paper-with-me

Papers

Inoculation Adapters: Improved Selective Generalization of Capabilities with Fewer Surprising Backdoors

2026-06-29 · Maxime Riché, Daniel Tan, Vili Kohonen, Niels Warncke arxiv

Inoculation prompting is a selective-generalization technique used against Emergent Misalignment. We introduce inoculation adapters (IA), a family of methods that similarly reduce the optimization pressure to learn undesired traits by strengthening those traits during training. Inoculation adapters are LoRAs that are trained and used in three steps: (1) trained on undesired traits; (2) attached frozen while a separate task adapter is trained on data exhibiting both desired and undesired traits; (3) the IA is discarded at deployment, while only the task adapter is kept. We compare inoculation adapters with four selective-generalization baselines: inoculation prompting, preventative steering, Concept Ablation Fine-Tuning (CAFT), and KL regularization. Across nine setups and five model families, the inoculation adapter family spans a new Pareto frontier of desired trait retention vs. undesired trait suppression, although given wide confidence intervals the magnitude of improvement remains uncertain. Inoculation adapters also avoid two drawbacks of inoculation prompting: they can suppress capabilities and traits that cannot be reliably elicited by a prompt, and they introduce fewer surprising backdoors. However, no IA variant optimizes all objectives perfectly; gains in desired-trait generalization are generally accompanied by weaker suppression of the undesired trait and increased backdoor occurrence.

📄 PDF Abstract BibTeX arXiv:2606.30252

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Inoculation Prompting: Eliciting traits from LLMs during training can suppress them at test-time

2025-10-05 · Daniel Tan, Anders Woodruff, Niels Warncke, Arun Jose 외 arxiv

Language model finetuning often results in learning undesirable traits in combination with desired ones. To address this, we propose inoculation prompting: modifying finetuning data by prepending a short system-prompt in…

Inoculation Prompting: Instructing LLMs to misbehave at train-time improves test-time alignment

2025-10-06 · Nevan Wichers, Aram Ebtekar, Ariana Azarbal, Victor Gillioz 외 arxiv

Large language models are sometimes trained with imperfect oversight signals, leading to undesired behaviors such as reward hacking and sycophancy. Improving oversight quality can be expensive or infeasible, motivating m…

Improving QA Model Performance with Cartographic Inoculation

2024-01-30 · Allen Chen, Okan Tanrikulu

QA models are faced with complex and open-ended contextual reasoning problems, but can often learn well-performing solution heuristics by exploiting dataset-specific patterns in their training data. These patterns, or "d…

model

Proactive Defense: Compound AI for Detecting Persuasion Attacks and Measuring Inoculation Effectiveness

2025-11-23 · Svitlana Volkova, Will Dupree, Hsien-Te Kao, Peter Bautista 외 arxiv

This paper introduces BRIES, a novel compound AI architecture designed to detect and measure the effectiveness of persuasion attacks across information environments. We present a system with specialized agents: a Twister…

Prompt EngineeringCausal Inference

FisherTune: Fisher-Guided Robust Tuning of Vision Foundation Models for Domain Generalized Segmentation

2025-03-23 · CVPR 2025 1 · Dong Zhao, Jinlong Li, Shuang Wang, Mengyao Wu 외

Vision Foundation Models (VFMs) excel in generalization due to large-scale pretraining, but fine-tuning them for Domain Generalized Semantic Segmentation (DGSS) while maintaining this ability remains challenging. Existin…

Semantic SegmentationVariational Inference