Module 08
Large Language Models
The commercially hot layer. The skill that separates people here is evaluation and cost control, not prompting.
14 topics ~81 h estimated learning time
Comes after07. Natural Language Processing
Sign in to keep your progress.
Ready to start
- 8.1 How LLMs are trained
Core · 7 h
- 8.2 Tokenizers, context and positional schemes
Core · 5 h
- 8.3 Decoding and sampling
Core · 4 h
- 8.4 Prompt engineering that survives users
Core · 6 h
- 8.5 Structured output and tool schemas
Core · 5 h
- 8.6 Embeddings and vector databases
Core · 5 h
- 8.7 Retrieval-augmented generation
Core · 8 h
- 8.8 Advanced RAG
Advanced · 6 h
- 8.9 Fine-tuning: LoRA, QLoRA, PEFT
Advanced · 7 h
- 8.10 Quantization, distillation and local models
Advanced · 6 h
- 8.11 Serving LLMs: latency, throughput, cost
Advanced · 6 h
- 8.12 Evaluating LLM systems
Advanced · 6 h
- 8.13 Safety, hallucination and guardrails
Core · 5 h
- 8.14 Multimodal models
Advanced · 5 h