Datasets:
Tasks:
Text Classification
Modalities:
Text
Formats:
parquet
Sub-tasks:
acceptability-classification
Languages:
English
Size:
10K - 100K
ArXiv:
License:
Commit
•
d22e0d9
1
Parent(s):
c353d7f
Add coordinate_structure_constraint_object_extraction data files
Browse files
README.md
CHANGED
@@ -268,10 +268,10 @@ dataset_info:
|
|
268 |
dtype: int32
|
269 |
splits:
|
270 |
- name: train
|
271 |
-
num_bytes:
|
272 |
num_examples: 1000
|
273 |
-
download_size:
|
274 |
-
dataset_size:
|
275 |
- config_name: determiner_noun_agreement_1
|
276 |
features:
|
277 |
- name: sentence_good
|
@@ -1929,6 +1929,10 @@ configs:
|
|
1929 |
data_files:
|
1930 |
- split: train
|
1931 |
path: coordinate_structure_constraint_complex_left_branch/train-*
|
|
|
|
|
|
|
|
|
1932 |
---
|
1933 |
|
1934 |
# Dataset Card for "blimp"
|
|
|
268 |
dtype: int32
|
269 |
splits:
|
270 |
- name: train
|
271 |
+
num_bytes: 171655
|
272 |
num_examples: 1000
|
273 |
+
download_size: 51584
|
274 |
+
dataset_size: 171655
|
275 |
- config_name: determiner_noun_agreement_1
|
276 |
features:
|
277 |
- name: sentence_good
|
|
|
1929 |
data_files:
|
1930 |
- split: train
|
1931 |
path: coordinate_structure_constraint_complex_left_branch/train-*
|
1932 |
+
- config_name: coordinate_structure_constraint_object_extraction
|
1933 |
+
data_files:
|
1934 |
+
- split: train
|
1935 |
+
path: coordinate_structure_constraint_object_extraction/train-*
|
1936 |
---
|
1937 |
|
1938 |
# Dataset Card for "blimp"
|
coordinate_structure_constraint_object_extraction/train-00000-of-00001.parquet
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:55d830a519e53f2400d3c436f751295a83a123f46084f247ce802ebec9585816
|
3 |
+
size 51584
|
dataset_infos.json
CHANGED
@@ -551,62 +551,50 @@
|
|
551 |
"features": {
|
552 |
"sentence_good": {
|
553 |
"dtype": "string",
|
554 |
-
"id": null,
|
555 |
"_type": "Value"
|
556 |
},
|
557 |
"sentence_bad": {
|
558 |
"dtype": "string",
|
559 |
-
"id": null,
|
560 |
"_type": "Value"
|
561 |
},
|
562 |
"field": {
|
563 |
"dtype": "string",
|
564 |
-
"id": null,
|
565 |
"_type": "Value"
|
566 |
},
|
567 |
"linguistics_term": {
|
568 |
"dtype": "string",
|
569 |
-
"id": null,
|
570 |
"_type": "Value"
|
571 |
},
|
572 |
"UID": {
|
573 |
"dtype": "string",
|
574 |
-
"id": null,
|
575 |
"_type": "Value"
|
576 |
},
|
577 |
"simple_LM_method": {
|
578 |
"dtype": "bool",
|
579 |
-
"id": null,
|
580 |
"_type": "Value"
|
581 |
},
|
582 |
"one_prefix_method": {
|
583 |
"dtype": "bool",
|
584 |
-
"id": null,
|
585 |
"_type": "Value"
|
586 |
},
|
587 |
"two_prefix_method": {
|
588 |
"dtype": "bool",
|
589 |
-
"id": null,
|
590 |
"_type": "Value"
|
591 |
},
|
592 |
"lexically_identical": {
|
593 |
"dtype": "bool",
|
594 |
-
"id": null,
|
595 |
"_type": "Value"
|
596 |
},
|
597 |
"pair_id": {
