{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 1.2, "eval_steps": 500, "global_step": 150, "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": 611.125, "completions/clipped_ratio": 0.0, "completions/max_length": 929.0, "completions/max_terminated_length": 929.0, "completions/mean_length": 611.125, "completions/mean_terminated_length": 611.125, "completions/min_length": 437.0, "completions/min_terminated_length": 437.0, "epoch": 0.008, "frac_reward_zero_std": 0.0, "grad_norm": 0.09433281421661377, "kl": 6.0034221291971335e-06, "learning_rate": 0.0, "loss": 2.9802322387695312e-08, "num_tokens": 11257.0, "reward": 0.2749999463558197, "reward_std": 1.5298373699188232, "rewards/reward_environment_execution/mean": -0.16249999403953552, "rewards/reward_environment_execution/std": 0.2794765532016754, "rewards/reward_format_compliance/mean": 0.25, "rewards/reward_format_compliance/std": 1.0350984334945679, "rewards/reward_investigation_quality/mean": 0.1875, "rewards/reward_investigation_quality/std": 0.15526476502418518, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 535.25, "completions/clipped_ratio": 0.0, "completions/max_length": 741.0, "completions/max_terminated_length": 741.0, "completions/mean_length": 535.25, "completions/mean_terminated_length": 535.25, "completions/min_length": 90.0, "completions/min_terminated_length": 90.0, "epoch": 0.016, "frac_reward_zero_std": 0.0, "grad_norm": 0.1564614325761795, "kl": 7.112513173979096e-06, "learning_rate": 1.0000000000000002e-06, "loss": 6.705522537231445e-08, "num_tokens": 21755.0, "reward": 0.3695124387741089, "reward_std": 1.6131230592727661, "rewards/reward_environment_execution/mean": -0.09298750758171082, "rewards/reward_environment_execution/std": 0.3859241008758545, "rewards/reward_format_compliance/mean": 0.25, "rewards/reward_format_compliance/std": 1.0350984334945679, "rewards/reward_investigation_quality/mean": 0.21250000596046448, "rewards/reward_investigation_quality/std": 0.18850919604301453, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 684.0, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 712.0, "completions/mean_length": 684.0, "completions/mean_terminated_length": 635.4285888671875, "completions/min_length": 554.0, "completions/min_terminated_length": 554.0, "epoch": 0.024, "frac_reward_zero_std": 0.0, "grad_norm": 0.10158229619264603, "kl": 6.621002114570729e-06, "learning_rate": 2.0000000000000003e-06, "loss": 5.21540641784668e-08, "num_tokens": 33423.0, "reward": 0.2749999463558197, "reward_std": 1.5298373699188232, "rewards/reward_environment_execution/mean": -0.16249999403953552, "rewards/reward_environment_execution/std": 0.2794765532016754, "rewards/reward_format_compliance/mean": 0.25, "rewards/reward_format_compliance/std": 1.0350984334945679, "rewards/reward_investigation_quality/mean": 0.1875, "rewards/reward_investigation_quality/std": 0.15526476502418518, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 571.75, "completions/clipped_ratio": 0.0, "completions/max_length": 788.0, "completions/max_terminated_length": 788.0, "completions/mean_length": 571.75, "completions/mean_terminated_length": 571.75, "completions/min_length": 430.0, "completions/min_terminated_length": 430.0, "epoch": 0.032, "frac_reward_zero_std": 0.0, "grad_norm": 0.11366094648838043, "kl": 6.710507079787931e-06, "learning_rate": 3e-06, "loss": 6.705522537231445e-08, "num_tokens": 44221.0, "reward": -0.4350000321865082, "reward_std": 1.5298373699188232, "rewards/reward_environment_execution/mean": -0.29750001430511475, "rewards/reward_environment_execution/std": 0.2794765532016754, "rewards/reward_format_compliance/mean": -0.25, "rewards/reward_format_compliance/std": 1.0350984334945679, "rewards/reward_investigation_quality/mean": 0.11250000447034836, "rewards/reward_investigation_quality/std": 0.15526476502418518, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 676.625, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 878.0, "completions/mean_length": 676.625, "completions/mean_terminated_length": 627.0, "completions/min_length": 509.0, "completions/min_terminated_length": 509.0, "epoch": 0.04, "frac_reward_zero_std": 0.0, "grad_norm": 0.08936131745576859, "kl": 5.939470554494619e-06, "learning_rate": 4.000000000000001e-06, "loss": 8.568167686462402e-08, "num_tokens": 55842.0, "reward": 0.6299999356269836, "reward_std": 1.4199999570846558, "rewards/reward_environment_execution/mean": -0.0949999988079071, "rewards/reward_environment_execution/std": 0.2499714195728302, "rewards/reward_format_compliance/mean": 0.5, "rewards/reward_format_compliance/std": 0.9258201122283936, "rewards/reward_investigation_quality/mean": 0.22500000894069672, "rewards/reward_investigation_quality/std": 0.13887302577495575, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 502.875, "completions/clipped_ratio": 0.0, "completions/max_length": 702.0, "completions/max_terminated_length": 702.0, "completions/mean_length": 502.875, "completions/mean_terminated_length": 502.875, "completions/min_length": 130.0, "completions/min_terminated_length": 130.0, "epoch": 0.048, "frac_reward_zero_std": 0.0, "grad_norm": 0.11681948602199554, "kl": 7.0123659838827734e-06, "learning_rate": 5e-06, "loss": 9.685754776000977e-08, "num_tokens": 66085.0, "reward": -0.4350000321865082, "reward_std": 1.5298373699188232, "rewards/reward_environment_execution/mean": -0.29750001430511475, "rewards/reward_environment_execution/std": 0.2794765532016754, "rewards/reward_format_compliance/mean": -0.25, "rewards/reward_format_compliance/std": 1.0350984334945679, "rewards/reward_investigation_quality/mean": 0.11250000447034836, "rewards/reward_investigation_quality/std": 0.15526476502418518, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 727.25, "completions/clipped_ratio": 0.0, "completions/max_length": 1006.0, "completions/max_terminated_length": 1006.0, "completions/mean_length": 727.25, "completions/mean_terminated_length": 727.25, "completions/min_length": 516.0, "completions/min_terminated_length": 516.0, "epoch": 0.056, "frac_reward_zero_std": 0.5, "grad_norm": 0.12677940726280212, "kl": 7.118041764897498e-06, "learning_rate": 6e-06, "loss": 6.705522537231445e-08, "num_tokens": 78275.0, "reward": 0.6299999356269836, "reward_std": 0.8198373317718506, "rewards/reward_environment_execution/mean": -0.0949999988079071, "rewards/reward_environment_execution/std": 0.2499714195728302, "rewards/reward_format_compliance/mean": 0.5, "rewards/reward_format_compliance/std": 0.9258201122283936, "rewards/reward_investigation_quality/mean": 0.22500000894069672, "rewards/reward_investigation_quality/std": 0.13887302577495575, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 574.5, "completions/clipped_ratio": 0.0, "completions/max_length": 681.0, "completions/max_terminated_length": 681.0, "completions/mean_length": 574.5, "completions/mean_terminated_length": 574.5, "completions/min_length": 463.0, "completions/min_terminated_length": 463.0, "epoch": 0.064, "frac_reward_zero_std": 0.0, "grad_norm": 0.09633345901966095, "kl": 7.368475223756832e-06, "learning_rate": 7e-06, "loss": 6.705522537231445e-08, "num_tokens": 89079.0, "reward": -0.04375004768371582, "reward_std": 1.4476807117462158, "rewards/reward_environment_execution/mean": -0.21875, "rewards/reward_environment_execution/std": 0.30258116126060486, "rewards/reward_format_compliance/mean": 0.0, "rewards/reward_format_compliance/std": 1.0690449476242065, "rewards/reward_investigation_quality/mean": 0.13750000298023224, "rewards/reward_investigation_quality/std": 0.1505940705537796, "rewards/reward_os_mechanics/mean": 0.03750000149011612, "rewards/reward_os_mechanics/std": 0.1060660257935524, "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, "completion_length": 611.75, "completions/clipped_ratio": 0.0, "completions/max_length": 727.0, "completions/max_terminated_length": 727.0, "completions/mean_length": 611.75, "completions/mean_terminated_length": 611.75, "completions/min_length": 522.0, "completions/min_terminated_length": 522.0, "epoch": 0.072, "frac_reward_zero_std": 0.0, "grad_norm": 0.08932310342788696, "kl": 7.581263787415082e-06, "learning_rate": 8.000000000000001e-06, "loss": 9.313225746154785e-08, "num_tokens": 100169.0, "reward": 0.2749999463558197, "reward_std": 1.5298373699188232, "rewards/reward_environment_execution/mean": -0.16249999403953552, "rewards/reward_environment_execution/std": 0.2794765532016754, "rewards/reward_format_compliance/mean": 0.25, "rewards/reward_format_compliance/std": 1.0350984334945679, "rewards/reward_investigation_quality/mean": 0.1875, "rewards/reward_investigation_quality/std": 0.15526476502418518, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 567.5, "completions/clipped_ratio": 0.0, "completions/max_length": 728.0, "completions/max_terminated_length": 728.0, "completions/mean_length": 567.5, "completions/mean_terminated_length": 567.5, "completions/min_length": 372.0, "completions/min_terminated_length": 372.0, "epoch": 0.08, "frac_reward_zero_std": 0.5, "grad_norm": 0.08864814788103104, "kl": 1.1717055826920841e-05, "learning_rate": 9e-06, "loss": 8.195638656616211e-08, "num_tokens": 110925.0, "reward": 0.6299999356269836, "reward_std": 0.8198373317718506, "rewards/reward_environment_execution/mean": -0.0949999988079071, "rewards/reward_environment_execution/std": 0.2499714195728302, "rewards/reward_format_compliance/mean": 0.5, "rewards/reward_format_compliance/std": 0.9258201122283936, "rewards/reward_investigation_quality/mean": 0.22500000894069672, "rewards/reward_investigation_quality/std": 0.13887302577495575, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 581.75, "completions/clipped_ratio": 0.0, "completions/max_length": 807.0, "completions/max_terminated_length": 807.0, "completions/mean_length": 581.75, "completions/mean_terminated_length": 581.75, "completions/min_length": 345.0, "completions/min_terminated_length": 345.0, "epoch": 0.088, "frac_reward_zero_std": 0.0, "grad_norm": 0.09192224591970444, "kl": 1.68721625186663e-05, "learning_rate": 1e-05, "loss": 1.7136335372924805e-07, "num_tokens": 121783.0, "reward": -0.4350000321865082, "reward_std": 1.5298373699188232, "rewards/reward_environment_execution/mean": -0.29750001430511475, "rewards/reward_environment_execution/std": 0.2794765532016754, "rewards/reward_format_compliance/mean": -0.25, "rewards/reward_format_compliance/std": 1.0350984334945679, "rewards/reward_investigation_quality/mean": 0.11250000447034836, "rewards/reward_investigation_quality/std": 0.15526476502418518, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 589.875, "completions/clipped_ratio": 0.0, "completions/max_length": 703.0, "completions/max_terminated_length": 703.0, "completions/mean_length": 589.875, "completions/mean_terminated_length": 589.875, "completions/min_length": 495.0, "completions/min_terminated_length": 495.0, "epoch": 0.096, "frac_reward_zero_std": 0.0, "grad_norm": 0.11080238968133926, "kl": 2.686391167117108e-05, "learning_rate": 1.1000000000000001e-05, "loss": 2.5331974029541016e-07, "num_tokens": 132710.0, "reward": 0.6299999356269836, "reward_std": 1.4199999570846558, "rewards/reward_environment_execution/mean": -0.0949999988079071, "rewards/reward_environment_execution/std": 0.2499714195728302, "rewards/reward_format_compliance/mean": 0.5, "rewards/reward_format_compliance/std": 0.9258201122283936, "rewards/reward_investigation_quality/mean": 0.22500000894069672, "rewards/reward_investigation_quality/std": 0.13887302577495575, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 671.25, "completions/clipped_ratio": 0.0, "completions/max_length": 828.0, "completions/max_terminated_length": 828.0, "completions/mean_length": 671.25, "completions/mean_terminated_length": 671.25, "completions/min_length": 370.0, "completions/min_terminated_length": 370.0, "epoch": 0.104, "frac_reward_zero_std": 0.0, "grad_norm": 0.1135329157114029, "kl": 3.927483930965536e-05, "learning_rate": 1.2e-05, "loss": 3.8743019104003906e-07, "num_tokens": 144288.0, "reward": 0.3623749613761902, "reward_std": 1.6417193412780762, "rewards/reward_environment_execution/mean": -0.10012500733137131, "rewards/reward_environment_execution/std": 0.37162211537361145, "rewards/reward_format_compliance/mean": 0.25, "rewards/reward_format_compliance/std": 1.0350984334945679, "rewards/reward_investigation_quality/mean": 0.21250000596046448, "rewards/reward_investigation_quality/std": 0.18850919604301453, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 635.875, "completions/clipped_ratio": 0.0, "completions/max_length": 790.0, "completions/max_terminated_length": 790.0, "completions/mean_length": 635.875, "completions/mean_terminated_length": 635.875, "completions/min_length": 513.0, "completions/min_terminated_length": 513.0, "epoch": 0.112, "frac_reward_zero_std": 0.0, "grad_norm": 