tyro

<p align="left"> <a href="https://codecov.io/gh/brentyi/tyro"> <img alt="codecov" src="https://codecov.io/gh/brentyi/tyro/branch/main/graph/badge.svg" /> </a> <a href="https://pypi.org/project/tyro/"> <img src="https://static.pepy.tech/personalized-badge/tyro?period=total&units=INTERNATIONAL_SYSTEM&left_color=GRAY&right_color=GREEN&left_text=downloads" alt="PyPI Downloads"> </a> <a href="https://pypi.org/project/tyro/"> <img alt="codecov" src="https://img.shields.io/pypi/pyversions/tyro" /> </a> </p> <p align="left"> <em><a href="https://brentyi.github.io/tyro">Documentation</a></em> &nbsp;&nbsp;&bull;&nbsp;&nbsp; <em><code>pip install tyro</code></em> </p>

<strong><code>tyro.cli()</code></strong> is a tool for generating CLI interfaces from type-annotated Python.

We can define configurable scripts using functions: <br /> <img width="836" height="322" alt="image" src="https://github.com/user-attachments/assets/5347693e-b126-4b05-8ef6-4ffe6cbb22f2" /> <br /> <br /> Or instantiate configs defined using tools like dataclasses, pydantic, and attrs: <br /> <img width="836" height="334" alt="image" src="https://github.com/user-attachments/assets/2c27065d-073e-4c59-9df9-20d153b4a33b" /> <br />

Other features include helptext generation, nested structures, subcommands, and shell completion. For examples and the API reference, see our documentation.

Why tyro?

1. Define things once. Standard Python type annotations, docstrings, and default values are parsed to automatically generate command-line interfaces with nice helptext.

2. Static types. Unlike tools dependent on dictionaries, YAML, or dynamic namespaces, arguments populated by tyro are better undestood by IDEs and language servers, as well as static checking tools like pyright and mypy.

3. Modularity. tyro supports hierarchical configurations, which make it easy to decentralize definitions, defaults, and documentation.

In the wild

tyro is designed to be lightweight for throwaway scripts, while improving maintainability for larger projects. Examples:

