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

Unittests to check model loading? #1690

Open
Muennighoff opened this issue Jan 2, 2025 · 4 comments · May be fixed by #1698
Open

Unittests to check model loading? #1690

Muennighoff opened this issue Jan 2, 2025 · 4 comments · May be fixed by #1698
Assignees

Comments

@Muennighoff
Copy link
Contributor

There are lots of models that fail at the basic step of loading (e.g. #1689). Maybe it would be good to have a unit test that checks that all models can be loaded? One problem may the size of some large models, but maybe there is a way to work around that / we only test models that can be loaded in a reasonable amount of time

@isaac-chung
Copy link
Collaborator

Maybe we can limit these by n_parameters? Idk something like 2B.

@isaac-chung isaac-chung linked a pull request Jan 3, 2025 that will close this issue
7 tasks
@isaac-chung
Copy link
Collaborator

Adding to @Samoed 's suggestion, is it okay for us to run this test once / add as a script, instead of running this test for every PR? Tests are coming up to 47min.

@KennethEnevoldsen
Copy link
Contributor

KennethEnevoldsen commented Jan 3, 2025

great idea,

generally think we could do a test that ensure that model can load (+ optionally additional checks like n_parameters match etc.). We coud then write these to a cache folder in repo (tests folder), which is invalidated based on a hash of the the model implementation (or model file). A simpler solution could also be to rerun after n-days? WDYT?

@isaac-chung
Copy link
Collaborator

Updated the PR checklist in #1698 to account for all suggestions 🙏

@isaac-chung isaac-chung self-assigned this Jan 4, 2025
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

Successfully merging a pull request may close this issue.

3 participants