Get started

Install the adapter with Core:

pnpm add @anvia/core @anvia/transformers

Initialize a local Transformers.js feature-extraction pipeline:

import { DEFAULT_TRANSFORMERS_EMBEDDING_MODEL, loadTransformersEmbeddingModel } from '@anvia/transformers'

const embeddings = await loadTransformersEmbeddingModel({ modelId: DEFAULT_TRANSFORMERS_EMBEDDING_MODEL })
const vectors = await embeddings.embedTexts([
  'Password reset links expire after thirty minutes.',
  'Enterprise customers receive priority support.',
])

The default is Xenova/all-MiniLM-L6-v2 with mean pooling, normalization, and a batch-size metadata value of 16.

Use with a vector store

import { embedDocuments } from '@anvia/core/embeddings';
import { InMemoryVectorStore, retrieveDocuments } from '@anvia/core/vector-store';
const { documents: embedded } = await embedDocuments({
    model: embeddings,
    documents: documents,
    id: (document) => document.id,
    content: (document) => document.text
});
const store = InMemoryVectorStore.fromDocuments({ documents: embedded });
const results = await retrieveDocuments({
    store,
    model: embeddings,
    query: 'priority support',
    topK: 5
});

Before production

• Warm model loading and cache assets.
• Verify the selected model supports feature-extraction.
• Keep model, pooling, and normalization identical for indexing and querying.
• Rebuild the index after any of those values changes.
• Measure local CPU, memory, startup, and request latency.
• Add application concurrency limits where needed.