": 0.36328125, "kl": 0.0035384253205847926, "learning_rate": 8.554421768707483e-07, "loss": -0.0028, "num_tokens": 15304892.0, "reward": 0.0625, "reward_std": 0.2745848298072815, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 256, "step_time": 63.003812201321125 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1463.0, "completions/max_terminated_length": 1463.0, "completions/mean_length": 674.3125, "completions/mean_terminated_length": 674.3125, "completions/min_length": 226.0, "completions/min_terminated_length": 226.0, "entropy": 0.8456391058862209, "epoch": 0.14569160997732428, "frac_reward_zero_std": 0.5, "grad_norm": 0.609375, "kl": 0.007904115656856447, "learning_rate": 8.54875283446712e-07, "loss": 0.0318, "num_tokens": 15360568.0, "reward": 0.2578125, "reward_std": 0.629891574382782, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 257, "step_time": 57.653264158405364 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1193.0, "completions/max_terminated_length": 1193.0, "completions/mean_length": 683.625, "completions/mean_terminated_length": 683.625, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "entropy": 0.7597847003489733, "epoch": 0.14625850340136054, "frac_reward_zero_std": 0.75, "grad_norm": 0.453125, "kl": 0.006292103018495254, "learning_rate": 8.543083900226758e-07, "loss": 0.0042, "num_tokens": 15413344.0, "reward": 0.234375, "reward_std": 0.6422230005264282, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 258, "step_time": 46.00685829296708 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1294.0, "completions/max_terminated_length": 1294.0, "completions/mean_length": 794.0625, "completions/mean_terminated_length": 794.0625, "completions/min_length": 500.0, "completions/min_terminated_length": 500.0, "entropy": 0.7875265162438154, "epoch": 0.14682539682539683, "frac_reward_zero_std": 0.75, "grad_norm": 0.404296875, "kl": 0.0028908635285915807, "learning_rate": 8.537414965986394e-07, "loss": -0.0122, "num_tokens": 15474628.0, "reward": 0.078125, "reward_std": 0.2847827076911926, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 259, "step_time": 51.2000568471849 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1213.0, "completions/max_terminated_length": 1213.0, "completions/mean_length": 707.296875, "completions/mean_terminated_length": 707.296875, "completions/min_length": 339.0, "completions/min_terminated_length": 339.0, "entropy": 0.8221424091607332, "epoch": 0.1473922902494331, "frac_reward_zero_std": 0.625, "grad_norm": 0.46875, "kl": 0.00474485175072914, "learning_rate": 8.531746031746031e-07, "loss": -0.0084, "num_tokens": 15529135.0, "reward": 0.3046875, "reward_std": 0.6993034482002258, "rewards/accuracy_reward_func/mean": 0.28125, "rewards/accuracy_reward_func/std": 0.7007648944854736, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 260, "step_time": 47.618589179590344 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1819.0, "completions/max_terminated_length": 1819.0, "completions/mean_length": 734.65625, "completions/mean_terminated_length": 734.65625, "completions/min_length": 251.0, "completions/min_terminated_length": 251.0, "entropy": 0.7079410217702389, "epoch": 0.14795918367346939, "frac_reward_zero_std": 0.25, "grad_norm": 0.65625, "kl": 0.0045277264434844255, "learning_rate": 8.526077097505668e-07, "loss": -0.0294, "num_tokens": 15587033.0, "reward": 0.3671875, "reward_std": 0.7676090598106384, "rewards/accuracy_reward_func/mean": 0.3125, "rewards/accuracy_reward_func/std": 0.7319250702857971, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 261, "step_time": 71.04479949083179 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1280.0, "completions/max_terminated_length": 1280.0, "completions/mean_length": 783.171875, "completions/mean_terminated_length": 783.171875, "completions/min_length": 278.0, "completions/min_terminated_length": 278.0, "entropy": 0.9606173038482666, "epoch": 0.14852607709750568, "frac_reward_zero_std": 0.875, "grad_norm": 0.244140625, "kl": 0.009063210411113687, "learning_rate": 8.520408163265306e-07, "loss": 0.0147, "num_tokens": 15648540.0, "reward": 0.03125, "reward_std": 0.25, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 262, "step_time": 50.15093679446727 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1643.0, "completions/max_terminated_length": 1643.0, "completions/mean_length": 833.578125, "completions/mean_terminated_length": 833.578125, "completions/min_length": 311.0, "completions/min_terminated_length": 311.0, "entropy": 0.7215248327702284, "epoch": 0.14909297052154194, "frac_reward_zero_std": 0.75, "grad_norm": 0.302734375, "kl": 0.006525891963974573, "learning_rate": 8.514739229024943e-07, "loss": -0.0078, "num_tokens": 15712585.0, "reward": 0.0546875, "reward_std": 0.26899558305740356, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 263, "step_time": 63.85956168174744 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1282.0, "completions/max_terminated_length": 1282.0, "completions/mean_length": 717.484375, "completions/mean_terminated_length": 717.484375, "completions/min_length": 466.0, "completions/min_terminated_length": 466.0, "entropy": 1.0237734578549862, "epoch": 0.14965986394557823, "frac_reward_zero_std": 0.75, "grad_norm": 0.357421875, "kl": 0.007683624091441743, "learning_rate": 8.509070294784581e-07, "loss": 0.0034, "num_tokens": 15768728.0, "reward": 0.28125, "reward_std": 0.6291528940200806, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 264, "step_time": 49.85610455926508 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 654.046875, "completions/mean_terminated_length": 654.046875, "completions/min_length": 262.0, "completions/min_terminated_length": 262.0, "entropy": 1.0060841478407383, "epoch": 0.15022675736961452, "frac_reward_zero_std": 0.625, "grad_norm": 0.55078125, "kl": 0.011602386686718091, "learning_rate": 8.503401360544217e-07, "loss": -0.0132, "num_tokens": 15821251.0, "reward": 0.3671875, "reward_std": 0.8273220658302307, "rewards/accuracy_reward_func/mean": 0.3125, "rewards/accuracy_reward_func/std": 0.7319250702857971, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 265, "step_time": 45.338741284795105 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 973.0, "completions/max_terminated_length": 973.0, "completions/mean_length": 668.828125, "completions/mean_terminated_length": 668.828125, "completions/min_length": 294.0, "completions/min_terminated_length": 294.0, "entropy": 1.005034700036049, "epoch": 0.15079365079365079, "frac_reward_zero_std": 0.5, "grad_norm": 0.57421875, "kl": 0.008998857665574178, "learning_rate": 8.497732426303855e-07, "loss": -0.0007, "num_tokens": 15873336.0, "reward": 0.125, "reward_std": 0.4364357888698578, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 266, "step_time": 38.137159804813564 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1368.0, "completions/max_terminated_length": 1368.0, "completions/mean_length": 717.421875, "completions/mean_terminated_length": 717.421875, "completions/min_length": 253.0, "completions/min_terminated_length": 253.0, "entropy": 0.7188418917357922, "epoch": 0.15136054421768708, "frac_reward_zero_std": 0.625, "grad_norm": 0.478515625, "kl": 0.007004673054325394, "learning_rate": 8.492063492063492e-07, "loss": -0.0062, "num_tokens": 15930243.0, "reward": 0.1171875, "reward_std": 0.37524792551994324, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 267, "step_time": 53.01031909137964 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1437.0, "completions/max_terminated_length": 1437.0, "completions/mean_length": 695.40625, "completions/mean_terminated_length": 695.40625, "completions/min_length": 236.0, "completions/min_terminated_length": 236.0, "entropy": 0.7823151145130396, "epoch": 0.15192743764172337, "frac_reward_zero_std": 0.625, "grad_norm": 0.5078125, "kl": 0.003142316978482995, "learning_rate": 8.486394557823128e-07, "loss": 0.0128, "num_tokens": 15988117.0, "reward": 0.09375, "reward_std": 0.426084041595459, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 268, "step_time": 56.694628211669624 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1636.0, "completions/max_terminated_length": 1636.0, "completions/mean_length": 605.09375, "completions/mean_terminated_length": 605.09375, "completions/min_length": 227.0, "completions/min_terminated_length": 227.0, "entropy": 0.8527481835335493, "epoch": 0.15249433106575963, "frac_reward_zero_std": 0.375, "grad_norm": 0.72265625, "kl": 0.004755638008646201, "learning_rate": 8.480725623582766e-07, "loss": 0.0051, "num_tokens": 16038539.0, "reward": 0.2734375, "reward_std": 0.6294977068901062, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 269, "step_time": 63.849821055307984 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1465.0, "completions/max_terminated_length": 1465.0, "completions/mean_length": 727.6875, "completions/mean_terminated_length": 727.6875, "completions/min_length": 253.0, "completions/min_terminated_length": 253.0, "entropy": 0.8521798010915518, "epoch": 0.15306122448979592, "frac_reward_zero_std": 0.625, "grad_norm": 0.4609375, "kl": 0.010075078469526488, "learning_rate": 8.475056689342403e-07, "loss": -0.0073, "num_tokens": 16096135.0, "reward": 0.203125, "reward_std": 0.5471335649490356, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 270, "step_time": 57.340402787551284 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1408.0, "completions/max_terminated_length": 1408.0, "completions/mean_length": 809.09375, "completions/mean_terminated_length": 809.09375, "completions/min_length": 264.0, "completions/min_terminated_length": 264.0, "entropy": 0.9269996285438538, "epoch": 0.1536281179138322, "frac_reward_zero_std": 0.5, "grad_norm": 0.51171875, "kl": 0.010795337875606492, "learning_rate": 8.469387755102041e-07, "loss": 0.0172, "num_tokens": 16158221.0, "reward": 0.2734375, "reward_std": 0.6662480235099792, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 271, "step_time": 55.47354524489492 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1257.0, "completions/max_terminated_length": 1257.0, "completions/mean_length": 714.09375, "completions/mean_terminated_length": 714.09375, "completions/min_length": 374.0, "completions/min_terminated_length": 374.0, "entropy": 0.9163622744381428, "epoch": 0.15419501133786848, "frac_reward_zero_std": 0.5, "grad_norm": 0.578125, "kl": 0.004687610104156192, "learning_rate": 8.463718820861677e-07, "loss": 0.0279, "num_tokens": 16214643.0, "reward": 0.2734375, "reward_std": 0.7123050093650818, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 272, "step_time": 49.16124189272523 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1902.0, "completions/max_terminated_length": 1902.0, "completions/mean_length": 791.703125, "completions/mean_terminated_length": 791.703125, "completions/min_length": 227.0, "completions/min_terminated_length": 227.0, "entropy": 0.7695765160024166, "epoch": 0.15476190476190477, "frac_reward_zero_std": 0.5, "grad_norm": 0.515625, "kl": 0.004302407833165489, "learning_rate": 8.458049886621315e-07, "loss": -0.0395, "num_tokens": 16275944.0, "reward": 0.1328125, "reward_std": 0.4386323094367981, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 273, "step_time": 74.4036480858922 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1376.0, "completions/max_terminated_length": 1376.0, "completions/mean_length": 726.25, "completions/mean_terminated_length": 726.25, "completions/min_length": 246.0, "completions/min_terminated_length": 246.0, "entropy": 0.7283195722848177, "epoch": 0.15532879818594103, "frac_reward_zero_std": 0.375, "grad_norm": 0.65234375, "kl": 0.008357068902114406, "learning_rate": 8.452380952380952e-07, "loss": 0.0142, "num_tokens": 16332280.0, "reward": 0.1875, "reward_std": 0.5455447435379028, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 274, "step_time": 53.547890301793814 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1329.0, "completions/max_terminated_length": 1329.0, "completions/mean_length": 689.5625, "completions/mean_terminated_length": 689.5625, "completions/min_length": 9.0, "completions/min_terminated_length": 9.0, "entropy": 0.7559501770883799, "epoch": 0.15589569160997732, "frac_reward_zero_std": 0.375, "grad_norm": 0.59375, "kl": 0.017642906168475747, "learning_rate": 8.44671201814059e-07, "loss": 0.0387, "num_tokens": 16385708.0, "reward": 0.328125, "reward_std": 0.7194505929946899, "rewards/accuracy_reward_func/mean": 0.28125, "rewards/accuracy_reward_func/std": 0.7007648944854736, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 275, "step_time": 51.140076864510775 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1377.0, "completions/max_terminated_length": 1377.0, "completions/mean_length": 801.453125, "completions/mean_terminated_length": 801.453125, "completions/min_length": 286.0, "completions/min_terminated_length": 286.0, "entropy": 0.76519157551229, "epoch": 0.1564625850340136, "frac_reward_zero_std": 0.75, "grad_norm": 0.349609375, "kl": 0.005709742763428949, "learning_rate": 8.441043083900226e-07, "loss": -0.0031, "num_tokens": 16448001.0, "reward": 0.1015625, "reward_std": 0.40999117493629456, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 276, "step_time": 53.25250900071114 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1350.0, "completions/max_terminated_length": 1350.0, "completions/mean_length": 624.234375, "completions/mean_terminated_length": 624.234375, "completions/min_length": 194.0, "completions/min_terminated_length": 194.0, "entropy": 0.7580607645213604, "epoch": 0.15702947845804988, "frac_reward_zero_std": 0.25, "grad_norm": 0.73828125, "kl": 0.01761455085943453, "learning_rate": 8.435374149659864e-07, "loss": -0.0033, "num_tokens": 16499128.0, "reward": 0.3359375, "reward_std": 0.7295067310333252, "rewards/accuracy_reward_func/mean": 0.3125, "rewards/accuracy_reward_func/std": 0.7319250702857971, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 277, "step_time": 52.124403435736895 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1474.0, "completions/max_terminated_length": 1474.0, "completions/mean_length": 888.296875, "completions/mean_terminated_length": 888.296875, "completions/min_length": 364.0, "completions/min_terminated_length": 364.0, "entropy": 0.7863531224429607, "epoch": 0.15759637188208617, "frac_reward_zero_std": 0.75, "grad_norm": 0.341796875, "kl": 0.006082025865907781, "learning_rate": 8.4297052154195e-07, "loss": -0.0108, "num_tokens": 16567819.0, "reward": 0.078125, "reward_std": 0.2847827076911926, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 278, "step_time": 58.55780995916575 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1131.0, "completions/max_terminated_length": 1131.0, "completions/mean_length": 710.3125, "completions/mean_terminated_length": 710.3125, "completions/min_length": 414.0, "completions/min_terminated_length": 414.0, "entropy": 0.8654900714755058, "epoch": 0.15816326530612246, "frac_reward_zero_std": 0.625, "grad_norm": 0.462890625, "kl": 0.004806106633623131, "learning_rate": 8.424036281179138e-07, "loss": 0.0177, "num_tokens": 16622871.0, "reward": 0.1171875, "reward_std": 0.43408530950546265, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 