output
20152025
most citedMulti-modal Understanding and Generation for Medical Images and Text via Vision-Language Pre-Training

213 citations

Showing 2018Show all

4 papers · 1 filter