<table> <tr> <td> <a href="https://github.com/nerfstudio-project/nerfstudio/"> nerfstudio-project/nerfstudio <br /><img alt="GitHub stars" src="https://img.shields.io/github/stars/nerfstudio-project/nerfstudio?style=social" /> </a> </td> <td> Open-source tools for neural radiance fields. </td> </tr> <tr> <td> <a href="https://github.com/Sea-Snell/JAXSeq/"> Sea-Snell/JAXSeq <br /><img alt="GitHub stars" src="https://img.shields.io/github/stars/Sea-Snell/JAXSeq?style=social" /> </a> </td> <td>Train very large language models in Jax.</td> </tr> <tr> <td> <a href="https://github.com/kevinzakka/obj2mjcf"> kevinzakka/obj2mjcf <br /><img alt="GitHub stars" src="https://img.shields.io/github/stars/kevinzakka/obj2mjcf?style=social" /> </a> </td> <td>Interface for processing OBJ files for Mujoco.</td> </tr> <tr> <td> <a href="https://github.com/blurgyy/jaxngp"> blurgyy/jaxngp <br /><img alt="GitHub stars" src="https://img.shields.io/github/stars/blurgyy/jaxngp?style=social" /> </a> </td> <td> CUDA-accelerated implementation of <a href="https://nvlabs.github.io/instant-ngp/">instant-ngp</a>, in JAX. </td> </tr> <tr> <td> <a href="https://github.com/NVIDIAGameWorks/kaolin-wisp"> NVIDIAGameWorks/kaolin-wisp <br /><img alt="GitHub stars" src="https://img.shields.io/github/stars/NVIDIAGameWorks/kaolin-wisp?style=social" /> </a> </td> <td>PyTorch library for neural fields.</td> </tr> <tr> <td> <a href="https://github.com/autonomousvision/sdfstudio"> autonomousvision/sdfstudio <br /><img alt="GitHub stars" src="https://img.shields.io/github/stars/autonomousvision/sdfstudio?style=social" /> </a> </td> <td>Unified framework for surface reconstruction.</td> </tr> <tr> <td> <a href="https://github.com/openrlbenchmark/openrlbenchmark"> openrlbenchmark/openrlbenchmark <br /><img alt="GitHub stars" src="https://img.shields.io/github/stars/openrlbenchmark/openrlbenchmark?style=social" /> </a> </td> <td>Collection of tracked experiments for reinforcement learning.</td> </tr> <tr> <td> <a href="https://github.com/vwxyzjn/cleanrl"> vwxyzjn/cleanrl <br /><img alt="GitHub stars" src="https://img.shields.io/github/stars/vwxyzjn/cleanrl?style=social" /> </a> </td> <td>Single-file implementation of deep RL algorithms.</td> </tr> <tr> <td> <a href="https://github.com/pytorch-labs/LeanRL/"> pytorch-labs/LeanRL <br /><img alt="GitHub stars" src="https://img.shields.io/github/stars/pytorch-labs/LeanRL?style=social" /> </a> </td> <td>Fork of CleanRL, optimized using PyTorch 2 features.</td> </tr> <tr> <td> <a href="https://github.com/pytorch/torchtitan"> pytorch/torchtitan <br /><img alt="GitHub stars" src="https://img.shields.io/github/stars/pytorch/torchtitan?style=social" /> </a> </td> <td>PyTorch-native platform for training generative AI models.</td> </tr> <tr> <td> <a href="https://github.com/KwaiVGI/LivePortrait"> KwaiVGI/LivePortrait <br /><img alt="GitHub stars" src="https://img.shields.io/github/stars/KwaiVGI/LivePortrait?style=social" /> </a> </td> <td>Stitching and retargeting for portraits.</td> </tr> <tr> <td> <a href="https://github.com/Physical-Intelligence/openpi/"> Physical-Intelligence/openpi <br /><img alt="GitHub stars" src="https://img.shields.io/github/stars/Physical-Intelligence/openpi?style=social" /> </a> </td> <td>Open-source models for robotics.</td> </tr> <tr> <td> <a href="https://github.com/NVIDIA/Isaac-GR00T"> NVIDIA/Isaac-GR00T <br /><img alt="GitHub stars" src="https://img.shields.io/github/stars/NVIDIA/Isaac-GR00T?style=social" /> </a> </td> <td>Foundation model for generalist robots.</td> </tr> <tr> <td> <a href="https://github.com/haosulab/ManiSkill"> haosulab/ManiSkill <br /><img alt="GitHub stars" src="https://img.shields.io/github/stars/haosulab/ManiSkill?style=social" /> </a> </td> <td>GPU-parallelized simulation + learning for robot manipulation.</td> </tr> <tr> <td> <a href="https://github.com/mujocolab/mjlab"> mujocolab/mjlab <br /><img alt="GitHub stars" src="https://img.shields.io/github/stars/mujocolab/mjlab?style=social" /> </a> </td> <td>Lightweight, modular abstractions for RL and sim-to-real robotics.</td> </tr> <tr> <td> <a href="https://github.com/amazon-far/holosoma"> amazon-far/holosoma <br /><img alt="GitHub stars" src="https://img.shields.io/github/stars/amazon-far/holosoma?style=social" /> </a> </td> <td>Humanoid robotics framework for RL training and deployment.</td> </tr> <tr> <td> <a href="https://github.com/MalcolmMielle/bark_monitor"> MalcolmMielle/bark_monitor <br /><img alt="GitHub stars" src="https://img.shields.io/github/stars/MalcolmMielle/bark_monitor?style=social" /> </a> </td> <td>Show your neighbor that your dog doesn't bark!</td> </tr> </table>

Alternatives

tyro is opinionated. If any design decisions don't make sense, feel free to file an issue!

You might also consider one of many alternative libraries. Some that we like:

• cappa offers a similar core feature set but with very different ergonomics. It looks polished and well-maintained!
• cyclopts and defopt has very comprehensive type annotation support and a heavier emphasis on subcommand generation.
• simple-parsing and jsonargparse provide deeper integration with configuration file formats like YAML and JSON.
• clipstick, which focuses on simplicity + generating CLIs from Pydantic models.
• datargs provides a minimal API for dataclasses.
• fire and clize support arguments without type annotations.

There are also some options that directly extend tyro:

• mininterface simultaneously generates GUI, TUI, web, CLI, and file-based program configuration.
• manuscript generates CLI interfaces from a simple block of configuration variables.

We also have some notes on tyro's design goals and other alternatives in the docs here.