class HyperParameters:
def __init__(self, default=None, parent_name=None, parent_values=None):
self._default = default
self._parent_name = parent_name
self._parent_values = parent_values
[docs]
class Boolean(HyperParameters):
"""
Define a boolean hyperparameter. This is used in neural network hyperparameter
tuning.
Refer to
`KerasTuner's documentation <https://keras.io/api/keras_tuner/hyperparameters/>`_
for information on the arguments :cite:`chollet2015keras`.
"""
[docs]
def __init__(self, default=None, parent_name=None, parent_values=None):
HyperParameters.__init__(self, default, parent_name, parent_values)
# ===========================================================
# Methods
[docs]
def hp(self, hp, hp_name):
"""
Create an instance of ``keras_tuner.HyperParameters.Boolean``.
Parameters
----------
hp: keras_tuner.HyperParameters
Base hyperparameter class.
hp_name: str
Name of the hyperparameter.
Returns
-------
boolean_hp: keras_tuner.HyperParameters.Boolean
Boolean KerasTuner hyperparameter.
"""
return hp.Boolean(
name=hp_name,
default=self._default,
parent_name=self._parent_name,
parent_values=self._parent_values,
)
[docs]
class Int(HyperParameters):
"""
Define an integer hyperparameter. This is used in neural network hyperparameter
tuning.
Refer to
`KerasTuner's documentation <https://keras.io/api/keras_tuner/hyperparameters/>`_
for information on the arguments :cite:`chollet2015keras`.
"""
[docs]
def __init__(
self,
min_value,
max_value,
step=None,
sampling="linear",
default=None,
parent_name=None,
parent_values=None,
):
self._min_value = min_value
self._max_value = max_value
self._step = step
self._sampling = sampling
HyperParameters.__init__(self, default, parent_name, parent_values)
# ===========================================================
# Methods
[docs]
def hp(self, hp, hp_name):
"""
Create an instance of ``keras_tuner.HyperParameters.Int``.
Parameters
----------
hp: keras_tuner.HyperParameters
Base hyperparameter class.
hp_name: str
Name of the hyperparameter.
Returns
-------
int_hp: keras_tuner.HyperParameters.Int
Integer KerasTuner hyperparameter.
"""
return hp.Int(
name=hp_name,
min_value=self._min_value,
max_value=self._max_value,
step=self._step,
sampling=self._sampling,
default=self._default,
parent_name=self._parent_name,
parent_values=self._parent_values,
)
[docs]
class Float(HyperParameters):
"""
Define an floating point hyperparameter. This is used in neural network
hyperparameter tuning.
Refer to
`KerasTuner's documentation <https://keras.io/api/keras_tuner/hyperparameters/>`_
for information on the arguments :cite:`chollet2015keras`.
"""
[docs]
def __init__(
self,
min_value,
max_value,
step=None,
sampling="linear",
default=None,
parent_name=None,
parent_values=None,
):
self._min_value = min_value
self._max_value = max_value
self._step = step
self._sampling = sampling
HyperParameters.__init__(self, default, parent_name, parent_values)
# ===========================================================
# Methods
[docs]
def hp(self, hp, hp_name):
"""
Create an instance of ``keras_tuner.HyperParameters.Float``.
Parameters
----------
hp: keras_tuner.HyperParameters
Base hyperparameter class.
hp_name: str
Name of the hyperparameter.
Returns
-------
float_hp: keras_tuner.HyperParameters.Float
Float KerasTuner hyperparameter.
"""
return hp.Float(
name=hp_name,
min_value=self._min_value,
max_value=self._max_value,
step=self._step,
sampling=self._sampling,
default=self._default,
parent_name=self._parent_name,
parent_values=self._parent_values,
)
[docs]
class Choice(HyperParameters):
"""
Define choice hyperparameter. This is used in neural network hyperparameter tuning.
This can be used for string or other parameters where a range is not applicable.
Refer to
`KerasTuner's documentation <https://keras.io/api/keras_tuner/hyperparameters/>`_
for information on the arguments :cite:`chollet2015keras`.
"""
[docs]
def __init__(
self, values, ordered=None, default=None, parent_name=None, parent_values=None
):
self._values = values
self._ordered = ordered
HyperParameters.__init__(self, default, parent_name, parent_values)
# ===========================================================
# Methods
[docs]
def hp(self, hp, hp_name):
"""
Create an instance of ``keras_tuner.HyperParameters.Choice``.
Parameters
----------
hp: keras_tuner.HyperParameters
Base hyperparameter class.
hp_name: str
Name of the hyperparameter.
Returns
-------
choice_hp: keras_tuner.HyperParameters.Choice
Choice KerasTuner hyperparameter.
"""
return hp.Choice(
name=hp_name,
values=self._values,
ordered=self._ordered,
default=self._default,
parent_name=self._parent_name,
parent_values=self._parent_values,
)
# ===========================================================
# Getters
@property
def values(self):
"""
: list of int, float, str, or bool: The possible choices for the hyperparameter.
"""
return self._values
[docs]
class Fixed(HyperParameters):
"""
Define fixed hyperparameter. This is used in neural network hyperparameter tuning.
Refer to
`KerasTuner's documentation <https://keras.io/api/keras_tuner/hyperparameters/>`_
for information on the arguments :cite:`chollet2015keras`.
"""
[docs]
def __init__(self, value, parent_name=None, parent_values=None):
self._value = value
HyperParameters.__init__(
self, parent_name=parent_name, parent_values=parent_values
)
# ===========================================================
# Methods
[docs]
def hp(self, hp, hp_name):
"""
Create an instance of ``keras_tuner.HyperParameters.Fixed``.
Parameters
----------
hp: keras_tuner.HyperParameters
Base hyperparameter class.
hp_name: str
Name of the hyperparameter.
Returns
-------
fixed_hp: keras_tuner.HyperParameters.Fixed
Fixed KerasTuner hyperparameter.
"""
return hp.Fixes(
name=hp_name,
value=self._value,
parent_name=self._parent_name,
parent_values=self._parent_values,
)