trewby/ppo-CartPole-v1

PPO Agent playing CartPole-v1

This is a trained model of a PPO agent playing CartPole-v1 using the stable-baselines3 library.

Trained on CartPole-v1 in Colab as part of the Hugging Face deep RL course; Box2D environments (for example LunarLander) can be unreliable on Colab’s current Python runtime, so CartPole is used for a clean, reproducible baseline.

Usage (with Stable-baselines3)

import gymnasium as gym
from stable_baselines3 import PPO
from huggingface_sb3 import load_from_hub

# Download model files from the Hub
checkpoint_path = load_from_hub(
    repo_id="trewby/ppo-CartPole-v1",
    filename="ppo-CartPole-v1.zip",
)

# Load the model
model = PPO.load(checkpoint_path)

# Run one episode
env = gym.make("CartPole-v1", render_mode="human")  # use "rgb_array" if you do not want a window
obs, info = env.reset()

done = False
while not done:
    action, _ = model.predict(obs, deterministic=True)
    obs, reward, terminated, truncated, info = env.step(action)
    done = terminated or truncated

env.close()