{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.125, "eval_steps": 500, "global_step": 100, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1619.0, "completions/mean_length": 656.890625, "completions/mean_terminated_length": 588.4754028320312, "completions/min_length": 9.0, "completions/min_terminated_length": 9.0, "epoch": 0.00125, "frac_reward_zero_std": 0.0, "grad_norm": 3.1868386268615723, "learning_rate": 0.0, "loss": 0.0, "num_tokens": 131897.0, "reward": 0.16484373807907104, "reward_std": 0.3002462685108185, "rewards/format_reward/mean": 0.3984375, "rewards/format_reward/std": 0.26991602778434753, "rewards/mcq_exact_match_reward/mean": 0.125, "rewards/mcq_exact_match_reward/std": 0.3333333432674408, "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": 1554.0, "completions/max_terminated_length": 1554.0, "completions/mean_length": 333.40625, "completions/mean_terminated_length": 333.40625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0025, "frac_reward_zero_std": 0.125, "grad_norm": 8.810633659362793, "learning_rate": 1e-07, "loss": -0.0, "num_tokens": 241027.0, "reward": 0.23593749105930328, "reward_std": 0.2982994318008423, "rewards/format_reward/mean": 0.328125, "rewards/format_reward/std": 0.25539806485176086, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1284.0, "completions/mean_length": 545.34375, "completions/mean_terminated_length": 471.4425964355469, "completions/min_length": 68.0, "completions/min_terminated_length": 68.0, "epoch": 0.00375, "frac_reward_zero_std": 0.0, "grad_norm": 2.1099283695220947, "learning_rate": 2e-07, "loss": 0.0, "num_tokens": 353969.0, "reward": 0.3570312261581421, "reward_std": 0.35539060831069946, "rewards/format_reward/mean": 0.4453125, "rewards/format_reward/std": 0.29703813791275024, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1360.0, "completions/mean_length": 467.21875, "completions/mean_terminated_length": 389.47540283203125, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.005, "frac_reward_zero_std": 0.125, "grad_norm": 12.928600311279297, "learning_rate": 3e-07, "loss": -0.0, "num_tokens": 462719.0, "reward": 0.2124999761581421, "reward_std": 0.29265305399894714, "rewards/format_reward/mean": 0.25, "rewards/format_reward/std": 0.28171807527542114, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "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": 1713.0, "completions/max_terminated_length": 1713.0, "completions/mean_length": 460.0625, "completions/mean_terminated_length": 460.0625, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.00625, "frac_reward_zero_std": 0.0, "grad_norm": 11.410627365112305, "learning_rate": 4e-07, "loss": 0.0, "num_tokens": 569571.0, "reward": 0.18671873211860657, "reward_std": 0.21998099982738495, "rewards/format_reward/mean": 0.3046875, "rewards/format_reward/std": 0.2615155577659607, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "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": 0.078125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1707.0, "completions/mean_length": 578.375, "completions/mean_terminated_length": 453.83050537109375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0075, "frac_reward_zero_std": 0.0, "grad_norm": 10.172877311706543, "learning_rate": 5e-07, "loss": -0.0, "num_tokens": 715995.0, "reward": 0.39687496423721313, "reward_std": 0.3704853653907776, "rewards/format_reward/mean": 0.375, "rewards/format_reward/std": 0.26726123690605164, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 1362.0, "completions/max_terminated_length": 1362.0, "completions/mean_length": 390.0625, "completions/mean_terminated_length": 390.0625, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.00875, "frac_reward_zero_std": 0.0, "grad_norm": 18.49201774597168, "learning_rate": 6e-07, "loss": 0.0, "num_tokens": 819415.0, "reward": 0.18125000596046448, "reward_std": 0.32663917541503906, "rewards/format_reward/mean": 0.25, "rewards/format_reward/std": 0.28171807527542114, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "step": 7 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 2042.0, "completions/max_terminated_length": 2042.0, "completions/mean_length": 394.875, "completions/mean_terminated_length": 394.875, "completions/min_length": 10.0, "completions/min_terminated_length": 10.0, "epoch": 0.01, "frac_reward_zero_std": 0.0, "grad_norm": 10.525629997253418, "learning_rate": 7e-07, "loss": -0.0, "num_tokens": 936127.0, "reward": 0.17734375596046448, "reward_std": 0.3208874762058258, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.3362041711807251, "rewards/mcq_exact_match_reward/mean": 0.140625, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "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": 0.0, "completions/max_length": 1945.0, "completions/max_terminated_length": 1945.0, "completions/mean_length": 492.34375, "completions/mean_terminated_length": 492.34375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.01125, "frac_reward_zero_std": 0.125, "grad_norm": 10.087387084960938, "learning_rate": 8e-07, "loss": -0.0, "num_tokens": 1039077.0, "reward": 0.2085937261581421, "reward_std": 0.32167965173721313, "rewards/format_reward/mean": 0.3671875, "rewards/format_reward/std": 0.28510910272598267, "rewards/mcq_exact_match_reward/mean": 0.171875, "rewards/mcq_exact_match_reward/std": 0.38025420904159546, "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": 1439.0, "completions/mean_length": 499.171875, "completions/mean_terminated_length": 449.20965576171875, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "epoch": 0.0125, "frac_reward_zero_std": 0.25, "grad_norm": 7.938133239746094, "learning_rate": 9e-07, "loss": 0.0, "num_tokens": 1146576.0, "reward": 0.30546873807907104, "reward_std": 0.23584464192390442, "rewards/format_reward/mean": 0.3984375, "rewards/format_reward/std": 0.26991602778434753, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 1620.0, "completions/mean_length": 544.21875, "completions/mean_terminated_length": 520.3492431640625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.01375, "frac_reward_zero_std": 0.0, "grad_norm": 5.112438201904297, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 1258870.0, "reward": 0.27031248807907104, "reward_std": 0.31277522444725037, "rewards/format_reward/mean": 0.359375, "rewards/format_reward/std": 0.24346621334552765, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1300.0, "completions/mean_length": 