configs/datasets.json

{
  "official_test": {
    "dataset_id": "pipecat-ai/smart-turn-data-v3.2-test",
    "expected_parquet_shards": 10,
    "expected_rows": 31527,
    "policy": "sealed_until_model_preprocessing_controller_and_threshold_are_frozen",
    "revision": "0500378e8ed6d38e37b016e24d261e8e6c6a6859"
  },
  "train": {
    "dataset_id": "pipecat-ai/smart-turn-data-v3.2-train",
    "expected_parquet_shards": 83,
    "expected_rows": 270946,
    "license_note": "No conventional dataset license field was declared when this snapshot was inspected; raw audio is not redistributed.",
    "revision": "e564e2ac567f774d1880aa1db6ce97afb8c519b7"
  }
}