SOURCE-LINKED INTELLIGENCE

An Efficient and Modular Framework for Targeted Harm Mitigation in LLMS

arXiv · AI, language, vision and robotics · article · Sep 12, 2026 · UTC

Large Language Models (LLMs) are powerful zero-shot learners but remain prone to misalignment with human preferences, often producing biased, toxic, or otherwise harmful outputs. Existing alignment methods, while effective, are costly and tightly coupled to the model, limiting flexibility and scalability. We propose a modular correction framework that augments pretrained LLMs with Activated LoRA (aLoRA) adapters and a context-aware routing mechanism to eliminate harms from misaligned model responses. Our approach enables expert adapters to activate mid-sequence without invalidating the KV cach

Read original source ↗ Open in workspace

recordType

paper

region

Global