Merge pull request SciSharp#224 from henon/master

Oceania2018 · web-flow · commit 0513d985f0fd · 2019-04-11T11:49:02.000-05:00
unit test suite: added lightweight version of all examples
diff --git a/src/TensorFlowNET.Core/Python.cs b/src/TensorFlowNET.Core/Python.cs
@@ -46,7 +46,7 @@ public static void with(IPython py, Action<IPython> action)
             catch (Exception ex)
             {
                 Console.WriteLine(ex.ToString());
-                throw ex;
+                throw;
             }
             finally
             {
@@ -65,7 +65,7 @@ public static void with<T>(T py, Action<T> action) where T : IPython
             catch (Exception ex)
             {
                 Console.WriteLine(ex.ToString());
-                throw ex;
+                throw;
             }
             finally
             {
@@ -83,10 +83,8 @@ public static TOut with<TIn, TOut>(TIn py, Func<TIn, TOut> action) where TIn : I
             }
             catch (Exception ex)
             {
-                Console.WriteLine(ex.ToString());
-#if DEBUG
-                Debugger.Break();
-#endif
+                Console.WriteLine(ex.ToString());
+                throw;
                 return default(TOut);
             }
             finally
diff --git a/test/TensorFlowNET.Examples/BasicEagerApi.cs b/test/TensorFlowNET.Examples/BasicEagerApi.cs
@@ -12,7 +12,7 @@ namespace TensorFlowNET.Examples
     public class BasicEagerApi : IExample
     {
         public int Priority => 100;
-        public bool Enabled => false;
+        public bool Enabled { get; set; } = false;
         public string Name => "Basic Eager";
 
         private Tensor a, b, c, d;
diff --git a/test/TensorFlowNET.Examples/BasicOperations.cs b/test/TensorFlowNET.Examples/BasicOperations.cs
@@ -12,7 +12,7 @@ namespace TensorFlowNET.Examples
     /// </summary>
     public class BasicOperations : Python, IExample
     {
-        public bool Enabled => true;
+        public bool Enabled { get; set; } = true;
         public int Priority => 2;
         public string Name => "Basic Operations";
 
diff --git a/test/TensorFlowNET.Examples/HelloWorld.cs b/test/TensorFlowNET.Examples/HelloWorld.cs
@@ -12,7 +12,7 @@ namespace TensorFlowNET.Examples
     public class HelloWorld : Python, IExample
     {
         public int Priority => 1;
-        public bool Enabled => true;
+        public bool Enabled { get; set; } = true;
         public string Name => "Hello World";
 
         public bool Run()
diff --git a/test/TensorFlowNET.Examples/IExample.cs b/test/TensorFlowNET.Examples/IExample.cs
@@ -17,7 +17,7 @@ public interface IExample
         /// <summary>
         /// True to run example
         /// </summary>
-        bool Enabled { get; }
+        bool Enabled { get; set; }
 
         string Name { get; }
 
diff --git a/test/TensorFlowNET.Examples/ImageRecognition.cs b/test/TensorFlowNET.Examples/ImageRecognition.cs
@@ -13,7 +13,7 @@ namespace TensorFlowNET.Examples
     public class ImageRecognition : Python, IExample
     {
         public int Priority => 7;
-        public bool Enabled => true;
+        public bool Enabled { get; set; } = true;
         public string Name => "Image Recognition";
 
         string dir = "ImageRecognition";
diff --git a/test/TensorFlowNET.Examples/InceptionArchGoogLeNet.cs b/test/TensorFlowNET.Examples/InceptionArchGoogLeNet.cs
@@ -19,7 +19,7 @@ namespace TensorFlowNET.Examples
     /// </summary>
     public class InceptionArchGoogLeNet : Python, IExample
     {
-        public bool Enabled => false;
+        public bool Enabled { get; set; } = false;
         public int Priority => 100;
         public string Name => "Inception Arch GoogLeNet";
 
diff --git a/test/TensorFlowNET.Examples/KMeansClustering.cs b/test/TensorFlowNET.Examples/KMeansClustering.cs
@@ -16,13 +16,17 @@ namespace TensorFlowNET.Examples
     public class KMeansClustering : Python, IExample
     {
         public int Priority => 8;
-        public bool Enabled => true;
+        public bool Enabled { get; set; } = true;
         public string Name => "K-means Clustering";
 
