Mixa-Q: Revisiting Activation Sparsity for Vision Transformers From a Mixed-Precision Quantization Perspective.

Weitian Wang, Shubham Rai, Cecilia De la Parra, Akash Kumar

Browse the full ICCV paper archive.