0.10361003875732422, "kl": 0.00011971511844421912, "learning_rate": 1.3000000000000001e-05, "loss": 1.1771917343139648e-06, "num_tokens": 155579.0, "reward": 0.2749999463558197, "reward_std": 1.5298373699188232, "rewards/reward_environment_execution/mean": -0.16249999403953552, "rewards/reward_environment_execution/std": 0.2794765532016754, "rewards/reward_format_compliance/mean": 0.25, "rewards/reward_format_compliance/std": 1.0350984334945679, "rewards/reward_investigation_quality/mean": 0.1875, "rewards/reward_investigation_quality/std": 0.15526476502418518, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 605.5, "completions/clipped_ratio": 0.0, "completions/max_length": 882.0, "completions/max_terminated_length": 882.0, "completions/mean_length": 605.5, "completions/mean_terminated_length": 605.5, "completions/min_length": 457.0, "completions/min_terminated_length": 457.0, "epoch": 0.12, "frac_reward_zero_std": 0.0, "grad_norm": 0.11182169616222382, "kl": 0.00015782948776177363, "learning_rate": 1.4e-05, "loss": 1.5720725059509277e-06, "num_tokens": 166631.0, "reward": -0.08000004291534424, "reward_std": 1.6396746635437012, "rewards/reward_environment_execution/mean": -0.23000000417232513, "rewards/reward_environment_execution/std": 0.28864213824272156, "rewards/reward_format_compliance/mean": 0.0, "rewards/reward_format_compliance/std": 1.0690449476242065, "rewards/reward_investigation_quality/mean": 0.15000000596046448, "rewards/reward_investigation_quality/std": 0.16035674512386322, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 698.875, "completions/clipped_ratio": 0.0, "completions/max_length": 974.0, "completions/max_terminated_length": 974.0, "completions/mean_length": 698.875, "completions/mean_terminated_length": 698.875, "completions/min_length": 573.0, "completions/min_terminated_length": 573.0, "epoch": 0.128, "frac_reward_zero_std": 0.0, "grad_norm": 0.09252525866031647, "kl": 0.00016497836440976243, "learning_rate": 1.5000000000000002e-05, "loss": 3.6209821701049805e-06, "num_tokens": 178430.0, "reward": 0.6262499690055847, "reward_std": 0.8273373246192932, "rewards/reward_environment_execution/mean": -0.09875000268220901, "rewards/reward_environment_execution/std": 0.24787309765815735, "rewards/reward_format_compliance/mean": 0.5, "rewards/reward_format_compliance/std": 0.9258201122283936, "rewards/reward_investigation_quality/mean": 0.22500000894069672, "rewards/reward_investigation_quality/std": 0.13887302577495575, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 680.625, "completions/clipped_ratio": 0.0, "completions/max_length": 815.0, "completions/max_terminated_length": 815.0, "completions/mean_length": 680.625, "completions/mean_terminated_length": 680.625, "completions/min_length": 458.0, "completions/min_terminated_length": 458.0, "epoch": 0.136, "frac_reward_zero_std": 0.0, "grad_norm": 0.09367026388645172, "kl": 0.000305861960441689, "learning_rate": 1.6000000000000003e-05, "loss": 3.03611159324646e-06, "num_tokens": 190251.0, "reward": 0.2749999463558197, "reward_std": 1.5298373699188232, "rewards/reward_environment_execution/mean": -0.16249999403953552, "rewards/reward_environment_execution/std": 0.2794765532016754, "rewards/reward_format_compliance/mean": 0.25, "rewards/reward_format_compliance/std": 1.0350984334945679, "rewards/reward_investigation_quality/mean": 0.1875, "rewards/reward_investigation_quality/std": 0.15526476502418518, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 675.125, "completions/clipped_ratio": 0.0, "completions/max_length": 861.0, "completions/max_terminated_length": 861.0, "completions/mean_length": 675.125, "completions/mean_terminated_length": 675.125, "completions/min_length": 544.0, "completions/min_terminated_length": 544.0, "epoch": 0.144, "frac_reward_zero_std": 0.5, "grad_norm": 0.062321849167346954, "kl": 0.00029328936034289654, "learning_rate": 1.7e-05, "loss": 2.9280781745910645e-06, "num_tokens": 201860.0, "reward": 0.6299999356269836, "reward_std": 0.8198373317718506, "rewards/reward_environment_execution/mean": -0.0949999988079071, "rewards/reward_environment_execution/std": 0.2499714195728302, "rewards/reward_format_compliance/mean": 0.5, "rewards/reward_format_compliance/std": 0.9258201122283936, "rewards/reward_investigation_quality/mean": 0.22500000894069672, "rewards/reward_investigation_quality/std": 0.13887302577495575, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 700.375, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 828.0, "completions/mean_length": 700.375, "completions/mean_terminated_length": 654.1428833007812, "completions/min_length": 417.0, "completions/min_terminated_length": 417.0, "epoch": 0.152, "frac_reward_zero_std": 0.0, "grad_norm": 0.09901226311922073, "kl": 0.0005741232034779387, "learning_rate": 1.8e-05, "loss": 5.729496479034424e-06, "num_tokens": 213851.0, "reward": 0.2749999463558197, "reward_std": 1.5298373699188232, "rewards/reward_environment_execution/mean": -0.16250000894069672, "rewards/reward_environment_execution/std": 0.2794765532016754, "rewards/reward_format_compliance/mean": 0.25, "rewards/reward_format_compliance/std": 1.0350984334945679, "rewards/reward_investigation_quality/mean": 0.1875, "rewards/reward_investigation_quality/std": 0.15526476502418518, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 629.375, "completions/clipped_ratio": 0.0, "completions/max_length": 801.0, "completions/max_terminated_length": 801.0, "completions/mean_length": 629.375, "completions/mean_terminated_length": 629.375, "completions/min_length": 516.0, "completions/min_terminated_length": 516.0, "epoch": 0.16, "frac_reward_zero_std": 0.5, "grad_norm": 0.0724208727478981, "kl": 0.0002936920918727992, "learning_rate": 1.9e-05, "loss": 2.8994754757150076e-06, "num_tokens": 225110.0, "reward": 0.9849998950958252, "reward_std": 0.7099999785423279, "rewards/reward_environment_execution/mean": -0.027499999850988388, "rewards/reward_environment_execution/std": 0.19091883301734924, "rewards/reward_format_compliance/mean": 0.75, "rewards/reward_format_compliance/std": 0.7071067690849304, "rewards/reward_investigation_quality/mean": 0.26250001788139343, "rewards/reward_investigation_quality/std": 0.1060660257935524, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 680.75, "completions/clipped_ratio": 0.0, "completions/max_length": 1015.0, "completions/max_terminated_length": 1015.0, "completions/mean_length": 680.75, "completions/mean_terminated_length": 680.75, "completions/min_length": 468.0, "completions/min_terminated_length": 468.0, "epoch": 0.168, "frac_reward_zero_std": 0.0, "grad_norm": 0.107891745865345, "kl": 0.001748525508446619, "learning_rate": 2e-05, "loss": 1.1645257472991943e-05, "num_tokens": 236760.0, "reward": 0.9837498664855957, "reward_std": 0.7124999761581421, "rewards/reward_environment_execution/mean": -0.028749998658895493, "rewards/reward_environment_execution/std": 0.1904458999633789, "rewards/reward_format_compliance/mean": 0.75, "rewards/reward_format_compliance/std": 0.7071067690849304, "rewards/reward_investigation_quality/mean": 0.26250001788139343, "rewards/reward_investigation_quality/std": 0.1060660257935524, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 669.5, "completions/clipped_ratio": 0.0, "completions/max_length": 785.0, "completions/max_terminated_length": 785.0, "completions/mean_length": 669.5, "completions/mean_terminated_length": 669.5, "completions/min_length": 557.0, "completions/min_terminated_length": 557.0, "epoch": 0.176, "frac_reward_zero_std": 0.5, "grad_norm": 0.0779874250292778, "kl": 0.0005601062766800169, "learning_rate": 1.9999067159645222e-05, "loss": 5.599111318588257e-06, "num_tokens": 248320.0, "reward": 0.9849998950958252, "reward_std": 0.7099999785423279, "rewards/reward_environment_execution/mean": -0.027499999850988388, "rewards/reward_environment_execution/std": 0.19091883301734924, "rewards/reward_format_compliance/mean": 0.75, "rewards/reward_format_compliance/std": 0.7071067690849304, "rewards/reward_investigation_quality/mean": 0.26250001788139343, "rewards/reward_investigation_quality/std": 0.1060660257935524, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 654.125, "completions/clipped_ratio": 0.0, "completions/max_length": 751.0, "completions/max_terminated_length": 751.0, "completions/mean_length": 654.125, "completions/mean_terminated_length": 654.125, "completions/min_length": 587.0, "completions/min_terminated_length": 587.0, "epoch": 0.184, "frac_reward_zero_std": 0.5, "grad_norm": 0.0675627738237381, "kl": 0.0004718652198789641, "learning_rate": 1.999626881261911e-05, "loss": 4.708766937255859e-06, "num_tokens": 259937.0, "reward": 0.6299999356269836, "reward_std": 0.8198373317718506, "rewards/reward_environment_execution/mean": -0.0949999988079071, "rewards/reward_environment_execution/std": 0.2499714195728302, "rewards/reward_format_compliance/mean": 0.5, "rewards/reward_format_compliance/std": 0.9258201122283936, "rewards/reward_investigation_quality/mean": 0.22500000894069672, "rewards/reward_investigation_quality/std": 0.13887302577495575, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 675.125, "completions/clipped_ratio": 0.0, "completions/max_length": 847.0, "completions/max_terminated_length": 847.0, "completions/mean_length": 675.125, "completions/mean_terminated_length": 675.125, "completions/min_length": 485.0, "completions/min_terminated_length": 485.0, "epoch": 0.192, "frac_reward_zero_std": 1.0, "grad_norm": 0.00018513108079787344, "kl": 0.0006105515094532166, "learning_rate": 1.9991605481003865e-05, "loss": 6.105515240051318e-06, "num_tokens": 271542.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 675.625, "completions/clipped_ratio": 0.0, "completions/max_length": 820.0, "completions/max_terminated_length": 820.0, "completions/mean_length": 675.625, "completions/mean_terminated_length": 675.625, "completions/min_length": 483.0, "completions/min_terminated_length": 483.0, "epoch": 0.2, "frac_reward_zero_std": 0.5, "grad_norm": 0.06521738320589066, "kl": 0.0008642823413538281, "learning_rate": 1.998507803482828e-05, "loss": 8.620321750640869e-06, "num_tokens": 283343.0, "reward": 0.6299999356269836, "reward_std": 0.8198373317718506, "rewards/reward_environment_execution/mean": -0.0949999988079071, "rewards/reward_environment_execution/std": 0.2499714195728302, "rewards/reward_format_compliance/mean": 0.5, "rewards/reward_format_compliance/std": 0.9258201122283936, "rewards/reward_investigation_quality/mean": 0.22500000894069672, "rewards/reward_investigation_quality/std": 0.13887302577495575, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 750.0, "completions/clipped_ratio": 0.0, "completions/max_length": 964.0, "completions/max_terminated_length": 964.0, "completions/mean_length": 750.0, "completions/mean_terminated_length": 750.0, "completions/min_length": 503.0, "completions/min_terminated_length": 503.0, "epoch": 0.208, "frac_reward_zero_std": 0.5, "grad_norm": 0.06086711958050728, "kl": 0.0008782152945059352, "learning_rate": 1.9976687691905394e-05, "loss": 8.761882781982422e-06, "num_tokens": 295539.0, "reward": 0.9849998950958252, "reward_std": 0.7099999785423279, "rewards/reward_environment_execution/mean": -0.027499999850988388, "rewards/reward_environment_execution/std": 0.19091883301734924, "rewards/reward_format_compliance/mean": 0.75, "rewards/reward_format_compliance/std": 0.7071067690849304, "rewards/reward_investigation_quality/mean": 0.26250001788139343, "rewards/reward_investigation_quality/std": 0.1060660257935524, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 670.75, "completions/clipped_ratio": 0.0, "completions/max_length": 808.0, "completions/max_terminated_length": 808.0, "completions/mean_length": 670.75, "completions/mean_terminated_length": 670.75, "completions/min_length": 500.0, "completions/min_terminated_length": 500.0, "epoch": 0.216, "frac_reward_zero_std": 1.0, "grad_norm": 0.0002566609182395041, "kl": 0.0010894250153796747, "learning_rate": 1.9966436017605296e-05, "loss": 1.0894251317949966e-05, "num_tokens": 307113.