Datasets:

rafmacalaba
/

data-use-sft-v2

license: cc-by-4.0
task_categories:
  - text-generation
tags:
  - sft
  - data-mentions
  - provenance
  - usage
  - impact
  - bibtex
configs:
  - config_name: real_v2
    data_files:
      - split: train
        path: real_v2/train.jsonl
      - split: val
        path: real_v2/val.jsonl
      - split: holdout
        path: real_v2/holdout.jsonl

data-use-sft v2 (config: real_v2)

Multitask SFT rows derived from the v2 relabeling. Same ChatML format as rafmacalaba/data-use-sft (real), plus a task field:

provenance — exact-substring {producer, year, geography, acronym}; for tier1∧named mentions the assistant JSON additionally carries bibtex (a single @misc entry assembled from those same metadata fields)
usage_impact — {data_type, usage_action, impact_label, usage_summary}

Splits: val+holdout = dual-pass consensus; train = single-pass (upgraded in v2.1). Only tier1/tier2 spans generate SFT rows; tier3/junk excluded by design.