Skip to content

superpositions-kit

Hardware-agnostic variational quantum circuits as PyTorch nn.Module layers.

superpositions-kit lets you build, train, and export quantum ML models with a single Circuit() factory call — backed by PennyLane, Qiskit, Amazon Braket, or the custom Spsim simulator.

Quick look

from superpositions_kit.configs import Encoding, Variational, Measurement, Device
from superpositions_kit.circuit import Circuit
import torch

circuit = Circuit(
    blocks=[
        Encoding(in_features=4, rotation="X"),
        Variational(variational_type="basic", depth=2, rotation="Z"),
        Measurement(measurement_mode="all", basis="PauliZ"),
    ],
    device=Device(provider="Pennylane", kind="ideal_simulator",
                  name="lightning.qubit", interface="torch",
                  diff_method="adjoint", shots=None),
)

out = circuit(torch.randn(8, 4))  # (batch=8, features=4) → (batch=8, qubits=4)