Saved in:
Bibliographic Details
Main Authors: Zhang, Caiqi, Liu, Fangyu, Basaldella, Marco, Collier, Nigel
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2403.20279
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912057797902336
author Zhang, Caiqi
Liu, Fangyu
Basaldella, Marco
Collier, Nigel
author_facet Zhang, Caiqi
Liu, Fangyu
Basaldella, Marco
Collier, Nigel
contents Large Language Models (LLMs) have demonstrated remarkable capability in a variety of NLP tasks. However, LLMs are also prone to generate nonfactual content. Uncertainty Quantification (UQ) is pivotal in enhancing our understanding of a model's confidence on its generation, thereby aiding in the mitigation of nonfactual outputs. Existing research on UQ predominantly targets short text generation, typically yielding brief, word-limited responses. However, real-world applications frequently necessitate much longer responses. Our study first highlights the limitations of current UQ methods in handling long text generation. We then introduce \textsc{Luq} and its two variations, a series of novel sampling-based UQ approaches specifically designed for long text. Our findings reveal that \textsc{Luq} outperforms existing baseline methods in correlating with the model's factuality scores (negative coefficient of -0.85 observed for Gemini Pro). To further improve the factuality of LLM responses, we propose \textsc{Luq-Ensemble}, a method that ensembles responses from multiple models and selects the response with the lowest uncertainty. The ensembling method greatly improves the response factuality upon the best standalone LLM.
format Preprint
id arxiv_https___arxiv_org_abs_2403_20279
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle LUQ: Long-text Uncertainty Quantification for LLMs
Zhang, Caiqi
Liu, Fangyu
Basaldella, Marco
Collier, Nigel
Computation and Language
Large Language Models (LLMs) have demonstrated remarkable capability in a variety of NLP tasks. However, LLMs are also prone to generate nonfactual content. Uncertainty Quantification (UQ) is pivotal in enhancing our understanding of a model's confidence on its generation, thereby aiding in the mitigation of nonfactual outputs. Existing research on UQ predominantly targets short text generation, typically yielding brief, word-limited responses. However, real-world applications frequently necessitate much longer responses. Our study first highlights the limitations of current UQ methods in handling long text generation. We then introduce \textsc{Luq} and its two variations, a series of novel sampling-based UQ approaches specifically designed for long text. Our findings reveal that \textsc{Luq} outperforms existing baseline methods in correlating with the model's factuality scores (negative coefficient of -0.85 observed for Gemini Pro). To further improve the factuality of LLM responses, we propose \textsc{Luq-Ensemble}, a method that ensembles responses from multiple models and selects the response with the lowest uncertainty. The ensembling method greatly improves the response factuality upon the best standalone LLM.
title LUQ: Long-text Uncertainty Quantification for LLMs
topic Computation and Language
url https://arxiv.org/abs/2403.20279