473.0625, "completions/mean_terminated_length": 448.0635070800781, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.015, "frac_reward_zero_std": 0.0, "grad_norm": 12.289234161376953, "learning_rate": 9.999316524962345e-07, "loss": -0.0, "num_tokens": 1371578.0, "reward": 0.24687498807907104, "reward_std": 0.3092396855354309, "rewards/format_reward/mean": 0.28125, "rewards/format_reward/std": 0.25, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1756.0, "completions/mean_length": 706.9375, "completions/mean_terminated_length": 663.6774291992188, "completions/min_length": 50.0, "completions/min_terminated_length": 50.0, "epoch": 0.01625, "frac_reward_zero_std": 0.0, "grad_norm": 1.1752923727035522, "learning_rate": 9.99726628670463e-07, "loss": -0.0, "num_tokens": 1510414.0, "reward": 0.12734374403953552, "reward_std": 0.2132968008518219, "rewards/format_reward/mean": 0.3359375, "rewards/format_reward/std": 0.2366211861371994, "rewards/mcq_exact_match_reward/mean": 0.09375, "rewards/mcq_exact_match_reward/std": 0.29378482699394226, "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": 1546.0, "completions/max_terminated_length": 1546.0, "completions/mean_length": 440.75, "completions/mean_terminated_length": 440.75, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0175, "frac_reward_zero_std": 0.25, "grad_norm": 3.877164840698242, "learning_rate": 9.993849845741523e-07, "loss": 0.0, "num_tokens": 1607894.0, "reward": 0.24140623211860657, "reward_std": 0.13719820976257324, "rewards/format_reward/mean": 0.3828125, "rewards/format_reward/std": 0.2313210517168045, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "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": 1463.0, "completions/max_terminated_length": 1463.0, "completions/mean_length": 562.546875, "completions/mean_terminated_length": 562.546875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.01875, "frac_reward_zero_std": 0.125, "grad_norm": 5.995397090911865, "learning_rate": 9.989068136093872e-07, "loss": -0.0, "num_tokens": 1719121.0, "reward": 0.2281249761581421, "reward_std": 0.2842724323272705, "rewards/format_reward/mean": 0.40625, "rewards/format_reward/std": 0.19669894874095917, "rewards/mcq_exact_match_reward/mean": 0.1875, "rewards/mcq_exact_match_reward/std": 0.39339789748191833, "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": 1096.0, "completions/max_terminated_length": 1096.0, "completions/mean_length": 494.921875, "completions/mean_terminated_length": 494.921875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.02, "frac_reward_zero_std": 0.0, "grad_norm": 6.576966762542725, "learning_rate": 9.982922465033348e-07, "loss": 0.0, "num_tokens": 1820220.0, "reward": 0.4296875, "reward_std": 0.34125807881355286, "rewards/format_reward/mean": 0.390625, "rewards/format_reward/std": 0.28824523091316223, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 1762.0, "completions/max_terminated_length": 1762.0, "completions/mean_length": 595.21875, "completions/mean_terminated_length": 595.21875, "completions/min_length": 194.0, "completions/min_terminated_length": 194.0, "epoch": 0.02125, "frac_reward_zero_std": 0.0, "grad_norm": 0.8279478549957275, "learning_rate": 9.975414512725056e-07, "loss": 0.0, "num_tokens": 1936634.0, "reward": 0.29218748211860657, "reward_std": 0.30033600330352783, "rewards/format_reward/mean": 0.421875, "rewards/format_reward/std": 0.2221602201461792, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 1525.0, "completions/max_terminated_length": 1525.0, "completions/mean_length": 531.453125, "completions/mean_terminated_length": 531.453125, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.0225, "frac_reward_zero_std": 0.0, "grad_norm": 10.704648971557617, "learning_rate": 9.966546331768192e-07, "loss": -0.0, "num_tokens": 2073071.0, "reward": 0.41718748211860657, "reward_std": 0.4428107738494873, "rewards/format_reward/mean": 0.421875, "rewards/format_reward/std": 0.25539806485176086, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1751.0, "completions/mean_length": 742.0625, "completions/mean_terminated_length": 699.9354858398438, "completions/min_length": 202.0, "completions/min_terminated_length": 202.0, "epoch": 0.02375, "frac_reward_zero_std": 0.125, "grad_norm": 0.6757264137268066, "learning_rate": 9.956320346634875e-07, "loss": 0.0, "num_tokens": 2196875.0, "reward": 0.2929687201976776, "reward_std": 0.34113290905952454, "rewards/format_reward/mean": 0.4296875, "rewards/format_reward/std": 0.1751912236213684, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1936.0, "completions/mean_length": 651.578125, "completions/mean_terminated_length": 629.4127197265625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.025, "frac_reward_zero_std": 0.125, "grad_norm": 8.654264450073242, "learning_rate": 9.944739353007341e-07, "loss": -0.0, "num_tokens": 2343992.0, "reward": 0.30937498807907104, "reward_std": 0.3132135272026062, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1966.0, "completions/mean_length": 787.265625, "completions/mean_terminated_length": 746.5967407226562, "completions/min_length": 230.0, "completions/min_terminated_length": 230.0, "epoch": 0.02625, "frac_reward_zero_std": 0.125, "grad_norm": 0.5615260004997253, "learning_rate": 9.931806517013612e-07, "loss": -0.0, "num_tokens": 2474425.0, "reward": 0.18281248211860657, "reward_std": 0.276297926902771, "rewards/format_reward/mean": 0.421875, "rewards/format_reward/std": 0.18298126757144928, "rewards/mcq_exact_match_reward/mean": 0.140625, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "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": 2035.0, "completions/max_terminated_length": 2035.0, "completions/mean_length": 703.59375, "completions/mean_terminated_length": 703.59375, "completions/min_length": 140.0, "completions/min_terminated_length": 140.0, "epoch": 0.0275, "frac_reward_zero_std": 0.25, "grad_norm": 0.801341712474823, "learning_rate": 9.917525374361911e-07, "loss": -0.0, "num_tokens": 2615423.0, "reward": 0.36562496423721313, "reward_std": 0.30971288681030273, "rewards/format_reward/mean": 0.375, "rewards/format_reward/std": 0.2182178944349289, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 0.09375, "completions/max_length": 2048.0, "completions/max_terminated_length": 1535.0, "completions/mean_length": 777.203125, "completions/mean_terminated_length": 645.7413940429688, "completions/min_length": 