ResearchTrend.AI
  • Papers
  • Communities
  • Events
  • Blog
  • Pricing
Papers
Communities
Social Events
Terms and Conditions
Pricing
Parameter LabParameter LabTwitterGitHubLinkedInBlueskyYoutube

© 2025 ResearchTrend.AI, All rights reserved.

  1. Home
  2. Papers
  3. 2401.10471
77
28
v1v2v3v4 (latest)

DeepEdit: Knowledge Editing as Decoding with Constraints

19 January 2024
Yiwei Wang
Muhao Chen
Nanyun Peng
Kai-Wei Chang
    KELM
ArXiv (abs)PDFHTML
Abstract

Answering multi-hop questions involving new knowledge is a challenging task in evaluating large language models' (LLMs) knowledge editing (KE) methods. This task is rather difficult because the LLMs' hallucinations on new knowledge would harm the logical coherence of LLMs' multi-hop reasoning and lead to incorrect answers. To address this issue, we design decoding constraints to "regulate" LLMs' reasoning, enhancing logical coherence when incorporating new knowledge. We incorporate the constraints into a new KE framework: DEEPEDIT (Depth-first Search-based Constrained Decoding for Knowledge Editing), which enhances LLMs to generate coherent reasoning chains with new knowledge through a depth-first search. Our search selects the most important knowledge that satisfies our constraints as the reasoning step to efficiently increase the reasoning depth. In addition to DEEPEDIT, we propose two new KE benchmarks: MQUAKE-2002 and MQUAKE-HARD, which provide more precise and challenging assessments of KE approaches. Qualitatively, DEEPEDIT enables LLMs to produce succinct and coherent reasoning chains involving new knowledge. Quantitatively, it yields significant improvements on multiple KE benchmarks.

View on arXiv
Comments on this paper