EuroLLM-9B: Technical Report
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918060343951360 |
|---|---|
| author | Martins, Pedro Henrique Alves, João Fernandes, Patrick Guerreiro, Nuno M. Rei, Ricardo Farajian, Amin Klimaszewski, Mateusz Alves, Duarte M. Pombal, José Boizard, Nicolas Faysse, Manuel Colombo, Pierre Yvon, François Haddow, Barry de Souza, José G. C. Birch, Alexandra Martins, André F. T. |
| author_facet | Martins, Pedro Henrique Alves, João Fernandes, Patrick Guerreiro, Nuno M. Rei, Ricardo Farajian, Amin Klimaszewski, Mateusz Alves, Duarte M. Pombal, José Boizard, Nicolas Faysse, Manuel Colombo, Pierre Yvon, François Haddow, Barry de Souza, José G. C. Birch, Alexandra Martins, André F. T. |
| contents | This report presents EuroLLM-9B, a large language model trained from scratch to support the needs of European citizens by covering all 24 official European Union languages and 11 additional languages. EuroLLM addresses the issue of European languages being underrepresented and underserved in existing open large language models. We provide a comprehensive overview of EuroLLM-9B's development, including tokenizer design, architectural specifications, data filtering, and training procedures. We describe the pre-training data collection and filtering pipeline, including the creation of EuroFilter, an AI-based multilingual filter, as well as the design of EuroBlocks-Synthetic, a novel synthetic dataset for post-training that enhances language coverage for European languages. Evaluation results demonstrate EuroLLM-9B's competitive performance on multilingual benchmarks and machine translation tasks, establishing it as the leading open European-made LLM of its size. To support open research and adoption, we release all major components of this work, including the base and instruction-tuned models, the EuroFilter classifier, and the synthetic post-training dataset. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_04079 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | EuroLLM-9B: Technical Report Martins, Pedro Henrique Alves, João Fernandes, Patrick Guerreiro, Nuno M. Rei, Ricardo Farajian, Amin Klimaszewski, Mateusz Alves, Duarte M. Pombal, José Boizard, Nicolas Faysse, Manuel Colombo, Pierre Yvon, François Haddow, Barry de Souza, José G. C. Birch, Alexandra Martins, André F. T. Computation and Language Artificial Intelligence Machine Learning This report presents EuroLLM-9B, a large language model trained from scratch to support the needs of European citizens by covering all 24 official European Union languages and 11 additional languages. EuroLLM addresses the issue of European languages being underrepresented and underserved in existing open large language models. We provide a comprehensive overview of EuroLLM-9B's development, including tokenizer design, architectural specifications, data filtering, and training procedures. We describe the pre-training data collection and filtering pipeline, including the creation of EuroFilter, an AI-based multilingual filter, as well as the design of EuroBlocks-Synthetic, a novel synthetic dataset for post-training that enhances language coverage for European languages. Evaluation results demonstrate EuroLLM-9B's competitive performance on multilingual benchmarks and machine translation tasks, establishing it as the leading open European-made LLM of its size. To support open research and adoption, we release all major components of this work, including the base and instruction-tuned models, the EuroFilter classifier, and the synthetic post-training dataset. |
| title | EuroLLM-9B: Technical Report |
| topic | Computation and Language Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2506.04079 |