fix NDArray indexing.

src/TensorFlowNET.Console/MemoryBasicTest.cs

@@ -29,7 +29,7 @@ namespace Tensorflow
         public Action<int, int> Constant2x3
             => (epoch, iterate) =>
             {
-                var nd = np.arange(1000).reshape(10, 100);
+                var nd = np.arange(1000).reshape((10, 100));
                 var tensor = tf.constant(nd);
                 var data = tensor.numpy();
             };
@@ -51,14 +51,14 @@ namespace Tensorflow
         public Action<int, int> Variable
             => (epoch, iterate) =>
             {
-                var nd = np.arange(1 * 256 * 256 * 3).reshape(1, 256, 256, 3);
+                var nd = np.arange(1 * 256 * 256 * 3).reshape((1, 256, 256, 3));
                 ResourceVariable variable = tf.Variable(nd);
             };
 
         public Action<int, int> VariableRead
             => (epoch, iterate) =>
             {
-                var nd = np.zeros(1 * 256 * 256 * 3).astype(np.float32).reshape(1, 256, 256, 3);
+                var nd = np.zeros(1 * 256 * 256 * 3).astype(np.float32).reshape((1, 256, 256, 3));
                 ResourceVariable variable = tf.Variable(nd);
 
                 for (int i = 0; i< 10; i++)

src/TensorFlowNET.Core/Data/MnistDataSet.cs

@@ -17,7 +17,7 @@ namespace Tensorflow
 
             NumOfExamples = (int)images.dims[0];
 
-            images = images.reshape(images.dims[0], images.dims[1] * images.dims[2]);
+            images = images.reshape((images.dims[0], images.dims[1] * images.dims[2]));
             images = images.astype(dataType);
             // for debug np.multiply performance
             var sw = new Stopwatch();

src/TensorFlowNET.Core/Data/MnistModelLoader.cs

@@ -124,7 +124,7 @@ namespace Tensorflow
                 bytestream.Read(buf, 0, buf.Length);
 
                 var data = np.frombuffer(buf, np.@byte);
-                data = data.reshape(num_images, rows, cols, 1);
+                data = data.reshape((num_images, rows, cols, 1));
 
                 return data;
             }

src/TensorFlowNET.Core/Eager/EagerRunner.TFE_FastPathExecute.cs

@@ -360,7 +360,7 @@ namespace Tensorflow.Eager
                     c_api.TFE_OpSetAttrFloat(op, key, Convert.ToSingle(value));
                     break;
                 case TF_AttrType.TF_ATTR_SHAPE:
-                    var dims = (value as int[]).Select(x => (long)x).ToArray();
+                    var dims = (value as long[]).ToArray();
                     c_api.TFE_OpSetAttrShape(op, key, dims, dims.Length, status.Handle);
                     status.Check(true);
                     break;

src/TensorFlowNET.Core/Eager/EagerTensor.Creation.cs

@@ -24,6 +24,9 @@ namespace Tensorflow.Eager
             NewEagerTensorHandle(_handle);
         }
 
+        internal unsafe EagerTensor(string value) : base(value)
+            => NewEagerTensorHandle(_handle);
+
         internal unsafe EagerTensor(Array array, Shape shape) : base(array, shape)
             => NewEagerTensorHandle(_handle);
 

src/TensorFlowNET.Core/Framework/graph_util_impl.cs

@@ -135,8 +135,9 @@ namespace Tensorflow
             output_node.Attr["dtype"] = dtype;
             output_node.Attr["value"] = new AttrValue()
             {
-                Tensor = tensor_util.make_tensor_proto(
-                data, dtype: dtype.Type.as_tf_dtype(), shape: data_shape)
+                Tensor = tensor_util.make_tensor_proto(data, 
+                    dtype: dtype.Type.as_tf_dtype(), 
+                    shape: data_shape)
             };
 
             return output_node;

src/TensorFlowNET.Core/NumPy/NDArray.Equal.cs

@@ -1,5 +1,6 @@
 using System;
 using System.Collections.Generic;
+using System.Linq;
 using System.Text;
 using static Tensorflow.Binding;
 
@@ -15,8 +16,17 @@ namespace Tensorflow.NumPy
                 long val => GetAtIndex<long>(0) == val,
                 float val => GetAtIndex<float>(0) == val,
                 double val => GetAtIndex<double>(0) == val,
+                NDArray val => Equals(this, val),
                 _ => base.Equals(obj)
             };
         }
+
+        bool Equals(NDArray x, NDArray y)
+        {
+            if (x.ndim != y.ndim)
+                return false;
+
+            return Enumerable.SequenceEqual(x.ToByteArray(), y.ToByteArray());
+        }
     }
 }

src/TensorFlowNET.Core/NumPy/NDArray.Implicit.cs

@@ -18,6 +18,9 @@ namespace Tensorflow.NumPy
         public static implicit operator int(NDArray nd)
             => nd._tensor.ToArray<int>()[0];
 
+        public static implicit operator float(NDArray nd)
+            => nd._tensor.ToArray<float>()[0];
+
         public static implicit operator double(NDArray nd)
             => nd._tensor.ToArray<double>()[0];
 

src/TensorFlowNET.Core/NumPy/NDArray.Index.cs

@@ -0,0 +1,63 @@
+using System;
+using System.Collections.Generic;
+using System.Linq;
+using System.Text;
+using static Tensorflow.Binding;
+
+namespace Tensorflow.NumPy
+{
+    public partial class NDArray
+    {
+        public NDArray this[int index]
+        {
+            get
+            {
+                return _tensor[index];
+            }
+
+            set
+            {
+
+            }
+        }
+
+        public NDArray this[params int[] index]
+        {
+            get
+            {
+                return _tensor[index.Select(x => new Slice(x, x + 1)).ToArray()];
+            }
+
+            set
+            {
+
+            }
+        }
+
+        public NDArray this[params Slice[] slices]
+        {
+            get
+            {
+                return _tensor[slices];
+            }
+
+            set
+            {
+
+            }
+        }
+
+        public NDArray this[NDArray mask]
+        {
+            get
+            {
+                throw new NotImplementedException("");
+            }
+
+            set
+            {
+
+            }
+        }
+    }
+}

src/TensorFlowNET.Core/Numpy/NDArray.cs

@@ -2,6 +2,7 @@
 using System.Collections.Generic;
 using System.Linq;
 using System.Text;
+using static Tensorflow.Binding;
 
 namespace Tensorflow.NumPy
 {
@@ -55,48 +56,12 @@ namespace Tensorflow.NumPy
             Initialize(shape, dtype: dtype);
         }
 
-        public NDArray(Tensor value)
+        public NDArray(Tensor value, Shape? shape = null)
         {
-            _tensor = value;
-        }
-
-        public NDArray this[params int[] index]
-        {
-            get
-            {
-                throw new NotImplementedException("");
-            }
-
-            set
-            {
-
-            }
-        }
-
-        public NDArray this[params Slice[] slices]
-        {
-            get
-            {
-                throw new NotImplementedException("");
-            }
-
-            set
-            {
-
-            }
-        }
-
-        public NDArray this[NDArray mask]
-        {
-            get
-            {
-                throw new NotImplementedException("");
-            }
-
-            set
-            {
-
-            }
+            if (shape is not null)
+                _tensor = tf.reshape(value, shape);
+            else
+                _tensor = value;
         }
 
         public static NDArray Scalar<T>(T value) where T : unmanaged
@@ -129,15 +94,14 @@ namespace Tensorflow.NumPy
 
         public bool HasNext() => throw new NotImplementedException("");
         public T MoveNext<T>() => throw new NotImplementedException("");
-        public NDArray reshape(params int[] shape) => throw new NotImplementedException("");
-        public NDArray reshape(params long[] shape) => throw new NotImplementedException("");
+        public NDArray reshape(Shape newshape) => new NDArray(_tensor, newshape);
         public NDArray astype(Type type) => throw new NotImplementedException("");
         public NDArray astype(NumpyDType type) => throw new NotImplementedException("");
         public bool array_equal(NDArray rhs) => throw new NotImplementedException("");
         public NDArray ravel() => throw new NotImplementedException("");
         public void shuffle(NDArray nd) => throw new NotImplementedException("");
         public Array ToMuliDimArray<T>() => throw new NotImplementedException("");
-        public byte[] ToByteArray() => _tensor.ToArray<byte>();
+        public byte[] ToByteArray() => _tensor.BufferToArray();
         public static string[] AsStringArray(NDArray arr) => throw new NotImplementedException("");
 
         public T[] Data<T>() where T : unmanaged

src/TensorFlowNET.Core/Numpy/Shape.cs

@@ -48,6 +48,8 @@ namespace Tensorflow
         public static implicit operator Shape((long, long, long, long) dims)
             => new Shape(dims.Item1, dims.Item2, dims.Item3, dims.Item4);
 
+        public bool IsEmpty => size == 0;
+
         public bool IsScalar => ndim == 0;
 
         public static Shape Scalar
@@ -73,8 +75,6 @@ namespace Tensorflow
             }
         }
 
-        public bool IsEmpty => throw new NotImplementedException("");
-
         public override bool Equals(object obj)
         {
             if(obj is Shape shape)

src/TensorFlowNET.Core/Tensors/Tensor.Assign.cs

@@ -1,4 +1,4 @@
-using Tensorflow.NumPy;
+using System;
 
 namespace Tensorflow
 {
@@ -12,6 +12,9 @@ namespace Tensorflow
 
         public ResourceVariable assign(Tensor tensor)
         {
+            if (tensor.dtype != dtype)
+                throw new ArrayTypeMismatchException("");
+
             if (OriginalVar != null)
             {
                 OriginalVar.StridedSliceAssign(tensor, OriginalVarSlice);

src/TensorFlowNET.Core/Tensors/Tensor.Creation.cs

@@ -465,7 +465,7 @@ namespace Tensorflow
         /// <summary>
         ///     Create a string Tensor from the given string
         /// </summary>
-        public unsafe Tensor(string str)
+        public Tensor(string str)
         {
             _handle = StringTensor(new string[] { str }, TensorShape.Scalar);
 #if TRACK_TENSOR_LIFE
@@ -473,7 +473,7 @@ namespace Tensorflow
 #endif
         }
 
-        public unsafe Tensor(string[] strings)
+        public Tensor(string[] strings)
         {
             _handle = StringTensor(strings, new TensorShape(strings.Length));
 #if TRACK_TENSOR_LIFE

src/TensorFlowNET.Core/Tensors/Tensor.Value.cs

@@ -7,109 +7,34 @@ namespace Tensorflow
 {
     public partial class Tensor
     {
-        [Obsolete("Please use ToArray<T>() instead.", false)]
-        public T[] Data<T>() where T : unmanaged
-        {
-            return ToArray<T>();
-        }
-
         /// <summary>
         ///     
         /// </summary>
         /// <typeparam name="T"></typeparam>
         /// <returns></returns>
-        public T[] ToArray<T>() where T : unmanaged
+        public unsafe T[] ToArray<T>() where T : unmanaged
         {
             //Are the types matching?
-            if (typeof(T).as_tf_dtype() == dtype)
-            {
-                if (NDims == 0 && size == 1)  //is it a scalar?
-                {
-                    unsafe
-                    {
-                        return new T[] { *(T*)buffer };
-                    }
-                }
+            if (typeof(T).as_tf_dtype() != dtype)
+                throw new ArrayTypeMismatchException($"dtype {dtype} mismatch.");
 
