Completes OPEN-3366 Refactor dataset validations #82
Conversation
openlayer/__init__.py Outdated
| file_path: str, | ||
| class_names: List[str], | ||
| label_column_name: str, | ||
| dataset_type: DatasetType = DatasetType.Validation, |
There was a problem hiding this comment.
i wouldn't set a default for this to encourage users to think about it
There was a problem hiding this comment.
good point! I'll change it
| dataset_type: DatasetType = DatasetType.Validation, | ||
| feature_names: List[str] = [], | ||
| text_column_name: Optional[str] = None, | ||
| categorical_feature_names: List[str] = [], |
There was a problem hiding this comment.
why'd we decide to not use a .yaml for this one but use it for model? bc it could be in memory?
There was a problem hiding this comment.
this is not final. I feel like these args are still a bit awkward. I added the support for a yaml for consistency with the model upload. Since there, the user is constructing a model package, it made sense to ask for files.
In the dataset case, the user is not preparing a package and, as you said, a lot of this info could be in memory (especially if they're using the add_dataframe method). I'm going to be re-thinking how to pass these args in the next few days.
| language: str = "en", | ||
| sep: str = ",", | ||
| commit_message: Optional[str] = None, | ||
| dataset_config_file_path: Optional[str] = None, |
There was a problem hiding this comment.
need the docstring for this prop
There was a problem hiding this comment.
Actually, since this is prob. not be final, I'll update the docstrings later. I'll have to re-write the docstrings to follow the new commit-style anyway, so I'll make sure to update it all together. (Wrote down on my to-dos so I don't forget)
| ) | ||
| except ma.ValidationError as err: | ||
| # ---------------------------- Dataset validations --------------------------- # | ||
| # TODO: re-think the way the arguments are passed for the dataset upload |
gustavocidornelas force-pushed the cid/refactor-dataset-validations branch from 76ffacd to 5a98e10 Compare 
Refactors the dataset validations and introduces dataset types (for training and validation sets).
Summary
DatasetValidatorclass, analogous to theModelValidatorintroduced in a previous PR.dataset_typeargument, which can be eitherDatasetType.TrainingorDatasetType.Validation.Additional comments
add_dataset. I introduced the possibility of passing it via adataset_config.yamlfile (analogous to themodel_config.yamlfor models), but it still feels a bit awkward. I'll continue thinking about it.