Package 

Class Tensor

  • 
public abstract class Tensor

    Representation of an ExecuTorch Tensor. Behavior is similar to PyTorch's tensor objects.

    Most tensors will be constructed as {@code Tensor.fromBlob(data, shape)}, where {@code data} can be an array or a direct Buffer (of the proper subclass). Helper methods are provided to allocate buffers properly.

    To access Tensor data, see dtype, shape, and various {@code getDataAs*} methods.

    When constructing {@code Tensor} objects with {@code data} as an array, it is not specified whether this data is copied or retained as a reference so it is recommended not to modify it after constructing. {@code data} passed as a Buffer is not copied, so it can be modified between Module calls to avoid reallocation. Data retrieved from {@code Tensor} objects may be copied or may be a reference to the {@code Tensor}'s internal data buffer. {@code shape} is always copied.

    Warning: These APIs are experimental and subject to change without notice

    • Method Detail

      • allocateByteBuffer

         static ByteBuffer allocateByteBuffer(int numElements)

        Allocates a new direct ByteBuffer with native byte order with specified capacity thatcan be used in fromBlob, .

        Parameters:
        numElements - capacity (number of elements) of result buffer.

      • allocateIntBuffer

         static IntBuffer allocateIntBuffer(int numElements)

        Allocates a new direct IntBuffer with native byte order with specified capacity thatcan be used in fromBlob.

        Parameters:
        numElements - capacity (number of elements) of result buffer.

      • allocateFloatBuffer

         static FloatBuffer allocateFloatBuffer(int numElements)

        Allocates a new direct FloatBuffer with native byte order with specified capacity thatcan be used in fromBlob.

        Parameters:
        numElements - capacity (number of elements) of result buffer.

      • allocateLongBuffer

         static LongBuffer allocateLongBuffer(int numElements)

        Allocates a new direct LongBuffer with native byte order with specified capacity thatcan be used in fromBlob.

        Parameters:
        numElements - capacity (number of elements) of result buffer.

      • allocateDoubleBuffer

         static DoubleBuffer allocateDoubleBuffer(int numElements)

        Allocates a new direct DoubleBuffer with native byte order with specified capacity thatcan be used in fromBlob.

        Parameters:
        numElements - capacity (number of elements) of result buffer.

      • fromBlobUnsigned

         static Tensor fromBlobUnsigned(Array<byte> data, Array<long> shape)

        Creates a new Tensor instance with dtype torch.uint8 with specified shape and data as array ofbytes.

        Parameters:
        data - Tensor elements
        shape - Tensor shape

      • fromBlob

         static Tensor fromBlob(Array<byte> data, Array<long> shape)

        Creates a new Tensor instance with dtype torch.int8 with specified shape and data as array ofbytes.

        Parameters:
        data - Tensor elements
        shape - Tensor shape

      • fromBlob

         static Tensor fromBlob(Array<int> data, Array<long> shape)

        Creates a new Tensor instance with dtype torch.int32 with specified shape and data as array ofints.

        Parameters:
        data - Tensor elements
        shape - Tensor shape

      • fromBlob

         static Tensor fromBlob(Array<float> data, Array<long> shape)

        Creates a new Tensor instance with dtype torch.float32 with specified shape and data as arrayof floats.

        Parameters:
        data - Tensor elements
        shape - Tensor shape

      • fromBlob

         static Tensor fromBlob(Array<long> data, Array<long> shape)

        Creates a new Tensor instance with dtype torch.int64 with specified shape and data as array oflongs.

        Parameters:
        data - Tensor elements
        shape - Tensor shape

      • fromBlob

         static Tensor fromBlob(Array<double> data, Array<long> shape)

        Creates a new Tensor instance with dtype torch.float64 with specified shape and data as arrayof doubles.

        Parameters:
        data - Tensor elements
        shape - Tensor shape

      • fromBlobUnsigned

         static Tensor fromBlobUnsigned(ByteBuffer data, Array<long> shape)

        Creates a new Tensor instance with dtype torch.uint8 with specified shape and data.

        Parameters:
        data - Direct buffer with native byte order that contains {@code Tensor.numel(shape)} elements.
        shape - Tensor shape

      • fromBlob

         static Tensor fromBlob(ByteBuffer data, Array<long> shape)

        Creates a new Tensor instance with dtype torch.int8 with specified shape and data.

        Parameters:
        data - Direct buffer with native byte order that contains {@code Tensor.numel(shape)} elements.
        shape - Tensor shape

      • fromBlob

         static Tensor fromBlob(IntBuffer data, Array<long> shape)

        Creates a new Tensor instance with dtype torch.int32 with specified shape and data.

        Parameters:
        data - Direct buffer with native byte order that contains {@code Tensor.numel(shape)} elements.
        shape - Tensor shape

      • fromBlob

         static Tensor fromBlob(FloatBuffer data, Array<long> shape)

        Creates a new Tensor instance with dtype torch.float32 with specified shape and data.

        Parameters:
        data - Direct buffer with native byte order that contains {@code Tensor.numel(shape)} elements.
        shape - Tensor shape

      • fromBlob

         static Tensor fromBlob(LongBuffer data, Array<long> shape)

        Creates a new Tensor instance with dtype torch.int64 with specified shape and data.

        Parameters:
        data - Direct buffer with native byte order that contains {@code Tensor.numel(shape)} elements.
        shape - Tensor shape

      • fromBlob

         static Tensor fromBlob(DoubleBuffer data, Array<long> shape)

        Creates a new Tensor instance with dtype torch.float64 with specified shape and data.

        Parameters:
        data - Direct buffer with native byte order that contains {@code Tensor.numel(shape)} elements.
        shape - Tensor shape

      • numel

         long numel()

        Returns the number of elements in this tensor.

      • numel

         static long numel(Array<long> shape)

        Calculates the number of elements in a tensor with the specified shape.

      • shape

         Array<long> shape()

        Returns the shape of this tensor. (The array is a fresh copy.)

      • toByteArray

         Array<byte> toByteArray()

        Serializes a {@code Tensor} into a byte array. Note: This method is experimental and subject tochange without notice. This does NOT supoprt list type.

      • fromByteArray

         static Tensor fromByteArray(Array<byte> bytes)

        Deserializes a {@code Tensor} from a byte[]. Note: This method is experimental and subject tochange without notice. This does NOT supoprt list type.

        Parameters:
        bytes - The byte array to deserialize from.