English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Despite the significant progress made in practical applications of alignedlanguage models (LMs), they tend to be overconfident in output answers comparedto the corresponding pre-trained LMs. In this work, we systematically evaluatethe impact of the alignment process on logit-based uncertainty calibration ofLMs under the multiple-choice setting. We first conduct a thoughtful empiricalstudy on how aligned LMs differ in calibration from their pre-trainedcounterparts. Experimental results reveal that there are two distinctuncertainties in LMs under the multiple-choice setting, which are responsiblefor the answer decision and the format preference of the LMs, respectively.Then, we investigate the role of these two uncertainties on aligned LM'scalibration through fine-tuning in simple synthetic alignment schemes andconclude that one reason for aligned LMs' overconfidence is the conflation ofthese two types of uncertainty. Furthermore, we examine the utility of commonpost-hoc calibration methods for aligned LMs and propose an easy-to-implementand sample-efficient method to calibrate aligned LMs. We hope our findingscould provide insights into the design of more reliable alignment processes forLMs.

Investigating Uncertainty Calibration of Aligned Language Models under the Multiple-Choice Setting