TinyInferCpp-Lab

We hand-roll a toy neural-network inference engine from scratch in C++23 (Input → Dense → ReLU → Dense → Argmax): float32, fixed structure, and a core path with no heap allocation, no exceptions, and no RTTI. It's a teaching project — not a replacement for TFLite Micro / STM32Cube.AI / CMSIS-NN / TinyMaix / NNoM / emlearn.

Articles

• Stage 0 · Project scaffold — A standalone CMake23 project + a Catch2 smoke test; pour the toolchain foundation first.
• Stage 1 · Fixed-dimension Tensor — A compile-time fixed-dimension, row-major, std::array-backed Tensor + std::expected.
• Stage 2 · Dense and Weight Layout — The fully-connected layer y=W·x+b, std::span view storage, weights laid out [Out,In].

What's next (in progress)

Stage 3 (ReLU / Argmax) → Stage 4 (DemoModel wiring) → Stage 5 (NumPy training & export) → Stage 6 (Python/C++ golden-test cross-check) → Stage 7 (embedded-friendliness audit) → Stage 8 (MCU porting plan). The companion project lives at code/volumn_codes/vol8-labs/ai/tiny_ml/.