{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.05714285714285714, "eval_steps": 500, "global_step": 50, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.671875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1734.0, "completions/mean_length": 1702.03125, "completions/mean_terminated_length": 993.6190795898438, "completions/min_length": 483.0, "completions/min_terminated_length": 483.0, "epoch": 0.001142857142857143, "frac_reward_zero_std": 0.0, "grad_norm": 0.25444135069847107, "learning_rate": 0.0, "loss": -0.0, "num_tokens": 118418.0, "reward": 0.17899775505065918, "reward_std": 0.7650213241577148, "rewards/cosine_scaled_reward/mean": -0.09800112992525101, "rewards/cosine_scaled_reward/std": 0.37953105568885803, "rewards/format_reward/mean": 0.375, "rewards/format_reward/std": 0.48795005679130554, "step": 1 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.71875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1894.0, "completions/mean_length": 1738.90625, "completions/mean_terminated_length": 949.0, "completions/min_length": 435.0, "completions/min_terminated_length": 435.0, "epoch": 0.002285714285714286, "frac_reward_zero_std": 0.0, "grad_norm": 0.24364067614078522, "learning_rate": 5e-08, "loss": -0.0, "num_tokens": 239748.0, "reward": 0.3848632574081421, "reward_std": 0.9111153483390808, "rewards/cosine_scaled_reward/mean": 0.020556632429361343, "rewards/cosine_scaled_reward/std": 0.4492928683757782, "rewards/format_reward/mean": 0.34375, "rewards/format_reward/std": 0.4787135720252991, "step": 2 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.890625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1405.0, "completions/mean_length": 1930.609375, "completions/mean_terminated_length": 974.71435546875, "completions/min_length": 477.0, "completions/min_terminated_length": 477.0, "epoch": 0.0034285714285714284, "frac_reward_zero_std": 0.0, "grad_norm": 0.27834266424179077, "learning_rate": 1e-07, "loss": -0.0, "num_tokens": 373779.0, "reward": -0.3227587938308716, "reward_std": 0.45940712094306946, "rewards/cosine_scaled_reward/mean": -0.2160668969154358, "rewards/cosine_scaled_reward/std": 0.21890601515769958, "rewards/format_reward/mean": 0.109375, "rewards/format_reward/std": 0.3145764470100403, "step": 3 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.515625, "completions/max_length": 2048.0, "completions/max_terminated_length": 2039.0, "completions/mean_length": 1596.75, "completions/mean_terminated_length": 1116.3870849609375, "completions/min_length": 474.0, "completions/min_terminated_length": 474.0, "epoch": 0.004571428571428572, "frac_reward_zero_std": 0.0, "grad_norm": 0.2799243628978729, "learning_rate": 1.5e-07, "loss": -0.0, "num_tokens": 485779.0, "reward": 0.27003082633018494, "reward_std": 0.7608597874641418, "rewards/cosine_scaled_reward/mean": -0.11498458683490753, "rewards/cosine_scaled_reward/std": 0.36645373702049255, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.5039526224136353, "step": 4 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.96875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1755.0, "completions/mean_length": 2035.46875, "completions/mean_terminated_length": 1647.0, "completions/min_length": 1539.0, "completions/min_terminated_length": 1539.0, "epoch": 0.005714285714285714, "frac_reward_zero_std": 0.0, "grad_norm": 0.24311119318008423, "learning_rate": 2e-07, "loss": -0.0, "num_tokens": 626865.0, "reward": -0.4839385151863098, "reward_std": 0.34498828649520874, "rewards/cosine_scaled_reward/mean": -0.2732192277908325, "rewards/cosine_scaled_reward/std": 0.18402352929115295, "rewards/format_reward/mean": 0.0625, "rewards/format_reward/std": 0.24397502839565277, "step": 5 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.859375, "completions/max_length": 2048.0, "completions/max_terminated_length": 1862.0, "completions/mean_length": 1884.109375, "completions/mean_terminated_length": 882.5555419921875, "completions/min_length": 524.0, "completions/min_terminated_length": 524.0, "epoch": 0.006857142857142857, "frac_reward_zero_std": 0.0, "grad_norm": 0.2741600275039673, "learning_rate": 2.5e-07, "loss": -0.0, "num_tokens": 759096.0, "reward": -0.2049689143896103, "reward_std": 0.639178991317749, "rewards/cosine_scaled_reward/mean": -0.18060946464538574, "rewards/cosine_scaled_reward/std": 0.2599981129169464, "rewards/format_reward/mean": 0.15625, "rewards/format_reward/std": 0.36596253514289856, "step": 6 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1795.0, "completions/mean_length": 1959.84375, "completions/mean_terminated_length": 