diff --git a/src/TensorFlowNET.Core/Tensors/Tensor.Creation.cs b/src/TensorFlowNET.Core/Tensors/Tensor.Creation.cs
index 81c8006b..b90c4e69 100644
--- a/src/TensorFlowNET.Core/Tensors/Tensor.Creation.cs
+++ b/src/TensorFlowNET.Core/Tensors/Tensor.Creation.cs
@@ -1,5 +1,7 @@
 /*****************************************************************************
    Copyright 2018 The TensorFlow.NET Authors. All Rights Reserved.
+   
+   Portions of this file have been adapted from TensorFlowSharp, authored by Miguel de Icaza (miguel@microsoft.com)
 
    Licensed under the Apache License, Version 2.0 (the "License");
    you may not use this file except in compliance with the License.
@@ -18,6 +20,8 @@ using NumSharp;
 using System;
 using System.Collections.Generic;
 using System.Linq;
+using System.Numerics;
+using System.Runtime.CompilerServices;
 using System.Runtime.InteropServices;
 using System.Text;
 using static Tensorflow.c_api;
@@ -36,39 +40,246 @@ namespace Tensorflow
             _handle = handle;
         }
 
-        public Tensor(NDArray nd, TF_DataType? tensorDType = null)
+#if _REGEN
+        %types=["sbyte", "byte", "short", "ushort", "int", "uint", "long", "ulong", "float", "double", "Complex"]
+        %foreach types%
+        
+        /// <summary>
+        /// Create a 1d Tensor from the given linear array and shape
+        /// </summary>
+        public Tensor(#1[] data)
         {
-            _handle = Allocate(nd, tensorDType: tensorDType);
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(#1)), new long[]{data.Length}, data, Marshal.SizeOf<#1>());
+        }
+
+        /// <summary>
+        /// Create a N-dimensional Tensor from the given array
+        /// </summary>
+        public Tensor(#1[] data, long[] shape)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(#1)), shape, data, Marshal.SizeOf<#1>());
+        }
+
+        %
+#else
+        
+        
+        /// <summary>
+        /// Create a 1d Tensor from the given linear array and shape
+        /// </summary>
+        public Tensor(sbyte[] data)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(sbyte)), new long[]{data.Length}, data, Marshal.SizeOf<sbyte>());
+        }
+
+        /// <summary>
+        /// Create a N-dimensional Tensor from the given array
+        /// </summary>
+        public Tensor(sbyte[] data, long[] shape)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(sbyte)), shape, data, Marshal.SizeOf<sbyte>());
+        }
+
+        
+        /// <summary>
+        /// Create a 1d Tensor from the given linear array and shape
+        /// </summary>
+        public Tensor(byte[] data)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(byte)), new long[]{data.Length}, data, Marshal.SizeOf<byte>());
+        }
+
+        /// <summary>
+        /// Create a N-dimensional Tensor from the given array
+        /// </summary>
+        public Tensor(byte[] data, long[] shape)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(byte)), shape, data, Marshal.SizeOf<byte>());
+        }
+
+        
+        /// <summary>
+        /// Create a 1d Tensor from the given linear array and shape
+        /// </summary>
+        public Tensor(short[] data)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(short)), new long[]{data.Length}, data, Marshal.SizeOf<short>());
+        }
+
+        /// <summary>
+        /// Create a N-dimensional Tensor from the given array
+        /// </summary>
+        public Tensor(short[] data, long[] shape)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(short)), shape, data, Marshal.SizeOf<short>());
+        }
+
+        
+        /// <summary>
+        /// Create a 1d Tensor from the given linear array and shape
+        /// </summary>
+        public Tensor(ushort[] data)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(ushort)), new long[]{data.Length}, data, Marshal.SizeOf<ushort>());
+        }
+
+        /// <summary>
+        /// Create a N-dimensional Tensor from the given array
+        /// </summary>
+        public Tensor(ushort[] data, long[] shape)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(ushort)), shape, data, Marshal.SizeOf<ushort>());
+        }
+
+        
+        /// <summary>
+        /// Create a 1d Tensor from the given linear array and shape
+        /// </summary>
+        public Tensor(int[] data)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(int)), new long[]{data.Length}, data, Marshal.SizeOf<int>());
+        }
+
+        /// <summary>
+        /// Create a N-dimensional Tensor from the given array
+        /// </summary>
+        public Tensor(int[] data, long[] shape)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(int)), shape, data, Marshal.SizeOf<int>());
+        }
+
+        
+        /// <summary>
+        /// Create a 1d Tensor from the given linear array and shape
+        /// </summary>
+        public Tensor(uint[] data)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(uint)), new long[]{data.Length}, data, Marshal.SizeOf<uint>());
+        }
+
+        /// <summary>
+        /// Create a N-dimensional Tensor from the given array
+        /// </summary>
+        public Tensor(uint[] data, long[] shape)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(uint)), shape, data, Marshal.SizeOf<uint>());
         }
 
