DiffusionTrialConfig<T>

Options & Configuration · AiDotNet.AutoML

Configuration for a diffusion model trial in AutoML.

Constructors

Constructor Summary
DiffusionTrialConfig()

Properties

Property Summary
DiffusionSchedulerType SchedulerType { get; set; } Gets or sets the scheduler type.
NoisePredictorType NoisePredictorType { get; set; } Gets or sets the noise predictor type.
double ConditioningDropoutProbability { get; set; } Gets or sets the probability that a training step drops its condition.
double GuidanceScale { get; set; } Gets or sets the guidance scale for classifier-free guidance.
double LearningRate { get; set; } Gets or sets the learning rate for training.
int BaseChannels { get; set; } Gets or sets the base channels for the noise predictor.
int ConditioningDim { get; set; } Gets or sets the width of the conditioning vector, which is the context width of the noise predictor’s cross-attention.
int ImageChannels { get; set; } Gets or sets the number of image channels the autoencoder reconstructs.
int InferenceSteps { get; set; } Gets or sets the number of inference steps.
int LatentDim { get; set; } Gets or sets the latent dimension (channels).
int LatentHeight { get; set; } Gets or sets the latent spatial height (default: 64 for 512x512 images with 8x downscaling).
int LatentWidth { get; set; } Gets or sets the latent spatial width (default: 64 for 512x512 images with 8x downscaling).
int NumHeads { get; set; } Gets or sets the number of attention heads in the noise predictor.
int NumResBlocks { get; set; } Gets or sets the number of residual blocks per level.
int TransformerDepth { get; set; } Gets or sets the number of transformer blocks in the DiT and U-ViT noise predictors.
int? Seed { get; set; } Gets or sets the optional random seed.

Methods

Method Summary
Dictionary<string, object> ToDictionary() Converts the configuration to a dictionary of parameters.
DiffusionTrialConfig<T> FromDictionary(Dictionary<string, object> parameters) Creates a configuration from a dictionary of parameters.