shibing624 commited on
Commit
d7acc86
1 Parent(s): 29f6c7f

Create app.py

Browse files
Files changed (1) hide show
  1. app.py +51 -0
app.py ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # -*- coding: utf-8 -*-
2
+ """
3
+ @author:XuMing(xuming624@qq.com)
4
+ @description: text similarity example, fine-tuned by CoSENT model
5
+ """
6
+ import gradio as gr
7
+ from similarities import Similarity
8
+
9
+ # 中文句向量模型
10
+ sim_model = Similarity(model_name_or_path='shibing624/text2vec-base-chinese')
11
+
12
+
13
+ def load_file(path):
14
+ with open(path, 'r', encoding='utf-8') as f:
15
+ return f.read().split('\n')
16
+
17
+
18
+ sim_model.add_corpus(load_file('corpus.txt'))
19
+
20
+
21
+ def ai_text(query):
22
+ res = sim_model.most_similar(queries=query, topn=5)
23
+ print(res)
24
+ for q_id, c in res.items():
25
+ print('query:', query)
26
+ print("search top 5:")
27
+ for corpus_id, s in c.items():
28
+ print(f'\t{sim_model.corpus[corpus_id]}: {s:.4f}')
29
+ res_show = '\n'.join(['search top5:'] + [f'text: {sim_model.corpus[corpus_id]} score: {s:.4f}' for corpus_id, s in
30
+ list(res.values())[0].items()])
31
+ return res_show
32
+
33
+
34
+ if __name__ == '__main__':
35
+ examples = [
36
+ ['星巴克被嘲笑了'],
37
+ ['西班牙失业率超过50%'],
38
+ ['她在看书'],
39
+ ['一个人弹琴'],
40
+ ]
41
+ input = gr.inputs.Textbox(lines=2, placeholder="Enter Query")
42
+ output_text = gr.outputs.Textbox()
43
+ gr.Interface(ai_text,
44
+ inputs=[input],
45
+ outputs=[output_text],
46
+ theme="grass",
47
+ title="Chinese Text Semantic Search Model",
48
+ description="Copy or input Chinese text here. Submit and the machine will find the most similarity texts.",
49
+ article="Link to <a href='https://github.com/shibing624/similarities' style='color:blue;' target='_blank\'>Github REPO</a>",
50
+ examples=examples
51
+ ).launch()