Aya Model: An Instruction Finetuned Open-Access Multilingual Language Model
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866910327023599616 |
|---|---|
| author | Üstün, Ahmet Aryabumi, Viraat Yong, Zheng-Xin Ko, Wei-Yin D'souza, Daniel Onilude, Gbemileke Bhandari, Neel Singh, Shivalika Ooi, Hui-Lee Kayid, Amr Vargus, Freddie Blunsom, Phil Longpre, Shayne Muennighoff, Niklas Fadaee, Marzieh Kreutzer, Julia Hooker, Sara |
| author_facet | Üstün, Ahmet Aryabumi, Viraat Yong, Zheng-Xin Ko, Wei-Yin D'souza, Daniel Onilude, Gbemileke Bhandari, Neel Singh, Shivalika Ooi, Hui-Lee Kayid, Amr Vargus, Freddie Blunsom, Phil Longpre, Shayne Muennighoff, Niklas Fadaee, Marzieh Kreutzer, Julia Hooker, Sara |
| contents | Recent breakthroughs in large language models (LLMs) have centered around a handful of data-rich languages. What does it take to broaden access to breakthroughs beyond first-class citizen languages? Our work introduces Aya, a massively multilingual generative language model that follows instructions in 101 languages of which over 50% are considered as lower-resourced. Aya outperforms mT0 and BLOOMZ on the majority of tasks while covering double the number of languages. We introduce extensive new evaluation suites that broaden the state-of-art for multilingual eval across 99 languages -- including discriminative and generative tasks, human evaluation, and simulated win rates that cover both held-out tasks and in-distribution performance. Furthermore, we conduct detailed investigations on the optimal finetuning mixture composition, data pruning, as well as the toxicity, bias, and safety of our models. We open-source our instruction datasets and our model at https://hf.co/CohereForAI/aya-101 |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_07827 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Aya Model: An Instruction Finetuned Open-Access Multilingual Language Model Üstün, Ahmet Aryabumi, Viraat Yong, Zheng-Xin Ko, Wei-Yin D'souza, Daniel Onilude, Gbemileke Bhandari, Neel Singh, Shivalika Ooi, Hui-Lee Kayid, Amr Vargus, Freddie Blunsom, Phil Longpre, Shayne Muennighoff, Niklas Fadaee, Marzieh Kreutzer, Julia Hooker, Sara Computation and Language Recent breakthroughs in large language models (LLMs) have centered around a handful of data-rich languages. What does it take to broaden access to breakthroughs beyond first-class citizen languages? Our work introduces Aya, a massively multilingual generative language model that follows instructions in 101 languages of which over 50% are considered as lower-resourced. Aya outperforms mT0 and BLOOMZ on the majority of tasks while covering double the number of languages. We introduce extensive new evaluation suites that broaden the state-of-art for multilingual eval across 99 languages -- including discriminative and generative tasks, human evaluation, and simulated win rates that cover both held-out tasks and in-distribution performance. Furthermore, we conduct detailed investigations on the optimal finetuning mixture composition, data pruning, as well as the toxicity, bias, and safety of our models. We open-source our instruction datasets and our model at https://hf.co/CohereForAI/aya-101 |
| title | Aya Model: An Instruction Finetuned Open-Access Multilingual Language Model |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2402.07827 |