TransformerLens

Neel Nanda and open-source contributors, GitHub, 2022

A Python library for mechanistic interpretability of GPT-style language models that makes it easy to cache, inspect and edit internal activations.

Code repositoryIntermediateA few hours to get started