279, "step_time": 44.18114794231951 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1269.0, "completions/max_terminated_length": 1269.0, "completions/mean_length": 694.734375, "completions/mean_terminated_length": 694.734375, "completions/min_length": 280.0, "completions/min_terminated_length": 280.0, "entropy": 1.0109095703810453, "epoch": 0.15873015873015872, "frac_reward_zero_std": 0.5, "grad_norm": 0.62890625, "kl": 0.007879828044679016, "learning_rate": 8.418367346938775e-07, "loss": 0.0182, "num_tokens": 16676726.0, "reward": 0.2734375, "reward_std": 0.6294977068901062, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 280, "step_time": 49.75455924682319 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1565.0, "completions/max_terminated_length": 1565.0, "completions/mean_length": 772.671875, "completions/mean_terminated_length": 772.671875, "completions/min_length": 238.0, "completions/min_terminated_length": 238.0, "entropy": 0.7377403303980827, "epoch": 0.159297052154195, "frac_reward_zero_std": 0.625, "grad_norm": 0.435546875, "kl": 0.006127516506239772, "learning_rate": 8.412698412698413e-07, "loss": -0.008, "num_tokens": 16736313.0, "reward": 0.1796875, "reward_std": 0.5445777773857117, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 281, "step_time": 61.17428585886955 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1397.0, "completions/max_terminated_length": 1397.0, "completions/mean_length": 787.421875, "completions/mean_terminated_length": 787.421875, "completions/min_length": 298.0, "completions/min_terminated_length": 298.0, "entropy": 0.6864061839878559, "epoch": 0.1598639455782313, "frac_reward_zero_std": 0.375, "grad_norm": 0.5390625, "kl": 0.0049530532414792106, "learning_rate": 8.40702947845805e-07, "loss": 0.0018, "num_tokens": 16796828.0, "reward": 0.203125, "reward_std": 0.5471335649490356, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 282, "step_time": 54.73281847964972 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1169.0, "completions/max_terminated_length": 1169.0, "completions/mean_length": 652.75, "completions/mean_terminated_length": 652.75, "completions/min_length": 254.0, "completions/min_terminated_length": 254.0, "entropy": 0.7306120228022337, "epoch": 0.16043083900226757, "frac_reward_zero_std": 0.5, "grad_norm": 0.4921875, "kl": 0.013816178601700813, "learning_rate": 8.401360544217688e-07, "loss": -0.0133, "num_tokens": 16850964.0, "reward": 0.2734375, "reward_std": 0.6542273163795471, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 283, "step_time": 45.850550681352615 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1697.0, "completions/max_terminated_length": 1697.0, "completions/mean_length": 670.9375, "completions/mean_terminated_length": 670.9375, "completions/min_length": 229.0, "completions/min_terminated_length": 229.0, "entropy": 0.9432363584637642, "epoch": 0.16099773242630386, "frac_reward_zero_std": 0.625, "grad_norm": 0.474609375, "kl": 0.0031036645887070335, "learning_rate": 8.395691609977323e-07, "loss": -0.0136, "num_tokens": 16903624.0, "reward": 0.3125, "reward_std": 0.7099072337150574, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 284, "step_time": 65.68877389188856 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1536.0, "completions/max_terminated_length": 1536.0, "completions/mean_length": 724.8125, "completions/mean_terminated_length": 724.8125, "completions/min_length": 192.0, "completions/min_terminated_length": 192.0, "entropy": 0.7398256752640009, "epoch": 0.16156462585034015, "frac_reward_zero_std": 0.375, "grad_norm": 0.57421875, "kl": 0.0050950217846548185, "learning_rate": 8.39002267573696e-07, "loss": -0.01, "num_tokens": 16960140.0, "reward": 0.296875, "reward_std": 0.6649904847145081, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 285, "step_time": 60.66464094724506 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1093.0, "completions/max_terminated_length": 1093.0, "completions/mean_length": 664.828125, "completions/mean_terminated_length": 664.828125, "completions/min_length": 256.0, "completions/min_terminated_length": 256.0, "entropy": 1.1280731745064259, "epoch": 0.1621315192743764, "frac_reward_zero_std": 0.625, "grad_norm": 0.5, "kl": 0.00682494361535646, "learning_rate": 8.384353741496598e-07, "loss": -0.0094, "num_tokens": 17013465.0, "reward": 0.3203125, "reward_std": 0.6978833675384521, "rewards/accuracy_reward_func/mean": 0.28125, "rewards/accuracy_reward_func/std": 0.7007648944854736, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 286, "step_time": 40.771853673271835 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1941.0, "completions/max_terminated_length": 1941.0, "completions/mean_length": 774.5625, "completions/mean_terminated_length": 774.5625, "completions/min_length": 277.0, "completions/min_terminated_length": 277.0, "entropy": 0.9306526109576225, "epoch": 0.1626984126984127, "frac_reward_zero_std": 0.5, "grad_norm": 0.5625, "kl": 0.014398715313291177, "learning_rate": 8.378684807256235e-07, "loss": 0.0087, "num_tokens": 17073797.0, "reward": 0.2265625, "reward_std": 0.6167611479759216, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 287, "step_time": 71.92694858182222 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1740.0, "completions/max_terminated_length": 1740.0, "completions/mean_length": 921.953125, "completions/mean_terminated_length": 921.953125, "completions/min_length": 394.0, "completions/min_terminated_length": 394.0, "entropy": 0.6797251738607883, "epoch": 0.16326530612244897, "frac_reward_zero_std": 0.75, "grad_norm": 0.37109375, "kl": 0.004947350826114416, "learning_rate": 8.373015873015873e-07, "loss": -0.0098, "num_tokens": 17144482.0, "reward": 0.0859375, "reward_std": 0.28942590951919556, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 288, "step_time": 64.15770970564336 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1693.0, "completions/max_terminated_length": 1693.0, "completions/mean_length": 753.03125, "completions/mean_terminated_length": 753.03125, "completions/min_length": 410.0, "completions/min_terminated_length": 410.0, "entropy": 0.9771784581243992, "epoch": 0.16383219954648526, "frac_reward_zero_std": 0.625, "grad_norm": 0.474609375, "kl": 0.007738123604212888, "learning_rate": 8.36734693877551e-07, "loss": -0.0303, "num_tokens": 17203300.0, "reward": 0.1171875, "reward_std": 0.37524792551994324, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 289, "step_time": 61.81910063419491 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1395.0, "completions/max_terminated_length": 1395.0, "completions/mean_length": 695.84375, "completions/mean_terminated_length": 695.84375, "completions/min_length": 390.0, "completions/min_terminated_length": 390.0, "entropy": 0.8843433819711208, "epoch": 0.16439909297052155, "frac_reward_zero_std": 0.625, "grad_norm": 0.51171875, "kl": 0.007000370969763026, "learning_rate": 8.361678004535147e-07, "loss": 0.0139, "num_tokens": 17258106.0, "reward": 0.1328125, "reward_std": 0.4386323094367981, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 290, "step_time": 50.315598805435 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1386.0, "completions/max_terminated_length": 1386.0, "completions/mean_length": 687.65625, "completions/mean_terminated_length": 687.65625, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "entropy": 0.8731617815792561, "epoch": 0.1649659863945578, "frac_reward_zero_std": 0.875, "grad_norm": 0.263671875, "kl": 0.009098797789192759, "learning_rate": 8.356009070294783e-07, "loss": 0.0123, "num_tokens": 17313788.0, "reward": 0.0234375, "reward_std": 0.10652101039886475, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 291, "step_time": 51.22578653600067 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1331.0, "completions/max_terminated_length": 1331.0, "completions/mean_length": 705.53125, "completions/mean_terminated_length": 705.53125, "completions/min_length": 228.0, "completions/min_terminated_length": 228.0, "entropy": 1.038501761853695, "epoch": 0.1655328798185941, "frac_reward_zero_std": 0.875, "grad_norm": 0.26171875, "kl": 0.01143992722791154, "learning_rate": 8.350340136054421e-07, "loss": -0.0001, "num_tokens": 17373430.0, "reward": 0.09375, "reward_std": 0.3435921370983124, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 292, "step_time": 50.04920950811356 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1142.0, "completions/max_terminated_length": 1142.0, "completions/mean_length": 639.75, "completions/mean_terminated_length": 639.75, "completions/min_length": 199.0, "completions/min_terminated_length": 199.0, "entropy": 0.9801824130117893, "epoch": 0.1660997732426304, "frac_reward_zero_std": 0.625, "grad_norm": 0.51953125, "kl": 0.006710244415444322, "learning_rate": 8.344671201814058e-07, "loss": -0.0121, "num_tokens": 17425430.0, "reward": 0.09375, "reward_std": 0.36596253514289856, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 293, "step_time": 42.188250828534365 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1590.0, "completions/max_terminated_length": 1590.0, "completions/mean_length": 809.890625, "completions/mean_terminated_length": 809.890625, "completions/min_length": 312.0, "completions/min_terminated_length": 312.0, "entropy": 0.9223752617835999, "epoch": 0.16666666666666666, "frac_reward_zero_std": 0.625, "grad_norm": 0.396484375, "kl": 0.006054805249732453, "learning_rate": 8.339002267573696e-07, "loss": 0.0244, "num_tokens": 17490111.0, "reward": 0.28125, "reward_std": 0.7339553236961365, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 294, "step_time": 58.92305904533714 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2058.0, "completions/max_terminated_length": 2058.0, "completions/mean_length": 762.875, "completions/mean_terminated_length": 762.875, "completions/min_length": 221.0, "completions/min_terminated_length": 221.0, "entropy": 0.7460720557719469, "epoch": 0.16723356009070295, "frac_reward_zero_std": 0.625, "grad_norm": 0.416015625, "kl": 0.004036792801343836, "learning_rate": 8.333333333333333e-07, "loss": -0.0032, "num_tokens": 17549871.0, "reward": 0.3515625, "reward_std": 0.7274640202522278, "rewards/accuracy_reward_func/mean": 0.3125, "rewards/accuracy_reward_func/std": 0.7319250702857971, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 295, "step_time": 75.95836392231286 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1319.0, "completions/max_terminated_length": 1319.0, "completions/mean_length": 680.90625, "completions/mean_terminated_length": 680.90625, "completions/min_length": 289.0, "completions/min_terminated_length": 289.0, "entropy": 0.7744119726121426, "epoch": 0.16780045351473924, "frac_reward_zero_std": 0.375, "grad_norm": 0.64453125, "kl": 0.004806325436220504, "learning_rate": 8.327664399092971e-07, "loss": 0.0217, "num_tokens": 17603681.0, "reward": 0.1640625, "reward_std": 0.5422958731651306, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 296, "step_time": 48.54740091320127 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1365.0, "completions/max_terminated_length": 1365.0, "completions/mean_length": 772.546875, "completions/mean_terminated_length": 772.546875, "completions/min_length": 322.0, "completions/min_terminated_length": 322.0, "entropy": 0.7225805297493935, "epoch": 0.1683673469387755, "frac_reward_zero_std": 0.375, "grad_norm": 0.5703125, "kl": 0.004341167637903709, "learning_rate": 8.321995464852607e-07, "loss": 0.0546, "num_tokens": 17663668.0, "reward": 0.2109375, "reward_std": 0.5477564930915833, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 297, "step_time": 49.85294338222593 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 4096.0, "completions/max_terminated_length": 1479.0, "completions/mean_length": 856.109375, "completions/mean_terminated_length": 804.6825561523438, "completions/min_length": 371.0, "completions/min_terminated_length": 371.0, "entropy": 0.819762421771884, "epoch": 0.1689342403628118, "frac_reward_zero_std": 0.625, "grad_norm": 0.427734375, "kl": 0.006123719067545608, "learning_rate": 8.316326530612245e-07, "loss": 0.0162, "num_tokens": 17728627.0, "reward": 0.1484375, "reward_std": 0.49344661831855774, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 298, "step_time": 173.51606942806393 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1389.0, "completions/max_terminated_length": 1389.0, "completions/mean_length": 711.5, "completions/mean_terminated_length": 711.5, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "entropy": 1.0076992735266685, "epoch": 0.16950113378684808, "frac_reward_zero_std": 0.75, "grad_norm": 0.42578125, "kl": 0.015792973979841918, "learning_rate": 8.310657596371881e-07, "loss": -0.0079, "num_tokens": 17785603.0, "reward": 0.1484375, "reward_std": 0.44257262349128723, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 299, "step_time": 50.72464693058282 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1502.0, "completions/max_terminated_length": 1502.0, "completions/mean_length": 806.609375, "completions/mean_terminated_length": 806.609375, "completions/min_length": 321.0, "completions/min_terminated_length": 321.0, "entropy": 0.6680117640644312, "epoch": 0.17006802721088435, "frac_reward_zero_std": 0.5, "grad_norm": 0.474609375, "kl": 0.00916920488816686, "learning_rate": 8.304988662131519e-07, "loss": 0.0101, "num_tokens": 17849186.0, "reward": 0.171875, "reward_std": 0.5434950590133667, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 300, "step_time": 54.90682506840676 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1255.0, "completions/max_terminated_length": 1255.0, "completions/mean_length": 699.234375, "completions/mean_terminated_length": 699.234375, "completions/min_length": 30.0, "completions/min_terminated_length": 30.0, "entropy": 0.8960948511958122, "epoch": 0.17063492063492064, "frac_reward_zero_std": 0.625, "grad_norm": 0.62109375, "kl": 0.009604600025340915, "learning_rate": 8.299319727891156e-07, "loss": -0.0302, "num_tokens": 17905065.0, "reward": 0.15625, "reward_std": 0.4443204402923584, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 301, "step_time": 45.49867108184844 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1305.0, "completions/max_terminated_length": 1305.0, "completions/mean_length": 689.890625, "completions/mean_terminated_length": 689.890625, "completions/min_length": 26.0, "completions/min_terminated_length": 26.0, "entropy": 0.8179099150002003, "epoch": 0.1712018140589569, "frac_reward_zero_std": 0.75, "grad_norm": 0.453125, "kl": 0.01156878836627584, "learning_rate": 8.293650793650794e-07, "loss": 0.0077, "num_tokens": 17960138.0, "reward": 0.109375, "reward_std": 0.2083333283662796, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.109375, "rewards/format_reward_func/std": 0.2083333432674408, "step": 302, "step_time": 46.90971658192575 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1887.0, "completions/max_terminated_length": 1887.0, "completions/mean_length": 793.046875, "completions/mean_terminated_length": 793.046875, "completions/min_length": 11.0, "completions/min_terminated_length": 11.0, "entropy": 0.8214219212532043, "epoch": 0.1717687074829932, "frac_reward_zero_std": 