Update app.py

app.py

@@ -22,7 +22,8 @@ def generate_response(passage: str, question: str) -> str:
     response = model.generate(**inputs, max_new_tokens=100)
     
     response = tokenizer.batch_decode(response, skip_special_tokens=True)[0]
-    
+
+    response = response[len(message[0]):].strip().split('\n')[0]
     
     return response