跳到正文
/AI 前沿

外观

快速抵达

↑ ↓ 选择 · Enter 打开 · Esc 关闭MyHOT
LABS & BUILDERS

AI 大厂动态

最新动态
最新阅读今日重点
筛选
通义千问 · 官方发布

SAPO:一种稳定且高性能的大语言模型强化学习方法

强化学习(Reinforcement Learning, RL)已经成为提升大语言模型(Large Language Models, LLM)推理能力的核心技术之一。现代 RL 训练流程使模型能够解决困难的数学问题、编写复杂代码和进行多模态推理。实践中,一种被广泛采用的方法是基于组的策略优化(group‑based policy optimization):对每个提示采样多个回复,并在组内进行奖励归一化。

原文
速览MYHOT
通义千问 · 官方发布

SAPO:一种稳定且高性能的大语言模型强化学习方法

强化学习(Reinforcement Learning, RL)已经成为提升大语言模型(Large Language Models, LLM)推理能力的核心技术之一。现代 RL 训练流程使模型能够解决困难的数学问题、编写复杂代码和进行多模态推理。实践中,一种被广泛采用的方法是基于组的策略优化(group‑based policy optimization):对每个提示采样多个回复,并在组内进行奖励归一化。

OpenAI News

Introducing OpenAI for Australia

OpenAI is launching OpenAI for Australia to build sovereign AI infrastructure, upskill more than 1.5 million workers, and accelerate innovation across the country’s growing AI ecosystem.

原文
速览MYHOT
OpenAI News

Introducing OpenAI for Australia

OpenAI is launching OpenAI for Australia to build sovereign AI infrastructure, upskill more than 1.5 million workers, and accelerate innovation across the country’s growing AI ecosystem.

通义千问 · 官方发布

Qwen3-TTS 全面升级!49种音色 + 10种语言 + 9种方言

Qwen3-TTS 是支持多音色、多语种和多方言的旗舰语音合成模型,致力于实现稳定、自然和高效的语音生成,目前可通过Qwen API访问。 主要改进: 更加丰富的音色支持:Qwen3-TTS 提供超过49种高品质音色,涵盖不同性别、年龄、地域特征与角色设定,满足多样化的场景需求。撒娇搞怪-茉兔,陪伴感满满的青梅竹马-小野杏,傲娇率性的女汉子-十三 ,严厉老师-墨讲师 ,智慧老者-沧明子,萝莉萌妹-萌小姬等众多角色等你探索。

原文
速览MYHOT
通义千问 · 官方发布

Qwen3-TTS 全面升级!49种音色 + 10种语言 + 9种方言

Qwen3-TTS 是支持多音色、多语种和多方言的旗舰语音合成模型,致力于实现稳定、自然和高效的语音生成,目前可通过Qwen API访问。 主要改进: 更加丰富的音色支持:Qwen3-TTS 提供超过49种高品质音色,涵盖不同性别、年龄、地域特征与角色设定,满足多样化的场景需求。撒娇搞怪-茉兔,陪伴感满满的青梅竹马-小野杏,傲娇率性的女汉子-十三 ,严厉老师-墨讲师 ,智慧老者-沧明子,萝莉萌妹-萌小姬等众多角色等你探索。

OpenAI News

OpenAI to acquire Neptune

OpenAI is acquiring Neptune to deepen visibility into model behavior and strengthen the tools researchers use to track experiments and monitor training.

原文
速览MYHOT
OpenAI News

OpenAI to acquire Neptune

OpenAI is acquiring Neptune to deepen visibility into model behavior and strengthen the tools researchers use to track experiments and monitor training.

OpenAI News

How confessions can keep language models honest

OpenAI researchers are testing “confessions,” a method that trains models to admit when they make mistakes or act undesirably, helping improve AI honesty, transparency, and trust in model outputs.

原文
速览MYHOT
OpenAI News

How confessions can keep language models honest

OpenAI researchers are testing “confessions,” a method that trains models to admit when they make mistakes or act undesirably, helping improve AI honesty, transparency, and trust in model outputs.

OpenAI News

Announcing the initial People-First AI Fund grantees

The OpenAI Foundation announces the initial recipients of the People-First AI Fund, awarding $40.5M in unrestricted grants to 208 nonprofits supporting community innovation and opportunity.

原文
速览MYHOT
OpenAI News

Announcing the initial People-First AI Fund grantees

The OpenAI Foundation announces the initial recipients of the People-First AI Fund, awarding $40.5M in unrestricted grants to 208 nonprofits supporting community innovation and opportunity.

OpenAI News

Inside Mirakl's agentic commerce vision

Mirakl is redefining commerce through AI agents and ChatGPT Enterprise—achieving faster documentation, smarter customer support, and building toward agent-native commerce with Mirakl Nexus.

原文
速览MYHOT
OpenAI News

Inside Mirakl's agentic commerce vision

Mirakl is redefining commerce through AI agents and ChatGPT Enterprise—achieving faster documentation, smarter customer support, and building toward agent-native commerce with Mirakl Nexus.

OpenAI News

Funding grants for new research into AI and mental health

OpenAI is awarding up to $2 million in grants for research at the intersection of AI and mental health. The program supports projects that study real-world risks, benefits, and applications to improve safety and well-being.

原文
速览MYHOT
OpenAI News

Funding grants for new research into AI and mental health

OpenAI is awarding up to $2 million in grants for research at the intersection of AI and mental health. The program supports projects that study real-world risks, benefits, and applications to improve safety and well-being.

OpenAI News

OpenAI and NORAD team up to bring new magic to “NORAD Tracks Santa”

OpenAI and NORAD are bringing new magic to “NORAD Tracks Santa” with three ChatGPT holiday tools that let families create festive elves, toy coloring pages, and custom Christmas stories.

原文
速览MYHOT
OpenAI News

OpenAI and NORAD team up to bring new magic to “NORAD Tracks Santa”

OpenAI and NORAD are bringing new magic to “NORAD Tracks Santa” with three ChatGPT holiday tools that let families create festive elves, toy coloring pages, and custom Christmas stories.

OpenAI News

Accenture and OpenAI accelerate enterprise AI success

Accenture and OpenAI are collaborating to help enterprises bring agentic AI capabilities into the core of their business and unlock new levels of growth.

原文
速览MYHOT
OpenAI News

Accenture and OpenAI accelerate enterprise AI success

Accenture and OpenAI are collaborating to help enterprises bring agentic AI capabilities into the core of their business and unlock new levels of growth.

OpenAI News

OpenAI takes an ownership stake in Thrive Holdings to accelerate enterprise AI adoption

OpenAI takes an ownership stake in Thrive Holdings to accelerate enterprise AI adoption, embedding frontier research and engineering directly into accounting and IT services to boost speed, accuracy, and efficiency while creating a scalabl…

原文
速览MYHOT
OpenAI News

OpenAI takes an ownership stake in Thrive Holdings to accelerate enterprise AI adoption

OpenAI takes an ownership stake in Thrive Holdings to accelerate enterprise AI adoption, embedding frontier research and engineering directly into accounting and IT services to boost speed, accuracy, and efficiency while creating a scalabl…

OpenAI News

Mixpanel security incident: what OpenAI users need to know

OpenAI shares details about a Mixpanel security incident involving limited API analytics data. No API content, credentials, or payment details were exposed. Learn what happened and how we’re protecting users.

原文
速览MYHOT
OpenAI News

Mixpanel security incident: what OpenAI users need to know

OpenAI shares details about a Mixpanel security incident involving limited API analytics data. No API content, credentials, or payment details were exposed. Learn what happened and how we’re protecting users.