Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
language:
|
3 |
+
- sk
|
4 |
+
tags:
|
5 |
+
- pos
|
6 |
+
license: cc
|
7 |
+
datasets:
|
8 |
+
- universal_dependencies
|
9 |
+
metrics:
|
10 |
+
- accuracy
|
11 |
+
widget:
|
12 |
+
- text: "Kde tá ľudská duša drieme?"
|
13 |
+
---
|
14 |
+
|
15 |
+
|
16 |
+
# POS tagger based on SlovakBERT
|
17 |
+
|
18 |
+
This is a POS tagger based on [SlovakBERT](https://huggingface.co/gerulata/slovakbert). The model uses [Universal POS tagset (UPOS)](https://universaldependencies.org/u/pos/). The model was fine-tuned using Slovak part of [Universal Dependencies dataset](https://universaldependencies.org/) [Zeman 2017] containing 10k manually annotated Slovak sentences.
|
19 |
+
|
20 |
+
## Results
|
21 |
+
|
22 |
+
The model was evaluated in [our paper](https://arxiv.org/abs/2109.15254) [Pikuliak et al 2021, Section 4.2]. It achieves \\(97.84\%\\) accuracy.
|
23 |
+
|
24 |
+
## Cite
|
25 |
+
|
26 |
+
```
|
27 |
+
@article{DBLP:journals/corr/abs-2109-15254,
|
28 |
+
author = {Mat{\'{u}}s Pikuliak and
|
29 |
+
Stefan Grivalsky and
|
30 |
+
Martin Konopka and
|
31 |
+
Miroslav Blst{\'{a}}k and
|
32 |
+
Martin Tamajka and
|
33 |
+
Viktor Bachrat{\'{y}} and
|
34 |
+
Mari{\'{a}}n Simko and
|
35 |
+
Pavol Bal{\'{a}}zik and
|
36 |
+
Michal Trnka and
|
37 |
+
Filip Uhl{\'{a}}rik},
|
38 |
+
title = {SlovakBERT: Slovak Masked Language Model},
|
39 |
+
journal = {CoRR},
|
40 |
+
volume = {abs/2109.15254},
|
41 |
+
year = {2021},
|
42 |
+
url = {https://arxiv.org/abs/2109.15254},
|
43 |
+
eprinttype = {arXiv},
|
44 |
+
eprint = {2109.15254},
|
45 |
+
}
|
46 |
+
```
|