NuPIC 1.0 has been released

Engineering NuPIC
, ,
1

NuPIC 1.0 is here!

NuPIC 1.0 is a stable, well-documented release of HTM sequence memory and its applications to prediction and anomaly detection. Now that NuPIC 1.0 is released, we (Numenta) want to put all our efforts into sensorimotor integration research, which builds upon foundational HTM algorithms developed in NuPIC. Since we started talking about this theory, many folks in the community have been asking for implementations, examples, etc. This is where we want to put our efforts.

So… we are putting NuPIC and NuPIC Core into “maintenance mode”. This means that for now we plan to do minor releases only, and limit changes in NuPIC and NuPIC Core to:

  • Fixing critical bugs.
  • Features needed to support ongoing research.

Maintaining transparency is still important for Numenta and HTM theory. We will continue to create experimental research code, educational materials and sample code, but these will be in our other open source repositories. There will be interesting things coming that you can all play with, but they will not be focused on applications. We continue to encourage community contributions in htm-community and maintain active discussions on HTM Forum.

I look forward to sharing with you the new sensorimotor work and getting your ongoing feedback and insight as we push the frontiers of understanding how the brain works.

6 Likes
2

Congratulation to NUPIC 1.0!

1 Like
3

Great news! Cheers to NuPIC 1.0!

Looking forward to hearing more about the sensorimotor integration research… ravenous :eagle: (closest thing to a raven I could find) you could say. For instance, I was attempting to get the “hand-crafted objects” (mentioned in the infamous Matt-Jeff Youtube series) example running (https://github.com/numenta/htmresearch/blob/master/projects/l2_pooling/infer_hand_crafted_objects.py) but couldn’t. Something related to how htmresearch depends on htmresearch-core… (Probably lacking some knowledge about swig and/or the build process).

Along these lines, @rhyolight, when there’s something sensorimotor (or on the build process) to be shared, I’d be happy to give feedback and help develop a small tutorial and/or documentation on how to build and use it. Cheers. :smiley:

1 Like