Pandipedia entry
Quantum Benchmark Datasets Compared
Choose a quantum benchmark resource by the task you want to evaluate. A dataset of simulated qubit dynamics, an optimization problem library and a molecular-property dataset answer different research questions. The comparison below separates their purposes and practical access requirements.
Compare the resources
| Resource | Purpose and contents | Access and requirements | Who it suits |
|---|---|---|---|
| QDataSet [1] | 52 simulated one- and two-qubit datasets for quantum control, tomography and noise-related machine-learning tasks. | Public documentation, notebooks and simulation code. The full collection is about 14 TB compressed. Check the dataset-host link before planning a download; start with the examples and select only the data your task needs. | Researchers learning models of quantum-system dynamics. |
| QOBLIB [2] | Ten optimization problem classes with instances, mathematical models, solution records and feasibility checks. | Public repository and results website. Read each problem’s format and dependencies. Code uses Apache 2.0; data uses CC BY 4.0, with separate attributions for third-party files. Running on paid quantum hardware is optional. | Teams comparing classical and quantum optimization methods on the same instances. |
| Quantum Algorithms Benchmarks [3] | Community problems and solutions for machine learning, optimization and Monte Carlo simulation. | Public Apache-2.0 repository, archived on 4 September 2026. Treat it as a historical reference; check submodule instructions and software compatibility before running it. | Researchers reproducing earlier benchmark approaches. |
| QM7 / QM7b / QM9 [4] | Molecular structures and computed properties for molecular machine learning. These are quantum-chemistry datasets, not a ready-made quantum-computer performance benchmark. | Creator documentation links to dataset downloads, formats and papers to cite. Check each dataset’s reuse terms and train/test protocol. | Chemistry and materials researchers predicting molecular properties. |
How to choose a benchmark
- Match the task. Use dynamics data for control or tomography, optimization instances for solver comparisons, and molecular data for property prediction.
- Make the comparison reproducible. Record dataset version, instance identifiers, preprocessing and software versions. Report hardware or simulator settings and the metric being compared.
- Keep claims within the experiment. A good result on one dataset does not establish broad quantum advantage. Compare against a suitable classical baseline under comparable conditions.
Resource documentation and repository status checked on 11 October 2026. Full datasets and benchmark programs were not downloaded or executed.
Sorry, Pandi could not find an answer.
Let's look at alternatives:
- Modify the query.
- Start a new thread.
- Remove sources (if manually added).