Compute recall accuracy. Args: pred (Numpy ndarray): Prediction array, should be in shape (B, C) target (Numpy ndarray): Ground t

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

Performs all reduce of a tensor after converting to FP16. Args: tensor(Tensor): a tensor to be all-reduced

Performs a single optimization step. Args: param_name(String): the name of the param param_value(Tensor): par

Get arguments for atan2 function. \param abs half-precision floating-point value \return \a abs and sqrt(1 - \a abs^2) as Q0.30

Generate batch arrays from transformed image list. Args: indexes (Sequence): current batch indexes list, e.g. [n, n + 1, ..., n +

The operator casts the elements of a given input tensor to a data type specified by the 'to' argument and returns an output tensor of the sam

download data into tmp

download data into tmp

3x3 convolution with padding