{"success":true,"data":{"author":"Warstadt, Parrish, Liu, Mohananey, Peng, Wang, Bowman (New York University)","bucketAccessType":"public","bucketName":"nyu-mll/blimp","bucketPrefix":"","category":"Natural Language Processing","cloudProvider":"huggingface","createdAt":"2026-11-21T22:16:38.820Z","description":"The Benchmark of Linguistic Minimal Pairs (BLiMP) tests what language models know about English grammar. It has 67 sub-datasets of 1,000 minimal pairs each (67,000 pairs). A minimal pair is two sentences that differ in one point, one grammatically acceptable and one not, which isolates a phenomenon of syntax, morphology or semantics. Examples are subject-verb and anaphor agreement, argument structure, binding, control and raising, determiner-noun agreement, ellipsis, filler-gap dependencies, island effects, negative polarity item licensing, quantifiers and irregular forms. The sentences were generated automatically from grammars written by linguists, and human agreement with the labels is 96.4%. Each row has sentence_good, sentence_bad, the field (syntax, morphology, semantics), the linguistic term, the UID of the paradigm and flags. Psycholinguists and NLP researchers score a model by whether it assigns higher probability to the acceptable sentence. There is one Parquet file per phenomenon; load it with Hugging Face datasets by config name. BLiMP is distributed under CC BY 4.0. Cite Warstadt et al., BLiMP: The Benchmark of Linguistic Minimal Pairs for English, TACL 2020.","displayTitle":"BLiMP: Benchmark of Linguistic Minimal Pairs for English Grammar (67 Phenomena, NYU)","huggingFaceUrl":"https://huggingface.co/datasets/nyu-mll/blimp","id":"95452e86-5723-5d9f-a208-805d20f0dd59","license":"CC-BY-4.0","licenseEvidenceUrl":"https://github.com/alexwarstadt/blimp#license","objectCountEstimate":69,"probeStatus":"ok","referenceUrl":"https://arxiv.org/abs/1912.00582","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"nyu-mll/blimp\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"nyu-mll/blimp\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=nyu-mll, license:cc-by-4.0, size_categories:10K<n<100K, format:parquet, downloads=46460"],"format":"mixed","generatedAt":"2026-11-21T22:31:57.977Z","layout":{"example":"adjunct_island/train-00000-of-00001.parquet","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"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","encoding":"hex","key":"adjunct_island/train-00000-of-00001.parquet","truncatedAt":512}],"sampleObjects":["adjunct_island/train-00000-of-00001.parquet","anaphor_gender_agreement/train-00000-of-00001.parquet","anaphor_number_agreement/train-00000-of-00001.parquet","animate_subject_passive/train-00000-of-00001.parquet","animate_subject_trans/train-00000-of-00001.parquet"],"sidecars":[{"bytes":"---\nannotations_creators:\n- crowdsourced\nlanguage_creators:\n- machine-generated\nlanguage:\n- en\nlicense:\n- cc-by-4.0\nmultilinguality:\n- monolingual\nsize_categories:\n- 10K<n<100K\nsource_datasets:\n- original\ntask_categories:\n- text-classification\ntask_ids:\n- acceptability-classification\npaperswithcode_id: blimp\npretty_name: BLiMP\ndataset_info:\n- config_name: adjunct_island\n  features:\n  - name: sentence_good\n    dtype: string\n  - name: sentence_bad\n    dtype: string\n  - name: field\n    dtype: string\n  - name: linguistics_term\n    dtype: string\n  - name: UID\n    dtype: string\n  - name: simple_LM_method\n    dtype: bool\n  - name: one_prefix_method\n    dtype: bool\n  - name: two_prefix_method\n    dtype: bool\n  - name: lexically_identical\n    dtype: bool\n  - name: pair_id\n    dtype: int32\n  splits:\n  - name: train\n    num_bytes: 165894\n    num_examples: 1000\n  download_size: 62231\n  dataset_size: 165894\n- config_name: anaphor_gender_agreement\n  features:\n  - name: sentence_good\n    dtype: string\n  - name: sentence_bad\n    dtype: string\n  - name: field\n    