Mika, did Chetan's tip to patch the SP before running Cerebro fix your problem? --------- Matt Taylor OS Community Flag-Bearer Numenta
On Mon, Oct 20, 2014 at 9:15 PM, Chetan Surpur <[email protected]> wrote: > Ah, so you have to patch the SP > before > you run it. > > > > On Sun, Oct 19, 2014 at 4:12 PM, Mika Schiller <[email protected]> > wrote: >> >> I'm trying to make the sp learn an input 20 times as you can see in the >> code below. First I go into nupic.cerebro2.server directory and run python >> server.py 9090. Then I go into nupic.cerebro2 directory, then /static and >> then run python -m SimpleHTTPServer 8000. When I go to >> http://localhost:8000/ I see the display, but no cells or anything. I've >> attached a screenshot of what I'm seeing. Any idea what might be going on >> here? Did I patch the sp correctly? Thanks! >> >>>> >>>> from nupic.encoders import ScalarEncoder >>>> >>>> from nupic.research.spatial_pooler import SpatialPooler >>>> >>>> import numpy >>>> >>>> from patcher import Patcher >>>> >>>> >>>> #create an encoder >>>> >>>> encoder = ScalarEncoder(n=22, w=3, minval=2.5, maxval=97.5, >>>> clipInput=True, forced=True) >>>> >>>> >>>> #create a spatial pooler >>>> >>>> sp = SpatialPooler(inputDimensions=(22), >>>> >>>> columnDimensions=(4,), potentialRadius=22, numActiveColumnsPerInhArea=1, >>>> >>>> globalInhibition=True, synPermActiveInc=0.03, potentialPct=1.0) >>>> >>>> >>>> >>>> >>>> #make sp learn input 20 times >>>> >>>> output = numpy.zeros((4,), dtype='int') >>>> >>>> for _ in xrange(20): >>>> >>>> sp.compute(encoder.encode(1), learn=True, activeArray=output) >> >> >>>> >>>> >>>> >>>> if __name__ == "__main__": >>>> >>>> >>>> >>>> Patcher().patchSP(sp) >> >> >> <Screen Shot 2014-10-19 at 7.08.50 PM.png> > >
