shibing624 commited on
Commit
fdab1ea
1 Parent(s): 04a455f

update demo.

Browse files
Files changed (1) hide show
  1. app.py +15 -19
app.py CHANGED
@@ -7,9 +7,9 @@ tokenizer = BertTokenizer.from_pretrained("shibing624/macbert4csc-base-chinese")
7
  model = BertForMaskedLM.from_pretrained("shibing624/macbert4csc-base-chinese")
8
 
9
 
10
- def ai_text(texts):
11
  with torch.no_grad():
12
- outputs = model(**tokenizer(texts, padding=True, return_tensors='pt'))
13
 
14
  def get_errors(corrected_text, origin_text):
15
  sub_details = []
@@ -29,30 +29,26 @@ def ai_text(texts):
29
  sub_details = sorted(sub_details, key=operator.itemgetter(2))
30
  return corrected_text, sub_details
31
 
32
- result = []
33
- for ids, text in zip(outputs.logits, texts):
34
- _text = tokenizer.decode(torch.argmax(ids, dim=-1), skip_special_tokens=True).replace(' ', '')
35
- corrected_text = _text[:len(text)]
36
- corrected_text, details = get_errors(corrected_text, text)
37
- print(text, ' => ', corrected_text, details)
38
- result.append((corrected_text, details))
39
- print(result)
40
- return result
41
 
42
 
43
  if __name__ == '__main__':
44
- print(ai_text(['少先队员因该为老人让坐']))
45
 
46
  examples = [
47
- [['真麻烦你了。希望你们好好的跳无']],
48
- [['少先队员因该为老人让坐']],
49
- [['机七学习是人工智能领遇最能体现智能的一个分知']],
50
- [['今天心情很好']],
51
- [['他法语说的很好,的语也不错']],
52
- [['他们的吵翻很不错,再说他们做的咖喱鸡也好吃']],
53
  ]
54
 
55
  output_text = gr.outputs.Textbox()
56
- gr.Interface(ai_text, "textbox", output_text, title="Chinese Text Correction shibing624/macbert4csc-base-chinese",
57
  description="Copy or input error Chinese text. Submit and the machine will correct text.",
58
  examples=examples).launch()
 
7
  model = BertForMaskedLM.from_pretrained("shibing624/macbert4csc-base-chinese")
8
 
9
 
10
+ def ai_text(text):
11
  with torch.no_grad():
12
+ outputs = model(**tokenizer(text, padding=True, return_tensors='pt'))
13
 
14
  def get_errors(corrected_text, origin_text):
15
  sub_details = []
 
29
  sub_details = sorted(sub_details, key=operator.itemgetter(2))
30
  return corrected_text, sub_details
31
 
32
+ _text = tokenizer.decode(torch.argmax(outputs.logits, dim=-1), skip_special_tokens=True).replace(' ', '')
33
+ corrected_text = _text[:len(text)]
34
+ corrected_text, details = get_errors(corrected_text, text)
35
+ print(text, ' => ', corrected_text, details)
36
+ return corrected_text, details
 
 
 
 
37
 
38
 
39
  if __name__ == '__main__':
40
+ print(ai_text('少先队员因该为老人让坐'))
41
 
42
  examples = [
43
+ ['真麻烦你了。希望你们好好的跳无'],
44
+ ['少先队员因该为老人让坐'],
45
+ ['机七学习是人工智能领遇最能体现智能的一个分知'],
46
+ ['今天心情很好'],
47
+ ['他法语说的很好,的语也不错'],
48
+ ['他们的吵翻很不错,再说他们做的咖喱鸡也好吃'],
49
  ]
50
 
51
  output_text = gr.outputs.Textbox()
52
+ gr.Interface(ai_text, "textbox", "textbox", title="Chinese Text Correction shibing624/macbert4csc-base-chinese",
53
  description="Copy or input error Chinese text. Submit and the machine will correct text.",
54
  examples=examples).launch()