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 623.5, "completions/clipped_ratio": 0.0, "completions/max_length": 750.0, "completions/max_terminated_length": 750.0, "completions/mean_length": 623.5, "completions/mean_terminated_length": 623.5, "completions/min_length": 407.0, "completions/min_terminated_length": 407.0, "epoch": 0.224, "frac_reward_zero_std": 0.5, "grad_norm": 0.06886623799800873, "kl": 0.0011775159764511045, "learning_rate": 1.9954324924563088e-05, "loss": 1.178685124614276e-05, "num_tokens": 318481.0, "reward": 0.9849998950958252, "reward_std": 0.7099999785423279, "rewards/reward_environment_execution/mean": -0.027499999850988388, "rewards/reward_environment_execution/std": 0.19091883301734924, "rewards/reward_format_compliance/mean": 0.75, "rewards/reward_format_compliance/std": 0.7071067690849304, "rewards/reward_investigation_quality/mean": 0.26250001788139343, "rewards/reward_investigation_quality/std": 0.1060660257935524, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 581.0, "completions/clipped_ratio": 0.0, "completions/max_length": 671.0, "completions/max_terminated_length": 671.0, "completions/mean_length": 581.0, "completions/mean_terminated_length": 581.0, "completions/min_length": 450.0, "completions/min_terminated_length": 450.0, "epoch": 0.232, "frac_reward_zero_std": 0.5, "grad_norm": 0.07106765359640121, "kl": 0.0015617744356859475, "learning_rate": 1.9940356672322037e-05, "loss": 1.5626415915903635e-05, "num_tokens": 329341.0, "reward": 0.6299999356269836, "reward_std": 0.8198373317718506, "rewards/reward_environment_execution/mean": -0.0949999988079071, "rewards/reward_environment_execution/std": 0.2499714195728302, "rewards/reward_format_compliance/mean": 0.5, "rewards/reward_format_compliance/std": 0.9258201122283936, "rewards/reward_investigation_quality/mean": 0.22500000894069672, "rewards/reward_investigation_quality/std": 0.13887302577495575, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 724.875, "completions/clipped_ratio": 0.0, "completions/max_length": 922.0, "completions/max_terminated_length": 922.0, "completions/mean_length": 724.875, "completions/mean_terminated_length": 724.875, "completions/min_length": 645.0, "completions/min_terminated_length": 645.0, "epoch": 0.24, "frac_reward_zero_std": 0.5, "grad_norm": 0.06778275966644287, "kl": 0.007046287857519928, "learning_rate": 1.9924533866912017e-05, "loss": 7.044896483421326e-05, "num_tokens": 341692.0, "reward": 0.9849998950958252, "reward_std": 0.7099999785423279, "rewards/reward_environment_execution/mean": -0.027499999850988388, "rewards/reward_environment_execution/std": 0.19091883301734924, "rewards/reward_format_compliance/mean": 0.75, "rewards/reward_format_compliance/std": 0.7071067690849304, "rewards/reward_investigation_quality/mean": 0.26250001788139343, "rewards/reward_investigation_quality/std": 0.1060660257935524, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 654.0, "completions/clipped_ratio": 0.0, "completions/max_length": 886.0, "completions/max_terminated_length": 886.0, "completions/mean_length": 654.0, "completions/mean_terminated_length": 654.0, "completions/min_length": 320.0, "completions/min_terminated_length": 320.0, "epoch": 0.248, "frac_reward_zero_std": 0.5, "grad_norm": 0.06569182872772217, "kl": 0.0018616169109009206, "learning_rate": 1.9906859460363307e-05, "loss": 1.860725024016574e-05, "num_tokens": 353120.0, "reward": 0.9849998950958252, "reward_std": 0.7099999785423279, "rewards/reward_environment_execution/mean": -0.027499999850988388, "rewards/reward_environment_execution/std": 0.19091883301734924, "rewards/reward_format_compliance/mean": 0.75, "rewards/reward_format_compliance/std": 0.7071067690849304, "rewards/reward_investigation_quality/mean": 0.26250001788139343, "rewards/reward_investigation_quality/std": 0.1060660257935524, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 691.375, "completions/clipped_ratio": 0.0, "completions/max_length": 803.0, "completions/max_terminated_length": 803.0, "completions/mean_length": 691.375, "completions/mean_terminated_length": 691.375, "completions/min_length": 378.0, "completions/min_terminated_length": 378.0, "epoch": 0.256, "frac_reward_zero_std": 0.5, "grad_norm": 0.0545475035905838, "kl": 0.0018646777753019705, "learning_rate": 1.988733675015585e-05, "loss": 1.8647670003701933e-05, "num_tokens": 365179.0, "reward": 0.9849998950958252, "reward_std": 0.7099999785423279, "rewards/reward_environment_execution/mean": -0.027499999850988388, "rewards/reward_environment_execution/std": 0.19091883301734924, "rewards/reward_format_compliance/mean": 0.75, "rewards/reward_format_compliance/std": 0.7071067690849304, "rewards/reward_investigation_quality/mean": 0.26250001788139343, "rewards/reward_investigation_quality/std": 0.1060660257935524, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 665.25, "completions/clipped_ratio": 0.0, "completions/max_length": 975.0, "completions/max_terminated_length": 975.0, "completions/mean_length": 665.25, "completions/mean_terminated_length": 665.25, "completions/min_length": 481.0, "completions/min_terminated_length": 481.0, "epoch": 0.264, "frac_reward_zero_std": 0.5, "grad_norm": 0.08151132613420486, "kl": 0.0020467515059863217, "learning_rate": 1.9865969378604023e-05, "loss": 2.0466744899749756e-05, "num_tokens": 376697.0, "reward": 0.9849998950958252, "reward_std": 0.7099999785423279, "rewards/reward_environment_execution/mean": -0.027499999850988388, "rewards/reward_environment_execution/std": 0.19091883301734924, "rewards/reward_format_compliance/mean": 0.75, "rewards/reward_format_compliance/std": 0.7071067690849304, "rewards/reward_investigation_quality/mean": 0.26250001788139343, "rewards/reward_investigation_quality/std": 0.1060660257935524, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 774.0, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 962.0, "completions/mean_length": 774.0, "completions/mean_terminated_length": 738.2857666015625, "completions/min_length": 556.0, "completions/min_terminated_length": 556.0, "epoch": 0.272, "frac_reward_zero_std": 1.0, "grad_norm": 0.0003262929094489664, "kl": 0.0015715667832409963, "learning_rate": 1.9842761332177115e-05, "loss": 1.571566826896742e-05, "num_tokens": 389465.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 647.5, "completions/clipped_ratio": 0.0, "completions/max_length": 842.0, "completions/max_terminated_length": 842.0, "completions/mean_length": 647.5, "completions/mean_terminated_length": 647.5, "completions/min_length": 383.0, "completions/min_terminated_length": 383.0, "epoch": 0.28, "frac_reward_zero_std": 0.0, "grad_norm": 0.1350807547569275, "kl": 0.0030019799596630037, "learning_rate": 1.9817716940755586e-05, "loss": 3.001093864440918e-05, "num_tokens": 400845.0, "reward": 0.6299999356269836, "reward_std": 1.4199999570846558, "rewards/reward_environment_execution/mean": -0.0949999988079071, "rewards/reward_environment_execution/std": 0.2499714195728302, "rewards/reward_format_compliance/mean": 0.5, "rewards/reward_format_compliance/std": 0.9258201122283936, "rewards/reward_investigation_quality/mean": 0.22500000894069672, "rewards/reward_investigation_quality/std": 0.13887302577495575, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 706.0, "completions/clipped_ratio": 0.0, "completions/max_length": 879.0, "completions/max_terminated_length": 879.0, "completions/mean_length": 706.0, "completions/mean_terminated_length": 706.0, "completions/min_length": 493.0, "completions/min_terminated_length": 493.0, "epoch": 0.288, "frac_reward_zero_std": 0.5, "grad_norm": 0.07488328218460083, "kl": 0.0025595930637791753, "learning_rate": 1.979084087682323e-05, "loss": 2.5570392608642578e-05, "num_tokens": 412881.0, "reward": 0.9849998950958252, "reward_std": 0.7099999785423279, "rewards/reward_environment_execution/mean": -0.027499999850988388, "rewards/reward_environment_execution/std": 0.19091883301734924, "rewards/reward_format_compliance/mean": 0.75, "rewards/reward_format_compliance/std": 0.7071067690849304, "rewards/reward_investigation_quality/mean": 0.26250001788139343, "rewards/reward_investigation_quality/std": 0.1060660257935524, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 724.75, "completions/clipped_ratio": 0.0, "completions/max_length": 1021.0, "completions/max_terminated_length": 1021.0, "completions/mean_length": 724.75, "completions/mean_terminated_length": 724.75, "completions/min_length": 541.0, "completions/min_terminated_length": 541.0, "epoch": 0.296, "frac_reward_zero_std": 1.0, "grad_norm": 0.0008801713702268898, "kl": 0.003089791862294078, "learning_rate": 1.9762138154595448e-05, "loss": 3.089791789534502e-05, "num_tokens": 425227.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 665.375, "completions/clipped_ratio": 0.0, "completions/max_length": 794.0, "completions/max_terminated_length": 794.0, "completions/mean_length": 665.375, "completions/mean_terminated_length": 665.375, "completions/min_length": 584.0, "completions/min_terminated_length": 584.0, "epoch": 0.304, "frac_reward_zero_std": 1.0, "grad_norm": 0.0004410329565871507, "kl": 0.0021486460755113512, "learning_rate": 1.9731614129083756e-05, "loss": 2.1486459445441142e-05, "num_tokens": 436754.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 670.375, "completions/clipped_ratio": 0.0, "completions/max_length": 808.0, "completions/max_terminated_length": 808.0, "completions/mean_length": 670.375, "completions/mean_terminated_length": 670.375, "completions/min_length": 466.0, "completions/min_terminated_length": 466.0, "epoch": 0.312, "frac_reward_zero_std": 0.5, "grad_norm": 0.08116427809000015, "kl": 0.0025287687312811613, "learning_rate": 1.9699274495096712e-05, "loss": 2.52649188041687e-05, "num_tokens": 448321.0, "reward": 0.9849998950958252, "reward_std": 0.7099999785423279, "rewards/reward_environment_execution/mean": -0.027499999850988388, "rewards/reward_environment_execution/std": 0.19091883301734924, "rewards/reward_format_compliance/mean": 0.75, "rewards/reward_format_compliance/std": 0.7071067690849304, "rewards/reward_investigation_quality/mean": 0.26250001788139343, "rewards/reward_investigation_quality/std": 0.1060660257935524, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 696.625, "completions/clipped_ratio": 0.0, "completions/max_length": 907.0, "completions/max_terminated_length": 907.0, "completions/mean_length": 696.625, "completions/mean_terminated_length": 696.625, "completions/min_length": 430.0, "completions/min_terminated_length": 430.0, "epoch": 0.32, "frac_reward_zero_std": 1.0, "grad_norm": 0.0004672395880334079, "kl": 0.002916595127317123, "learning_rate": 1.9665125286177448e-05, "loss": 2.9165948944864795e-05, "num_tokens": 460114.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 625.5, "completions/clipped_ratio": 0.0, "completions/max_length": 781.0, "completions/max_terminated_length": 781.0, "completions/mean_length": 625.5, "completions/mean_terminated_length": 625.5, "completions/min_length": 523.0, "completions/min_terminated_length": 523.0, "epoch": 0.328, "frac_reward_zero_std": 1.0, "grad_norm": 0.0008413000032305717, "kl": 0.0035934818733949214, "learning_rate": 1.9629172873477995e-05, "loss": 3.593481960706413e-05, "num_tokens": 471342.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 745.875, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 902.0, "completions/mean_length": 745.875, "completions/mean_terminated_length": 706.1428833007812, "completions/min_length": 524.0, "completions/min_terminated_length": 524.0, "epoch": 0.336, "frac_reward_zero_std": 1.0, "grad_norm": 0.0004357508150860667, "kl": 0.0025905392394633964, "learning_rate": 1.9591423964570634e-05, "loss": 2.590539224911481e-05, "num_tokens": 483537.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 640.75, "completions/clipped_ratio": 0.0, "completions/max_length": 785.0, "completions/max_terminated_length": 785.0, "completions/mean_length": 640.75, "completions/mean_terminated_length": 640.75, "completions/min_length": 512.0, "completions/min_terminated_length": 512.0, "epoch": 0.344, "frac_reward_zero_std": 1.0, "grad_norm": 0.0004914621822535992, "kl": 0.003195204713847488, "learning_rate": 1.9551885602196482e-05, "loss": 3.1952044082572684e-05, "num_tokens": 494859.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 720.25, "completions/clipped_ratio": 0.0, "completions/max_length": 880.0, "completions/max_terminated_length": 880.0, "completions/mean_length": 720.25, "completions/mean_terminated_length": 720.25, "completions/min_length": 601.0, "completions/min_terminated_length": 601.0, "epoch": 0.352, "frac_reward_zero_std": 1.0, "grad_norm": 0.0007082747761160135, "kl": 0.0027196566516067833, "learning_rate": 1.9510565162951538e-05, "loss": 2.7196565497433767e-05, "num_tokens": 507169.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 749.5, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 883.0, "completions/mean_length": 749.5, "completions/mean_terminated_length": 710.2857666015625, "completions/min_length": 606.0, "completions/min_terminated_length": 606.0, "epoch": 0.36, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005853723268955946, "kl": 0.0031615312909707427, "learning_rate": 1.9467470355910438e-05, "loss": 3.1615312764188275e-05, "num_tokens": 519529.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 675.875, "completions/clipped_ratio": 0.0, "completions/max_length": 878.0, "completions/max_terminated_length": 878.0, "completions/mean_length": 675.875, "completions/mean_terminated_length": 675.875, "completions/min_length": 539.0, "completions/min_terminated_length": 539.0, "epoch": 0.368, "frac_reward_zero_std": 1.0, "grad_norm": 0.0008831220911815763, "kl": 0.0038355615688487887, "learning_rate": 1.9422609221188208e-05, "loss": 3.8355618016794324e-05, "num_tokens": 531484.