1342.75, "completions/min_length": 974.0, "completions/min_terminated_length": 974.0, "epoch": 0.008, "frac_reward_zero_std": 0.0, "grad_norm": 0.21986258029937744, "learning_rate": 3e-07, "loss": -0.0, "num_tokens": 894934.0, "reward": -0.11210991442203522, "reward_std": 0.6349427103996277, "rewards/cosine_scaled_reward/mean": -0.14199243485927582, "rewards/cosine_scaled_reward/std": 0.3749195337295532, "rewards/format_reward/mean": 0.171875, "rewards/format_reward/std": 0.38025420904159546, "step": 7 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.71875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1895.0, "completions/mean_length": 1717.78125, "completions/mean_terminated_length": 873.888916015625, "completions/min_length": 342.0, "completions/min_terminated_length": 342.0, "epoch": 0.009142857142857144, "frac_reward_zero_std": 0.0, "grad_norm": 0.23102505505084991, "learning_rate": 3.5e-07, "loss": -0.0, "num_tokens": 1015288.0, "reward": 0.12653985619544983, "reward_std": 0.4742490351200104, "rewards/cosine_scaled_reward/mean": -0.09298005700111389, "rewards/cosine_scaled_reward/std": 0.39157670736312866, "rewards/format_reward/mean": 0.3125, "rewards/format_reward/std": 0.467176616191864, "step": 8 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.828125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1815.0, "completions/mean_length": 1928.53125, "completions/mean_terminated_length": 1352.9091796875, "completions/min_length": 999.0, "completions/min_terminated_length": 999.0, "epoch": 0.010285714285714285, "frac_reward_zero_std": 0.0, "grad_norm": 0.2600339353084564, "learning_rate": 4e-07, "loss": -0.0, "num_tokens": 1150170.0, "reward": -0.14216071367263794, "reward_std": 0.702994704246521, "rewards/cosine_scaled_reward/mean": -0.17264285683631897, "rewards/cosine_scaled_reward/std": 0.33145979046821594, "rewards/format_reward/mean": 0.203125, "rewards/format_reward/std": 0.40550529956817627, "step": 9 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.71875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1432.0, "completions/mean_length": 1699.84375, "completions/mean_terminated_length": 810.1111450195312, "completions/min_length": 337.0, "completions/min_terminated_length": 337.0, "epoch": 0.011428571428571429, "frac_reward_zero_std": 0.0, "grad_norm": 0.2749471366405487, "learning_rate": 4.5e-07, "loss": -0.0, "num_tokens": 1269792.0, "reward": -0.13922849297523499, "reward_std": 0.4937349855899811, "rewards/cosine_scaled_reward/mean": -0.2102392464876175, "rewards/cosine_scaled_reward/std": 0.30274781584739685, "rewards/format_reward/mean": 0.28125, "rewards/format_reward/std": 0.4531635046005249, "step": 10 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.9375, "completions/max_length": 2048.0, "completions/max_terminated_length": 1614.0, "completions/mean_length": 1994.453125, "completions/mean_terminated_length": 1191.25, "completions/min_length": 916.0, "completions/min_terminated_length": 916.0, "epoch": 0.012571428571428572, "frac_reward_zero_std": 0.0, "grad_norm": 0.2305486500263214, "learning_rate": 5e-07, "loss": 0.0, "num_tokens": 1409109.0, "reward": -0.39525067806243896, "reward_std": 0.3650783896446228, "rewards/cosine_scaled_reward/mean": -0.2288753092288971, "rewards/cosine_scaled_reward/std": 0.22182811796665192, "rewards/format_reward/mean": 0.0625, "rewards/format_reward/std": 0.24397502839565277, "step": 11 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.609375, "completions/max_length": 2048.0, "completions/max_terminated_length": 2024.0, "completions/mean_length": 1706.6875, "completions/mean_terminated_length": 1174.239990234375, "completions/min_length": 319.0, "completions/min_terminated_length": 319.0, "epoch": 0.013714285714285714, "frac_reward_zero_std": 0.0, "grad_norm": 0.2918066382408142, "learning_rate": 5.5e-07, "loss": 0.0, "num_tokens": 1529281.0, "reward": 0.08787664026021957, "reward_std": 0.7579531073570251, "rewards/cosine_scaled_reward/mean": -0.18262416124343872, "rewards/cosine_scaled_reward/std": 0.37901216745376587, "rewards/format_reward/mean": 0.453125, "rewards/format_reward/std": 0.501733124256134, "step": 12 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.765625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1943.0, "completions/mean_length": 1820.828125, "completions/mean_terminated_length": 1078.7333984375, "completions/min_length": 527.0, "completions/min_terminated_length": 527.0, "epoch": 0.014857142857142857, "frac_reward_zero_std": 0.0, "grad_norm": 0.27849823236465454, "learning_rate": 6e-07, "loss": -0.0, "num_tokens": 1656854.0, "reward": 