Workshop - XAI for LLMs: looking under the hood of Large Language Models - ft. Serg Masis - Lead Data Scientist at Syngenta

​Join us for a riveting exploration into the world of Explainable Artificial Intelligence (XAI) as we uncover the secrets of Large Language Models (LLMs) like BERT, Llama, and GPT. In this session, we’ll kick things off with a clear and concise introduction to what LLMs are and delve into their sophisticated transformer architectures.

Using the powerful Bertviz tool, we will take you on a visual journey “under the hood” to examine the intricate attention mechanisms that drive both encoder-based models like BERT and decoder-based models like Llama and GPT. This exploration will highlight how these mechanisms function in real-world applications, from text classification to generating compelling text.

Next, we’ll harness the capabilities of the SHAP library to reveal feature importance values in text generation tasks and use Integrated Gradients to shed light on text classification processes. Whether you’re a tech enthusiast, a seasoned AI professional, or simply curious about how AI can be made understandable, this session will equip you with a deeper understanding of these cutting-edge technologies and their practical implications.

​Serg Masís is a Lead Data Scientist in agriculture with a lengthy background in entrepreneurship and web and mobile development, and the author of the bestselling book "Interpretable Machine Learning with Python", and the upcoming book "DIY AI". He's passionate about data-driven decision-making, Responsible AI, behavioral economics, and making AI more accessible.