Yann LeCun@ylecun

Elon Musk: This false nomenclature of "researcher" and "engineer", which is a thinly-masked way of describing a two-tier engineerin...

Saining Xie@sainingxie

Lucas Beyer (bl16): Ok this makes me super happy. The "NoFilter" work, paper, and advocacy that @angelinepouget and I argued so hard for is ...

Jim Fan@DrJimFan

AI at Meta: We're excited to have @shengjia_zhao at the helm as Chief Scientist of Meta Superintelligence Labs. Big things are comin...

Noam Brown@polynoamial

Nat McAleese: We are seeing much faster AI progress than **Paul Christiano** and **Yudkowsky** predicted, who had gold in 2025 at 8% a...

Noam Brown@polynoamial

Ravid Shwartz Ziv: So, all the models underperform humans on the new International Mathematical Olympiad questions, and Grok-4 is especiall...

Yann LeCun@ylecun

Micah Goldblum: 🚨 Did you know that small-batch vanilla SGD without momentum (i.e. the first optimizer you learn about in intro ML) is ...

Saining Xie@sainingxie

Xichen Pan: The code and instruction-tuning data for MetaQuery are now open-sourced! Code: https://github.com/facebookresearch/metaq...

Saining Xie@sainingxie

Rob Fergus: 1/ Excited to share that I'm taking on the role of leading Fundamental AI Research (FAIR) at Meta. Huge thanks to Joelle...

Saining Xie@sainingxie

Peter Tong: We are open-sourcing all the models in Web-SSL, from ViT-L to ViT-7B! It was super fun to train and play with these mass...

OpenRouter 宣布为 Llama 3.3 70b 降价，同时提供该模型的六个版本及对应提供商。

OpenRouter 宣布对 Llama 3.3 70b 进行价格下调，同时新增六个模型及相应供应商。此次调整扩大了 Llama 3.3 70b 的可选提供商范围，并降低了调用成本。

Meta 于 4 月 18 日发布的开源模型 Llama 3-70B 在 Chatbot Arena 排行榜迅速登顶，参与超 5 万次对战。该模型在开放式写作和创意任务上表现突出，胜率达 60%，但在数学、编码等封闭式技术任务上逊于 GPT-4-Turbo 和 Claude 3 Opus。随着提示难度增加，其胜率从 50% 显著下降至 40%。分析显示，Llama 3 的输出风格更友好且具对话性，这成为其获得用户偏好的关键因素。

文章针对《纽约时报》关于 Yi-34B 与 Llama 2 关系的报道进行事实核查，澄清 Yi-34B 在架构设计、训练数据及分词器实现上与 Llama 2 的实际差异，同时系统梳理了当前大语言模型训练领域的行业常见实践，强调在遵循开源协议前提下基于现有架构进行技术迭代是 AI 社区的标准做法。