google-deepmind/penzai
A JAX library for building, editing, and visualizing neural networks as legible functional pytree data structures.
Not currently ranked — collecting fresh signals.
star history
Penzai provides tools for working with neural networks after training, supporting tasks like model surgery, ablation studies, and internal activation inspection. It includes an interactive tree pretty-printer called Treescope for visualizing deeply-nested JAX pytrees and arrays. The library offers pytree manipulation utilities and a named axis system for flexible JAX programming.
Frequently asked
- What is google-deepmind/penzai?
- A JAX library for building, editing, and visualizing neural networks as legible functional pytree data structures.
- Is penzai open source?
- Yes — google-deepmind/penzai is open source, released under the Apache-2.0 license.
- What language is penzai written in?
- google-deepmind/penzai is primarily written in Python.
- How popular is penzai?
- google-deepmind/penzai has 1.9k stars on GitHub.
- Where can I find penzai?
- google-deepmind/penzai is on GitHub at https://github.com/google-deepmind/penzai.