paper-with-me

홈 › Papers

Circuit Tracing in Vision-Language Models: Understanding the Internal Mechanisms of Multimodal Thinking

2026-02-23 · Jingcheng Yang, Tianhu Xiong, Shengyi Qian, Klara Nahrstedt, Mingyuan Wu arxiv

Vision-language models (VLMs) are powerful but remain opaque black boxes. We introduce the first framework for transparent circuit tracing in VLMs to systematically analyze multimodal reasoning. By utilizing transcoders, attribution graphs, and attention-based methods, we uncover how VLMs hierarchically integrate visual and semantic concepts. We reveal that distinct visual feature circuits can handle mathematical reasoning and support cross-modal associations. Validated through feature steering and circuit patching, our framework proves these circuits are causal and controllable, laying the groundwork for more explainable and reliable VLMs.

📄 PDF Abstract BibTeX arXiv:2602.20330

Code (0)

등록된 구현이 없습니다.

Tasks

Mathematical ReasoningMultimodal Reasoning

Similar Papers 제목 키워드 기반

Detecting Unfaithful Chain-of-Thought via Circuit-Guided Internal-External Discrepancy

2026-05-25 · Xu Shen, Zhen Tan, Song Wang, Pingjun Hong 외 arxiv

Chain-of-thought (CoT) reasoning improves the problem-solving ability of large language models (LLMs), but generated reasoning traces may not faithfully reflect the model's actual decision process. Existing CoT unfaithfu…

Pierce the Mists, Greet the Sky: Decipher Knowledge Overshadowing via Knowledge Circuit Analysis

2025-05-20 · Haoming Huang, Yibo Yan, Jiahao Huo, Xin Zou 외

Large Language Models (LLMs), despite their remarkable capabilities, are hampered by hallucinations. A particularly challenging variant, knowledge overshadowing, occurs when one piece of activated knowledge inadvertently…

Hallucination

Language over Content: Tracing Cultural Understanding in Multilingual Large Language Models

2025-10-18 · Seungho Cho, Changgeon Ko, Eui Jun Hwang, Junmyeong Lee 외 arxiv

Large language models (LLMs) are increasingly used across diverse cultural contexts, making accurate cultural understanding essential. Prior evaluations have mostly focused on output-level performance, obscuring the fact…

ADAG: Automatically Describing Attribution Graphs

2026-04-08 · Aryaman Arora, Zhengxuan Wu, Jacob Steinhardt, Sarah Schwettmann arxiv

In language model interpretability research, \textbf{circuit tracing} aims to identify which internal features causally contributed to a particular output and how they affected each other, with the goal of explaining the…

CircuitProbe: Tracing Visual Temporal Evidence Flow in Video Language Models

2025-07-25 · Yiming Zhang, Zhuokai Zhao, Chengzhang Yu, Kun Wang 외 arxiv

Autoregressive large vision--language models (LVLMs) interface video and language by projecting video features into the LLM's embedding space as continuous visual token embeddings. However, it remains unclear where tempo…