Adding `gradients` library as a dependency

Run this command in a terminal, in your project's directory:

cargo add gradients

To add it manually, edit your project's Cargo.toml file and add to the [dependencies] section:

gradients = "0.3.4"

The gradients library will be automatically available globally. Read the gradients library documentation.

Back to the crate overview.

Readme

gradients

Deep Learning library using custos and custos-math.

external (C) dependencies: OpenCL, CUDA, nvrtc, cublas, a BLAS lib (OpenBLAS, Intel MKL, ...)

Installation

There are two features available that are enabled by default:

cuda ... CUDA, nvrtc and cublas must be installed
opencl ... OpenCL is needed

If you deactivate them (add default-features = false and provide no additional features), only the CPU device can be used.

For all feature-configurations, a BLAS library needs to be installed on the system.

[dependencies]
gradients = "0.3.4"

# to disable the default features (cuda, opencl) and use an own set of features:
#gradients = {version = "0.3.4", default-features = false, features=["opencl"]}

MNIST example

(if this example does not compile, consider looking here)

Use a struct that implements the NeuralNetwork trait (it is implemented via the network attribute) to define which layers you want to use:

use gradients::purpur::{CSVLoader, CSVReturn, Converter};
use gradients::OneHotMat;
use gradients::{
    correct_classes,
    nn::{cce, cce_grad},
    range, Adam, CLDevice, Linear, network, ReLU, Softmax,
};

#[network]
pub struct Network {
    lin1: Linear<784, 128>,
    relu1: ReLU,
    lin2: Linear<128, 10>,
    relu2: ReLU,
    lin3: Linear<10, 10>,
    softmax: Softmax,
}

Load data and create an instance of Network:

You can download the mnist dataset here.

// use cpu (no features enabled): let device = gradients::CPU::new().select();
// use cuda device (cuda feature enabled): let device = gradients::CudaDevice::new(0).unwrap().select();
// use opencl device (opencl feature enabled):
let device = CLDevice::new(0)?;

let mut net = Network::with_device(&device);

let loader = CSVLoader::new(true);
let loaded_data: CSVReturn<f32> = loader.load("PATH/TO/DATASET/mnist_train.csv")?;

let i = Matrix::from((
    &device,
    (loaded_data.sample_count, loaded_data.features),
    &loaded_data.x,
));
let i = i / 255.;

let y = Matrix::from((&device, (loaded_data.sample_count, 1), &loaded_data.y));
let y = y.onehot();

Training loop:

let mut opt = Adam::new(0.01);

for epoch in range(200) {
    let preds = net.forward(&i);
    let correct_training = correct_classes(&loaded_data.y.as_usize(), &preds) as f32;

    let loss = cce(&device, &preds, &y);
    println!(
        "epoch: {epoch}, loss: {loss}, training_acc: {acc}",
        acc = correct_training / loaded_data.sample_count() as f32
    );

    let grad = cce_grad(&device, &preds, &y);
    net.backward(&grad);
    opt.step(&device, net.params());
}

Adding gradients library as a dependency

Adding `gradients` library as a dependency