Skip to main content
Tags add key=value identifiers to platform resources, facilitating classification, filtering, and governance in training, inference, image, and other tag-supported modules.

​
Usage recommendations

  • Label keys should represent classification dimensions, such as model, trainingType, and type.
  • Label values should represent specific values, such as test, llm, and production.
  • For the same category of labels, keep key names consistent whenever possible. Avoid using multiple naming conventions such as trainingType, train_type, and taskType.

​
Label validation

The label management page provides the Enable label validation switch. When enabled, modules that support label validation require the use of maintained labels to help standardize label usage. When disabled, label usage rules depend on the actual behavior of each business module. Before enabling label validation, we recommend maintaining commonly used labels and confirming naming conventions with your team to avoid affecting subsequent resource creation or updates.

​
Create tags

  1. Go to Config Management > Label Management.
  2. Click Create.
  3. Enter the label key and value. Tags are typically displayed in the key=value format.
  4. Click Create.
After creation, modules that support label validation can use the label in accordance with the label management rules.

​
Manage tags

After finding the target tag in Config Management > Label Management, you can update or delete it in the Actions column.
  • Update: Modify the tag key or tag value. If the tag is already referenced by resources, confirm before updating whether it will affect resource filtering, statistics, or governance rules.
  • Delete: Delete tags that are no longer needed. Before deletion, confirm that the tag no longer needs to be selected or referenced in business modules. For historical tags already applied to resources, post-deletion handling is subject to the rules of the corresponding module.