Background

A Jupyter Notebook provides an interactive way for new users to try PromptCoT's problem generation without dealing with command-line complexity.

Tasks

1. Create notebooks/quickstart.ipynb

The notebook should:

• Install dependencies
• Check GPU availability
• Load sample data from notebooks/sample_concepts.jsonl
• Run problem generation (reuse code from existing scripts)
• Display generated results clearly

2. Create notebooks/sample_concepts.jsonl

Provide minimal sample data for the notebook to process.

3. Update README.md

Add a link to the quickstart notebook in the "Getting Started" section.

Implementation Notes

• Reuse existing functionality from problem_generation.py or related scripts
• If the code isn't easily importable, consider refactoring into callable functions
• Add explanatory comments throughout the notebook for beginners
• Display results using DataFrames or formatted Markdown

Definition of Done

• notebooks/quickstart.ipynb runs without errors from start to finish
• notebooks/sample_concepts.jsonl contains valid sample data
• README.md links to the notebook
• Results are clearly visualized

Happy Hacking! This task is associated with "OceanBase Developer Carnival" event.