| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.995499549954996, | |
| "eval_steps": 15, | |
| "global_step": 395, | |
| "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": 0.0279017857142857, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1978.0, | |
| "completions/mean_length": 538.3058471679688, | |
| "completions/mean_terminated_length": 494.97357177734375, | |
| "completions/min_length": 3.0, | |
| "completions/min_terminated_length": 3.0, | |
| "epoch": 0.012601260126012601, | |
| "grad_norm": 0.602237343788147, | |
| "learning_rate": 0.0, | |
| "loss": 0.0039, | |
| "num_tokens": 609906.0, | |
| "reward": 0.2795497477054596, | |
| "reward_std": 0.41905471682548523, | |
| "rewards/accuracy_reward/mean": 0.1685267835855484, | |
| "rewards/accuracy_reward/std": 0.3745425045490265, | |
| "rewards/cosine_scaled_reward/mean": 0.1684744656085968, | |
| "rewards/cosine_scaled_reward/std": 0.37442949414253235, | |
| "rewards/format_reward/mean": 0.0535714291036129, | |
| "rewards/format_reward/std": 0.2252957671880722, | |
| "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": 0.021484375, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1946.5, | |
| "completions/mean_length": 523.0030899047852, | |
| "completions/mean_terminated_length": 489.49188232421875, | |
| "completions/min_length": 2.5, | |
| "completions/min_terminated_length": 2.5, | |
| "epoch": 0.063006300630063, | |
| "grad_norm": 0.32099631428718567, | |
| "learning_rate": 1e-07, | |
| "loss": 0.0035, | |
| "num_tokens": 3031821.0, | |
| "reward": 0.2492716759443283, | |
| "reward_std": 0.3989013135433197, | |
| "rewards/accuracy_reward/mean": 0.1512276753783226, | |
| "rewards/accuracy_reward/std": 0.3584018871188164, | |
| "rewards/cosine_scaled_reward/mean": 0.15116609632968903, | |
| "rewards/cosine_scaled_reward/std": 0.35826219618320465, | |
| "rewards/format_reward/mean": 0.044921875931322575, | |
| "rewards/format_reward/std": 0.20668350905179977, | |
| "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": 0.024107142857142883, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1979.6, | |
| "completions/mean_length": 547.3105102539063, | |
| "completions/mean_terminated_length": 510.2965393066406, | |
| "completions/min_length": 4.8, | |
| "completions/min_terminated_length": 4.8, | |
| "epoch": 0.126012601260126, | |
| "grad_norm": 0.4403175711631775, | |
| "learning_rate": 2.25e-07, | |
| "loss": 0.0027, | |
| "num_tokens": 6168628.0, | |
| "reward": 0.2597510188817978, | |
| "reward_std": 0.39840335249900816, | |
| "rewards/accuracy_reward/mean": 0.15892857015132905, | |
| "rewards/accuracy_reward/std": 0.365299665927887, | |
| "rewards/cosine_scaled_reward/mean": 0.15878771841526032, | |
| "rewards/cosine_scaled_reward/std": 0.3649946331977844, | |
| "rewards/format_reward/mean": 0.04285714328289032, | |
| "rewards/format_reward/std": 0.2020670711994171, | |
| "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": 0.02589285714285714, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1994.4, | |
| "completions/mean_length": 555.5024780273437, | |
| "completions/mean_terminated_length": 515.7921813964844, | |
| "completions/min_length": 6.2, | |
| "completions/min_terminated_length": 6.2, | |
| "epoch": 0.18901890189018902, | |
| "grad_norm": 1.8242275714874268, | |
| "learning_rate": 3.5e-07, | |
| "loss": 0.0031, | |
| "num_tokens": 9301535.0, | |
| "reward": 0.27981547117233274, | |
| "reward_std": 0.40177412033081056, | |
| "rewards/accuracy_reward/mean": 0.16138392984867095, | |
| "rewards/accuracy_reward/std": 0.36725109815597534, | |
| "rewards/cosine_scaled_reward/mean": 0.16119343042373657, | |
| "rewards/cosine_scaled_reward/std": 0.36687323451042175, | |
| "rewards/format_reward/mean": 0.07566964402794837, | |
| "rewards/format_reward/std": 0.2617058604955673, | |
| "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": 0.02209821428571428, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 2017.0, | |
| "completions/mean_length": 548.893994140625, | |
| "completions/mean_terminated_length": 515.1586364746094, | |
| "completions/min_length": 3.0, | |
| "completions/min_terminated_length": 3.0, | |
| "epoch": 0.252025202520252, | |
| "grad_norm": 0.6227929592132568, | |
| "learning_rate": 4.7499999999999995e-07, | |
| "loss": -0.0047, | |
| "num_tokens": 12415812.0, | |
| "reward": 0.32004910707473755, | |
| "reward_std": 0.4481643557548523, | |
| "rewards/accuracy_reward/mean": 0.17611607313156127, | |
| "rewards/accuracy_reward/std": 0.3801506757736206, | |
| "rewards/cosine_scaled_reward/mean": 0.1758124738931656, | |
| "rewards/cosine_scaled_reward/std": 0.37958953976631166, | |
| "rewards/format_reward/mean": 0.11205357164144517, | |
| "rewards/format_reward/std": 0.3129072666168213, | |
| "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": 0.02075892857142858, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1966.2, | |
| "completions/mean_length": 565.0823852539063, | |
| "completions/mean_terminated_length": 533.6526733398438, | |
| "completions/min_length": 2.8, | |
| "completions/min_terminated_length": 2.8, | |
| "epoch": 0.31503150315031503, | |
| "grad_norm": 0.38101691007614136, | |
| "learning_rate": 6e-07, | |
| "loss": -0.0097, | |
| "num_tokens": 15816741.0, | |
| "reward": 0.5020053923130036, | |
| "reward_std": 0.6075942158699036, | |
| "rewards/accuracy_reward/mean": 0.2622948855161667, | |
| "rewards/accuracy_reward/std": 0.4395259916782379, | |
| "rewards/cosine_scaled_reward/mean": 0.26026553511619566, | |
| "rewards/cosine_scaled_reward/std": 0.43781105875968934, | |
| "rewards/format_reward/mean": 0.22053571343421935, | |
| "rewards/format_reward/std": 0.40703149437904357, | |
| "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": 0.017187499999999977, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1967.0, | |
| "completions/mean_length": 578.9469116210937, | |
| "completions/mean_terminated_length": 553.24716796875, | |
| "completions/min_length": 6.8, | |
| "completions/min_terminated_length": 6.8, | |
| "epoch": 0.37803780378037805, | |
| "grad_norm": 0.33144351840019226, | |
| "learning_rate": 7.249999999999999e-07, | |
| "loss": -0.0055, | |
| "num_tokens": 19248647.0, | |
| "reward": 0.7754053592681884, | |
| "reward_std": 0.677242922782898, | |
| "rewards/accuracy_reward/mean": 0.35334821343421935, | |
| "rewards/accuracy_reward/std": 0.4751245081424713, | |
| "rewards/cosine_scaled_reward/mean": 0.3530427932739258, | |
| "rewards/cosine_scaled_reward/std": 0.4747708261013031, | |
| "rewards/format_reward/mean": 0.4910714328289032, | |
| "rewards/format_reward/std": 0.48938958048820497, | |
| "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": 0.014285714285714258, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1959.6, | |
| "completions/mean_length": 608.9741455078125, | |
| "completions/mean_terminated_length": 588.129736328125, | |
| "completions/min_length": 12.0, | |
| "completions/min_terminated_length": 12.0, | |
| "epoch": 0.44104410441044106, | |
| "grad_norm": 0.2479850798845291, | |
| "learning_rate": 8.499999999999999e-07, | |
| "loss": 0.0144, | |
| "num_tokens": 22822291.0, | |
| "reward": 1.1359302043914794, | |
| "reward_std": 0.633991277217865, | |
| "rewards/accuracy_reward/mean": 0.5194196462631225, | |
| "rewards/accuracy_reward/std": 0.49612280130386355, | |
| "rewards/cosine_scaled_reward/mean": 0.518735259771347, | |
| "rewards/cosine_scaled_reward/std": 0.495624178647995, | |
| "rewards/format_reward/mean": 0.7142857193946839, | |
| "rewards/format_reward/std": 0.4489860713481903, | |
| "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": 0.0125, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1967.4, | |
| "completions/mean_length": 672.230615234375, | |
| "completions/mean_terminated_length": 654.8196044921875, | |
| "completions/min_length": 70.6, | |
| "completions/min_terminated_length": 70.6, | |
| "epoch": 0.504050405040504, | |
| "grad_norm": 0.19197942316532135, | |
| "learning_rate": 9.75e-07, | |
| "loss": 0.0261, | |
| "num_tokens": 26671876.0, | |
| "reward": 1.3178162574768066, | |
| "reward_std": 0.5191608130931854, | |
| "rewards/accuracy_reward/mean": 0.5977678537368775, | |
| "rewards/accuracy_reward/std": 0.48948062062263487, | |
| "rewards/cosine_scaled_reward/mean": 0.5967931866645813, | |
| "rewards/cosine_scaled_reward/std": 0.4888714671134949, | |
| "rewards/format_reward/mean": 0.8433035612106323, | |
| "rewards/format_reward/std": 0.35585846900939944, | |
| "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": 0.01383928571428572, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 2024.0, | |
| "completions/mean_length": 691.8661010742187, | |
| "completions/mean_terminated_length": 672.869873046875, | |
| "completions/min_length": 127.6, | |
| "completions/min_terminated_length": 127.6, | |
| "epoch": 0.5670567056705671, | |
| "grad_norm": 0.12823984026908875, | |
| "learning_rate": 9.99686774088805e-07, | |
| "loss": 0.0293, | |
| "num_tokens": 30644708.0, | |
| "reward": 1.4662254571914672, | |
| "reward_std": 0.4329299807548523, | |
| "rewards/accuracy_reward/mean": 0.6647321343421936, | |
| "rewards/accuracy_reward/std": 0.47048051953315734, | |
| "rewards/cosine_scaled_reward/mean": 0.6637008666992188, | |
| "rewards/cosine_scaled_reward/std": 0.46995849609375, | |
| "rewards/format_reward/mean": 0.9392857074737548, | |
| "rewards/format_reward/std": 0.23625308275222778, | |
| "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": 0.01517857142857142, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 2006.2, | |
| "completions/mean_length": 724.4636352539062, | |
