A new Korean grammar benchmark shows that LLMs score lowest on phonological and pronunciation tasks, and feeding them pronunciation or morpheme annotations reduces that gap.
7https://huggingface.co/yanolja/ EEVE-Korean-10.8B-v1.0 8https://github.com/axolotl-ai-cloud/ axolotl 9https://huggingface.co/LGAI-EXAONE/ EXAONE-3.5-7.8B-Instruct
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Polishing Every Facet of the GEM: Testing Linguistic Competence of LLMs and Humans in Korean
A new Korean grammar benchmark shows that LLMs score lowest on phonological and pronunciation tasks, and feeding them pronunciation or morpheme annotations reduces that gap.