VivaKitNatural Language to SQL Assistant
Production conversational database assistant translating plain English questions into executable SQLite queries with automated query explanations and read-only safety guards. Powered by local Qwen2.5-1.5B-Instruct under a 100% commercially compliant Apache-2.0 license (zero copyleft restrictions). Evaluated across 30 held-out self-authored benchmark queries on the kit's own college database, achieving 46.67% Execution Accuracy (EX) and 80% SQL syntax validity with sub-second local CPU execution (~0.5 s/query) and zero external API dependencies - an honest score for a 1.5B fully-offline model, with every query shown alongside its explanation for manual verification.
Secure checkout by Dodo Payments (UPI, cards). Your personalised download link is e-mailed within 24 hours. One purchase covers one project team of up to 4 students. See terms and refunds.
Our honest results
| Execution Accuracy (EX) | 46.67% (30 held-out self-authored test queries) |
| Valid SQL Syntax Rate | 80.00% (Executes cleanly against SQLite 3) |
| Exact Match Rate (EM) | 40.00% (Normalized string alignment) |
| CPU Generation Latency | ~525 ms/query (4-thread quantized GGUF inference) |
| SQLite Query Execution | < 1 ms (Near-instantaneous relational execution) |
Evaluated strictly on 30 held-out relational test queries across single-table filters, aggregations, and multi-table joins. Includes 5-table college relational database, Gradio 6 console with live query runner and relational schema visualizer, read-only safety sandbox, automated pytest suite, and 15 viva voce questions on Text-to-SQL, Spider benchmark, schema linking, and Q4_K_M quantization.
What's included
- Readable source codeSmall, commented modules you can explain line by line in your viva.
- Demo appA local web app to show your examiners - runs offline on your laptop.
- Trained model + reference resultsSo you can compare your own runs fairly.
- Learning guide with viva Q&ABackground, design choices explained, limitations and extension ideas.
- One-click setup & testsWindows setup script and tests that prove it works on your PC.
- Dataset download scriptFetches the public data from its original source, with the citation.
Dataset credit
College Academic Relational Database & Benchmark Queries by VivaKit (Public Domain / CC0). Qwen2.5-1.5B-Instruct by Alibaba Cloud (Apache-2.0).
Academic integrity
We sell code and teaching, not finished projects. You must understand, extend and write up the work yourself, acknowledge the kit and cite the dataset. No ready-made reports.