Struct UniPCSchedulerConfig 

pub struct UniPCSchedulerConfig {
    pub corrector: CorrectorConfiguration,
    pub sigma_schedule: SigmaSchedule,
    pub timestep_schedule: TimestepSchedule,
    pub solver_order: usize,
    pub prediction_type: PredictionType,
    pub num_training_timesteps: usize,
    pub thresholding: bool,
    pub dynamic_thresholding_ratio: f64,
    pub sample_max_value: f64,
    pub solver_type: SolverType,
    pub lower_order_final: bool,
}

Fields

corrector: CorrectorConfiguration

Configure the UNIC corrector. By default it is disabled

sigma_schedule: SigmaSchedule

Determines how sigma relates to a given timestep

timestep_schedule: TimestepSchedule

Determines the points

solver_order: usize

The solver order which can be 1 or higher. It is recommended to use solver_order=2 for guided sampling, and solver_order=3 for unconditional sampling.

prediction_type: PredictionType

Prediction type of the scheduler function

num_training_timesteps: usize

thresholding: bool

Whether to use the “dynamic thresholding” method. This is unsuitable for latent-space diffusion models such as Stable Diffusion.

dynamic_thresholding_ratio: f64

The ratio for the dynamic thresholding method. Valid only when thresholding=True.

sample_max_value: f64

The threshold value for dynamic thresholding.

solver_type: SolverType

lower_order_final: bool

Whether to use lower-order solvers in the final steps.

Trait Implementations

impl Clone for UniPCSchedulerConfig

fn clone(&self) -> UniPCSchedulerConfig

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for UniPCSchedulerConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for UniPCSchedulerConfig

fn default() -> Self

Returns the “default value” for a type.

impl SchedulerConfig for UniPCSchedulerConfig

fn build(&self, inference_steps: usize) -> Result<Box<dyn Scheduler>>