global model, tokenizer

app.py

@@ -62,6 +62,9 @@ def generate(
     top_k: int = 50,
     repetition_penalty: float = 1.2,
 ) -> Iterator[str]:
+    global model, tokenizer
+    if model is None or tokenizer is None:
+        load_model()
     conversation = []
     if system_prompt:
         conversation.append({"role": "system", "content": system_prompt})