Visual Gradio Improvements

app.py

@@ -18,7 +18,7 @@ logger = logging.getLogger(__name__)
 
 # Paths to the static image and GIF
 README_IMAGE_PATH = os.path.join("figs", "saliencies-merit-dataset.png")
-GIF_PATH = os.path.join("figs", "demo_samples.gif")
+GIF_PATH = os.path.join("figs", "demo-samples.gif")
 
 # Global variables for Donut model, processor, and dataset
 donut_model = None
@@ -134,12 +134,12 @@ if __name__ == "__main__":
     with gr.Blocks() as demo:
         gr.Markdown("# Saliency Maps with the MERIT Dataset 🎒📃🏆")
 
-        gr.Image(value=README_IMAGE_PATH, label="Example Document")
+        gr.Image(value=README_IMAGE_PATH, width=400, height=300)
 
         with gr.Tab("Introduction"):
             gr.Markdown(
                 """
-            ## Welcome to Saliency Maps with the [MERIT Dataset](https://huggingface.co/datasets/de-Rodrigo/merit)
+            ## Welcome to Saliency Maps with the [MERIT Dataset](https://huggingface.co/datasets/de-Rodrigo/merit 🎒📃🏆)
 
             This space demonstrates the capabilities of different Vision Language models 
             for document understanding tasks.
@@ -151,7 +151,7 @@ if __name__ == "__main__":
             """
             )
 
-            gr.Image(value=GIF_PATH, label="Document Understanding Process")
+            gr.Image(value=GIF_PATH, label="Dataset samples you can process")
 
         with gr.Tab("Try It Yourself"):
             gr.Markdown(