Update handler.py

handler.py

@@ -9,7 +9,7 @@ class EndpointHandler():
         self.device = "cuda:0" if torch.cuda.is_available() else "cpu"
         self.torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
 
-        self.model = AutoModelForCausalLM.from_pretrained("microsoft/Florence-2-large", torch_dtype=torch_dtype, trust_remote_code=True).to(device)
+        self.model = AutoModelForCausalLM.from_pretrained("microsoft/Florence-2-large", torch_dtype=self.torch_dtype, trust_remote_code=True).to(self.device)
         self.processor = AutoProcessor.from_pretrained("microsoft/Florence-2-large", trust_remote_code=True)