MonarchAttention: Zero-Shot Conversion to Fast, Hardware-Aware Structured Attention

NeurIPSSpotlight2025

Authors
Can Yaras, Alec S Xu, Pierre Abillama, Changwoo Lee, Laura Balzano
Venue
NeurIPS 2025
Track
Spotlight

TL;DR

Transformers have achieved state-of-the-art performance across various tasks, but suffer from a notable quadratic complexity in sequence length due to the attention mechanism…

Opening excerpt from the authors’ abstract. source

Read the paper

Topics

transformer attention zero-shot

← All NeurIPS 2025 Spotlight papers · Browse the whole archive