Examples
Every runnable example in examples/ trains or runs a model entirely in C++, writes a header-row CSV to its output/ directory, and ships a plot.py that renders the result. The pages in this section document each example: what it does, how it works, the exact build and run commands, and its graphical output.
Reproduce any plot with:
cd examples/<name>
make release
make run # writes output/*.csv
make plot # writes output/*.png (needs matplotlib in a venv/pyenv)
The plot scripts share one style module, examples/plotting/tinymind_plot.py (matplotlib only, headless-safe). The CSV-first contract means you can also drop the data into pandas or a spreadsheet and build your own visualizations. The plots on these pages use the dark theme to match this site; make plot defaults to a light theme — set TINYMIND_PLOT_THEME=dark to reproduce them exactly.
See the Example Gallery for every plot at a glance, or pick an example from the navigation. Dataset-driven examples are surveyed in the UCI Dataset Capability Report.
Table of contents
- Example Gallery
- XOR (fixed-point MLP)
- KAN XOR
- GRU XOR
- LSTM Sinusoid
- Q-Learning Maze
- DQN Maze
- LTC Sequence (Liquid Time-Constant)
- CfC Sequence (Closed-form Continuous-time)
- int8 QCfC Liquid Cell
- PINN: 1-D Heat Equation
- KWS Cortex-M (float pipeline)
- KWS Cortex-M (int8 pipeline)
- int8 ResNet Residual Block
- int8 ResNet-18 Stem + Block
- int8 MobileNetV2 Pipeline
- int8 Transformer Encoder Block
- SIMD Performance Matrix
- Mixed-Precision KWS (int8 + fp16)
- Mixed-Precision MLP (int8 + Q-format)
- PyTorch Importer Demo
- PyTorch → int8 XOR
- PyTorch → Q-format XOR
- Predictive Maintenance (AI4I 2020)
- Iris Species Classifier
- Energy Efficiency Regression
- Optical Handwritten Digits
- Human Activity Recognition (LSTM)
- Gas Sensor Array Drift
- Air Quality Forecasting (LSTM)
- LSTM Sinusoid (Float vs Q16.16)
- Elman Temporal XOR
- Elman Japanese Vowels
- int8 Transformer Encoder Stack
- int8 Transformer Encoder Stack (Softmax Attention)