indonli

ds = tfds.load('huggingface:indonli/indonli')

IndoNLI is the first human-elicited Natural Language Inference (NLI) dataset for Indonesian.
  IndoNLI is annotated by both crowd workers and experts. The expert-annotated data is used exclusively as a test set.
  It is designed to provide a challenging test-bed for Indonesian NLI by explicitly incorporating various linguistic phenomena such as numerical reasoning, structural changes, idioms, or temporal and spatial reasoning.

{
    "premise": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "hypothesis": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "label": {
        "num_classes": 3,
        "names": [
            "entailment",
            "neutral",
            "contradiction"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    }
}