Skip to main content
In GPUhub instances, data (including environments) is saved even after shutdown. You needn’t reconfigure or reupload data upon restart. However, instances will be released if shut down continuously for 15 days. For details, see Instance Data Retention Policy.

​
Before You Start

Please read this carefully before using GPUHub. These are the essential things every user needs to know before getting started.
Please review these 10 points before using GPUHub:
  1. Storage — System Disk, Data Disk, and File Storage serve different purposes. Images do not include Data Disk data.
  2. Billing — Stopping an instance stops GPU charges, but storage charges may continue.
  3. Data — Back up important data before releasing an instance. Deleted data may not be recoverable.
  4. Access — GPUHub provides a container environment, not host or Docker daemon access.
  5. Images — Custom Docker images and private registries are not supported.
  6. Ports — Each instance supports 2 public ports only. Additional ports cannot be added.
  7. Network — Public Internet performance depends on routing, ISPs, and destination services. End-to-end performance is not guaranteed.
  8. Elastic Deployment — Elastic Deployment has no dedicated Data Disk. Use images or File Storage for persistent data.
  9. Payments — Automatic payments and cryptocurrency payments, including USDT, are not supported.
  10. Refunds — GPUHub is prepaid. Please review the billing and refund policies before adding funds.

​
Create an Instance

After registration, enter the console. Under My Instances, click “Launch Instance”. On the instance creating page, select billing method, region, GPU type and count. Then choose a suitable idle host and image (with basic Deep Learning frameworks or community images) to create the instance.
For larger data storage, check the expandable disk size. Refer to the Instance Overview for the data disk path.
After creation, wait for startup. The main operation entry points are in the item.
Note that billing starts when the instance status shows ‘Running’. Shut it down promptly to stop billing when not in use. For billing rules, see Billing Description.

​
Upload Data

Once the instance is running, find the “JupyterLab” shortcut tool and click to open it. Locate the upload button in the screenshot to upload data. For folder uploads or advanced methods, refer to the Data Upload documentation.

​
Terminal Training

Open the terminal in the JupyterLab page. For remote development with other IDEs, refer to VSCode (recommended) and PyCharm. Execute Python commands in the terminal to complete training.

​
Advanced Learning

GPU Selection

Select GPU type and count

Environment Configration

Config your development environment

Upload Data

Upload data with FileZilla
Securely access your instance via SSH

Open Ports

Expose custom ports to make your applications accessible

VSCode

Remote development with VSCode

PyCharm

Remote development with PyCharm

Run a Daemon Process

Keep alive with a daemon process

Linux Basics

Basic Linux commands

FAQ

Frequently asked questions