Rune is a self-contained, language-agnostic executable containing a compiled ML pipeline. The project is open-source and available on GitHub.
What is a machine learning (ML) pipeline?
ML pipeline is a collection of inputs, models, pre- and post-processing steps, and outputs that are joined together to solve a problem.
What is a model?
A model is the brains behind a ML pipeline. Several “layers” of numbers connected using mathematical operations that gets trained to generate desired outputs.
What ML frameworks do you support?
Right now we support Tensorflow Lite and Tensorflow JS models. We are actively working to add support for ONNX framework which allows us to bring a wide variety of frameworks to the ecosystem.
What platforms do you support?
Hammer Forge supports edge deployment to iOS, Android, web, mobile web, and desktop environments today. We are adding support for Arduino, Coral, Jetson Nano, and Raspberry PI soon. For enterprise customers we are also supporting additional devices based on custom needs.
What is Hammer Forge?
Hammer Forge is our SAAS platform that consists of Hammer Forge Studio and Foundry. Studio is the a drag-and-drop editor for creating Runefiles and automating the build, testing, and deployment process. Foundry is rock solid build system for Runes.
What is a Runefile?
Runefile is a declarative YAML file which tells the Rune compiler how to build a Rune. It is open source.
What is a deployment?
When you successfully deploy a project pipeline to a device (or web), you have created a deployment. When a device uses your edge app, it becomes an active deployment. We have designed our platform so that you can create any number of projects during development without having to deploy anything live. When you feel confident you can deploy.