Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Marks, Samuel, Rager, Can, Michaud, Eric J., Belinkov, Yonatan, Bau, David, Mueller, Aaron
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913760398016512
author Marks, Samuel
Rager, Can
Michaud, Eric J.
Belinkov, Yonatan
Bau, David
Mueller, Aaron
author_facet Marks, Samuel
Rager, Can
Michaud, Eric J.
Belinkov, Yonatan
Bau, David
Mueller, Aaron
contents We introduce methods for discovering and applying sparse feature circuits. These are causally implicated subnetworks of human-interpretable features for explaining language model behaviors. Circuits identified in prior work consist of polysemantic and difficult-to-interpret units like attention heads or neurons, rendering them unsuitable for many downstream applications. In contrast, sparse feature circuits enable detailed understanding of unanticipated mechanisms. Because they are based on fine-grained units, sparse feature circuits are useful for downstream tasks: We introduce SHIFT, where we improve the generalization of a classifier by ablating features that a human judges to be task-irrelevant. Finally, we demonstrate an entirely unsupervised and scalable interpretability pipeline by discovering thousands of sparse feature circuits for automatically discovered model behaviors.
format Preprint
id arxiv_https___arxiv_org_abs_2403_19647
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Models
Marks, Samuel
Rager, Can
Michaud, Eric J.
Belinkov, Yonatan
Bau, David
Mueller, Aaron
Machine Learning
Artificial Intelligence
Computation and Language
We introduce methods for discovering and applying sparse feature circuits. These are causally implicated subnetworks of human-interpretable features for explaining language model behaviors. Circuits identified in prior work consist of polysemantic and difficult-to-interpret units like attention heads or neurons, rendering them unsuitable for many downstream applications. In contrast, sparse feature circuits enable detailed understanding of unanticipated mechanisms. Because they are based on fine-grained units, sparse feature circuits are useful for downstream tasks: We introduce SHIFT, where we improve the generalization of a classifier by ablating features that a human judges to be task-irrelevant. Finally, we demonstrate an entirely unsupervised and scalable interpretability pipeline by discovering thousands of sparse feature circuits for automatically discovered model behaviors.
title Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Models
topic Machine Learning
Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2403.19647