Skip to content

chore: Added default string to HuggingfacePostTrainingConfig #3109

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

Open
wants to merge 9 commits into
base: main
Choose a base branch
from
+2 −2
Open

chore: Added default string to HuggingfacePostTrainingConfig #3109

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
db6693e
chore: Added default string to HuggingfacePostTrainingConfig
Aug 12, 2025
7713142
chore:Updated empty string to None
Aug 12, 2025
8768432
Merge branch 'main' into HuggingfacePostTrainingConfig-branch
cheesecake100201 Aug 12, 2025
6520a6d
Merge branch 'main' into HuggingfacePostTrainingConfig-branch
Aug 14, 2025
5ce2f00
Merge branch 'main' into HuggingfacePostTrainingConfig-branch
cheesecake100201 Aug 19, 2025
75bdc7b
chore:Added a check for if dpo_output_dir exists in config
Aug 19, 2025
d0d7376
Merge branch 'main' into HuggingfacePostTrainingConfig-branch
cheesecake100201 Aug 25, 2025
0390054
Merge branch 'main' into HuggingfacePostTrainingConfig-branch
cheesecake100201 Aug 26, 2025
66f4af7
Merge branch 'main' into HuggingfacePostTrainingConfig-branch
cheesecake100201 Aug 28, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion llama_stack/providers/inline/post_training/huggingface/config.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -71,7 +71,7 @@ class HuggingFacePostTrainingConfig(BaseModel):
dpo_beta: float = 0.1
use_reference_model: bool = True
dpo_loss_type: Literal["sigmoid", "hinge", "ipo", "kto_pair"] = "sigmoid"
dpo_output_dir: str
dpo_output_dir: str | None = None
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I see, yes this is a backwards incompatible change as you noted. We should make that a runtime check where the provider yells if this value continues to be None at runtime as @cdoern notes.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

So do we need to change anything else here or is this good ?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Currently its throwing stack trace error, do we need to show this a log error only or something else?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

you'll need to add checks in the code which uses dpo_output_dir. This is the source of the errors you are seeing, also please run pre-commit to re-generate the openAPI schema

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Could see only one place where dpo_output_dir was being used have added a check there also, please review again once

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@cdoern I am unable to run pre-commit because of some dependencies issues coming up because of upgrading of my laptop, solely because of change in architecture


@classmethod
def sample_run_config(cls, __distro_dir__: str, **kwargs: Any) -> dict[str, Any]:
Expand Down
2 changes: 1 addition & 1 deletion llama_stack/providers/inline/post_training/huggingface/post_training.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -135,7 +135,7 @@ async def handler(on_log_message_cb, on_status_change_cb, on_artifact_collected_

resources_allocated, checkpoints = await recipe.train(
model=finetuned_model,
output_dir=f"{self.config.dpo_output_dir}/{job_uuid}",
output_dir=f"{self.config.dpo_output_dir}/{job_uuid}" if self.config.dpo_output_dir else None,
job_uuid=job_uuid,
dpo_config=algorithm_config,
config=training_config,
Expand Down
Loading