py-coursera-neural
py-coursera-neural copied to clipboard
Bug truth value
Hi,
There seems to be a bug in the (activation >= 0) condition. Here is the traceback: http://sprunge.us/cEfE?pytb
Pull request is about some minor improvements, see log messages for descriptions. I'll add later a howto in the readme to install a virtualenv and the packages needed with pinned versions.
Hey sorry I'm just getting to this - can you print out your activation value as well? This bug suggests a dimensionality problem. For example, see this code snippet
In [7]: NP.dot(NP.array([[1],[2]]),NP.array([[3,4]]))
Out[7]:
array([[3, 4],
[6, 8]])
activation >= 0 will then complain that this is not a single number.