Aya Expanse:融合研究突破,开拓多语言新前沿
Aya Expanse: Combining Research Breakthroughs for a New Multilingual Frontier
AI总结:
本文提出 8B/32B 参数的 Aya Expanse 多语言模型,融合数据套利、偏好训练与模型融合,在 23 种语言评测中超越同规模开放权重模型及更大参数模型,并开放权重和 m-ArenaHard 数据集。
摘要:
我们推出 Aya Expanse 模型系列,这是包含 8B 和 32B 参数的新一代多语言语言模型,旨在应对一项关键挑战:开发性能卓越、能力匹敌甚至超越单语言模型的多语言模型。借助 Cohere For AI 和 Cohere 多年来的研究积累,包括数据套利、多语言偏好训练和模型融合方面的进展,Aya Expanse 在多语言性能上树立了新的最优水平。我们在被翻译为 23 种语言的 Arena-Hard-Auto 数据集上进行评估,结果表明 Aya Expanse 8B 和 32B 在各自参数类别中优于领先的开放权重模型,包括 Gemma 2、Qwen 2.5 和 Llama 3.1,胜率最高达到 76.6%。值得注意的是,Aya Expanse 32B 优于参数量为其两倍的 Llama 3.1 70B,取得 54.0% 的胜率。在这份简短技术报告中,我们展示了 Aya Expanse 模型系列的扩展评估结果,并发布其开放权重,同时推出新的多语言评估数据集 m-ArenaHard。
英文摘要:
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.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。