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.

Sign up for GitHub

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

Jump to bottom

about hyper parameter settings in my own dataset #182

Open

momohuangsha opened this issue Dec 3, 2021 · 1 comment

momohuangsha commented Dec 3, 2021 •

edited

Loading

hi, @TreB1eN Thank you for sharing.

I set up a dataset from YouTube video, AFDB, CNBC etc, with a total of 6587 categories. Use MTCNN to detect face and align

Training data like this:

use MobileFaceNet and ArcFace head
milestones = [30, 70, 200] lr=0.01 batch size =64 Train in my notebook(GTX1050ti 4G)

After 70 epochs, the train loss stable at about 3 :

So is the hyper parameter setting appropriate?
Do you have any good suggestions about hyper parameters,please?

KevinRK27 commented Nov 21, 2022

How did you make the .idx, .lst, and .bin file?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment