GraphModelOptions

Options & Configuration · AiDotNet.Models.Options

Shared hyperparameters for the graph task models — the node, link and graph-level heads that sit on a message-passing encoder.

Inheritance: ModelOptions → NeuralNetworkOptions → ModelHyperparameterOptions → GraphModelOptions

Known derived types: GraphClassificationOptions, LinkPredictionOptions, NodeClassificationOptions

For Beginners

These models all learn from a graph: a set of items (nodes) joined by relationships (edges). The settings here are the ones every one of them has. Each model’s own options class ships the values it was published with, so you do not normally set any of them.

How It Works

The layer count is deliberately NOT here. The three models name it differently (NumLayers for the node and link heads, NumGnnLayers for the graph-level one) and those names are part of each model’s published vocabulary, so the property lives on each leaf rather than being renamed to a common word.

The maximum gradient norm is not here either — it is already on MaxGradNorm, which every model options class inherits.

Properties

Property Summary
double DropoutRate { get; set; } Gets or sets the dropout rate applied between layers.
int HiddenDim { get; set; } Gets or sets the width of the hidden message-passing layers.