deleted the device placement

handler.py

@@ -7,11 +7,6 @@ class EndpointHandler:
         # Load model and tokenizer
         self.tokenizer = AutoTokenizer.from_pretrained(path)
         self.model = AutoModelForCausalLM.from_pretrained(path, device_map="auto", torch_dtype=torch.float16)
-        self.pipeline = TextGenerationPipeline(
-            model=self.model,
-            tokenizer=self.tokenizer,
-            device=0 if torch.cuda.is_available() else -1
-        )
 
     def __call__(self, data: Dict[str, Any]) -> List[Dict[str, str]]:
         prompt_input = data.get("inputs", "")