LXMERTOptions
Options & Configuration · AiDotNet.VisionLanguage.Foundational
Configuration options for LXMERT cross-modal encoder.
Inheritance: ModelOptions → FoundationalVLMOptions → LXMERTOptions
How It Works
LXMERT (Tan and Bansal, EMNLP 2019) has three encoder types: object relationship encoder, language encoder, and cross-modality encoder with cross-attention layers.
Constructors
| Constructor | Summary |
|---|---|
LXMERTOptions() |
Initializes a new instance with default values. |
LXMERTOptions(LXMERTOptions other) |
Initializes a new instance by copying from another instance. |
Properties
| Property | Summary |
|---|---|
int MaxVisualObjects { get; set; } |
Gets or sets the maximum number of visual objects. |
int NumCrossModalityLayers { get; set; } |
Gets or sets the number of cross-modality encoder layers. |
int NumRelationshipLayers { get; set; } |
Gets or sets the number of object relationship encoder layers. |