arXiv subjects

Alexander Abramov

Publications and source records attributed to Alexander Abramov.

4 recordsLinked to original sources

arXiv2025

The Russian-focused embedders' exploration: ruMTEB benchmark and Russian embedding model design

Embedding models play a crucial role in Natural Language Processing (NLP) by creating text embeddings used in various tasks such as information retrieval and assessing semantic text similarity. This paper focuses on research related to embedding models in the Russian language. It introduces a new Russian-focused embedding model called ru-en-RoSBERTa and the ruMTEB benchmark, the Russian version extending the Massive Text Embedding Benchmark (MTEB). Our benchmark includes seven categories of tasks, such as semantic textual similarity, text classification, reranking, and retrieval.The research also assesses a representative set of Russian and multilingual models on the proposed benchmark. The findings indicate that the new model achieves results that are on par with state-of-the-art models in Russian. We release the model ru-en-RoSBERTa, and the ruMTEB framework comes with open-source code, integration into the original framework and a public leaderboard.

cs.CL↗

arXiv2024

A Family of Pretrained Transformer Language Models for Russian

Transformer language models (LMs) are fundamental to NLP research methodologies and applications in various languages. However, developing such models specifically for the Russian language has received little attention. This paper introduces a collection of 13 Russian Transformer LMs, which spans encoder (ruBERT, ruRoBERTa, ruELECTRA), decoder (ruGPT-3), and encoder-decoder (ruT5, FRED-T5) architectures. We provide a report on the model architecture design and pretraining, and the results of evaluating their generalization abilities on Russian language understanding and generation datasets and benchmarks. By pretraining and releasing these specialized Transformer LMs, we aim to broaden the scope of the NLP research directions and enable the development of industrial solutions for the Russian language.

cs.CL↗

arXiv2021

A Combined Theoretical and Experimental Study of the Phase Coexistence and Morphotropic Boundaries in Ferroelectric-Antiferroelectric-Antiferrodistortive Multiferroics

The physical nature of the ferroelectric (FE), ferrielectric (FEI) and antiferroelectric (AFE) phases, their coexistence and spatial distributions underpin the functionality of antiferrodistortive (AFD) multiferroics in the vicinity of morphotropic phase transitions. Using Landau-Ginzburg-Devonshire (LGD) phenomenology and a semi-microscopic four sublattice model (FSM), we explore the behavior of different AFE, FEI and FE long-range orderings and their coexistence at the morphotropic phase boundaries in FE-AFE-AFD multiferroics. These theoretical predictions are compared with the experimental observations for dense Bi1-yRyFeO3 ceramics, where R is Sm or La atoms with the fraction 0 < y< 0.25, as confirmed by the X-ray diffraction (XRD) and Piezoresponse Force Microscopy (PFM). These complementary measurements were used to study the macroscopic and nanoscopic transformation of the crystal structure with the doping. The comparison of the measured and calculated AFE/FE phase fractions demonstrate that the LGD-FSM approach well describes the experimental results obtained by XRD and PFM for Bi1-yRyFeO3. Hence, this combined theoretical and experimental approach provides further insight into the origin of the morphotropic boundaries and coexisting FE and AFE states in model rare-earth doped multiferroics.

cond-mat.mtrl-sci↗

arXiv2019

Polarization-dependent conductivity of grain boundaries in BiFeO3 thin films

Charge transport across the interfaces in complex oxides attracts a lot of attention because it allows creating novel functionalities useful for device applications. In particular, it has been observed that movable domain walls in epitaxial BiFeO3 films possess enhanced conductivity that can be used for read out in ferroelectric-based memories. In this work, the relation between the polarization and conductivity in sol-gel BiFeO3 films with special emphasis on grain boundaries as natural interfaces in polycrystalline ferroelectrics is investigated. The grains exhibit self-organized domain structure in these films, so that the "domain clusters" consisting of several grains with aligned polarization directions are formed. Surprisingly, grain boundaries between these clusters (with antiparallel polarization direction) have significantly higher electrical conductivity in comparison to "inter-cluster" grain boundaries, in which the conductivity was even smaller than in the bulk. As such, polarization-dependent conductivity of the grain boundaries was observed for the first time in ferroelectric thin films. The results are rationalized by thermodynamic modelling combined with finite element simulations of the charge and stress accumulation at the grain boundaries giving major contribution to conductivity. The observed polarization-dependent conductivity of grain boundaries in ferroelectrics opens up a new avenue for exploiting these materials in electronic devices.

cond-mat.mtrl-sci↗