Update app.py

app.py

@@ -160,7 +160,7 @@ with gr.Blocks(title="BPE Visualization Demo") as demo:
             )
         
         with gr.Column(scale=2):
-            gr.Markdown("<p style='font-size:20px;'><span style='color:red;'>If the input text is not included in the image</span>, the attention map will show a lot of noise (the actual response value is very low), since we normalize the attention map according to the relative value.</p>")
+            gr.Markdown("<p style='font-size:20px;'><span style='color:red;'>If the input text is not included in the image</span>, the attention map will show a lot of noise (the actual response value is very low), since we normalize the attention map according to the relative value. Since there are fewer Chinese images in public data than English, <span style='color:red;'>we recommend you use the TokenFD-4096-English-seg version.</span></p>")
 
             with gr.Row():
                 orig_img = gr.Image(label="Original picture", interactive=False)