Query Information
Query information research focuses on improving how information is retrieved and utilized, particularly within complex systems like large language models (LLMs) and databases. Current research emphasizes enhancing query processing through techniques like query rewriting, debiasing, and augmentation using LLMs and various attention mechanisms (e.g., grouped query attention), often incorporating multi-modal data and feedback mechanisms for improved accuracy and efficiency. This work is significant for advancing information retrieval across diverse applications, including question answering, video moment retrieval, and even enhancing the security and reliability of LLMs themselves.
Papers
WordGame: Efficient & Effective LLM Jailbreak via Simultaneous Obfuscation in Query and Response
Tianrong Zhang, Bochuan Cao, Yuanpu Cao, Lu Lin, Prasenjit Mitra, Jinghui Chen
Just rephrase it! Uncertainty estimation in closed-source language models via multiple rephrased queries
Adam Yang, Chen Chen, Konstantinos Pitas