Activations

class serket.nn.CeLU(alpha=1.0)

Celu activation function

Parameters:

alpha (float)

alpha: float

Field Information:

Name: alpha Default: 1.0

Callbacks:

• On setting attribute:

  • ScalarLike()

• On getting attribute:

  • <bound method Primitive.bind of stop_gradient>

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.ELU(alpha=1.0)

Exponential linear unit

Parameters:

alpha (float)

alpha: float

Field Information:

Name: alpha Default: 1.0

Callbacks:

• On setting attribute:

  • ScalarLike()

• On getting attribute:

  • <bound method Primitive.bind of stop_gradient>

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.GELU(approximate=False)

Gaussian error linear unit

Parameters:

approximate (bool)

approximate: bool

Field Information:

Name: approximate Default: False

Callbacks:

• On setting attribute:

  • IsInstance(klass=<class 'bool'>)

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.GLU

Gated linear unit

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.HardShrink(alpha=0.5)

Hard shrink activation function

Parameters:

alpha (float)

alpha: float

Field Information:

Name: alpha Default: 0.5

Callbacks:

• On setting attribute:

  • Range(min_val=0, max_val=inf, min_inclusive=True, max_inclusive=True)

  • ScalarLike()

• On getting attribute:

  • <bound method Primitive.bind of stop_gradient>

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.HardSigmoid(*a, **k)

Hard sigmoid activation function

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.HardSwish(*a, **k)

Hard swish activation function

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.HardTanh(*a, **k)

Hard tanh activation function

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.LeakyReLU(negative_slope=0.01)

Leaky ReLU activation function

Parameters:

negative_slope (float)

negative_slope: float

Field Information:

Name: negative_slope Default: 0.01

Callbacks:

• On setting attribute:

  • Range(min_val=0, max_val=inf, min_inclusive=True, max_inclusive=True)

  • ScalarLike()

• On getting attribute:

  • <bound method Primitive.bind of stop_gradient>

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.LogSigmoid(*a, **k)

Log sigmoid activation function

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.LogSoftmax(*a, **k)

Log softmax activation function

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.Mish(*a, **k)

Mish activation function https://arxiv.org/pdf/1908.08681.pdf.

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.PReLU(a=0.25)

Parametric ReLU activation function

Parameters:

a (float)

a: float

Field Information:

Name: a Default: 0.25

Callbacks:

• On setting attribute:

  • Range(min_val=0, max_val=inf, min_inclusive=True, max_inclusive=True)

  • ScalarLike()

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.ReLU(*a, **k)

ReLU activation function

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.ReLU6(*a, **k)

ReLU6 activation function

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.SeLU(*a, **k)

Scaled Exponential Linear Unit

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.Sigmoid(*a, **k)

Sigmoid activation function

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.SoftPlus(*a, **k)

SoftPlus activation function

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.SoftShrink(alpha=0.5)

SoftShrink activation function

Parameters:

alpha (float)

alpha: float

Field Information:

Name: alpha Default: 0.5

Callbacks:

• On setting attribute:

  • Range(min_val=0, max_val=inf, min_inclusive=True, max_inclusive=True)

  • ScalarLike()

• On getting attribute:

  • <bound method Primitive.bind of stop_gradient>

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.SoftSign(*a, **k)

SoftSign activation function

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.SquarePlus(*a, **k)

SquarePlus activation function

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.Swish(*a, **k)

Swish activation function

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.Tanh(*a, **k)

Tanh activation function

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.TanhShrink(*a, **k)

TanhShrink activation function

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array

class serket.nn.ThresholdedReLU(theta=1.0)

Thresholded ReLU activation function.

Parameters:

theta (float)

theta: float

Field Information:

Name: theta Default: 1.0

Callbacks:

• On setting attribute:

  • Range(min_val=0, max_val=inf, min_inclusive=True, max_inclusive=True)

  • ScalarLike()

• On getting attribute:

  • <bound method Primitive.bind of stop_gradient>

__call__(input)

Call self as a function.

Parameters:

input (Array)

Return type:

Array