AI model lightweighting and optimisation technology company Nota said on July 7 it will launch a tuning service that lightweight customer AI models for Amazon Web Services' in-house AI chips.
Nota, as an AWS certified partner, provides lightweighting and tuning services tailored to AWS AI chip hardware such as Trainium and Inferentia. Nota's lightweighting and deployment platform, Netspresso, will be used. Netspresso can cut an AI model's size by up to more than 90 percent while maintaining accuracy, it said.
The service consists of technology verification and diagnosis, model porting and lightweighting, and performance tuning.
The target is corporate customers that lack AI optimisation specialists. Nota emphasised it can support cost efficiency and improved inference performance by tuning AI models for AWS AI chip environments.
Chae Myung-soo (채명수), Nota's chief executive, said, "AWS AI chips deliver excellent price-performance." He added, "We will support tuning so customers can fully leverage AWS chip performance in their models and deliver tangible cost-efficiency results."
Seul-gi Son
sageson@d-today.co.kr