ixa-ehu
commited on
Commit
•
75ed8dd
1
Parent(s):
845d307
README updated
Browse files- README.md~ +0 -41
README.md~
DELETED
@@ -1,41 +0,0 @@
|
|
1 |
-
---
|
2 |
-
language: eu
|
3 |
-
license: cc-by-4.0
|
4 |
-
tags:
|
5 |
-
- basque
|
6 |
-
- roberta
|
7 |
-
---
|
8 |
-
|
9 |
-
# Roberta-eus Euscrawl base cased
|
10 |
-
|
11 |
-
This is a RoBERTa model for Basque model presented in [Does corpus quality really matter for low-resource languages?](https://arxiv.org/abs/2203.08111). There are several models for Basque using the RoBERTa architecture, which are pre-trained using different corpora:
|
12 |
-
|
13 |
-
- roberta-eus-euscrawl-base-cased: Basque RoBERTa trained on Euscrawl, a corpus created using tailored crawling from Basque sites, and which is distributed under a CC-BY license. EusCrawl It contains 12,528k documents and 423M tokens.
|
14 |
-
- roberta-eus-euscrawl-large-cased: Basque RoBERTa large trained on EusCrawl.
|
15 |
-
- roberta-eus-mC4-base-cased: Basque RoBERTa trained on the Basque portion of mc4 dataset.
|
16 |
-
- roberta-eus-CC100-base-cased: Basque RoBERTa trained on Basque portion of cc100 dataset.
|
17 |
-
|
18 |
-
The models have been tested on five different downstream tasks for Basque: Topic classification, Sentiment analysis, Stance detection, Named Entity Recognition (NER), and Question Answering (refer to the [paper](https://arxiv.org/abs/2203.08111) for more details). See summary of results below:
|
19 |
-
|
20 |
-
|
21 |
-
| Model | Topic class. | Sentiment | Stance det. | NER | QA | Average |
|
22 |
-
|----------------------------------|--------------|-----------|-------------|----------|----------|----------|
|
23 |
-
| roberta-eus-euscrawl-base-cased | 76.2 | 77.7 | 57.4 | 86.8 | 34.6 | 66.5 |
|
24 |
-
| roberta-eus-euscrawl-large-cased | **77.6** | 78.8 | 62.9 | **87.2** | **38.3** | **69.0** |
|
25 |
-
| roberta-eus-mC4-base-cased | 75.3 | **80.4** | 59.1 | 86.0 | 35.2 | 67.2 |
|
26 |
-
| roberta-eus-CC100-base-cased | 76.2 | 78.8 | **63.4** | 85.2 | 35.8 | 67.9 |
|
27 |
-
|
28 |
-
|
29 |
-
If you use any of these models, please cite the following paper:
|
30 |
-
|
31 |
-
```
|
32 |
-
@misc{artetxe2022euscrawl,
|
33 |
-
title={Does corpus quality really matter for low-resource languages?},
|
34 |
-
author={Mikel Artetxe, Itziar Aldabe, Rodrigo Agerri,
|
35 |
-
Olatz Perez-de-Viñaspre, Aitor Soroa},
|
36 |
-
year={2022},
|
37 |
-
eprint={2203.08111},
|
38 |
-
archivePrefix={arXiv},
|
39 |
-
primaryClass={cs.CL}
|
40 |
-
}
|
41 |
-
```
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|