VisionConfig class

Configuration for initializing vision inference

Vision context is separate from text context - developers control when the ~540MB vision models load into memory.

Constructors

VisionConfig({required String modelPath, required String mmprojPath, int numThreads = 4, int contextSize = 0, bool useGpu = true, int maxMemoryMb = 1536})

const

Properties

contextSize → int

Context size in tokens (0 = auto based on model)

final

hashCode → int

The hash code for this object.

no setter, inherited

maxMemoryMb → int

Maximum memory budget in MB

final

mmprojPath → String

Path to the mmproj GGUF file (multimodal projector)

final

modelPath → String

Path to the VLM GGUF model file

final

numThreads → int

Number of threads for inference (defaults to 4)

final

runtimeType → Type

A representation of the runtime type of the object.

no setter, inherited

useGpu → bool

Enable GPU acceleration (defaults to true)

final

Methods

noSuchMethod(Invocation invocation) → dynamic

Invoked when a nonexistent method or property is accessed.

inherited

toJson() → Map<String, dynamic>

toString() → String

A string representation of this object.

override

Operators

operator ==(Object other) → bool

The equality operator.

inherited