Customizing Amazon Nova models
You can customize Amazon Nova models with Amazon Bedrock or SageMaker AI, depending on the requirements of your use case, to improve model performance and create a better customer experience.
Customization for the Amazon Nova models is provided with responsible AI considerations. The following table summarizes the availability of customization and distillation for Amazon Nova.
Model Name |
Model ID |
Amazon Bedrock Fine-tuning |
Amazon Bedrock Distillation |
Sagemaker Training Job Fine-tuning |
|---|---|---|---|---|
Amazon Nova Micro |
amazon.nova-micro-v1:0:128k |
Yes |
Student |
Yes |
Amazon Nova Lite |
amazon.nova-lite-v1:0:300k |
Yes |
Student |
Yes |
Amazon Nova Pro |
amazon.nova-pro-v1:0:300k |
Yes |
Teacher and student |
Yes |
Amazon Nova Premier |
amazon.nova-premier-v1:0:1000k |
No |
Teacher |
No |
Amazon Nova Canvas |
amazon.nova-canvas-v1:0 |
Yes |
No |
No |
Amazon Nova Reel |
amazon.nova-reel-v1:1 |
No |
No |
No |
The following image demonstrates the customization paths available for Amazon Nova models:
The following table summarizes the training recipe options available. The table includes information about both the service you can use and the inference technique available.
Training recipe |
Amazon Bedrock |
SageMaker AI Training Jobs |
SageMaker AI Hyperpod |
On demand |
Provision throughput |
|---|---|---|---|---|---|
Parameter-efficient supervised fine-tuning |
Yes |
Yes |
Yes |
Yes |
Yes |
Full rank supervised fine-tuning |
No |
Yes |
Yes |
No |
Yes |
Parameter-efficient fine-tuning Direct Preference Optimization |
No |
Yes |
Yes |
Yes |
Yes |
Full rank Direct Preference Optimization |
No |
Yes |
Yes |
No |
Yes |
Proximal policy optimization reinforcement learning |
No |
No |
Yes |
No |
Yes |
Distillation - Amazon Nova Premier as teacher |
Yes |
No |
Yes |
Yes |
Yes |
Distillation - Amazon Nova Pro as teacher |
Yes |
No |
Yes |
Yes |
Yes |
Continuous pre-training |
No |
No |
Yes |
No |
Yes |
