{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.125, "eval_steps": 500, "global_step": 50, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 2006.0, "completions/mean_length": 531.53125, "completions/mean_terminated_length": 482.6128845214844, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.0025, "frac_reward_zero_std": 0.0, "grad_norm": 4.296548366546631, "learning_rate": 0.0, "loss": 0.0, "num_tokens": 134762.0, "reward": 0.328125, "reward_std": 0.39572834968566895, "rewards/format_reward/mean": 0.3125, "rewards/format_reward/std": 0.24397502839565277, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "step": 1 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1242.0, "completions/max_terminated_length": 1242.0, "completions/mean_length": 395.703125, "completions/mean_terminated_length": 395.703125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.005, "frac_reward_zero_std": 0.0, "grad_norm": 4.7831130027771, "learning_rate": 2e-07, "loss": -0.0, "num_tokens": 218591.0, "reward": 0.25468748807907104, "reward_std": 0.26610028743743896, "rewards/format_reward/mean": 0.359375, "rewards/format_reward/std": 0.28824523091316223, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "step": 2 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.09375, "completions/max_length": 2048.0, "completions/max_terminated_length": 960.0, "completions/mean_length": 622.3125, "completions/mean_terminated_length": 474.82757568359375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0075, "frac_reward_zero_std": 0.0, "grad_norm": 10.044818878173828, "learning_rate": 4e-07, "loss": -0.0, "num_tokens": 346243.0, "reward": 0.3742187023162842, "reward_std": 0.3915485441684723, "rewards/format_reward/mean": 0.3046875, "rewards/format_reward/std": 0.29028159379959106, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "step": 3 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1710.0, "completions/mean_length": 470.125, "completions/mean_terminated_length": 445.0793762207031, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.01, "frac_reward_zero_std": 0.0, "grad_norm": 4.091139793395996, "learning_rate": 6e-07, "loss": -0.0, "num_tokens": 476139.0, "reward": 0.29374998807907104, "reward_std": 0.39774471521377563, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.2597312331199646, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 4 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 997.0, "completions/max_terminated_length": 997.0, "completions/mean_length": 412.8125, "completions/mean_terminated_length": 412.8125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0125, "frac_reward_zero_std": 0.0, "grad_norm": 3.120988368988037, "learning_rate": 8e-07, "loss": 0.0, "num_tokens": 581439.0, "reward": 0.15859374403953552, "reward_std": 0.26927319169044495, "rewards/format_reward/mean": 0.3359375, "rewards/format_reward/std": 0.2680720090866089, "rewards/mcq_exact_match_reward/mean": 0.125, "rewards/mcq_exact_match_reward/std": 0.3333333432674408, "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": 0.09375, "completions/max_length": 2048.0, "completions/max_terminated_length": 1770.0, "completions/mean_length": 624.140625, "completions/mean_terminated_length": 476.8448181152344, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.015, "frac_reward_zero_std": 0.0, "grad_norm": 10.825153350830078, "learning_rate": 1e-06, "loss": 0.0, "num_tokens": 726536.0, "reward": 0.13437499105930328, "reward_std": 0.25146162509918213, "rewards/format_reward/mean": 0.25, "rewards/format_reward/std": 0.29546841979026794, "rewards/mcq_exact_match_reward/mean": 0.109375, "rewards/mcq_exact_match_reward/std": 0.3145764470100403, "step": 6 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1289.0, "completions/max_terminated_length": 1289.0, "completions/mean_length": 455.296875, "completions/mean_terminated_length": 455.296875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0175, "frac_reward_zero_std": 0.0, "grad_norm": 7.804375171661377, "learning_rate": 9.99726628670463e-07, "loss": 0.0, "num_tokens": 843627.0, "reward": 0.24531248211860657, "reward_std": 0.2808409631252289, "rewards/format_reward/mean": 0.421875, "rewards/format_reward/std": 0.2221602201461792, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "step": 7 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1622.0, "completions/mean_length": 