Fine-Tuning of Large Language Models in Terms of Architectures of Monitoring of Scientific Information
Abstract
The paper is devoted to the research and experimental implementation of algorithms and architectures for monitoring scientific and technical information aimed at minimizing monitoring maintenance time and increasing the availability of monitoring as a tool for personally oriented informing and supporting the analytical activity of the user on issues of scientific and technical development of interest to him. The algorithm of the model for monitoring scientific and technical publications on the Internet is proposed, which allows generating forecast reports in parallel with processing the metadata of the knowledge base. A mechanism has been developed for the automatic output of predictive information on the development of the branch of science and technology, built on the basis of an artificial neural network based on the principles of a large language model, further trained on facts from the network of concepts of the developed knowledge base. An algorithm has been developed for using a pre-trained artificial neural network of the "Large Language Model" type on a 1000-node video processor to recognize scientific and technical problems, as well as generate annotations and reports on the state of the art, allowing to speed up the monitoring process tenfold compared with the use of publicly available untrained networks running on the CPU. A method of operating a pre-trained network is proposed, based on processing a language model using a video processor with support for CUDA technology and placing a large language model locally in RAM. The results obtained by the author allow us to put forward a new class of information and reference systems - personally adaptable information and reference and analytical monitoring systems that ensure the accumulation and updating of knowledge on the development of science and technology in accordance with the individual interests and preferences of the user.
This work is licensed under a Creative Commons Attribution 4.0 International License.
Publication policy of the journal is based on traditional ethical principles of the Russian scientific periodicals and is built in terms of ethical norms of editors and publishers work stated in Code of Conduct and Best Practice Guidelines for Journal Editors and Code of Conduct for Journal Publishers, developed by the Committee on Publication Ethics (COPE). In the course of publishing editorial board of the journal is led by international rules for copyright protection, statutory regulations of the Russian Federation as well as international standards of publishing.
Authors publishing articles in this journal agree to the following: They retain copyright and grant the journal right of first publication of the work, which is automatically licensed under the Creative Commons Attribution License (CC BY license). Users can use, reuse and build upon the material published in this journal provided that such uses are fully attributed.
