dirigeable

Références:

île_adjointe

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

ds = tfds.load('huggingface:blimp/adjunct_island')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

anaphor_gender_agreement

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

ds = tfds.load('huggingface:blimp/anaphor_gender_agreement')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

anaphor_number_agreement

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

ds = tfds.load('huggingface:blimp/anaphor_number_agreement')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

animer_sujet_passif

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

ds = tfds.load('huggingface:blimp/animate_subject_passive')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

animer_sujet_trans

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

ds = tfds.load('huggingface:blimp/animate_subject_trans')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

causal

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

ds = tfds.load('huggingface:blimp/causative')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

complexe_NP_island

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

ds = tfds.load('huggingface:blimp/complex_NP_island')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

ordinate_structure_constraint_complex_left_branch

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

ds = tfds.load('huggingface:blimp/coordinate_structure_constraint_complex_left_branch')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

coordonnée_structure_contrainte_object_extraction

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

ds = tfds.load('huggingface:blimp/coordinate_structure_constraint_object_extraction')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

determiner_noun_agreement_1

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

ds = tfds.load('huggingface:blimp/determiner_noun_agreement_1')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

determiner_noun_agreement_2

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

ds = tfds.load('huggingface:blimp/determiner_noun_agreement_2')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

determiner_noun_agreement_irregular_1

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

ds = tfds.load('huggingface:blimp/determiner_noun_agreement_irregular_1')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

determiner_noun_agreement_irregular_2

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

ds = tfds.load('huggingface:blimp/determiner_noun_agreement_irregular_2')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

determiner_noun_agreement_with_adj_2

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

ds = tfds.load('huggingface:blimp/determiner_noun_agreement_with_adj_2')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

determiner_noun_agreement_with_adj_irregular_1

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

ds = tfds.load('huggingface:blimp/determiner_noun_agreement_with_adj_irregular_1')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

determiner_noun_agreement_with_adj_irregular_2

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

ds = tfds.load('huggingface:blimp/determiner_noun_agreement_with_adj_irregular_2')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

determiner_noun_agreement_with_adjective_1

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

ds = tfds.load('huggingface:blimp/determiner_noun_agreement_with_adjective_1')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

distractor_agreement_relational_nom

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

ds = tfds.load('huggingface:blimp/distractor_agreement_relational_noun')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

distractor_agreement_relative_clause

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

ds = tfds.load('huggingface:blimp/distractor_agreement_relative_clause')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

drop_argument

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

ds = tfds.load('huggingface:blimp/drop_argument')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

points de suspension_n_bar_1

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

ds = tfds.load('huggingface:blimp/ellipsis_n_bar_1')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

points de suspension_n_bar_2

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

ds = tfds.load('huggingface:blimp/ellipsis_n_bar_2')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

existential_there_object_raising

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

ds = tfds.load('huggingface:blimp/existential_there_object_raising')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

existential_there_quantifiers_1

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

ds = tfds.load('huggingface:blimp/existential_there_quantifiers_1')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

existential_there_quantifiers_2

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

ds = tfds.load('huggingface:blimp/existential_there_quantifiers_2')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

existential_there_subject_raising

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

ds = tfds.load('huggingface:blimp/existential_there_subject_raising')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

expletive_it_object_raising

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

ds = tfds.load('huggingface:blimp/expletive_it_object_raising')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

inchoatif

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

ds = tfds.load('huggingface:blimp/inchoative')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

intransitif

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

ds = tfds.load('huggingface:blimp/intransitive')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

participe_passé_irrégulier_adjectifs

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

ds = tfds.load('huggingface:blimp/irregular_past_participle_adjectives')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

verbes_participes_passés_irréguliers

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

ds = tfds.load('huggingface:blimp/irregular_past_participle_verbs')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

irrégulier_pluriel_subject_verb_agreement_1

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

ds = tfds.load('huggingface:blimp/irregular_plural_subject_verb_agreement_1')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

irrégulier_pluriel_subject_verb_agreement_2

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

ds = tfds.load('huggingface:blimp/irregular_plural_subject_verb_agreement_2')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

left_branch_island_echo_question

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

ds = tfds.load('huggingface:blimp/left_branch_island_echo_question')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

left_branch_island_simple_question

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

ds = tfds.load('huggingface:blimp/left_branch_island_simple_question')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

