Understand what happens between tokens and text.
Build a connected mental model of tokenization, embeddings, attention, transformer blocks, training, decoding, post-training, and interpretability.
Explain how text becomes tokens and vectors
Trace information through attention and transformer blocks
Connect training objectives to model behavior
Reason about decoding, post-training, and interpretability
Small token sequences and model traces that make each internal operation visible before scaling the idea up.
No prior LLM course is required. Basic algebra and Python familiarity help with optional code.
Open any module to review its exact sections. Progress and completion follow you through the course.
Topics include BPE, Vocabulary, Glitch tokens, and more.
Topics include Word embeddings, Sinusoidal positions, RoPE, and more.
Topics include Q/K/V, Scaled dot-product, Multi-head, and more.
Topics include LayerNorm vs RMSNorm, SwiGLU FFN, Pre-norm, and more.
Topics include Next-token loss, AdamW, Scaling laws, and more.
Topics include Greedy vs sampling, Temperature, Top-k / top-p / min-p, and more.
Topics include SFT, DPO, GRPO / RLVR, and more.
Topics include Mechanistic interpretability, Induction heads, Sparse autoencoders, and more.
Engineers entering LLM application work
Data professionals who need model-level intuition
Technical leaders evaluating LLM claims
These are independent LDS courses that extend the closest skills. Choose the direction that matches what you want to build next.
Build a working language model from scratch.
Build retrieval systems you can measure.
Build the machinery around an agent loop.
The first module establishes the language and example used throughout the rest of the course.
