Understanding Multimodal LLMs: the Mechanistic Interpretability of Llava in Visual Question Answering
원문 초록
Understanding the mechanisms behind Large Language Models (LLMs) is crucial for designing improved models and strategies. While recent studies have yielded valuable insights into the mechanisms of textual LLMs, the mechanisms of Multi-modal Large Language Models (MLLMs) remain underexplored. In this paper, we apply mechanistic interpretability methods to analyze the visual question answering (VQA) mechanisms in the first MLLM, Llava. We compare the mechanisms between VQA and textual QA (TQA) in color answering tasks and find that: a) VQA exhibits a mechanism similar to the in-context learning mechanism observed in TQA; b) the visual features exhibit significant interpretability when projecting the visual embeddings into the embedding space; and c) Llava enhances the existing capabilities of the corresponding textual LLM Vicuna during visual instruction tuning. Based on these findings, we develop an interpretability tool to help users and researchers identify important visual locations for final predictions, aiding in the understanding of visual hallucination. Our method demonstrates faster and more effective results compared to existing interpretability approaches. Code: \url{https://github.com/zepingyu0512/llava-mechanism}
함께 살펴볼 논문
공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.
SMMILE: An Expert-Driven Benchmark for Multimodal Medical In-Context Learning
HalLoc: Token-level Localization of Hallucinations for Vision Language Models
Provoking Multi-modal Few-Shot LVLM via Exploration-Exploitation In-Context Learning
Describe Anything Model for Visual Question Answering on Text-rich Images
DrishtiKon: Multi-Granular Visual Grounding for Text-Rich Document Images
HRIBench: Benchmarking Vision-Language Models for Real-Time Human Perception in Human-Robot Interaction
토론
이 논문과 연결된 포럼 글이 아직 없습니다.