LLM Foundations
Transformer layers, tokenization, training stages, MoE vs Dense models.
What is an LLM & How It works
beginner~8hLearn the core neural architectures, conditional probability, tokens, and dense vs MoE configurations.
Training Stages: Pretraining, SFT, RLHF, GRPO
intermediate~12hDeconstruct the four stages of building a production reasoning LLM from scratch.
Text Generation: Decoding & Sampling
beginner~6hLearn generation parameters (temperature, top-p, top-k) and local execution runtimes.
7 LLM Generation Parameters
beginner~3hThe 7 levers that control every LLM generation — max tokens, temperature, top-k, top-p, frequency penalty, presence penalty, and stop sequences — plus bonus min-p sampling.
4 LLM Text Generation Strategies + SLED
intermediate~3hGreedy, Multinomial Sampling, Beam Search, and Contrastive Search — the 4 core strategies for picking the next token — plus SLED (Self-Logits Evolution Decoding), a bonus technique that improves factuality using every layer's predictions.
3 Techniques to Train an LLM Using Another LLM (Distillation)
advanced~4hSoft-label, hard-label, and co-distillation — the three ways real production LLMs (Llama 4, Gemma, DeepSeek) train a smaller student model using a larger teacher.
4 Ways to Run LLMs Locally
beginner~2hOllama, LM Studio, vLLM, and llama.cpp — the four most common ways to run LLMs on your own machine, compared for privacy, ease of use, and performance.
Mixture of Experts: Router Training Challenges & Solutions
advanced~3hWhy naively training a Mixture-of-Experts router causes expert under-training, and the two concrete fixes (noise injection + top-K masking, token capacity limits) that solve it.