mteb/biblenlp-corpus

This dataset pre-computes all English-centric directions from bible-nlp/biblenlp-corpus, and as a result loading is significantly faster.

Loading example:

>>> from datasets import load_dataset
>>> dataset = load_dataset("davidstap/biblenlp-corpus-mmteb", "eng-arb", trust_remote_code=True)
>>> dataset
DatasetDict({
    train: Dataset({
        features: ['eng', 'arb'],
        num_rows: 28723
    })
    validation: Dataset({
        features: ['eng', 'arb'],
        num_rows: 1578
    })
    test: Dataset({
        features: ['eng', 'arb'],
        num_rows: 1551
    })
})
>>> 

Note that in all possible configurations, eng comes before the other language.