Zpět na vyhledávání

Bc. Martin Geletka

Master's thesis

Speeding up inference time of neural machine translation

Speeding up inference time of neural machine translation
Abstract:
Vďaka takzvaným Transformer modelom sa nedávno dosiahli signifikantné posuny v úlohách strojového prekladu. V praxi však tieto modely trpia vysokou latenciou, takže sú často nepoužiteľné v praktických aplikáciách. Táto práca študuje dôvody tejto vysokej latencie a zhrnuje, aplikuje a porovnáva techniky strojového učenia, ktorých cieľom je skrátenie inferenčného času Transformer modelov použitých pri …more
Abstract:
Large qualitative gains were recently made in machine translation tasks thanks to Transformers models. However, in practice, these models suffer from high latency, such that they often are hardly usable in practical applications. This thesis study studies the reasons behind high latency time and tries to overview, employ and compare techniques, which tries to decrease the inference time of the Transformer …more
 

Keywords

Machine Translation Neural Networks Transformers Neural Machine Translation Optimization
 
Language used: English
Date on which the thesis was submitted / produced: 14. 12. 2021
Identifier: https://is.muni.cz/th/if8vh/

Thesis defence

  • Date of defence: 1. 2. 2022
  • Supervisor: doc. RNDr. Petr Sojka, Ph.D.
  • Reader: RNDr. Vít Novotný

Citation record

Cite this text

Full text of thesis

Contents of on-line thesis archive
Published in Theses:
  • světu
Other ways of accessing the text
Institution archiving the thesis and making it accessible: Masarykova univerzita, Fakulta informatiky
Složka Reference to the local database directory of the institution