ShowO2<T>

Models & Types · AiDotNet.VisionLanguage.Unified

Show-o2: improved native unified multimodal model.

Inheritance: NeuralNetworkBase<T> → VisionLanguageModelBase<T> → ShowO2<T>

Implements: IUnifiedVisionModel<T>

For Beginners

Show-o2 is an improved version of Show-o with enhanced understanding and generation quality. Default values follow the original paper settings.

How It Works

Show-o2 (NUS, 2025) improves upon Show-o with enhanced native unified multimodal capabilities. It refines the discrete diffusion approach for image generation with improved visual token quality and strengthens the multimodal understanding pathway through better cross-modal attention and scaled training on larger multimodal datasets.

References:

  • Paper: “Show-o2: Improved Unified Multimodal Understanding and Generation” (NUS, 2025)

Example

// Create a Show-o2 model for improved unified multimodal processing
// with enhanced native understanding and generation capabilities
var architecture = new NeuralNetworkArchitecture<double>(
    inputType: InputType.TwoDimensional,
    taskType: NeuralNetworkTaskType.ImageClassification,
    inputHeight: 224, inputWidth: 224, inputDepth: 3, outputSize: 512);
            
// ONNX inference mode with pre-trained model
var model = new ShowO2<double>(architecture, "showo2.onnx");
            
// Training mode with native layers
var trainModel = new ShowO2<double>(architecture, new ShowO2Options());

Constructors

Constructor Summary
ShowO2(NeuralNetworkArchitecture<T> architecture, ShowO2Options options, IGradientBasedOptimizer<T, Tensor<T>, Tensor<T>> optimizer)
ShowO2(NeuralNetworkArchitecture<T> architecture, string modelPath, ShowO2Options options)

Properties

Property Summary
bool SupportsGeneration { get; }
int DecoderEmbeddingDim { get; }
int EmbeddingDimension { get; }
int MaxGenerationLength { get; }

Methods

Method Summary
ModelMetadata<T> GetModelMetadata()
Tensor<T> EncodeImage(Tensor<T> image)
Tensor<T> GenerateFromImage(Tensor<T> image, string prompt) Generates text from image using Show-o2’s improved unified transformer.
Tensor<T> GenerateImage(string textDescription) Generates an image from text using Show-o2’s improved discrete diffusion.