Google’s DeepMind announced that it has released and open sourced Sonnet – a new open source library that allows developers to quickly create complex neural networks.
DeepMind defines Sonnet as a “library built on top of TensorFlow for building complex neural networks.” Sonnet, therefore, does not replace TensorFlow – another open source library for creating neural networks shared to the public by Google. In order to use the Sonnet library, one has to install TensorFlow first.
For nearly a year, researchers at DeepMind have used the TensorFlow library for developing neural networks. DeepMind researchers, in a blog post, wrote that neural networks created using Sonnet can be “freely mixed” with raw TensorFlow code and other high-level neural network libraries.
“Making Sonnet public allows other models created within DeepMind to be easily shared with the community, and we also hope that the community will use Sonnet to take their own research forwards,” DeepMind researchers said.DeepMind has shared its new open source library for creating complex neural networks through GitHub.
In recent months, DeepMind has also open-sourced its flagship platform DeepMind Lab and is also presently working with Blizzard to develop an open source API that supports AI research in StarCraft II.