[2405.10523] Adaptable and Reliable Text Classification using Large Language Models

View a PDF of the paper titled Adaptable and Dependable Textual content Classification utilizing Giant Language Fashions, by Zhiqiang Wang and three different authors

View PDF
HTML (experimental)

Summary:Textual content classification is prime in Pure Language Processing (NLP), and the appearance of Giant Language Fashions (LLMs) has revolutionized the sphere. This paper introduces an adaptable and dependable textual content classification paradigm, which leverages LLMs because the core element to deal with textual content classification duties. Our system simplifies the normal textual content classification workflows, decreasing the necessity for intensive preprocessing and domain-specific experience to ship adaptable and dependable textual content classification outcomes. We evaluated the efficiency of a number of LLMs, machine studying algorithms, and neural network-based architectures on 4 various datasets. Outcomes reveal that sure LLMs surpass conventional strategies in sentiment evaluation, spam SMS detection, and multi-label classification. Moreover, it’s proven that the system’s efficiency may be additional enhanced by way of few-shot or fine-tuning methods, making the fine-tuned mannequin the highest performer throughout all datasets. Supply code and datasets can be found on this GitHub repository: this https URL.