-                //types match, no need to perform cast
-                var ret = new T[size];
+            if (NDims == 0 && size == 1)  //is it a scalar?
+            {
                 unsafe
                 {
-                    var len = (long)size;
-                    fixed (T* dst = ret)
-                    {
-                        //T can only be unmanaged, I believe it is safe to say that MemoryCopy is valid for all cases this method can be called.
-                        var src = (T*)buffer;
-                        len *= (long)itemsize;
-                        System.Buffer.MemoryCopy(src, dst, len, len);
-                    }
+                    return new T[] { *(T*)buffer };
                 }
-
-                return ret;
             }
-            else
-            {
 
-                //types do not match, need to perform cast
-                if (NDims == 0 && size == 1) //is it a scalar?
-                {
-                    unsafe
-                    {
-                        throw new NotImplementedException("");
-                        #region Compute
-                        /*switch (dtype.as_numpy_dtype().GetTypeCode())
-                        {
-                            case NPTypeCode.Boolean: return new T[] { Converts.ChangeType<T>(*(bool*)buffer) };
-                            case NPTypeCode.Byte: return new T[] { Converts.ChangeType<T>(*(byte*)buffer) };
-                            case NPTypeCode.Int16: return new T[] { Converts.ChangeType<T>(*(short*)buffer) };
-                            case NPTypeCode.UInt16: return new T[] { Converts.ChangeType<T>(*(ushort*)buffer) };
-                            case NPTypeCode.Int32: return new T[] { Converts.ChangeType<T>(*(int*)buffer) };
-                            case NPTypeCode.UInt32: return new T[] { Converts.ChangeType<T>(*(uint*)buffer) };
-                            case NPTypeCode.Int64: return new T[] { Converts.ChangeType<T>(*(long*)buffer) };
-                            case NPTypeCode.UInt64: return new T[] { Converts.ChangeType<T>(*(ulong*)buffer) };
-                            case NPTypeCode.Char: return new T[] { Converts.ChangeType<T>(*(char*)buffer) };
-                            case NPTypeCode.Double: return new T[] { Converts.ChangeType<T>(*(double*)buffer) };
-                            case NPTypeCode.Single: return new T[] { Converts.ChangeType<T>(*(float*)buffer) };
-                            case NPTypeCode.String: return new T[] { Converts.ChangeType<T>((string)this) };
-                            default:
-                                throw new NotSupportedException();
-                        }*/
-                        #endregion
-                    }
-                }
+            //types match, no need to perform cast
+            var ret = new T[size];
+            var len = (long)(size * itemsize);
+            var src = (T*)buffer;
 
-                var ret = new T[size];
-                unsafe
-                {
-                    var len = (long)size;
-                    fixed (T* dstRet = ret)
-                    {
-                        T* dst = dstRet; //local stack copy
-                        throw new NotImplementedException("");
-                        #region Compute
-                        /*switch (dtype.as_numpy_dtype().GetTypeCode())
-                        {
-                            case NPTypeCode.Boolean: new UnmanagedMemoryBlock<bool>((bool*)buffer, len).CastTo(new UnmanagedMemoryBlock<T>(dst, len), null, null); break;
-                            case NPTypeCode.Byte: new UnmanagedMemoryBlock<byte>((byte*)buffer, len).CastTo(new UnmanagedMemoryBlock<T>(dst, len), null, null); break;
-                            case NPTypeCode.Int16: new UnmanagedMemoryBlock<short>((short*)buffer, len).CastTo(new UnmanagedMemoryBlock<T>(dst, len), null, null); break;
-                            case NPTypeCode.UInt16: new UnmanagedMemoryBlock<ushort>((ushort*)buffer, len).CastTo(new UnmanagedMemoryBlock<T>(dst, len), null, null); break;
-                            case NPTypeCode.Int32: new UnmanagedMemoryBlock<int>((int*)buffer, len).CastTo(new UnmanagedMemoryBlock<T>(dst, len), null, null); break;
-                            case NPTypeCode.UInt32: new UnmanagedMemoryBlock<uint>((uint*)buffer, len).CastTo(new UnmanagedMemoryBlock<T>(dst, len), null, null); break;
-                            case NPTypeCode.Int64: new UnmanagedMemoryBlock<long>((long*)buffer, len).CastTo(new UnmanagedMemoryBlock<T>(dst, len), null, null); break;
-                            case NPTypeCode.UInt64: new UnmanagedMemoryBlock<ulong>((ulong*)buffer, len).CastTo(new UnmanagedMemoryBlock<T>(dst, len), null, null); break;
-                            case NPTypeCode.Char: new UnmanagedMemoryBlock<char>((char*)buffer, len).CastTo(new UnmanagedMemoryBlock<T>(dst, len), null, null); break;
-                            case NPTypeCode.Double: new UnmanagedMemoryBlock<double>((double*)buffer, len).CastTo(new UnmanagedMemoryBlock<T>(dst, len), null, null); break;
-                            case NPTypeCode.Single: new UnmanagedMemoryBlock<float>((float*)buffer, len).CastTo(new UnmanagedMemoryBlock<T>(dst, len), null, null); break;
-                            case NPTypeCode.String: throw new NotSupportedException("Unable to convert from string to other dtypes"); //TODO! this should call Converts.To<T> 
-                            default:
-                                throw new NotSupportedException();
-                        }*/
-                        #endregion
-                    }
-                }
+            fixed (T* dst = ret)
+                System.Buffer.MemoryCopy(src, dst, len, len);
 
