{ "adjunct_island": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "adjunct_island", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 165894, "num_examples": 1000, "dataset_name": null } }, "download_size": 62231, "dataset_size": 165894, "size_in_bytes": 228125 }, "anaphor_gender_agreement": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "anaphor_gender_agreement", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 130918, "num_examples": 1000, "dataset_name": null } }, "download_size": 39201, "dataset_size": 130918, "size_in_bytes": 170119 }, "anaphor_number_agreement": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "anaphor_number_agreement", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 139879, "num_examples": 1000, "dataset_name": null } }, "download_size": 41547, "dataset_size": 139879, "size_in_bytes": 181426 }, "animate_subject_passive": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "animate_subject_passive", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 144423, "num_examples": 1000, "dataset_name": null } }, "download_size": 47282, "dataset_size": 144423, "size_in_bytes": 191705 }, "animate_subject_trans": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "animate_subject_trans", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 127798, "num_examples": 1000, "dataset_name": null } }, "download_size": 49651, "dataset_size": 127798, "size_in_bytes": 177449 }, "causative": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "causative", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 122772, "num_examples": 1000, "dataset_name": null } }, "download_size": 48963, "dataset_size": 122772, "size_in_bytes": 171735 }, "complex_NP_island": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "complex_NP_island", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 198972, "num_examples": 1000, "dataset_name": null } }, "download_size": 78211, "dataset_size": 198972, "size_in_bytes": 277183 }, "coordinate_structure_constraint_complex_left_branch": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "coordinate_structure_constraint_complex_left_branch", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 210912, "num_examples": 1000, "dataset_name": null } }, "download_size": 67908, "dataset_size": 210912, "size_in_bytes": 278820 }, "coordinate_structure_constraint_object_extraction": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "coordinate_structure_constraint_object_extraction", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 171655, "num_examples": 1000, "dataset_name": null } }, "download_size": 51584, "dataset_size": 171655, "size_in_bytes": 223239 }, "determiner_noun_agreement_1": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "determiner_noun_agreement_1", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 156120, "num_examples": 1000, "dataset_name": null } }, "download_size": 49893, "dataset_size": 156120, "size_in_bytes": 206013 }, "determiner_noun_agreement_2": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "determiner_noun_agreement_2", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 156204, "num_examples": 1000, "dataset_name": null } }, "download_size": 49527, "dataset_size": 156204, "size_in_bytes": 205731 }, "determiner_noun_agreement_irregular_1": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "determiner_noun_agreement_irregular_1", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 164473, "num_examples": 1000, "dataset_name": null } }, "download_size": 47274, "dataset_size": 164473, "size_in_bytes": 211747 }, "determiner_noun_agreement_irregular_2": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "determiner_noun_agreement_irregular_2", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 161074, "num_examples": 1000, "dataset_name": null } }, "download_size": 47422, "dataset_size": 161074, "size_in_bytes": 208496 }, "determiner_noun_agreement_with_adj_2": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "determiner_noun_agreement_with_adj_2", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 179666, "num_examples": 1000, "dataset_name": null } }, "download_size": 56346, "dataset_size": 179666, "size_in_bytes": 236012 }, "determiner_noun_agreement_with_adj_irregular_1": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "determiner_noun_agreement_with_adj_irregular_1", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 184529, "num_examples": 1000, "dataset_name": null } }, "download_size": 54405, "dataset_size": 184529, "size_in_bytes": 238934 }, "determiner_noun_agreement_with_adj_irregular_2": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "determiner_noun_agreement_with_adj_irregular_2", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 184396, "num_examples": 1000, "dataset_name": null } }, "download_size": 54064, "dataset_size": 184396, "size_in_bytes": 238460 }, "determiner_noun_agreement_with_adjective_1": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "determiner_noun_agreement_with_adjective_1", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 185126, "num_examples": 1000, "dataset_name": null } }, "download_size": 55682, "dataset_size": 185126, "size_in_bytes": 240808 }, "distractor_agreement_relational_noun": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "distractor_agreement_relational_noun", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 191473, "num_examples": 1000, "dataset_name": null } }, "download_size": 59641, "dataset_size": 191473, "size_in_bytes": 251114 }, "distractor_agreement_relative_clause": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "distractor_agreement_relative_clause", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 216756, "num_examples": 1000, "dataset_name": null } }, "download_size": 77897, "dataset_size": 216756, "size_in_bytes": 294653 }, "drop_argument": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "drop_argument", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 109806, "num_examples": 1000, "dataset_name": null } }, "download_size": 39961, "dataset_size": 109806, "size_in_bytes": 149767 }, "ellipsis_n_bar_1": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "ellipsis_n_bar_1", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 217590, "num_examples": 1000, "dataset_name": null } }, "download_size": 92776, "dataset_size": 217590, "size_in_bytes": 310366 }, "ellipsis_n_bar_2": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "ellipsis_n_bar_2", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 233161, "num_examples": 1000, "dataset_name": null } }, "download_size": 98882, "dataset_size": 233161, "size_in_bytes": 332043 }, "existential_there_object_raising": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "existential_there_object_raising", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 223741, "num_examples": 1000, "dataset_name": null } }, "download_size": 76641, "dataset_size": 223741, "size_in_bytes": 300382 }, "existential_there_quantifiers_1": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "existential_there_quantifiers_1", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 162931, "num_examples": 1000, "dataset_name": null } }, "download_size": 51576, "dataset_size": 162931, "size_in_bytes": 214507 }, "existential_there_quantifiers_2": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "existential_there_quantifiers_2", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 164826, "num_examples": 1000, "dataset_name": null } }, "download_size": 52092, "dataset_size": 164826, "size_in_bytes": 216918 }, "existential_there_subject_raising": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "existential_there_subject_raising", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 200063, "num_examples": 1000, "dataset_name": null } }, "download_size": 59519, "dataset_size": 200063, "size_in_bytes": 259582 }, "expletive_it_object_raising": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "expletive_it_object_raising", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 238615, "num_examples": 1000, "dataset_name": null } }, "download_size": 88607, "dataset_size": 238615, "size_in_bytes": 327222 }, "inchoative": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "inchoative", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 104319, "num_examples": 1000, "dataset_name": null } }, "download_size": 39842, "dataset_size": 104319, "size_in_bytes": 144161 }, "intransitive": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "intransitive", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 111097, "num_examples": 1000, "dataset_name": null } }, "download_size": 42387, "dataset_size": 111097, "size_in_bytes": 153484 }, "irregular_past_participle_adjectives": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "irregular_past_participle_adjectives", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 144661, "num_examples": 1000, "dataset_name": null } }, "download_size": 36654, "dataset_size": 144661, "size_in_bytes": 181315 }, "irregular_past_participle_verbs": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "irregular_past_participle_verbs", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 125692, "num_examples": 1000, "dataset_name": null } }, "download_size": 37297, "dataset_size": 125692, "size_in_bytes": 162989 }, "irregular_plural_subject_verb_agreement_1": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "irregular_plural_subject_verb_agreement_1", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 165584, "num_examples": 1000, "dataset_name": null } }, "download_size": 50725, "dataset_size": 165584, "size_in_bytes": 216309 }, "irregular_plural_subject_verb_agreement_2": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "irregular_plural_subject_verb_agreement_2", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 153843, "num_examples": 1000, "dataset_name": null } }, "download_size": 42707, "dataset_size": 153843, "size_in_bytes": 196550 }, "left_branch_island_echo_question": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "left_branch_island_echo_question", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 147840, "num_examples": 1000, "dataset_name": null } }, "download_size": 50481, "dataset_size": 147840, "size_in_bytes": 198321 }, "left_branch_island_simple_question": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "sentence_good": { "dtype": "string", "_type": "Value" }, "sentence_bad": { "dtype": "string", "_type": "Value" }, "field": { "dtype": "string", "_type": "Value" }, "linguistics_term": { "dtype": "string", "_type": "Value" }, "UID": { "dtype": "string", "_type": "Value" }, "simple_LM_method": { "dtype": "bool", "_type": "Value" }, "one_prefix_method": { "dtype": "bool", "_type": "Value" }, "two_prefix_method": { "dtype": "bool", "_type": "Value" }, "lexically_identical": { "dtype": "bool", "_type": "Value" }, "pair_id": { "dtype": "int32", "_type": "Value" } }, "builder_name": "blimp", "dataset_name": "blimp", "config_name": "left_branch_island_simple_question", "version": { "version_str": "0.1.0", "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 150060, "num_examples": 1000, "dataset_name": null } }, "download_size": 50293, "dataset_size": 150060, "size_in_bytes": 200353 }, "matrix_question_npi_licensor_present": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "matrix_question_npi_licensor_present", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 154657, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/matrix_question_npi_licensor_present.jsonl": { "num_bytes": 457806, "checksum": "657c9d6e6d50a9cd651881fcbc57cef2fd998611f6f53158ed05e48822fcd2b5" } }, "download_size": 457806, "dataset_size": 154657, "size_in_bytes": 612463 }, "npi_present_1": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "npi_present_1", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 139860, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/npi_present_1.jsonl": { "num_bytes": 438013, "checksum": "9aa85dc5c2b0d85ea51f21db50d6d327ab5b45d1669e386d54b1360d3ac954af" } }, "download_size": 438013, "dataset_size": 139860, "size_in_bytes": 577873 }, "npi_present_2": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "npi_present_2", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 129031, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/npi_present_2.jsonl": { "num_bytes": 422136, "checksum": "52fcfa230a40ae27998273cc08ecd062aad17025c92389df3404a85ed519909b" } }, "download_size": 422136, "dataset_size": 129031, "size_in_bytes": 551167 }, "only_npi_licensor_present": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "only_npi_licensor_present", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 149911, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/only_npi_licensor_present.jsonl": { "num_bytes": 459170, "checksum": "475c6595187a32e4dcfb868d3db90394e89669a94a9312ebd60f88dad1665f1c" } }, "download_size": 459170, "dataset_size": 149911, "size_in_bytes": 609081 }, "only_npi_scope": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "only_npi_scope", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 210297, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/only_npi_scope.jsonl": { "num_bytes": 583720, "checksum": "ca1f0671643e37936515f02125036bc2904ab9b35c4689b0e7a3ea3914e571f6" } }, "download_size": 583720, "dataset_size": 210297, "size_in_bytes": 794017 }, "passive_1": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "passive_1", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 147277, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/passive_1.jsonl": { "num_bytes": 340272, "checksum": "e9374307bca5c24e9df18597a0859f7fb5ca4cfe5f136ec894d1ba69f4c90116" } }, "download_size": 340272, "dataset_size": 147277, "size_in_bytes": 487549 }, "passive_2": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "passive_2", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 115355, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/passive_2.jsonl": { "num_bytes": 308350, "checksum": "4d49e438185bed4122a252dd9ee69bc8eccb2fe6c1b09ccc78198fd925fed8c5" } }, "download_size": 308350, "dataset_size": 115355, "size_in_bytes": 423705 }, "principle_A_c_command": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "principle_A_c_command", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 189885, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/principle_A_c_command.jsonl": { "num_bytes": 527689, "checksum": "74c8111666542c797dee1c38c9e432b0b65751dfa0faa51874a990f2cbfc76fc" } }, "download_size": 527689, "dataset_size": 189885, "size_in_bytes": 717574 }, "principle_A_case_1": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "principle_A_case_1", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 171793, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/principle_A_case_1.jsonl": { "num_bytes": 477239, "checksum": "7efb5246fe315dc2cf56b80739783cffcddee9e52650716cf5aa0f6987f7644d" } }, "download_size": 477239, "dataset_size": 171793, "size_in_bytes": 649032 }, "principle_A_case_2": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "principle_A_case_2", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 171807, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/principle_A_case_2.jsonl": { "num_bytes": 492973, "checksum": "5de6db4c21a8433bc1f1030e81d5ee151efce02078460f7de7ca5e9566afd914" } }, "download_size": 492973, "dataset_size": 171807, "size_in_bytes": 664780 }, "principle_A_domain_1": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "principle_A_domain_1", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 