The Gigantum Client is an open source web application that you can install and run locally to manage your work in Python and R. It is self-service and can be installed in minutes pretty much anywhere you can run Docker, from a laptop to a commercial cloud instance.
The browser based interface is used to create, manage, and interact with Gigantum Projects and Datasets. The Client runs in the background while you work in Jupyter or RStudio to automate file management, versioning and containerization without any extra effort.
This automation makes it easy to move and share totally functioning and reproducible work between machines and people. All work done in Gigantum is reproducible and portable from the moment you start working.
Gigantum doesn't dictate where you compute because that choice is better left up to you. We just make it easy for you to deploy where you want while maintaining easy reproducibility. The Client is containerized and easy to install and use anywhere.
Gigantum supports computing with GPUs, and you can easily create Projects that support different CUDA versions. When run on a GPU instance, Gigantum will automatically ensure driver versions are compatible, and if so configure Docker to use the GPU for you. At the moment, Gigantum only supports Nvidia GPUs running on a Linux host OS.
The Client is a decentralized application, like Git. So when you run the Client on different machines, you may have different versions of Projects and Datasets on those local Clients.
The Client provides one click publish and sync with the Hub so that you can maintain consistency of versions across machines and people. You can publish Projects either publicly or privately, and there are a variety of permission options for working with collaborators.
Updated about a year ago