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

question about train loss and train time #2

Open
jojosama opened this issue Aug 23, 2024 · 1 comment
Open

question about train loss and train time #2

jojosama opened this issue Aug 23, 2024 · 1 comment

Comments

@jojosama
Copy link

1.When I train , I find the train loss is a negative number, does it is normal?
like this: " loss: -1.92 v_num: o08v train_loss_step: -2.13 "
Could you please provide the training log?

2.And how long does it take to train with vox2(56-hours data) ?
my gpu is 4 3090 and 1 epoch need 13hours (about 60-hours data), it spend too much time.

thankyou

@JusperLee
Copy link
Owner

  1. We use snr as the loss, which needs to be set to a negative number in the carrier degree descent method.
image
  1. 3 hours for 8 V100 40G

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

2 participants