Benchmark RadarStocBench
StocBench evaluates generative models for probabilistic forecasting of stochastic fluid flows, using a two-dimensional Kolmogorov flow with stochastic forcing, measuring one-step distributional accuracy and preservation of the enstrophy spectrum during autoregressive rollouts.
- Released
- 2026-08-23
- Readiness
- Runnable
- Primary field
- General AI
Why it matters
It provides a controlled stochastic dynamics testbed to compare transport-based and distillation-based generative models under limited inference budgets, distinguishing aleatoric and epistemic uncertainty through a deterministic control variant.
Motivation
We benchmark transport-based generative models as well as distillation-based few-step methods for the probabilistic forecasting of stochastic fluid flows, with a particular focus on performance under limited inference budgets.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.