ReBased (Language Model Architecture)

Main article: Training artificial intelligence

2024: Product Announcement

Russian scientists from the T-Bank AI Research laboratory have developed a new ReBased technology for accelerated processing of long texts by artificial intelligence. This innovation will significantly reduce the cost of using AI in word processing with almost no loss in quality, the press service of T-Bank reported in August 2024.

ReBased is a new language model architecture that optimizes the mechanism for extracting information from text. The technology adds new learning parameters that are responsible for the optimal search for relationships between parts of the text, which improves the process of processing it and allows you to get more accurate answers.

It is noteworthy that in parallel with the release of our article, a group of researchers from Stanford released a study on the same topic, but with a different approach to the solution. Now this is one of the most interesting areas of research at NLP around the world, - said Yaroslav Aksenov, a researcher of natural language processing at T-Bank AI Research, in August 2024.

ReBased can reduce the cost of using artificial intelligence for specialized tasks with a specific application. For example, in medicine, such a task can be considered the classification of texts based on symptoms and diagnoses.

Experiments have shown that the new technology improves understanding of relationships in text by an average of 10%. At the same time, ReBased allows you to generate texts with lower resource requirements with almost no loss of quality.

The development of Russian scientists is of great importance for business, since now the use of language models limits the high costs of computing power. In addition, reducing energy consumption in computing, especially in large data centers, will help reduce the negative impact on the environment and reduce greenhouse gas emissions.