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
Issues setting up working environment #96
Comments
Hey @dtch1997 this is a known issue. Simply run The error was caused by torch-scatter didn’t specify torch as a dependency. |
hey @vwxyzjn thanks for the response. It didn't work for me, after running the second poetry install I get this error:
|
I re-tried the double poetry install in a fresh virtualenv and I got the CUDA version mismatch error again
|
The following is my installation trace. By comparing it with yours, it seems that you are not using a virtualized environment, which may be the cause of the issue. Mine was Alternatively, could you check if
|
This comment has been minimized.
This comment has been minimized.
I re-tried the double install outside of my conda env. This time we can see it's using the virtualenv correctly as the command is
|
I see. You can try installing torch compiled with a newer version of cuda. You can try overriding the
|
What worked for me in the end was to delete the
|
Perhaps that should be the default way to install, I believe there's currently no good way to automatically choose the right CUDA version in poetry. Keeping this issue open to track updating pyproject.toml, CI, and docs. |
@vwxyzjn Would be curious to get your thoughts on what the best approach would be. |
Also, @dtch1997 thanks for the sleuthing, having a seamless setup process is super important, and smoothing out any kinks will make it much easier for new contributors to get ramped up. |
This is a little bit unfortunate where some of torch's version is not on PyPi such as the Not exactly sure what would be the best solution here. In https://poets-ai.github.io/elegy/contributing/, they use |
After thinking and playing it a bit, I think @dtch1997's approach is more preferable. It avoids many headaches at the cost of not exactly locked dependencies.
|
I'm Python
3.8.12
which should be supported according to thepyproject.toml
.I'm also using Poetry
1.1.7
withpoetry-core==1.0.4
, because of this related issue" https://stackoverflow.com/questions/69836936/poetry-attributeerror-link-object-has-no-attribute-nameHowever, I get weird errors concerning the concurrent installation of
torch
andtorch-scatter
. It looks liketorch-scatter
is trying to be installed before torch. See log.txt for full stack trace.Tried installing torch first and then re-installing torch-scatter but got a different error:
log3.txt
Help appreciated
The text was updated successfully, but these errors were encountered: