A processor is a package that contains online prediction logic, including model loading logic and request prediction logic. If the official processors provided by Elastic Algorithm Service (EAS) cannot meet the model deployment requirements, you can create a custom processor based on the processor development standard.

Prerequisites

To ensure the model service security, you cannot use custom processors in shared resource groups. Therefore, a dedicated resource group must be purchased before you use a custom processor to deploy a model service. For more information about how to purchase or create a dedicated resource group, see Purchase a dedicated resource group.

Custom processor development manuals

The following programming languages can be used to develop custom processors:

Deploy a model by using a custom processor

We recommend that you debug a custom processor before you use it to deploy a model. You can deploy a model by uploading it in the Machine Learning Platform for AI (PAI) console or by using the EASCMD client.
  • Upload a model in the console

    Set Resource Group Type to Dedicated Resource Group and Processor Type to Self-definition processor. For more information, see Upload and deploy models in the console.

  • Deploy a model by using EASCMD

    During service deployment, set the resource parameter to the ID of the purchased dedicated resource group. For more information, see Use the EASCMD client.