Dataset Management

Creation of datasets and uploading of data

see Data upload.

Bind dataset to container

When creating a container. Binding data allows you to select the dataset you create and bind it to the specified directory in the container.

When binding data. Need to select the directory mapped to the container. As shown in the following figure, take cifar_10 The data has been bound to /input0 Means passing through the container /input0 The directory provides access to the contents of the dataset.

for example cifar10 There are files in the dataset meta.csv, If it is bound to a container /input0 catalogue. So we can go through the path /input0/meta.csv Read the file.

You can also see the specific binding directory of the container in the container details. As shown in the following figure:

Change of type

stay "data set" Under the settings page. Allow changing the type of dataset:

Change it to "Model" after. It will be in "Model" Display in a type of data warehouse.

Deletion of data versions and deletion of the entire dataset

As mentioned earlier. Multiple versions can be included in the dataset. If a specific version is no longer needed. Can be in "edition" On the right side of the list "Delete the current version" Delete at the button. After this version was deleted. The storage resources it occupies will be released.

Of course, the entire dataset can also be permanently deleted if it is no longer needed. After deletion, all versions under this dataset will be deleted. It will no longer occupy users' storage quotas.