10月15日——欧洲物理研究机构欧洲核子研究中心（CERN）与Mistral AI宣布建立长期战略合作伙伴关系，共同开发用于科学发现的通用基础模型。
Oct 15 - European physics research centre CERN and Mistral AI have established a long-term strategic partnership to develop a foundation model for scientific discovery, the two organisations said on Thursday.
根据这份合作意向书，这家法国人工智能企业将得以使用全球规模最大的科学数据集之一，从而在前沿科学领域提升竞争力。
The statement of intent gives the French AI company access to one of the world's largest scientific datasets as it seeks to compete on frontier science.
Mistral首席执行官亚瑟·门施周四在《世界报》发表的文章中指出，前沿人工智能模型“如今已成为科学进步不可或缺的工具”，他还批评了那些将自身技术优势据为己用的企业。
Mistral Chief Executive Arthur Mensch wrote in an article published in Le Monde on Thursday that frontier models were "now indispensable to scientific progress" and criticised companies that keep their advantages to themselves.
欧洲核子研究中心首席信息官恩里卡·波尔卡里表示，此次合作将“推动人工智能技术的进步，使整个物理学界都能从中受益并共享成果”。
CERN Chief Information Officer Enrica Porcari said the new partnership would "advance AI capabilities that can serve the whole physics community and be shared".
双方计划共同打造一款名为“Flare”的大型物理模型。据欧洲核子研究中心介绍，该模型将是首个基于数十亿次模拟粒子碰撞数据训练而成的对撞物理领域基础模型。
The partners aim to build a "Large Physics Model", called "Flare", which CERN said in a statement would be the first foundation model for collider physics, trained on billions of simulated particle collisions.
该中心发言人对路透社表示，这一模型的运作原理与生成式人工智能所依赖的大型语言模型类似，只是训练数据由文本换成了对撞实验数据。该模型旨在取代目前用于粒子重建与识别的多款专用算法。
It will use the same idea as the large language models behind generative AI but be trained on collision data rather than text, a CERN spokesperson told Reuters. It is meant to replace some specialised algorithms now used to reconstruct and identify particles.
该发言人还提到，过去十年间，欧洲核子研究中心一直借助人工智能技术将实验产生的原始数据转化为事件图像，不过每个处理环节都需使用不同的工具。
The spokesperson said CERN has been using AI to turn raw data from experiments into pictures of the events for a decade, but with multiple tools used for each step.
目前，欧洲核子研究中心已积累了约1艾字节（相当于100万太字节）的大型强子对撞机实验数据；预计在2030年左右完成升级后，对撞事件的数据量还将再增长十倍。
CERN holds about one exabyte, or 1 million terabytes, of experimental data from the Large Hadron Collider, and expects the volume of collisions to increase tenfold after its ongoing upgrade, expected around 2030.
波尔卡里指出：“高能物理学为人工智能技术的研发与测试提供了极为严苛且独特的实验环境。”
"High-energy physics provides a unique environment in which artificial intelligence can be developed and tested with exceptional rigour," Porcari said.
目前双方尚未公布具体的开发时间表与预算。不过双方均表示，未来这些模型还有望应用于其他科学领域。
No timeline or budget was disclosed. Both sides said the models could later serve other sciences.
这家位于巴黎的实验室被广泛视为欧洲追赶美国和中国竞争对手的最佳希望。本月，该实验室发布了一款开源权重模型，称其有能力与竞争对手一较高下。该模型将于10月27日正式向公众发布。
The Paris-based lab, broadly viewed as Europe's best chance to catch up with US and Chinese competitors, this month unveiled a new open-weight model it says can challenge its rivals. It will be released to the public on October 27.
用户可从相关平台下载这些开源权重模型，并在自己的硬件设备上使用。
Users can download open-weight models from platforms and use them on their own hardware.