测试环境:

vistual studio 2019

win10  64位

.net framework 4.7.2

安装参考:TensorFlow.NET

测试程序参考:TensorFlow.NET

1  在包管理工具中安装:Install-Package TensorFlow.NET

本次测试安装的版本是:0.70.2

TensorFlow.NET机器学习环境搭建(1)C#-LMLPHP

 

2  在包管理中安装:Install-Package TensorFlow.Keras

本次测试安装的版本是:

TensorFlow.NET机器学习环境搭建(1)C#-LMLPHP

 

3  在包管理中安装:Install-Package SciSharp.TensorFlow.Redist

本次测试安装的版本是:2.10.1.0

TensorFlow.NET机器学习环境搭建(1)C#-LMLPHP

 

4  选择项目生成的目标平台,选Any CPU好像不行,根据window平台选择x64或者x86

TensorFlow.NET机器学习环境搭建(1)C#-LMLPHP

 

5  输入测试代码:

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
using Tensorflow;

using Tensorflow.NumPy;

using static Tensorflow.Binding;

using static Tensorflow.KerasApi;

namespace TFDemo
{
    class Program
    {
        static void Main(string[] args)
        {
            // Parameters        
            var training_steps = 1000;
            var learning_rate = 0.01f;
            var display_step = 100;

            // Sample data
            var X = np.array(3.3f, 4.4f, 5.5f, 6.71f, 6.93f, 4.168f, 9.779f, 6.182f, 7.59f, 2.167f,
                         7.042f, 10.791f, 5.313f, 7.997f, 5.654f, 9.27f, 3.1f);
            var Y = np.array(1.7f, 2.76f, 2.09f, 3.19f, 1.694f, 1.573f, 3.366f, 2.596f, 2.53f, 1.221f,
                         2.827f, 3.465f, 1.65f, 2.904f, 2.42f, 2.94f, 1.3f);
            var n_samples = X.shape[0];

            // We can set a fixed init value in order to demo
            var W = Tensorflow.Binding.tf.Variable(-0.06f, name: "weight");
            var b = Tensorflow.Binding.tf.Variable(-0.73f, name: "bias");
            var optimizer = Tensorflow.KerasApi.keras.optimizers.SGD(learning_rate);

            // Run training for the given number of steps.
            foreach (var step in Tensorflow.Binding.range(1, training_steps + 1))
            {
                // Run the optimization to update W and b values.
                // Wrap computation inside a GradientTape for automatic differentiation.
                var g = Tensorflow.Binding.tf.GradientTape();
                // Linear regression (Wx + b).
                var pred = W * X + b;
                // Mean square error.
                var loss = Tensorflow.Binding.tf.reduce_sum(Tensorflow.Binding.tf.pow(pred - Y, 2)) / (2 * n_samples);
                // should stop recording
                // Compute gradients.
                var gradients = g.gradient(loss, (W, b));

                // Update W and b following gradients.
                optimizer.apply_gradients(Tensorflow.Binding.zip(gradients, (W, b)));

                if (step % display_step == 0)
                {
                    pred = W * X + b;
                    loss = Tensorflow.Binding.tf.reduce_sum(Tensorflow.Binding.tf.pow(pred - Y, 2)) / (2 * n_samples);
                    Console.WriteLine($"step: {step}, loss: {loss.numpy()}, W: {W.numpy()}, b: {b.numpy()}");
                }
            }

            Console.ReadLine();
        }
    }
}

如果没有报错的话,运行结果如下:

TensorFlow.NET机器学习环境搭建(1)C#-LMLPHP

 

当我把生成的dll拷贝到另外一台电脑运行时,很不幸,报了如下的错误:

未经处理的异常:  System.DllNotFoundException: 无法加载 DLL“tensorflow”: 找不到指定的模块。 (异常来自 HRESULT:0x8007007E)。
   在 Tensorflow.c_api.TF_AllocateTensor(TF_DataType dtype, Int64[] dims, Int32 num_dims, UInt64 len)
   在 Tensorflow.c_api.TF_NewTensor(Shape shape, TF_DataType dtype, Void* data)
   在 Tensorflow.Tensor.InitTensor(Array array, Shape shape)
   在 Tensorflow.NumPy.np.array[T](T[] data)
   在 TFDemo.Program.Main(String[] args) 位置 E:\VSProject\TFDemo\Program.cs:行号 26

TensorFlow.NET机器学习环境搭建(1)C#-LMLPHP

查阅stack overflow,找到了相应的解决方案:原来是要安装:vc_redist.x64.exe

下载链接:https://download.visualstudio.microsoft.com/download/pr/89a3b9df-4a09-492e-8474-8f92c115c51d/B1A32C71A6B7D5978904FB223763263EA5A7EB23B2C44A0D60E90D234AD99178/VC_redist.x64.exe 

stack overflow原链接如下:c# - Unable to load DLL 'tensorflow' or one of its dependencies (ML.NET) - Stack Overflow

TensorFlow.NET机器学习环境搭建(1)C#-LMLPHP

 

11-15 10:35