Transfer Learning for Natural Language Understanding Tasks

Student: Florinskiy Mikhail

Supervisor: Ekaterina Artemova

Educational Programme: Data Science (Master)

Year of Graduation: 2021

This paper studies the transfer learning method between English and Russian for Natural Language Understanding tasks. SuperGLUE benchmark, as well as its Russian-language analogue Russian SuperGLUE are chosen for research as NLU task sets. The implemented experiments include zero-shot and few-shot approaches in training multilingual mBERT, XML-R and mBART models to solve these problems. In order to work with the models listed, a modified Jiant library was used with supplementary support for additional tasks and models. This study is aimed at obtaining, comparing and evaluating the results when transferring learning from English to Russian, as well as identifying models that adapt to the set learning conditions for specific target tasks.

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