1 paper
Nhat Thanh Tran, Jack Xin
We study a fast local-global window-based attention method to accelerate Informer for long sequence time-series forecasting. While window attention being local is a considerable co…