-
Notifications
You must be signed in to change notification settings - Fork 3.5k
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
Release Pytorch Lightning as a conda package #794
Comments
@dangpzanco agreed. Is this something you can take care of? at least the config. then we can set up what needs to happen for conda release. |
@williamFalcon I'm not sure if I'll be able to take care of this (I don't have much spare time). But if nothing comes up in a couple of weeks, I'll probably look into it. |
@Borda want to look at this for this release? |
Depends how much time will leave, I don't have any experience with conda packaging yet... |
According to @hadim it is not possible to release as conda package: |
Apparently it's possible to add pytorch packages to conda-forge following this example: conda-forge/staged-recipes#11762 The pytorch version available on conda-forge is 1.1 CPU so tests won't pass but tests are already performed on this repo anyway. |
would be great to have it in conda! Thanks hadim |
You can close here. |
we should probably add instructions in the readme. |
🚀 Feature
Please make the
pytorch-lightning
package available from the conda package manager. This would probably be done through conda-forge:conda install pytorch-lightning -c conda-forge
Motivation
The default way of installing Pytorch is via their conda channel, so (probably) most users of Lightning already use the conda package manager.
Conda packages provide automatic updates (
conda update --all
) and better package dependencies management, which could make the user's life a lot easier.Pitch
I'd like to install
pytorch-lightning
and its dependencies via this command:Ideally, this wouldn't force reinstall the
pytorch
package from its channel toconda-forge
.Alternatives
My personal way of installing
pytorch-lightning
is running:pip install --no-deps pytorch-lightning test-tube
And then
conda install
the missing dependencies that show up onpip check
.On version 0.5.3.x this was mandatory, since the required
pytorch-lightning
packages would mess upnumpy
and other conda packages.The text was updated successfully, but these errors were encountered: