Anvia

@anvia/transformers ​

Transformers.js provides local dense embeddings through Hugging Face feature-extraction pipelines. It is a compact choice for development and lightweight retrieval workflows that should not call a remote embedding API.

SupportFirst-party
Version1.1.1
RuntimeESM, runtimes supported by Transformers.js
PeerMatching @anvia/core stable release

Install ​

bash
pnpm add @anvia/transformers @anvia/core

@huggingface/transformers is a runtime dependency of the adapter. Model assets are resolved by Transformers.js when the model is first initialized.

Create local embeddings ​

ts
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 30 minutes.',
  'Enterprise customers receive priority support.',
])

console.log(vectors[0].vector)

Capabilities ​

CapabilitySupport
Dense text embeddingsYes
Configurable modelAny compatible feature-extraction model
Poolingmean or cls
NormalizationConfigurable, enabled by default
Sparse embeddingsNo
Remote provider callNo; inference runs through Transformers.js

Common patterns ​

Configure the feature extractor ​

ts
const embeddings = await loadTransformersEmbeddingModel({
  modelId: 'Xenova/all-MiniLM-L6-v2',
  pooling: 'mean',
  normalize: true,
  maxBatchSize: 16,
})

Inject a pipeline ​

ts
import { adaptTransformersEmbeddingModel } from '@anvia/transformers'

const embeddings = adaptTransformersEmbeddingModel({
  runtime: featureExtractionPipeline,
  modelId: 'my-local-model',
  pooling: 'cls',
  normalize: true,
})

The adapter is useful when the application owns model loading or tests need a deterministic pipeline. For ordinary use, prefer the async loader.

Use the same model, pooling, normalization, and preprocessing for ingestion and query embeddings. A change to any of them usually requires rebuilding the vector index.

Compatibility ​

@anvia/transformers is ESM and uses @huggingface/transformers. The chosen model must support the feature-extraction task and return one numeric vector per input. Runtime support, model caching, and hardware acceleration follow Transformers.js.

Continue ​

Built for Anvia.