AI News

Prism Attention: coarse-to-fine index sharing to trade per-layer accuracy for faster long-context decoding

ABC2 min read
Research