This repository contains all the Python code for the Udemy course:
Deep Understanding of Large Language Models (LLMs): Architecture, Training, and Mechanisms.
Course URL: https://www.udemy.com/course/dullms_x/?couponCode=202508 (90+ hours of intensive video lectures — theory, math, and code)
This course is a comprehensive deep dive into AI large language models — the kind of models that power ChatGPT, Claude, Gemini, and other cutting-edge systems.
It combines mathematical foundations, architectural details, and hands-on PyTorch coding to give you the deepest possible understanding of how LLMs work.
See the list of lectures and topics here: https://docs.google.com/spreadsheets/d/1jTwXQJBh2aGFdazuVsnS98dcwwjGWL1Uw__U5fOuhEE/edit?usp=sharing
- Transformer architecture and self-attention mechanisms
- Tokenization, embeddings, and encoding layers
- Training LLMs from scratch and fine-tuning pre-trained models
- Inference and decoding strategies
- Optimization, regularization, and scaling laws
- Evaluation metrics and understanding LLM limitations
- PyTorch implementation of core components
- Code challenges with full explanations and solutions
- All lecture code organized by course section and video
- Code Challenge solutions — complete, commented Python scripts
- Custom PyTorch implementations of LLM building blocks
- Data preprocessing and visualization utilities
While the code here is 100% free, the Udemy course includes:
- 90+ hours of video content
- Step-by-step walkthroughs of every code example
- Mathematical derivations for all key concepts
- Tips for real-world AI and NLP projects
- Unlike many AI tutorials, this course goes far beyond just calling APIs:
- Understand LLM internals — not just how to use them
- Build your own Transformer components from scratch in PyTorch
- Learn the why, not just the how
- Develop skills for research, engineering, and advanced ML work
Mike X Cohen (that's me :D ) Neuroscientist, educator, and author of multiple books and courses on data science, machine learning, calculus, linear algebra, signal processing, and statistics