Scout Before You Attend: Sketch-and-Walk Sparse Attention for Efficient LLM Inference

Add code
Feb 07, 2026

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: