Spaces:
Runtime error
Runtime error
Commit History
-
feat(p2): add download_data.sh + gitignore data/
sourabh gupta- scripts/download_data.sh: idempotent Kaggle CLI download with DATA_URL override - .gitignore: exclude data/ (reproducible via script, not committed) - Inspected 209,527 rows: p99=379 chars, only 0.13% exceed 512-char chunk size → Chunking decision: RecursiveCharacterTextSplitter as safety net only; 99.87% of docs produce exactly 1 chunk (no semantic chunker needed)6e0e5a7