Amazon today introduced AWS SageMaker Ground Truth to provide data labeling for training AI models through humans or through a custom AI model.

Also announced today: Inferentia, a new AWS chip for the deployment of AI; Elastic Inference to provide more efficient model inference; SageMaker RL for reinforcement learning; and the opening of a common marketplace for developers to sell their AI models.

SageMaker is Amazon’s service to build, train, and deploy machine learning models.

SageMaker was first introduced at re:Invent one year ago and competes with services to build AI like Microsoft’s Azure Machine Learning and Google’s AutoML.

This follows the addition of a GitHub integration and built-in algorithms for SageMaker introduced last week. The ability to train models locally on your own machine was introduced earlier this year.

Several AI-related news announcements have been made this week at re:Invent, an annual AWS conference being held in Las Vegas.

On Monday, AWS introduced RoboMaker, a service to help developers test and deploy robotic hardware, and the Gunrock team was named winner of the Alexa Prize, a university student team challenge to make conversational AI capable of maintaining conversation with humans for 20 minutes.

On Tuesday, Amazon opened a marketplace for Docker containers, which includes six Nvidia AI solutions. Amazon also announced the launch of AWS Ground Station, a service for businesses and governments to transmit data from satellites in orbit around the Earth back to antennas at datacenters around the world.

More to come