Code and Results for "A Comparative Study of Incremental vs Prompt-based Training for Code Summarization across Transformer Architectures"
Year: 2026
Type: software
Abstract: Supplementary material for the paper *A Comparative Study of Incremental vs Prompt-based Training for Code Summarization across Transformer Architectures*. This archive contains **both** parts of the study's code: - **`Confirmatory_code/`** — the seed experiment reported in Appendix C. Self-contained and reproducible: the scripts here regenerate Appendix C's table from scratch, and the released prediction and score files allow the reported numbers to be checked without re-running any training. - **`Original_code/`** — the notebooks used for the experiments reported in Section 4 and Table 3, as executed. Released for transparency. **These do not reproduce Table 3's values exactly, and are not presented as doing so**; the reasons are documented in Appendix B of the paper and summarised below. The two are kept separate because they answer different questions. The confirmatory code shows that the reported configuration is stable across random seeds. The notebooks show what was actually run.
Authors: Archana Praveen Kumar, Namratha M, Pratheeksha S, Rashmitha R Bangera, Rajashree Krishna, Ashwini B, Sayed Mohsin Reza
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7208812378