Update app.py

app.py

@@ -107,15 +107,20 @@ def get_answer(image, url, question) -> str:
 
 
 with gr.Blocks() as demo:
+    with gr.Row():
+        gr.Markdown(
+            """
+            ## Donut-OCR-VQA
+            - This demo uses fine-tuned OCR-VQA-Donut model on OCR-VQA-200k to answer questions about images.
+            
+            ## IO description
+            - Input is an image or URL that represents a book cover (recommended) and a question that asks about information on the image.
+            - Output: an answer to the question.
+            """
+        )
+
     with gr.Row():
         with gr.Column():
-            gr.Markdown(
-                """
-                # Donut-OCR-VQA
-                This demo uses fine-tuned OCR-VQA-Donut model on OCR-VQA-200k to answer questions about images.
-                Feel free to try it out!
-                """
-            )
             
             image = gr.Image(shape=(224, 224), type="pil")
             image_url = gr.Textbox(lines=1, label="Image URL", placeholder="Or, paste the image URL here")
@@ -124,14 +129,6 @@ with gr.Blocks() as demo:
             ask = gr.Button(label="Get the answer")
 
         with gr.Column():
-            gr.Markdown(
-                """
-                # IO description
-                - Input is an image or URL that represents a book cover (recommended) and a question that asks about information on the image.
-                - Output: an answer to the question.
-                """
-            )
-            
             answer = gr.Label(label="Answer")
 
     ask.click(get_answer, inputs=[image, image_url, question], outputs=[answer])