0.03077489137649536, "reward_std": 0.6479229927062988, "rewards/cosine_scaled_reward/mean": -0.12523755431175232, "rewards/cosine_scaled_reward/std": 0.34234777092933655, "rewards/format_reward/mean": 0.28125, "rewards/format_reward/std": 0.4531635046005249, "step": 13 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.6875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1972.0, "completions/mean_length": 1772.296875, "completions/mean_terminated_length": 1165.75, "completions/min_length": 605.0, "completions/min_terminated_length": 605.0, "epoch": 0.016, "frac_reward_zero_std": 0.0, "grad_norm": 0.25037428736686707, "learning_rate": 6.5e-07, "loss": 0.0, "num_tokens": 1780889.0, "reward": 0.3261271119117737, "reward_std": 0.6276673078536987, "rewards/cosine_scaled_reward/mean": -0.008811453357338905, "rewards/cosine_scaled_reward/std": 0.46767035126686096, "rewards/format_reward/mean": 0.34375, "rewards/format_reward/std": 0.4787135720252991, "step": 14 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.703125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1831.0, "completions/mean_length": 1715.5625, "completions/mean_terminated_length": 928.2105102539062, "completions/min_length": 413.0, "completions/min_terminated_length": 413.0, "epoch": 0.017142857142857144, "frac_reward_zero_std": 0.0, "grad_norm": 0.26902371644973755, "learning_rate": 7e-07, "loss": -0.0, "num_tokens": 1901605.0, "reward": 0.3007117211818695, "reward_std": 0.3918319642543793, "rewards/cosine_scaled_reward/mean": -0.005894124507904053, "rewards/cosine_scaled_reward/std": 0.4677385091781616, "rewards/format_reward/mean": 0.3125, "rewards/format_reward/std": 0.467176616191864, "step": 15 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.984375, "completions/max_length": 2048.0, "completions/max_terminated_length": 757.0, "completions/mean_length": 2027.828125, "completions/mean_terminated_length": 757.0, "completions/min_length": 757.0, "completions/min_terminated_length": 757.0, "epoch": 0.018285714285714287, "frac_reward_zero_std": 0.0, "grad_norm": 0.25064370036125183, "learning_rate": 7.5e-07, "loss": -0.0, "num_tokens": 2041826.0, "reward": -0.499896764755249, "reward_std": 0.34189552068710327, "rewards/cosine_scaled_reward/mean": -0.2577608823776245, "rewards/cosine_scaled_reward/std": 0.18115636706352234, "rewards/format_reward/mean": 0.015625, "rewards/format_reward/std": 0.125, "step": 16 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.546875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1816.0, "completions/mean_length": 1530.796875, "completions/mean_terminated_length": 906.586181640625, "completions/min_length": 378.0, "completions/min_terminated_length": 378.0, "epoch": 0.019428571428571427, "frac_reward_zero_std": 0.0, "grad_norm": 0.3018316924571991, "learning_rate": 8e-07, "loss": -0.0, "num_tokens": 2150317.0, "reward": 0.23110359907150269, "reward_std": 0.6260336637496948, "rewards/cosine_scaled_reward/mean": -0.12663568556308746, "rewards/cosine_scaled_reward/std": 0.39377179741859436, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.5037065148353577, "step": 17 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.796875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1689.0, "completions/mean_length": 1813.671875, "completions/mean_terminated_length": 894.3846435546875, "completions/min_length": 505.0, "completions/min_terminated_length": 505.0, "epoch": 0.02057142857142857, "frac_reward_zero_std": 0.0, "grad_norm": 0.23236438632011414, "learning_rate": 8.499999999999999e-07, "loss": 0.0, "num_tokens": 2276768.0, "reward": -0.10029121488332748, "reward_std": 0.7172800302505493, "rewards/cosine_scaled_reward/mean": -0.18295811116695404, "rewards/cosine_scaled_reward/std": 0.3038564622402191, "rewards/format_reward/mean": 0.265625, "rewards/format_reward/std": 0.44515693187713623, "step": 18 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.78125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1697.0, "completions/mean_length": 1843.15625, "completions/mean_terminated_length": 1111.571533203125, "completions/min_length": 484.0, "completions/min_terminated_length": 484.0, "epoch": 0.021714285714285714, "frac_reward_zero_std": 0.0, "grad_norm": 0.2313074916601181, "learning_rate": 9e-07, "loss": -0.0, "num_tokens": 2405986.0, "reward": 0.09310440719127655, "reward_std": 0.7020131349563599, "rewards/cosine_scaled_reward/mean": -0.08626029640436172, "rewards/cosine_scaled_reward/std": 0.44063708186149597, "rewards/format_reward/mean": 0.265625, "rewards/format_reward/std": 0.44515693187713623, "step": 19 