Datasets:
project README
Browse files
README.md
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language: [rw]
|
| 3 |
+
tags: [kinyarwanda, text, pretraining]
|
| 4 |
+
---
|
| 5 |
+
# wandaa-kin-traindata
|
| 6 |
+
|
| 7 |
+
Training text for the wandaa-kin-gpt Kinyarwanda model.
|
| 8 |
+
|
| 9 |
+
- kinya_clean.txt, kinya_merged.txt, train_blocks*.npy : first corpus (about 2.9M tokens, mostly textbook text), tokenized with the 8000-token tokenizer map-boy/wandaa-kin-tokenizer
|
| 10 |
+
- v3/clean_docs_v3.jsonl : cleaned documents (12,212 docs; fields: source, text) from kigalitoday, igihe, umuryango, cpt_train, firecrawl and the textbook corpus. Markdown, urls, menus, English/French lines and repeated boilerplate removed; curly apostrophes converted to straight ones
|
| 11 |
+
- v3/corpus_v3_train.npy, v3/corpus_v3_val.npy : 129-token blocks (6.6M train tokens, 0.19M validation tokens; validation split by whole document)
|
| 12 |
+
|
| 13 |
+
Warning: sources include scraped news sites. Check each site's terms before reusing the text beyond research.
|