{
  "labels": [
    "harmful_behaviour",
    "hate_speech",
    "neutral",
    "offensive_language"
  ]
}