Denoising Diffusion Implicit Models (DDIM) scheduler.

References

• [2010.02502] Denoising Diffusion Implicit Models

Denoising Diffusion Probabilistic Models (DDPM) scheduler.

References

• [2006.11239] Denoising Diffusion Probabilistic Models

Abstract type for schedulers.

Add noise to clean data using the forward diffusion process.

Input

• scheduler::Scheduler: scheduler to use
• x₀::AbstractArray: clean data to add noise to
• ϵ::AbstractArray: noise to add to clean data
• t::AbstractArray: timesteps used to weight the noise

Output

• xₜ::AbstractArray: noisy data at the given timesteps

Compute the velocity of the diffusion process.

Input

• scheduler::Scheduler: scheduler to use
• x₀::AbstractArray: clean data to add noise to
• ϵ::AbstractArray: noise to add to clean data
• t::AbstractArray: timesteps used to weight the noise

Output

• vₜ::AbstractArray: velocity at the given timesteps

References

[1, Ann. D]

Remove noise from model output using the backward diffusion process.

Input

• scheduler::Scheduler: scheduler to use
• xₜ::AbstractArray: sample to be denoised
• ϵᵧ::AbstractArray: predicted noise to remove
• t::AbstractArray: timestep t of xₜ

Output

• xₜ₋₁::AbstractArray: denoised sample at t=t-1
• x̂₀::AbstractArray: denoised sample at t=0