Update README.md

README.md

@@ -3,6 +3,8 @@ language: Chinese
 datasets: CLUECorpusSmall
 widget: 
 - text: "北京是[MASK]国的首都。"
+
+
 ---
 
 
@@ -16,26 +18,27 @@ This is the set of 24 Chinese RoBERTa models pre-trained by [UER-py](https://www
 
 You can download the 24 Chinese RoBERTa miniatures either from the [UER-py Github page](https://github.com/dbiir/UER-py/), or via HuggingFace from the links below:
 
-|   |H=128|H=256|H=512|H=768|
-|---|:---:|:---:|:---:|:---:|
-| **L=2**  |[**2/128 (Tiny)**][2_128]|[2/256]|[2/512]|[2/768]|
-| **L=4**  |[4/128]|[**4/256 (Mini)**][4_256]|[**4/512 (Small)**][4_512]|[4/768]|
-| **L=6**  |[6/128]|[6/256]|[6/512]|[6/768]|
-| **L=8**  |[8/128]|[8/256]|[**8/512 (Medium)**][8_512]|[8/768]|
-| **L=10** |[10/128]|[10/256]|[10/512]|[10/768]|
-| **L=12** |[12/128]|[12/256]|[12/512]|[**12/768 (Base)**][12_768]|
+|          |           H=128           |           H=256           |            H=512            |            H=768            |
+| -------- | :-----------------------: | :-----------------------: | :-------------------------: | :-------------------------: |
+| **L=2**  | [**2/128 (Tiny)**][2_128] |          [2/256]          |           [2/512]           |           [2/768]           |
+| **L=4**  |          [4/128]          | [**4/256 (Mini)**][4_256] | [**4/512 (Small)**][4_512]  |           [4/768]           |
+| **L=6**  |          [6/128]          |          [6/256]          |           [6/512]           |           [6/768]           |
+| **L=8**  |          [8/128]          |          [8/256]          | [**8/512 (Medium)**][8_512] |           [8/768]           |
+| **L=10** |         [10/128]          |         [10/256]          |          [10/512]           |          [10/768]           |
+| **L=12** |         [12/128]          |         [12/256]          |          [12/512]           | [**12/768 (Base)**][12_768] |
 
 Here are scores on the devlopment set of six Chinese tasks:
 
-|Model|Score|douban|chnsenticorp|lcqmc|tnews(CLUE)|iflytek(CLUE)|ocnli(CLUE)|
-|---|:---:|:---:|:---:|:---:|:---:|:---:|:---:|
-|RoBERTa-Tiny|72.3|83.0|91.4|81.8|62.0|55.0|60.3|
-|RoBERTa-Mini|75.7|84.8|93.7|86.1|63.9|58.3|67.4|
-|RoBERTa-Small|76.8|86.5|93.4|86.5|65.1|59.4|69.7|
-|RoBERTa-Medium|77.8|87.6|94.8|88.1|65.6|59.5|71.2|
-|RoBERTa-Base|79.5|89.1|95.2|89.2|67.0|60.9|75.5|
+| Model          | Score | douban | chnsenticorp | lcqmc | tnews(CLUE) | iflytek(CLUE) | ocnli(CLUE) |
+| -------------- | :---: | :----: | :----------: | :---: | :---------: | :-----------: | :---------: |
+| RoBERTa-Tiny   | 72.3  |  83.0  |     91.4     | 81.8  |    62.0     |     55.0      |    60.3     |
+| RoBERTa-Mini   | 75.7  |  84.8  |     93.7     | 86.1  |    63.9     |     58.3      |    67.4     |
+| RoBERTa-Small  | 76.8  |  86.5  |     93.4     | 86.5  |    65.1     |     59.4      |    69.7     |
+| RoBERTa-Medium | 77.8  |  87.6  |     94.8     | 88.1  |    65.6     |     59.5      |    71.2     |
+| RoBERTa-Base   | 79.5  |  89.1  |     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:
 
-For each task, we selected the best fine-tuning hyperparameters from the lists below:
 - epochs: 3, 5, 8
 - batch sizes: 32, 64
 - learning rates: 3e-5, 1e-4, 3e-4
@@ -96,7 +99,7 @@ output = model(encoded_input)
 
 ## Training data
 
-CLUECorpusSmall 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.
+[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
 
@@ -105,41 +108,54 @@ Models are pre-trained by [UER-py](https://github.com/dbiir/UER-py/) on [Tencent
 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 --target mlm
+                      --dataset_path cluecorpussmall_seq128_dataset.pt \
+                      --processes_num 32 --seq_length 128 \
+                      --dynamic_masking --target 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 \
-					--tie_weights --embedding word_pos_seg --encoder transformer --mask fully_visible --target mlm
+                    --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 \
+                    --tie_weights --embedding word_pos_seg --encoder transformer --mask fully_visible --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 --target mlm
+                      --dataset_path cluecorpussmall_seq512_dataset.pt \
+                      --processes_num 32 --seq_length 512 \
+                      --dynamic_masking --target mlm
 ```
+
 ```
 python3 pretrain.py --dataset_path cluecorpussmall_seq512_dataset.pt \
                     --pretrained_model_path models/cluecorpussmall_roberta_medium_seq128_model.bin-1000000 \
-					--vocab_path models/google_zh_vocab.txt \
-					--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 \
-					--tie_weights --embedding word_pos_seg --encoder transformer --mask fully_visible --target mlm
+                    --vocab_path models/google_zh_vocab.txt \
+                    --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 \
+                    --tie_weights --embedding word_pos_seg --encoder transformer --mask fully_visible --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 --target mlm
 ```
 
 ### BibTeX entry and citation info
@@ -158,5 +174,4 @@ python3 pretrain.py --dataset_path cluecorpussmall_seq512_dataset.pt \
 [4_256]: https://huggingface.co/uer/chinese_roberta_L-4_H-256
 [4_512]: https://huggingface.co/uer/chinese_roberta_L-4_H-512
 [8_512]: https://huggingface.co/uer/chinese_roberta_L-8_H-512
-[12_768]: https://huggingface.co/uer/chinese_roberta_L-12_H-768
-
+[12_768]: https://huggingface.co/uer/chinese_roberta_L-12_H-768