qailab.torch.qmodel

qailab.torch.qmodel#

Module with QModel

Summary#

Classes:

HybridOptimizer

Dummy HO

QModel

Quantum model class

Reference#

class qailab.torch.qmodel.HybridOptimizer(model, lr_classical=0.01, lr_quantum=0.01, optimizer_quantum='SGD', optimizer_classical='Adam', betas=(0.9, 0.999), spsa_resamplings=1, spsa_gamma_decay=0.101, spsa_alpha_decay=0.602)[source]#

Bases: object

Dummy HO

class qailab.torch.qmodel.QModel(module: Module, loss: Callable, optimizer_type: type[Optimizer] | type[HybridOptimizer] | str = 'adamw', learning_rate: float | Literal['auto'] = 'auto', quantum_learning_rate: float | Literal['auto'] = 'auto', batch_size: int = 1, epochs: int = 1, validation_fraction: float = 0.2, shuffle: bool = True, device: Literal['cpu', 'cuda', 'mps'] = 'cpu', metric: Literal['accuracy', 'mse'] = 'accuracy')[source]#

Bases: BaseEstimator

Quantum model class

Parameters#

module: nn.Module

pytorch Module representing the quantum or classical neural network.

loss: Callable

pytorch loss function to be used during training.

optimizer_type: type[Optimizer] | str, default = “adamw”

pytorch Optimizer class to be used during training.

learning_rate: float | Literal[‘auto’], default = “auto”

learning rate used by the optimizer, “auto” sets it to optimizer’s default one.

quantum_learning_rate: float | Literal[‘auto’], default = ‘auto’

learning rate for quantum layers used by the HybridOptimizer, “auto” sets it to optimizer’s default one.

batch_size: int, default = 1

number of training examples in batch.

epochs: int, default = 1

number of epochs to train the model.

validation_fraction: float, default = 0.2

share of the training dataset to be used for validation.

shuffle: bool, default = True

whether to shuffle data every epoch.

device: {“cpu”,”cuda”,”mps”}, default=”cpu”

the device neural network will be trained on.

Attributes#

optimizer: Optimizer

pytorch optimizer object used during training

loss_history: dict[str,list]

history of loss values from the last fit call. dict contains keys ‘training’ and ‘validation’

module: Module#
loss: Callable#
optimizer_type: type[Optimizer] | type[HybridOptimizer]#
learning_rate: float | Literal['auto']#
quantum_learning_rate: float | Literal['auto']#
optimizer: Optimizer#
batch_size: int#
epochs: int#
validation_fraction: float#
shuffle: bool#
device: Literal['cpu', 'cuda', 'mps'] = 'cpu'#
set_fit_request(*, x: bool | None | str = '$UNCHANGED$') → QModel#

Request metadata passed to the fit method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to fit if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to fit.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters#

xstr, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED

Metadata routing for x parameter in fit.

Returns#

selfobject

The updated object.

set_predict_request(*, x: bool | None | str = '$UNCHANGED$') → QModel#

Request metadata passed to the predict method.

Note that this method is only relevant if enable_metadata_routing=True (see sklearn.set_config()). Please see User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to predict if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to predict.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Note

This method is only relevant if this estimator is used as a sub-estimator of a meta-estimator, e.g. used inside a Pipeline. Otherwise it has no effect.

Parameters#

xstr, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED

Metadata routing for x parameter in predict.

Returns#

selfobject

The updated object.

reset_parameters() → None[source]#

Resets parameters of layers

fit(x: Tensor | ndarray | DataFrame, y: Tensor | ndarray | DataFrame | Series) → QModel[source]#

scikit-learn like fit method trains the neural network based on training set (x,y).

Parameters#

x: Tensor | np.ndarray | pd.DataFrame

The training input samples of shape (n_samples, n_features).

y: Tensor | np.array | pd.DataFrame | pd.Series

The training target values of shape (n_samples,) or (n_samples, n_outputs)

Returns#

self: QModel

trained NN model

fit_predict(x: Tensor | ndarray | DataFrame, y: Tensor | ndarray | DataFrame | Series) → Tensor[source]#

scikit-learn like fit_predict method trains the neural network based on training set (x,y) and predicts values for training examples x combines fit and predict methods into one.

Parameters#

x: Tensor | np.ndarray | pd.DataFrame

The training input samples of shape (n_samples, n_features).

y: Tensor | np.array | pd.DataFrame | pd.Series

The training target values of shape (n_samples,) or (n_samples, n_outputs).

Returns#

y_pred: Tensor

The predicted values for the training examples x.

predict(x: Tensor | ndarray | DataFrame) → Tensor[source]#

scikit-learn like predict method predicts values for examples input examples x.

Parameters#

x: Tensor | np.ndarray | pd.DataFrame

The input samples of shape (n_samples, n_features).

Returns#

y_pred: Tensor | np.ndarray | pd.DataFrame

The predicted values for examples x.

set_params(**params)[source]#

scikit-learn like param setting method allows changing parameters of the model set in constructor.

Parameters#

**params: dict

Keyword arguments representing the parameters to be set.

to_torch_module() → Module[source]#

Returns QModel’s module with torch neural network.

Returns:

Torch neural network.

Return type:

nn.Module

Contents