钟意博客 ·

大语言模型的不确定性

💡 原文中文，约4200字，阅读约需10分钟。

📝

内容提要

即使 temperature=0、seed=0 也无法保证相同输出

🏷️

Xbox’s latest games showcase had Hades 2, The Expanse, and Bluey
It's a weird time for Xbox, with a big leadership shakeup that has many w...
My brief, weird time with the Samsung TriFold
I was ready to give up on the Galaxy Z TriFold. We had a strong case for gett...
Architecting for agentic AI development on AWS
In this post, we demonstrate how to architect AWS systems that enable AI agen...
片段：3月26日
Anthropic对8万用户进行了调查，发现人们对AI的看法各异，但发展中国家对AI持更乐观态度。同时，Julias Shaw提出了将规格文档转化为可执行测试的五步清单。
Mozilla and Mila announce strategic research partnership to advance open source and sovereign AI capabilities
The future of AI should belong to all of humanity, well beyond a handful of c...
我们GitHub Actions 2026安全路线图将迎来什么
软件供应链攻击日益严重，GitHub Actions计划在2026年加强安全，重点在依赖锁定、政策驱动执行和CI/CD监控。新功能将确保工作流可审计、凭证...