我是AI和Python的新手，我试图只运行一个批次以达到过拟合的目的。我找到了以下代码：iter(train_loader).next()

但我不确定该在我的代码中哪里实现它。即使我实现了，我如何在每次迭代后检查以确保我在训练同一个批次？

train_loader = torch.utils.data.DataLoader(    dataset_train,    batch_size=48,    shuffle=True,    num_workers=2)net = nn.Sequential(    nn.Flatten(),    nn.Linear(128*128*3,10))nepochs = 3statsrec = np.zeros((3,nepochs))loss_fn = nn.CrossEntropyLoss()optimizer = optim.Adam(net.parameters(), lr=0.001)for epoch in range(nepochs):  # 多次遍历数据集    running_loss = 0.0    n = 0    for i, data in enumerate(train_loader, 0):        inputs, labels = data                 # 清零参数梯度        optimizer.zero_grad()         # 前向传播、反向传播和更新参数        outputs = net(inputs)        loss = loss_fn(outputs, labels)        loss.backward()        optimizer.step()            # 累积损失        running_loss += loss.item()        n += 1        ltrn = running_loss/n    ltst, atst = stats(train_loader, net)    statsrec[:,epoch] = (ltrn, ltst, atst)    print(f"epoch: {epoch} 训练损失: {ltrn: .3f} 测试损失: {ltst: .3f} 测试准确率: {atst: .1%}")

请给我一些提示

回答：

如果你想在一个批次上进行训练，那么请移除你对数据加载器的循环：

for i, data in enumerate(train_loader, 0):    inputs, labels = data

并且在遍历epoch之前简单地获取train_loader迭代器的第一个元素，否则next会在每次迭代时被调用，你将会在每个epoch运行不同的批次：

inputs, labels = next(iter(train_loader))i = 0for epoch in range(nepochs):    optimizer.zero_grad()     outputs = net(inputs)    loss = loss_fn(outputs, labels)    loss.backward()    optimizer.step()    # ...