To BERT or Not to BERT: Comparing Task-specific and Task-agnostic Semi-Supervised Approaches for Sequence Tagging
This addresses the issue of high computational costs for NLP practitioners, though it is incremental as it builds on existing semi-supervised methods.
The paper tackled the problem of costly training of large Transformer models like BERT by comparing task-specific semi-supervised approaches with task-agnostic ones for sequence tagging, showing that Cross-View Training achieves similar performance with reduced financial and environmental impact.
Leveraging large amounts of unlabeled data using Transformer-like architectures, like BERT, has gained popularity in recent times owing to their effectiveness in learning general representations that can then be further fine-tuned for downstream tasks to much success. However, training these models can be costly both from an economic and environmental standpoint. In this work, we investigate how to effectively use unlabeled data: by exploring the task-specific semi-supervised approach, Cross-View Training (CVT) and comparing it with task-agnostic BERT in multiple settings that include domain and task relevant English data. CVT uses a much lighter model architecture and we show that it achieves similar performance to BERT on a set of sequence tagging tasks, with lesser financial and environmental impact.