r"""Use :func:`~.fake_quant_bias` to process ``target``. Only valid when ``act_fake_quant`` and ``weight_fake_quant`` are both enabled.

r"""Returns a :class:`Tensor` with the same data and number of elements with the specified :attr:`~.Tensor.dtype`.

r"""Compute gradients (or vector-Jacobian product) for all attached tensors, accumulate to corresponding .grad attribute, and release resource

Read an environment variable and interpret it as a boolean. True values are (case insensitive): 'y', 'yes', 't', 'true', 'on', and '1'; false

r""" Args: version: version to deprecate this function origin: origin module path name: function name tbd: to be

get device tensor

whether endpoint contain a given var node

Apply fake quantization to the inp tensor. Args: inp: the input tensor which need to be faked. qparams: to get mode, qmin, qmax,

! * \brief fill zeros in the content of a dev tensor */

user interface for deconv

* AES-ECB block encryption/decryption */

! * \brief offset on allocated block in bytes */

get the opr impl so setting other than param() can be modified

record an event on cn

r"""Replaces vars in the graph. Args: repl_dict: the map {old_var: new_var} that specifies how to replace the vars.

r"""Sets whether tensor.shape returns a tensor instead of a tuple

size() in log2

set underlying device memory manager

take a -128..127 value and stbi__clamp it and convert to 0..255

returns 1 if "a*b*c + add" has no negative terms/factors and doesn't overflow