0.5, "grad_norm": 0.50390625, "kl": 0.011117888643639162, "learning_rate": 8.28798185941043e-07, "loss": 0.0095, "num_tokens": 18021213.0, "reward": 0.2578125, "reward_std": 0.629891574382782, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 303, "step_time": 67.9431212414056 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1247.0, "completions/max_terminated_length": 1247.0, "completions/mean_length": 699.15625, "completions/mean_terminated_length": 699.15625, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "entropy": 0.8105713315308094, "epoch": 0.17233560090702948, "frac_reward_zero_std": 0.625, "grad_norm": 0.47265625, "kl": 0.003976978739956394, "learning_rate": 8.282312925170068e-07, "loss": 0.0067, "num_tokens": 18074143.0, "reward": 0.1484375, "reward_std": 0.49344661831855774, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 304, "step_time": 44.68646402191371 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1519.0, "completions/max_terminated_length": 1519.0, "completions/mean_length": 770.28125, "completions/mean_terminated_length": 770.28125, "completions/min_length": 243.0, "completions/min_terminated_length": 243.0, "entropy": 0.8100260365754366, "epoch": 0.17290249433106575, "frac_reward_zero_std": 0.875, "grad_norm": 0.279296875, "kl": 0.005946670476987492, "learning_rate": 8.276643990929705e-07, "loss": 0.0033, "num_tokens": 18134369.0, "reward": 0.046875, "reward_std": 0.14689241349697113, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 305, "step_time": 55.88502377551049 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1386.0, "completions/max_terminated_length": 1386.0, "completions/mean_length": 792.5625, "completions/mean_terminated_length": 792.5625, "completions/min_length": 327.0, "completions/min_terminated_length": 327.0, "entropy": 0.6959674190729856, "epoch": 0.17346938775510204, "frac_reward_zero_std": 0.5, "grad_norm": 0.44921875, "kl": 0.010840905888471752, "learning_rate": 8.270975056689343e-07, "loss": 0.0076, "num_tokens": 18197653.0, "reward": 0.1640625, "reward_std": 0.496453195810318, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 306, "step_time": 50.81538745574653 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1070.0, "completions/max_terminated_length": 1070.0, "completions/mean_length": 738.96875, "completions/mean_terminated_length": 738.96875, "completions/min_length": 238.0, "completions/min_terminated_length": 238.0, "entropy": 0.9731320925056934, "epoch": 0.17403628117913833, "frac_reward_zero_std": 0.5, "grad_norm": 0.546875, "kl": 0.005864134756848216, "learning_rate": 8.26530612244898e-07, "loss": -0.0048, "num_tokens": 18255139.0, "reward": 0.234375, "reward_std": 0.5907269716262817, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 307, "step_time": 38.74946460779756 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1482.0, "completions/max_terminated_length": 1482.0, "completions/mean_length": 744.671875, "completions/mean_terminated_length": 744.671875, "completions/min_length": 293.0, "completions/min_terminated_length": 293.0, "entropy": 0.7935100309550762, "epoch": 0.1746031746031746, "frac_reward_zero_std": 0.5, "grad_norm": 0.46484375, "kl": 0.0056961430091178045, "learning_rate": 8.259637188208616e-07, "loss": 0.0037, "num_tokens": 18319262.0, "reward": 0.34375, "reward_std": 0.7811011672019958, "rewards/accuracy_reward_func/mean": 0.28125, "rewards/accuracy_reward_func/std": 0.7007648944854736, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 308, "step_time": 55.75887096114457 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1193.0, "completions/max_terminated_length": 1193.0, "completions/mean_length": 667.40625, "completions/mean_terminated_length": 667.40625, "completions/min_length": 237.0, "completions/min_terminated_length": 237.0, "entropy": 0.9020283985882998, "epoch": 0.17517006802721088, "frac_reward_zero_std": 0.625, "grad_norm": 0.515625, "kl": 0.03261678959825076, "learning_rate": 8.253968253968253e-07, "loss": 0.0068, "num_tokens": 18373248.0, "reward": 0.109375, "reward_std": 0.4315783977508545, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 309, "step_time": 42.62671375833452 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1227.0, "completions/max_terminated_length": 1227.0, "completions/mean_length": 694.5, "completions/mean_terminated_length": 694.5, "completions/min_length": 142.0, "completions/min_terminated_length": 142.0, "entropy": 0.8163422644138336, "epoch": 0.17573696145124718, "frac_reward_zero_std": 0.25, "grad_norm": 0.69921875, "kl": 0.007384991717117373, "learning_rate": 8.24829931972789e-07, "loss": 0.0036, "num_tokens": 18428016.0, "reward": 0.375, "reward_std": 0.7559289336204529, "rewards/accuracy_reward_func/mean": 0.34375, "rewards/accuracy_reward_func/std": 0.7605084180831909, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 310, "step_time": 44.21994534600526 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 719.0, "completions/mean_terminated_length": 719.0, "completions/min_length": 327.0, "completions/min_terminated_length": 327.0, "entropy": 0.9615583010017872, "epoch": 0.17630385487528344, "frac_reward_zero_std": 0.5, "grad_norm": 0.58984375, "kl": 0.011452683342213277, "learning_rate": 8.242630385487528e-07, "loss": 0.0143, "num_tokens": 18484128.0, "reward": 0.28125, "reward_std": 0.665922224521637, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 311, "step_time": 53.601689458824694 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 747.96875, "completions/mean_terminated_length": 747.96875, "completions/min_length": 205.0, "completions/min_terminated_length": 205.0, "entropy": 0.8322487156838179, "epoch": 0.17687074829931973, "frac_reward_zero_std": 0.375, "grad_norm": 0.625, "kl": 0.008456386131001636, "learning_rate": 8.236961451247165e-07, "loss": -0.0169, "num_tokens": 18542350.0, "reward": 0.453125, "reward_std": 0.8296047449111938, "rewards/accuracy_reward_func/mean": 0.4375, "rewards/accuracy_reward_func/std": 0.8333333730697632, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 312, "step_time": 45.08718290459365 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1363.0, "completions/max_terminated_length": 1363.0, "completions/mean_length": 692.75, "completions/mean_terminated_length": 692.75, "completions/min_length": 270.0, "completions/min_terminated_length": 270.0, "entropy": 0.8413090296089649, "epoch": 0.17743764172335602, "frac_reward_zero_std": 0.5, "grad_norm": 0.58203125, "kl": 0.007160043605836108, "learning_rate": 8.231292517006803e-07, "loss": -0.006, "num_tokens": 18596598.0, "reward": 0.1953125, "reward_std": 0.5463964939117432, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 313, "step_time": 48.35842065606266 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1331.0, "completions/max_terminated_length": 1331.0, "completions/mean_length": 770.21875, "completions/mean_terminated_length": 770.21875, "completions/min_length": 398.0, "completions/min_terminated_length": 398.0, "entropy": 0.8696554526686668, "epoch": 0.17800453514739228, "frac_reward_zero_std": 0.625, "grad_norm": 0.423828125, "kl": 0.00865155813517049, "learning_rate": 8.22562358276644e-07, "loss": 0.025, "num_tokens": 18655588.0, "reward": 0.2265625, "reward_std": 0.6419816017150879, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 314, "step_time": 46.93548165168613 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 971.0, "completions/max_terminated_length": 971.0, "completions/mean_length": 614.375, "completions/mean_terminated_length": 614.375, "completions/min_length": 294.0, "completions/min_terminated_length": 294.0, "entropy": 0.9113731905817986, "epoch": 0.17857142857142858, "frac_reward_zero_std": 0.5, "grad_norm": 0.6328125, "kl": 0.0057736381713766605, "learning_rate": 8.219954648526076e-07, "loss": -0.0173, "num_tokens": 18705884.0, "reward": 0.296875, "reward_std": 0.6649904847145081, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 315, "step_time": 35.24758036993444 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1689.0, "completions/max_terminated_length": 1689.0, "completions/mean_length": 785.59375, "completions/mean_terminated_length": 785.59375, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "entropy": 0.7394520230591297, "epoch": 0.17913832199546487, "frac_reward_zero_std": 0.75, "grad_norm": 0.32421875, "kl": 0.004618886479875073, "learning_rate": 8.214285714285713e-07, "loss": 0.004, "num_tokens": 18767962.0, "reward": 0.109375, "reward_std": 0.41277948021888733, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 316, "step_time": 60.534888616763055 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1130.0, "completions/max_terminated_length": 1130.0, "completions/mean_length": 690.171875, "completions/mean_terminated_length": 690.171875, "completions/min_length": 297.0, "completions/min_terminated_length": 297.0, "entropy": 0.9527055807411671, "epoch": 0.17970521541950113, "frac_reward_zero_std": 0.5, "grad_norm": 0.59765625, "kl": 0.007261134553118609, "learning_rate": 8.208616780045351e-07, "loss": -0.0121, "num_tokens": 18822853.0, "reward": 0.0859375, "reward_std": 0.36247265338897705, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 317, "step_time": 39.881969602778554 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1123.0, "completions/max_terminated_length": 1123.0, "completions/mean_length": 664.859375, "completions/mean_terminated_length": 664.859375, "completions/min_length": 343.0, "completions/min_terminated_length": 343.0, "entropy": 0.765125572681427, "epoch": 0.18027210884353742, "frac_reward_zero_std": 0.5, "grad_norm": 0.6796875, "kl": 0.005805548498756252, "learning_rate": 8.202947845804988e-07, "loss": -0.0292, "num_tokens": 18876444.0, "reward": 0.1484375, "reward_std": 0.44257262349128723, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 318, "step_time": 40.361245087347925 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1142.0, "completions/max_terminated_length": 1142.0, "completions/mean_length": 643.703125, "completions/mean_terminated_length": 643.703125, "completions/min_length": 279.0, "completions/min_terminated_length": 279.0, "entropy": 1.0180951990187168, "epoch": 0.18083900226757368, "frac_reward_zero_std": 0.625, "grad_norm": 0.55078125, "kl": 0.012444952561054379, "learning_rate": 8.197278911564626e-07, "loss": 0.0094, "num_tokens": 18927193.0, "reward": 0.1015625, "reward_std": 0.3692515790462494, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 319, "step_time": 39.848433456383646 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1394.0, "completions/max_terminated_length": 1394.0, "completions/mean_length": 742.8125, "completions/mean_terminated_length": 742.8125, "completions/min_length": 411.0, "completions/min_terminated_length": 411.0, "entropy": 0.879846004769206, "epoch": 0.18140589569160998, "frac_reward_zero_std": 0.75, "grad_norm": 0.37109375, "kl": 0.006181372940773144, "learning_rate": 8.191609977324263e-07, "loss": -0.0046, "num_tokens": 18984853.0, "reward": 0.171875, "reward_std": 0.4647814631462097, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.109375, "rewards/format_reward_func/std": 0.2083333432674408, "step": 320, "step_time": 48.91933018993586 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1921.0, "completions/max_terminated_length": 1921.0, "completions/mean_length": 762.21875, "completions/mean_terminated_length": 762.21875, "completions/min_length": 347.0, "completions/min_terminated_length": 347.0, "entropy": 0.671691881492734, "epoch": 0.18197278911564627, "frac_reward_zero_std": 0.25, "grad_norm": 0.58203125, "kl": 0.007276202712091617, "learning_rate": 8.1859410430839e-07, "loss": 0.0197, "num_tokens": 19044075.0, "reward": 0.3203125, "reward_std": 0.6978833675384521, "rewards/accuracy_reward_func/mean": 0.28125, "rewards/accuracy_reward_func/std": 0.7007648944854736, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 321, "step_time": 67.99173118919134 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1506.0, "completions/max_terminated_length": 1506.0, "completions/mean_length": 685.0625, "completions/mean_terminated_length": 685.0625, "completions/min_length": 211.0, "completions/min_terminated_length": 211.0, "entropy": 0.7449827585369349, "epoch": 0.18253968253968253, "frac_reward_zero_std": 0.5, "grad_norm": 0.54296875, "kl": 0.00876174186123535, "learning_rate": 8.180272108843537e-07, "loss": -0.0183, "num_tokens": 19098599.0, "reward": 0.1640625, "reward_std": 0.5422958731651306, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 322, "step_time": 53.3912045750767 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1408.0, "completions/max_terminated_length": 1408.0, "completions/mean_length": 853.375, "completions/mean_terminated_length": 853.375, "completions/min_length": 427.0, "completions/min_terminated_length": 427.0, "entropy": 0.6776500511914492, "epoch": 0.18310657596371882, "frac_reward_zero_std": 0.625, "grad_norm": 0.341796875, "kl": 0.004744594931253232, "learning_rate": 8.174603174603174e-07, "loss": 0.0253, "num_tokens": 19162807.0, "reward": 0.1796875, "reward_std": 0.5445777773857117, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 323, "step_time": 49.32134121377021 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1729.0, "completions/max_terminated_length": 1729.0, "completions/mean_length": 770.234375, "completions/mean_terminated_length": 770.234375, "completions/min_length": 255.0, "completions/min_terminated_length": 255.0, "entropy": 0.8353247307240963, "epoch": 0.1836734693877551, "frac_reward_zero_std": 0.625, "grad_norm": 0.380859375, "kl": 0.005294523012707941, "learning_rate": 8.168934240362811e-07, "loss": -0.034, "num_tokens": 19221598.0, "reward": 0.25, "reward_std": 0.6666666865348816, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.0, "rewards/format_reward_func/std": 0.0, "step": 324, "step_time": 60.75746985152364 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1198.0, "completions/max_terminated_length": 1198.0, "completions/mean_length": 710.0625, "completions/mean_terminated_length": 710.0625, "completions/min_length": 231.0, "completions/min_terminated_length": 231.0, "entropy": 0.8704922962933779, "epoch": 0.18424036281179137, "frac_reward_zero_std": 0.375, "grad_norm": 0.6640625, "kl": 0.006477083035861142, "learning_rate": 8.163265306122449e-07, "loss": -0.0105, "num_tokens": 19278218.0, "reward": 0.328125, "reward_std": 0.6970388293266296, "rewards/accuracy_reward_func/mean": 0.28125, "rewards/accuracy_reward_func/std": 0.7007648944854736, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 325, "step_time": 42.49967502243817 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1223.0, "completions/max_terminated_length": 1223.0, "completions/mean_length": 693.765625, "completions/mean_terminated_length": 693.765625, "completions/min_length": 278.0, "completions/min_terminated_length": 278.0, "entropy": 0.7533231135457754, "epoch": 0.18480725623582767, "frac_reward_zero_std": 0.625, "grad_norm": 0.5, "kl": 0.014457460885751061, "learning_rate": 8.157596371882086e-07, "loss": 0.0182, "num_tokens": 19333523.0, "reward": 0.09375, "reward_std": 0.36596253514289856, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 