---
license: cc-by-4.0
task_categories:
- translation
language:
- gl
- es
- eu
- ca
- en
- pt
pretty_name: finetuningMT
size_categories:
- 100K<n<1M
---

]

### Description

This dataset features a multilingual architecture specifically designed to strengthen the grammatical understanding of the Galician-Portuguese system.
In its initial sections, it employs Galician and Portuguese as pivot languages to connect other Iberian languages and English, a structure that prioritizes the learning of Galician-Portuguese linguistic nuances over other language pairs.

The final section incorporates instruction-tuning datasets for translation-related tasks—such as post-editing, gender evaluation, and entity recognition—originally unavailable in Galician. To bridge this gap, these materials were generated through a pipeline combining lexical adaptation and localization from Portuguese, the use of the Apertium symbolic translator, and a final normalization post-editing phase to ensure high-quality, idiomatic Galician.




## Paragraphs

These are a combination of sentence  taken from our [MT corpus] (https://github.com/proxectonos/corpora)


| Dataset            | Langpair | Size |
|--------------------|----------|------|
| gl-ca_paragraphs   | gl-ca    |  10k    |
| es-gl_paragraphs   | es-gl    |  10k    |
| ca-gl_paragraphs   | ca-gl    |  10k    |
| eu-gl_paragraphs   | eu-gl    |  10k    |
| gl-es_paragraphs   | gl-es    |  10k    |
| gl-pt_paragraphs   | gl-pt    |  10k    |
| pt-gl_paragraphs   | pt-gl    |  10k    |
| gl-eu_paragraphs   | gl-eu    |  10k    |
| gl-en_paragraphs   | gl-en    |  10k    |
| en-gl_paragraphs   | en-gl    |  10k    |

## Sentences


10k samples from our [MT corpus] (https://github.com/proxectonos/corpora)

| Dataset   | Langpair | Size |
|-----------|----------|------|
| gl-pt     | gl-pt    |  10k    |
| pt-gl     | pt-gl    |  10k    |
| pt-es     | pt-es    |  10k    |
| gl-es     | gl-es    |  10k    |
| pt-en     | pt-en    |  10k    |
| en-gl     | en-gl    |  10k    |
| es-gl     | es-gl    |  10k    |
| gl-en     | gl-en    |  10k    |

## Towerblock

These datasets are taken from [TowerBlocks](https://huggingface.co/datasets/Unbabel/TowerBlocks-v0.2) The Galician part of the dataset is originally written in Portuguese.
| Dataset Path                              | Langpair / Type | Size |
|------------------------------------------------------------|-----------------|------|
| multiconer2023-transliterate_fastchat     | transliterate   |  854    |
| general_mt_clean-transliterate_fastchat   | transliterate   |  8239    |
| gender-eval-context-transliterate_fastchat| transliterate   |  1666    |
| opus_filtered-transliterate_fastchat      | transliterate   |  2000    |
| apequest-transliterate_fastchat           | transliterate   |  5760   |
## Funding
This work is funded by the Ministerio para la Transformación Digital y de la Función Pública - Funded by EU – NextGenerationEU within the framework of the project Desarrollo de Modelos ALIA. 
Esta publicación del proyecto Desarrollo de Modelos ALIA está financiada por el Ministerio para la Transformación Digital y de la Función Pública y por el Plan de Recuperación, Transformación y Resiliencia – Financiado por la Unión Europea – NextGenerationEU.