File size: 2,664 Bytes
ca6a6ea
 
85a8879
 
 
 
ca6a6ea
85a8879
2ae9b05
85a8879
2ae9b05
 
 
 
 
80bff58
2ae9b05
 
 
 
 
 
 
 
 
 
 
 
 
a1b8ea8
 
 
d5a5078
 
2ae9b05
 
 
 
 
be00833
310c312
 
 
 
 
 
be00833
310c312
2ae9b05
 
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
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
---
license: cc-by-4.0
language: eu
tags:
- basque
- bert
---

# ElhBERTeu

This is a BERT model for Basque introduced in [BasqueGLUE: A Natural Language Understanding Benchmark for Basque]().

To train ElhBERTeu, we collected different corpora sources from several domains: updated (2021) national and local news sources, Basque Wikipedia, as well as novel news sources and texts from other domains, such as science (both academic and divulgative), literature or subtitles. More details about the corpora used and their sizes are shown in the following table. Texts from news sources were oversampled (duplicated) as done during the training of BERTeus. In total 575M tokens were used for pre-training ElhBERTeu.

|Domain     | Size     |
|-----------|----------|
|News       | 2 x 224M |
|Wikipedia  | 40M      |
|Science    | 58M      |
|Literature | 24M      |
|Others     | 7M       |
|Total      | 575M     |

ElhBERTeu is a base, uncased monolingual BERT model for Basque, with a vocab size of 50K.

ElhBERTeu was trained following the design decisions for [BERTeus](https://huggingface.co/ixa-ehu/berteus-base-cased). The tokenizer and the hyper-parameter settings remained the same, with the only difference being that the full pre-training of the model (1M steps) was performed with a sequence length of 512 on a v3-8 TPU.

The model has been evaluated on the recently created BasqueGLUE NLU benchmark:

| Model     |  AVG      |  NERC     |  F_intent | F_slot  |  BHTC   |  BEC    |  Vaxx   |  QNLI   |  WiC    | coref   |
|-----------|:---------:|:---------:|:---------:|:-------:|:-------:|:-------:|:-------:|:-------:|:-------:|:-------:|
|           |           |   F1      |    F1     |   F1    |   F1    |   F1    |  MF1    |  acc    |  acc    |  acc    |
| BERTeus   |   73.23   |   81.92   | **82.52** |  74.34  |**78.26**|  69.43  |  59.30  |**74.26**|  70.71  |**68.31**|
| ElhBERTeu | **73.71** | **82.30** |   82.24   |**75.64**|  78.05  |**69.89**|**63.81**|  73.84  |**71.71**|  65.93  |

If you use this model, please cite the following paper:

- G. Urbizu, I. San Vicente, X. Saralegi, R. Agerri, A. Soroa. BasqueGLUE: A Natural Language Understanding Benchmark for Basque. In proceedings of the 13th Language Resources and Evaluation Conference (LREC 2022). June 2022. Marseille, France

```
@inproceedings{urbizu2022basqueglue,
  title={BasqueGLUE: A Natural Language Understanding Benchmark for Basque},
  author={Gorka Urbizu and I{\~n}aki San Vicente and Xabier Saralegi and Rodrigo Agerri and Aitor Soroa},
  booktitle={Proceedings of the 13th Language Resources and Evaluation Conference},
  year={2022}
}
```

License:
CC BY 4.0