Tiled Flash Linear Attention: More Efficient Linear RNN and xLSTM Kernels

Open in new window