326, "step_time": 42.95775222312659 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1570.0, "completions/max_terminated_length": 1570.0, "completions/mean_length": 838.71875, "completions/mean_terminated_length": 838.71875, "completions/min_length": 260.0, "completions/min_terminated_length": 260.0, "entropy": 0.6676177363842726, "epoch": 0.18537414965986396, "frac_reward_zero_std": 0.625, "grad_norm": 0.373046875, "kl": 0.006262856913963333, "learning_rate": 8.151927437641724e-07, "loss": 0.0079, "num_tokens": 19398801.0, "reward": 0.171875, "reward_std": 0.49776285886764526, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 327, "step_time": 56.06219778954983 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1752.0, "completions/max_terminated_length": 1752.0, "completions/mean_length": 826.015625, "completions/mean_terminated_length": 826.015625, "completions/min_length": 434.0, "completions/min_terminated_length": 434.0, "entropy": 0.8412331230938435, "epoch": 0.18594104308390022, "frac_reward_zero_std": 0.625, "grad_norm": 0.431640625, "kl": 0.010855631437152624, "learning_rate": 8.14625850340136e-07, "loss": 0.0193, "num_tokens": 19461722.0, "reward": 0.234375, "reward_std": 0.5907269716262817, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 328, "step_time": 61.37109683454037 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1725.0, "completions/max_terminated_length": 1725.0, "completions/mean_length": 770.890625, "completions/mean_terminated_length": 770.890625, "completions/min_length": 269.0, "completions/min_terminated_length": 269.0, "entropy": 0.8527063429355621, "epoch": 0.1865079365079365, "frac_reward_zero_std": 0.625, "grad_norm": 0.45703125, "kl": 0.007462741399649531, "learning_rate": 8.140589569160997e-07, "loss": 0.004, "num_tokens": 19522291.0, "reward": 0.1171875, "reward_std": 0.43408530950546265, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 329, "step_time": 61.21692004892975 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1513.0, "completions/max_terminated_length": 1513.0, "completions/mean_length": 742.6875, "completions/mean_terminated_length": 742.6875, "completions/min_length": 46.0, "completions/min_terminated_length": 46.0, "entropy": 0.8760765083134174, "epoch": 0.1870748299319728, "frac_reward_zero_std": 0.375, "grad_norm": 0.5859375, "kl": 0.007629668689332902, "learning_rate": 8.134920634920635e-07, "loss": 0.043, "num_tokens": 19579751.0, "reward": 0.25, "reward_std": 0.6299408078193665, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 330, "step_time": 52.7058973480016 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1631.0, "completions/max_terminated_length": 1631.0, "completions/mean_length": 683.75, "completions/mean_terminated_length": 683.75, "completions/min_length": 121.0, "completions/min_terminated_length": 121.0, "entropy": 1.0992595627903938, "epoch": 0.18764172335600907, "frac_reward_zero_std": 0.625, "grad_norm": 0.4609375, "kl": 0.008189230851712637, "learning_rate": 8.129251700680271e-07, "loss": -0.0046, "num_tokens": 19633055.0, "reward": 0.109375, "reward_std": 0.4315783977508545, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.015625, "rewards/format_reward_func/std": 0.08768405020236969, "step": 331, "step_time": 58.39010221231729 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1540.0, "completions/max_terminated_length": 1540.0, "completions/mean_length": 854.65625, "completions/mean_terminated_length": 854.65625, "completions/min_length": 236.0, "completions/min_terminated_length": 236.0, "entropy": 0.8076948560774326, "epoch": 0.18820861678004536, "frac_reward_zero_std": 0.875, "grad_norm": 0.2119140625, "kl": 0.004547693082713522, "learning_rate": 8.123582766439909e-07, "loss": 0.0008, "num_tokens": 19701201.0, "reward": 0.0625, "reward_std": 0.32732683420181274, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 332, "step_time": 54.694669464603066 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1356.0, "completions/max_terminated_length": 1356.0, "completions/mean_length": 837.5625, "completions/mean_terminated_length": 837.5625, "completions/min_length": 192.0, "completions/min_terminated_length": 192.0, "entropy": 0.7439503911882639, "epoch": 0.18877551020408162, "frac_reward_zero_std": 0.75, "grad_norm": 0.35546875, "kl": 0.007064882040140219, "learning_rate": 8.117913832199546e-07, "loss": 0.0165, "num_tokens": 19765029.0, "reward": 0.0703125, "reward_std": 0.27984103560447693, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 333, "step_time": 47.44372510071844 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1374.0, "completions/max_terminated_length": 1374.0, "completions/mean_length": 748.5625, "completions/mean_terminated_length": 748.5625, "completions/min_length": 287.0, "completions/min_terminated_length": 287.0, "entropy": 0.8740064445883036, "epoch": 0.1893424036281179, "frac_reward_zero_std": 0.75, "grad_norm": 0.3984375, "kl": 0.012891482037957758, "learning_rate": 8.112244897959183e-07, "loss": 0.0077, "num_tokens": 19822689.0, "reward": 0.25, "reward_std": 0.7126966714859009, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 334, "step_time": 47.68842428550124 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1891.0, "completions/max_terminated_length": 1891.0, "completions/mean_length": 837.1875, "completions/mean_terminated_length": 837.1875, "completions/min_length": 262.0, "completions/min_terminated_length": 262.0, "entropy": 0.655430793762207, "epoch": 0.1899092970521542, "frac_reward_zero_std": 0.5, "grad_norm": 0.484375, "kl": 0.006139033401268534, "learning_rate": 8.10657596371882e-07, "loss": 0.0307, "num_tokens": 19887005.0, "reward": 0.359375, "reward_std": 0.789156436920166, "rewards/accuracy_reward_func/mean": 0.3125, "rewards/accuracy_reward_func/std": 0.7319250702857971, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 335, "step_time": 65.92373467236757 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1099.0, "completions/max_terminated_length": 1099.0, "completions/mean_length": 646.640625, "completions/mean_terminated_length": 646.640625, "completions/min_length": 183.0, "completions/min_terminated_length": 183.0, "entropy": 0.8803813941776752, "epoch": 0.19047619047619047, "frac_reward_zero_std": 0.5, "grad_norm": 0.55859375, "kl": 0.013101798918796703, "learning_rate": 8.100907029478458e-07, "loss": 0.0343, "num_tokens": 19938014.0, "reward": 0.203125, "reward_std": 0.5471335649490356, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 336, "step_time": 38.181463305838406 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1683.0, "completions/max_terminated_length": 1683.0, "completions/mean_length": 890.3125, "completions/mean_terminated_length": 890.3125, "completions/min_length": 439.0, "completions/min_terminated_length": 439.0, "entropy": 0.6142123676836491, "epoch": 0.19104308390022676, "frac_reward_zero_std": 0.75, "grad_norm": 0.298828125, "kl": 0.009792614495381713, "learning_rate": 8.095238095238095e-07, "loss": 0.0013, "num_tokens": 20005442.0, "reward": 0.1484375, "reward_std": 0.49344661831855774, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 337, "step_time": 59.30274276807904 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1184.0, "completions/max_terminated_length": 1184.0, "completions/mean_length": 712.53125, "completions/mean_terminated_length": 712.53125, "completions/min_length": 156.0, "completions/min_terminated_length": 156.0, "entropy": 0.7328694984316826, "epoch": 0.19160997732426305, "frac_reward_zero_std": 0.875, "grad_norm": 0.310546875, "kl": 0.030268765331129543, "learning_rate": 8.089569160997733e-07, "loss": 0.005, "num_tokens": 20063588.0, "reward": 0.09375, "reward_std": 0.29378482699394226, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 338, "step_time": 42.308118044398725 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1431.0, "completions/max_terminated_length": 1431.0, "completions/mean_length": 682.65625, "completions/mean_terminated_length": 682.65625, "completions/min_length": 346.0, "completions/min_terminated_length": 346.0, "entropy": 0.9973875358700752, "epoch": 0.1921768707482993, "frac_reward_zero_std": 0.625, "grad_norm": 0.49609375, "kl": 0.009495670878095552, "learning_rate": 8.083900226757369e-07, "loss": -0.0035, "num_tokens": 20119038.0, "reward": 0.296875, "reward_std": 0.7111289501190186, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 339, "step_time": 51.06384574435651 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1514.0, "completions/max_terminated_length": 1514.0, "completions/mean_length": 837.703125, "completions/mean_terminated_length": 837.703125, "completions/min_length": 337.0, "completions/min_terminated_length": 337.0, "entropy": 0.7334924042224884, "epoch": 0.1927437641723356, "frac_reward_zero_std": 0.625, "grad_norm": 0.421875, "kl": 0.006166985287563875, "learning_rate": 8.078231292517006e-07, "loss": -0.0146, "num_tokens": 20183195.0, "reward": 0.1171875, "reward_std": 0.46922925114631653, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 340, "step_time": 53.17411183193326 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2742.0, "completions/max_terminated_length": 2742.0, "completions/mean_length": 880.5, "completions/mean_terminated_length": 880.5, "completions/min_length": 399.0, "completions/min_terminated_length": 399.0, "entropy": 1.0167470518499613, "epoch": 0.1933106575963719, "frac_reward_zero_std": 0.625, "grad_norm": 0.51171875, "kl": 0.01515770039986819, "learning_rate": 8.072562358276643e-07, "loss": -0.024, "num_tokens": 20252307.0, "reward": 0.140625, "reward_std": 0.4406772553920746, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 341, "step_time": 102.54252031445503 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1389.0, "completions/max_terminated_length": 1389.0, "completions/mean_length": 827.375, "completions/mean_terminated_length": 827.375, "completions/min_length": 334.0, "completions/min_terminated_length": 334.0, "entropy": 0.7021002303808928, "epoch": 0.19387755102040816, "frac_reward_zero_std": 0.75, "grad_norm": 0.349609375, "kl": 0.00620406138477847, "learning_rate": 8.066893424036281e-07, "loss": 0.0052, "num_tokens": 20315067.0, "reward": 0.09375, "reward_std": 0.36596253514289856, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 342, "step_time": 48.8435224480927 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1466.0, "completions/max_terminated_length": 1466.0, "completions/mean_length": 900.28125, "completions/mean_terminated_length": 900.28125, "completions/min_length": 565.0, "completions/min_terminated_length": 565.0, "entropy": 0.6297467797994614, "epoch": 0.19444444444444445, "frac_reward_zero_std": 0.75, "grad_norm": 0.361328125, "kl": 0.008732411442906596, "learning_rate": 8.061224489795918e-07, "loss": -0.0083, "num_tokens": 20383693.0, "reward": 0.09375, "reward_std": 0.36596253514289856, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 343, "step_time": 51.75457154493779 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1693.0, "completions/max_terminated_length": 1693.0, "completions/mean_length": 737.515625, "completions/mean_terminated_length": 737.515625, "completions/min_length": 259.0, "completions/min_terminated_length": 259.0, "entropy": 0.6419767197221518, "epoch": 0.19501133786848074, "frac_reward_zero_std": 0.5, "grad_norm": 0.5546875, "kl": 0.005702765440219082, "learning_rate": 8.055555555555556e-07, "loss": 0.0057, "num_tokens": 20440726.0, "reward": 0.34375, "reward_std": 0.728528618812561, "rewards/accuracy_reward_func/mean": 0.3125, "rewards/accuracy_reward_func/std": 0.7319250702857971, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 344, "step_time": 59.728486637584865 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 752.296875, "completions/mean_terminated_length": 752.296875, "completions/min_length": 435.0, "completions/min_terminated_length": 435.0, "entropy": 0.9790338799357414, "epoch": 0.195578231292517, "frac_reward_zero_std": 0.625, "grad_norm": 0.435546875, "kl": 0.009619498610845767, "learning_rate": 8.049886621315193e-07, "loss": -0.0008, "num_tokens": 20501713.0, "reward": 0.1015625, "reward_std": 0.3692515790462494, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 345, "step_time": 43.7907804697752 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1285.0, "completions/max_terminated_length": 1285.0, "completions/mean_length": 673.359375, "completions/mean_terminated_length": 673.359375, "completions/min_length": 40.0, "completions/min_terminated_length": 40.0, "entropy": 0.9337127301841974, "epoch": 0.1961451247165533, "frac_reward_zero_std": 0.25, "grad_norm": 0.66796875, "kl": 0.00934422598220408, "learning_rate": 8.04421768707483e-07, "loss": 0.0401, "num_tokens": 20555952.0, "reward": 0.4375, "reward_std": 0.8237544298171997, "rewards/accuracy_reward_func/mean": 0.40625, "rewards/accuracy_reward_func/std": 0.8110105991363525, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 346, "step_time": 45.38670400623232 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2688.0, "completions/max_terminated_length": 2688.0, "completions/mean_length": 761.03125, "completions/mean_terminated_length": 761.03125, "completions/min_length": 292.0, "completions/min_terminated_length": 292.0, "entropy": 0.4654211699962616, "epoch": 0.19671201814058956, "frac_reward_zero_std": 0.375, "grad_norm": 0.51953125, "kl": 0.0064781456894706935, "learning_rate": 8.038548752834466e-07, "loss": 0.0359, "num_tokens": 20613442.0, "reward": 0.328125, "reward_std": 0.7194505929946899, "rewards/accuracy_reward_func/mean": 0.28125, "rewards/accuracy_reward_func/std": 0.7007648944854736, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 347, "step_time": 98.53148745372891 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1560.0, "completions/max_terminated_length": 1560.0, "completions/mean_length": 800.109375, "completions/mean_terminated_length": 800.109375, "completions/min_length": 465.0, "completions/min_terminated_length": 465.0, "entropy": 0.7911105696111917, "epoch": 0.19727891156462585, "frac_reward_zero_std": 0.5, "grad_norm": 0.462890625, "kl": 0.007789595707436092, "learning_rate": 8.032879818594103e-07, "loss": -0.0085, "num_tokens": 20674321.0, "reward": 0.171875, "reward_std": 0.49776285886764526, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 348, "step_time": 54.41751306131482 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1034.0, "completions/max_terminated_length": 1034.0, "completions/mean_length": 695.703125, "completions/mean_terminated_length": 695.703125, "completions/min_length": 290.0, "completions/min_terminated_length": 290.0, "entropy": 1.1095048561692238, "epoch": 0.19784580498866214, "frac_reward_zero_std": 0.625, "grad_norm": 0.5546875, "kl": 0.012152556111686863, "learning_rate": 8.027210884353741e-07, "loss": 0.0108, "num_tokens": 20730230.0, "reward": 0.21875, "reward_std": 0.6164092421531677, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 349, "step_time": 36.84613507427275 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1679.0, "completions/max_terminated_length": 1679.0, "completions/mean_length": 767.109375, "completions/mean_terminated_length": 767.109375, "completions/min_length": 