paper-with-me

홈 › Papers

Adversarial Contrastive Decoding: Boosting Safety Alignment of Large Language Models via Opposite Prompt Optimization

2024-06-24 · Zhengyue Zhao, Xiaoyun Zhang, Kaidi Xu, Xing Hu, Rui Zhang, Zidong Du, Qi Guo, Yunji Chen

With the widespread application of Large Language Models (LLMs), it has become a significant concern to ensure their safety and prevent harmful responses. While current safe-alignment methods based on instruction fine-tuning and Reinforcement Learning from Human Feedback (RLHF) can effectively reduce harmful responses from LLMs, they often require high-quality datasets and heavy computational overhead during model training. Another way to align language models is to modify the logit of tokens in model outputs without heavy training. Recent studies have shown that contrastive decoding can enhance the performance of language models by reducing the likelihood of confused tokens. However, these methods require the manual selection of contrastive models or instruction templates. To this end, we propose Adversarial Contrastive Decoding (ACD), an optimization-based framework to generate two opposite system prompts for prompt-based contrastive decoding. ACD only needs to apply a lightweight prompt tuning on a rather small anchor dataset (< 3 min for each model) without training the target model. Experiments conducted on extensive models and benchmarks demonstrate that the proposed method achieves much better safety performance than previous model training-free decoding methods without sacrificing its original generation ability.

📄 PDF Abstract BibTeX arXiv:2406.16743

Code (0)

등록된 구현이 없습니다.

Tasks

Safety Alignment

Methods 이 논문이 사용한 방법론

ALIGN In the ALIGN method, visual and language representations are jointly trained from noisy image alt-text data. The image and text encoders are learned via contrastive loss…

Similar Papers 제목 키워드 기반

ROSE Doesn't Do That: Boosting the Safety of Instruction-Tuned Large Language Models with Reverse Prompt Contrastive Decoding

2024-02-19 · Qihuang Zhong, Liang Ding, Juhua Liu, Bo Du 외

With the development of instruction-tuned large language models (LLMs), improving the safety of LLMs has become more critical. However, the current approaches for aligning the LLMs output with expected safety usually req…

Please refuse to answer me! Mitigating Over-Refusal in Large Language Models via Adaptive Contrastive Decoding

2026-04-18 · Yupeng Qi, Ziyu Lyu, Lixin Cui, Lu Bai 외 arxiv

Safety-aligned large language models (LLMs) often generate refusal responses to harmless queries due to the over-refusal problem. However, existing methods for mitigating over-refusal cannot maintain a low refusal ratio …

Steering Multimodal Large Language Models Decoding for Context-Aware Safety

2025-09-23 · Zheyuan Liu, Zhangchen Xu, Guangyao Dou, Xiangchi Yuan 외 arxiv

Multimodal Large Language Models (MLLMs) are increasingly deployed in real-world applications, yet their ability to make context-aware safety decisions remains limited. Existing methods often fail to balance oversensitiv…

CARE: Decoding Time Safety Alignment via Rollback and Introspection Intervention

2025-09-01 · Xiaomeng Hu, Fei Huang, Chenhan Yuan, Junyang Lin 외 arxiv

As large language models (LLMs) are increasingly deployed in real-world applications, ensuring the safety of their outputs during decoding has become a critical challenge. However, existing decoding-time interventions, s…

TRACEALIGN -- Tracing the Drift: Attributing Alignment Failures to Training-Time Belief Sources in LLMs

2025-08-04 · Amitava Das, Vinija Jain, Aman Chadha arxiv

Large Language Models (LLMs) fine-tuned to align with human values often exhibit alignment drift, producing unsafe or policy-violating completions when exposed to adversarial prompts, decoding perturbations, or paraphras…