from torch import nn 


# 定義神經(jīng)網(wǎng)絡(luò)模型
class NeuralNetwork(nn.Module):
    def __init__(self):
        super().__init__()
        self.flatten = nn.Flatten()  # 將 1x28x28 展平為 784
        self.linear_relu_stack = nn.Sequential(
            nn.Linear(28*28, 128),
            nn.ReLU(),
            nn.Linear(128, 64),
            nn.ReLU(),
            nn.Linear(64, 10)  # 最終10類輸出
        )

    def forward(self, x):
        x = self.flatten(x)
        logits = self.linear_relu_stack(x)
        return logits

import torch
from torch import nn
from torch.utils.data import DataLoader
from torchvision import datasets, transforms
import matplotlib.pyplot as plt
from neural_network_model import NeuralNetwork
from torchvision.transforms import ToTensor


# 下載 FashionMNIST 訓(xùn)練集和測(cè)試集
training_data = datasets.FashionMNIST(
    root="data",
    train=True,
    download=True,
    transform=ToTensor()
)

test_data = datasets.FashionMNIST(
    root="data",
    train=False,
    download=True,
    transform=ToTensor()
)

# 用 DataLoader 封裝
train_dataloader = DataLoader(training_data, batch_size=64, shuffle=True)
# batch_size=64：每次迭代從訓(xùn)練集中取出 64 個(gè)樣本。
# shuffle=True：每輪訓(xùn)練（epoch）前會(huì)打亂數(shù)據(jù)順序，提高訓(xùn)練效果，防止模型記住順序。
test_dataloader = DataLoader(test_data, batch_size=64, shuffle=True)


# 設(shè)置訓(xùn)練設(shè)備、實(shí)例化模型和損失函數(shù)
device = "cuda" if torch.cuda.is_available() else "cpu"
model = NeuralNetwork().to(device) # 創(chuàng)建該網(wǎng)絡(luò)的一個(gè)實(shí)例對(duì)象并存儲(chǔ)到設(shè)備

loss_fn = nn.CrossEntropyLoss()  #設(shè)置損失函數(shù)
optimizer = torch.optim.SGD(model.parameters(), lr=1e-2)


# 訓(xùn)練函數(shù)
def train_loop(dataloader, model, loss_fn, optimizer):
    size = len(dataloader.dataset)
    num_batches = len(dataloader)
    model.train()

    total_loss = 0
    correct = 0

    for X, y in dataloader:
        X, y = X.to(device), y.to(device)

        pred = model(X)
        loss = loss_fn(pred, y)

        optimizer.zero_grad()
        loss.backward()
        optimizer.step()

        total_loss += loss.item()
        correct += (pred.argmax(1) == y).type(torch.float).sum().item()

    avg_loss = total_loss / num_batches
    accuracy = correct / size
    return avg_loss, accuracy

# 測(cè)試函數(shù)
def test_loop(dataloader, model, loss_fn):
    size = len(dataloader.dataset)
    num_batches = len(dataloader)
    model.eval()

    total_loss = 0
    correct = 0

    with torch.no_grad():
        for X, y in dataloader:
            X, y = X.to(device), y.to(device)

            pred = model(X)
            loss = loss_fn(pred, y)

            total_loss += loss.item()
            correct += (pred.argmax(1) == y).type(torch.float).sum().item()

    avg_loss = total_loss / num_batches
    accuracy = correct / size
    return avg_loss, accuracy



# 準(zhǔn)備畫(huà)圖數(shù)據(jù)
train_losses, train_accuracies = [], []
test_losses, test_accuracies = [], []


# 訓(xùn)練過(guò)程
epochs = 10
for epoch in range(epochs):
    print(f"Epoch {epoch+1}/{epochs}")

    train_loss, train_acc = train_loop(train_dataloader, model, loss_fn, optimizer)
    test_loss, test_acc = test_loop(test_dataloader, model, loss_fn)

    train_losses.append(train_loss)
    train_accuracies.append(train_acc)
    test_losses.append(test_loss)
    test_accuracies.append(test_acc)

    print(f"Train Loss: {train_loss:.4f}, Train Acc: {train_acc:.4f}")
    print(f"Test Loss: {test_loss:.4f}, Test Acc: {test_acc:.4f}")
    print("-" * 50)


# 設(shè)置支持中文顯示
plt.rcParams['font.family'] = 'SimHei'
plt.rcParams['axes.unicode_minus'] = False

# 繪制 loss / accuracy 曲線
epochs_range = range(1, epochs + 1)

plt.figure(figsize=(10, 4))

# Loss 曲線
plt.subplot(1, 2, 1)
plt.plot(epochs_range, train_losses, label="Train Loss 訓(xùn)練損失")
plt.plot(epochs_range, test_losses, label="Test Loss 測(cè)試損失")
plt.xlabel("Epoch 訓(xùn)練輪數(shù)")
plt.ylabel("Loss 損失")
plt.title("Loss Curve 損失曲線")
plt.legend()
plt.grid(True)

# Accuracy 曲線
plt.subplot(1, 2, 2)
plt.plot(epochs_range, train_accuracies, label="Train Acc 訓(xùn)練準(zhǔn)確率")
plt.plot(epochs_range, test_accuracies, label="Test Acc 測(cè)試準(zhǔn)確率")
plt.xlabel("Epoch 訓(xùn)練輪數(shù)")
plt.ylabel("Accuracy 準(zhǔn)確率")
plt.title("Accuracy Curve 準(zhǔn)確率曲線")
plt.legend()
plt.grid(True)

plt.tight_layout()
plt.show()