dtype: string\n  - name: linguistics_term\n    dtype: string\n  - name: UID\n    dtype: string\n  - name: simple_LM_method\n    dtype: bool\n  - name: one_prefix_method\n    dtype: bool\n  - name: two_prefix_method\n    dtype: bool\n  - name: lexically_identical\n    dtype: bool\n  - name: pair_id\n    dtype: int32\n  splits:\n  - name: train\n    num_bytes: 130918\n    num_examples: 1000\n  download_size: 39201\n  dataset_size: 130918\n- config_name: anaphor_number_agreement\n  features:\n  - name: sentence_good\n    dtype: string\n  - name: sentence_bad\n    dtype: string\n  - name: field\n    dtype: string\n  - name: linguistics_term\n    dtype: string\n  - name: UID\n    dtype: string\n  - name: simple_LM_method\n    dtype: bool\n  - name: one_prefix_method\n    dtype: bool\n  - name: two_prefix_method\n    dtype: bool\n  - name: lexically_identical\n    dtype: bool\n  - name: pair_id\n    dtype: int32\n  splits:\n  - name: train\n    num_bytes: 139879\n    num_examples: 1000\n  download_size: 41547\n  dataset_size: 139879\n- config_name: animate_subject_passive\n  features:\n  - name: sentence_good\n    dtype: string\n  - name: sentence_bad\n    dtype: string\n  - name: field\n    dtype: string\n  - name: linguistics_term\n    dtype: string\n  - name: UID\n    dtype: string\n  - name: simple_LM_method\n    dtype: bool\n  - name: one_prefix_method\n    dtype: bool\n  - name: two_prefix_method\n    dtype: bool\n  - name: lexically_identical\n    dtype: bool\n  - name: pair_id\n    dtype: int32\n  splits:\n  - name: train\n    num_bytes: 144423\n    num_examples: 1000\n  download_size: 47282\n  dataset_size: 144423\n- config_name: animate_subject_trans\n  features:\n  - name: sentence_good\n    dtype: string\n  - name: sentence_bad\n    dtype: string\n  - name: field\n    dtype: string\n  - name: linguistics_term\n    dtype: string\n  - name: UID\n    dtype: string\n  - name: simple_LM_method\n    dtype: bool\n  - name: one_prefix_method\n    dtype: bool\n  - name: two_prefix_method\n    dtype: bool\n  - name: lexically_identical\n    dtype: bool\n  - name: pair_id\n    dtype: int32\n  splits:\n  - name: train\n    num_bytes: 127798\n    num_examples: 1000\n  download_size: 49651\n  dataset_size: 127798\n- config_name: causative\n  features:\n  - name: sentence_good\n    dtype: string\n  - name: sentence_bad\n    dtype: string\n  - name: field\n    dtype: string\n  - name: linguistics_term\n    dtype: string\n  - name: UID\n    dtype: string\n  - name: simple_LM_method\n    dtype: bool\n  - name: one_prefix_method\n    dtype: bool\n  - name: two_prefix_method\n    dtype: bool\n  - name: lexically_identical\n    dtype: bool\n  - name: pair_id\n    dtype: int32\n  splits:\n  - name: train\n    num_bytes: 122772\n    num_examples: 1000\n  download_size: 48963\n  dataset_size: 122772\n- config_name: complex_NP_island\n  features:\n  - name: sentence_good\n    dtype: string\n  - name: sentence_bad\n    dtype: string\n  - name: field\n    dtype: string\n  - name: linguistics_term\n    dtype: string\n  - name: UID\n    dtype: string\n  - name: simple_LM_method\n    dtype: bool\n  - name: one_prefix_method\n    dtype: bool\n  - na","encoding":"utf8","key":"README.md","name":"README.md","size":58554,"truncated":true}],"sourceObjectKey":"adjunct_island/train-00000-of-00001.parquet","starterCell":"from datasets import load_dataset\nds = load_dataset(\"nyu-mll/blimp\")\nprint(ds)","starterSamplePath":"adjunct_island/train-00000-of-00001.parquet","totalObjects":69},"summary":"The Benchmark of Linguistic Minimal Pairs (BLiMP) tests what language models know about English grammar. It has 67 sub-datasets of 1,000 minimal pairs each (67,000 pairs).","title":"BLiMP: Benchmark of Linguistic Minimal Pairs for English Grammar (67 Phenomena, NYU)","updatedAt":"2026-11-22T00:56:18.555Z","visibility":"public","websiteUrl":"https://huggingface.co/datasets/nyu-mll/blimp"}}