mindspore.dataset.Dataset.repeat

mindspore.dataset.Dataset.repeat(count=None)[源代码]

重复此数据集 count 次。如果 countNone-1 ,则无限重复迭代。

说明

repeat和batch的顺序反映了batch的数量。建议:repeat操作在batch操作之后使用。

参数:
  • count (int) - 数据集重复的次数。默认值: None

返回:

Dataset,应用了上述操作的新数据集对象。

样例:

>>> import mindspore.dataset as ds
>>>
>>> # Create a dataset with 10 elements
>>> dataset = ds.GeneratorDataset([i for i in range(10)], "column1")
>>> ori_size = dataset.get_dataset_size()
>>>
>>> # Repeat the dataset 50 times.
>>> dataset = dataset.repeat(50)
>>> repeated_size = dataset.get_dataset_size()
>>> print("ori_size", ori_size, ", repeated_size", repeated_size)
ori_size 10 , repeated_size 500
>>>
>>> # Since the original dataset size is less than batch_size, thus no data is returned
>>> dataset1 = ds.GeneratorDataset([i for i in range(10)], "column1")
>>> dataset1 = dataset1.batch(batch_size=20, drop_remainder=True)
>>> dataset1 = dataset1.repeat(6)
>>>
>>> # Repeat the original dataset to 60 elements, thus 3 batches are returned
>>> dataset2 = ds.GeneratorDataset([i for i in range(10)], "column1")
>>> dataset2 = dataset2.repeat(6)
>>> dataset2 = dataset2.batch(batch_size=20, drop_remainder=True)
>>> print("dataset1 size", dataset1.get_dataset_size(), ", dataset2 size", dataset2.get_dataset_size())
dataset1 size 0 , dataset2 size 3