537.78125, "completions/mean_terminated_length": 437.10003662109375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.02, "frac_reward_zero_std": 0.0, "grad_norm": 8.704272270202637, "learning_rate": 9.989068136093872e-07, "loss": -0.0, "num_tokens": 974469.0, "reward": 0.3265624940395355, "reward_std": 0.4057665169239044, "rewards/format_reward/mean": 0.453125, "rewards/format_reward/std": 0.3177144229412079, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 8 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1529.0, "completions/mean_length": 604.953125, "completions/mean_terminated_length": 582.0476684570312, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0225, "frac_reward_zero_std": 0.0, "grad_norm": 8.170771598815918, "learning_rate": 9.975414512725056e-07, "loss": 0.0, "num_tokens": 1115738.0, "reward": 0.25312498211860657, "reward_std": 0.3531861901283264, "rewards/format_reward/mean": 0.34375, "rewards/format_reward/std": 0.233588308095932, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "step": 9 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1773.0, "completions/mean_length": 526.203125, "completions/mean_terminated_length": 477.1128845214844, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.025, "frac_reward_zero_std": 0.0, "grad_norm": 5.386010646820068, "learning_rate": 9.956320346634875e-07, "loss": -0.0, "num_tokens": 1243599.0, "reward": 0.28593748807907104, "reward_std": 0.32800400257110596, "rewards/format_reward/mean": 0.359375, "rewards/format_reward/std": 0.2592533528804779, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 10 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1511.0, "completions/mean_length": 537.46875, "completions/mean_terminated_length": 513.4920654296875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0275, "frac_reward_zero_std": 0.125, "grad_norm": 14.920385360717773, "learning_rate": 9.931806517013612e-07, "loss": -0.0, "num_tokens": 1363173.0, "reward": 0.29374998807907104, "reward_std": 0.3397524952888489, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.24397502839565277, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 11 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1409.0, "completions/max_terminated_length": 1409.0, "completions/mean_length": 414.171875, "completions/mean_terminated_length": 414.171875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.03, "frac_reward_zero_std": 0.125, "grad_norm": 24.134349822998047, "learning_rate": 9.901899829374047e-07, "loss": -0.0, "num_tokens": 1498360.0, "reward": 0.22265625, "reward_std": 0.32389262318611145, "rewards/format_reward/mean": 0.5078125, "rewards/format_reward/std": 0.24384792149066925, "rewards/mcq_exact_match_reward/mean": 0.171875, "rewards/mcq_exact_match_reward/std": 0.38025420904159546, "step": 12 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1405.0, "completions/max_terminated_length": 1405.0, "completions/mean_length": 456.578125, "completions/mean_terminated_length": 456.578125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0325, "frac_reward_zero_std": 0.125, "grad_norm": 9.119366645812988, "learning_rate": 9.866632986240029e-07, "loss": 0.0, "num_tokens": 1616277.0, "reward": 0.26953125, "reward_std": 0.30530208349227905, "rewards/format_reward/mean": 0.5078125, "rewards/format_reward/std": 0.2744719088077545, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "step": 13 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1842.0, "completions/max_terminated_length": 1842.0, "completions/mean_length": 334.71875, "completions/mean_terminated_length": 334.71875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.035, "frac_reward_zero_std": 0.125, "grad_norm": 14.80947494506836, "learning_rate": 9.826044551386742e-07, "loss": -0.0, "num_tokens": 1744323.0, "reward": 0.30156248807907104, "reward_std": 0.4018325209617615, "rewards/format_reward/mean": 0.515625, "rewards/format_reward/std": 0.1985812783241272, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 14 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1220.0, "completions/max_terminated_length": 1220.0, "completions/mean_length": 259.90625, "completions/mean_terminated_length": 259.90625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0375, "frac_reward_zero_std": 0.125, "grad_norm": 12.551422119140625, "learning_rate": 