The Nengo team is stoked to announce the release of Nengo 2.4.0!
Nengo is a Python library for building and simulating large-scale neural models for AI and robotics. It can be thought of as a neural compiler, transforming a functional description of a neural model to a network of spiking or non-spiking neurons that can run on multiple backends including GPUs and neuromorphic hardware.
The biggest change in 2.4.0 is an optimization step
in the build process.
When you do
Nengo sets up some internal data structures
to run the model quickly.
Now, after setting up those internal data structures,
we inspect them and merge together similar operations
in order to speed up the simulation.
The end result is that many common types of models
will be faster in Nengo 2.4.0,
so we recommend upgrading as soon as possible!
Note, however, that optimizing models does take some time, so if you’re running into long build times and you’re not running your model for very long, you can turn off the optimizer like so:
Along with adding the optimizer,
we have fixed a few bugs,
and now raise exceptions when
model act strangely.
For example, if a node returns
we will now raise a
as these issues are hard to track down and debug.
To see the full list of changes in Nengo 2.4.0, head to the Github release page.
To get the new version of Nengo, use
pip install --upgrade nengo
Please come to the Nengo forum! We welcome any questions and suggestions you might have, and invite you to share your Nengo creations there.