Small and Practical BERT Models for Sequence Labeling

Abstract

We propose a practical scheme to train a single multilingual sequence labeling model that yields state of the art results and is small and fast enough to run on a single CPU. Starting from a public multilingual BERT checkpoint, our final model is 6x smaller and 27x faster, and has higher accuracy than a state-of-the-art multilingual baseline. We show that our model especially outperforms on low-resource languages, and works on codemixed input text without being explicitly trained on codemixed examples. We showcase the effectiveness of our method by reporting on part-of-speech tagging and morphological prediction on 70 treebanks and 48 languages.

arXiv comment11 pages including appendices; accepted to appear at EMNLP-IJCNLP 2019

References in corpus (1)

• Distilling the Knowledge in a Neural Network

Cited by in corpus (9)

• MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices
• Structured Pruning of a BERT-based Question Answering Model
• Distilling Large Language Models into Tiny and Effective Students using pQRNN
• Revisiting One-vs-All Classifiers for Predictive Uncertainty and Out-of-Distribution Detection in Neural Networks
• Load What You Need: Smaller Versions of Multilingual BERT
• LRC-BERT: Latent-representation Contrastive Knowledge Distillation for Natural Language Understanding
• MergeDistill: Merging Pre-trained Language Models using Distillation
• RuSentEval: Linguistic Source, Encoder Force!
• One Network Fits All? Modular versus Monolithic Task Formulations in Neural Networks