search

MatthieuCourbariaux / BinaryNet

Training Deep Neural Networks with Weights and Activations Constrained to +1 or -1
BSD 3-Clause "New" or "Revised" License
1.04k stars 346 forks source link

ImportError: No module named six.moves #29

Open JiyuFang-git opened 5 years ago

JiyuFang-git commented 5 years ago

I installed pylearn2 and theano(bleeding-edge version) following the links given on the README. When running mnist.py I encountered the following error:

Using cuDNN version 5005 on context None Preallocating 3027/4036 Mb (0.750000) on cuda Mapped name None to device cuda: GeForce GTX 970 (0000:0F:00.0) Traceback (most recent call last): File "mnist.py", line 24, in from pylearn2.datasets.mnist import MNIST File "/home/jiyu/fyp/pylearn2/pylearn2/init.py", line 4, in from pylearn2.utils.logger import configure_custom File "/home/jiyu/fyp/pylearn2/pylearn2/utils/init.py", line 11, in from theano.compat.six.moves import input, zip as izip ImportError: No module named six.moves

I tried to solve the issue by installing Theano 0.8, but then a new problem occurred:

ERROR (theano.sandbox.gpuarray): Could not initialize pygpu, support disabled Traceback (most recent call last): File "/home/jiyu/miniconda2/envs/py27/lib/python2.7/site-packages/theano/sandbox/gpuarray/init.py", line 95, in init_dev(config.device) File "/home/jiyu/miniconda2/envs/py27/lib/python2.7/site-packages/theano/sandbox/gpuarray/init.py", line 46, in init_dev "Make sure Theano and libgpuarray/pygpu " RuntimeError: ('Wrong major API version for gpuarray:', 2, 'Make sure Theano and libgpuarray/pygpu are in sync.')

Could you make sure that you are suggesting correct versions of Theano and Pylearn2? Thanks

iamweiweishi commented 5 years ago

from six.moves import input, zip as izip

roshan-gopalakrishnan commented 5 years ago

Anybody resolved this issue ?

nickfraser commented 5 years ago

The environment for this can be quite tricky to set up, very specific versions of some packages are required.

I project I worked on BNN-PYNQ used slightly modified versions of this BinaryNet repo to train the networks. The dockerfile in that repo and instructions are still (at least the last time I checked) compatible with this project.

I'd recommend the dockerfiles, but you can also try to follow the instructions directly as well.