Update README.md

README.md

@@ -17,9 +17,9 @@ tags:
 
 ## 简介 Brief Introduction
 
-将自然语言理解任务转化为multiple choice任务，并且使用 48 个 NLU 任务进行预训练
+UniMC 核心思想是将自然语言理解任务转化为 multiple choice 任务，并且使用多个 NLU 任务来进行预训练。我们在英文数据集实验结果表明仅含有 2.35 亿参数的 [ALBERT模型](https://huggingface.co/IDEA-CCNL/Erlangshen-UniMC-Albert-235M-English)的zero-shot性能可以超越众多千亿的模型。并在中文测评基准 FewCLUE 和 ZeroCLUE 两个榜单中，13亿的[二郎神](https://huggingface.co/IDEA-CCNL/Erlangshen-UniMC-MegatronBERT-1.3B-Chinese)获得了第一的成绩。
 
-Convert natural language understanding tasks into multiple choice tasks, and use 48 NLU task for pre-training
+The core idea of UniMC is to convert natural language understanding tasks into multiple choice tasks and use multiple NLU tasks for pre-training. Our experimental results on the English dataset show that the zero-shot performance of a [ALBERT](https://huggingface.co/IDEA-CCNL/Erlangshen-UniMC-Albert-235M-English) model with only 235 million parameters can surpass that of many hundreds of billions of models. And in the Chinese evaluation benchmarks FewCLUE and ZeroCLUE two lists, 1.3 billion [Erlangshen](https://huggingface.co/IDEA-CCNL/Erlangshen-UniMC-MegatronBERT-1.3B-Chinese) won the first result.
 
 ## 模型分类 Model Taxonomy
 
@@ -37,14 +37,15 @@ avoiding problems in commonly used large generative models such as FLAN. It not
 
 ### 下游效果 Performance
 
+
 **Few-shot**
 | Model      | eprstmt    | csldcp   | tnews     | iflytek  | ocnli     | bustm     | chid      | csl      | wsc       | Avg       |
 |------------|------------|----------|-----------|----------|-----------|-----------|-----------|----------|-----------|-----------|
-| Finetuning | 65.4       | 35.5     | 49        | 32.8     | 33        | 60.7      | 14.9      | 50       | 55.6      | 44.1      |
-| PET        | 86.7       | 51.7     | 54.5      | 46       | 44        | 56        | 61.2      | 59.4     | 57.5      | 57.44     |
-| LM-BFF     | 85.6       | 54.4     | 53        | 47.1     | 41.6      | 57.6      | 61.2      | 51.7     | 54.7      | 56.32     |
-| P-tuning   | 88.3       | 56       | 54.2      | **57.6** | 41.9      | 60.9      | 59.3      | **62.9** | 58.1      | 59.91     |
-| EFL        | 84.9       | 45       | 52.1      | 42.7     | 66.2      | 71.8      | 30.9      | 56.6     | 53        | 55.91     |
+| [FineTuning](https://arxiv.org/pdf/2107.07498.pdf)-RoBERTa-110M  | 65.4       | 35.5     | 49        | 32.8     | 33        | 60.7      | 14.9      | 50       | 55.6      | 44.1      |
+| [FineTuning](https://arxiv.org/pdf/2107.07498.pdf)-ERNIE1.0-110M | 66.5 | 57   | 516  | 42.1 | 32   | 60.4 | 15    | 60.1 | 50.3 | 48.34 |
+| [PET](https://arxiv.org/pdf/2107.07498.pdf)-ERNIE1.0-110M        | 84   | 59.9 | 56.4 | 50.3 | 38.1 | 58.4 | 40.6  | 61.1 | 58.7 | 56.39 |
+| [P-tuning](https://arxiv.org/pdf/2107.07498.pdf)-ERNIE1.0-110M   | 80.6 | 56.6 | 55.9 | 52.6 | 35.7 | 60.8 | 39.61 | 51.8 | 55.7 | 54.37 |
+| [EFL](https://arxiv.org/pdf/2107.07498.pdf)-ERNIE1.0-110M        | 76.7 | 47.9 | 56.3 | 52.1 | 48.7 | 54.6 | 30.3  | 52.8 | 52.3 | 52.7  |
 | [UniMC-RoBERTa-110M](https://huggingface.co/IDEA-CCNL/Erlangshen-UniMC-RoBERTa-110M-Chinese) | 88.64      | 54.08    | 54.32     | 48.6     | 66.55     | 73.76     | 67.71     | 52.54    | 59.92     | 62.86     |
 | [UniMC-RoBERTa-330M](https://huggingface.co/IDEA-CCNL/Erlangshen-UniMC-RoBERTa-330M-Chinese) | 89.53      | 57.3     | 54.25     | 50       | 70.59     | 77.49     | 78.09     | 55.73    | 65.16     | 66.46     |
 | [UniMC-MegatronBERT-1.3B](https://huggingface.co/IDEA-CCNL/Erlangshen-UniMC-MegatronBERT-1.3B-Chinese) | **89.278** | **60.9** | **57.46** | 52.89    | **76.33** | **80.37** | **90.33** | 61.73    | **79.15** | **72.05** |
@@ -53,10 +54,10 @@ avoiding problems in commonly used large generative models such as FLAN. It not
 
 | Model         | eprstmt   | csldcp    | tnews     | iflytek   | ocnli     | bustm    | chid     | csl      | wsc       | Avg       |
 |---------------|-----------|-----------|-----------|-----------|-----------|----------|----------|----------|-----------|-----------|
-| GPT-zero      | 57.5      | 26.2      | 37        | 19        | 34.4      | 50       | 65.6     | 50.1     | 50.3      | 43.4      |
-| PET-zero      | 85.2      | 12.6      | 26.1      | 26.6      | 40.3      | 50.6     | 57.6     | 52.2     | 54.7      | 45.1      |
-| NSP-BERT      | 86.9      | 47.6      | 51        | 41.6      | 37.4      | 63.4     | 52       | **64.4** | 59.4      | 55.96     |
-| ZeroPrompt    | -         | -         | -         | 16.14     | 46.16     | -        | -        | -        | 47.98     | -         |
+| [GPT](https://arxiv.org/pdf/2107.07498.pdf)-110M      | 57.5      | 26.2      | 37        | 19        | 34.4      | 50       | 65.6     | 50.1     | 50.3      | 43.4      |
+| [PET](https://arxiv.org/pdf/2107.07498.pdf)-RoBERTa-110M      | 85.2      | 12.6      | 26.1      | 26.6      | 40.3      | 50.6     | 57.6     | 52.2     | 54.7      | 45.1      |
+| [NSP-BERT](https://arxiv.org/abs/2109.03564)-110M      | 86.9      | 47.6      | 51        | 41.6      | 37.4      | 63.4     | 52       | **64.4** | 59.4      | 55.96     |
+| [ZeroPrompt](https://arxiv.org/abs/2201.06910)-T5-1.5B    | -         | -         | -         | 16.14     | 46.16     | -        | -        | -        | 47.98     | -         |
 |  Yuan1.0-13B  | 88.13     | 38.99     | 57.47     | 38.82     | 48.13     | 59.38    | 86.14    | 50       | 38.99     | 56.22     |
 | ERNIE3.0-240B | 88.75     | **50.97** | **57.83** | **40.42** | 53.57     | 64.38    | 87.13    | 56.25    | 53.46     | 61.41     |
 | [UniMC-RoBERTa-110M](https://huggingface.co/IDEA-CCNL/Erlangshen-UniMC-RoBERTa-110M-Chinese)    | 86.16     | 31.26     | 46.61     | 26.54     | 66.91     | 73.34    | 66.68    | 50.09    | 53.66     | 55.7      |
@@ -64,7 +65,6 @@ avoiding problems in commonly used large generative models such as FLAN. It not
 | [UniMC-MegatronBERT-1.3B](https://huggingface.co/IDEA-CCNL/Erlangshen-UniMC-MegatronBERT-1.3B-Chinese)     | **88.79** | 42.06     | 55.21     | 33.93     | **75.57** | **79.5** | **89.4** | 50.25    | **66.67** | **64.53** |
 
 
-
 ## 使用 Usage
 ```shell
 git clone https://github.com/IDEA-CCNL/Fengshenbang-LM.git
@@ -75,11 +75,11 @@ pip install --editable .
 
 ```python3
 import argparse
-from fengshen.pipelines.multiplechoice import UniMCPiplines
+from fengshen.pipelines.multiplechoice import UniMCPipelines
 
 
 total_parser = argparse.ArgumentParser("TASK NAME")
-total_parser = UniMCPiplines.piplines_args(total_parser)
+total_parser = UniMCPipelines.piplines_args(total_parser)
 args = total_parser.parse_args()
 args.pretrained_model_path = 'IDEA-CCNL/Erlangshen-UniMC-RoBERTa-110M-Chinese'
 args.learning_rate=2e-5
@@ -87,7 +87,7 @@ args.max_length=512
 args.max_epochs=3
 args.batchsize=8
 args.default_root_dir='./'
-model = UniMCPiplines(args)
+model = UniMCPipelines(args)
 
 train_data = []
 dev_data = []