| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.25, |
| "eval_steps": 500, |
| "global_step": 25, |
| "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": 659.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 821.0, |
| "completions/max_terminated_length": 821.0, |
| "completions/mean_length": 659.0, |
| "completions/mean_terminated_length": 659.0, |
| "completions/min_length": 516.0, |
| "completions/min_terminated_length": 516.0, |
| "epoch": 0.01, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.13492469489574432, |
| "kl": 6.197717993927654e-06, |
| "learning_rate": 0.0, |
| "loss": 2.8312206268310547e-07, |
| "num_tokens": 5736.0, |
| "reward": 0.1550000160932541, |
| "reward_std": 0.7699999809265137, |
| "rewards/reward_environment_execution/mean": -0.0949999988079071, |
| "rewards/reward_environment_execution/std": 0.26999998092651367, |
| "rewards/reward_format_compliance/mean": 0.02499999850988388, |
| "rewards/reward_format_compliance/std": 0.3499999940395355, |
| "rewards/reward_investigation_quality/mean": 0.22500000894069672, |
| "rewards/reward_investigation_quality/std": 0.15000002086162567, |
| "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": 543.75, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 693.0, |
| "completions/max_terminated_length": 693.0, |
| "completions/mean_length": 543.75, |
| "completions/mean_terminated_length": 543.75, |
| "completions/min_length": 426.0, |
| "completions/min_terminated_length": 426.0, |
| "epoch": 0.02, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.1295754313468933, |
| "kl": 7.463787596861948e-06, |
| "learning_rate": 5.000000000000001e-07, |
| "loss": 3.725290298461914e-07, |
| "num_tokens": 11019.0, |
| "reward": 0.15250001847743988, |
| "reward_std": 0.7683477997779846, |
| "rewards/reward_environment_execution/mean": -0.09749999642372131, |
| "rewards/reward_environment_execution/std": 0.2683747410774231, |
| "rewards/reward_format_compliance/mean": 0.02499999850988388, |
| "rewards/reward_format_compliance/std": 0.3499999940395355, |
| "rewards/reward_investigation_quality/mean": 0.22500000894069672, |
| "rewards/reward_investigation_quality/std": 0.15000002086162567, |
| "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": 572.75, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 677.0, |
| "completions/max_terminated_length": 677.0, |
| "completions/mean_length": 572.75, |
| "completions/mean_terminated_length": 572.75, |
| "completions/min_length": 490.0, |
| "completions/min_terminated_length": 490.0, |
| "epoch": 0.03, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.1402619630098343, |
| "kl": 5.991519174131099e-06, |
| "learning_rate": 1.0000000000000002e-06, |
| "loss": 2.2351741790771484e-07, |
| "num_tokens": 16418.0, |
| "reward": 0.5375000238418579, |
| "reward_std": 0.004999995231628418, |
| "rewards/reward_environment_execution/mean": 0.03750000149011612, |
| "rewards/reward_environment_execution/std": 0.004999999888241291, |
| "rewards/reward_format_compliance/mean": 0.20000000298023224, |
| "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": 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": 705.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 783.0, |
| "completions/max_terminated_length": 783.0, |
| "completions/mean_length": 705.0, |
| "completions/mean_terminated_length": 705.0, |
| "completions/min_length": 634.0, |
| "completions/min_terminated_length": 634.0, |
| "epoch": 0.04, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.13879841566085815, |
| "kl": 6.360480483635911e-06, |
| "learning_rate": 1.5e-06, |
| "loss": 2.980232238769531e-07, |
| "num_tokens": 22514.0, |
| "reward": 0.24392502009868622, |
| "reward_std": 0.8475051522254944, |
| "rewards/reward_environment_execution/mean": 0.04392500966787338, |
| "rewards/reward_environment_execution/std": 0.4515362083911896, |
| "rewards/reward_format_compliance/mean": 0.02499999850988388, |
| "rewards/reward_format_compliance/std": 0.3499999940395355, |
| "rewards/reward_investigation_quality/mean": 0.17500001192092896, |
| "rewards/reward_investigation_quality/std": 0.15000002086162567, |
| "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": 677.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 785.0, |
| "completions/max_terminated_length": 785.0, |
| "completions/mean_length": 677.0, |
| "completions/mean_terminated_length": 677.0, |
| "completions/min_length": 623.0, |
| "completions/min_terminated_length": 623.0, |
| "epoch": 0.05, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.10656019300222397, |
| "kl": 4.833805178350303e-06, |
| "learning_rate": 2.0000000000000003e-06, |
| "loss": 2.2351741790771484e-07, |
| "num_tokens": 28326.0, |
| "reward": 0.15250001847743988, |
| "reward_std": 0.7683477997779846, |
| "rewards/reward_environment_execution/mean": -0.09750000387430191, |
| "rewards/reward_environment_execution/std": 0.2683747410774231, |
| "rewards/reward_format_compliance/mean": 0.02499999850988388, |
| "rewards/reward_format_compliance/std": 0.3499999940395355, |
| "rewards/reward_investigation_quality/mean": 0.22500000894069672, |
| "rewards/reward_investigation_quality/std": 0.15000002086162567, |
| "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": 701.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 913.0, |
| "completions/max_terminated_length": 913.0, |
| "completions/mean_length": 701.0, |
| "completions/mean_terminated_length": 701.0, |
| "completions/min_length": 553.0, |
| "completions/min_terminated_length": 553.0, |
| "epoch": 0.06, |
| "frac_reward_zero_std": 1.0, |
| "grad_norm": 6.133589340606704e-05, |
| "kl": 6.346002919599414e-06, |
| "learning_rate": 2.5e-06, |
| "loss": 2.5384014179508085e-07, |
| "num_tokens": 34230.0, |
| "reward": -1.0, |
| "reward_std": 0.0, |
| "rewards/reward_environment_execution/mean": -0.5, |
| "rewards/reward_environment_execution/std": 0.0, |
| "rewards/reward_format_compliance/mean": -0.5, |
| "rewards/reward_format_compliance/std": 0.0, |
| "rewards/reward_investigation_quality/mean": 0.0, |
| "rewards/reward_investigation_quality/std": 0.0, |
| "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": 524.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 640.0, |
| "completions/max_terminated_length": 640.0, |
| "completions/mean_length": 524.0, |
| "completions/mean_terminated_length": 524.0, |
| "completions/min_length": 450.0, |
| "completions/min_terminated_length": 450.0, |
| "epoch": 0.07, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.13970911502838135, |
| "kl": 6.830848860772676e-06, |
| "learning_rate": 3e-06, |
| "loss": 2.8312206268310547e-07, |
| "num_tokens": 39426.0, |
| "reward": -0.22999998927116394, |
| "reward_std": 0.8891193866729736, |
| "rewards/reward_environment_execution/mean": -0.23000000417232513, |
| "rewards/reward_environment_execution/std": 0.31176915764808655, |
| "rewards/reward_format_compliance/mean": -0.15000000596046448, |
| "rewards/reward_format_compliance/std": 0.404145210981369, |
| "rewards/reward_investigation_quality/mean": 0.15000000596046448, |
| "rewards/reward_investigation_quality/std": 0.17320507764816284, |
| "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": 626.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 725.0, |
| "completions/max_terminated_length": 725.0, |
| "completions/mean_length": 626.0, |
| "completions/mean_terminated_length": 626.0, |
| "completions/min_length": 537.0, |
| "completions/min_terminated_length": 537.0, |
| "epoch": 0.08, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.12764990329742432, |
| "kl": 5.7269965054729255e-06, |
| "learning_rate": 3.5e-06, |
| "loss": 2.1606683731079102e-07, |
| "num_tokens": 45026.0, |
| "reward": 0.1550000160932541, |
| "reward_std": 0.7699999809265137, |
| "rewards/reward_environment_execution/mean": -0.0949999988079071, |
| "rewards/reward_environment_execution/std": 0.26999998092651367, |
| "rewards/reward_format_compliance/mean": 0.02499999850988388, |
| "rewards/reward_format_compliance/std": 0.3499999940395355, |
| "rewards/reward_investigation_quality/mean": 0.22500000894069672, |
| "rewards/reward_investigation_quality/std": 0.15000002086162567, |
| "rewards/reward_os_mechanics/mean": 0.0, |
| "rewards/reward_os_mechanics/std": 0.0, |
| "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": 669.5, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 752.0, |
| "completions/max_terminated_length": 752.0, |
| "completions/mean_length": 669.5, |
| "completions/mean_terminated_length": 669.5, |
| "completions/min_length": 555.0, |
| "completions/min_terminated_length": 555.0, |
| "epoch": 0.09, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.1321861445903778, |
| "kl": 5.624260097647493e-06, |
| "learning_rate": 4.000000000000001e-06, |
| "loss": 2.086162567138672e-07, |
| "num_tokens": 50812.0, |
| "reward": -0.22999998927116394, |
| "reward_std": 0.8891193866729736, |
| "rewards/reward_environment_execution/mean": -0.23000000417232513, |
| "rewards/reward_environment_execution/std": 0.31176915764808655, |
| "rewards/reward_format_compliance/mean": -0.15000000596046448, |
| "rewards/reward_format_compliance/std": 0.404145210981369, |
| "rewards/reward_investigation_quality/mean": 0.15000000596046448, |
| "rewards/reward_investigation_quality/std": 0.17320507764816284, |
| "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": 640.25, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 790.0, |
| "completions/max_terminated_length": 790.0, |
| "completions/mean_length": 640.25, |
| "completions/mean_terminated_length": 640.25, |
| "completions/min_length": 471.0, |
| "completions/min_terminated_length": 471.0, |
| "epoch": 0.1, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.12317467480897903, |
| "kl": 6.860862526991696e-06, |
| "learning_rate": 4.5e-06, |
| "loss": 3.0547380447387695e-07, |
| "num_tokens": 56485.0, |
| "reward": 0.1550000160932541, |
| "reward_std": 0.7699999809265137, |
| "rewards/reward_environment_execution/mean": -0.0949999988079071, |
| "rewards/reward_environment_execution/std": 0.26999998092651367, |
| "rewards/reward_format_compliance/mean": 0.02499999850988388, |
| "rewards/reward_format_compliance/std": 0.3499999940395355, |
| "rewards/reward_investigation_quality/mean": 0.22500000894069672, |
| "rewards/reward_investigation_quality/std": 0.15000002086162567, |
| "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": 664.75, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 855.0, |
| "completions/max_terminated_length": 855.0, |
| "completions/mean_length": 664.75, |
| "completions/mean_terminated_length": 664.75, |
| "completions/min_length": 559.0, |
| "completions/min_terminated_length": 559.0, |
| "epoch": 0.11, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.1155354306101799, |
| "kl": 6.481316631834488e-06, |
| "learning_rate": 5e-06, |
| "loss": 2.682209014892578e-07, |
| "num_tokens": 62252.0, |
| "reward": -0.6150000095367432, |
| "reward_std": 0.7699999809265137, |
| "rewards/reward_environment_execution/mean": -0.36500000953674316, |
| "rewards/reward_environment_execution/std": 0.26999998092651367, |
| "rewards/reward_format_compliance/mean": -0.32499998807907104, |
| "rewards/reward_format_compliance/std": 0.3499999940395355, |
| "rewards/reward_investigation_quality/mean": 0.07500000298023224, |
| "rewards/reward_investigation_quality/std": 0.15000002086162567, |
| "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": 699.5, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 880.0, |
| "completions/max_terminated_length": 880.0, |
| "completions/mean_length": 699.5, |
| "completions/mean_terminated_length": 699.5, |
| "completions/min_length": 539.0, |
| "completions/min_terminated_length": 539.0, |
| "epoch": 0.12, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.11765824258327484, |
| "kl": 5.136041295372706e-06, |
| "learning_rate": 4.99847706754774e-06, |
| "loss": 1.862645149230957e-07, |
| "num_tokens": 68150.0, |
| "reward": -0.6150000095367432, |
| "reward_std": 0.7699999809265137, |
| "rewards/reward_environment_execution/mean": -0.36500000953674316, |
| "rewards/reward_environment_execution/std": 0.26999998092651367, |
| "rewards/reward_format_compliance/mean": -0.32499998807907104, |
| "rewards/reward_format_compliance/std": 0.3499999940395355, |
| "rewards/reward_investigation_quality/mean": 0.07500000298023224, |
| "rewards/reward_investigation_quality/std": 0.15000002086162567, |
| "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": 753.5, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 913.0, |
| "completions/max_terminated_length": 913.0, |
| "completions/mean_length": 753.5, |
| "completions/mean_terminated_length": 753.5, |
| "completions/min_length": 543.0, |
| "completions/min_terminated_length": 543.0, |
| "epoch": 0.13, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.1163109764456749, |
| "kl": 6.4769044456625124e-06, |
| "learning_rate": 4.993910125649561e-06, |
| "loss": 2.682209014892578e-07, |
| "num_tokens": 74436.0, |
| "reward": 0.1550000160932541, |
| "reward_std": 0.7699999809265137, |
| "rewards/reward_environment_execution/mean": -0.0949999988079071, |
