FusionType
Enums · AiDotNet.VisionLanguage.Encoders
Specifies how vision and language features are fused in a VLM.
Fields
| Field | Summary |
|---|---|
FusionType BridgeLayers |
Bridge layers: explicit bridge connections between encoders. |
FusionType CoAttention |
Co-attention: parallel streams with co-attention layers. |
FusionType CrossModal |
Cross-modal encoder: dedicated cross-modal transformer layers. |
FusionType DualStream |
Dual stream: separate encoders with cross-attention bridges. |
FusionType SingleStream |
Single stream: visual and text tokens concatenated in one transformer. |
int value__ |