Automated API Docs Generator using Generative AI

Our study provides an improvement on the creation of Application Programming Interfaces (APIs) usage documentation using the efficiency and power of Generative AI. APIs play an important role in software integration and software maintenance but the process of API documentation creation has been traditional and did not evolve with time, this paper employs Generative AI to enhance the accuracy, speed, and scale of API documentation generation. The automated API documentation generator is created using natural language processing applied through a large language model (TinyPixel/Llama-2-7B-bf16-sharded model). Training data was created by applying web scraping on various large tech companies' documentation web pages to get a good quality and industry-standard documentation dataset. It was further diversified and increased using the GPT model to handle a wide range of API scenarios. The fine-tuning greatly enhanced the TinyPixel/Llama-2-7B-bf16-sharded model's efficiency and quality of output which is proven by the reduced response time and the accuracy of documentation generated. Our study's comparative study confirms the effectiveness of the approach used. Our study's conclusion offers a comprehensive approach that should improve software development processes and pave the way for additional developments in API documentation.