Lysine crotonylation (Kcr) is an important post-translational modification (PTM) that plays a crucial role in gene regulation, chromatin remodeling, and cellular metabolism. Dysregulation of Kcr has been associated with various diseases, including cancer, neurodegenerative disorders, and inflammatory conditions. Therefore, accurate prediction of Kcr sites is essential for understanding it’s biological implications and identifying potential therapeutic targets. In this study, we propose CLGPT-Kcr, a novel deep learning model that integrates the GPT-2 architecture and simultaneously performs feature extraction and classification within a unified framework for predicting protein Kcr sites. Experimental results show that CLGPT-Kcr outperforms recent deep learning-based methods on the same dataset. Specifically, our model achieves the highest performance, with ACC of 0.838 and MCC of 0.681, surpassing both Bert-Kcr (with ACC of 0.820 and MCC of 0.640) and Deep-Kcr (with ACC of 0.751 and MCC of 0.516). These results highlight the effectiveness of CLGPT-Kcr in Kcr site prediction, providing a more accurate and reliable tool for biological research and disease studies. To better support scientists in their research, we have made our source code and relevant data publicly available on GitHub at: https://github.com/nuinvtnu/CLGPT_Kcr .