+        public int? train_size = null;
+        public int validation_size = 5000;
+        public int? test_size = null;
+        public int batch_size = 1024; // The number of samples per batch
+
         Datasets mnist;
         NDArray full_data_x;
         int num_steps = 50; // Total steps to train
-        int batch_size = 1024; // The number of samples per batch
         int k = 25; // The number of clusters
         int num_classes = 10; // The 10 digits
         int num_features = 784; // Each image is 28x28 pixels
@@ -45,7 +49,7 @@ public bool Run()
 
         public void PrepareData()
         {
-            mnist = MnistDataSet.read_data_sets("mnist", one_hot: true);
+            mnist = MnistDataSet.read_data_sets("mnist", one_hot: true, train_size: train_size, validation_size:validation_size, test_size:test_size);
             full_data_x = mnist.train.images;
         }
     }
diff --git a/test/TensorFlowNET.Examples/LinearRegression.cs b/test/TensorFlowNET.Examples/LinearRegression.cs
@@ -13,16 +13,16 @@ namespace TensorFlowNET.Examples
     public class LinearRegression : Python, IExample
     {
         public int Priority => 3;
-        public bool Enabled => true;
+        public bool Enabled { get; set; } = true;
         public string Name => "Linear Regression";
 
-        NumPyRandom rng = np.random;
+        public int training_epochs = 1000;
 
         // Parameters
         float learning_rate = 0.01f;
-        int training_epochs = 1000;
         int display_step = 50;
 
+        NumPyRandom rng = np.random;
         NDArray train_X, train_Y;
         int n_samples;
 
diff --git a/test/TensorFlowNET.Examples/LogisticRegression.cs b/test/TensorFlowNET.Examples/LogisticRegression.cs
@@ -17,12 +17,16 @@ namespace TensorFlowNET.Examples
     public class LogisticRegression : Python, IExample
     {
         public int Priority => 4;
-        public bool Enabled => true;
+        public bool Enabled { get; set; } = true;
         public string Name => "Logistic Regression";
 
+        public int training_epochs = 10;
+        public int? train_size = null;
+        public int validation_size = 5000;
+        public int? test_size = null;
+        public int batch_size = 100;
+
         private float learning_rate = 0.01f;
-        private int training_epochs = 10;
-        private int batch_size = 100;
         private int display_step = 1;
 
         Datasets mnist;
@@ -96,7 +100,7 @@ public bool Run()
 
         public void PrepareData()
         {
-            mnist = MnistDataSet.read_data_sets("mnist", one_hot: true);
+            mnist = MnistDataSet.read_data_sets("mnist", one_hot: true, train_size: train_size, validation_size: validation_size, test_size: test_size);
         }
 
         public void SaveModel(Session sess)
@@ -139,7 +143,7 @@ public void Predict()
                 if (results.argmax() == (batch_ys[0] as NDArray).argmax())
                     print("predicted OK!");
                 else
-                    throw new ValueError("predict error, maybe 90% accuracy");
+                    throw new ValueError("predict error, should be 90% accuracy");
             });
         }
     }
diff --git a/test/TensorFlowNET.Examples/MetaGraph.cs b/test/TensorFlowNET.Examples/MetaGraph.cs
@@ -10,7 +10,7 @@ namespace TensorFlowNET.Examples
     public class MetaGraph : Python, IExample
     {
         public int Priority => 100;
-        public bool Enabled => false;
+        public bool Enabled { get; set; } = false;
         public string Name => "Meta Graph";
 
         public bool Run()
diff --git a/test/TensorFlowNET.Examples/NaiveBayesClassifier.cs b/test/TensorFlowNET.Examples/NaiveBayesClassifier.cs
@@ -13,7 +13,7 @@ namespace TensorFlowNET.Examples
     public class NaiveBayesClassifier : Python, IExample
     {
         public int Priority => 6;
-        public bool Enabled => true;
+        public bool Enabled { get; set; } = true;
         public string Name => "Naive Bayes Classifier";
 
         public Normal dist { get; set; }
diff --git a/test/TensorFlowNET.Examples/NamedEntityRecognition.cs b/test/TensorFlowNET.Examples/NamedEntityRecognition.cs
@@ -11,7 +11,7 @@ namespace TensorFlowNET.Examples
     public class NamedEntityRecognition : Python, IExample
     {
         public int Priority => 100;
-        public bool Enabled => false;
+        public bool Enabled { get; set; } = false;
         public string Name => "NER";
 
