TY  - JOUR
T1  - PEFT Methods for Domain Adaptation
AU  - Jin, Lee You 
AU  - Koo, Yoon Kyung 
AU  - Dam, Chung Woo 
JO  - The Transactions of the Korea Information Processing Society
PY  - 2025
DA  - 2025/1/30
DO  - https://doi.org/10.3745/TKIPS.2025.14.4.239
KW  - Parameter-Efficient Fine-tuning
KW  - Domain Adaptation
KW  - Deep Learning
KW  - Large Language Model
KW  - LoRA
KW  - MoRA
AB  - This study analyzed that the biggest obstacle in deploying Large Language Models (LLMs) in industrial settings is incorporating domain 
specificity into the models. To mitigate this issue, the study compared model performance when domain knowledge was additionally 
trained using MoRA, which enables learning more knowledge information, and LoRA, which is the most common among various PEFT 
methods. Along with this, training time was reduced through securing high-quality data and efficient data loading. The findings of this 
research will provide practical guidelines for developing efficient domain-specific language models with limited computing resources.