Update app.py

app.py

@@ -223,14 +223,15 @@ with gr.Blocks(css=css) as demo :
         <h2 style="text-align: center;">Soft Video Understanding</h2>
         <p style="text-align: center;">
             An experiment to try to achieve what i call "soft video understanding" with open-source available models. <br />
-            We use moondream1 to caption extracted frames, salmonn to analyze extracted audio, then send visual and audio details to Zephyr which is instructed to resume what it understood.
-            Instructions prompt is available for further discussion with the Community. Note that audio is crucial for better overall vision. Video longer than 10 seconds will be cut.
+            We use moondream1 to caption extracted frames, salmonn to analyze extracted audio, then give visual and audio details to Zephyr which is instructed to resume what it understood.<br />
+            Instructions prompt is available for further discussion with the Community. <br />
+            Note that audio is crucial for better overall vision. Video longer than 10 seconds will be cut.
         </p>
         """)
         with gr.Row():
             with gr.Column():
                 video_in = gr.Video(label="Video input")
-                with gr.Accordion("System Instructions", open=False):
+                with gr.Accordion("System Instructions (for your curiosity)", open=False):
                     system_instruction = gr.Markdown(
                         value = standard_sys
                     )