| "completions/mean_terminated_length": 704.2110107421875, | |
| "completions/min_length": 183.8, | |
| "completions/min_terminated_length": 183.8, | |
| "epoch": 0.6300630063006301, | |
| "grad_norm": 0.11830825358629227, | |
| "learning_rate": 9.984149663879992e-07, | |
| "loss": 0.0277, | |
| "num_tokens": 34704233.0, | |
| "reward": 1.4932293176651001, | |
| "reward_std": 0.39716021418571473, | |
| "rewards/accuracy_reward/mean": 0.6720982074737549, | |
| "rewards/accuracy_reward/std": 0.46862396597862244, | |
| "rewards/cosine_scaled_reward/mean": 0.6715031266212463, | |
| "rewards/cosine_scaled_reward/std": 0.4683354377746582, | |
| "rewards/format_reward/mean": 0.9707589268684387, | |
| "rewards/format_reward/std": 0.1669136554002762, | |
| "step": 50 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.009821428571428559, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1976.2, | |
| "completions/mean_length": 713.6201171875, | |
| "completions/mean_terminated_length": 700.3927490234375, | |
| "completions/min_length": 167.4, | |
| "completions/min_terminated_length": 167.4, | |
| "epoch": 0.693069306930693, | |
| "grad_norm": 0.10666404664516449, | |
| "learning_rate": 9.961674879848956e-07, | |
| "loss": 0.0233, | |
| "num_tokens": 38681571.0, | |
| "reward": 1.5286698818206788, | |
| "reward_std": 0.36640719771385194, | |
| "rewards/accuracy_reward/mean": 0.6915178418159484, | |
| "rewards/accuracy_reward/std": 0.4611486494541168, | |
| "rewards/cosine_scaled_reward/mean": 0.6908218264579773, | |
| "rewards/cosine_scaled_reward/std": 0.4608599662780762, | |
| "rewards/format_reward/mean": 0.9834821462631226, | |
| "rewards/format_reward/std": 0.12500424087047576, | |
| "step": 55 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.012053571428571419, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1924.6, | |
| "completions/mean_length": 711.6618530273438, | |
| "completions/mean_terminated_length": 695.3810913085938, | |
| "completions/min_length": 164.0, | |
| "completions/min_terminated_length": 164.0, | |
| "epoch": 0.7560756075607561, | |
| "grad_norm": 0.1039816215634346, | |
| "learning_rate": 9.929487384240101e-07, | |
| "loss": 0.0233, | |
| "num_tokens": 42628176.0, | |
| "reward": 1.5549091339111327, | |
| "reward_std": 0.3488038718700409, | |
| "rewards/accuracy_reward/mean": 0.7089285612106323, | |
| "rewards/accuracy_reward/std": 0.45444890260696413, | |
| "rewards/cosine_scaled_reward/mean": 0.7075859427452087, | |
| "rewards/cosine_scaled_reward/std": 0.4539263606071472, | |
| "rewards/format_reward/mean": 0.984375, | |
| "rewards/format_reward/std": 0.1233037456870079, | |
| "step": 60 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.01339285714285714, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1963.2, | |
| "completions/mean_length": 713.0527099609375, | |
| "completions/mean_terminated_length": 694.9294799804687, | |
| "completions/min_length": 205.6, | |
| "completions/min_terminated_length": 205.6, | |
| "epoch": 0.819081908190819, | |
| "grad_norm": 0.11131048947572708, | |
| "learning_rate": 9.88765018558761e-07, | |
| "loss": 0.0214, | |
| "num_tokens": 46613412.0, | |
| "reward": 1.553797197341919, | |
| "reward_std": 0.34091995358467103, | |
| "rewards/accuracy_reward/mean": 0.7078125, | |
| "rewards/accuracy_reward/std": 0.45394955277442933, | |
| "rewards/cosine_scaled_reward/mean": 0.7067014455795289, | |
| "rewards/cosine_scaled_reward/std": 0.4534592032432556, | |
| "rewards/format_reward/mean": 0.9852678537368774, | |
| "rewards/format_reward/std": 0.11978632807731629, | |
| "step": 65 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.010267857142857162, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1950.2, | |
| "completions/mean_length": 687.0343994140625, | |
| "completions/mean_terminated_length": 672.891162109375, | |
| "completions/min_length": 155.0, | |
| "completions/min_terminated_length": 155.0, | |
| "epoch": 0.8820882088208821, | |
| "grad_norm": 0.11210864782333374, | |
| "learning_rate": 9.836245182172519e-07, | |
| "loss": 0.0192, | |
| "num_tokens": 50425302.0, | |
| "reward": 1.568692970275879, | |
| "reward_std": 0.3520396828651428, | |
| "rewards/accuracy_reward/mean": 0.7180039405822753, | |
| "rewards/accuracy_reward/std": 0.44985153079032897, | |
| "rewards/cosine_scaled_reward/mean": 0.7167718410491943, | |
| "rewards/cosine_scaled_reward/std": 0.44936810731887816, | |
| "rewards/format_reward/mean": 0.9883928537368775, | |
| "rewards/format_reward/std": 0.10585158914327622, | |
| "step": 70 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.007366071428571397, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1996.0, | |
| "completions/mean_length": 702.3578369140625, | |
| "completions/mean_terminated_length": 692.3953125, | |
| "completions/min_length": 167.2, | |
| "completions/min_terminated_length": 167.2, | |
| "epoch": 0.9450945094509451, | |
| "grad_norm": 0.11102982610464096, | |
| "learning_rate": 9.775373001702997e-07, | |
| "loss": 0.0159, | |
| "num_tokens": 54367609.0, | |
| "reward": 1.5509341955184937, | |
| "reward_std": 0.3260269582271576, | |
| "rewards/accuracy_reward/mean": 0.7037946343421936, | |
| "rewards/accuracy_reward/std": 0.45627689361572266, | |
| "rewards/cosine_scaled_reward/mean": 0.7023146390914917, | |
| "rewards/cosine_scaled_reward/std": 0.45560853481292723, | |
| "rewards/format_reward/mean": 0.9919642925262451, | |
| "rewards/format_reward/std": 0.08732087463140488, | |
| "step": 75 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.008453970180305114, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1951.4, | |
| "completions/mean_length": 707.547412109375, | |
| "completions/mean_terminated_length": 696.137548828125, | |
| "completions/min_length": 175.0, | |
| "completions/min_terminated_length": 175.0, | |
| "epoch": 1.0126012601260126, | |
| "grad_norm": 0.10652182251214981, | |
| "learning_rate": 9.705152804330871e-07, | |
| "loss": 0.0192, | |
| "num_tokens": 58375847.0, | |
| "reward": 1.5461820125579835, | |
| "reward_std": 0.3652195274829865, | |
| "rewards/accuracy_reward/mean": 0.7006696343421936, | |
| "rewards/accuracy_reward/std": 0.4574929714202881, | |
| "rewards/cosine_scaled_reward/mean": 0.6995066642761231, | |
| "rewards/cosine_scaled_reward/std": 0.45710301399230957, | |
| "rewards/format_reward/mean": 0.9915178656578064, | |
| "rewards/format_reward/std": 0.0895774394273758, | |
| "step": 80 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.005580357142857162, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1953.8, | |
| "completions/mean_length": 682.2417724609375, | |
| "completions/mean_terminated_length": 674.6181396484375, | |
| "completions/min_length": 172.0, | |
| "completions/min_terminated_length": 172.0, | |
| "epoch": 1.0756075607560756, | |
| "grad_norm": 0.11581874638795853, | |
| "learning_rate": 9.625722049390045e-07, | |
| "loss": 0.0092, | |
| "num_tokens": 62230690.0, | |
| "reward": 1.5838431358337401, | |
| "reward_std": 0.3170156180858612, | |
| "rewards/accuracy_reward/mean": 0.7247767925262452, | |
| "rewards/accuracy_reward/std": 0.4461586236953735, | |
| "rewards/cosine_scaled_reward/mean": 0.723936116695404, | |
| "rewards/cosine_scaled_reward/std": 0.4458396315574646, | |
| "rewards/format_reward/mean": 0.9941964268684387, | |
| "rewards/format_reward/std": 0.07385859042406082, | |
| "step": 85 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.010714285714285721, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 2015.2, | |
| "completions/mean_length": 713.357861328125, | |
| "completions/mean_terminated_length": 698.86904296875, | |
| "completions/min_length": 212.2, | |
| "completions/min_terminated_length": 212.2, | |
| "epoch": 1.1386138613861387, | |
| "grad_norm": 0.10194957256317139, | |
| "learning_rate": 9.537236226313388e-07, | |
| "loss": 0.0201, | |
| "num_tokens": 66250349.0, | |
| "reward": 1.5542301416397095, | |
| "reward_std": 0.33158175349235536, | |
| "rewards/accuracy_reward/mean": 0.7066964268684387, | |
| "rewards/accuracy_reward/std": 0.453377491235733, | |
| "rewards/cosine_scaled_reward/mean": 0.7060048222541809, | |
| "rewards/cosine_scaled_reward/std": 0.45305579900741577, | |
| "rewards/format_reward/mean": 0.989062488079071, | |
| "rewards/format_reward/std": 0.1036216139793396, | |
| "step": 90 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.00959821428571428, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1982.4, | |
| "completions/mean_length": 696.29912109375, | |
| "completions/mean_terminated_length": 683.2262084960937, | |
| "completions/min_length": 205.0, | |
| "completions/min_terminated_length": 205.0, | |
| "epoch": 1.2016201620162017, | |
| "grad_norm": 0.11849887669086456, | |
| "learning_rate": 9.439868550254868e-07, | |
| "loss": 0.0179, | |
| "num_tokens": 70144897.0, | |
| "reward": 1.5924715518951416, | |
| "reward_std": 0.3465422332286835, | |
| "rewards/accuracy_reward/mean": 0.7321428537368775, | |
| "rewards/accuracy_reward/std": 0.4421205699443817, | |
| "rewards/cosine_scaled_reward/mean": 0.7307018756866455, | |
| "rewards/cosine_scaled_reward/std": 0.44154341220855714, | |
| "rewards/format_reward/mean": 0.9899553656578064, | |
| "rewards/format_reward/std": 0.09479753226041794, | |
| "step": 95 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0046875, | |
| "completions/max_length": 2023.8, | |
| "completions/max_terminated_length": 1962.0, | |
| "completions/mean_length": 685.8980224609375, | |
| "completions/mean_terminated_length": 679.4762451171875, | |
| "completions/min_length": 209.0, | |
| "completions/min_terminated_length": 209.0, | |
| "epoch": 1.2646264626462647, | |
