Update models/gpt4o.py

models/gpt4o.py

@@ -58,7 +58,7 @@ def extract_all_text_pix2struct(image: Image.Image):
         processor = Pix2StructProcessor.from_pretrained("google/pix2struct-textcaps-base")
         pix2struct_model = Pix2StructForConditionalGeneration.from_pretrained(
             "google/pix2struct-textcaps-base"
-        ).to("cpu")
+        ).to("cuda" if torch.cuda.is_available() else "cpu")
 
     inputs = processor(images=image, return_tensors="pt").to(pix2struct_model.device)
     predictions = pix2struct_model.generate(**inputs, max_new_tokens=512)