1 paper · 1 filter
Ziyu Liu, Shengyuan Ding, Xinyu Fang +7
Vision-to-code tasks require models to reconstruct structured visual inputs, such as charts, tables, and SVGs, into executable or structured representations with high visual fideli…