内容简介 分享两篇RL在LLM中有效性的论文总结要点：GRPO为什么这么好？并非来自奖励正则化，而是“筛选”掉了模型做得全对（太简单）和全错（太难）的样本。（数据依旧是天花板，决定因素）RL真能提升LLM能力吗？不能，只是让 Base Model ...

Sarvam AI is racing to build India’s first sovereign foundational AI model—an ambitious, 70-billion-parameter system designed ...

Technological advancements in artificial intelligence, AI, have marked a significant appeal in human interaction. DeepSeek s a pioneer in providing AI-interactions using their robots at little or no ...

This post is part of MoFo’s 2025 Intersection of AI and Life Sciences blog series. In this blog series, we explore how artificial intelligence ...

From the dangers of uncontrolled agentic AI to reining in non-human identities, next week’s RSAC Conference covers every ...

April 14, 2025, ChatGPT became available for all students and faculty members at CSUSB. It is available in the myCoyote ...

Full Stack LLM Safty团队 投稿量子位 | 公众号 QbitAI 随着人工智能技术迅猛发展，大模型（如GPT-4、文心一言等）正逐步渗透至社会生活的各个领域，从医疗、教育到金融、政务，其影响力与日俱增。 然而，技术的进步也伴随着潜在风险——大模型安全这一议题正成为全球科技界关注的焦点。 南洋理工大学、新加坡国立大学等全球40余所顶尖机构的67位学者联袂打造大模型全链路安全综述，综 ...

Leveraged ETFs offer flexibility and opportunity, but there is a smart way to use them that begins with education.

PhD 这些年即将告一段落，这几个月梳理先前的工作，准备 Tutorial，借鉴了不少去年从 RLC 上听 David Silver 讲过的思想，在这个 “RL Finally Generalizes (Shunyu Yao)” 的时代到来之际，也一直想写一篇文章作为整理，恰好最近读 Silver 和 Sutton 一起写的《经验时代》 (Welcome to the era of ...

Researchers developed a more efficient way to control the outputs of a large language model, guiding it to generate text that adheres to a certain structure, like a programming language, and remains ...

Kurt Muehmel is the head of AI strategy at Dataiku. He is a creative and analytical executive with 15+ years of experience ...

Despite being rich in AI talent, India faces a significant barrier in scaling its AI ambitions: a shortage of affordable ...