Assessing the Reliability of Large Language Model Knowledge
This addresses the issue of LLM hallucination for researchers and practitioners by providing a tool to assess model reliability, though it is incremental as it builds on existing evaluation methods.
The paper tackles the problem of evaluating the factual reliability of large language models (LLMs) beyond accuracy, proposing MONITOR as a metric to measure consistency across different prompts and contexts, with experiments on 12 LLMs showing effectiveness and low computational overhead.
Large language models (LLMs) have been treated as knowledge bases due to their strong performance in knowledge probing tasks. LLMs are typically evaluated using accuracy, yet this metric does not capture the vulnerability of LLMs to hallucination-inducing factors like prompt and context variability. How do we evaluate the capabilities of LLMs to consistently produce factually correct answers? In this paper, we propose MOdel kNowledge relIabiliTy scORe (MONITOR), a novel metric designed to directly measure LLMs' factual reliability. MONITOR computes the distance between the probability distributions of a valid output and its counterparts produced by the same LLM probing the same fact using different styles of prompts and contexts.Experiments on a comprehensive range of 12 LLMs demonstrate the effectiveness of MONITOR in evaluating the factual reliability of LLMs while maintaining a low computational overhead. In addition, we release the FKTC (Factual Knowledge Test Corpus) test set, containing 210,158 prompts in total to foster research along this line (https://github.com/Vicky-Wil/MONITOR).