305.0, "completions/min_terminated_length": 305.0, "entropy": 0.8563258647918701, "epoch": 0.1984126984126984, "frac_reward_zero_std": 0.625, "grad_norm": 0.369140625, "kl": 0.011192118050530553, "learning_rate": 8.021541950113378e-07, "loss": 0.0244, "num_tokens": 20794885.0, "reward": 0.109375, "reward_std": 0.37234508991241455, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 350, "step_time": 63.66168717946857 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1726.0, "completions/max_terminated_length": 1726.0, "completions/mean_length": 766.4375, "completions/mean_terminated_length": 766.4375, "completions/min_length": 304.0, "completions/min_terminated_length": 304.0, "entropy": 0.808762526139617, "epoch": 0.1989795918367347, "frac_reward_zero_std": 0.375, "grad_norm": 0.63671875, "kl": 0.004394636191136669, "learning_rate": 8.015873015873016e-07, "loss": -0.0018, "num_tokens": 20855457.0, "reward": 0.2890625, "reward_std": 0.6655030846595764, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 351, "step_time": 61.878230896778405 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1161.0, "completions/max_terminated_length": 1161.0, "completions/mean_length": 730.625, "completions/mean_terminated_length": 730.625, "completions/min_length": 369.0, "completions/min_terminated_length": 369.0, "entropy": 0.7505794577300549, "epoch": 0.19954648526077098, "frac_reward_zero_std": 0.625, "grad_norm": 0.45703125, "kl": 0.0053577764047076926, "learning_rate": 8.010204081632652e-07, "loss": -0.004, "num_tokens": 20912017.0, "reward": 0.1640625, "reward_std": 0.496453195810318, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 352, "step_time": 40.289197850972414 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1783.0, "completions/max_terminated_length": 1783.0, "completions/mean_length": 776.0625, "completions/mean_terminated_length": 776.0625, "completions/min_length": 357.0, "completions/min_terminated_length": 357.0, "entropy": 1.0009315647184849, "epoch": 0.20011337868480725, "frac_reward_zero_std": 0.875, "grad_norm": 0.27734375, "kl": 0.005335954439942725, "learning_rate": 8.00453514739229e-07, "loss": -0.0115, "num_tokens": 20971821.0, "reward": 0.0390625, "reward_std": 0.13524484634399414, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 353, "step_time": 62.82513804361224 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 4096.0, "completions/max_terminated_length": 1211.0, "completions/mean_length": 826.515625, "completions/mean_terminated_length": 721.04833984375, "completions/min_length": 8.0, "completions/min_terminated_length": 8.0, "entropy": 0.6306920368224382, "epoch": 0.20068027210884354, "frac_reward_zero_std": 0.25, "grad_norm": 0.578125, "kl": 0.004599000429152511, "learning_rate": 7.998866213151927e-07, "loss": -0.0351, "num_tokens": 21034734.0, "reward": 0.6875, "reward_std": 1.005934715270996, "rewards/accuracy_reward_func/mean": 0.625, "rewards/accuracy_reward_func/std": 0.934353232383728, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 354, "step_time": 170.80954848509282 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1284.0, "completions/max_terminated_length": 1284.0, "completions/mean_length": 793.71875, "completions/mean_terminated_length": 793.71875, "completions/min_length": 348.0, "completions/min_terminated_length": 348.0, "entropy": 0.9095571152865887, "epoch": 0.20124716553287983, "frac_reward_zero_std": 0.75, "grad_norm": 0.41796875, "kl": 0.006502260177512653, "learning_rate": 7.993197278911564e-07, "loss": -0.0001, "num_tokens": 21094500.0, "reward": 0.078125, "reward_std": 0.2847827076911926, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 355, "step_time": 45.50296873413026 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1178.0, "completions/max_terminated_length": 1178.0, "completions/mean_length": 699.28125, "completions/mean_terminated_length": 699.28125, "completions/min_length": 253.0, "completions/min_terminated_length": 253.0, "entropy": 0.7517442852258682, "epoch": 0.2018140589569161, "frac_reward_zero_std": 0.875, "grad_norm": 0.2265625, "kl": 0.009225394751410931, "learning_rate": 7.987528344671201e-07, "loss": -0.0013, "num_tokens": 21153798.0, "reward": 0.1328125, "reward_std": 0.5058560967445374, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 356, "step_time": 43.04798846133053 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1961.0, "completions/max_terminated_length": 1961.0, "completions/mean_length": 805.625, "completions/mean_terminated_length": 805.625, "completions/min_length": 273.0, "completions/min_terminated_length": 273.0, "entropy": 0.7604002766311169, "epoch": 0.20238095238095238, "frac_reward_zero_std": 0.375, "grad_norm": 0.53125, "kl": 0.006553398416144773, "learning_rate": 7.981859410430839e-07, "loss": 0.0395, "num_tokens": 21214686.0, "reward": 0.4765625, "reward_std": 0.8424721956253052, "rewards/accuracy_reward_func/mean": 0.4375, "rewards/accuracy_reward_func/std": 0.8333333730697632, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 357, "step_time": 68.64221620652825 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1548.0, "completions/max_terminated_length": 1548.0, "completions/mean_length": 774.65625, "completions/mean_terminated_length": 774.65625, "completions/min_length": 141.0, "completions/min_terminated_length": 141.0, "entropy": 0.6257712636142969, "epoch": 0.20294784580498867, "frac_reward_zero_std": 0.625, "grad_norm": 0.427734375, "kl": 0.009716352680698037, "learning_rate": 7.976190476190476e-07, "loss": 0.0088, "num_tokens": 21275408.0, "reward": 0.1328125, "reward_std": 0.4386323094367981, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 358, "step_time": 54.75614361651242 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1500.0, "completions/max_terminated_length": 1500.0, "completions/mean_length": 647.3125, "completions/mean_terminated_length": 647.3125, "completions/min_length": 311.0, "completions/min_terminated_length": 311.0, "entropy": 0.9415576122701168, "epoch": 0.20351473922902494, "frac_reward_zero_std": 0.625, "grad_norm": 0.58984375, "kl": 0.008676898840349168, "learning_rate": 7.970521541950113e-07, "loss": -0.0108, "num_tokens": 21327556.0, "reward": 0.1875, "reward_std": 0.5455447435379028, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 359, "step_time": 52.70056934468448 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1533.0, "completions/max_terminated_length": 1533.0, "completions/mean_length": 738.34375, "completions/mean_terminated_length": 738.34375, "completions/min_length": 267.0, "completions/min_terminated_length": 267.0, "entropy": 0.6640665493905544, "epoch": 0.20408163265306123, "frac_reward_zero_std": 0.5, "grad_norm": 0.51953125, "kl": 0.0048632768739480525, "learning_rate": 7.96485260770975e-07, "loss": 0.0163, "num_tokens": 21384250.0, "reward": 0.375, "reward_std": 0.7237468957901001, "rewards/accuracy_reward_func/mean": 0.3125, "rewards/accuracy_reward_func/std": 0.7319250702857971, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 360, "step_time": 53.82986585609615 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1365.0, "completions/max_terminated_length": 1365.0, "completions/mean_length": 806.40625, "completions/mean_terminated_length": 806.40625, "completions/min_length": 438.0, "completions/min_terminated_length": 438.0, "entropy": 0.897387720644474, "epoch": 0.2046485260770975, "frac_reward_zero_std": 0.625, "grad_norm": 0.466796875, "kl": 0.004769071820192039, "learning_rate": 7.959183673469388e-07, "loss": -0.006, "num_tokens": 21446972.0, "reward": 0.1484375, "reward_std": 0.49344661831855774, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 361, "step_time": 49.43968720827252 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1389.0, "completions/max_terminated_length": 1389.0, "completions/mean_length": 751.515625, "completions/mean_terminated_length": 751.515625, "completions/min_length": 381.0, "completions/min_terminated_length": 381.0, "entropy": 0.9682449884712696, "epoch": 0.20521541950113378, "frac_reward_zero_std": 0.5, "grad_norm": 0.494140625, "kl": 0.0074831047968473285, "learning_rate": 7.953514739229025e-07, "loss": 0.0206, "num_tokens": 21504861.0, "reward": 0.1484375, "reward_std": 0.44257262349128723, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 362, "step_time": 48.7774790963158 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1300.0, "completions/max_terminated_length": 1300.0, "completions/mean_length": 673.171875, "completions/mean_terminated_length": 673.171875, "completions/min_length": 302.0, "completions/min_terminated_length": 302.0, "entropy": 0.8318615090101957, "epoch": 0.20578231292517007, "frac_reward_zero_std": 0.5, "grad_norm": 0.59765625, "kl": 0.006348504466586746, "learning_rate": 7.947845804988662e-07, "loss": -0.0042, "num_tokens": 21558000.0, "reward": 0.3046875, "reward_std": 0.710561990737915, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 363, "step_time": 45.67016400489956 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1382.0, "completions/max_terminated_length": 1382.0, "completions/mean_length": 620.296875, "completions/mean_terminated_length": 620.296875, "completions/min_length": 175.0, "completions/min_terminated_length": 175.0, "entropy": 0.9658241774886847, "epoch": 0.20634920634920634, "frac_reward_zero_std": 0.75, "grad_norm": 0.451171875, "kl": 0.019720278680324554, "learning_rate": 7.942176870748299e-07, "loss": -0.0023, "num_tokens": 21610395.0, "reward": 0.0625, "reward_std": 0.2745848298072815, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 364, "step_time": 49.3195380019024 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1189.0, "completions/max_terminated_length": 1189.0, "completions/mean_length": 728.140625, "completions/mean_terminated_length": 728.140625, "completions/min_length": 388.0, "completions/min_terminated_length": 388.0, "entropy": 0.876466816291213, "epoch": 0.20691609977324263, "frac_reward_zero_std": 0.375, "grad_norm": 0.671875, "kl": 0.018591600528452545, "learning_rate": 7.936507936507936e-07, "loss": -0.0426, "num_tokens": 21665964.0, "reward": 0.2578125, "reward_std": 0.629891574382782, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 365, "step_time": 44.70258327387273 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1513.0, "completions/max_terminated_length": 1513.0, "completions/mean_length": 734.28125, "completions/mean_terminated_length": 734.28125, "completions/min_length": 310.0, "completions/min_terminated_length": 310.0, "entropy": 0.7373211551457644, "epoch": 0.20748299319727892, "frac_reward_zero_std": 0.5, "grad_norm": 0.5078125, "kl": 0.03787548135733232, "learning_rate": 7.930839002267573e-07, "loss": 0.0167, "num_tokens": 21725278.0, "reward": 0.1328125, "reward_std": 0.489915668964386, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0078125, "rewards/format_reward_func/std": 0.0625, "step": 366, "step_time": 55.486514929682016 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1356.0, "completions/max_terminated_length": 1356.0, "completions/mean_length": 826.609375, "completions/mean_terminated_length": 826.609375, "completions/min_length": 299.0, "completions/min_terminated_length": 299.0, "entropy": 0.803842194378376, "epoch": 0.20804988662131518, "frac_reward_zero_std": 0.5, "grad_norm": 0.49609375, "kl": 0.01093247470271308, "learning_rate": 7.92517006802721e-07, "loss": 0.0045, "num_tokens": 21788645.0, "reward": 0.1796875, "reward_std": 0.49894481897354126, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 367, "step_time": 48.00471571832895 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1389.0, "completions/max_terminated_length": 1389.0, "completions/mean_length": 688.078125, "completions/mean_terminated_length": 688.078125, "completions/min_length": 367.0, "completions/min_terminated_length": 367.0, "entropy": 1.0153057016432285, "epoch": 0.20861678004535147, "frac_reward_zero_std": 0.625, "grad_norm": 0.52734375, "kl": 0.009997964443755336, "learning_rate": 7.919501133786848e-07, "loss": 0.0096, "num_tokens": 21843770.0, "reward": 0.1640625, "reward_std": 0.496453195810318, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 368, "step_time": 49.48769662436098 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1204.0, "completions/max_terminated_length": 1204.0, "completions/mean_length": 751.9375, "completions/mean_terminated_length": 751.9375, "completions/min_length": 419.0, "completions/min_terminated_length": 419.0, "entropy": 0.7951037138700485, "epoch": 0.20918367346938777, "frac_reward_zero_std": 0.75, "grad_norm": 0.380859375, "kl": 0.006458746662246995, "learning_rate": 7.913832199546485e-07, "loss": -0.0013, "num_tokens": 21902414.0, "reward": 0.078125, "reward_std": 0.2847827076911926, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 369, "step_time": 42.71605859789997 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1887.0, "completions/max_terminated_length": 1887.0, "completions/mean_length": 770.671875, "completions/mean_terminated_length": 770.671875, "completions/min_length": 160.0, "completions/min_terminated_length": 160.0, "entropy": 0.8654769137501717, "epoch": 0.20975056689342403, "frac_reward_zero_std": 0.75, "grad_norm": 0.353515625, "kl": 0.007016115239821374, "learning_rate": 7.908163265306123e-07, "loss": -0.0292, "num_tokens": 21962993.0, "reward": 0.203125, "reward_std": 0.6529466509819031, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 370, "step_time": 66.75023751053959 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1402.0, "completions/max_terminated_length": 1402.0, "completions/mean_length": 711.1875, "completions/mean_terminated_length": 711.1875, "completions/min_length": 363.0, "completions/min_terminated_length": 363.0, "entropy": 0.9247559905052185, "epoch": 0.21031746031746032, "frac_reward_zero_std": 0.75, "grad_norm": 0.4296875, "kl": 0.01341819655499421, "learning_rate": 7.902494331065758e-07, "loss": -0.01, "num_tokens": 22018997.0, "reward": 0.078125, "reward_std": 0.2847827076911926, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 371, "step_time": 49.6273089479655 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1664.0, "completions/max_terminated_length": 1664.0, "completions/mean_length": 824.03125, "completions/mean_terminated_length": 824.03125, "completions/min_length": 375.0, "completions/min_terminated_length": 375.0, "entropy": 0.782563166692853, "epoch": 0.2108843537414966, "frac_reward_zero_std": 0.875, "grad_norm": 0.2060546875, "kl": 0.008505658013746142, "learning_rate": 7.896825396825396e-07, "loss": 0.0049, "num_tokens": 22080911.0, "reward": 0.21875, "reward_std": 0.6777355074882507, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 372, "step_time": 58.256893769837916 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1581.0, "completions/max_terminated_length": 1581.0, "completions/mean_length": 856.078125, "completions/mean_terminated_length": 856.078125, "completions/min_length": 337.0, "completions/min_terminated_length": 337.0, "entropy": 0.7458044551312923, "epoch": 0.21145124716553287, "frac_reward_zero_std": 0.625, "grad_norm": 0.392578125, "kl": 0.0071403429537895136, "learning_rate": 7.891156462585033e-07, "loss": -0.006, "num_tokens": 