Fine-grained Hallucination Detection and Editing for Language Models

2024/01/12 by Abhika Mishra, Akari Asai, Mishra, Abhika +11 · 23 citations
Computer Science · Psychology · #Topic Modeling #Mental Health via Writing #Machine Learning in Healthcare

paper · pdf · doi:10.48550/arxiv.2401.06855

Abstract

Large language models (LMs) are prone to generate factual errors, which are often called hallucinations. In this paper, we introduce a comprehensive taxonomy of hallucinations and argue that hallucinations manifest in diverse forms, each requiring varying degrees of careful assessments to verify factuality. We propose a novel task of automatic fine-grained hallucination detection and construct a new evaluation benchmark, FavaBench, that includes about one thousand fine-grained human judgments on three LM outputs across various domains. Our analysis reveals that ChatGPT and Llama2-Chat (70B, 7B) exhibit diverse types of hallucinations in the majority of their outputs in information-seeking scenarios. We train FAVA, a retrieval-augmented LM by carefully creating synthetic data to detect and correct fine-grained hallucinations. On our benchmark, our automatic and human evaluations show that FAVA significantly outperforms ChatGPT and GPT-4 on fine-grained hallucination detection, and edits suggested by FAVA improve the factuality of LM-generated text.

Cited by

• D-Score: A Spectral Hidden-State Signal for Hallucination Detection in Large Language Models
• A Unified Definition of Hallucination, Or: It's the World Model, Stupid
• AlignBench: Benchmarking Fine-Grained Image-Text Alignment with Synthetic Image-Caption Pairs
• SynClaimEval: A Framework for Evaluating the Utility of Synthetic Data in Long-Context Claim Verification
• Hallucination Benchmark for Speech Foundation Models
• Hallucination Localization in Video Captioning
• Mitigating Hallucination in Large Language Models (LLMs): An Application-Oriented Survey on RAG, Reasoning, and Agentic Systems
• HACK: Hallucinations Along Certainty and Knowledge Axes
• Confabulations from ACL Publications (CAP): A Dataset for Scientific Hallucination Detection
• HAD: HAllucination Detection Language Models Based on a Comprehensive Hallucination Taxonomy
• Can LLMs Correct Themselves? A Benchmark of Self-Correction in LLMs
• Detecting Hallucinations in Authentic LLM-Human Interactions
• VisRAG2.0: Mitigating Visual Hallucinations via Evidence-Guided Multi-Image Reasoning in Visual Retrieval-Augmented Generation
• When Models Lie, We Learn: Multilingual Span-Level Hallucination Detection with PsiloQA
• Large Language Models Hallucination: A Comprehensive Survey
• ReFACT: A Benchmark for Scientific Confabulation Detection with Positional Error Annotations
• Fine-Grained Detection of Context-Grounded Hallucinations Using LLMs
• Decomposing and Revising What Language Models Generate
• Synthesizing scientific literature with retrieval-augmented language models
• Investigating Hallucination in Conversations for Low Resource Languages
• FRED: Financial Retrieval-Enhanced Detection and Editing of Hallucinations in Language Models
• First Hallucination Tokens Are Different from Conditional Ones
• Hallucination Detection and Mitigation with Diffusion in Multi-Variate Time-Series Foundation Models

Related

• Large Language Models Hallucination: A Comprehensive Survey
• HaluEval: A Large-Scale Hallucination Evaluation Benchmark for Large Language Models
• The Dawn After the Dark: An Empirical Study on Factuality Hallucination in Large Language Models
• PFME: A Modular Approach for Fine-grained Hallucination Detection and Editing of Large Language Models
• 🧜Siren’s Song in the AI Ocean: A Survey on Hallucination in Large Language Models