北航、人大和九坤投资共同撰写的论文 《Scaling Laws for Code: Every Programming Language Matters》 整理而成。 在代码大模型(Code LLMs)的预训练中,行业内长期存在一种惯性思维,即把所有编程语言的代码都视为同质化的文本数据,主要关注数据总量的堆叠。然而,现代软件开发本质上是多语言混合的,不同语言的语法特性、语料规模和应用场景差异巨大。
在代码大模型(Code ...
Citing issues with logic, correctness, and security, a new report recommends specific guardrails for AI-generated code.
科技行者 on MSN
北航团队首次揭秘多语言编程的奥秘:为什么Python比Rust更“饿”数据?
这项由北京航空航天大学的杨健、国鑫、林静等研究者联合优矿公司和中国人民大学人工智能学院团队完成的突破性研究,发表于2025年12月的arXiv预印本(论文编号:2512.13472v1),是全球首次系统性探索多语言编程训练规律的重要成果。
昨天,MiniMax M2.1 发布。前脚 MiniMax 刚传出通过港交所聆讯的消息,后脚就直接发布了新一代模型 —— M2.1。巧的是 GLM-4.7 ...
Aider is a “pair-programming” tool that can use various providers as the AI back end, including a locally running instance of ...
Entering Week 15, the Kansas City Chiefs will be in an unfamiliar territory for the first time in the Patrick Mahomes era. Kansas City dropped to 6-7 following a 20-10 loss to the Houston Texans on ...
The biggest stories of the day delivered to your inbox.
Philip Rivers made his return to the NFL after five years away, but he could not lead the Indianapolis Colts to a win over the Seattle Seahawks. The Seahawks earned an 18-16 victory over the Colts in ...
The biggest stories of the day delivered to your inbox.
There is no bigger game this weekend than the AFC East rivalry clash between the Buffalo Bills and New England Patriots. Both teams are Super Bowl contenders, and both quarterbacks are playing at an ...
一些您可能无法访问的结果已被隐去。
显示无法访问的结果