| "grad_norm": 0.11047360301017761, | |
| "learning_rate": 9.333809623012763e-07, | |
| "loss": 0.0137, | |
| "num_tokens": 73963520.0, | |
| "reward": 1.6014429330825806, | |
| "reward_std": 0.3224481403827667, | |
| "rewards/accuracy_reward/mean": 0.7363839387893677, | |
| "rewards/accuracy_reward/std": 0.4394433259963989, | |
| "rewards/cosine_scaled_reward/mean": 0.7348053455352783, | |
| "rewards/cosine_scaled_reward/std": 0.4390797853469849, | |
| "rewards/format_reward/mean": 0.995312511920929, | |
| "rewards/format_reward/std": 0.060620172321796416, | |
| "step": 100 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.006026785714285698, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1955.6, | |
| "completions/mean_length": 685.2373168945312, | |
| "completions/mean_terminated_length": 677.0635864257813, | |
| "completions/min_length": 165.2, | |
| "completions/min_terminated_length": 165.2, | |
| "epoch": 1.3276327632763276, | |
| "grad_norm": 0.11911053210496902, | |
| "learning_rate": 9.219267059917681e-07, | |
| "loss": 0.0179, | |
| "num_tokens": 77867639.0, | |
| "reward": 1.5839551210403442, | |
| "reward_std": 0.3121181547641754, | |
| "rewards/accuracy_reward/mean": 0.725, | |
| "rewards/accuracy_reward/std": 0.44613986015319823, | |
| "rewards/cosine_scaled_reward/mean": 0.7243833780288697, | |
| "rewards/cosine_scaled_reward/std": 0.4459415376186371, | |
| "rewards/format_reward/mean": 0.9935267806053162, | |
| "rewards/format_reward/std": 0.07562160044908524, | |
| "step": 105 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0049107142857142795, | |
| "completions/max_length": 1989.2, | |
| "completions/max_terminated_length": 1903.0, | |
| "completions/mean_length": 677.8884155273438, | |
| "completions/mean_terminated_length": 671.1434326171875, | |
| "completions/min_length": 192.4, | |
| "completions/min_terminated_length": 192.4, | |
| "epoch": 1.3906390639063906, | |
| "grad_norm": 0.12140108644962311, | |
| "learning_rate": 9.096465083415807e-07, | |
| "loss": 0.0172, | |
| "num_tokens": 81711467.0, | |
| "reward": 1.6193986415863038, | |
| "reward_std": 0.3027470111846924, | |
| "rewards/accuracy_reward/mean": 0.7484375, | |
| "rewards/accuracy_reward/std": 0.4324600279331207, | |
| "rewards/cosine_scaled_reward/mean": 0.7475025296211243, | |
| "rewards/cosine_scaled_reward/std": 0.43219619393348696, | |
| "rewards/format_reward/mean": 0.9944196462631225, | |
| "rewards/format_reward/std": 0.07256761714816093, | |
| "step": 110 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.00825892857142858, | |
| "completions/max_length": 2032.2, | |
| "completions/max_terminated_length": 1932.0, | |
| "completions/mean_length": 692.1948974609375, | |
| "completions/mean_terminated_length": 680.9073486328125, | |
| "completions/min_length": 187.8, | |
| "completions/min_terminated_length": 187.8, | |
| "epoch": 1.4536453645364538, | |
| "grad_norm": 0.13913048803806305, | |
| "learning_rate": 8.96564408414293e-07, | |
| "loss": 0.014, | |
| "num_tokens": 85604932.0, | |
| "reward": 1.5589273691177368, | |
| "reward_std": 0.3327997088432312, | |
| "rewards/accuracy_reward/mean": 0.7130610585212708, | |
| "rewards/accuracy_reward/std": 0.4512468934059143, | |
| "rewards/cosine_scaled_reward/mean": 0.70667005777359, | |
| "rewards/cosine_scaled_reward/std": 0.45351189374923706, | |
| "rewards/format_reward/mean": 0.9915178656578064, | |
| "rewards/format_reward/std": 0.07997382432222366, | |
| "step": 115 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.005357142857142861, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1993.8, | |
| "completions/mean_length": 657.3136474609375, | |
| "completions/mean_terminated_length": 649.76474609375, | |
| "completions/min_length": 174.0, | |
| "completions/min_terminated_length": 174.0, | |
| "epoch": 1.5166516651665165, | |
| "grad_norm": 0.11739910393953323, | |
| "learning_rate": 8.827060150348489e-07, | |
| "loss": 0.0121, | |
| "num_tokens": 89320697.0, | |
| "reward": 1.6037935495376587, | |
| "reward_std": 0.2636448621749878, | |
| "rewards/accuracy_reward/mean": 0.7381696462631225, | |
| "rewards/accuracy_reward/std": 0.4392316222190857, | |
| "rewards/cosine_scaled_reward/mean": 0.7377209424972534, | |
| "rewards/cosine_scaled_reward/std": 0.4390567898750305, | |
| "rewards/format_reward/mean": 0.9935267925262451, | |
| "rewards/format_reward/std": 0.07422313913702965, | |
| "step": 120 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.009821428571428559, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1931.4, | |
| "completions/mean_length": 687.1069580078125, | |
| "completions/mean_terminated_length": 673.6233276367187, | |
| "completions/min_length": 177.8, | |
| "completions/min_terminated_length": 177.8, | |
| "epoch": 1.5796579657965797, | |
| "grad_norm": 0.11269731819629669, | |
| "learning_rate": 8.680984566590792e-07, | |
| "loss": 0.0104, | |
| "num_tokens": 93191536.0, | |
| "reward": 1.6061813592910767, | |
| "reward_std": 0.3020260453224182, | |
| "rewards/accuracy_reward/mean": 0.7410714387893677, | |
| "rewards/accuracy_reward/std": 0.4377245783805847, | |
| "rewards/cosine_scaled_reward/mean": 0.7404875159263611, | |
| "rewards/cosine_scaled_reward/std": 0.43749932646751405, | |
| "rewards/format_reward/mean": 0.9897321462631226, | |
| "rewards/format_reward/std": 0.10028065890073776, | |
| "step": 125 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0078124999999999774, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1875.6, | |
| "completions/mean_length": 655.5540405273438, | |
| "completions/mean_terminated_length": 644.6221923828125, | |
| "completions/min_length": 189.6, | |
| "completions/min_terminated_length": 189.6, | |
| "epoch": 1.6426642664266426, | |
| "grad_norm": 0.13505014777183533, | |
| "learning_rate": 8.527703282684765e-07, | |
| "loss": 0.014, | |
| "num_tokens": 96899722.0, | |
| "reward": 1.6154436826705934, | |
| "reward_std": 0.30025253891944886, | |
| "rewards/accuracy_reward/mean": 0.7470982193946838, | |
| "rewards/accuracy_reward/std": 0.43418333530426023, | |
| "rewards/cosine_scaled_reward/mean": 0.7465121984481812, | |
| "rewards/cosine_scaled_reward/std": 0.43400861620903014, | |
| "rewards/format_reward/mean": 0.9901785850524902, | |
| "rewards/format_reward/std": 0.09706691503524781, | |
| "step": 130 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.00401785714285714, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1906.4, | |
| "completions/mean_length": 651.9241333007812, | |
| "completions/mean_terminated_length": 646.2785034179688, | |
| "completions/min_length": 181.4, | |
| "completions/min_terminated_length": 181.4, | |
| "epoch": 1.7056705670567056, | |
| "grad_norm": 0.1649669110774994, | |
| "learning_rate": 8.36751635394175e-07, | |
| "loss": 0.0071, | |
| "num_tokens": 100597958.0, | |
| "reward": 1.621180534362793, | |
| "reward_std": 0.27085027396678923, | |
| "rewards/accuracy_reward/mean": 0.7497767925262451, | |
| "rewards/accuracy_reward/std": 0.4303715705871582, | |
| "rewards/cosine_scaled_reward/mean": 0.7490574598312378, | |
| "rewards/cosine_scaled_reward/std": 0.43027289509773253, | |
| "rewards/format_reward/mean": 0.993749988079071, | |
| "rewards/format_reward/std": 0.07439046055078506, | |
| "step": 135 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.00625, | |
| "completions/max_length": 1999.2, | |
| "completions/max_terminated_length": 1840.8, | |
| "completions/mean_length": 629.5085205078125, | |
| "completions/mean_terminated_length": 620.7006591796875, | |
| "completions/min_length": 158.0, | |
| "completions/min_terminated_length": 158.0, | |
| "epoch": 1.7686768676867688, | |
| "grad_norm": 0.15633437037467957, | |
| "learning_rate": 8.200737353797148e-07, | |
| "loss": 0.0149, | |
| "num_tokens": 104226908.0, | |
| "reward": 1.6223199844360352, | |
| "reward_std": 0.28274320662021635, | |
| "rewards/accuracy_reward/mean": 0.7506696343421936, | |
| "rewards/accuracy_reward/std": 0.43105919361114503, | |
| "rewards/cosine_scaled_reward/mean": 0.7499970197677612, | |
| "rewards/cosine_scaled_reward/std": 0.43102300763130186, | |
| "rewards/format_reward/mean": 0.9933035612106323, | |
| "rewards/format_reward/std": 0.06884920671582222, | |
| "step": 140 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.004241071428571441, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1935.0, | |
| "completions/mean_length": 616.7654296875, | |
| "completions/mean_terminated_length": 610.688623046875, | |
| "completions/min_length": 171.4, | |
| "completions/min_terminated_length": 171.4, | |
| "epoch": 1.8316831683168315, | |
| "grad_norm": 0.16371622681617737, | |
| "learning_rate": 8.027692759975693e-07, | |
| "loss": 0.0091, | |
| "num_tokens": 107827689.0, | |
| "reward": 1.5669083833694457, | |
| "reward_std": 0.308967524766922, | |
| "rewards/accuracy_reward/mean": 0.7129464268684387, | |
| "rewards/accuracy_reward/std": 0.44984045028686526, | |
| "rewards/cosine_scaled_reward/mean": 0.7126113176345825, | |
| "rewards/cosine_scaled_reward/std": 0.44967737793922424, | |
| "rewards/format_reward/mean": 0.995312488079071, | |
| "rewards/format_reward/std": 0.06629456728696823, | |
| "step": 145 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.00379464285714286, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1910.4, | |
| "completions/mean_length": 605.9803833007812, | |
| "completions/mean_terminated_length": 600.4914794921875, | |
| "completions/min_length": 137.0, | |
| "completions/min_terminated_length": 137.0, | |
| "epoch": 1.8946894689468947, | |
| "grad_norm": 0.16951541602611542, | |
| "learning_rate": 7.848721315395974e-07, | |
| "loss": 0.0083, | |
| "num_tokens": 111359681.0, | |
| "reward": 1.6116993427276611, | |
| "reward_std": 0.3007046103477478, | |
| "rewards/accuracy_reward/mean": 0.7428571343421936, | |
| "rewards/accuracy_reward/std": 0.43679211139678953, | |
| "rewards/cosine_scaled_reward/mean": 0.7423717975616455, | |
| "rewards/cosine_scaled_reward/std": 0.4366111934185028, | |
| "rewards/format_reward/mean": 0.995312488079071, | |
| "rewards/format_reward/std": 0.06531310081481934, | |
| "step": 150 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.00401785714285714, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1928.0, | |
| "completions/mean_length": 610.7569458007813, | |
| "completions/mean_terminated_length": 604.954296875, | |
| "completions/min_length": 176.0, | |
| "completions/min_terminated_length": 176.0, | |
| "epoch": 1.9576957695769577, | |
| "grad_norm": 0.17909295856952667, | |
| "learning_rate": 7.664173365065265e-07, | |
| "loss": 0.0103, | |
| "num_tokens": 114894120.0, | |
| "reward": 1.6185973644256593, | |
| "reward_std": 0.29384649991989137, | |
| "rewards/accuracy_reward/mean": 0.7475446581840515, | |
| "rewards/accuracy_reward/std": 0.4323570549488068, | |
| "rewards/cosine_scaled_reward/mean": 0.7470160365104676, | |
| "rewards/cosine_scaled_reward/std": 0.4321545004844666, | |
| "rewards/format_reward/mean": 0.9950892806053162, | |
| "rewards/format_reward/std": 0.06607652679085732, | |
| "step": 155 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0036299410540915344, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1968.8, | |
| "completions/mean_length": 619.8013061523437, | |
| "completions/mean_terminated_length": 614.59990234375, | |
| "completions/min_length": 153.6, | |
| "completions/min_terminated_length": 153.6, | |
| "epoch": 2.025202520252025, | |
| "grad_norm": 0.14987502992153168, | |
| "learning_rate": 7.474410170262718e-07, | |
| "loss": 0.0109, | |
| "num_tokens": 118441067.0, | |
| "reward": 1.608110237121582, | |
| "reward_std": 0.2891895234584808, | |
| "rewards/accuracy_reward/mean": 0.7404017925262452, | |
| "rewards/accuracy_reward/std": 0.437438040971756, | |
| "rewards/cosine_scaled_reward/mean": 0.7396578192710876, | |
| "rewards/cosine_scaled_reward/std": 0.4371993124485016, | |
| "rewards/format_reward/mean": 0.9957589268684387, | |
| "rewards/format_reward/std": 0.0631766937673092, | |
| "step": 160 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.006026785714285721, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1911.4, | |
| "completions/mean_length": 626.1058227539063, | |
| "completions/mean_terminated_length": 617.5382690429688, | |
| "completions/min_length": 183.0, | |
| "completions/min_terminated_length": 183.0, | |
| "epoch": 2.0882088208820884, | |
| "grad_norm": 0.17082159221172333, | |
| "learning_rate": 7.279803201353449e-07, | |
| "loss": 0.0123, | |
| "num_tokens": 122031925.0, | |
| "reward": 1.6282689809799193, | |
| "reward_std": 0.2621865808963776, | |
| "rewards/accuracy_reward/mean": 0.754687511920929, | |
| "rewards/accuracy_reward/std": 0.42926387786865233, | |
| "rewards/cosine_scaled_reward/mean": 0.7534128069877625, | |
| "rewards/cosine_scaled_reward/std": 0.4291880249977112, | |
| "rewards/format_reward/mean": 0.99375, | |
| "rewards/format_reward/std": 0.07523396834731103, | |
| "step": 165 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.002678571428571419, | |
| "completions/max_length": 2017.2, | |
| "completions/max_terminated_length": 1854.8, | |
| "completions/mean_length": 626.8839477539062, | |
| "completions/mean_terminated_length": 623.02763671875, | |
| "completions/min_length": 191.6, | |
| "completions/min_terminated_length": 191.6, | |
| "epoch": 2.151215121512151, | |
| "grad_norm": 0.14197254180908203, | |
| "learning_rate": 7.080733410617851e-07, | |
| "loss": 0.0094, | |
| "num_tokens": 125637397.0, | |
| "reward": 1.6435120105743408, | |
| "reward_std": 0.2684548318386078, | |
| "rewards/accuracy_reward/mean": 0.7638392806053161, | |
| "rewards/accuracy_reward/std": 0.42319408655166624, | |
| "rewards/cosine_scaled_reward/mean": 0.7626935601234436, | |
| "rewards/cosine_scaled_reward/std": 0.4231561243534088, | |
| "rewards/format_reward/mean": 0.9966517925262451, | |
| "rewards/format_reward/std": 0.04984662607312203, | |
| "step": 170 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0033482142857143016, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1970.8, | |
| "completions/mean_length": 644.8076293945312, | |
| "completions/mean_terminated_length": 640.0797729492188, | |
| "completions/min_length": 151.6, | |
| "completions/min_terminated_length": 151.6, | |
| "epoch": 2.2142214221422143, | |
| "grad_norm": 0.16446265578269958, | |
| "learning_rate": 6.87759048651962e-07, | |
| "loss": 0.0102, | |
| "num_tokens": 129311063.0, | |
| "reward": 1.6173259973526002, | |
| "reward_std": 0.29779149889945983, | |
| "rewards/accuracy_reward/mean": 0.7466517806053161, | |
| "rewards/accuracy_reward/std": 0.43190850019454957, | |
| "rewards/cosine_scaled_reward/mean": 0.7460358500480652, | |
| "rewards/cosine_scaled_reward/std": 0.43167824745178224, | |
| "rewards/format_reward/mean": 0.9953125, | |
| "rewards/format_reward/std": 0.06686954125761986, | |
| "step": 175 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.005357142857142861, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1889.6, | |
| "completions/mean_length": 629.3937744140625, | |
| "completions/mean_terminated_length": 621.8263305664062, | |
| "completions/min_length": 170.8, | |
| "completions/min_terminated_length": 170.8, | |
| "epoch": 2.2772277227722775, | |
| "grad_norm": 0.13965024054050446, | |
| "learning_rate": 6.670772090872278e-07, | |
| "loss": 0.0114, | |
| "num_tokens": 132892795.0, | |
| "reward": 1.6336923837661743, | |
| "reward_std": 0.26696730852127076, | |
| "rewards/accuracy_reward/mean": 0.7592241644859314, | |
| "rewards/accuracy_reward/std": 0.424642938375473, | |
| "rewards/cosine_scaled_reward/mean": 0.7585270166397095, | |
| "rewards/cosine_scaled_reward/std": 0.4244441330432892, | |
| "rewards/format_reward/mean": 0.9946428418159485, | |
| "rewards/format_reward/std": 0.07060872837901115, | |
| "step": 180 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.00401785714285714, | |
| "completions/max_length": 1994.0, | |
| "completions/max_terminated_length": 1871.0, | |
| "completions/mean_length": 623.195556640625, | |
| "completions/mean_terminated_length": 617.4927368164062, | |
| "completions/min_length": 170.2, | |
| "completions/min_terminated_length": 170.2, | |
| "epoch": 2.34023402340234, | |
| "grad_norm": 0.13541771471500397, | |
| "learning_rate": 6.460683080397508e-07, | |
| "loss": 0.0108, | |
| "num_tokens": 136497919.0, | |
| "reward": 1.638669991493225, | |
| "reward_std": 0.2737217009067535, | |
| "rewards/accuracy_reward/mean": 0.7620797157287598, | |
| "rewards/accuracy_reward/std": 0.4253371238708496, | |
| "rewards/cosine_scaled_reward/mean": 0.7601523399353027, | |
| "rewards/cosine_scaled_reward/std": 0.42597259283065797, | |
| "rewards/format_reward/mean": 0.9957589387893677, | |
| "rewards/format_reward/std": 0.05754158571362496, | |
| "step": 185 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.004687500000000022, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1951.2, | |
| "completions/mean_length": 617.7080688476562, | |
| "completions/mean_terminated_length": 610.93955078125, | |
| "completions/min_length": 158.4, | |
| "completions/min_terminated_length": 158.4, | |
| "epoch": 2.4032403240324034, | |
| "grad_norm": 0.1562889963388443, | |
| "learning_rate": 6.247734714199105e-07, | |
| "loss": 0.018, | |
| "num_tokens": 140050915.0, | |
| "reward": 1.6390478134155273, | |
| "reward_std": 0.2830404281616211, | |
| "rewards/accuracy_reward/mean": 0.7611607193946839, | |
| "rewards/accuracy_reward/std": 0.4261560559272766, | |
| "rewards/cosine_scaled_reward/mean": 0.7606848120689392, | |
| "rewards/cosine_scaled_reward/std": 0.42603325843811035, | |
| "rewards/format_reward/mean": 0.995089304447174, | |
| "rewards/format_reward/std": 0.06784935072064399, | |
| "step": 190 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0049107142857142795, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1952.4, | |
| "completions/mean_length": 629.9161010742188, | |
| "completions/mean_terminated_length": 622.9476318359375, | |
| "completions/min_length": 170.0, | |
| "completions/min_terminated_length": 170.0, | |
| "epoch": 2.466246624662466, | |
| "grad_norm": 0.16037462651729584, | |
| "learning_rate": 6.032343848703969e-07, | |
| "loss": 0.0111, | |
| "num_tokens": 143659179.0, | |
| "reward": 1.6348266124725341, | |
| "reward_std": 0.2890359699726105, | |
| "rewards/accuracy_reward/mean": 0.7587053537368774, | |
| "rewards/accuracy_reward/std": 0.4270632088184357, | |
| "rewards/cosine_scaled_reward/mean": 0.7578226804733277, | |
| "rewards/cosine_scaled_reward/std": 0.4268454909324646, | |
| "rewards/format_reward/mean": 0.9944196343421936, | |
| "rewards/format_reward/std": 0.0683063194155693, | |
| "step": 195 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0046875, | |
| "completions/max_length": 2011.6, | |
| "completions/max_terminated_length": 1926.0, | |
| "completions/mean_length": 632.8710083007812, | |
| "completions/mean_terminated_length": 626.33193359375, | |
| "completions/min_length": 161.6, | |
| "completions/min_terminated_length": 161.6, | |
| "epoch": 2.5292529252925293, | |
| "grad_norm": 0.14154884219169617, | |
| "learning_rate": 5.814932121646073e-07, | |
| "loss": 0.0079, | |
| "num_tokens": 147280177.0, | |
