QSBench's picture
Update README.md
eec428e verified
metadata
license: cc-by-nc-4.0
task_categories:
  - tabular-regression
  - feature-extraction
language:
  - en
tags:
  - qiskit
  - quantum-circuits
  - synthetic-dataset
  - benchmark
  - expectation-values
  - quantum-computing
  - qml-benchmark
  - quantum dataset
  - qml dataset
  - quantum benchmark
  - noisy quantum data
  - device noise
  - hardware-mimic
  - thermal relaxation
  - error mitigation
  - noise robustness
pretty_name: >-
  QSBench Device Demo v1.0.0 – Realistic Device-like Noise (GenericBackendV2,
  n=10)
size_categories:
  - 1K<n<10K

QSBench Logo 🌐 Website | 🤗 Dataset | 🛠️ GitHub | 🚀 Interactive Demo

QSBench Device Demo v1.0.0

Realistic hardware-mimic quantum dataset — the most physically accurate noise demo in the QSBench family.

This release uses device noise model based on GenericBackendV2, which simulates a full set of realistic hardware errors (T1/T2 relaxation, gate errors, readout errors, and crosstalk-like effects).

2048 high-quality synthetic quantum circuits with realistic device-like noise.

Designed for researchers working on sim-to-real transfer, hardware-aware quantum ML, and benchmarking models under conditions closest to real quantum processors.

Why this dataset?

Most synthetic datasets use simplified noise (depolarizing or amplitude damping).
Device noise is much closer to what you see on actual IBM, Rigetti or IonQ hardware.
This dataset helps close the sim-to-real gap.

Use Cases

  • Sim-to-real transfer learning
  • Hardware-aware model benchmarking
  • Testing robustness under realistic multi-source noise
  • Error mitigation research (including crosstalk approximation)
  • Comparing simplified noise vs real-device noise

Dataset Overview

  • Samples: 2048
  • Qubits: 10
  • Depth: 8
  • Circuit Families: Mixed (HEA, RealAmplitudes, QFT, Efficient SU(2), Random)
  • Entanglement: Full
  • Noise: device (GenericBackendV2 — realistic device-like noise)
  • Observables: Z, X, Y in mixed mode (global + per-qubit)
  • Shots: 1024
  • Splits: Train / Validation / Test — deterministic hash-based

What's Inside Each Sample

  • Raw and transpiled QASM
  • Circuit adjacency matrix
  • Detailed gate statistics
  • Structural metrics (gate entropy, Meyer-Wallach entanglement)
  • Ideal expectation values
  • Noisy expectation values (after realistic device noise)
  • Error targets: error_<label>
  • Full generation metadata

Key Advantage

Unlike single-channel noise models, device noise combines multiple realistic error sources simultaneously — exactly what happens on real quantum hardware.

Load the Dataset

from datasets import load_dataset

dataset = load_dataset("QSBench/QSBench-Device-Demo-v1.0.0", split="train")
print(dataset[0])

Using pandas:

import pandas as pd
df = pd.read_parquet("data/shards/*.parquet")
print(df[["ideal_expval_Z_global", "noisy_expval_Z_global", "error_Z_global"]].head())

Repository Structure

Data is stored in the main branch:

QSBench-Device-Demo-v1.0.0/
├── README.md
└── data/
    └── shards/
        └── *.parquet

Metadata files are available in the metadata branch.

Related QSBench Datasets

Part of the QSBench Family

This is a public demo version. Full-scale Device Noise Pack and other specialized releases are available via the QSBench Generator.

Notes

  • Fully synthetic, generated with Qiskit Aer + GenericBackendV2

  • License: CC BY-NC 4.0 (Personal & Research Use)

Questions or custom requests? Visit qsbench.github.io or open an issue on GitHub.

Support QSBench

You can support the project directly on this Giveth page:
https://giveth.io/project/qsbench

Your donations help us generate larger datasets, cover GPU costs, and continue developing new realistic noise models.


Generated with QSBench Generator v5.1.0