ACL Findings2026
FABLE: Fine-grained Fact Anchoring for Unstructured Model Editing
Abstract
Unstructured model editing aims to update models with real-world text, yet existing methods often memorize text holistically without reliable fine-grained fact access. To address this, we propose FABLE, a hierarchical framework that decouples fine-grained fact injection from holistic text generation. FABLE follows a two-stage, fact-first strategy: discrete facts are anchored in shallow layers, followed by minimal updates to deeper layers to produce coherent text. This decoupling resolves the mismatch between holistic recall and fine-grained fact access, reflecting the unidirectional Transformer flow in which surfaceform generation amplifies rather than corrects underlying fact representations. We also introduce UnFine, a diagnostic benchmark with fine-grained question-answer pairs and factlevel metrics for systematic evaluation. Experiments show that FABLE substantially improves fine-grained question answering while maintaining state-of-the-art holistic editing performance. Our code is publicly available at https://github.com/caskcsg/FABLE.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
- Aging with GRACE: Lifelong Model Editing with Discrete Key-Value Adaptors cites Locating and Editing Factual Associations in GPT
- Aging with GRACE: Lifelong Model Editing with Discrete Key-Value Adaptors cites Language Models are Few-Shot Learners
- Locating and Editing Factual Associations in GPT cites Language Models are Few-Shot Learners
- Measuring Massive Multitask Language Understanding cites Language Models are Few-Shot Learners
Builds on24
- Language Models are Few-Shot LearnersTom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah et al.NeurIPS 2020 · 64,255 citations
- BERTScore: Evaluating Text Generation with BERTTianyi Zhang, Varsha Kishore, Felix Wu, Kilian Q. Weinberger et al.ICLR 2020 · 8,443 citations
- Measuring Massive Multitask Language UnderstandingDan Hendrycks, Collin Burns, Steven Basart, Andy Zou et al.ICLR 2021 · 7,905 citations
- Locating and Editing Factual Associations in GPTKevin Meng, David Bau, Alex Andonian, Yonatan BelinkovNeurIPS 2022 · 3,415 citations
- Aging with GRACE: Lifelong Model Editing with Discrete Key-Value AdaptorsTom Hartvigsen, Swami Sankaranarayanan, Hamid Palangi, Yoon Kim et al.NeurIPS 2023 · 349 citations
Related papers
- StepKE: Stepwise Knowledge Editing for Multi-Hop Question AnsweringJaewook Lee, Dahyun Jung, Heuiseok LimEMNLP Findings 2025
- Avoiding Knowledge Edit Skipping in Multi-hop Question Answering with Guided DecompositionYi Liu, Xiangrong Zhu, Xiangyu Liu, Wei Wei et al.EMNLP Findings 2025
- Locate-then-edit for Multi-hop Factual Recall under Knowledge EditingZhuoran Zhang, Yongxiang Li, Zijian Kan, Keyuan Cheng et al.ICML 2025
- Better Call SAUL: Fluent and Consistent Language Model Editing with Generation RegularizationMingyang Wang, Lukas Lange, Heike Adel, Jannik Strötgen et al.EMNLP Findings 2024
- MULFE: A Multi-Level Benchmark for Free Text Model EditingChenhao Wang, Pengfei Cao, Zhuoran Jin, Yubo Chen et al.ACL 2024
