Datasets:
Tasks:
Question Answering
Modalities:
Text
Formats:
parquet
Sub-tasks:
multiple-choice-qa
Languages:
tu
Size:
10K - 100K
ArXiv:
License:
metadata
annotations_creators:
- machine-generated
language_creators:
- machine-translated
language:
- tu
multilinguality:
- monolingual
size_categories:
- 10K<n<100K
source_datasets:
- mmlu
- arabicmmlu
- DarijaMMlu
task_categories:
- question-answering
task_ids:
- multiple-choice-qa
dataset_info:
- config_name: accounting
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 29430
num_examples: 74
- name: dev
num_bytes: 1076
num_examples: 3
download_size: 20994
dataset_size: 30506
- config_name: arabic_language
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 238173
num_examples: 669
- name: dev
num_bytes: 2590
num_examples: 9
download_size: 86977
dataset_size: 240763
- config_name: arabic_language_(general)
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 1446694
num_examples: 612
- name: dev
num_bytes: 6308
num_examples: 3
download_size: 303041
dataset_size: 1453002
- config_name: arabic_language_(grammar)
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 123836
num_examples: 365
- name: dev
num_bytes: 796
num_examples: 3
download_size: 34132
dataset_size: 124632
- config_name: biology
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 427458
num_examples: 1409
- name: dev
num_bytes: 964
num_examples: 3
download_size: 159873
dataset_size: 428422
- config_name: civics
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 104820
num_examples: 323
- name: dev
num_bytes: 1765
num_examples: 6
download_size: 45019
dataset_size: 106585
- config_name: computer_science
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 151830
num_examples: 542
- name: dev
num_bytes: 3832
num_examples: 12
download_size: 59956
dataset_size: 155662
- config_name: driving_test
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 400625
num_examples: 1211
- name: dev
num_bytes: 929
num_examples: 3
download_size: 142002
dataset_size: 401554
- config_name: economics
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 220639
num_examples: 584
- name: dev
num_bytes: 2644
num_examples: 9
download_size: 85240
dataset_size: 223283
- config_name: general_knowledge
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 311288
num_examples: 1198
- name: dev
num_bytes: 2812
num_examples: 9
download_size: 114404
dataset_size: 314100
- config_name: geography
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 402361
num_examples: 1367
- name: dev
num_bytes: 2614
num_examples: 9
download_size: 130186
dataset_size: 404975
- config_name: global_facts
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 33173
num_examples: 100
- name: dev
num_bytes: 1776
num_examples: 5
download_size: 21666
dataset_size: 34949
- config_name: high_school_european_history
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 310781
num_examples: 165
- name: dev
num_bytes: 13235
num_examples: 5
download_size: 156010
dataset_size: 324016
- config_name: high_school_geography
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 76163
num_examples: 198
- name: dev
num_bytes: 2368
num_examples: 5
download_size: 39213
dataset_size: 78531
- config_name: high_school_government_and_politics
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 115184
num_examples: 193
- name: dev
num_bytes: 3078
num_examples: 5
download_size: 55948
dataset_size: 118262
- config_name: high_school_psychology
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 260934
num_examples: 545
- name: dev
num_bytes: 2811
num_examples: 5
download_size: 114905
dataset_size: 263745
- config_name: high_school_statistics
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 165860
num_examples: 216
- name: dev
num_bytes: 3834
num_examples: 5
download_size: 82190
dataset_size: 169694
- config_name: high_school_world_history
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 539950
num_examples: 237
- name: dev
num_bytes: 7418
num_examples: 5
download_size: 264937
dataset_size: 547368
- config_name: history
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 316416
num_examples: 1065
- name: dev
num_bytes: 2283
num_examples: 9
download_size: 117684
dataset_size: 318699
- config_name: human_aging
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 77145
num_examples: 223
- name: dev
num_bytes: 1681
num_examples: 5
download_size: 42019
dataset_size: 78826
- config_name: international_law
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 82207
num_examples: 121
- name: dev
num_bytes: 3553
num_examples: 5
download_size: 42384
dataset_size: 85760
- config_name: islamic_studies
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 612921
num_examples: 2210
- name: dev
num_bytes: 4052
num_examples: 12
download_size: 204687
dataset_size: 616973
- config_name: jurisprudence
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 51680
num_examples: 108
- name: dev
num_bytes: 2009
num_examples: 5
download_size: 33351
dataset_size: 53689
- config_name: law
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 139087
num_examples: 314
- name: dev
num_bytes: 1843
