This is a collection of various client-requested use-cases. Each use-case is accompanied by a short live-demo video, along with all the files used. As we learn more about our clients' specific needs, this page will be updated with new content.
- MLflow with run:ai: experiment management is important for Data Scientist. This is a demo of how to set up and use mlflow with run:ai.
- Docker intro: run:ai runs using Docker images. This is a (very) brief introduction to Docker, image creation, and how to use them in the context of run:ai. Please also check out the Persistent Environments use-case if you wish to keep the creation of Docker images to a minimum.
- Tensorboard with Jupyter (ResNet demo): Many Data Scientist like to use Tensorboard to keep an eye on the their current training experiments. They also like to have it side-by-side with Jupyter. In this demo, we will show how to integrate Tensorboard and Jupyter Lab within the context of run:ai.
- Persistent environments (with conda/mamba & Jupyter): Some Data Scientist find creating Docker images for every single one of their environments a bit of a hindrance. They would often prefer the ability to create and alter environments on the fly, and to have those environments remain, even after an image has finished running in a job. This demo shows users how they can create and persist conda/mamba environemnts using an NFS.
- Weights & Biases with run:ai: W&B is one of the best (if not the best) tools for experiment tracking and management, which is why run:ai is proud to have them as an official partner. In this tutorial, we'll demo how to use W&B alongside run:ai.
- Virtual environments intro with conda: A short tutorial on what virtual environments are, why you should use them, and how to work with them using conda/mamba.