Emotion Detection with Transformers: A Comparative Study

arXiv:2403.15454v1 Announce Type: new
Abstract: In this study, we explore the application of transformer-based models for emotion classification on text data. We train and evaluate several pre-trained transformer models, on the Emotion dataset using different variants of transformers. The paper also analyzes some factors that in-fluence the performance of the model, such as the fine-tuning of the transformer layer, the trainability of the layer, and the preprocessing of the text data. Our analysis reveals that commonly applied techniques like removing …

abstract application arxiv classification cs.cl data dataset detection emotion emotion detection explore paper performance stat.ap study text train transformer transformer models transformers type variants