Automatically Identifying and Interpreting Sparse Circuits with Hierarchical Tracing
Keywords: Mechanistic Interpretability; Sparse Autoencoder; Circuit Analysis; Large Language Model
TL;DR: We introduce a method named Hierachical Tracing using Sparse Autoencoders and Transcoders to improve the interpretability of Transformer models by isolating key circuits and leveraging automated tools for scalable analysis.
Abstract: We present a novel approach to Transformer circuit analysis using Sparse Autoencoders (SAEs) and Transcoders. SAEs allow fine-grained feature extraction from model activations, while Transcoders handle non-linear MLP outputs for deterministic circuit tracing. Our Hierarchical Tracing method isolates interpretable circuits at both local and global levels, enabling deeper insights into tasks like subject-verb agreement and indirect object identification. Additionally, we introduce an automated workflow leveraging GPT-4o for scalable circuit analysis. This framework provides a clearer understanding of Transformer model behavior and its underlying mechanisms.
Primary Area: interpretability and explainable AI
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 6192
Loading