| task_categories: | |
| - text-classification | |
| tags: | |
| - data-use | |
| - gate | |
| - domain | |
| - multi-label | |
| license: cc-by-4.0 | |
| configs: | |
| - config_name: default | |
| data_files: | |
| - split: train | |
| path: encoder_train.jsonl | |
| - split: val | |
| path: encoder_val.jsonl | |
| - split: holdout | |
| path: encoder_holdout.jsonl | |
| # datause-encoder-data | |
| Training data for the data-use encoder: a page-level `has_data` gate plus a | |
| document-level `teratopic` domain classifier, in one joint dataset. | |
| Columns (same schema on every row): | |
| - `task` — `gate` (page-level binary) or `domain` (document-level multi-label) | |
| - `doc_id` — source document id | |
| - `text` — page text (gate) or title+abstract (domain) | |
| - `has_data` — 0/1 for gate rows (0 placeholder on domain rows) | |
| - `labels` — `teratopic` label list for domain rows (empty on gate rows) | |
| Splits are document-id disjoint across both tasks. The 30 teratopic labels | |
| (in column order) are in `encoder_labels.json`. | |