AIpom at SemEval-2024 Task 8: Detecting AI-produced Outputs in M4

Thus, employing the pipeline of decoder and encoder models proves to be an effective solution. SemEval-2024 Task 8 (Wang et al., 2024a) focuses Additionally, these studies highlight domain on multigenerator, multidomain, and multilingual shift issues, as there is a significant score disparity machine-generated text detection based on the M4 between the development and official evaluation corpus (Wang et al., 2024b). The shared task offers sets. Future efforts should focus on enhancing the three subtasks, which correspond to standard AIpom robustness with respect to the text domain task formulations in the rapidly developing field and text generator. The codebase and models are of artificial text detection (Jawahar et al., 2020; publicly released