I’ve been playing around with NuPIC for a little while, watching the videos and reading through much of the literature. While working through the Algorithms API Quickstart, I made a Jupyter notebook to document the process, adding simple examples where I could. I took a little time to polish it up if other people want to use it as a resource. It borrows very heavily from the Quickstart code, so I hope I’m not stepping on any toes. Very big thanks to Matt for all the time and effort he’s put into making this material accessible.
I love this! Thanks for sharing. Looks very thorough. One bit of feedback on terminology… especially in the ‘A Note on the Language of “Columns”’ section, use the term minicolumn instead of column.
Thanks for the responses. I forgot to mention that any and all feedback is welcome, especially to point out any mistakes. I don’t want to add any confusion.
Thanks @rhyolight for the tip; I’ll make the language clearer.
This is a really cool contribution @learner, especially now that quick and easy hosted notebooks exist with offerings like AWS Sagemaker and Databricks etc, it’s a good way for beginners to quickly try it all out.
I kept getting np (your numpy import) mentally confused with nupic but that’s probably just me
but after install this, and type “pip install nupic” in cmd, I saw this error:
“annaconda-client 1.6.14 has requirement python-dateutil>=2.6.1, but you will have python-dateutil 2.1 which is incompatible."
and also this error:
“annaconda-client 1.6.14 has requirement pyYAML>=3.12, but you will have pyyaml 3.10 which is incompatible.”
please help me, what should i do?
i really like to see this visualization but i cant:pensive:
[/quote]
I’m sorry, but we don’t support Anaconda. None of us run Anaconda so we can’t debug with setting up a new environment, and I just don’t have the time to do it these days.
It’s also a great tool for collaboration - you don’t need to resend a notebook after any update.
Plus, it will be computed on the Google Cloud for free, including using GPU or TPU (unfortunately it doesn’t supported by Neupic, but it’s really nice opportunity).