agiats commited on
Commit
c34afbe
1 Parent(s): be1a3db

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +9 -6
app.py CHANGED
@@ -8,7 +8,7 @@ from sudachipy import dictionary
8
  from sudachipy import tokenizer as sudachi_tokenizer
9
  from transformers import AutoModelForCausalLM, PreTrainedTokenizer, T5Tokenizer
10
 
11
- # trained model
12
  model_dir = Path(__file__).parents[0] / "model"
13
  device = torch.device("cuda:0") if torch.cuda.is_available() else torch.device("cpu")
14
  tokenizer = T5Tokenizer.from_pretrained(model_dir)
@@ -120,7 +120,9 @@ def create_highlighted_text(
120
  if mean_surprisal is None:
121
  highlighted_text = "<h2><b>" + label + "</b></h2>"
122
  else:
123
- highlighted_text = "<h2><b>" + label + f"</b>(サプライザル平均値: {mean_surprisal:.3f})</h2>"
 
 
124
  for token, score in tokens2scores:
125
  highlighted_text += highlight_token(token, score)
126
  return highlighted_text
@@ -168,6 +170,7 @@ def main(input_text: str) -> Tuple[str, str, str]:
168
  offsets = calc_offsets(sudachi_tokenize(input_text))
169
  tokens2surprisal = aggregate_surprisals_by_offset(char2surprisal, offsets)
170
  tokens2surprisal = normalize_surprisals(tokens2surprisal)
 
171
  highlighted_text = create_highlighted_text(
172
  "学習後モデル", tokens2surprisal, mean_surprisal
173
  )
@@ -200,13 +203,13 @@ def main(input_text: str) -> Tuple[str, str, str]:
200
  if __name__ == "__main__":
201
  demo = gr.Interface(
202
  fn=main,
203
- title="文章の読みやすさを自動評価する言語処理システム",
204
- description="テキストを入力すると、読みにくさに応じてハイライトされて出力されます。",
205
  show_label=True,
206
  inputs=gr.Textbox(
207
  lines=5,
208
- label="テキスト",
209
- placeholder="ここにテキストを入力してください。",
210
  ),
211
  outputs=[
212
  gr.HTML(label="学習前モデル", show_label=True),
 
8
  from sudachipy import tokenizer as sudachi_tokenizer
9
  from transformers import AutoModelForCausalLM, PreTrainedTokenizer, T5Tokenizer
10
 
11
+
12
  model_dir = Path(__file__).parents[0] / "model"
13
  device = torch.device("cuda:0") if torch.cuda.is_available() else torch.device("cpu")
14
  tokenizer = T5Tokenizer.from_pretrained(model_dir)
 
120
  if mean_surprisal is None:
121
  highlighted_text = "<h2><b>" + label + "</b></h2>"
122
  else:
123
+ highlighted_text = (
124
+ "<h2><b>" + label + f"</b>(サプライザル平均値: {mean_surprisal:.3f})</h2>"
125
+ )
126
  for token, score in tokens2scores:
127
  highlighted_text += highlight_token(token, score)
128
  return highlighted_text
 
170
  offsets = calc_offsets(sudachi_tokenize(input_text))
171
  tokens2surprisal = aggregate_surprisals_by_offset(char2surprisal, offsets)
172
  tokens2surprisal = normalize_surprisals(tokens2surprisal)
173
+
174
  highlighted_text = create_highlighted_text(
175
  "学習後モデル", tokens2surprisal, mean_surprisal
176
  )
 
203
  if __name__ == "__main__":
204
  demo = gr.Interface(
205
  fn=main,
206
+ title="文章の読みやすさを自動評価するAI",
207
+ description="文章を入力すると、読みづらい表現は赤く、読みやすい表現は青くハイライトされて出力されます。",
208
  show_label=True,
209
  inputs=gr.Textbox(
210
  lines=5,
211
+ label="文章",
212
+ placeholder="ここに文章を入力してください。",
213
  ),
214
  outputs=[
215
  gr.HTML(label="学習前モデル", show_label=True),