Improving The Quality Of Large GPT-like Language Models

Student: Ivanova Alesia

Educational Programme: Applied Mathematics and Information Science (Bachelor)

Final Grade: 10

Year of Graduation: 2024

The development of language models capable of providing complete, reliable, practically useful, and safe responses to user queries is a current challenge in machine learning. One such model is YandexGPT, developed by the company “Yandex”. This paper proposes methods to improve the quality of the model in question by adding specially formatted data to the training set, integrating several training stages into one, and using additional information when forming the training set. The quality of the model has been improved for tasks requiring a specified response format. A system for automatic detection and elimination of data leaks from the test set into the training set has been implemented. The results have been confirmed by experimental research conducted during an internship at “Yandex”.

Full text (added May 20, 2024)

Student Theses at HSE must be completed in accordance with the University Rules and regulations specified by each educational programme.

Summaries of all theses must be published and made freely available on the HSE website.

The full text of a thesis can be published in open access on the HSE website only if the authoring student (copyright holder) agrees, or, if the thesis was written by a team of students, if all the co-authors (copyright holders) agree. After a thesis is published on the HSE website, it obtains the status of an online publication.

Student theses are objects of copyright and their use is subject to limitations in accordance with the Russian Federation’s law on intellectual property.

In the event that a thesis is quoted or otherwise used, reference to the author’s name and the source of quotation is required.

Search all student theses