Khinh khí cầu

Người giới thiệu:

đảo_phụ

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/adjunct_island')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/anaphor_gender_agreement')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/anaphor_number_agreement')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

animate_subject_passive

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/animate_subject_passive')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

animate_subject_trans

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/animate_subject_trans')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

nguyên nhân

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/causative')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

phức hợp_NP_island

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/complex_NP_island')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

tọa độ_structure_constraint_complex_left_branch

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/coordinate_structure_constraint_complex_left_branch')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

tọa độ_structure_constraint_object_extraction

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/coordinate_structure_constraint_object_extraction')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

yếu tố xác định_danh từ_thỏa thuận_1

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/determiner_noun_agreement_1')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

yếu tố quyết định_danh từ_thỏa thuận_2

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/determiner_noun_agreement_2')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

bộ xác định_danh từ_agreement_irregular_1

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/determiner_noun_agreement_irregular_1')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

bộ xác định_danh từ_agreement_irregular_2

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/determiner_noun_agreement_irregular_2')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

bộ xác định_danh từ_agreement_with_adj_2

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/determiner_noun_agreement_with_adj_2')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

bộ xác định_danh từ_agreement_with_adj_irregular_1

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/determiner_noun_agreement_with_adj_irregular_1')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

bộ xác định_danh từ_agreement_with_adj_irregular_2

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/determiner_noun_agreement_with_adj_irregular_2')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

từ xác định_danh từ_thỏa thuận_với_tính từ_1

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/determiner_noun_agreement_with_adjective_1')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

kẻ gây phân tâm_thỏa thuận_quan hệ_danh từ

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/distractor_agreement_relational_noun')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

người phân tâm_thỏa thuận_relative_clause

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/distractor_agreement_relative_clause')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/drop_argument')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

dấu chấm lửng_n_bar_1

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/ellipsis_n_bar_1')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

dấu chấm lửng_n_bar_2

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/ellipsis_n_bar_2')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

hiện sinh_there_object_raise

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/existential_there_object_raising')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

tồn tại_there_quantifiers_1

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/existential_there_quantifiers_1')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

tồn tại_there_quantifiers_2

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/existential_there_quantifiers_2')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

hiện sinh_there_subject_raise

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/existential_there_subject_raising')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

expultive_it_object_raise

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/expletive_it_object_raising')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

gợi ý

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/inchoative')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

nội động từ

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/intransitive')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

bất quy tắc_quá khứ_phân từ_tính từ

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/irregular_past_participle_adjectives')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

bất quy tắc_quá khứ_phân từ_động từ

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/irregular_past_participle_verbs')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

không đều_số nhiều_chủ đề_động từ_thỏa thuận_1

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/irregular_plural_subject_verb_agreement_1')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

không đều_số nhiều_chủ đề_động từ_thỏa thuận_2

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/irregular_plural_subject_verb_agreement_2')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/left_branch_island_echo_question')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/left_branch_island_simple_question')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

ma trận_question_npi_liccensor_hiện tại

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/matrix_question_npi_licensor_present')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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_hiện tại_1

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/npi_present_1')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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_hiện tại_2

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/npi_present_2')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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_liccensor_hiện tại

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/only_npi_licensor_present')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

chỉ_npi_scope

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/only_npi_scope')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

thụ động_1

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/passive_1')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

thụ động_2

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/passive_2')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

nguyên tắc_A_c_command

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/principle_A_c_command')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

nguyên tắc_A_case_1

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/principle_A_case_1')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

nguyên tắc_A_case_2

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/principle_A_case_2')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

nguyên tắc_A_domain_1

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/principle_A_domain_1')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

nguyên tắc_A_domain_2

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/principle_A_domain_2')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

nguyên tắc_A_domain_3

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/principle_A_domain_3')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

nguyên tắc_A_xây dựng lại

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/principle_A_reconstruction')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

thường_plural_subject_verb_agreement_1

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/regular_plural_subject_verb_agreement_1')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

thường_plural_subject_verb_agreement_2

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/regular_plural_subject_verb_agreement_2')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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_liccensor_hiện tại

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/sentential_negation_npi_licensor_present')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/sentential_negation_npi_scope')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/sentential_subject_island')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

so sánh nhất_quantifiers_1

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/superlative_quantifiers_1')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

so sánh nhất_quantifiers_2

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/superlative_quantifiers_2')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

khó khăn_vs_raise_1

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/tough_vs_raising_1')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

khó khăn_vs_raise_2

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/tough_vs_raising_2')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

Bắc cầu

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/transitive')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}

đảo_wh

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/wh_island')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/wh_questions_object_gap')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/wh_questions_subject_gap')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/wh_questions_subject_gap_long_distance')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/wh_vs_that_no_gap')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/wh_vs_that_no_gap_long_distance')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/wh_vs_that_with_gap')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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

Sử dụng lệnh sau để tải tập dữ liệu này trong TFDS:

ds = tfds.load('huggingface:blimp/wh_vs_that_with_gap_long_distance')
  • Sự miêu tả :
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.
  • Giấy phép : Không có giấy phép được biết đến
  • Phiên bản : 0.1.0
  • Chia tách :
Tách ra Ví dụ
'train' 1000
  • Đặc trưng :
{
    "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"
    }
}