Quantization of Neural Networks Used for NLP Tasks

Student: Senin Artem

Educational Programme: Financial Technology and Data Analysis (Master)

Year of Graduation: 2021

This paper examines the possibility of decreasing size and speeding up inference of neural networks used for NLP tasks. Apart from classical quantization technique, this paper includes the analysis of a new quantization algorithm named QSin. Quantization techniques were compared to post-training quantization techniques and quantization using straight-through estimator by training a large text model named GPT2 and making predictions on dataset DBpedia for different degrees of quantization intensity. Study shows that QSin algorithm is more effective at decreasing model’s size than post-training quantization techniques and quantization using straight-through estimator.

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