• A
  • A
  • A
  • ABC
  • ABC
  • ABC
  • А
  • А
  • А
  • А
  • А
Regular version of the site
  • HSE University
  • Student Theses
  • Development of an Improved System of Grammatical Errors Correction Used Methods of Quantization and Distillation of Neural Networks

Development of an Improved System of Grammatical Errors Correction Used Methods of Quantization and Distillation of Neural Networks

Student: Sorokin Nikita

Supervisor: Eduard Klyshinskiy

Faculty: HSE Tikhonov Moscow Institute of Electronics and Mathematics (MIEM HSE)

Educational Programme: Information Science and Computation Technology (Bachelor)

Year of Graduation: 2021

The object of the development is a neural network model for correcting grammatical errors in the text in English, of varying degrees of complexity. The goal is to implement a neural network for correcting grammatical errors, research and application of compression methods(distillation and quantization) for the T5 neural network. In the course of the work, a neural network based on the T5 architecture was implemented and trained, the methods of quantization and distillation of neural networks based on the T5 architecture were studied, and the optimal algorithms for our task were selected, which improved the algorithm's operating time by 7 times and reduced the amount of disk space occupied by 9 times. We also managed to achieve a model quality that exceeds the best or is close to the best quality of the currently existing neural network models for correcting grammatical errors on the BEA and JFLEG benchmarks. The model was also implemented in a demo website. The resulting model can be implemented and used on mobile devices or users ' computers, due to its high efficiency, and used for its intended purpose. The volume of the final qualification work is 39 pages, the number of illustrations is 27, the number of tables is 0, the number of sources used is 30.

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