hate_speech_offensive

References:

• Code
• Huggingface

Use the following command to load this dataset in TFDS:

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

• Description:

An annotated dataset for hate speech and offensive language detection on tweets.

• License: MIT
• Version: 1.0.0
• Splits:

• Features:

{
    "count": {
        "dtype": "int64",
        "id": null,
        "_type": "Value"
    },
    "hate_speech_count": {
        "dtype": "int64",
        "id": null,
        "_type": "Value"
    },
    "offensive_language_count": {
        "dtype": "int64",
        "id": null,
        "_type": "Value"
    },
    "neither_count": {
        "dtype": "int64",
        "id": null,
        "_type": "Value"
    },
    "class": {
        "num_classes": 3,
        "names": [
            "hate speech",
            "offensive language",
            "neither"
        ],
        "names_file": null,
        "id": null,
        "_type": "ClassLabel"
    },
    "tweet": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}