https://github.com/lommix/snail_nn

Neural network and matrix library with parallized learning build from the ground up. Educational project.
https://github.com/lommix/snail_nn

ai calculus gradient-descent machine-learning matrix-library neural-network parallelism

Last synced: 2 months ago
JSON representation

Neural network and matrix library with parallized learning build from the ground up. Educational project.

• Host: GitHub
• URL: https://github.com/lommix/snail_nn
• Owner: Lommix
• License: apache-2.0
• Created: 2023-07-13T09:51:40.000Z (over 1 year ago)
• Default Branch: master
• Last Pushed: 2023-10-17T22:03:41.000Z (about 1 year ago)
• Last Synced: 2024-09-19T03:14:55.161Z (4 months ago)
• Topics: ai, calculus, gradient-descent, machine-learning, matrix-library, neural-network, parallelism
• Language: Rust
• Homepage:
• Size: 618 KB
• Stars: 3
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# [WIP] Snail NN - smol neural network library

Minimalistic CPU based neural network library with backpropagation and parallelized stochastic gradient descent.

## Examples

Storing images inside the neural network, upscaling and interpolate between them.

```bash

cargo run --example imagepol --release

```

![image](docs/example_interpolation.png)

---

The mandatory xor example

```bash

cargo run --example xor --release

```

![image](docs/xor.png)

---

Example Code:

```rust

use snail_nn::prelude::*;

fn main(){

    let mut nn = Model::new(&[2, 3, 1]);

    nn.set_activation(Activation::Sigmoid)

    let mut batch = TrainingBatch::empty(2, 1);

    let rate = 1.0;

    // AND - training data

    batch.add(&[0.0, 0.0], &[0.0]);

    batch.add(&[1.0, 0.0], &[0.0]);

    batch.add(&[0.0, 1.0], &[0.0]);

    batch.add(&[1.0, 1.0], &[1.0]);

    for _ in 0..10000 {

        let (w_gradient, b_gradient) = nn.gradient(&batch.random_chunk(2));

        nn.learn(w_gradient, b_gradient, rate);

    }

    println!("ouput {:?} expected: 0.0", nn.forward(&[0.0, 0.0]));

    println!("ouput {:?} expected: 0.0", nn.forward(&[1.0, 0.0]));

    println!("ouput {:?} expected: 0.0", nn.forward(&[0.0, 1.0]));

    println!("ouput {:?} expected: 1.0", nn.forward(&[1.0, 1.0]));

}

```

## Features

-   Sigmoid, Tanh & Relu activation functions

-   Parallelized stochastic gradient descent

## Todos

-   Wgpu compute shaders