9.780178907671788e-07, "loss": -0.0, "num_tokens": 1840917.0, "reward": 0.16484373807907104, "reward_std": 0.18668311834335327, "rewards/format_reward/mean": 0.5546875, "rewards/format_reward/std": 0.29703813791275024, "rewards/mcq_exact_match_reward/mean": 0.109375, "rewards/mcq_exact_match_reward/std": 0.3145764470100403, "step": 15 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 907.0, "completions/max_terminated_length": 907.0, "completions/mean_length": 56.75, "completions/mean_terminated_length": 56.75, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.04, "frac_reward_zero_std": 0.375, "grad_norm": 18.445289611816406, "learning_rate": 9.729086208503173e-07, "loss": -0.0, "num_tokens": 1932021.0, "reward": 0.29921871423721313, "reward_std": 0.27164503931999207, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.1406387835741043, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 16 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 741.0, "completions/max_terminated_length": 741.0, "completions/mean_length": 184.375, "completions/mean_terminated_length": 184.375, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.0425, "frac_reward_zero_std": 0.25, "grad_norm": 36.467376708984375, "learning_rate": 9.672822322997304e-07, "loss": -0.0, "num_tokens": 2016877.0, "reward": 0.47734373807907104, "reward_std": 0.22254440188407898, "rewards/format_reward/mean": 0.5546875, "rewards/format_reward/std": 0.2538151443004608, "rewards/mcq_exact_match_reward/mean": 0.421875, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 17 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 608.0, "completions/max_terminated_length": 608.0, "completions/mean_length": 78.265625, "completions/mean_terminated_length": 78.265625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.045, "frac_reward_zero_std": 0.25, "grad_norm": 29.991186141967773, "learning_rate": 9.611448774886923e-07, "loss": 0.0, "num_tokens": 2112086.0, "reward": 0.30937498807907104, "reward_std": 0.21851179003715515, "rewards/format_reward/mean": 0.59375, "rewards/format_reward/std": 0.19669894874095917, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 18 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 689.0, "completions/max_terminated_length": 689.0, "completions/mean_length": 61.96875, "completions/mean_terminated_length": 61.96875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0475, "frac_reward_zero_std": 0.5, "grad_norm": 21.87519645690918, "learning_rate": 9.545032675245813e-07, "loss": 0.0, "num_tokens": 2210404.0, "reward": 0.35078123211860657, "reward_std": 0.23000042140483856, "rewards/format_reward/mean": 0.5390625, "rewards/format_reward/std": 0.13524486124515533, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "step": 19 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 525.0, "completions/max_terminated_length": 525.0, "completions/mean_length": 30.4375, "completions/mean_terminated_length": 30.4375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05, "frac_reward_zero_std": 0.375, "grad_norm": 35.56521224975586, "learning_rate": 9.473646649103817e-07, "loss": -0.0, "num_tokens": 2301520.0, "reward": 0.3374999761581421, "reward_std": 0.28328073024749756, "rewards/format_reward/mean": 0.5625, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 20 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 500.0, "completions/max_terminated_length": 500.0, "completions/mean_length": 44.578125, "completions/mean_terminated_length": 44.578125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0525, "frac_reward_zero_std": 0.375, "grad_norm": 19.46839714050293, "learning_rate": 9.397368756032444e-07, "loss": -0.0, "num_tokens": 2382741.0, "reward": 0.3499999940395355, "reward_std": 0.2742733359336853, "rewards/format_reward/mean": 0.53125, "rewards/format_reward/std": 0.19669894874095917, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "step": 21 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1183.0, "completions/max_terminated_length": 1183.0, "completions/mean_length": 36.796875, "completions/mean_terminated_length": 36.796875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.055, "frac_reward_zero_std": 0.625, "grad_norm": 20.99708366394043, "learning_rate": 9.316282404787869e-07, "loss": -0.0, "num_tokens": 2474712.0, "reward": 0.7117186784744263, "reward_std": 0.17411330342292786, "rewards/format_reward/mean": 