MVR-cache raises semantic cache hit rates by up to 37% on benchmarks using learned prompt segmentation, multi-vector retrieval, and RL training while preserving correctness.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IR 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
MVR-cache: Optimizing Semantic Caching via Multi-Vector Retrieval and Learned Prompt Segmentation
MVR-cache raises semantic cache hit rates by up to 37% on benchmarks using learned prompt segmentation, multi-vector retrieval, and RL training while preserving correctness.