-        public unsafe Tensor(byte[] buffer)
+        
+        /// <summary>
+        /// Create a 1d Tensor from the given linear array and shape
+        /// </summary>
+        public Tensor(long[] data)
         {
-            var size = c_api.TF_StringEncodedSize((UIntPtr)buffer.Length);
-            _handle = TF_AllocateTensor(TF_DataType.TF_STRING, IntPtr.Zero, 0, (UIntPtr)((ulong)size + 8));
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(long)), new long[]{data.Length}, data, Marshal.SizeOf<long>());
+        }
 
-            IntPtr tensor = c_api.TF_TensorData(_handle);
-            Marshal.WriteInt64(tensor, 0);
-            fixed (byte* src = &buffer[0])
-                c_api.TF_StringEncode(src, (UIntPtr)buffer.Length, (sbyte*)(tensor + sizeof(Int64)), size, status);
+        /// <summary>
+        /// Create a N-dimensional Tensor from the given array
+        /// </summary>
+        public Tensor(long[] data, long[] shape)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(long)), shape, data, Marshal.SizeOf<long>());
+        }
+
+        
+        /// <summary>
+        /// Create a 1d Tensor from the given linear array and shape
+        /// </summary>
+        public Tensor(ulong[] data)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(ulong)), new long[]{data.Length}, data, Marshal.SizeOf<ulong>());
+        }
 
-            status.Check(true);
+        /// <summary>
+        /// Create a N-dimensional Tensor from the given array
+        /// </summary>
+        public Tensor(ulong[] data, long[] shape)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(ulong)), shape, data, Marshal.SizeOf<ulong>());
         }
 
-        private IntPtr Allocate(NDArray nd, TF_DataType? tensorDType = null)
+        
+        /// <summary>
+        /// Create a 1d Tensor from the given linear array and shape
+        /// </summary>
+        public Tensor(float[] data)
         {
-            if (tensorDType == TF_DataType.TF_STRING &&
-                nd.dtype.Name == "Byte")
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(float)), new long[]{data.Length}, data, Marshal.SizeOf<float>());
+        }
+
+        /// <summary>
+        /// Create a N-dimensional Tensor from the given array
+        /// </summary>
+        public Tensor(float[] data, long[] shape)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(float)), shape, data, Marshal.SizeOf<float>());
+        }
+
+        
+        /// <summary>
+        /// Create a 1d Tensor from the given linear array and shape
+        /// </summary>
+        public Tensor(double[] data)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(double)), new long[]{data.Length}, data, Marshal.SizeOf<double>());
+        }
+
+        /// <summary>
+        /// Create a N-dimensional Tensor from the given array
+        /// </summary>
+        public Tensor(double[] data, long[] shape)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(double)), shape, data, Marshal.SizeOf<double>());
+        }
+
+        
+        /// <summary>
+        /// Create a 1d Tensor from the given linear array and shape
+        /// </summary>
+        public Tensor(Complex[] data)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(Complex)), new long[]{data.Length}, data, Marshal.SizeOf<Complex>());
+        }
+
+        /// <summary>
+        /// Create a N-dimensional Tensor from the given array
+        /// </summary>
+        public Tensor(Complex[] data, long[] shape)
+        {
+            _handle = CreateTensorWithoutCopying(dtypes.as_dtype(typeof(Complex)), shape, data, Marshal.SizeOf<Complex>());
+        }
+#endif
+
+        public Tensor(NDArray nd, TF_DataType? tensorDType = null)
+        {
+            _handle = Allocate(nd, tensorDType: tensorDType);
+        }
+
+        private unsafe IntPtr Allocate(NDArray nd, TF_DataType? tensorDType = null)
+        {
+            if (tensorDType == TF_DataType.TF_STRING && nd.dtype.Name == "Byte")
             {
-                return new Tensor(nd.Data<byte>());
+                var buffer=nd.Data<byte>();
+                var size = c_api.TF_StringEncodedSize((UIntPtr)buffer.Length);
+                var handle = TF_AllocateTensor(TF_DataType.TF_STRING, IntPtr.Zero, 0, (UIntPtr)((ulong)size + 8));
+
+                IntPtr tensor = c_api.TF_TensorData(handle);
+                Marshal.WriteInt64(tensor, 0);
+                fixed (byte* src = &buffer[0])
+                    c_api.TF_StringEncode(src, (UIntPtr)buffer.Length, (sbyte*)(tensor + sizeof(Int64)), size, status);
+
+                status.Check(true);
+                return handle;
             }
 
