Help Center/ ModelArts/ Service Overview/ ModelArts and Other Services
Updated on 2025-02-07 GMT+08:00
View PDF
Share

ModelArts and Other Services

IAM

ModelArts uses Identity and Access Management (IAM) for authentication and authorization. For more information about IAM, see the Identity and Access Management Documentation.

OBS

ModelArts uses Object Storage Service (OBS) to securely and reliably store data and models at low costs. For more information about OBS, see Object Storage Service Documentation.

Table 1 Relationship between ModelArts and OBS

Function

Task

Relationship

ExeML

Data labeling

The data labeled on ModelArts is stored in OBS.

Auto training

After a training job is complete, the generated model is stored in OBS.

Model deployment

ModelArts deploys models stored in OBS as real-time services.

AI development lifecycle

Data management
  • Datasets are stored in OBS.
  • Dataset labeling information is stored in OBS.
  • Data can be imported from OBS.

Development environment

Data or code files in a notebook instance are stored in OBS.

Model training
  • The datasets used by training jobs are stored in OBS.
  • The running scripts for training jobs are stored in OBS.
  • The models generated by training jobs are stored in the specified OBS paths.
  • The run logs of training jobs are stored in the specified OBS paths.

Model management

After a training job is completed, the generated model is stored in OBS. You can import the model from OBS.

Model deployment

Models stored in OBS can be deployed as services.

System management

N/A

ModelArts is authorized to access OBS (using an agency or access key) so that ModelArts can use OBS to store data and create notebook instances.

EVS

ModelArts uses Elastic Volume Service (EVS) to store created notebook instances. For details, see Elastic Volume Service User Guide.

CCE

ModelArts deploys models as real-time services using Cloud Container Engine (CCE), which supports high concurrency and auto scaling. For more information about CCE, see Cloud Container Engine User Guide.

SWR

To use an AI framework that is not supported by ModelArts, use Software Repository for Container (SWR) to customize an image and import the image to ModelArts for training or inference. For details about SWR, see Software Repository for Container User Guide.

Cloud Eye

ModelArts uses Cloud Eye to monitor real-time services and model loads in real time and send alarms and notifications automatically. For details about Cloud Eye, see Cloud Eye User Guide.