MPLUGDocOwl<T>

Models & Types · AiDotNet.VisionLanguage.Document

mPLUG-DocOwl: modular MLLM for document understanding with visual abstractor.

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

Implements: IDocumentUnderstandingModel<T>

For Beginners

mPLUG-DocOwl is a document understanding model from Alibaba with a visual abstractor for efficient document processing. Default values follow the original paper settings.

How It Works

mPLUG-DocOwl (Alibaba, 2023) is a modularized multimodal large language model for document understanding. It uses a visual abstractor module to compress high-resolution document image features into a compact set of visual tokens, bridging the ViT encoder and LLM decoder for efficient document VQA, information extraction, and document classification.

References:

  • Paper: “mPLUG-DocOwl: Modularized Multimodal Large Language Model for Document Understanding” (Alibaba, 2023)

Example

// Create an mPLUG-DocOwl model for document understanding
// with visual abstractor for efficient document feature compression
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 MPLUGDocOwl<double>(architecture, "mplugdocowl.onnx");
            
// Training mode with native layers
var trainModel = new MPLUGDocOwl<double>(architecture, new MPLUGDocOwlOptions());

Constructors

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

Properties

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

Methods

Method Summary
ModelMetadata<T> GetModelMetadata()
Tensor<T> AnswerDocumentQuestion(Tensor<T> documentImage, string question)
Tensor<T> EncodeImage(Tensor<T> image)
Tensor<T> ExtractText(Tensor<T> documentImage)
Tensor<T> GenerateFromImage(Tensor<T> image, string prompt) Generates text from a document image using mPLUG-DocOwl’s visual abstractor pipeline.