Results¶

These reference results are generated from the public package APIs used by the notebooks. Notebook-derived result pages are generated separately from committed notebook outputs:

The configurations are intentionally small, deterministic smoke-scale examples. Notebook-derived result pages are extracted from outputs already committed in the notebooks. They are reproducible reference outputs, not quantum-advantage claims.

Environment¶

Generated: stable
Git commit: stable
Python: 3.12.1
Package version: 0.2.14
PennyLane: 0.45.0
Matplotlib backend: Agg
Default execution: analytic default.qubit unless a shot count is listed

Summary¶

Workflow	Primary metric	Value	Runtime
Variational quantum classifier	`train_accuracy`	0.4595	not recorded
Variational quantum regression	`train_mse`	0.9098	not recorded
Quantum convolutional neural network	`train_accuracy`	0.8333	not recorded
Quantum autoencoder	`test_compression_fidelity`	0.7014	not recorded
Quantum kernel classifier	`train_accuracy`	0.8519	not recorded
Trainable quantum kernel	`train_accuracy`	0.7333	not recorded
Trainable quantum kernel regressor	`train_mse`	0.0125	not recorded
Quantum metric learning	`train_accuracy`	0.5946	not recorded

Variational quantum classifier¶

Configuration:

dataset=moons, n_samples=50, noise=0.1000, seed=123, n_layers=1, steps=8, shots=analytic

Metric	Value
`train_accuracy`	0.4595
`test_accuracy`	0.6154
`final_loss`	1.4790
`runtime_seconds`	not recorded

Images:

moons embedangle layers1 steps8 samples50 noise0p1 seed123 analytic noiseless dataset

Variational quantum regression¶

Configuration:

dataset=linear, n_samples=50, noise=0.1000, seed=123, n_layers=1, steps=8, shots=analytic

Metric	Value
`train_mse`	0.9098
`test_mse`	0.3316
`final_loss`	0.9841
`runtime_seconds`	not recorded

Images:

linear layers1 steps8 samples50 noise0p1 seed123 analytic noiseless dataset

Quantum convolutional neural network¶

Configuration:

dataset=moons, n_samples=40, noise=0.1000, seed=123, steps=6, shots=analytic

Metric	Value
`train_accuracy`	0.8333
`test_accuracy`	0.9000
`final_loss`	0.4556
`runtime_seconds`	not recorded

Images:

moons steps6 samples40 noise0p1 seed123 analytic noiseless dataset

Quantum autoencoder¶

Configuration:

family=correlated, n_samples=32, noise=0.0500, seed=123, n_layers=1, latent_qubits=2, steps=6

Metric	Value
`test_compression_fidelity`	0.7014
`test_reconstruction_fidelity`	0.7014
`final_loss`	0.3676
`runtime_seconds`	not recorded

Images:

correlated layers1 latent2 steps6 samples32 noise0p05 seed123 loss

Quantum kernel classifier¶

Configuration:

dataset=moons, n_samples=36, noise=0.1000, seed=123, shots=analytic

Metric	Value
`train_accuracy`	0.8519
`test_accuracy`	0.8889
`runtime_seconds`	not recorded

Images:

moons samples36 noise0p1 seed123 analytic noiseless dataset

Trainable quantum kernel¶

Configuration:

dataset=moons, n_samples=20, noise=0.1000, seed=123, embedding_layers=1, steps=2, shots_train=analytic, shots_kernel=analytic

Metric	Value
`train_accuracy`	0.7333
`test_accuracy`	0.8000
`final_alignment`	0.1755
`final_loss`	-0.1755
`runtime_seconds`	not recorded

Images:

moons trainable kernel embdata reupload layers1 steps2 samples20 noise0p1 seed123 analytic analytic noiseless alignment

Trainable quantum kernel regressor¶

Configuration:

dataset=sine, n_samples=20, noise=0.1000, seed=123, embedding_layers=1, steps=2, shots_train=analytic, shots_kernel=analytic, alpha=0.0010

Metric	Value
`train_mse`	0.0125
`test_mse`	0.4359
`final_alignment`	0.4287
`final_loss`	-0.4287
`runtime_seconds`	not recorded

Quantum metric learning¶

Configuration:

dataset=moons, samples=50, seed=42, layers=1, steps=8, pairs_per_step=16, log_every=0

Metric	Value
`train_accuracy`	0.5946
`test_accuracy`	0.6923
`final_loss`	0.1152
`runtime_seconds`	not recorded

Images:

moons layers1 steps8 samples50 margin0p5 seed42 embeddings

Reproduce¶

Regenerate this file and notebook-result pages from the repository root:

python docs/pages/generate_results.py

Notebook result pages are extracted from outputs already committed in the notebooks. Execute notebooks locally first when notebook outputs need to change, commit those notebook outputs, then regenerate and commit the derived result pages and assets. Generated images are written under docs/pages/assets/reference-results/ and embedded above.