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 702.75, "completions/clipped_ratio": 0.0, "completions/max_length": 842.0, "completions/max_terminated_length": 842.0, "completions/mean_length": 702.75, "completions/mean_terminated_length": 702.75, "completions/min_length": 331.0, "completions/min_terminated_length": 331.0, "epoch": 0.376, "frac_reward_zero_std": 0.5, "grad_norm": 0.10168822109699249, "kl": 0.0033635341096669436, "learning_rate": 1.9375990128440205e-05, "loss": 2.777203917503357e-05, "num_tokens": 543478.0, "reward": 1.338749885559082, "reward_std": 0.002499997615814209, "rewards/reward_environment_execution/mean": 0.038750000298023224, "rewards/reward_environment_execution/std": 0.0035355337895452976, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 606.5, "completions/clipped_ratio": 0.0, "completions/max_length": 736.0, "completions/max_terminated_length": 736.0, "completions/mean_length": 606.5, "completions/mean_terminated_length": 606.5, "completions/min_length": 541.0, "completions/min_terminated_length": 541.0, "epoch": 0.384, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005576707189902663, "kl": 0.00318920097197406, "learning_rate": 1.932762177530064e-05, "loss": 3.1892010156298056e-05, "num_tokens": 554542.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 826.875, "completions/clipped_ratio": 0.25, "completions/max_length": 1024.0, "completions/max_terminated_length": 921.0, "completions/mean_length": 826.875, "completions/mean_terminated_length": 761.1666870117188, "completions/min_length": 474.0, "completions/min_terminated_length": 474.0, "epoch": 0.392, "frac_reward_zero_std": 1.0, "grad_norm": 0.0004255030071362853, "kl": 0.003332219843287021, "learning_rate": 1.9277513185759847e-05, "loss": 3.33221978507936e-05, "num_tokens": 567537.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 625.375, "completions/clipped_ratio": 0.0, "completions/max_length": 768.0, "completions/max_terminated_length": 768.0, "completions/mean_length": 625.375, "completions/mean_terminated_length": 625.375, "completions/min_length": 520.0, "completions/min_terminated_length": 520.0, "epoch": 0.4, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005748015246354043, "kl": 0.003716042934684083, "learning_rate": 1.922567370848072e-05, "loss": 3.7160429201321676e-05, "num_tokens": 578764.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 613.0, "completions/clipped_ratio": 0.0, "completions/max_length": 826.0, "completions/max_terminated_length": 826.0, "completions/mean_length": 613.0, "completions/mean_terminated_length": 613.0, "completions/min_length": 174.0, "completions/min_terminated_length": 174.0, "epoch": 0.408, "frac_reward_zero_std": 0.0, "grad_norm": 0.11565838754177094, "kl": 0.004887384086032398, "learning_rate": 1.917211301505453e-05, "loss": 4.3004751205444336e-05, "num_tokens": 589876.0, "reward": 0.9837498664855957, "reward_std": 0.7124999761581421, "rewards/reward_environment_execution/mean": -0.028749998658895493, "rewards/reward_environment_execution/std": 0.1904458999633789, "rewards/reward_format_compliance/mean": 0.75, "rewards/reward_format_compliance/std": 0.7071067690849304, "rewards/reward_investigation_quality/mean": 0.26250001788139343, "rewards/reward_investigation_quality/std": 0.1060660257935524, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 662.0, "completions/clipped_ratio": 0.0, "completions/max_length": 785.0, "completions/max_terminated_length": 785.0, "completions/mean_length": 662.0, "completions/mean_terminated_length": 662.0, "completions/min_length": 463.0, "completions/min_terminated_length": 463.0, "epoch": 0.416, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005133210797794163, "kl": 0.0035537532239686698, "learning_rate": 1.9116841098196538e-05, "loss": 3.553753049345687e-05, "num_tokens": 601384.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 608.875, "completions/clipped_ratio": 0.0, "completions/max_length": 779.0, "completions/max_terminated_length": 779.0, "completions/mean_length": 608.875, "completions/mean_terminated_length": 608.875, "completions/min_length": 444.0, "completions/min_terminated_length": 444.0, "epoch": 0.424, "frac_reward_zero_std": 1.0, "grad_norm": 0.0008719735778868198, "kl": 0.004926819354295731, "learning_rate": 1.9059868269881637e-05, "loss": 4.92681901960168e-05, "num_tokens": 612651.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 643.625, "completions/clipped_ratio": 0.0, "completions/max_length": 836.0, "completions/max_terminated_length": 836.0, "completions/mean_length": 643.625, "completions/mean_terminated_length": 643.625, "completions/min_length": 343.0, "completions/min_terminated_length": 343.0, "epoch": 0.432, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005881673423573375, "kl": 0.003908562459400855, "learning_rate": 1.9001205159420512e-05, "loss": 3.908562212018296e-05, "num_tokens": 623988.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 644.75, "completions/clipped_ratio": 0.0, "completions/max_length": 751.0, "completions/max_terminated_length": 751.0, "completions/mean_length": 644.75, "completions/mean_terminated_length": 644.75, "completions/min_length": 511.0, "completions/min_terminated_length": 511.0, "epoch": 0.44, "frac_reward_zero_std": 1.0, "grad_norm": 0.000741733645554632, "kl": 0.004500469920458272, "learning_rate": 1.8940862711476515e-05, "loss": 4.500470095081255e-05, "num_tokens": 635686.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 721.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1003.0, "completions/max_terminated_length": 1003.0, "completions/mean_length": 721.0, "completions/mean_terminated_length": 721.0, "completions/min_length": 528.0, "completions/min_terminated_length": 528.0, "epoch": 0.448, "frac_reward_zero_std": 1.0, "grad_norm": 0.0006931996904313564, "kl": 0.0035460830258671194, "learning_rate": 1.8878852184023754e-05, "loss": 3.5460830986266956e-05, "num_tokens": 647662.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 733.625, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 830.0, "completions/mean_length": 733.625, "completions/mean_terminated_length": 692.1428833007812, "completions/min_length": 556.0, "completions/min_terminated_length": 556.0, "epoch": 0.456, "frac_reward_zero_std": 0.5, "grad_norm": 0.06685224920511246, "kl": 0.0071609383303439245, "learning_rate": 1.8815185146246718e-05, "loss": 7.160753011703491e-05, "num_tokens": 659887.0, "reward": 0.9849998950958252, "reward_std": 0.7099999785423279, "rewards/reward_environment_execution/mean": -0.027499999850988388, "rewards/reward_environment_execution/std": 0.19091883301734924, "rewards/reward_format_compliance/mean": 0.75, "rewards/reward_format_compliance/std": 0.7071067690849304, "rewards/reward_investigation_quality/mean": 0.26250001788139343, "rewards/reward_investigation_quality/std": 0.1060660257935524, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 701.625, "completions/clipped_ratio": 0.0, "completions/max_length": 773.0, "completions/max_terminated_length": 773.0, "completions/mean_length": 701.625, "completions/mean_terminated_length": 701.625, "completions/min_length": 516.0, "completions/min_terminated_length": 516.0, "epoch": 0.464, "frac_reward_zero_std": 1.0, "grad_norm": 0.000510046782437712, "kl": 0.0030409153987420723, "learning_rate": 1.8749873476381827e-05, "loss": 3.040915180463344e-05, "num_tokens": 671868.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 717.125, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 809.0, "completions/mean_length": 717.125, "completions/mean_terminated_length": 673.2857666015625, "completions/min_length": 526.0, "completions/min_terminated_length": 526.0, "epoch": 0.472, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005348174017854035, "kl": 0.003948982805013657, "learning_rate": 1.8682929359501338e-05, "loss": 3.948983066948131e-05, "num_tokens": 683981.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 665.125, "completions/clipped_ratio": 0.0, "completions/max_length": 911.0, "completions/max_terminated_length": 911.0, "completions/mean_length": 665.125, "completions/mean_terminated_length": 665.125, "completions/min_length": 389.0, "completions/min_terminated_length": 389.0, "epoch": 0.48, "frac_reward_zero_std": 1.0, "grad_norm": 0.0006783441640436649, "kl": 0.004887666902504861, "learning_rate": 1.8614365285240002e-05, "loss": 4.887667091679759e-05, "num_tokens": 695514.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 702.875, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 806.0, "completions/mean_length": 702.875, "completions/mean_terminated_length": 657.0, "completions/min_length": 517.0, "completions/min_terminated_length": 517.0, "epoch": 0.488, "frac_reward_zero_std": 1.0, "grad_norm": 0.0004363492480479181, "kl": 0.0033030706690624356, "learning_rate": 1.8544194045464888e-05, "loss": 3.3030708436854184e-05, "num_tokens": 707349.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 667.5, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 812.0, "completions/mean_length": 667.5, "completions/mean_terminated_length": 616.5714721679688, "completions/min_length": 424.0, "completions/min_terminated_length": 424.0, "epoch": 0.496, "frac_reward_zero_std": 1.0, "grad_norm": 0.0008751379209570587, "kl": 0.0047638857358833775, "learning_rate": 1.8472428731888836e-05, "loss": 4.7638852265663445e-05, "num_tokens": 718877.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 701.0, "completions/clipped_ratio": 0.0, "completions/max_length": 1000.0, "completions/max_terminated_length": 1000.0, "completions/mean_length": 701.0, "completions/mean_terminated_length": 701.0, "completions/min_length": 525.0, "completions/min_terminated_length": 525.0, "epoch": 0.504, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005635059787891805, "kl": 0.004204738041153178, "learning_rate": 1.8399082733627967e-05, "loss": 4.204738070257008e-05, "num_tokens": 730845.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 683.25, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 791.0, "completions/mean_length": 683.25, "completions/mean_terminated_length": 634.5714721679688, "completions/min_length": 500.0, "completions/min_terminated_length": 500.0, "epoch": 0.512, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005522868596017361, "kl": 0.0036764607211807743, "learning_rate": 1.832416973470368e-05, "loss": 3.676460255519487e-05, "num_tokens": 742683.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 760.0, "completions/clipped_ratio": 0.25, "completions/max_length": 1024.0, "completions/max_terminated_length": 777.0, "completions/mean_length": 760.0, "completions/mean_terminated_length": 672.0, "completions/min_length": 449.0, "completions/min_terminated_length": 449.0, "epoch": 0.52, "frac_reward_zero_std": 0.5, "grad_norm": 0.0575263537466526, "kl": 0.004310534393880516, "learning_rate": 1.8247703711489684e-05, "loss": 3.725290298461914e-05, "num_tokens": 754971.0, "reward": 1.338749885559082, "reward_std": 0.002499997615814209, "rewards/reward_environment_execution/mean": 0.038750000298023224, "rewards/reward_environment_execution/std": 0.0035355337895452976, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 585.625, "completions/clipped_ratio": 0.0, "completions/max_length": 748.0, "completions/max_terminated_length": 748.0, "completions/mean_length": 585.625, "completions/mean_terminated_length": 585.625, "completions/min_length": 452.0, "completions/min_terminated_length": 452.0, "epoch": 0.528, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005746590322814882, "kl": 0.0049333779024891555, "learning_rate": 1.816969893010442e-05, "loss": 4.933377567795105e-05, "num_tokens": 765856.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 665.75, "completions/clipped_ratio": 0.0, "completions/max_length": 970.0, "completions/max_terminated_length": 970.0, "completions/mean_length": 665.75, "completions/mean_terminated_length": 665.75, "completions/min_length": 479.0, "completions/min_terminated_length": 479.0, "epoch": 0.536, "frac_reward_zero_std": 1.0, "grad_norm": 0.0006066298228688538, "kl": 0.004155741800786927, "learning_rate": 1.8090169943749477e-05, "loss": 4.155741771683097e-05, "num_tokens": 777562.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 650.875, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 797.0, "completions/mean_length": 650.875, "completions/mean_terminated_length": 597.5714721679688, "completions/min_length": 347.0, "completions/min_terminated_length": 347.0, "epoch": 0.544, "frac_reward_zero_std": 1.0, "grad_norm": 0.000580288004130125, "kl": 0.004517835797742009, "learning_rate": 1.8009131589994418e-05, "loss": 4.5178356231190264e-05, "num_tokens": 788973.