             IntPtr dotHandle = IntPtr.Zero;
-            ulong size = 0;
+            int buffersize = 0;
 
             if (nd.dtype.Name != "String")
             {
-                dotHandle = Marshal.AllocHGlobal(nd.dtypesize * nd.size);
-                size = (ulong)(nd.size * nd.dtypesize);
+                buffersize = (nd.size * nd.dtypesize);
+                dotHandle = Marshal.AllocHGlobal(buffersize);
             }
 
             var dataType = ToTFDataType(nd.dtype);
@@ -116,7 +327,7 @@ namespace Tensorflow
                 dims,
                 dims.Length,
                 dotHandle,
-                (UIntPtr)size,
+                (UIntPtr)buffersize,
                 deallocator,
                 ref deallocator_called);
 
@@ -130,5 +341,59 @@ namespace Tensorflow
             _dtype = dtype;
             _id = ops.uid();
         }
+
+        /// <summary>
+        /// Creates a new tensor from the given array without copying memory. The array is pinned down and the pointer passed on.
+        /// </summary>
+        /// <param name="shape">Represents the tensor shape.</param>
+        /// <param name="data">The linear array of data, the data must fit in the tensor with the specified dimensions.</param>
+        /// <param name="element_size">The number of bytes in memory of a single array element</param>
+        /// <remarks>
+        /// Use the FromBuffer method to create a tensor that has the specified dimensions
+        /// and is initialized with data from the data array.   The data is copied starting
+        /// at the start offset, for count bytes and is laid out into the tensor following the
+        /// specified dimensions.
+        /// </remarks>
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        protected IntPtr CreateTensorWithoutCopying(TF_DataType dt, long[] shape, Array data, int element_size)
+        {
+            return CreateTensorWithoutCopying(dt, shape, data, 0, data.Length, element_size);
+        }
+
+        /// <summary>
+        /// Creates a new tensor from a subsection of the given array without copying memory. The array is pinned down and the pointer passed on.
+        /// </summary>
+        /// <param name="shape">Represents the tensor shape.</param>
+        /// <param name="data">The linear array of data, the data must fit in the tensor with the specified dimensions.</param>
+        /// <param name="start">The offset into the provided data array where the data resides.</param>
+        /// <param name="count">The number of elements to copy from data.</param>
+        /// <param name="element_size">The number of bytes in memory of a single array element</param>
+        /// <remarks>
+        /// Use the FromBuffer method to create a tensor that has the specified dimensions
+        /// and is initialized with data from the data array.   The data is copied starting
+        /// at the start offset, for count bytes and is laid out into the tensor following the
+        /// specified dimensions.
+        /// </remarks>
+        [MethodImpl(MethodImplOptions.AggressiveInlining)]
+        protected IntPtr CreateTensorWithoutCopying(TF_DataType dt, long[] shape, Array data, int start, int count, int element_size)
+        {
+            if (start < 0 || start > data.Length - count)
+                throw new ArgumentException("start + count > Array size");
+            
+            // get a handle to the pinned array which we will pass on to the tensor computation engine to use
+            var dataHandle = GCHandle.Alloc(data, GCHandleType.Pinned);
+
+            // Free the original buffer and set flag
+            Deallocator deallocator = (IntPtr values, IntPtr len, ref bool closure) =>
+            {
+                dataHandle.Free();
+                closure = true;
+            };
+
+            if (shape == null)
+                return TF_NewTensor(dt, null, 0, dataHandle.AddrOfPinnedObject() + start * element_size, (UIntPtr)(count * element_size), deallocator, ref deallocator_called);
+            else
+                return TF_NewTensor(dt, shape, shape.Length, dataHandle.AddrOfPinnedObject() + start * element_size, (UIntPtr)(count * element_size), deallocator, ref deallocator_called);
+        }
     }
 }
diff --git a/src/TensorFlowNET.Core/Tensors/Tensor.cs b/src/TensorFlowNET.Core/Tensors/Tensor.cs
index 786daa5c..cc54c667 100644
--- a/src/TensorFlowNET.Core/Tensors/Tensor.cs
+++ b/src/TensorFlowNET.Core/Tensors/Tensor.cs
@@ -31,7 +31,7 @@ namespace Tensorflow
     /// </summary>
     public partial class Tensor : IDisposable, ITensorOrOperation
     {
-        private readonly IntPtr _handle;
+        private IntPtr _handle;
 
