TFDS supporte désormais le format Croissant 🥐 ! Lisez la documentation pour en savoir plus.

Cette page a été traduite par l'API Cloud Translation.

tapaco

Les références:

toutes les langues

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/all_languages')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	1926192

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

un F

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/af')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	307

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

ar

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/ar')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	6446

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

az

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/az')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	624

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

être

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/be')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	1512

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

ber

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/ber')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	67484

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

bg

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/bg')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	6324

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

milliard

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/bn')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	1440

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

br

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/br')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	2536

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

Californie

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/ca')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	518

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

cbk

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/cbk')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	262

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

cmn

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/cmn')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	12549

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

cs

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/cs')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	6659

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

papa

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/da')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	11220

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

de

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/de')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	125091

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

el

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/el')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	10072

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

fr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/en')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	158053

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

eo

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/eo')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	207105

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

es

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/es')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	85064

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

et

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/et')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	241

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

UE

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/eu')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	573

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

Fi

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/fi')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	31753

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

fr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/fr')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	116733

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

gl

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/gl')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	351

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

va

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/gos')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	279

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

il

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/he')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	68350

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

Salut

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/hi')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	1913

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

heure

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/hr')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	505

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

hein

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/hu')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	67964

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

salut

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/hy')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	603

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

je

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/ia')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	2548

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

identifiant

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/id')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	1602

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

c'est à dire

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/ie')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	488

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

io

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/io')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	480

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

est

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/is')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	1641

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

il

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/it')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	198919

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

oui

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/ja')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	44267

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

jbo

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/jbo')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	2704

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

kab

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/kab')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	15944

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

ko

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/ko')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	503

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

kW

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/kw')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	1328

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

la

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/la')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	6889

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

lfn

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/lfn')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	2313

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

lt

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/lt')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	8042

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

mk

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/mk')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	14678

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

M.

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/mr')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	16413

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

nb

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/nb')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	1094

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

nds

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/nds')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	2633

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

nl

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/nl')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	23561

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

orv

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/orv')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	471

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

ota

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/ota')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	486

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

pes

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/pes')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	4285

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

PL

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/pl')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	22391

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

pt

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/pt')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	78430

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

rn

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/rn')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	648

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

ro

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/ro')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	2092

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

ru

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/ru')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	251263

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

sl

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/sl')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	706

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

sr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/sr')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	8175

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

sv

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/sv')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	7005

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

merci

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/tk')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	1165

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

télé

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/tl')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	1017

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

bah

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/tlh')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	2804

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

Toki

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/toki')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	3738

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

tr

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/tr')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	142088

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

tt

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/tt')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	2398

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

pouah

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/ug')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	1183

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

Royaume-Uni

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/uk')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	54431

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

ton

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/ur')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	252

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

vi

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/vi')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	962

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

vo

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/vo')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	328

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

guerre

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/war')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	327

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

wuu

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/wuu')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	408

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

yue

Utilisez la commande suivante pour charger cet ensemble de données dans TFDS :

ds = tfds.load('huggingface:tapaco/yue')

Description :

A freely available paraphrase corpus for 73 languages extracted from the Tatoeba database. Tatoeba is a
crowdsourcing project mainly geared towards language learners. Its aim is to provide example sentences and translations for particular
linguistic constructions and words. The paraphrase corpus is created by populating a graph with Tatoeba sentences and equivalence links
between sentences “meaning the same thing”. This graph is then traversed to extract sets of paraphrases. Several language-independent
filters and pruning steps are applied to remove uninteresting sentences. A manual evaluation performed on three languages shows
that between half and three quarters of inferred paraphrases are correct and that most remaining ones are either correct but trivial, or
near-paraphrases that neutralize a morphological distinction. The corpus contains a total of 1.9 million sentences, with 200 – 250 000
sentences per language. It covers a range of languages for which, to our knowledge, no other paraphrase dataset exists.

Licence : Creative Commons Attribution 2.0 Générique
Version : 1.0.0
Divisions :

Diviser	Exemples
`'train'`	561

Caractéristiques :

{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}