docling-project / docling-project/docling-sdg

Introduce more flexibility for QA generation output persistence

Open
#14 1 comment 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
51
Forks
21
PR merge metrics
No merged PRs in 30d

Description

Right now, the output file for QA generation [is specified in `GenerationOptions`](https://github.com/docling-project/docling-sdg/blob/main/docling_sdg/qa/base.py#L110), as used [in QA gen here](https://github.com/docling-project/docling-sdg/blob/main/docling_sdg/qa/generate.py#L234).

When using this to write a demo notebook in #11 and wanting to output multiple datasets from multiple documents in a 1:1 fashion, encapsulating the output method into `GenerateOptions` requires more complexity than allowing `generate_from_chunks` and friends to take a parameter specifying where to output, which is some usability friction.

A more general question is visiting which options should be in the `*Options` classes passed to constructors and which should be method parameters. Things like `api_key` definitely make sense where they are, but maybe others - `temperature` is another example - should probably have the existing defaults but also be overwritable in function calls.

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.