LinguaLens: Towards Interpreting Linguistic Mechanisms of Large Language Models via Sparse Auto-Encoder
Published in EMNLP, 2025
We introduce LinguaLens, a sparse auto-encoder–based framework for analyzing linguistic mechanisms in large language models.
Recommended citation: Yi Jing, Zijun Yao, Hongzhu Guo, Lingxu Ran, Xiaozhi Wang, Lei Hou, Juanzhi Li. 2025. LinguaLens: Towards Interpreting Linguistic Mechanisms of Large Language Models via Sparse Auto-Encoder. EMNLP 2025.
Download Paper