Overview
This section provides an overview of the AI computing platform, including pre-set images and custom image repository options.
Introduction
The AI computing platform integrates a built-in container image repository, providing users with ready-to-use pre-built image repositories, including TensorFlow, Pytorch, Jupyter, etc. At the same time, it supports users to customize image repositories and upload their local computing environment to the image repository for use.
Pre-set image
On the preset image list page, users can directly use preset images to submit training tasks. These preset images have been fully verified and have built-in common installation packages or toolkits in advance, so users do not need to spend too much time on environment configuration. The following table lists the preset images currently available.
| Image Name | Version | Built-in |
|---|---|---|
| Pytorch | 2.0.1-cuda11.7-cudnn8-runtime | Python 3.10.11, cuda 11.7, Size: 6.48GB |
| Tensorflow | 2.14.0rc1-gpu | Python 3.11.0rc1, cuda 11.8, Size: 7.37GB |
Custom image repository
If the preset image cannot meet the actual development needs, users can build a custom image. The AI intelligent computing platform supports users to configure environment dependencies based on preset images or Dockerfile to build custom images for creating training tasks.