Fix import_meta_graph exception if variables is empty.

Author: Oceania2018

docs/source/MnistInRnn.md

@@ -0,0 +1,5 @@
+# Chapter. MNIST In RNN
+
+### Recurrent Neural Networks
+
+Recurrent Neural Networks (RNNs) are popular models that have shown great promise in sequential data classification task. The traditional neural network model cannot make the next prediction input based on the knowledge that has been learned before.

src/TensorFlowNET.Core/Framework/meta_graph.cs

@@ -142,7 +142,8 @@ public static (Dictionary<string, VariableV1>, ITensorOrOperation[]) import_scop
 
                         break;
                     default:
-                        throw new NotImplementedException("import_scoped_meta_graph_with_return_elements");
+                        Console.WriteLine($"Cannot identify data type for collection {col.Key}. Skipping.");
+                        break;
                 }
             }
 

src/TensorFlowNET.Core/TensorFlow.Binding.csproj

@@ -25,7 +25,8 @@ https://tensorflownet.readthedocs.io</Description>
 3: Add tf.nn.rnn_cell.BasicRNNCell and tf.nn.dynamic_rnn.
 4: Add EstimatorSpec.
 5: Add rnn.static_rnn.
-6: Add array_grad._SplitGrad().</PackageReleaseNotes>
+6: Add array_grad._SplitGrad().
+7: Fix import_meta_graph exception if variables is empty.</PackageReleaseNotes>
     <LangVersion>7.3</LangVersion>
     <FileVersion>0.13.0.0</FileVersion>
     <PackageLicenseFile>LICENSE</PackageLicenseFile>

src/TensorFlowNET.Core/Variables/variables.py.cs

@@ -61,7 +61,6 @@ public static VariableV1[] _all_saveable_objects(string scope = "")
         public static List<VariableV1> global_variables(string scope = null)
         {
             return ops.get_collection<VariableV1>(tf.GraphKeys.GLOBAL_VARIABLES, scope);
-
         }
 
         /// <summary>