DynamicViT: Efficient Vision Transformers with Dynamic Token Sparsification
Primary LanguageJupyter NotebookMIT LicenseMIT