22145620.0, "reward": 0.0859375, "reward_std": 0.36247265338897705, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 373, "step_time": 55.691850893199444 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 675.375, "completions/mean_terminated_length": 675.375, "completions/min_length": 287.0, "completions/min_terminated_length": 287.0, "entropy": 1.13796192035079, "epoch": 0.21201814058956917, "frac_reward_zero_std": 0.5, "grad_norm": 0.5234375, "kl": 0.0060508267633849755, "learning_rate": 7.885487528344671e-07, "loss": -0.0124, "num_tokens": 22198964.0, "reward": 0.3125, "reward_std": 0.7319250702857971, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 374, "step_time": 36.025930147618055 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1204.0, "completions/max_terminated_length": 1204.0, "completions/mean_length": 728.03125, "completions/mean_terminated_length": 728.03125, "completions/min_length": 280.0, "completions/min_terminated_length": 280.0, "entropy": 0.7827387042343616, "epoch": 0.21258503401360543, "frac_reward_zero_std": 0.75, "grad_norm": 0.392578125, "kl": 0.006068907270673662, "learning_rate": 7.879818594104308e-07, "loss": 0.0027, "num_tokens": 22255678.0, "reward": 0.1015625, "reward_std": 0.3692515790462494, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 375, "step_time": 42.63181810081005 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1404.0, "completions/max_terminated_length": 1404.0, "completions/mean_length": 736.28125, "completions/mean_terminated_length": 736.28125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "entropy": 0.6721552610397339, "epoch": 0.21315192743764172, "frac_reward_zero_std": 0.375, "grad_norm": 0.6171875, "kl": 0.005795677818241529, "learning_rate": 7.874149659863946e-07, "loss": 0.0195, "num_tokens": 22313056.0, "reward": 0.3828125, "reward_std": 0.795534074306488, "rewards/accuracy_reward_func/mean": 0.34375, "rewards/accuracy_reward_func/std": 0.7605084180831909, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 376, "step_time": 49.154366455972195 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1287.0, "completions/max_terminated_length": 1287.0, "completions/mean_length": 644.5, "completions/mean_terminated_length": 644.5, "completions/min_length": 297.0, "completions/min_terminated_length": 297.0, "entropy": 0.6348938029259443, "epoch": 0.213718820861678, "frac_reward_zero_std": 0.75, "grad_norm": 0.4453125, "kl": 0.006758220930350944, "learning_rate": 7.868480725623582e-07, "loss": 0.0116, "num_tokens": 22362848.0, "reward": 0.09375, "reward_std": 0.36596253514289856, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 377, "step_time": 45.09674393199384 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2051.0, "completions/max_terminated_length": 2051.0, "completions/mean_length": 705.109375, "completions/mean_terminated_length": 705.109375, "completions/min_length": 342.0, "completions/min_terminated_length": 342.0, "entropy": 0.6979906857013702, "epoch": 0.21428571428571427, "frac_reward_zero_std": 0.375, "grad_norm": 0.609375, "kl": 0.020916220819344744, "learning_rate": 7.86281179138322e-07, "loss": 0.0318, "num_tokens": 22418207.0, "reward": 0.453125, "reward_std": 0.8003905415534973, "rewards/accuracy_reward_func/mean": 0.40625, "rewards/accuracy_reward_func/std": 0.8110105991363525, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 378, "step_time": 72.91525127086788 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1368.0, "completions/max_terminated_length": 1368.0, "completions/mean_length": 740.421875, "completions/mean_terminated_length": 740.421875, "completions/min_length": 331.0, "completions/min_terminated_length": 331.0, "entropy": 1.0521539971232414, "epoch": 0.21485260770975056, "frac_reward_zero_std": 0.5, "grad_norm": 0.7421875, "kl": 0.00853558856761083, "learning_rate": 7.857142857142856e-07, "loss": -0.0424, "num_tokens": 22477218.0, "reward": 0.1796875, "reward_std": 0.49894481897354126, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 379, "step_time": 48.239318737760186 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2350.0, "completions/max_terminated_length": 2350.0, "completions/mean_length": 785.5, "completions/mean_terminated_length": 785.5, "completions/min_length": 5.0, "completions/min_terminated_length": 5.0, "entropy": 0.8634990677237511, "epoch": 0.21541950113378686, "frac_reward_zero_std": 0.625, "grad_norm": 0.423828125, "kl": 0.007020699333224911, "learning_rate": 7.851473922902494e-07, "loss": 0.0021, "num_tokens": 22542362.0, "reward": 0.140625, "reward_std": 0.4406772553920746, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 380, "step_time": 87.2783026220277 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1341.0, "completions/max_terminated_length": 1341.0, "completions/mean_length": 739.515625, "completions/mean_terminated_length": 739.515625, "completions/min_length": 284.0, "completions/min_terminated_length": 284.0, "entropy": 0.6452418379485607, "epoch": 0.21598639455782312, "frac_reward_zero_std": 0.375, "grad_norm": 0.55078125, "kl": 0.008970728755230084, "learning_rate": 7.845804988662131e-07, "loss": 0.0187, "num_tokens": 22599899.0, "reward": 0.2421875, "reward_std": 0.5908843874931335, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 381, "step_time": 47.75332001410425 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1873.0, "completions/max_terminated_length": 1873.0, "completions/mean_length": 758.328125, "completions/mean_terminated_length": 758.328125, "completions/min_length": 228.0, "completions/min_terminated_length": 228.0, "entropy": 0.7167969793081284, "epoch": 0.2165532879818594, "frac_reward_zero_std": 0.625, "grad_norm": 0.4296875, "kl": 0.014076261446462013, "learning_rate": 7.840136054421769e-07, "loss": -0.0125, "num_tokens": 22658704.0, "reward": 0.2109375, "reward_std": 0.5477564930915833, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 382, "step_time": 65.95075592491776 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1037.0, "completions/max_terminated_length": 1037.0, "completions/mean_length": 569.5625, "completions/mean_terminated_length": 569.5625, "completions/min_length": 207.0, "completions/min_terminated_length": 207.0, "entropy": 0.9594349786639214, "epoch": 0.2171201814058957, "frac_reward_zero_std": 0.375, "grad_norm": 0.796875, "kl": 0.009265722255804576, "learning_rate": 7.834467120181405e-07, "loss": -0.0062, "num_tokens": 22706340.0, "reward": 0.2265625, "reward_std": 0.5904644727706909, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 383, "step_time": 37.67073441017419 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1525.0, "completions/max_terminated_length": 1525.0, "completions/mean_length": 876.65625, "completions/mean_terminated_length": 876.65625, "completions/min_length": 447.0, "completions/min_terminated_length": 447.0, "entropy": 0.887593649327755, "epoch": 0.21768707482993196, "frac_reward_zero_std": 0.75, "grad_norm": 0.296875, "kl": 0.005941338909906335, "learning_rate": 7.828798185941043e-07, "loss": 0.006, "num_tokens": 22775126.0, "reward": 0.1484375, "reward_std": 0.5540592670440674, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 384, "step_time": 54.38487575855106 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1332.0, "completions/max_terminated_length": 1332.0, "completions/mean_length": 742.078125, "completions/mean_terminated_length": 742.078125, "completions/min_length": 368.0, "completions/min_terminated_length": 368.0, "entropy": 0.9018192701041698, "epoch": 0.21825396825396826, "frac_reward_zero_std": 0.75, "grad_norm": 0.373046875, "kl": 0.0055784515861887485, "learning_rate": 7.82312925170068e-07, "loss": 0.0013, "num_tokens": 22833043.0, "reward": 0.1796875, "reward_std": 0.49894481897354126, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 385, "step_time": 48.398129280656576 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1386.0, "completions/max_terminated_length": 1386.0, "completions/mean_length": 725.171875, "completions/mean_terminated_length": 725.171875, "completions/min_length": 8.0, "completions/min_terminated_length": 8.0, "entropy": 0.9289520047605038, "epoch": 0.21882086167800455, "frac_reward_zero_std": 0.5, "grad_norm": 0.5703125, "kl": 0.011218311949050985, "learning_rate": 7.817460317460318e-07, "loss": -0.0105, "num_tokens": 22894374.0, "reward": 0.203125, "reward_std": 0.5471335649490356, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 386, "step_time": 52.46727229747921 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 664.46875, "completions/mean_terminated_length": 664.46875, "completions/min_length": 165.0, "completions/min_terminated_length": 165.0, "entropy": 0.8146755807101727, "epoch": 0.2193877551020408, "frac_reward_zero_std": 0.5, "grad_norm": 0.5390625, "kl": 0.020864911552052945, "learning_rate": 7.811791383219954e-07, "loss": 0.021, "num_tokens": 22949100.0, "reward": 0.2890625, "reward_std": 0.688941478729248, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 387, "step_time": 71.70860512368381 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1321.0, "completions/max_terminated_length": 1321.0, "completions/mean_length": 740.78125, "completions/mean_terminated_length": 740.78125, "completions/min_length": 9.0, "completions/min_terminated_length": 9.0, "entropy": 0.8137216381728649, "epoch": 0.2199546485260771, "frac_reward_zero_std": 0.5, "grad_norm": 0.546875, "kl": 0.007592617155751213, "learning_rate": 7.806122448979591e-07, "loss": -0.0027, "num_tokens": 23006438.0, "reward": 0.1640625, "reward_std": 0.496453195810318, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 388, "step_time": 46.02304187230766 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1633.0, "completions/max_terminated_length": 1633.0, "completions/mean_length": 886.15625, "completions/mean_terminated_length": 886.15625, "completions/min_length": 493.0, "completions/min_terminated_length": 493.0, "entropy": 0.8164814542979002, "epoch": 0.2205215419501134, "frac_reward_zero_std": 0.625, "grad_norm": 0.412109375, "kl": 0.004828404598811176, "learning_rate": 7.800453514739229e-07, "loss": -0.0077, "num_tokens": 23075376.0, "reward": 0.2265625, "reward_std": 0.6167611479759216, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 389, "step_time": 57.176508142612875 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1641.0, "completions/max_terminated_length": 1641.0, "completions/mean_length": 741.0, "completions/mean_terminated_length": 741.0, "completions/min_length": 329.0, "completions/min_terminated_length": 329.0, "entropy": 0.8530206680297852, "epoch": 0.22108843537414966, "frac_reward_zero_std": 0.5, "grad_norm": 0.59765625, "kl": 0.010848585545318201, "learning_rate": 7.794784580498865e-07, "loss": 0.0302, "num_tokens": 23132752.0, "reward": 0.3203125, "reward_std": 0.6978833675384521, "rewards/accuracy_reward_func/mean": 0.28125, "rewards/accuracy_reward_func/std": 0.7007648944854736, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 390, "step_time": 57.22498300112784 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1255.0, "completions/max_terminated_length": 1255.0, "completions/mean_length": 608.375, "completions/mean_terminated_length": 608.375, "completions/min_length": 320.0, "completions/min_terminated_length": 320.0, "entropy": 0.8223625533282757, "epoch": 0.22165532879818595, "frac_reward_zero_std": 0.5, "grad_norm": 0.59765625, "kl": 0.015227242663968354, "learning_rate": 7.789115646258503e-07, "loss": 0.0455, "num_tokens": 23181792.0, "reward": 0.2421875, "reward_std": 0.5908843874931335, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 391, "step_time": 43.47986264619976 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1616.0, "completions/max_terminated_length": 1616.0, "completions/mean_length": 780.921875, "completions/mean_terminated_length": 780.921875, "completions/min_length": 394.0, "completions/min_terminated_length": 394.0, "entropy": 0.7135511264204979, "epoch": 0.2222222222222222, "frac_reward_zero_std": 0.75, "grad_norm": 0.37109375, "kl": 0.010050005716038868, "learning_rate": 7.78344671201814e-07, "loss": 0.0072, "num_tokens": 23241811.0, "reward": 0.34375, "reward_std": 0.8207381367683411, "rewards/accuracy_reward_func/mean": 0.28125, "rewards/accuracy_reward_func/std": 0.7007648944854736, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 392, "step_time": 56.368631350807846 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1227.0, "completions/max_terminated_length": 1227.0, "completions/mean_length": 735.453125, "completions/mean_terminated_length": 735.453125, "completions/min_length": 365.0, "completions/min_terminated_length": 365.0, "entropy": 0.9050820115953684, "epoch": 0.2227891156462585, "frac_reward_zero_std": 0.625, "grad_norm": 0.466796875, "kl": 0.005837492266437039, "learning_rate": 7.777777777777778e-07, "loss": 0.0221, "num_tokens": 23298648.0, "reward": 0.1953125, "reward_std": 0.6016985774040222, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 393, "step_time": 43.485394665040076 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1343.0, "completions/max_terminated_length": 1343.0, "completions/mean_length": 807.890625, "completions/mean_terminated_length": 807.890625, "completions/min_length": 376.0, "completions/min_terminated_length": 376.0, "entropy": 0.8196720741689205, "epoch": 0.2233560090702948, "frac_reward_zero_std": 0.625, "grad_norm": 0.423828125, "kl": 0.009971795996534638, "learning_rate": 7.772108843537415e-07, "loss": 0.0204, "num_tokens": 23361753.0, "reward": 0.1875, "reward_std": 0.5455447435379028, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 394, "step_time": 47.62207393441349 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1605.0, "completions/max_terminated_length": 1605.0, "completions/mean_length": 822.5625, "completions/mean_terminated_length": 822.5625, "completions/min_length": 371.0, "completions/min_terminated_length": 371.0, "entropy": 0.5809940975159407, "epoch": 0.22392290249433106, "frac_reward_zero_std": 0.5, "grad_norm": 0.5078125, "kl": 0.004477146649151109, "learning_rate": 7.766439909297052e-07, "loss": 0.0224, "num_tokens": 23425413.0, "reward": 0.2109375, "reward_std": 0.5477564930915833, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 395, "step_time": 56.65555914957076 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1438.0, "completions/max_terminated_length": 1438.0, "completions/mean_length": 715.125, "completions/mean_terminated_length": 715.125, "completions/min_length": 183.0, "completions/min_terminated_length": 183.0, "entropy": 0.9844364896416664, "epoch": 0.22448979591836735, "frac_reward_zero_std": 0.375, "grad_norm": 0.6328125, "kl": 0.007322670266148634, "learning_rate": 7.760770975056688e-07, "loss": 0.0234, "num_tokens": 23482165.0, "reward": 0.5078125, "reward_std": 0.8796291947364807, "rewards/accuracy_reward_func/mean": 0.46875, "rewards/accuracy_reward_func/std": 0.8539125919342041, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 396, "step_time": 50.7606395855546 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1679.0, "completions/max_terminated_length": 1679.0, "completions/mean_length": 638.4375, "completions/mean_terminated_length": 