ConvNN-Attention

Convolutional Nearest Neighbor Attention (ConvNN-Attention) for Transformers

TL;DR

Convolutional Nearest Neighbor Attention (ConvNN-Attention) is a novel attention mechanism that combines convolutional operations with nearest neighbor search to enhance the performance of Transformer models in both vision and language tasks.

Paper

The paper on Attention Via Convolutional Nearest Neighbors is available on arXiv: here.