num_examples: 3
download_size: 61065
dataset_size: 140930
- config_name: logical_fallacies
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 77315
num_examples: 163
- name: dev
num_bytes: 2386
num_examples: 5
download_size: 39026
dataset_size: 79701
- config_name: management
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 31077
num_examples: 103
- name: dev
num_bytes: 1431
num_examples: 5
download_size: 22592
dataset_size: 32508
- config_name: management_ar
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 21513
num_examples: 75
- name: dev
num_bytes: 784
num_examples: 3
download_size: 17156
dataset_size: 22297
- config_name: marketing
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 95342
num_examples: 234
- name: dev
num_bytes: 2436
num_examples: 5
download_size: 50326
dataset_size: 97778
- config_name: math
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 65200
num_examples: 409
- name: dev
num_bytes: 521
num_examples: 3
download_size: 26865
dataset_size: 65721
- config_name: moral_disputes
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 170285
num_examples: 346
- name: dev
num_bytes: 2899
num_examples: 5
download_size: 81210
dataset_size: 173184
- config_name: moral_scenarios
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 601102
num_examples: 895
- name: dev
num_bytes: 3226
num_examples: 5
download_size: 143240
dataset_size: 604328
- config_name: natural_science
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 159752
num_examples: 578
- name: dev
num_bytes: 1390
num_examples: 6
download_size: 61357
dataset_size: 161142
- config_name: nutrition
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 148940
num_examples: 306
- name: dev
num_bytes: 3393
num_examples: 5
download_size: 73626
dataset_size: 152333
- config_name: philosophy
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 119272
num_examples: 311
- name: dev
num_bytes: 1477
num_examples: 5
download_size: 61274
dataset_size: 120749
- config_name: philosophy_ar
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 17043
num_examples: 39
- name: dev
num_bytes: 1242
num_examples: 3
download_size: 18669
dataset_size: 18285
- config_name: physics
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 85558
num_examples: 255
- name: dev
num_bytes: 1091
num_examples: 3
download_size: 40315
dataset_size: 86649
- config_name: political_science
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 64237
num_examples: 210
- name: dev
num_bytes: 895
num_examples: 3
download_size: 30217
dataset_size: 65132
- config_name: professional_law
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 2251189
num_examples: 1534
- name: dev
num_bytes: 7218
num_examples: 5
download_size: 1019946
dataset_size: 2258407
- config_name: professional_psychology
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 345414
num_examples: 612
- name: dev
num_bytes: 3533
num_examples: 5
download_size: 158991
dataset_size: 348947
- config_name: public_relations
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 45790
num_examples: 110
- name: dev
num_bytes: 2352
num_examples: 5
download_size: 29615
dataset_size: 48142
- config_name: security_studies
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 297804
num_examples: 245
- name: dev
num_bytes: 7454
num_examples: 5
download_size: 143008
dataset_size: 305258
- config_name: social_science
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 223619
num_examples: 946
- name: dev
num_bytes: 1386
num_examples: 6
download_size: 76151
dataset_size: 225005
- config_name: sociology
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 102564
num_examples: 201
- name: dev
num_bytes: 2553
num_examples: 5
download_size: 57047
dataset_size: 105117
- config_name: world_religions
features:
- name: question
dtype: string
- name: context
dtype: string
- name: choices
sequence: string
- name: answer
dtype: int64
- name: subject
dtype: string
- name: subject_darija
dtype: string
- name: source
dtype: string
splits:
- name: test
num_bytes: 46118
num_examples: 171
- name: dev
num_bytes: 1215
num_examples: 5
download_size: 27597
dataset_size: 47333
configs:
- config_name: accounting
data_files:
- split: test
path: accounting/test-*
- split: dev
path: accounting/dev-*
- config_name: arabic_language
data_files:
- split: test
path: arabic_language/test-*
- split: dev
path: arabic_language/dev-*
- config_name: arabic_language_(general)
data_files:
- split: test
path: arabic_language_(general)/test-*
- split: dev
path: arabic_language_(general)/dev-*
- config_name: arabic_language_(grammar)
data_files:
- split: test
path: arabic_language_(grammar)/test-*
- split: dev
path: arabic_language_(grammar)/dev-*
- config_name: biology
data_files:
- split: test
path: biology/test-*
- split: dev
path: biology/dev-*
- config_name: civics
data_files:
- split: test
path: civics/test-*
- split: dev
path: civics/dev-*
- config_name: computer_science
data_files:
- split: test
path: computer_science/test-*
- split: dev
path: computer_science/dev-*
- config_name: driving_test
data_files:
- split: test
path: driving_test/test-*
- split: dev
