Genie 2:大型基础世界模型(large-scale foundation world model)
Google DeepMind刚推出了Genie 2。Genie 2是一种基础世界模型,能够生成无限多样的、可...
Read MoreVLsI模型:逐层蒸馏,逐层对齐,实现从大规模到小规模视觉-语言模型(VLM)的高效知识迁移
论文《VLsI: Verbalized Layers-to-Interactions from Large t...
Read More通过多目标优化(multi-objective optimization)自动发现最佳元解算器(meta-solvers)
论文Automatic discovery of optimal meta-solvers via multi...
Read MoreSTAR(Synthesis of Tailored Architectures):深度学习的合成定制化架构
论文STAR: Synthesis of Tailored Architectures(《STAR: 合成定制...
Read MoreModel Context Protocol(MCP):模型上下文协议
近日,Anthropic公司发布了“模型上下文协议(Model Context Protocol,MCP)”。...
Read More多种大语言模型(LLMs)在磁共振成像(MRI)技术问题回答中的表现
论文Performance of Large Language Models in Technical MRI...
Read MoreStar Attention算法:有效提升大型语言模型(LLM)在长序列推理任务中的效率
论文《Star Attention: Efficient LLM Inference over Long Se...
Read MoreHAI-DEF(Health AI Developer Foundations):健康AI开发基础模型集
论文《Health AI Developer Foundations》详细描述了由Google研究团队和Dee...
Read MoreFlamingo-CXR:结合视觉和语言表示,可自动化生成放射报告的先进AI模型
论文Collaboration between clinicians and vision–language...
Read More