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 689.5, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 766.0, "completions/mean_length": 689.5, "completions/mean_terminated_length": 641.7142944335938, "completions/min_length": 389.0, "completions/min_terminated_length": 389.0, "epoch": 0.552, "frac_reward_zero_std": 1.0, "grad_norm": 0.0007625495200045407, "kl": 0.004702944046584889, "learning_rate": 1.7926598988008584e-05, "loss": 4.7029439883772284e-05, "num_tokens": 800709.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 687.375, "completions/clipped_ratio": 0.0, "completions/max_length": 1019.0, "completions/max_terminated_length": 1019.0, "completions/mean_length": 687.375, "completions/mean_terminated_length": 687.375, "completions/min_length": 478.0, "completions/min_terminated_length": 478.0, "epoch": 0.56, "frac_reward_zero_std": 1.0, "grad_norm": 0.0006017910782247782, "kl": 0.003289921907708049, "learning_rate": 1.7842587535740315e-05, "loss": 3.289921733085066e-05, "num_tokens": 812752.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 707.875, "completions/clipped_ratio": 0.0, "completions/max_length": 1004.0, "completions/max_terminated_length": 1004.0, "completions/mean_length": 707.875, "completions/mean_terminated_length": 707.875, "completions/min_length": 545.0, "completions/min_terminated_length": 545.0, "epoch": 0.568, "frac_reward_zero_std": 0.5, "grad_norm": 0.07566738873720169, "kl": 0.004167352628428489, "learning_rate": 1.77571129070442e-05, "loss": 3.58390825567767e-05, "num_tokens": 824819.0, "reward": 1.338749885559082, "reward_std": 0.002499997615814209, "rewards/reward_environment_execution/mean": 0.038750000298023224, "rewards/reward_environment_execution/std": 0.0035355337895452976, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 660.125, "completions/clipped_ratio": 0.0, "completions/max_length": 815.0, "completions/max_terminated_length": 815.0, "completions/mean_length": 660.125, "completions/mean_terminated_length": 660.125, "completions/min_length": 239.0, "completions/min_terminated_length": 239.0, "epoch": 0.576, "frac_reward_zero_std": 1.0, "grad_norm": 0.0011171274818480015, "kl": 0.0053537557541858405, "learning_rate": 1.7670191048756827e-05, "loss": 5.353755477699451e-05, "num_tokens": 836312.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 679.75, "completions/clipped_ratio": 0.0, "completions/max_length": 793.0, "completions/max_terminated_length": 793.0, "completions/mean_length": 679.75, "completions/mean_terminated_length": 679.75, "completions/min_length": 512.0, "completions/min_terminated_length": 512.0, "epoch": 0.584, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005916557274758816, "kl": 0.00392132272827439, "learning_rate": 1.758183817772163e-05, "loss": 3.921322786482051e-05, "num_tokens": 848126.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 752.25, "completions/clipped_ratio": 0.0, "completions/max_length": 940.0, "completions/max_terminated_length": 940.0, "completions/mean_length": 752.25, "completions/mean_terminated_length": 752.25, "completions/min_length": 664.0, "completions/min_terminated_length": 664.0, "epoch": 0.592, "frac_reward_zero_std": 1.0, "grad_norm": 0.0008296913001686335, "kl": 0.0038888415147084743, "learning_rate": 1.749207077776331e-05, "loss": 3.8888414565008134e-05, "num_tokens": 860696.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 741.875, "completions/clipped_ratio": 0.0, "completions/max_length": 1018.0, "completions/max_terminated_length": 1018.0, "completions/mean_length": 741.875, "completions/mean_terminated_length": 741.875, "completions/min_length": 501.0, "completions/min_terminated_length": 501.0, "epoch": 0.6, "frac_reward_zero_std": 1.0, "grad_norm": 0.0008696228032931685, "kl": 0.0057216553832404315, "learning_rate": 1.740090559661252e-05, "loss": 5.721655543311499e-05, "num_tokens": 873003.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 642.875, "completions/clipped_ratio": 0.0, "completions/max_length": 940.0, "completions/max_terminated_length": 940.0, "completions/mean_length": 642.875, "completions/mean_terminated_length": 642.875, "completions/min_length": 527.0, "completions/min_terminated_length": 527.0, "epoch": 0.608, "frac_reward_zero_std": 1.0, "grad_norm": 0.005777810234576464, "kl": 0.014752466755453497, "learning_rate": 1.730835964278124e-05, "loss": 0.00014752465358469635, "num_tokens": 884350.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 762.875, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 878.0, "completions/mean_length": 762.875, "completions/mean_terminated_length": 725.5714721679688, "completions/min_length": 541.0, "completions/min_terminated_length": 541.0, "epoch": 0.616, "frac_reward_zero_std": 1.0, "grad_norm": 0.0008344343514181674, "kl": 0.0048244605713989586, "learning_rate": 1.721445018238956e-05, "loss": 4.824460484087467e-05, "num_tokens": 897017.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 745.25, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 835.0, "completions/mean_length": 745.25, "completions/mean_terminated_length": 705.4285888671875, "completions/min_length": 552.0, "completions/min_terminated_length": 552.0, "epoch": 0.624, "frac_reward_zero_std": 1.0, "grad_norm": 0.001024937373585999, "kl": 0.00511919436394237, "learning_rate": 1.7119194735944336e-05, "loss": 5.119194611324929e-05, "num_tokens": 909195.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 711.875, "completions/clipped_ratio": 0.0, "completions/max_length": 981.0, "completions/max_terminated_length": 981.0, "completions/mean_length": 711.875, "completions/mean_terminated_length": 711.875, "completions/min_length": 594.0, "completions/min_terminated_length": 594.0, "epoch": 0.632, "frac_reward_zero_std": 1.0, "grad_norm": 0.00042360470979474485, "kl": 0.003573804016923532, "learning_rate": 1.7022611075070476e-05, "loss": 3.573803405743092e-05, "num_tokens": 921258.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 725.125, "completions/clipped_ratio": 0.0, "completions/max_length": 885.0, "completions/max_terminated_length": 885.0, "completions/mean_length": 725.125, "completions/mean_terminated_length": 725.125, "completions/min_length": 600.0, "completions/min_terminated_length": 600.0, "epoch": 0.64, "frac_reward_zero_std": 0.5, "grad_norm": 0.06427791714668274, "kl": 0.004669759000535123, "learning_rate": 1.6924717219195258e-05, "loss": 4.6700239181518555e-05, "num_tokens": 933427.0, "reward": 0.9849998950958252, "reward_std": 0.7099999785423279, "rewards/reward_environment_execution/mean": -0.027499999850988388, "rewards/reward_environment_execution/std": 0.19091883301734924, "rewards/reward_format_compliance/mean": 0.75, "rewards/reward_format_compliance/std": 0.7071067690849304, "rewards/reward_investigation_quality/mean": 0.26250001788139343, "rewards/reward_investigation_quality/std": 0.1060660257935524, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 704.0, "completions/clipped_ratio": 0.0, "completions/max_length": 877.0, "completions/max_terminated_length": 877.0, "completions/mean_length": 704.0, "completions/mean_terminated_length": 704.0, "completions/min_length": 566.0, "completions/min_terminated_length": 566.0, "epoch": 0.648, "frac_reward_zero_std": 1.0, "grad_norm": 0.000489928585011512, "kl": 0.003910054118023254, "learning_rate": 1.6825531432186545e-05, "loss": 3.910053783329204e-05, "num_tokens": 945415.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 675.5, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 865.0, "completions/mean_length": 675.5, "completions/mean_terminated_length": 625.7142944335938, "completions/min_length": 468.0, "completions/min_terminated_length": 468.0, "epoch": 0.656, "frac_reward_zero_std": 1.0, "grad_norm": 0.0009050219086930156, "kl": 0.005692193168215454, "learning_rate": 1.6725072218945274e-05, "loss": 5.692193371942267e-05, "num_tokens": 957203.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 739.0, "completions/clipped_ratio": 0.0, "completions/max_length": 883.0, "completions/max_terminated_length": 883.0, "completions/mean_length": 739.0, "completions/mean_terminated_length": 739.0, "completions/min_length": 515.0, "completions/min_terminated_length": 515.0, "epoch": 0.664, "frac_reward_zero_std": 1.0, "grad_norm": 0.0009260174119845033, "kl": 0.0054808445274829865, "learning_rate": 1.662335832195308e-05, "loss": 5.480844993144274e-05, "num_tokens": 969663.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 722.875, "completions/clipped_ratio": 0.25, "completions/max_length": 1024.0, "completions/max_terminated_length": 837.0, "completions/mean_length": 722.875, "completions/mean_terminated_length": 622.5, "completions/min_length": 323.0, "completions/min_terminated_length": 323.0, "epoch": 0.672, "frac_reward_zero_std": 1.0, "grad_norm": 0.0007106768898665905, "kl": 0.004766069527249783, "learning_rate": 1.6520408717775507e-05, "loss": 4.7660694690421224e-05, "num_tokens": 981662.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 808.125, "completions/clipped_ratio": 0.25, "completions/max_length": 1024.0, "completions/max_terminated_length": 1002.0, "completions/mean_length": 808.125, "completions/mean_terminated_length": 736.1666870117188, "completions/min_length": 563.0, "completions/min_terminated_length": 563.0, "epoch": 0.68, "frac_reward_zero_std": 1.0, "grad_norm": 0.000751954154111445, "kl": 0.004348799702711403, "learning_rate": 1.6416242613521612e-05, "loss": 4.3487998482305557e-05, "num_tokens": 994351.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 629.5, "completions/clipped_ratio": 0.0, "completions/max_length": 745.0, "completions/max_terminated_length": 745.0, "completions/mean_length": 629.5, "completions/mean_terminated_length": 629.5, "completions/min_length": 485.0, "completions/min_terminated_length": 485.0, "epoch": 0.688, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005383202224038541, "kl": 0.00451987877022475, "learning_rate": 1.631087944326053e-05, "loss": 4.519878712017089e-05, "num_tokens": 1005755.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 578.0, "completions/clipped_ratio": 0.0, "completions/max_length": 849.0, "completions/max_terminated_length": 849.0, "completions/mean_length": 578.0, "completions/mean_terminated_length": 578.0, "completions/min_length": 268.0, "completions/min_terminated_length": 268.0, "epoch": 0.696, "frac_reward_zero_std": 1.0, "grad_norm": 0.0022287180181592703, "kl": 0.00829805782996118, "learning_rate": 1.6204338864395683e-05, "loss": 8.298057946376503e-05, "num_tokens": 1016587.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 723.875, "completions/clipped_ratio": 0.0, "completions/max_length": 893.0, "completions/max_terminated_length": 893.0, "completions/mean_length": 723.875, "completions/mean_terminated_length": 723.875, "completions/min_length": 545.0, "completions/min_terminated_length": 545.0, "epoch": 0.704, "frac_reward_zero_std": 1.0, "grad_norm": 0.0007930471329018474, "kl": 0.005153974459972233, "learning_rate": 1.609664075399735e-05, "loss": 5.153974416316487e-05, "num_tokens": 1028570.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 652.25, "completions/clipped_ratio": 0.0, "completions/max_length": 769.0, "completions/max_terminated_length": 769.0, "completions/mean_length": 652.25, "completions/mean_terminated_length": 652.25, "completions/min_length": 525.0, "completions/min_terminated_length": 525.0, "epoch": 0.712, "frac_reward_zero_std": 1.0, "grad_norm": 0.0006688805297017097, "kl": 0.004801850765943527, "learning_rate": 1.5987805205094225e-05, "loss": 4.801850445801392e-05, "num_tokens": 1040144.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 625.375, "completions/clipped_ratio": 0.0, "completions/max_length": 864.0, "completions/max_terminated_length": 864.0, "completions/mean_length": 625.375, "completions/mean_terminated_length": 625.375, "completions/min_length": 435.0, "completions/min_terminated_length": 435.0, "epoch": 0.72, "frac_reward_zero_std": 1.0, "grad_norm": 0.000412154447985813, "kl": 0.0038969786401139572, "learning_rate": 1.5877852522924733e-05, "loss": 3.896978887496516e-05, "num_tokens": 1051363.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 659.5, "completions/clipped_ratio": 0.0, "completions/max_length": 891.0, "completions/max_terminated_length": 891.0, "completions/mean_length": 659.5, "completions/mean_terminated_length": 659.5, "completions/min_length": 457.0, "completions/min_terminated_length": 457.0, "epoch": 0.728, "frac_reward_zero_std": 1.0, "grad_norm": 0.0015461925650015473, "kl": 0.006992757291300222, "learning_rate": 1.5766803221148676e-05, "loss": 6.992756971158087e-05, "num_tokens": 1063027.