layer/
layer.rs

1/// Example of a simple neural network layer with a layer of neurons.
2///
3/// This example demonstrates how to create a simple neural network layer with a layer of neurons.
4/// The layer has three inputs and one output. The output is the weighted sum of the inputs.
5///
6/// Contrast this example with the `neuron` example.
7/// 
8/// For more information, see [`Layer`].
9extern crate alpha_micrograd_rust;
10
11use alpha_micrograd_rust::nn::{Activation, Layer};
12use alpha_micrograd_rust::value::Expr;
13
14fn main() {
15    let mut target = vec![Expr::new_leaf(15.0, "t1"), Expr::new_leaf(85.0, "t2")];
16    target[0].is_learnable = false;
17    target[1].is_learnable = false;
18
19    let layer = Layer::new(3, 2, Activation::None);
20    println!("Initial values: {:}", layer);
21
22    let mut inputs = vec![
23        Expr::new_leaf(1.0, "x_1"),
24        Expr::new_leaf(2.0, "x_2"),
25        Expr::new_leaf(3.0, "x_3"),
26    ];
27
28    inputs.iter_mut().for_each(|input| {
29        input.is_learnable = false;
30    });
31
32    let mut y = layer.forward(inputs);
33    let mut y1 = y.remove(0);
34    y1.name = "y1".to_string();
35    let mut y2 = y.remove(0);
36    y2.name = "y2".to_string();
37
38    let d1 = y1 - target[0].clone();
39    let mut sqr1 = Expr::new_leaf(2.0, "square_exponent1");
40    sqr1.is_learnable = false;
41
42    let d2 = y2 - target[1].clone();
43    let mut sqr2 = Expr::new_leaf(2.0, "square_exponent2");
44    sqr2.is_learnable = false;
45
46    let mut loss = d1.pow(sqr1, "diff1") + d2.pow(sqr2, "diff2");
47
48    let t1 = loss.find("t1").unwrap();
49    let t2 = loss.find("t2").unwrap();
50    let y1 = loss.find("y1").unwrap();
51    let y2 = loss.find("y2").unwrap();
52
53    println!("Initial targets: {:.2}, {:.2}", t1.result, t2.result);
54    println!("Predicted: {:.2}, {:.2}", y1.result, y2.result);
55    println!("Initial loss: {:.2}", loss.result);
56
57    println!("\nTraining:");
58    let learning_rate = 0.004;
59    for i in 1..=100 {
60        loss.learn(learning_rate);
61        loss.recalculate();
62
63        let t1 = loss.find("t1").unwrap();
64        let t2 = loss.find("t2").unwrap();
65        let y1 = loss.find("y1").unwrap();
66        let y2 = loss.find("y2").unwrap();
67
68        println!(
69            "Iteration {:3}, loss: {:9.4} / predicted: {:5.2}, {:5.2} (targets: {:5.2}, {:5.2})",
70            i, loss.result, y1.result, y2.result, t1.result, t2.result
71        );
72    }
73
74    println!("Final values: {:}", layer);
75}
layer/layer.rs

layer/
layer.rs
