Open source artificial intelligence startup develops fundamental models Utilizing Amazon SageMaker on AWS results in a 58% increase in speed and cost effectiveness.
The open source artificial intelligence (AI) company Stability AI has chosen Amazon Web Services, Inc. (AWS), an Amazon.com, Inc. company (NASDAQ: AMZN), as its preferred cloud provider to build and scale its AI models for the creation of image, language, audio, video, and 3D content, the company announced today at AWS re:Invent. Stability AI accelerates its work on open source generative AI models by utilising Amazon SageMaker (AWS’s end-to-end machine learning service), as well as AWS’s established compute and storage infrastructure. Additionally, Stability AI will work with AWS to make its open source tools and models accessible to worldwide startups, academics, and businesses.
Generative AI models from Stability AI can generate text, images, music, video, code, and more from straightforward language instructions. Models that are flexible to a range of tasks in areas like language, picture, audio, and video are known as generative AI or fundamental models. These models need a high-performance computing cluster with thousands of GPUs or AWS Trainium chips, extensive knowledge, and months of training. Stable Diffusion Version 2.0, which includes open source models for cutting-edge picture synthesis and enables users to produce new images from user-supplied text and image inputs, was just released by the company. In addition, models to upgrade image resolution and infer depth information to create new images are introduced in this most recent open source release.
When compared to self-managed machine learning infrastructure, Stability AI will save 58% on training time and costs by using Amazon SageMaker to develop and test machine learning models. The business is leveraging Amazon Elastic Compute Cloud (Amazon EC2) to set up one of the biggest clusters of machine learning instances to run its open source models on AWS, making it simpler for developers to access those models. Stability AI intends to construct and train future models using AWS machine learning and AI services in order to obtain the greatest performance at the most affordable price. Additionally, Stability AI will make its open-source models accessible to all AWS customers on Amazon SageMaker JumpStart, the model hub of Amazon SageMaker.
“Only AWS offers the scale, reliability, and efficiency to deliver the next generation of AI with Stability AI, which will enable billions of people to create stunning art within seconds,” said Bratin Saha, vice president of Machine Learning and AI Services at Amazon. “Amazon SageMaker will help Stability AI scale its use of machine learning with end-to-end tooling, empowering developers to innovate AI applications and dream up new use cases. We are excited to help them invent new products, services, and experiences for the world with the proven performance of one of the largest-ever clusters of machine learning instances in the cloud.”