Back to News
ImportantSourceODAILY

资源充足却无法自行检测?Benchmark合伙人质疑Anthropic推动限制模型蒸馏

Odaily星球日报讯 Benchmark 合伙人 Chetan Puttagunta 在 X 平台发文表示,Anthropic 公开呼吁加强对 AI 模型蒸馏的监管令人感到困惑,他表示 Anthropic 目前是一家估值约 1 万亿美元、拥有大量技术和资金资源的公司,按照其描述的规模,所谓“大规模蒸馏攻击”理论上应该较容易被识别和追踪,如果 Anthropic 选择限制相关行为,真正的成本可能并非技术能力不足,而是放弃部分 API 收入。“唯一需要付出的成本似乎是减少相关 API 业务收入。” 此前,Anthropic 等 AI 公司持续关注模型蒸馏问题。模型蒸馏通常指利用一个大型模型(教师模型)的输出结果训练另一个模型(学生模型),以降低成本、提升效率。部分 AI 公司担忧,竞争对手可能通过大量调用 API 获取模型输出,用于训练自身模型,从而绕过原始研发投入。而 Puttagunta 的观点认为,AI 行业围绕模型蒸馏的争议,本质上涉及商业利益、开放竞争和知识产权保护之间的平衡。对于领先 AI 公司而言,如何在保护核心技术与维持开放生态之间找到平衡,将成为未来行业竞争的重要议题。
Disclaimer: The views above are the author's only and do not represent 711BTC. Nothing here constitutes investment advice.

Related

07-01 19:18Important

Anthropic's main model, Claude Sonnet 5, officially launched | _2024111120230_ | API Platform

Odaily Odaily reported on July 1st that Anthropic's latest Sonnet series masterpiece, Claude Sonnet 5, has been officially released to developers via its API platform. This model has achieved groundbreaking progress in code generation and complex agent tasks. Leveraging a massive 1MB context window and innovative adaptive thinking capabilities, Claude Sonnet 5 can handle large and complex projects more stably and efficiently. Meanwhile, to meet the needs of different business scenarios, [2024111120231] has taken the lead in the industry by launching a dual-channel access mode of "official direct connection" and "self-selected service provider," comprehensively helping developers flexibly balance ultimate performance and operating costs. Starting today, users can log in to the [2024111120232] console to apply for an API key and quickly access the platform.

07-02 13:59Important

Anthropic's main model, Claude Sonnet 5, officially launched on the API platform | _2024111120230_

ChainCatcher reports that Anthropic's latest Sonnet series masterpiece, Claude Sonnet 5, is now officially available to developers via its API platform. This model achieves groundbreaking advancements in code generation and complex agent tasks. Leveraging a massive 1MB context window and innovative adaptive thinking capabilities, Claude Sonnet 5 can handle large and complex projects more stably and efficiently. Furthermore, to meet the needs of different business scenarios, Claude Sonnet 5 is the first in the industry to launch a dual-channel access model: "Official Direct Connection" and "Self-Selected Service Provider," comprehensively helping developers flexibly balance ultimate performance and operational costs. Starting today, users can log in to the console to apply for an API key for quick access and experience. For detailed technical guides and documentation, please refer to the official website.

06-29 23:05

Galaxy has been officially included in the Russell 1000 Index, joining the list of benchmark large-cap U.S. stocks.

Odaily Odaily reports that Galaxy announced on the X platform that it has been officially included in the Russell 1000 Index, which covers approximately 1,000 of the largest U.S. listed companies by market capitalization and is one of the important benchmarks for measuring the performance of large-cap stocks. Galaxy stated that its business has achieved scaled operation in the institutional-grade crypto market and is rapidly expanding its data center infrastructure construction, becoming a participant connecting the crypto finance and AI computing power industries.

06-28 11:39

Serenity: Anthropic should strengthen model protection instead of blaming policy restrictions on AI competition.

Mars Finance reported on June 28th that, in response to a WSJ report that China's Zhipu AI has caught up with Anthropic in some cybersecurity benchmark tests, market analyst Serenity stated that rather than attributing this result to US government policy, it's more accurate to believe that Anthropic failed to establish a sufficiently robust model protection mechanism to prevent model distillation. Serenity noted that even before Anthropic released Fable, rumors circulated in the industry that Chinese teams were capable of distilling its models. He believes that the key to AI competition lies not only in investing heavily in capital expenditure (CapEx) to build technological moats, but also in protecting core model capabilities and avoiding the "handing over" of achievements through excessive API calls.

07-08 19:31Important

Momenta, dubbed the "first stock in physics AI," has completed its IPO, with its market capitalization once exceeding HK$70 billion.

Odaily Odaily reports that Momenta, dubbed the "first stock in physical AI," has completed its IPO on the Hong Kong Stock Exchange. The offering price was HK$295.6 per share, and the public offering was oversubscribed by 414 times. The stock opened slightly higher, with its market capitalization briefly exceeding HK$70 billion. Assuming the "greenshoe option" (over-allotment option) is fully exercised, the global offering consisted of approximately 22.93 million shares, raising approximately HK$6.8 billion. According to ZhenFund, Momenta announced that its mass-produced business has surpassed 1 million units equipped with its technology.

07-08 07:52

Anthropic reportedly plans to expand in Manhattan, doubling its New York City staff to 1,000.

According to Odaily Odaily, Anthropic reportedly plans to expand in Manhattan, doubling its New York City workforce to 1,000; it plans to lease a 16-story building in Hudson Square. The company will begin relocating to New York City in the summer of 2026. (Cailian Press)