Update app.py

app.py

@@ -20,7 +20,6 @@ model = LlavaNextVideoForConditionalGeneration.from_pretrained(
     device_map='auto'
 )
 
-@spaces.GPU
 def read_video_pyav(container, indices):
     '''
     Decode the video with PyAV decoder.
@@ -73,20 +72,19 @@ def process_video(video_file, question_parts):
     
     return generated_text.split("ASSISTANT: ", 1)[-1].strip()
 
-@spaces.GPU
-def process_videos(video_files, question_parts):
+def process_videos(video_files, question):
     """Processes multiple videos and answers a single question for each."""
     answers = []
     for video_file in video_files:
         video_name = os.path.basename(video_file.name)
-        answer = process_video(video_file, question_parts)
+        answer = process_video(video_file, question)
         answers.append(f"**Video: {video_name}**\n{answer}\n")
     return "\n---\n".join(answers)
 
 # Define Gradio interface for multiple videos
 def gradio_interface(videos, indoors_outdoors, standing_sitting, hands_free, interacting_screen):
 
-    question = "Is the subject in the video "
+    question = "Is the subject in the video"
     if indoors_outdoors:
         question += "present indoors or outdoors? "
     if standing_sitting: