TabMOptions<T>
Options & Configuration · AiDotNet.Models.Options
Configuration options for TabM, a parameter-efficient ensemble model for tabular data.
Inheritance: ModelOptions → NeuralNetworkOptions → ModelHyperparameterOptions → FinancialNeuralNetworkOptions → RiskModelOptions<T> → TabMOptions<T>
For Beginners
TabM is a smart way to get the benefits of model ensembles (multiple models voting together) without the huge computational cost.
Key concepts:
- Ensembles: Multiple models combined usually give better predictions
- Parameter Sharing: TabM shares most weights across ensemble members
- Rank Vectors: Small per-member vectors that customize the shared weights
- Efficient: Gets ensemble benefits with minimal extra parameters
Why this matters:
- Traditional ensembles need N times the parameters for N models
- TabM only adds about 1-5% extra parameters per ensemble member
- Same computational cost as a single model during inference
- Often outperforms both single models and traditional ensembles
Example usage:
How It Works
TabM uses BatchEnsemble-style parameter sharing to create multiple ensemble members with minimal parameter overhead. Each member shares base weights but has its own small rank vectors that modulate the shared weights.
Reference: “TabM: Advancing Tabular Deep Learning With Parameter-Efficient Ensembling” (2024)
Constructors
| Constructor | Summary |
|---|---|
TabMOptions() |
Properties
| Property | Summary |
|---|---|
bool AverageEnsemble { get; set; } |
Gets or sets whether to average ensemble predictions or concatenate. |
bool UseBias { get; set; } |
Gets or sets whether to use bias in the ensemble layers. |
bool UseFeatureEmbeddings { get; set; } |
Gets or sets whether to use numerical feature embeddings. |
bool UseLayerNorm { get; set; } |
Gets or sets whether to use layer normalization. |
double DropoutRate { get; set; } |
Gets or sets the dropout rate. |
double RankInitScale { get; set; } |
Gets or sets the initialization scale for rank vectors. |
double WeightDecay { get; set; } |
Gets or sets the weight decay (L2 regularization) coefficient. |
int FeatureEmbeddingDimension { get; set; } |
Gets or sets the embedding dimension for feature embeddings. |
int NumEnsembleMembers { get; set; } |
Gets or sets the number of ensemble members. |
int[] HiddenDimensions { get; set; } |
Gets or sets the hidden layer dimensions. |
Methods
| Method | Summary |
|---|---|
TabMOptions<T> Clone() |
Creates a copy of the options. |