“It’s now nearly a year since DeepMind made the decision to switch the entire research organisation to using TensorFlow (TF). It’s proven to be a good choice - many of our models learn significantly faster, and the built-in features for distributed training have hugely simplified our code. Along the way, we found that the flexibility and adaptiveness of TF lends itself to building higher level frameworks for specific purposes, and we’ve written one for quickly building neural network modules with TF. We are actively developing this codebase, but what we have so far fits our research needs well, and we’re excited to announce that today we are open sourcing it. We call this framework Sonnet. Since its initial launch in November 2015, a diverse ecosystem of higher level libraries has sprung up around TensorFlow enabling common tasks to be accomplished quicker. Sonnet shares many similarities with some of these existing neural network libraries, but has some features specifically designed around our research requirements. The code release accompanying our Learning to learn paper included a preliminary version of Sonnet, and other forthcoming code releases will be built on top of the full library we are releasing today.”
Link to article