<!--
{
  "availability" : [
    "iOS: 17.0.0 -",
    "iPadOS: 17.0.0 -",
    "macCatalyst: 17.0.0 -",
    "macOS: 14.0.0 -",
    "visionOS: 1.0.0 -"
  ],
  "documentType" : "symbol",
  "framework" : "CreateML",
  "identifier" : "/documentation/CreateML/MLTextClassifier/FeatureExtractorType/bertEmbedding",
  "metadataVersion" : "0.1.0",
  "role" : "Case",
  "symbol" : {
    "kind" : "Case",
    "modules" : [
      "Create ML"
    ],
    "preciseIdentifier" : "s:8CreateML16MLTextClassifierV20FeatureExtractorTypeO13bertEmbeddingyA2EmF"
  },
  "title" : "MLTextClassifier.FeatureExtractorType.bertEmbedding"
}
-->

# MLTextClassifier.FeatureExtractorType.bertEmbedding

A feature extractor that provides BERT contextual word embeddings.

```
case bertEmbedding
```

## Discussion

The embeddings consider the context from left-to-right and right-to-left simultaneously.

BERT embedding requires certain downloadable assets to be present on device at training
time. Training will throw an error if the specified language is unavailable at runtime. Asset downloads are
managed in the background automatically by the OS when a new language is configured in device
settings, such as when adding a new keyboard language or changing the preferred language.

---

Copyright &copy; 2026 Apple Inc. All rights reserved. | [Terms of Use](https://www.apple.com/legal/internet-services/terms/site.html) | [Privacy Policy](https://www.apple.com/privacy/privacy-policy)