306.0, "completions/min_terminated_length": 306.0, "epoch": 0.02875, "frac_reward_zero_std": 0.125, "grad_norm": 0.6975756287574768, "learning_rate": 9.901899829374047e-07, "loss": -0.0, "num_tokens": 2754036.0, "reward": 0.4484374523162842, "reward_std": 0.39573222398757935, "rewards/format_reward/mean": 0.421875, "rewards/format_reward/std": 0.18298126757144928, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "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": 1564.0, "completions/max_terminated_length": 1564.0, "completions/mean_length": 672.171875, "completions/mean_terminated_length": 672.171875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.03, "frac_reward_zero_std": 0.125, "grad_norm": 4.384822368621826, "learning_rate": 9.884934153917996e-07, "loss": 0.0, "num_tokens": 2869375.0, "reward": 0.25078123807907104, "reward_std": 0.31022295355796814, "rewards/format_reward/mean": 0.4765625, "rewards/format_reward/std": 0.10652101784944534, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "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": 1670.0, "completions/max_terminated_length": 1670.0, "completions/mean_length": 675.125, "completions/mean_terminated_length": 675.125, "completions/min_length": 189.0, "completions/min_terminated_length": 189.0, "epoch": 0.03125, "frac_reward_zero_std": 0.25, "grad_norm": 0.7445923089981079, "learning_rate": 9.866632986240029e-07, "loss": -0.0, "num_tokens": 2982655.0, "reward": 0.27968746423721313, "reward_std": 0.34310489892959595, "rewards/format_reward/mean": 0.453125, "rewards/format_reward/std": 0.14689241349697113, "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": 1086.0, "completions/max_terminated_length": 1086.0, "completions/mean_length": 609.734375, "completions/mean_terminated_length": 609.734375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0325, "frac_reward_zero_std": 0.25, "grad_norm": 2.850458860397339, "learning_rate": 9.847001329696652e-07, "loss": -0.0, "num_tokens": 3104054.0, "reward": 0.3101562261581421, "reward_std": 0.2846561074256897, "rewards/format_reward/mean": 0.4453125, "rewards/format_reward/std": 0.15728822350502014, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 1914.0, "completions/max_terminated_length": 1914.0, "completions/mean_length": 694.21875, "completions/mean_terminated_length": 694.21875, "completions/min_length": 230.0, "completions/min_terminated_length": 230.0, "epoch": 0.03375, "frac_reward_zero_std": 0.125, "grad_norm": 0.7950528860092163, "learning_rate": 9.826044551386742e-07, "loss": -0.0, "num_tokens": 3215052.0, "reward": 0.38515621423721313, "reward_std": 0.3771534562110901, "rewards/format_reward/mean": 0.4140625, "rewards/format_reward/std": 0.19012710452079773, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1550.0, "completions/mean_length": 744.640625, "completions/mean_terminated_length": 723.952392578125, "completions/min_length": 227.0, "completions/min_terminated_length": 227.0, "epoch": 0.035, "frac_reward_zero_std": 0.125, "grad_norm": 0.7383192181587219, "learning_rate": 9.803768380684242e-07, "loss": -0.0, "num_tokens": 3350693.0, "reward": 0.31328123807907104, "reward_std": 0.40903252363204956, "rewards/format_reward/mean": 0.4765625, "rewards/format_reward/std": 0.10652101784944534, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "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": 1275.0, "completions/max_terminated_length": 1275.0, "completions/mean_length": 516.5, "completions/mean_terminated_length": 516.5, "completions/min_length": 85.0, "completions/min_terminated_length": 85.0, "epoch": 0.03625, "frac_reward_zero_std": 0.25, "grad_norm": 0.8985627889633179, "learning_rate": 9.780178907671788e-07, "loss": 0.0, "num_tokens": 3457773.0, "reward": 0.3765624761581421, "reward_std": 0.2873324751853943, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.08768405020236969, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 1133.0, "completions/max_terminated_length": 1133.0, "completions/mean_length": 651.0, "completions/mean_terminated_length": 651.0, "completions/min_length": 221.0, "completions/min_terminated_length": 221.0, "epoch": 0.0375, "frac_reward_zero_std": 0.375, "grad_norm": 0.5748822093009949, "learning_rate": 9.755282581475767e-07, "loss": -0.0, "num_tokens": 3572461.0, "reward": 0.3140624761581421, "reward_std": 0.3108360469341278, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.08768405020236969, "rewards/mcq_exact_match_reward/mean": 0.265625, "rewards/mcq_exact_match_reward/std": 0.44515693187713623, "step": 30 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1473.0, "completions/mean_length": 569.15625, "completions/mean_terminated_length": 545.6825561523438, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.03875, "frac_reward_zero_std": 0.125, "grad_norm": 2.3605189323425293, "learning_rate": 9.729086208503173e-07, "loss": -0.0, "num_tokens": 3680727.0, "reward": 0.4390624761581421, "reward_std": 0.3992472290992737, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.08768405020236969, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 1728.0, "completions/max_terminated_length": 1728.0, "completions/mean_length": 652.421875, "completions/mean_terminated_length": 652.421875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.04, "frac_reward_zero_std": 0.375, "grad_norm": 1.3452924489974976, "learning_rate": 9.701596950580807e-07, "loss": -0.0, "num_tokens": 3793314.0, "reward": 0.5804687142372131, "reward_std": 0.27674588561058044, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 32 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1577.0, "completions/mean_length": 599.953125, "completions/mean_terminated_length": 576.96826171875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.04125, "frac_reward_zero_std": 0.25, "grad_norm": 8.010074615478516, "learning_rate": 9.672822322997304e-07, "loss": -0.0, "num_tokens": 3908543.0, "reward": 0.36171871423721313, "reward_std": 0.35437196493148804, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1449.0, "completions/mean_length": 651.953125, "completions/mean_terminated_length": 629.793701171875, "completions/min_length": 163.0, "completions/min_terminated_length": 163.0, "epoch": 0.0425, "frac_reward_zero_std": 0.375, "grad_norm": 0.7337175011634827, "learning_rate": 9.642770192448535e-07, "loss": -0.0, "num_tokens": 4013956.0, "reward": 0.2523437440395355, "reward_std": 0.2831503748893738, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "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": 1362.0, "completions/max_terminated_length": 1362.0, "completions/mean_length": 673.015625, "completions/mean_terminated_length": 673.015625, "completions/min_length": 82.0, "completions/min_terminated_length": 82.0, "epoch": 0.04375, "frac_reward_zero_std": 0.0, "grad_norm": 0.9999064803123474, "learning_rate": 9.611448774886923e-07, "loss": 0.0, "num_tokens": 4129581.0, "reward": 0.26484376192092896, "reward_std": 0.326660692691803, "rewards/format_reward/mean": 0.4609375, "rewards/format_reward/std": 0.13524486124515533, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "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": 1408.0, "completions/max_terminated_length": 1408.0, "completions/mean_length": 632.578125, "completions/mean_terminated_length": 632.578125, "completions/min_length": 270.0, "completions/min_terminated_length": 270.0, "epoch": 0.045, "frac_reward_zero_std": 0.0, "grad_norm": 0.7939808368682861, "learning_rate": 9.578866633275286e-07, "loss": -0.0, "num_tokens": 4244626.0, "reward": 0.4390624463558197, "reward_std": 0.4785521626472473, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.08768405020236969, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1457.0, "completions/mean_length": 668.109375, "completions/mean_terminated_length": 600.245849609375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.04625, "frac_reward_zero_std": 0.125, "grad_norm": 5.269177436828613, "learning_rate": 9.545032675245813e-07, "loss": -0.0, "num_tokens": 4386809.0, "reward": 0.3531249761581421, "reward_std": 0.3662275969982147, "rewards/format_reward/mean": 0.40625, "rewards/format_reward/std": 0.19669894874095917, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 1874.0, "completions/max_terminated_length": 1874.0, "completions/mean_length": 717.5, "completions/mean_terminated_length": 717.5, "completions/min_length": 253.0, "completions/min_terminated_length": 253.0, "epoch": 0.0475, "frac_reward_zero_std": 0.0, "grad_norm": 0.7175169587135315, "learning_rate": 9.509956150664795e-07, "loss": 0.0, "num_tokens": 4501393.0, "reward": 0.24531248211860657, "reward_std": 0.29694393277168274, "rewards/format_reward/mean": 0.421875, "rewards/format_reward/std": 0.18298126757144928, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "step": 38 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1690.0, "completions/max_terminated_length": 1690.0, "completions/mean_length": 697.09375, "completions/mean_terminated_length": 697.09375, "completions/min_length": 306.0, "completions/min_terminated_length": 306.0, "epoch": 0.04875, "frac_reward_zero_std": 0.375, "grad_norm": 0.5538021922111511, "learning_rate": 9.473646649103817e-07, "loss": 0.0, "num_tokens": 4612431.0, "reward": 0.13984373211860657, "reward_std": 0.1723020374774933, "rewards/format_reward/mean": 0.4609375, "rewards/format_reward/std": 0.13524486124515533, "rewards/mcq_exact_match_reward/mean": 0.09375, "rewards/mcq_exact_match_reward/std": 0.29378482699394226, "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": 0.046875, "completions/max_length": 2048.0, "completions/max_terminated_length": 1946.0, "completions/mean_length": 914.203125, "completions/mean_terminated_length": 858.4425659179688, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05, "frac_reward_zero_std": 0.25, "grad_norm": 12.278266906738281, "learning_rate": 9.436114097218058e-07, "loss": -0.0, "num_tokens": 4762100.0, "reward": 0.26249998807907104, "reward_std": 0.3294534683227539, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.21875, "rewards/mcq_exact_match_reward/std": 0.4166666865348816, "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": 1638.0, "completions/max_terminated_length": 1638.0, "completions/mean_length": 576.984375, "completions/mean_terminated_length": 576.984375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05125, "frac_reward_zero_std": 0.25, "grad_norm": 2.0300309658050537, "learning_rate": 9.397368756032444e-07, "loss": 0.0, "num_tokens": 4893027.0, "reward": 0.3765624761581421, "reward_std": 0.2795547842979431, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.08768405020236969, "rewards/mcq_exact_match_reward/mean": 0.328125, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "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": 2000.0, "completions/mean_length": 727.34375, "completions/mean_terminated_length": 706.3809814453125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0525, "frac_reward_zero_std": 0.25, "grad_norm": 2.269731044769287, "learning_rate": 9.357421218136386e-07, "loss": -0.0, "num_tokens": 5047777.0, "reward": 0.40625, "reward_std": 0.3007018566131592, "rewards/format_reward/mean": 0.46875, "rewards/format_reward/std": 0.12198751419782639, "rewards/mcq_exact_match_reward/mean": 0.359375, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "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": 1376.0, "completions/max_terminated_length": 1376.0, "completions/mean_length": 559.90625, "completions/mean_terminated_length": 559.90625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05375, "frac_reward_zero_std": 0.25, "grad_norm": 5.062185764312744, "learning_rate": 9.316282404787869e-07, "loss": -0.0, "num_tokens": 5172019.0, "reward": 0.4210937023162842, "reward_std": 0.27887406945228577, "rewards/format_reward/mean": 0.4609375, "rewards/format_reward/std": 0.13524486124515533, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 0.03125, "completions/max_length": 2048.0, "completions/max_terminated_length": 1803.0, "completions/mean_length": 544.40625, "completions/mean_terminated_length": 495.9031982421875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.055, "frac_reward_zero_std": 0.125, "grad_norm": 3.2864391803741455, "learning_rate": 9.273963562927694e-07, "loss": -0.0, "num_tokens": 5282253.0, "reward": 0.3570312261581421, "reward_std": 0.2898201048374176, "rewards/format_reward/mean": 0.4453125, "rewards/format_reward/std": 0.15728822350502014, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "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": 1704.0, "completions/max_terminated_length": 1704.0, "completions/mean_length": 478.953125, "completions/mean_terminated_length": 478.953125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05625, "frac_reward_zero_std": 0.125, "grad_norm": 15.53862476348877, "learning_rate": 