ClipOptions
Options & Configuration · AiDotNet.NeuralNetworks.Options
Input and embedding configuration for the ONNX-only ClipNeuralNetwork.
Inheritance: ModelOptions → NeuralNetworkOptions → ModelHyperparameterOptions → VisionLanguageInputOptions → ClipOptions
How It Works
The three dimensions must match the supplied ONNX encoders; they do not resize the graphs. The image path accepts three RGB channels. Patch sizes, vocabularies, tower widths, heads, and layer counts are properties of those graphs, not configurable native layers, so this options type does not expose setters for them.
Constructors
| Constructor | Summary |
|---|---|
ClipOptions() |
Initializes a new instance of the ClipOptions class carrying this model’s shipped defaults. |
ClipOptions(ClipOptions other) |
Copies the CLIP input dimensions and inherited model settings. |
Methods
| Method | Summary |
|---|---|
void Validate() |
Throws if a value this model requires has been left unset or is not positive. |