Datasets:

Multilinguality:
multilingual
Size Categories:
1K<n<10K
Annotations Creators:
crowdsourced
ArXiv:
Tags:
License:
albertvillanova HF staff commited on
Commit
63a5ce5
1 Parent(s): ab11c86

Add X-CSQA-jap data files

Browse files
README.md CHANGED
@@ -709,13 +709,13 @@ dataset_info:
709
  dtype: string
710
  splits:
711
  - name: test
712
- num_bytes: 251148
713
  num_examples: 1074
714
  - name: validation
715
- num_bytes: 240686
716
  num_examples: 1000
717
- download_size: 7519903
718
- dataset_size: 491834
719
  - config_name: X-CSQA-nl
720
  features:
721
  - name: id
@@ -963,6 +963,12 @@ configs:
963
  path: X-CSQA-it/test-*
964
  - split: validation
965
  path: X-CSQA-it/validation-*
 
 
 
 
 
 
966
  - config_name: X-CSQA-zh
967
  data_files:
968
  - split: test
 
709
  dtype: string
710
  splits:
711
  - name: test
712
+ num_bytes: 250846
713
  num_examples: 1074
714
  - name: validation
715
+ num_bytes: 240404
716
  num_examples: 1000
717
+ download_size: 249420
718
+ dataset_size: 491250
719
  - config_name: X-CSQA-nl
720
  features:
721
  - name: id
 
963
  path: X-CSQA-it/test-*
964
  - split: validation
965
  path: X-CSQA-it/validation-*
966
+ - config_name: X-CSQA-jap
967
+ data_files:
968
+ - split: test
969
+ path: X-CSQA-jap/test-*
970
+ - split: validation
971
+ path: X-CSQA-jap/validation-*
972
  - config_name: X-CSQA-zh
973
  data_files:
974
  - split: test
X-CSQA-jap/test-00000-of-00001.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e95aab82e0977b0987ac2ee33696580ea1b5b18d43a6008982f2ced1b3ef2011
3
+ size 128992
X-CSQA-jap/validation-00000-of-00001.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:11c79d3036e27139bb62c1ad7b78f8275fe2bb710033e25b0fb30a605dc950d5
3
+ size 120428
dataset_infos.json CHANGED
@@ -403,48 +403,38 @@
403
  "features": {
404
  "id": {
405
  "dtype": "string",
406
- "id": null,
407
  "_type": "Value"
408
  },
409
  "lang": {
410
  "dtype": "string",
411
- "id": null,
412
  "_type": "Value"
413
  },
414
  "question": {
415
  "stem": {
416
  "dtype": "string",
417
- "id": null,
418
  "_type": "Value"
419
  },
420
  "choices": {
421
  "feature": {
422
  "label": {
423
  "dtype": "string",
424
- "id": null,
425
  "_type": "Value"
426
  },
427
  "text": {
428
  "dtype": "string",
429
- "id": null,
430
  "_type": "Value"
431
  }
432
  },
433
- "length": -1,
434
- "id": null,
435
  "_type": "Sequence"
436
  }
437
  },
438
  "answerKey": {
439
  "dtype": "string",
440
- "id": null,
441
  "_type": "Value"
442
  }
443
  },
444
- "post_processed": null,
445
- "supervised_keys": null,
446
- "task_templates": null,
447
  "builder_name": "xcsr",
 
448
  "config_name": "X-CSQA-jap",
449
  "version": {
450
  "version_str": "1.1.0",
@@ -456,27 +446,20 @@
456
  "splits": {
457
  "test": {
458
  "name": "test",
459
- "num_bytes": 251148,
460
  "num_examples": 1074,
461
- "dataset_name": "xcsr"
462
  },
463
  "validation": {
464
  "name": "validation",
465
- "num_bytes": 240686,
466
  "num_examples": 1000,
467
- "dataset_name": "xcsr"
468
- }
469
- },
470
- "download_checksums": {
471
- "https://inklab.usc.edu/XCSR/xcsr_datasets.zip": {
472
- "num_bytes": 7519903,
473
- "checksum": "c45b29ece740643252d5402e76be1e33f96f9d6910053f79e80d39887f10c85e"
474
  }
475
  },
476
- "download_size": 7519903,
477
- "post_processing_size": null,
478
- "dataset_size": 491834,
479
- "size_in_bytes": 8011737
480
  },
481
  "X-CSQA-nl": {
482
  "description": "To evaluate multi-lingual language models (ML-LMs) for commonsense reasoning in a cross-lingual zero-shot transfer setting (X-CSR), i.e., training in English and test in other languages, we create two benchmark datasets, namely X-CSQA and X-CODAH. Specifically, we automatically translate the original CSQA and CODAH datasets, which only have English versions, to 15 other languages, forming development and test sets for studying X-CSR. As our goal is to evaluate different ML-LMs in a unified evaluation protocol for X-CSR, we argue that such translated examples, although might contain noise, can serve as a starting benchmark for us to obtain meaningful analysis, before more human-translated datasets will be available in the future.\n",
 
403
  "features": {
404
  "id": {
405
  "dtype": "string",
 
406
  "_type": "Value"
407
  },
408
  "lang": {
409
  "dtype": "string",
 
410
  "_type": "Value"
411
  },
412
  "question": {
413
  "stem": {
414
  "dtype": "string",
 
415
  "_type": "Value"
416
  },
417
  "choices": {
418
  "feature": {
419
  "label": {
420
  "dtype": "string",
 
421
  "_type": "Value"
422
  },
423
  "text": {
424
  "dtype": "string",
 
425
  "_type": "Value"
426
  }
427
  },
 
 
428
  "_type": "Sequence"
429
  }
430
  },
431
  "answerKey": {
432
  "dtype": "string",
 
433
  "_type": "Value"
434
  }
435
  },
 
 
 
436
  "builder_name": "xcsr",
437
+ "dataset_name": "xcsr",
438
  "config_name": "X-CSQA-jap",
439
  "version": {
440
  "version_str": "1.1.0",
 
446
  "splits": {
447
  "test": {
448
  "name": "test",
449
+ "num_bytes": 250846,
450
  "num_examples": 1074,
451
+ "dataset_name": null
452
  },
453
  "validation": {
454
  "name": "validation",
455
+ "num_bytes": 240404,
456
  "num_examples": 1000,
457
+ "dataset_name": null
 
 
 
 
 
 
458
  }
459
  },
460
+ "download_size": 249420,
461
+ "dataset_size": 491250,
462
+ "size_in_bytes": 740670
 
463
  },
464
  "X-CSQA-nl": {
465
  "description": "To evaluate multi-lingual language models (ML-LMs) for commonsense reasoning in a cross-lingual zero-shot transfer setting (X-CSR), i.e., training in English and test in other languages, we create two benchmark datasets, namely X-CSQA and X-CODAH. Specifically, we automatically translate the original CSQA and CODAH datasets, which only have English versions, to 15 other languages, forming development and test sets for studying X-CSR. As our goal is to evaluate different ML-LMs in a unified evaluation protocol for X-CSR, we argue that such translated examples, although might contain noise, can serve as a starting benchmark for us to obtain meaningful analysis, before more human-translated datasets will be available in the future.\n",