using Microsoft.ML;

using Microsoft.ML.Data;

using System;

using System.Collections.Generic;

using System.IO;

using TextTransferLearning.DataStructures;

namespace TextTransferLearning

{

class Program

{

static readonly string _modelPath = Path.Combine(Environment.CurrentDirectory, "Model");

static void Main(string[] args)

{

var context = new MLContext();

var tensorFlowModel = context.Model.LoadTensorFlowModel(_modelPath);

DataViewSchema schema = tensorFlowModel.GetModelSchema();

Console.WriteLine(" =============== TensorFlow Model Schema =============== ");

var featuresType = (VectorDataViewType)schema["Features"].Type;

Console.WriteLine($"Name: Features, Type: {featuresType.ItemType.RawType}, Size: ({featuresType.Dimensions[0]})");

var predictionType = (VectorDataViewType)schema["Prediction/Softmax"].Type;

Console.WriteLine($"Name: Prediction/Softmax, Type: {predictionType.ItemType.RawType}, Size: ({predictionType.Dimensions[0]})");

// Run here to show tensorflow error

var wordLookup = context.Data.LoadFromTextFile(Path.Combine(_modelPath, "imdb_word_index.csv"),

columns: new[]

{

new TextLoader.Column("Words", DataKind.String, 0),

new TextLoader.Column("Ids", DataKind.Int32, 1),

},

separatorChar: ',');

Action<VariableLengthVector, FixedLengthVector> ResizeFeaturesAction = (s, f) =>

{

var features = s.VariableLengthFeatures;

Array.Resize(ref features, 600);

f.Features = features;

};

var pipeline = context.Transforms.Text.TokenizeIntoWords("Tokens", "Review")

.Append(context.Transforms.Conversion.MapValue("VariableLengthFeatures", wordLookup, wordLookup.Schema["Words"], wordLookup.Schema["Ids"], "Tokens"))

.Append(context.Transforms.CustomMapping(ResizeFeaturesAction, "Resize"))

.Append(tensorFlowModel.ScoreTensorFlowModel("Prediction/Softmax", "Features"))

.Append(context.Transforms.CopyColumns("Prediction", "Prediction/Softmax"));

var data = context.Data.LoadFromEnumerable(new List<MovieReview>());

var model = pipeline.Fit(data);

var predictionEngine = context.Model.CreatePredictionEngine<MovieReview, MovieReviewSentiment>(model);

var review = new MovieReview

{

Review = "I like the action"

};

var prediction = predictionEngine.Predict(review);

Console.WriteLine(Environment.NewLine);

Console.WriteLine("Number of classes: {0}", prediction.Prediction.Length);

Console.WriteLine("Sentiment? {0}", prediction.Prediction[1] > 0.5 ? "Positive" : "Negative");

}

}

}