Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Better specification of setup requirements #346

Closed
maxbartolo opened this issue Dec 2, 2017 · 6 comments
Closed

Better specification of setup requirements #346

maxbartolo opened this issue Dec 2, 2017 · 6 comments

Comments

@maxbartolo
Copy link
Collaborator

I think it requires cuDNN > v6 because v5 didn’t work for me and also an explanation of why CUDA 8.0 is specified (ie will 7.5 work fine or is that there so you don’t update to 9.0? - I’m not sure because I just updated to 8.0). Also, what Python version? I updated to 3.6 since it uses tensorflow 1.4 but getting a runtime warning which seems to indicate that I should have left it at 3.5. Also, you have to manually install tensorflow-gpu to get it running on the GPUs (I think) and it doesn’t get installed with the requirements.

@dirkweissenborn
Copy link
Collaborator

That is correct. The setup is quite messy at the moment. I think we should work this over. We should probably also switch from egg to whl when building dists. What do you think @pminervini ? Can you handle this?

@pminervini
Copy link
Collaborator

pminervini commented Dec 19, 2017

What do you think @pminervini ? Can you handle this?

Yup, let me check what's the best practice

I updated to 3.6 since it uses tensorflow 1.4 but getting a runtime warning which seems to indicate that I should have left it at 3.5.

It's fine - I have it as well.

@pminervini
Copy link
Collaborator

I can set up a nightly cron job for building a jack wheel:

http://jack.neuralnoise.com/dist/

Then it would be possible to install jack as follows:

pip3 install http://jack.neuralnoise.com/dist/jack-0.1.0-py2.py3-none-any.whl

But maybe there's a better way of handling this whole thing.

@dirkweissenborn
Copy link
Collaborator

I think jack will primarily be used as a whitebox framework. I, for instance, never use the setup.py script because I don't want jack to get installed on my machine. What I would like is a good documentation for how to setup jack properly without having to install it. I mean, we already have a requirements.txt which should be enough in addition with a good documentation on how to setup jack with CPU or GPU, respectively.

@pminervini
Copy link
Collaborator

pminervini commented Dec 19, 2017

@dirkweissenborn ok - I'm now working on using pip install -e .[tf] and pip install -e .[tf_gpu] for installing jack without/with GPU suport (instead of running setup.py); I just need to fix a couple of bugs and merge

@pminervini
Copy link
Collaborator

Bam: #349

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants