from CalcGrad import Network, Value

net = Network([2, 4, 1])  # 2 inputs → 4 hidden → 1 output

data = [
    ([0.0, 0.0], [0.0]),
    ([0.0, 1.0], [1.0]),
    ([1.0, 0.0], [1.0]),
    ([1.0, 1.0], [0.0]),
]

lr = 0.05
for epoch in range(1000):
    total_loss = Value(0.0)
    for x, y in data:
        inputs = [Value(i) for i in x]
        targets = [Value(t) for t in y]
        outputs = net(inputs)
        loss = net.calc_loss(targets, outputs)

        for p in net.parameters(): p.zero_grad()
        loss.backward()
        for p in net.parameters(): p.data -= lr * p.grad

        total_loss += loss

    if epoch % 100 == 0:
        print(f"Epoch {epoch} Loss: {total_loss.data:.4f}")

for x, _ in data:
    out = net([Value(i) for i in x])
    print(f"{x} → {[o.data for o in out]}")