Search:
Match:
2 results

Korean Legal Reasoning Benchmark for LLMs

Published:Dec 31, 2025 02:35
1 min read
ArXiv

Analysis

This paper introduces a new benchmark, KCL, specifically designed to evaluate the legal reasoning abilities of LLMs in Korean. The key contribution is the focus on knowledge-independent evaluation, achieved through question-level supporting precedents. This allows for a more accurate assessment of reasoning skills separate from pre-existing knowledge. The benchmark's two components, KCL-MCQA and KCL-Essay, offer both multiple-choice and open-ended question formats, providing a comprehensive evaluation. The release of the dataset and evaluation code is a valuable contribution to the research community.
Reference

The paper highlights that reasoning-specialized models consistently outperform general-purpose counterparts, indicating the importance of specialized architectures for legal reasoning.

Analysis

This paper investigates the use of scaled charges in force fields for modeling NaCl and KCl in water. It evaluates the performance of different scaled charge values (0.75, 0.80, 0.85, 0.92) in reproducing various experimental properties like density, structure, transport properties, surface tension, freezing point depression, and maximum density. The study highlights that while scaled charges improve the accuracy of electrolyte modeling, no single charge value can perfectly replicate all properties. This suggests that the choice of scaled charge depends on the specific property of interest.
Reference

The use of a scaled charge of 0.75 is able to reproduce with high accuracy the viscosities and diffusion coefficients of NaCl solutions by the first time.