MAD-HD: Multi-Agent Debate-Driven Ungrounded Hallucination Detection for Large Language Models

We are working on releasing the code and datasets

Introduction

Python implementation of our paper: [MAD-HD: Multi-Agent Debate-Driven Ungrounded Hallucination Detection for Large Language Models].

We propose a Multi-Agent Debate framework based on Qwen2-72b-Instruct to effectively detect hallucinations.

MAD-HD

The code for the paper 'MAD-HD: Multi-Agent Debate-Driven Ungrounded Hallucination Detection for Large Language Models.'