Addressing Uncertainty in LLMs to Enhance Reliability in Generative AI

Ramneet Kaur, Colin Samplawski, Adam D. Cobb, Anirban Roy, Brian Matejek, Manoj Acharya, Daniel Elenius, Alexander Michael Berenbeim, John A. Pavlik, Nathaniel D. Bastian, Susmit Jha

Keywords: LLMs, uncertainty quantification, conformal prediction, semantic equivalence, dynamic clustering

TL;DR: This paper takes a step towards enhancing reliability in generative AI by addressing uncertainty in LLMs on a given query. A dynamic semantic clustering algorithm is proposed for enhancing UQ, along with a novel non-conformity score in CP for LLMs.

Abstract: In this paper, we present a dynamic semantic clustering approach inspired by the Chinese Restaurant Process, aimed at addressing uncertainty in the inference of Large Language Models (LLMs). We quantify uncertainty of an LLM on a given query by calculating entropy of the generated semantic clusters. Further, we propose leveraging the (negative) likelihood of these clusters as the (non)conformity score within Conformal Prediction framework, allowing the model to predict a set of responses instead of a single output, thereby accounting for uncertainty in its predictions. We demonstrate the effectiveness of our uncertainty quantification (UQ) technique on two well-known question-answering benchmarks, COQA and TriviaQA, utilizing two LLMs—Llama-2-13b and Mistral-7b. Our approach achieves state-of-the-art (SOTA) performance in UQ, as assessed by metrics such as AUROC, AUARC, and AURAC. The proposed conformal predictor is also shown to produce smaller prediction sets while maintaining the same probabilistic guarantee of including the correct response, in comparison to existing SOTA conformal prediction baseline. Our code is publicly accessible at https://shorturl.at/7yHSq.

Submission Number: 117