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

Restart from ckpt: set training data #39

Open
Doris2022 opened this issue Feb 19, 2025 · 3 comments
Open

Restart from ckpt: set training data #39

Doris2022 opened this issue Feb 19, 2025 · 3 comments

Comments

@Doris2022
Copy link

Hi~ Great job!(^_^)
I have a few questions about reproducing your results. After about 200 steps, the program crashed because of OOM. Do you know if I want to restart from the last ckpt, where can I set the training data——not using previews training data in the 200 steps? Is that possible for the current implementation?
Thank you very much for your help!!!

@michaelzhiluo
Copy link
Contributor

Not currently possible in this current implementation; however, the main branch in VERL should have some code that can be ported over for saved optimizer and data loader states.

Our internal version is up to date with verl, watch out for our future releases ;)

@Doris2022
Copy link
Author

Not currently possible in this current implementation; however, the main branch in VERL should have some code that can be ported over for saved optimizer and data loader states.

Our internal version is up to date with verl, watch out for our future releases ;)

Thank you ! Looking forward to that!

@SwordFaith
Copy link

Is their any easy way to merge verl subdirectory of deepscaler with verl new release ?

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