Timezone: »

Generating Training Data with Language Models: Towards Zero-Shot Language Understanding
Yu Meng · Jiaxin Huang · Yu Zhang · Jiawei Han

Tue Nov 29 09:00 AM -- 11:00 AM (PST) @ Hall J #520

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.

Author Information

Yu Meng (University of Illinois at Urbana-Champaign)
Jiaxin Huang (University of Illinois Urbana-Champaign)
Yu Zhang (University of Illinois, Urbana Champaign)
Jiawei Han (University of Illinois at Urbana-Champaign)

More from the Same Authors