Update app.py

app.py

@@ -47,6 +47,7 @@ h1 {
 tokenizer = AutoTokenizer.from_pretrained("alfredplpl/sarashina2-7b-it-test")
 model = AutoPeftModelForCausalLM.from_pretrained("alfredplpl/sarashina2-7b-it-test")
 tokenizer.chat_template = "{% for message in messages %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}"
+model=model.to("cuda:0")
 
 @spaces.GPU()
 def chat_llama3_8b(message: str, 
@@ -64,14 +65,12 @@ def chat_llama3_8b(message: str,
     Returns:
         str: The generated response.
     """
-    global model
     conversation = []
     conversation.append({"role": "system", "content": "あなたは優秀なアシスタントです。"})
     for user, assistant in history:
         conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}])
     conversation.append({"role": "user", "content": message})
 
-    model=model.to("cuda:0")
     input_ids = tokenizer.apply_chat_template(conversation, return_tensors="pt")
     
     streamer = TextIteratorStreamer(tokenizer, timeout=10.0, skip_prompt=True, skip_special_tokens=True)