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 685.125, "completions/clipped_ratio": 0.0, "completions/max_length": 895.0, "completions/max_terminated_length": 895.0, "completions/mean_length": 685.125, "completions/mean_terminated_length": 685.125, "completions/min_length": 475.0, "completions/min_terminated_length": 475.0, "epoch": 0.736, "frac_reward_zero_std": 1.0, "grad_norm": 0.0011083967983722687, "kl": 0.005801332445116714, "learning_rate": 1.565467801802006e-05, "loss": 5.801332008559257e-05, "num_tokens": 1074688.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 658.0, "completions/clipped_ratio": 0.0, "completions/max_length": 817.0, "completions/max_terminated_length": 817.0, "completions/mean_length": 658.0, "completions/mean_terminated_length": 658.0, "completions/min_length": 547.0, "completions/min_terminated_length": 547.0, "epoch": 0.744, "frac_reward_zero_std": 1.0, "grad_norm": 0.0015490923542529345, "kl": 0.0067122922773705795, "learning_rate": 1.554149783252175e-05, "loss": 6.71229136059992e-05, "num_tokens": 1086328.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 694.125, "completions/clipped_ratio": 0.0, "completions/max_length": 954.0, "completions/max_terminated_length": 954.0, "completions/mean_length": 694.125, "completions/mean_terminated_length": 694.125, "completions/min_length": 471.0, "completions/min_terminated_length": 471.0, "epoch": 0.752, "frac_reward_zero_std": 0.5, "grad_norm": 0.10838653147220612, "kl": 0.0066940476826857775, "learning_rate": 1.542728378046262e-05, "loss": 6.692749593639746e-05, "num_tokens": 1098085.0, "reward": 0.9849998950958252, "reward_std": 0.7099999785423279, "rewards/reward_environment_execution/mean": -0.027499999850988388, "rewards/reward_environment_execution/std": 0.19091883301734924, "rewards/reward_format_compliance/mean": 0.75, "rewards/reward_format_compliance/std": 0.7071067690849304, "rewards/reward_investigation_quality/mean": 0.26250001788139343, "rewards/reward_investigation_quality/std": 0.1060660257935524, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 593.625, "completions/clipped_ratio": 0.0, "completions/max_length": 694.0, "completions/max_terminated_length": 694.0, "completions/mean_length": 593.625, "completions/mean_terminated_length": 593.625, "completions/min_length": 501.0, "completions/min_terminated_length": 501.0, "epoch": 0.76, "frac_reward_zero_std": 1.0, "grad_norm": 0.000631893053650856, "kl": 0.004648085654480383, "learning_rate": 1.5312057170538033e-05, "loss": 4.64808581455145e-05, "num_tokens": 1109034.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 681.5, "completions/clipped_ratio": 0.0, "completions/max_length": 780.0, "completions/max_terminated_length": 780.0, "completions/mean_length": 681.5, "completions/mean_terminated_length": 681.5, "completions/min_length": 471.0, "completions/min_terminated_length": 471.0, "epoch": 0.768, "frac_reward_zero_std": 1.0, "grad_norm": 0.0010983948595821857, "kl": 0.004860512548475526, "learning_rate": 1.5195839500354337e-05, "loss": 4.8605124902678654e-05, "num_tokens": 1120678.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 687.875, "completions/clipped_ratio": 0.0, "completions/max_length": 846.0, "completions/max_terminated_length": 846.0, "completions/mean_length": 687.875, "completions/mean_terminated_length": 687.875, "completions/min_length": 433.0, "completions/min_terminated_length": 433.0, "epoch": 0.776, "frac_reward_zero_std": 1.0, "grad_norm": 0.0006151673151180148, "kl": 0.004180053161690012, "learning_rate": 1.5078652452418063e-05, "loss": 4.180052565061487e-05, "num_tokens": 1132557.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 687.125, "completions/clipped_ratio": 0.0, "completions/max_length": 949.0, "completions/max_terminated_length": 949.0, "completions/mean_length": 687.125, "completions/mean_terminated_length": 687.125, "completions/min_length": 567.0, "completions/min_terminated_length": 567.0, "epoch": 0.784, "frac_reward_zero_std": 1.0, "grad_norm": 0.0006311318138614297, "kl": 0.004457214614376426, "learning_rate": 1.4960517890090683e-05, "loss": 4.4572145270649344e-05, "num_tokens": 1144610.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 684.5, "completions/clipped_ratio": 0.0, "completions/max_length": 881.0, "completions/max_terminated_length": 881.0, "completions/mean_length": 684.5, "completions/mean_terminated_length": 684.5, "completions/min_length": 520.0, "completions/min_terminated_length": 520.0, "epoch": 0.792, "frac_reward_zero_std": 1.0, "grad_norm": 0.0011139169801026583, "kl": 0.005598273302894086, "learning_rate": 1.4841457853509606e-05, "loss": 5.598273128271103e-05, "num_tokens": 1156474.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "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, "completion_length": 638.5, "completions/clipped_ratio": 0.0, "completions/max_length": 813.0, "completions/max_terminated_length": 813.0, "completions/mean_length": 638.5, "completions/mean_terminated_length": 638.5, "completions/min_length": 521.0, "completions/min_terminated_length": 521.0, "epoch": 0.8, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005143387825228274, "kl": 0.0034461540053598583, "learning_rate": 1.4721494555476189e-05, "loss": 3.446154005359858e-05, "num_tokens": 1167790.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 100 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 592.375, "completions/clipped_ratio": 0.0, "completions/max_length": 859.0, "completions/max_terminated_length": 859.0, "completions/mean_length": 592.375, "completions/mean_terminated_length": 592.375, "completions/min_length": 439.0, "completions/min_terminated_length": 439.0, "epoch": 0.808, "frac_reward_zero_std": 1.0, "grad_norm": 0.0007209504256024957, "kl": 0.004973333969246596, "learning_rate": 1.4600650377311523e-05, "loss": 4.973334216629155e-05, "num_tokens": 1178745.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 101 }, { "clip_ratio/high_max": 0.0, "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": 700.125, "completions/clipped_ratio": 0.0, "completions/max_length": 861.0, "completions/max_terminated_length": 861.0, "completions/mean_length": 700.125, "completions/mean_terminated_length": 700.125, "completions/min_length": 523.0, "completions/min_terminated_length": 523.0, "epoch": 0.816, "frac_reward_zero_std": 1.0, "grad_norm": 0.0006949947564862669, "kl": 0.00382522409199737, "learning_rate": 1.4478947864680821e-05, "loss": 3.825223393505439e-05, "num_tokens": 1190726.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 102 }, { "clip_ratio/high_max": 0.0, "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": 704.625, "completions/clipped_ratio": 0.0, "completions/max_length": 958.0, "completions/max_terminated_length": 958.0, "completions/mean_length": 704.625, "completions/mean_terminated_length": 704.625, "completions/min_length": 549.0, "completions/min_terminated_length": 549.0, "epoch": 0.824, "frac_reward_zero_std": 1.0, "grad_norm": 0.0015731051098555326, "kl": 0.005999519751640037, "learning_rate": 1.4356409723387092e-05, "loss": 5.999519635224715e-05, "num_tokens": 1202899.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 103 }, { "clip_ratio/high_max": 0.0, "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": 603.875, "completions/clipped_ratio": 0.0, "completions/max_length": 720.0, "completions/max_terminated_length": 720.0, "completions/mean_length": 603.875, "completions/mean_terminated_length": 603.875, "completions/min_length": 430.0, "completions/min_terminated_length": 430.0, "epoch": 0.832, "frac_reward_zero_std": 0.5, "grad_norm": 0.07669439911842346, "kl": 0.004357149155111983, "learning_rate": 1.4233058815134978e-05, "loss": 3.7732555938418955e-05, "num_tokens": 1214118.0, "reward": 1.338749885559082, "reward_std": 0.002499997615814209, "rewards/reward_environment_execution/mean": 0.038750000298023224, "rewards/reward_environment_execution/std": 0.0035355337895452976, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 104 }, { "clip_ratio/high_max": 0.0, "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": 663.25, "completions/clipped_ratio": 0.0, "completions/max_length": 764.0, "completions/max_terminated_length": 764.0, "completions/mean_length": 663.25, "completions/mean_terminated_length": 663.25, "completions/min_length": 561.0, "completions/min_terminated_length": 561.0, "epoch": 0.84, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005208631628192961, "kl": 0.003520077880239114, "learning_rate": 1.4108918153265485e-05, "loss": 3.520077734719962e-05, "num_tokens": 1225792.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 105 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 628.0, "completions/clipped_ratio": 0.0, "completions/max_length": 799.0, "completions/max_terminated_length": 799.0, "completions/mean_length": 628.0, "completions/mean_terminated_length": 628.0, "completions/min_length": 519.0, "completions/min_terminated_length": 519.0, "epoch": 0.848, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005702024209313095, "kl": 0.003354142216267064, "learning_rate": 1.3984010898462417e-05, "loss": 3.354142245370895e-05, "num_tokens": 1237192.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 106 }, { "clip_ratio/high_max": 0.0, "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": 776.125, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 1007.0, "completions/mean_length": 776.125, "completions/mean_terminated_length": 740.7142944335938, "completions/min_length": 509.0, "completions/min_terminated_length": 509.0, "epoch": 0.856, "frac_reward_zero_std": 1.0, "grad_norm": 0.000770260114222765, "kl": 0.0030931195215089247, "learning_rate": 1.3858360354431355e-05, "loss": 3.0931194487493485e-05, "num_tokens": 1249597.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 107 }, { "clip_ratio/high_max": 0.0, "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": 620.125, "completions/clipped_ratio": 0.0, "completions/max_length": 928.0, "completions/max_terminated_length": 928.0, "completions/mean_length": 620.125, "completions/mean_terminated_length": 620.125, "completions/min_length": 421.0, "completions/min_terminated_length": 421.0, "epoch": 0.864, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005046819569543004, "kl": 0.003935346117941663, "learning_rate": 1.3731989963551916e-05, "loss": 3.935345739591867e-05, "num_tokens": 1260770.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 108 }, { "clip_ratio/high_max": 0.0, "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": 571.75, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 665.0, "completions/mean_length": 571.75, "completions/mean_terminated_length": 507.14288330078125, "completions/min_length": 284.0, "completions/min_terminated_length": 284.0, "epoch": 0.872, "frac_reward_zero_std": 0.5, "grad_norm": 0.136940136551857, "kl": 0.004765524790855125, "learning_rate": 1.3604923302504146e-05, "loss": 4.182383418083191e-05, "num_tokens": 1271728.0, "reward": 1.338749885559082, "reward_std": 0.002499997615814209, "rewards/reward_environment_execution/mean": 0.038750000298023224, "rewards/reward_environment_execution/std": 0.0035355337895452976, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 109 }, { "clip_ratio/high_max": 0.0, "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": 689.5, "completions/clipped_ratio": 0.0, "completions/max_length": 853.0, "completions/max_terminated_length": 853.0, "completions/mean_length": 689.5, "completions/mean_terminated_length": 689.5, "completions/min_length": 429.0, "completions/min_terminated_length": 429.0, "epoch": 0.88, "frac_reward_zero_std": 0.5, "grad_norm": 0.10756723582744598, "kl": 0.003721862827660516, "learning_rate": 1.3477184077869892e-05, "loss": 3.7223100662231445e-05, "num_tokens": 1283808.0, "reward": 0.9837499260902405, "reward_std": 0.7091705799102783, "rewards/reward_environment_execution/mean": -0.028749998658895493, "rewards/reward_environment_execution/std": 0.1904458999633789, "rewards/reward_format_compliance/mean": 0.75, "rewards/reward_format_compliance/std": 0.7071067690849304, "rewards/reward_investigation_quality/mean": 0.26250001788139343, "rewards/reward_investigation_quality/std": 0.1060660257935524, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 110 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 718.875, "completions/clipped_ratio": 0.0, "completions/max_length": 814.0, "completions/max_terminated_length": 814.0, "completions/mean_length": 718.875, "completions/mean_terminated_length": 718.875, "completions/min_length": 516.0, "completions/min_terminated_length": 516.0, "epoch": 0.888, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005541781429201365, "kl": 0.003651737206382677, "learning_rate": 1.3348796121709862e-05, "loss": 3.6517369153443724e-05, "num_tokens": 1295927.