Tiny LLM from Scratch¶
Build a small language model directly on your laptop. Collect data, train a tokenizer, hand-code a transformer, and run a 10M-parameter model end-to-end in under four hours — then quantize and serve it with llama.cpp.
What this book covers / doesn't cover¶
Covered
nanoGPT-style transformer · BPE · TinyStories/Cosmopedia · AdamW · mixed precision · perplexity · GGUF · llama.cpp
Mentioned only
RoPE · RMSNorm · SwiGLU · GQA · KV cache · LoRA
Out of scope
MoE · RLHF · DPO/GRPO · multi-node · FSDP · 70B+ scale
Prerequisites
Python · intro PyTorch · matrix-multiply intuition · Colab or M1+ Mac
Where to go¶
- Learning system — how each chapter is structured
- Curriculum — all 32 chapters + capstone
- Start Part 1 — why small models, why now