Machine translation with LLMs / Masintõlge suurte keelemudelite abil
Organization
TartuNLP
Abstract
LLMs like ChatGPT have recently emerged, yet machine translation (like Google Translate, Neurotõlge, etc) is still mostly done with sequence-to-sequence neural networks. The goal of this thesis is to train both LLMs and sequence-to-sequence models to perform machine translation and compare the results.
Graduation Theses defence year
2024-2025
Supervisor
Mark Fišel
Spoken language (s)
Estonian, English
Requirements for candidates
Some familiarity with machine learning and neural networks. Confident Python skills. Experience with HuggingFace/PyTorch is a plus.
Level
Masters
Keywords
Application of contact
Name
Mark Fišel
Phone
E-mail