-                return ret;
-            }
+            return ret;
         }
 
         /// <summary>

src/TensorFlowNET.Core/Tensors/TensorShape.cs

@@ -160,7 +160,7 @@ namespace Tensorflow
         {
             if (dims != null && shape2.dims != null)
             {
-                if (dims.Length != shape2.dims.Length)
+                if (shape.size != (ulong)shape2.size)
                     return false;
             }
 

src/TensorFlowNET.Core/Tensors/constant_op.cs

@@ -165,7 +165,7 @@ namespace Tensorflow
                 case TensorShape val:
                     return new EagerTensor(val.dims, ctx.DeviceName);
                 case string val:
-                    return new EagerTensor(val, ctx.DeviceName);
+                    return new EagerTensor(val);
                 case string[] val:
                     return new EagerTensor(val, ctx.DeviceName);
                 case bool val:

src/TensorFlowNET.Core/Tensors/tensor_util.cs

@@ -100,7 +100,7 @@ namespace Tensorflow
         };
 
         /// <summary>
-        /// Create a TensorProto.
+        /// Create a TensorProto, invoked in graph mode
         /// </summary>
         /// <param name="values"></param>
         /// <param name="dtype"></param>
@@ -108,89 +108,22 @@ namespace Tensorflow
         /// <param name="verify_shape"></param>
         /// <param name="allow_broadcast"></param>
         /// <returns></returns>
-        public static TensorProto make_tensor_proto(object values, TF_DataType dtype = TF_DataType.DtInvalid, int[] shape = null, bool verify_shape = false, bool allow_broadcast = false)
+        public static TensorProto make_tensor_proto(object values, TF_DataType dtype = TF_DataType.DtInvalid, int[]? shape = null, bool verify_shape = false, bool allow_broadcast = false)
         {
             if (allow_broadcast && verify_shape)
                 throw new ValueError("allow_broadcast and verify_shape are not both allowed.");
             if (values is TensorProto tp)
                 return tp;
 
+            dtype = values.GetType().as_tf_dtype();
             // We first convert value to a numpy array or scalar.
-            NDArray nparray = null;
-            var np_dt = dtype.as_system_dtype();
-
-            if (values is NDArray nd)
-            {
-                nparray = nd;
-            }
-            else if(values is string str)
-            {
-                // scalar string
-                nparray = convert_to_numpy_ndarray(values);
-                shape = new int[0];
-            }
-            else if(values is string[] strings)
-            {
-                nparray = convert_to_numpy_ndarray(values);
-                shape = new[] { strings.Length };
-            }
-            else
-            {
-                if (values == null)
-                    throw new ValueError("None values not supported.");
-
-                nparray = convert_to_numpy_ndarray(values);
-
-                if (np_dt != null && np_dt != typeof(string))
-                    nparray = nparray.astype(np_dt);
-            }
-
-            var numpy_dtype = nparray.dtype.as_tf_dtype();
-            if (numpy_dtype == TF_DataType.DtInvalid)
-                throw new TypeError($"Unrecognized data type: {nparray.dtype}");
-
-            // If dtype was specified and is a quantized type, we convert
-            // numpy_dtype back into the quantized version.
-            if (quantized_types.Contains(dtype))
-                numpy_dtype = dtype;
-
-            bool is_same_size = false;
-            ulong shape_size = 0;
-
-            // If shape is not given, get the shape from the numpy array.
-            if (shape == null)
-            {
-                if (numpy_dtype == TF_DataType.TF_STRING)
-                {
-                    if (nparray.ndim == 0)
-                    {
-                        // scalar string
-                        shape = new int[0];
-                        shape_size = 0;
-                    }
-                    else
-                        throw new NotImplementedException($"Not implemented for {nparray.ndim} dims string array.");
-                }
-                else
-                {
-                    shape = nparray.dims.Select(x => Convert.ToInt32(x)).ToArray();
-                    is_same_size = true;
-                    shape_size = nparray.size;
-                }
-            }
-            else
-            {
-                shape_size = (ulong)new Shape(shape.Select(x => Convert.ToInt64(x)).ToArray()).size;
-                is_same_size = shape_size == nparray.size;
-            }
-
             var tensor_proto = new TensorProto
             {
-                Dtype = numpy_dtype.as_datatype_enum(),
-                TensorShape = tensor_util.as_shape(shape)
+                Dtype = dtype.as_datatype_enum(),
+                // TensorShape = tensor_util.as_shape(shape.dims)
             };
 
-            if (is_same_size && _TENSOR_CONTENT_TYPES.Contains(numpy_dtype) && shape_size > 1)
+            /*if (is_same_size && _TENSOR_CONTENT_TYPES.Contains(numpy_dtype) && shape_size > 1)
             {
                 byte[] bytes = nparray.ToByteArray();
                 tensor_proto.TensorContent = Google.Protobuf.ByteString.CopyFrom(bytes.ToArray());
@@ -212,11 +145,15 @@ namespace Tensorflow
                 return tensor_proto;
             }
 
-            var proto_values = nparray.ravel();
-
-            switch (nparray.dtype.ToString())
+            var proto_values = nparray.ravel();*/
+            
+            switch (values)
             {
-                case "Bool":
+                case float val:
+                    tensor_proto.TensorShape = tensor_util.as_shape(new int[0]);
+                    tensor_proto.FloatVal.AddRange(new[] { val });
+                    break;
+                /*case "Bool":
                 case "Boolean":
                     tensor_proto.BoolVal.AddRange(proto_values.Data<bool>());
                     break;
@@ -226,13 +163,10 @@ namespace Tensorflow
                 case "Int64":
                     tensor_proto.Int64Val.AddRange(proto_values.Data<long>());
                     break;
-                case "Single":
-                    tensor_proto.FloatVal.AddRange(proto_values.Data<float>());
-                    break;
                 case "Double":
                     tensor_proto.DoubleVal.AddRange(proto_values.Data<double>());
                     break;
-                /*case "String":
+                case "String":
                     tensor_proto.StringVal.AddRange(proto_values.Data<string>().Select(x => Google.Protobuf.ByteString.CopyFromUtf8(x.ToString())));
                     break;*/
                 default:
@@ -435,76 +369,6 @@ would not be rank 1.", tensor.op.get_attr("axis")));
             return ret;
         }
 