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.578125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1676.0, "completions/mean_length": 1523.03125, "completions/mean_terminated_length": 803.629638671875, "completions/min_length": 395.0, "completions/min_terminated_length": 395.0, "epoch": 0.022857142857142857, "frac_reward_zero_std": 0.0, "grad_norm": 0.295642226934433, "learning_rate": 9.499999999999999e-07, "loss": -0.0, "num_tokens": 2514812.0, "reward": 0.3644811511039734, "reward_std": 0.7943294048309326, "rewards/cosine_scaled_reward/mean": -0.03650941699743271, "rewards/cosine_scaled_reward/std": 0.44610291719436646, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.5, "step": 20 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.703125, "completions/max_length": 2048.0, "completions/max_terminated_length": 2002.0, "completions/mean_length": 1793.328125, "completions/mean_terminated_length": 1190.157958984375, "completions/min_length": 455.0, "completions/min_terminated_length": 455.0, "epoch": 0.024, "frac_reward_zero_std": 0.0, "grad_norm": 0.2961376905441284, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 2640393.0, "reward": 0.06134350597858429, "reward_std": 0.6498202085494995, "rewards/cosine_scaled_reward/mean": -0.14120325446128845, "rewards/cosine_scaled_reward/std": 0.3548509478569031, "rewards/format_reward/mean": 0.34375, "rewards/format_reward/std": 0.4787135720252991, "step": 21 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.4375, "completions/max_length": 2048.0, "completions/max_terminated_length": 1925.0, "completions/mean_length": 1386.75, "completions/mean_terminated_length": 872.4444580078125, "completions/min_length": 271.0, "completions/min_terminated_length": 271.0, "epoch": 0.025142857142857144, "frac_reward_zero_std": 0.0, "grad_norm": 0.34918317198753357, "learning_rate": 9.99931462820376e-07, "loss": 0.0, "num_tokens": 2738161.0, "reward": 0.5064569711685181, "reward_std": 0.7104054689407349, "rewards/cosine_scaled_reward/mean": -0.035834040492773056, "rewards/cosine_scaled_reward/std": 0.4265843331813812, "rewards/format_reward/mean": 0.578125, "rewards/format_reward/std": 0.49776285886764526, "step": 22 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.59375, "completions/max_length": 2048.0, "completions/max_terminated_length": 1986.0, "completions/mean_length": 1656.59375, "completions/mean_terminated_length": 1084.5384521484375, "completions/min_length": 364.0, "completions/min_terminated_length": 364.0, "epoch": 0.026285714285714287, "frac_reward_zero_std": 0.0, "grad_norm": 0.26697081327438354, "learning_rate": 9.997258721585931e-07, "loss": -0.0, "num_tokens": 2854975.0, "reward": 0.2737857699394226, "reward_std": 0.6956006288528442, "rewards/cosine_scaled_reward/mean": -0.0896696150302887, "rewards/cosine_scaled_reward/std": 0.3913433253765106, "rewards/format_reward/mean": 0.453125, "rewards/format_reward/std": 0.501733124256134, "step": 23 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.6875, "completions/max_length": 2048.0, "completions/max_terminated_length": 2004.0, "completions/mean_length": 1790.421875, "completions/mean_terminated_length": 1223.75, "completions/min_length": 421.0, "completions/min_terminated_length": 421.0, "epoch": 0.027428571428571427, "frac_reward_zero_std": 0.0, "grad_norm": 0.24950818717479706, "learning_rate": 9.993832906395582e-07, "loss": -0.0, "num_tokens": 2980490.0, "reward": -0.08990197628736496, "reward_std": 0.7724581956863403, "rewards/cosine_scaled_reward/mean": -0.21682599186897278, "rewards/cosine_scaled_reward/std": 0.35711658000946045, "rewards/format_reward/mean": 0.34375, "rewards/format_reward/std": 0.4787135720252991, "step": 24 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.703125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1985.0, "completions/mean_length": 1703.953125, "completions/mean_terminated_length": 889.1052856445312, "completions/min_length": 427.0, "completions/min_terminated_length": 427.0, "epoch": 0.02857142857142857, "frac_reward_zero_std": 0.0, "grad_norm": 0.28078693151474, "learning_rate": 9.989038226169207e-07, "loss": 0.0, "num_tokens": 3099839.0, "reward": -0.12643180787563324, "reward_std": 0.6687923669815063, "rewards/cosine_scaled_reward/mean": -0.21946589648723602, "rewards/cosine_scaled_reward/std": 0.30431970953941345, "rewards/format_reward/mean": 0.3125, "rewards/format_reward/std": 0.467176616191864, "step": 25 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.75, "completions/max_length": 2048.0, "completions/max_terminated_length": 1997.0, "completions/mean_length": 