{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.125, "eval_steps": 500, "global_step": 50, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1298.0, "completions/max_terminated_length": 1298.0, "completions/mean_length": 395.375, "completions/mean_terminated_length": 395.375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0025, "frac_reward_zero_std": 0.0, "grad_norm": 7.455569267272949, "learning_rate": 0.0, "loss": -0.0, "num_tokens": 115752.0, "reward": 0.4414061903953552, "reward_std": 0.45246145129203796, "rewards/format_reward/mean": 0.3515625, "rewards/format_reward/std": 0.24688033759593964, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 1 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1574.0, "completions/max_terminated_length": 1574.0, "completions/mean_length": 484.578125, "completions/mean_terminated_length": 484.578125, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.005, "frac_reward_zero_std": 0.0, "grad_norm": 27.786325454711914, "learning_rate": 3.333333333333333e-07, "loss": 0.0, "num_tokens": 236917.0, "reward": 0.36796873807907104, "reward_std": 0.3844963312149048, "rewards/format_reward/mean": 0.3984375, "rewards/format_reward/std": 0.22146137058734894, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 2 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1453.0, "completions/mean_length": 467.78125, "completions/mean_terminated_length": 416.8064270019531, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.0075, "frac_reward_zero_std": 0.0, "grad_norm": 17.03249168395996, "learning_rate": 6.666666666666666e-07, "loss": -0.0, "num_tokens": 351783.0, "reward": 0.29765623807907104, "reward_std": 0.35281550884246826, "rewards/format_reward/mean": 0.3203125, "rewards/format_reward/std": 0.24180518090724945, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1445.0, "completions/mean_length": 451.015625, "completions/mean_terminated_length": 425.66668701171875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.01, "frac_reward_zero_std": 0.125, "grad_norm": 11.641621589660645, "learning_rate": 1e-06, "loss": 0.0, "num_tokens": 464432.0, "reward": 0.21015624701976776, "reward_std": 0.27983880043029785, "rewards/format_reward/mean": 0.3828125, "rewards/format_reward/std": 0.2478829026222229, "rewards/mcq_exact_match_reward/mean": 0.171875, "rewards/mcq_exact_match_reward/std": 0.38025420904159546, "step": 4 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1653.0, "completions/max_terminated_length": 1653.0, "completions/mean_length": 462.53125, "completions/mean_terminated_length": 462.53125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0125, "frac_reward_zero_std": 0.125, "grad_norm": 7.59962797164917, "learning_rate": 9.988834393115767e-07, "loss": 0.0, "num_tokens": 592362.0, "reward": 0.43437498807907104, "reward_std": 0.3136149048805237, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "step": 5 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1724.0, "completions/mean_length": 513.34375, "completions/mean_terminated_length": 488.9841613769531, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.015, "frac_reward_zero_std": 0.0, "grad_norm": 8.707071304321289, "learning_rate": 9.9553874407739e-07, "loss": -0.0, "num_tokens": 698584.0, "reward": 0.28828123211860657, "reward_std": 0.29159048199653625, "rewards/format_reward/mean": 0.3828125, "rewards/format_reward/std": 0.2478829026222229, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1537.0, "completions/mean_length": 522.984375, "completions/mean_terminated_length": 473.7903137207031, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0175, "frac_reward_zero_std": 0.0, "grad_norm": 9.272878646850586, "learning_rate": 9.899808525182934e-07, "loss": -0.0, "num_tokens": 828047.0, "reward": 0.2203124761581421, "reward_std": 0.33064454793930054, "rewards/format_reward/mean": 0.328125, "rewards/format_reward/std": 0.23935678601264954, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "step": 7 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1684.0, "completions/max_terminated_length": 1684.0, "completions/mean_length": 496.140625, "completions/mean_terminated_length": 496.140625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.02, "frac_reward_zero_std": 0.25, "grad_norm": 9.264406204223633, "learning_rate": 9.822345875271883e-07, "loss": -0.0, "num_tokens": 962776.0, "reward": 0.27031248807907104, "reward_std": 0.260608434677124, "rewards/format_reward/mean": 0.359375, "rewards/format_reward/std": 0.22658175230026245, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "step": 8 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 2022.0, "completions/mean_length": 584.421875, "completions/mean_terminated_length": 537.2096557617188, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0225, "frac_reward_zero_std": 