arXivMaryam Haghifam, Zifan He, Jason Cong, Yizhou SunSun, May 24, 2026, 1:22 AM PDT
score 16.1
Hierarchical memory helps transformers handle long documents faster
Original: H$^{2}$MT: Semantic Hierarchy-Aware Hierarchical Memory Transformer
Source: arxiv.org ↗
Writing ELI5 summary…