Understanding Geometric Representations in Self-Supervised Vision Transformers via Subspace Intervention figure
AlphaXiv 中文概览(可滚动查看)