Haiyang Xu

A new approach improves efficiency in Vision-Language Pre-training tasks.

TRIPS enhances efficiency in vision-language tasks by selecting relevant image patches.

This article discusses a new framework for assessing hallucinations in LVLMs.

MIBench tests multimodal models' performance on multiple images.

mPLUG-Owl3 improves understanding of images and videos for better responses.

MaVEn enhances AI's ability to process multiple images for better reasoning.