KAIST XAI Tutorial 2025 | XAI Evaluation | Haksoo Lim (SAIL, KAIST)

Recent advances in large language models (LLMs) have led to remarkable performance gains and rapid adoption across a wide range of applications. Despite this success, their intricate architectures and high-dimensional internal representations make it difficult to clearly understand how these models process information and make decisions. Gaining insight into the inner mechanisms of language models is essential for recognizing their limitations and ensuring their safe and responsible use. In this talk, I will survey major approaches to explaining language models and illustrate how these techniq…