|
598 |
"dtype": "int32",
|
599 |
-
"id": null,
|
600 |
"_type": "Value"
|
601 |
}
|
602 |
},
|
603 |
-
"supervised_keys": null,
|
604 |
"builder_name": "blimp",
|
|
|
605 |
"config_name": "coordinate_structure_constraint_object_extraction",
|
606 |
"version": {
|
607 |
"version_str": "0.1.0",
|
608 |
-
"description": null,
|
609 |
-
"datasets_version_to_prepare": null,
|
610 |
"major": 0,
|
611 |
"minor": 1,
|
612 |
"patch": 0
|
@@ -614,20 +602,14 @@
|
|
614 |
"splits": {
|
615 |
"train": {
|
616 |
"name": "train",
|
617 |
-
"num_bytes":
|
618 |
"num_examples": 1000,
|
619 |
-
"dataset_name":
|
620 |
-
}
|
621 |
-
},
|
622 |
-
"download_checksums": {
|
623 |
-
"https://raw.githubusercontent.com/alexwarstadt/blimp/master/data/coordinate_structure_constraint_object_extraction.jsonl": {
|
624 |
-
"num_bytes": 366045,
|
625 |
-
"checksum": "d01c27a7a893f3ec071a7913efa504d297671d86a987f26ddb565de914db2fd7"
|
626 |
}
|
627 |
},
|
628 |
-
"download_size":
|
629 |
-
"dataset_size":
|
630 |
-
"size_in_bytes":
|
631 |
},
|
632 |
"determiner_noun_agreement_1": {
|
633 |
"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",
|
|
|
551 |
"features": {
|
552 |
"sentence_good": {
|
553 |
"dtype": "string",
|
|
|
554 |
"_type": "Value"
|
555 |
},
|
556 |
"sentence_bad": {
|
557 |
"dtype": "string",
|
|
|
558 |
"_type": "Value"
|
559 |
},
|
560 |
"field": {
|
561 |
"dtype": "string",
|
|
|
562 |
"_type": "Value"
|
563 |
},
|
564 |
"linguistics_term": {
|
565 |
"dtype": "string",
|
|
|
566 |
"_type": "Value"
|
567 |
},
|
568 |
"UID": {
|
569 |
"dtype": "string",
|
|
|
570 |
"_type": "Value"
|
571 |
},
|
572 |
"simple_LM_method": {
|
573 |
"dtype": "bool",
|
|
|
574 |
"_type": "Value"
|
575 |
},
|
576 |
"one_prefix_method": {
|
577 |
"dtype": "bool",
|
|
|
578 |
"_type": "Value"
|
579 |
},
|
580 |
"two_prefix_method": {
|
581 |
"dtype": "bool",
|
|
|
582 |
"_type": "Value"
|
583 |
},
|
584 |
"lexically_identical": {
|
585 |
"dtype": "bool",
|
|
|
586 |
"_type": "Value"
|
587 |
},
|
588 |
"pair_id": {
|
589 |
"dtype": "int32",
|
|
|
590 |
"_type": "Value"
|
591 |
}
|
592 |
},
|
|
|
593 |
"builder_name": "blimp",
|
594 |
+
"dataset_name": "blimp",
|
595 |
"config_name": "coordinate_structure_constraint_object_extraction",
|
596 |
"version": {
|
597 |
"version_str": "0.1.0",
|
|
|
|
|
598 |
"major": 0,
|
599 |
"minor": 1,
|
600 |
"patch": 0
|
|
|
602 |
"splits": {
|
603 |
"train": {
|
604 |
"name": "train",
|
605 |
+
"num_bytes": 171655,
|
606 |
"num_examples": 1000,
|
607 |
+
"dataset_name": null
|
|
|
|
|
|
|
|
|
|
|
|
|
608 |
}
|
609 |
},
|
610 |
+
"download_size": 51584,
|
611 |
+
"dataset_size": 171655,
|
612 |
+
"size_in_bytes": 223239
|
613 |
},
|
614 |
"determiner_noun_agreement_1": {
|
615 |
"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",
|