Automated Evaluation of Language Model Performance – Machine Translation from English to Montenegrin Language and English to Thai

Linguistic models represent a fundamental component of Natural Language Processing (NLP) technologies. NLP has been in the spotlight in the post-COVID-19 pandemic period, attracting attention from both experts and the general public. This is due to significant advancements in the performance of language models and the applications they enable. The reasons for this progress can be found in the increased availability of data, computational power, and innovations in the field of machine learning. The paper focuses on the specificity of the Montenegrin and Thai language and the ability of language models to comprehend, process, or generate it. In conclusion, findings and recommendations are presented regarding the selection of language models for specific tasks and the automatic evaluation of performance.