2 days ago
-
Shorten licensing description in dataset cards
-
License dataset compilation under ODC-By 1.0
-
Publish English and Russian dataset cards
3 days ago
-
Refresh unfiltered and replace all filtered versions with current filtered/train
6 days ago
-
Refresh unfiltered and replace all filtered versions with current filtered/train
20 days ago
-
Add filtered3/train: three audited cleanup cycles and publisher-aware paragraph LID
-
Add filtered2/train pilot: boilerplate cleanup, verified lexical near-dedup and full fragment LID
-
Add conservative filtered/train with dedup lineage and per-language before/after tokens
-
Add shuffled unfiltered/train corpus with provenance, language ID and Gemma 4 counts
-
initial commit
