Pre-trained language models achieve outstanding performance in NLP tasks. Various knowledge distillation methods have been proposed to reduce the heavy computation and storage requirements of pre-trained language models. However, from our observations, student models acquired by knowledge distillation suffer from adversarial attacks, which limits their usage in security sensitive scenarios. In order to overcome these security problems, RoSearch is proposed as a comprehensive framework to search the student models with better adversarial robustness when performing knowledge distillation. A directed acyclic graph based search space is built and an evolutionary search strategy is utilized to guide the searching approach. Each searched architecture is trained by knowledge distillation on pre-trained language model and then evaluated under a robustness-, accuracy- and efficiency-aware metric as environmental fitness. Experimental results show that RoSearch can improve robustness of student models from 7%~18% up to 45.8%~47.8% on different datasets with comparable weight compression ratio to existing distillation methods (4.6$\times$~6.5$\times$ improvement from teacher model BERT_BASE) and low accuracy drop. In addition, we summarize the relationship between student architecture and robustness through statistics of searched models.
翻译:培训前语言模型在NLP任务中取得了杰出的成绩。提出了各种知识蒸馏方法,以减少培训前语言模型的繁重计算和储存要求。然而,根据我们的观察,通过知识蒸馏获得的学生模型受到对抗性攻击,这限制了其在安全敏感情景中的使用。为了克服这些安全问题,建议RoSearch是一个综合框架,用以在进行知识蒸馏时,以更好的对抗性强力搜索学生模型。建立了一个定向的以循环图为基础的搜索空间,并使用进化搜索战略来指导搜索方法。每个搜索结构都通过对预先培训的语言模型进行知识蒸馏培训,然后根据稳健、准确、高效的环保度衡量标准进行评估。实验结果表明,RoSearch可以提高学生模型的稳健性,从7 ⁇ 18%到45.8 ⁇ 47.8%,用于与现有蒸馏方法相比具有可比重量压缩率的不同数据集(4.6美元)和教师模型BERT_BASEE)相比,以及低精确度下降后的改进时间(4.5美元)。此外,我们通过搜索模型总结了学生之间牢固的关系。