Update app.py

app.py

@@ -32,7 +32,7 @@ tokenizer = AutoTokenizer.from_pretrained(model_name)
 
 def generate_answer(question):
     inputs = tokenizer.encode("Question: " + question, return_tensors="pt")
-    outputs = model.generate(inputs, max_length=100, num_return_sequences=1, early_stopping=True)
+    outputs = model.generate(inputs, max_length=100, num_return_sequences=1, early_stopping=True, do_sample=True)
     answer = tokenizer.decode(outputs[0], skip_special_tokens=True)
     return answer