LLM Mastery: Hands-on Code, Align and Master LLMs

Code LLMs and alignment from scratch with Python and PyTorch. And explore generative AI and deep learning using Origami

Created by: Javier Ideami
Multidisciplinary engineer, researcher & creative director

What Will I Learn?

• Code and train an LLM from scratch, line by line, understanding every concept in detail
• Understand and analyze in depth an advanced LLM architecture based on the Llama system
• Code and train an alignment process from scratch, to align an LLM with a preferred form of interaction
• Understand in great depth key concepts like the Attention Mechanisms, the Cross Entropy Loss, the way neural networks learn and many more
• Explore in depth insights about deep learning and neural networks through the use of Origami
• Visual understanding: Dive deep into the core of an LLM and visually explore how its attention matrices evolve throughout the training process.
• In addition to the coding, every section includes in-depth explanations of key concepts related to these architectures and generative AI

Requirements

• Basic knowledge of python. It's enough with the very basics, as we will code every little thing together, line by line
• You can code either on a free online platform like Google Colab, or using your local Laptop or Desktop
• Lots of enthusiasm. We are going to go really deep into both the code and the concepts around the code. You can grow and stretch your knowledge and experience a lot if you stay focused and motivated throughout the course, let's do it 🙂

Target audience

• AI Enthusiasts and Developers: Individuals with a passion for artificial intelligence and a basic knowledge of Python who want to dive deep into the world of LLMs and generative AI.
• Tech Innovators and Creators: Aspiring AI developers who wish to be at the cutting edge of technology, learning to build and understand complex AI systems from the ground up.
• Students and Professionals in AI: Those studying or working in AI-related fields who seek to enhance their practical skills and theoretical understanding of LLMs and deep learning.
• Curious Minds with Creative Flair: Learners interested in a unique blend of technology and art, who are eager to explore deep learning concepts through innovative methods like origami.
• Software engineers: interested in understanding and coding LLMs as well as implementing AI alignment techniques with LLMs
• General: People that are curious about how LLMs work and want to understand them in great depth
• General: People that want to stretch their generative AI experience using Python and Pytorch to program LLMs and alignment processes
• General: People that want to know all the key parts of how ChatGPT, Gemini or Claude work internally, in order to get inspired about new ways of applying this technology
• General: Those that want a deep introduction to Generative AI, deep learning and neural networks
• General: Those that want to deeply understand how neural networks learn in a fun way and unique way through Origami