Image de couverture de livre

Merging Triggers, Breaking Backdoors: Defensive Poisoning for Instruction-Tuned Language Models

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Kim, San, Lee, Gary Geunbae
Format:	Preprint
Publié:	2026
Sujets:	Computation and Language Artificial Intelligence
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

Safeguarding RAG Pipelines with GMTP: A Gradient-based Masked Token Probability Method for Poisoned Document Detection
par: Kim, San, et autres
Publié: (2025)

Adversarial DPO: Harnessing Harmful Data for Reducing Toxicity with Minimal Impact on Coherence and Evasiveness in Dialogue Agents
par: Kim, San, et autres
Publié: (2024)

Ontology-Free General-Domain Knowledge Graph-to-Text Generation Dataset Synthesis using Large Language Model
par: Kim, Daehee, et autres
Publié: (2024)

Mixture-of-Experts with Intermediate CTC Supervision for Accented Speech Recognition
par: Lee, Wonjun, et autres
Publié: (2026)

Key-Element-Informed sLLM Tuning for Document Summarization
par: Ryu, Sangwon, et autres
Publié: (2024)

Instructions as Backdoors: Backdoor Vulnerabilities of Instruction Tuning for Large Language Models
par: Xu, Jiashu, et autres
Publié: (2023)

Learning When to Translate for Multilingual Reasoning
par: Kang, Deokhyung, et autres
Publié: (2026)

Autoregressive Score Generation for Multi-trait Essay Scoring
par: Do, Heejin, et autres
Publié: (2024)

K-COMP: Retrieval-Augmented Medical Domain Question Answering With Knowledge-Injected Compressor
par: Cho, Jeonghun, et autres
Publié: (2025)

P2P: A Poison-to-Poison Remedy for Reliable Backdoor Defense in LLMs
par: Zhao, Shuai, et autres
Publié: (2025)

Why Do Multilingual Reasoning Gaps Emerge in Reasoning Language Models?
par: Kang, Deokhyung, et autres
Publié: (2025)

Multimodal Cognitive Reframing Therapy via Multi-hop Psychotherapeutic Reasoning
par: Kim, Subin, et autres
Publié: (2025)

Exploring Iterative Controllable Summarization with Large Language Models
par: Ryu, Sangwon, et autres
Publié: (2024)

Denoising Table-Text Retrieval for Open-Domain Question Answering
par: Kang, Deokhyung, et autres
Publié: (2024)

Cross-lingual Back-Parsing: Utterance Synthesis from Meaning Representation for Zero-Resource Semantic Parsing
par: Kang, Deokhyung, et autres
Publié: (2024)

Behavior-Aware Item Modeling via Dynamic Procedural Solution Representations for Knowledge Tracing
par: Seo, Jun, et autres
Publié: (2026)

Leveraging What's Overfixed: Post-Correction via LLM Grammatical Error Overcorrection
par: Park, Taehee, et autres
Publié: (2025)

Teach-to-Reason with Scoring: Self-Explainable Rationale-Driven Multi-Trait Essay Scoring
par: Do, Heejin, et autres
Publié: (2025)

Autoregressive Multi-trait Essay Scoring via Reinforcement Learning with Scoring-aware Multiple Rewards
par: Do, Heejin, et autres
Publié: (2024)

Prompt as Triggers for Backdoor Attack: Examining the Vulnerability in Language Models
par: Zhao, Shuai, et autres
Publié: (2023)

Retrieval-Augmented Fine-Tuning With Preference Optimization For Visual Program Generation
par: Kang, Deokhyung, et autres
Publié: (2025)

Speak & Spell: LLM-Driven Controllable Phonetic Error Augmentation for Robust Dialogue State Tracking
par: Lee, Jihyun, et autres
Publié: (2024)

EXAONE 3.0 7.8B Instruction Tuned Language Model
par: An, Soyoung, et autres
Publié: (2024)

From Shortcuts to Triggers: Backdoor Defense with Denoised PoE
par: Liu, Qin, et autres
Publié: (2023)

Adaptive Planning for Multi-Attribute Controllable Summarization with Monte Carlo Tree Search
par: Ryu, Sangwon, et autres
Publié: (2025)

Multi-Dimensional Optimization for Text Summarization via Reinforcement Learning
par: Ryu, Sangwon, et autres
Publié: (2024)

Defending Against Weight-Poisoning Backdoor Attacks for Parameter-Efficient Fine-Tuning
par: Zhao, Shuai, et autres
Publié: (2024)

Neuro-RIT: Neuron-Guided Instruction Tuning for Robust Retrieval-Augmented Language Model
par: Kim, Jaemin, et autres
Publié: (2026)

Multi-Facet Blending for Faceted Query-by-Example Retrieval
par: Do, Heejin, et autres
Publié: (2024)

Towards Prompt Generalization: Grammar-aware Cross-Prompt Automated Essay Scoring
par: Do, Heejin, et autres
Publié: (2025)

Disperse-Then-Merge: Pushing the Limits of Instruction Tuning via Alignment Tax Reduction
par: Fu, Tingchen, et autres
Publié: (2024)

A Survey of Recent Backdoor Attacks and Defenses in Large Language Models
par: Zhao, Shuai, et autres
Publié: (2024)

Acquiring Clean Language Models from Backdoor Poisoned Datasets by Downscaling Frequency Space
par: Wu, Zongru, et autres
Publié: (2024)

Acoustic Feature Mixup for Balanced Multi-aspect Pronunciation Assessment
par: Do, Heejin, et autres
Publié: (2024)

Emerging Safety Attack and Defense in Federated Instruction Tuning of Large Language Models
par: Ye, Rui, et autres
Publié: (2024)

When Emotion Becomes Trigger: Emotion-style dynamic Backdoor Attack Parasitising Large Language Models
par: Liu, Ziyu, et autres
Publié: (2026)

Phased Instruction Fine-Tuning for Large Language Models
par: Pang, Wei, et autres
Publié: (2024)

Investigating Instruction Tuning Large Language Models on Graphs
par: Zhu, Kerui, et autres
Publié: (2024)

Model Fusion through Bayesian Optimization in Language Model Fine-Tuning
par: Jang, Chaeyun, et autres
Publié: (2024)

Understanding LLM Behavior in Multi-Target Cross-Lingual Summarization
par: Ryu, Sangwon, et autres
Publié: (2026)