Can large language models create new Croatian words?
March 17, 2026
As part of the celebration of the Month of the Croatian Language on March 13, a lecture on large language models and artificial intelligence was held at the Dubrovnik Libraries. The lecture was given by Professor Marko Tadić, who presented the first Croatian monolingual language model HR-GPT Beta, trained on almost eight billion Croatian words and available in the HR-CLARIN language data repository.