9.230476262104676e-07, "loss": 0.0, "num_tokens": 5411594.0, "reward": 0.27968746423721313, "reward_std": 0.3150333762168884, "rewards/format_reward/mean": 0.453125, "rewards/format_reward/std": 0.14689241349697113, "rewards/mcq_exact_match_reward/mean": 0.234375, "rewards/mcq_exact_match_reward/std": 0.42695629596710205, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1733.0, "completions/mean_length": 374.21875, "completions/mean_terminated_length": 347.65081787109375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0575, "frac_reward_zero_std": 0.125, "grad_norm": 7.419568061828613, "learning_rate": 9.185832391312642e-07, "loss": -0.0, "num_tokens": 5524720.0, "reward": 0.35859373211860657, "reward_std": 0.3421567678451538, "rewards/format_reward/mean": 0.4609375, "rewards/format_reward/std": 0.13524486124515533, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "step": 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": 1451.0, "completions/max_terminated_length": 1451.0, "completions/mean_length": 326.015625, "completions/mean_terminated_length": 326.015625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.05875, "frac_reward_zero_std": 0.375, "grad_norm": 15.159464836120605, "learning_rate": 9.1400441557401e-07, "loss": -0.0, "num_tokens": 5639889.0, "reward": 0.20624998211860657, "reward_std": 0.25513988733291626, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "step": 47 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1600.0, "completions/max_terminated_length": 1600.0, "completions/mean_length": 293.96875, "completions/mean_terminated_length": 293.96875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06, "frac_reward_zero_std": 0.375, "grad_norm": 14.93795108795166, "learning_rate": 9.093124073433462e-07, "loss": 0.0, "num_tokens": 5758167.0, "reward": 0.5343749523162842, "reward_std": 0.2867125868797302, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.484375, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "step": 48 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1434.0, "completions/max_terminated_length": 1434.0, "completions/mean_length": 244.828125, "completions/mean_terminated_length": 244.828125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06125, "frac_reward_zero_std": 0.5, "grad_norm": 16.188343048095703, "learning_rate": 9.045084971874737e-07, "loss": -0.0, "num_tokens": 5836036.0, "reward": 0.4874999523162842, "reward_std": 0.24511480331420898, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.4375, "rewards/mcq_exact_match_reward/std": 0.5, "step": 49 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 966.0, "completions/max_terminated_length": 966.0, "completions/mean_length": 98.4375, "completions/mean_terminated_length": 98.4375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0625, "frac_reward_zero_std": 0.5, "grad_norm": 10.723838806152344, "learning_rate": 8.995939984474623e-07, "loss": 0.0, "num_tokens": 5937280.0, "reward": 0.5492187142372131, "reward_std": 0.17791283130645752, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 1634.0, "completions/mean_length": 88.4375, "completions/mean_terminated_length": 57.333335876464844, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06375, "frac_reward_zero_std": 0.75, "grad_norm": 7.924827575683594, "learning_rate": 8.945702546981968e-07, "loss": -0.0, "num_tokens": 6024276.0, "reward": 0.39296871423721313, "reward_std": 0.1099943146109581, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.34375, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "step": 51 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 835.0, "completions/max_terminated_length": 835.0, "completions/mean_length": 48.03125, "completions/mean_terminated_length": 48.03125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.065, "frac_reward_zero_std": 0.25, "grad_norm": 10.166023254394531, "learning_rate": 8.894386393810562e-07, "loss": 0.0, "num_tokens": 6125534.0, "reward": 0.4859374463558197, "reward_std": 0.2746109664440155, "rewards/format_reward/mean": 0.484375, "rewards/format_reward/std": 0.08768405020236969, "rewards/mcq_exact_match_reward/mean": 0.4375, "rewards/mcq_exact_match_reward/std": 0.5, "step": 52 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 893.0, "completions/max_terminated_length": 893.0, "completions/mean_length": 32.25, "completions/mean_terminated_length": 32.25, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06625, "frac_reward_zero_std": 0.625, "grad_norm": 19.42243766784668, "learning_rate": 8.842005554284295e-07, "loss": 0.0, "num_tokens": 6208166.0, "reward": 0.6710937023162842, "reward_std": 0.09162302315235138, "rewards/format_reward/mean": 0.4609375, "rewards/format_reward/std": 0.13524486124515533, "rewards/mcq_exact_match_reward/mean": 0.625, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 53 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 109.0, "completions/max_terminated_length": 109.0, "completions/mean_length": 9.40625, "completions/mean_terminated_length": 9.40625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0675, "frac_reward_zero_std": 0.5, "grad_norm": 22.67922019958496, "learning_rate": 8.788574348801674e-07, "loss": -0.0, "num_tokens": 6304904.0, "reward": 0.453125, "reward_std": 0.22660629451274872, "rewards/format_reward/mean": 0.46875, "rewards/format_reward/std": 0.12198751419782639, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 54 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.06875, "frac_reward_zero_std": 0.75, "grad_norm": 14.258320808410645, "learning_rate": 8.734107384920769e-07, "loss": -0.0, "num_tokens": 6385392.0, "reward": 0.5968749523162842, "reward_std": 0.12255740165710449, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.546875, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 0.0, "completions/max_length": 363.0, "completions/max_terminated_length": 363.0, "completions/mean_length": 11.640625, "completions/mean_terminated_length": 11.640625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07, "frac_reward_zero_std": 0.625, "grad_norm": 19.662662506103516, "learning_rate": 8.678619553365658e-07, "loss": 0.0, "num_tokens": 6482561.0, "reward": 0.5804687142372131, "reward_std": 0.11109752953052521, "rewards/format_reward/mean": 0.4921875, "rewards/format_reward/std": 0.0625, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 56 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 286.0, "completions/max_terminated_length": 