Skip to chapter content
The Incin Book
Summary
Introduction
Getting started
Installation
Editor integrations
Quickstart
Core concepts
Tensors
Shapes: static, dynamic, and mixed
Advanced shapes
Autograd
Errors
Building models
Layers and `#[module]`
Sequential models
A small Transformer-style block
Training
Losses, optimizers, and schedulers
Data loading
Quantization
Distributed planning
Metrics
Saving and loading
Backends
CPU, and what actually runs on GPU today
The target API and canonical dispatch
Backend authoring
From proofs to execution
Custom and fused operations
Reference
The macro reference
Every feature flag
Invariants and proof types
Experimental surfaces
Coming from PyTorch
0.1.0 release notes
What's not finished yet
Deep dive
The layered architecture
Type semantics
Lowering: from descriptor to kernel
Proofs: how claims are checked
What the macros guarantee
Light
Rust
Coal
Navy
Ayu
The Incin Book
ESC
Loading chapter...