General prerequisites
The customization process involves several key stages including model training, evaluation, and deployment for inference, each requiring specific resources and configurations. Before beginning your Amazon Nova model customization on SageMaker AI, ensure you have the following general prerequisites.
-
An AWS account. If you don't have an AWS account, follow these instructions to sign up for one.
-
Access to the base Amazon Nova model customization recipes
. -
Familiarity of YAML
configuration files. -
Familiarity of how to run a Jupyter notebook in your environment.
-
Familiarity of how to create AWS resources like Amazon S3 buckets and IAM roles with appropriate permissions.
-
Familiarity of how to train a model with Amazon SageMaker AI.
-
Familiarity of Amazon SageMaker HyperPod with EKS orchestration.
-
Familiarity of Amazon SageMaker HyperPod CLI.
-
Familiarity of Amazon Nova foundational models.
-
Familiarity of available Amazon Nova models and algorithms for customization.
-
Familiarity of Amazon Bedrock inference.
