create-rl-app

A CLI application to bootstrap reinforcement learning applications within the Jaxnasium ecosystem. Quickly scaffold new RL projects for either developing environments with baseline algorithms or for altering existing baselines.

What it does

create-rl-app is a command-line tool that helps you quickly set up new reinforcement learning projects using the Jaxnasium framework. It has three commands:

• create-rl-app init <project_name> - bootstrap a new project (uv init --package + uv add jaxnasium + a training script and, optionally, an example environment and algorithm source code).
• create-rl-app add <item> - scaffold an individual file (algorithm, architecture, ...) from jaxnasium into the current directory. Run create-rl-app add --list to see what's available.
• <package> add <item> - the same, but for any other installed package that ships its own create-rl-app.toml (e.g. uvx jaxnasium add ppo).

Useage

uvx (Recommended)

uvx create-rl-app init <project_name>
cd <project_name>
uv run train.py

pipx

pipx run create-rl-app <project_name>
cd <project_name>
# Create a new environment (e.g. conda, venv, etc.)
# source .../bin/activate
python train.py

Or Install Globally

uv tool install create-rl-app

pip install create-rl-app

Dependencies

None.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.