Updated feature: You can now use a previously customized model (fine-tuned or distilled) as the base model for further customization.

AWS Release Notes Summary

New Features

You can now create a custom model by using Fine-tuning or Continued Pre-training in the Amazon Bedrock console or API. You can further fine-tune an existing custom model.

Prerequisites

• IAM Service Role: Create an IAM service role to access the S3 bucket for storing your model customization data.
• Encryption: (Optional) Encrypt input and output data, customization job, or inference requests.
• VPC: (Optional) Create a VPC to protect your customization job.

Submit Your Job

Console

Follow these steps to submit a model customization job:

1. Sign in to the AWS Management Console and open the Amazon Bedrock console.
2. Choose Custom models under Tune.
3. In the Models tab, select Customize model and then Create Fine-tuning job or Create Continued Pre-training job.
4. Configure model details, job configuration, input data, hyperparameters, output data, and service access.
5. Choose Fine-tune model or Create Continued Pre-training job to start the job.

API

Send a CreateModelCustomizationJob request with the following fields:

• roleArn: The ARN of the service role with permissions to customize models.
• baseModelIdentifier: The model ID or ARN of the foundation model or previously customized model.
• customModelName: The name for the newly customized model.
• jobName: The name for the training job.
• hyperParameters: Hyperparameters for the model customization process.
• trainingDataConfig: The Amazon S3 URI of the training dataset.
• validationDataConfig: (Optional) The Amazon S3 URI of the validation dataset.
• outputDataConfig: The Amazon S3 URI to write the output data to.

The response returns a jobArn to monitor or stop the job.

Source: AWS release notes

If you need further guidance on AWS, our experts are available at AWS@westloop.io. You may also reach us by submitting the Contact Us form.