Introduces NPAS and AV Filter using LLM attention weights to defend RAG against poisoning, reporting up to 20% accuracy gains while adaptive attacks reach 35% success.
Zipcache: Accurate and efficient kv cache quantization with salient token identification
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
representative citing papers
citing papers explorer
-
Through the Stealth Lens: Attention-Aware Defenses Against Poisoning in RAG
Introduces NPAS and AV Filter using LLM attention weights to defend RAG against poisoning, reporting up to 20% accuracy gains while adaptive attacks reach 35% success.
- Meta-Soft: Leveraging Composable Meta-Tokens for Context-Preserving KV Cache Compression
- SPHERICAL KV: Angle-Domain Attention and Rate-Distortion Retention for Efficient Long-Context Inference