{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 2.5, "eval_steps": 500, "global_step": 300, "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, "completion_length": 459.225, "completions/clipped_ratio": 0.45, "completions/max_length": 504.6, "completions/max_terminated_length": 405.6, "completions/mean_length": 459.225, "completions/mean_terminated_length": 376.68333740234374, "completions/min_length": 401.6, "completions/min_terminated_length": 350.4, "epoch": 0.08333333333333333, "frac_reward_zero_std": 0.7, "grad_norm": 0.0, "kl": 0.0, "learning_rate": 1.5e-06, "loss": 0.00675208568572998, "num_tokens": 23133.0, "reward": 0.2900000035762787, "reward_std": 0.362790584564209, "rewards/execution_unit_test_reward/mean": 0.03999999687075615, "rewards/execution_unit_test_reward/std": 0.03154700621962547, "rewards/format_and_syntax_reward/mean": 0.25, "rewards/format_and_syntax_reward/std": 0.3943375706672668, "step": 10, "step_time": 48.20472573229999 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 480.95, "completions/clipped_ratio": 0.65, "completions/max_length": 512.0, "completions/max_terminated_length": 306.9, "completions/mean_length": 480.95, "completions/mean_terminated_length": 293.51666870117185, "completions/min_length": 433.2, "completions/min_terminated_length": 279.6, "epoch": 0.16666666666666666, "frac_reward_zero_std": 0.5, "grad_norm": 0.09039819240570068, "kl": 0.0, "learning_rate": 3.1666666666666667e-06, "loss": 0.014680254459381103, "num_tokens": 47503.0, "reward": -0.2174999862909317, "reward_std": 0.5750000238418579, "rewards/execution_unit_test_reward/mean": -0.15500000342726708, "rewards/execution_unit_test_reward/std": 0.05000000074505806, "rewards/format_and_syntax_reward/mean": -0.0625, "rewards/format_and_syntax_reward/std": 0.625, "step": 20, "step_time": 46.95116089439996 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 468.375, "completions/clipped_ratio": 0.525, "completions/max_length": 503.2, "completions/max_terminated_length": 354.6, "completions/mean_length": 468.375, "completions/mean_terminated_length": 333.78333435058596, "completions/min_length": 409.9, "completions/min_terminated_length": 307.5, "epoch": 0.25, "frac_reward_zero_std": 0.5, "grad_norm": 0.0, "kl": 0.0, "learning_rate": 4.833333333333333e-06, "loss": 0.007300767302513123, "num_tokens": 71658.0, "reward": -0.26249998807907104, "reward_std": 0.5677905738353729, "rewards/execution_unit_test_reward/mean": -0.16250000298023223, "rewards/execution_unit_test_reward/std": 0.05654700547456741, "rewards/format_and_syntax_reward/mean": -0.1, "rewards/format_and_syntax_reward/std": 0.5943375706672669, "step": 30, "step_time": 47.35723500230008 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 457.625, "completions/clipped_ratio": 0.55, "completions/max_length": 508.0, "completions/max_terminated_length": 391.4, "completions/mean_length": 457.625, "completions/mean_terminated_length": 361.40000305175784, "completions/min_length": 378.6, "completions/min_terminated_length": 327.4, "epoch": 0.3333333333333333, "frac_reward_zero_std": 0.5, "grad_norm": 0.07833444327116013, "kl": 0.0, "learning_rate": 4.986304738420684e-06, "loss": 0.007216435670852661, "num_tokens": 95007.0, "reward": -0.25, "reward_std": 0.5427905917167664, "rewards/execution_unit_test_reward/mean": -0.1750000037252903, "rewards/execution_unit_test_reward/std": 0.06154700443148613, "rewards/format_and_syntax_reward/mean": -0.075, "rewards/format_and_syntax_reward/std": 0.5443375706672668, "step": 40, "step_time": 48.140497195699844 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 439.3, "completions/clipped_ratio": 0.35, "completions/max_length": 504.1, "completions/max_terminated_length": 