Workspace Management

ModelArts Studio provides flexible and efficient workspace management. You can create multiple workspaces based on different application scenarios, project types, or team requirements. Each workspace is independent, ensuring that assets in the workspace are not affected by other workspaces. This ensures data and resource isolation and security. You can flexibly divide workspaces as required to implement orderly resource management and optimal configuration, maximizing resource utilization in different scenarios. To further optimize resource management, the platform provides multiple role permission systems. You can configure permissions for different roles, from managers to module personnel, to ensure that each user has appropriate access and operation permissions in the specified workspace. This fine-grained permission management mode ensures data security and improves resource utilization.

On the platform, workspace assets refer to all stored resources within the workspace, including data assets and model assets. These assets are the basis for users to develop and manage data on the platform. Centralized storage and unified management enhance operation efficiency while ensuring standardized and secure resource handling.

• Data assets: refer to all datasets published by users on the platform. These datasets are stored in the data asset repository, where you can view detailed information such as format, size, and ratio at any time. The platform also automatically logs the operational history of each dataset, including creation, publishing, and rollout. To simplify management, the platform supports dataset deletion, allowing you to flexibly manage and adjust your datasets. During model training and data analysis, these datasets can be called as needed to ensure data accuracy and security and enhance data asset utilization. Additionally, datasets can be published to and subscribed from the Gallery.
• Model assets: include models used for trials, subscriptions, or trained and published on the platform. These models are stored in the model asset repository for centralized management. You can review all historical versions and operation records of a model to track its evolution. The platform also offers a range of convenient operations, such as model training, compression, and deployment, simplifying the model development and application process. Additionally, the platform provides import and export functionalities, enabling you to migrate Pangu models between different sites, enhancing flexibility and efficiency in model asset sharing and management. Models can also be published to and subscribed from the Gallery.

By centrally managing workspace assets, the platform enables you to efficiently organize and utilize resources while ensuring their security, consistency, and flexibility. This integrated approach guarantees effective resource utilization and intelligent configuration, offering you a more convenient development and management experience.