Summary of Genaudit: Fixing Factual Errors in Language Model Outputs with Evidence, by Kundan Krishna et al.
GenAudit: Fixing Factual Errors in Language Model Outputs with Evidence
by Kundan Krishna, Sanjana Ramprasad, Prakhar Gupta, Byron C. Wallace, Zachary C. Lipton, Jeffrey P. Bigham
First submitted to arxiv on: 19 Feb 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper presents a novel tool, GenAudit, designed to assist fact-checking large language model (LLM) responses for document-grounded tasks. GenAudit suggests edits to the LLM response by revising or removing claims not supported by the reference document and provides evidence from the reference for facts that do appear to have support. The tool is trained on various models and interfaces, aiming to improve human performance in detecting errors in LLM-generated summaries. Comprehensive evaluations and user studies demonstrate GenAudit’s effectiveness in reducing errors in diverse domains. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary GenAudit helps fix mistakes made by language models when summarizing documents. This tool checks if what the model says is true or not, using the original document as proof. It even suggests changes to make the summary more accurate. In tests with many different LLMs and types of documents, GenAudit did a great job detecting errors in their summaries. Using this tool can also help humans find mistakes made by language models. |
Keywords
* Artificial intelligence * Large language model
