{"product_id":"building-a-frontier-llm-from-chong-lip-phang-9798182010832","title":"Building a Frontier LLM from Scratch: Architecture, Training, Alignment, and Serving of a DeepSeek-Style Mixture-of-Experts Reasoning Model","description":"\u003cp\u003e\u003cb\u003eMost \"build an LLM\" books stop at a small GPT. This one takes you all the way to the frontier.\u003c\/b\u003e\u003c\/p\u003e\u003cp\u003eToday's leading models - DeepSeek-V3, GLM, and the reasoning systems behind them - are not just bigger GPTs. They are sparse Mixture-of-Experts networks with Multi-head Latent Attention, trained in FP8 across thousands of GPUs and taught to reason with reinforcement learning. This book builds that entire modern stack from first principles, one component at a time.\u003c\/p\u003e\u003cp\u003eStarting from tensors and automatic differentiation, you'll implement and understand every layer of a contemporary large language model - tokenization, attention, the transformer block, rotary positions, a decoder-only architecture - and then the techniques that define the frontier: fine-grained Mixture-of-Experts, Multi-head Latent Attention, Multi-Token Prediction, and sparse attention. From there it covers what it actually takes to train, align, and serve such a model at scale.\u003c\/p\u003e\u003cp\u003e\u003cb\u003eWhat you'll understand and build: \u003c\/b\u003e\u003cbr\u003e- The full architecture of a modern MoE language model, component by component\u003cbr\u003e- Pretraining at scale - FP8 training, distributed and pipeline parallelism, stability, and the systems that keep a run alive\u003cbr\u003e- Alignment from SFT and RLHF to DPO and GRPO - the reinforcement-learning recipe behind reasoning models\u003cbr\u003e- Inference and serving - KV-cache optimization, paged attention, quantization, continuous batching\u003cbr\u003e- The research frontier - reasoning, agents, multimodality, and extreme efficiency\u003cbr\u003e- Two full case studies dissecting real frontier models: DeepSeek-V3 and GLM\u003c\/p\u003e\u003cp\u003e\u003cbr\u003e\u003cb\u003eWho it's for: \u003c\/b\u003e engineers, researchers, and serious students who know some Python and want to understand modern LLMs deeply enough to \u003ci\u003ebuild\u003c\/i\u003e one - not just call an API.\u003c\/p\u003e\u003cp\u003eEvery chapter pairs clear explanation with worked examples, illustrative code, and reference tables, and ends with exercises. The result is a single, self-contained path from import torch to a DeepSeek-style Mixture-of-Experts reasoning model.\u003c\/p\u003e\u003cp\u003e\u003cb\u003eStop treating large language models as black boxes. Build one.\u003c\/b\u003e\u003c\/p\u003e\u003cbr\u003e\u003cbr\u003e\u003cb\u003eAuthor:\u003c\/b\u003e Chong Lip Phang\u003cbr\u003e\u003cb\u003eISBN-13:\u003c\/b\u003e 9798182010832\u003cbr\u003e\u003cb\u003ePublisher:\u003c\/b\u003e Independently Published\u003cbr\u003e\u003cb\u003eLanguage:\u003c\/b\u003e English\u003cbr\u003e\u003cb\u003ePublished:\u003c\/b\u003e 06\/17\/2026\u003cbr\u003e\u003cb\u003ePages:\u003c\/b\u003e 422\u003cbr\u003e\u003cb\u003eFormat:\u003c\/b\u003e Paperback\u003cbr\u003e\u003cb\u003eWeight:\u003c\/b\u003e 1.12lbs\u003cbr\u003e\u003cb\u003eSize:\u003c\/b\u003e 9.00h x 6.00w x 1.05d","brand":"Chong Lip Phang","offers":[{"title":"Paperback","offer_id":48997792579839,"sku":"9798182010832","price":19.99,"currency_code":"USD","in_stock":true}],"url":"https:\/\/www.whiterainbookhouse.com\/products\/building-a-frontier-llm-from-chong-lip-phang-9798182010832","provider":"WR Book House","version":"1.0","type":"link"}