Interpretability of LLMs - CLT to Attribution Graph - Spring 2026

Explore the transition from Causal Language Modeling (CLT) to attribution graphs in this graduate-level computer science lecture from the University of Utah's CS 6966 course on Large Language Model interpretability. Delve into advanced techniques for understanding how LLMs process and generate text by examining the causal relationships between tokens and learning how these relationships can be visualized and analyzed through attribution graph methodologies. Gain insights into cutting-edge research approaches for making black-box language models more transparent and interpretable, with particular focus on how causal language modeling principles can be mapped to graph-based attribution systems that reveal the decision-making processes within neural networks.