# Latin (demo) model for [BabyLemmatizer](https://github.com/asahala/BabyLemmatizer)
This is a demo model for Latin based on [PROIEL data](https://github.com/UniversalDependencies/UD_Latin-PROIEL). Model uses alphabetic tokenization and requires BabyLemmatizer 2.1 or newer.

## Evaluation results

```
Neural Net Evaluation
COMPONENT       AVG     CI       MODEL0
POS-tagger      95.38   ±0.00    95.38
Lemmatizer      96.47   ±0.00    96.47
Combined        94.39   ±0.00    94.39
POS-tagger OOV  87.12   ±0.00    87.12
Lemmatizer OOV  82.70   ±0.00    82.70
Combined   OOV  80.28   ±0.00    80.28
-----------------------------------------------
OOV input rate  10.58            10.58

Post-correct Evaluation
COMPONENT       AVG     CI       MODEL0
POS-tagger      95.38   ±0.00    95.38
Lemmatizer      96.49   ±0.00    96.49
Combined        94.42   ±0.00    94.42
POS-tagger OOV  87.12   ±0.00    87.12
Lemmatizer OOV  82.70   ±0.00    82.70
Combined   OOV  80.28   ±0.00    80.28
-----------------------------------------------
OOV input rate  10.58            10.58
```