fix name_scope default_name.

Oceania2018 · Oceania2018 · commit 649623b003e9 · 2019-02-21T09:08:38.000-06:00
diff --git a/data/linear_regression.zip b/data/linear_regression.zip
diff --git a/src/TensorFlowNET.Core/Operations/OpDefLibrary.cs b/src/TensorFlowNET.Core/Operations/OpDefLibrary.cs
@@ -193,7 +193,9 @@ public Operation _apply_op_helper(string op_type_name, string name = "", dynamic
                 }
 
                 // Add Op to graph
-                var op = g.create_op(op_type_name, inputs.ToArray(), output_types.ToArray(),
+                var op = g.create_op(op_type_name, 
+                    inputs.ToArray(), 
+                    output_types.ToArray(),
                     name: scope,
                     input_types: input_types.ToArray(),
                     attrs: attr_protos,
diff --git a/src/TensorFlowNET.Core/Train/Saving/BaseSaverBuilder.cs b/src/TensorFlowNET.Core/Train/Saving/BaseSaverBuilder.cs
@@ -61,7 +61,7 @@ public virtual SaverDef _build_internal(RefVariable[] names_to_saveables,
             bool sharded = false,
             int max_to_keep = 5,
             float keep_checkpoint_every_n_hours = 10000,
-            string name = "",
+            string name = null,
             bool restore_sequentially = false,
             string filename = "model",
             bool build_save = true,
diff --git a/src/TensorFlowNET.Core/Train/Saving/Saver.cs b/src/TensorFlowNET.Core/Train/Saving/Saver.cs
@@ -37,7 +37,7 @@ public Saver(RefVariable[] var_list = null,
             bool sharded = false,
             int max_to_keep = 5,
             float keep_checkpoint_every_n_hours = 10000,
-            string name = "",
+            string name = null,
             bool restore_sequentially = false,
             SaverDef saver_def = null,
             ISaverBuilder builder = null,
diff --git a/src/TensorFlowNET.Core/ops.name_scope.cs b/src/TensorFlowNET.Core/ops.name_scope.cs
@@ -27,10 +27,7 @@ public name_scope(string name, string default_name = "", object values = null)
 
             public void __enter__()
             {
-                if (String.IsNullOrEmpty(_name))
-                {
-                    _name = _default_name;
-                }
+                _name = _name == null ? _default_name : _name;
 
                 Graph g = null;
                 if (_values is List<Tensor> values)
diff --git a/test/TensorFlowNET.Examples/LinearRegression.cs b/test/TensorFlowNET.Examples/LinearRegression.cs
@@ -57,7 +57,7 @@ public void Run()
             var grad = tf.train.GradientDescentOptimizer(learning_rate);
             var optimizer = grad.minimize(cost);*/
 
-            var new_saver = tf.train.import_meta_graph("save_model.meta", import_scope: "import");
+            var new_saver = tf.train.import_meta_graph("linear_regression.meta");
 
             var X = graph.OperationByName("Placeholder");
             var Y = graph.OperationByName("Placeholder_1");
diff --git a/test/TensorFlowNET.Examples/python/linear_regression.py b/test/TensorFlowNET.Examples/python/linear_regression.py
@@ -14,7 +14,7 @@
 # Parameters
 learning_rate = 0.01
 training_epochs = 1000
-display_step = 50
+display_step = 10
 
 # Training Data
 train_X = numpy.asarray([3.3,4.4,5.5,6.71,6.93,4.168,9.779,6.182,7.59,2.167,
@@ -23,28 +23,41 @@
                          2.827,3.465,1.65,2.904,2.42,2.94,1.3])
 n_samples = train_X.shape[0]
 
-# tf Graph Input
-X = tf.placeholder("float")
-Y = tf.placeholder("float")
-
-# Set model weights
-W = tf.Variable(rng.randn(), name="weight")
-b = tf.Variable(rng.randn(), name="bias")
-
-# Construct a linear model
-mul = tf.multiply(X, W)
-pred = tf.add(mul, b)
-
-# Mean squared error
-sub = pred-Y
-pow = tf.pow(sub, 2)
-
-reduce = tf.reduce_sum(pow)
-cost = reduce/(2*n_samples)
-# Gradient descent
-#  Note, minimize() knows to modify W and b because Variable objects are trainable=True by default
-grad = tf.train.GradientDescentOptimizer(learning_rate)
-optimizer = grad.minimize(cost)
+if False:
+    # tf Graph Input
+    X = tf.placeholder("float")
+    Y = tf.placeholder("float")
+
+    # Set model weights
+    W = tf.Variable(-0.06, name="weight")
+    b = tf.Variable(-0.73, name="bias")
+
+    # Construct a linear model
+    mul = tf.multiply(X, W)
+    pred = tf.add(mul, b)
+
+    # Mean squared error
+    sub = pred-Y
+    pow = tf.pow(sub, 2)
+
+    reduce = tf.reduce_sum(pow)
+    cost = reduce/(2*n_samples)
+    # Gradient descent
+    #  Note, minimize() knows to modify W and b because Variable objects are trainable=True by default
+    grad = tf.train.GradientDescentOptimizer(learning_rate)
+    optimizer = grad.minimize(cost)
+    # tf.train.export_meta_graph(filename='save_model.meta');
+else:
+    # tf Graph Input
+    new_saver = tf.train.import_meta_graph("save_model.meta")
+    nodes = tf.get_default_graph()._nodes_by_name;
+    optimizer = nodes["GradientDescent"]
+    cost = nodes["truediv"].outputs[0]
+    X = nodes["Placeholder"].outputs[0]
+    Y = nodes["Placeholder_1"].outputs[0]
+    W = nodes["weight"].outputs[0]
+    b = nodes["bias"].outputs[0]
+    pred = nodes["Add"].outputs[0]
 
 # Initialize the variables (i.e. assign their default value)
 init = tf.global_variables_initializer()
