1 paper across 1 session
Our approach improves Vision Transformer Dense Representations via Self-Distillation