How to implement XOR gate nural network with torch in .NET 6

TorchSharpTest220910.cs

// EXCEEDSYSTEM Sample of the XOR gate neural network with the TorchSharp 
// https://www.exceedsystem.net/2022/09/12/how-to-implement-xor-gate-neural-network-with-torch-in-dotnet-6
// License: MIT
using TorchSharp;
using static TorchSharp.torch;
using static TorchSharp.torch.nn;
using static TorchSharp.torch.nn.functional;

// Fix the random number seed to be used for weight and bias
torch.random.manual_seed(1);

// Training data(xa, xb)
var trainData = new float[,]
{
    {0, 0},
    {1, 0},
    {0, 1},
    {1, 1}
};

// Training label(t)
var trainLabel = new float[,]
{
    {0},
    {1},
    {1},
    {0}
};

// Create a newral network
var nn = Sequential(
    // Linear transformation 1
    Linear(2, 2),
    // Non-Linear transformation
    Sigmoid(),
    // Linear transformation 2
    Linear(2, 1)
);

// Set the modules in training mode
nn.train();

// Create tensors for training data and labels
var x = tensor(trainData);
var y = tensor(trainLabel);

// Number of epochs
const int epochs = 10000;

// Learning rate
const double lr = 0.1;

// Create a loss function
var criterion = mse_loss(Reduction.Sum);

// Create an optimizer
var optimizer = optim.SGD(nn.parameters(), lr);

// Training
for (var epoch = 1; epoch <= epochs; ++epoch)
{
    // Computes prediction
    var eval = nn.forward(x);

    // Computes MSE loss
    var loss = criterion(eval, y);

    // Sets gradients to zero
    optimizer.zero_grad();

    // Computes gradients
    loss.backward();

    // Updates parameters
    optimizer.step();

    if (epoch % 100 == 0)
        Console.WriteLine($"Epoch:{epoch} Loss:{loss.ToSingle()}");
}

// Set the modules in evaluation mode
nn.eval();

// Make predictions using trained model
// 0 ⊻ 0
nn.forward(torch.tensor(new float[] { 0, 0 })).print();
// 1 ⊻ 0
nn.forward(torch.tensor(new float[] { 1, 0 })).print();
// 0 ⊻ 1
nn.forward(torch.tensor(new float[] { 0, 1 })).print();
// 1 ⊻ 1
nn.forward(torch.tensor(new float[] { 1, 1 })).print();