Course provider: University of Ljubljana, Faculty of Computer and Information Science (UL FRI)
Instructors: Marko Robnik-Šikonja (UL FRI)
Learning objectives: Get acquainted with large language models (LLMs), their architecture and training, leading to their effective use.
Course contents: Large language models are changing the way we write, read, and do intellectual jobs. The lecture presents the working of the transformer architecture of neural networks and focuses on the decoder models, which are used in generative language models, such as ChatGPT. Explaining their construction, pretraining, instruction following, preference alignment, and fine-tuning, the lecture gives the necessary background to understand their behaviour. Based on this, it explains prompting strategies, such as in-context learning and chain-of-thought reasoning. The contents are based on examples from business and science.
Learning outcomes: Knowledge of LLM construction and recommendations for their use.