---
license: cc-by-nc-sa-4.0
task_categories:
  - text-classification
  - question-answering
language:
  - ar
tags:
  - MMLU
  - reading-comprehension
  - commonsense-reasoning
  - capabilities
  - cultural-understanding
  - world-knowledge
pretty_name: 'AraDiCE -- Arabic Dialect and Cultural Evaluation'
size_categories:
  - 10K<n<100K
dataset_info:
- config_name: ArabicMMLU-lev
  splits:
    - name: test
      num_examples: 14455
- config_name: ArabicMMLU-lev
  splits:
    - name: test
      num_examples: 14455
- config_name: PIQA-msa
  splits:
    - name: test
      num_examples: 1838
- config_name: PIQA-lev
  splits:
    - name: test
      num_examples: 1838
- config_name: PIQA-egy
  splits:
    - name: test
      num_examples: 1838
- config_name: OBQA-msa
  splits:
    - name: test
      num_examples: 497
- config_name: OBQA-lev
  splits:
    - name: test
      num_examples: 497
- config_name: OBQA-egy
  splits:
    - name: test
      num_examples: 497
- config_name: Winogrande-msa
  splits:
    - name: test
      num_examples: 1267
- config_name: Winogrande-lev
  splits:
    - name: test
      num_examples: 1267
- config_name: Winogrande-egy
  splits:
    - name: test
      num_examples: 1267
- config_name: TruthfulQA-msa
  splits:
    - name: test
      num_examples: 780
- config_name: TruthfulQA-lev
  splits:
    - name: test
      num_examples: 780
- config_name: TruthfulQA-egy
  splits:
    - name: test
      num_examples: 780
- config_name: BoolQ-msa
  splits:
    - name: test
      num_examples: 892
- config_name: BoolQ-lev
  splits:
    - name: test
      num_examples: 892
- config_name: BoolQ-egy
  splits:
    - name: test
      num_examples: 892
- config_name: BoolQ-eng
  splits:
    - name: test
      num_examples: 892
- config_name: AraDiCE-Culture-glf
  splits:
    - name: test
      num_examples: 30
- config_name: AraDiCE-Culture-lev
  splits:
    - name: test
      num_examples: 120
- config_name: AraDiCE-Culture-egy
  splits:
    - name: test
      num_examples: 30      
configs:
- config_name: ArabicMMLU-egy
  data_files:
    - split: test
      path: ArabicMMLU_egy/test.json
- config_name: ArabicMMLU_lev
  data_files:
    - split: test
      path: ArabicMMLU_lev/test.json
- config_name: PIQA-msa
  data_files:
    - split: test
      path: PIQA_msa/validation.json
- config_name: PIQA-lev
  data_files:
    - split: test
      path: PIQA_lev/validation.json
- config_name: PIQA-egy
  data_files:
    - split: test
      path: PIQA_egy/validation.json
- config_name: OBQA-msa
  data_files:
    - split: test
      path: OBQA_msa/test.json
- config_name: OBQA-lev
  data_files:
    - split: test
      path: OBQA_lev/test.json
- config_name: OBQA-egy
  data_files:
    - split: test
      path: OBQA_egy/test.json
- config_name: Winogrande-msa
  data_files:
    - split: test
      path: Winogrande_msa/dev.json
- config_name: Winogrande-lev
  data_files:
    - split: test
      path: Winogrande_lev/dev.json
- config_name: Winogrande-egy
  data_files:
    - split: test
      path: Winogrande_egy/dev.json
- config_name: TruthfulQA-msa
  data_files:
    - split: test
      path: TruthfulQA_msa/test.json
- config_name: TruthfulQA-lev
  data_files:
    - split: test
      path: TruthfulQA_lev/test.json
- config_name: TruthfulQA-egy
  data_files:
    - split: test
      path: TruthfulQA_egy/test.json
- config_name: BoolQ-msa
  data_files:
    - split: test
      path: BoolQ_msa/validation.json
- config_name: BoolQ-lev
  data_files:
    - split: test
      path: BoolQ_lev/validation.json
- config_name: BoolQ-egy
  data_files:
    - split: test
      path: BoolQ_egy/validation.json
- config_name: BoolQ-eng
  data_files:
    - split: test
      path: BoolQ_eng/validation.json
- config_name: AraDiCE-Culture-glf
  data_files:
    - split: test
      path: AraDiCE-Culture_glf/test.json
- config_name: AraDiCE-Culture-lev
  data_files:
    - split: test
      path: AraDiCE-Culture_lev/test.json
- config_name: AraDiCE-Culture-egy
  data_files:
    - split: test
      path: AraDiCE-Culture_egy/test.json
---

# AraDiCE: Benchmarks for Dialectal and Cultural Capabilities in LLMs

## Overview

The **AraDiCE** dataset is designed to evaluate dialectal and cultural capabilities in large language models (LLMs). The dataset consists of post-edited versions of various benchmark datasets, curated for validation in cultural and dialectal contexts relevant to Arabic.

As part of the supplemental materials, we have selected a few datasets (see below) for the reader to review. We will make the full AraDiCE benchmarking suite publicly available to the community.

<!-- ## File/Directory

TO DO:

- **licenses_by-nc-sa_4.0_legalcode.txt** License information.
- **README.md** This file. -->
## Dataset Statistics

The datasets used in this study include: *i)* four existing Arabic datasets for understanding and generation: *Arabic Dialects Dataset (ADD)*, *ADI*, *QADI*, along with a dialectal response generation dataset, and *MADAR*; *ii)* seven datasets translated and post-edited into MSA and dialects (Levantine and Egyptian), which include *ArabicMMLU*, *BoolQ*, *PIQA*, *OBQA*, *Winogrande*, *Belebele*, and *TruthfulQA*; and *iii)* *AraDiCE-Culture*, an in-house developed regional Arabic cultural understanding dataset. Please find below the types of dataset and their statistics benchmarked in **AraDiCE**.

<p align="left"> <img src="./benchmarking_tasks_datasets.png" style="width: 40%;" id="title-icon"> </p>

<p align="left"> <img src="./data_stat_table.png" style="width: 40%;" id="title-icon"> </p>


## Dataset Usage

The AraDiCE dataset is intended to be used for benchmarking and evaluating large language models, specifically focusing on:

- Assessing the performance of LLMs on Arabic-specific dialect and cultural specifics.
- Dialectal variations in the Arabic language.
- Cultural context awareness in reasoning.

## Evaluation
We have used [lm-harness](https://github.com/EleutherAI/lm-evaluation-harness) eval framework to for the benchmarking. We will soon release them. Stay tuned!!

## Machine Translation Models
We will soon be releasing all our *machine translation models*. Stay tuned! For early access, feel free to contact us.



## License

The dataset is distributed under the **Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License (CC BY-NC-SA 4.0)**. The full license text can be found in the accompanying `licenses_by-nc-sa_4.0_legalcode.txt` file.


## Citation
Please find the paper <a href="https://arxiv.org/pdf/2409.11404" target="_blank" style="margin-right: 15px; margin-left: 10px">here.</a>

```
@article{mousi2024aradicebenchmarksdialectalcultural,
      title={{AraDiCE}: Benchmarks for Dialectal and Cultural Capabilities in LLMs},
      author={Basel Mousi and Nadir Durrani and Fatema Ahmad and Md. Arid Hasan and Maram Hasanain and Tameem Kabbani and Fahim Dalvi and Shammur Absar Chowdhury and Firoj Alam},
      year={2024},
      publisher={arXiv:2409.11404},
      url={https://arxiv.org/abs/2409.11404},
}
```