447.9, "completions/mean_length": 439.3, "completions/mean_terminated_length": 406.78334045410156, "completions/min_length": 372.4, "completions/min_terminated_length": 372.4, "epoch": 0.4166666666666667, "frac_reward_zero_std": 0.6, "grad_norm": 0.07586903125047684, "kl": 0.0, "learning_rate": 4.939155700460536e-06, "loss": 0.012726987898349761, "num_tokens": 117279.0, "reward": 0.35750000476837157, "reward_std": 0.46010414361953733, "rewards/execution_unit_test_reward/mean": 0.08249999731779098, "rewards/execution_unit_test_reward/std": 0.15061553120613097, "rewards/format_and_syntax_reward/mean": 0.275, "rewards/format_and_syntax_reward/std": 0.3724744915962219, "step": 50, "step_time": 47.27600140520017 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 423.875, "completions/clipped_ratio": 0.275, "completions/max_length": 488.4, "completions/max_terminated_length": 413.1, "completions/mean_length": 423.875, "completions/mean_terminated_length": 360.375, "completions/min_length": 359.5, "completions/min_terminated_length": 308.3, "epoch": 0.5, "frac_reward_zero_std": 0.5, "grad_norm": 0.0987296998500824, "kl": 0.0, "learning_rate": 4.859021265983959e-06, "loss": 0.0107418492436409, "num_tokens": 139234.0, "reward": 0.08750000298023224, "reward_std": 0.5320170760154724, "rewards/execution_unit_test_reward/mean": -0.025000005960464478, "rewards/execution_unit_test_reward/std": 0.2770170696079731, "rewards/format_and_syntax_reward/mean": 0.1125, "rewards/format_and_syntax_reward/std": 0.275, "step": 60, "step_time": 45.66706426839983 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 466.9, "completions/clipped_ratio": 0.475, "completions/max_length": 498.0, "completions/max_terminated_length": 398.7, "completions/mean_length": 466.9, "completions/mean_terminated_length": 381.125, "completions/min_length": 413.9, "completions/min_terminated_length": 362.7, "epoch": 0.5833333333333334, "frac_reward_zero_std": 0.7, "grad_norm": 0.10045502334833145, "kl": 0.0, "learning_rate": 4.746985115747918e-06, "loss": 0.011861865222454072, "num_tokens": 162934.0, "reward": -0.0449999988079071, "reward_std": 0.39837169647216797, "rewards/execution_unit_test_reward/mean": -0.17000000253319741, "rewards/execution_unit_test_reward/std": 0.03464101776480675, "rewards/format_and_syntax_reward/mean": 0.125, "rewards/format_and_syntax_reward/std": 0.4330127120018005, "step": 70, "step_time": 49.01302265149984 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 466.4, "completions/clipped_ratio": 0.5, "completions/max_length": 510.7, "completions/max_terminated_length": 404.5, "completions/mean_length": 466.4, "completions/mean_terminated_length": 380.0166717529297, "completions/min_length": 405.7, "completions/min_terminated_length": 354.5, "epoch": 0.6666666666666666, "frac_reward_zero_std": 0.4, "grad_norm": 0.07955718040466309, "kl": 0.0, "learning_rate": 4.604562346488144e-06, "loss": 0.029622507095336915, "num_tokens": 186506.0, "reward": 0.06250001192092895, "reward_std": 0.6850000381469726, "rewards/execution_unit_test_reward/mean": 1.1175870895385742e-09, "rewards/execution_unit_test_reward/std": 0.16000000312924384, "rewards/format_and_syntax_reward/mean": 0.0625, "rewards/format_and_syntax_reward/std": 0.625, "step": 80, "step_time": 48.009539030099766 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 426.5, "completions/clipped_ratio": 0.3, "completions/max_length": 486.0, "completions/max_terminated_length": 401.7, "completions/mean_length": 426.5, "completions/mean_terminated_length": 347.5083374023437, "completions/min_length": 345.0, "completions/min_terminated_length": 293.8, "epoch": 0.75, "frac_reward_zero_std": 0.5, "grad_norm": 0.08607563376426697, "kl": 