Speaker
Mr
Mikhail Andreev
(National Research Nuclear University MEPhI (Moscow Engineering Physics Institute))
Description
The article discusses the development of a request distribution system for large language models (LLMs) specialized in various scientific fields. The primary focus is on fine-tuning a controlling large language model that will identify scientific disciplines represented in textual data, thereby optimizing the processing and execution of requests while considering the specifics and requirements of particular scientific disciplines. The paper presents methods and algorithms aimed at improving the precision and speed of request processing, as well as enhancing interaction between the controlling language model and specialized scientific domain language models, for the efficient processing of large volumes of textual data.
Author
Mr
Mikhail Andreev
(National Research Nuclear University MEPhI (Moscow Engineering Physics Institute))