arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2202.04538cs.CLcs.LG
收藏

用语言模型生成训练数据:走向Zero-Shot语言理解

Generating Training Data with Language Models: Towards Zero-Shot Language Understanding

University of Illinois at Urbana-Champaign(伊利诺伊大学厄巴纳-香槟分校)

Yu Meng, Jiaxin Huang, Yu Zhang, Jiawei Han

摘要:

预训练的语言模型(PLM)在各种自然语言处理任务中表现出了卓越的性能:单向PLM(例如,GPT)以其卓越的文本生成能力而闻名;双向PLM(例如,BERT)已经成为自然语言理解(NLU)任务的突出选择。虽然这两种类型的模型都取得了有前途少样本学习性能,他们的潜力零样本学习一直未被充分发掘。在本文中,我们提出了一种简单的方法,使用这两种类型的PLM进行NLU任务的完全零样本学习,而不需要任何特定于任务的数据:单向PLM生成由提示引导的类条件文本,这些文本用作微调双向PLM的训练数据。通过基于生成概率选择的质量训练数据和应用于微调阶段的正则化技术(标签平滑和时间集成)以获得更好的泛化和稳定性,我们的方法在GLUE基准的七个分类任务(例如,在MNLI-m/mm上为72.3/73.8,在SST-2上为92.8),显著优于zero-shot提示方法,并且使用每类32个训练样本实现甚至与强Few-Shot方法相当的结果。

英文摘要:

Pretrained language models (PLMs) have demonstrated remarkable performance in various natural language processing tasks: Unidirectional PLMs (e.g., GPT) are well known for their superior text generation capabilities; bidirectional PLMs (e.g., BERT) have been the prominent choice for natural language understanding (NLU) tasks. While both types of models have achieved promising few-shot learning performance, their potential for zero-shot learning has been underexplored. In this paper, we present a simple approach that uses both types of PLMs for fully zero-shot learning of NLU tasks without requiring any task-specific data: A unidirectional PLM generates class-conditioned texts guided by prompts, which are used as the training data for fine-tuning a bidirectional PLM. With quality training data selected based on the generation probability and regularization techniques (label smoothing and temporal ensembling) applied to the fine-tuning stage for better generalization and stability, our approach demonstrates strong performance across seven classification tasks of the GLUE benchmark (e.g., 72.3/73.8 on MNLI-m/mm and 92.8 on SST-2), significantly outperforming zero-shot prompting methods and achieving even comparable results to strong few-shot approaches using 32 training samples per class.

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。

↑