ML4ITS / TimeVQVAE-AnomalyDetection

[official] PyTorch implementation of TimeVQVAE-AD, a time series anomaly detection model.
MIT License
42 stars 9 forks source link

Issue running steps in README - Please could you confirm package version requirements? #4

Closed daniel-mannion closed 1 month ago

daniel-mannion commented 1 month ago

Hi, first off, thanks for providing the code to your model! I've followed the steps provided in the README but the training of stage 1 does not run. I get the following error:

image

This seems it could be an issue with which versions of packages are being used. I started with an empty venv and installed packages using the provided requirements.txt. The pip freeze from the resultant environment is attached below.m Not all package versions have been specified, I'm wondering if the most recent version of these packages are incompatible... Please could you confirm which versions of packages you are using? I can then try these and test if that was the issue or not.

packages.txt

daniel-mannion commented 1 month ago

Ahh I've just tried reverting numpy from 2.0 to 1.26.4 and it corrected the issue. I'll submit a pull request with a modified requirements file if that helps.

danelee2601 commented 1 month ago

thanks for the contribution! never thought numpy version would be an issue.

daniel-mannion commented 1 month ago

No worries I had a similar problem with a repo recently. It looks like numpy 2.0 has a tendency to break things.