TFDS ora supporta il formato Croissant 🥐 ! Leggi la documentazione per saperne di più.

Questa pagina è stata tradotta dall'API Cloud Translation.

tapaco

Riferimenti:

tutte_lingue

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	1926192

Caratteristiche :

{
    "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"
    }
}

ef

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	307

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	6446

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	624

Caratteristiche :

{
    "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"
    }
}

Essere

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	1512

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	67484

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	6324

Caratteristiche :

{
    "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"
    }
}

miliardo

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	1440

Caratteristiche :

{
    "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"
    }
}

fratello

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	2536

Caratteristiche :

{
    "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"
    }
}

ca

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	518

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	262

Caratteristiche :

{
    "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"
    }
}

cmq

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	12549

Caratteristiche :

{
    "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.s

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	6659

Caratteristiche :

{
    "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"
    }
}

da

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	11220

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	125091

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	10072

Caratteristiche :

{
    "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"
    }
}

en

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	158053

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	207105

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	85064

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	241

Caratteristiche :

{
    "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"
    }
}

Unione Europea

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	573

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	31753

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	116733

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	351

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	279

Caratteristiche :

{
    "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"
    }
}

Lui

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	68350

Caratteristiche :

{
    "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"
    }
}

CIAO

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	1913

Caratteristiche :

{
    "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"
    }
}

ora

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	505

Caratteristiche :

{
    "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"
    }
}

eh

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	67964

Caratteristiche :

{
    "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"
    }
}

ciao

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	603

Caratteristiche :

{
    "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"
    }
}

e.a

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	2548

Caratteristiche :

{
    "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"
    }
}

id

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	1602

Caratteristiche :

{
    "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"
    }
}

cioè

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	488

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	480

Caratteristiche :

{
    "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"
    }
}

È

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	1641

Caratteristiche :

{
    "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"
    }
}

Esso

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	198919

Caratteristiche :

{
    "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"
    }
}

sì

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	44267

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	2704

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	15944

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	503

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	1328

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	6889

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	2313

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	8042

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	14678

Caratteristiche :

{
    "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"
    }
}

Sig

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	16413

Caratteristiche :

{
    "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"
    }
}

n.b

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	1094

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	2633

Caratteristiche :

{
    "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"
    }
}

n.l

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	23561

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	471

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	486

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	4285

Caratteristiche :

{
    "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"
    }
}

per favore

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	22391

Caratteristiche :

{
    "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"
    }
}

punto

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	78430

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	648

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	2092

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	251263

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	706

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	8175

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	7005

Caratteristiche :

{
    "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"
    }
}

ok

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	1165

Caratteristiche :

{
    "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"
    }
}

tl

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	1017

Caratteristiche :

{
    "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"
    }
}

tlh

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	2804

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	3738

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	142088

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	2398

Caratteristiche :

{
    "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"
    }
}

ug

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	1183

Caratteristiche :

{
    "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"
    }
}

Regno Unito

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	54431

Caratteristiche :

{
    "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 tuo

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	252

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	962

Caratteristiche :

{
    "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"
    }
}

voi

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	328

Caratteristiche :

{
    "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"
    }
}

guerra

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	327

Caratteristiche :

{
    "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

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	408

Caratteristiche :

{
    "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"
    }
}

sì

Utilizzare il comando seguente per caricare questo set di dati in TFDS:

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

Descrizione :

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.

Licenza : Creative Commons Attribuzione 2.0 Generica
Versione : 1.0.0
Divide :

Diviso	Esempi
`'train'`	561

Caratteristiche :

{
    "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"
    }
}