New ModelArts Version with New Features

ModelArts is a one-stop development platform for AI developers. With distributed training, automated model building, and model deployment, ModelArts helps AI developers build models quickly and manage the AI development lifecycle.

This new major version upgrade includes following new features and bug fixes.

New Features:

• Online services support rolling upgrades.
• Allows training jobs with customized images to use images shared by others.
• The development environment supports the new OS dedicated resource pool.
• The inference service supports the new OS dedicated resource pool.
• The training job supports the new OS dedicated resource pool.
• Online services support HTTPS and Websockets protocols. You can open the websocket switch when creating an online service.

• Online services support CLI script health check. You can do the health check when creating an AI Application and choose model source as “container image”.

• The new version of the OS dedicated resource pool is supported.
• The new version of the OS dedicated resource pool supports job priority queuing.
• The new version of the OS dedicated resource pool supports node inspection and alarm reporting.

Fixed Features:

• Starting the Online Inference Service in the Stopped State Fails Occasionally
• When the OS dedicated resource pool is created, a training job fails to be delivered and remains in the waiting state.
• Fixed an issue where the job list interface fails because pool Info is occasionally lost.

Further information can be found in the ModelArts area of the Help Centers