Make it more obvious which names are public

instructlab / sdg

Python library for Synthetic Data Generation

Apache License 2.0

5 stars 13 forks source link

Make it more obvious which names are public #30

Open tiran opened 1 week ago

tiran commented 1 week ago

The synthetic data generation code was moved to a separate package instructlab.sdg so it can be consumed by multiple projects. The current project layout does not make it obvious which names are designed for public consumption with a stable API, and which are internal implementation details.

I recommend:

prefix all modules with _, e.g. _generate_data.py and _utils.py
import public names in __init__.py and include them in __all__ variable.

russellb commented 1 week ago

Thanks, @tiran.

I expect this code to receive a significant overhaul in the next week or two. I will be sure to ensure these suggestions are incorporated.

FYI @oindrillac @aakankshaduggal @shivchander