.

app.py

@@ -61,8 +61,8 @@ def create_ui(examples_path: str):
                         interactive=False
                     )
                     with gr.Accordion("Chain of thought details", open=False):
-                        gr.Markdown(f"### Summary: {example['analysis']['video_description']}")
-                        gr.Markdown(f"### Highlights to search for: {example['analysis']['highlight_types']}")
+                        gr.Markdown(f"### Summary:\n{example['analysis']['video_description']}")
+                        gr.Markdown(f"### Highlights to search for:\n{example['analysis']['highlight_types']}")
 
         gr.Markdown("## Try It Yourself!")
         with gr.Row():
@@ -94,6 +94,14 @@ def create_ui(examples_path: str):
 
         @spaces.GPU
         def on_process(video):
+            # Clear all components when starting new processing
+            yield [
+                "",  # Clear status
+                "",  # Clear video description
+                "",  # Clear highlight types
+                gr.update(value=None, visible=False),  # Clear video
+                gr.update(visible=False)  # Hide accordion
+            ]
             if not video:
                 yield [
                     "Please upload a video",
@@ -129,7 +137,7 @@ def create_ui(examples_path: str):
                 detector = BatchedVideoHighlightDetector(
                     model, 
                     processor, 
-                    batch_size=12
+                    batch_size=8
                 )
 
                 yield [
@@ -141,7 +149,7 @@ def create_ui(examples_path: str):
                 ]
                 
                 video_desc = detector.analyze_video_content(video)
-                formatted_desc = f"### Summary: {video_desc[:500] + '...' if len(video_desc) > 500 else video_desc}"
+                formatted_desc = f"### Summary:\n {video_desc[:500] + '...' if len(video_desc) > 500 else video_desc}"
                 
                 yield [
                     "Determining highlight types...",
@@ -152,7 +160,7 @@ def create_ui(examples_path: str):
                 ]
                 
                 highlights = detector.determine_highlights(video_desc)
-                formatted_highlights = f"### Highlights to search for: {highlights[:500] + '...' if len(highlights) > 500 else highlights}"
+                formatted_highlights = f"### Highlights to search for:\n {highlights[:500] + '...' if len(highlights) > 500 else highlights}"
                 
                 # Get all segments
                 segments = get_fixed_30s_segments(video)
@@ -208,6 +216,11 @@ def create_ui(examples_path: str):
                     gr.update(visible=False),
                     gr.update(visible=False)
                 ]
+            finally:
+                if model is not None:
+                    del model
+                    torch.cuda.empty_cache()
+
 
         process_btn.click(
             on_process,