path: driving_test/dev-*
- config_name: economics
data_files:
- split: test
path: economics/test-*
- split: dev
path: economics/dev-*
- config_name: general_knowledge
data_files:
- split: test
path: general_knowledge/test-*
- split: dev
path: general_knowledge/dev-*
- config_name: geography
data_files:
- split: test
path: geography/test-*
- split: dev
path: geography/dev-*
- config_name: global_facts
data_files:
- split: test
path: global_facts/test-*
- split: dev
path: global_facts/dev-*
- config_name: high_school_european_history
data_files:
- split: test
path: high_school_european_history/test-*
- split: dev
path: high_school_european_history/dev-*
- config_name: high_school_geography
data_files:
- split: test
path: high_school_geography/test-*
- split: dev
path: high_school_geography/dev-*
- config_name: high_school_government_and_politics
data_files:
- split: test
path: high_school_government_and_politics/test-*
- split: dev
path: high_school_government_and_politics/dev-*
- config_name: high_school_psychology
data_files:
- split: test
path: high_school_psychology/test-*
- split: dev
path: high_school_psychology/dev-*
- config_name: high_school_statistics
data_files:
- split: test
path: high_school_statistics/test-*
- split: dev
path: high_school_statistics/dev-*
- config_name: high_school_world_history
data_files:
- split: test
path: high_school_world_history/test-*
- split: dev
path: high_school_world_history/dev-*
- config_name: history
data_files:
- split: test
path: history/test-*
- split: dev
path: history/dev-*
- config_name: human_aging
data_files:
- split: test
path: human_aging/test-*
- split: dev
path: human_aging/dev-*
- config_name: international_law
data_files:
- split: test
path: international_law/test-*
- split: dev
path: international_law/dev-*
- config_name: islamic_studies
data_files:
- split: test
path: islamic_studies/test-*
- split: dev
path: islamic_studies/dev-*
- config_name: jurisprudence
data_files:
- split: test
path: jurisprudence/test-*
- split: dev
path: jurisprudence/dev-*
- config_name: law
data_files:
- split: test
path: law/test-*
- split: dev
path: law/dev-*
- config_name: logical_fallacies
data_files:
- split: test
path: logical_fallacies/test-*
- split: dev
path: logical_fallacies/dev-*
- config_name: management
data_files:
- split: test
path: management/test-*
- split: dev
path: management/dev-*
- config_name: management_ar
data_files:
- split: test
path: management_ar/test-*
- split: dev
path: management_ar/dev-*
- config_name: marketing
data_files:
- split: test
path: marketing/test-*
- split: dev
path: marketing/dev-*
- config_name: math
data_files:
- split: test
path: math/test-*
- split: dev
path: math/dev-*
- config_name: moral_disputes
data_files:
- split: test
path: moral_disputes/test-*
- split: dev
path: moral_disputes/dev-*
- config_name: moral_scenarios
data_files:
- split: test
path: moral_scenarios/test-*
- split: dev
path: moral_scenarios/dev-*
- config_name: natural_science
data_files:
- split: test
path: natural_science/test-*
- split: dev
path: natural_science/dev-*
- config_name: nutrition
data_files:
- split: test
path: nutrition/test-*
- split: dev
path: nutrition/dev-*
- config_name: philosophy
data_files:
- split: test
path: philosophy/test-*
- split: dev
path: philosophy/dev-*
- config_name: philosophy_ar
data_files:
- split: test
path: philosophy_ar/test-*
- split: dev
path: philosophy_ar/dev-*
- config_name: physics
data_files:
- split: test
path: physics/test-*
- split: dev
path: physics/dev-*
- config_name: political_science
data_files:
- split: test
path: political_science/test-*
- split: dev
path: political_science/dev-*
- config_name: professional_law
data_files:
- split: test
path: professional_law/test-*
- split: dev
path: professional_law/dev-*
- config_name: professional_psychology
data_files:
- split: test
path: professional_psychology/test-*
- split: dev
path: professional_psychology/dev-*
- config_name: public_relations
data_files:
- split: test
path: public_relations/test-*
- split: dev
path: public_relations/dev-*
- config_name: security_studies
data_files:
- split: test
path: security_studies/test-*
- split: dev
path: security_studies/dev-*
- config_name: social_science
data_files:
- split: test
path: social_science/test-*
- split: dev
path: social_science/dev-*
- config_name: sociology
data_files:
- split: test
path: sociology/test-*
- split: dev
path: sociology/dev-*
- config_name: world_religions
data_files:
- split: test
path: world_religions/test-*
- split: dev
path: world_religions/dev-*
license: cc-by-sa-4.0
Dataset Description
- Homepage: https://hf.co/datasets/linagora/MMLU_tunisian
- Repository: https://github.com/linagora-labs/lighteval_tunisian_derja
Dataset Summary
MMLU_tunisian is an evaluation benchmark designed to assess large language models' (LLM) performance in Tunisian Derja, a variety of Arabic. It consists of 22,027 multiple-choice questions, translated from selected subsets of the Massive Multitask Language Understanding (MMLU), ArabicMMLU and DarijaMMLU benchmarks to measure model performance on 44 subjects in Darija.
