paper-with-me

홈 › Papers

Self and Cross-Model Distillation for LLMs: Effective Methods for Refusal Pattern Alignment

2024-06-17 · Jie Li, Yi Liu, Chongyang Liu, Xiaoning Ren, Ling Shi, Weisong Sun, Yinxing Xue

Large Language Models (LLMs) like OpenAI's GPT series, Anthropic's Claude, and Meta's LLaMa have shown remarkable capabilities in text generation. However, their susceptibility to toxic prompts presents significant security challenges. This paper investigates alignment techniques, including Supervised Fine-Tuning (SFT) and Reinforcement Learning from Human Feedback (RLHF), to mitigate these risks. We conduct an empirical study on refusal patterns across nine LLMs, revealing that models with uniform refusal patterns, such as Claude3, exhibit higher security. Based on these findings, we propose self-distilling and cross-model distilling methods to enhance LLM security. Our results show that these methods significantly improve refusal rates and reduce unsafe content, with cross-model distilling achieving refusal rates close to Claude3's 94.51%. These findings underscore the potential of distillation-based alignment in securing LLMs against toxic prompts.

📄 PDF Abstract BibTeX arXiv:2406.11285

Code (0)

등록된 구현이 없습니다.

Tasks

Text Generation

Methods 이 논문이 사용한 방법론

Attention 설명 없음
Cosine Annealing Cosine Annealing is a type of learning rate schedule that has the effect of starting with a large learning rate that is relatively rapidly decreased to a minimum value before…
BPE Byte Pair Encoding, or BPE, is a subword segmentation algorithm that encodes rare and unknown words as sequences of subword units. The intuition is that various word…
Attention Dropout Attention Dropout is a type of dropout used in attention-based architectures, where elements are randomly dropped out of the…
Weight Decay 설명 없음
Dropout Dropout is a regularization technique for neural networks that drops a unit (along with connections) at training time with a specified probability $p$ (a common value is…
Adam 설명 없음
Linear Warmup With Cosine Annealing Linear Warmup With Cosine Annealing is a learning rate schedule where we increase the learning rate linearly for $n$ updates and then anneal according to a cosine schedule…

Similar Papers 제목 키워드 기반

SelfCodeAlign: Self-Alignment for Code Generation

2024-10-31 · Yuxiang Wei, Federico Cassano, Jiawei Liu, Yifeng Ding 외

Instruction tuning is a supervised fine-tuning approach that significantly improves the ability of large language models (LLMs) to follow human instructions. We propose SelfCodeAlign, the first fully transparent and perm…

Code GenerationHumanEval

UniSD: Towards a Unified Self-Distillation Framework for Large Language Models

2026-05-07 · Yiqiao Jin, Yiyang Wang, Lucheng Fu, Yijia Xiao 외 arxiv

Self-distillation (SD) offers a promising path for adapting large language models (LLMs) without relying on stronger external teachers. However, SD in autoregressive LLMs remains challenging because self-generated trajec…

Contrastive Learning

Learning from the Self-future: On-policy Self-distillation for dLLMs

2026-06-16 · Yifu Luo, Zeyu Chen, Haoyu Wang, Xinhao Hu 외 arxiv

On-policy self-distillation (OPSD) has proven effective for post-training large language models (LLMs), yet its application to diffusion LLMs (dLLMs) remains unexplored. Existing OPSD methods are inherently autoregressiv…

Multilingual Safety Alignment via Self-Distillation

2026-05-03 · Ruiyang Qin, Qingzhuo Wang, Dongrui Liu, Qiang Li 외 arxiv

Large language models (LLMs) exhibit severe multilingual safety misalignment: they possess strong safeguards in high-resource languages but remain highly vulnerable to jailbreak attacks in low-resource languages. Current…

OPSDL: On-Policy Self-Distillation for Long-Context Language Models

2026-04-19 · Xinsen Zhang, Zhenkai Ding, Tianjun Pan, Run Yang 외 arxiv

Extending the effective context length of large language models (LLMs) remains a central challenge for real-world applications. While recent post-training methods have made progress in long-context scaling, they either r…