module `npy.numpy_onnx_impl`#

Short summary#

module mlprodict.npy.numpy_onnx_impl

numpy functions implemented with onnx.

Functions#

function	truncated documentation
`abs`	See `numpy.abs()`.
`acos`	See `numpy.acos()`.
`acosh`	See `numpy.acosh()`.
`amax`	See `numpy.amax()`.
`amin`	See `numpy.amin()`.
`arange`	See `numpy.arange()`, start, stop must be specified.
`argmax`	See `numpy.argmax()`.
`argmin`	See `numpy.argmin()`.
`asin`	See `numpy.asin()`.
`asinh`	See `numpy.asinh()`.
`atan`	See `numpy.atan()`.
`atanh`	See `numpy.atanh()`.
`ceil`	See `numpy.ceil()`.
`clip`	See `numpy.clip()`.
`compress`	See `numpy.compress()`. numpy.compress(condition, x) or npnx.compress(x, condition).
`concat`	Operator concat, handle `numpy.vstack()` and `numpy.hstack()`.
`cos`	See `numpy.cos()`.
`cosh`	See `numpy.cosh()`.
`cst`	Creates a constant. log(x) + numpy.float32(1) works but numpy.float32(32) + log(x) fails because Python calls …
`cumsum`	See `numpy.cumsum()`.
`det`	See `numpy.linalg:det()`.
`dot`	See `numpy.dot()`
`einsum`	See `numpy.einsum()`.
`erf`	See scipy.special.erf.
`exp`	See `numpy.exp()`.
`expand_dims`	See `numpy.expand_dims()`.
`expit`	See scipy.special.expit.
`floor`	See `numpy.floor()`.
`hstack`	See `numpy.hstack()`.
`identity`	Identity.
`isnan`	See `numpy.isnan()`.
`log`	See `numpy.log()`.
`log1p`	See `numpy.log1p()`.
`matmul`	See `numpy.matmul()`.
`mean`	See `numpy.mean()`.
`onnx_if`	Implements a test with onnx syntax.
`pad`	It does not implement `numpy.pad()` but the ONNX version `onnx_pad`. …
`prod`	See `numpy.prod()`.
`reciprocal`	See `numpy.reciprocal()`.
`relu`	relu
`round`	See `numpy.round()`.
`sigmoid`	See scipy.special.expit.
`sign`	See `numpy.sign()`.
`sin`	See `numpy.sin()`.
`sinh`	See `numpy.sinh()`.
`sqrt`	See `numpy.sqrt()`.
`squeeze`	See `numpy.squeeze()`.
`sum`	See `numpy.sum()`.
`tan`	See `numpy.tan()`.
`tanh`	See `numpy.tanh()`.
`topk`	See `numpy.argsort()`.
`transpose`	See `numpy.transpose()`.
`unsqueeze`	See `numpy.expand_dims()`.
`vstack`	See `numpy.vstack()`.
`where`	See `numpy.where()`.

Documentation#

numpy functions implemented with onnx.

New in version 0.6.

Changed in version 0.7.

source on GitHub

mlprodict.npy.numpy_onnx_impl.abs(x)#: See numpy.abs().

mlprodict.npy.numpy_onnx_impl.acos(x)#: See numpy.acos().

mlprodict.npy.numpy_onnx_impl.acosh(x)#: See numpy.acosh().

mlprodict.npy.numpy_onnx_impl.amax(x, axis=None, keepdims=0)#: See numpy.amax().

mlprodict.npy.numpy_onnx_impl.amin(x, axis=None, keepdims=0)#: See numpy.amin().

mlprodict.npy.numpy_onnx_impl.arange(start, stop, step=1)#: See numpy.arange(), start, stop must be specified.

mlprodict.npy.numpy_onnx_impl.argmax(x, axis=0, keepdims=0)#

See numpy.argmax().

Warning

ONNX does not implement default value axis=None.

source on GitHub

mlprodict.npy.numpy_onnx_impl.argmin(x, axis=0, keepdims=0)#

See numpy.argmin().

Warning

ONNX does not implement default value axis=None.

source on GitHub

mlprodict.npy.numpy_onnx_impl.asin(x)#: See numpy.asin().

mlprodict.npy.numpy_onnx_impl.asinh(x)#: See numpy.asinh().

mlprodict.npy.numpy_onnx_impl.atan(x)#: See numpy.atan().

mlprodict.npy.numpy_onnx_impl.atanh(x)#: See numpy.atanh().

mlprodict.npy.numpy_onnx_impl.ceil(x)#: See numpy.ceil().

mlprodict.npy.numpy_onnx_impl.clip(x, a_min=None, a_max=None)#: See numpy.clip().

mlprodict.npy.numpy_onnx_impl.compress(condition, x, axis=None)#

See numpy.compress(). numpy.compress(condition, x) or npnx.compress(x, condition).

source on GitHub

mlprodict.npy.numpy_onnx_impl.concat(*x, axis=0)#

Operator concat, handle numpy.vstack() and numpy.hstack().

source on GitHub

mlprodict.npy.numpy_onnx_impl.cos(x)#: See numpy.cos().

mlprodict.npy.numpy_onnx_impl.cosh(x)#: See numpy.cosh().

mlprodict.npy.numpy_onnx_impl.cst(x, dtype=None)#

Creates a constant. log(x) + numpy.float32(1) works but numpy.float32(32) + log(x) fails because Python calls numpy.float32.__add__ instead of OnnxVar.__add__. With this function, expression cst(1.) + log(x) is valid. Parameter dtype is used to overwrite the default dtype (numpy.float32 for floats and numpy.int64 for ints.

source on GitHub

mlprodict.npy.numpy_onnx_impl.cumsum(x, axis)#: See numpy.cumsum().

mlprodict.npy.numpy_onnx_impl.det(x)#: See numpy.linalg:det().

mlprodict.npy.numpy_onnx_impl.dot(a, b)#: See numpy.dot()

mlprodict.npy.numpy_onnx_impl.einsum(*x, equation=None)#: See numpy.einsum().

mlprodict.npy.numpy_onnx_impl.erf(x)#: See scipy.special.erf.

mlprodict.npy.numpy_onnx_impl.exp(x)#: See numpy.exp().

mlprodict.npy.numpy_onnx_impl.expand_dims(x, axis)#: See numpy.expand_dims().

mlprodict.npy.numpy_onnx_impl.expit(x)#: See scipy.special.expit.

mlprodict.npy.numpy_onnx_impl.floor(x)#: See numpy.floor().

mlprodict.npy.numpy_onnx_impl.hstack(*x)#: See numpy.hstack().

mlprodict.npy.numpy_onnx_impl.identity(x)#: Identity.

mlprodict.npy.numpy_onnx_impl.isnan(x)#: See numpy.isnan().

mlprodict.npy.numpy_onnx_impl.log(x)#: See numpy.log().

mlprodict.npy.numpy_onnx_impl.log1p(x)#: See numpy.log1p().

mlprodict.npy.numpy_onnx_impl.matmul(a, b)#: See numpy.matmul().

mlprodict.npy.numpy_onnx_impl.mean(x, axis=None, keepdims=0)#: See numpy.mean().

mlprodict.npy.numpy_onnx_impl.onnx_if(condition, then_branch, else_branch)#

Implements a test with onnx syntax.

Parameters:

condition – condition (OnnxVar)
then_branch – then branch, of type if_then_else
else_branch – else branch, of type if_then_else

Returns:

result (OnnxVar)

source on GitHub

mlprodict.npy.numpy_onnx_impl.pad(x, pads, constant_value=None, mode='constant')#

It does not implement numpy.pad() but the ONNX version onnx_pad.

source on GitHub

mlprodict.npy.numpy_onnx_impl.prod(x, axis=None, keepdims=0)#: See numpy.prod().

mlprodict.npy.numpy_onnx_impl.reciprocal(x)#: See numpy.reciprocal().

mlprodict.npy.numpy_onnx_impl.relu(x)#

mlprodict.npy.numpy_onnx_impl.round(x)#: See numpy.round().

mlprodict.npy.numpy_onnx_impl.sigmoid(x)#: See scipy.special.expit.

mlprodict.npy.numpy_onnx_impl.sign(x)#: See numpy.sign().

mlprodict.npy.numpy_onnx_impl.sin(x)#: See numpy.sin().

mlprodict.npy.numpy_onnx_impl.sinh(x)#: See numpy.sinh().

mlprodict.npy.numpy_onnx_impl.sqrt(x)#: See numpy.sqrt().

mlprodict.npy.numpy_onnx_impl.squeeze(x, axis=None)#: See numpy.squeeze().

mlprodict.npy.numpy_onnx_impl.sum(x, axis=None, keepdims=0)#: See numpy.sum().

mlprodict.npy.numpy_onnx_impl.tan(x)#: See numpy.tan().

mlprodict.npy.numpy_onnx_impl.tanh(x)#: See numpy.tanh().

mlprodict.npy.numpy_onnx_impl.topk(x, k, axis=-1, largest=1, sorted=1)#: See numpy.argsort().

mlprodict.npy.numpy_onnx_impl.transpose(x, perm=(1, 0))#: See numpy.transpose().

mlprodict.npy.numpy_onnx_impl.unsqueeze(x, axes)#: See numpy.expand_dims().

mlprodict.npy.numpy_onnx_impl.vstack(*x)#: See numpy.vstack().

mlprodict.npy.numpy_onnx_impl.where(cond, x, y)#: See numpy.where().

module npy.numpy_onnx_impl#

Short summary#

Functions#

Documentation#

module `npy.numpy_onnx_impl`#