raj-tomar001 commited on
Commit
cdbcd6b
1 Parent(s): a4ddd85

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +12 -17
app.py CHANGED
@@ -1,18 +1,18 @@
1
  import gradio as gr
2
- from transformers import DebertaTokenizer, DebertaForSequenceClassification
3
  from transformers import pipeline
4
  import json
5
 
6
 
7
- save_path_abstract = './fine-tuned-deberta'
8
- model_abstract = DebertaForSequenceClassification.from_pretrained(save_path_abstract)
9
- tokenizer_abstract = DebertaTokenizer.from_pretrained(save_path_abstract)
10
 
11
  classifier_abstract = pipeline('text-classification', model=model_abstract, tokenizer=tokenizer_abstract)
12
 
13
- save_path_essay = './fine-tuned-deberta'
14
- model_essay = DebertaForSequenceClassification.from_pretrained(save_path_essay)
15
- tokenizer_essay = DebertaTokenizer.from_pretrained(save_path_essay)
16
 
17
  classifier_essay = pipeline('text-classification', model=model_essay, tokenizer=tokenizer_essay)
18
 
@@ -36,18 +36,13 @@ def process_result_detection_tab(text):
36
  'Human Written, Machine Polished': float: the probability that the text is human written and machine polished
37
  '''
38
  mapping = {'llm': 'Machine Generated', 'human':'Human Written', 'machine-humanized': 'Machine Written, Machine Humanized', 'machine-polished': 'Human Written, Machine Polished'}
 
 
39
 
40
- # Initialize scores for all classes
41
- final_results = {label: 0.0 for label in mapping.values()}
42
-
43
- # Add scores from classifier_abstract
44
- if result['label'] in mapping:
45
- final_results[mapping[result['label']]] += 0.5 * result['score']
46
-
47
- # Add scores from classifier_essay
48
- if result_r['label'] in mapping:
49
- final_results[mapping[result_r['label']]] += 0.5 * result_r['score']
50
 
 
51
  print(final_results)
52
  return final_results
53
 
 
1
  import gradio as gr
2
+ from transformers import DebertaTokenizer, DebertaForSequenceClassification, DistilBertTokenizer, DistilBertForSequenceClassification
3
  from transformers import pipeline
4
  import json
5
 
6
 
7
+ save_path_abstract = './fine-tuned-distillberta'
8
+ model_abstract = DistilBertForSequenceClassification.from_pretrained(save_path_abstract)
9
+ tokenizer_abstract = DistilBertTokenizer.from_pretrained(save_path_abstract)
10
 
11
  classifier_abstract = pipeline('text-classification', model=model_abstract, tokenizer=tokenizer_abstract)
12
 
13
+ save_path_essay = './fine-tuned-distillberta'
14
+ model_essay = DistilBertForSequenceClassification.from_pretrained(save_path_essay)
15
+ tokenizer_essay = DistilBertTokenizer.from_pretrained(save_path_essay)
16
 
17
  classifier_essay = pipeline('text-classification', model=model_essay, tokenizer=tokenizer_essay)
18
 
 
36
  'Human Written, Machine Polished': float: the probability that the text is human written and machine polished
37
  '''
38
  mapping = {'llm': 'Machine Generated', 'human':'Human Written', 'machine-humanized': 'Machine Written, Machine Humanized', 'machine-polished': 'Human Written, Machine Polished'}
39
+ result = classifier_abstract(text)
40
+ result_r = classifier_essay(text)
41
 
42
+ labels = [mapping[x['label']] for x in result]
43
+ scores = list(0.5 * np.array([x['score'] for x in result]) + 0.5 * np.array([x['score'] for x in result_r]))
 
 
 
 
 
 
 
 
44
 
45
+ final_results = dict(zip(labels, scores))
46
  print(final_results)
47
  return final_results
48