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

This article announces reinforcement fine-tuning support in Amazon Bedrock, enabling developers to improve model accuracy without extensive ML expertise or large labeled datasets.

• Delivers 66% average accuracy gains over base models
• Automates reinforcement fine-tuning workflow for accessibility
• Requires only small prompt sets instead of large labeled datasets
• Supports both objective tasks (code generation) and subjective tasks (chatbot interactions)
• Proprietary data remains secure within AWS environment
• Available via Amazon Bedrock console and APIs
• Initially supports Amazon Nova 2 Lite with more models coming

Reinforcement fine-tuning in Amazon Bedrock simplifies model customization, allowing organizations to adapt AI models to specific business needs without specialized talent or complex infrastructure.