Fine-tuned KLUE-BERT outperforms GPT-4 for Korean legal text classification

A new study published on arXiv evaluates various AI models for classifying Korean sexual offense cases, finding that fine-tuned smaller models like KLUE-BERT outperform larger, general-purpose models such as GPT-3.5 and GPT-4.0. KLUE-BERT achieved a 99.3% accuracy rate, demonstrating the effectiveness of domain adaptation for legal text classification. The research also utilized explainable AI (XAI) techniques to analyze model predictions and identify linguistic features influencing decisions, highlighting the need for both performance and interpretability in legal AI applications. AI

IMPACT Highlights the importance of domain-specific fine-tuning over raw model size for specialized AI tasks like legal text classification.

RANK_REASON The cluster contains an academic paper detailing AI model performance evaluation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• alphaXiv
• arXiv
• CatalyzeX
• Connected Papers
• DagsHub
• Gotit.pub
• GPT-3.5
• GPT-4.0
• Hugging Face
• KLUE-BERT
• Litmaps
• ScienceCast
• scite Smart Citations

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →