File size: 10,443 Bytes
f8c3cd5
b082602
6e8e3b2
f8c3cd5
 
3a33988
 
4ebd28b
 
6e8e3b2
 
 
 
 
3eaf964
6e8e3b2
 
 
3eaf964
6e8e3b2
3a33988
 
06bfe1c
 
 
 
 
 
6e8e3b2
 
 
5a94b33
3a33988
5a94b33
 
 
 
 
3a33988
 
6e8e3b2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3a33988
6e8e3b2
 
 
3eaf964
6e8e3b2
 
 
 
3a33988
6e8e3b2
 
 
3a33988
 
47b8a7f
6e8e3b2
3a33988
6e8e3b2
 
 
d68fc5e
3a33988
 
 
 
47b8a7f
6e8e3b2
3a33988
6e8e3b2
3a33988
6e8e3b2
 
 
3a33988
 
47b8a7f
6e8e3b2
3a33988
6e8e3b2
 
3a33988
47b8a7f
d68fc5e
3a33988
 
 
 
47b8a7f
3a33988
 
 
 
 
3eaf964
3a33988
47b8a7f
6e8e3b2
 
 
 
 
fac54e5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6e8e3b2
 
 
 
 
 
 
 
 
d6ff93f
06bfe1c
 
 
 
d6ff93f
06bfe1c
47b8a7f
06bfe1c
 
 
 
 
 
d6ff93f
06bfe1c
 
 
 
 
 
 
 
8060dc2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
---
language: zh
datasets: CLUECorpusSmall
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](https://github.com/dbiir/UER-py/), which is introduced in [this paper](https://arxiv.org/abs/1909.05658).

[Turc et al.](https://arxiv.org/abs/1908.08962) have shown that the standard BERT recipe is effective on a wide range of model sizes. Following their paper, we released the 24 Chinese RoBERTa models. In order to facilitate users to reproduce the results, we used the publicly available corpus and provided all training details.

You can download the 24 Chinese RoBERTa miniatures either from the [UER-py Modelzoo page](https://github.com/dbiir/UER-py/wiki/Modelzoo), or via HuggingFace from the links below:

|          |           H=128           |           H=256           |            H=512            |            H=768            |
| -------- | :-----------------------: | :-----------------------: | :-------------------------: | :-------------------------: |
| **L=2**  | [**2/128 (Tiny)**][2_128] |      [2/256][2_256]       |       [2/512][2_512]        |       [2/768][2_768]        |
| **L=4**  |      [4/128][4_128]       | [**4/256 (Mini)**][4_256] | [**4/512 (Small)**][4_512]  |       [4/768][4_768]        |
| **L=6**  |      [6/128][6_128]       |      [6/256][6_256]       |       [6/512][6_512]        |       [6/768][6_768]        |
| **L=8**  |      [8/128][8_128]       |      [8/256][8_256]       | [**8/512 (Medium)**][8_512] |       [8/768][8_768]        |
| **L=10** |     [10/128][10_128]      |     [10/256][10_256]      |      [10/512][10_512]       |      [10/768][10_768]       |
| **L=12** |     [12/128][12_128]      |     [12/256][12_256]      |      [12/512][12_512]       | [**12/768 (Base)**][12_768] |

Here are scores on the devlopment set of six Chinese tasks:

| Model          | Score | book_review | chnsenticorp | lcqmc | tnews(CLUE) | iflytek(CLUE) | ocnli(CLUE) |
| -------------- | :---: | :----: | :----------: | :---: | :---------: | :-----------: | :---------: |
| RoBERTa-Tiny   | 72.3  |  83.4  |     91.4     | 81.8  |    62.0     |     55.0      |    60.3     |
| RoBERTa-Mini   | 75.9  |  85.7  |     93.7     | 86.1  |    63.9     |     58.3      |    67.4     |
| RoBERTa-Small  | 76.9  |  87.5  |     93.4     | 86.5  |    65.1     |     59.4      |    69.7     |
| RoBERTa-Medium | 78.0  |  88.7  |     94.8     | 88.1  |    65.6     |     59.5      |    71.2     |
| RoBERTa-Base   | 79.7  |  90.1  |     95.2     | 89.2  |    67.0     |     60.9      |    75.5     |

For each task, we selected the best fine-tuning hyperparameters from the lists below, and trained with the sequence length of 128:

- epochs: 3, 5, 8
- batch sizes: 32, 64
- learning rates: 3e-5, 1e-4, 3e-4

## How to use

You can use this model directly with a pipeline for masked language modeling (take the case of RoBERTa-Medium):

```python
>>> from transformers import pipeline
>>> unmasker = pipeline('fill-mask', model='uer/chinese_roberta_L-8_H-512')
>>> unmasker("中国的首都是[MASK]京。")
[
    {'sequence': '[CLS] 中 国 的 首 都 是 北 京 。 [SEP]', 
     'score': 0.8701988458633423, 
     'token': 1266, 
     'token_str': '北'},
    {'sequence': '[CLS] 中 国 的 首 都 是 南 京 。 [SEP]',
     'score': 0.1194809079170227, 
     'token': 1298, 
     'token_str': '南'},
    {'sequence': '[CLS] 中 国 的 首 都 是 东 京 。 [SEP]', 
     'score': 0.0037803512532263994, 
     'token': 691, 
     'token_str': '东'},
    {'sequence': '[CLS] 中 国 的 首 都 是 普 京 。 [SEP]',
     'score': 0.0017127094324678183, 
     'token': 3249,
     'token_str': '普'},
    {'sequence': '[CLS] 中 国 的 首 都 是 望 京 。 [SEP]',
     'score': 0.001687526935711503,
     'token': 3307, 
     'token_str': '望'}
]
```

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

```python
from transformers import BertTokenizer, BertModel
tokenizer = BertTokenizer.from_pretrained('uer/chinese_roberta_L-8_H-512')
model = BertModel.from_pretrained("uer/chinese_roberta_L-8_H-512")
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:

```python
from transformers import BertTokenizer, TFBertModel
tokenizer = BertTokenizer.from_pretrained('uer/chinese_roberta_L-8_H-512')
model = TFBertModel.from_pretrained("uer/chinese_roberta_L-8_H-512")
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

[CLUECorpusSmall](https://github.com/CLUEbenchmark/CLUECorpus2020/) is used as training data. We found that models pre-trained on CLUECorpusSmall outperform those pre-trained on CLUECorpus2020, although CLUECorpus2020 is much larger than CLUECorpusSmall.

## Training procedure

Models are pre-trained by [UER-py](https://github.com/dbiir/UER-py/) on [Tencent Cloud](https://cloud.tencent.com/). 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. We use the same hyper-parameters on different model sizes.

Taking the case of RoBERTa-Medium

Stage1:

```
python3 preprocess.py --corpus_path corpora/cluecorpussmall.txt \
                      --vocab_path models/google_zh_vocab.txt \
                      --dataset_path cluecorpussmall_seq128_dataset.pt \
                      --processes_num 32 --seq_length 128 \
                      --dynamic_masking --data_processor mlm
```

```
python3 pretrain.py --dataset_path cluecorpussmall_seq128_dataset.pt \
                    --vocab_path models/google_zh_vocab.txt \
                    --config_path models/bert/medium_config.json \
                    --output_model_path models/cluecorpussmall_roberta_medium_seq128_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 \
                    --data_processor mlm --target mlm
```

Stage2:

```
python3 preprocess.py --corpus_path corpora/cluecorpussmall.txt \
                      --vocab_path models/google_zh_vocab.txt \
                      --dataset_path cluecorpussmall_seq512_dataset.pt \
                      --processes_num 32 --seq_length 512 \
                      --dynamic_masking --data_processor mlm
```

```
python3 pretrain.py --dataset_path cluecorpussmall_seq512_dataset.pt \
                    --vocab_path models/google_zh_vocab.txt \
                    --pretrained_model_path models/cluecorpussmall_roberta_medium_seq128_model.bin-1000000 \
                    --config_path models/bert/medium_config.json \
                    --output_model_path models/cluecorpussmall_roberta_medium_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 \
                    --data_processor mlm --target mlm
```

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

```
python3 scripts/convert_bert_from_uer_to_huggingface.py --input_model_path models/cluecorpussmall_roberta_medium_seq512_model.bin-250000 \                                                        
                                                        --output_model_path pytorch_model.bin \
                                                        --layers_num 8 --type mlm
```

### BibTeX entry and citation info

```
@article{devlin2018bert,
  title={Bert: Pre-training of deep bidirectional transformers for language understanding},
  author={Devlin, Jacob and Chang, Ming-Wei and Lee, Kenton and Toutanova, Kristina},
  journal={arXiv preprint arXiv:1810.04805},
  year={2018}
}

@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{turc2019,
  title={Well-Read Students Learn Better: On the Importance of Pre-training Compact Models},
  author={Turc, Iulia and Chang, Ming-Wei and Lee, Kenton and Toutanova, Kristina},
  journal={arXiv preprint arXiv:1908.08962v2 },
  year={2019}
}

@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}
}
```

[2_128]:https://huggingface.co/uer/chinese_roberta_L-2_H-128
[2_256]:https://huggingface.co/uer/chinese_roberta_L-2_H-256
[2_512]:https://huggingface.co/uer/chinese_roberta_L-2_H-512
[2_768]:https://huggingface.co/uer/chinese_roberta_L-2_H-768
[4_128]:https://huggingface.co/uer/chinese_roberta_L-4_H-128
[4_256]:https://huggingface.co/uer/chinese_roberta_L-4_H-256
[4_512]:https://huggingface.co/uer/chinese_roberta_L-4_H-512
[4_768]:https://huggingface.co/uer/chinese_roberta_L-4_H-768
[6_128]:https://huggingface.co/uer/chinese_roberta_L-6_H-128
[6_256]:https://huggingface.co/uer/chinese_roberta_L-6_H-256
[6_512]:https://huggingface.co/uer/chinese_roberta_L-6_H-512
[6_768]:https://huggingface.co/uer/chinese_roberta_L-6_H-768
[8_128]:https://huggingface.co/uer/chinese_roberta_L-8_H-128
[8_256]:https://huggingface.co/uer/chinese_roberta_L-8_H-256
[8_512]:https://huggingface.co/uer/chinese_roberta_L-8_H-512
[8_768]:https://huggingface.co/uer/chinese_roberta_L-8_H-768
[10_128]:https://huggingface.co/uer/chinese_roberta_L-10_H-128
[10_256]:https://huggingface.co/uer/chinese_roberta_L-10_H-256
[10_512]:https://huggingface.co/uer/chinese_roberta_L-10_H-512
[10_768]:https://huggingface.co/uer/chinese_roberta_L-10_H-768
[12_128]:https://huggingface.co/uer/chinese_roberta_L-12_H-128
[12_256]:https://huggingface.co/uer/chinese_roberta_L-12_H-256
[12_512]:https://huggingface.co/uer/chinese_roberta_L-12_H-512
[12_768]:https://huggingface.co/uer/chinese_roberta_L-12_H-768