DatasetDistillerOptions
Options & Configuration · AiDotNet.Data.Quality
Configuration options for dataset distillation.
How It Works
Dataset distillation synthesizes a small set of examples that capture the essence of the full training set. Training on the distilled set produces similar model quality to training on the full dataset.
Constructors
| Constructor | Summary |
|---|---|
DatasetDistillerOptions() |
Properties
| Property | Summary |
|---|---|
double DistillLearningRate { get; set; } |
Learning rate for optimizing distilled samples. |
int NumSteps { get; set; } |
Number of distillation optimization steps. |
int SamplesPerClass { get; set; } |
Number of synthetic samples per class to generate. |
int? Seed { get; set; } |
Random seed for reproducibility. |
Methods
| Method | Summary |
|---|---|
void Validate() |
Validates that all option values are within acceptable ranges. |