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 111 }, { "clip_ratio/high_max": 0.0, "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": 616.875, "completions/clipped_ratio": 0.0, "completions/max_length": 903.0, "completions/max_terminated_length": 903.0, "completions/mean_length": 616.875, "completions/mean_terminated_length": 616.875, "completions/min_length": 341.0, "completions/min_terminated_length": 341.0, "epoch": 0.896, "frac_reward_zero_std": 1.0, "grad_norm": 0.0007263852749019861, "kl": 0.004626723180990666, "learning_rate": 1.3219783387117384e-05, "loss": 4.6267228754004464e-05, "num_tokens": 1307066.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 112 }, { "clip_ratio/high_max": 0.0, "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": 748.75, "completions/clipped_ratio": 0.25, "completions/max_length": 1024.0, "completions/max_terminated_length": 913.0, "completions/mean_length": 748.75, "completions/mean_terminated_length": 657.0, "completions/min_length": 497.0, "completions/min_terminated_length": 497.0, "epoch": 0.904, "frac_reward_zero_std": 1.0, "grad_norm": 0.0007187369046732783, "kl": 0.003865452017635107, "learning_rate": 1.3090169943749475e-05, "loss": 3.865452163154259e-05, "num_tokens": 1319456.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 113 }, { "clip_ratio/high_max": 0.0, "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": 712.25, "completions/clipped_ratio": 0.0, "completions/max_length": 910.0, "completions/max_terminated_length": 910.0, "completions/mean_length": 712.25, "completions/mean_terminated_length": 712.25, "completions/min_length": 543.0, "completions/min_terminated_length": 543.0, "epoch": 0.912, "frac_reward_zero_std": 1.0, "grad_norm": 0.0006156337331049144, "kl": 0.0037589431449305266, "learning_rate": 1.2959979973336236e-05, "loss": 3.758943057619035e-05, "num_tokens": 1331550.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 114 }, { "clip_ratio/high_max": 0.0, "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": 580.0, "completions/clipped_ratio": 0.0, "completions/max_length": 755.0, "completions/max_terminated_length": 755.0, "completions/mean_length": 580.0, "completions/mean_terminated_length": 580.0, "completions/min_length": 403.0, "completions/min_terminated_length": 403.0, "epoch": 0.92, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005986274918541312, "kl": 0.004748013074276969, "learning_rate": 1.28292377651693e-05, "loss": 4.7480127250310034e-05, "num_tokens": 1342406.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 115 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 652.25, "completions/clipped_ratio": 0.0, "completions/max_length": 832.0, "completions/max_terminated_length": 832.0, "completions/mean_length": 652.25, "completions/mean_terminated_length": 652.25, "completions/min_length": 432.0, "completions/min_terminated_length": 432.0, "epoch": 0.928, "frac_reward_zero_std": 1.0, "grad_norm": 0.0007125227712094784, "kl": 0.00425728945992887, "learning_rate": 1.2697967711570243e-05, "loss": 4.257289401721209e-05, "num_tokens": 1354168.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 116 }, { "clip_ratio/high_max": 0.0, "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": 829.375, "completions/clipped_ratio": 0.0, "completions/max_length": 960.0, "completions/max_terminated_length": 960.0, "completions/mean_length": 829.375, "completions/mean_terminated_length": 829.375, "completions/min_length": 683.0, "completions/min_terminated_length": 683.0, "epoch": 0.936, "frac_reward_zero_std": 1.0, "grad_norm": 0.0008483250858262181, "kl": 0.003871161781717092, "learning_rate": 1.2566194303339738e-05, "loss": 3.871161607094109e-05, "num_tokens": 1367007.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 117 }, { "clip_ratio/high_max": 0.0, "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": 695.375, "completions/clipped_ratio": 0.0, "completions/max_length": 795.0, "completions/max_terminated_length": 795.0, "completions/mean_length": 695.375, "completions/mean_terminated_length": 695.375, "completions/min_length": 607.0, "completions/min_terminated_length": 607.0, "epoch": 0.944, "frac_reward_zero_std": 1.0, "grad_norm": 0.0004924535169266164, "kl": 0.003306248632725328, "learning_rate": 1.2433942125188359e-05, "loss": 3.306248618173413e-05, "num_tokens": 1379130.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 118 }, { "clip_ratio/high_max": 0.0, "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": 716.875, "completions/clipped_ratio": 0.0, "completions/max_length": 908.0, "completions/max_terminated_length": 908.0, "completions/mean_length": 716.875, "completions/mean_terminated_length": 716.875, "completions/min_length": 561.0, "completions/min_terminated_length": 561.0, "epoch": 0.952, "frac_reward_zero_std": 1.0, "grad_norm": 0.0006755761569365859, "kl": 0.0034013454569503665, "learning_rate": 1.2301235851149867e-05, "loss": 3.4013453841907904e-05, "num_tokens": 1391421.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 119 }, { "clip_ratio/high_max": 0.0, "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": 680.25, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 827.0, "completions/mean_length": 680.25, "completions/mean_terminated_length": 631.1428833007812, "completions/min_length": 483.0, "completions/min_terminated_length": 483.0, "epoch": 0.96, "frac_reward_zero_std": 0.5, "grad_norm": 0.09553336352109909, "kl": 0.004561332025332376, "learning_rate": 1.216810023997781e-05, "loss": 4.557903957902454e-05, "num_tokens": 1403243.0, "reward": 0.9849998950958252, "reward_std": 0.7099999785423279, "rewards/reward_environment_execution/mean": -0.027499999850988388, "rewards/reward_environment_execution/std": 0.19091883301734924, "rewards/reward_format_compliance/mean": 0.75, "rewards/reward_format_compliance/std": 0.7071067690849304, "rewards/reward_investigation_quality/mean": 0.26250001788139343, "rewards/reward_investigation_quality/std": 0.1060660257935524, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 120 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 685.125, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 746.0, "completions/mean_length": 685.125, "completions/mean_terminated_length": 636.7142944335938, "completions/min_length": 539.0, "completions/min_terminated_length": 539.0, "epoch": 0.968, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005767505499534309, "kl": 0.004357667494332418, "learning_rate": 1.2034560130526341e-05, "loss": 4.357667421572842e-05, "num_tokens": 1414948.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 121 }, { "clip_ratio/high_max": 0.0, "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": 547.75, "completions/clipped_ratio": 0.0, "completions/max_length": 739.0, "completions/max_terminated_length": 739.0, "completions/mean_length": 547.75, "completions/mean_terminated_length": 547.75, "completions/min_length": 466.0, "completions/min_terminated_length": 466.0, "epoch": 0.976, "frac_reward_zero_std": 1.0, "grad_norm": 0.001038038870319724, "kl": 0.004890444863121957, "learning_rate": 1.1900640437116074e-05, "loss": 4.890445416094735e-05, "num_tokens": 1425714.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 122 }, { "clip_ratio/high_max": 0.0, "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": 642.125, "completions/clipped_ratio": 0.0, "completions/max_length": 859.0, "completions/max_terminated_length": 859.0, "completions/mean_length": 642.125, "completions/mean_terminated_length": 642.125, "completions/min_length": 290.0, "completions/min_terminated_length": 290.0, "epoch": 0.984, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005056797526776791, "kl": 0.0035681982844835147, "learning_rate": 1.1766366144885877e-05, "loss": 3.5681980079971254e-05, "num_tokens": 1437231.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 123 }, { "clip_ratio/high_max": 0.0, "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": 716.25, "completions/clipped_ratio": 0.0, "completions/max_length": 1023.0, "completions/max_terminated_length": 1023.0, "completions/mean_length": 716.25, "completions/mean_terminated_length": 716.25, "completions/min_length": 511.0, "completions/min_terminated_length": 511.0, "epoch": 0.992, "frac_reward_zero_std": 1.0, "grad_norm": 0.0006406443426385522, "kl": 0.004348908376414329, "learning_rate": 1.1631762305131424e-05, "loss": 4.348907896201126e-05, "num_tokens": 1449337.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 124 }, { "clip_ratio/high_max": 0.0, "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": 748.625, "completions/clipped_ratio": 0.0, "completions/max_length": 1010.0, "completions/max_terminated_length": 1010.0, "completions/mean_length": 748.625, "completions/mean_terminated_length": 748.625, "completions/min_length": 524.0, "completions/min_terminated_length": 524.0, "epoch": 1.0, "frac_reward_zero_std": 1.0, "grad_norm": 0.0004692629154305905, "kl": 0.0033065118768718094, "learning_rate": 1.1496854030631443e-05, "loss": 3.306511644041166e-05, "num_tokens": 1461526.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 125 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 661.875, "completions/clipped_ratio": 0.0, "completions/max_length": 776.0, "completions/max_terminated_length": 776.0, "completions/mean_length": 661.875, "completions/mean_terminated_length": 661.875, "completions/min_length": 545.0, "completions/min_terminated_length": 545.0, "epoch": 1.008, "frac_reward_zero_std": 1.0, "grad_norm": 0.0004876362218055874, "kl": 0.003236054108128883, "learning_rate": 1.1361666490962468e-05, "loss": 3.236054180888459e-05, "num_tokens": 1473197.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 126 }, { "clip_ratio/high_max": 0.0, "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": 617.125, "completions/clipped_ratio": 0.0, "completions/max_length": 1005.0, "completions/max_terminated_length": 1005.0, "completions/mean_length": 617.125, "completions/mean_terminated_length": 617.125, "completions/min_length": 452.0, "completions/min_terminated_length": 452.0, "epoch": 1.016, "frac_reward_zero_std": 1.0, "grad_norm": 0.0007722392329014838, "kl": 0.005528639652766287, "learning_rate": 1.1226224907802986e-05, "loss": 5.5286396673182026e-05, "num_tokens": 1484346.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 127 }, { "clip_ratio/high_max": 0.0, "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": 682.375, "completions/clipped_ratio": 0.0, "completions/max_length": 839.0, "completions/max_terminated_length": 839.0, "completions/mean_length": 682.375, "completions/mean_terminated_length": 682.375, "completions/min_length": 538.0, "completions/min_terminated_length": 538.0, "epoch": 1.024, "frac_reward_zero_std": 1.0, "grad_norm": 0.00045290120760910213, "kl": 0.003257805190514773, "learning_rate": 1.1090554550227899e-05, "loss": 3.257804928580299e-05, "num_tokens": 1496333.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 128 }, { "clip_ratio/high_max": 0.0, "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": 631.375, "completions/clipped_ratio": 0.0, "completions/max_length": 789.0, "completions/max_terminated_length": 789.0, "completions/mean_length": 631.375, "completions/mean_terminated_length": 631.375, "completions/min_length": 466.0, "completions/min_terminated_length": 466.0, "epoch": 1.032, "frac_reward_zero_std": 1.0, "grad_norm": 0.0008484916761517525, "kl": 0.004940687998896465, "learning_rate": 1.0954680729994103e-05, "loss": 4.940688086207956e-05, "num_tokens": 1507584.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 129 }, { "clip_ratio/high_max": 0.0, "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": 698.375, "completions/clipped_ratio": 0.0, "completions/max_length": 847.0, "completions/max_terminated_length": 847.0, "completions/mean_length": 698.375, "completions/mean_terminated_length": 698.375, "completions/min_length": 526.0, "completions/min_terminated_length": 526.0, "epoch": 1.04, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005641288007609546, "kl": 0.003881108801579103, "learning_rate": 1.0818628796818134e-05, "loss": 3.881108568748459e-05, "num_tokens": 1519535.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 130 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 635.625, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 955.0, "completions/mean_length": 635.625, "completions/mean_terminated_length": 580.1428833007812, "completions/min_length": 413.0, "completions/min_terminated_length": 413.0, "epoch": 1.048, "frac_reward_zero_std": 1.0, "grad_norm": 0.0007799405138939619, "kl": 0.004543178831227124, "learning_rate": 1.0682424133646712e-05, "loss": 4.543178147287108e-05, "num_tokens": 1531000.