Program.cs

using System;
using System.Diagnostics;
using System.Linq;
using Numpy.Models;
using Python.Runtime;
using Torch;

namespace SimpleNeuralNetworkExample
{
 partial class Program
 {
 static void Main(string[] args)
 {
 Console.WriteLine("Importing torch ...");

 var dtype = torch.@float;
 var device = torch.device("cuda:0"); // "cuda:0" or "cpu"
 // N is batch size; D_in is input dimension;
 // H is hidden dimension; D_out is output dimension.

 var (N, D_in, H, D_out) = (64, 1000, 100, 10);
 // Create random Tensors to hold input and outputs.
 // Setting requires_grad=False indicates that we do not need to compute gradients
 // with respect to these Tensors during the backward pass.
 Console.WriteLine("Creating random data ...");
 var x = torch.randn(new Shape(N, D_in), device: device, dtype: dtype);
 var y = torch.randn(new Shape(N, D_out), device: device, dtype: dtype);

 LearnManualBackprop(dtype, device, x, y);

 LearnWithAutoGrad(dtype, device, x, y);

 LearnWithNnModules(x, y);

 LearnWithCustomAutoGrad(dtype, device, x, y);

 LearnWithOptimizer(x, y);

 Console.Write("Hit any key to exit: ");
 Console.ReadKey();
 }


 }
}