Closed rhyolight closed 9 years ago
I will try to find out which SHA this started, then I'll probably need to defer to @scottpurdy for analysis.
I think we can update the test since there have been a couple known changes since this test was last run. Here is one case that caused changes: #1972
I've updated it.
NuPIC used to be 100% correct at predicting cellular automata within a 500-element window after 581 rows of data. Now it is only 98.8% correct.
See failure at: https://travis-ci.org/numenta/nupic.regression/builds/73279926