| "rewards/reward_environment_execution/std": 0.26999998092651367, |
| "rewards/reward_format_compliance/mean": 0.02499999850988388, |
| "rewards/reward_format_compliance/std": 0.3499999940395355, |
| "rewards/reward_investigation_quality/mean": 0.22500000894069672, |
| "rewards/reward_investigation_quality/std": 0.15000002086162567, |
| "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": 691.25, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 870.0, |
| "completions/max_terminated_length": 870.0, |
| "completions/mean_length": 691.25, |
| "completions/mean_terminated_length": 691.25, |
| "completions/min_length": 596.0, |
| "completions/min_terminated_length": 596.0, |
| "epoch": 0.14, |
| "frac_reward_zero_std": 1.0, |
| "grad_norm": 5.699428584193811e-05, |
| "kl": 4.6748977524657676e-06, |
| "learning_rate": 4.986304738420684e-06, |
| "loss": 1.869958907718683e-07, |
| "num_tokens": 80465.0, |
| "reward": 0.5400000214576721, |
| "reward_std": 0.0, |
| "rewards/reward_environment_execution/mean": 0.03999999910593033, |
| "rewards/reward_environment_execution/std": 0.0, |
| "rewards/reward_format_compliance/mean": 0.20000000298023224, |
| "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": 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": 588.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 832.0, |
| "completions/max_terminated_length": 832.0, |
| "completions/mean_length": 588.0, |
| "completions/mean_terminated_length": 588.0, |
| "completions/min_length": 432.0, |
| "completions/min_terminated_length": 432.0, |
| "epoch": 0.15, |
| "frac_reward_zero_std": 1.0, |
| "grad_norm": 0.00010395531717222184, |
| "kl": 9.099214707930514e-06, |
| "learning_rate": 4.975670171853926e-06, |
| "loss": 3.639685814960103e-07, |
| "num_tokens": 86089.0, |
| "reward": -1.0, |
| "reward_std": 0.0, |
| "rewards/reward_environment_execution/mean": -0.5, |
| "rewards/reward_environment_execution/std": 0.0, |
| "rewards/reward_format_compliance/mean": -0.5, |
| "rewards/reward_format_compliance/std": 0.0, |
| "rewards/reward_investigation_quality/mean": 0.0, |
| "rewards/reward_investigation_quality/std": 0.0, |
| "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": 607.75, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 673.0, |
| "completions/max_terminated_length": 673.0, |
| "completions/mean_length": 607.75, |
| "completions/mean_terminated_length": 607.75, |
| "completions/min_length": 517.0, |
| "completions/min_terminated_length": 517.0, |
| "epoch": 0.16, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.1489061564207077, |
| "kl": 9.158140301224194e-06, |
| "learning_rate": 4.962019382530521e-06, |
| "loss": 3.2782554626464844e-07, |
| "num_tokens": 91620.0, |
| "reward": 0.1550000160932541, |
| "reward_std": 0.7699999809265137, |
| "rewards/reward_environment_execution/mean": -0.0949999988079071, |
| "rewards/reward_environment_execution/std": 0.26999998092651367, |
| "rewards/reward_format_compliance/mean": 0.02499999850988388, |
| "rewards/reward_format_compliance/std": 0.3499999940395355, |
| "rewards/reward_investigation_quality/mean": 0.22500000894069672, |
| "rewards/reward_investigation_quality/std": 0.15000002086162567, |
| "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": 657.25, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 737.0, |
| "completions/max_terminated_length": 737.0, |
| "completions/mean_length": 657.25, |
| "completions/mean_terminated_length": 657.25, |
| "completions/min_length": 528.0, |
| "completions/min_terminated_length": 528.0, |
| "epoch": 0.17, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.14814995229244232, |
| "kl": 1.14353706521797e-05, |
| "learning_rate": 4.9453690018345144e-06, |
| "loss": 4.470348358154297e-07, |
| "num_tokens": 97513.0, |
| "reward": -0.22999998927116394, |
| "reward_std": 0.8891193866729736, |
| "rewards/reward_environment_execution/mean": -0.23000000417232513, |
| "rewards/reward_environment_execution/std": 0.31176915764808655, |
| "rewards/reward_format_compliance/mean": -0.15000000596046448, |
| "rewards/reward_format_compliance/std": 0.404145210981369, |
| "rewards/reward_investigation_quality/mean": 0.15000000596046448, |
| "rewards/reward_investigation_quality/std": 0.17320507764816284, |
| "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": 549.25, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 621.0, |