0.0, "learning_rate": 4.4336789818349105e-06, "loss": 0.029112178087234496, "num_tokens": 208250.0, "reward": -0.019999998807907104, "reward_std": 0.5427905917167664, "rewards/execution_unit_test_reward/mean": -0.19500000402331352, "rewards/execution_unit_test_reward/std": 0.06154700443148613, "rewards/format_and_syntax_reward/mean": 0.175, "rewards/format_and_syntax_reward/std": 0.5443375706672668, "step": 90, "step_time": 44.824996925399866 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 435.775, "completions/clipped_ratio": 0.3, "completions/max_length": 491.1, "completions/max_terminated_length": 452.1, "completions/mean_length": 435.775, "completions/mean_terminated_length": 413.025, "completions/min_length": 370.6, "completions/min_terminated_length": 370.6, "epoch": 0.8333333333333334, "frac_reward_zero_std": 0.6, "grad_norm": 0.0, "kl": 0.0, "learning_rate": 4.236645926147493e-06, "loss": 0.008134901523590088, "num_tokens": 230813.0, "reward": 0.23750000298023224, "reward_std": 0.47201708555221555, "rewards/execution_unit_test_reward/mean": -0.07500000596046448, "rewards/execution_unit_test_reward/std": 0.15701706930994988, "rewards/format_and_syntax_reward/mean": 0.3125, "rewards/format_and_syntax_reward/std": 0.375, "step": 100, "step_time": 45.33207831669988 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 458.225, "completions/clipped_ratio": 0.375, "completions/max_length": 499.5, "completions/max_terminated_length": 451.7, "completions/mean_length": 458.225, "completions/mean_terminated_length": 422.8, "completions/min_length": 393.4, "completions/min_terminated_length": 393.4, "epoch": 0.9166666666666666, "frac_reward_zero_std": 0.5, "grad_norm": 0.0, "kl": 0.0, "learning_rate": 4.016127713497034e-06, "loss": 0.015300160646438599, "num_tokens": 253874.0, "reward": 0.2575000077486038, "reward_std": 0.5794448137283326, "rewards/execution_unit_test_reward/mean": 0.007499999552965164, "rewards/execution_unit_test_reward/std": 0.3105276867747307, "rewards/format_and_syntax_reward/mean": 0.25, "rewards/format_and_syntax_reward/std": 0.3943375706672668, "step": 110, "step_time": 46.56039770629995 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 434.725, "completions/clipped_ratio": 0.375, "completions/max_length": 504.3, "completions/max_terminated_length": 425.3, "completions/mean_length": 434.725, "completions/mean_terminated_length": 393.9250030517578, "completions/min_length": 360.3, "completions/min_terminated_length": 360.3, "epoch": 1.0, "frac_reward_zero_std": 0.6, "grad_norm": 0.08591324090957642, "kl": 0.0, "learning_rate": 3.7751064744141886e-06, "loss": 0.013795450329780579, "num_tokens": 276019.0, "reward": 0.21250000298023225, "reward_std": 0.6501041293144226, "rewards/execution_unit_test_reward/mean": 0.01249999850988388, "rewards/execution_unit_test_reward/std": 0.25144598111510275, "rewards/format_and_syntax_reward/mean": 0.2, "rewards/format_and_syntax_reward/std": 0.45907703042030334, "step": 120, "step_time": 45.14645728140022 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 451.9, "completions/clipped_ratio": 0.425, "completions/max_length": 506.5, "completions/max_terminated_length": 383.3, "completions/mean_length": 451.9, "completions/mean_terminated_length": 332.2500030517578, "completions/min_length": 381.0, "completions/min_terminated_length": 278.6, "epoch": 1.0833333333333333, "frac_reward_zero_std": 0.6, "grad_norm": 0.0, "kl": 0.0, "learning_rate": 3.516841607689501e-06, "loss": -0.015290644764900208, "num_tokens": 299035.0, "reward": 0.1550000086426735, "reward_std": 0.4900000154972076, "rewards/execution_unit_test_reward/mean": 0.004999993741512299, "rewards/execution_unit_test_reward/std": 