Polina Tabakova decided to apply for a Philology degree at HSE in Nizhny Novgorod because she grew up in Mari El and did not want to move far away from the Russian forests. In an interview for the Young Scientists of HSE University project, she spoke about the genre of the campus novel, the existential drama of Kolobok, and a blackout version of Eugene Onegin.
Researchers from the AI and Digital Science Institute at the HSE Faculty of Computer Science have developed a new compression method for large language models such as GPT and LLaMA that reduces their size by 25–36% without additional training or significant loss of accuracy. This is the first approach to use mathematical transformations—specifically, rotations of model weights—to make models more amenable to compression with structured matrices. The study results have been published in ACL Findings 2025. The code is available on GitHub.
Gizdatullin D., Ignatov D. I., Baixeries J., В кн.: Программа секций XVIII Апрельской конференции. Сессия P-04. Исследование демографических последовательностей.: НИУ ВШЭ, 2017. С. 1–11.
В данной работе представлены результаты применения узорных структур (pattern structures) и “контрастных” закономер- ностей (emerging patterns) в анализе демографических последова- тельностей для данных по России. Панельные данные Российской части исследования GGS (Generation and Gender Survey) на основе трех волн опроса в 2004, 2007, и 2011 описывают 11 поколений ре- спондентов, начиная с 1930 по 1984. ...
Gizdatullin D., Baixeries J., Ignatov D. I. et al., , in: Intelligent Data Processing 11th International Conference, IDP 2016, Barcelona, Spain, October 10–14, 2016, Revised Selected PapersVol. 794.: Switzerland: Springer, 2019. Ch. 6 P. 74–91.
There are many different methods for computing relevant
patterns in sequential data and interpreting the results. In this paper,
we compute emerging patterns (EP) in demographic sequences using
sequence-based pattern structures, along with different algorithmic solutions.
The purpose of this method is to meet the following domain
requirement: the obtained patterns must be (closed) frequent contiguous
prefixes of the input sequences. ...