EDMDPMSolverMultistepScheduler
EDMDPMSolverMultistepScheduler
is a Karras formulation of DPMSolverMultistep
, a multistep scheduler from DPM-Solver: A Fast ODE Solver for Diffusion Probabilistic Model Sampling in Around 10 Steps and DPM-Solver++: Fast Solver for Guided Sampling of Diffusion Probabilistic Models by Cheng Lu, Yuhao Zhou, Fan Bao, Jianfei Chen, Chongxuan Li, and Jun Zhu.
DPMSolver (and the improved version DPMSolver++) is a fast dedicated high-order solver for diffusion ODEs with convergence order guarantee. Empirically, DPMSolver sampling with only 20 steps can generate high-quality samples, and it can generate quite good samples even in 10 steps.
EDMDPMSolverMultistepScheduler
class diffusers.EDMDPMSolverMultistepScheduler
< source >( sigma_min: float = 0.002 sigma_max: float = 80.0 sigma_data: float = 0.5 sigma_schedule: str = 'karras' num_train_timesteps: int = 1000 prediction_type: str = 'epsilon' rho: float = 7.0 solver_order: int = 2 thresholding: bool = False dynamic_thresholding_ratio: float = 0.995 sample_max_value: float = 1.0 algorithm_type: str = 'dpmsolver++' solver_type: str = 'midpoint' lower_order_final: bool = True euler_at_final: bool = False final_sigmas_type: Optional = 'zero' )
Parameters
- sigma_min (
float
, optional, defaults to 0.002) — Minimum noise magnitude in the sigma schedule. This was set to 0.002 in the EDM paper [1]; a reasonable range is [0, 10]. - sigma_max (
float
, optional, defaults to 80.0) — Maximum noise magnitude in the sigma schedule. This was set to 80.0 in the EDM paper [1]; a reasonable range is [0.2, 80.0]. - sigma_data (
float
, optional, defaults to 0.5) — The standard deviation of the data distribution. This is set to 0.5 in the EDM paper [1]. - sigma_schedule (
str
, optional, defaults tokarras
) — Sigma schedule to compute thesigmas
. By default, we the schedule introduced in the EDM paper (https://arxiv.org/abs/2206.00364). Other acceptable value is “exponential”. The exponential schedule was incorporated in this model: https://huggingface.co/stabilityai/cosxl. - num_train_timesteps (
int
, defaults to 1000) — The number of diffusion steps to train the model. - solver_order (
int
, defaults to 2) — The DPMSolver order which can be1
or2
or3
. It is recommended to usesolver_order=2
for guided sampling, andsolver_order=3
for unconditional sampling. - prediction_type (
str
, defaults toepsilon
, optional) — Prediction type of the scheduler function; can beepsilon
(predicts the noise of the diffusion process),sample
(directly predicts the noisy sample) or
v_prediction` (see section 2.4 of Imagen Video paper). - thresholding (
bool
, defaults toFalse
) — Whether to use the “dynamic thresholding” method. This is unsuitable for latent-space diffusion models such as Stable Diffusion. - dynamic_thresholding_ratio (
float
, defaults to 0.995) — The ratio for the dynamic thresholding method. Valid only whenthresholding=True
. - sample_max_value (
float
, defaults to 1.0) — The threshold value for dynamic thresholding. Valid only whenthresholding=True
andalgorithm_type="dpmsolver++"
. - algorithm_type (
str
, defaults todpmsolver++
) — Algorithm type for the solver; can bedpmsolver++
orsde-dpmsolver++
. Thedpmsolver++
type implements the algorithms in the DPMSolver++ paper. It is recommended to usedpmsolver++
orsde-dpmsolver++
withsolver_order=2
for guided sampling like in Stable Diffusion. - solver_type (
str
, defaults tomidpoint
) — Solver type for the second-order solver; can bemidpoint
orheun
. The solver type slightly affects the sample quality, especially for a small number of steps. It is recommended to usemidpoint
solvers. - lower_order_final (
bool
, defaults toTrue
) — Whether to use lower-order solvers in the final steps. Only valid for < 15 inference steps. This can stabilize the sampling of DPMSolver for steps < 15, especially for steps <= 10. - euler_at_final (
bool
, defaults toFalse
) — Whether to use Euler’s method in the final step. It is a trade-off between numerical stability and detail richness. This can stabilize the sampling of the SDE variant of DPMSolver for small number of inference steps, but sometimes may result in blurring. - final_sigmas_type (
str
, defaults to"zero"
) — The finalsigma
value for the noise schedule during the sampling process. If"sigma_min"
, the final sigma is the same as the last sigma in the training schedule. Ifzero
, the final sigma is set to 0.
Implements DPMSolverMultistepScheduler in EDM formulation as presented in Karras et al. 2022 [1].
EDMDPMSolverMultistepScheduler
is a fast dedicated high-order solver for diffusion ODEs.
[1] Karras, Tero, et al. “Elucidating the Design Space of Diffusion-Based Generative Models.” https://arxiv.org/abs/2206.00364
This model inherits from SchedulerMixin and ConfigMixin. Check the superclass documentation for the generic methods the library implements for all schedulers such as loading and saving.
convert_model_output
< source >( model_output: Tensor sample: Tensor = None ) → torch.Tensor
Convert the model output to the corresponding type the DPMSolver/DPMSolver++ algorithm needs. DPM-Solver is designed to discretize an integral of the noise prediction model, and DPM-Solver++ is designed to discretize an integral of the data prediction model.
The algorithm and model type are decoupled. You can use either DPMSolver or DPMSolver++ for both noise prediction and data prediction models.
dpm_solver_first_order_update
< source >( model_output: Tensor sample: Tensor = None noise: Optional = None ) → torch.Tensor
One step for the first-order DPMSolver (equivalent to DDIM).
multistep_dpm_solver_second_order_update
< source >( model_output_list: List sample: Tensor = None noise: Optional = None ) → torch.Tensor
One step for the second-order multistep DPMSolver.
multistep_dpm_solver_third_order_update
< source >( model_output_list: List sample: Tensor = None ) → torch.Tensor
One step for the third-order multistep DPMSolver.
scale_model_input
< source >( sample: Tensor timestep: Union ) → torch.Tensor
Ensures interchangeability with schedulers that need to scale the denoising model input depending on the
current timestep. Scales the denoising model input by (sigma**2 + 1) ** 0.5
to match the Euler algorithm.
set_begin_index
< source >( begin_index: int = 0 )
Sets the begin index for the scheduler. This function should be run from pipeline before the inference.
set_timesteps
< source >( num_inference_steps: int = None device: Union = None )
Sets the discrete timesteps used for the diffusion chain (to be run before inference).
step
< source >( model_output: Tensor timestep: int sample: Tensor generator = None return_dict: bool = True ) → SchedulerOutput or tuple
Parameters
- model_output (
torch.Tensor
) — The direct output from learned diffusion model. - timestep (
int
) — The current discrete timestep in the diffusion chain. - sample (
torch.Tensor
) — A current instance of a sample created by the diffusion process. - generator (
torch.Generator
, optional) — A random number generator. - return_dict (
bool
) — Whether or not to return a SchedulerOutput ortuple
.
Returns
SchedulerOutput or tuple
If return_dict is True
, SchedulerOutput is returned, otherwise a
tuple is returned where the first element is the sample tensor.
Predict the sample from the previous timestep by reversing the SDE. This function propagates the sample with the multistep DPMSolver.
SchedulerOutput
class diffusers.schedulers.scheduling_utils.SchedulerOutput
< source >( prev_sample: Tensor )
Base class for the output of a scheduler’s step
function.