Tensor

class Tensor(info: Optional[TensorInfo] = None, array: Optional[ndarray] = None, tensor: Optional[Tensor] = None)

Represents an input, output, or intermediate computation result.

Constructor for the Tensor class.

Parameters:

• info (TensorInfo object) -- The information of the tensor.

• array (numpy.ndarray) -- The data of the tensor.

• tensor (pytcim.Tensor) -- The existed pytcim tensor. Info and array will be ignored when tensor is set.

Methods

__init__	Constructor for the Tensor class.
astype	Creates a new tensor with the specified target data type, and optionally casts the data of original tensor with the target data type, stores casted data in the new tenor.
buffer	Retrieves the buffer that holds the data of the tensor.
cast	Creates a new tensor with the specified target data type, and optionally casts the data of original tensor with the target data type, stores casted data in the new tenor.
cast_to	Converts tensor data to special data type.
clone	Allocates memory for a new current Tensor object that is a copy of the current Tensor object if auto_copy is true.
copy_to	Copies the tensor data of the current Tensor object to another Tensor object.
device	Retrieves the device type on which the tensor data is stored.
device_id	Retrieves the logical ID of the Houmo device on which the tensor is stored.
info	Retrieves the information of the tensor associated with the current Tensor object.
numpy	Retrieves the data of the tensor.
select_batch	Selects a batch of elements along the batch dimension from the original Tensor object, and returns a new Tensor object that shares the underlying memory with the original Tensor object.
split_yuv	Splits a YUV-formatted tensor into separate Y and UV tensors without additional memory allocation.
to_host	Retrieves or creates a tensor in host memory with tensor data based on the current Tensor object.