Diffusers.Schedulers.DDIM
— TypeDenoising Diffusion Implicit Models (DDIM) scheduler.
References
Diffusers.Schedulers.DDPM
— TypeDenoising Diffusion Probabilistic Models (DDPM) scheduler.
References
Diffusers.Schedulers.Scheduler
— TypeAbstract type for schedulers.
Diffusers.Schedulers.forward
— MethodAdd noise to clean data using the forward diffusion process.
Input
scheduler::Scheduler
: scheduler to usex₀::AbstractArray
: clean data to add noise toϵ::AbstractArray
: noise to add to clean datat::AbstractArray
: timesteps used to weight the noise
Output
xₜ::AbstractArray
: noisy data at the given timesteps
Diffusers.Schedulers.get_velocity
— MethodCompute the velocity of the diffusion process.
Input
scheduler::Scheduler
: scheduler to usex₀::AbstractArray
: clean data to add noise toϵ::AbstractArray
: noise to add to clean datat::AbstractArray
: timesteps used to weight the noise
Output
vₜ::AbstractArray
: velocity at the given timesteps
References
Diffusers.Schedulers.reverse
— MethodRemove noise from model output using the backward diffusion process.
Input
scheduler::Scheduler
: scheduler to usexₜ::AbstractArray
: sample to be denoisedϵᵧ::AbstractArray
: predicted noise to removet::AbstractArray
: timestep t ofxₜ
Output
xₜ₋₁::AbstractArray
: denoised sample at t=t-1x̂₀::AbstractArray
: denoised sample at t=0