Due to limited server storage space, could the author conveniently provide the processed dataset files required for training and testing?