Update pipeline.py

pipeline.py

@@ -50,21 +50,12 @@ class SuperDiffSDXLPipeline(DiffusionPipeline, ConfigMixin):
         """
         super().__init__()
         device = "cuda" if torch.cuda.is_available() else "cpu"
-        print("decice", device)
 
         vae.to(device)
         unet.to(device)
         text_encoder.to(device)
         text_encoder_2.to(device)
-        #dtype = torch.float16
-        #vae = AutoencoderKL.from_pretrained(model_path, subfolder="vae").to(device)
-        #tokenizer = CLIPTokenizer.from_pretrained(model_path, subfolder="tokenizer")
-        #tokenizer_2 = CLIPTokenizer.from_pretrained(model_path, subfolder="tokenizer_2")
-        #text_encoder = CLIPTextModel.from_pretrained(model_path, subfolder="text_encoder").to(device, dtype=dtype)
-        #text_encoder_2 = CLIPTextModelWithProjection.from_pretrained(model_path, subfolder="text_encoder_2").to(device, dtype=dtype)
-        #unet = UNet2DConditionModel.from_pretrained(model_path, subfolder="unet").to(device, dtype=dtype)
-        #vae.eval()
-        #unet.eval()
+
 
         self.register_modules(unet=unet,
                               vae=vae,
@@ -73,9 +64,8 @@ class SuperDiffSDXLPipeline(DiffusionPipeline, ConfigMixin):
                               tokenizer=tokenizer,
                               tokenizer_2=tokenizer_2,
                               )
-        print("decice2", self.device)
+        
     def prepare_prompt_input(self, prompt_o, prompt_b, batch_size, height, width):
-        print("self.device", self.device)
         text_input = self.tokenizer(prompt_o* batch_size, padding="max_length", max_length=self.tokenizer.model_max_length, truncation=True, return_tensors="pt")
         text_input_2 = self.tokenizer_2(prompt_o* batch_size, padding="max_length", max_length=self.tokenizer_2.model_max_length, truncation=True, return_tensors="pt")
         with torch.no_grad():