Decoupled Alignment for Robust Plug-and-Play Adaptation

Authors: Haozheng Luo, Jiahao Yu, Wenxin Zhang, Jialong Li, Jerry Yao-Chieh Hu, Xinyu Xing, Han Liu
Published: 2024-06-03 | Updated: 2024-06-06

Source: https://arxiv.org/abs/2406.01514

PDF: https://arxiv.org/pdf/2406.01514

AIにより推定されたラベル

モデル性能評価 プロンプトインジェクション LLM性能評価

※ こちらのラベルはAIによって自動的に追加されました。そのため、正確でないことがあります。
詳細は文献データベースについてをご覧ください。

Abstract

We introduce a low-resource safety enhancement method for aligning large language models (LLMs) without the need for supervised fine-tuning (SFT) or reinforcement learning from human feedback (RLHF). Our main idea is to exploit knowledge distillation to extract the alignment information from existing well-aligned LLMs and integrate it into unaligned LLMs in a plug-and-play fashion. Methodology, we employ delta debugging to identify the critical components of knowledge necessary for effective distillation. On the harmful question dataset, our method significantly enhances the average defense success rate by approximately 14.41 pre-trained LLMs, without compromising performance.