Thanks. @rhyolight why are these NuPIC versions so unstable?
I had built a sample code-base for current NuPIC anomaly-detection for any sample data using that one-hot-gym example, will share it once accomplished and re-factored.
NuPIC is not unstable, it’s just that this sample application is dependent on NuPIC 0.5.6 (we are at NuPIC 1.0+).
We have a bunch of older example projects we created several years ago. Since then we have focused almost entirely on research again, so they never got updated when we released NuPIC 1.0. There are a few API changes that need to be made, but nothing major. Here is the list, if you are interested in trying: