Hello!
I’m sharing a small open-source plugin I put together while working with NVIDIA Data Designer (NeMo) and generating sample data for LLM-related experiments.
The plugin adds a declarative way to define columns, aiming to reduce repetition when working with larger schemas and to make column intent more explicit in code. Instead of configuring each column imperatively, the focus is on describing structure and constraints in a compact, readable form.
This is very much a practical, hands-on experiment driven by my own usage, not a polished product. Feedback, questions, and suggestions are very welcome, especially from others using Data Designer for synthetic data generation.
Source and details:
https://github.com/webmaxru/data-designer-declarative-columns
Hope it’s useful to someone else exploring similar workflows.