added requirements memory speed xformers

handler.py

@@ -119,8 +119,9 @@ class EndpointHandler:
         )
 
         # boosts performance by another 20%
-        # self.pipe.enable_xformers_memory_efficient_attention()
-        # self.pipe.enable_attention_slicing()
+        self.pipe.enable_xformers_memory_efficient_attention()
+        self.pipe.enable_attention_slicing()
+        # may need a requirement in the root with xformer
 
     def load_lora(self, pipeline, lora_path, lora_weight=0.5):
         state_dict = load_file(lora_path)