Batch获取数据的写法

批量获取数据的方法：

def shuffle(*args):
    "Shuffles list of NumPy arrays in unison"
    state = np.random.get_state()
    for array in args:
        np.random.set_state(state)
        np.random.shuffle(array)

def grouper(iter_, n):
    """Collect data into fixed-length chunks or blocks
     grouper('ABCDEFG', 3, 'x') --> ABC DEF Gxx
     from python itertools docs"""
    args = [iter(iter_)] * n
    return zip(*args)

def batches(data, labels, batch_size, randomize=True):
    if len(data) != len(labels):
        raise ValueError('Image data and label data must be same size')
    if batch_size > len(data):
        raise ValueError('Batch size cannot be larger than size of datasets')
    if randomize: 
        shuffle(data, labels)
    for res in zip(grouper(data, batch_size),
                   grouper(labels, batch_size)):
        yield res

for b in batches(list(range(10)), 
                     list(range(100,110)), 
                     3, randomize=True):
    print(b)