As developers, we understand the desire to have full control and access over every aspect of your project, all the way down to the codebase.

Here's a FAQ, for clarity on our project moving forward.

DataTorch Beta is the current version of DataTorch used by enterprises and in the cloud here at DataTorch.io. We renamed it to DataTorch Beta to distinguish it from the new, open-sourced rewrite of our platform.

DataTorch 1.0 is the new, open-source rewrite of DataTorch. The code is available on our GitHub and it is currently a work in progress. We expect it to be functional in about 3 months and up to par with DataTorch Beta in about 6 months.

DataTorch 1.0 will have all of the functionality of DataTorch Beta and more. DataTorch Beta already has a robust data annotator, cloud file browser, and agent/pipeline functionality. As we began to add more features to DataTorch Beta, we began to reach limitations that necessitated a rewrite of the underlying architecture. We envision DataTorch 1.0 as a total ML development suite and community, with features such as experiment tracking, artifacts, notebooks, and more.

The official reason is "to make machine learning development as flexible and accessible to everyone as possible". Unofficially, it has always been something we considered, given our roots in the open-source COCO Annotator. As we evaluated what we wanted the true scope of DataTorch to be, it was determined that a rewrite of the codebase was necessary, and so it made sense to simultaneously begin switching to an open-source model at the same time.

Due to previous contracts, we are unable to open the code for DatatTorch Beta. However, we can make it available to individual clients on a case-by-case basis. Please contact us for more information if that is something you are interested in.

For our enterprise customers, we offer support contracts for DataTorch Beta and will aid in moving your organiztion over to DataTorch 1.0 when the time comes. DataTorch Beta will continue to be available for use in the cloud here at DataTorch.io, and will continue to be available after the release of DataTorch 1.0.