- Load dataset using new JSONL format (split="train")
- Reconstruct nested data structure from flattened records
- Add modality key mapping for consistent naming
- Parse JSON strings for characteristics and relationships
This update makes the Space compatible with the migrated dataset
that no longer uses a Python loading script.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>