0.3100000061094761, "rewards/format_and_syntax_reward/mean": 0.15, "rewards/format_and_syntax_reward/std": 0.2, "step": 130, "step_time": 46.26916497119991 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 431.275, "completions/clipped_ratio": 0.25, "completions/max_length": 478.5, "completions/max_terminated_length": 461.8, "completions/mean_length": 431.275, "completions/mean_terminated_length": 412.9916717529297, "completions/min_length": 372.0, "completions/min_terminated_length": 372.0, "epoch": 1.1666666666666667, "frac_reward_zero_std": 0.9, "grad_norm": 0.0, "kl": 0.0, "learning_rate": 3.244825702596205e-06, "loss": 0.0023515313863754273, "num_tokens": 321226.0, "reward": 0.2425000101327896, "reward_std": 0.11500000953674316, "rewards/execution_unit_test_reward/mean": -0.19500000327825545, "rewards/execution_unit_test_reward/std": 0.010000000149011612, "rewards/format_and_syntax_reward/mean": 0.4375, "rewards/format_and_syntax_reward/std": 0.125, "step": 140, "step_time": 43.70449590049984 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 421.875, "completions/clipped_ratio": 0.3, "completions/max_length": 492.7, "completions/max_terminated_length": 454.8, "completions/mean_length": 421.875, "completions/mean_terminated_length": 389.3666687011719, "completions/min_length": 325.5, "completions/min_terminated_length": 325.5, "epoch": 1.25, "frac_reward_zero_std": 0.6, "grad_norm": 0.0, "kl": 0.0, "learning_rate": 2.9627373076117864e-06, "loss": -0.0014450401067733765, "num_tokens": 342765.0, "reward": 0.18000000417232515, "reward_std": 0.5600000381469726, "rewards/execution_unit_test_reward/mean": -0.07000000476837158, "rewards/execution_unit_test_reward/std": 0.15151817128062248, "rewards/format_and_syntax_reward/mean": 0.25, "rewards/format_and_syntax_reward/std": 0.5, "step": 150, "step_time": 44.239437705200636 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 447.85, "completions/clipped_ratio": 0.375, "completions/max_length": 504.4, "completions/max_terminated_length": 371.0, "completions/mean_length": 447.85, "completions/mean_terminated_length": 323.7166687011719, "completions/min_length": 383.0, "completions/min_terminated_length": 280.6, "epoch": 1.3333333333333333, "frac_reward_zero_std": 0.6, "grad_norm": 0.0, "kl": 0.0, "learning_rate": 2.6743911843603134e-06, "loss": 0.013647961616516113, "num_tokens": 365563.0, "reward": 0.09000000357627869, "reward_std": 0.5937306761741639, "rewards/execution_unit_test_reward/mean": -0.08500000238418579, "rewards/execution_unit_test_reward/std": 0.1527905598282814, "rewards/format_and_syntax_reward/mean": 0.175, "rewards/format_and_syntax_reward/std": 0.48094010949134824, "step": 160, "step_time": 45.15049445520026 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 460.475, "completions/clipped_ratio": 0.4, "completions/max_length": 496.4, "completions/max_terminated_length": 333.8, "completions/mean_length": 460.475, "completions/mean_terminated_length": 299.8833343505859, "completions/min_length": 409.9, "completions/min_terminated_length": 256.3, "epoch": 1.4166666666666667, "frac_reward_zero_std": 0.7, "grad_norm": 0.0, "kl": 0.0, "learning_rate": 2.3836867195056333e-06, "loss": 0.004179513454437256, "num_tokens": 388898.0, "reward": -0.21749999821186067, "reward_std": 0.3805811405181885, "rewards/execution_unit_test_reward/mean": -0.155000002682209, "rewards/execution_unit_test_reward/std": 0.03309401199221611, "rewards/format_and_syntax_reward/mean": -0.0625, "rewards/format_and_syntax_reward/std": 0.4136751413345337, "step": 170, "step_time": 44.43520084530046 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 420.525, "completions/clipped_ratio": 0.2, "completions/max_length": 468.3, "completions/max_terminated_length": 