BUFS_qa / README.md
geumdam's picture
Upload README.md
0047115 verified
|
Raw History Blame Contribute Delete
3.16 kB
metadata
license: cc-by-4.0
language:
  - ko
pretty_name: 부산외국어대학교 안내 Instruction 데이터셋
size_categories:
  - n<1K
task_categories:
  - question-answering
  - text-generation
tags:
  - korean
  - instruction-tuning
  - alpaca
  - university
  - question-answering
configs:
  - config_name: default
    data_files:
      - split: train
        path: train_json_1.json
      - split: test
        path: test_json_1.json
dataset_info:
  features:
    - name: instruction
      dtype: string
    - name: input
      dtype: string
    - name: output
      dtype: string
  splits:
    - name: train
      num_examples: 24
      num_bytes: 9223
    - name: test
      num_examples: 6
      num_bytes: 2288
  download_size: 11511
  dataset_size: 11511

부산외국어대학교 안내 Instruction 데이터셋

부산외국어대학교의 학사, 전공, 국제교류, 입학, 캠퍼스, 장학·취업 안내를 한국어 질의응답으로 정리한 instruction 데이터셋입니다. 각 행은 Alpaca 형식의 instruction, input, output 세 필드로 구성됩니다.

이 데이터는 대학 공식 문서가 아닙니다. 모집 요강, 장학 기준, 학사 일정은 바뀔 수 있으므로 사실 확인은 부산외국어대학교와 입학처 안내를 기준으로 합니다.

파일

파일 split 행 수
train_json_1.json train 24
test_json_1.json test 6

두 파일 모두 JSON 배열입니다. input은 비어 있는 문자열이며, 질문은 instruction에, 답은 output에 있습니다.

{
  "instruction": "질문",
  "input": "",
  "output": "답변"
}

주제

train은 학사 제도와 전공, 수강·학점교류, 교환학생과 해외 프로그램, 수시·정시, 캠퍼스와 학생 생활을 다룹니다. test는 장학금, 어학 자격, 취업 지원, 현장실습, 지역학 지원, 졸업 후 진출 분야를 다룹니다.

불러오기

저장소 루트에 이 README.md와 두 JSON 파일을 함께 올리면, Hub가 위 YAML의 train / test split으로 읽습니다.

from datasets import load_dataset

dataset = load_dataset("사용자명/저장소명")
print(dataset["train"][0]["instruction"])
print(dataset["train"][0]["output"])

로컬 폴더에서 확인할 때는 아래처럼 읽습니다.

from datasets import load_dataset

dataset = load_dataset(
    "json",
    data_files={
        "train": "train_json_1.json",
        "test": "test_json_1.json",
    },
)

업로드

데이터셋 저장소를 만든 뒤, 이 폴더의 README.md, train_json_1.json, test_json_1.json을 저장소 루트에 올립니다. 하위 폴더로 옮기면 카드에 적은 경로와 어긋납니다.

huggingface-cli login
huggingface-cli upload 사용자명/저장소명 . . --repo-type dataset --include "README.md" --include "train_json_1.json" --include "test_json_1.json"

라이선스

CC BY 4.0입니다. 이용 시 출처를 표시합니다.