local-attention
LEAP: Linear Explainable Attention in Parallel for causal language modeling with O(1) path length, and O(1) inference
Implementation of LA_MIL, Local Attention Graph-based Transformer for WSIs, PyTorch
