cifar10数据集是否少一个参数， 决定是读取训练数据还是测试数据？ about mindspore

[DouSong]

For example, you can setting param training=True or training=False in function vgg_create_dataset

You can visit this page to get more information

[leo038]

ok，in this way, I have to put train data and valid data in different dir. This is not a good way because the standard cifar10 dataset put the train data and valid data in the same dir. Like the cirfar10 dataset in other frameworks, this work should be done in de.Cifar10Dataset, rather than the users themselves.

[liucunwei321]

Good advice
At first, we are considering the unification of API parameters. API parameter unification is also a way of user experience.
If some API user experience is not good, we will improve it.
Thanks