variance_scaling_initializer

Author: Oceania2018

src/TensorFlowNET.Core/APIs/tf.init.cs

@@ -66,20 +66,20 @@ public IInitializer random_normal_initializer(float mean = 0.0f,
         /// <summary>
         /// Initializer capable of adapting its scale to the shape of weights tensors.
         /// </summary>
-        /// <param name="scale"></param>
+        /// <param name="factor"></param>
         /// <param name="mode"></param>
         /// <param name="distribution"></param>
         /// <param name="seed"></param>
         /// <param name="dtype"></param>
         /// <returns></returns>
-        public IInitializer variance_scaling_initializer(float scale = 1.0f,
-            string mode = "fan_in",
-            string distribution = "truncated_normal",
+        public IInitializer variance_scaling_initializer(float factor = 1.0f,
+            string mode = "FAN_IN",
+            bool uniform = false,
             int? seed = null,
             TF_DataType dtype = TF_DataType.TF_FLOAT) => new VarianceScaling(
-                scale: scale,
+                factor: factor,
                 mode: mode,
-                distribution: distribution,
+                uniform: uniform,
                 seed: seed,
                 dtype: dtype);
     }

src/TensorFlowNET.Core/APIs/tf.random.cs

@@ -28,21 +28,21 @@ public partial class tensorflow
         /// <param name="seed"></param>
         /// <param name="name"></param>
         /// <returns></returns>
-        public Tensor random_normal(int[] shape,
+        public Tensor random_normal(TensorShape shape,
             float mean = 0.0f,
             float stddev = 1.0f,
             TF_DataType dtype = TF_DataType.TF_FLOAT,
             int? seed = null,
             string name = null) => random_ops.random_normal(shape, mean, stddev, dtype, seed, name);
 
-        public Tensor random_uniform(int[] shape,
+        public Tensor random_uniform(TensorShape shape,
             float minval = 0,
             float maxval = 1,
             TF_DataType dtype = TF_DataType.TF_FLOAT,
             int? seed = null,
             string name = null) => random_ops.random_uniform(shape, minval, maxval, dtype, seed, name);
 
-        public Tensor truncated_normal(int[] shape,
+        public Tensor truncated_normal(TensorShape shape,
             float mean = 0.0f,
             float stddev = 1.0f,
             TF_DataType dtype = TF_DataType.TF_FLOAT,
@@ -62,5 +62,8 @@ public Tensor truncated_normal(int[] shape,
         /// </returns>
         public Tensor random_shuffle(Tensor value, int? seed = null, string name = null)
             => random_ops.random_shuffle(value, seed: seed, name: name);
+
+        public void set_random_seed(int seed)
+            => ops.get_default_graph().seed = seed;
     }
 }

src/TensorFlowNET.Core/Binding.Util.cs

@@ -273,6 +273,9 @@ public static double sum(IEnumerable enumerable)
             return sum;
         }
 
+        public static double sum(IEnumerable<int> enumerable)
+            => enumerable.Sum();
+
         public static double sum<TKey, TValue>(Dictionary<TKey, TValue> values)
         {
             return sum(values.Keys);

src/TensorFlowNET.Core/Data/DatasetV2.cs

@@ -1,4 +1,6 @@
-namespace Tensorflow.Data
+using System;
+
+namespace Tensorflow.Data
 {
     /// <summary>
     /// Represents a potentially large set of elements.
@@ -11,5 +13,9 @@
     /// </summary>
     public class DatasetV2
     {
+        public static DatasetV2 from_generator()
+        {
+            throw new NotImplementedException("");
+        }
     }
 }

src/TensorFlowNET.Core/Graphs/Graph.cs

@@ -107,6 +107,16 @@ public partial class Graph : DisposableObject
 
         public bool building_function;
 
+        int _seed;
+        public int seed
+        {
+            get => _seed;
+            set
+            {
+                _seed = value;
+            }
+        }
+
         public Graph()
         {
             _handle = c_api.TF_NewGraph();

src/TensorFlowNET.Core/Keras/Initializers.cs

@@ -27,7 +27,7 @@ public class Initializers
         /// <returns></returns>
         public IInitializer he_normal(int? seed = null)
         {
-            return new VarianceScaling(scale: 2.0f, mode: "fan_in", distribution: "truncated_normal", seed: seed);
+            return new VarianceScaling(factor: 2.0f, mode: "fan_in", seed: seed);
         }
     }
 }

src/TensorFlowNET.Core/Operations/Initializers/GlorotUniform.cs

@@ -22,7 +22,10 @@ public GlorotUniform(float scale = 1.0f,
             string mode = "fan_avg",
             string distribution = "uniform",
             int? seed = null,
-            TF_DataType dtype = TF_DataType.TF_FLOAT) : base(scale, mode, distribution, seed, dtype)
+            TF_DataType dtype = TF_DataType.TF_FLOAT) : base(factor: scale, 
+                mode: mode, 
+                seed: seed, 
+                dtype: dtype)
         {
 
         }

src/TensorFlowNET.Core/Operations/Initializers/VarianceScaling.cs

@@ -31,17 +31,22 @@ public class VarianceScaling : IInitializer
         protected int? _seed;
         protected TF_DataType _dtype;
 
-        public VarianceScaling(float scale = 1.0f,
-            string mode = "fan_in",
-            string distribution = "truncated_normal",
+        public VarianceScaling(float factor = 2.0f,
+            string mode = "FAN_IN",
+            bool uniform = false,
             int? seed = null,
             TF_DataType dtype = TF_DataType.TF_FLOAT)
         {
-            if (scale < 0)
+            if (!dtype.is_floating())
+                throw new TypeError("Cannot create initializer for non-floating point type.");
+            if (!new string[] { "FAN_IN", "FAN_OUT", "FAN_AVG" }.Contains(mode))
+                throw new TypeError($"Unknown {mode} %s [FAN_IN, FAN_OUT, FAN_AVG]");
+
+            if (factor < 0)
                 throw new ValueError("`scale` must be positive float.");
-            _scale = scale;
+
+            _scale = factor;
             _mode = mode;
-            _distribution = distribution;
             _seed = seed;
             _dtype = dtype;
         }