Skip to main content

Get started

The Deploy option allows users to create DevPod instances using image templates, storage volumes and related configurations.

Configure DevPod​

Planning an appropriate DevPod instance is a crucial step in deploying DevPod. The choice of GPU and storage volume can significantly affect the performance and efficiency of your project. This page provides guidance on how to configure the DevPod specifications. However, these are just general guidelines. Keep your specific requirements in mind and plan accordingly.

It's crucial to understand the specific resource requirements of a model. Usually, you can find detailed information in the model card's description on platforms like Hugging Face or in the config.json model file.

There are some tools that can assist you in evaluating and calculating the specific resource requirements of a model, such as:

Using these tools and resources should enable you to have a clearer understanding of what aspects need to be focused on in a DevPod.

Create a new DevPods​

  1. Navigate to DevPods and select + Deploy.

  2. Select the region for deploying DevPod.

  3. Specify a Network volume (optional).

  4. Select an instance type. For example, H100 SXM.

  5. Select an image (template).

  6. Configure the resources required for the normal operation of DevPod.

  7. Review your configuration and select Deploy On-Demand.

Charges occur after the DevPod build is complete.

Access DevPod​

After creating a DevPod and waiting for the instance to operate normally, click on the drop-down box on the right side of the instance to display the details of the DevPod. Then click on the "Connect" button, and a display box will pop up. According to your needs, you can choose either HTTPS access or TCP (SSH) access or Web Terminal.