用稀疏自编码器让视觉语言模型的神经元更单一、可解释。
Sparse Autoencoders Learn Monosemantic Features in Vision-Language Models
- 在视觉语言模型上训练稀疏自编码器，提升神经元语义单一性。
- 稀疏性和宽潜空间是提升单一语义的关键因素。
- 无需修改语言模型，就能通过干预视觉编码器控制输出。
稀疏自编码器（SAEs）近年来被用于提升大语言模型的可解释性与可控性，这对人工智能安全至关重要。本文将SAEs应用于视觉语言模型（如CLIP），提出一套完整的神经元级单一语义评估框架。为确保评估贴近人类感知，我们基于大规模用户研究构建了基准测试。实验表明，对VLM训练SAEs显著提升了神经元的单一语义性，其中稀疏性和宽潜空间影响最显著。此外，我们证明对CLIP视觉编码器施加SAE干预，可直接引导多模态大语言模型（如LLaVA）的输出，且无需修改底层语言模型。这些发现凸显了SAEs作为无监督工具在增强VLM可解释性与控制力方面的实用性与有效性。代码与基准数据见https://github.com/ExplainableML/sae-for-vlm。
原文摘要 · Abstract (English)
Sparse Autoencoders (SAEs) have recently gained attention as a means to improve the interpretability and steerability of Large Language Models (LLMs), both of which are essential for AI safety. In this work, we extend the application of SAEs to Vision-Language Models (VLMs), such as CLIP, and introduce a comprehensive framework for evaluating monosemanticity at the neuron-level in visual representations. To ensure that our evaluation aligns with human perception, we propose a benchmark derived from a large-scale user study. Our experimental results reveal that SAEs trained on VLMs significantly enhance the monosemanticity of individual neurons, with sparsity and wide latents being the most influential factors. Further, we demonstrate that applying SAE interventions on CLIP's vision encoder directly steers multimodal LLM outputs (e.g., LLaVA), without any modifications to the underlying language model. These findings emphasize the practicality and efficacy of SAEs as an unsupervised tool for enhancing both interpretability and control of VLMs. Code and benchmark data are available at https://github.com/ExplainableML/sae-for-vlm.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。