Weights sum up to zero

[mominchen]

Hello,
I got recently this error message, although I read it's been fixed. I'm running python 2.7.13 on a Macbook Air and I've installed the following versions: nolearn 0.6.1.dev0, Theano 0.8.1, Lasagne 0.2.dev1, numpy 1.12.0

The following code was running and if I set up train_split to 0.01, it works. The line below ... is causing the error message below. I'd really appriciate any help I can get, to make this code run:

train_split=TrainSplit(eval_size=0.0), 
update=nesterov_momentum,
update_learning_rate=theano.shared(float32(0.01)),
update_momentum=theano.shared(float32(0.9)),
regression=True,
max_epochs=5000,
verbose=1,
....
net.fit(X, y)

File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/nolearn/lasagne/base.py", line 693, in fit
self.train_loop(X, y, epochs=epochs)
File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/nolearn/lasagne/base.py", line 796, in train_loop
custom_scores, weights=batch_valid_sizes, axis=1)
File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/numpy/lib/function_base.py", line 1140, in average
"Weights sum to zero, can't be normalized")
ZeroDivisionError: Weights sum to zero, can't be normalized