Year: 2025
Type: article
Abstract: Automatic code comment generation is an emerging area of research at the intersection of software engineering and natural language processing. This work presents the design and implementation of an AI-driven comment generation system fine-tuned on the CodeT5-base transformer architecture, tailored for Python code. The approach leverages recent advances in sequence-to-sequence modeling to produce concise, contextually relevant, and semantically accurate comments in real time. By addressing key challenges in preprocessing, training optimization, and model integration, the system achieves high-quality outputs with fast inference performance. This study demonstrates the potential of transformer-based models to enhance software doc umentation, support code comprehension, and improve developer productivity in modern software engineering workflows.
Authors: FARHAN A, Vishnu C
Institution: Sacred Heart College
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4415480719