Are LLMs Aware that Some Questions are not Open-ended?

TL;DR

This paper investigates whether Large Language Models (LLMs) recognize the nature of questions they face and proposes a method to improve their question awareness, reducing hallucinations and enhancing response quality.

Contribution

The paper introduces Question Awareness Temperature Sampling (QuATS), a novel technique that adaptively adjusts output distributions to improve LLMs' question awareness without manual tuning.

Findings

LLMs often lack awareness of question types, leading to hallucinations.

QuATS improves LLM performance across multiple benchmarks.

Automatic temperature adjustment enhances response appropriateness.

Abstract

Large Language Models (LLMs) have shown the impressive capability of answering questions in a wide range of scenarios. However, when LLMs face different types of questions, it is worth exploring whether LLMs are aware that some questions have limited answers and need to respond more deterministically but some do not. We refer to this as question awareness of LLMs. The lack of question awareness in LLMs leads to two phenomena that LLMs are: (1) too casual to answer non-open-ended questions or (2) too boring to answer open-ended questions. In this paper, we first evaluate the question awareness in LLMs. The experimental results show that LLMs have the issues of lacking awareness of questions in certain domains, e.g. factual knowledge, resulting in hallucinations during the generation. To mitigate these, we propose a method called Question Awareness Temperature Sampling (QuATS). This method enhances the question awareness of LLMs by adaptively adjusting the output distributions based on question features. The automatic adjustment in QuATS eliminates the need for manual temperature tuning in text generation and consistently improves model performance in various benchmarks.