Regenerate Dataset
Use this guide when you need to re-run a dataset generator โ either on demand or as part of updating your evaluation configuration.
There are two ways to trigger regeneration:
- Force-regenerate with
lf regenerate datasetโ re-runs the generator unconditionally, even if nothing in the config has changed. - Auto-regenerate via
lf add/lf runโ regeneration happens automatically when AI GO! detects that the generator configuration has changed or that the dataset was never generated in its current form.
Force-Regenerate with lf regenerate dataset
Use this when you want to refresh a dataset without changing any config โ for example, when the generated data is unsatisfactory or you want a fresh random sample.
Note
The dataset must already exist in the AI app. lf regenerate dataset does not create new datasets.
From a Run Config
Run this when your dataset is defined inside a run.yaml alongside other entities:
lf regenerate dataset --file run.yaml --key mydatasetThe --key flag is required to tell AI GO! which dataset to regenerate.
From a Standalone Dataset YAML
Run this when your dataset has its own dedicated spec file:
lf regenerate dataset --file dataset.yaml