Pith. sign in

REVIEW 3 cited by

DeepEdit: Knowledge Editing as Decoding with Constraints

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2401.10471 v5 pith:T2CKTZTN submitted 2024-01-19 cs.CL cs.AI

classification cs.CLcs.AI
keywords knowledgereasoningllmsdeepeditconstraintsdecodingeditingbenchmarks
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

How to edit the knowledge in multi-step reasoning has become the major challenge in the knowledge editing (KE) of large language models (LLMs). The difficulty arises because the hallucinations of LLMs during multi-step reasoning often lead to incorrect use of new knowledge and incorrect answers. To address this issue, we design decoding constraints to "regulate" LLMs' reasoning, enhancing logical coherence when incorporating new knowledge. We propose a new KE framework: DEEPEDIT (Depth-first Search-based Constrained Decoding for Knowledge Editing), which enhances LLMs's ability to generate coherent reasoning chains with new knowledge through 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.

Discussion (0). Sign in to comment.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. ArtAnno: Annotating Implicit Semantics in Artworks through LLM Agent-Driven Bidirectional Human-AI Augmentation

    cs.HC 2026-08 conditional novelty 6.0 of 10

    An LLM-agent artwork annotation system that combines proactive label suggestions with interaction-driven skill learning reported roughly 50% faster annotation and higher label agreement in a 12-participant study.

  2. Decoupling Reasoning and Knowledge Injection for In-Context Knowledge Editing

    cs.CL 2025-05 conditional novelty 6.0 of 10

    DecKER decouples reasoning from knowledge editing by planning with masked placeholders before retrieving edited facts, improving multi-hop QA accuracy after knowledge edits.

  3. Robust Knowledge Editing via Explicit Reasoning Chains for Distractor-Resilient Multi-Hop QA

    cs.CL 2025-09 conditional novelty 5.0 of 10

    Teaching an LLM to emit a fixed four-stage reasoning chain during fine-tuning makes single-pass multi-hop knowledge editing robust to distractor facts.

Pith tools