The Mirage of Performance Gains: Why Contrastive Decoding Fails to Mitigate Object Hallucinations in MLLMs? 文章

ArXiv CS.CV2026-06-05NEWSen作者: Hao Yin, Guangzong Si, Zilei Wang
查看原文 →
关系图谱

The Mirage of Performance Gains: Why Contrastive Decoding Fails to Mitigate Object Hallucinations in MLLMs? · 相关技术

相关技术

POPE BenchmarkAdaptive Plausibility ConstraintMultimodal Large Language Models (MLLMs)contrastive decodingLLM