Update app.py

app.py

@@ -181,11 +181,14 @@ if uploaded_file is not None:
             text = pg.get_text()
             sentences = sentence_tokenize(text)
             for sent in sentences:
-              # x = mask_generation(sent)
-              # t5_words = list(set(sent.split(" ")).difference(set(x.split(" "))))
-              
+              x = mask_generation(sent)
+              t5=[]  
+              t5_words = list(set(sent.split("\n")))
+              for i in t5_words:
+                  t5+=i.split(" ")
+              t5_words=set(t5).difference(set(x.split(" "))))        
               entities,words_out = extract_entities(sent)
-              # words_out+=t5_words
+              words_out+=t5_words
               bert_words = words_red_bert(sent)
               new=[]
               for w in words_out: