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

Language models (LMs) now excel at many tasks such as few-shot learning, 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-...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2023-05
Main Authors: Gao, Luyu, Dai, Zhuyun, Pasupat, Panupong, Chen, Anthony, Arun Tejasvi Chaganty, Fan, Yicheng, Zhao, Vincent Y, Lao, Ni, Lee, Hongrae, Da-Cheng, Juan, Guu, Kelvin
Format: Article
Language:English
Subjects:
Retrofitting
Revisions
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Language models (LMs) now excel at many tasks such as few-shot learning, 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.
ISSN:2331-8422