{ "level": "high", "problem": 3, "stage": 1, "stage_name": "s1-single", "algorithm": "maskable_ppo", "feature_version": 4, "action_layout_version": 1, "reward_version": 1, "obs_dim": 1248, "act_dim": 120, "seed": 7, "requested_timesteps": 32, "actual_timesteps": 48, "n_envs": 1, "vec_env": "dummy", "config_profile": "production", "fidelity": "production", "reward_profile": "time", "local_selector": "geometry", "local_model": null, "device": "mps", "wall_seconds": 17.241, "max_wall_s": 120.0, "created_unix_s": 1789128385.1594582, "python": "3.13.9", "platform": "macOS-27.0-arm64-arm-64bit-Mach-O", "torch": "2.11.0", "numpy": "2.5.3", "stable_baselines3": "2.9.0", "sb3_contrib": "2.9.0", "mp_actor": true, "model_sha256": "ccea1b1c47d6542ad7ee1f9f1a7e7b23db6228a257a608a07850f47c3e388247", "monitor_files": [ "output/hrl/p1-resume-smoke/high_curve_q3_env0.monitor.csv" ], "reward_note": "reward_profile=time 使用 HRLReward.time_only(P0.4 纯时间口径);reward_profile=prompt 使用原始 §5.2 塑形。模型元数据已固化口径。", "evaluation": { "episodes": 0, "successes": 0, "success_rate": null, "rows": [] }, "eval_manifest": null }