-
Notifications
You must be signed in to change notification settings - Fork 100
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
Failure to Start GUI #1476
Comments
Hi @Nike-Nay, From reading through some of the linked issues below, it seems this is an error with a failed pip install during the SLEAP installation process. Thanks for sending the output of I'm linking some related issues here:
Thanks, |
Hi Liezl, thanks for the quick response. I have added the output for both pip freeze and mamba list here: pip freeze
mamba list
|
I see you have tensorflow version 2.6.3. After the conda install, did you downgrade tensorflow via pip? I believe the issue is due to issues with dependency management between conda and pip. My solution was to install SLEAP via pip and then downgrade tf. This resulted in both the GPU being detected by tf and the gui loading. |
Hi @rgough5, Did you have a similar problem with the conda package? On Windows as well? @Nike-Nay we released a new package and I am wondering if this would bypass your problem? mamba create -y -n sleap -c conda-forge -c nvidia -c sleap -c anaconda sleap=1.3.2 If it does end up in a failure, would you mind posting the output of the installation command and the sleap-label command. Thanks, |
@roomrys I had a similar issue on SLEAP v1.3.2 although on Fedora Linux. I tried installing with conda from both the package and source, but in both cases tensorflow could not detect our GPU (NVIDIA GeForce RTX 4090). Downgrading tensorflow via pip did allow for the GPU to be detected, but the GUI would not open with |
Hi @rgough5, Could you try:
You could also try the same method but with Let us know what you get! Cheers, Talmo |
Unforunately, it did not work. I did get a different error
as opposed to the previous error when I just used
If it helps, when I downgrade using the second method I get the warning
Which is why I switched to installing everything via pip. Installing with pip came with it's own set of issues, but it didn't take too much effort to get it working from there. |
Linux-5.15.0-78-generic-x86_64-with-debian-bookworm-sid
|
We released a new version at the end of last week which aimed to address the problems @rgough5 brought up and put restrictions on mamba create -y -n sleap -c conda-forge -c nvidia -c sleap -c anaconda sleap=1.3.3 is working for both of you? The pip package has also been updated @rgough5. Thanks! |
@roomrys No issues here: one line install and SLEAP work with our GPUs! Thank you so much. |
Bug description
I have installed sleap using mambaforge and installed it via the 'conda package' instructions at https://sleap.ai/installation.html
I believe everything is installed correctly as I can activate sleap in the terminal.
However, when I attempt to load up the GUI using 'sleap-label' it returns the following:
FileNotFoundError: [Errno 2] No such file or directory: 'c:\users\asus\mambaforge\envs\sleap\lib\site-packages\tensorflow_estimator-2.6.0.dist-info\METADATA'
I am not sure why there is no METADATA file. Is this an issue with the downloaded package or an issue on my system?
Expected behaviour
Actual behaviour
Your personal set up
Environment packages
Logs
Screenshots
How to reproduce
The text was updated successfully, but these errors were encountered: