Update app.py

app.py

@@ -98,7 +98,7 @@ def mode_load(path):
 
 @spaces.GPU()
 def stream_chat(message, history: list, temperature: float, max_length: int, top_p: float, top_k: int, penalty: float):
-    if len(history)==0:
+    if model=="":
         model = AutoModelForCausalLM.from_pretrained(
             MODEL_ID,
             torch_dtype=torch.bfloat16,
@@ -246,4 +246,5 @@ with gr.Blocks(css=CSS, theme="soft", fill_height=True) as demo:
     gr.Examples(EXAMPLES, [chat_input])
 
 if __name__ == "__main__":
+    model=""
     demo.queue(api_open=False).launch(show_api=False, share=False, )#server_name="0.0.0.0", )