diff --git a/src/main/java/io/bioimage/modelrunner/tensorflow/v2/api050/tensor/TensorBuilder.java b/src/main/java/io/bioimage/modelrunner/tensorflow/v2/api050/tensor/TensorBuilder.java
index bc0a7c0..f0eb079 100644
--- a/src/main/java/io/bioimage/modelrunner/tensorflow/v2/api050/tensor/TensorBuilder.java
+++ b/src/main/java/io/bioimage/modelrunner/tensorflow/v2/api050/tensor/TensorBuilder.java
@@ -140,7 +140,7 @@ public static TUint8 buildUByte(RandomAccessibleInterval<UnsignedByteType> tenso
 		int[] sArr = new int[tensorShape.length];
 		for (int i = 0; i < sArr.length; i ++)
 			sArr[i] = (int) tensorShape[i];
-		blocks.copy( new long[tensorShape.length], flatArr, sArr );
+		blocks.copy( tensor.minAsLongArray(), flatArr, sArr );
 		ByteDataBuffer dataBuffer = RawDataBufferFactory.create(flatArr, false);
 		TUint8 ndarray = Tensor.of(TUint8.class, Shape.of(ogShape), dataBuffer);
 		return ndarray;
@@ -169,7 +169,7 @@ public static TInt32 buildInt(RandomAccessibleInterval<IntType> tensor)
 		int[] sArr = new int[tensorShape.length];
 		for (int i = 0; i < sArr.length; i ++)
 			sArr[i] = (int) tensorShape[i];
-		blocks.copy( new long[tensorShape.length], flatArr, sArr );
+		blocks.copy( tensor.minAsLongArray(), flatArr, sArr );
 		IntDataBuffer dataBuffer = RawDataBufferFactory.create(flatArr, false);
 		TInt32 ndarray = TInt32.tensorOf(Shape.of(ogShape),
 			dataBuffer);
@@ -199,7 +199,7 @@ private static TInt64 buildLong(RandomAccessibleInterval<LongType> tensor)
 		int[] sArr = new int[tensorShape.length];
 		for (int i = 0; i < sArr.length; i ++)
 			sArr[i] = (int) tensorShape[i];
-		blocks.copy( new long[tensorShape.length], flatArr, sArr );
+		blocks.copy( tensor.minAsLongArray(), flatArr, sArr );
 		LongDataBuffer dataBuffer = RawDataBufferFactory.create(flatArr, false);
 		TInt64 ndarray = TInt64.tensorOf(Shape.of(ogShape),
 			dataBuffer);
@@ -230,7 +230,7 @@ public static TFloat32 buildFloat(
 		int[] sArr = new int[tensorShape.length];
 		for (int i = 0; i < sArr.length; i ++)
 			sArr[i] = (int) tensorShape[i];
-		blocks.copy( new long[tensorShape.length], flatArr, sArr );
+		blocks.copy( tensor.minAsLongArray(), flatArr, sArr );
 		FloatDataBuffer dataBuffer = RawDataBufferFactory.create(flatArr, false);
 		TFloat32 ndarray = TFloat32.tensorOf(Shape.of(ogShape), dataBuffer);
 		return ndarray;
@@ -260,7 +260,7 @@ private static TFloat64 buildDouble(
 		int[] sArr = new int[tensorShape.length];
 		for (int i = 0; i < sArr.length; i ++)
 			sArr[i] = (int) tensorShape[i];
-		blocks.copy( new long[tensorShape.length], flatArr, sArr );
+		blocks.copy( tensor.minAsLongArray(), flatArr, sArr );
 		DoubleDataBuffer dataBuffer = RawDataBufferFactory.create(flatArr, false);
 		TFloat64 ndarray = TFloat64.tensorOf(Shape.of(ogShape), dataBuffer);
 		return ndarray;