Aya Expanse: Combining Research Breakthroughs for a New Multilingual Frontier
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866929616931782656 |
|---|---|
| author | Dang, John Singh, Shivalika D'souza, Daniel Ahmadian, Arash Salamanca, Alejandro Smith, Madeline Peppin, Aidan Hong, Sungjin Govindassamy, Manoj Zhao, Terrence Kublik, Sandra Amer, Meor Aryabumi, Viraat Campos, Jon Ander Tan, Yi-Chern Kocmi, Tom Strub, Florian Grinsztajn, Nathan Flet-Berliac, Yannis Locatelli, Acyr Lin, Hangyu Talupuru, Dwarak Venkitesh, Bharat Cairuz, David Yang, Bowen Chung, Tim Ko, Wei-Yin Shi, Sylvie Shang Shukayev, Amir Bae, Sammie Piktus, Aleksandra Castagné, Roman Cruz-Salinas, Felipe Kim, Eddie Crawhall-Stein, Lucas Morisot, Adrien Roy, Sudip Blunsom, Phil Zhang, Ivan Gomez, Aidan Frosst, Nick Fadaee, Marzieh Ermis, Beyza Üstün, Ahmet Hooker, Sara |
| author_facet | Dang, John Singh, Shivalika D'souza, Daniel Ahmadian, Arash Salamanca, Alejandro Smith, Madeline Peppin, Aidan Hong, Sungjin Govindassamy, Manoj Zhao, Terrence Kublik, Sandra Amer, Meor Aryabumi, Viraat Campos, Jon Ander Tan, Yi-Chern Kocmi, Tom Strub, Florian Grinsztajn, Nathan Flet-Berliac, Yannis Locatelli, Acyr Lin, Hangyu Talupuru, Dwarak Venkitesh, Bharat Cairuz, David Yang, Bowen Chung, Tim Ko, Wei-Yin Shi, Sylvie Shang Shukayev, Amir Bae, Sammie Piktus, Aleksandra Castagné, Roman Cruz-Salinas, Felipe Kim, Eddie Crawhall-Stein, Lucas Morisot, Adrien Roy, Sudip Blunsom, Phil Zhang, Ivan Gomez, Aidan Frosst, Nick Fadaee, Marzieh Ermis, Beyza Üstün, Ahmet Hooker, Sara |
| contents | We introduce the Aya Expanse model family, a new generation of 8B and 32B parameter multilingual language models, aiming to address the critical challenge of developing highly performant multilingual models that match or surpass the capabilities of monolingual models. By leveraging several years of research at Cohere For AI and Cohere, including advancements in data arbitrage, multilingual preference training, and model merging, Aya Expanse sets a new state-of-the-art in multilingual performance. Our evaluations on the Arena-Hard-Auto dataset, translated into 23 languages, demonstrate that Aya Expanse 8B and 32B outperform leading open-weight models in their respective parameter classes, including Gemma 2, Qwen 2.5, and Llama 3.1, achieving up to a 76.6% win-rate. Notably, Aya Expanse 32B outperforms Llama 3.1 70B, a model with twice as many parameters, achieving a 54.0% win-rate. In this short technical report, we present extended evaluation results for the Aya Expanse model family and release their open-weights, together with a new multilingual evaluation dataset m-ArenaHard. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2412_04261 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Aya Expanse: Combining Research Breakthroughs for a New Multilingual Frontier Dang, John Singh, Shivalika D'souza, Daniel Ahmadian, Arash Salamanca, Alejandro Smith, Madeline Peppin, Aidan Hong, Sungjin Govindassamy, Manoj Zhao, Terrence Kublik, Sandra Amer, Meor Aryabumi, Viraat Campos, Jon Ander Tan, Yi-Chern Kocmi, Tom Strub, Florian Grinsztajn, Nathan Flet-Berliac, Yannis Locatelli, Acyr Lin, Hangyu Talupuru, Dwarak Venkitesh, Bharat Cairuz, David Yang, Bowen Chung, Tim Ko, Wei-Yin Shi, Sylvie Shang Shukayev, Amir Bae, Sammie Piktus, Aleksandra Castagné, Roman Cruz-Salinas, Felipe Kim, Eddie Crawhall-Stein, Lucas Morisot, Adrien Roy, Sudip Blunsom, Phil Zhang, Ivan Gomez, Aidan Frosst, Nick Fadaee, Marzieh Ermis, Beyza Üstün, Ahmet Hooker, Sara Computation and Language We introduce the Aya Expanse model family, a new generation of 8B and 32B parameter multilingual language models, aiming to address the critical challenge of developing highly performant multilingual models that match or surpass the capabilities of monolingual models. By leveraging several years of research at Cohere For AI and Cohere, including advancements in data arbitrage, multilingual preference training, and model merging, Aya Expanse sets a new state-of-the-art in multilingual performance. Our evaluations on the Arena-Hard-Auto dataset, translated into 23 languages, demonstrate that Aya Expanse 8B and 32B outperform leading open-weight models in their respective parameter classes, including Gemma 2, Qwen 2.5, and Llama 3.1, achieving up to a 76.6% win-rate. Notably, Aya Expanse 32B outperforms Llama 3.1 70B, a model with twice as many parameters, achieving a 54.0% win-rate. In this short technical report, we present extended evaluation results for the Aya Expanse model family and release their open-weights, together with a new multilingual evaluation dataset m-ArenaHard. |
| title | Aya Expanse: Combining Research Breakthroughs for a New Multilingual Frontier |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2412.04261 |