With Amazon SageMaker AI
Note
This documentation is for Amazon Nova Version 1. Amazon Nova 2 is now available with new models and enhanced capabilities. For information on how to customize Amazon Nova 2, visit Customizing Amazon Nova 2 models.
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 or launch jobs on SageMaker AI Hyperpod.
Topics
Amazon Amazon Nova Customization SDK
The Amazon Nova Customization SDK is a comprehensive Python SDK that provides a unified, programmatic interface for the complete Amazon Amazon Nova model customization lifecycle. The SDK simplifies model customization by offering a single, consistent API for training, evaluation, monitoring, deployment, and inference across SageMaker AI and Amazon Bedrock platforms. For more information, go to Amazon Amazon Nova Customization SDK.