| "reward": 1.6299890995025634, | |
| "reward_std": 0.27802737057209015, | |
| "rewards/accuracy_reward/mean": 0.7558035850524902, | |
| "rewards/accuracy_reward/std": 0.42795242071151735, | |
| "rewards/cosine_scaled_reward/mean": 0.7550673365592957, | |
| "rewards/cosine_scaled_reward/std": 0.4278149545192719, | |
| "rewards/format_reward/mean": 0.9933035731315613, | |
| "rewards/format_reward/std": 0.06978957653045655, | |
| "step": 200 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0051339285714286035, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1918.8, | |
| "completions/mean_length": 615.8734741210938, | |
| "completions/mean_terminated_length": 608.469384765625, | |
| "completions/min_length": 168.8, | |
| "completions/min_terminated_length": 168.8, | |
| "epoch": 2.592259225922592, | |
| "grad_norm": 0.12673905491828918, | |
| "learning_rate": 5.595925126690801e-07, | |
| "loss": 0.0128, | |
| "num_tokens": 150787250.0, | |
| "reward": 1.6726323127746583, | |
| "reward_std": 0.2540514647960663, | |
| "rewards/accuracy_reward/mean": 0.7837053537368774, | |
| "rewards/accuracy_reward/std": 0.40963547229766845, | |
| "rewards/cosine_scaled_reward/mean": 0.782987654209137, | |
| "rewards/cosine_scaled_reward/std": 0.40952636003494264, | |
| "rewards/format_reward/mean": 0.9948660850524902, | |
| "rewards/format_reward/std": 0.06850367933511733, | |
| "step": 205 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.003794642857142838, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1862.4, | |
| "completions/mean_length": 612.8245849609375, | |
| "completions/mean_terminated_length": 607.3509399414063, | |
| "completions/min_length": 166.8, | |
| "completions/min_terminated_length": 166.8, | |
| "epoch": 2.6552655265526552, | |
| "grad_norm": 0.12310470640659332, | |
| "learning_rate": 5.375751580315354e-07, | |
| "loss": 0.0103, | |
| "num_tokens": 154286136.0, | |
| "reward": 1.634846067428589, | |
| "reward_std": 0.25745274126529694, | |
| "rewards/accuracy_reward/mean": 0.7580357193946838, | |
| "rewards/accuracy_reward/std": 0.42683074474334715, | |
| "rewards/cosine_scaled_reward/mean": 0.7574151754379272, | |
| "rewards/cosine_scaled_reward/std": 0.42679900527000425, | |
| "rewards/format_reward/mean": 0.9962053537368775, | |
| "rewards/format_reward/std": 0.0598295733332634, | |
| "step": 210 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0035714285714285587, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1913.4, | |
| "completions/mean_length": 600.7576049804687, | |
| "completions/mean_terminated_length": 595.5764038085938, | |
| "completions/min_length": 178.4, | |
| "completions/min_terminated_length": 178.4, | |
| "epoch": 2.7182718271827184, | |
| "grad_norm": 0.12419384717941284, | |
| "learning_rate": 5.154842482576126e-07, | |
| "loss": 0.0126, | |
| "num_tokens": 157779418.0, | |
| "reward": 1.658141851425171, | |
| "reward_std": 0.259933602809906, | |
| "rewards/accuracy_reward/mean": 0.7734375, | |
| "rewards/accuracy_reward/std": 0.41526424288749697, | |
| "rewards/cosine_scaled_reward/mean": 0.7729800581932068, | |
| "rewards/cosine_scaled_reward/std": 0.4152462720870972, | |
| "rewards/format_reward/mean": 0.9964285612106323, | |
| "rewards/format_reward/std": 0.05749904364347458, | |
| "step": 215 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0040178571428571615, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1845.8, | |
| "completions/mean_length": 606.7765747070313, | |
| "completions/mean_terminated_length": 600.97294921875, | |
| "completions/min_length": 168.0, | |
| "completions/min_terminated_length": 168.0, | |
| "epoch": 2.781278127812781, | |
| "grad_norm": 0.16448700428009033, | |
| "learning_rate": 4.933630273405846e-07, | |
| "loss": 0.0162, | |
| "num_tokens": 161255897.0, | |
| "reward": 1.6536934852600098, | |
| "reward_std": 0.2780122488737106, | |
| "rewards/accuracy_reward/mean": 0.7709821581840515, | |
| "rewards/accuracy_reward/std": 0.41886167526245116, | |
| "rewards/cosine_scaled_reward/mean": 0.7701100707054138, | |
| "rewards/cosine_scaled_reward/std": 0.4186771154403687, | |
| "rewards/format_reward/mean": 0.9953125, | |
| "rewards/format_reward/std": 0.06595941931009293, | |
| "step": 220 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.002901785714285743, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 2013.6, | |
| "completions/mean_length": 597.0652099609375, | |
| "completions/mean_terminated_length": 592.8518188476562, | |
| "completions/min_length": 162.6, | |
| "completions/min_terminated_length": 162.6, | |
| "epoch": 2.8442844284428443, | |
| "grad_norm": 0.14639931917190552, | |
| "learning_rate": 4.712547986092099e-07, | |
| "loss": 0.0067, | |
| "num_tokens": 164732445.0, | |
| "reward": 1.6427595376968385, | |
| "reward_std": 0.27772180140018465, | |
| "rewards/accuracy_reward/mean": 0.7633928656578064, | |
| "rewards/accuracy_reward/std": 0.42306692004203794, | |
| "rewards/cosine_scaled_reward/mean": 0.7627511620521545, | |
| "rewards/cosine_scaled_reward/std": 0.4228369653224945, | |
| "rewards/format_reward/mean": 0.9959821462631225, | |
| "rewards/format_reward/std": 0.062224668264389035, | |
| "step": 225 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0008928571428571397, | |
| "completions/max_length": 1944.6, | |
| "completions/max_terminated_length": 1889.6, | |
| "completions/mean_length": 599.8812866210938, | |
| "completions/mean_terminated_length": 598.5843017578125, | |
| "completions/min_length": 169.6, | |
| "completions/min_terminated_length": 169.6, | |
| "epoch": 2.9072907290729075, | |
| "grad_norm": 0.1395055204629898, | |
| "learning_rate": 4.492028399594338e-07, | |
| "loss": 0.0043, | |
| "num_tokens": 168207073.0, | |
| "reward": 1.6468029737472534, | |
| "reward_std": 0.26623707115650175, | |
| "rewards/accuracy_reward/mean": 0.7649553537368774, | |
| "rewards/accuracy_reward/std": 0.42270631790161134, | |
| "rewards/cosine_scaled_reward/mean": 0.764587914943695, | |
| "rewards/cosine_scaled_reward/std": 0.42256171107292173, | |
| "rewards/format_reward/mean": 0.9991071343421936, | |
| "rewards/format_reward/std": 0.022806894034147263, | |
| "step": 230 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.006026785714285721, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1864.6, | |
| "completions/mean_length": 641.97099609375, | |
| "completions/mean_terminated_length": 633.473095703125, | |
| "completions/min_length": 186.6, | |
| "completions/min_terminated_length": 186.6, | |
| "epoch": 2.9702970297029703, | |
| "grad_norm": 0.12765838205814362, | |
| "learning_rate": 4.2725031913587426e-07, | |
| "loss": 0.0096, | |
| "num_tokens": 171879167.0, | |
| "reward": 1.6136202335357666, | |
| "reward_std": 0.26737120449543, | |
| "rewards/accuracy_reward/mean": 0.7448660731315613, | |
| "rewards/accuracy_reward/std": 0.43231003880500796, | |
| "rewards/cosine_scaled_reward/mean": 0.7442046284675599, | |
| "rewards/cosine_scaled_reward/std": 0.43226224184036255, | |
| "rewards/format_reward/mean": 0.9933035612106323, | |
| "rewards/format_reward/std": 0.07773758172988891, | |
| "step": 235 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.005047243411927904, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1930.4, | |
| "completions/mean_length": 646.4539672851563, | |
| "completions/mean_terminated_length": 639.4364501953125, | |
| "completions/min_length": 183.2, | |
| "completions/min_terminated_length": 183.2, | |
| "epoch": 3.037803780378038, | |
| "grad_norm": 0.1862199455499649, | |
| "learning_rate": 4.054402092289354e-07, | |
| "loss": 0.0103, | |
| "num_tokens": 175632429.0, | |
| "reward": 1.5804502010345458, | |
| "reward_std": 0.29361810684204104, | |
| "rewards/accuracy_reward/mean": 0.7220982074737549, | |
| "rewards/accuracy_reward/std": 0.44716750979423525, | |
| "rewards/cosine_scaled_reward/mean": 0.7216144323348999, | |
| "rewards/cosine_scaled_reward/std": 0.44697086215019227, | |
| "rewards/format_reward/mean": 0.9950892925262451, | |
| "rewards/format_reward/std": 0.06746494099497795, | |
| "step": 240 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0033482142857142795, | |
| "completions/max_length": 2020.6, | |
| "completions/max_terminated_length": 1812.2, | |
| "completions/mean_length": 615.8056030273438, | |
| "completions/mean_terminated_length": 611.0432861328125, | |
| "completions/min_length": 194.2, | |
| "completions/min_terminated_length": 194.2, | |
| "epoch": 3.1008100810081007, | |
| "grad_norm": 0.16215628385543823, | |
| "learning_rate": 3.838152045529636e-07, | |
| "loss": 0.0143, | |
| "num_tokens": 179124806.0, | |
| "reward": 1.6801328897476195, | |
| "reward_std": 0.2461833655834198, | |
| "rewards/accuracy_reward/mean": 0.7881696343421936, | |
| "rewards/accuracy_reward/std": 0.4080040454864502, | |
| "rewards/cosine_scaled_reward/mean": 0.7877209305763244, | |
| "rewards/cosine_scaled_reward/std": 0.4078561782836914, | |
| "rewards/format_reward/mean": 0.9962053537368775, | |
| "rewards/format_reward/std": 0.05407858937978745, | |
| "step": 245 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.002678571428571419, | |
| "completions/max_length": 2029.4, | |
| "completions/max_terminated_length": 1823.4, | |
| "completions/mean_length": 606.2763671875, | |
| "completions/mean_terminated_length": 602.3788940429688, | |
| "completions/min_length": 189.6, | |
| "completions/min_terminated_length": 189.6, | |
| "epoch": 3.163816381638164, | |
| "grad_norm": 0.14493921399116516, | |
| "learning_rate": 3.624176370701235e-07, | |
| "loss": 0.0047, | |
| "num_tokens": 182588044.0, | |
| "reward": 1.6811640739440918, | |
