Skip to content
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

Cannot evaluate inp query #2

Open
slnxyr opened this issue Jul 12, 2023 · 5 comments
Open

Cannot evaluate inp query #2

slnxyr opened this issue Jul 12, 2023 · 5 comments

Comments

@slnxyr
Copy link

slnxyr commented Jul 12, 2023

Hi, I follow your guide to train and evaluate the model by 1p-train, but there is something wrong when evaluate inp query. Could you help me? Thanks.
Beside, how long does it take to train all queries?
image

@LHRLAB
Copy link
Owner

LHRLAB commented Jul 12, 2023

It's strange that you may have a problem with the version of the package.

@slnxyr
Copy link
Author

slnxyr commented Jul 12, 2023

But the other queries can get the evaluation result in the same code.

@LHRLAB
Copy link
Owner

LHRLAB commented Jul 12, 2023

That may be the format problem of the inp data set. We will check it, and you can try to download the data set again.

@YiZuMonash
Copy link

Same question as @slnxyr , I ran the code based on the full queries, but it seems to take a long time. May I know how long will it takes because it still in epoch 1 for nearly 1 day which confused me a lot. my settings as follows. Thanks.

parser.add_argument("--train_tasks", default="1p,2p,3p,2i,3i,pi,ip,2u,up,2cp,3cp,2in,3in,inp,pin,pni", type=str)

@LHRLAB
Copy link
Owner

LHRLAB commented Jul 12, 2023

Yes, full-query training does take a long time, but because NQE is based on Transformer, you can achieve a similar effect of full-query training with only 1p training.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants