A semantic cache for LLMs that verifies the first k response tokens on cache hits reduces cache-collision attack hit rate to near zero while preserving benign cache utility.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
LaCache: Robust Semantic Caching for LLM Serving
A semantic cache for LLMs that verifies the first k response tokens on cache hits reduces cache-collision attack hit rate to near zero while preserving benign cache utility.