Backpropagation

When the output is wrong, how much is each weight to blame?

A tiny network, one training example, and every number shown. The table at the bottom is the heart of how all neural networks learn: for each weight, backprop multiplies a few local slopes together to get "how much does the error change if I nudge this weight?"

The network after 2440 updates

w1 = 0.321 w2 = 0.541 w3 = 0.419 w4 = 0.638 w5 = −3.424 w6 = −3.397 w7 = 2.397 w8 = 2.458 i1 0.050 i2 0.100 h1 0.604 h2 0.607 o1 0.029 o2 0.972 target 0.01 target 0.99 bias b1 = 0.35 output bias b2 = 0.6

Train it

0.5
0.000338total error E
0.0285, 0.9718outputs (want 0.01, 0.99)
0.00000.0756100.1491,2200.2241,8300.2982,440updates

1. Forward pass: compute the answer

net_h1 = w1·i1 + w2·i2 + b1 = 0.3205·0.05 + 0.5411·0.10 + 0.35 = 0.420132
h1 = σ(net_h1) = 0.603515    h2 = σ(0.434792) = 0.607017
net_o1 = w5·h1 + w6·h2 + b2 = −3.528782    o1 = 0.028504
net_o2 = w7·h1 + w8·h2 + b2 = 3.538211    o2 = 0.971756
E = ½(t1 − o1)² + ½(t2 − o2)² = 0.000171 + 0.000166 = 0.000337632

What to notice

Output weights are easy. w5's blame is just three numbers multiplied: how wrong o1 is, how steep o1's sigmoid is, and how big h1 was (a big input means the weight mattered more).

Hidden weights reuse that work. w1 can't see the error directly. Its blame arrives through both outputs: output 1's blame times w5, plus output 2's blame times w7. Then the same local slopes. That reuse is what makes backprop fast enough for billions of weights.

The check column agrees. "Nudge and measure" gives the same numbers without any calculus, but it needs two extra forward passes per weight. For an LLM that would be billions of passes per step; backprop needs one.

2. Backward pass: blame for every weight

WeightLocal slopes multiplied together (chain rule)∂E/∂wCheck (nudge)New value = w − 0.50 × ∂E/∂w
w5 (o₁ − t₁) 0.01850 × o₁(1 − o₁) 0.02769 × h₁ 0.60351 0.0003093 0.0003093 −3.42439 → −3.42454663
w6 (o₁ − t₁) 0.01850 × o₁(1 − o₁) 0.02769 × h₂ 0.60702 0.0003110 0.0003110 −3.39712 → −3.39727586
w7 (o₂ − t₂) −0.01824 × o₂(1 − o₂) 0.02745 × h₁ 0.60351 −0.0003022 −0.0003022 2.39653 → 2.39667615
w8 (o₂ − t₂) −0.01824 × o₂(1 − o₂) 0.02745 × h₂ 0.60702 −0.0003040 −0.0003040 2.45771 → 2.45786302
w1 δo₁·w5 + δo₂·w7 −0.00295 × h₁(1 − h₁) 0.23928 × i₁ 0.05000 −0.0000354 −0.0000354 0.32053 → 0.32054507
w2 δo₁·w5 + δo₂·w7 −0.00295 × h₁(1 − h₁) 0.23928 × i₂ 0.10000 −0.0000707 −0.0000707 0.54105 → 0.54109014
w3 δo₁·w6 + δo₂·w8 −0.00297 × h₂(1 − h₂) 0.23855 × i₁ 0.05000 −0.0000354 −0.0000354 0.41917 → 0.41918414
w4 δo₁·w6 + δo₂·w8 −0.00297 × h₂(1 − h₂) 0.23855 × i₂ 0.10000 −0.0000709 −0.0000709 0.63833 → 0.63836828

At the start with learning rate 0.5 you should see w5 → 0.35891648 and w1 → 0.14978072, the same as the original worked example. δo means (o − t)·o(1 − o), the output neuron's blame.

The code that just ran

Everything on this page came from calling this code, which lives at the top of this same file.

Show or hide the source
final class TinyNet
{
    public array $w = ['w1' => 0.15, 'w2' => 0.20, 'w3' => 0.25, 'w4' => 0.30,
                       'w5' => 0.40, 'w6' => 0.45, 'w7' => 0.50, 'w8' => 0.55,
                       'b1' => 0.35, 'b2' => 0.60];   // biases stay fixed, as in the original example
    public array $in = [0.05, 0.10];
    public array $target = [0.01, 0.99];

    public function forward(): array
    {
        $w = $this->w; [$i1, $i2] = $this->in; [$t1, $t2] = $this->target;
        $nh1 = $w['w1'] * $i1 + $w['w2'] * $i2 + $w['b1'];  $h1 = sigmoid($nh1);
        $nh2 = $w['w3'] * $i1 + $w['w4'] * $i2 + $w['b1'];  $h2 = sigmoid($nh2);
        $no1 = $w['w5'] * $h1 + $w['w6'] * $h2 + $w['b2'];  $o1 = sigmoid($no1);
        $no2 = $w['w7'] * $h1 + $w['w8'] * $h2 + $w['b2'];  $o2 = sigmoid($no2);
        $E1 = 0.5 * ($t1 - $o1) ** 2;
        $E2 = 0.5 * ($t2 - $o2) ** 2;
        return compact('i1', 'i2', 't1', 't2', 'nh1', 'h1', 'nh2', 'h2', 'no1', 'o1', 'no2', 'o2', 'E1', 'E2') + ['E' => $E1 + $E2];
    }

    public function loss(): float { return $this->forward()['E']; }

    // BACKPROPAGATION. Each gradient is a product of simple local slopes:
    //   (o − t)      how the error changes with the output
    //   o·(1 − o)    how the sigmoid output changes with its input
    //   h or i       how that input changes with the weight
    // Hidden weights get the SUM of the blame flowing back from both outputs.
    public function gradients(): array
    {
        $f = $this->forward(); $w = $this->w;
        $do1 = ($f['o1'] - $f['t1']) * $f['o1'] * (1 - $f['o1']);   // δ of output 1
        $do2 = ($f['o2'] - $f['t2']) * $f['o2'] * (1 - $f['o2']);   // δ of output 2
        $back1 = $do1 * $w['w5'] + $do2 * $w['w7'];                 // blame reaching h1
        $back2 = $do1 * $w['w6'] + $do2 * $w['w8'];                 // blame reaching h2
        $s1 = $f['h1'] * (1 - $f['h1']);
        $s2 = $f['h2'] * (1 - $f['h2']);

        $g = fn($factors) => ['factors' => $factors, 'value' => array_product(array_column($factors, 1))];
        return [
            'w5' => $g([['(o₁ − t₁)', $f['o1'] - $f['t1']], ['o₁(1 − o₁)', $f['o1'] * (1 - $f['o1'])], ['h₁', $f['h1']]]),
            'w6' => $g([['(o₁ − t₁)', $f['o1'] - $f['t1']], ['o₁(1 − o₁)', $f['o1'] * (1 - $f['o1'])], ['h₂', $f['h2']]]),
            'w7' => $g([['(o₂ − t₂)', $f['o2'] - $f['t2']], ['o₂(1 − o₂)', $f['o2'] * (1 - $f['o2'])], ['h₁', $f['h1']]]),
            'w8' => $g([['(o₂ − t₂)', $f['o2'] - $f['t2']], ['o₂(1 − o₂)', $f['o2'] * (1 - $f['o2'])], ['h₂', $f['h2']]]),
            'w1' => $g([['δo₁·w5 + δo₂·w7', $back1], ['h₁(1 − h₁)', $s1], ['i₁', $f['i1']]]),
            'w2' => $g([['δo₁·w5 + δo₂·w7', $back1], ['h₁(1 − h₁)', $s1], ['i₂', $f['i2']]]),
            'w3' => $g([['δo₁·w6 + δo₂·w8', $back2], ['h₂(1 − h₂)', $s2], ['i₁', $f['i1']]]),
            'w4' => $g([['δo₁·w6 + δo₂·w8', $back2], ['h₂(1 − h₂)', $s2], ['i₂', $f['i2']]]),
        ];
    }

    // Sanity check without any calculus: nudge the weight up and down a hair
    // and see how much the loss moves. Should match backprop almost exactly.
    public function numericalGradient(string $name, float $eps = 1e-5): float
    {
        $keep = $this->w[$name];
        $this->w[$name] = $keep + $eps; $up = $this->loss();
        $this->w[$name] = $keep - $eps; $down = $this->loss();
        $this->w[$name] = $keep;
        return ($up - $down) / (2 * $eps);
    }

    public function step(float $lr): void
    {
        foreach ($this->gradients() as $name => $g) $this->w[$name] -= $lr * $g['value'];
    }
}