WRSWorld Research StateContribute
WORLD RESEARCH STATE

Zhao and colleagues: UniPC diffusion predictor-corrector sampling

Contribution grade

Generative-model methods and numerical sampling · Human-designed research method or resource

Reference score80,000Provisional editorial estimate of this exact contribution; its grade is assessed separately.

What changed

Combined a unified higher-order predictor and corrector, reusing model evaluations to improve few-step diffusion sampling.

Scientific scope

Combined a unified higher-order predictor and corrector, reusing model evaluations to improve few-step diffusion sampling. Order improvements have numerical assumptions; correction without another model evaluation is not an unconditional zero-cost or exact finite-step generation guarantee.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Order improvements have numerical assumptions; correction without another model evaluation is not an unconditional zero-cost or exact finite-step generation guarantee.

Contributors

Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.

Full scientific record · Browse contribution grades · Request a correction