refrescado

ds = tfds.load('huggingface:refresd')

The Rationalized English-French Semantic Divergences (REFreSD) dataset consists of 1,039 
English-French sentence-pairs annotated with sentence-level divergence judgments and token-level 
rationales. For any questions, write to ebriakou@cs.umd.edu.

{
    "sentence_en": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_fr": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "label": {
        "num_classes": 2,
        "names": [
            "divergent",
            "equivalent"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    },
    "all_labels": {
        "num_classes": 3,
        "names": [
            "unrelated",
            "some_meaning_difference",
            "no_meaning_difference"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    },
    "rationale_en": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "rationale_fr": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}