Commit
·
5b794a2
1
Parent(s):
e3ae526
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,62 @@
|
|
| 1 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
license: mit
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
tags:
|
| 3 |
+
- flair
|
| 4 |
+
- token-classification
|
| 5 |
+
- sequence-tagger-model
|
| 6 |
+
language: uk
|
| 7 |
+
model-index:
|
| 8 |
+
- name: flair-uk-pos
|
| 9 |
+
results:
|
| 10 |
+
- task:
|
| 11 |
+
name: POS
|
| 12 |
+
type: token-classification
|
| 13 |
+
metrics:
|
| 14 |
+
- name: POS F Score
|
| 15 |
+
type: f_score
|
| 16 |
+
value: 0.9793
|
| 17 |
+
widget:
|
| 18 |
+
- text: "Президент Володимир Зеленський пояснив, що наразі діалог із режимом Володимира путіна неможливий, адже агресор обрав курс на знищення українського народу. За словами Зеленського цей режим РФ виявляє неповагу до суверенітету і територіальної цілісності України."
|
| 19 |
license: mit
|
| 20 |
---
|
| 21 |
+
|
| 22 |
+
# flair-uk-ner
|
| 23 |
+
|
| 24 |
+
## Model description
|
| 25 |
+
|
| 26 |
+
**flair-uk-pos** is a Flair model that is ready to use for part-of-speech (upos) tagging. It is based on flair embeddings, that I've trained for Ukrainian language (available [here](https://huggingface.co/dchaplinsky/flair-uk-backward) and [here](https://huggingface.co/dchaplinsky/flair-uk-forward)) and has superior performance and a very **small size** (just 72mb!).
|
| 27 |
+
|
| 28 |
+
|
| 29 |
+
Results:
|
| 30 |
+
- F-score (micro) **0.9793**
|
| 31 |
+
- F-score (macro) **0.9275**
|
| 32 |
+
- Accuracy **0.9793**
|
| 33 |
+
|
| 34 |
+
| By class: | precision | recall | f1-score | support |
|
| 35 |
+
|--------------|-----------|--------|----------|---------|
|
| 36 |
+
| NOUN | 0.9857 | 0.9851 | 0.9854 | 4549 |
|
| 37 |
+
| PUNCT | 0.9984 | 1.0000 | 0.9992 | 3097 |
|
| 38 |
+
| ADJ | 0.9772 | 0.9852 | 0.9812 | 1959 |
|
| 39 |
+
| ADP | 0.9956 | 0.9968 | 0.9962 | 1584 |
|
| 40 |
+
| VERB | 0.9891 | 0.9910 | 0.9900 | 1552 |
|
| 41 |
+
| ADV | 0.9630 | 0.9118 | 0.9367 | 714 |
|
| 42 |
+
| CCONJ | 0.9685 | 0.9746 | 0.9715 | 630 |
|
| 43 |
+
| PROPN | 0.9279 | 0.9472 | 0.9375 | 625 |
|
| 44 |
+
| DET | 0.9729 | 0.9698 | 0.9713 | 629 |
|
| 45 |
+
| PRON | 0.9706 | 0.9631 | 0.9669 | 515 |
|
| 46 |
+
| PART | 0.9235 | 0.8693 | 0.8956 | 375 |
|
| 47 |
+
| NUM | 0.9722 | 0.9804 | 0.9763 | 357 |
|
| 48 |
+
| SCONJ | 0.8768 | 0.9577 | 0.9154 | 260 |
|
| 49 |
+
| AUX | 0.8906 | 0.9500 | 0.9194 | 120 |
|
| 50 |
+
| X | 0.9833 | 0.9593 | 0.9712 | 123 |
|
| 51 |
+
| SYM | 1.0000 | 0.7059 | 0.8276 | 17 |
|
| 52 |
+
| INTJ | 0.5556 | 0.5000 | 0.5263 | 10 |
|
| 53 |
+
| accuracy | | | 0.9793 | 17116 |
|
| 54 |
+
| macro avg | 0.9383 | 0.9204 | 0.9275 | 17116 |
|
| 55 |
+
| weighted avg | 0.9794 | 0.9793 | 0.9792 | 17116 |
|
| 56 |
+
|
| 57 |
+
|
| 58 |
+
The model was fine-tuned on the [Ukrainian (UD) corpus](https://universaldependencies.org/treebanks/uk_iu/index.html), released by the [non-profit organization Institute for Ukrainian](https://mova.institute).
|
| 59 |
+
Training code is also available [here](https://github.com/lang-uk/flair-pos).
|
| 60 |
+
|
| 61 |
+
|
| 62 |
+
Copyright: [Dmytro Chaplynskyi](https://twitter.com/dchaplinsky), [lang-uk project](https://lang.org.ua), 2022
|