With Amazon SageMaker AI

Developers sometimes require greater control and flexibility when customizing Amazon Nova models. With SageMaker AI, you can leverage larger labeled datasets, perform deeper customization (such as preference optimization or full rank fine-tuning), and access enhanced capabilities for hyperparameter optimization and active train-loss curve analysis. You can perform these advanced customization features on Amazon Nova models in the SageMaker AI platform.

With Amazon Nova customization using SageMaker AI, you can launch SageMaker training jobs to customize Amazon Nova models with the following techniques: