File size: 756 Bytes
583bd82 98ef316 583bd82 3d6a747 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 |
---
datasets:
- oscar
- srwac
- leipzig
- cc100
- hrwac
language:
- hr
- sr
- multilingual
tags:
- masked-lm
widget:
- text: "Ovo je početak <mask>."
license: apache-2.0
---
# Transformer language model for Croatian and Serbian
Trained on 28GB datasets that contain Croatian and Serbian language for one epochs (3 mil. steps).
Leipzig Corpus, OSCAR, srWac, hrWac, cc100-hr and cc100-sr datasets
| Model | #params | Arch. | Training data |
|--------------------------------|--------------------------------|-------|-----------------------------------|
| `Andrija/SRoBERTa-XL` | 80M | Forth | Leipzig Corpus, OSCAR, srWac, hrWac, cc100-hr and cc100-sr (28 GB of text) | |