{
  "model_type": "sentiment-analysis",
  "labels": ["negative", "neutral", "positive"],
  "framework": "pytorch"
}