1938.078125, "completions/mean_terminated_length": 1608.3125, "completions/min_length": 1087.0, "completions/min_terminated_length": 1087.0, "epoch": 0.029714285714285714, "frac_reward_zero_std": 0.0, "grad_norm": 0.21486051380634308, "learning_rate": 9.982876141412855e-07, "loss": 0.0, "num_tokens": 3234508.0, "reward": 0.05503671616315842, "reward_std": 0.6532000303268433, "rewards/cosine_scaled_reward/mean": -0.1287316530942917, "rewards/cosine_scaled_reward/std": 0.36068078875541687, "rewards/format_reward/mean": 0.3125, "rewards/format_reward/std": 0.467176616191864, "step": 26 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.765625, "completions/max_length": 2048.0, "completions/max_terminated_length": 2037.0, "completions/mean_length": 1896.375, "completions/mean_terminated_length": 1401.0667724609375, "completions/min_length": 568.0, "completions/min_terminated_length": 568.0, "epoch": 0.030857142857142857, "frac_reward_zero_std": 0.0, "grad_norm": 0.2675936222076416, "learning_rate": 9.975348529157229e-07, "loss": -0.0, "num_tokens": 3366164.0, "reward": -0.02987336739897728, "reward_std": 0.5919089913368225, "rewards/cosine_scaled_reward/mean": -0.1633741855621338, "rewards/cosine_scaled_reward/std": 0.3508918881416321, "rewards/format_reward/mean": 0.296875, "rewards/format_reward/std": 0.4604927599430084, "step": 27 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.8125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1951.0, "completions/mean_length": 1832.96875, "completions/mean_terminated_length": 901.1666870117188, "completions/min_length": 450.0, "completions/min_terminated_length": 450.0, "epoch": 0.032, "frac_reward_zero_std": 0.0, "grad_norm": 0.2518174946308136, "learning_rate": 9.96645768238595e-07, "loss": 0.0, "num_tokens": 3493810.0, "reward": 0.08577289432287216, "reward_std": 0.6993601322174072, "rewards/cosine_scaled_reward/mean": -0.08211354911327362, "rewards/cosine_scaled_reward/std": 0.45168522000312805, "rewards/format_reward/mean": 0.25, "rewards/format_reward/std": 0.4364357888698578, "step": 28 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.859375, "completions/max_length": 2048.0, "completions/max_terminated_length": 1779.0, "completions/mean_length": 1921.1875, "completions/mean_terminated_length": 1146.2222900390625, "completions/min_length": 710.0, "completions/min_terminated_length": 710.0, "epoch": 0.03314285714285714, "frac_reward_zero_std": 0.0, "grad_norm": 0.25027790665626526, "learning_rate": 9.956206309337066e-07, "loss": -0.0, "num_tokens": 3627238.0, "reward": -0.3098237216472626, "reward_std": 0.4339829087257385, "rewards/cosine_scaled_reward/mean": -0.2330368608236313, "rewards/cosine_scaled_reward/std": 0.17332859337329865, "rewards/format_reward/mean": 0.15625, "rewards/format_reward/std": 0.36596253514289856, "step": 29 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.71875, "completions/max_length": 2048.0, "completions/max_terminated_length": 2048.0, "completions/mean_length": 1891.109375, "completions/mean_terminated_length": 1490.1666259765625, "completions/min_length": 401.0, "completions/min_terminated_length": 401.0, "epoch": 0.03428571428571429, "frac_reward_zero_std": 0.0, "grad_norm": 0.24189673364162445, "learning_rate": 9.944597532678119e-07, "loss": 0.0, "num_tokens": 3758805.0, "reward": -0.08874380588531494, "reward_std": 0.5923835635185242, "rewards/cosine_scaled_reward/mean": -0.18499691784381866, "rewards/cosine_scaled_reward/std": 0.27955111861228943, "rewards/format_reward/mean": 0.28125, "rewards/format_reward/std": 0.4531635046005249, "step": 30 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.8125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1467.0, "completions/mean_length": 1818.8125, "completions/mean_terminated_length": 825.6666870117188, "completions/min_length": 444.0, "completions/min_terminated_length": 444.0, "epoch": 0.03542857142857143, "frac_reward_zero_std": 0.0, "grad_norm": 0.24893531203269958, "learning_rate": 9.931634888554935e-07, "loss": 0.0, "num_tokens": 3885705.0, "reward": -0.18628405034542084, "reward_std": 0.5522075891494751, "rewards/cosine_scaled_reward/mean": -0.20251703262329102, "rewards/cosine_scaled_reward/std": 0.37591472268104553, "rewards/format_reward/mean": 0.21875, "rewards/format_reward/std": 0.4166666865348816, "step": 31 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.796875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1867.0, "completions/mean_length": 1878.140625, "completions/mean_terminated_length": 1211.769287109375, "completions/min_length": 654.0, "completions/min_terminated_length": 