Cannot install 0.3.0 version and Colab examples are failed to run

[alysiayx]

1. try to run pip install semilearn==0.3.0 but 0.2.4 downloaded instead, not sure if that is correct.

2. the Colab example "Custom_Dataset.ipynb" is failed to run as NotImplementedError raised after defining configs and creating config

Can anyone help with these issues?

[Hhhhhhao]

I found this is due to mkl-fft==1.3.0 previously in requirements.txt has no longer correct version in pip. I have already updated this in requirement (#122). But to use 0.3.0, you have to install from source now:

git clone https://github.com/microsoft/Semi-supervised-learning.git
pip install -e ./

[alysiayx]

I found this is due to mkl-fft==1.3.0 previously in requirements.txt has no longer correct version in pip. I have already updated this in requirement (#122). But to use 0.3.0, you have to install from source now:

git clone https://github.com/microsoft/Semi-supervised-learning.git
pip install -e ./

Thank you so much for the prompt reply. but unfortunately, mkl-fft>=1.3.0 will still raise an error in colab. I think mkl-fft might unnecessary?

and I also find that semilearn 0.3.0 is incompatible with Apple M1 chip, specifically for the tensorflow-macos (protobuf should be >= 3.20.3 instead of 3.19.6). and mkt is also not available in Apple M1.

I guess it would be possible to loosen the version requirements?

[github-actions]

This issue is stale because it has been open 30 days with no activity. Remove stale label or comment or this will be closed in 5 days.

[ZahraaHM]

I have the same problem , it is not resolved for sorry

[Hhhhhhao]

I have the same problem , it is not resolved for sorry

Have you tried to install the latest semilearn==0.3.1? This problem should be addressed in the new version.

[ZahraaHM]

yes, I used the last version linked in README, just with last two hours

[Hhhhhhao]

yes, I used the last version linked in README, just with last two hours

I see. This might be caused by the update on the colab side. If you create a new conda environment and install semilearn, there won't be any problem:

conda create -n test_semilearn python=3.10
conda activate test_semilearn
pip install -U semilearn==0.3.1

[github-actions]

This issue is stale because it has been open 30 days with no activity. Remove stale label or comment or this will be closed in 5 days.

[github-actions]

This issue was closed because it has been stalled for 5 days with no activity.