Sign In

JMed48k: A Multi-Profession Japanese Medical Licensing Benchmark for Vision-Language Model Evaluation

์ž‘์„ฑ์ž
  • Haebom
์นดํ…Œ๊ณ ๋ฆฌ
Empty

์ €์ž

Yue Xun, Junyu Liu, Qian Niu, Xinyi Wang, Zheng Yuan, Zirui Li, Zequn Zhang, Bowen Zhao, Shujun Wang, Irene Li, Kan Hatakeyama-Sato, Yusuke Iwasawa, Yutaka Matsuo

๐Ÿ’ก ๊ฐœ์š”

๋ณธ ๋…ผ๋ฌธ์€ ์ผ๋ณธ ์˜๋ฃŒ ๋ฉดํ—ˆ ์‹œํ—˜์— ์‚ฌ์šฉ๋˜๋Š” ๊ฐ๊ด€์‹ ๋ฌธ์ œ์™€ ์ด๋ฏธ์ง€๋ฅผ ํฌํ•จํ•˜๋Š” 48,862๊ฐœ์˜ ์งˆ๋ฌธ์œผ๋กœ ๊ตฌ์„ฑ๋œ JMed48k์ด๋ผ๋Š” ์ƒˆ๋กœ์šด ๋‹ค์ค‘ ์ง์—… ์˜๋ฃŒ ๋ผ์ด์„ ์‹ฑ ๋ฒค์น˜๋งˆํฌ๋ฅผ ์ œ์•ˆํ•œ๋‹ค. 2005๋…„๋ถ€ํ„ฐ 2025๋…„๊นŒ์ง€ 11๊ฐœ ๊ตญ๊ฐ€ ๋ฉดํ—ˆ ์‹œํ—˜์—์„œ ์ˆ˜์ง‘๋œ ์ด ๋ฒค์น˜๋งˆํฌ๋Š” ํ…์ŠคํŠธ์™€ ์ด๋ฏธ์ง€ ๊ธฐ๋ฐ˜ ์งˆ๋ฌธ์— ๋Œ€ํ•œ ๋น„์ „-์–ธ์–ด ๋ชจ๋ธ์˜ ์„ฑ๋Šฅ์„ ํ‰๊ฐ€ํ•˜๋„๋ก ์„ค๊ณ„๋˜์—ˆ๋‹ค. ์ด๋ฏธ์ง€ ์ œ๊ฑฐ ๊ฐ์‚ฌ ๊ฒฐ๊ณผ, ๋ฒ”์šฉ ๋ชจ๋ธ์€ ์ด๋ฏธ์ง€๋กœ๋ถ€ํ„ฐ ์ƒ๋‹นํ•œ ์ด๋“์„ ์–ป๋Š” ๋ฐ˜๋ฉด, ์˜๋ฃŒ ํŠนํ™” ๋ชจ๋ธ์€ ์‹œ๊ฐ ์ •๋ณด ํ™œ์šฉ๋„๊ฐ€ ๋‚ฎ์œผ๋ฉฐ, ์ด๋ฏธ์ง€ ์ œ๊ฑฐ ํ›„์—๋„ ์ •๋‹ต๋ฅ ์ด ํฌ๊ฒŒ ๋ณ€ํ•˜์ง€ ์•Š๋Š” ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚ฌ๋‹ค.

๐Ÿ”‘ ์‹œ์‚ฌ์  ๋ฐ ํ•œ๊ณ„

โ€ข
์˜๋ฃŒ ๋ถ„์•ผ ๋น„์ „-์–ธ์–ด ๋ชจ๋ธ์˜ ์„ฑ๋Šฅ ํ‰๊ฐ€๋ฅผ ์œ„ํ•œ ํ˜„์‹ค์ ์ด๊ณ  ๋‹ค์ธต์ ์ธ ๋ฒค์น˜๋งˆํฌ๋ฅผ ์ œ๊ณตํ•œ๋‹ค.
โ€ข
์ด๋ฏธ์ง€ ์ •๋ณด๊ฐ€ ๋ชจ๋ธ์˜ ๋‹ต๋ณ€์— ๋ฏธ์น˜๋Š” ์˜ํ–ฅ์„ ์ง์—…๋ณ„๋กœ ์ •๋Ÿ‰์ ์œผ๋กœ ๋ถ„์„ํ•˜๊ณ , ์˜๋ฃŒ ํŠนํ™” ๋ชจ๋ธ์˜ ์‹œ๊ฐ ์ •๋ณด ํ™œ์šฉ ๋Šฅ๋ ฅ์— ๋Œ€ํ•œ ํ†ต์ฐฐ์„ ์ œ์‹œํ•œ๋‹ค.
โ€ข
๋ฒค์น˜๋งˆํฌ์— ํฌํ•จ๋œ ์งˆ๋ฌธ์˜ ์‹ค์ œ ์‹œํ—˜ ๋‚œ์ด๋„ ๋ฐ ์ตœ์‹  ๋™ํ–ฅ ๋ฐ˜์˜์— ๋Œ€ํ•œ ์ง€์†์ ์ธ ๊ฒ€์ฆ ๋ฐ ์—…๋ฐ์ดํŠธ๊ฐ€ ํ•„์š”ํ•˜๋ฉฐ, ๋‹ค์–‘ํ•œ ์–ธ์–ด ๋ฐ ๋ฌธํ™”๊ถŒ ์˜๋ฃŒ ์‹œ์Šคํ…œ์— ๋Œ€ํ•œ ์ผ๋ฐ˜ํ™” ๊ฐ€๋Šฅ์„ฑ์„ ํƒ๊ตฌํ•ด์•ผ ํ•œ๋‹ค.
๐Ÿ‘
Made with Slashpage