654.0, "epoch": 0.036571428571428574, "frac_reward_zero_std": 0.0, "grad_norm": 0.25341352820396423, "learning_rate": 9.917322325514487e-07, "loss": 0.0, "num_tokens": 4016258.0, "reward": -0.14861394464969635, "reward_std": 0.5451517105102539, "rewards/cosine_scaled_reward/mean": -0.19149449467658997, "rewards/cosine_scaled_reward/std": 0.3489256203174591, "rewards/format_reward/mean": 0.234375, "rewards/format_reward/std": 0.42695629596710205, "step": 32 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.890625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1823.0, "completions/mean_length": 1969.03125, "completions/mean_terminated_length": 1326.0, "completions/min_length": 998.0, "completions/min_terminated_length": 998.0, "epoch": 0.037714285714285714, "frac_reward_zero_std": 0.0, "grad_norm": 0.23557375371456146, "learning_rate": 9.901664203302124e-07, "loss": -0.0, "num_tokens": 4153492.0, "reward": -0.3634287118911743, "reward_std": 0.5462046265602112, "rewards/cosine_scaled_reward/mean": -0.25983935594558716, "rewards/cosine_scaled_reward/std": 0.3271723687648773, "rewards/format_reward/mean": 0.15625, "rewards/format_reward/std": 0.36596253514289856, "step": 33 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.53125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1998.0, "completions/mean_length": 1568.296875, "completions/mean_terminated_length": 1024.6334228515625, "completions/min_length": 506.0, "completions/min_terminated_length": 506.0, "epoch": 0.038857142857142854, "frac_reward_zero_std": 0.0, "grad_norm": 0.2892495095729828, "learning_rate": 9.88466529153356e-07, "loss": 0.0, "num_tokens": 4263415.0, "reward": 0.575156569480896, "reward_std": 0.8866004347801208, "rewards/cosine_scaled_reward/mean": 0.045390784740448, "rewards/cosine_scaled_reward/std": 0.5505619645118713, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.5037065148353577, "step": 34 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.84375, "completions/max_length": 2048.0, "completions/max_terminated_length": 1834.0, "completions/mean_length": 1830.65625, "completions/mean_terminated_length": 657.0, "completions/min_length": 371.0, "completions/min_terminated_length": 371.0, "epoch": 0.04, "frac_reward_zero_std": 0.0, "grad_norm": 0.28274399042129517, "learning_rate": 9.866330768241983e-07, "loss": -0.0, "num_tokens": 4392073.0, "reward": -0.1704331934452057, "reward_std": 0.7666259407997131, "rewards/cosine_scaled_reward/mean": -0.18677911162376404, "rewards/cosine_scaled_reward/std": 0.36125659942626953, "rewards/format_reward/mean": 0.203125, "rewards/format_reward/std": 0.40550529956817627, "step": 35 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.890625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1962.0, "completions/mean_length": 1950.671875, "completions/mean_terminated_length": 1158.1429443359375, "completions/min_length": 669.0, "completions/min_terminated_length": 669.0, "epoch": 0.04114285714285714, "frac_reward_zero_std": 0.0, "grad_norm": 0.2504905164241791, "learning_rate": 9.846666218300807e-07, "loss": -0.0, "num_tokens": 4528028.0, "reward": -0.49544650316238403, "reward_std": 0.3493530750274658, "rewards/cosine_scaled_reward/mean": -0.302410751581192, "rewards/cosine_scaled_reward/std": 0.17342224717140198, "rewards/format_reward/mean": 0.109375, "rewards/format_reward/std": 0.3145764470100403, "step": 36 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.859375, "completions/max_length": 2048.0, "completions/max_terminated_length": 1879.0, "completions/mean_length": 1956.546875, "completions/mean_terminated_length": 1397.6666259765625, "completions/min_length": 789.0, "completions/min_terminated_length": 789.0, "epoch": 0.04228571428571429, "frac_reward_zero_std": 0.0, "grad_norm": 0.24223695695400238, "learning_rate": 9.825677631722435e-07, "loss": 0.0, "num_tokens": 4664271.0, "reward": -0.2983526587486267, "reward_std": 0.45510220527648926, "rewards/cosine_scaled_reward/mean": -0.22730132937431335, "rewards/cosine_scaled_reward/std": 0.21532759070396423, "rewards/format_reward/mean": 0.15625, "rewards/format_reward/std": 0.36596253514289856, "step": 37 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.8125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1844.0, "completions/mean_length": 1872.0625, "completions/mean_terminated_length": 1109.666748046875, "completions/min_length": 799.0, "completions/min_terminated_length": 799.0, "epoch": 0.04342857142857143, "frac_reward_zero_std": 0.0, "grad_norm": 0.22518785297870636, "learning_rate": 9.80337140183366e-07, "loss": 