Supported Tasks
- Task Category: Multiple-choice question answering
- Task: Answering multiple-choice questions in Derja
Languages
The dataset is available in Tunisian Arabic (Derja).
Dataset Structure
The dataset consists of 44 folders covering the 44 subjects included in the dataset.
Data Instances
Each data instance of each subject contains a multiple-choice question with 2 to 5 answer options. The structure includes:
- question: The multiple-choice question in Darija.
- context: Additional contextual information that may be useful for answering the question.
- choices: A list of possible answer options.
- answer: The correct answer to the question (0, 1, 2, 3, or 4).
- subject: The subject category for the question.
- subject_darija: The subject category in Darija.
- source: The source from which the question was derived (either MMLU or ArabicMMLU).
Example:
{
"question": "العضوية اللي تمثلها فلسطين في الأمم المتحدة.",
"context": "",
"choices": [
"عضو",
"مراقب",
"دائم",
],
"answer": 1,
"subject": "geography",
"subject_darija": "الجغرافيا",
"source": "arabic_mmlu",
"split": "test"
}
Data Splits
The dataset consists of two main splits: test and development.
Dataset Creation
Curation Rationale
The dataset was created to address the need for high-quality, culturally relevant benchmarks for evaluating language models in Tunisian Derja. By translating and adapting established benchmarks, it allows for consistent evaluation across languages and domains.
Source Data
Initial Data Collection and Normalization
The data was derived from two major benchmarks:
- Massive Multitask Language Understanding (MMLU): A large benchmark for multiple-choice question answering.
- ArabicMMLU: An Arabic version of MMLU.
- DarijaMMLU: An Maroccan Arabic Darija version of MMLU. The selected subsets were translated into Tunisian Derja using facebook/nllb-200-3.3B and CODA normalisation.
Personal and Sensitive Information
The dataset does not contain personal or sensitive information.
Considerations for Using the Data
Social Impact of Dataset
This dataset promotes the development of language models capable of understanding and responding in Tunisian Derja, contributing to the advancement of NLP for underrepresented languages.
Limitations
- The dataset is limited to the topics and domains covered by MMLU and ArabicMMLU.
Additional Information
Dataset Curators
Linagora
Citation Information
@model{linagora2025LLM-tn,
author = {Wajdi Ghezaiel and Jean-Pierre Lorré},
title = {Labess-7b-chat:Tunisian Derja LLM},
year = {2025},
month = {January},
url = {https://huggingface.co/datasets/Wajdi1976/Labess-7b-chat}
}
@article{hendryckstest2021,
title={Measuring Massive Multitask Language Understanding},
author={Dan Hendrycks and Collin Burns and Steven Basart and Andy Zou and Mantas Mazeika and Dawn Song and Jacob Steinhardt},
journal={Proceedings of the International Conference on Learning Representations (ICLR)},
year={2021}
}
@article{koto2024arabicmmlu,
title={Arabicmmlu: Assessing massive multitask language understanding in arabic},
author={Koto, Fajri and Li, Haonan and Shatnawi, Sara and Doughman, Jad and Sadallah, Abdelrahman Boda and Alraeesi, Aisha and Almubarak, Khalid and Alyafeai, Zaid and Sengupta, Neha and Shehata, Shady and others},
journal={arXiv preprint arXiv:2402.12840},
year={2024}
}
@article{shang2024atlaschatadaptinglargelanguage,
title={Atlas-Chat: Adapting Large Language Models for Low-Resource Moroccan Arabic Dialect},
author={Guokan Shang and Hadi Abdine and Yousef Khoubrane and Amr Mohamed and Yassine Abbahaddou and Sofiane Ennadir and Imane Momayiz and Xuguang Ren and Eric Moulines and Preslav Nakov and Michalis Vazirgiannis and Eric Xing},
year={2024},
eprint={2409.17912},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2409.17912},
}