-        public static NDArray convert_to_numpy_ndarray(object values)
-        {
-            NDArray nd;
-
-            switch (values)
-            {
-                case NDArray val:
-                    nd = val;
-                    break;
-                case TensorShape val:
-                    nd = val.dims;
-                    break;
-                case bool boolVal:
-                    nd = boolVal;
-                    break;
-                case int intVal:
-                    nd = intVal;
-                    break;
-                case int[] intVals:
-                    nd = np.array(intVals);
-                    break;
-                case int[,] intVals:
-                    nd = np.array(intVals);
-                    break;
-                case long intVal:
-                    nd = intVal;
-                    break;
-                case long[] intVals:
-                    nd = np.array(intVals);
-                    break;
-                case long[,] intVals:
-                    nd = np.array(intVals);
-                    break;
-                case float floatVal:
-                    nd = floatVal;
-                    break;
-                case float[] floatVals:
-                    nd = floatVals;
-                    break;
-                case float[,] floatVals:
-                    nd = np.array(floatVals);
-                    break;
-                case double doubleVal:
-                    nd = doubleVal;
-                    break;
-                case double[] doubleVals:
-                    nd = np.array(doubleVals);
-                    break;
-                case double[,] doubleVals:
-                    nd = np.array(doubleVals);
-                    break;
-                case string strVal:
-                    nd = new NDArray(Encoding.ASCII.GetBytes(strVal));
-                    break;
-                case string[] strVals:
-                    nd = np.array(strVals);
-                    break;
-                case byte[] byteValues:
-                    nd = byteValues;
-                    break;
-                case byte[,] byteValues:
-                    nd = np.array(byteValues);
-                    break;
-                default:
-                    throw new NotImplementedException($"convert_to_numpy_ndarray: Support for type {values.GetType()} Not Implemented");
-            }
-
-            return nd;
-        }
-
         public static TensorShapeProto as_shape<T>(T[] dims)
         {
             TensorShapeProto shape = new TensorShapeProto();

src/TensorFlowNET.Keras/Datasets/Cifar10.cs

@@ -118,7 +118,7 @@ namespace Tensorflow.Keras.Datasets
             var value = new Span<byte>(pickle, start_pos, value_length).ToArray();
             start_pos += value_length;
 
-            return (key, np.array(value).reshape(10000, 3, 32, 32));
+            return (key, np.array(value).reshape((10000, 3, 32, 32)));
         }
 
         string Download()

src/TensorFlowNet.Benchmarks/Crash/RepeatDataSetCrash.cs

@@ -12,7 +12,7 @@ namespace Tensorflow.Benchmark.Crash
         [Benchmark]
         public void Run()
         {
-            var data = tf.convert_to_tensor(np.arange(0, 50000 * 10).astype(np.float32).reshape(50000, 10));
+            var data = tf.convert_to_tensor(np.arange(0, 50000 * 10).astype(np.float32).reshape((50000, 10)));
 
             var dataset = keras.preprocessing.timeseries_dataset_from_array(data,
                 sequence_length: 10,

src/TensorFlowNet.Benchmarks/Leak/GpuLeakByCNN.cs

@@ -24,7 +24,7 @@ namespace Tensorflow.Benchmark.Leak
 
             var bytes = new byte[num * width * height * 3];
             var inputImages = np.array(bytes) / 255.0f;
-            inputImages = inputImages.reshape(num, height, width, 3);
+            inputImages = inputImages.reshape((num, height, width, 3));
 
             bytes = new byte[num];
             var outLables = np.array(bytes);

test/TensorFlowNET.Keras.UnitTest/Layers/Layers.Convolution.Test.cs

@@ -16,7 +16,7 @@ namespace TensorFlowNET.Keras.UnitTest
 
             var conv = keras.layers.Conv1D(filters, activation: "linear");
 
-            var x = np.arange(256.0f).reshape(8, 8, 4);
+            var x = np.arange(256.0f).reshape((8, 8, 4));
             var y = conv.Apply(x);
 
             Assert.AreEqual(3, y.shape.ndim);
@@ -32,7 +32,7 @@ namespace TensorFlowNET.Keras.UnitTest
 
             var conv = keras.layers.Conv1D(filters, kernel_size: 3, activation: "linear");
 
-            var x = np.arange(256.0f).reshape(8, 8, 4);
+            var x = np.arange(256.0f).reshape((8, 8, 4));
             var y = conv.Apply(x);
 
             Assert.AreEqual(3, y.shape.ndim);
@@ -48,7 +48,7 @@ namespace TensorFlowNET.Keras.UnitTest
 
             var conv = keras.layers.Conv1D(filters, kernel_size: 3, padding: "same", activation: "linear");
 
-            var x = np.arange(256.0f).reshape(8, 8, 4);
+            var x = np.arange(256.0f).reshape((8, 8, 4));
             var y = conv.Apply(x);
 
             Assert.AreEqual(3, y.shape.ndim);
@@ -63,7 +63,7 @@ namespace TensorFlowNET.Keras.UnitTest
             var filters = 8;
             var conv = keras.layers.Conv1D(filters, kernel_size: 3, strides: 2, activation: "linear");
 
-            var x = np.arange(256.0f).reshape(8, 8, 4);
+            var x = np.arange(256.0f).reshape((8, 8, 4));
             var y = conv.Apply(x);
 
