Update app.py

app.py

@@ -40,9 +40,10 @@ def predict(question: str, option_a: str, option_b: str, option_c: str, option_d
     with torch.no_grad():
         outputs = model.generate(
             **inputs,
-            max_new_tokens=10,
-            temperature=0.7,
-            do_sample=False,
+            max_new_tokens=256,
+            temperature=0.6,
+            top_p=0.9,
+            do_sample=True,
             pad_token_id=tokenizer.eos_token_id
         )