SCoNE: Selective Context-aware Neuron Editing for Robust Retrieval-Augmented Generation
Mirrored from arXiv — NLP / Computation & Language for archival readability. Support the source by reading on the original site.
Computer Science > Computation and Language
Title:SCoNE: Selective Context-aware Neuron Editing for Robust Retrieval-Augmented Generation
Abstract:Retrieval-Augmented Generation (RAG) is highly sensitive to retrieval noise: when retrieved documents mix informative and irrelevant context, LLMs are easily distracted, leading to hallucinations. To overcome this, we propose SCoNE (Selective Context-aware Neuron Editing), a training-free model editing approach that improves retrieval noise robustness by selectively strengthening context-aware FFN neurons that are identified by both high attribution and high cross-input variability. SCoNE requires only a small number of mining samples, no fine-tuning, and no inference-time overhead. Across various knowledge-intensive question-answering benchmarks and two LLM backbones, SCoNE consistently outperforms competitive baseline methods. Our code is available at this https URL.
| Comments: | Accepted to EMNLP 2026 Main Conference |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2609.00689 [cs.CL] |
| (or arXiv:2609.00689v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2609.00689
arXiv-issued DOI via DataCite (pending registration)
|
Access Paper:
- View PDF
- HTML (experimental)
- TeX Source
References & Citations
Bookmark
Bibliographic and Citation Tools
Code, Data and Media Associated with this Article
Demos
Recommenders and Search Tools
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
More from arXiv — NLP / Computation & Language
-
Evaluating Large Language Models for Forced Outage Risk Prediction: Benefits and Comparison to Machine Learning
Sep 7
-
SharedSAE: One Feature Dictionary Across Language Models
Sep 7
-
Conformity Breaks Conformal Prediction
Sep 7
-
When Load-Balancing Goes Too Far: Expert Pruning in Over-Dispersed Mixture-of-Experts Models
Sep 7
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.