Amazon Bedrock now supports reinforcement fine-tuning delivering 66% accuracy gains on average over base models

Amazon Bedrock Reinforcement Fine-Tuning

Amazon Bedrock now supports reinforcement fine-tuning, enabling developers to improve model accuracy with minimal labeled data. This feature automates the fine-tuning workflow, allowing teams to quickly adapt models to specific requirements using a small set of prompts. Reinforcement fine-tuning delivers an average of 66% accuracy gains over base models.

This new capability addresses the challenge of adapting AI models to unique business needs, offering a fast, automated, and secure customization process. You can train models using data from your computer or Amazon S3, and define reward functions with rule-based graders or AI-based judges. Your data remains within AWS's secure environment throughout the process.

What to do

• Start using reinforcement fine-tuning via the Amazon Bedrock console or Amazon Bedrock APIs.
• Explore the launch blog, pricing page, and documentation for more information.

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.