Class ExecuTorchEmbeddings

ExecuTorch-based implementation of Embeddings for React Native.

Implements

Embeddings

Index

Constructors

Methods

embed load unload

Constructors

constructor

new ExecuTorchEmbeddings(
params: ExecuTorchEmbeddingsParams,
): ExecuTorchEmbeddings
Creates a new ExecuTorch embeddings instance.
Parameters
- params: ExecuTorchEmbeddingsParams
  Parameters for the instance.
  - modelPath
    Path or URL of the embedding model.
  - tokenizerPath
    Path or URL of the tokenizer.
  - defaultPrompt
    Prompt prepended to inputs without a more specific prompt.
  - documentPrompt
    Prompt prepended to documents being indexed.
  - queryPrompt
    Prompt prepended to search queries.
  - onDownloadProgress
    Download progress callback (0-1).
Returns ExecuTorchEmbeddings

Methods

embed

embed(text: string, options?: EmbedOptions): Promise<number[]>
Generates an embedding vector for the given text.
Parameters
- text: string
  Input string to embed.
- Optionaloptions: EmbedOptions
  Whether text is a document or a query, which selects the prompt.
Returns Promise<number[]>
Promise that resolves to the embedding vector.

Implementation of Embeddings.embed

load

load(): Promise<ExecuTorchEmbeddings>
Downloads (if needed) and loads the model and tokenizer via react-native-executorch. Overlapping calls share one load. An unload issued while the load is in flight wins unless load is called again before it settles.

Returns Promise<ExecuTorchEmbeddings>
Promise that resolves to the same instance.

Implementation of Embeddings.load

unload

unload(): Promise<void>
Unloads the underlying model and releases its native resources. A load still in flight is awaited first so that nothing is left behind.

Returns Promise<void>
Implementation of Embeddings.unload