---
library_name: stable-baselines3
tags:
- Pusher-v4
- deep-reinforcement-learning
- reinforcement-learning
- stable-baselines3
model-index:
- name: PPO
  results:
  - task:
      type: reinforcement-learning
      name: reinforcement-learning
    dataset:
      name: Pusher-v4
      type: Pusher-v4
    metrics:
    - type: mean_reward
      value: -34.22 +/- 3.25
      name: mean_reward
      verified: false
---

# **PPO** Agent playing **Pusher-v4**
This is a trained model of a **PPO** agent playing **Pusher-v4**
using the [stable-baselines3 library](https://github.com/DLR-RM/stable-baselines3).

## Usage (with Stable-baselines3)
TODO: Add your code


```python
# Usage code
import gymnasium as gym
import renderlab as rl
from huggingface_sb3 import load_from_hub
from stable_baselines3 import PPO
from stable_baselines3.common.vec_env import DummyVecEnv
from stable_baselines3.common.evaluation import evaluate_policy
from stable_baselines3.common.monitor import Monitor

repo_id = "VinayHajare/ppo-Pusher-v4"
filename = "ppo-Pusher-v4.zip"

eval_env = gym.make("Pusher-v4",render_mode="rgb_array")
checkpoint = load_from_hub(repo_id, filename)
model = PPO.load(checkpoint,env=eval_env,print_system_info=True)

mean_reward, std_reward = evaluate_policy(model,eval_env, n_eval_episodes=10, deterministic=True)
print(f"mean_reward={mean_reward:.2f} +/- {std_reward}")

# Enjoy trained agent
env = eval_env
env = rl.RenderFrame(env,"./output")
observation, info = env.reset()
for _ in range(1000):
    action, _states = model.predict(observation, deterministic=True)
    observation, rewards, terminated, truncated, info = env.step(action)
env.play()
```