Metrics methods skeletonizing completed.

src/TensorFlowNET.Keras/Metrics/BinaryCrossentropy.cs

@@ -4,7 +4,16 @@ using System.Text;
 
 namespace Tensorflow.Keras.Metrics
 {
-    class BinaryCrossentropy
+    public class BinaryCrossentropy : MeanMetricWrapper
     {
+        public BinaryCrossentropy(string name = "binary_crossentropy", string dtype = null, bool from_logits = false, float label_smoothing = 0)
+            : base(Fn, name, dtype)
+        {
+        }
+
+        internal static Tensor Fn(Tensor y_true, Tensor y_pred)
+        {
+            return Losses.Loss.binary_crossentropy(y_true, y_pred);
+        }
     }
 }

src/TensorFlowNET.Keras/Metrics/CategoricalCrossentropy.cs

@@ -4,7 +4,16 @@ using System.Text;
 
 namespace Tensorflow.Keras.Metrics
 {
-    class CategoricalCrossentropy
+    public class CategoricalCrossentropy : MeanMetricWrapper
     {
+        public CategoricalCrossentropy(string name = "categorical_crossentropy", string dtype = null, bool from_logits = false, float label_smoothing = 0)
+            : base(Fn, name, dtype)
+        {
+        }
+
+        internal static Tensor Fn(Tensor y_true, Tensor y_pred)
+        {
+            return Losses.Loss.categorical_crossentropy(y_true, y_pred);
+        }
     }
 }

src/TensorFlowNET.Keras/Metrics/KLDivergence.cs

@@ -4,7 +4,11 @@ using System.Text;
 
 namespace Tensorflow.Keras.Metrics
 {
-    class KLDivergence
+    public class KLDivergence : MeanMetricWrapper
     {
+        public KLDivergence(string name = "kullback_leibler_divergence", string dtype = null)
+            : base(Losses.Loss.logcosh, name, dtype)
+        {
+        }
     }
 }

src/TensorFlowNET.Keras/Metrics/MeanIoU.cs

@@ -1,10 +1,34 @@
 using System;
+using System.Collections;
 using System.Collections.Generic;
 using System.Text;
 
 namespace Tensorflow.Keras.Metrics
 {
-    class MeanIoU
+    public class MeanIoU : Metric
     {
+        public MeanIoU(int num_classes, string name, string dtype) : base(name, dtype)
+        {
+        }
+
+        public override void reset_states()
+        {
+            throw new NotImplementedException();
+        }
+
+        public override Hashtable get_config()
+        {
+            throw new NotImplementedException();
+        }
+
+        public override Tensor result()
+        {
+            throw new NotImplementedException();
+        }
+
+        public override void update_state(Args args, KwArgs kwargs)
+        {
+            throw new NotImplementedException();
+        }
     }
 }

src/TensorFlowNET.Keras/Metrics/MeanTensor.cs

@@ -4,7 +4,44 @@ using System.Text;
 
 namespace Tensorflow.Keras.Metrics
 {
-    class MeanTensor
+    public class MeanTensor : Metric
     {
+        public int total
+        {
+            get
+            {
+                throw new NotImplementedException();
+            }
+        }
+
+        public int count
+        {
+            get
+            {
+                throw new NotImplementedException();
+            }
+        }
+
+        public MeanTensor(int num_classes, string name = "mean_tensor", string dtype) : base(name, dtype)
+        {
+        }
+
+
+        private void _build(TensorShape shape) => throw new NotImplementedException();
+
+        public override void reset_states()
+        {
+            throw new NotImplementedException();
+        }
+
+        public override Tensor result()
+        {
+            throw new NotImplementedException();
+        }
+
+        public override void update_state(Args args, KwArgs kwargs)
+        {
+            throw new NotImplementedException();
+        }
     }
 }

src/TensorFlowNET.Keras/Metrics/SparseCategoricalCrossentropy.cs

@@ -4,7 +4,16 @@ using System.Text;
 
 namespace Tensorflow.Keras.Metrics
 {
-    class SparseCategoricalCrossentropy
+    public class SparseCategoricalCrossentropy : MeanMetricWrapper
     {
+        public SparseCategoricalCrossentropy(string name = "sparse_categorical_crossentropy", string dtype = null, bool from_logits = false, int axis = -1)
+            : base(Fn, name, dtype)
+        {
+        }
+
+        internal static Tensor Fn(Tensor y_true, Tensor y_pred)
+        {
+            return Losses.Loss.sparse_categorical_crossentropy(y_true, y_pred);
+        }
     }
 }

src/TensorFlowNET.Keras/Metrics/SumOverBatchSize.cs

@@ -4,7 +4,10 @@ using System.Text;
 
 namespace Tensorflow.Keras.Metrics
 {
-    class SumOverBatchSize
+    public class SumOverBatchSize : Reduce
     {
+        public SumOverBatchSize(string name = "sum_over_batch_size", string dtype = null) : base(Reduction.SUM_OVER_BATCH_SIZE, name, dtype)
+        {
+        }
     }
 }

src/TensorFlowNET.Keras/Metrics/SumOverBatchSizeMetricWrapper.cs

@@ -1,10 +1,25 @@
 using System;
+using System.Collections;
 using System.Collections.Generic;
 using System.Text;
 
 namespace Tensorflow.Keras.Metrics
 {
-    class SumOverBatchSizeMetricWrapper
+    public class SumOverBatchSizeMetricWrapper : SumOverBatchSize
     {
+        public SumOverBatchSizeMetricWrapper(Func<Tensor, Tensor, Tensor> fn, string name, string dtype = null)
+        {
+            throw new NotImplementedException();
+        }
+
+        public override void update_state(Args args, KwArgs kwargs)
+        {
+            throw new NotImplementedException();
+        }
+
+        public override Hashtable get_config()
+        {
+            throw new NotImplementedException();
+        }
     }
 }