README.md

metadata

title: SQLCoder Studio
emoji: ⚡
colorFrom: blue
colorTo: purple
sdk: gradio
app_file: app.py
pinned: false
license: mit

⚡ SQLCoder Studio

A Natural Language → SQL → Results interface powered by defog/sqlcoder-7b-2.

What it does

1. You type a plain-English question about the database
2. SQLCoder-7B-2 generates the SQL query
3. The query executes against a local SQLite database
4. Results are shown as a table (with CSV download)

Sample Database

Table	Description
employees	12 employees with name, dept, salary, hire date
departments	5 departments with budget & location
projects	6 projects with status & budget
sales	10 sales records with product, amount, region

Example Questions

• "Show me all employees in Engineering with salary above 120000"
• "Which department has the highest total salary budget?"
• "Who are the top 3 sales performers by total amount?"
• "List all active projects with their budgets"

Hardware

Runs best on GPU (T4/A10G). CPU inference is possible but slow (~1–2 min per query).