Prabhat17/ppo-LunarLander-v2

PPO LunarLander-v2

This repository contains a Proximal Policy Optimization (PPO) agent trained using Stable-Baselines3.

Evaluation results

The agent was evaluated over 10 episodes using deterministic action selection.

Metric	Value
Mean reward	224.86
Standard deviation	24.08
Mean reward minus standard deviation	200.78
Evaluation episodes	10

Environment compatibility

Important: The model was trained and evaluated using Gymnasium LunarLander-v3. It has not been trained or evaluated using LunarLander-v2.

The results above are self-reported results from the LunarLander-v3 environment and have not been independently verified for LunarLander-v2.

Model file

The trained model is available in this repository as ppo-lunarlander.zip.

Usage

import gymnasium as gym
from stable_baselines3 import PPO

env = gym.make("LunarLander-v3")
model = PPO.load("ppo-lunarlander.zip")

observation, info = env.reset()
terminated = False
truncated = False

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

env.close()