Show HN: UQLM – Closed-book hallucination detection with UQ

Category: library

Tags: hallucination-detection, uncertainty-quantification, llm

Score: 7.8/10 (Innovation: 7, Technical: 8, Documentation: 9, Utility: 7)

UQLM is a Python library for detecting hallucinations in large language model outputs using state-of-the-art uncertainty quantification techniques, including black-box, white-box, and LLM-as-a-judge scorers. It offers a comprehensive, modular framework that combines multiple research-backed methods into a single, easy-to-use API, making it a practical tool for improving LLM reliability in production settings.

Target audience: backend devs, data engineers, ai-ml researchers

Repository: https://github.com/cvs-health/uqlm · Python · Apache-2.0 · 1160 stars

View on Hacker News