286.0, "completions/mean_length": 10.375, "completions/mean_terminated_length": 10.375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07125, "frac_reward_zero_std": 0.625, "grad_norm": 24.95022201538086, "learning_rate": 8.622126023955445e-07, "loss": 0.0, "num_tokens": 6574897.0, "reward": 0.6906249523162842, "reward_std": 0.15992167592048645, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.640625, "rewards/mcq_exact_match_reward/std": 0.4836103618144989, "step": 57 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 170.0, "completions/mean_length": 42.421875, "completions/mean_terminated_length": 10.587302207946777, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0725, "frac_reward_zero_std": 0.5, "grad_norm": 26.536001205444336, "learning_rate": 8.564642241456986e-07, "loss": -0.0, "num_tokens": 6650804.0, "reward": 0.37812498211860657, "reward_std": 0.19939781725406647, "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": 58 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 532.0, "completions/max_terminated_length": 532.0, "completions/mean_length": 14.21875, "completions/mean_terminated_length": 14.21875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07375, "frac_reward_zero_std": 0.625, "grad_norm": 24.962839126586914, "learning_rate": 8.506183921362442e-07, "loss": -0.0, "num_tokens": 6735930.0, "reward": 0.5656249523162842, "reward_std": 0.17358145117759705, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "step": 59 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 281.0, "completions/max_terminated_length": 281.0, "completions/mean_length": 10.296875, "completions/mean_terminated_length": 10.296875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.075, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 8.446767045592829e-07, "loss": 0.0, "num_tokens": 6817557.0, "reward": 0.6749999523162842, "reward_std": 0.0, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.625, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 0.0, "completions/max_length": 231.0, "completions/max_terminated_length": 231.0, "completions/mean_length": 9.90625, "completions/mean_terminated_length": 9.90625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07625, "frac_reward_zero_std": 0.75, "grad_norm": 22.505226135253906, "learning_rate": 8.386407858128706e-07, "loss": 0.0, "num_tokens": 6928655.0, "reward": 0.34687498211860657, "reward_std": 0.10205791890621185, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.296875, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "step": 61 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 67.0, "completions/max_terminated_length": 67.0, "completions/mean_length": 7.515625, "completions/mean_terminated_length": 7.515625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0775, "frac_reward_zero_std": 0.625, "grad_norm": 18.1346492767334, "learning_rate": 8.325122860569241e-07, "loss": -0.0, "num_tokens": 7009872.0, "reward": 0.25312498211860657, "reward_std": 0.15992167592048645, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.203125, "rewards/mcq_exact_match_reward/std": 0.40550529956817627, "step": 62 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.07875, "frac_reward_zero_std": 0.75, "grad_norm": 17.632652282714844, "learning_rate": 8.262928807620843e-07, "loss": -0.0, "num_tokens": 7090744.0, "reward": 0.7218749523162842, "reward_std": 0.12255740165710449, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.671875, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 63 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08, "frac_reward_zero_std": 0.5, "grad_norm": 31.994657516479492, "learning_rate": 8.199842702516582e-07, "loss": -0.0, "num_tokens": 7164832.0, "reward": 0.3781249523162842, "reward_std": 0.24039676785469055, "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": 64 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08125, "frac_reward_zero_std": 0.625, "grad_norm": 28.0701904296875, "learning_rate": 8.135881792367685e-07, "loss": -0.0, "num_tokens": 7258464.0, "reward": 0.7999999523162842, "reward_std": 0.16675157845020294, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.75, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "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": 0.015625, "completions/max_length": 2048.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 37.90625, "completions/mean_terminated_length": 6.000000476837158, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0825, "frac_reward_zero_std": 0.875, "grad_norm": 22.904006958007812, "learning_rate": 8.071063563448339e-07, "loss": 0.0, "num_tokens": 7324714.0, "reward": 0.4406249523162842, "reward_std": 0.04419416934251785, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "step": 66 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08375, "frac_reward_zero_std": 0.625, "grad_norm": 18.512020111083984, "learning_rate": 8.005405736415125e-07, "loss": -0.0, "num_tokens": 7401690.0, "reward": 0.20624998211860657, "reward_std": 0.1962025910615921, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.15625, "rewards/mcq_exact_match_reward/std": 0.36596253514289856, "step": 67 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.085, "frac_reward_zero_std": 0.75, "grad_norm": 30.331026077270508, "learning_rate": 7.938926261462365e-07, "loss": -0.0, "num_tokens": 7475922.0, "reward": 0.37812498211860657, "reward_std": 0.10205792635679245, "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": 68 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08625, "frac_reward_zero_std": 0.625, "grad_norm": 17.431846618652344, "learning_rate": 7.871643313414718e-07, "loss": 0.0, "num_tokens": 7544194.0, "reward": 0.36249998211860657, "reward_std": 0.1462520956993103, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.3125, "rewards/mcq_exact_match_reward/std": 0.467176616191864, "step": 69 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0875, "frac_reward_zero_std": 0.875, "grad_norm": 20.139265060424805, "learning_rate": 7.803575286758363e-07, "loss": -0.0, "num_tokens": 7608914.0, "reward": 0.5812499523162842, "reward_std": 0.0578637532889843, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "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": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.08875, "frac_reward_zero_std": 0.875, "grad_norm": 14.362000465393066, "learning_rate": 7.734740790612136e-07, "loss": -0.0, "num_tokens": 