NCSOFT Launches 'VARCO Judge LLM,' South Korea's First LLM Performance Evaluation Model

On September 23, NCSOFT (NC) announced the launch of 'VARCO Judge LLM,' South Korea's first evaluation model designed to assess the performance and capabilities of large language models (LLM).'VARCO Judge LLM' evaluates how quickly and accurately other LLMs perform tasks. Recentl

featured-image

On September 23, NCSOFT (NC) announced the launch of 'VARCO Judge LLM,' South Korea's first evaluation model designed to assess the performance and capabilities of large language models (LLM). 'VARCO Judge LLM' evaluates how quickly and accurately other LLMs perform tasks. Recently, companies have spent a significant amount of time identifying the most suitable LLM models from the diverse array available in the market.

By using 'VARCO Judge LLM,' companies can more effectively verify which LLM models are best suited for their AI services. NC's 'VARCO Judge LLM' is noted for its superior ability to address LLM bias issues and its outstanding performance in the Korean language. It is among the top-performing models in its class.



Notably, NC successfully demonstrated its technological prowess on the global stage by publishing a paper on LLM evaluation models at the prestigious NLP conference, 'EMNLP (Empirical Methods in Natural Language Processing),' this year. Using NC's evaluation model, companies developing AI-based services can swiftly compare and assess the quality of various LLMs, choosing the model best optimized for their services. AI model R&D companies can use the evaluation model to verify their LLM's performance, demonstrate superiority over other models, or quickly identify and address weaknesses.

Model hub operators can accelerate the process of selecting, optimizing, and deploying LLMs to provide more effective models. As the first domestic game company to develop its own language model, NC actively incorporates AI technology throughout its game development process and internal operations. With the release of this evaluation model, the company aims to enhance the quality of its proprietary LLM, 'VARCO,' and establish a leading position in the LLM model evaluation sector.

Lee Yeon-soo, Head of NC Research, stated, "In the rapidly evolving AI market, services that select and apply the optimal models for specific industries are becoming increasingly important. VARCO Judge LLM will establish itself as an essential tool in AI business, extending beyond improving the quality of existing LLM-based services.".