0.125, "grad_norm": 7.400536060333252, "learning_rate": 9.723345458039593e-07, "loss": 0.0, "num_tokens": 1085083.0, "reward": 0.35390621423721313, "reward_std": 0.33069658279418945, "rewards/format_reward/mean": 0.4140625, "rewards/format_reward/std": 0.209963858127594, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1590.0, "completions/mean_length": 536.625, "completions/mean_terminated_length": 512.6349487304688, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.025, "frac_reward_zero_std": 0.0, "grad_norm": 15.998021125793457, "learning_rate": 9.603249433382144e-07, "loss": -0.0, "num_tokens": 1229939.0, "reward": 0.2874999940395355, "reward_std": 0.32263603806495667, "rewards/format_reward/mean": 0.375, "rewards/format_reward/std": 0.30860671401023865, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 10 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1772.0, "completions/max_terminated_length": 1772.0, "completions/mean_length": 489.875, "completions/mean_terminated_length": 489.875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0275, "frac_reward_zero_std": 0.125, "grad_norm": 4.511173248291016, "learning_rate": 9.462594179299405e-07, "loss": -0.0, "num_tokens": 1355115.0, "reward": 0.23359373211860657, "reward_std": 0.33054599165916443, "rewards/format_reward/mean": 0.4609375, "rewards/format_reward/std": 0.20518454909324646, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "step": 11 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1588.0, "completions/max_terminated_length": 1588.0, "completions/mean_length": 380.65625, "completions/mean_terminated_length": 380.65625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.03, "frac_reward_zero_std": 0.0, "grad_norm": 17.27495002746582, "learning_rate": 9.302007896300697e-07, "loss": 0.0, "num_tokens": 1481285.0, "reward": 0.4273437559604645, "reward_std": 0.39148083329200745, "rewards/format_reward/mean": 0.5234375, "rewards/format_reward/std": 0.24282869696617126, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 12 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1059.0, "completions/max_terminated_length": 1059.0, "completions/mean_length": 310.609375, "completions/mean_terminated_length": 310.609375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0325, "frac_reward_zero_std": 0.5, "grad_norm": 2.3885817527770996, "learning_rate": 9.122207801708801e-07, "loss": 0.0, "num_tokens": 1586116.0, "reward": 0.29609373211860657, "reward_std": 0.13455916941165924, "rewards/format_reward/mean": 0.4609375, "rewards/format_reward/std": 0.13524486124515533, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1610.0, "completions/mean_length": 194.828125, "completions/mean_terminated_length": 165.41270446777344, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.035, "frac_reward_zero_std": 0.125, "grad_norm": 25.50330924987793, "learning_rate": 8.923996926393305e-07, "loss": 0.0, "num_tokens": 1695009.0, "reward": 0.33515626192092896, "reward_std": 0.27259403467178345, "rewards/format_reward/mean": 0.5390625, "rewards/format_reward/std": 0.2236899733543396, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 14 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1318.0, "completions/max_terminated_length": 1318.0, "completions/mean_length": 356.875, "completions/mean_terminated_length": 356.875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0375, "frac_reward_zero_std": 0.0, "grad_norm": 11.296743392944336, "learning_rate": 8.708260528239788e-07, "loss": -0.0, "num_tokens": 1782273.0, "reward": 0.41718751192092896, "reward_std": 0.3614344298839569, "rewards/format_reward/mean": 0.578125, "rewards/format_reward/std": 0.3238992393016815, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "step": 15 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 909.0, "completions/max_terminated_length": 909.0, "completions/mean_length": 78.140625, "completions/mean_terminated_length": 78.140625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.04, "frac_reward_zero_std": 0.75, "grad_norm": 13.47980785369873, "learning_rate": 8.475962138373212e-07, "loss": 0.0, "num_tokens": 1893482.0, "reward": 0.3179687261581421, "reward_std": 0.048844318836927414, "rewards/format_reward/mean": 0.5234375, "rewards/format_reward/std": 0.13886408507823944, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "step": 16 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 715.0, "completions/max_terminated_length": 715.0, "completions/mean_length": 121.09375, "completions/mean_terminated_length": 121.09375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0425, "frac_reward_zero_std": 0.5, "grad_norm": 19.67048454284668, "learning_rate": 8.228139257794012e-07, "loss": 0.0, "num_tokens": 1971344.0, "reward": 0.47343748807907104, "reward_std": 0.12049457430839539, "rewards/format_reward/mean": 