Skip to main content
This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub
Implements the Epsilon Scaling method from the research paper “Elucidating the Exposure Bias in Diffusion Models.” This method improves sample quality by scaling the predicted noise during the sampling process. It uses a uniform schedule to mitigate exposure bias in diffusion models.

Inputs

ParameterData TypeRequiredRangeDescription
modelMODELYes-The model to apply epsilon scaling to
scaling_factorFLOATNo0.5 - 1.5The factor used to scale the predicted noise (default: 1.005)

Outputs

Output NameData TypeDescription
modelMODELThe model with epsilon scaling applied