uomnf97's picture
Update README.md
0e5dc5f
|
raw
history blame
1.08 kB
metadata
language: ko
tags:
  - korean
  - klue
  - korquad
metrics:
  - exact_match
  - f1
library_name: transformers

๐Ÿง‘๐Ÿปโ€๐Ÿ’ป KLUE RoBERTa Large

  • ์ด ๋ชจ๋ธ์€ klue/roberta-large๋ฅผ ํ•œ๊ตญ์–ด Machine Reading Comprehension๋ฅผ ์œ„ํ•ด KorQuAD ๋ฐ์ดํ„ฐ 2.1 version 27,423๊ฐœ์˜ ๋ฐ์ดํ„ฐ๋ฅผ ํ•™์Šต์‹œ์ผœ ๋งŒ๋“  ๋ชจ๋ธ์ž…๋‹ˆ๋‹ค.

๐Ÿ“ What Should Know

  • KorQuAD v2.1์˜ ์›๋ณธ ๋ฐ์ดํ„ฐ๊ฐ€ ์•„๋‹Œ ํ•˜์ดํผ๋งํฌ, ํƒœ๊ทธ, ์œ ๋‹ˆ์ฝ”๋“œ BOM๋ฅผ ์ œ๊ฑฐํ•˜์—ฌ ์ „์ฒ˜๋ฆฌ๋ฅผ ํ•˜์˜€๊ณ , context ๊ธธ์ด๊ฐ€ 7500์ด ๋„˜์–ด๊ฐ„ ๋ฐ์ดํ„ฐ๋“ค์€ ์ œ์™ธํ•˜์—ฌ 27,423๊ฐœ์˜ ๋ฐ์ดํ„ฐ์…‹์„ ์ด์šฉํ•˜์—ฌ ํ•™์Šต์‹œ์ผฐ์Šต๋‹ˆ๋‹ค.
  • ์›๋ณธ ๋ฐ์ดํ„ฐ ๋งํฌ : https://korquad.github.io/

๐Ÿ“ Getting Started

from transformers import AutoConfig, AutoModelForQuestionAnswering, AutoTokenizer

config = AutoConfig.from_pretrained('uomnf97/klue-roberta-finetuned-korquad-v2')
tokenizer = AutoTokenizer.from_pretrained('uomnf97/klue-roberta-finetuned-korquad-v2')
model = AutoModelForQuestionAnswering.from_pretrained('uomnf97/klue-roberta-finetuned-korquad-v2',config=config)