matrix_question_npi_licensor_present

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

ds = tfds.load('huggingface:blimp/matrix_question_npi_licensor_present')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

npi_présent_1

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

ds = tfds.load('huggingface:blimp/npi_present_1')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

npi_présent_2

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

ds = tfds.load('huggingface:blimp/npi_present_2')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

only_npi_licensor_present

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

ds = tfds.load('huggingface:blimp/only_npi_licensor_present')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

only_npi_scope

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

ds = tfds.load('huggingface:blimp/only_npi_scope')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

passif_1

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

ds = tfds.load('huggingface:blimp/passive_1')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

passif_2

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

ds = tfds.load('huggingface:blimp/passive_2')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

principe_A_c_commande

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

ds = tfds.load('huggingface:blimp/principle_A_c_command')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

principe_A_cas_1

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

ds = tfds.load('huggingface:blimp/principle_A_case_1')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

principe_A_cas_2

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

ds = tfds.load('huggingface:blimp/principle_A_case_2')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

principe_A_domaine_1

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

ds = tfds.load('huggingface:blimp/principle_A_domain_1')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

principe_A_domaine_2

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

ds = tfds.load('huggingface:blimp/principle_A_domain_2')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

principe_A_domaine_3

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

ds = tfds.load('huggingface:blimp/principle_A_domain_3')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

principe_A_reconstruction

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

ds = tfds.load('huggingface:blimp/principle_A_reconstruction')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

regular_plural_subject_verb_agreement_1

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

ds = tfds.load('huggingface:blimp/regular_plural_subject_verb_agreement_1')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

regular_plural_subject_verb_agreement_2

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

ds = tfds.load('huggingface:blimp/regular_plural_subject_verb_agreement_2')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

sentential_negation_npi_licensor_present

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

ds = tfds.load('huggingface:blimp/sentential_negation_npi_licensor_present')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

sentential_negation_npi_scope

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

ds = tfds.load('huggingface:blimp/sentential_negation_npi_scope')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

sentential_subject_island

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

ds = tfds.load('huggingface:blimp/sentential_subject_island')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

quantificateurs_superlatifs_1

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

ds = tfds.load('huggingface:blimp/superlative_quantifiers_1')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

quantificateurs_superlatifs_2

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

ds = tfds.load('huggingface:blimp/superlative_quantifiers_2')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

dur_vs_raising_1

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

ds = tfds.load('huggingface:blimp/tough_vs_raising_1')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

dur_vs_raising_2

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

ds = tfds.load('huggingface:blimp/tough_vs_raising_2')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

transitif

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

ds = tfds.load('huggingface:blimp/transitive')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

wh_island

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

ds = tfds.load('huggingface:blimp/wh_island')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

wh_questions_object_gap

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

ds = tfds.load('huggingface:blimp/wh_questions_object_gap')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

wh_questions_subject_gap

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

ds = tfds.load('huggingface:blimp/wh_questions_subject_gap')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

wh_questions_subject_gap_long_distance

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

ds = tfds.load('huggingface:blimp/wh_questions_subject_gap_long_distance')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

wh_vs_that_no_gap

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

ds = tfds.load('huggingface:blimp/wh_vs_that_no_gap')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

wh_vs_that_no_gap_long_distance

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

ds = tfds.load('huggingface:blimp/wh_vs_that_no_gap_long_distance')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

wh_vs_that_with_gap

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

ds = tfds.load('huggingface:blimp/wh_vs_that_with_gap')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}

wh_vs_that_with_gap_long_distance

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

ds = tfds.load('huggingface:blimp/wh_vs_that_with_gap_long_distance')
  • Descriptif :
BLiMP is a challenge set for evaluating what language models (LMs) know about
major grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each
containing 1000 minimal pairs isolating specific contrasts in syntax,
morphology, or semantics. The data is automatically generated according to
expert-crafted grammars.
  • Licence : Aucune licence connue
  • Version : 0.1.0
  • Fractionnements :
Diviser Exemples
'train' 1000
  • Caractéristiques :
{
    "sentence_good": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "sentence_bad": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "field": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "linguistics_term": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "UID": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "simple_LM_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "one_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "two_prefix_method": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "lexically_identical": {
        "dtype": "bool",
        "id": null,
        "_type": "Value"
    },
    "pair_id": {
        "dtype": "int32",
        "id": null,
        "_type": "Value"
    }
}