         public bool Run()
diff --git a/test/TensorFlowNET.Examples/NearestNeighbor.cs b/test/TensorFlowNET.Examples/NearestNeighbor.cs
@@ -15,10 +15,13 @@ namespace TensorFlowNET.Examples
     public class NearestNeighbor : Python, IExample
     {
         public int Priority => 5;
-        public bool Enabled => true;
+        public bool Enabled { get; set; } = true;
         public string Name => "Nearest Neighbor";
         Datasets mnist;
         NDArray Xtr, Ytr, Xte, Yte;
+        public int? TrainSize = null;
+        public int ValidationSize = 5000;
+        public int? TestSize = null;
 
         public bool Run()
         {
@@ -62,10 +65,10 @@ public bool Run()
 
         public void PrepareData()
         {
-            mnist = MnistDataSet.read_data_sets("mnist", one_hot: true);
+            mnist = MnistDataSet.read_data_sets("mnist", one_hot: true, train_size: TrainSize, validation_size:ValidationSize, test_size:TestSize);
             // In this example, we limit mnist data
-            (Xtr, Ytr) = mnist.train.next_batch(5000); // 5000 for training (nn candidates)
-            (Xte, Yte) = mnist.test.next_batch(200); // 200 for testing
+            (Xtr, Ytr) = mnist.train.next_batch(TrainSize==null ? 5000 : TrainSize.Value / 100); // 5000 for training (nn candidates)
+            (Xte, Yte) = mnist.test.next_batch(TestSize==null ? 200 : TestSize.Value / 100); // 200 for testing
         }
     }
 }
diff --git a/test/TensorFlowNET.Examples/TextClassification/DataHelpers.cs b/test/TensorFlowNET.Examples/TextClassification/DataHelpers.cs
@@ -13,7 +13,7 @@ public class DataHelpers
         private const string TRAIN_PATH = "text_classification/dbpedia_csv/train.csv";
         private const string TEST_PATH = "text_classification/dbpedia_csv/test.csv";
 
-        public static (int[][], int[], int) build_char_dataset(string step, string model, int document_max_len)
+        public static (int[][], int[], int) build_char_dataset(string step, string model, int document_max_len, int? limit=null)
         {
             string alphabet = "abcdefghijklmnopqrstuvwxyz0123456789-,;.!?:’'\"/|_#$%ˆ&*˜‘+=<>()[]{} ";
             /*if (step == "train")
@@ -25,10 +25,11 @@ public static (int[][], int[], int) build_char_dataset(string step, string model
                 char_dict[c.ToString()] = char_dict.Count;
 
             var contents = File.ReadAllLines(TRAIN_PATH);
-            
-            var x = new int[contents.Length][];
-            var y = new int[contents.Length];
-            for (int i = 0; i < contents.Length; i++)
+            var size = limit == null ? contents.Length : limit.Value;
+
+            var x = new int[size][];
+            var y = new int[size];
+            for (int i = 0; i < size; i++)
             {
                 string[] parts = contents[i].ToLower().Split(",\"").ToArray();
                 string content = parts[2];
diff --git a/test/TensorFlowNET.Examples/TextClassification/TextClassificationTrain.cs b/test/TensorFlowNET.Examples/TextClassification/TextClassificationTrain.cs
@@ -15,8 +15,9 @@ namespace TensorFlowNET.Examples.CnnTextClassification
     public class TextClassificationTrain : Python, IExample
     {
         public int Priority => 100;
-        public bool Enabled => false;
+        public bool Enabled { get; set; }= false;
         public string Name => "Text Classification";
+        public int? DataLimit = null;
 
         private string dataDir = "text_classification";
         private string dataFileName = "dbpedia_csv.tar.gz";
@@ -28,7 +29,7 @@ public bool Run()
         {
             PrepareData();
             Console.WriteLine("Building dataset...");
-            var (x, y, alphabet_size) = DataHelpers.build_char_dataset("train", "vdcnn", CHAR_MAX_LEN);
+            var (x, y, alphabet_size) = DataHelpers.build_char_dataset("train", "vdcnn", CHAR_MAX_LEN, DataLimit);
 
             var (train_x, valid_x, train_y, valid_y) = train_test_split(x, y, test_size: 0.15f);
 
