---
configs:
- config_name: default
  data_files:
  - split: train
    path: train/*.parquet
  - split: validation_ranking
    path: validation_ranking/*.parquet
  - split: test_ranking
    path: test_ranking/*.parquet
---
# DILI record-level V27-general

This level-agnostic release uses current V10 physical evidence and Gold-v1
parent-disjoint validation/test queries. Training pairs stay inside an intrinsic
V10 assay bucket and cross normalized parents. Gold level is optional provenance,
not a pairing or calibration boundary.

- support gate: at least 10 records and 4 parents
- variance ratio: undefined allowed; finite values above 0.75 excluded
- train record-role degree cap: 96
- ID/OOD validation and test query cap: 500 per regime
- ID ranking cap: 50; OOD ranking cap: 75
- display: semantic categorical adapter where defined, otherwise canonical V10 pair

Rows: {"test_ranking": {"files": ["part-00000.parquet"], "rows": 3027}, "train": {"files": ["part-00000.parquet"], "rows": 123958}, "validation_ranking": {"files": ["part-00000.parquet"], "rows": 3294}}