README.md

metadata

language: Chinese
datasets: CLUECorpus
widget:
  - text: 北京是[MASK]国的首都。

Chinese RoBERTa Miniatures

Model description

This is the set of 24 Chinese RoBERTa models pre-trained by UER-py.

You can download the 24 Chinese RoBERTa miniatures either from the UER-py Github page, or via HuggingFace from the links below:

	H=128	H=256	H=512	H=768
L=2	2/128 (Tiny)	2/256	[2/512]	2/768
L=4	[4/128]	4/256 (Mini)	4/512 (Small)	[4/768]
L=6	[6/128]	[6/256]	[6/512]	[6/768]
L=8	[8/128]	[8/256]	8/512 (Medium)	[8/768]
L=10	[10/128]	[10/256]	[10/512]	[10/768]
L=12	12/128	[12/256]	[12/512]	[12/768 (Base)]

How to use

You can use this model directly with a pipeline for masked language modeling:

>>> from transformers import pipeline
>>> unmasker = pipeline('fill-mask', model='uer/chinese_roberta_L-2_H-768')
>>> unmasker("中国的首都是[MASK]京。")
[
    {'sequence': '[CLS] 中 国 的 首 都 是 北 京 。 [SEP]',
     'score': 0.6976630091667175, 
     'token': 1266,
     'token_str': '北'}, 
    {'sequence': '[CLS] 中 国 的 首 都 是 东 京 。 [SEP]', 
     'score': 0.2517661452293396,
     'token': 691, 
     'token_str': '东'}, 
    {'sequence': '[CLS] 中 国 的 首 都 是 南 京 。 [SEP]',
     'score': 0.04122703894972801,
     'token': 1298,
     'token_str': '南'}, 
    {'sequence': '[CLS] 中 国 的 首 都 是 吴 京 。 [SEP]',
     'score': 0.0015233848243951797,
     'token': 1426,
     'token_str': '吴'}, 
    {'sequence': '[CLS] 中 国 的 首 都 是 普 京 。 [SEP]', 
     'score': 0.001429844880476594, 
     'token': 3249, 
     'token_str': '普'}
]

Here is how to use this model to get the features of a given text in PyTorch:

from transformers import BertTokenizer, BertModel
tokenizer = BertTokenizer.from_pretrained('uer/chinese_roberta_L-2_H-768')
model = BertModel.from_pretrained("uer/chinese_roberta_L-2_H-768")
text = "用你喜欢的任何文本替换我。"
encoded_input = tokenizer(text, return_tensors='pt')
output = model(**encoded_input)

and in TensorFlow:

from transformers import BertTokenizer, TFBertModel
tokenizer = BertTokenizer.from_pretrained('uer/chinese_roberta_L-2_H-768')
model = TFBertModel.from_pretrained("uer/chinese_roberta_L-2_H-768")
text = "用你喜欢的任何文本替换我。"
encoded_input = tokenizer(text, return_tensors='tf')
output = model(encoded_input)

Training data

CLUECorpus2020 and CLUECorpusSmall are used as training data.

Training procedure

Models are pre-trained by UER-py on Tencent Cloud TI-ONE. We pre-train 1,000,000 steps with a sequence length of 128 and then pre-train 250,000 additional steps with a sequence length of 512.

Stage1:

python3 preprocess.py --corpus_path corpora/cluecorpus.txt \
                      --vocab_path models/google_zh_vocab.txt \
                      --dataset_path cluecorpus_seq128_dataset.pt \
                      --processes_num 32 --seq_length 128 \
                      --dynamic_masking --target mlm

python3 pretrain.py --dataset_path cluecorpus_seq128_dataset.pt \
                    --vocab_path models/google_zh_vocab.txt \
                    --config_path models/bert_l2h768_config.json \
                    --output_model_path models/cluecorpus_roberta_l2h768_seq512_model.bin \
                    --world_size 8 --gpu_ranks 0 1 2 3 4 5 6 7 \
                    --total_steps 1000000 --save_checkpoint_steps 100000 --report_steps 50000 \
                    --learning_rate 1e-4 --batch_size 64 \
                    --tie_weights --encoder bert --target mlm

Stage2:

python3 preprocess.py --corpus_path corpora/cluecorpus.txt \
                      --vocab_path models/google_zh_vocab.txt \
                      --dataset_path cluecorpus_seq512_dataset.pt \
                      --processes_num 32 --seq_length 512 \
                      --dynamic_masking --target mlm

python3 pretrain.py --dataset_path cluecorpus_seq512_dataset.pt \
                    --pretrained_model_path models/cluecorpus_roberta_l2h768_seq512_model.bin-1000000 \
                    --vocab_path models/google_zh_vocab.txt \
                    --config_path models/bert_l2h768_config.json \
                    --output_model_path models/cluecorpus_roberta_l2h768_seq512_model.bin \
                    --world_size 8 --gpu_ranks 0 1 2 3 4 5 6 7 \
                    --total_steps 250000 --save_checkpoint_steps 50000 --report_steps 10000 \
                    --learning_rate 5e-5 --batch_size 16 \
                    --tie_weights --encoder bert --target mlm

BibTeX entry and citation info

@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}
}