638.4375, "completions/min_length": 21.0, "completions/min_terminated_length": 21.0, "entropy": 0.8209603633731604, "epoch": 0.22505668934240364, "frac_reward_zero_std": 0.625, "grad_norm": 0.50390625, "kl": 0.006006026364048012, "learning_rate": 7.755102040816326e-07, "loss": 0.0226, "num_tokens": 23532073.0, "reward": 0.2578125, "reward_std": 0.7014281749725342, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 397, "step_time": 58.50951211899519 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 745.4375, "completions/mean_terminated_length": 745.4375, "completions/min_length": 315.0, "completions/min_terminated_length": 315.0, "entropy": 1.0559663251042366, "epoch": 0.2256235827664399, "frac_reward_zero_std": 0.75, "grad_norm": 0.37890625, "kl": 0.011853849471663125, "learning_rate": 7.749433106575963e-07, "loss": 0.005, "num_tokens": 23591981.0, "reward": 0.0859375, "reward_std": 0.28942590951919556, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 398, "step_time": 46.28473854716867 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1452.0, "completions/max_terminated_length": 1452.0, "completions/mean_length": 732.6875, "completions/mean_terminated_length": 732.6875, "completions/min_length": 406.0, "completions/min_terminated_length": 406.0, "entropy": 0.9582856111228466, "epoch": 0.2261904761904762, "frac_reward_zero_std": 0.5, "grad_norm": 0.5703125, "kl": 0.010223097779089585, "learning_rate": 7.743764172335601e-07, "loss": 0.002, "num_tokens": 23648993.0, "reward": 0.2421875, "reward_std": 0.5908843874931335, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 399, "step_time": 50.92035705037415 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1604.0, "completions/max_terminated_length": 1604.0, "completions/mean_length": 701.421875, "completions/mean_terminated_length": 701.421875, "completions/min_length": 316.0, "completions/min_terminated_length": 316.0, "entropy": 0.708191528916359, "epoch": 0.22675736961451248, "frac_reward_zero_std": 0.625, "grad_norm": 0.4296875, "kl": 0.00698961055604741, "learning_rate": 7.738095238095238e-07, "loss": -0.0018, "num_tokens": 23702196.0, "reward": 0.265625, "reward_std": 0.6040127277374268, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.109375, "rewards/format_reward_func/std": 0.2083333432674408, "step": 400, "step_time": 55.5334142036736 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1712.0, "completions/max_terminated_length": 1712.0, "completions/mean_length": 774.65625, "completions/mean_terminated_length": 774.65625, "completions/min_length": 315.0, "completions/min_terminated_length": 315.0, "entropy": 0.886969679966569, "epoch": 0.22732426303854875, "frac_reward_zero_std": 0.375, "grad_norm": 0.578125, "kl": 0.007826441855286248, "learning_rate": 7.732426303854876e-07, "loss": -0.0255, "num_tokens": 23763430.0, "reward": 0.171875, "reward_std": 0.528690755367279, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 401, "step_time": 59.97146640624851 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1648.0, "completions/max_terminated_length": 1648.0, "completions/mean_length": 803.59375, "completions/mean_terminated_length": 803.59375, "completions/min_length": 266.0, "completions/min_terminated_length": 266.0, "entropy": 0.810698663815856, "epoch": 0.22789115646258504, "frac_reward_zero_std": 0.75, "grad_norm": 0.3125, "kl": 0.012139400845626369, "learning_rate": 7.726757369614512e-07, "loss": -0.0093, "num_tokens": 23827900.0, "reward": 0.1015625, "reward_std": 0.3692515790462494, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 402, "step_time": 58.216640568338335 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1508.0, "completions/max_terminated_length": 1508.0, "completions/mean_length": 700.5, "completions/mean_terminated_length": 700.5, "completions/min_length": 261.0, "completions/min_terminated_length": 261.0, "entropy": 0.7563704717904329, "epoch": 0.22845804988662133, "frac_reward_zero_std": 0.625, "grad_norm": 0.48828125, "kl": 0.01042965950910002, "learning_rate": 7.721088435374149e-07, "loss": 0.0104, "num_tokens": 23881908.0, "reward": 0.1796875, "reward_std": 0.5298036932945251, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 403, "step_time": 52.62817795109004 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2535.0, "completions/max_terminated_length": 2535.0, "completions/mean_length": 730.875, "completions/mean_terminated_length": 730.875, "completions/min_length": 283.0, "completions/min_terminated_length": 283.0, "entropy": 0.6416552513837814, "epoch": 0.2290249433106576, "frac_reward_zero_std": 0.375, "grad_norm": 0.57421875, "kl": 0.012166628352133557, "learning_rate": 7.715419501133786e-07, "loss": 0.0248, "num_tokens": 23940236.0, "reward": 0.40625, "reward_std": 0.7911965847015381, "rewards/accuracy_reward_func/mean": 0.34375, "rewards/accuracy_reward_func/std": 0.7605084180831909, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 404, "step_time": 94.3658363269642 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 982.0, "completions/max_terminated_length": 982.0, "completions/mean_length": 618.953125, "completions/mean_terminated_length": 618.953125, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "entropy": 1.278845801949501, "epoch": 0.22959183673469388, "frac_reward_zero_std": 0.75, "grad_norm": 0.470703125, "kl": 0.015799203421920538, "learning_rate": 7.709750566893424e-07, "loss": 0.0101, "num_tokens": 23992065.0, "reward": 0.0546875, "reward_std": 0.26899558305740356, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 405, "step_time": 34.95195194054395 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1614.0, "completions/max_terminated_length": 1614.0, "completions/mean_length": 777.9375, "completions/mean_terminated_length": 777.9375, "completions/min_length": 314.0, "completions/min_terminated_length": 314.0, "entropy": 0.7842423766851425, "epoch": 0.23015873015873015, "frac_reward_zero_std": 0.75, "grad_norm": 0.341796875, "kl": 0.006712102811434306, "learning_rate": 7.704081632653061e-07, "loss": -0.0187, "num_tokens": 24050733.0, "reward": 0.1953125, "reward_std": 0.6016985774040222, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 406, "step_time": 57.17906874511391 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 809.640625, "completions/mean_terminated_length": 809.640625, "completions/min_length": 372.0, "completions/min_terminated_length": 372.0, "entropy": 0.7890525497496128, "epoch": 0.23072562358276644, "frac_reward_zero_std": 0.875, "grad_norm": 0.23046875, "kl": 0.0075322547781979665, "learning_rate": 7.698412698412699e-07, "loss": 0.0, "num_tokens": 24113734.0, "reward": 0.0546875, "reward_std": 0.15728822350502014, "rewards/accuracy_reward_func/mean": 0.0, "rewards/accuracy_reward_func/std": 0.0, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 407, "step_time": 46.28891652356833 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1545.0, "completions/max_terminated_length": 1545.0, "completions/mean_length": 834.359375, "completions/mean_terminated_length": 834.359375, "completions/min_length": 473.0, "completions/min_terminated_length": 473.0, "entropy": 0.7317435592412949, "epoch": 0.23129251700680273, "frac_reward_zero_std": 0.5, "grad_norm": 0.494140625, "kl": 0.018764967753668316, "learning_rate": 7.692743764172335e-07, "loss": -0.019, "num_tokens": 24176309.0, "reward": 0.296875, "reward_std": 0.6649904847145081, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 408, "step_time": 53.98831983841956 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1609.0, "completions/max_terminated_length": 1609.0, "completions/mean_length": 939.53125, "completions/mean_terminated_length": 939.53125, "completions/min_length": 319.0, "completions/min_terminated_length": 319.0, "entropy": 0.6552373021841049, "epoch": 0.231859410430839, "frac_reward_zero_std": 0.625, "grad_norm": 0.3203125, "kl": 0.008924480978748761, "learning_rate": 7.687074829931972e-07, "loss": 0.0026, "num_tokens": 24247775.0, "reward": 0.15625, "reward_std": 0.4787135720252991, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 409, "step_time": 56.90790192503482 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1414.0, "completions/max_terminated_length": 1414.0, "completions/mean_length": 703.5625, "completions/mean_terminated_length": 703.5625, "completions/min_length": 13.0, "completions/min_terminated_length": 13.0, "entropy": 0.769635409116745, "epoch": 0.23242630385487528, "frac_reward_zero_std": 0.75, "grad_norm": 0.37109375, "kl": 0.007143637565604877, "learning_rate": 7.68140589569161e-07, "loss": 0.0177, "num_tokens": 24306219.0, "reward": 0.125, "reward_std": 0.37796446681022644, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 410, "step_time": 51.87743657454848 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1533.0, "completions/max_terminated_length": 1533.0, "completions/mean_length": 877.0, "completions/mean_terminated_length": 877.0, "completions/min_length": 269.0, "completions/min_terminated_length": 269.0, "entropy": 0.6306070797145367, "epoch": 0.23299319727891157, "frac_reward_zero_std": 0.5, "grad_norm": 0.42578125, "kl": 0.005516977093066089, "learning_rate": 7.675736961451246e-07, "loss": 0.0086, "num_tokens": 24373091.0, "reward": 0.2578125, "reward_std": 0.629891574382782, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 411, "step_time": 54.269387434236705 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1518.0, "completions/max_terminated_length": 1518.0, "completions/mean_length": 737.828125, "completions/mean_terminated_length": 737.828125, "completions/min_length": 270.0, "completions/min_terminated_length": 270.0, "entropy": 0.8014046140015125, "epoch": 0.23356009070294784, "frac_reward_zero_std": 0.625, "grad_norm": 0.447265625, "kl": 0.006993270544626284, "learning_rate": 7.670068027210884e-07, "loss": -0.008, "num_tokens": 24429456.0, "reward": 0.28125, "reward_std": 0.7446901798248291, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 412, "step_time": 53.018550055101514 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1667.0, "completions/max_terminated_length": 1667.0, "completions/mean_length": 828.375, "completions/mean_terminated_length": 828.375, "completions/min_length": 349.0, "completions/min_terminated_length": 349.0, "entropy": 0.7223452497273684, "epoch": 0.23412698412698413, "frac_reward_zero_std": 0.375, "grad_norm": 0.59765625, "kl": 0.0052251661400077865, "learning_rate": 7.664399092970521e-07, "loss": -0.0082, "num_tokens": 24491352.0, "reward": 0.25, "reward_std": 0.549169659614563, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.09375, "rewards/format_reward_func/std": 0.19669894874095917, "step": 413, "step_time": 57.776623263955116 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1483.0, "completions/max_terminated_length": 1483.0, "completions/mean_length": 830.328125, "completions/mean_terminated_length": 830.328125, "completions/min_length": 309.0, "completions/min_terminated_length": 309.0, "entropy": 0.5297314096242189, "epoch": 0.23469387755102042, "frac_reward_zero_std": 0.375, "grad_norm": 0.49609375, "kl": 0.011464409122709185, "learning_rate": 7.658730158730159e-07, "loss": 0.0144, "num_tokens": 24554581.0, "reward": 0.2734375, "reward_std": 0.6662480235099792, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 414, "step_time": 52.02922047767788 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1472.0, "completions/max_terminated_length": 1472.0, "completions/mean_length": 751.3125, "completions/mean_terminated_length": 751.3125, "completions/min_length": 295.0, "completions/min_terminated_length": 295.0, "entropy": 0.9294396489858627, "epoch": 0.23526077097505668, "frac_reward_zero_std": 0.625, "grad_norm": 0.5546875, "kl": 0.006961827792110853, "learning_rate": 7.653061224489795e-07, "loss": 0.0455, "num_tokens": 24613457.0, "reward": 0.2734375, "reward_std": 0.6294977068901062, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 415, "step_time": 51.68498680740595 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1395.0, "completions/max_terminated_length": 1395.0, "completions/mean_length": 708.25, "completions/mean_terminated_length": 708.25, "completions/min_length": 271.0, "completions/min_terminated_length": 271.0, "entropy": 0.7721049040555954, "epoch": 0.23582766439909297, "frac_reward_zero_std": 0.875, "grad_norm": 0.2177734375, "kl": 0.006501360003312584, "learning_rate": 7.647392290249433e-07, "loss": -0.0016, "num_tokens": 24671417.0, "reward": 0.109375, "reward_std": 0.4495919644832611, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 416, "step_time": 49.7948232954368 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1853.0, "completions/max_terminated_length": 1853.0, "completions/mean_length": 753.90625, "completions/mean_terminated_length": 753.90625, "completions/min_length": 203.0, "completions/min_terminated_length": 203.0, "entropy": 0.7060510329902172, "epoch": 0.23639455782312926, "frac_reward_zero_std": 0.75, "grad_norm": 0.296875, "kl": 0.011613724156632088, "learning_rate": 7.64172335600907e-07, "loss": -0.0015, "num_tokens": 24731715.0, "reward": 0.2265625, "reward_std": 0.5904644727706909, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 417, "step_time": 65.67481262143701 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1218.0, "completions/max_terminated_length": 1218.0, "completions/mean_length": 648.90625, "completions/mean_terminated_length": 648.90625, "completions/min_length": 185.0, "completions/min_terminated_length": 185.0, "entropy": 0.9807036016136408, "epoch": 0.23696145124716553, "frac_reward_zero_std": 0.625, "grad_norm": 0.578125, "kl": 0.010876500498852693, "learning_rate": 7.636054421768708e-07, "loss": -0.0033, "num_tokens": 24784693.0, "reward": 0.1328125, "reward_std": 0.4386323094367981, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 418, "step_time": 43.0141840679571 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1202.0, "completions/max_terminated_length": 1202.0, "completions/mean_length": 695.703125, "completions/mean_terminated_length": 695.703125, "completions/min_length": 268.0, "completions/min_terminated_length": 268.0, "entropy": 0.700308034196496, "epoch": 0.23752834467120182, "frac_reward_zero_std": 0.75, "grad_norm": 0.40234375, "kl": 0.011756473148125224, "learning_rate": 7.630385487528344e-07, "loss": -0.0051, "num_tokens": 24839458.0, "reward": 0.15625, "reward_std": 0.4443204402923584, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 419, "step_time": 42.459158713929355 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1592.0, "completions/max_terminated_length": 1592.0, "completions/mean_length": 766.3125, "completions/mean_terminated_length": 766.3125, "completions/min_length": 237.0, "completions/min_terminated_length": 237.0, "entropy": 0.8335917517542839, "epoch": 0.23809523809523808, "frac_reward_zero_std": 0.75, "grad_norm": 0.3515625, "kl": 0.007984273630427197, "learning_rate": 7.624716553287982e-07, "loss": -0.0152, "num_tokens": 24899534.0, "reward": 0.1484375, "reward_std": 0.5092765688896179, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 420, "step_time": 56.33396484144032 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2090.0, "completions/max_terminated_length": 