envquest 0.0.2

A collection of Reinforcement Learning algorithms to train autonomous agents in different environments.

RL Studio

Play with reinforcement learning algorithms.

Requirements

You need these requirements to run the project:

• python >= 3.8
• uv
• ffmpeg

# Linux
sudo apt install ffmpeg
# MacOS
brew install ffmpeg

Installation

Install the python dependencies

uv venv .venv
source .venv/bin/activate
uv sync --frozen

Start a local WandB server to track your experiments

wandb server start

Usage

Run a random policy agent

Run a random agent in a Gym environment

python -m scripts.run_random_agent

Read the python script for more information: scripts/run_random_agent.py

Train Agents

DQN Agent

Train a DQN policy agent in a Gym environment

python -m scripts.train_dqn

Read the python script for more information: scripts/train_dqn.py

Sarsa Agent

Train a Sarsa policy agent in a Gym environment

python -m scripts.train_sarsa

Read the python script for more information: scripts/train_sarsa.py