Annalyn Ng commited on
Commit
ecfe3e6
1 Parent(s): 93be2eb

add preprocessing to add mask

Browse files
Files changed (1) hide show
  1. app.py +9 -3
app.py CHANGED
@@ -12,12 +12,18 @@ mask_token = tokenizer.mask_token
12
  text = f"雨天,我整个人就便{mask_token}了,不想出外,甚至不想去上课。"
13
  target_word = "懒惰"
14
 
 
 
 
 
15
  def eval_prob(target_word, text):
 
 
16
  # Get index of target_word
17
  idx = tokenizer.encode(target_word)[2]
18
 
19
  # Get logits
20
- inputs = tokenizer(text, return_tensors="pt")
21
  token_logits = model(**inputs).logits
22
 
23
  # Find the location of the MASK and extract its logits
@@ -36,8 +42,8 @@ def eval_prob(target_word, text):
36
  gr.Interface(
37
  fn=eval_prob,
38
  inputs=[
39
- gr.Textbox(placeholder="懒惰"),
40
- gr.Textbox(placeholder=f"雨天,我整个人就便{mask_token}了,不想出外,甚至不想去上课。"),
41
  ],
42
  outputs="text",
43
  title="Chinese Sentence Grading",
 
12
  text = f"雨天,我整个人就便{mask_token}了,不想出外,甚至不想去上课。"
13
  target_word = "懒惰"
14
 
15
+ def add_mask(target_word, text):
16
+ text_mask = text.replace(target_word, mask_token)
17
+ return text_mask
18
+
19
  def eval_prob(target_word, text):
20
+
21
+ text_mask = add_mask(target_word, text)
22
  # Get index of target_word
23
  idx = tokenizer.encode(target_word)[2]
24
 
25
  # Get logits
26
+ inputs = tokenizer(text_mask, return_tensors="pt")
27
  token_logits = model(**inputs).logits
28
 
29
  # Find the location of the MASK and extract its logits
 
42
  gr.Interface(
43
  fn=eval_prob,
44
  inputs=[
45
+ gr.Textbox(placeholder="夸大"),
46
+ gr.Textbox(placeholder=f"我们使用 Generative AI 已经很长时间了,所以他们最近的媒体报道可能被夸大了。"),
47
  ],
48
  outputs="text",
49
  title="Chinese Sentence Grading",