Sonnet is a library built on top of TensorFlow for building complex neural networks.

Sonnet can be installed from pip, with or without GPU support.

The version with GPU depends on tensorflow-gpu and will install it if you have not done so. The version without GPU support will use the version of tensorflow you have installed, and will not work if you don't have any version installed.

To install sonnet without gpu support, run

$ pip install dm-sonnet

To install it with gpu support:

$ pip install dm-sonnet-gpu

The following code constructs a Linear module and connects it to multiple inputs. The variables (i.e., the weights and biases of the linear transformation) are automatically shared.

import sonnet as snt

train_data = get_training_data()
test_data = get_test_data()

# Construct the module, providing any configuration necessary.
linear_regression_module = snt.Linear(output_size=FLAGS.output_size)

# Connect the module to some inputs, any number of times.
train_predictions = linear_regression_module(train_data)
test_predictions = linear_regression_module(test_data)

Check out the full documentation page here.