[人人能懂] 从统一语言、认知陷阱到犯错的艺术要让AI真正变聪明，是该为它发明一
[人人能懂] 从统一语言、认知陷阱到犯错的艺术要让AI真正变聪明，是该为它发明一
爱生活爱珂珂的文章
![[人人能懂] 从统一语言、认知陷阱到犯错的艺术要让AI真正变聪明，是该为它发明一](http://dayu-img.uc.cn/columbus/img/oc/1002/149ab563f8e118bbae9a584b8e409ad6.jpg)
![[LG]《Learning to Make MISTAKEs: Modeling](http://dayu-img.uc.cn/columbus/img/oc/1002/8bb5f3481c4fa78a598f0db7240e5fbf.jpg)
[LG]《Learning to Make MISTAKEs: Modeling
[LG]《Learning to Make MISTAKEs: Modeling
![[CL]《The Curious Case of Curiosity acros](http://dayu-img.uc.cn/columbus/img/oc/1002/9b47a8f4009a7665ce276097305dc005.jpg)
[CL]《The Curious Case of Curiosity acros
[CL]《The Curious Case of Curiosity acros
![[LG]《Not All Bits Are Equal: Scale-Depen](http://dayu-img.uc.cn/columbus/img/oc/1002/2e1ecc60c14c6f35cca3a8a8d287cf46.jpg)
[LG]《Not All Bits Are Equal: Scale-Depen
[LG]《Not All Bits Are Equal: Scale-Depen
![[CL]《LLMs Can Get "Brain Rot"!》S Xing, J](http://dayu-img.uc.cn/columbus/img/oc/1002/283b236b6b5d6f42d3f05bf0f595ebe3.jpg)
[CL]《LLMs Can Get "Brain Rot"!》S Xing, J
[CL]《LLMs Can Get "Brain Rot"!》S Xing, J
![[LG]《Tensor Logic: The Language of AI》P](http://dayu-img.uc.cn/columbus/img/oc/1002/35e0ff00e2a05dca75d7e4f8fea6bcad.jpg)
[LG]《Tensor Logic: The Language of AI》P
[LG]《Tensor Logic: The Language of AI》P
![早！[太阳] 早安 ](http://dayu-img.uc.cn/columbus/img/oc/1002/f7ca8528ac3f4522698989638e25fa17.jpg)
早！[太阳] 早安
早！[太阳] 早安
如果你想让大型语言模型（LLMs）运行得更快、更省钱，这16项技术值得深入掌握：
如果你想让大型语言模型（LLMs）运行得更快、更省钱，这16项技术值得深入掌握：
卅载躬耕育栋材，七旬弦歌向未来。邮传薪火承初心，智启新程向远开。祝北邮未来更美好
卅载躬耕育栋材，七旬弦歌向未来。邮传薪火承初心，智启新程向远开。祝北邮未来更美好
![[CL]《Catch Your Breath: Adaptive Computa](http://dayu-img.uc.cn/columbus/img/oc/1002/53d48cd420a7314e2bf5ea273ae38a55.jpg)
[CL]《Catch Your Breath: Adaptive Computa
[CL]《Catch Your Breath: Adaptive Computa
![[LG]《Beyond Multi-Token Prediction: Pret](http://dayu-img.uc.cn/columbus/img/oc/1002/db3b59fbd8e7cf605a42584bba1e31f4.jpg)
[LG]《Beyond Multi-Token Prediction: Pret
[LG]《Beyond Multi-Token Prediction: Pret
![[LG]《BitNet Distillation》X Wu, S Huang,](http://dayu-img.uc.cn/columbus/img/oc/1002/9a3f4f61ef539b4a43725f792639efde.jpg)
[LG]《BitNet Distillation》X Wu, S Huang,
[LG]《BitNet Distillation》X Wu, S Huang,
![[CL]《Midtraining Bridges Pretraining and](http://dayu-img.uc.cn/columbus/img/oc/1002/3191250df76e0da158d4286f656091a8.jpg)
[CL]《Midtraining Bridges Pretraining and
[CL]《Midtraining Bridges Pretraining and
![[LG]《Optimal Control Theoretic Neural Op](http://dayu-img.uc.cn/columbus/img/oc/1002/56f3e884b96d0537e33a9bebb6134931.jpg)
[LG]《Optimal Control Theoretic Neural Op
[LG]《Optimal Control Theoretic Neural Op
![早！[太阳] 早安 ](http://dayu-img.uc.cn/columbus/img/oc/1002/644c8420bdc0a3d7b52d2c3b5b7304e6.jpg)
早！[太阳] 早安
早！[太阳] 早安
![晚安～ [月亮] 晚安 ](http://dayu-img.uc.cn/columbus/img/oc/1002/1a731a725f0eb9783b1908527c02f7a8.jpg)
晚安～ [月亮] 晚安
晚安～ [月亮] 晚安
《爱可可微博热门分享(10.17)》
《爱可可微博热门分享(10.17)》
《It takes two to think》这篇文章强调了创造性思维背后的核心
《It takes two to think》这篇文章强调了创造性思维背后的核心
作为LLM/AI从业者，要在快速演变的职场中保持竞争力，必须掌握以下20项关键技
作为LLM/AI从业者，要在快速演变的职场中保持竞争力，必须掌握以下20项关键技
「生成多个带概率的答案」斯坦福最新论文提出“言语采样”（Verbalized S
「生成多个带概率的答案」斯坦福最新论文提出“言语采样”（Verbalized S
作者信息
