The NengoDL team is jazzed to announce the release of NengoDL 1.2.0.
NengoDL is a backend for Nengo that integrates deep learning methods (supported by the TensorFlow framework) with other Nengo modelling tools. This allows users to optimize their models using deep learning training methods, improves simulation speed (on CPU or GPU), and makes it easy to insert TensorFlow models (such as a convolutional neural network) into Nengo networks.
To use NengoDL, replace instances of
For example, if you have a network called
net and you run it as
with nengo.Simulator(net) as sim: sim.run(10)
you would change that to
with nengo_dl.Simulator(net) as sim: sim.run(10)
and that’s it!
Information on accessing the more advanced features of NengoDL can be found in the documentation.
The 1.2.0 release has made it much easier to train spiking neural networks. NengoDL will now automatically use a differentiable, rate-based approximation for spiking neuron models during training (if one is known), rather than the old approach where users had to swap neuron types manually. Check out the updated spiking MNIST example to see how this looks in practice. We’ve also expanded the config system, and made improvements to both training and inference speed. Check out the GitHub release page for a full changelog.
To install NengoDL, we recommend using
pip install nengo-dl
More detailed installation instructions can be found here.
Please come to the Nengo forum! We welcome any questions and suggestions you might have.