Hello, thanks for your excellent work! Is this model theoretically capable of handling infinitely long videos?