172565, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/principle_A_domain_1.jsonl": { "num_bytes": 499865, "checksum": "2276dc35a51396c7d11a401e378e0f4b31cd10d7916c73e9a361703cb96eddc5" } }, "download_size": 499865, "dataset_size": 172565, "size_in_bytes": 672430 }, "principle_A_domain_2": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "principle_A_domain_2", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 166728, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/principle_A_domain_2.jsonl": { "num_bytes": 493189, "checksum": "9d14e5ae911fb8ddf571484ef2d6a6e6a053765c1c6d34253aec2c11f7d7306a" } }, "download_size": 493189, "dataset_size": 166728, "size_in_bytes": 659917 }, "principle_A_domain_3": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "principle_A_domain_3", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 160393, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/principle_A_domain_3.jsonl": { "num_bytes": 513886, "checksum": "a0d66b9d8541ab1df08f629161c6c4e8bca3aa2693eaa342edd2097e6b579bdd" } }, "download_size": 513886, "dataset_size": 160393, "size_in_bytes": 674279 }, "principle_A_reconstruction": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "principle_A_reconstruction", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 153499, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/principle_A_reconstruction.jsonl": { "num_bytes": 345494, "checksum": "65dc403c4d6e9a609f2c4cc2ec6c8baa05cf6fce89b920a816b0d9932204b2a3" } }, "download_size": 345494, "dataset_size": 153499, "size_in_bytes": 498993 }, "regular_plural_subject_verb_agreement_1": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "regular_plural_subject_verb_agreement_1", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 160214, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/regular_plural_subject_verb_agreement_1.jsonl": { "num_bytes": 451850, "checksum": "1a18d94062c8e792c0a200a3b00dff0e051a92bb084c05e9b5e5c73bee32a620" } }, "download_size": 451850, "dataset_size": 160214, "size_in_bytes": 612064 }, "regular_plural_subject_verb_agreement_2": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "regular_plural_subject_verb_agreement_2", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 155004, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/regular_plural_subject_verb_agreement_2.jsonl": { "num_bytes": 456477, "checksum": "e5b4668811e63b00ac4c9569cb9553281a39c21df43256905ad78e59ae24942c" } }, "download_size": 456477, "dataset_size": 155004, "size_in_bytes": 611481 }, "sentential_negation_npi_licensor_present": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "sentential_negation_npi_licensor_present", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 173259, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/sentential_negation_npi_licensor_present.jsonl": { "num_bytes": 490996, "checksum": "c43405aca13ba4aa6ad87f9b051e4ba5fc458f24e683cfe0af23e6e09ec7268b" } }, "download_size": 490996, "dataset_size": 173259, "size_in_bytes": 664255 }, "sentential_negation_npi_scope": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "sentential_negation_npi_scope", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 233493, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/sentential_negation_npi_scope.jsonl": { "num_bytes": 614930, "checksum": "6c341ccc39c94bea45c0412986751bf2f6eca48626728e30e23d2b6a22c8b949" } }, "download_size": 614930, "dataset_size": 233493, "size_in_bytes": 848423 }, "sentential_subject_island": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "sentential_subject_island", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 173827, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/sentential_subject_island.jsonl": { "num_bytes": 365822, "checksum": "a286b67cc5912aafeb06750b38923abf3338786ffa4309e0a6a07b1dbe2444bd" } }, "download_size": 365822, "dataset_size": 173827, "size_in_bytes": 539649 }, "superlative_quantifiers_1": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "superlative_quantifiers_1", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 160685, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/superlative_quantifiers_1.jsonl": { "num_bytes": 381189, "checksum": "bebc5c5ff28cf226cc35293c60cde1fcf7565a7b58cb219d01ac8908b952a497" } }, "download_size": 381189, "dataset_size": 160685, "size_in_bytes": 541874 }, "superlative_quantifiers_2": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "superlative_quantifiers_2", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 160735, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/superlative_quantifiers_2.jsonl": { "num_bytes": 516120, "checksum": "4bef08f3e6bd3fbb46021dda34000f3c00222c5576babe4882a8541f6820a106" } }, "download_size": 516120, "dataset_size": 160735, "size_in_bytes": 676855 }, "tough_vs_raising_1": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "tough_vs_raising_1", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 150031, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/tough_vs_raising_1.jsonl": { "num_bytes": 343026, "checksum": "8eb3a1ecbdbd969c5b77e2f85b12a39d33723bdc58d673e5067bb456da91d0e5" } }, "download_size": 343026, "dataset_size": 150031, "size_in_bytes": 493057 }, "tough_vs_raising_2": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "tough_vs_raising_2", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 171079, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/tough_vs_raising_2.jsonl": { "num_bytes": 364074, "checksum": "a6db2f24a8aee0c90a9be63e6bb6787fe04e990c518761115f69f27c7438b5ae" } }, "download_size": 364074, "dataset_size": 171079, "size_in_bytes": 535153 }, "transitive": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "transitive", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 134499, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/transitive.jsonl": { "num_bytes": 460291, "checksum": "48de2025571897c7590659c793d2d1ce85c247808d4199bad60f66cf2eb179ef" } }, "download_size": 460291, "dataset_size": 134499, "size_in_bytes": 594790 }, "wh_island": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "wh_island", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 143735, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/wh_island.jsonl": { "num_bytes": 448630, "checksum": "77824ca8c0008fb7a204e785ca2dcee1b2803f666bb545bd95cb8a51e04f1a4c" } }, "download_size": 448630, "dataset_size": 143735, "size_in_bytes": 592365 }, "wh_questions_object_gap": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "wh_questions_object_gap", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 194440, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/wh_questions_object_gap.jsonl": { "num_bytes": 387435, "checksum": "685c74eb8c199f30a24bba1bfabf9d9dff7962a64430e62b9b5f61d5369241ff" } }, "download_size": 387435, "dataset_size": 194440, "size_in_bytes": 581875 }, "wh_questions_subject_gap": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "wh_questions_subject_gap", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 196988, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/wh_questions_subject_gap.jsonl": { "num_bytes": 389983, "checksum": "3c33e4efc7182824211ff3288694cf010f4ce84f85c8df67410d784139c722b3" } }, "download_size": 389983, "dataset_size": 196988, "size_in_bytes": 586971 }, "wh_questions_subject_gap_long_distance": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "wh_questions_subject_gap_long_distance", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 269665, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/wh_questions_subject_gap_long_distance.jsonl": { "num_bytes": 462660, "checksum": "985bae202dbc5566b9b5c51bd77dd9d1dbcd87e0f38e008c34a11d7aa07db788" } }, "download_size": 462660, "dataset_size": 269665, "size_in_bytes": 732325 }, "wh_vs_that_no_gap": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "wh_vs_that_no_gap", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 190267, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/wh_vs_that_no_gap.jsonl": { "num_bytes": 383262, "checksum": "6a25d0d1499e9f522696a62ce9dba8c49f58c5446fddc6d852bf562d1016163c" } }, "download_size": 383262, "dataset_size": 190267, "size_in_bytes": 573529 }, "wh_vs_that_no_gap_long_distance": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "wh_vs_that_no_gap_long_distance", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 248434, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/wh_vs_that_no_gap_long_distance.jsonl": { "num_bytes": 441429, "checksum": "968faf59602a67f9cfd7c2e91a314a3f7342fd5daf41c21cadd25288fc85fdcf" } }, "download_size": 441429, "dataset_size": 248434, "size_in_bytes": 689863 }, "wh_vs_that_with_gap": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "wh_vs_that_with_gap", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 174781, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/wh_vs_that_with_gap.jsonl": { "num_bytes": 367776, "checksum": "56f1ca988a8ca368234f12359792c414fdfec2891154e3f8ce6404da1d3f3212" } }, "download_size": 367776, "dataset_size": 174781, "size_in_bytes": 542557 }, "wh_vs_that_with_gap_long_distance": { "description": "\nBLiMP is a challenge set for evaluating what language models (LMs) know about\nmajor grammatical phenomena in English. BLiMP consists of 67 sub-datasets, each\ncontaining 1000 minimal pairs isolating specific contrasts in syntax,\nmorphology, or semantics. The data is automatically generated according to\nexpert-crafted grammars.\n", "citation": "\n@article{warstadt2019blimp,\n title={BLiMP: A Benchmark of Linguistic Minimal Pairs for English},\n author={Warstadt, Alex and Parrish, Alicia and Liu, Haokun and Mohananey, Anhad and Peng, Wei, and Wang, Sheng-Fu and Bowman, Samuel R},\n journal={arXiv preprint arXiv:1912.00582},\n year={2019}\n}\n", "homepage": "https://github.com/alexwarstadt/blimp/tree/master/", "license": "", "features": { "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" } }, "supervised_keys": null, "builder_name": "blimp", "config_name": "wh_vs_that_with_gap_long_distance", "version": { "version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 232990, "num_examples": 1000, "dataset_name": "blimp" } }, "download_checksums": { "https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/wh_vs_that_with_gap_long_distance.jsonl": { "num_bytes": 425985, "checksum": "3cb19100d6384264e512801e96c20e172a3c0c7778276b80ed60c65e97b86ab5" } }, "download_size": 425985, "dataset_size": 232990, "size_in_bytes": 658975 } }