Table of Contents

Property VisionParser

Namespace
LMKit.Retrieval
Assembly
LM-Kit.NET.dll

VisionParser

Gets or sets the vision language model (VLM) parser used for advanced document understanding.

public VlmOcr VisionParser { get; set; }

Property Value

VlmOcr

The VLM-based OCR instance, or null if vision parsing is not configured.

Examples

var docRag = new DocumentRag(embeddingModel);

// Configure vision-based document understanding
LM visionModel = LM.LoadFromModelID("gemma3:4b");
docRag.VisionParser = new VlmOcr(visionModel);

Remarks

When ProcessingMode is set to DocumentUnderstanding or when Auto selects vision-based processing, this parser analyzes page images using a vision language model to extract structured content.

Vision-based parsing typically produces markdown-formatted output that preserves document structure, including headings, tables, lists, and other layout elements. This approach is particularly effective for complex documents with mixed content types.

Share