0.515625, "rewards/format_reward/std": 0.23517554998397827, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 17 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1030.0, "completions/max_terminated_length": 1030.0, "completions/mean_length": 95.875, "completions/mean_terminated_length": 95.875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.045, "frac_reward_zero_std": 0.5, "grad_norm": 8.539053916931152, "learning_rate": 7.965898723646776e-07, "loss": -0.0, "num_tokens": 2071648.0, "reward": 0.4601562023162842, "reward_std": 0.20276136696338654, "rewards/format_reward/mean": 0.5390625, "rewards/format_reward/std": 0.16194961965084076, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 18 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 543.0, "completions/max_terminated_length": 543.0, "completions/mean_length": 44.34375, "completions/mean_terminated_length": 44.34375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0475, "frac_reward_zero_std": 0.75, "grad_norm": 7.316840648651123, "learning_rate": 7.690411765816864e-07, "loss": 0.0, "num_tokens": 2186862.0, "reward": 0.33281248807907104, "reward_std": 0.09033423662185669, "rewards/format_reward/mean": 0.515625, "rewards/format_reward/std": 0.08768405020236969, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 19 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05, "frac_reward_zero_std": 0.5, "grad_norm": 18.261327743530273, "learning_rate": 7.402908775933419e-07, "loss": 0.0, "num_tokens": 2316110.0, "reward": 0.4249999523162842, "reward_std": 0.1767766922712326, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 20 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 500.0, "completions/max_terminated_length": 500.0, "completions/mean_length": 51.53125, "completions/mean_terminated_length": 51.53125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0525, "frac_reward_zero_std": 0.25, "grad_norm": 21.777469635009766, "learning_rate": 7.104673812141675e-07, "loss": 0.0, "num_tokens": 2425600.0, "reward": 0.5523437261581421, "reward_std": 0.31216710805892944, "rewards/format_reward/mean": 0.5234375, "rewards/format_reward/std": 0.10652101784944534, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "step": 21 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 884.0, "completions/max_terminated_length": 884.0, "completions/mean_length": 63.96875, "completions/mean_terminated_length": 63.96875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.055, "frac_reward_zero_std": 0.625, "grad_norm": 13.770291328430176, "learning_rate": 6.797038864187563e-07, "loss": -0.0, "num_tokens": 2551366.0, "reward": 0.6148437261581421, "reward_std": 0.16732138395309448, "rewards/format_reward/mean": 0.5234375, "rewards/format_reward/std": 0.10652101784944534, "rewards/mcq_exact_match_reward/mean": 0.5625, "rewards/mcq_exact_match_reward/std": 0.5, "step": 22 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 595.0, "completions/max_terminated_length": 595.0, "completions/mean_length": 114.4375, "completions/mean_terminated_length": 114.4375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0575, "frac_reward_zero_std": 0.125, "grad_norm": 26.109661102294922, "learning_rate": 6.48137790442817e-07, "loss": -0.0, "num_tokens": 2640258.0, "reward": 0.5617187023162842, "reward_std": 0.3653518557548523, "rewards/format_reward/mean": 0.6171875, "rewards/format_reward/std": 0.2313210517168045, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "step": 23 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06, "frac_reward_zero_std": 0.75, "grad_norm": 13.819435119628906, "learning_rate": 6.159100751337641e-07, "loss": -0.0, "num_tokens": 2748562.0, "reward": 0.3937499523162842, "reward_std": 0.0883883461356163, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "step": 24 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 722.0, "completions/max_terminated_length": 722.0, "completions/mean_length": 51.71875, "completions/mean_terminated_length": 51.71875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0625, "frac_reward_zero_std": 0.625, "grad_norm": 16.029680252075195, "learning_rate": 5.83164677291565e-07, "loss": 0.0, "num_tokens": 2828080.0, "reward": 0.5062499642372131, "reward_std": 0.10539869219064713, "rewards/format_reward/mean": 0.53125, "rewards/format_reward/std": 0.12198751419782639, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 25 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 588.0, "completions/max_terminated_length": 588.0, "completions/mean_length": 149.046875, "completions/mean_terminated_length": 149.046875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.065, "frac_reward_zero_std": 0.375, "grad_norm": 10.594514846801758, "learning_rate": 5.500478458120493e-07, "loss": -0.0, "num_tokens": 2906387.0, "reward": 0.3062499761581421, "reward_std": 