Announcing new ways to customize models with SageMaker AI, without managing GPUs

Deploy SageMaker AI Serverless Training Sessions to leverage automated GPU allocation, distributed execution, and fault recovery, allowing your team to focus on writing and coordinating the training loop without managing underlying infrastructure.

Adopt SageMaker AI Serverless Training Sessions to integrate advanced model customization capabilities into your AI development workflow, reducing operational toil and enabling continuous customization with techniques like SFT, DPO, RLVR, RLAIF, or multi-turn RL.

Evaluate the new SageMaker AI Serverless Training Sessions to streamline your AI model customization process, reducing infrastructure management overhead and enabling more accurate, cost-effective models tailored to your business context.