Nutrient Java SDK/io.nutrient.sdk.settings/LocalVlmSettings

LocalVlmSettings

Settings for LocalVlm. Values fall back through three levels: document → SDK → built-in default. Writes target the document only when set on a document's settings, otherwise the SDK globally when set on SdkSettings.

Functions

getBatchSize

public final int getBatchSize()

Number of zone images to include per VLM call.

getClassificationStrategy

public final VlmClassificationStrategy getClassificationStrategy()

Strategy for how zones are sent to the VLM for classification.

getContextTokens

public final int getContextTokens()

Context size in tokens; 0 uses the engine default (4096).

getGpuLayers

public final int getGpuLayers()

Model layers to offload to GPU; 0 is pure CPU.

getImageMinTokens

public final int getImageMinTokens()

Minimum image tokens for dynamic-resolution vision models; 0 uses the model default.

getMaxConcurrency

public final int getMaxConcurrency()

Maximum number of concurrent VLM calls.

getMaxTokens

public final int getMaxTokens()

Maximum tokens in response (-1 = engine default)

getMmprojPath

public final String getMmprojPath()

Optional filesystem path to the matching multimodal projector.

getModelPath

public final String getModelPath()

Optional filesystem path to the text model.

getSendFullPageReference

public final boolean getSendFullPageReference()

Whether to send the full page image as reference context alongside cropped zones.

getSystemPrompt

public final String getSystemPrompt()

Optional custom system prompt for zone classification.

getTemperature

public final double getTemperature()

Temperature for response generation (0.0 = deterministic, 1.0 = creative).

getThreads

public final int getThreads()

Inference threads; 0 uses the engine default.

setBatchSize

public final void setBatchSize(int value)

Number of zone images to include per VLM call.

setClassificationStrategy

public final void setClassificationStrategy(VlmClassificationStrategy value)

Strategy for how zones are sent to the VLM for classification.

setContextTokens

public final void setContextTokens(int value)

Context size in tokens; 0 uses the engine default (4096).

setGpuLayers

public final void setGpuLayers(int value)

Model layers to offload to GPU; 0 is pure CPU.

setImageMinTokens

public final void setImageMinTokens(int value)

Minimum image tokens for dynamic-resolution vision models; 0 uses the model default.

setMaxConcurrency

public final void setMaxConcurrency(int value)

Maximum number of concurrent VLM calls.

setMaxTokens

public final void setMaxTokens(int value)

Maximum tokens in response (-1 = engine default)

setMmprojPath

public final void setMmprojPath(String value)

Optional filesystem path to the matching multimodal projector.

setModelPath

public final void setModelPath(String value)

Optional filesystem path to the text model.

setSendFullPageReference

public final void setSendFullPageReference(boolean value)

Whether to send the full page image as reference context alongside cropped zones.

setSystemPrompt

public final void setSystemPrompt(String value)

Optional custom system prompt for zone classification.

setTemperature

public final void setTemperature(double value)

Temperature for response generation (0.0 = deterministic, 1.0 = creative).

setThreads

public final void setThreads(int value)

Inference threads; 0 uses the engine default.