# Is there an HTM version that could integrate with otsad (e.g. Python3)?

**URL:** <https://discourse.numenta.org/t/is-there-an-htm-version-that-could-integrate-with-otsad-e-g-python3/9097>\
**Category:** Implementations\
**Created:** [November 2, 2021, 12:33pm UTC](https://discourse.numenta.org/t/is-there-an-htm-version-that-could-integrate-with-otsad-e-g-python3/9097 "2021-11-02T12:33:38Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![markNZed](https://avatars.discourse-cdn.com/v4/letter/m/c5a1d2/32.png) [@markNZed](https://discourse.numenta.org/u/markNZed)\
**Post date:** [November 2, 2021, 12:33pm UTC](https://discourse.numenta.org/t/is-there-an-htm-version-that-could-integrate-with-otsad-e-g-python3/9097/1 "2021-11-02T12:33:38Z")

</div>

I came across the otsad project for benchmarking anomaly detectors based on NAB and it is a shame that HTM is not included. As mentioned in this issue the reason is because nupic requires Python2 [Learning more about otsad · Issue #2 · alaineiturria/otsad · GitHub](https://github.com/alaineiturria/otsad/issues/2) Perhaps someone know of a solution to get HTM into otsad? Was there work to port nupic to Python3?

---

<div class="post-metadata">

**Author:** ![DanML](https://avatars.discourse-cdn.com/v4/letter/d/edb3f5/32.png) [@DanML](https://discourse.numenta.org/u/DanML)\
**Post date:** [November 3, 2021, 8:39am UTC](https://discourse.numenta.org/t/is-there-an-htm-version-that-could-integrate-with-otsad-e-g-python3/9097/2 "2021-11-03T08:39:30Z")

</div>

This version is a Python3 API around the C++ core, which might do what you need:

> **[GitHub - htm-community/htm.core: Actively developed Hierarchical Temporal...](https://github.com/htm-community/htm.core)**
>
> Actively developed Hierarchical Temporal Memory (HTM) community fork (continuation) of NuPIC. Implementation for C++ and Python - GitHub - htm-community/htm.core: Actively developed Hierarchical T...

I have used this successfully for Python3 and Jupyter experiments.

---

<div class="post-metadata">

**Author:** ![markNZed](https://avatars.discourse-cdn.com/v4/letter/m/c5a1d2/32.png) [@markNZed](https://discourse.numenta.org/u/markNZed)\
**Post date:** [November 3, 2021, 8:46am UTC](https://discourse.numenta.org/t/is-there-an-htm-version-that-could-integrate-with-otsad-e-g-python3/9097/3 "2021-11-03T08:46:04Z")

</div>

Any idea of how it performs on NAB? It seems Numenta might have added some code specifically for NAB that improved the score significantly (related to likelihood calculation I think) I don’t see an entry for htm.core at [GitHub - numenta/NAB: The Numenta Anomaly Benchmark](https://github.com/numenta/NAB) Maybe it should have an entry?

---

<div class="post-metadata">

**Author:** ![DanML](https://avatars.discourse-cdn.com/v4/letter/d/edb3f5/32.png) [@DanML](https://discourse.numenta.org/u/DanML)\
**Post date:** [November 3, 2021, 9:43am UTC](https://discourse.numenta.org/t/is-there-an-htm-version-that-could-integrate-with-otsad-e-g-python3/9097/4 "2021-11-03T09:43:41Z")

</div>

Sorry, I’ve never used NAB.

---

<div class="post-metadata">

**Author:** ![markNZed](https://avatars.discourse-cdn.com/v4/letter/m/c5a1d2/32.png) [@markNZed](https://discourse.numenta.org/u/markNZed)\
**Post date:** [November 3, 2021, 9:45am UTC](https://discourse.numenta.org/t/is-there-an-htm-version-that-could-integrate-with-otsad-e-g-python3/9097/5 "2021-11-03T09:45:45Z")

</div>

I’ve come across some work at [GitHub - htm-community/NAB: The Numenta Anomaly Benchmark](https://github.com/htm-community/NAB) which looks to be what is needed. Bu it highlights issues with the liklihood estimation that was added to the original code - the scores for htm.core seem to be much lower