Datasets:

badblli
/

HOSIA-Intent

Tasks:

Text Classification

Languages:

Size:

License:

Dataset card Files Files and versions

badblli commited on Sep 25

Commit

186c373

·

verified ·

1 Parent(s): 3dc057b

Update v0.3 HF publication status

Files changed (1) hide show

PROGRESS.md +1 -1

PROGRESS.md CHANGED Viewed

@@ -10,7 +10,7 @@ Updated: 2026-09-25
 - One repeated Russian phrasing was rewritten with EVREN and revalidated. The extension generator now recovers successful candidate batches from its ignored audit log after an interrupted run and records actual generator/translator models.
 - Ontology validation passed (40 concepts, 40 relations, 80 lexical mappings); the 2,000-row combined dataset passed the balanced v0.2-intent validator. All records remain `reviewed: false` and `allowed_for_training: false`; AI review is not human review.
 - The public hard-negative set still covers only the original 10 intents and is not an undisclosed benchmark. The new 20-intent dataset draft needs a matching hard-negative review set before model-quality conclusions are broadened.
-- Local output: `data/generated/evren-v0.1/hosia_v0.3_multilingual_ai_reviewed.jsonl` (ignored by Git). The Dataset Card is updated; v0.3 data files still need to be uploaded to HF.
 ## Quality harness update

 - One repeated Russian phrasing was rewritten with EVREN and revalidated. The extension generator now recovers successful candidate batches from its ignored audit log after an interrupted run and records actual generator/translator models.
 - Ontology validation passed (40 concepts, 40 relations, 80 lexical mappings); the 2,000-row combined dataset passed the balanced v0.2-intent validator. All records remain `reviewed: false` and `allowed_for_training: false`; AI review is not human review.
 - The public hard-negative set still covers only the original 10 intents and is not an undisclosed benchmark. The new 20-intent dataset draft needs a matching hard-negative review set before model-quality conclusions are broadened.
+- Local output: `data/generated/evren-v0.1/hosia_v0.3_multilingual_ai_reviewed.jsonl` (ignored by Git). The v0.3 data, 20-intent config, and Dataset Card were published to `badblli/HOSIA-Intent` in HF commit `3dc057b33b9c1a1ad79fd977d87f10dc1d0c3c53`.
 ## Quality harness update