Publications / Domestic Conferences & Theses
Comparative Analysis of Korean Quality in Large-Scale Language Models Based on Zero-Shot Learning
Abstract
Compared the performance of Korean-based LLMs and English-based LLMs on four KoBEST tasks to investigate whether language influences LLM performance. The results confirmed that adding prior knowledge about Korean affects LLM performance.
Key Contribution
Investigated language-specific influences on LLM performance across Korean benchmarks.