schneiderkamplab/dfm12-dolci-pl

dfm12-dolci-pl

Published accepted-only DFM12 subset. Local audit-snapshot fields describe the pre-publication build, not Hub publication status. Only completed kept decisions with all three scores at least 4 are included, after deterministic gates. Automated review is not native-speaker certification. Exclusion metadata contains only IDs/status/errors/scores/reasons, never excluded conversations. Full native messages and explicit assistant target indices are preserved. OPUS pairs produce both directions, linked by parentpairid; both directions share the same pair audit. Training input must use only messages and optional targetmessageindex. Accepted references, stored audit decisions, original provenance and portable attribution are under metadata, never concatenated into training messages. Full excluded records remain only in the separate, local non-upload snapshot, not this package.

Licensing and Attribution

No blanket license is asserted. Preserve the exact per-record source license, URL, title, history and attribution in metadata/audits.jsonl.gz (export_provenance). Local attribution files are bundled in metadata/evidence and referenced by relative paths. Original raw records remain unchanged for traceability. License labels among accepted records:

• —unknown; see source evidence: 284453

Validation

Run python validate_dataset.py using Python 3.9+; only the standard library is required.