C-ML: C Machine Learning Library

A machine learning library written in pure C
Automatic differentiation, neural networks, optimizers, datasets, and training utilities

---

Features

• Automatic Differentiation -- Dynamic computation graphs with automatic gradient computation
• Neural Network Layers -- Linear, Conv1d/2d/3d, RNN/LSTM/GRU, Transformer, Embedding, BatchNorm, LayerNorm, GroupNorm, Pooling, Dropout, 12 activation functions
• Containers -- Sequential, ModuleList, ModuleDict
• Optimizers -- SGD, Adam, AdamW, RMSprop, Adagrad, AdaDelta + LR schedulers (Step, Exponential, Cosine, ReduceOnPlateau, MultiStep)
• Loss Functions -- MSE, MAE, BCE, CrossEntropy, Huber, KL Divergence, Hinge, Focal, SmoothL1
• Dataset Hub -- One-liner loading: cml_dataset_load("iris") with auto-download and caching
• Model Zoo -- Pre-built architectures: MLP, ResNet, VGG, GPT-2, BERT
• Model I/O -- Save/load models and training checkpoints
• Tensor Operations -- Comprehensive ops with NumPy-style broadcasting
• SIMD Vectorization -- SSE/AVX/AVX-512/NEON with runtime detection
• BLAS Integration -- Dynamic loading of MKL/OpenBLAS/Accelerate
• Memory Management -- Automatic cleanup, memory pools, graph allocator
• Python Bindings -- CFFI-based Python interface

---

Quick Start

Build

git clone https://github.com/jaywyawhare/C-ML.git
cd C-ML

# CMake build
mkdir -p build && cd build
cmake -DBUILD_EXAMPLES=ON -DBUILD_TESTS=ON ..
make -j$(nproc)

Hello World

#include "cml.h"
#include <stdio.h>

int main(void) {
    cml_init();

    // Load a dataset
    Dataset* ds = cml_dataset_load("iris");
    dataset_normalize(ds, "minmax");
    Dataset *train, *test;
    dataset_split(ds, 0.8f, &train, &test);

    // Build a model
    Sequential* model = cml_nn_sequential();
    DeviceType dev = cml_get_default_device();
    DType dt = cml_get_default_dtype();
    model = cml_nn_sequential_add(model, (Module*)cml_nn_linear(4, 16, dt, dev, true));
    model = cml_nn_sequential_add(model, (Module*)cml_nn_relu(false));
    model = cml_nn_sequential_add(model, (Module*)cml_nn_linear(16, 3, dt, dev, true));

    // Train
    Optimizer* opt = cml_optim_adam_for_model((Module*)model, 0.01f, 0.0f, 0.9f, 0.999f, 1e-8f);
    cml_nn_module_set_training((Module*)model, true);

    for (int epoch = 0; epoch < 100; epoch++) {
        cml_optim_zero_grad(opt);
        Tensor* out = cml_nn_module_forward((Module*)model, train->X);
        Tensor* loss = cml_nn_mse_loss(out, train->y);
        cml_backward(loss, NULL, false, false);
        cml_optim_step(opt);
        tensor_free(loss);
        tensor_free(out);
    }

    cml_cleanup();
    return 0;
}

gcc -std=c11 -O2 example.c -I./include -L./build/lib -lcml_static -lm -ldl -o example
./example

---

Examples

C-ML includes 17 example programs covering common ML tasks:

Example	Description	Dataset
ex01_tensor_ops	Basic tensor creation, arithmetic, reductions	None
ex02_linear_regression	Linear regression with SGD	Boston Housing
ex03_logistic_regression	Binary classification with BCE loss	Breast Cancer
ex04_mlp_classifier	Multi-class MLP classifier	Iris
ex05_autoencoder	Autoencoder with bottleneck	Digits 8x8
ex06_conv_net	Image classification MLP	Digits 8x8
ex07_rnn_sequence	RNN time series prediction	Airline
ex08_lstm_timeseries	LSTM time series forecasting	Airline
ex09_gru_classifier	GRU sequence classifier	Iris
ex10_embedding	Embedding lookup table demo	None
ex11_gan	Generative Adversarial Network	Digits 8x8
ex12_multi_task	Multi-task learning	Wine
ex13_transformer	Transformer encoder with self-attention	None
ex14_lr_scheduler	LR scheduler comparison	Boston Housing
ex15_activations	Activation function showcase	Breast Cancer
hello_cml	Minimal forward pass	None
simple_xor	XOR problem with training loop	XOR

# Run after building
./build/bin/ex04_mlp_classifier

See examples/README.md for details.

---

Dataset Hub

Load datasets with a single function call:

Dataset* ds = cml_dataset_load("iris");        // 150 samples, 4 features, 3 classes
Dataset* ds = cml_dataset_load("mnist");       // 70k samples, 784 features, 10 classes
Dataset* ds = cml_dataset_load("cifar10");     // 60k samples, 3072 features, 10 classes
Dataset* ds = cml_dataset_from_csv("data.csv", -1);  // Custom CSV

Supported datasets: iris, wine, breast_cancer, boston, mnist, fashion_mnist, cifar10, airline, digits

Datasets are automatically downloaded and cached in ~/.cml/datasets/. See docs/datasets.md.

---

Documentation

Guide	Description
Getting Started	Build, install, first program
API Reference	Complete API documentation
Neural Network Layers	All layers with signatures and examples
Training Guide	Optimizers, schedulers, training loops
Datasets	Dataset hub and custom data loading
Autograd	Automatic differentiation guide
Graph Mode	Lazy execution and IR optimization
Optimizations	SIMD, BLAS, fusion, caching internals
IR Graph Management	Memory optimization and kernel export
Kernel Studio	Optimization pass visualization
External Dependencies	Optional library integration
Python Bindings	Python CFFI setup

---

Architecture

C-ML/
├── include/            # Public API headers
│   ├── cml.h           # Main header (include this)
│   ├── tensor/         # Tensor operations
│   ├── autograd/       # Automatic differentiation
│   ├── nn/             # Neural network layers, containers, model I/O
│   ├── ops/ir/         # IR graph, optimization, execution, LLVM backend
│   ├── optim/          # Optimizers and LR schedulers
│   ├── datasets/       # Dataset hub
│   └── zoo/            # Model zoo
├── src/                # Implementation
├── examples/           # 17 example programs
├── tests/              # Test suite
├── python/             # Python CFFI bindings
└── docs/               # Documentation

---

Build Options

# CMake (recommended)
mkdir -p build && cd build
cmake -DBUILD_EXAMPLES=ON -DBUILD_TESTS=ON -DCMAKE_BUILD_TYPE=Release ..
make -j$(nproc)

# Makefile shortcuts
make              # Standard build
make release      # Release build with optimizations
make debug        # Debug build with sanitizers
make test         # Build and run tests

Integration

# Static library
gcc your_program.c -I./include -L./build/lib -lcml_static -lm -ldl -o your_program

# Shared library
gcc your_program.c -I./include -L./build/lib -lcml -lm -ldl -o your_program
export LD_LIBRARY_PATH=./build/lib:$LD_LIBRARY_PATH

---

Testing

cd build && ctest --output-on-failure
# Or: make test

---

Contributing

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

---

License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Made with C

Back to top