[Doc] Fix VLM prompt placeholder sample bug (#9170)

docs/source/models/vlm.rst

@@ -116,7 +116,7 @@ Instead of passing in a single image, you can pass in a list of images.
 .. code-block:: python
 
     # Refer to the HuggingFace repo for the correct format to use
-    prompt = "<|user|>\n<image_1>\n<image_2>\nWhat is the content of each image?<|end|>\n<|assistant|>\n"
+    prompt = "<|user|>\n<|image_1|>\n<|image_2|>\nWhat is the content of each image?<|end|>\n<|assistant|>\n"
 
     # Load the images using PIL.Image
     image1 = PIL.Image.open(...)