Fetching the paper…

DearKD: Data-Efficient Early Knowledge Distillation for Vision Transformers · Around