ShowOOptions
Options & Configuration · AiDotNet.VisionLanguage.Unified
Configuration options for Show-o: single transformer for unified understanding and generation.
Inheritance: ModelOptions → GenerativeVLMOptions → UnifiedVisionOptions → ShowOOptions
For Beginners
These options configure the released Show-o architecture and can be changed for other checkpoints.
How It Works
Defaults follow the released 256px Show-o checkpoint: Phi-1.5 at 2048 hidden dimensions, 24 layers and 32 heads, an 8192-entry MAGVIT-v2 codebook, 256 image tokens, and the published AdamW 1e-4/0.01 training recipe.
Constructors
| Constructor | Summary |
|---|---|
ShowOOptions() |
Initializes a new instance with default values. |
ShowOOptions(ShowOOptions other) |
Initializes a new instance by copying from another instance. |
Properties
| Property | Summary |
|---|---|
int DiffusionSteps { get; set; } |
Gets or sets the number of discrete mask-prediction diffusion steps. |
int ImageTokenCount { get; set; } |
Gets or sets the number of MAGVIT-v2 image-token positions. |