LoRARecycleOptions<T, TInput, TOutput>

Options & Configuration · AiDotNet.MetaLearning.Options

Configuration options for LoRA-Recycle (Hu et al., CVPR 2025).

Inheritance: ModelOptions → LoRARecycleOptions<T, TInput, TOutput>

Implements: IMetaLearnerOptions<T>

How It Works

LoRA-Recycle distills a “meta-LoRA” from diverse pre-tuned LoRA adapters without accessing their private training data. It supports tuning-free few-shot adaptation by recycling knowledge from previously-learned tasks via prototype-based matching.

Key Parameters:

  • NumRecycledAdapters — number of previously-learned LoRA adapters to maintain
  • Rank — rank of each LoRA adapter
  • PrototypeDim — dimensionality of prototype embeddings for adapter selection

Constructors

Constructor Summary
LoRARecycleOptions(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; } Optimizer for task adaptation.
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 KLWeight { get; set; } KL divergence weight for distillation loss.
double OuterLearningRate { get; set; }
double SelectionTemperature { get; set; } Temperature parameter for softmax-based adapter weighting.
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 NumRecycledAdapters { get; set; } Number of recycled LoRA adapters to maintain in the adapter bank.
int PrototypeDim { get; set; } Dimensionality of the prototype embedding used for adapter selection.
int Rank { get; set; } Rank of each LoRA adapter (controls adapter capacity).
int? RandomSeed { get; set; }

Methods

Method Summary
IMetaLearnerOptions<T> Clone()
bool IsValid()