Linear Attention for Efficient Bidirectional Sequence Modeling | Read Paper on Bytez