Reduced max GPU duration

app.py

@@ -51,7 +51,7 @@ model, tokenizer = load_trained_lora_model(
     tasks_config=serve_args.tasks_config
 )
 
-@spaces.GPU(duration=300)
+@spaces.GPU(duration=150)
 def caption_audio(audio_file):
     chunk_audio_files = split_audio(audio_file, CHUNK_LENGTH)
     chunk_captions = []