LGAICLFeb 24, 2025

FADE: Why Bad Descriptions Happen to Good Features

arXiv:2502.16994v29 citationsh-index: 32Has CodeACL
Originality Incremental advance
AI Analysis

This work addresses a critical gap in automated interpretability pipelines for researchers, though it is incremental as it builds on existing methods to improve evaluation.

The paper tackles the lack of standardized evaluation for feature descriptions in mechanistic interpretability by introducing FADE, a scalable framework that quantifies alignment across four metrics, revealing challenges in generating descriptions, especially for SAEs compared to MLP neurons.

Recent advances in mechanistic interpretability have highlighted the potential of automating interpretability pipelines in analyzing the latent representations within LLMs. While this may enhance our understanding of internal mechanisms, the field lacks standardized evaluation methods for assessing the validity of discovered features. We attempt to bridge this gap by introducing FADE: Feature Alignment to Description Evaluation, a scalable model-agnostic framework for automatically evaluating feature-to-description alignment. FADE evaluates alignment across four key metrics - Clarity, Responsiveness, Purity, and Faithfulness - and systematically quantifies the causes of the misalignment between features and their descriptions. We apply FADE to analyze existing open-source feature descriptions and assess key components of automated interpretability pipelines, aiming to enhance the quality of descriptions. Our findings highlight fundamental challenges in generating feature descriptions, particularly for SAEs compared to MLP neurons, providing insights into the limitations and future directions of automated interpretability. We release FADE as an open-source package at: https://github.com/brunibrun/FADE

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes