README.md

EECS Cleaned Corpus

Cleaned UC Berkeley EECS crawl data exported from the CS288 assignment repository.

Contents

• data/pages.jsonl: 4848 cleaned pages
• data/chunks.jsonl: 22266 retrieval chunks
• data/qa_live_benchmark.jsonl: 220 benchmark examples

Schema

• pages.jsonl: url, title, text, page_id
• chunks.jsonl: chunk-level retrieval records from the offline build pipeline

This export is produced offline from the repository's cleaned corpus artifacts.