|
--- |
|
tags: |
|
- deep-reinforcement-learning |
|
- reinforcement-learning |
|
- stable-baselines3 |
|
--- |
|
# ThomasSimonini/ppo-SpaceInvadersNoFrameskip-v4 |
|
|
|
This is a pre-trained model of a PPO agent playing SpaceInvadersNoFrameskip using the [stable-baselines3](https://github.com/DLR-RM/stable-baselines3) library. It is taken from [RL-trained-agents](https://github.com/DLR-RM/rl-trained-agents) |
|
|
|
|
|
### Usage (with Stable-baselines3) |
|
Using this model becomes easy when you have stable-baselines3 and huggingface_sb3 installed: |
|
``` |
|
pip install stable-baselines3 |
|
pip install huggingface_sb3 |
|
``` |
|
|
|
Then, you can use the model like this: |
|
|
|
```python |
|
import gym |
|
|
|
from huggingface_sb3 import load_from_hub |
|
from stable_baselines3 import PPO |
|
from stable_baselines3.common.evaluation import evaluate_policy |
|
from stable_baselines3.common.env_util import make_atari_env |
|
from stable_baselines3.common.vec_env import VecFrameStack |
|
|
|
# Retrieve the model from the hub |
|
## repo_id = id of the model repository from the Hugging Face Hub (repo_id = {organization}/{repo_name}) |
|
## filename = name of the model zip file from the repository |
|
checkpoint = load_from_hub(repo_id="ThomasSimonini/ppo-SpaceInvadersNoFrameskip-v4", filename="ppo-SpaceInvadersNoFrameskip-v4.zip") |
|
model = PPO.load(checkpoint) |
|
``` |
|
|
|
### Evaluation Results |
|
Mean_reward: 627.160 (162 eval episodes) |
|
|
|
|