Memory-Efficient Vision Transformers: An Activation-Aware Mixed-Rank Compression Strategy.

Seyedarmin Azizi, Mahdi Nazemi, Massoud Pedram

Browse the full ECCV paper archive.