316usman/sanctions-screening-match
SANCTIONS_SCREENING_MATCH A preference dataset for SANCTIONS_SCREENING_MATCH, harvested from real, human-labelled sources and curated by an automated harvesting harness with an LLM quality gate. Format Standard preference / DPO schema — each row: column meaning prompt the request (originally prompt) chosen the human-preferred response rejected a worse response to the same prompt source the dataset/URL the row was harvested from… See the full description on the dataset page: https://huggingface.co/datasets/316usman/sanctions-screening-match.
SANCTIONSSCREENINGMATCH
A preference dataset for SANCTIONS_SCREENING_MATCH, harvested from real, human-labelled sources and curated by an automated harvesting harness with an LLM quality gate.
Format
Standard preference / DPO schema — each row:
Splits
80/10/10 train / validation / test (seeded shuffle): train:1201 / validation:150 / test:151
Stats
- Rows: 1502
- Distinct sources: 77
Sources
https://law.stackexchange.com/questions/103190https://law.stackexchange.com/questions/103256https://law.stackexchange.com/questions/104418https://law.stackexchange.com/questions/104851https://law.stackexchange.com/questions/105647https://law.stackexchange.com/questions/105745https://law.stackexchange.com/questions/106488https://law.stackexchange.com/questions/106609https://law.stackexchange.com/questions/106690https://law.stackexchange.com/questions/107621https://law.stackexchange.com/questions/108258https://law.stackexchange.com/questions/108537https://law.stackexchange.com/questions/110330https://law.stackexchange.com/questions/110502https://law.stackexchange.com/questions/113683https://law.stackexchange.com/questions/114326https://law.stackexchange.com/questions/114596https://law.stackexchange.com/questions/114622https://law.stackexchange.com/questions/115315https://law.stackexchange.com/questions/115385
Provenance
Each row's chosen/rejected distinction comes from a real human signal (upvotes, accepted answers, ratings, or a real strong-vs-weak reply). Rows passed an automated quality gate checking that chosen is a clean response (not a transcript), the chosen/rejected contrast is about quality (not length), and the row is on-intent.