OnnxAdapter class

AIModelAdapter implementation backed by the ONNX Runtime.

Basic usage

final bytes = await File('model.onnx').readAsBytes();
final adapter = OnnxAdapter(modelBytes: bytes, modelId: 'my-model');

final report = await SateAI.stress(
  model: adapter,
  injectors: [MemoryPressureInjector(limitMb: 200)],
);
adapter.dispose();

The adapter accepts raw model bytes instead of a file path so it works uniformly with Flutter asset bundles, network fetches, and local storage without coupling to a specific I/O API.

Implemented types

• AIModelAdapter

Constructors

OnnxAdapter({required Uint8List modelBytes, required String modelId, OrtSessionFactory? sessionFactory})

Constructs an OnnxAdapter.

Properties

currentMemoryMB → double

Estimated memory currently consumed by the model in megabytes.

no setter, override

hashCode → int

The hash code for this object.

no setter, inherited

isDegraded → bool

Whether the model is in a degraded / unreliable state.

no setter, override

modelId → String

Stable identifier for this model (used in reports).

no setter, override

runtimeType → Type

A representation of the runtime type of the object.

no setter, inherited

Methods

dispose() → void

Releases all ONNX Runtime resources held by this adapter.

isHealthy() → Future<bool>

Returns true when the adapter can safely accept inference calls.

override

noSuchMethod(Invocation invocation) → dynamic

Invoked when a nonexistent method or property is accessed.

inherited

reset() → Future<void>

Resets the adapter to a healthy, non-degraded state.

override

runInference(AIInput input) → Future<AIOutput>

Runs a single ONNX inference call.

override

simulateMemoryPressure(int mb) → Future<void>

Increases the simulated memory footprint by mb megabytes.

override

toString() → String

A string representation of this object.

inherited

Operators

operator ==(Object other) → bool

The equality operator.

inherited