name	description	tags	icon
Nvidia NGC Deeplearning Template	Use this template to create a deeplearning workspace on Coder based on Nvidia NGC containers.	docker deeplearning pytorch tensorflow jupyter nvidia ngc	https://raw.githubusercontent.com/matifali/logos/main/nvidia.svg

Deep Learning Workspaces Using Nvidia NGC Containers

Coder Setup

Follow these steps to configure accessing your workspaces locally on any machine.

Linux/MacOS

1. Open a terminal and run

   curl -L https://coder.com/install.sh | sh

Windows

1. Open a powershell window and run

   winget install Coder.Coder

Usage

1. Clone this repository

   git clone https://github,com/matifali/coder-templates
   cd coder-templates/deeplearning

2. Login to coder

   coder login CODER_URL

   Replace coder.example.com with your coder deplyment URL or IP

3. Create a template

   coder templates create deeplearning-nvidia

4. Create a workspace

   Go to https://CODER_URL/workspaces and click on Create Workspace and select deeplearning-nvidia template.

Note: Do not forget to change the CODER_URL to your coder deployment URL.

Connecting

There are multiple options to connect to your workspace using local clients or browser.

VS Code Browser

Click on the VS Code Browser icon to launch a VS Code server that you can connect from your browser.

Jupyter Lab

Click on the jupyter icon to launch a jupyter lab server that you can connect from your browser.

Also, you can connect using the Web Terminal or SSH by clicking on the above buttons.

VS Code Desktop

Click on the VS Code Desktop icon to conNect to your workspace using VS Code Desktop.

JetBrains Gateway

Follow the instructions to use the JetBrains Gateway with Coder.

Persistent Storage

There will be a ~/data inside every workspace. All files placed here will survive reboots and be available to all workspaces.

Docker Image

This template uses the Nvidia NGC container for

1. PyTorch as the base image.
2. TensorFlow as the base image.