BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//cfp.pydata.org//pydataglobal2025//speaker//ANPNZZ
BEGIN:VEVENT
SUMMARY:UQLM: Detecting LLM Hallucinations with Uncertainty Quantification
  in Python - Dylan Bouchard\, Mohit Singh Chauhan
DTSTART:20251210T183000Z
DTEND:20251210T190000Z
DTSTAMP:20261006T024228Z
UID:pretalx-pydataglobal2025-8U7WLS@cfp.pydata.org
DESCRIPTION:As LLMs become increasingly embedded in critical applications 
 across healthcare\, legal\, and financial domains\, their tendency to gene
 rate plausible-sounding but false information poses significant risks. Thi
 s talk introduces UQLM\, an open-source Python package for uncertainty-awa
 re generation that flags likely hallucinations without requiring ground tr
 uth data. UQLM computes response-level confidence scores from token probab
 ilities\, consistency across sampled responses\, LLM judges\, and tunable 
 ensembles. Attendees will learn practical strategies for implementing hall
 ucination detection in production systems and leave with code examples the
 y can immediately apply to improve the reliability of their LLM-powered ap
 plications. No prior uncertainty quantification background required.
LOCATION:Machine Learning & AI
URL:https://cfp.pydata.org/pydataglobal2025/talk/8U7WLS/
END:VEVENT
END:VCALENDAR