Google DeepMind Open sourcing Sonnet - a new library for constructing neural networks
For Those who don’t know about Google DeepMind, Google DeepMind is the most advanced and powerful machine learning based Artificial Intelligence computer. Beating Chinese game Go’s world champion is one of the biggest achievement of DeepMind. Research of Go game known as AlphaGo.
Ater launching Google’s own open-source website, Now Google DeepMind open-sourcing its Sonnet framework library. Sonnet framework was developed by Google DeepMind developers using TensorFlow for quickly building neural network modules.
By open-sourcing Sonnet DeepMind wants to expand the community of DeepMind, the main motive behind this release is that DeepMind developers want to help other to make complex neural networks for their projects/researches.
Sonnet uses an object-oriented approach, similar to Torch/NN, allowing modules to be created which define the forward pass of some computation, said DeepMind in their blog post
“Modules are ‘called’ with some input Tensors, which adds ops to the Graph and returns output Tensors”, Bringing transparently to variable sharing is achieved by automatically reusing variables on subsequent calls to the same module.
DeepMind have made changes to core TF to consider models as hierarchies, This will make easier for users to switching between modules while doing experiments.
Sonnet is available on Github Repository, DeepMind has also published a paper in which it’s describing the initial version of Sonnet.
0 comments