Generate, but Verify: Reducing Hallucination in Vision-Language Models with Retrospective Resampling

Open in new window