0.22346952557563782, "rewards/format_reward/mean": 0.5625, "rewards/format_reward/std": 0.24397502839565277, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 26 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 604.0, "completions/max_terminated_length": 604.0, "completions/mean_length": 18.9375, "completions/mean_terminated_length": 18.9375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0675, "frac_reward_zero_std": 0.625, "grad_norm": 15.209108352661133, "learning_rate": 5.167074885038372e-07, "loss": -0.0, "num_tokens": 3006191.0, "reward": 0.23906248807907104, "reward_std": 0.16869747638702393, "rewards/format_reward/mean": 0.515625, "rewards/format_reward/std": 0.08768405020236969, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "step": 27 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 486.0, "completions/max_terminated_length": 486.0, "completions/mean_length": 19.40625, "completions/mean_terminated_length": 19.40625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07, "frac_reward_zero_std": 0.375, "grad_norm": 25.709003448486328, "learning_rate": 4.832925114961628e-07, "loss": -0.0, "num_tokens": 3114025.0, "reward": 0.3585937023162842, "reward_std": 0.23680154979228973, "rewards/format_reward/mean": 0.4609375, "rewards/format_reward/std": 0.13524486124515533, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "step": 28 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 711.0, "completions/max_terminated_length": 711.0, "completions/mean_length": 136.21875, "completions/mean_terminated_length": 136.21875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0725, "frac_reward_zero_std": 0.375, "grad_norm": 13.170354843139648, "learning_rate": 4.499521541879508e-07, "loss": -0.0, "num_tokens": 3211447.0, "reward": 0.46796873211860657, "reward_std": 0.29294508695602417, "rewards/format_reward/mean": 0.6171875, "rewards/format_reward/std": 0.2313210517168045, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 29 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1564.0, "completions/max_terminated_length": 1564.0, "completions/mean_length": 63.53125, "completions/mean_terminated_length": 63.53125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.075, "frac_reward_zero_std": 0.375, "grad_norm": 20.386463165283203, "learning_rate": 4.1683532270843495e-07, "loss": -0.0, "num_tokens": 3341129.0, "reward": 0.3179687261581421, "reward_std": 0.3107113242149353, "rewards/format_reward/mean": 0.5234375, "rewards/format_reward/std": 0.10652101784944534, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "step": 30 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 866.0, "completions/max_terminated_length": 866.0, "completions/mean_length": 96.9375, "completions/mean_terminated_length": 96.9375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0775, "frac_reward_zero_std": 0.375, "grad_norm": 17.929813385009766, "learning_rate": 3.840899248662358e-07, "loss": -0.0, "num_tokens": 3421261.0, "reward": 0.24453124403953552, "reward_std": 0.30016404390335083, "rewards/format_reward/mean": 0.5703125, "rewards/format_reward/std": 0.1751912236213684, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "step": 31 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 922.0, "completions/max_terminated_length": 922.0, "completions/mean_length": 131.859375, "completions/mean_terminated_length": 131.859375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08, "frac_reward_zero_std": 0.5, "grad_norm": 13.730548858642578, "learning_rate": 3.5186220955718303e-07, "loss": -0.0, "num_tokens": 3520652.0, "reward": 0.52734375, "reward_std": 0.14764924347400665, "rewards/format_reward/mean": 0.5859375, "rewards/format_reward/std": 0.19012710452079773, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 32 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 461.0, "completions/max_terminated_length": 461.0, "completions/mean_length": 24.640625, "completions/mean_terminated_length": 24.640625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0825, "frac_reward_zero_std": 0.5, "grad_norm": 16.63532829284668, "learning_rate": 3.2029611358124365e-07, "loss": -0.0, "num_tokens": 3666669.0, "reward": 0.4257812201976776, "reward_std": 0.24793623387813568, "rewards/format_reward/mean": 0.5078125, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 33 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 474.0, "completions/max_terminated_length": 474.0, "completions/mean_length": 67.59375, "completions/mean_terminated_length": 67.59375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.085, "frac_reward_zero_std": 0.5, "grad_norm": 16.327985763549805, "learning_rate": 2.895326187858326e-07, "loss": 0.0, "num_tokens": 3765659.0, "reward": 0.5695312023162842, "reward_std": 0.16235783696174622, "rewards/format_reward/mean": 