Update handler

handler.py

@@ -20,11 +20,11 @@ class EndpointHandler():
     def __init__(self, path=""):
         
         self.processor = Blip2Processor.from_pretrained(path)
-        self.model = Blip2ForConditionalGeneration.from_pretrained(path, load_in_8bit=True, device_map="auto")
+        self.model = Blip2ForConditionalGeneration.from_pretrained(path, torch_dtype=torch.float16 , device_map="auto")
 
         self.device = "cuda" if torch.cuda.is_available() else "cpu"
 
-        # self.model.to(self.device)
+        self.model.to(self.device)
 
         logging.info('Model moved to device-' + self.device)