7714186.0, "reward": 0.4562499523162842, "reward_std": 0.0578637532889843, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 71 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.09, "frac_reward_zero_std": 0.375, "grad_norm": 36.90115737915039, "learning_rate": 7.665158643639969e-07, "loss": -0.0, "num_tokens": 7789674.0, "reward": 0.7218749523162842, "reward_std": 0.29826050996780396, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.671875, "rewards/mcq_exact_match_reward/std": 0.4732423722743988, "step": 72 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.09125, "frac_reward_zero_std": 0.75, "grad_norm": 25.01017951965332, "learning_rate": 7.594847868906076e-07, "loss": -0.0, "num_tokens": 7855418.0, "reward": 0.6437499523162842, "reward_std": 0.0883883461356163, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.59375, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 73 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 8.0, "completions/max_terminated_length": 8.0, "completions/mean_length": 6.03125, "completions/mean_terminated_length": 6.03125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0925, "frac_reward_zero_std": 0.875, "grad_norm": 9.84754467010498, "learning_rate": 7.523827688674219e-07, "loss": -0.0, "num_tokens": 7951604.0, "reward": 0.9093749523162842, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.859375, "rewards/mcq_exact_match_reward/std": 0.3503824472427368, "step": 74 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.09375, "frac_reward_zero_std": 0.875, "grad_norm": 23.88617706298828, "learning_rate": 7.452117519152541e-07, "loss": -0.0, "num_tokens": 8031388.0, "reward": 0.33124998211860657, "reward_std": 0.0578637532889843, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.28125, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "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": 0.0, "completions/max_length": 8.0, "completions/max_terminated_length": 8.0, "completions/mean_length": 6.0625, "completions/mean_terminated_length": 6.0625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.095, "frac_reward_zero_std": 0.875, "grad_norm": 14.253681182861328, "learning_rate": 7.379736965185368e-07, "loss": 0.0, "num_tokens": 8109160.0, "reward": 0.44062498211860657, "reward_std": 0.04419416934251785, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "step": 76 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.09625, "frac_reward_zero_std": 0.625, "grad_norm": 29.047731399536133, "learning_rate": 7.306705814893439e-07, "loss": -0.0, "num_tokens": 8204000.0, "reward": 0.5656249523162842, "reward_std": 0.16887323558330536, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.515625, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "step": 77 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.0975, "frac_reward_zero_std": 0.75, "grad_norm": 24.563663482666016, "learning_rate": 7.233044034264033e-07, "loss": -0.0, "num_tokens": 8316568.0, "reward": 0.5187499523162842, "reward_std": 0.1293872892856598, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.46875, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 78 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.09875, "frac_reward_zero_std": 0.75, "grad_norm": 15.42125415802002, "learning_rate": 7.158771761692464e-07, "loss": -0.0, "num_tokens": 8399208.0, "reward": 0.44062498211860657, "reward_std": 0.10205792635679245, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.390625, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "step": 79 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1, "frac_reward_zero_std": 0.625, "grad_norm": 24.344844818115234, "learning_rate": 7.083909302476452e-07, "loss": -0.0, "num_tokens": 8471472.0, "reward": 0.6749999523162842, "reward_std": 0.16675157845020294, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.625, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "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": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.10125, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 7.008477123264847e-07, "loss": 0.0, "num_tokens": 8542920.0, "reward": 0.4249999523162842, "reward_std": 0.0, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.375, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 81 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 13.0, "completions/max_terminated_length": 13.0, "completions/mean_length": 6.6875, "completions/mean_terminated_length": 6.6875, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1025, "frac_reward_zero_std": 0.875, "grad_norm": 26.638757705688477, "learning_rate": 6.932495846462261e-07, "loss": -0.0, "num_tokens": 8620268.0, "reward": 0.5749999284744263, "reward_std": 0.0578637532889843, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 82 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.10375, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 6.855986244591103e-07, "loss": 0.0, "num_tokens": 8725580.0, "reward": 0.7999999523162842, "reward_std": 0.0, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.75, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 83 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 207.0, "completions/max_terminated_length": 207.0, "completions/mean_length": 9.140625, "completions/mean_terminated_length": 9.140625, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.105, "frac_reward_zero_std": 0.875, "grad_norm": 20.473995208740234, "learning_rate": 6.778969234612583e-07, "loss": -0.0, "num_tokens": 8804525.0, "reward": 0.7687499523162842, "reward_std": 0.0578637532889843, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.71875, "rewards/mcq_exact_match_reward/std": 0.4531635046005249, "step": 84 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.10625, "frac_reward_zero_std": 0.5, "grad_norm": 32.69689178466797, "learning_rate": 6.701465872208216e-07, "loss": -0.0, "num_tokens": 8862133.0, "reward": 0.4874999523162842, "reward_std": 0.19727616012096405, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.4375, "rewards/mcq_exact_match_reward/std": 0.5, "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": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1075, "frac_reward_zero_std": 0.75, "grad_norm": 32.79877853393555, "learning_rate": 6.623497346023417e-07, "loss": -0.0, "num_tokens": 8948557.0, "reward": 0.5812499523162842, "reward_std": 0.12467905879020691, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "step": 