0.0, "num_tokens": 4795795.0, "reward": -0.0591111034154892, "reward_std": 0.38858330249786377, "rewards/cosine_scaled_reward/mean": -0.1311180591583252, "rewards/cosine_scaled_reward/std": 0.32316854596138, "rewards/format_reward/mean": 0.203125, "rewards/format_reward/std": 0.40550529956817627, "step": 38 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.71875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1982.0, "completions/mean_length": 1674.8125, "completions/mean_terminated_length": 721.1111450195312, "completions/min_length": 358.0, "completions/min_terminated_length": 358.0, "epoch": 0.044571428571428574, "frac_reward_zero_std": 0.0, "grad_norm": 0.26911357045173645, "learning_rate": 9.779754323328192e-07, "loss": -0.0, "num_tokens": 4913767.0, "reward": 0.14183415472507477, "reward_std": 0.6081592440605164, "rewards/cosine_scaled_reward/mean": -0.09314543008804321, "rewards/cosine_scaled_reward/std": 0.3410241901874542, "rewards/format_reward/mean": 0.328125, "rewards/format_reward/std": 0.4732423722743988, "step": 39 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.5625, "completions/max_length": 2048.0, "completions/max_terminated_length": 2018.0, "completions/mean_length": 1549.328125, "completions/mean_terminated_length": 908.1785888671875, "completions/min_length": 204.0, "completions/min_terminated_length": 204.0, "epoch": 0.045714285714285714, "frac_reward_zero_std": 0.0, "grad_norm": 0.2770562767982483, "learning_rate": 9.754833590196926e-07, "loss": 0.0, "num_tokens": 5022996.0, "reward": 0.3034515678882599, "reward_std": 0.5147567987442017, "rewards/cosine_scaled_reward/mean": -0.09827423095703125, "rewards/cosine_scaled_reward/std": 0.39993754029273987, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.5039526224136353, "step": 40 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1819.0, "completions/mean_length": 1768.609375, "completions/mean_terminated_length": 1302.9583740234375, "completions/min_length": 584.0, "completions/min_terminated_length": 584.0, "epoch": 0.046857142857142854, "frac_reward_zero_std": 0.0, "grad_norm": 0.23544111847877502, "learning_rate": 9.728616793536587e-07, "loss": 0.0, "num_tokens": 5147339.0, "reward": 0.05204566568136215, "reward_std": 0.7308298349380493, "rewards/cosine_scaled_reward/mean": -0.18491466343402863, "rewards/cosine_scaled_reward/std": 0.3467314541339874, "rewards/format_reward/mean": 0.421875, "rewards/format_reward/std": 0.49776285886764526, "step": 41 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.703125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1894.0, "completions/mean_length": 1662.234375, "completions/mean_terminated_length": 748.5789794921875, "completions/min_length": 164.0, "completions/min_terminated_length": 164.0, "epoch": 0.048, "frac_reward_zero_std": 0.0, "grad_norm": 0.3946230709552765, "learning_rate": 9.701111919237408e-07, "loss": 0.0, "num_tokens": 5264082.0, "reward": -0.1084136962890625, "reward_std": 0.35625624656677246, "rewards/cosine_scaled_reward/mean": -0.21045684814453125, "rewards/cosine_scaled_reward/std": 0.17068159580230713, "rewards/format_reward/mean": 0.3125, "rewards/format_reward/std": 0.467176616191864, "step": 42 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.609375, "completions/max_length": 2048.0, "completions/max_terminated_length": 2044.0, "completions/mean_length": 1628.796875, "completions/mean_terminated_length": 974.8399658203125, "completions/min_length": 387.0, "completions/min_terminated_length": 387.0, "epoch": 0.04914285714285714, "frac_reward_zero_std": 0.0, "grad_norm": 0.2622542679309845, "learning_rate": 9.672327345550543e-07, "loss": -0.0, "num_tokens": 5379941.0, "reward": 0.24864289164543152, "reward_std": 0.622364342212677, "rewards/cosine_scaled_reward/mean": -0.08661604672670364, "rewards/cosine_scaled_reward/std": 0.3968709111213684, "rewards/format_reward/mean": 0.421875, "rewards/format_reward/std": 0.49776285886764526, "step": 43 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1813.0, "completions/mean_length": 1598.265625, "completions/mean_terminated_length": 848.7083740234375, "completions/min_length": 233.0, "completions/min_terminated_length": 233.0, "epoch": 0.05028571428571429, "frac_reward_zero_std": 0.0, "grad_norm": 0.3861124813556671, "learning_rate": 9.64227184053598e-07, "loss": -0.0, "num_tokens": 5492926.0, "reward": 0.17736095190048218, "reward_std": 0.5736653804779053, "rewards/cosine_scaled_reward/mean": -0.09881951659917831, "rewards/cosine_scaled_reward/std": 0.4637540578842163, "rewards/format_reward/mean": 0.375, "rewards/format_reward/std": 