在您的例子中，PyTorch只是抱怨nn.BatchNorm1d的输入的形状，这里的输入应该具有(B, C, L)的形状。C是嵌入维数，L是输入序列的长度/时间步。通常，PyTorch中的1d指的是一个序列，例如一个标记化的句子，其中的每个标记（总共L）表示为C维向量，并堆叠在dim 1上。要修复此错误，可以执行以下操作

tensor_x = torch.as_tensor(X).unsqueeze(1)  # use as_tensor to avoid unnecessary data copy
tensor_y = torch.as_tensor(y)

完整修改代码：

import numpy as np
import torch
from torch.utils.data import TensorDataset, DataLoader

inputDim = 10
n = 1000
X = np.random.rand(n, inputDim)
y = np.random.rand(0, 2, n)

tensor_x = torch.as_tensor(X).unsqueeze(1)
tensor_y = torch.as_tensor(y)
Xy = (tensor_x, tensor_y)
XyLoader = DataLoader(Xy, batch_size = 16, shuffle = True, drop_last = True)

model = torch.nn.Sequential(
  torch.nn.Linear(inputDim, 200),
  torch.nn.ReLU(),
  torch.nn.BatchNorm1d(num_features=1),
  torch.nn.Linear(200, 100),
  torch.nn.Tanh(),
  torch.nn.BatchNorm1d(num_features=1),
  torch.nn.Linear(100,1),
  torch.nn.Sigmoid()
)

optimizer = torch.optim.Adam(model.parameters(), lr= 0.001)
loss_fn = torch.nn.BCELoss()

nepochs = 1000
for epochs in range(nepochs):
    for X, y in XyLoader:
        batch_size = X.shape[0]
        y_hat = model(X.view(batch_size,-1))
        loss = loss_fn(y_hat, y)
        optimizer.zero_grad()
        loss.backward()
        optimizer.step()

with torch.no_grad():
    xt = torch.as_tensor(np.random.rand(1, inputDim), dtype=torch.float32).unsqueeze(1)
    y2 = model(xt.float())
    print(y2.detach().numpy()[0][0])