Saved in:
Bibliographic Details
Main Authors: Wang, Shun, Loakman, Tyler, Lei, Youbo, Liu, Yi, Yang, Bohao, Zhao, Yuting, Yang, Dong, Lin, Chenghua
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2507.06427
Tags: Add Tag
No Tags, Be the first to tag this record!
Table of Contents:
  • Large Language Models (LLMs) are traditionally viewed as black-box algorithms, therefore reducing trustworthiness and obscuring potential approaches to increasing performance on downstream tasks. In this work, we apply an effective LLM decomposition method using a dictionary-learning approach with sparse autoencoders. This helps extract monosemantic features from polysemantic LLM neurons. Remarkably, our work identifies model-internal misunderstanding, allowing the automatic reformulation of the prompts with additional annotations to improve the interpretation by LLMs. Moreover, this approach demonstrates a significant performance improvement in downstream tasks, such as mathematical reasoning and metaphor detection.