All Products
Search

This Product

    Platform For AI:Install a Pai-Megatron-Patch image

    Document Center

    Platform For AI:Install a Pai-Megatron-Patch image

    Last Updated:Jan 26, 2026

    This topic describes how to install the Pai-Megatron-Patch image in DLC or DSW to accelerate model training.

    Limits

    • Pai-Megatron-Patch requires GPU-accelerated instances.

    • The GPU driver version must be 460.32 or later.

    Procedure

    Install a Pai-Megatron-Patch image in DLC

    Deep Learning Containers (DLC) is a cloud-native deep learning training platform that supports custom images, distributed training, and multiple frameworks.

    DLC lets you load custom images for Pai-Megatron-Patch deployment. After installation, you can run large-scale distributed training on multi-GPU servers.

    Perform the following steps:

    1. Log on to the PAI console.

    2. In the left pane, click Workspace List. On the Workspace List page, click a workspace.

    3. In the left pane, choose Model Development and Training > Deep Learning Containers (DLC), and click Create Job.

    4. Configure the following parameters. For other parameters, see Create a training job.

      • Environment Information: Set Node Image to Image Address, and enter the following address: pai-image-manage-registry.cn-wulanchabu.cr.aliyuncs.com/pai/pytorch-training:2.0-ubuntu20.04-py3.10-cuda11.8-megatron-patch-llm

      • Resource Information:

        • Set Framework to PyTorch.

        • Job Resource: Click image in the Resource Specification column, and select a GPU-accelerated node type and specifications.

      image

      image

    5. Click OK.

    Install a Pai-Megatron-Patch image in DSW

    Data Science Workshop (DSW) is a cloud-based deep learning development environment that integrates JupyterLab and supports custom plug-ins without O&M configuration.

    DSW also supports custom images. After installation, you can debug Pai-Megatron-Patch training acceleration programs.

    Perform the following steps:

    1. Log on to the PAI console.

    2. In the left pane, click Workspace List. On the Workspace List page, click a workspace.

    3. In the left pane, choose Model Development and Training > Data Science Workshop (DSW), and click Create Instance.

    4. Configure the following parameters. For other parameters, see Create a DSW instance.

      • Resource Quota: Select Public Resources (Pay-as-you-go).

      • Resource Specification: Click image and select a GPU-accelerated instance specification.

      • Image: Enter the following address: pai-image-manage-registry.cn-wulanchabu.cr.aliyuncs.com/pai/pytorch-training:2.0-ubuntu20.04-py3.10-cuda11.8-megatron-patch-llm

      image

    5. Click OK to create the DSW instance.

    Post-installation usage

    After installation, find examples in the examples folder of Pai-Megatron-Patch.