Enhancing Database Accessibility : A Llama-2 Powered Interface For Intelligent Database Querying
NEELAM PAWAR, ISHA WAGHULDE, SOHAM NIMBALKAR, PRATHAMESH SURYAWANSHI, Manish Sharma · International Journal For Multidisciplinary Research · 2025
This research work introduces an AI-powered natural language to SQL (NL2SQL) interface to make database querying easy for end-users. Our system, which uses a fine-tuned LLaMA-2 model, generates SQL queries dynamically, with better adaptability and accuracy than rule-based systems. Based on a curated dataset from Hugging Face’s synthetic text-to-SQL corpus, our system maintains robust performance on various query structures. Implemented using Python, PyTorch, and Flask, and deployed using Gunicorn and Tensor Dock, the system exhibits high accuracy in converting natural language to executable SQL. Experimental results demonstrate improvements in query precision, execution efficiency, and deployment scalability. Future development will aim to improve contextual comprehension and real-time query performance.