Enregistré dans:
Détails bibliographiques
Auteurs principaux: Gulko, Alex, Peng, Yusen, Kumar, Sachin
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2509.00691
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866918149599789056
author Gulko, Alex
Peng, Yusen
Kumar, Sachin
author_facet Gulko, Alex
Peng, Yusen
Kumar, Sachin
contents Sparse autoencoders (SAEs) are a promising approach for uncovering interpretable features in large language models (LLMs). While several automated evaluation methods exist for SAEs, most rely on external LLMs. In this work, we introduce CE-Bench, a novel and lightweight contrastive evaluation benchmark for sparse autoencoders, built on a curated dataset of contrastive story pairs. We conduct comprehensive evaluation studies to validate the effectiveness of our approach. Our results show that CE-Bench reliably measures the interpretability of sparse autoencoders and aligns well with existing benchmarks without requiring an external LLM judge, achieving over 70% Spearman correlation with results in SAEBench. The official implementation and evaluation dataset are open-sourced and publicly available.
format Preprint
id arxiv_https___arxiv_org_abs_2509_00691
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle CE-Bench: Towards a Reliable Contrastive Evaluation Benchmark of Interpretability of Sparse Autoencoders
Gulko, Alex
Peng, Yusen
Kumar, Sachin
Computation and Language
Sparse autoencoders (SAEs) are a promising approach for uncovering interpretable features in large language models (LLMs). While several automated evaluation methods exist for SAEs, most rely on external LLMs. In this work, we introduce CE-Bench, a novel and lightweight contrastive evaluation benchmark for sparse autoencoders, built on a curated dataset of contrastive story pairs. We conduct comprehensive evaluation studies to validate the effectiveness of our approach. Our results show that CE-Bench reliably measures the interpretability of sparse autoencoders and aligns well with existing benchmarks without requiring an external LLM judge, achieving over 70% Spearman correlation with results in SAEBench. The official implementation and evaluation dataset are open-sourced and publicly available.
title CE-Bench: Towards a Reliable Contrastive Evaluation Benchmark of Interpretability of Sparse Autoencoders
topic Computation and Language
url https://arxiv.org/abs/2509.00691