CacheTrap achieves 100% targeted attack success on five open-source LLMs by using an efficient search to locate and flip a single bit in the KV cache as a transient trigger, while preserving normal accuracy without the trigger.
Sbfa: Single sneaky bit flip attack to break large language models
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
representative citing papers
A data-parameter correspondence unifies data-centric and parameter-centric LLM optimizations as dual geometric operations on the statistical manifold via Fisher-Rao metric and Legendre duality.
citing papers explorer
-
CacheTrap: Unveiling a Stealthier Gray-Box Trojan against LLMs
CacheTrap achieves 100% targeted attack success on five open-source LLMs by using an efficient search to locate and flip a single bit in the KV cache as a transient trigger, while preserving normal accuracy without the trigger.
-
Towards a Data-Parameter Correspondence for LLMs: A Preliminary Discussion
A data-parameter correspondence unifies data-centric and parameter-centric LLM optimizations as dual geometric operations on the statistical manifold via Fisher-Rao metric and Legendre duality.