CIFAR10 script dtype error

[jaxony]

I receive the following TypeError when I try to import CIFAR10 data via the pylearn2 API:

Traceback (most recent call last):
File "my_binary_net.py", line 27, in
train_set = CIFAR10(which_set="train", start=0, stop=train_set_size)
File "/Users/Jackson/pylearn2/pylearn2/datasets/cifar10.py", line 85, in init
x = numpy.zeros((lenx, self.img_size), dtype=dtype)
TypeError: 'numpy.float64' object cannot be interpreted as an integer

I fixed the error by typecasting lenx, which is originally np.float64, to int:
lenx = int(numpy.ceil((ntrain + nvalid) / 10000.) * 10000)

[nouiz]

Pylearn2 don't have maintainer. This error is comming from newer numpy. If you make a PR to fix, it, I can merge it, but we don't maintain it anymore. There is other framework on top of Theano like Lasagne, Keras and blocks that are still alive.

…

On Mon, Jun 5, 2017 at 9:47 AM Jackson Huang ***@***.***> wrote: I receive the following TypeError when I try to import CIFAR10 data via the pylearn2 API: Traceback (most recent call last): File "my_binary_net.py", line 27, in train_set = CIFAR10(which_set="train", start=0, stop=train_set_size) File "/Users/Jackson/pylearn2/pylearn2/datasets/cifar10.py", line 85, in *init* x = numpy.zeros((lenx, self.img_size), dtype=dtype) TypeError: 'numpy.float64' object cannot be interpreted as an integer I fixed the error by typecasting lenx, which is originally np.float64 to int: lenx = int(numpy.ceil((ntrain + nvalid) / 10000.) * 10000) — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#1587>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AALC-xXsZYyDJoyp1YIR0IGrLVMNTaeoks5sBAbhgaJpZM4NwEAN> .