72
This talk will provide an introduction to Large Language Models (LLMs), offering some insights into their internal mechanisms, training processes, and knowledge representation within neural networks.
We will also discuss the capabilities and limitations of language models.
Among other things, I will present new results on the control of features such as “emotion” or “truthfulness” and results on how LLMs discriminate humans.
Ada Bohm is a researcher in the field of AI and high-performance computing. Her interests include AI safety, computational ergonomics, and various programming projects. She has a PhD in computer science and works as a research engineer in the area of AI safety and as a developer of a scheduling system for supercomputers. She is employed by the Alignment of Complex Systems Research Group at Charles University and It4Innovations – National Supercomputing Center. She also co-founded the company lern.cz, which helps others to apply machine learning methods in practice.
