Update src/app/response.py

src/app/response.py

@@ -35,7 +35,7 @@ def caption_image(image: PIL.Image.Image, max_new_tokens: int, sampling: bool) -
             gr.Warning("Please provide an image.")
 
         # Prepare the inputs
-        prompt = "caption en"
+        prompt = "<image>caption en"
         model_inputs = (
             processor(text=prompt, images=image, return_tensors="pt")
             .to(torch.bfloat16)