Innovator: Scientific Continued Pretraining with Fine-grained MoE Upcycling
"混合专家模型" 相关文章
Qwen Releases Qwen3-Coder-480B-A35B-Instruct: Its Most Powerful Open Agentic Code Model Yet
[推广] Kimi k2 online
[推广] Kimi k2 online
MoSE: Skill-by-Skill Mixture-of-Expert Learning for Autonomous Driving
每日互动，Deepseek R2有望本月发布，二波可期
Efficient Training of Large-Scale AI Models Through Federated Mixture-of-Experts: A System-Level Approach
City-Level Foreign Direct Investment Prediction with Tabular Learning on Judicial Data
盘古团队回应抄袭指控：并非基于其他模型训练 已标注开源代码版权声明
华为盘古首次露出，昇腾原生72B MoE架构，SuperCLUE千亿内模型并列国内第一
The Rise of Mixture-of-Experts: How Sparse AI Models Are Shaping the Future of Machine Learning
消息称DeepSeek R2下月发：成本较GPT降97%、华为芯片性能不输英伟达
Qwen3强势来袭：推理力爆表、语言超百种、智能体协作领先，引领AI开源大模型
消息称DeepSeek R2下月发：成本较GPT降97%、华为芯片性能不输英伟达
Alibaba Qwen Team Just Released Qwen3: The Latest Generation of Large Language Models in Qwen Series, Offering a Comprehensive Suite of Dense and Mixture-of-Experts (MoE) Models
Multimodal Models Don’t Need Late Fusion: Apple Researchers Show Early-Fusion Architectures are more Scalable, Efficient, and Modality-Agnostic
Meta公布MoE架構開發的Llama 4 開源4000億、1090億參數的Maverick
消息称蚂蚁集团采用阿里、华为等国产芯片训练 AI：性能匹敌英伟达 H800，成本降低 20%
Observations About LLM Inference Pricing
DeepSeek开源放大招：FlashMLA让H800算力狂飙！曝光低成本秘笈
