Update handler.py

handler.py

@@ -52,14 +52,12 @@ def load_models():
         "lllyasviel/control_v11f1e_sd15_tile",
         torch_dtype=torch.float16,
     )
-    controlnet_tile.enable_xformers_memory_efficient_attention(attention_op=MemoryEfficientAttentionFlashAttentionOp)
-
+    
     controlnet_brightness = ControlNetModel.from_pretrained(
         "ioclab/control_v1p_sd15_brightness",
         torch_dtype=torch.float16,
     )
-    controlnet_brightness.enable_xformers_memory_efficient_attention(attention_op=MemoryEfficientAttentionFlashAttentionOp)
-
+    
     pipe = StableDiffusionControlNetPipeline.from_pretrained(
         MODEL_ID,
         controlnet=[
@@ -74,6 +72,7 @@ def load_models():
     pipe.to("cuda")
 
     pipe.scheduler = EulerAncestralDiscreteScheduler.from_config(pipe.scheduler.config)
+    pipe.enable_vae_slicing()
     pipe.enable_xformers_memory_efficient_attention(attention_op=MemoryEfficientAttentionFlashAttentionOp)
     return pipe