Chinese RoBERTa-Base Models for Text Classification

Model description

This is the set of 5 Chinese RoBERTa-Base classification models fine-tuned by UER-py, which is introduced in this paper. Besides, the models could also be fine-tuned by TencentPretrain introduced in this paper, which inherits UER-py to support models with parameters above one billion, and extends it to a multimodal pre-training framework.

You can download the 5 Chinese RoBERTa-Base classification models either from the UER-py Modelzoo page, or via Model Database from the links below:

How to use

You can use this model directly with a pipeline for text classification (take the case of roberta-base-finetuned-chinanews-chinese):

>>> from transformers import AutoModelForSequenceClassification,AutoTokenizer,pipeline
>>> model = AutoModelForSequenceClassification.from_pretrained('uer/roberta-base-finetuned-chinanews-chinese')
>>> tokenizer = AutoTokenizer.from_pretrained('uer/roberta-base-finetuned-chinanews-chinese')
>>> text_classification = pipeline('sentiment-analysis', model=model, tokenizer=tokenizer)
>>> text_classification("北京上个月召开了两会")
    [{'label': 'mainland China politics', 'score': 0.7211663722991943}]

Training data

5 Chinese text classification datasets are used. JD full, JD binary, and Dianping datasets consist of user reviews of different sentiment polarities. Ifeng and Chinanews consist of first paragraphs of news articles of different topic classes. They are collected by Glyph project and more details are discussed in the corresponding paper.

Training procedure

Models are fine-tuned by UER-py on Tencent Cloud. We fine-tune three epochs with a sequence length of 512 on the basis of the pre-trained model chinese_roberta_L-12_H-768. At the end of each epoch, the model is saved when the best performance on development set is achieved. We use the same hyper-parameters on different models.

Taking the case of roberta-base-finetuned-chinanews-chinese

python3 run_classifier.py --pretrained_model_path models/cluecorpussmall_roberta_base_seq512_model.bin-250000 \
                          --vocab_path models/google_zh_vocab.txt \
                          --train_path datasets/glyph/chinanews/train.tsv \
                          --dev_path datasets/glyph/chinanews/dev.tsv \
                          --output_model_path models/chinanews_classifier_model.bin \
                          --learning_rate 3e-5 --epochs_num 3 --batch_size 32 --seq_length 512

Finally, we convert the pre-trained model into Model Database's format:

python3 scripts/convert_bert_text_classification_from_uer_to_Model Database.py --input_model_path models/chinanews_classifier_model.bin \
                                                                            --output_model_path pytorch_model.bin \
                                                                            --layers_num 12

BibTeX entry and citation info

@article{liu2019roberta,
  title={Roberta: A robustly optimized bert pretraining approach},
  author={Liu, Yinhan and Ott, Myle and Goyal, Naman and Du, Jingfei and Joshi, Mandar and Chen, Danqi and Levy, Omer and Lewis, Mike and Zettlemoyer, Luke and Stoyanov, Veselin},
  journal={arXiv preprint arXiv:1907.11692},
  year={2019}
}

@article{zhang2017encoding,
  title={Which encoding is the best for text classification in chinese, english, japanese and korean?},
  author={Zhang, Xiang and LeCun, Yann},
  journal={arXiv preprint arXiv:1708.02657},
  year={2017}
}

@article{zhao2019uer,
  title={UER: An Open-Source Toolkit for Pre-training Models},
  author={Zhao, Zhe and Chen, Hui and Zhang, Jinbin and Zhao, Xin and Liu, Tao and Lu, Wei and Chen, Xi and Deng, Haotang and Ju, Qi and Du, Xiaoyong},
  journal={EMNLP-IJCNLP 2019},
  pages={241},
  year={2019}
}

@article{zhao2023tencentpretrain,
  title={TencentPretrain: A Scalable and Flexible Toolkit for Pre-training Models of Different Modalities},
  author={Zhao, Zhe and Li, Yudong and Hou, Cheng and Zhao, Jing and others},
  journal={ACL 2023},
  pages={217},
  year={2023}