Revisiting Pre-Trained Models for Chinese Natural Language Processing
This work addresses the effectiveness of pre-trained models for Chinese NLP, providing resources and insights for the community, though it is incremental as it builds on existing methods like RoBERTa.
The paper revisits Chinese pre-trained language models, proposing MacBERT with a masking strategy called MLM as correction, and shows it achieves state-of-the-art performance on eight Chinese NLP tasks.
Bidirectional Encoder Representations from Transformers (BERT) has shown marvelous improvements across various NLP tasks, and consecutive variants have been proposed to further improve the performance of the pre-trained language models. In this paper, we target on revisiting Chinese pre-trained language models to examine their effectiveness in a non-English language and release the Chinese pre-trained language model series to the community. We also propose a simple but effective model called MacBERT, which improves upon RoBERTa in several ways, especially the masking strategy that adopts MLM as correction (Mac). We carried out extensive experiments on eight Chinese NLP tasks to revisit the existing 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. Resources available: https://github.com/ymcui/MacBERT