CharBERT: Character-aware Pre-trained Language Model

Abstract

Most pre-trained language models (PLMs) construct word representations at subword level with Byte-Pair Encoding (BPE) or its variations, by which OOV (out-of-vocab) words are almost avoidable. However, those methods split a word into subword units and make the representation incomplete and fragile. In this paper, we propose a character-aware pre-trained language model named CharBERT improving on the previous methods (such as BERT, RoBERTa) to tackle these problems. We first construct the contextual word embedding for each token from the sequential character representations, then fuse the representations of characters and the subword representations by a novel heterogeneous interaction module. We also propose a new pre-training task named NLM (Noisy LM) for unsupervised character representation learning. We evaluate our method on question answering, sequence labeling, and text classification tasks, both on the original datasets and adversarial misspelling test sets. The experimental results show that our method can significantly improve the performance and robustness of PLMs simultaneously. Pretrained models, evaluation sets, and code are available at https://github.com/wtma/CharBERT

arXiv comment12 pages, to appear at COLING 2020

References in corpus (8)

• Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation
• Gaussian Error Linear Units (GELUs)
• Bidirectional Attention Flow for Machine Comprehension
• Introduction to the CoNLL-2002 Shared Task: Language-Independent Named Entity Recognition
• Synthetic and Natural Noise Both Break Neural Machine Translation
• On Adversarial Examples for Character-Level Neural Machine Translation
• Words or Characters? Fine-grained Gating for Reading Comprehension
• Benchmarking Robustness of Machine Reading Comprehension Models

Cited by in corpus (9)

• Between words and characters: A Brief History of Open-Vocabulary Modeling and Tokenization in NLP
• AMMUS : A Survey of Transformer-based Pretrained Models in Natural Language Processing
• Better Robustness by More Coverage: Adversarial Training with Mixup Augmentation for Robust Fine-tuning
• Language Modelling with Pixels
• DomURLs_BERT: Pre-trained BERT-based Model for Malicious Domains and URLs Detection and Classification
• Character-level HyperNetworks for Hate Speech Detection
• LadRa-Net: Locally-Aware Dynamic Re-read Attention Net for Sentence Semantic Matching
• Learning to Look Inside: Augmenting Token-Based Encoders with Character-Level Information
• Using Distributional Principles for the Semantic Study of Contextual Language Models