Home Install Gnomehat Demo Projects Star it on Github About

Simple experiment management for deep learning

Run your Python programs with GnomeHat...



... and they become distributed, reproducible experiments




An easy environment for deep learning



Gnomehat installs an up-to-date PyTorch/Tensorflow/CUDA stack automatically, so you can get started right away. Gnomehat queues and distributes experiments across the GPUs on your local machine, or across multiple machines on a local network.



Run hundreds of experiments, and never lose track of one


Gnomehat keeps a snapshot of your source code every time you run an experiment, so you'll never forget which version of your code generated which result.



Control your own infrastructure

GnomeHat is not a cloud service. It's an open source tool that you can run on your own computer. Whether you're running an individual workstation or a cluster of GPU servers, GnomeHat keeps you in control.



One-Click Deep Learning Demonstrations


Run the latest open source deep learning code directly from Github, with no setup or configuration required.

Progressive Growing of Generative Adversarial Networks
Generate photorealistic faces of fake celebrities
Tero Karras, Timo Aila, Samuli Laine & Jakko Lehtinen (ICLR 2018)
https://www.github.com/gnomehat/progressive_growing_of_gans

Semantic Segmentation
Detect people, cars, and more in input images.
Waleed Abdulla's implementation of He et al. (CVPR 2017)
https://www.github.com/gnomehat/Mask_RCNN
and many more!




Easy to Install, Simple to Use

GnomeHat is built for graduate students and hobbyist Python programmers, and doesn't require any advanced knowledge of DevOps or software workflows. All you need is a Ubuntu desktop or server with one or more NVIDIA GPUs. Follow the installation guide to get running in minutes.





Install GnomeHat now to get started.