0.5546875, "rewards/format_reward/std": 0.15728822350502014, "rewards/mcq_exact_match_reward/mean": 0.65625, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "step": 22 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 858.0, "completions/max_terminated_length": 858.0, "completions/mean_length": 149.296875, "completions/mean_terminated_length": 149.296875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0575, "frac_reward_zero_std": 0.5, "grad_norm": 17.8493595123291, "learning_rate": 9.230476262104676e-07, "loss": 0.0, "num_tokens": 2565059.0, "reward": 0.5648437738418579, "reward_std": 0.2229262888431549, "rewards/format_reward/mean": 0.6484375, "rewards/format_reward/std": 0.2302463799715042, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "step": 23 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 383.0, "completions/max_terminated_length": 383.0, "completions/mean_length": 17.96875, "completions/mean_terminated_length": 17.96875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06, "frac_reward_zero_std": 0.375, "grad_norm": 19.127931594848633, "learning_rate": 9.1400441557401e-07, "loss": -0.0, "num_tokens": 2680977.0, "reward": 0.26640623807907104, "reward_std": 0.22785000503063202, "rewards/format_reward/mean": 0.4765625, "rewards/format_reward/std": 0.10652101784944534, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "step": 24 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 485.0, "completions/max_terminated_length": 485.0, "completions/mean_length": 47.15625, "completions/mean_terminated_length": 47.15625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0625, "frac_reward_zero_std": 0.5, "grad_norm": 18.63139533996582, "learning_rate": 9.045084971874737e-07, "loss": 0.0, "num_tokens": 2761971.0, "reward": 0.29765623807907104, "reward_std": 0.2175242304801941, "rewards/format_reward/mean": 0.6328125, "rewards/format_reward/std": 0.22257846593856812, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "step": 25 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 260.0, "completions/max_terminated_length": 260.0, "completions/mean_length": 11.484375, "completions/mean_terminated_length": 11.484375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.065, "frac_reward_zero_std": 0.625, "grad_norm": 16.90754508972168, "learning_rate": 8.945702546981968e-07, "loss": 0.0, "num_tokens": 2843962.0, "reward": 0.46796876192092896, "reward_std": 0.14599090814590454, "rewards/format_reward/mean": 0.6171875, "rewards/format_reward/std": 0.21347814798355103, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 26 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 490.0, "completions/max_terminated_length": 490.0, "completions/mean_length": 19.515625, "completions/mean_terminated_length": 19.515625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0675, "frac_reward_zero_std": 0.75, "grad_norm": 12.182069778442383, "learning_rate": 8.842005554284295e-07, "loss": -0.0, "num_tokens": 2934115.0, "reward": 0.699999988079071, "reward_std": 0.10215429216623306, "rewards/format_reward/mean": 0.59375, "rewards/format_reward/std": 0.19669894874095917, "rewards/mcq_exact_match_reward/mean": 0.640625, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "step": 27 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 527.0, "completions/max_terminated_length": 527.0, "completions/mean_length": 25.90625, "completions/mean_terminated_length": 25.90625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07, "frac_reward_zero_std": 0.25, "grad_norm": 26.69842529296875, "learning_rate": 8.734107384920769e-07, "loss": -0.0, "num_tokens": 3016237.0, "reward": 0.71484375, "reward_std": 0.356208860874176, "rewards/format_reward/mean": 0.5859375, "rewards/format_reward/std": 0.244862899184227, "rewards/mcq_exact_match_reward/mean": 0.65625, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "step": 28 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 229.0, "completions/max_terminated_length": 229.0, "completions/mean_length": 16.734375, "completions/mean_terminated_length": 16.734375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0725, "frac_reward_zero_std": 0.5, "grad_norm": 17.901775360107422, "learning_rate": 8.622126023955445e-07, "loss": -0.0, "num_tokens": 3111692.0, "reward": 0.589062511920929, "reward_std": 0.22337424755096436, "rewards/format_reward/mean": 