Error Correction of Technical Document Using LLaMA3

The linguistic errors in technical documents can lead to confusion in product usage, which can be particularly problematic in technology-intensive industries such as robotics and automation systems. In this study, we developed a model that automatically corrects linguistic errors in doc- uments by fine-tuning the LLaMA3 model, and we evaluated the model’s performance across various types of errors. A comparison of the model’s performance before and after fine-tuning revealed that the fine-tuned LLaMA3 model performed better in all error categories. Further- more, the fine-tuned LLaMA3 model demonstrated competitive performance compared to ChatGPT, confirming that fine-tuning can significantly enhance performance. The findings of this study provide important insights into the development of automated error correction sys- tems for technical documents. Future research will focus on improving the model’s perfor- mance by utilizing more diverse datasets and exploring its applicability to other types of professional documents, such as academic papers and white papers.