LordN77/ppo-LunarLander-v2

PPO LunarLander-v2

A Stable-Baselines3 PPO agent trained from scratch for 1,015,808 timesteps using Gymnasium 0.29.1 and Stable-Baselines3 2.3.2 for Deep Reinforcement Learning Course Unit 1.

The model achieved 247.41 ± 20.43 across 10 deterministic evaluation episodes with evaluation seed 1000. The course score (mean minus standard deviation) is 226.98. The replay shows the first evaluation episode.

Load the model

Install Gymnasium 0.29.1 with Box2D, Stable-Baselines3 2.3.2, and huggingface_hub.

from huggingface_hub import hf_hub_download
from stable_baselines3 import PPO

checkpoint = hf_hub_download("LordN77/ppo-LunarLander-v2", "ppo-LunarLander-v2.zip")
model = PPO.load(checkpoint, device="cpu")