Why should a cached binary LightGBM Dataset be treated as a versioned preprocessing artifact?
Classic MLLevel: IntermediateLightGBM · Tuning
Why should a cached binary LightGBM Dataset be treated as a versioned preprocessing artifact?
Hints
- What work does the cache avoid?
- That avoided work encodes preprocessing choices.
Answer
It contains constructed/binned data, not just a raw file path. Changes to binning, feature schema or source data can invalidate reuse; record the construction settings and rebuild when their contract changes.
Source and licence: LightGBM · Tuning · MIT · answer: Quantoros editorial