Built independently by an author, for readers. Read the story and support ChapterPal
ChapterPal Logo
Log inSign up

keyword

Selective QA

Selective question answering is a natural language processing framework in which a model evaluates whether to provide an answer to a given query or abstain from responding based on its estimated confidence or uncertainty. Instead of attempting to answer every prompt unconditionally, a system operating under this paradigm uses uncertainty quantification or confidence estimation to withhold responses when the risk of generating an incorrect or nonfactual output is high. By filtering out low-confidence predictions, selective question answering improves the overall accuracy and reliability of the delivered answers, effectively trading total question coverage for higher precision and safety.

1 item

LUQ: Long-text Uncertainty Quantification for LLMs

LUQ: Long-text Uncertainty Quantification for LLMs

Caiqi Zhang, Fangyu Liu, Marco Basaldella, Nigel Collier

Why you should read this

Introduces LUQ, a sampling-based uncertainty quantification framework designed for long-form language model outputs, demonstrating strong negative correlation with factual errors and enabling multi-model ensembles that boost overall response factuality.

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 LUQ with its two variations: LUQ-ATOMIC and LUQ-PAIR, a series of novel sampling-based UQ approaches specifically designed for long text. Our findings reveal that 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 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.

Added

2026-10-02