arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2412.17846cs.CL
收藏

利用响应启动提示增强大型语言模型的知识蒸馏

Enhancing Knowledge Distillation for LLMs with Response-Priming Prompting

Algoverse AI Research(Algoverse AI研究院)

Vijay Goyal, Mustafa Khan, Aprameya Tirupati, Harveer Saini, Michael Lam, Kevin Zhu

AI总结:

本文提出新型响应启动提示策略,通过从Llama 3.1 405B蒸馏至8B模型,显著提升了学生模型在GSM8K上的性能,并发现提示能诱导注意力头产生积极行为从而提高准确率。

摘要:

大型语言模型(LLMs)在广泛的自然语言处理(NLP)任务中展现出了卓越的性能。然而,由于巨大的计算需求和资源限制,这些模型通常难以部署。知识蒸馏(KD)是一种将较大LLMs性能转移到较小模型的有效技术。传统的KD方法主要关注教师模型的直接输出,很少强调提示在知识转移过程中的作用。在本文中,我们提出了一组应用于知识蒸馏流程的新型响应启动提示策略,以提升学生模型的性能。我们的方法通过从量化的Llama 3.1 405B Instruct教师模型中蒸馏知识,微调了一个较小的Llama 3.1 8B Instruct模型。我们应用了LoRA优化,并在GSM8K基准上进行评估。实验结果表明,将推理引导提示整合到所提出的KD流程中能显著提升学生模型性能,为在资源受限环境中部署强大模型提供了高效途径。我们发现,与未使用提示蒸馏的相同模型相比,Ground Truth提示使蒸馏后的Llama 3.1 8B Instruct在GSM8K上的性能提升了55%。对学生模型自注意力层的深入调查表明,更成功的提示模型往往在其注意力头内部表现出某些积极行为,这与它们提高的准确率密切相关。我们的实现可在https://github.com/alonso130r/knowledge-distillation找到。

英文摘要:

Large language models (LLMs) have demonstrated remarkable performance across a wide range of natural language processing (NLP) tasks. However, these models are often difficult to deploy due to significant computational requirements and resource constraints. Knowledge distillation (KD) is an effective technique for transferring the performance of larger LLMs to smaller models. Traditional KD methods primarily focus on the direct output of the teacher model, with little emphasis on the role of prompting during knowledge transfer. In this paper, we propose a set of novel response-priming prompting strategies applied in the knowledge distillation pipeline to enhance the performance of student models. Our approach fine-tunes a smaller Llama 3.1 8B Instruct model by distilling knowledge from a quantized Llama 3.1 405B Instruct teacher model. We apply LoRA optimization and evaluate on the GSM8K benchmark. Experimental results demonstrate that integrating reasoning-eliciting prompting into the proposed KD pipeline significantly improves student model performance, offering an efficient way to deploy powerful models in resource-constrained environments. We find that Ground Truth prompting results in a 55\% performance increase on GSM8K for a distilled Llama 3.1 8B Instruct compared to the same model distilled without prompting. A thorough investigation into the self-attention layers of the student models indicates that the more successful prompted models tend to exhibit certain positive behaviors inside their attention heads which can be tied to their increased accuracy. Our implementation can be found at https://github.com/alonso130r/knowledge-distillation.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