You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this dataset content.

ko-ged-mc-2501-elementary

ko-ged-mc-2501-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 first GED exam of 2025. All exam items are sourced from the official archives managed by the Korean Ministry of Education.

Subject Distribution

Subject Count
Ethics 20
Korean 20
Practical Arts 20
Social Studies 19
Science 19
English 18
Art 16
Math 16
Physical Education 14
Music 12

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": "์‚ฌํšŒ",
    "Question": "ใ‰ ์— ๋“ค์–ด๊ฐˆ ๊ธฐ๊ด€์œผ๋กœ ์•Œ๋งž์€ ๊ฒƒ์€?\nโ—ฆ ์งˆ๋ฌธ: ๋ถ€๋ชจ๋‹˜๊ป˜์„œ ๋™์ƒ์˜ ์ถœ์ƒ ์‹ ๊ณ ๋ฅผ ํ•˜๋ ค๊ณ  ํ•ด. ์–ด๋–ค ๊ณต๊ณต๊ธฐ๊ด€์— ๊ฐ€๋Š” ๊ฒƒ์ด ๊ฐ€์žฅ ์ข‹์„๊นŒ?\nโ—ฆ ๋‹ต๋ณ€: ์งˆ๋ฌธ์— ํ•ด๋‹นํ•˜๋Š” ๊ณต๊ณต๊ธฐ๊ด€์€ ใ‰ ์ž…๋‹ˆ๋‹ค. ์ด๊ณณ์—์„œ๋Š” ์ถœ์ƒ ์‹ ๊ณ ๋ฅผ ๋น„๋กฏํ•˜์—ฌ ์ฃผ๋ฏผ๋“ฑ๋ก์ฆ ๋ฐœ๊ธ‰ ๋“ฑ ์ฃผ๋ฏผ๋“ค์˜ ์ƒํ™œ์— ํ•„์š”ํ•œ ๊ฒƒ๋“ค์„ ์ฒ˜๋ฆฌํ•ฉ๋‹ˆ๋‹ค.",
    "Correct Answer": "ํ–‰์ •๋ณต์ง€์„ผํ„ฐ",
    "Incorrect Answer 1": "๊ฒฝ์ฐฐ์„œ",
    "Incorrect Answer 2": "๋„์„œ๊ด€",
    "Incorrect Answer 3": "์†Œ๋ฐฉ์„œ"
  }
]
Downloads last month
17

Collection including davidkim205/ko-ged-2501-mc-elementary