GPT-5实现多模态医学推理能力质的飞跃 • 首次系统评估GPT-5在统一协议
GPT-5实现多模态医学推理能力质的飞跃 • 首次系统评估GPT-5在统一协议
爱生活爱珂珂的文章
Wilson Lin两个月打造的自研神经搜索引擎，性能与成本优化令人瞩目：• 集
Wilson Lin两个月打造的自研神经搜索引擎，性能与成本优化令人瞩目：• 集
![[人人能懂] AI的“小灶、泥人与神级助理” （扫码收听） ](http://dayu-img.uc.cn/columbus/img/oc/1002/97c105417a386393d6cb62decbf9216d.jpg)
[人人能懂] AI的“小灶、泥人与神级助理” （扫码收听）
[人人能懂] AI的“小灶、泥人与神级助理” （扫码收听）
![[CL]Speed Always Wins: A Survey on Effic](http://dayu-img.uc.cn/columbus/img/oc/1002/9fd5bf1699bdb30e2f5c34a50192db73.jpg)
[CL]Speed Always Wins: A Survey on Effic
[CL]Speed Always Wins: A Survey on Effic
![[LG]《Beyond Scaling Law: A Data-Efficien](http://dayu-img.uc.cn/columbus/img/oc/1002/44a899d6f4657e8a3fb844dbb79042fd.jpg)
[LG]《Beyond Scaling Law: A Data-Efficien
[LG]《Beyond Scaling Law: A Data-Efficien
![[CV]《VertexRegen: Mesh Generation with C](http://dayu-img.uc.cn/columbus/img/oc/1002/4678ab07bd19d95e87b512677fedf141.jpg)
[CV]《VertexRegen: Mesh Generation with C
[CV]《VertexRegen: Mesh Generation with C
![[LG]《Noise Hypernetworks: Amortizing Tes](http://dayu-img.uc.cn/columbus/img/oc/1002/389e5e3bac7a9dc083b1e9e77a5af5a9.jpg)
[LG]《Noise Hypernetworks: Amortizing Tes
[LG]《Noise Hypernetworks: Amortizing Tes
![[LG]《μ-Parametrization for Mixture of Ex](http://dayu-img.uc.cn/columbus/img/oc/1002/2f394609e96b3deab4709d7959a645c2.jpg)
[LG]《μ-Parametrization for Mixture of Ex
[LG]《μ-Parametrization for Mixture of Ex
![[LG]《Part I: Tricks or Traps? A Deep Div](http://dayu-img.uc.cn/columbus/img/oc/1002/ea5574909033391e2eb25f455f2e1856.jpg)
[LG]《Part I: Tricks or Traps? A Deep Div
[LG]《Part I: Tricks or Traps? A Deep Div
![早！[太阳] 早安 ](http://dayu-img.uc.cn/columbus/img/oc/1002/2eaedd83a0e3e0ca61061aa1b624421b.jpg)
早！[太阳] 早安
早！[太阳] 早安
深入解析扩散模型本质，《Generative modelling in late
深入解析扩散模型本质，《Generative modelling in late
OpenAI 推出的 GPT-OSS 模型融合了多篇前沿论文的关键创新，构建了一
OpenAI 推出的 GPT-OSS 模型融合了多篇前沿论文的关键创新，构建了一
终端效率提升利器盘点，适合追求极致CLI体验的你：• 🔍 fzf：模糊查找神器
终端效率提升利器盘点，适合追求极致CLI体验的你：• 🔍 fzf：模糊查找神器
Marc Andreessen 2025 年职场建议，深刻洞察 AI 时代下的职
Marc Andreessen 2025 年职场建议，深刻洞察 AI 时代下的职
![[CL]《A Rose by Any Other Name Would Smel](http://dayu-img.uc.cn/columbus/img/oc/1002/27b924af2ae74a9b71d8398fa1d4aafd.jpg)
[CL]《A Rose by Any Other Name Would Smel
[CL]《A Rose by Any Other Name Would Smel
![[LG]《Constrained Decoding of Diffusion L](http://dayu-img.uc.cn/columbus/img/oc/1002/12d3adb87d795b7d1f420b5b7680f10e.jpg)
[LG]《Constrained Decoding of Diffusion L
[LG]《Constrained Decoding of Diffusion L
![[CL]《SSRL: Self-Search Reinforcement Lea](http://dayu-img.uc.cn/columbus/img/oc/1002/be9621f2e4b3f8f4d46cfd6a57e588ba.jpg)
[CL]《SSRL: Self-Search Reinforcement Lea
[CL]《SSRL: Self-Search Reinforcement Lea
![[LG]《SoK: Data Minimization in Machine L](http://dayu-img.uc.cn/columbus/img/oc/1002/e3d8386a17fae701569d047e5d9743ed.jpg)
[LG]《SoK: Data Minimization in Machine L
[LG]《SoK: Data Minimization in Machine L
![[LG]《XQuant: Breaking the Memory Wall fo](http://dayu-img.uc.cn/columbus/img/oc/1002/abefc9ee14ddec5e8de5f6598ee207f0.jpg)
[LG]《XQuant: Breaking the Memory Wall fo
[LG]《XQuant: Breaking the Memory Wall fo
![早！[太阳] 早安 ](http://dayu-img.uc.cn/columbus/img/oc/1002/1b514fa4847364b108258946642a85a7.jpg)
早！[太阳] 早安
早！[太阳] 早安
作者信息
