提出新方法精准定位多任务视觉语言模型中关键神经元。
From Heads to Neurons: Causal Attribution and Steering in Multi-Task Vision-Language Models
- 基于注意力头条件下的因果贡献评估神经元重要性
- 在四个任务上验证，显著提升关键神经元识别准确率
- 适合研究模型可解释性与可控生成的学者使用
近期研究致力于在视觉语言模型（VLMs）中进行神经元级解释，以识别影响最终预测的关键神经元。然而，现有方法多聚焦单一任务，难以跨任务比较神经元重要性；且排名策略通常孤立评分神经元，忽视任务相关注意力头如何塑造前馈网络（FFN）神经元的写入效应。这一疏漏会加剧多任务场景下神经元的多义性，导致关键神经元识别与干预引入噪声。本文提出无梯度框架 HONES（Head-Oriented Neuron Explanation & Steering），实现多任务 VLM 的任务感知神经元归因与调控。HONEs 根据任务相关注意力头条件，评估 FFN 神经元的因果写入贡献，并通过轻量缩放调节显著神经元。在四个多样化多模态任务及两个主流 VLM 上的实验表明，HONEs 在识别任务关键神经元方面优于现有方法，并在调控后提升模型性能。代码已开源：https://github.com/petergit1/HONES。
原文摘要 · Abstract (English)
Recent work has increasingly explored neuron-level interpretation in vision-language models (VLMs) to identify neurons critical to final predictions. However, existing neuron analyses generally focus on single tasks, limiting the comparability of neuron importance across tasks. Moreover, ranking strategies tend to score neurons in isolation, overlooking how task-dependent information pathways shape the write-in effects of feed-forward network (FFN) neurons. This oversight can exacerbate neuron polysemanticity in multi-task settings, introducing noise into the identification and intervention of task-critical neurons. In this study, we propose HONES (Head-Oriented Neuron Explanation & Steering), a gradient-free framework for task-aware neuron attribution and steering in multi-task VLMs. HONES ranks FFN neurons by their causal write-in contributions conditioned on task-relevant attention heads, and further modulates salient neurons via lightweight scaling. Experiments on four diverse multimodal tasks and two popular VLMs show that HONES outperforms existing methods in identifying task-critical neurons and improves model performance after steering. Our source code is released at: https://github.com/petergit1/HONES.
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。