Theano for Pylearn2

[VidushB]

For a research project, my group and I are struggling with imports. None of us are computer science majors and so we really feel out of our depth.

I have imported theano, anaconda and pylearn2 to use in Jupyter Notebooks for Ian Goodfellow's original GAN paper. When trying to import
from pylearn2.space import VectorSpace however, I am getting an error associated with Theano GPU backend

The original code from the download had a theano line:

from theano.sandbox.cuda.type import CudaNdarrayType

which to my knowledge aims to bring in GPU backend.

SkipTest: You are importing theano.sandbox.cuda. This is the old GPU back-end and is removed from Theano. Use Theano 0.9 to use it. Even better, transition to the new GPU back-end! See https://github.com/Theano/Theano/wiki/Converting-to-the-new-gpu-back-end%28gpuarray%29

is the error that I get. It seems that theano.sandbox has been replaced? I tried getting up to date:

I applied:
conda install theano pygpu
conda install -c mila-udem -c mila-udem/label/pre theano pygpu
to get theano up to date

Doing so, I replaced code with:
from theano.gpuarray.dnn import cuDNN

hoping for a smooth execution.

The error seen was:

cannot import name 'cuDNN' from 'theano.gpuarray.dnn'
as per what I have read from theano based pages, that is a very legal line of code.

Any help is appreciated and I will try to answer any follow up to the best of my limited ability

[Frédéric Bastien]

Ian’s original script or Pylearn2 does not seem to support the most recent Theano version. Try older Theano version. Otherwise, you can try to port them to the more recent Theano version, but this will probably be more difficult. From: VidushB <[e-mail]> Sent: Thursday, January 9, 2020 7:22 AM To: Theano/Theano <[e-mail]> Cc: Subscribed <[e-mail]> Subject: [Theano/Theano] Theano for Pylearn2 (#6736) For a research project, my group and I are struggling with imports. None of us are computer science majors and so we really feel out of our depth. I have imported theano, anaconda and pylearn2 to use in Jupyter Notebooks for Ian Goodfellow's original GAN paper. When trying to import from pylearn2.space import VectorSpace however, I am getting an error associated with Theano GPU backend The original code from the download had a theano line: from theano.sandbox.cuda.type import CudaNdarrayType which to my knowledge aims to bring in GPU backend. SkipTest: You are importing theano.sandbox.cuda. This is the old GPU back-end and is removed from Theano. Use Theano 0.9 to use it. Even better, transition to the new GPU back-end! See https://github.com/Theano/Theano/wiki/Converting-to-the-new-gpu-back-end%28gpuarray%29 is the error that I get. It seems that theano.sandbox has been replaced? I tried getting up to date: I applied: conda install theano pygpu conda install -c mila-udem -c mila-udem/label/pre theano pygpu to get theano up to date Doing so, I replaced code with: from theano.gpuarray.dnn import cuDNN hoping for a smooth execution. The error seen was: cannot import name 'cuDNN' from 'theano.gpuarray.dnn' as per what I have read from theano based pages, that is a very legal line of code. Any help is appreciated and I will try to answer any follow up to the best of my limited ability — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub<#6736?email_source=notifications&email_token=AABMF66S2EQEBOHTYC5EWYTQ44JHPA5CNFSM4KEXNQBKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4IFBLQFQ>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AABMF675YAU6Q7W6BVRCMELQ44JHPANCNFSM4KEXNQBA>.

…

----------------------------------------------------------------------------------- This email message is for the sole use of the intended recipient(s) and may contain confidential information. Any unauthorized review, use, disclosure or distribution is prohibited. If you are not the intended recipient, please contact the sender by reply email and destroy all copies of the original message. -----------------------------------------------------------------------------------