fix scalar ndarray to tensor proto.

src/TensorFlowNET.Core/NumPy/Numpy.Manipulation.cs

@@ -18,7 +18,7 @@ namespace Tensorflow.NumPy
         public static NDArray squeeze(NDArray x1, Axis? axis = null) => new NDArray(array_ops.squeeze(x1, axis));
 
         [AutoNumPy]
-        public static NDArray stack(NDArray arrays, Axis axis = null) => new NDArray(array_ops.stack(arrays, axis ?? 0));
+        public static NDArray stack(params NDArray[] arrays) => new NDArray(array_ops.stack(arrays));
 
         [AutoNumPy]
         public static NDArray dstack(params NDArray[] tup) => throw new NotImplementedException("");

src/TensorFlowNET.Core/NumPy/ShapeHelper.cs

@@ -94,7 +94,7 @@ namespace Tensorflow.NumPy
             {
                 -1 => "<unknown>",
                 0 => "()",
-                1 => $"({shape.dims[0]},)",
+                1 => $"({shape.dims[0].ToString().Replace("-1", "None")},)",
                 _ => $"({string.Join(", ", shape.dims).Replace("-1", "None")})"
             };
         }

src/TensorFlowNET.Core/Tensors/tensor_util.cs

@@ -134,12 +134,41 @@ namespace Tensorflow
                 TensorShape = shape.as_shape_proto()
             };
 
-            // scalar
             if (values is NDArray nd)
             {
-                var len = nd.dtypesize * nd.size;
-                byte[] bytes = nd.ToByteArray();
-                tensor_proto.TensorContent = Google.Protobuf.ByteString.CopyFrom(bytes);
+                // scalar
+                if (nd.shape.IsScalar)
+                {
+                    switch (nd.dtype)
+                    {
+                        case TF_DataType.TF_BOOL:
+                            tensor_proto.BoolVal.AddRange(nd.ToArray<bool>());
+                            break;
+                        case TF_DataType.TF_UINT8:
+                            tensor_proto.IntVal.AddRange(nd.ToArray<byte>().Select(x => (int)x).ToArray());
+                            break;
+                        case TF_DataType.TF_INT32:
+                            tensor_proto.IntVal.AddRange(nd.ToArray<int>());
+                            break;
+                        case TF_DataType.TF_INT64:
+                            tensor_proto.Int64Val.AddRange(nd.ToArray<long>());
+                            break;
+                        case TF_DataType.TF_FLOAT:
+                            tensor_proto.FloatVal.AddRange(nd.ToArray<float>());
+                            break;
+                        case TF_DataType.TF_DOUBLE:
+                            tensor_proto.DoubleVal.AddRange(nd.ToArray<double>());
+                            break;
+                        default:
+                            throw new Exception("make_tensor_proto Not Implemented");
+                    }
+                }
+                else
+                {
+                    var len = nd.dtypesize * nd.size;
+                    byte[] bytes = nd.ToByteArray();
+                    tensor_proto.TensorContent = Google.Protobuf.ByteString.CopyFrom(bytes);
+                }
             }
             else if (dtype == TF_DataType.TF_STRING && !(values is NDArray))
             {

test/TensorFlowNET.UnitTest/Numpy/Array.Creation.Test.cs

@@ -21,6 +21,7 @@ namespace TensorFlowNET.UnitTest.NumPy
             var zeros = np.zeros((2, 2));
             var ones = np.ones((2, 2));
             var full = np.full((2, 2), 0.1f);
+            Assert.AreEqual(np.float32, full.dtype);
         }
 
         [TestMethod]