Part 4 Advanced concepts

This last part of the book covers the most advanced concepts. Chapters 9 and 10 explain advanced quantization techniques for larger SLMs and one way of implementing friendly model profiling insights. Chapters 11 and 12 showcase multiple ways to deploy and serve SLMs in hardware-constrained environments, laptops, and devices (specifically presenting options for the Android OS). Chapters 13 and 14 cover the integration of SLMs with RAG and agentic AI. Finally, chapter 15 talks about test-time compute and SLMs, and it concludes the book with an end-to-end example of tuning an SLM for reasoning on a specific domain through GRPO.