List your company

Semantic Textual Similarity on KorSTS (dev)

84.47Spearman Correlation (100x)KOMBO_Jamo
55.516463.033270.5578.0668Apr 27, 2026

Evaluation results

MethodLinks
KOMBO_JamoTokenization=Jamo, Vocab Size=1702026.0484.47
BERTTokenization=MorSubword, Vocab Size=32K2026.0484.26
BERTTokenization=Character, Vocab Size=2K2026.0483.98
BERTTokenization=Subword, Vocab Size=32K2026.0483.8
KOMBO_CjiTokenization=Cji, Vocab Size=1362026.0483.37
KOMBO_StrokeTokenization=Stroke, Vocab Size=1302026.0482.42
BERTTokenization=Morpheme, Vocab Size=32K2026.0482.19
KOMBO_BTSTokenization=BTS, Vocab Size=1122026.0479.73
BERTTokenization=Jamo, Vocab Size=1702026.0477.22
BERTTokenization=Word, Vocab Size=64K2026.0472.2
BERTTokenization=Stroke, Vocab Size=1302026.0467.99
BERTTokenization=Cji, Vocab Size=1362026.0463.86
BERTTokenization=BTS, Vocab Size=1122026.0456.63