2090.0, "completions/mean_length": 784.921875, "completions/mean_terminated_length": 784.921875, "completions/min_length": 10.0, "completions/min_terminated_length": 10.0, "entropy": 0.8108249437063932, "epoch": 0.23866213151927437, "frac_reward_zero_std": 0.75, "grad_norm": 0.34765625, "kl": 0.009463348469580524, "learning_rate": 7.619047619047618e-07, "loss": 0.0444, "num_tokens": 24960313.0, "reward": 0.09375, "reward_std": 0.36596253514289856, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 421, "step_time": 73.98894651699811 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1189.0, "completions/max_terminated_length": 1189.0, "completions/mean_length": 680.71875, "completions/mean_terminated_length": 680.71875, "completions/min_length": 251.0, "completions/min_terminated_length": 251.0, "entropy": 0.7651320341974497, "epoch": 0.23922902494331066, "frac_reward_zero_std": 0.5, "grad_norm": 0.546875, "kl": 0.009368772603920661, "learning_rate": 7.613378684807256e-07, "loss": -0.0073, "num_tokens": 25013415.0, "reward": 0.3359375, "reward_std": 0.7614454030990601, "rewards/accuracy_reward_func/mean": 0.28125, "rewards/accuracy_reward_func/std": 0.7007648944854736, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 422, "step_time": 42.23136337008327 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 3049.0, "completions/max_terminated_length": 3049.0, "completions/mean_length": 744.796875, "completions/mean_terminated_length": 744.796875, "completions/min_length": 334.0, "completions/min_terminated_length": 334.0, "entropy": 0.8581236302852631, "epoch": 0.23979591836734693, "frac_reward_zero_std": 0.375, "grad_norm": 0.60546875, "kl": 0.016731746160076, "learning_rate": 7.607709750566893e-07, "loss": 0.0067, "num_tokens": 25071746.0, "reward": 0.296875, "reward_std": 0.6649904847145081, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 423, "step_time": 116.03805009834468 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1223.0, "completions/max_terminated_length": 1223.0, "completions/mean_length": 708.03125, "completions/mean_terminated_length": 708.03125, "completions/min_length": 318.0, "completions/min_terminated_length": 318.0, "entropy": 0.8269235882908106, "epoch": 0.24036281179138322, "frac_reward_zero_std": 0.375, "grad_norm": 0.70703125, "kl": 0.009408657337189652, "learning_rate": 7.602040816326531e-07, "loss": 0.0151, "num_tokens": 25127276.0, "reward": 0.421875, "reward_std": 0.7777600884437561, "rewards/accuracy_reward_func/mean": 0.375, "rewards/accuracy_reward_func/std": 0.7867957949638367, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 424, "step_time": 43.01042407378554 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1607.0, "completions/max_terminated_length": 1607.0, "completions/mean_length": 730.46875, "completions/mean_terminated_length": 730.46875, "completions/min_length": 269.0, "completions/min_terminated_length": 269.0, "entropy": 0.7529134526848793, "epoch": 0.2409297052154195, "frac_reward_zero_std": 0.375, "grad_norm": 0.58203125, "kl": 0.00982889297301881, "learning_rate": 7.596371882086168e-07, "loss": 0.003, "num_tokens": 25185394.0, "reward": 0.1953125, "reward_std": 0.5463964939117432, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 425, "step_time": 56.57126920670271 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1328.0, "completions/max_terminated_length": 1328.0, "completions/mean_length": 634.703125, "completions/mean_terminated_length": 634.703125, "completions/min_length": 302.0, "completions/min_terminated_length": 302.0, "entropy": 1.2237685769796371, "epoch": 0.24149659863945577, "frac_reward_zero_std": 0.625, "grad_norm": 0.58203125, "kl": 0.008560267873690464, "learning_rate": 7.590702947845806e-07, "loss": -0.0192, "num_tokens": 25239415.0, "reward": 0.265625, "reward_std": 0.6544642448425293, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 426, "step_time": 49.03863408975303 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1526.0, "completions/max_terminated_length": 1526.0, "completions/mean_length": 710.359375, "completions/mean_terminated_length": 710.359375, "completions/min_length": 244.0, "completions/min_terminated_length": 244.0, "entropy": 0.8540695123374462, "epoch": 0.24206349206349206, "frac_reward_zero_std": 0.75, "grad_norm": 0.30078125, "kl": 0.00960883658262901, "learning_rate": 7.585034013605441e-07, "loss": 0.003, "num_tokens": 25296710.0, "reward": 0.234375, "reward_std": 0.6898858547210693, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 427, "step_time": 54.59573978651315 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 796.84375, "completions/mean_terminated_length": 796.84375, "completions/min_length": 250.0, "completions/min_terminated_length": 250.0, "entropy": 0.6341959312558174, "epoch": 0.24263038548752835, "frac_reward_zero_std": 0.625, "grad_norm": 0.42578125, "kl": 0.00874024344375357, "learning_rate": 7.579365079365078e-07, "loss": 0.0029, "num_tokens": 25358604.0, "reward": 0.28125, "reward_std": 0.6291528940200806, "rewards/accuracy_reward_func/mean": 0.21875, "rewards/accuracy_reward_func/std": 0.6291528940200806, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 428, "step_time": 49.703380913473666 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1209.0, "completions/max_terminated_length": 1209.0, "completions/mean_length": 720.515625, "completions/mean_terminated_length": 720.515625, "completions/min_length": 319.0, "completions/min_terminated_length": 319.0, "entropy": 0.8529839962720871, "epoch": 0.24319727891156462, "frac_reward_zero_std": 0.625, "grad_norm": 0.48828125, "kl": 0.010199644399108365, "learning_rate": 7.573696145124716e-07, "loss": 0.0177, "num_tokens": 25414237.0, "reward": 0.171875, "reward_std": 0.49776285886764526, "rewards/accuracy_reward_func/mean": 0.125, "rewards/accuracy_reward_func/std": 0.48795005679130554, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 429, "step_time": 42.44065989367664 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2672.0, "completions/max_terminated_length": 2672.0, "completions/mean_length": 770.734375, "completions/mean_terminated_length": 770.734375, "completions/min_length": 374.0, "completions/min_terminated_length": 374.0, "entropy": 0.9049090035259724, "epoch": 0.2437641723356009, "frac_reward_zero_std": 0.5, "grad_norm": 0.51171875, "kl": 0.006098392186686397, "learning_rate": 7.568027210884353e-07, "loss": 0.0221, "num_tokens": 25476180.0, "reward": 0.1875, "reward_std": 0.5455447435379028, "rewards/accuracy_reward_func/mean": 0.15625, "rewards/accuracy_reward_func/std": 0.5409794449806213, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 430, "step_time": 99.39775457978249 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1474.0, "completions/max_terminated_length": 1474.0, "completions/mean_length": 709.59375, "completions/mean_terminated_length": 709.59375, "completions/min_length": 265.0, "completions/min_terminated_length": 265.0, "entropy": 0.8665502294898033, "epoch": 0.2443310657596372, "frac_reward_zero_std": 0.375, "grad_norm": 0.625, "kl": 0.017350808397168294, "learning_rate": 7.562358276643991e-07, "loss": -0.0523, "num_tokens": 25531130.0, "reward": 0.5546875, "reward_std": 0.8550373315811157, "rewards/accuracy_reward_func/mean": 0.5, "rewards/accuracy_reward_func/std": 0.8728715777397156, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 431, "step_time": 51.904258251190186 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1440.0, "completions/max_terminated_length": 1440.0, "completions/mean_length": 713.40625, "completions/mean_terminated_length": 713.40625, "completions/min_length": 276.0, "completions/min_terminated_length": 276.0, "entropy": 0.8070783540606499, "epoch": 0.24489795918367346, "frac_reward_zero_std": 0.125, "grad_norm": 0.7265625, "kl": 0.010728749562986195, "learning_rate": 7.556689342403628e-07, "loss": -0.0235, "num_tokens": 25586036.0, "reward": 0.2265625, "reward_std": 0.5904644727706909, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0390625, "rewards/format_reward_func/std": 0.13524486124515533, "step": 432, "step_time": 50.71470108628273 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2009.0, "completions/max_terminated_length": 2009.0, "completions/mean_length": 799.0, "completions/mean_terminated_length": 799.0, "completions/min_length": 239.0, "completions/min_terminated_length": 239.0, "entropy": 0.5964866243302822, "epoch": 0.24546485260770975, "frac_reward_zero_std": 0.5, "grad_norm": 0.51171875, "kl": 0.0044653531949734315, "learning_rate": 7.551020408163265e-07, "loss": -0.0234, "num_tokens": 25648468.0, "reward": 0.453125, "reward_std": 0.8199823498725891, "rewards/accuracy_reward_func/mean": 0.40625, "rewards/accuracy_reward_func/std": 0.8110105991363525, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 433, "step_time": 73.19168344326317 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1266.0, "completions/max_terminated_length": 1266.0, "completions/mean_length": 702.15625, "completions/mean_terminated_length": 702.15625, "completions/min_length": 286.0, "completions/min_terminated_length": 286.0, "entropy": 0.996532890945673, "epoch": 0.24603174603174602, "frac_reward_zero_std": 0.625, "grad_norm": 0.498046875, "kl": 0.010547072102781385, "learning_rate": 7.545351473922901e-07, "loss": 0.0006, "num_tokens": 25712806.0, "reward": 0.2109375, "reward_std": 0.6159563660621643, "rewards/accuracy_reward_func/mean": 0.1875, "rewards/accuracy_reward_func/std": 0.5875696539878845, "rewards/format_reward_func/mean": 0.0234375, "rewards/format_reward_func/std": 0.10652101784944534, "step": 434, "step_time": 52.20338438823819 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1473.0, "completions/max_terminated_length": 1473.0, "completions/mean_length": 796.34375, "completions/mean_terminated_length": 796.34375, "completions/min_length": 311.0, "completions/min_terminated_length": 311.0, "entropy": 0.8812602311372757, "epoch": 0.2465986394557823, "frac_reward_zero_std": 0.5, "grad_norm": 0.546875, "kl": 0.0062492226279573515, "learning_rate": 7.539682539682539e-07, "loss": 0.0032, "num_tokens": 25773204.0, "reward": 0.4296875, "reward_std": 0.7759642601013184, "rewards/accuracy_reward_func/mean": 0.375, "rewards/accuracy_reward_func/std": 0.7867957949638367, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 435, "step_time": 54.18101462442428 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 4096.0, "completions/max_terminated_length": 1388.0, "completions/mean_length": 809.109375, "completions/mean_terminated_length": 756.9365234375, "completions/min_length": 225.0, "completions/min_terminated_length": 225.0, "entropy": 0.8817069754004478, "epoch": 0.2471655328798186, "frac_reward_zero_std": 0.5, "grad_norm": 0.462890625, "kl": 0.010766757986857556, "learning_rate": 7.534013605442176e-07, "loss": 0.0167, "num_tokens": 25834115.0, "reward": 0.140625, "reward_std": 0.4406772553920746, "rewards/accuracy_reward_func/mean": 0.09375, "rewards/accuracy_reward_func/std": 0.42608407139778137, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 436, "step_time": 170.6649263864383 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1468.0, "completions/max_terminated_length": 1468.0, "completions/mean_length": 763.84375, "completions/mean_terminated_length": 763.84375, "completions/min_length": 320.0, "completions/min_terminated_length": 320.0, "entropy": 0.7432105969637632, "epoch": 0.24773242630385486, "frac_reward_zero_std": 0.625, "grad_norm": 0.396484375, "kl": 0.006353874458000064, "learning_rate": 7.528344671201814e-07, "loss": -0.0116, "num_tokens": 25892729.0, "reward": 0.1171875, "reward_std": 0.37524792551994324, "rewards/accuracy_reward_func/mean": 0.0625, "rewards/accuracy_reward_func/std": 0.35073620080947876, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 437, "step_time": 52.915428700856864 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1375.0, "completions/max_terminated_length": 1375.0, "completions/mean_length": 864.5625, "completions/mean_terminated_length": 864.5625, "completions/min_length": 375.0, "completions/min_terminated_length": 375.0, "entropy": 0.6446567680686712, "epoch": 0.24829931972789115, "frac_reward_zero_std": 0.5, "grad_norm": 0.435546875, "kl": 0.005845273204613477, "learning_rate": 7.522675736961451e-07, "loss": 0.0277, "num_tokens": 25957949.0, "reward": 0.4375, "reward_std": 0.8614310622215271, "rewards/accuracy_reward_func/mean": 0.40625, "rewards/accuracy_reward_func/std": 0.8110105991363525, "rewards/format_reward_func/mean": 0.03125, "rewards/format_reward_func/std": 0.12198751419782639, "step": 438, "step_time": 47.994716121815145 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1348.0, "completions/max_terminated_length": 1348.0, "completions/mean_length": 655.546875, "completions/mean_terminated_length": 655.546875, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "entropy": 0.6887895595282316, "epoch": 0.24886621315192745, "frac_reward_zero_std": 0.5, "grad_norm": 0.5078125, "kl": 0.013109318068018183, "learning_rate": 7.517006802721088e-07, "loss": 0.0194, "num_tokens": 26011384.0, "reward": 0.3125, "reward_std": 0.6636837720870972, "rewards/accuracy_reward_func/mean": 0.25, "rewards/accuracy_reward_func/std": 0.6666666865348816, "rewards/format_reward_func/mean": 0.0625, "rewards/format_reward_func/std": 0.1666666716337204, "step": 439, "step_time": 47.03840684890747 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1645.0, "completions/max_terminated_length": 1645.0, "completions/mean_length": 670.484375, "completions/mean_terminated_length": 670.484375, "completions/min_length": 373.0, "completions/min_terminated_length": 373.0, "entropy": 0.866967786103487, "epoch": 0.2494331065759637, "frac_reward_zero_std": 0.625, "grad_norm": 0.478515625, "kl": 0.008332332065037917, "learning_rate": 7.511337868480725e-07, "loss": -0.0031, "num_tokens": 26064231.0, "reward": 0.3359375, "reward_std": 0.6961041688919067, "rewards/accuracy_reward_func/mean": 0.28125, "rewards/accuracy_reward_func/std": 0.7007648944854736, "rewards/format_reward_func/mean": 0.0546875, "rewards/format_reward_func/std": 0.15728822350502014, "step": 440, "step_time": 57.855160099454224 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 2774.0, "completions/max_terminated_length": 2774.0, "completions/mean_length": 838.234375, "completions/mean_terminated_length": 838.234375, "completions/min_length": 273.0, "completions/min_terminated_length": 273.0, "entropy": 0.8465369511395693, "epoch": 0.25, "frac_reward_zero_std": 0.75, "grad_norm": 0.357421875, "kl": 0.010944681576802395, "learning_rate": 7.505668934240363e-07, "loss": -0.0077, "num_tokens": 26128918.0, "reward": 0.078125, "reward_std": 0.2847827076911926, "rewards/accuracy_reward_func/mean": 0.03125, "rewards/accuracy_reward_func/std": 0.25, "rewards/format_reward_func/mean": 0.046875, "rewards/format_reward_func/std": 0.14689241349697113, "step": 441, "step_time": 103.54677613079548 } ], "logging_steps": 1, "max_steps": 1764, "num_input_tokens_seen": 26128918, "num_train_epochs": 1, "save_steps": 500, "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": 4, "trial_name": null, "trial_params": null }