Update README.md
Browse files
README.md
CHANGED
@@ -36,17 +36,19 @@ widget:
|
|
36 |
|
37 |
## 模型信息 Model Information
|
38 |
|
|
|
39 |
|
|
|
40 |
|
41 |
-
|
42 |
|
43 |
-
|
44 |
|
45 |
-
We
|
46 |
|
47 |
| gram | 1-gram | 2-gram | 3-gram | 4-gram |
|
48 |
-
|
|
49 |
-
|
|
50 |
|
51 |
## 使用 Usage
|
52 |
|
@@ -85,7 +87,11 @@ for idx,sentence in enumerate(generation_output.sequences):
|
|
85 |
|
86 |
```
|
87 |
|
88 |
-
|
|
|
|
|
|
|
|
|
89 |
|
90 |
![avatar](https://huggingface.co/IDEA-CCNL/YuyuanQA-GPT2-3.5B/resolve/main/QA-DEMO.png)
|
91 |
|
|
|
36 |
|
37 |
## 模型信息 Model Information
|
38 |
|
39 |
+
问答在自然语言处理领域中反映AI系统的知识水平的重要任务。为了可以在医疗领域中使用强大的问答能力的语言模型,我们基于Yuyuan-GPT2-3.5B,对其使用了10K条医疗的问答对进行微调。我们希望探索一种简单、有效的方式直接实现问答系统而不需要额外的设计,即利用大模型强大的记忆力和理解能力。
|
40 |
|
41 |
+
Question answering (QA) is an important task in the Natural Language Processing to present the knowledge level of AI systems. To provide a language model with powerful QA capability in the medical domain, we fine-tuned Yuyuan-GPT2-3.5B on 10K medical Q&A pairs.
|
42 |
|
43 |
+
### 下游任务 Performance
|
44 |
|
45 |
+
我们测试了该模型在未见过的100条QA对上的表现:
|
46 |
|
47 |
+
We tested the model on 100 unseen QA pairs:
|
48 |
|
49 |
| gram | 1-gram | 2-gram | 3-gram | 4-gram |
|
50 |
+
| :----: | :----: |:----: | :----: | :----: |
|
51 |
+
| blue score | 0.357727 | 0.2713 | 0.22304 | 0.19099 |
|
52 |
|
53 |
## 使用 Usage
|
54 |
|
|
|
87 |
|
88 |
```
|
89 |
|
90 |
+
### 演示 Demo
|
91 |
+
|
92 |
+
我们用该模型做了一个医疗问答演示。将来,我们会将这款产品做成微信小程序与大家见面。
|
93 |
+
|
94 |
+
We made a demo of medical QA system with this model. In the future, we will make this product into a wechat app to meet you.
|
95 |
|
96 |
![avatar](https://huggingface.co/IDEA-CCNL/YuyuanQA-GPT2-3.5B/resolve/main/QA-DEMO.png)
|
97 |
|