Property VlmMaximumCompletionTokens

Namespace

LMKit.Document.Conversion

Assembly

LM-Kit.NET.dll

VlmMaximumCompletionTokens

Gets or sets the maximum number of completion tokens emitted by the vision model per page. Use -1 to disable the limit. Defaults to 8192, sized so real-world pages (dense tables, multi-column text) transcribe without truncation.

public int VlmMaximumCompletionTokens { get; set; }

Property Value

int

Examples

// No cap: let the VLM emit as many tokens as the page needs.
var options = new DocumentToMarkdownOptions { VlmMaximumCompletionTokens = -1 };

Remarks

Compare GeneratedTokenCount against this value to detect a page that still hit the cap. Lowering the cap speeds up conversion at the risk of missing tail content on long pages, and under SharedSlotPool a lower cap leaves more of the slot window to the image.