try to debug

app.py

@@ -181,7 +181,7 @@ def answer(system: str, context: str, question: str,
 
         # --- generate ------------------------------------------------------
         output = chat_model.generate(
-            prompt,
+            **tokens,
             max_new_tokens=512,
             max_length=MAX_PROMPT_TOKENS + 512,
             do_sample=True,