| "reward_std": 0.24738809168338777, | |
| "rewards/accuracy_reward/mean": 0.7888392806053162, | |
| "rewards/accuracy_reward/std": 0.4081935465335846, | |
| "rewards/cosine_scaled_reward/mean": 0.7882209300994873, | |
| "rewards/cosine_scaled_reward/std": 0.40814857482910155, | |
| "rewards/format_reward/mean": 0.9964285731315613, | |
| "rewards/format_reward/std": 0.053148042410612106, | |
| "step": 250 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0022321428571428605, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1797.2, | |
| "completions/mean_length": 610.3625122070313, | |
| "completions/mean_terminated_length": 607.1727905273438, | |
| "completions/min_length": 166.6, | |
| "completions/min_terminated_length": 166.6, | |
| "epoch": 3.2268226822682267, | |
| "grad_norm": 0.12905967235565186, | |
| "learning_rate": 3.412893935235936e-07, | |
| "loss": 0.0075, | |
| "num_tokens": 186106108.0, | |
| "reward": 1.664694619178772, | |
| "reward_std": 0.2503175616264343, | |
| "rewards/accuracy_reward/mean": 0.7774553537368775, | |
| "rewards/accuracy_reward/std": 0.41434150338172915, | |
| "rewards/cosine_scaled_reward/mean": 0.7767105340957642, | |
| "rewards/cosine_scaled_reward/std": 0.4142517387866974, | |
| "rewards/format_reward/mean": 0.9977678418159485, | |
| "rewards/format_reward/std": 0.04592652022838593, | |
| "step": 255 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.003571428571428581, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1915.6, | |
| "completions/mean_length": 619.6020385742188, | |
| "completions/mean_terminated_length": 614.4874267578125, | |
| "completions/min_length": 146.0, | |
| "completions/min_terminated_length": 146.0, | |
| "epoch": 3.28982898289829, | |
| "grad_norm": 0.1830361932516098, | |
| "learning_rate": 3.204718334422998e-07, | |
| "loss": 0.0095, | |
| "num_tokens": 189646853.0, | |
| "reward": 1.6342453956604004, | |
| "reward_std": 0.30342464447021483, | |
| "rewards/accuracy_reward/mean": 0.7575892806053162, | |
| "rewards/accuracy_reward/std": 0.42738271951675416, | |
| "rewards/cosine_scaled_reward/mean": 0.7568834662437439, | |
| "rewards/cosine_scaled_reward/std": 0.4271435678005219, | |
| "rewards/format_reward/mean": 0.9964285731315613, | |
| "rewards/format_reward/std": 0.059244754165410994, | |
| "step": 260 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0020089285714285808, | |
| "completions/max_length": 1969.0, | |
| "completions/max_terminated_length": 1903.2, | |
| "completions/mean_length": 618.0531494140625, | |
| "completions/mean_terminated_length": 615.2063110351562, | |
| "completions/min_length": 160.0, | |
| "completions/min_terminated_length": 160.0, | |
| "epoch": 3.352835283528353, | |
| "grad_norm": 0.13739930093288422, | |
| "learning_rate": 3.0000570817769464e-07, | |
| "loss": 0.0078, | |
| "num_tokens": 193209859.0, | |
| "reward": 1.6563403844833373, | |
| "reward_std": 0.2727595895528793, | |
| "rewards/accuracy_reward/mean": 0.7716517925262452, | |
| "rewards/accuracy_reward/std": 0.4175321161746979, | |
| "rewards/cosine_scaled_reward/mean": 0.7711627960205079, | |
| "rewards/cosine_scaled_reward/std": 0.4173714339733124, | |
| "rewards/format_reward/mean": 0.9982142806053161, | |
| "rewards/format_reward/std": 0.031032296270132063, | |
| "step": 265 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.002678571428571441, | |
| "completions/max_length": 1993.2, | |
| "completions/max_terminated_length": 1915.8, | |
| "completions/mean_length": 627.6781494140625, | |
| "completions/mean_terminated_length": 623.89052734375, | |
| "completions/min_length": 156.4, | |
| "completions/min_terminated_length": 156.4, | |
| "epoch": 3.4158415841584158, | |
| "grad_norm": 0.14828138053417206, | |
| "learning_rate": 2.7993108113107443e-07, | |
| "loss": 0.0066, | |
| "num_tokens": 196821017.0, | |
| "reward": 1.6634482383728026, | |
| "reward_std": 0.23966350257396699, | |
| "rewards/accuracy_reward/mean": 0.7767857193946839, | |
| "rewards/accuracy_reward/std": 0.41428372263908386, | |
| "rewards/cosine_scaled_reward/mean": 0.7762267231941223, | |
| "rewards/cosine_scaled_reward/std": 0.41410806179046633, | |
| "rewards/format_reward/mean": 0.9970982074737549, | |
| "rewards/format_reward/std": 0.0474703922867775, | |
| "step": 270 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.00401785714285714, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1884.6, | |
| "completions/mean_length": 609.0343994140625, | |
| "completions/mean_terminated_length": 603.2148071289063, | |
| "completions/min_length": 169.0, | |
| "completions/min_terminated_length": 169.0, | |
| "epoch": 3.478847884788479, | |
| "grad_norm": 0.1737632006406784, | |
| "learning_rate": 2.602872493275881e-07, | |
| "loss": 0.0051, | |
| "num_tokens": 200339011.0, | |
| "reward": 1.6778597354888916, | |
| "reward_std": 0.25451319515705106, | |
| "rewards/accuracy_reward/mean": 0.7868303537368775, | |
| "rewards/accuracy_reward/std": 0.4092303454875946, | |
| "rewards/cosine_scaled_reward/mean": 0.7860764384269714, | |
| "rewards/cosine_scaled_reward/std": 0.4091233015060425, | |
| "rewards/format_reward/mean": 0.9959821343421936, | |
| "rewards/format_reward/std": 0.056319963186979294, | |
| "step": 275 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0024553571428571398, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1807.8, | |
| "completions/mean_length": 627.6569458007813, | |
| "completions/mean_terminated_length": 624.14736328125, | |
| "completions/min_length": 171.6, | |
| "completions/min_terminated_length": 171.6, | |
| "epoch": 3.5418541854185417, | |
| "grad_norm": 0.14920973777770996, | |
| "learning_rate": 2.411126664904665e-07, | |
| "loss": 0.0122, | |
| "num_tokens": 203921794.0, | |
| "reward": 1.640373420715332, | |
| "reward_std": 0.27687210142612456, | |
| "rewards/accuracy_reward/mean": 0.7618303537368775, | |
| "rewards/accuracy_reward/std": 0.424312424659729, | |
| "rewards/cosine_scaled_reward/mean": 0.7613270163536072, | |
| "rewards/cosine_scaled_reward/std": 0.42429112195968627, | |
| "rewards/format_reward/mean": 0.9957589149475098, | |
| "rewards/format_reward/std": 0.06131577789783478, | |
| "step": 280 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0033482142857142795, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1812.4, | |
| "completions/mean_length": 599.8631958007812, | |
| "completions/mean_terminated_length": 595.0171997070313, | |
| "completions/min_length": 178.0, | |
| "completions/min_terminated_length": 178.0, | |
| "epoch": 3.604860486048605, | |
| "grad_norm": 0.16235579550266266, | |
| "learning_rate": 2.2244486776605241e-07, | |
| "loss": 0.0082, | |
| "num_tokens": 207393885.0, | |
| "reward": 1.6748189449310302, | |
| "reward_std": 0.25130158960819243, | |
| "rewards/accuracy_reward/mean": 0.784374988079071, | |
| "rewards/accuracy_reward/std": 0.4097891986370087, | |
| "rewards/cosine_scaled_reward/mean": 0.7840126991271973, | |
| "rewards/cosine_scaled_reward/std": 0.40970866084098817, | |
| "rewards/format_reward/mean": 0.996875, | |
| "rewards/format_reward/std": 0.05381677523255348, | |
| "step": 285 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0020089285714285808, | |
| "completions/max_length": 1973.4, | |
| "completions/max_terminated_length": 1706.2, | |
| "completions/mean_length": 593.27236328125, | |
| "completions/mean_terminated_length": 590.3545043945312, | |
| "completions/min_length": 170.4, | |
| "completions/min_terminated_length": 170.4, | |
| "epoch": 3.667866786678668, | |
| "grad_norm": 0.14174528419971466, | |
| "learning_rate": 2.0432039624699015e-07, | |
| "loss": 0.0072, | |
| "num_tokens": 210886881.0, | |
| "reward": 1.6687217950820923, | |
| "reward_std": 0.2413029581308365, | |
| "rewards/accuracy_reward/mean": 0.7799107074737549, | |
| "rewards/accuracy_reward/std": 0.41347917914390564, | |
| "rewards/cosine_scaled_reward/mean": 0.779630982875824, | |
| "rewards/cosine_scaled_reward/std": 0.4133819997310638, | |
| "rewards/format_reward/mean": 0.9979910731315613, | |
| "rewards/format_reward/std": 0.038909842073917386, | |
| "step": 290 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.004241071428571419, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1969.2, | |
| "completions/mean_length": 635.64912109375, | |
| "completions/mean_terminated_length": 629.61572265625, | |
| "completions/min_length": 174.6, | |
| "completions/min_terminated_length": 174.6, | |
| "epoch": 3.730873087308731, | |
| "grad_norm": 0.13457916676998138, | |
| "learning_rate": 1.867747314374041e-07, | |
| "loss": 0.014, | |
| "num_tokens": 214550253.0, | |
| "reward": 1.6441724061965943, | |
| "reward_std": 0.25683614015579226, | |
| "rewards/accuracy_reward/mean": 0.7647321343421936, | |
| "rewards/accuracy_reward/std": 0.42399376034736636, | |
| "rewards/cosine_scaled_reward/mean": 0.7633446335792542, | |
| "rewards/cosine_scaled_reward/std": 0.4237023591995239, | |
| "rewards/format_reward/mean": 0.9955357193946839, | |
| "rewards/format_reward/std": 0.06432178840041161, | |
| "step": 295 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.006026785714285698, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1899.4, | |
| "completions/mean_length": 625.6288208007812, | |
| "completions/mean_terminated_length": 617.0573120117188, | |
| "completions/min_length": 183.8, | |
| "completions/min_terminated_length": 183.8, | |
| "epoch": 3.793879387938794, | |
| "grad_norm": 0.16838262975215912, | |
| "learning_rate": 1.6984221980010454e-07, | |
| "loss": 0.0107, | |
| "num_tokens": 218094638.0, | |
| "reward": 1.6642211198806762, | |
| "reward_std": 0.2543728470802307, | |
