Interface: ModelRunner

April 26, 2025 · View on GitHub

Abstraction that allows for running a generated model on the JS thread or asynchronously (e.g. in a Web Worker), depending on the implementation.

Methods

createOutputs(): Outputs

Create an Outputs instance that is sized to accommodate the output variable data stored by the model.

A new Outputs instance.

runModel(inputs, outputs, options?): Promise<Outputs>

Run the model.

Name	Type	Description
`inputs`	`number`[] \| `InputValue`[]	The model input values (must be in the same order as in the spec file).
`outputs`	`Outputs`	The structure into which the model outputs will be stored.
`options?`	`RunModelOptions`	Additional options that influence the model run.

Promise<Outputs>

A promise that resolves with the outputs when the model run is complete.

terminate(): Promise<void>

Terminate the runner by releasing underlying resources (e.g., the worker thread or Wasm module/buffers).

Promise<void>