README.md

metadata

library_name: stable-baselines3
tags:
  - FetchPushDense-v4
  - deep-reinforcement-learning
  - reinforcement-learning
  - stable-baselines3
model-index:
  - name: TQC
    results:
      - task:
          type: reinforcement-learning
          name: reinforcement-learning
        dataset:
          name: FetchPushDense-v4
          type: FetchPushDense-v4
        metrics:
          - type: mean_reward
            value: '-2.02 +/- 1.19'
            name: mean_reward
            verified: false
license: mit
language:
  - en

TQC Agent playing FetchPushDense-v4

• Github Repository
• Google Colab Notebook
• Finding Theta - Blog Post

Then, you can load the model using the following Python code:

import gymnasium as gym
from sb3_contrib import TQC
from stable_baselines3.common.env_util import make_vec_env

gymnasium.register_envs(gymnasium_robotics)

# Load the trained model
model = TQC.load("best-model.zip")

# Create the environment
env = make_vec_env("FetchPushDense-v4", n_envs=1)

# Reset the environment
obs, info = env.reset()

# Enjoy the trained agent
for _ in range(1000):
    action, _states = model.predict(obs, deterministic=True)
    obs, rewards, terminated, truncated, info = env.step(action)
    if terminated or truncated:
        obs, info = env.reset()
    env.render()
env.close()

Hugging Face Hub

You can also use the Hugging Face Hub to load the model. First, you need to install the Hugging Face Hub library:

pip install huggingface_hub

Then, you can load the model from the hub using the following code:

from huggingface_hub import hf_hub_download
import torch as th
from sb3_contrib import TQC
from stable_baselines3.common.env_util import make_vec_env

gymnasium.register_envs(gymnasium_robotics)

# Download the model from the Hub
model_path = hf_hub_download(repo_id="kuds/fetch-push-dense-tqc", filename="best-model.zip")

# Load the model
model = TQC.load(model_path)

# Create the environment
env = make_vec_env("FetchPushDense-v4", n_envs=1)

# Enjoy the trained agent
obs = env.reset()
for i in range(1000):
    action, _states = model.predict(obs, deterministic=True)
    obs, rewards, dones, info = env.step(action)
    env.render("human")