             Assert.AreEqual(3, y.shape.ndim);
@@ -78,7 +78,7 @@ namespace TensorFlowNET.Keras.UnitTest
             var filters = 8;
             var conv = keras.layers.Conv1D(filters, kernel_size: 3, dilation_rate: 2, activation: "linear");
 
-            var x = np.arange(256.0f).reshape(8, 8, 4);
+            var x = np.arange(256.0f).reshape((8, 8, 4));
             var y = conv.Apply(x);
 
             Assert.AreEqual(3, y.shape.ndim);
@@ -93,7 +93,7 @@ namespace TensorFlowNET.Keras.UnitTest
             var filters = 8;
             var conv = keras.layers.Conv1D(filters, kernel_size: 3, dilation_rate: 2, padding: "same", activation: "linear");
 
-            var x = np.arange(256.0f).reshape(8, 8, 4);
+            var x = np.arange(256.0f).reshape((8, 8, 4));
             var y = conv.Apply(x);
 
             Assert.AreEqual(3, y.shape.ndim);
@@ -108,7 +108,7 @@ namespace TensorFlowNET.Keras.UnitTest
             var filters = 8;
             var conv = keras.layers.Conv2D(filters, activation: "linear");
 
-            var x = np.arange(256.0f).reshape(1,8,8,4);
+            var x = np.arange(256.0f).reshape((1, 8, 8, 4));
             var y = conv.Apply(x);
 
             Assert.AreEqual(4, y.shape.ndim);
@@ -124,7 +124,7 @@ namespace TensorFlowNET.Keras.UnitTest
             var filters = 8;
             var conv = keras.layers.Conv2D(filters, kernel_size: 3, activation: "linear");
 
-            var x = np.arange(256.0f).reshape(1, 8, 8, 4);
+            var x = np.arange(256.0f).reshape((1, 8, 8, 4));
             var y = conv.Apply(x);
 
             Assert.AreEqual(4, y.shape.ndim);
@@ -140,7 +140,7 @@ namespace TensorFlowNET.Keras.UnitTest
             var filters = 8;
             var conv = keras.layers.Conv2D(filters, kernel_size: 3, padding: "same", activation: "linear");
 
-            var x = np.arange(256.0f).reshape(1, 8, 8, 4);
+            var x = np.arange(256.0f).reshape((1, 8, 8, 4));
             var y = conv.Apply(x);
 
             Assert.AreEqual(4, y.shape.ndim);
@@ -156,7 +156,7 @@ namespace TensorFlowNET.Keras.UnitTest
             var filters = 8;
             var conv = keras.layers.Conv2D(filters, kernel_size: 3, strides: 2, activation: "linear");
 
-            var x = np.arange(256.0f).reshape(1, 8, 8, 4);
+            var x = np.arange(256.0f).reshape((1, 8, 8, 4));
             var y = conv.Apply(x);
 
             Assert.AreEqual(4, y.shape.ndim);
@@ -172,7 +172,7 @@ namespace TensorFlowNET.Keras.UnitTest
             var filters = 8;
             var conv = keras.layers.Conv2D(filters, kernel_size: 3, dilation_rate: 2, activation: "linear");
 
-            var x = np.arange(256.0f).reshape(1, 8, 8, 4);
+            var x = np.arange(256.0f).reshape((1, 8, 8, 4));
             var y = conv.Apply(x);
 
             Assert.AreEqual(4, y.shape.ndim);
@@ -188,7 +188,7 @@ namespace TensorFlowNET.Keras.UnitTest
             var filters = 8;
             var conv = keras.layers.Conv2D(filters, kernel_size: 3, dilation_rate: 2, padding: "same", activation: "linear");
 
-            var x = np.arange(256.0f).reshape(1, 8, 8, 4);
+            var x = np.arange(256.0f).reshape((1, 8, 8, 4));
             var y = conv.Apply(x);
 
             Assert.AreEqual(4, y.shape.ndim);

test/TensorFlowNET.Keras.UnitTest/Layers/Layers.Merging.Test.cs

@@ -11,8 +11,8 @@ namespace TensorFlowNET.Keras.UnitTest
         [TestMethod]
         public void Concatenate()
         {
-            var x = np.arange(20).reshape(2, 2, 5);
-            var y = np.arange(20, 30).reshape(2, 1, 5);
+            var x = np.arange(20).reshape((2, 2, 5));
+            var y = np.arange(20, 30).reshape((2, 1, 5));
             var z = keras.layers.Concatenate(axis: 1).Apply(new Tensors(x, y));
             Assert.AreEqual((2, 3, 5), z.shape);
         }

test/TensorFlowNET.Keras.UnitTest/Layers/LayersTest.cs

@@ -84,7 +84,7 @@ namespace TensorFlowNET.Keras.UnitTest
         public void Embedding_Simple()
         {
             var emb = keras.layers.Embedding(256, 12, input_length: 4);
-            var input_array = np.arange(12).reshape(3, 4).astype(np.float32);
+            var input_array = np.arange(12).reshape((3, 4)).astype(np.float32);
             var output = emb.Apply(input_array);
             Assert.AreEqual(new TensorShape(3, 4, 12), output.shape);
         }

test/TensorFlowNET.Native.UnitTest/Gradients/GradientsTest.cs

@@ -226,7 +226,7 @@ namespace Tensorflow.Native.UnitTest
             //long[] dims = { 2, 2 };
             //Tensor t = c_api.TF_AllocateTensor(TF_FLOAT, dims, 2, sizeof(float) * 4);
             //Marshal.Copy(values, 0, t, 4);
-            Tensor t = new Tensor(new NDArray(values).reshape(2, 2));
+            Tensor t = new Tensor(new NDArray(values).reshape((2, 2)));
             return t;
         }
 

