Fabrik is an open-source, browser-based platform for collaboratively building, visualising and designing neural networks through a drag-and-drop interface. Developed by the CloudCV community, the GitHub project has about 1,100 stars and targets the friction of writing model code by hand when you just want to sketch an architecture.
In Fabrik you assemble a network visually: drag layers onto a canvas, connect them, set parameters, and see the architecture as a graph. Models can be imported from, edited and exported to popular frameworks โ the project supports Keras, TensorFlow and Caffe โ so a design sketched in the browser can move into real training code, and existing models can be brought in for inspection or modification. Because it runs as a web app, Fabrik is inherently collaborative: teams and classmates can share links, review designs and iterate together, much like a design tool but for neural nets. It was built with teaching and rapid prototyping in mind, lowering the barrier for students to experiment with architectures without first mastering a framework’s API.
Fabrik grew out of the CloudCV ecosystem of browser-based deep learning tools. Its export path means the visual editor is not a dead end: designs can graduate into Keras, TensorFlow or Caffe code for real training, and existing models can be imported back for visualisation and tweaking.
Fabrik is completely free and open source under the GPL-3.0 license. You can self-host the web application from the GitHub repository at no cost; there is no paid tier.
Students, educators and ML practitioners who want to prototype, teach or review neural network architectures visually before committing to framework code.