0.578125, "rewards/format_reward/std": 0.18298126757144928, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 29 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 383.0, "completions/max_terminated_length": 383.0, "completions/mean_length": 17.546875, "completions/mean_terminated_length": 17.546875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.075, "frac_reward_zero_std": 0.25, "grad_norm": 33.41448211669922, "learning_rate": 8.506183921362442e-07, "loss": 0.0, "num_tokens": 3201151.0, "reward": 0.69140625, "reward_std": 0.2782713770866394, "rewards/format_reward/mean": 0.6640625, "rewards/format_reward/std": 0.2366211861371994, "rewards/mcq_exact_match_reward/mean": 0.625, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 30 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 184.0, "completions/max_terminated_length": 184.0, "completions/mean_length": 12.265625, "completions/mean_terminated_length": 12.265625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0775, "frac_reward_zero_std": 0.375, "grad_norm": 30.0745792388916, "learning_rate": 8.386407858128706e-07, "loss": -0.0, "num_tokens": 3316152.0, "reward": 0.6242187023162842, "reward_std": 0.21998533606529236, "rewards/format_reward/mean": 0.6171875, "rewards/format_reward/std": 0.21347814798355103, "rewards/mcq_exact_match_reward/mean": 0.5625, "rewards/mcq_exact_match_reward/std": 0.5, "step": 31 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1141.0, "completions/max_terminated_length": 1141.0, "completions/mean_length": 25.3125, "completions/mean_terminated_length": 25.3125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08, "frac_reward_zero_std": 0.75, "grad_norm": 8.674514770507812, "learning_rate": 8.262928807620843e-07, "loss": -0.0, "num_tokens": 3410396.0, "reward": 0.328125, "reward_std": 0.12255740165710449, "rewards/format_reward/mean": 0.625, "rewards/format_reward/std": 0.2182178944349289, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "step": 32 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 176.0, "completions/max_terminated_length": 176.0, "completions/mean_length": 9.171875, "completions/mean_terminated_length": 9.171875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0825, "frac_reward_zero_std": 0.75, "grad_norm": 17.30906105041504, "learning_rate": 8.135881792367685e-07, "loss": 0.0, "num_tokens": 3511927.0, "reward": 0.6617187261581421, "reward_std": 0.047428861260414124, "rewards/format_reward/mean": 0.5234375, "rewards/format_reward/std": 0.10652101784944534, "rewards/mcq_exact_match_reward/mean": 0.609375, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "step": 33 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 681.0, "completions/max_terminated_length": 681.0, "completions/mean_length": 19.75, "completions/mean_terminated_length": 19.75, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.085, "frac_reward_zero_std": 0.625, "grad_norm": 13.118045806884766, "learning_rate": 8.005405736415125e-07, "loss": 0.0, "num_tokens": 3639415.0, "reward": 0.3140625059604645, "reward_std": 0.09128153324127197, "rewards/format_reward/mean": 0.640625, "rewards/format_reward/std": 0.22658175230026245, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 34 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 533.0, "completions/max_terminated_length": 533.0, "completions/mean_length": 23.265625, "completions/mean_terminated_length": 23.265625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0875, "frac_reward_zero_std": 0.75, "grad_norm": 13.433796882629395, "learning_rate": 7.871643313414718e-07, "loss": -0.0, "num_tokens": 3718976.0, "reward": 0.796875, "reward_std": 0.048613592982292175, "rewards/format_reward/mean": 0.625, "rewards/format_reward/std": 0.2182178944349289, "rewards/mcq_exact_match_reward/mean": 0.734375, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "step": 35 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 389.0, "completions/max_terminated_length": 389.0, "completions/mean_length": 14.171875, "completions/mean_terminated_length": 14.171875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.09, "frac_reward_zero_std": 0.625, "grad_norm": 10.209205627441406, "learning_rate": 7.734740790612136e-07, "loss": 0.0, "num_tokens": 3837411.0, "reward": 0.37812498211860657, "reward_std": 0.15992167592048645, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 