attention-mining

Tag

Cards List
#attention-mining

REVA: Reusable Evidence View Aggregation for Context-Efficient RAG Serving

arXiv cs.LG · 2026-09-11 Cached

REVA is a framework that mines historical attention traces to create reusable evidence views for context-efficient RAG serving, improving generation quality while reducing compression overhead and latency.

0 favorites 0 likes
← Back to home

Submit Feedback