3 ms·Constructing Transformers for Longer Sequences with Sparse Attention Methods1 points by theafh 6y ago