36 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 203.0, "completions/max_terminated_length": 203.0, "completions/mean_length": 12.34375, "completions/mean_terminated_length": 12.34375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0925, "frac_reward_zero_std": 0.375, "grad_norm": 26.544673919677734, "learning_rate": 7.594847868906076e-07, "loss": -0.0, "num_tokens": 3989545.0, "reward": 0.38593748211860657, "reward_std": 0.15750139951705933, "rewards/format_reward/mean": 0.578125, "rewards/format_reward/std": 0.18298126757144928, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 37 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1000.0, "completions/max_terminated_length": 1000.0, "completions/mean_length": 42.859375, "completions/mean_terminated_length": 42.859375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.095, "frac_reward_zero_std": 0.625, "grad_norm": 14.327924728393555, "learning_rate": 7.452117519152541e-07, "loss": 0.0, "num_tokens": 4074032.0, "reward": 0.4203125238418579, "reward_std": 0.1469651460647583, "rewards/format_reward/mean": 0.765625, "rewards/format_reward/std": 0.2812775671482086, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "step": 38 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1182.0, "completions/mean_length": 114.015625, "completions/mean_terminated_length": 83.31746673583984, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0975, "frac_reward_zero_std": 0.5, "grad_norm": 15.019013404846191, "learning_rate": 7.306705814893439e-07, "loss": 0.0, "num_tokens": 4158713.0, "reward": 0.44296878576278687, "reward_std": 0.19020721316337585, "rewards/format_reward/mean": 0.8359375, "rewards/format_reward/std": 0.2680720090866089, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "step": 39 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 368.0, "completions/max_terminated_length": 368.0, "completions/mean_length": 25.953125, "completions/mean_terminated_length": 25.953125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1, "frac_reward_zero_std": 0.625, "grad_norm": 17.713464736938477, "learning_rate": 7.158771761692464e-07, "loss": -0.0, "num_tokens": 4238038.0, "reward": 0.653124988079071, "reward_std": 0.11186130344867706, "rewards/format_reward/mean": 0.90625, "rewards/format_reward/std": 0.19669894874095917, "rewards/mcq_exact_match_reward/mean": 0.5625, "rewards/mcq_exact_match_reward/std": 0.5, "step": 40 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1458.0, "completions/max_terminated_length": 1458.0, "completions/mean_length": 90.984375, "completions/mean_terminated_length": 90.984375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1025, "frac_reward_zero_std": 0.25, "grad_norm": 16.038522720336914, "learning_rate": 7.008477123264847e-07, "loss": 0.0, "num_tokens": 4347693.0, "reward": 0.3648437559604645, "reward_std": 0.20585379004478455, "rewards/format_reward/mean": 0.8359375, "rewards/format_reward/std": 0.2680720090866089, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 41 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 296.0, "completions/mean_length": 54.796875, "completions/mean_terminated_length": 23.15873146057129, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.105, "frac_reward_zero_std": 0.125, "grad_norm": 25.373247146606445, "learning_rate": 6.855986244591103e-07, "loss": 0.0, "num_tokens": 4443288.0, "reward": 0.6656250357627869, "reward_std": 0.3440704941749573, "rewards/format_reward/mean": 0.875, "rewards/format_reward/std": 0.2357022762298584, "rewards/mcq_exact_match_reward/mean": 0.578125, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 42 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 613.0, "completions/max_terminated_length": 613.0, "completions/mean_length": 62.9375, "completions/mean_terminated_length": 62.9375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1075, "frac_reward_zero_std": 0.25, "grad_norm": 24.26311492919922, "learning_rate": 6.701465872208216e-07, "loss": -0.0, "num_tokens": 4557476.0, "reward": 0.668749988079071, "reward_std": 0.266130268573761, "rewards/format_reward/mean": 0.90625, "rewards/format_reward/std": 0.21593283116817474, "rewards/mcq_exact_match_reward/mean": 0.578125, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 43 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 703.0, "completions/max_terminated_length": 