TorchRuntimeClient

class TorchRuntimeClient(model, optimizer, loss_func, epochs=10, shots=1024, measurement_error_mitigation=False, provider=None, backend=None)

Bases: object

Deprecation: TorchRuntimeClient

The Qiskit Machine Learning Torch runtime client to call the Torch runtime.

Parametreler:

• model (torch.nn.Module) – A PyTorch module to be trained.

• optimizer (torch.optim.Optimizer) – A PyTorch optimizer for the model parameters.

• loss_func (Callable) – A PyTorch-compatible loss function. Can be one of the official PyTorch loss functions from torch.nn.loss or a custom function defined by the user.

• epochs (int) – The maximum number of training epochs. By default, 10.

• shots (int) – The number of shots for the quantum backend. By default, 1024.

• measurement_error_mitigation (bool) – Whether or not to use measurement error mitigation.

• default (By) –

• False. –

• provider (Provider | None) – IBMQ provider that supports runtime services.

• backend (Backend | None) – Selected quantum backend.

Attributes

loss_func	Return the loss function.
measurement_error_mitigation	Returns whether or not to use measurement error mitigation.
model	Return the model.
optimizer	Return the optimizer.
provider	Return the provider.
shots	Return the number of shots.

Methods

fit(train_loader[, val_loader, hooks, ...])	Train the model using the Torch Train Runtime ('torch-train').
predict(data_loader)	Perform prediction on the passed data using the trained model and the Torch Infer Runtime ('torch-infer').
score(data_loader, score_func)	Calculate a score using the trained model and the Torch Infer Runtime ('torch-infer').