How to use and interpret activation patching
Machine Learning
2024-04-24 v1
Abstract
Activation patching is a popular mechanistic interpretability technique, but has many subtleties regarding how it is applied and how one may interpret the results. We provide a summary of advice and best practices, based on our experience using this technique in practice. We include an overview of the different ways to apply activation patching and a discussion on how to interpret the results. We focus on what evidence patching experiments provide about circuits, and on the choice of metric and associated pitfalls.
Cite
@article{arxiv.2404.15255,
title = {How to use and interpret activation patching},
author = {Stefan Heimersheim and Neel Nanda},
journal= {arXiv preprint arXiv:2404.15255},
year = {2024}
}
Comments
A tutorial on activation patching. 13 pages, 2 figures