ro_sts

ds = tfds.load('huggingface:ro_sts/ro_sts')

The RO-STS (Romanian Semantic Textual Similarity) dataset contains 8628 pairs of sentences with their similarity score. It is a high-quality translation of the STS benchmark dataset.

{
    "score": {
        "dtype": "float32",
        "id": null,
        "_type": "Value"
    },
    "sentence1": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence2": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}