تاباكو

مراجع:

all_languages

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/all_languages')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 1926192
  • سمات :
{
    "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"
    }
}

بالعربية

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/af')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 307
  • سمات :
{
    "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"
    }
}

ع

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/ar')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 6446
  • سمات :
{
    "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"
    }
}

من الألف إلى الياء

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/az')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 624
  • سمات :
{
    "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"
    }
}

يكون

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/be')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 1512
  • سمات :
{
    "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"
    }
}

بير

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/ber')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 67484
  • سمات :
{
    "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"
    }
}

بغ

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/bg')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 6324
  • سمات :
{
    "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"
    }
}

مليار

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/bn')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 1440
  • سمات :
{
    "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"
    }
}

ر

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/br')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 2536
  • سمات :
{
    "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"
    }
}

كاليفورنيا

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/ca')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 518
  • سمات :
{
    "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

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/cbk')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 262
  • سمات :
{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

cmn

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/cmn')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 12549
  • سمات :
{
    "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"
    }
}

خدمات العملاء

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/cs')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 6659
  • سمات :
{
    "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"
    }
}

دا

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/da')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 11220
  • سمات :
{
    "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"
    }
}

دي

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/de')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 125091
  • سمات :
{
    "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"
    }
}

ش

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/el')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 10072
  • سمات :
{
    "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"
    }
}

أون

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/en')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 158053
  • سمات :
{
    "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

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/eo')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 207105
  • سمات :
{
    "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"
    }
}

وفاق

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/es')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 85064
  • سمات :
{
    "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"
    }
}

وآخرون

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/et')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 241
  • سمات :
{
    "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"
    }
}

الاتحاد الأوروبي

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/eu')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 573
  • سمات :
{
    "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"
    }
}

فاي

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/fi')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 31753
  • سمات :
{
    "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"
    }
}

الاب

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/fr')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 116733
  • سمات :
{
    "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

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/gl')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 351
  • سمات :
{
    "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"
    }
}

go.go

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/gos')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 279
  • سمات :
{
    "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"
    }
}

هو

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/he')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 68350
  • سمات :
{
    "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"
    }
}

أهلاً

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/hi')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 1913
  • سمات :
{
    "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"
    }
}

ساعة

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/hr')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 505
  • سمات :
{
    "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"
    }
}

هو

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/hu')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 67964
  • سمات :
{
    "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"
    }
}

مرحبا

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/hy')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 603
  • سمات :
{
    "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"
    }
}

Ia

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/ia')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 2548
  • سمات :
{
    "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"
    }
}

بطاقة تعريف

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/id')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 1602
  • سمات :
{
    "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"
    }
}

أي

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/ie')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 488
  • سمات :
{
    "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

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/io')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 480
  • سمات :
{
    "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"
    }
}

يكون

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/is')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 1641
  • سمات :
{
    "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"
    }
}

هو - هي

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/it')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 198919
  • سمات :
{
    "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"
    }
}

جا

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/ja')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 44267
  • سمات :
{
    "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

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/jbo')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 2704
  • سمات :
{
    "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"
    }
}

كاب

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/kab')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 15944
  • سمات :
{
    "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"
    }
}

كو

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/ko')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 503
  • سمات :
{
    "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"
    }
}

كيلووات

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/kw')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 1328
  • سمات :
{
    "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"
    }
}

لا

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/la')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 6889
  • سمات :
{
    "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

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/lfn')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 2313
  • سمات :
{
    "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"
    }
}

لتر

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/lt')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 8042
  • سمات :
{
    "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"
    }
}

عضو الكنيست

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/mk')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 14678
  • سمات :
{
    "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"
    }
}

السيد

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/mr')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 16413
  • سمات :
{
    "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"
    }
}

ملحوظة:

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/nb')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 1094
  • سمات :
{
    "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

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/nds')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 2633
  • سمات :
{
    "paraphrase_set_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "paraphrase": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "lists": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "tags": {
        "feature": {
            "dtype": "string",
            "id": null,
            "_type": "Value"
        },
        "length": -1,
        "id": null,
        "_type": "Sequence"
    },
    "language": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}

nl

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/nl')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 23561
  • سمات :
{
    "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

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/orv')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 471
  • سمات :
{
    "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"
    }
}

أوتا

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/ota')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 486
  • سمات :
{
    "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"
    }
}

بيس

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/pes')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 4285
  • سمات :
{
    "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"
    }
}

رر

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/pl')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 22391
  • سمات :
{
    "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"
    }
}

نقطة

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/pt')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 78430
  • سمات :
{
    "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"
    }
}

آر إن

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/rn')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 648
  • سمات :
{
    "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"
    }
}

ريال عماني

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/ro')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 2092
  • سمات :
{
    "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"
    }
}

رو

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/ru')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 251263
  • سمات :
{
    "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

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/sl')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 706
  • سمات :
{
    "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"
    }
}

ريال

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/sr')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 8175
  • سمات :
{
    "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"
    }
}

القديس

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/sv')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 7005
  • سمات :
{
    "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"
    }
}

تك

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/tk')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 1165
  • سمات :
{
    "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"
    }
}

ليرة تركية

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/tl')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 1017
  • سمات :
{
    "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

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/tlh')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 2804
  • سمات :
{
    "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"
    }
}

توكي

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/toki')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 3738
  • سمات :
{
    "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"
    }
}

آر

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/tr')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 142088
  • سمات :
{
    "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"
    }
}

تي تي

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/tt')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 2398
  • سمات :
{
    "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

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/ug')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 1183
  • سمات :
{
    "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"
    }
}

المملكة المتحدة

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/uk')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 54431
  • سمات :
{
    "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"
    }
}

أور

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/ur')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 252
  • سمات :
{
    "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"
    }
}

سادسا

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/vi')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 962
  • سمات :
{
    "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"
    }
}

صوت

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/vo')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 328
  • سمات :
{
    "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"
    }
}

حرب

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/war')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 327
  • سمات :
{
    "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"
    }
}

ووو

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/wuu')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 408
  • سمات :
{
    "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"
    }
}

يو

استخدم الأمر التالي لتحميل مجموعة البيانات هذه في TFDS:

ds = tfds.load('huggingface:tapaco/yue')
  • وصف :
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.
  • الترخيص : إسناد المشاع الإبداعي 2.0 عام
  • الإصدار : 1.0.0
  • الإنشقاقات :
ينقسم أمثلة
'train' 561
  • سمات :
{
    "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"
    }
}