86 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.10875, "frac_reward_zero_std": 0.875, "grad_norm": 36.093441009521484, "learning_rate": 6.545084971874736e-07, "loss": -0.0, "num_tokens": 9014869.0, "reward": 0.7531249523162842, "reward_std": 0.0646936446428299, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.703125, "rewards/mcq_exact_match_reward/std": 0.4604927599430084, "step": 87 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 6.0, "completions/mean_length": 37.90625, "completions/mean_terminated_length": 6.000000476837158, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.11, "frac_reward_zero_std": 0.75, "grad_norm": 39.25770568847656, "learning_rate": 6.466250186922324e-07, "loss": 0.0, "num_tokens": 9078183.0, "reward": 0.6593749523162842, "reward_std": 0.10205791890621185, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.609375, "rewards/mcq_exact_match_reward/std": 0.4917473793029785, "step": 88 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 15.0, "completions/max_terminated_length": 15.0, "completions/mean_length": 7.125, "completions/mean_terminated_length": 7.125, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.11125, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 6.387014543809223e-07, "loss": 0.0, "num_tokens": 9167535.0, "reward": 0.6687499284744263, "reward_std": 0.0, "rewards/format_reward/mean": 0.4375, "rewards/format_reward/std": 0.1666666716337204, "rewards/mcq_exact_match_reward/mean": 0.625, "rewards/mcq_exact_match_reward/std": 0.48795005679130554, "step": 89 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1125, "frac_reward_zero_std": 0.875, "grad_norm": 10.918007850646973, "learning_rate": 6.307399704769098e-07, "loss": -0.0, "num_tokens": 9270359.0, "reward": 0.5812499523162842, "reward_std": 0.0578637532889843, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.53125, "rewards/mcq_exact_match_reward/std": 0.5029674172401428, "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": 0.0, "completions/max_length": 410.0, "completions/max_terminated_length": 410.0, "completions/mean_length": 15.34375, "completions/mean_terminated_length": 15.34375, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.11375, "frac_reward_zero_std": 0.75, "grad_norm": 29.343151092529297, "learning_rate": 6.227427435703995e-07, "loss": -0.0, "num_tokens": 9360037.0, "reward": 0.7062499523162842, "reward_std": 0.12467905879020691, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.65625, "rewards/mcq_exact_match_reward/std": 0.4787135720252991, "step": 91 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.115, "frac_reward_zero_std": 0.75, "grad_norm": 37.439308166503906, "learning_rate": 6.147119600233758e-07, "loss": -0.0, "num_tokens": 9456917.0, "reward": 1.0187499523162842, "reward_std": 0.0883883461356163, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.96875, "rewards/mcq_exact_match_reward/std": 0.17536810040473938, "step": 92 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.11625, "frac_reward_zero_std": 0.75, "grad_norm": 39.252254486083984, "learning_rate": 6.066498153718734e-07, "loss": 0.0, "num_tokens": 9535725.0, "reward": 0.7999999523162842, "reward_std": 0.0883883386850357, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.75, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 93 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1175, "frac_reward_zero_std": 0.875, "grad_norm": 21.52398681640625, "learning_rate": 5.985585137257401e-07, "loss": -0.0, "num_tokens": 9612989.0, "reward": 1.0187499523162842, "reward_std": 0.0578637532889843, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.96875, "rewards/mcq_exact_match_reward/std": 0.17536810040473938, "step": 94 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.11875, "frac_reward_zero_std": 0.75, "grad_norm": 36.34272766113281, "learning_rate": 5.90440267166055e-07, "loss": -0.0, "num_tokens": 9680565.0, "reward": 0.5968749523162842, "reward_std": 0.12255740165710449, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.546875, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "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": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.12, "frac_reward_zero_std": 0.875, "grad_norm": 39.669925689697266, "learning_rate": 5.82297295140367e-07, "loss": -0.0, "num_tokens": 9764157.0, "reward": 0.5968749523162842, "reward_std": 0.0646936446428299, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.546875, "rewards/mcq_exact_match_reward/std": 0.501733124256134, "step": 96 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.12125, "frac_reward_zero_std": 0.875, "grad_norm": 26.796714782714844, "learning_rate": 5.741318238559209e-07, "loss": -0.0, "num_tokens": 9849037.0, "reward": 0.4562499523162842, "reward_std": 0.0578637532889843, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.40625, "rewards/mcq_exact_match_reward/std": 0.49501484632492065, "step": 97 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.1225, "frac_reward_zero_std": 0.875, "grad_norm": 21.199926376342773, "learning_rate": 5.659460856710345e-07, "loss": -0.0, "num_tokens": 9924885.0, "reward": 0.5343749523162842, "reward_std": 0.04419417306780815, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.484375, "rewards/mcq_exact_match_reward/std": 0.5037065148353577, "step": 98 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.12375, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 5.577423184847931e-07, "loss": 0.0, "num_tokens": 9989325.0, "reward": 0.29999998211860657, "reward_std": 0.0, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.25, "rewards/mcq_exact_match_reward/std": 0.4364357888698578, "step": 99 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 6.0, "completions/max_terminated_length": 6.0, "completions/mean_length": 6.0, "completions/mean_terminated_length": 6.0, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "epoch": 0.125, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "learning_rate": 5.495227651252315e-07, "loss": 0.0, "num_tokens": 10083349.0, "reward": 0.5499999523162842, "reward_std": 0.0, "rewards/format_reward/mean": 0.5, "rewards/format_reward/std": 0.0, "rewards/mcq_exact_match_reward/mean": 0.5, "rewards/mcq_exact_match_reward/std": 0.5039526224136353, "step": 100 } ], "logging_steps": 1, "max_steps": 200, "num_input_tokens_seen": 10083349, "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 }