Fetching the paper…

H-Transformer-1D: Fast One-Dimensional Hierarchical Attention for Sequences · Around