test/TensorFlowNET.Native.UnitTest/Tensors/TensorTest.cs

@@ -94,7 +94,7 @@ namespace Tensorflow.Native.UnitTest.Tensors
         [TestMethod]
         public void Tensor()
         {
-            var nd = np.array(1f, 2f, 3f, 4f, 5f, 6f).reshape(2, 3);
+            var nd = np.array(1f, 2f, 3f, 4f, 5f, 6f).reshape((2, 3));
 
             var tensor = new Tensor(nd);
             var array = tensor.ToArray<float>();

test/TensorFlowNET.UnitTest/Basics/RandomTest.cs

@@ -17,7 +17,7 @@ namespace TensorFlowNET.UnitTest.Basics
         [TestMethod]
         public void TFRandomSeedTest()
         {
-            var initValue = np.arange(6).reshape(3, 2);
+            var initValue = np.arange(6).reshape((3, 2));
             tf.set_random_seed(1234);
             var a1 = tf.random_uniform(1);
             var b1 = tf.random_shuffle(tf.constant(initValue));
@@ -40,7 +40,7 @@ namespace TensorFlowNET.UnitTest.Basics
         [TestMethod, Ignore]
         public void TFRandomSeedTest2()
         {
-            var initValue = np.arange(6).reshape(3, 2);
+            var initValue = np.arange(6).reshape((3, 2));
             tf.set_random_seed(1234);
             var a1 = tf.random_uniform(1, seed:1234);
             var b1 = tf.random_shuffle(tf.constant(initValue), seed: 1234);

test/TensorFlowNET.UnitTest/Basics/SessionTest.cs

@@ -18,8 +18,8 @@ namespace TensorFlowNET.UnitTest
         {
             lock (this)
             {
-                var a = constant_op.constant(np.array(3.0).reshape(1, 1));
-                var b = constant_op.constant(np.array(2.0).reshape(1, 1));
+                var a = constant_op.constant(np.array(3.0).reshape((1, 1)));
+                var b = constant_op.constant(np.array(2.0).reshape((1, 1)));
                 var c = math_ops.matmul(a, b, name: "matmul");
                 using (var sess = tf.Session())
                 {

test/TensorFlowNET.UnitTest/Basics/TensorTest.cs

@@ -54,7 +54,7 @@ namespace TensorFlowNET.UnitTest
         [TestMethod]
         public void batch_to_space_nd()
         {
-            var inputs = np.arange(24).reshape(4, 2, 3);
+            var inputs = np.arange(24).reshape((4, 2, 3));
             var block_shape = new[] { 2, 2 };
             int[,] crops = { { 0, 0 }, { 0, 0 } };
             var tensor = tf.batch_to_space_nd(inputs, block_shape, crops);

test/TensorFlowNET.UnitTest/Basics/VariableTest.cs

@@ -2,6 +2,7 @@
 using Tensorflow.NumPy;
 using System.Linq;
 using static Tensorflow.Binding;
+using System;
 
 namespace TensorFlowNET.UnitTest.Basics
 {
@@ -13,7 +14,7 @@ namespace TensorFlowNET.UnitTest.Basics
         {
             var x = tf.Variable(10, name: "x");
             Assert.AreEqual(0, x.shape.ndim);
-            Assert.AreEqual(10, (int)x.numpy());
+            Assert.AreEqual(x.numpy(), 10);
         }
 
         [TestMethod]
@@ -28,7 +29,7 @@ namespace TensorFlowNET.UnitTest.Basics
         {
             var x = tf.constant(3, name: "x");
             var y = tf.Variable(x + 1, name: "y");
-            Assert.AreEqual(4, (int)y.numpy());
+            Assert.AreEqual(y.numpy(), 4);
         }
 
         [TestMethod]
@@ -36,7 +37,7 @@ namespace TensorFlowNET.UnitTest.Basics
         {
             var variable = tf.Variable(31, name: "tree");
             var unread = variable.assign(12);
-            Assert.AreEqual(12, (int)unread.numpy());
+            Assert.AreEqual(unread.numpy(), 12);
         }
 
         [TestMethod]
@@ -45,7 +46,7 @@ namespace TensorFlowNET.UnitTest.Basics
             var v1 = tf.Variable(10.0f, name: "v1");
             var v2 = v1.assign(v1 + 1.0f);
             Assert.AreEqual(v1.numpy(), v2.numpy());
-            Assert.AreEqual(11f, (float)v1.numpy());
+            Assert.AreEqual(v1.numpy(), 11f);
         }
 
         [TestMethod]
@@ -71,6 +72,7 @@ namespace TensorFlowNET.UnitTest.Basics
                 { 4, 5, 6 },
                 { 7, 8, 9 }
             };
+
             var x = tf.Variable(nd);
 
             // get slice form variable
@@ -93,7 +95,8 @@ namespace TensorFlowNET.UnitTest.Basics
             Assert.AreEqual(nd[2], x[2].numpy());
         }
 
-        [TestMethod, Ignore]
+        [TestMethod]
+        [ExpectedException(typeof(ArrayTypeMismatchException))]
         public void TypeMismatchedSliceAssign()
         {
             NDArray intNd = new int[]
@@ -105,12 +108,7 @@ namespace TensorFlowNET.UnitTest.Basics
                 -5, 6, -7
             };
             var x = tf.Variable(doubleNd);
-
-            var slice = x[":"];
-            Assert.ThrowsException<System.Exception>(
-                // this statement exit without throwing any exception but the "test execution summary" seems not able to detect that.
-                () => slice.assign(intNd)
-            );
+            x[":"].assign(intNd);
         }
 
         [TestMethod]
@@ -120,7 +118,7 @@ namespace TensorFlowNET.UnitTest.Basics
             for (int i = 0; i < 5; i++)
                 x.assign(x + 1);
 
