Fred Rotbart
Fred Rotbart
Exact same problem on OSX. Was okay up to today.
Thanks @liberty-developer, that solved the problem for me.
The code https://github.com/htm-community/htm.core/blob/f6fdfaa0067636b41023d83a249a33b77804e031/py/htm/advanced/frameworks/location/location_network_creation.py#L649 uses SDR overlap successfully to classify objects.
Continuing discussion from #668. >column? (imho too much threads, see above), or a layer (that would make sense), I'd like this to be enabled generally #255 where each "main class"...
@dkeeney thanks for the update. In case that this turns out to be intractable, as insurance, I have just started investigating whether I can use [Ray](https://ray.readthedocs.io/en/latest/index.html) which does not depend...
After looking at Ray, though it may work, @dkeeney I really hope that you do get pickle working. Futures or Dask are a much easier path to multiprocessing the htm.
The important thing to check for is for location similarity. The best way to confirm this is to use the coordinate_test.py tests that I have pushed into htm.core.
@dkeeney if I was comfortable enough with C++ I would have been happy to. But it is way too far out of my comfort zone so I must decline.
Thanks. It would be much appreciated.
BTW this is part of my efforts to get around the performance bottleneck of the ColumnPooler.