| "rewards/accuracy_reward/mean": 0.7783482193946838, | |
| "rewards/accuracy_reward/std": 0.41493446230888364, | |
| "rewards/cosine_scaled_reward/mean": 0.7779957056045532, | |
| "rewards/cosine_scaled_reward/std": 0.41481839418411254, | |
| "rewards/format_reward/mean": 0.993749988079071, | |
| "rewards/format_reward/std": 0.07161349579691886, | |
| "step": 300 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.004464285714285721, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1934.0, | |
| "completions/mean_length": 627.9725708007812, | |
| "completions/mean_terminated_length": 621.595361328125, | |
| "completions/min_length": 186.0, | |
| "completions/min_terminated_length": 186.0, | |
| "epoch": 3.856885688568857, | |
| "grad_norm": 0.14313475787639618, | |
| "learning_rate": 1.5355600752177312e-07, | |
| "loss": 0.0125, | |
| "num_tokens": 221700083.0, | |
| "reward": 1.6535644292831422, | |
| "reward_std": 0.2506650030612946, | |
| "rewards/accuracy_reward/mean": 0.7707589268684387, | |
| "rewards/accuracy_reward/std": 0.4196486592292786, | |
| "rewards/cosine_scaled_reward/mean": 0.7700751662254334, | |
| "rewards/cosine_scaled_reward/std": 0.41945862770080566, | |
| "rewards/format_reward/mean": 0.9955357074737549, | |
| "rewards/format_reward/std": 0.06590693667531014, | |
| "step": 305 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.003125, | |
| "completions/max_length": 2014.6, | |
| "completions/max_terminated_length": 1926.0, | |
| "completions/mean_length": 613.800927734375, | |
| "completions/mean_terminated_length": 609.321435546875, | |
| "completions/min_length": 179.8, | |
| "completions/min_terminated_length": 179.8, | |
| "epoch": 3.91989198919892, | |
| "grad_norm": 0.13859793543815613, | |
| "learning_rate": 1.3794797562774596e-07, | |
| "loss": 0.0068, | |
| "num_tokens": 225211495.0, | |
| "reward": 1.6338534832000733, | |
| "reward_std": 0.2609659105539322, | |
| "rewards/accuracy_reward/mean": 0.7595358729362488, | |
| "rewards/accuracy_reward/std": 0.4260308504104614, | |
| "rewards/cosine_scaled_reward/mean": 0.7553341269493103, | |
| "rewards/cosine_scaled_reward/std": 0.4275141417980194, | |
| "rewards/format_reward/mean": 0.9970982074737549, | |
| "rewards/format_reward/std": 0.0464889258146286, | |
| "step": 310 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.00625, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1900.4, | |
| "completions/mean_length": 611.440869140625, | |
| "completions/mean_terminated_length": 602.3813110351563, | |
| "completions/min_length": 181.0, | |
| "completions/min_terminated_length": 181.0, | |
| "epoch": 3.9828982898289826, | |
| "grad_norm": 0.1344553679227829, | |
| "learning_rate": 1.230486775734097e-07, | |
| "loss": 0.013, | |
| "num_tokens": 228791646.0, | |
| "reward": 1.6422872066497802, | |
| "reward_std": 0.23485693633556365, | |
| "rewards/accuracy_reward/mean": 0.7667692184448243, | |
| "rewards/accuracy_reward/std": 0.42179493308067323, | |
| "rewards/cosine_scaled_reward/mean": 0.761570417881012, | |
| "rewards/cosine_scaled_reward/std": 0.4240230083465576, | |
| "rewards/format_reward/mean": 0.9935267806053162, | |
| "rewards/format_reward/std": 0.07590627297759056, | |
| "step": 315 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0031640083217753024, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1952.0, | |
| "completions/mean_length": 617.7764038085937, | |
| "completions/mean_terminated_length": 613.2669799804687, | |
| "completions/min_length": 164.4, | |
| "completions/min_terminated_length": 164.4, | |
| "epoch": 4.05040504050405, | |
| "grad_norm": 0.15917585790157318, | |
| "learning_rate": 1.0888727943437748e-07, | |
| "loss": 0.0084, | |
| "num_tokens": 232312551.0, | |
| "reward": 1.6375758409500123, | |
| "reward_std": 0.2651157915592194, | |
| "rewards/accuracy_reward/mean": 0.7629625201225281, | |
| "rewards/accuracy_reward/std": 0.42385138273239137, | |
| "rewards/cosine_scaled_reward/mean": 0.7563670516014099, | |
| "rewards/cosine_scaled_reward/std": 0.4267165422439575, | |
| "rewards/format_reward/mean": 0.996875, | |
| "rewards/format_reward/std": 0.052468103170394895, | |
| "step": 320 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0022321428571428826, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1919.4, | |
| "completions/mean_length": 608.4109619140625, | |
| "completions/mean_terminated_length": 605.1753173828125, | |
| "completions/min_length": 191.8, | |
| "completions/min_terminated_length": 191.8, | |
| "epoch": 4.113411341134113, | |
| "grad_norm": 0.1608886420726776, | |
| "learning_rate": 9.549150281252632e-08, | |
| "loss": 0.0043, | |
| "num_tokens": 235787696.0, | |
| "reward": 1.66022846698761, | |
| "reward_std": 0.2489331156015396, | |
| "rewards/accuracy_reward/mean": 0.7745535850524903, | |
| "rewards/accuracy_reward/std": 0.41753928661346434, | |
| "rewards/cosine_scaled_reward/mean": 0.7740282416343689, | |
| "rewards/cosine_scaled_reward/std": 0.4174140512943268, | |
| "rewards/format_reward/mean": 0.9973214268684387, | |
| "rewards/format_reward/std": 0.05046965554356575, | |
| "step": 325 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.003794642857142838, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1945.6, | |
| "completions/mean_length": 622.467431640625, | |
| "completions/mean_terminated_length": 617.0385375976563, | |
| "completions/min_length": 152.6, | |
| "completions/min_terminated_length": 152.6, | |
| "epoch": 4.176417641764177, | |
| "grad_norm": 0.15062691271305084, | |
| "learning_rate": 8.288757056966017e-08, | |
| "loss": 0.0112, | |
| "num_tokens": 239378142.0, | |
| "reward": 1.6499414920806885, | |
| "reward_std": 0.2437061160802841, | |
| "rewards/accuracy_reward/mean": 0.7680803537368774, | |
| "rewards/accuracy_reward/std": 0.4208991706371307, | |
| "rewards/cosine_scaled_reward/mean": 0.7675167083740234, | |
| "rewards/cosine_scaled_reward/std": 0.4206974029541016, | |
| "rewards/format_reward/mean": 0.9962053418159484, | |
| "rewards/format_reward/std": 0.05708769783377647, | |
| "step": 330 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.005133928571428558, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1929.2, | |
| "completions/mean_length": 633.4506958007812, | |
| "completions/mean_terminated_length": 626.1871459960937, | |
| "completions/min_length": 175.4, | |
| "completions/min_terminated_length": 175.4, | |
| "epoch": 4.2394239423942395, | |
| "grad_norm": 0.14156123995780945, | |
| "learning_rate": 7.110015549502546e-08, | |
| "loss": 0.0082, | |
| "num_tokens": 243021681.0, | |
| "reward": 1.6500544548034668, | |
| "reward_std": 0.26026159822940825, | |
| "rewards/accuracy_reward/mean": 0.768749988079071, | |
| "rewards/accuracy_reward/std": 0.42072436213493347, | |
| "rewards/cosine_scaled_reward/mean": 0.7679658651351928, | |
| "rewards/cosine_scaled_reward/std": 0.4205094575881958, | |
| "rewards/format_reward/mean": 0.9946428418159485, | |
| "rewards/format_reward/std": 0.06923034116625786, | |
| "step": 335 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0024553571428571177, | |
| "completions/max_length": 1975.6, | |
| "completions/max_terminated_length": 1895.2, | |
| "completions/mean_length": 627.3937744140625, | |
| "completions/mean_terminated_length": 623.917529296875, | |
| "completions/min_length": 161.8, | |
| "completions/min_terminated_length": 161.8, | |
| "epoch": 4.302430243024302, | |
| "grad_norm": 0.13976019620895386, | |
| "learning_rate": 6.015233200716874e-08, | |
| "loss": 0.0064, | |
| "num_tokens": 246632541.0, | |
| "reward": 1.6928518533706665, | |
| "reward_std": 0.2341456115245819, | |
| "rewards/accuracy_reward/mean": 0.7964285731315612, | |
| "rewards/accuracy_reward/std": 0.40205827355384827, | |
| "rewards/cosine_scaled_reward/mean": 0.7953016996383667, | |
| "rewards/cosine_scaled_reward/std": 0.40191030502319336, | |
| "rewards/format_reward/mean": 0.9975446343421936, | |
| "rewards/format_reward/std": 0.04161064326763153, | |
| "step": 340 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.005133928571428581, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1957.8, | |
| "completions/mean_length": 618.9397705078125, | |
| "completions/mean_terminated_length": 611.7075561523437, | |
| "completions/min_length": 167.0, | |
| "completions/min_terminated_length": 167.0, | |
| "epoch": 4.365436543654366, | |
| "grad_norm": 0.15124908089637756, | |
| "learning_rate": 5.006553098467764e-08, | |
| "loss": 0.0103, | |
| "num_tokens": 250178143.0, | |
| "reward": 1.6755474090576172, | |
| "reward_std": 0.2548744171857834, | |
| "rewards/accuracy_reward/mean": 0.7857142925262451, | |
| "rewards/accuracy_reward/std": 0.4092557728290558, | |
| "rewards/cosine_scaled_reward/mean": 0.7848000407218934, | |
| "rewards/cosine_scaled_reward/std": 0.4092385172843933, | |
| "rewards/format_reward/mean": 0.9948660850524902, | |
| "rewards/format_reward/std": 0.06760494336485863, | |
| "step": 345 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.004464285714285721, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1865.8, | |
| "completions/mean_length": 626.4183349609375, | |
| "completions/mean_terminated_length": 620.0564453125, | |
| "completions/min_length": 182.0, | |
| "completions/min_terminated_length": 182.0, | |
| "epoch": 4.428442844284429, | |
| "grad_norm": 0.14921754598617554, | |
| "learning_rate": 4.085949781423048e-08, | |
| "loss": 0.007, | |
| "num_tokens": 253769705.0, | |
| "reward": 1.6789589166641234, | |
| "reward_std": 0.23221377432346343, | |
| "rewards/accuracy_reward/mean": 0.7877232074737549, | |
| "rewards/accuracy_reward/std": 0.40675231218338015, | |
