Type Alias: LoadModelOptions

LoadModelOptions = object

Defined in: core/model.ts:83

Configuration options for loading an ExecuTorch model into native memory.

Properties

eagerLoadMethods?

readonly optional eagerLoadMethods: boolean

Defined in: core/model.ts:95

Whether to eagerly load and compile all model methods and delegate subgraphs during initialization.

When true (the default), all exported methods are fully loaded and backend delegates (such as CoreML or Vulkan) are compiled into memory upfront, guaranteeing instantaneous first-inference latency.

Set to false to lazily load and compile methods on their first execution.

Default

true