atomistic-machine-learning / schnetpack-gschnet

G-SchNet extension for SchNetPack
MIT License
46 stars 8 forks source link

Question about QM9 usage #16

Closed loilisxka closed 1 month ago

loilisxka commented 1 month ago

I'm executing gschnet_train --config-dir=/home/huzheyu/ai_lab/gener_ai/test experiment=gschnet_qm9 but get the error

Cannot find primary config 'train'. Check that it's in your config search path.

Config search path:
      provider=hydra, path=pkg://hydra.conf
      provider=main, path=pkg://schnetpack.configs
      provider=hydra-colorlog, path=pkg://hydra_plugins.hydra_colorlog.conf
      provider=command-line, path=file:///home/huzheyu/ai_lab/gener_ai/test
      provider=schema, path=structured://

Set the environment variable HYDRA_FULL_ERROR=1 for a complete stack trace.

I don't know what causes this error, if you guys can answer I'd be very grateful.

NiklasGebauer commented 1 month ago

Hello @loilisxka ,

thanks for reporting this issue! We rely on the training procedure give in schnetpack and it seems like something is not working as intended. I was able to reproduce the bug by installing schnetpack-gschnet with the newest version of schnetpack (2.1.0). This version of schnetpack was released last week and I suppose some changes have led to this error. I couldn't pinpoint the exact cause yet, but I will try to find a proper solution as soon as possible.

As a workaround, you can uninstall schnetpack pip uninstall schnetpack and install the previous version in your environment pip install schnetpack==2.0.4 This should enable you to properly start the training.

Hope this helps and sorry for the hiccups with the new version! Best regards, Niklas

NiklasGebauer commented 1 month ago

The bug was caused by the setup of schnetpack, where config files were not properly copied into the package. We fixed this in the schnetpack repository and released version 2.1.1.

So it should be possible to upgrade to schnetpack 2.1.1 now and run the training as expected. Please let me know in case this fix does not work in your case.

Best regards, Niklas

loilisxka commented 1 month ago

OK, It's working, thanks for you very much.

------------------ 原始邮件 ------------------ 发件人: "atomistic-machine-learning/schnetpack-gschnet" @.>; 发送时间: 2024年9月5日(星期四) 晚上7:58 @.>; @.**@.>; 主题: Re: [atomistic-machine-learning/schnetpack-gschnet] Question about QM9 usage (Issue #16)

The bug was caused by the setup of schnetpack, where config files were not properly copied into the package. We fixed this in the schnetpack repository and released version 2.1.1.

So it should be possible to upgrade to schnetpack 2.1.1 now and run the training as expected. Please let me know in case this fix does resolve your problem.

Best regards, Niklas

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: @.***>