| "completions/max_terminated_length": 621.0, |
| "completions/mean_length": 549.25, |
| "completions/mean_terminated_length": 549.25, |
| "completions/min_length": 510.0, |
| "completions/min_terminated_length": 510.0, |
| "epoch": 0.18, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.13028942048549652, |
| "kl": 1.1230136237827537e-05, |
| "learning_rate": 4.925739315689991e-06, |
| "loss": 4.6193599700927734e-07, |
| "num_tokens": 102810.0, |
| "reward": -0.22999998927116394, |
| "reward_std": 0.8891193866729736, |
| "rewards/reward_environment_execution/mean": -0.23000000417232513, |
| "rewards/reward_environment_execution/std": 0.31176915764808655, |
| "rewards/reward_format_compliance/mean": -0.15000000596046448, |
| "rewards/reward_format_compliance/std": 0.404145210981369, |
| "rewards/reward_investigation_quality/mean": 0.15000000596046448, |
| "rewards/reward_investigation_quality/std": 0.17320507764816284, |
| "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": 565.5, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 686.0, |
| "completions/max_terminated_length": 686.0, |
| "completions/mean_length": 565.5, |
| "completions/mean_terminated_length": 565.5, |
| "completions/min_length": 288.0, |
| "completions/min_terminated_length": 288.0, |
| "epoch": 0.19, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.2010871171951294, |
| "kl": 1.3259304523671744e-05, |
| "learning_rate": 4.903154239845798e-06, |
| "loss": 5.438923835754395e-07, |
| "num_tokens": 108344.0, |
| "reward": 0.1550000160932541, |
| "reward_std": 0.7699999809265137, |
| "rewards/reward_environment_execution/mean": -0.0949999988079071, |
| "rewards/reward_environment_execution/std": 0.26999998092651367, |
| "rewards/reward_format_compliance/mean": 0.02499999850988388, |
| "rewards/reward_format_compliance/std": 0.3499999940395355, |
| "rewards/reward_investigation_quality/mean": 0.22500000894069672, |
| "rewards/reward_investigation_quality/std": 0.15000002086162567, |
| "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": 692.25, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 777.0, |
| "completions/max_terminated_length": 777.0, |
| "completions/mean_length": 692.25, |
| "completions/mean_terminated_length": 692.25, |
| "completions/min_length": 633.0, |
| "completions/min_terminated_length": 633.0, |
| "epoch": 0.2, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.12407222390174866, |
| "kl": 1.1413530501158675e-05, |
| "learning_rate": 4.8776412907378845e-06, |
| "loss": 4.6938657760620117e-07, |
| "num_tokens": 114393.0, |
| "reward": 0.1550000160932541, |
| "reward_std": 0.7699999809265137, |
| "rewards/reward_environment_execution/mean": -0.0949999988079071, |
| "rewards/reward_environment_execution/std": 0.26999998092651367, |
| "rewards/reward_format_compliance/mean": 0.02499999850988388, |
| "rewards/reward_format_compliance/std": 0.3499999940395355, |
| "rewards/reward_investigation_quality/mean": 0.22500000894069672, |
| "rewards/reward_investigation_quality/std": 0.15000002086162567, |
| "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": 633.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 702.0, |
| "completions/max_terminated_length": 702.0, |
| "completions/mean_length": 633.0, |
| "completions/mean_terminated_length": 633.0, |
| "completions/min_length": 568.0, |
| "completions/min_terminated_length": 568.0, |
| "epoch": 0.21, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.14354389905929565, |
| "kl": 1.5171975746852695e-05, |
| "learning_rate": 4.849231551964771e-06, |
| "loss": 6.407499313354492e-07, |
| "num_tokens": 120177.0, |
| "reward": 0.1550000160932541, |
| "reward_std": 0.7699999809265137, |
| "rewards/reward_environment_execution/mean": -0.0949999988079071, |
| "rewards/reward_environment_execution/std": 0.26999998092651367, |
| "rewards/reward_format_compliance/mean": 0.02499999850988388, |
| "rewards/reward_format_compliance/std": 0.3499999940395355, |
| "rewards/reward_investigation_quality/mean": 0.22500000894069672, |
| "rewards/reward_investigation_quality/std": 0.15000002086162567, |
| "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": 669.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 978.0, |
| "completions/max_terminated_length": 978.0, |
| "completions/mean_length": 669.0, |
| "completions/mean_terminated_length": 669.0, |
