NSMC

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

This is a movie review dataset in the Korean language. Reviews were scraped from Naver movies. The dataset construction is based on the method noted in Large movie review dataset from Maas et al., 2011.

{
    "id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "document": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "label": {
        "num_classes": 2,
        "names": [
            "negative",
            "positive"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    }
}