Type Alias: Model

Model = object

Defined in: core/model.ts:38

A compiled, ready-to-run ExecuTorch model loaded into native memory.

A Model exposes the raw ExecuTorch execution interface. It is intentionally low-level and domain-agnostic; higher-level task pipelines build on top of this interface.

Obtain a Model instance via the loadModel function. When the model is no longer needed, call Model.dispose to release native memory.

Properties

backends

readonly backends: Record<string, readonly string[]>

Defined in: core/model.ts:44

ExecuTorch backends used by a given method.

---

path

readonly path: string

Defined in: core/model.ts:40

The local filesystem path of the .pte model file.

---

schema

readonly schema: ModelSpec<ConcreteDim>

Defined in: core/model.ts:42

The exported schema of this model.

Methods

dispose()

dispose(): void

Defined in: core/model.ts:70

Releases the native ExecuTorch model and frees all associated resources.

After calling dispose, this model instance must not be used again.

Returns

void

---

execute()

execute(methodName, inputs, outputTensors): ModelOutput[]

Defined in: core/model.ts:63

Executes a named model method synchronously.

Inputs are provided in the same order as the method's input slots. Output tensors must be pre-allocated and passed in outputTensors; the runtime writes results into them in-place and also returns them as the function result.

Parameters

methodName

string

The exported method to run (e.g. 'forward').

inputs

ModelInput[]

The list of input values to pass to the method, in order.

outputTensors

Tensor[]

Pre-allocated tensors for the method to write outputs into, in order.

Returns

ModelOutput[]

Throws

Thrown with code EXECUTION_FAILED if inference fails, SCHEMA_MISMATCH if runtime constraints fail, RESOURCE_BUSY if the model or a tensor is in use, RESOURCE_DISPOSED if disposed, or INVALID_ARGUMENT if inputs or output placeholders are invalid.