| "rewards/cosine_scaled_reward/mean": 0.7873820304870606, | |
| "rewards/cosine_scaled_reward/std": 0.4066896378993988, | |
| "rewards/format_reward/mean": 0.9950892806053162, | |
| "rewards/format_reward/std": 0.06925405710935592, | |
| "step": 350 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.002678571428571419, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1934.2, | |
| "completions/mean_length": 618.8587280273438, | |
| "completions/mean_terminated_length": 615.0157958984375, | |
| "completions/min_length": 160.0, | |
| "completions/min_terminated_length": 160.0, | |
| "epoch": 4.491449144914491, | |
| "grad_norm": 0.16535188257694244, | |
| "learning_rate": 3.2552253738074074e-08, | |
| "loss": 0.008, | |
| "num_tokens": 257334160.0, | |
| "reward": 1.6443732500076294, | |
| "reward_std": 0.2480471521615982, | |
| "rewards/accuracy_reward/mean": 0.7647321343421936, | |
| "rewards/accuracy_reward/std": 0.423707377910614, | |
| "rewards/cosine_scaled_reward/mean": 0.7641927599906921, | |
| "rewards/cosine_scaled_reward/std": 0.4235422372817993, | |
| "rewards/format_reward/mean": 0.9950892925262451, | |
| "rewards/format_reward/std": 0.06574137881398201, | |
| "step": 355 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0024553571428571398, | |
| "completions/max_length": 1972.6, | |
| "completions/max_terminated_length": 1843.2, | |
| "completions/mean_length": 613.258740234375, | |
| "completions/mean_terminated_length": 609.7564453125, | |
| "completions/min_length": 191.4, | |
| "completions/min_terminated_length": 191.4, | |
| "epoch": 4.554455445544555, | |
| "grad_norm": 0.14730912446975708, | |
| "learning_rate": 2.5160060576596365e-08, | |
| "loss": 0.0072, | |
| "num_tokens": 260865495.0, | |
| "reward": 1.6269155740737915, | |
| "reward_std": 0.2688676297664642, | |
| "rewards/accuracy_reward/mean": 0.7524553418159485, | |
| "rewards/accuracy_reward/std": 0.42926363945007323, | |
| "rewards/cosine_scaled_reward/mean": 0.7520452857017517, | |
| "rewards/cosine_scaled_reward/std": 0.4292009472846985, | |
| "rewards/format_reward/mean": 0.996874988079071, | |
| "rewards/format_reward/std": 0.053170456737279895, | |
| "step": 360 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0029017857142857205, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1827.2, | |
| "completions/mean_length": 618.671240234375, | |
| "completions/mean_terminated_length": 614.5585571289063, | |
| "completions/min_length": 169.0, | |
| "completions/min_terminated_length": 169.0, | |
| "epoch": 4.617461746174618, | |
| "grad_norm": 0.1736821085214615, | |
| "learning_rate": 1.8697388895048515e-08, | |
| "loss": 0.01, | |
| "num_tokens": 264474382.0, | |
| "reward": 1.6391252517700194, | |
| "reward_std": 0.25250489115715025, | |
| "rewards/accuracy_reward/mean": 0.760937488079071, | |
| "rewards/accuracy_reward/std": 0.4259650409221649, | |
| "rewards/cosine_scaled_reward/mean": 0.7601699948310852, | |
| "rewards/cosine_scaled_reward/std": 0.42582063674926757, | |
| "rewards/format_reward/mean": 0.9962053656578064, | |
| "rewards/format_reward/std": 0.05821031853556633, | |
| "step": 365 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0013392857142857206, | |
| "completions/max_length": 1962.0, | |
| "completions/max_terminated_length": 1836.8, | |
| "completions/mean_length": 595.1326293945312, | |
| "completions/mean_terminated_length": 593.2115600585937, | |
| "completions/min_length": 174.4, | |
| "completions/min_terminated_length": 174.4, | |
| "epoch": 4.68046804680468, | |
| "grad_norm": 0.16435028612613678, | |
| "learning_rate": 1.3176889676734692e-08, | |
| "loss": 0.0088, | |
| "num_tokens": 267903232.0, | |
| "reward": 1.6919838905334472, | |
| "reward_std": 0.2522643059492111, | |
| "rewards/accuracy_reward/mean": 0.7953125, | |
| "rewards/accuracy_reward/std": 0.40310133099555967, | |
| "rewards/cosine_scaled_reward/mean": 0.7946819305419922, | |
| "rewards/cosine_scaled_reward/std": 0.40297635197639464, | |
| "rewards/format_reward/mean": 0.9986607074737549, | |
| "rewards/format_reward/std": 0.027685175836086272, | |
| "step": 370 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.00379464285714286, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1986.6, | |
| "completions/mean_length": 633.7323852539063, | |
| "completions/mean_terminated_length": 628.3538696289063, | |
| "completions/min_length": 181.8, | |
| "completions/min_terminated_length": 181.8, | |
| "epoch": 4.743474347434743, | |
| "grad_norm": 0.16870158910751343, | |
| "learning_rate": 8.60936955811753e-09, | |
| "loss": 0.0053, | |
| "num_tokens": 271506625.0, | |
| "reward": 1.677209448814392, | |
| "reward_std": 0.24304699897766113, | |
| "rewards/accuracy_reward/mean": 0.7866071343421936, | |
| "rewards/accuracy_reward/std": 0.4092015206813812, | |
| "rewards/cosine_scaled_reward/mean": 0.7849991202354432, | |
| "rewards/cosine_scaled_reward/std": 0.409130734205246, | |
| "rewards/format_reward/mean": 0.9962053537368775, | |
| "rewards/format_reward/std": 0.058644138276576996, | |
| "step": 375 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.003125, | |
| "completions/max_length": 2033.8, | |
| "completions/max_terminated_length": 1926.0, | |
| "completions/mean_length": 596.5832885742187, | |
| "completions/mean_terminated_length": 592.036572265625, | |
| "completions/min_length": 161.2, | |
| "completions/min_terminated_length": 161.2, | |
| "epoch": 4.806480648064807, | |
| "grad_norm": 0.18181964755058289, | |
| "learning_rate": 5.003769674320368e-09, | |
| "loss": 0.011, | |
| "num_tokens": 274925870.0, | |
| "reward": 1.6703053951263427, | |
| "reward_std": 0.24658992886543274, | |
| "rewards/accuracy_reward/mean": 0.7816964268684388, | |
| "rewards/accuracy_reward/std": 0.41211633682250975, | |
| "rewards/cosine_scaled_reward/mean": 0.7807891845703125, | |
| "rewards/cosine_scaled_reward/std": 0.4120659828186035, | |
| "rewards/format_reward/mean": 0.9964285731315613, | |
| "rewards/format_reward/std": 0.05088246762752533, | |
| "step": 380 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.00401785714285714, | |
| "completions/max_length": 2025.6, | |
| "completions/max_terminated_length": 1959.0, | |
| "completions/mean_length": 629.3553833007812, | |
| "completions/mean_terminated_length": 623.6705200195313, | |
| "completions/min_length": 182.2, | |
| "completions/min_terminated_length": 182.2, | |
| "epoch": 4.8694869486948695, | |
| "grad_norm": 0.15454868972301483, | |
| "learning_rate": 2.3671481564347396e-09, | |
| "loss": 0.0112, | |
| "num_tokens": 278501238.0, | |
| "reward": 1.6728654146194457, | |
| "reward_std": 0.24723841547966002, | |
| "rewards/accuracy_reward/mean": 0.7837053537368774, | |
| "rewards/accuracy_reward/std": 0.41015875339508057, | |
| "rewards/cosine_scaled_reward/mean": 0.7823377847671509, | |
| "rewards/cosine_scaled_reward/std": 0.40985342860221863, | |
| "rewards/format_reward/mean": 0.9959821343421936, | |
| "rewards/format_reward/std": 0.04546051248908043, | |
| "step": 385 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.0033482142857142795, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1895.4, | |
| "completions/mean_length": 639.8073974609375, | |
| "completions/mean_terminated_length": 635.0597290039062, | |
| "completions/min_length": 183.0, | |
| "completions/min_terminated_length": 183.0, | |
| "epoch": 4.932493249324932, | |
| "grad_norm": 0.15312208235263824, | |
| "learning_rate": 7.046663148979615e-10, | |
| "loss": 0.0143, | |
| "num_tokens": 282142359.0, | |
| "reward": 1.6685362339019776, | |
| "reward_std": 0.25656993985176085, | |
| "rewards/accuracy_reward/mean": 0.7805803537368774, | |
| "rewards/accuracy_reward/std": 0.4132432878017426, | |
| "rewards/cosine_scaled_reward/mean": 0.7792598366737366, | |
| "rewards/cosine_scaled_reward/std": 0.4132158994674683, | |
| "rewards/format_reward/mean": 0.9966517806053161, | |
| "rewards/format_reward/std": 0.05593275725841522, | |
| "step": 390 | |
| }, | |
| { | |
| "clip_ratio/high_max": 0.0, | |
| "clip_ratio/high_mean": 0.0, | |
| "clip_ratio/low_mean": 0.0, | |
| "clip_ratio/low_min": 0.0, | |
| "clip_ratio/region_mean": 0.0, | |
| "completions/clipped_ratio": 0.003833651178918163, | |
| "completions/max_length": 2048.0, | |
| "completions/max_terminated_length": 1935.2, | |
| "completions/mean_length": 615.3019409179688, | |
| "completions/mean_terminated_length": 609.7786987304687, | |
| "completions/min_length": 197.0, | |
| "completions/min_terminated_length": 197.0, | |
| "epoch": 4.995499549954996, | |
| "grad_norm": 0.15891171991825104, | |
| "learning_rate": 1.95785359849987e-11, | |
| "loss": 0.01, | |
| "num_tokens": 285681294.0, | |
| "reward": 1.6595622539520263, | |
| "reward_std": 0.2522480070590973, | |
| "rewards/accuracy_reward/mean": 0.7747767686843872, | |
| "rewards/accuracy_reward/std": 0.41717290282249453, | |
| "rewards/cosine_scaled_reward/mean": 0.7740350365638733, | |
| "rewards/cosine_scaled_reward/std": 0.4170451521873474, | |
| "rewards/format_reward/mean": 0.9955357193946839, | |
| "rewards/format_reward/std": 0.06459032222628594, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 4.995499549954996, | |
| "step": 395, | |
| "total_flos": 0.0, | |
| "train_loss": 0.011041237461038782, | |
| "train_runtime": 79107.9261, | |
| "train_samples_per_second": 0.562, | |
| "train_steps_per_second": 0.005 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 395, | |
| "num_input_tokens_seen": 285681294, | |
| "num_train_epochs": 5, | |
| "save_steps": 15, | |
| "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": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |