{ "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": 320001, "requested_timesteps": 2048, "actual_timesteps": 2048, "n_envs": 1, "vec_env": "dummy", "config_profile": "production", "fidelity": "production", "reward_profile": "time", "local_selector": "geometry", "local_model": null, "device": "mps", "wall_seconds": 260.069, "max_wall_s": 800.0, "created_unix_s": 1789128140.691232, "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": "d3c0ba5cfa947896b0d80d590cb5159c9c301c204e7884d00a1e0e35a66d9499", "monitor_files": [ "output/hrl/p1-smoke-time/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": 2, "successes": 2, "success_rate": 1.0, "rows": [ { "episode": 0, "seed": 1320001, "steps": 12, "cleared": 1, "accounted": 20, "virtual_time_s": 3435.320338, "stop_reason": null, "anomalies": [], "success": true }, { "episode": 1, "seed": 1320002, "steps": 9, "cleared": 1, "accounted": 20, "virtual_time_s": 3220.677424, "stop_reason": null, "anomalies": [], "success": true } ] }, "eval_manifest": null }