mxnet下手动下载FashionMNIST到本地之后进行读取

1. 数据准备
   直接从github上下载数据:

   

   
   将数据放置在~/.mxnet/datasets/fashion-mnist/文件夹下。
2. 获取sha
   个人觉得最简单的办法是修改：/path/mxnet/gluon/utils.py的download函数

    if not verify_ssl:
        warnings.warn(
            'Unverified HTTPS request is being made (verify_ssl=False). '
            'Adding certificate verification is strongly advised.')
    
    sha1 = hashlib.sha1()
    with open(fname, 'rb') as f:
        while True:
            data = f.read(1048576)
            if not data:
                break
            sha1.update(data)
    print('hash name', sha1.hexdigest())

将输出的hash值填入到/path/mxnet/gluon/data/vision/datasets.py的FashionMNIST函数中即可：

        self._train_data = ('train-images-idx3-ubyte.gz',
                            '95978b76b6897f6ca69a25145d01716efb615989')
        self._train_label = ('train-labels-idx1-ubyte.gz',
                             '09814cfef5a041118ceace42f8dae995319d331a')
        self._test_data = ('t10k-images-idx3-ubyte.gz',
                           '5edda96c6d8c36ff915115a0e8136d370a021576')
        self._test_label = ('t10k-labels-idx1-ubyte.gz',
                            '9caad14e1aff9adac77d3744963212d36af15bee')