Documentation / @agentick/apple / appleEmbedding

Function: appleEmbedding()

appleEmbedding(config?): EmbeddingModel

Defined in: embedding.ts:44

Create an Apple on-device embedding model.

Uses NLContextualEmbedding under the hood — mean-pooled token vectors producing a single vector per input text.

Parameters

config?

AppleEmbeddingConfig = {}

Returns

EmbeddingModel

Example

const embedder = appleEmbedding({ script: "latin" });
const { embeddings } = await embedder.embed({ input: ["machine learning"] });
console.log(embeddings[0].length); // 512