439.9, "completions/mean_length": 420.525, "completions/mean_terminated_length": 403.23333435058595, "completions/min_length": 358.4, "completions/min_terminated_length": 358.4, "epoch": 1.5, "frac_reward_zero_std": 0.8, "grad_norm": 0.0, "kl": 0.0, "learning_rate": 2.094555192236082e-06, "loss": 0.004060398042201996, "num_tokens": 410731.0, "reward": 0.24000000953674316, "reward_std": 0.24279056787490844, "rewards/execution_unit_test_reward/mean": -0.13500000312924385, "rewards/execution_unit_test_reward/std": 0.12154700830578805, "rewards/format_and_syntax_reward/mean": 0.375, "rewards/format_and_syntax_reward/std": 0.14433757066726685, "step": 180, "step_time": 41.72829634220015 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 452.15, "completions/clipped_ratio": 0.4, "completions/max_length": 495.2, "completions/max_terminated_length": 366.9, "completions/mean_length": 452.15, "completions/mean_terminated_length": 333.0, "completions/min_length": 392.6, "completions/min_terminated_length": 290.2, "epoch": 1.5833333333333335, "frac_reward_zero_std": 0.5, "grad_norm": 0.0, "kl": 0.0, "learning_rate": 1.8109066104575023e-06, "loss": -0.005810551345348358, "num_tokens": 433893.0, "reward": -0.1949999988079071, "reward_std": 0.5678947031497955, "rewards/execution_unit_test_reward/mean": -0.1200000025331974, "rewards/execution_unit_test_reward/std": 0.16716253608465195, "rewards/format_and_syntax_reward/mean": -0.075, "rewards/format_and_syntax_reward/std": 0.46681206226348876, "step": 190, "step_time": 44.95948694490016 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 406.925, "completions/clipped_ratio": 0.175, "completions/max_length": 460.7, "completions/max_terminated_length": 449.9, "completions/mean_length": 406.925, "completions/mean_terminated_length": 391.90833740234376, "completions/min_length": 337.4, "completions/min_terminated_length": 337.4, "epoch": 1.6666666666666665, "frac_reward_zero_std": 0.6, "grad_norm": 0.0, "kl": 0.0, "learning_rate": 1.5365768346438798e-06, "loss": 0.01931442618370056, "num_tokens": 454926.0, "reward": 0.2900000035762787, "reward_std": 0.46000003814697266, "rewards/execution_unit_test_reward/mean": 0.03999999612569809, "rewards/execution_unit_test_reward/std": 0.04000000059604645, "rewards/format_and_syntax_reward/mean": 0.25, "rewards/format_and_syntax_reward/std": 0.5, "step": 200, "step_time": 43.089426999100034 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 417.675, "completions/clipped_ratio": 0.25, "completions/max_length": 466.8, "completions/max_terminated_length": 400.0, "completions/mean_length": 417.675, "completions/mean_terminated_length": 357.225, "completions/min_length": 366.6, "completions/min_terminated_length": 315.4, "epoch": 1.75, "frac_reward_zero_std": 0.9, "grad_norm": 0.0929790586233139, "kl": 0.0, "learning_rate": 1.275275704404783e-06, "loss": 0.0020782992243766783, "num_tokens": 476781.0, "reward": -0.044999992847442626, "reward_std": 0.13279056549072266, "rewards/execution_unit_test_reward/mean": -0.17000000253319741, "rewards/execution_unit_test_reward/std": 0.011547005921602248, "rewards/format_and_syntax_reward/mean": 0.125, "rewards/format_and_syntax_reward/std": 0.14433757066726685, "step": 210, "step_time": 42.20883124400079 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 452.25, "completions/clipped_ratio": 0.5, "completions/max_length": 508.5, "completions/max_terminated_length": 388.3, "completions/mean_length": 452.25, "completions/mean_terminated_length": 351.2416687011719, "completions/min_length": 365.0, "completions/min_terminated_length": 313.8, "epoch": 1.8333333333333335, "frac_reward_zero_std": 0.3, "grad_norm": 0.09507235884666443, "kl": 0.0, "learning_rate": 