Udemy
Turn what you know into an opportunity and reach millions around the world.
Learn More
Your cart is empty.
Keep shopping
Large Language Models Demystified: Transformers, RAG, Agents
New
Rating: 5.0 out of 5(5 ratings)
30 students

Large Language Models Demystified: Transformers, RAG, Agents

Learn tokenization, Transformers, attention, training, RLHF, RAG and agents, and see exactly how LLMs generate answers.
Last updated 11/2026
English
English [Auto],

What you'll learn

  • Explain how an LLM turns a prompt into an answer, from tokenization to the final output
  • Understand the key math behind LLMs: vectors, probability, gradients, and backpropagation
  • Break down self-attention, multi-head attention, and the full Transformer architecture
  • Describe how LLMs are trained at scale with next-token prediction, GPUs, and scaling laws
  • Compare fine-tuning, RLHF, and DPO to see how a base model becomes an AI assistant
  • Control generated text with temperature, top-k, and top-p, and learn how the KV cache speeds up inference
  • Explore RAG, tool use, AI agents, and Mixture-of-Experts architectures
  • Recognize why LLMs hallucinate, and how bias, prompt injection, and evaluation work
  • Make models smaller and faster using quantization, distillation, and LoRA
  • Build clear intuition for tokenization methods (BPE, WordPiece, SentencePiece) and embeddings

Course content

5 sections • 39 lectures • 4h 31m total length
  • 1. From Artificial Intelligence to Large Language Models7:07

    Establish the hierarchy AI → Machine Learning → Deep Learning → Generative AI → Foundation Models → LLMs. Introduce NLP, generative models, foundation models, and what makes a model “large.” Explain how an LLM differs from a traditional task-specific ML model.

  • 2. What Is a Language Model Actually Learning?6:52
  • 3. From N-Grams to Neural Language Models to Transformers6:45
  • 4. Anatomy of a Modern Large Language Model7:00
  • 5. How an LLM Learns: From Raw Data to a Base Model6:35
  • 6. From a Base Model to an AI Assistant and Reasoning Model7:16
  • 7. The Complete LLM Lifecycle: From Internet Text to User Answer6:38

Requirements

  • No prior AI experience needed
  • Basic comfort with high-school-level math is helpful (everything is explained from the ground up)
  • Curiosity about how AI works

Description

Have you ever wondered what really happens inside ChatGPT, Claude, or Gemini after you press Enter? This course takes you inside a Large Language Model, step by step, and explains the mechanisms in plain language, without hiding behind buzzwords.

You will start with the basics of AI, machine learning, and language modeling. Then you’ll build the math you need (vectors, probability, gradients, backpropagation) so that nothing feels like magic. From there, you’ll follow the full journey of a prompt: tokenization, embeddings, the Transformer architecture, self-attention, positional encoding, and how models are trained at scale.

You’ll also learn how a raw base model becomes a helpful AI assistant through fine-tuning, RLHF, DPO, and alignment. You’ll see how answers are generated token by token with temperature, top-k, top-p, and the KV cache. Finally, you’ll explore modern topics such as in-context learning, RAG, tool use and AI agents, Mixture-of-Experts, hallucinations, bias, security, evaluation, quantization, LoRA, and the future of LLMs.

What you’ll learn

  • How LLMs work from input text to final output

  • The math behind neural networks, explained simply

  • How tokenization and embeddings turn text into numbers

  • How self-attention and multi-head attention work

  • How LLMs are trained, fine-tuned, and aligned

  • How decoding strategies shape the model’s answers

  • How RAG, tools, and agents extend what LLMs can do

  • Why LLMs hallucinate, and how they are evaluated and made faster

Who this course is for
Students, developers, data professionals, and curious learners who want to truly understand LLMs, not just use them.

Who this course is for:

  • Beginners who want to understand how ChatGPT-style models work
  • Software developers moving into AI and LLM engineering
  • Data science and machine learning students
  • Professionals who want to talk about AI with confidence
  • Anyone preparing for AI/ML interviews