config.json
1.3 KB · 44 lines · json Raw
1 {
2 "algorithm": "PPO",
3 "library": "stable-baselines3",
4 "base_environment": "LunarLander-v3",
5 "model_filename": "ppo-LunarLander-v3-flip-128x128.zip",
6 "reward_version": "v4-soft-touchdown",
7 "reward_config_filename": "reward_config.json",
8 "reward_wrapper_filename": "flip_landing_reward_wrapper.py",
9 "observation_dimensions": 11,
10 "architecture": {
11 "actor_layers": [
12 128,
13 128
14 ],
15 "critic_layers": [
16 128,
17 128
18 ]
19 },
20 "evaluation": {
21 "n_episodes": 100,
22 "mean_shaped_reward": -325.2546577557927,
23 "std_shaped_reward": 435.5396409711088,
24 "shaped_course_score": -760.7942987269015,
25 "min_shaped_reward": -3342.837273847694,
26 "max_shaped_reward": -88.07963784059791,
27 "mean_original_reward": -706.4599452970039,
28 "std_original_reward": 73.7189566670546,
29 "original_course_score": -780.1789019640585,
30 "min_original_reward": -818.5765463852929,
31 "max_original_reward": -463.0414724621158,
32 "original_reward_200_rate": 0.0,
33 "flip_completion_rate": 0.36,
34 "recovery_completion_rate": 0.12,
35 "recovery_given_flip_rate": 0.3333333333333333,
36 "safe_landing_rate": 0.0,
37 "flip_landing_rate": 0.0,
38 "landing_given_flip_rate": 0.0,
39 "stable_landing_rate": 0.0,
40 "inside_landing_zone_rate": 0.2,
41 "outside_zone_landing_rate": 0.0,
42 "in_zone_crash_rate": 0.08
43 }
44 }