README.md

metadata

language:
  - ko
tags:
  - sentence_transformers
  - cross_encoder
pipeline_tag: text-ranking
library_name: sentence-transformers

Example

from sentence_transformers import CrossEncoder

model = CrossEncoder('ddobokki/electra-small-sts-cross-encoder')
model.predict(["그녀는 행복해서 웃었다.", "그녀는 웃겨서 눈물이 났다."])
-> 0.8206561

Dataset

• KorSTS
  • Train
  • Test
• KLUE STS
  • Train
  • Test

Performance

Dataset	Pearson corr.	Spearman corr.
KorSTS(test) + KLUE STS(test)	0.8528	0.8504

TODO

Using KLUE 1.1 train, dev data