Consume a tensorflow regression model from .NET

[jbilbrey]

System information

using Visual Studio 2022
Python 3.8.12
Tensorflow 2.5
Runtime Environment:
OS Name: Windows
OS Version: 10.0.19043
.NET SDKs installed:
5.0.400 [C:\Program Files\dotnet\sdk]
6.0.100 [C:\Program Files\dotnet\sdk]
Issue
I am trying to create a simple end-to-end problem. I create a simple model in Python and import the saved_model.pb into .NET. I would like to consume this model. Please help me by providing the solution that I can solve this problem.

What did you do?
I created a simple Python script in for the data set. Successfuly trained and evaluated the model and then exported it. The model can be loaded but I cannot configure the pipeline to make prediction.
What happened?
Now I try to consume the model with ML.NET Console.
What did you expect?
Cannot consume the model, because not able to create the proper pipeline.

I included the python script and the .NET program file in this repo.
https://github.com/jbilbrey/repo

using Microsoft.ML;
using Microsoft.ML.Data;
using System;
using System.IO;
using Microsoft.ML.Transforms;
using System.Linq;

namespace TestDeleteMe
{
class Program
{

    static string dataset = Path.Combine(Directory.GetCurrentDirectory(), "mpg.txt");        
    static readonly string _modelPath = Path.Combine(Environment.CurrentDirectory, "model");

    static void Main(string[] args)
    {
        var mlContext = new MLContext();

        Console.WriteLine("The path to the model is... " + _modelPath.ToString());  // debug only
        //load tensorflow model
        var tensorFlowModel = mlContext.Model.LoadTensorFlowModel(_modelPath);
        
        var schema = tensorFlowModel.GetModelSchema();
        var inputSchema = tensorFlowModel.GetInputSchema();
                           
        Console.WriteLine("The data is..." + inputSchema.ToString());

        var reader = mlContext.Data.CreateTextLoader(new[] {
            new TextLoader.Column("dense_input_1", DataKind.Single, new[] {new TextLoader.Range(1,9)}),          
        }, separatorChar: '\t', hasHeader: true);

        // read the data
        var data = reader.Load(dataset);

        // print data to screen
        var inputs = mlContext.Data.CreateEnumerable<InputData>(data, reuseRowObject: false).ToArray();

        // print the data to the console
        for (int i = 0; i < inputs.Length; i++)
        {
            //var predictedLabel = engine.Predict(inputs[i]);

            for (int j = 0; j < inputs[i].Features.Length; j++)
            {
                Console.Write(inputs[i].Features[j]);
                Console.Write(" ");
            }
            //Console.WriteLine(predictedLabel.Output[0]);
        }

        ////////////// fit the model /////////////////// NEED HELP HERE to consume/use the model!!!

        ///var estimator = tensorFlowModel.ScoreTensorFlowModel("Predict", "dense_input_1").Fit(data);

    }
}

class InputData
{
    [ColumnName("dense_input_1"), VectorType(9)]
    public float[]? Features { get; set; }
}


class OutputData
{
    [ColumnName("Output"), VectorType(1)]
    public float[]? Prediction { get; set; }
}

}

[jbilbrey]

I found the solution in this post.

#5487