0.5390625, "rewards/format_reward/std": 0.16194961965084076, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_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": 0.0, "completions/max_length": 1286.0, "completions/max_terminated_length": 1286.0, "completions/mean_length": 74.296875, "completions/mean_terminated_length": 74.296875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0875, "frac_reward_zero_std": 0.375, "grad_norm": 18.27692413330078, "learning_rate": 2.597091224066581e-07, "loss": 0.0, "num_tokens": 3869942.0, "reward": 0.3828125, "reward_std": 0.16247877478599548, "rewards/format_reward/mean": 0.546875, "rewards/format_reward/std": 0.14689241349697113, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 724.0, "completions/mean_length": 107.390625, "completions/mean_terminated_length": 76.5873031616211, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.09, "frac_reward_zero_std": 0.75, "grad_norm": 2.7832350730895996, "learning_rate": 2.309588234183137e-07, "loss": 0.0, "num_tokens": 3966567.0, "reward": 0.5968749523162842, "reward_std": 0.06629125773906708, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.08908708393573761, "rewards/mcq_exact_match_reward/mean": 0.546875, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 506.0, "completions/mean_length": 98.609375, "completions/mean_terminated_length": 67.66667175292969, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0925, "frac_reward_zero_std": 0.25, "grad_norm": 32.34717559814453, "learning_rate": 2.0341012763532239e-07, "loss": 0.0, "num_tokens": 4062662.0, "reward": 0.6203124523162842, "reward_std": 0.33224427700042725, "rewards/format_reward/mean": 0.578125, "rewards/format_reward/std": 0.20351573824882507, "rewards/mcq_exact_match_reward/mean": 0.5625, "rewards/mcq_exact_match_reward/std": 0.5, "step": 37 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 449.0, "completions/max_terminated_length": 449.0, "completions/mean_length": 23.40625, "completions/mean_terminated_length": 23.40625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.095, "frac_reward_zero_std": 0.25, "grad_norm": 25.669404983520508, "learning_rate": 1.7718607422059879e-07, "loss": -0.0, "num_tokens": 4156112.0, "reward": 0.1601562350988388, "reward_std": 0.24368003010749817, "rewards/format_reward/mean": 0.5078125, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.109375, "rewards/mcq_exact_match_reward/std": 0.3145764470100403, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 599.0, "completions/mean_length": 111.546875, "completions/mean_terminated_length": 80.80953216552734, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0975, "frac_reward_zero_std": 0.5, "grad_norm": 16.772607803344727, "learning_rate": 1.5240378616267886e-07, "loss": -0.0, "num_tokens": 4268163.0, "reward": 0.34062498807907104, "reward_std": 0.23483146727085114, "rewards/format_reward/mean": 0.59375, "rewards/format_reward/std": 0.21593283116817474, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 39 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 712.0, "completions/max_terminated_length": 712.0, "completions/mean_length": 79.75, "completions/mean_terminated_length": 79.75, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1, "frac_reward_zero_std": 0.375, "grad_norm": 10.34507942199707, "learning_rate": 1.291739471760212e-07, "loss": -0.0, "num_tokens": 4366523.0, "reward": 0.33906248211860657, "reward_std": 0.20761553943157196, "rewards/format_reward/mean": 0.578125, "rewards/format_reward/std": 0.18298126757144928, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 40 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 938.0, "completions/max_terminated_length": 938.0, "completions/mean_length": 102.140625, "completions/mean_terminated_length": 102.140625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1025, "frac_reward_zero_std": 0.375, "grad_norm": 13.970519065856934, "learning_rate": 1.076003073606695e-07, "loss": -0.0, "num_tokens": 4480932.0, "reward": 0.6039062142372131, "reward_std": 0.2498142421245575, "rewards/format_reward/mean": 0.5703125, "rewards/format_reward/std": 0.1751912236213684, "rewards/mcq_exact_match_reward/mean": 0.546875, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 41 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1215.0, "completions/max_terminated_length": 1215.0, "completions/mean_length": 145.484375, "completions/mean_terminated_length": 145.484375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.105, "frac_reward_zero_std": 0.375, "grad_norm": 17.185466766357422, "learning_rate": 8.777921982911996e-08, "loss": 0.0, "num_tokens": 4551787.0, "reward": 0.6265624761581421, "reward_std": 0.15359337627887726, "rewards/format_reward/mean": 0.640625, "rewards/format_reward/std": 0.22658175230026245, "rewards/mcq_exact_match_reward/mean": 