GotOcr2ProjectorLayer<T>
Layers · AiDotNet.NeuralNetworks.Layers
GOT-OCR2’s vision-to-language projector (HF GotOcr2MultiModalProjector). Two bias-free stride-2 3x3
convolutions (C to 2C, then 2C to textDim) cut the map 4x per side. The tokens are then flattened and
pass a linear layer. ViTDet-B’s 64x64x256 neck output becomes 256 tokens.
Inheritance: LayerBase<T> → GotOcr2ProjectorLayer<T>
Implements: IShapeContract
Constructors
| Constructor | Summary |
|---|---|
GotOcr2ProjectorLayer(int visionChannels, int textDim, int grid) |
Properties
| Property | Summary |
|---|---|
bool SupportsTraining { get; } |
|
int TokenCount { get; } |
Number of image tokens produced. |
Methods
| Method | Summary |
|---|---|
IReadOnlyList<ILayer<T>> GetSubLayers() |
|
IReadOnlyList<OutputAxisContract> OutputAxesFor(int inputRank) |
|
void ResetState() |