The guest lecture will provide a practical insight into the algorithmic use of large language models (LLM). The lecture will examine how large basic language models differ from instructed LLMs. Examples of practical use of instructed LLMs will be examined and analyzed. We will also pay attention to mechanisms that allow us to protect against prompt injections using the capabilities of LLM programming interfaces. At the end of the lecture, an insight will be provided into how the capabilities of LLMs are analyzed for solving practical problems, using the translation task as an example.
Dr.sc.comp. Mārcis Pinnis is the Head of Artificial Intelligence (AI) Development at SIA “Tilde”, a researcher at the Linguistic Computing Laboratory of the University of Latvia and an assistant professor at the School of Economics and Culture. He holds a PhD in Computer Science from the University of Latvia (obtained in 2015) and a Master’s degree in Computational Linguistics from St. Edmund’s College (University of Cambridge, UK; obtained in 2009). Mārcis has been developing various language technologies for 20 years and has been leading the Tilde research team for the last four years. He is the author and co-author of more than 80 scientific publications and has recently been focusing on the development, algorithmic use and quality assessment of large language models.