Update app.py

app.py

@@ -4,12 +4,6 @@ from transformers import (
       BartTokenizer
 )
 
-model_name = 'unlisboa/bart_qa_assistant'
-tokenizer = BartTokenizer.from_pretrained(model_name)
-device = get_device()
-model = BartForConditionalGeneration.from_pretrained(model_name).to(device)
-model.eval()
-
 def get_device():
     # If there's a GPU available...
     if torch.cuda.is_available():
@@ -24,12 +18,18 @@ def get_device():
         device = torch.device("cpu")
     return device
 
+model_name = 'unlisboa/bart_qa_assistant'
+tokenizer = BartTokenizer.from_pretrained(model_name)
+device = get_device()
+model = BartForConditionalGeneration.from_pretrained(model_name).to(device)
+model.eval()
+
 def run_bart(question, censor):
     print(question, censor)
     
     model_input = tokenizer(question_input, truncation=True, padding=True, return_tensors="pt")
     generated_answers_encoded = model.generate(input_ids=model_input["input_ids"].to(device),
-                                                                                      attention_mask=model_input["attention_mask"].to(device),
+                                                                                                                                                                                                                                                       attention_mask=model_input["attention_mask"].to(device),
                                                                                       #bad_words_ids=bad_words_ids,
                                                                                       force_words_ids=None,
                                                                                       min_length=1,