about [ai] physics
Mission, consortium governance, and how we evaluate models on the platform.
Mission
Surrogate benchmarks rarely share a yardstick. Problems ship with different datasets, meshes, splits, and metrics, so published scores are hard to compare, and hard to trust for engineering decisions.
AIPhysics.org provides transparent, standardized evaluation on predicted vs actual for the quantities that drive engineering decisions. Those are the scalars and fields teams need a surrogate to get right before they trust it in a workflow, not proxy metrics from training.
When using open community datasets, such as OpenRadioss Neon, DrivAerML, or HiLiftAeroML, we credit the original projects and link directly to their source downloads. For full models and corpora, see the attribution FAQs.
How we score
Each problem reports native predicted vs actual metrics on the quantities that matter for that physics: scalars (peak stress, Kt, Cl, Cd) and fields (stress, pressure, or displacement on the mesh). Recipes differ by problem; absolute errors stay in native units on the problem page.
The score arena maps those natives into a shared 0–1 playground: Scalars, Fields, and Explore = √(S × F) when both exist. Pin a run and open “How scored?” for the problem-specific math. Problem pages keep the full submission tables so you can see where each model lands and where it falls short.
Consortium
Early-stage community working toward shared benchmarks, metrics, and evaluation standards.
The consortium is in early stages. AIPhysics.org is operated by Rescale AI Engineering today. The outline below is where we are headed as partners get involved, not how the platform is run right now. If you want to contribute a benchmark dataset, reach out using the contact form below.
- Consortium council (planned): Equal representation from industry and academic members. Quarterly review of benchmarks and metrics.
- Metrics working groups (planned): Domain experts would define evaluation metrics and engineering thresholds for each benchmark.
- Transparent process (planned): Decisions documented publicly. Community input welcomed on benchmark proposals.
Contact
Consortium membership, benchmark contributions, general questions, and technical support.