AWS launches new tools to make it easier to create with generative AI

14 Apr. (EUROPE PRESS) –

Amazon World Services (AWS) has announced this Friday the launch of new services and tools that aim to democratizing generative artificial intelligence (AI) facilitating all developers access to a wide range of base or foundational models (FM), from which the AI ​​feeds to create the content.

Amazon has specialized in artificial intelligence and machine learning (ML) for twenty years, technologies that have allowed it to develop services such as ‘Alexa’ cloud voice assistanthe ecommerce recommendation engine ‘Prime Air’ and the computer vision technology (artificial vision) from Amazon Go.

As AWS explained in a statement, during this time it has innovated in order to offer a scalable and highly productive infrastructure for cost-effective ML training and inference, while also launching several services that allow customers add artificial intelligence capabilities to applications — such as image recognition, prediction, and smart search — through an application programming interface (API) call.

For generative AI, the company applies the same “democratizing approach” and, for this reason, it is now working to take these technologies out of the field of research and experimentation and extend their availability beyond ‘startups’ and large technology companies.

With this objective, it has now launched a series of innovations that, as it has indicated, will help their clients to use generative AI in their companies in a “practical and simple way“.

BEDROCK AND TITAN

On the one hand, they are the Amazon Bedrock and Amazon Titan modelswhich are used to develop and dimension generative artificial intelligence applications with foundational models or base models (FM).

Amazon Bedrock is a new service that makes FMs available from major AI startups (AI21 Labs, Anthropic, Stability AI) and Amazon via an API so that the developer can find the model that best suits their needs. needs.

Bedrock provides the option to access a wide range of high power FM for text and images, including Amazon’s Titan FM.

With Bedrock’s serverless experience, creators can privately personalize FMs with your own data and easily integrate and deploy them into your applications using familiar AWS tools and capabilities without having to manage any infrastructure.

Amazon Code Whisperer

In addition, AWS has announced the general availability of the new network-optimized Trn1n instancesdesigned to provide 20% higher performance than Trn1 for large models and intensive network use.

Trn1n instances are specifically designed for achieve high performance deep learning and reduce training times from months to weeks or even days, according to the technology company, which adds that with shorter training times you can iterate faster, create more innovative models and increase productivity.

In addition, Inf2 instances powered by AWS Inferentia2, specifically optimized for large-scale generative AI applications with models containing hundreds of billions of parameters.

According to AWS, Inf2 instances offer up to four times better throughput and up to ten times less latency compared to the previous generation based on Inferentia and have “high speed” connectivity between accelerators to enable large-scale distributed inference.

Inf2 instances can be used to run deep learning applications in natural language understanding, translation, video and image generation, speech recognition, or personalization.

Finally, AWS has announced the general availability, free to individual developers, of Amazon Code Whisperer. It’s about a artificial intelligence coding companion which uses a foundational model to improve developer productivity by generating real-time code hints based on programmer feedback in their natural language and previous code in their Integrated Development Environment (IDE).

Amazon CodeWhisperer is now available for Python, Java, JavaScript, TypeScript, and C#, plus ten new languages, including Go, Kotlin, Rust, PHP, and SQL.

The tech company has pointed out that generative AI can take the heavy lifting out of the equation by “writing” much of the undifferentiated code, allowing engineers to develop faster and, at the same time, focus on the more creative aspects of programming.