File size: 4,716 Bytes
96dcb13 2106d2c 96dcb13 31f4dcf 96dcb13 |
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 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 |
---
license: mit
language:
- gl
- es
- en
- eu
- ca
metrics:
- bleu average (Flores): 25.1
---
**English text [here](https://huggingface.co/proxectonos/NOS-MT-OpenNMT-gl-es/blob/main/README_English.md)**
**Model Description**
Model created with OpenNMT-py 3.2 for the languages of the Kingdom of Spain and English using a transformer architecture. The model was converted to the ctranslate2 format.
**How to Use This Model**
+ Install [Python 3.9](https://www.python.org/downloads/release/python-390/)
+ Install [ctranslate 3.2](https://github.com/OpenNMT/CTranslate2)
+ Install subword_nmt:
```bash
pip install subword-nmt
```
+ Translate an input.txt using the model with the following commands:
```bash
subword_nmt.apply_bpe -c ./bpe/mnmt_25.bpe < input.txt > input.bpe
```
```bash
python3 translate.py ./ct2-multi input.bpe > output.txt
```
```bash
': sed -i 's/@@ //g' output.txt
```
**Training**
In the training, we used authentic and synthetic corpora from the [ProxectoNós](https://github.com/proxectonos/corpora). The former are corpora of translations made directly by human translators. It is important to note that despite these texts being made by humans, they are not free from linguistic errors. The latter are corpora that contain Spanish or Portuguese, which we translated into Galician through automatic translation with Opentrad/Apertium and transliteration for out-of-vocabulary words.
**Training Procedure**
+ Tokenization of the datasets was done with the tokenizer (tokenizer.pl) from [linguakit](https://github.com/citiususc/Linguakit) which was modified to avoid line breaks per token from the original file.
+ The BPE vocabulary for the models was generated through the [learn_bpe.py](https://github.com/OpenNMT/OpenNMT-py/blob/master/tools/learn_bpe.py) script from OpenNMT.
**Evaluation**
The BLEU evaluation of the models is done with a mix of internally developed tests (gold1, gold2, test-suite) and other available datasets in Galician (Flores).
| ca-en | ca-es | ca-eu | ca-gl | en-ca | en-es | en-eu | en-gl | es-ca | es-en | es-eu | eu-es | es-gl | eu-ca | eu-en | eu-gl | gl-ca | gl-en | gl-es | gl-eu | AVERAGE |
|----------|----------|----------|----------|----------|----------|----------|----------|----------|-----------|-----------|-----------|-----------|-----------|-----------|-----------|-----------|-----------|-----------|-----------|-----------|
| 39.6 | 23.6 | 16.3 | 30.9 | 39.8 | 24.5 | 18.6 | 31.9 | 22.9 | 24.6 | 13.0 | 17.8 | 22.0 | 23.0 | 26.1 | 21.5 | 30.9 | 34.9 | 23.7 | 16.4 | 25.1 |
**Model Licenses**
MIT License
Copyright (c) 2023 Proxecto Nós
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
**Funding**
This research was funded by the project "Nós: Galician in the society and economy of artificial intelligence", resulting from an agreement between the Xunta de Galicia and the University of Santiago de Compostela, which resulted in the ED431G2019/04 grant from the Consellería de Educación, Universidade e Formación Profesional da Galiza, and by the European Regional Development Fund (ERDF/FEDER program), and Reference Groups: ED431C 2020/21.
**Cite This Work**
If you use this model in your work, please cite it as follows:
Daniel Bardanca Outeirinho, Pablo Gamallo Otero, Iria de-Dios-Flores, and José Ramom Pichel Campos. 2024.
Exploring the effects of vocabulary size in neural machine translation: Galician as a target language.
In Proceedings of the 16th International Conference on Computational Processing of Portuguese, pages 600–604,
Santiago de Compostela, Galiza. Association for Computational Linguistics.
|