Clustered Self-Assessment: A Simple yet Effective Method for Uncertainty Quantification in Large Language Models
This work addresses the need for reliable uncertainty estimates in LLM outputs, which is critical for users to judge response reliability, but the method is incremental as it combines existing clustering and self-assessment ideas.
The paper proposes a self-assessment method for uncertainty quantification in LLMs that clusters sampled generations into semantically distinct groups, converts them into multiple-choice options, and uses the LLM's assigned probabilities as confidence estimates. The method consistently outperforms baselines across multiple models and datasets, achieving competitive performance with as few as two additional samples.
Large language models (LLMs) demonstrate remarkable performance across diverse tasks, but they often generate responses that appear plausible while being factually incorrect. This problem is compounded by the lack of explicit uncertainty estimates, which makes it difficult for users to judge the reliability of model outputs. Existing uncertainty quantification methods typically rely on indirect signals, such as entropy across sampled generations. These signals can be difficult to interpret and do not fully leverage the model's ability to assess its own uncertainty. We propose a simple yet effective self-assessment method for uncertainty quantification in LLMs. Our approach groups sampled generations into semantically distinct clusters, converts them into answer options in a structured multiple-choice question, and uses the probability assigned by the LLM to each option as a confidence estimate. Experiments across multiple models and datasets show that our method consistently outperforms baseline approaches. Notably, it achieves competitive performance with as few as two additional samples, demonstrating both its effectiveness and efficiency.