odio_discurso_filipino

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

Contains 10k tweets (training set) that are labeled as hate speech or non-hate speech. Released with 4,232 validation and 4,232 testing samples. Collected during the 2016 Philippine Presidential Elections.

{
    "text": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "label": {
        "num_classes": 2,
        "names": [
            "0",
            "1"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    }
}