Masking (illustration)

Quantum mechanics

Linguistics

Voltage

Philosophy

Operating system

Computer science

Visual arts

Chinese language

Language model

Natural language processing

Physics

Artificial intelligence

Transformer

Word (group theory)

Encoder

Speech recognition

IEEE/ACM Transactions on Audio Speech and Language Processing

Bidirectional Encoder Representations from Transformers (BERT) has shown marvelous improvements across various NLP tasks, and its consecutive variants have been proposed to further improve the performance of the pre-trained language models. In this paper, we aim to first introduce the whole word masking (wwm) strategy for Chinese BERT, along with a series of Chinese pre-trained language models. Then we also propose a simple but effective model called MacBERT, which improves upon RoBERTa in several ways. Especially, we propose a new masking strategy called MLM as correction (Mac). To demonstrate the effectiveness of these models, we create a series of Chinese pre-trained language models as our baselines, including BERT, RoBERTa, ELECTRA, RBT, etc. We carried out extensive experiments on ten Chinese NLP tasks to evaluate the created Chinese pre-trained language models as well as the proposed MacBERT. Experimental results show that MacBERT could achieve state-of-the-art performances on many NLP tasks, and we also ablate details with several findings that may help future research. We open-source our pre-trained language models for further facilitating our research community.

Pre-Training With Whole Word Masking for Chinese BERT