Update modeling_mplug_owl2.py

modeling_mplug_owl2.py

@@ -270,7 +270,7 @@ class MPLUGOwl2LlamaForCausalLM(LlamaForCausalLM, MPLUGOwl2MetaForCausalLM):
     def score(self, images, 
               task_: str = "quality",
               input_: str = "image",
-              return_dict=False,
+              return_dict = False,
               image_tensor = None,
              ):
         if not hasattr(self, "weight_tensor"):
@@ -279,8 +279,8 @@ class MPLUGOwl2LlamaForCausalLM(LlamaForCausalLM, MPLUGOwl2MetaForCausalLM):
         if input_ == "image":
             if image_tensor is None:
                 images = [expand2square(img, tuple(int(x*255) for x in self.image_processor.image_mean)) for img in images]
-                input_ids = tokenizer_image_token(prompt, self.tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').unsqueeze(0).to(self.device)
                 image_tensor = self.image_processor.preprocess(images, return_tensors="pt")["pixel_values"].half().to(self.device)
+            input_ids = tokenizer_image_token(prompt, self.tokenizer, IMAGE_TOKEN_INDEX, return_tensors='pt').unsqueeze(0).to(self.device)
             with torch.inference_mode():
                 output_logits = self(input_ids.repeat(image_tensor.shape[0], 1),
                                 images=image_tensor)["logits"][:,-1, self.preferential_ids_]