metadata

dataset_info:
  features:
    - name: Source
      dtype: string
    - name: Sentence
      dtype: string
    - name: Topic
      dtype: string
  splits:
    - name: train
      num_bytes: 10696
      num_examples: 100
  download_size: 6725
  dataset_size: 10696
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
license: apache-2.0
task_categories:
  - text-generation
language:
  - ar
pretty_name: CIDAR-EVAL-100
size_categories:
  - n<1K

Dataset Card for "CIDAR-EVAL-100"

CIDAR-EVAL-100

CIDAR-EVAL-100 contains 100 instructions about Arabic culture. The dataset can be used to evaluate an LLM for culturally relevant answers.

📚 Datasets Summary

Name	Explanation
CIDAR	10,000 instructions and responses in Arabic
CIDAR-EVAL-100	100 instructions to evaluate LLMs on cultural relevance
CIDAR-MCQ-100	100 Multiple choice questions and answers to evaluate LLMs on cultural relevance

Category	CIDAR-EVAL-100	CIDAR-MCQ-100
Food&Drinks	14	8
Names	14	8
Animals	2	4
Language	10	20
Jokes&Puzzles	3	7
Religion	5	10
Business	6	7
Cloths	4	5
Science	3	4
Sports&Games	4	2
Tradition	4	10
Weather	4	2
Geography	7	8
General	4	3
Fonts	5	2
Literature	10	2
Plants	3	0
Total	100	100

📋 Dataset Structure

Source(str): Source of the instruction.
Sentence(str): Sentence of the instruction.
Topic(str): Topic covered by the instruction.

📁 Loading The Dataset

You can download the dataset directly from HuggingFace or use the following code:

from datasets import load_dataset
cidar = load_dataset('arbml/CIDAR-EVAL-100')

📄 Sample From The Dataset:

Source: Manual

Sentence: أخبرني عن أشهر أربعة حيوانات في المنطقة

Topic: Animals

🔑 License

The dataset is licensed under Apache-2.0. Apache-2.0.

Citation

@misc{alyafeai2024cidar,
      title={{CIDAR: Culturally Relevant Instruction Dataset For Arabic}}, 
      author={Zaid Alyafeai and Khalid Almubarak and Ahmed Ashraf and Deema Alnuhait and Saied Alshahrani and Gubran A. Q. Abdulrahman and Gamil Ahmed and Qais Gawah and Zead Saleh and Mustafa Ghaleb and Yousef Ali and Maged S. Al-Shaibani},
      year={2024},
      eprint={2402.03177},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}