Getting Started
These tutorials walk through complete, working examples that demonstrate TinyMind’s capabilities on real problems. Each tutorial includes source code, build instructions, and size analysis.
| Tutorial | What You’ll Learn | Final Size |
|---|---|---|
| Neural Network in Under 4KB | Feed-forward NN with fixed-point, XOR prediction | ~2.3 KB trainable (~1.7 KB inference-only) |
| Q-Learning in Under 1KB | Tabular Q-learning, maze solving | 869 bytes |
| DQN Maze Solver | Deep Q-Network with neural network function approximation | ~16 KB |
| Keyword Spotting CNN on a Cortex-M | Depthwise-separable 2D CNN, bench harness, MCU porting | ~19 KB static |
| Predictive Maintenance on AI4I 2020 | Q16.16 MLP, imbalanced binary classification, confusion matrix | ~35 KB static |
| PyTorch -> TinyMind int8 (XOR) | End-to-end post-training int8 quantization: PyTorch float training, per-tensor calibration, pure-integer C++ inference | Tiny |
| PyTorch -> TinyMind int8 (importer) | Production flow: torch.state_dict -> tinymind_import.py -> weights.hpp. PercentileObserver / KLDivergenceObserver / cross-layer equalization | Tiny |
| Keyword Spotting CNN (int8) | int8 quantized depthwise-separable CNN, per-channel depthwise, CSV cycle/byte report vs float | ~5 KB static |
| MobileNetV2-shaped int8 | int8 exemplar: stride-2 stem + inverted-residual blocks + GAP + dense, linear-bottleneck convention, golden-byte regression | Compact |