FmyzA's picture
Upload PPO agent trained locally (evaluate + replay video + model card)
0f8b335 verified
|
Raw History Blame Contribute Delete
1.9 kB
---
library_name: stable-baselines3
tags:
- InvertedDoublePendulum-v5
- deep-reinforcement-learning
- reinforcement-learning
- stable-baselines3
- gymnasium
- ppo
model-index:
- name: ppo-InvertedDoublePendulum-v5
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: InvertedDoublePendulum-v5
type: InvertedDoublePendulum-v5
metrics:
- type: mean_reward
value: 9358.13 +/- 0.39
name: mean_reward
verified: false
---
# **ppo-InvertedDoublePendulum-v5** Agent playing **InvertedDoublePendulum-v5**
This is a trained model of a **PPO** agent playing **InvertedDoublePendulum-v5**
using the [stable-baselines3 library](https://github.com/DLR-RM/stable-baselines3)
and [gymnasium](https://gymnasium.farama.org/).
## Evaluation
| metric | value |
|---|---|
| mean_reward | **9358.13** |
| std_reward | 0.39 |
| mean - std | 9357.74 |
| episodes | 10 |
## Usage (with Stable-Baselines3)
```python
import gymnasium as gym
from stable_baselines3 import PPO
from huggingface_sb3 import load_from_hub
repo_id = "ppo-InvertedDoublePendulum-v5" # or <your-user>/<repo-name>
filename = "ppo-InvertedDoublePendulum-v5.zip"
checkpoint = load_from_hub(repo_id, filename)
model = PPO.load(checkpoint)
env = gym.make("InvertedDoublePendulum-v5", render_mode="rgb_array")
obs, _ = env.reset()
for _ in range(1000):
action, _ = model.predict(obs, deterministic=True)
obs, reward, terminated, truncated, info = env.step(action)
if terminated or truncated:
obs, _ = env.reset()
env.close()
```
## Files
- `ppo-InvertedDoublePendulum-v5.zip` - the trained policy
- `replay.mp4` - a deterministic replay of one episode
- `results.json` - raw evaluation numbers
- `config.json` - the constructor arguments stored inside the checkpoint