Fetching the paper…

SwiftFormer: Efficient Additive Attention for Transformer-based Real-time Mobile Vision Applications · Around