We are interested in developing non-English LLMs and transferring knowledge from large language models to smaller ones.

• Development of Solar-based Self-Introduction Correction LLM Model using SFT&DPO 2024.02 ~ 2024.04
• Development of Gemma-2B Korean Pre-trained Model 2024.04 ~ 2024.06
• Application of Quantization to LLM Model using llama.cpp 2024.06 ~ 2024.06
• Development of Korean LLAVA Model using Chat-Vector 2024.06 ~ 2024.07
• Development of Korean Financial LLM Leaderboard 2024.07 ~ 2024.08
• Application of Quantization to LLM Model using TensorRT-LLM&Triton 2024.08 ~ 2024.09
• Development of Gemma2-2B Korean Pre-trained Model(Unsloth Continual Pretrain) 2024.09 ~ 2024.09

• Machine Learning / Deep Learning
• LLM Pretrain / Fine-Tuning
• Quantization / Knowledge Distillation

⚡ Skills ⚡

💻 Programming Language

📚 Stacks

🛠 Tools

• 🥇 2023 인공지능콘텐츠융합창작랩 AI 융합 콘텐츠 공모전(AI+ 콘텐츠 결과 부문) - 1nd placed [overview]
• 🥈 2024 국립국어원 인공지능의 한국어 능력 평가 경진대회 - 대화맥락추론(나) 부분 은상 수상 [overview]
• 🥉 2024 국립국어원 인공지능의 한국어 능력 평가 경진대회 - 일상대화요약(나) 부분 특별상 수상 [overview]
• 2024 데이콘 재정정보 AI 검색 알고리즘 경진대회 - Top 9.5% (34/359) [overview]