Skip to content

[BUG] Nanotron runner imports non-existant #555

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

Open
jquesnelle opened this issue Feb 13, 2025 · 3 comments
Open

[BUG] Nanotron runner imports non-existant #555

jquesnelle opened this issue Feb 13, 2025 · 3 comments
Labels
bug Something isn't working

Comments

@jquesnelle
Copy link

Describe the bug

Attempting to run a nanotron eval (lighteval nantron) will result in an error since hierarchical_logger does not exist. See https://github.com/huggingface/lighteval/blob/main/src/lighteval/main_nanotron.py#L55

To Reproduce

Run

lighteval nanotron --checkpoint-config-path some-nanotron-path --lighteval-config-path some-lighteval-path

Expected behavior

Nanotron eval works

Version info

Reproduced on fac17bb

@jquesnelle jquesnelle added the bug Something isn't working label Feb 13, 2025
@jquesnelle
Copy link
Author

jquesnelle commented Feb 13, 2025

Follow up -- if you remove hierarchical_logger, https://github.com/huggingface/lighteval/blob/main/src/lighteval/pipeline.py#L73 errors. Looks like it was renamed to lighteval.models.nanotron.nanotron_model

@peteboyd
Copy link

Have this issue too. Can confirm that the fixes provided by @jquesnelle work. 1) remove the htrack_block import and code block in main_nanotron.py and 2) rename the import in pipeline.py.

@NathanHB
Copy link
Member

great ! thanks for the fix, can you open a PR so that we can review and merge ? :)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working
Projects
None yet
Development

No branches or pull requests

3 participants