{
  "model_type": "text-classification",
  "labels": ["positive", "negative", "neutral"]
}