1.0305368692688175e-06, "loss": 0.03854071199893951, "num_tokens": 499931.0, "reward": 0.4500000089406967, "reward_std": 0.9227905988693237, "rewards/execution_unit_test_reward/mean": 0.32500000670552254, "rewards/execution_unit_test_reward/std": 0.36154701113700866, "rewards/format_and_syntax_reward/mean": 0.125, "rewards/format_and_syntax_reward/std": 0.6443375706672668, "step": 220, "step_time": 46.41162945159958 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 455.125, "completions/clipped_ratio": 0.525, "completions/max_length": 496.6, "completions/max_terminated_length": 403.2, "completions/mean_length": 455.125, "completions/mean_terminated_length": 374.40833740234376, "completions/min_length": 395.0, "completions/min_terminated_length": 343.8, "epoch": 1.9166666666666665, "frac_reward_zero_std": 0.4, "grad_norm": 0.09052669256925583, "kl": 0.0, "learning_rate": 8.056700021356695e-07, "loss": 0.014869977533817292, "num_tokens": 523036.0, "reward": -0.037499997019767764, "reward_std": 0.694807642698288, "rewards/execution_unit_test_reward/mean": -0.06250000558793545, "rewards/execution_unit_test_reward/std": 0.18356407433748245, "rewards/format_and_syntax_reward/mean": 0.025, "rewards/format_and_syntax_reward/std": 0.5943375706672669, "step": 230, "step_time": 45.12485095650008 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 447.975, "completions/clipped_ratio": 0.4, "completions/max_length": 498.3, "completions/max_terminated_length": 347.9, "completions/mean_length": 447.975, "completions/mean_terminated_length": 313.02500305175784, "completions/min_length": 381.5, "completions/min_terminated_length": 279.1, "epoch": 2.0, "frac_reward_zero_std": 0.8, "grad_norm": 0.0, "kl": 0.0, "learning_rate": 6.037160416278279e-07, "loss": 0.006232145428657532, "num_tokens": 545963.0, "reward": 0.19750000834465026, "reward_std": 0.20500001311302185, "rewards/execution_unit_test_reward/mean": -0.20250000357627868, "rewards/execution_unit_test_reward/std": 0.02499999925494194, "rewards/format_and_syntax_reward/mean": 0.4, "rewards/format_and_syntax_reward/std": 0.2, "step": 240, "step_time": 45.35569858550025 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 408.575, "completions/clipped_ratio": 0.325, "completions/max_length": 455.1, "completions/max_terminated_length": 371.2, "completions/mean_length": 408.575, "completions/mean_terminated_length": 338.81667175292966, "completions/min_length": 351.3, "completions/min_terminated_length": 300.1, "epoch": 2.0833333333333335, "frac_reward_zero_std": 0.5, "grad_norm": 0.08674506843090057, "kl": 0.0, "learning_rate": 4.27406068612396e-07, "loss": 0.009747776389122009, "num_tokens": 567190.0, "reward": -0.19749999046325684, "reward_std": 0.5752082467079163, "rewards/execution_unit_test_reward/mean": -0.0600000012665987, "rewards/execution_unit_test_reward/std": 0.2712310619652271, "rewards/format_and_syntax_reward/mean": -0.1375, "rewards/format_and_syntax_reward/std": 0.36994898319244385, "step": 250, "step_time": 42.926730689599935 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 451.725, "completions/clipped_ratio": 0.325, "completions/max_length": 499.9, "completions/max_terminated_length": 463.3, "completions/mean_length": 451.725, "completions/mean_terminated_length": 425.0416687011719, "completions/min_length": 384.5, "completions/min_terminated_length": 384.5, "epoch": 2.1666666666666665, "frac_reward_zero_std": 0.5, "grad_norm": 0.0, "kl": 0.0, "learning_rate": 2.791243730174903e-07, "loss": 0.02311110943555832, "num_tokens": 590499.0, "reward": 0.12249999940395355, "reward_std": 0.6048076510429382, "rewards/execution_unit_test_reward/mean": -0.06500000581145286, "rewards/execution_unit_test_reward/std": 