| "completions/min_length": 444.0, |
| "completions/min_terminated_length": 444.0, |
| "epoch": 0.22, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.12197419255971909, |
| "kl": 1.5716058214820805e-05, |
| "learning_rate": 4.817959636416969e-06, |
| "loss": 6.183981895446777e-07, |
| "num_tokens": 126125.0, |
| "reward": 0.1550000160932541, |
| "reward_std": 0.7699999809265137, |
| "rewards/reward_environment_execution/mean": -0.0949999988079071, |
| "rewards/reward_environment_execution/std": 0.26999998092651367, |
| "rewards/reward_format_compliance/mean": 0.02499999850988388, |
| "rewards/reward_format_compliance/std": 0.3499999940395355, |
| "rewards/reward_investigation_quality/mean": 0.22500000894069672, |
| "rewards/reward_investigation_quality/std": 0.15000002086162567, |
| "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": 774.25, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 1024.0, |
| "completions/max_terminated_length": 1024.0, |
| "completions/mean_length": 774.25, |
| "completions/mean_terminated_length": 774.25, |
| "completions/min_length": 484.0, |
| "completions/min_terminated_length": 484.0, |
| "epoch": 0.23, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.1378677934408188, |
| "kl": 2.0351369357740623e-05, |
| "learning_rate": 4.783863644106502e-06, |
| "loss": 8.493661880493164e-07, |
| "num_tokens": 132326.0, |
| "reward": -0.21249999105930328, |
| "reward_std": 0.9097756743431091, |
| "rewards/reward_environment_execution/mean": -0.23750001192092896, |
| "rewards/reward_environment_execution/std": 0.30335623025894165, |
| "rewards/reward_format_compliance/mean": -0.15000000596046448, |
| "rewards/reward_format_compliance/std": 0.404145210981369, |
| "rewards/reward_investigation_quality/mean": 0.125, |
| "rewards/reward_investigation_quality/std": 0.15000000596046448, |
| "rewards/reward_os_mechanics/mean": 0.05000000074505806, |
| "rewards/reward_os_mechanics/std": 0.10000000149011612, |
| "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": 565.75, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 732.0, |
| "completions/max_terminated_length": 732.0, |
| "completions/mean_length": 565.75, |
| "completions/mean_terminated_length": 565.75, |
| "completions/min_length": 314.0, |
| "completions/min_terminated_length": 314.0, |
| "epoch": 0.24, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.1561768352985382, |
| "kl": 2.7805408535641618e-05, |
| "learning_rate": 4.746985115747918e-06, |
| "loss": 1.1026859283447266e-06, |
| "num_tokens": 137685.0, |
| "reward": -0.6150000095367432, |
| "reward_std": 0.7699999809265137, |
| "rewards/reward_environment_execution/mean": -0.36500000953674316, |
| "rewards/reward_environment_execution/std": 0.26999998092651367, |
| "rewards/reward_format_compliance/mean": -0.32499998807907104, |
| "rewards/reward_format_compliance/std": 0.3499999940395355, |
| "rewards/reward_investigation_quality/mean": 0.07500000298023224, |
| "rewards/reward_investigation_quality/std": 0.15000002086162567, |
| "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": 668.75, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 909.0, |
| "completions/max_terminated_length": 909.0, |
| "completions/mean_length": 668.75, |
| "completions/mean_terminated_length": 668.75, |
| "completions/min_length": 528.0, |
| "completions/min_terminated_length": 528.0, |
| "epoch": 0.25, |
| "frac_reward_zero_std": 0.0, |
| "grad_norm": 0.13285094499588013, |
| "kl": 2.5312373509223107e-05, |
| "learning_rate": 4.707368982147318e-06, |
| "loss": 1.0132789611816406e-06, |
| "num_tokens": 143456.0, |
| "reward": 0.1550000160932541, |
| "reward_std": 0.7699999809265137, |
| "rewards/reward_environment_execution/mean": -0.0949999988079071, |
| "rewards/reward_environment_execution/std": 0.26999998092651367, |
| "rewards/reward_format_compliance/mean": 0.02499999850988388, |
| "rewards/reward_format_compliance/std": 0.3499999940395355, |
| "rewards/reward_investigation_quality/mean": 0.22500000894069672, |
| "rewards/reward_investigation_quality/std": 0.15000002086162567, |
| "rewards/reward_os_mechanics/mean": 0.0, |
| "rewards/reward_os_mechanics/std": 0.0, |
| "step": 25 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 100, |
| "num_input_tokens_seen": 143456, |
| "num_train_epochs": 1, |
| "save_steps": 25, |
| "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 |
| } |
|
|
