RARR: Researching and Revising What Language Models Say, Using Language Models

By Luyu Gao, Zhuyun Dai, Panupong Pasupat, Anthony Chen, Arun Tejasvi Chaganty, Yicheng Fan, Vincent Y. Zhao, Ni Lao, Hongrae Lee, Da-Cheng Juan and 1 more. Annual Meeting of the Association for Computational Linguistics, 2022

339 citations recorded by the source corpus.

Language models (LMs) now excel at many tasks such as question answering, reasoning, and dialog. However, they sometimes generate unsupported or misleading content. A user cannot easily determine whether their outputs are trustworthy or not, because most LMs do not have any built-in mechanism for attribution to external evidence. To enable attribution while still preserving all the powerful advantages of recent generation models, we propose RARR (Retrofit Attribution using Research and Revision), a system that 1) automatically finds attribution for the output of any text generation model, and 2) post-edits the output to fix unsupported content while preserving the original output as much as possible. When applied to the output of several state-of-the-art LMs on a diverse set of generation tasks, we find that RARR significantly improves attribution while otherwise preserving the original input to a much greater degree than previously explored edit models. Furthermore, the implementation of RARR requires only a handful of training examples, a large language model, and standard web search.

Subfields: Artificial Intelligence

View Paper: https://www.semanticscholar.org/paper/66242baf48b0f6b828e7547ac39ffaa5e1b2cb3e

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

Chain-of-Verification Reduces Hallucination in Large Language Models (2023, methodology): “Other approaches include using tools for fact-checking (Chern et al., 2023a; Galitsky, 2023; Peng et al., 2023), or linking to external documents with attribution (Menick et al., 2022; Rashkin et al., 2023; Gao et al., 2023).”
A Comprehensive Survey of Hallucination Mitigation Techniques in Large Language Models (2024, methodology): “…employ the information retrieval system after generating the entirety of its output: Retrofit Attribution using Research and Revision (RARR): (Gao et al., 2023) In the realm of LLMs, notable advancements have been achieved across various tasks; however, issues persist, such as generating…”
Knowledge Conflicts for LLMs: A Survey (2024, background): “Mean-while, LLMs continue to engage with external contextual knowledge after deployed (Pan et al., 2022), including user prompts (Liu et al., 2023a), documents from the Web (Shi et al., 2023c), or tools (Schick et al., 2023; Zhuang et al., 2023).”
A Survey of Hallucination in Large Foundation Models (2023, methodology): “RARR: Researching and Revising What Language Models Say, Using Language Models (Gao et al., 2023) Editing for Attribution NQ, SQA, QReCC Attributable to Identified Sources (Castaldo and Yang, 2007)”
RAGTruth: A Hallucination Corpus for Developing Trustworthy Retrieval-Augmented Language Models (2023, methodology): “It contains context, response generated by LLM with and span-level annotation. tual or accurate information (Rawte et al., 2023).”
DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelines (2023, background): “The formatting of these example prompts is adapted from Gao et al. (2023a).”
A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions (2023, neutral): “To enhance the trustworthiness and attribution of LLMs, Gao et al. (2023a) adopt the research-then-revise workflow, which initially research relevant evidence and subsequently revise the initial generation based on detected discrepancies with the evidence.”
ExpertQA: Expert-Curated Questions and Attributed Answers (2023, methodology): “…of attributable to identified sources (AIS) (Rashkin et al., 2021), i.e. judging whether model generated content can be verified against given attributions, previous work has found NLI models to be effective in providing automated AIS (AutoAIS) estimations (Gao et al., 2023; Bohnet et al., 2022).”
Enabling Large Language Models to Generate Text with Citations (2023, methodology): “TRUE targets factual correctness and has been used by previous works in similar context (Bohnet et al., 2022; Gao et al., 2022).”