         private int _id;
         private Operation _op;
@@ -351,6 +351,7 @@ namespace Tensorflow
         public void Dispose()
         {
             c_api.TF_DeleteTensor(_handle);
+            _handle = IntPtr.Zero;
             status.Dispose();
         }
 
diff --git a/src/TensorFlowNET.Core/Tensors/dtypes.cs b/src/TensorFlowNET.Core/Tensors/dtypes.cs
index ebe6e283..5000401a 100644
--- a/src/TensorFlowNET.Core/Tensors/dtypes.cs
+++ b/src/TensorFlowNET.Core/Tensors/dtypes.cs
@@ -52,32 +52,51 @@ namespace Tensorflow
             }
         }
 
+        // "sbyte", "byte", "short", "ushort", "int", "uint", "long", "ulong", "float", "double", "Complex"
         public static TF_DataType as_dtype(Type type)
         {
             TF_DataType dtype = TF_DataType.DtInvalid;
-
+            
             switch (type.Name)
             {
-                case "Boolean":
-                    dtype = TF_DataType.TF_BOOL;
+                case "SByte":
+                    dtype = TF_DataType.TF_INT8;
+                    break;
+                case "Byte":
+                    dtype = TF_DataType.TF_UINT8;
+                    break;
+                case "Int16":
+                    dtype = TF_DataType.TF_INT16;
+                    break;
+                case "UInt16":
+                    dtype = TF_DataType.TF_UINT16;
                     break;
                 case "Int32":
                     dtype = TF_DataType.TF_INT32;
                     break;
+                case "UInt32":
+                    dtype = TF_DataType.TF_UINT32;
+                    break;
                 case "Int64":
                     dtype = TF_DataType.TF_INT64;
                     break;
+                case "UInt64":
+                    dtype = TF_DataType.TF_UINT64;
+                    break;
                 case "Single":
                     dtype = TF_DataType.TF_FLOAT;
                     break;
                 case "Double":
                     dtype = TF_DataType.TF_DOUBLE;
                     break;
+                case "Complex":
+                    dtype = TF_DataType.TF_COMPLEX128;
+                    break;
                 case "String":
                     dtype = TF_DataType.TF_STRING;
                     break;
-                case "Byte":
-                    dtype = TF_DataType.TF_STRING;
+                case "Boolean":
+                    dtype = TF_DataType.TF_BOOL;
                     break;
                 default:
                     throw new Exception("as_dtype Not Implemented");