InferenceEngineProvider class abstract
A pluggable inference engine (MediaPipe .task, LiteRT-LM .litertlm, or a
third-party .onnx/.gguf engine).
Implemented in an engine package (e.g. flutter_gemma_mediapipe) and passed
to FlutterGemma.initialize via inferenceEngines:. Core selects an engine
by probing: the first registered engine (highest priority) whose
canHandle returns true for the active model spec wins. There is NO central
file-type map — a third-party engine self-selects with zero core changes.
Constructors
Properties
- hashCode → int
-
The hash code for this object.
no setterinherited
- name → String
-
Human-readable name for diagnostics / error messages (e.g. 'MediaPipe').
no setter
- priority → int
-
Selection precedence on overlap. Core engines use 0; a third party raises
this to take precedence for a spec both could handle.
no setter
- runtimeType → Type
-
A representation of the runtime type of the object.
no setterinherited
Methods
-
canHandle(
InferenceModelSpec spec) → bool -
Whether this engine can run
spec. Probed by the registry. -
createModel(
InferenceModelSpec spec, RuntimeConfig config) → Future< InferenceModel> -
Build a runtime InferenceModel for
spec+config. -
noSuchMethod(
Invocation invocation) → dynamic -
Invoked when a nonexistent method or property is accessed.
inherited
-
toString(
) → String -
A string representation of this object.
inherited
Operators
-
operator ==(
Object other) → bool -
The equality operator.
inherited