AI Book
Sign InStart Learning
New! Crystallography volume now in circulation

Read AI like a manuscript.Build it from scratch.

A catalogue of 17 technical volumes on transformers, deep learning, probability and the systems they run on. Each one written for the reader who intends to rebuild the thing.

Get started — it's freeBegin with Transformers
17
Volumes
452
Chapters
2464
Lessons
∞
Re-reads
How we write

Go from theory to mastery.

Each volume is engineered to make complex AI accessible — without losing the rigour that lets you actually rebuild the thing in production.

Read the machinery.

Attention, gradients, quantisation — every mechanism is reassembled from first principles, not summarised.

Math, code, diagram — together.

No tab-switching. The equation, its PyTorch translation, and the mental model live in the same paragraph.

Finish what tutorials won't.

Past the demo — into production tradeoffs, failure modes, cost curves, and the second-order tricks.

Features

Features that make the long read easier.

Notes, bookmarks, code playgrounds, 3D visualizations — every tool you need to actually finish a technical book.

Featured volume
QKVAOTRANSFORMERSFROM SCRATCH

Transformers from Scratch

Master the Transformer architecture from scratch. Learn attention mechanisms, positional encoding, and build a complete German-to-English translation model.

Read this volume
Interactive

3D visualizations for every concept.

Rotate a tensor, drag a vector, watch a transformer block fire — built with Three.js, never as static GIFs.

Code-trace

Every line of PyTorch, hand-explained.

Each function gets a teacher-grade walkthrough: arguments, shapes, actual values, edge cases. No skipped iterations.

Volumes in circulation

Pick a volume. Start a habit.

Browse full library
QKVAOTRANSFORMERSFROM SCRATCH
№ 01
AI & Machine Learning·Advanced

Transformers from Scratch

Master the Transformer architecture from scratch. Learn attention mechanisms, positional encoding, and build a complete German-to-English translation model.

18 chapters · 75 lessons
t=4t=3t=2t=1t=0x_t = √ᾱ_t x_0 + √(1-ᾱ_t) εDIFFUSIONMODELS
№ 02
AI & Machine Learning·Advanced

Denoising Diffusion Models from Scratch

Master diffusion models from mathematical foundations to image generation. Learn the theory, implement DDPM in PyTorch, and understand modern systems like Stable Diffusion.

18 chapters · 76 lessons
INPUTHIDDENOUTPUTDEEP LEARNINGNEURAL NETWORKS
№ 03
AI & Machine Learning·Intermediate

Deep Learning from Scratch with PyTorch

Master deep learning from mathematical foundations to production-ready models. Cover CNNs, RNNs, Transformers, GNNs, GANs, VAEs, Reinforcement Learning, and more with PyTorch implementations.

37 chapters · 179 lessons
σ(z)x₁w₁x₂w₂x₃w₃x₄w₄Σwx+bŷactivateFORWARDBACKWARDNEURAL NETWORKSHOW THEY WORK
№ 04
AI & Machine Learning·Beginner

How Neural Networks Work

Understand neural networks from the ground up with Python and PyTorch. From single neurons to CNNs and RNNs, build intuition through code and visualizations.

20 chapters · 65 lessons
🧠ReadWriteBashPerceive → Reason → ActAGENTIC AISYSTEMSClaude Code • Codex • Gemini
№ 05
AI & Machine Learning·Advanced

Building Agentic AI Systems

Master the art of building production-ready AI agents. Learn from Claude Code, Gemini, and Codex architectures. Build multi-agent systems with tool use, memory, and safety guardrails.

29 chapters · 175 lessons
Failure ZoneHealthyWarningCriticalRemaining Useful LifeRUL: 42 cyclesCNNBiLSTMAttentionDEEP LEARNING FORPREDICTIVE MAINTENANCEAMNL: State-of-the-Art on NASA C-MAPSS
№ 06
AI & Machine Learning·Advanced

Gradient-Aware Multi-Task Learning for Predictive Maintenance: AMNL, GABA, and GRACE

A research-grade walkthrough of three multi-task learning strategies for Remaining Useful Life prediction — AMNL (accuracy-first), GABA (safety-first), and GRACE (balanced) — all built on a shared CNN-BiLSTM-Attention backbone. Validated across 335 experiments on NASA C-MAPSS and N-CMAPSS DS02, beating the published SOTA (DKAMFormer) on multi-condition data.

29 chapters · 121 lessons
Pricing

Pricing so simple, you'd read instantly.

Start free. Upgrade only when you want progress sync across devices, premium 3D playgrounds, and the full code-trace playground.

Reader
$0/forever
Start reading
  • All published volumes
  • Interactive 3D visualizations
  • Code-trace walkthroughs
  • Community access
Scholar
Featured
$12/month
Get Scholar
  • Everything in Reader
  • Progress + notes across devices
  • Tensor & attention playgrounds
  • Downloadable PDFs
  • Priority email support
Team
$49/seat/mo
Talk to us
  • Everything in Scholar
  • Team reading-group rooms
  • Shared notebooks & bookmarks
  • SSO + invoicing
  • Onboarding session

Open a volume. Finish the chapter.

17 volumes · 452 chapters · 2464 lessons.

Open the libraryCreate a free reader
No credit card. Cancel anytime. All volumes available on free.