{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/detecting-instruction-fine-tuning-attack-on","title":"Detecting Instruction Fine-tuning Attack on Language Models with Influence Function","arxiv_id":"2504.09026","date":"2025-04-12","proceeding":null,"authors":["Jiawei Li"],"abstract":"Instruction fine-tuning attacks pose a significant threat to large language models (LLMs) by subtly embedding poisoned data in fine-tuning datasets, which can trigger harmful or unintended responses across a range of tasks. This undermines model alignment and poses security risks in real-world deployment. In this work, we present a simple and effective approach to detect and mitigate such attacks using influence functions, a classical statistical tool adapted for machine learning interpretation. Traditionally, the high computational costs of influence functions have limited their application to large models and datasets. The recent Eigenvalue-Corrected Kronecker-Factored Approximate Curvature (EK-FAC) approximation method enables efficient influence score computation, making it feasible for large-scale analysis. We are the first to apply influence functions for detecting language model instruction fine-tuning attacks on large-scale datasets, as both the instruction fine-tuning attack on language models and the influence calculation approximation technique are relatively new. Our large-scale empirical evaluation of influence functions on 50,000 fine-tuning examples and 32 tasks reveals a strong association between influence scores and sentiment. Building on this, we introduce a novel sentiment transformation combined with influence functions to detect and remove critical poisons -- poisoned data points that skew model predictions. Removing these poisons (only 1% of total data) recovers model performance to near-clean levels, demonstrating the effectiveness and efficiency of our approach. Artifact is available at https://github.com/lijiawei20161002/Poison-Detection. WARNING: This paper contains offensive data examples.","url_abs":"https://arxiv.org/abs/2504.09026v1","url_pdf":"https://arxiv.org/pdf/2504.09026v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"detecting-instruction-fine-tuning-attack-on","repo_url":"https://github.com/lijiawei20161002/poison-detection","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"jax","reach":null}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/2504.09026","atlas_url":"https://app.syntology.ai/?focus=2504.09026","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}