{
  "labels": [
    "computer vision",
    "natural language processing",
    "something else"
  ],
  "normalize_embeddings": false
}