703.0, "completions/mean_length": 30.671875, "completions/mean_terminated_length": 30.671875, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.11, "frac_reward_zero_std": 0.375, "grad_norm": 23.199234008789062, "learning_rate": 6.545084971874736e-07, "loss": 0.0, "num_tokens": 4645103.0, "reward": 0.2992187738418579, "reward_std": 0.2479499727487564, "rewards/format_reward/mean": 0.9609375, "rewards/format_reward/std": 0.18483558297157288, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "step": 44 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 872.0, "completions/max_terminated_length": 872.0, "completions/mean_length": 38.765625, "completions/mean_terminated_length": 38.765625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1125, "frac_reward_zero_std": 0.25, "grad_norm": 33.96176528930664, "learning_rate": 6.387014543809223e-07, "loss": 0.0, "num_tokens": 4734104.0, "reward": 0.9328124523162842, "reward_std": 0.24389895796775818, "rewards/format_reward/mean": 0.890625, "rewards/format_reward/std": 0.22658175230026245, "rewards/mcq_exact_match_reward/mean": 0.84375, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "step": 45 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 716.0, "completions/max_terminated_length": 716.0, "completions/mean_length": 65.640625, "completions/mean_terminated_length": 65.640625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.115, "frac_reward_zero_std": 0.125, "grad_norm": 24.17040252685547, "learning_rate": 6.227427435703995e-07, "loss": -0.0, "num_tokens": 4851193.0, "reward": 0.671875, "reward_std": 0.2817694842815399, "rewards/format_reward/mean": 0.9375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.578125, "rewards/mcq_exact_match_reward/std": 0.49776285886764526, "step": 46 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 544.0, "completions/max_terminated_length": 544.0, "completions/mean_length": 28.0625, "completions/mean_terminated_length": 28.0625, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.1175, "frac_reward_zero_std": 0.5, "grad_norm": 26.796812057495117, "learning_rate": 6.066498153718734e-07, "loss": 0.0, "num_tokens": 4927477.0, "reward": 0.7078125476837158, "reward_std": 0.2524387836456299, "rewards/format_reward/mean": 0.984375, "rewards/format_reward/std": 0.08768405020236969, "rewards/mcq_exact_match_reward/mean": 0.609375, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "step": 47 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1351.0, "completions/mean_length": 168.125, "completions/mean_terminated_length": 107.48387145996094, "completions/min_length": 12.0, "completions/min_terminated_length": 12.0, "epoch": 0.12, "frac_reward_zero_std": 0.125, "grad_norm": 20.72756576538086, "learning_rate": 5.90440267166055e-07, "loss": 0.0, "num_tokens": 5073125.0, "reward": 0.596875011920929, "reward_std": 0.318790078163147, "rewards/format_reward/mean": 0.96875, "rewards/format_reward/std": 0.17536810040473938, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "step": 48 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 412.0, "completions/mean_length": 57.03125, "completions/mean_terminated_length": 25.428573608398438, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1225, "frac_reward_zero_std": 0.5, "grad_norm": 21.319555282592773, "learning_rate": 5.741318238559209e-07, "loss": 0.0, "num_tokens": 5191855.0, "reward": 0.7445312738418579, "reward_std": 0.11654192954301834, "rewards/format_reward/mean": 0.8828125, "rewards/format_reward/std": 0.2313210517168045, "rewards/mcq_exact_match_reward/mean": 0.65625, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "step": 49 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 743.0, "completions/mean_length": 106.53125, "completions/mean_terminated_length": 43.90322494506836, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.125, "frac_reward_zero_std": 0.375, "grad_norm": 12.621797561645508, "learning_rate": 5.577423184847931e-07, "loss": 0.0, "num_tokens": 5291529.0, "reward": 0.41875001788139343, "reward_std": 0.2325233668088913, "rewards/format_reward/mean": 0.90625, "rewards/format_reward/std": 0.233588308095932, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 50 } ], "logging_steps": 1, "max_steps": 100, "num_input_tokens_seen": 5291529, "num_train_epochs": 1, "save_steps": 50, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }