Update README.md
Browse files
README.md
CHANGED
@@ -43,4 +43,32 @@ The evaluation dataset is in Chinese, and we used the same language model **RoBE
|
|
43 |
|
44 |
|
45 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
46 |
|
|
|
43 |
|
44 |
|
45 |
|
46 |
+
## Uses
|
47 |
+
To use the tool, first install the `promcse` package from [PyPI](https://pypi.org/project/promcse/)
|
48 |
+
```bash
|
49 |
+
pip install promcse
|
50 |
+
```
|
51 |
+
|
52 |
+
After installing the package, you can load our model by two lines of code
|
53 |
+
```python
|
54 |
+
from promcse import PromCSE
|
55 |
+
model = PromCSE("hellonlp/promcse-bert-base-zh", "cls", 10)
|
56 |
+
```
|
57 |
+
|
58 |
+
Then you can use our model for encoding sentences into embeddings
|
59 |
+
```python
|
60 |
+
embeddings = model.encode("武汉是一个美丽的城市。")
|
61 |
+
print(embeddings.shape)
|
62 |
+
#torch.Size([1024])
|
63 |
+
```
|
64 |
+
|
65 |
+
Compute the cosine similarities between two groups of sentences
|
66 |
+
```python
|
67 |
+
sentences_a = ['你好吗']
|
68 |
+
sentences_b = ['你怎么样','我吃了一个苹果','你过的好吗','你还好吗','你',
|
69 |
+
'你好不好','你好不好呢','我不开心','我好开心啊', '你吃饭了吗',
|
70 |
+
'你好吗','你现在好吗','你好个鬼']
|
71 |
+
similarities = model.similarity(sentences_a, sentences_b)
|
72 |
+
print(similarities)
|
73 |
+
```
|
74 |
|