This talk explores the process of training and fine-tuning BERT, a state-of-the-art language model, specifically for legal texts in a new language using HuggingFace transformers library and RoBERTa model. Discover the challenges of adapting BERT for legal jargon and nuances, including the difficulties of learning a language with limited resources and speakers. Gain insights into masked language modeling techniques and how they can be applied to improve language understanding in the legal domain. Explore the unique considerations and potential solutions when working with legal texts in lesser-known languages.
Nemanja Petrovic
Badin Soft
I am an experienced Engineering Manager and Technical Lead with a strong background in software development, specializing in backend development and machine learning. I possess highly skilled expertise in microservice architecture and have a proven track record of successfully leading development teams and delivering top-quality solutions. Alongside my technical responsibilities, I am also a co-founder and board member of the Nis Java User Group, actively staying up-to-date with the latest technologies and sharing knowledge within the community.