RED HAT announced about the readiness of the new edition of the distribution of Rhel- red house linux AI (Rhel AI), specially adapted for the performance of machine learning and designed to simplify the creation of server solutions using large language models.
The distribution includes a selection of tools and frameworks for machine learning, drivers for using various hardware accelerators AMD, Intel, and NVIDIA, and components to use the capabilities of the Dell, Cisco, HPE, Lenovo and Supermicro servers, optimized for AI systems. The assemblies are distributed through the Red Hat Customer Portal in the options for direct installation on servers and for deployment in the AWS and IBM Cloud cloud systems.
The platform can be used to develop corporate AI applications, implement content generation services, create dialogue systems and integration into the application of virtual assistants that support such skills as an opportunity to answer questions in natural language, solve mathematical problems, generate meaningful text on a given The topic, make a brief presentation of the contents, correct errors in the text, perform rewriting in other words, help in writing code in various programming languages, form letters and documents according to the template.
In Rhel AI, the components are integrated for creating, testing and performing machine learning systems on the basis of a large language model granite , The open company IBM under the Apache 2.0 license, capable of taking into account the text of up to 4 thousand tokens and covering 7 billion parameters. To interact with the Granite model, the open tools of instructs are integrated into the distribution t For Chatbots) to fit their needs and optimize models, as well as for adding additional knowledge and the implementation of the new skills in pre-trained models .