Track1: GenAI Mastery

Tuning Up: A Hands-On Guide to Creating Your Own Custom LLM

Curious about crafting your very own Large Language Model (LLM) capable of surpassing off-the-shelf solutions like ChatGPT, even when starting with a much smaller base model?

This workshop aims to demystify the process of creating a custom LLM. We will begin by exploring different approaches to building LLM-based solutions and their applications. The main focus will be on the exhilarating topic of fine-tuning, a process that goes beyond augmenting prompts with auxiliary information to transform the model itself. Participants will engage in hands-on exercises to build their own custom LLM by fine-tuning an open-source foundational model using their own data. The result will be a tailor-made, original LLM exhibiting properties that off-the-shelf models cannot offer.

Learning goal

After the workshop, participants will know how to make their own custom LLM by fine-tuning an open-source foundational model using their own data.

Pre-requisites

Basic knowledge of Python is required. Experience with deep learning and NLP  is a plus.

  • Dr. Bernhard Vennemann

    Bernhard received his Doctor of Science in 2019 and holds a M.Sc. and B.Sc. in Mechanical Engineering from the Swiss Federal Institute of Technology (ETH Zurich). Before joining D ONE, Bernhard worked as a University lecturer for machine learning at ETH Zurich. Bernhard has been with the team since 2021.

  • Lucas Zurbuchen

    Lucas holds a Bachelor of Science in Computer Science from Northwestern University, where he also studied Double Bass Performance. His expertise spans data visualization, software development, AI, and data analytics, with a strong foundation in both technical and creative disciplines. Lucas has demonstrated his ability to translate complex data into interactive visual experiences, making significant contributions in both academic and professional settings. Lucas has been with the team since 2024.