0.16856407523155212, "rewards/format_and_syntax_reward/mean": 0.1875, "rewards/format_and_syntax_reward/std": 0.5193375706672668, "step": 260, "step_time": 45.8208733094998 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 421.275, "completions/clipped_ratio": 0.3, "completions/max_length": 473.4, "completions/max_terminated_length": 398.6, "completions/mean_length": 421.275, "completions/mean_terminated_length": 356.5916717529297, "completions/min_length": 349.1, "completions/min_terminated_length": 297.9, "epoch": 2.25, "frac_reward_zero_std": 0.6, "grad_norm": 0.0, "kl": 0.0, "learning_rate": 1.6087621040117157e-07, "loss": -0.010004176199436188, "num_tokens": 612030.0, "reward": 0.01000000536441803, "reward_std": 0.5700000286102295, "rewards/execution_unit_test_reward/mean": -0.11500000357627868, "rewards/execution_unit_test_reward/std": 0.13000000044703483, "rewards/format_and_syntax_reward/mean": 0.125, "rewards/format_and_syntax_reward/std": 0.5, "step": 270, "step_time": 42.53594893429981 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 446.5, "completions/clipped_ratio": 0.25, "completions/max_length": 505.4, "completions/max_terminated_length": 481.8, "completions/mean_length": 446.5, "completions/mean_terminated_length": 425.9750061035156, "completions/min_length": 369.0, "completions/min_terminated_length": 369.0, "epoch": 2.3333333333333335, "frac_reward_zero_std": 0.8, "grad_norm": 0.0, "kl": 0.0, "learning_rate": 7.426068431000883e-08, "loss": 0.007316985726356506, "num_tokens": 634918.0, "reward": 0.41750000715255736, "reward_std": 0.20500001311302185, "rewards/execution_unit_test_reward/mean": 0.017499996721744536, "rewards/execution_unit_test_reward/std": 0.02499999925494194, "rewards/format_and_syntax_reward/mean": 0.4, "rewards/format_and_syntax_reward/std": 0.2, "step": 280, "step_time": 45.02933877839987 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 445.4, "completions/clipped_ratio": 0.35, "completions/max_length": 487.7, "completions/max_terminated_length": 366.8, "completions/mean_length": 445.4, "completions/mean_terminated_length": 331.9916687011719, "completions/min_length": 398.3, "completions/min_terminated_length": 295.9, "epoch": 2.4166666666666665, "frac_reward_zero_std": 0.6, "grad_norm": 0.0, "kl": 0.0, "learning_rate": 2.0449121146845773e-08, "loss": -0.007108403742313385, "num_tokens": 657610.0, "reward": 0.30000001192092896, "reward_std": 0.43000001311302183, "rewards/execution_unit_test_reward/mean": 0.08749999664723873, "rewards/execution_unit_test_reward/std": 0.14500000178813935, "rewards/format_and_syntax_reward/mean": 0.2125, "rewards/format_and_syntax_reward/std": 0.325, "step": 290, "step_time": 43.80834858289982 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 432.475, "completions/clipped_ratio": 0.25, "completions/max_length": 491.7, "completions/max_terminated_length": 464.8, "completions/mean_length": 432.475, "completions/mean_terminated_length": 413.8250030517578, "completions/min_length": 364.2, "completions/min_terminated_length": 364.2, "epoch": 2.5, "frac_reward_zero_std": 0.7, "grad_norm": 0.09202820807695389, "kl": 0.0, "learning_rate": 1.692299905944883e-10, "loss": 0.014088675379753113, "num_tokens": 679905.0, "reward": 0.18000001311302186, "reward_std": 0.4087344169616699, "rewards/execution_unit_test_reward/mean": -0.07000000514090061, "rewards/execution_unit_test_reward/std": 0.2270825244486332, "rewards/format_and_syntax_reward/mean": 0.25, "rewards/format_and_syntax_reward/std": 0.25, "step": 300, "step_time": 45.583720880399234 } ], "logging_steps": 10, "max_steps": 300, "num_input_tokens_seen": 679905, "num_train_epochs": 3, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }