Update app.py

app.py

@@ -32,7 +32,7 @@ def generate_response(prompt):
     # Generate
     generated_ids = model.generate(
         **model_inputs,
-        max_new_tokens=2048,
+        max_new_tokens=100,
         do_sample=True,
         temperature=0.7,
         top_p=0.9