JMPOptions<T, TInput, TOutput>
Options & Configuration ยท AiDotNet.MetaLearning.Options
Configuration options for JMP (Joint Multi-Phase meta-learning).
Inheritance: ModelOptions โ JMPOptions<T, TInput, TOutput>
Implements: IMetaLearnerOptions<T>
How It Works
JMP uses a multi-phase inner loop with separate learning rates and regularization. Phase 1 (coarse) uses a higher learning rate for fast, rough adaptation. Phase 2 (fine) uses a lower learning rate with stronger regularization toward the Phase 1 result for careful refinement.
Constructors
| Constructor | Summary |
|---|---|
JMPOptions(IFullModel<T, TInput, TOutput> metaModel) |
Creates options for meta-learning metaModel, with the default settings. |
Properties
| Property | Summary |
|---|---|
IEpisodicDataLoader<T, TInput, TOutput> DataLoader { get; set; } |
Episodic data loader that supplies the meta-training tasks; null when tasks are passed to training directly. |
IFullModel<T, TInput, TOutput> MetaModel { get; set; } |
The model whose parameters the meta-learner trains to adapt quickly to new tasks. |
IGradientBasedOptimizer<T, TInput, TOutput> InnerOptimizer { get; set; } |
Must stay null: this algorithm uses a two-phase inner learning-rate schedule, which assumes plain SGD inner steps, so a supplied optimizer is rejected when the learner is built. |
IGradientBasedOptimizer<T, TInput, TOutput> MetaOptimizer { get; set; } |
Optimizer for the meta-update. |
ILossFunction<T> LossFunction { get; set; } |
Loss function for meta-training; null uses the default multi-class classification loss. |
bool EnableCheckpointing { get; set; } |
|
bool UseFirstOrder { get; set; } |
Whether to use the first-order approximation of the meta-gradient. |
double InnerLearningRate { get; set; } |
|
double OuterLearningRate { get; set; } |
|
double Phase1Fraction { get; set; } |
Fraction of adaptation steps in Phase 1 (coarse). |
double Phase1LRMultiplier { get; set; } |
Learning rate multiplier for Phase 1 (coarse). |
double Phase2LRMultiplier { get; set; } |
Learning rate multiplier for Phase 2 (fine). |
double PhaseRegWeight { get; set; } |
L2 regularization during Phase 2 toward Phase 1 result. |
double? GradientClipThreshold { get; set; } |
|
int AdaptationSteps { get; set; } |
|
int CheckpointFrequency { get; set; } |
|
int EvaluationFrequency { get; set; } |
|
int EvaluationTasks { get; set; } |
|
int MetaBatchSize { get; set; } |
|
int NumMetaIterations { get; set; } |
|
int? RandomSeed { get; set; } |
Methods
| Method | Summary |
|---|---|
IMetaLearnerOptions<T> Clone() |
|
bool IsValid() |