Loads a data file into a list of `InputBatch`s.

shallow copy calls copy constructor of singa::Tensor Returns: new tensor copied

Copy data from other Tensor instance. Args: t (Tensor): source Tensor.

Copy the data from the numpy array. used as static method Args: data: singa ctensor np_array: source numpy array

Copy the data from the numpy array. used as static method Args: data: singa ctensor np_array: source numpy array

Cosine function. This function is exact to rounding for all rounding modes. See also:** Documentation for [std::cos](https://en.cppreference.com/w/cp

Create a CudaGPU device on the given device ID. Args: device_id (int): GPU card ID. Returns: a swig converted CudaGPU device

Create a list of CudaGPU devices. Args: device_ids (list): a list of GPU card IDs. Returns: a list of swig converted CudaGPU

Crop a max square patch of the input image at given position and scale it into given size. According to img size, crop position could

Crop a max square patch of the input image at given position and resize it into given size. Args: patch(tuple): width, height

Inject deep into distutils to customize how the dispatch to gcc/nvcc works. If you subclass UnixCCompiler, it's not trivial to get your subcla

To measure the io/compute time more acccuretely, we pick the data pre_processing here.

Same as clone(). Returns: a new Tensor

this function checks whether the next index falls on a special multiplier of the outer shape so the algorithm knows when to jump over/back to a starti

batch size need to be large to see all negative ans

Elementi-wise multiplication. Args: lhs (Tensor): lhs tensor rhs (Tensor): rhs tensor ret (Tensor, optional): if not None

Elementi-wise multiplication. Args: lhs (Tensor): lhs tensor rhs (Tensor): rhs tensor ret (Tensor, optional): if not None

Produces an embedding operator. Args: Returns: the output Tensor.

Apply random enhancement for Color,Contrast,Brightness,Sharpness. Args: scale(float): enhancement degree is from [1-scale, 1+scale]

Return `a=b`, where a and b are Tensor.

Sets the model in evaluation mode.

Raise exception flags. This raises the specified floating point exceptions and also invokes any additional automatic exception handling as configured

Fit the sampler with architecture's score. :param score: :return:

Flattens the input tensor into a 2D matrix. If input tensor has shape `(d_0, d_1, ... d_n)` then the output will have shape `(d_0 X d_1 ...

Forward propagation. Args: xs: input args consisting of only CTensors. Returns: CTensor instance(s)

Forward propagation Args: *input: input arguments consisting of only PyTensors Returns: PyTensor instance(s)

Args: img (PIL Image or numpy array): Image to be processed. Returns: PIL Image or numpy array: Processed im