Does Amazon SageMaker RL support heterogenous clusters?

Question

Does SageMaker RL support heterogenous clusters? I'd like to have our training to run on GPU and and SageMaker RL, and our inferences to run on CPUs.

Accepted Answer

Yes, Amazon SageMaker RL allows you to define training workers separately from inference workers.

For more information, see [Amazon SageMaker RL – Managed reinforcement learning with Amazon SageMaker](https://aws.amazon.com/blogs/aws/amazon-sagemaker-rl-managed-reinforcement-learning-with-amazon-sagemaker/) on the AWS News Blog. Also,  the [Build and Train Reinforcement Models with Amazon SageMaker RL](https://www.youtube.com/watch?v=xGD6rkPuQ7Q) AWS online tech talk (see minute 26).

Does Amazon SageMaker RL support heterogenous clusters?

Relevant content