This repository contains Kasm workspaces for AI workloads.
See workspace specific documentation:
NVIDIA CUDA base image - CUDA-enabled base image.
PyTorch Image - CUDA-enabled base image with PyTorch.
Tensorflow Image - CUDA-enabled base image with Tensorflow.
Please refer to the GPU Setup page for details on pre-requisites.
To build the provided images:
sudo docker build -t kasmweb/ubuntu-noble-nvidia:dev -f dockerfile-kasm-ubuntu-nvidia .
While these image are primarily built to run inside the Workspaces platform, they can also be executed manually. Please note that certain functionality, such as audio, uploads, downloads, and microphone pass-through are only available within the Kasm platform.
sudo docker run --rm -it --shm-size=512m -p 6901:6901 -e VNC_PW=password kasmweb/ubuntu-noble-nvidia:dev
The container is now accessible via a browser : https://<IP>:6901
- User :
kasm_user - Password:
password
Kasm Workspaces is a docker container streaming platform that enables you to deliver browser-based access to desktops, applications, and web services. Kasm uses a modern DevOps approach for programmatic delivery of services via Containerized Desktop Infrastructure (CDI) technology to create on-demand, disposable, docker containers that are accessible via web browser. The rendering of the graphical-based containers is powered by the open-source project KasmVNC
Kasm Workspaces was developed to meet the most demanding secure collaboration requirements that is highly scalable, customizable, and easy to maintain. Most importantly, Kasm provides a solution, rather than a service, so it is infinitely customizable to your unique requirements and includes a developer API so that it can be integrated with, rather than replace, your existing applications and workflows. Kasm can be deployed in the cloud (Public or Private), on-premise (Including Air-Gapped Networks), or in a hybrid configuration.
A self-guided on-demand demo is available at kasmweb.com