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 131 }, { "clip_ratio/high_max": 0.0, "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": 685.875, "completions/clipped_ratio": 0.0, "completions/max_length": 973.0, "completions/max_terminated_length": 973.0, "completions/mean_length": 685.875, "completions/mean_terminated_length": 685.875, "completions/min_length": 455.0, "completions/min_terminated_length": 455.0, "epoch": 1.056, "frac_reward_zero_std": 1.0, "grad_norm": 0.0008580928551964462, "kl": 0.004743750352645293, "learning_rate": 1.05460921519211e-05, "loss": 4.743750469060615e-05, "num_tokens": 1542695.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 132 }, { "clip_ratio/high_max": 0.0, "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": 886.0, "completions/clipped_ratio": 0.5, "completions/max_length": 1024.0, "completions/max_terminated_length": 913.0, "completions/mean_length": 886.0, "completions/mean_terminated_length": 748.0, "completions/min_length": 578.0, "completions/min_terminated_length": 578.0, "epoch": 1.064, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005378791829571128, "kl": 0.0036341611994430423, "learning_rate": 1.0409658286836144e-05, "loss": 3.634161112131551e-05, "num_tokens": 1555995.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 133 }, { "clip_ratio/high_max": 0.0, "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": 547.0, "completions/clipped_ratio": 0.0, "completions/max_length": 651.0, "completions/max_terminated_length": 651.0, "completions/mean_length": 547.0, "completions/mean_terminated_length": 547.0, "completions/min_length": 322.0, "completions/min_terminated_length": 322.0, "epoch": 1.072, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005891018081456423, "kl": 0.004448721068911254, "learning_rate": 1.0273147992594861e-05, "loss": 4.4487205741461366e-05, "num_tokens": 1566603.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 134 }, { "clip_ratio/high_max": 0.0, "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": 724.75, "completions/clipped_ratio": 0.0, "completions/max_length": 896.0, "completions/max_terminated_length": 896.0, "completions/mean_length": 724.75, "completions/mean_terminated_length": 724.75, "completions/min_length": 611.0, "completions/min_terminated_length": 611.0, "epoch": 1.08, "frac_reward_zero_std": 0.5, "grad_norm": 0.06136518716812134, "kl": 0.003305838443338871, "learning_rate": 1.013658673765951e-05, "loss": 3.3054500818252563e-05, "num_tokens": 1578617.0, "reward": 0.9849998950958252, "reward_std": 0.7099999785423279, "rewards/reward_environment_execution/mean": -0.027499999850988388, "rewards/reward_environment_execution/std": 0.19091883301734924, "rewards/reward_format_compliance/mean": 0.75, "rewards/reward_format_compliance/std": 0.7071067690849304, "rewards/reward_investigation_quality/mean": 0.26250001788139343, "rewards/reward_investigation_quality/std": 0.1060660257935524, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 135 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 707.625, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 836.0, "completions/mean_length": 707.625, "completions/mean_terminated_length": 662.4285888671875, "completions/min_length": 404.0, "completions/min_terminated_length": 404.0, "epoch": 1.088, "frac_reward_zero_std": 1.0, "grad_norm": 0.0006287655560299754, "kl": 0.003811593269347213, "learning_rate": 1e-05, "loss": 3.8115929783089086e-05, "num_tokens": 1590674.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 136 }, { "clip_ratio/high_max": 0.0, "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": 656.75, "completions/clipped_ratio": 0.0, "completions/max_length": 771.0, "completions/max_terminated_length": 771.0, "completions/mean_length": 656.75, "completions/mean_terminated_length": 656.75, "completions/min_length": 553.0, "completions/min_terminated_length": 553.0, "epoch": 1.096, "frac_reward_zero_std": 1.0, "grad_norm": 0.0006832216749899089, "kl": 0.004610352334566414, "learning_rate": 9.863413262340491e-06, "loss": 4.6103523345664144e-05, "num_tokens": 1602468.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 137 }, { "clip_ratio/high_max": 0.0, "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": 710.5, "completions/clipped_ratio": 0.0, "completions/max_length": 870.0, "completions/max_terminated_length": 870.0, "completions/mean_length": 710.5, "completions/mean_terminated_length": 710.5, "completions/min_length": 512.0, "completions/min_terminated_length": 512.0, "epoch": 1.104, "frac_reward_zero_std": 1.0, "grad_norm": 0.0004808783996850252, "kl": 0.0031752858485560864, "learning_rate": 9.726852007405144e-06, "loss": 3.17528574669268e-05, "num_tokens": 1614368.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 138 }, { "clip_ratio/high_max": 0.0, "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": 701.625, "completions/clipped_ratio": 0.0, "completions/max_length": 970.0, "completions/max_terminated_length": 970.0, "completions/mean_length": 701.625, "completions/mean_terminated_length": 701.625, "completions/min_length": 535.0, "completions/min_terminated_length": 535.0, "epoch": 1.112, "frac_reward_zero_std": 1.0, "grad_norm": 0.0008137078839354217, "kl": 0.004681420483393595, "learning_rate": 9.590341713163858e-06, "loss": 4.681420614360832e-05, "num_tokens": 1626365.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 139 }, { "clip_ratio/high_max": 0.0, "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": 673.875, "completions/clipped_ratio": 0.0, "completions/max_length": 898.0, "completions/max_terminated_length": 898.0, "completions/mean_length": 673.875, "completions/mean_terminated_length": 673.875, "completions/min_length": 497.0, "completions/min_terminated_length": 497.0, "epoch": 1.12, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005561233847402036, "kl": 0.004355500102974474, "learning_rate": 9.453907848078901e-06, "loss": 4.355500277597457e-05, "num_tokens": 1638132.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 140 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 656.25, "completions/clipped_ratio": 0.0, "completions/max_length": 813.0, "completions/max_terminated_length": 813.0, "completions/mean_length": 656.25, "completions/mean_terminated_length": 656.25, "completions/min_length": 490.0, "completions/min_terminated_length": 490.0, "epoch": 1.1280000000000001, "frac_reward_zero_std": 1.0, "grad_norm": 0.0006565627991221845, "kl": 0.005087505269329995, "learning_rate": 9.317575866353293e-06, "loss": 5.087504905532114e-05, "num_tokens": 1649766.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 141 }, { "clip_ratio/high_max": 0.0, "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": 734.625, "completions/clipped_ratio": 0.125, "completions/max_length": 1024.0, "completions/max_terminated_length": 950.0, "completions/mean_length": 734.625, "completions/mean_terminated_length": 693.2857666015625, "completions/min_length": 517.0, "completions/min_terminated_length": 517.0, "epoch": 1.1360000000000001, "frac_reward_zero_std": 1.0, "grad_norm": 0.0016482328064739704, "kl": 0.006351819261908531, "learning_rate": 9.181371203181873e-06, "loss": 6.351818592520431e-05, "num_tokens": 1662015.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 142 }, { "clip_ratio/high_max": 0.0, "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": 748.75, "completions/clipped_ratio": 0.0, "completions/max_length": 939.0, "completions/max_terminated_length": 939.0, "completions/mean_length": 748.75, "completions/mean_terminated_length": 748.75, "completions/min_length": 415.0, "completions/min_terminated_length": 415.0, "epoch": 1.144, "frac_reward_zero_std": 0.5, "grad_norm": 0.0668199434876442, "kl": 0.0046288135927170515, "learning_rate": 9.0453192700059e-06, "loss": 4.044175148010254e-05, "num_tokens": 1674397.0, "reward": 1.338749885559082, "reward_std": 0.002499997615814209, "rewards/reward_environment_execution/mean": 0.038750000298023224, "rewards/reward_environment_execution/std": 0.0035355337895452976, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 143 }, { "clip_ratio/high_max": 0.0, "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": 626.5, "completions/clipped_ratio": 0.0, "completions/max_length": 723.0, "completions/max_terminated_length": 723.0, "completions/mean_length": 626.5, "completions/mean_terminated_length": 626.5, "completions/min_length": 486.0, "completions/min_terminated_length": 486.0, "epoch": 1.152, "frac_reward_zero_std": 1.0, "grad_norm": 0.0008783203084021807, "kl": 0.005382616596762091, "learning_rate": 8.909445449772103e-06, "loss": 5.382616654969752e-05, "num_tokens": 1685789.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 144 }, { "clip_ratio/high_max": 0.0, "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": 683.0, "completions/clipped_ratio": 0.0, "completions/max_length": 803.0, "completions/max_terminated_length": 803.0, "completions/mean_length": 683.0, "completions/mean_terminated_length": 683.0, "completions/min_length": 615.0, "completions/min_terminated_length": 615.0, "epoch": 1.16, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005522130522876978, "kl": 0.003862932018819265, "learning_rate": 8.773775092197018e-06, "loss": 3.862931771436706e-05, "num_tokens": 1697773.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 145 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completion_length": 642.875, "completions/clipped_ratio": 0.0, "completions/max_length": 837.0, "completions/max_terminated_length": 837.0, "completions/mean_length": 642.875, "completions/mean_terminated_length": 642.875, "completions/min_length": 483.0, "completions/min_terminated_length": 483.0, "epoch": 1.168, "frac_reward_zero_std": 1.0, "grad_norm": 0.0010609590681269765, "kl": 0.005505072098458186, "learning_rate": 8.638333509037537e-06, "loss": 5.505072476807982e-05, "num_tokens": 1709116.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 146 }, { "clip_ratio/high_max": 0.0, "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": 653.125, "completions/clipped_ratio": 0.0, "completions/max_length": 758.0, "completions/max_terminated_length": 758.0, "completions/mean_length": 653.125, "completions/mean_terminated_length": 653.125, "completions/min_length": 519.0, "completions/min_terminated_length": 519.0, "epoch": 1.176, "frac_reward_zero_std": 1.0, "grad_norm": 0.000876356556545943, "kl": 0.0053305430337786674, "learning_rate": 8.503145969368562e-06, "loss": 5.330542626325041e-05, "num_tokens": 1720545.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 147 }, { "clip_ratio/high_max": 0.0, "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": 698.375, "completions/clipped_ratio": 0.0, "completions/max_length": 1009.0, "completions/max_terminated_length": 1009.0, "completions/mean_length": 698.375, "completions/mean_terminated_length": 698.375, "completions/min_length": 567.0, "completions/min_terminated_length": 567.0, "epoch": 1.184, "frac_reward_zero_std": 1.0, "grad_norm": 0.0005659233429469168, "kl": 0.004354778298875317, "learning_rate": 8.36823769486858e-06, "loss": 4.3547781388042495e-05, "num_tokens": 1732696.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 148 }, { "clip_ratio/high_max": 0.0, "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": 623.875, "completions/clipped_ratio": 0.0, "completions/max_length": 731.0, "completions/max_terminated_length": 731.0, "completions/mean_length": 623.875, "completions/mean_terminated_length": 623.875, "completions/min_length": 541.0, "completions/min_terminated_length": 541.0, "epoch": 1.192, "frac_reward_zero_std": 1.0, "grad_norm": 0.0007407800876535475, "kl": 0.004880121661699377, "learning_rate": 8.233633855114127e-06, "loss": 4.880120832240209e-05, "num_tokens": 1744071.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 149 }, { "clip_ratio/high_max": 0.0, "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": 710.25, "completions/clipped_ratio": 0.0, "completions/max_length": 849.0, "completions/max_terminated_length": 849.0, "completions/mean_length": 710.25, "completions/mean_terminated_length": 710.25, "completions/min_length": 588.0, "completions/min_terminated_length": 588.0, "epoch": 1.2, "frac_reward_zero_std": 1.0, "grad_norm": 0.0010268789483234286, "kl": 0.0045860171085223556, "learning_rate": 8.099359562883931e-06, "loss": 4.586016802932136e-05, "num_tokens": 1756129.0, "reward": 1.3399999141693115, "reward_std": 0.0, "rewards/reward_environment_execution/mean": 0.03999999910593033, "rewards/reward_environment_execution/std": 0.0, "rewards/reward_format_compliance/mean": 1.0, "rewards/reward_format_compliance/std": 0.0, "rewards/reward_investigation_quality/mean": 0.30000001192092896, "rewards/reward_investigation_quality/std": 0.0, "rewards/reward_os_mechanics/mean": 0.0, "rewards/reward_os_mechanics/std": 0.0, "step": 150 } ], "logging_steps": 1, "max_steps": 250, "num_input_tokens_seen": 1756129, "num_train_epochs": 2, "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 }