Factored Verification: Detecting and Reducing Hallucination in Summaries of Academic Papers

Charlie George, Andreas Stuhlmüller
10/16/2023
cs.CLcs.AI

No official code found. Best implementation: elicit/fave-dataset (13 stars, MIT, last commit 2024-10-20).

Abstract

Hallucination plagues even frontier LLMs--but how bad is it really for summarizing academic papers? We evaluate Factored Verification, a simple automated method for detecting hallucinations in abstractive summaries. This method sets a new SotA on hallucination detection in the summarization task of the HaluEval benchmark, achieving 76.2% accuracy. We then use this method to estimate how often language models hallucinate when summarizing across multiple academic papers and find 0.62 hallucinations in the average ChatGPT (16k) summary, 0.84 for GPT-4, and 1.55 for Claude 2. We ask models to self-correct using Factored Critiques and find that this lowers the number of hallucinations to 0.49 for ChatGPT, 0.46 for GPT-4, and 0.95 for Claude 2. The hallucinations we find are often subtle, so we advise caution when using models to synthesize academic papers.

View on arXivView PDF

Code Implementations(1)

elicit/fave-datasetCommunity100%
MIT

Paper dataset for "Factored Verification: Detecting and Reducing Hallucination in Summaries of Academic Papers"

133PythonOct 15, 20232 years agoMIT

Similar Papers

HaluEval: A Large-Scale Hallucination Evaluation Benchmark for Large Language Models

87% match1 repo5952023

Evaluating Hallucinations in Chinese Large Language Models

87% match1 repo1392023

Improving Factual Consistency of News Summarization by Contrastive Preference Optimization

87% match1 repo72023

Metric Ensembles For Hallucination Detection

87% match2023

From Single to Multi: How LLMs Hallucinate in Multi-Document Summarization

87% match1 repo72024

A Stitch in Time Saves Nine: Detecting and Mitigating Hallucinations of LLMs by Validating Low-Confidence Generation

86% match2023

Cite this paper

@article{george2023factored,
  title  = {Factored Verification: Detecting and Reducing Hallucination in Summaries of Academic Papers},
  author = {Charlie George and Andreas Stuhlmüller},
  year   = {2023},
  eprint = {2310.10627},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2310.10627v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2310.10627.svg)](https://codeofpaper.com/paper/2310.10627)

Discussion