diff --git a/test/TensorFlowNET.Examples/TextClassification/TextClassificationWithMovieReviews.cs b/test/TensorFlowNET.Examples/TextClassification/TextClassificationWithMovieReviews.cs
@@ -12,7 +12,7 @@ namespace TensorFlowNET.Examples
     public class TextClassificationWithMovieReviews : Python, IExample
     {
         public int Priority => 9;
-        public bool Enabled => false;
+        public bool Enabled { get; set; } = false;
         public string Name => "Movie Reviews";
 
         string dir = "text_classification_with_movie_reviews";
diff --git a/test/TensorFlowNET.Examples/Utility/MnistDataSet.cs b/test/TensorFlowNET.Examples/Utility/MnistDataSet.cs
@@ -15,29 +15,33 @@ public class MnistDataSet
         private const string TRAIN_LABELS = "train-labels-idx1-ubyte.gz";
         private const string TEST_IMAGES = "t10k-images-idx3-ubyte.gz";
         private const string TEST_LABELS = "t10k-labels-idx1-ubyte.gz";
-
         public static Datasets read_data_sets(string train_dir, 
             bool one_hot = false,
             TF_DataType dtype = TF_DataType.TF_FLOAT,
             bool reshape = true,
             int validation_size = 5000,
+            int? train_size = null,
+            int? test_size = null,
             string source_url = DEFAULT_SOURCE_URL)
         {
+            if (train_size!=null && validation_size >= train_size)
+                throw new ArgumentException("Validation set should be smaller than training set");
+
             Web.Download(source_url + TRAIN_IMAGES, train_dir, TRAIN_IMAGES);
             Compress.ExtractGZip(Path.Join(train_dir, TRAIN_IMAGES), train_dir);
-            var train_images = extract_images(Path.Join(train_dir, TRAIN_IMAGES.Split('.')[0]));
+            var train_images = extract_images(Path.Join(train_dir, TRAIN_IMAGES.Split('.')[0]), limit: train_size);
 
             Web.Download(source_url + TRAIN_LABELS, train_dir, TRAIN_LABELS);
             Compress.ExtractGZip(Path.Join(train_dir, TRAIN_LABELS), train_dir);
-            var train_labels = extract_labels(Path.Join(train_dir, TRAIN_LABELS.Split('.')[0]), one_hot: one_hot);
+            var train_labels = extract_labels(Path.Join(train_dir, TRAIN_LABELS.Split('.')[0]), one_hot: one_hot, limit: train_size);
 
             Web.Download(source_url + TEST_IMAGES, train_dir, TEST_IMAGES);
             Compress.ExtractGZip(Path.Join(train_dir, TEST_IMAGES), train_dir);
-            var test_images = extract_images(Path.Join(train_dir, TEST_IMAGES.Split('.')[0]));
+            var test_images = extract_images(Path.Join(train_dir, TEST_IMAGES.Split('.')[0]), limit: test_size);
 
             Web.Download(source_url + TEST_LABELS, train_dir, TEST_LABELS);
             Compress.ExtractGZip(Path.Join(train_dir, TEST_LABELS), train_dir);
-            var test_labels = extract_labels(Path.Join(train_dir, TEST_LABELS.Split('.')[0]), one_hot: one_hot);
+            var test_labels = extract_labels(Path.Join(train_dir, TEST_LABELS.Split('.')[0]), one_hot: one_hot, limit:test_size);
 
             int end = train_images.shape[0];
             var validation_images = train_images[np.arange(validation_size)];
@@ -52,14 +56,15 @@ public static Datasets read_data_sets(string train_dir,
             return new Datasets(train, validation, test);
         }
 
-        public static NDArray extract_images(string file)
+        public static NDArray extract_images(string file, int? limit=null)
         {
             using (var bytestream = new FileStream(file, FileMode.Open))
             {
                 var magic = _read32(bytestream);
                 if (magic != 2051)
                     throw new ValueError($"Invalid magic number {magic} in MNIST image file: {file}");
-                var num_images = _read32(bytestream);
+                var num_images =  _read32(bytestream);
+                num_images = limit == null ? num_images : Math.Min(num_images, (uint)limit);
                 var rows = _read32(bytestream);
                 var cols = _read32(bytestream);
                 var buf = new byte[rows * cols * num_images];
@@ -70,14 +75,15 @@ public static NDArray extract_images(string file)
             }
         }
 
