Data preparation
This section provides data preparation guides for each training technique and model version.
Topics
Validation tools
Before submitting a fine-tuning job, we recommend validating your dataset to catch formatting issues early. You have two options:
-
Dataset validation script – A standalone Python script hosted on GitHub. See the dataset validation script
. -
Amazon Nova Forge SDK – Validation support is built into the SDK. See Data preparation
in the Amazon Nova Forge SDK README.