ENERGY-DRINK-LOVE/komt_DPOv3

Our Team

• Youjin Chung
• Jingyeom Kim

Model

Base Model

• davidkim205/komt-solar-10.7b-sft-v5

Hardware and Software

• Hardware: A100 * 8 for training our model
• Deepspeed library & Huggingface TRL Trainer

Dataset

• DPO_dataset
  • 자체 제작 dpo dataset(AI-hub dataset 활용)
  • OpenOrca DPO 등 영어 데이터셋 번역(ENERGY-DRINK-LOVE/translate_share_gpt_dedup_llama_SFT_1024, 자체모델 활용)

Training Method

• DPO

Benchmark

Ko LM Eval Harness

Ko-LLM-Leaderboard

• (240316기준 4등)
• 

Average	Ko-ARC	Ko-HellaSwag	Ko-MMLU	Ko-TruthfulQA	Ko-CommonGen V2
61.20	57.51	70.33	53.34	68.49	56.32