Large Language Models for Text Classification: A New Era of Accuracy and Efficiency
摘要
This study evaluates the effectiveness of large language models (LLMs) in text classification tasks, comparing their performance with earlier models like BART and RoBERTa. We conducted an empirical analysis of various models, including BERT-base-unicase, RoBERTa-base, Mistal-nemo, and Llama3, using metrics such as precision, recall, and F1-score in categories like Business, Entertainment, Politics, Sport, and Tech. Our findings show that Llama3 outperforms other models, achieving an average precision and F1-score of 0.91%, excelling in Sport and Tech with precision scores of 0.97%. These results position Llama3 as a leading choice for text classification tasks.