ko-ged-mc-elementary

ko-ged-mc-elementary is a benchmark dataset designed to evaluate the comprehension and reasoning capabilities of large language models (LLMs) in the Korean language. The dataset comprises multiple-choice questions extracted from the General Equivalency Diploma (GED) exams. Specifically, this dataset contains questions from the second GED exam of 2024. All exam items are sourced from the official archives managed by the Korean Ministry of Education.

Subject Distribution

Subject	Count
Ethics	20
Korean	20
Social Studies	17
Physical Education	16
Practical Arts	16
Science	15
Math	12
English	11
Art	10
Music	7

Structure

• id: a unique positive integer identifier for each question
• Domain: subject name
• Question: the question prompt
• Correct Answer: the correct answer
• Incorrect Answer 1, Incorrect Answer 2, Incorrect Answer 3: three incorrect choices

[
  {
    "id": 1,
    "Domain": "영어",
    "Question": "다음 중 모두 소문자로 쓴 낱말은?",
    "Correct Answer": "jump",
    "Incorrect Answer 1": "Hat",
    "Incorrect Answer 2": "RED",
    "Incorrect Answer 3": "Grapes"
  }
]