0.48795005679130554, "step": 44 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1934.0, "completions/mean_length": 1945.546875, "completions/mean_terminated_length": 1228.375, "completions/min_length": 909.0, "completions/min_terminated_length": 909.0, "epoch": 0.05142857142857143, "frac_reward_zero_std": 0.0, "grad_norm": 0.2586025893688202, "learning_rate": 9.610954559391704e-07, "loss": -0.0, "num_tokens": 5629097.0, "reward": -0.2874904274940491, "reward_std": 0.4528215825557709, "rewards/cosine_scaled_reward/mean": -0.21405771374702454, "rewards/cosine_scaled_reward/std": 0.3033171594142914, "rewards/format_reward/mean": 0.140625, "rewards/format_reward/std": 0.3503824472427368, "step": 45 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.765625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1514.0, "completions/mean_length": 1772.890625, "completions/mean_terminated_length": 874.2000732421875, "completions/min_length": 597.0, "completions/min_terminated_length": 597.0, "epoch": 0.052571428571428575, "frac_reward_zero_std": 0.0, "grad_norm": 0.27347299456596375, "learning_rate": 9.578385041664925e-07, "loss": -0.0, "num_tokens": 5753730.0, "reward": -0.0957992672920227, "reward_std": 0.4836219251155853, "rewards/cosine_scaled_reward/mean": -0.17289963364601135, "rewards/cosine_scaled_reward/std": 0.3050842881202698, "rewards/format_reward/mean": 0.25, "rewards/format_reward/std": 0.4364357888698578, "step": 46 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.609375, "completions/max_length": 2048.0, "completions/max_terminated_length": 2020.0, "completions/mean_length": 1595.734375, "completions/mean_terminated_length": 890.199951171875, "completions/min_length": 379.0, "completions/min_terminated_length": 379.0, "epoch": 0.053714285714285714, "frac_reward_zero_std": 0.0, "grad_norm": 0.3536407947540283, "learning_rate": 9.54457320834625e-07, "loss": 0.0, "num_tokens": 5866257.0, "reward": -0.01777055859565735, "reward_std": 0.6523094177246094, "rewards/cosine_scaled_reward/mean": -0.22763527929782867, "rewards/cosine_scaled_reward/std": 0.3455982208251953, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.5, "step": 47 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.546875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1792.0, "completions/mean_length": 1581.84375, "completions/mean_terminated_length": 1019.2413940429688, "completions/min_length": 397.0, "completions/min_terminated_length": 397.0, "epoch": 0.054857142857142854, "frac_reward_zero_std": 0.0, "grad_norm": 0.26218104362487793, "learning_rate": 9.509529358847654e-07, "loss": -0.0, "num_tokens": 5978039.0, "reward": 0.36145922541618347, "reward_std": 0.8229352235794067, "rewards/cosine_scaled_reward/mean": -0.06145789101719856, "rewards/cosine_scaled_reward/std": 0.4491077661514282, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.5037065148353577, "step": 48 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.484375, "completions/max_length": 2048.0, "completions/max_terminated_length": 1525.0, "completions/mean_length": 1404.46875, "completions/mean_terminated_length": 799.9393920898438, "completions/min_length": 276.0, "completions/min_terminated_length": 276.0, "epoch": 0.056, "frac_reward_zero_std": 0.0, "grad_norm": 0.3138538897037506, "learning_rate": 9.473264167865171e-07, "loss": 0.0, "num_tokens": 6077989.0, "reward": 0.23753327131271362, "reward_std": 0.6856037378311157, "rewards/cosine_scaled_reward/mean": -0.1468583643436432, "rewards/cosine_scaled_reward/std": 0.36308491230010986, "rewards/format_reward/mean": 0.53125, "rewards/format_reward/std": 0.5029674172401428, "step": 49 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.65625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1834.0, "completions/mean_length": 1668.46875, "completions/mean_terminated_length": 943.9091186523438, "completions/min_length": 327.0, "completions/min_terminated_length": 327.0, "epoch": 0.05714285714285714, "frac_reward_zero_std": 0.0, "grad_norm": 0.2541959285736084, "learning_rate": 9.43578868212728e-07, "loss": 0.0, "num_tokens": 6195587.0, "reward": 0.2079824060201645, "reward_std": 0.6563009023666382, "rewards/cosine_scaled_reward/mean": -0.09132131934165955, "rewards/cosine_scaled_reward/std": 0.39781448245048523, "rewards/format_reward/mean": 0.390625, "rewards/format_reward/std": 0.4917473793029785, "step": 50 } ], "logging_steps": 1, "max_steps": 200, "num_input_tokens_seen": 6195587, "num_train_epochs": 1, "save_steps": 50, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 4, "trial_name": null, "trial_params": null }