Update README.md
Browse files
README.md
CHANGED
@@ -22,6 +22,19 @@ The current model is based on GreekBERT (https://huggingface.co/nlpaueb/bert-bas
|
|
22 |
The classification head consists of two linear layers of 256 neurons each. The model was trained for 1 epoch as it achieved high accuracy (94%)
|
23 |
without further improvement, likely due to GreekBERT's pre-training across various varieties of Greek.
|
24 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
25 |
**Φόρτωση μοντέλου/ Load Pretrained Model**
|
26 |
|
27 |
```python
|
|
|
22 |
The classification head consists of two linear layers of 256 neurons each. The model was trained for 1 epoch as it achieved high accuracy (94%)
|
23 |
without further improvement, likely due to GreekBERT's pre-training across various varieties of Greek.
|
24 |
|
25 |
+
**Δεδομένα / Resources**
|
26 |
+
|
27 |
+
Χρησιμοποιήσαμε γλωσσικά δεδομένα από το Project Gutenberg, Opus, Europarl, Wiki, HNC, Bible, GlobalVoices, ParaCrawl.
|
28 |
+
Επισημειώθηκαν από την ομάδα των γλωσσολόγων 5020 χωρία με τις τέσσερις ποικιλίες της Ελληνικής Γλώσσας, ΚΝΕ, Δημοτική, Καθαρεύυσα, Αρχαία Ελλήνική.
|
29 |
+
Όσον αφορά στην ετίκετα αρχαία ελληνική επισημειώθηκαν με αυτή όλα τα κείμενα που αφόρουν Κλασική Εποχή και Ελληνιστική Κοινή.
|
30 |
+
Οι παραπάνω ποικιλίες επιλέχθηκαν με γνώμονα την ιστορικότητα της Ελληνικής Γλώσσας.
|
31 |
+
|
32 |
+
We used data from Project Gutenberg, Opus, Europarl, Wiki, HNC, Bible, GlobalVoices, and ParaCrawl.
|
33 |
+
A team of linguists annotated 5,020 fragmentum across the four varieties of the Greek language: Modern Greek, Demotic, Katharevousa, and Ancient Greek.
|
34 |
+
Regarding the label ‘Ancient Greek,’ this was applied to texts from the Classical and Hellenistic periods.
|
35 |
+
These varieties were selected with an emphasis on the historical continuity of the Greek language.
|
36 |
+
|
37 |
+
|
38 |
**Φόρτωση μοντέλου/ Load Pretrained Model**
|
39 |
|
40 |
```python
|