Generate, Annotate, And Learn: NLP With Synthetic Text

He Xuanli, Nassar Islam, Kiros Jamie, Haffari Gholamreza, Norouzi Mohammad. Arxiv 2021

[Paper]    
Applications Distillation Efficiency And Optimization Few Shot Model Architecture Pretraining Methods Prompting Tools Training Techniques Transformer

This paper studies the use of language models as a source of synthetic unlabeled text for NLP. We formulate a general framework called ``generate, annotate, and learn (GAL)’’ to take advantage of synthetic text within knowledge distillation, self-training, and few-shot learning applications. To generate high-quality task-specific text, we either fine-tune LMs on inputs from the task of interest, or prompt large LMs with few examples. We use the best available classifier to annotate synthetic text with soft pseudo labels for knowledge distillation and self-training, and use LMs to obtain hard labels for few-shot learning. We train new supervised models on the combination of labeled and pseudo-labeled data, which results in significant gains across several applications. We investigate key components of GAL and present theoretical and empirical arguments against the use of class-conditional LMs to generate synthetic labeled text instead of unlabeled text. GAL achieves new state-of-the-art knowledge distillation results for 6-layer transformers on the GLUE leaderboard.

Similar Work