| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.4, |
| "eval_steps": 500, |
| "global_step": 50, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "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 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 250, |
| "num_input_tokens_seen": 578764, |
| "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 |
| } |
|
|