-            Assert.AreEqual(15, (int)x.numpy());
+            Assert.AreEqual(x.numpy(), 15);
         }
 
         [TestMethod]
@@ -138,8 +136,8 @@ namespace TensorFlowNET.UnitTest.Basics
             var a = tf.Variable(5);
             var a_identity = tf.identity(a);
             a.assign_add(1);
-            Assert.AreEqual(5, (int)a_identity.numpy());
-            Assert.AreEqual(6, (int)a.numpy());
+            Assert.AreEqual(a_identity.numpy(), 5);
+            Assert.AreEqual(a.numpy(), 6);
         }
     }
 }

test/TensorFlowNET.UnitTest/ManagedAPI/ArrayOpsTest.cs

@@ -15,11 +15,11 @@ namespace TensorFlowNET.UnitTest.ManagedAPI
         public void Slice()
         {
             // Tests based on example code in TF documentation
-            var input_array = tf.constant(np.array(new int[] { 1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 5, 5, 5, 6, 6, 6 }).reshape(3,2,3));
+            var input_array = tf.constant(np.array(new int[] { 1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 5, 5, 5, 6, 6, 6 }).reshape((3,2,3)));
             var indices = tf.constant(np.array(new int[] { 0, 2 }));
 
             var r1 = array_ops.slice(input_array, new int[] { 1, 0, 0 }, new int[] { 1, 1, 3 });
-            Assert.AreEqual(new TensorShape(1,1,3), r1.shape);
+            Assert.AreEqual(new Shape(1,1,3), r1.shape);
             var r1np = r1.numpy();
             Assert.AreEqual(r1np[0, 0, 0], 3);
             Assert.AreEqual(r1np[0, 0, 1], 3);
@@ -27,7 +27,7 @@ namespace TensorFlowNET.UnitTest.ManagedAPI
 
 
             var r2 = array_ops.slice(input_array, new int[] { 1, 0, 0 }, new int[] { 1, 2, 3 });
-            Assert.AreEqual(new TensorShape(1, 2, 3), r2.shape);
+            Assert.AreEqual(new Shape(1, 2, 3), r2.shape);
             var r2np = r2.numpy();
             Assert.AreEqual(r2np[0, 0, 0], 3);
             Assert.AreEqual(r2np[0, 0, 1], 3);
@@ -37,7 +37,7 @@ namespace TensorFlowNET.UnitTest.ManagedAPI
             Assert.AreEqual(r2np[0, 1, 2], 4);
 
             var r3 = array_ops.slice(input_array, new int[] { 1, 0, 0 }, new int[] { 2, 1, 3 });
-            Assert.AreEqual(new TensorShape(2, 1, 3), r3.shape);
+            Assert.AreEqual(new Shape(2, 1, 3), r3.shape);
             var r3np = r3.numpy();
             Assert.AreEqual(r3np[0, 0, 0], 3);
             Assert.AreEqual(r3np[0, 0, 1], 3);
@@ -53,29 +53,29 @@ namespace TensorFlowNET.UnitTest.ManagedAPI
         [TestMethod]
         public void Gather()
         {
-            var input_array = tf.constant(np.arange(12).reshape(3, 4).astype(np.float32));
+            var input_array = tf.constant(np.arange(12).reshape((3, 4)).astype(np.float32));
             var indices = tf.constant(np.array(new int[] { 0, 2 }));
 
             var result = array_ops.gather(input_array, indices);
-            Assert.AreEqual(new TensorShape(2, 4), result.shape);
+            Assert.AreEqual(new Shape(2, 4), result.shape);
             Assert.AreEqual(result.numpy()[0, 0], 0.0f);
             Assert.AreEqual(result.numpy()[0, 1], 1.0f);
             Assert.AreEqual(result.numpy()[1, 3], 11.0f);
 
             // Tests based on example code in Python doc string for tf.gather()
 
-            var p1 = tf.random.normal(new TensorShape(5, 6, 7, 8));
-            var i1 = tf.random_uniform(new TensorShape(10, 11), maxval: 7, dtype: tf.int32);
+            var p1 = tf.random.normal(new Shape(5, 6, 7, 8));
+            var i1 = tf.random_uniform(new Shape(10, 11), maxval: 7, dtype: tf.int32);
             var r1 = tf.gather(p1, i1, axis:2);
-            Assert.AreEqual(new TensorShape(5, 6, 10, 11, 8), r1.shape);
+            Assert.AreEqual(new Shape(5, 6, 10, 11, 8), r1.shape);
 
-            var p2 = tf.random.normal(new TensorShape(4,3));
+            var p2 = tf.random.normal(new Shape(4,3));
             var i2 = tf.constant(new int[,] { { 0, 2} });
             var r2 = tf.gather(p2, i2, axis: 0);
-            Assert.AreEqual(new TensorShape(1, 2, 3), r2.shape);
+            Assert.AreEqual(new Shape(1, 2, 3), r2.shape);
 
             var r3 = tf.gather(p2, i2, axis: 1);
-            Assert.AreEqual(new TensorShape(4,1,2), r3.shape);
+            Assert.AreEqual(new Shape(4,1,2), r3.shape);
         }
     }
 }

test/TensorFlowNET.UnitTest/ManagedAPI/ConstantTest.cs

@@ -136,10 +136,10 @@ namespace TensorFlowNET.UnitTest.Basics
         [TestMethod]
         public void NDimConst()
         {
-            var nd = np.array(new int[][]
+            var nd = np.array(new int[,]
             {
-                new int[]{ 3, 1, 1 },
-                new int[]{ 2, 1, 3 }
+                { 3, 1, 1 },
+                { 2, 1, 3 }
             });
 
             var tensor = tf.constant(nd);