0.5625, "rewards/mcq_exact_match_reward/std": 0.5, "step": 42 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 829.0, "completions/max_terminated_length": 829.0, "completions/mean_length": 88.96875, "completions/mean_terminated_length": 88.96875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1075, "frac_reward_zero_std": 0.625, "grad_norm": 5.71751594543457, "learning_rate": 6.979921036993041e-08, "loss": -0.0, "num_tokens": 4652017.0, "reward": 0.6585937142372131, "reward_std": 0.13295188546180725, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.609375, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "step": 43 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 729.0, "completions/max_terminated_length": 729.0, "completions/mean_length": 77.859375, "completions/mean_terminated_length": 77.859375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.11, "frac_reward_zero_std": 0.5, "grad_norm": 15.294127464294434, "learning_rate": 5.3740582070059435e-08, "loss": 0.0, "num_tokens": 4747488.0, "reward": 0.45937496423721313, "reward_std": 0.222185879945755, "rewards/format_reward/mean": 0.53125, "rewards/format_reward/std": 0.1510545015335083, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 44 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 314.0, "completions/max_terminated_length": 314.0, "completions/mean_length": 37.71875, "completions/mean_terminated_length": 37.71875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1125, "frac_reward_zero_std": 0.5, "grad_norm": 18.710546493530273, "learning_rate": 3.967505666178555e-08, "loss": 0.0, "num_tokens": 4835382.0, "reward": 0.40156251192092896, "reward_std": 0.19661885499954224, "rewards/format_reward/mean": 0.578125, "rewards/format_reward/std": 0.18298126757144928, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "step": 45 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 889.0, "completions/max_terminated_length": 889.0, "completions/mean_length": 81.5, "completions/mean_terminated_length": 81.5, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.115, "frac_reward_zero_std": 0.5, "grad_norm": 12.153166770935059, "learning_rate": 2.766545419604066e-08, "loss": -0.0, "num_tokens": 4935078.0, "reward": 0.28828123211860657, "reward_std": 0.20956924557685852, "rewards/format_reward/mean": 0.5390625, "rewards/format_reward/std": 0.13524486124515533, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "step": 46 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 665.0, "completions/max_terminated_length": 665.0, "completions/mean_length": 47.578125, "completions/mean_terminated_length": 47.578125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1175, "frac_reward_zero_std": 0.5, "grad_norm": 17.548736572265625, "learning_rate": 1.776541247281177e-08, "loss": 0.0, "num_tokens": 5038595.0, "reward": 0.3499999940395355, "reward_std": 0.2097943276166916, "rewards/format_reward/mean": 0.53125, "rewards/format_reward/std": 0.12198751419782639, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "step": 47 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 451.0, "completions/max_terminated_length": 451.0, "completions/mean_length": 46.875, "completions/mean_terminated_length": 46.875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.12, "frac_reward_zero_std": 0.625, "grad_norm": 34.39931106567383, "learning_rate": 1.0019147481706625e-08, "loss": -0.0, "num_tokens": 5123955.0, "reward": 0.5062499642372131, "reward_std": 0.18945617973804474, "rewards/format_reward/mean": 0.53125, "rewards/format_reward/std": 0.12198751419782639, "rewards/mcq_exact_match_reward/mean": 0.453125, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 48 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1225, "frac_reward_zero_std": 0.375, "grad_norm": 17.74892807006836, "learning_rate": 4.461255922609985e-09, "loss": 0.0, "num_tokens": 5247763.0, "reward": 0.5343749523162842, "reward_std": 0.25726157426834106, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.484375, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "step": 49 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1143.0, "completions/max_terminated_length": 1143.0, "completions/mean_length": 48.859375, "completions/mean_terminated_length": 48.859375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.125, "frac_reward_zero_std": 0.75, "grad_norm": 12.674071311950684, "learning_rate": 1.116560688423418e-09, "loss": -0.0, "num_tokens": 5352218.0, "reward": 0.3804687261581421, "reward_std": 0.11017291247844696, "rewards/format_reward/mean": 0.5234375, "rewards/format_reward/std": 0.10652101784944534, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 50 } ], "logging_steps": 1, "max_steps": 50, "num_input_tokens_seen": 5352218, "num_train_epochs": 1, "save_steps": 50, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }