optimizer.py

import torch
from torch.optim import Optimizer


class CustomAdamW(Optimizer):
    def __init__(
        self, params, lr=0.001, betas=(0.9, 0.999), eps=1e-8, weight_decay=0.01
    ):
        if not 0.0 <= lr:
            raise ValueError("Invalid learning rate: {}".format(lr))
        if not 0.0 <= eps:
            raise ValueError("Invalid epsilon value: {}".format(eps))
        if not 0.0 <= betas[0] < 1.0:
            raise ValueError("Invalid beta parameter at index 0: {}".format(betas[0]))
        if not 0.0 <= betas[1] < 1.0:
            raise ValueError("Invalid beta parameter at index 1: {}".format(betas[1]))
        if not 0.0 <= weight_decay:
            raise ValueError("Invalid weight_decay value: {}".format(weight_decay))

        defaults = dict(lr=lr, betas=betas, eps=eps, weight_decay=weight_decay)
        super(CustomAdamW, self).__init__(params, defaults)

    def step(self, closure=None):
        loss = None
        if closure is not None:
            loss = closure()

        for group in self.param_groups:
            for p in group["params"]:
                if p.grad is None:
                    continue
                grad = p.grad.data
                state = self.state[p]

                # State initialization
                if len(state) == 0:
                    state["step"] = 0
                    # Exponential moving average of gradient values
                    state["exp_avg"] = torch.zeros_like(p.data)
                    # Exponential moving average of squared gradient values
                    state["exp_avg_sq"] = torch.zeros_like(p.data)

                exp_avg, exp_avg_sq = state["exp_avg"], state["exp_avg_sq"]
                beta1, beta2 = group["betas"]

                state["step"] += 1

                # Decay the first and second moment running average coefficient
                exp_avg.mul_(beta1).add_(grad, alpha=1 - beta1)
                exp_avg_sq.mul_(beta2).addcmul_(grad, grad, value=1 - beta2)

                denom = exp_avg_sq.sqrt().add_(group["eps"])

                step_size = group["lr"]
                bias_correction1 = 1 - beta1 ** state["step"]
                bias_correction2 = 1 - beta2 ** state["step"]
                step_size = step_size * bias_correction2**0.5 / bias_correction1

                p.data.addcdiv_(exp_avg, denom, value=-step_size)

                # Weight decay
                if group["weight_decay"] > 0.0:
                    p.data.add_(p.data, alpha=-group["lr"] * group["weight_decay"])

        return loss


# Exemple d'utilisation
model = SimpleNN()
custom_adamw = CustomAdamW(model.parameters(), lr=0.001, weight_decay=0.01)


# Boucle d'entraînement (similaire à l'exemple précédent)
def train(model, device, train_loader, optimizer, criterion, epoch):
    model.train()
    for batch_idx, (data, target) in enumerate(train_loader):
        data, target = data.to(device), target.to(device)
        optimizer.zero_grad()
        output = model(data)
        loss = criterion(output, target)
        loss.backward()
        optimizer.step()
        if batch_idx % 100 == 0:
            print(
                f"Train Epoch: {epoch} [{batch_idx * len(data)}/{len(train_loader.dataset)} "
                f"({100. * batch_idx / len(train_loader):.0f}%)]\tLoss: {loss.item():.6f}"
            )


# Vérifier si un GPU est disponible
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.to(device)

# Entraîner le modèle avec l'optimiseur AdamW personnalisé
for epoch in range(1, 3):
    train(model, device, train_loader, custom_adamw, criterion, epoch)
    test(model, device, test_loader, criterion)