2026
Language Bias in LVLMs: From In-Depth Analysis to Simple and Effective Mitigation
ICML 2026poster
Large Vision-Language Models (LVLMs) extend large language models with visual understanding, but remain vulnerable to hallucination, where outputs are fluent yet inconsistent with images. Recent studies link this issue to language bias—the tendency of LVLMs to over-rely on text while neglecting visu…