| license: cc-by-sa-4.0 | |
| language: | |
| - ban | |
| task_categories: | |
| - text-generation | |
| # Balinese continued-pretraining corpus (v0) | |
| Cleaned monolingual Balinese for LLM continued-pretraining (FineWeb-2 ban_Latn + Balinese Wikipedia + story/text sets; normalize + quality-filter + exact-dedup). Part of [Open Indonesia Models](https://github.com/timothydillan/open-indonesia-models). | |
| ```json | |
| { | |
| "docs": 29825, | |
| "stats": { | |
| "in": 33670, | |
| "empty": 4, | |
| "rejected": 3539, | |
| "dup": 302, | |
| "kept": 29825 | |
| }, | |
| "per_source": { | |
| "HuggingFaceFW/fineweb-2": 9233, | |
| "wikimedia/wikipedia": 20986, | |
| "pradanaadn/balinese-story-texts-extends": 3338, | |
| "1ndianajones/balinese_story_telling_dataset": 89, | |
| "1ndianajones/balinese_semarandana_dataset": 24 | |
| }, | |
| "approx_words": 10318531 | |
| } | |
| ``` | |