Update app.py

change keep model

app.py

@@ -39,12 +39,13 @@ tokenizer = AutoTokenizer.from_pretrained(model_id, token=huggingface_token)
 #print(model_id,device,dtype)
 histories = []
 
-
-def generate_text(messages):
-    model = AutoModelForCausalLM.from_pretrained(
+model = AutoModelForCausalLM.from_pretrained(
                 model_id, token=huggingface_token ,torch_dtype=dtype,device_map=device
             )
-    model.to(device)
+model.to(device)
+
+def generate_text(messages):
+
     question = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
     question = tokenizer(question, return_tensors="pt").to(device)