VLRC: Vision-Language Reprojection Consistency as a scalable signal for better feed-forward 3D pretraining figure
AlphaXiv 中文概览(可滚动查看)