wiqa

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

The WIQA dataset V1 has 39705 questions containing a perturbation and a possible effect in the context of a paragraph. 
The dataset is split into 29808 train questions, 6894 dev questions and 3003 test questions.

{
    "question_stem": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "question_para_step": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "answer_label": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "answer_label_as_choice": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "choices": {
        "feature": {
            "text": {
                "dtype": "string",
                "id": null,
                "_type": "Value"
            },
            "label": {
                "dtype": "string",
                "id": null,
                "_type": "Value"
            }
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "metadata_question_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "metadata_graph_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "metadata_para_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "metadata_question_type": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "metadata_path_len": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}