DeepForge Design Studio

DeepForge is a development environment for deep learning designed for simplicity, collaboration and reproducibility of experiments. Visual editing allows users to quickly and easily design neural network architectures and machine learning pipelines using a simple, intuitive interface. Integrated Version Control means that every experiment can be easily reproduced as everything is automatically version controlled. Users can design, develop and iterate on neural network models from within the browser and train the models on remote machines and monitor all executions using real-time feedback!

DeepForge Design Studio

IMPORTANT!

To launch the tool, you must first be logged into the portal. If you have no account, you can create one here.

Demo Launch

Links

Website: https://github.com/deepforge-dev/deepforge
Documentation: https://deepforge.readthedocs.io/
Wiki: https://github.com/deepforge-dev/deepforge/wiki
Intro Slides: https://docs.google.com/presentation/d/10_y5O3gHXSATfjHVLJg7dOdrz-tAXNWjlxhJ5SlA0ic/edit?usp=sharing
Datamodel Developer Slides: https://docs.google.com/presentation/d/1hd3IyUlzW_TIPnzCnE-1pdz00Pw8WaIxYiOW_Hyog-M/edit#slide=id.p
Examples: https://github.com/deepforge-dev/examples
Publication: https://media.readthedocs.org/pdf/deepforge/v1.4.2/deepforge.pdf

Acknowledgements

DeepForge has been developed in collaboration with the National Science Foundation Office of Advanced Cyberinfrastructure (OAC) under award 1740151 as well as the following organizations: Digital Reasoning, Johns Hopkins, and SciServer.

Contacts

Brian Broll