Compressing Large-Scale Transformer-Based Models: A Case Study on BERT
This is a survey paper that synthesizes existing research, making it incremental in nature, aimed at researchers and practitioners working on efficient NLP model deployment.
The paper tackles the problem of compressing large-scale Transformer-based models like BERT to reduce their resource and computational demands for deployment on low-capability devices, by summarizing and analyzing state-of-the-art compression methods and providing best practices and insights.
Pre-trained Transformer-based models have achieved state-of-the-art performance for various Natural Language Processing (NLP) tasks. However, these models often have billions of parameters, and, thus, are too resource-hungry and computation-intensive to suit low-capability devices or applications with strict latency requirements. One potential remedy for this is model compression, which has attracted a lot of research attention. Here, we summarize the research in compressing Transformers, focusing on the especially popular BERT model. In particular, we survey the state of the art in compression for BERT, we clarify the current best practices for compressing large-scale Transformer models, and we provide insights into the workings of various methods. Our categorization and analysis also shed light on promising future research directions for achieving lightweight, accurate, and generic NLP models.