Update README.md
Browse files
README.md
CHANGED
|
@@ -4,5 +4,23 @@ license: cc-by-nc-sa-4.0
|
|
| 4 |
|
| 5 |
This repository contains KenLM models for the Ukrainian language
|
| 6 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
Attribution to the NEWS models:
|
| 8 |
- Chaplynskyi, D. et al. (2021) lang-uk Ukrainian Ubercorpus [Data set]. https://lang.org.ua/uk/corpora/#anchor4
|
|
|
|
| 4 |
|
| 5 |
This repository contains KenLM models for the Ukrainian language
|
| 6 |
|
| 7 |
+
Metrics (tested with acoustic model of wav2vec2-xls-r-300m model):
|
| 8 |
+
|
| 9 |
+
| Model | CER | WER |
|
| 10 |
+
|-|-|-|
|
| 11 |
+
| no LM | 0.0412 | 0.2206 |
|
| 12 |
+
| lm-3gram-50k | 0.0348 | 0.1826 |
|
| 13 |
+
| lm-4gram-50k | 0.0347 | 0.1818 |
|
| 14 |
+
| lm-5gram-50k | 0.0347 | 0.1821 |
|
| 15 |
+
| lm-3gram-100k | 0.031 | 0.1588 |
|
| 16 |
+
| lm-4gram-100k | 0.0308 | 0.1579 |
|
| 17 |
+
| lm-5gram-100k | 0.0308 | 0.1579 |
|
| 18 |
+
| lm-3gram-300k | 0.0261 | 0.1294 |
|
| 19 |
+
| lm-4gram-300k | 0.0261 | 0.1293 |
|
| 20 |
+
| lm-5gram-300k | 0.0261 | 0.1293 |
|
| 21 |
+
| lm-3gram-500k | 0.0248 | 0.1209 |
|
| 22 |
+
| lm-4gram-500k | 0.0247 | 0.1207 |
|
| 23 |
+
| lm-5gram-500k | 0.0247 | 0.1209 |
|
| 24 |
+
|
| 25 |
Attribution to the NEWS models:
|
| 26 |
- Chaplynskyi, D. et al. (2021) lang-uk Ukrainian Ubercorpus [Data set]. https://lang.org.ua/uk/corpora/#anchor4
|