interndata_merged

# InternData A1 — full merged dataset This repository contains all 38 validated merged InternData A1 schema groups, covering 712,240 episodes. Both real and simulation groups are included. Data is stored in independent uncompressed tar archives with a hard 10 GiB limit. Use `archive_manifest.jsonl`, `unit_manifest.jsonl`, and `SHA256SUMS` for selection and verification. `materialize.py` safely restores a complete group or one episode while preserving the merged directory layout.