-        public static NDArray extract_labels(string file, bool one_hot = false, int num_classes = 10)
+        public static NDArray extract_labels(string file, bool one_hot = false, int num_classes = 10, int? limit = null)
         {
             using (var bytestream = new FileStream(file, FileMode.Open))
             {
                 var magic = _read32(bytestream);
                 if (magic != 2049)
                     throw new ValueError($"Invalid magic number {magic} in MNIST label file: {file}");
                 var num_items = _read32(bytestream);
+                num_items = limit == null ? num_items : Math.Min(num_items,(uint) limit);
                 var buf = new byte[num_items];
                 bytestream.Read(buf, 0, buf.Length);
                 var labels = np.frombuffer(buf, np.uint8);
diff --git a/test/TensorFlowNET.UnitTest/ExamplesTests/ExamplesTest.cs b/test/TensorFlowNET.UnitTest/ExamplesTests/ExamplesTest.cs
diff --git a/test/TensorFlowNET.UnitTest/TensorFlowNET.UnitTest.csproj b/test/TensorFlowNET.UnitTest/TensorFlowNET.UnitTest.csproj

Original file line number	Diff line number	Diff line change
`@@ -46,7 +46,7 @@ public static void with(IPython py, Action<IPython> action)`
`46`	`46`	`catch (Exception ex)`
`47`	`47`	`{`
`48`	`48`	`Console.WriteLine(ex.ToString());`
`49`		`- throw ex;`
	`49`	`+ throw;`
`50`	`50`	`}`
`51`	`51`	`finally`
`52`	`52`	`{`
`@@ -65,7 +65,7 @@ public static void with<T>(T py, Action<T> action) where T : IPython`
`65`	`65`	`catch (Exception ex)`
`66`	`66`	`{`
`67`	`67`	`Console.WriteLine(ex.ToString());`
`68`		`- throw ex;`
	`68`	`+ throw;`
`69`	`69`	`}`
`70`	`70`	`finally`
`71`	`71`	`{`
`@@ -83,10 +83,8 @@ public static TOut with<TIn, TOut>(TIn py, Func<TIn, TOut> action) where TIn : I`
`83`	`83`	`}`
`84`	`84`	`catch (Exception ex)`
`85`	`85`	`{`
`86`		`- Console.WriteLine(ex.ToString());`
`87`		`-#if DEBUG`
`88`		`- Debugger.Break();`
`89`		`-#endif`
	`86`	`+ Console.WriteLine(ex.ToString());`
	`87`	`+ throw;`
`90`	`88`	`return default(TOut);`
`91`	`89`	`}`
`92`	`90`	`finally`
Original file line number	Diff line number	Diff line change
`@@ -12,7 +12,7 @@ namespace TensorFlowNET.Examples`
`12`	`12`	`public class BasicEagerApi : IExample`
`13`	`13`	`{`
`14`	`14`	`public int Priority => 100;`
`15`		`- public bool Enabled => false;`
	`15`	`+ public bool Enabled { get; set; } = false;`
`16`	`16`	`public string Name => "Basic Eager";`
`17`	`17`
`18`	`18`	`private Tensor a, b, c, d;`
Original file line number	Diff line number	Diff line change
`@@ -13,7 +13,7 @@ namespace TensorFlowNET.Examples`
`13`	`13`	`public class ImageRecognition : Python, IExample`
`14`	`14`	`{`
`15`	`15`	`public int Priority => 7;`
`16`		`- public bool Enabled => true;`
	`16`	`+ public bool Enabled { get; set; } = true;`
`17`	`17`	`public string Name => "Image Recognition";`
`18`	`18`
`19`	`19`	`string dir = "ImageRecognition";`
Original file line number	Diff line number	Diff line change
`@@ -19,7 +19,7 @@ namespace TensorFlowNET.Examples`
`19`	`19`	`/// </summary>`
`20`	`20`	`public class InceptionArchGoogLeNet : Python, IExample`
`21`	`21`	`{`
`22`		`- public bool Enabled => false;`
	`22`	`+ public bool Enabled { get; set; } = false;`
`23`	`23`	`public int Priority => 100;`
`24`	`24`	`public string Name => "Inception Arch GoogLeNet";`
`25`	`25`