tinyML

A lightweight C++ machine-learning library for embedded, edge and real-time workloads.

The repository contains two intentionally different surfaces. TinyML::Core is the supported, dependency-free library surface with an install/export contract and semantic-versioning guarantees. TinyML::Extended contains the xsimd-backed model stack and research modules. The extended target is built by default for source compatibility, but its individual modules have their own stability status documented in docs/STABILITY.md.

Requirements

The core requires a C++17 compiler and CMake 3.20 or newer. It has no third-party runtime or build dependency.

The extended target additionally requires xsimd 12.1 or newer. Top-level source builds can fetch xsimd automatically; package consumers are expected to provide it through their normal dependency manager.

Build

Core only:

cmake -S . -B build -DTINYML_BUILD_EXTENDED=OFF
cmake --build build --parallel
ctest --test-dir build --output-on-failure

Full source build:

cmake -S . -B build
cmake --build build --parallel
ctest --test-dir build --output-on-failure

Benchmarks, examples and the playground are deliberately excluded from ordinary library builds. Enable them explicitly with TINYML_BUILD_BENCHMARKS, TINYML_BUILD_EXAMPLES or TINYML_BUILD_PLAYGROUND.

Install and consume

cmake -S . -B build -DTINYML_BUILD_EXTENDED=OFF -DTINYML_BUILD_TESTS=OFF
cmake --build build --parallel
cmake --install build --prefix ./install

A downstream CMake project can then use the installed core without xsimd:

find_package(TinyML 1.0 CONFIG REQUIRED COMPONENTS Core)
target_link_libraries(my_target PRIVATE TinyML::Core)

#include <tinyml/core.hpp>

ML::Model model ({ 2, 4, 1 });
model.feedForward ({ 0.25, -0.5 });
auto output = model.getResult();

For the extended package:

find_package(TinyML 1.0 CONFIG REQUIRED COMPONENTS Extended)
target_link_libraries(my_target PRIVATE TinyML::Extended)

The installed package includes a generated <tinyml/version.hpp> with TINYML_VERSION_MAJOR, TINYML_VERSION_MINOR, TINYML_VERSION_PATCH and TINYML_VERSION_STRING.

Build options

Option	Default	Purpose
TINYML_BUILD_EXTENDED	ON	Build the xsimd-backed extended library
TINYML_BUILD_TESTS	top-level only	Build the test suite
TINYML_BUILD_BENCHMARKS	OFF	Build local benchmark executables
TINYML_BUILD_EXAMPLES	OFF	Build examples
TINYML_BUILD_PLAYGROUND	OFF	Build the playground server
TINYML_FETCH_DEPENDENCIES	top-level only	Fetch missing xsimd/GoogleTest dependencies
TINYML_ENABLE_WARNINGS	top-level only	Enable compiler warning flags on TinyML targets
TINYML_ENABLE_LTO	OFF	Enable IPO/LTO when supported

Testing and release guarantees

CI builds the zero-dependency core separately from the full library with GCC and Clang. Every configuration is installed into a staging prefix and then consumed by a fresh downstream CMake project through find_package, so packaging regressions fail before release.

Timing assertions are opt-in because shared CI hardware is unsuitable for performance gates. Benchmark executables are local-only. See TESTING.md.

Versioned releases are assembled from cmake --install, not by manually copying build artifacts. Core and extended archives are published separately, with SHA-256 checksums.

API stability

The compatibility contract is explicit rather than implied by the presence of a header in the repository. See docs/STABILITY.md for the stable, preview and source-only surfaces.

Changes to the stable core follow semantic versioning. Preview and source-only modules may change before they graduate into the stable surface.

Licensing

The installable core is MIT licensed. The extended model library contains commercially licensed modules and xsimd-backed functionality. Individual file licensing and the distinction between licensing and API stability are documented in LICENSING.md. Commercial terms are in LICENSE-COMMERCIAL.md.

Security reports should follow SECURITY.md. Contributions should follow CONTRIBUTING.md. Release history is tracked in CHANGELOG.md.