Planet covered with solar panels and AI datacenters
HARDWARE
-
Tesla Launches Actually Smart Summon Autonomous Vehicle Feature
By
–
La semana que viene se lanzará la actualización Tesla Actually Smart Summon para los vehículos de la compañía, lo que permitirá que el coche pueda ir a buscarte a tu ubicación de manera autónoma.
— Juan Merodio (@juanmerodio) 4 septembre 2024
Se acabó eso de cargar con las bolsas hasta tu coche en el aparcamiento del… pic.twitter.com/o0ADoaG0IkLa semana que viene se lanzará la actualización Tesla Actually Smart Summon para los vehículos de la compañía, lo que permitirá que el coche pueda ir a buscarte a tu ubicación de manera autónoma. Se acabó eso de cargar con las bolsas hasta tu coche en el aparcamiento del
-
Two-step DLP process for mask fabrication from 800nm to 200nm
By
–
Two step process maybe — DLP to 800nm mask, then use that mask to make a 200nm mask?
-
PhD Defense: Hardware-Efficient Training Pioneer Achievement
By
–
Congratulations @jiawzhao on an excellent PhD defense! Jiawei has been a pioneer in hardware-efficient training. When he started his PhD, everyone was focusing on inference efficiency, and training runs were small, Jiawei took the bold step to pursue training efficiency. Slides:
-
Wall Street’s Chip Stock Selloff: Nvidia, Amphenol, Intel Analysis
By
–
Wall Street’s Best Ideas: No one knows why the chips are down No particular reason for mass-selling of chip stocks, but suggestions about a detail of Nvidia’s Blackwell chip are offered, as Amphenol takes it on the chin, and rumors continue to swirl about Intel, this time about
-
Training 124M Parameter LLMs on MacBook M3 in Real-Time
By
–
It is remarkable that anyone can now train a 124M parameter LLM in about real-time on a MacBook M3. So easy to experiment. This would have been the stuff of dreams when I was in school. I training neural nets, but I really admire the people who build the hardware.
-
Older Chip Fabs Economics and Potential Applications
By
–
Sometimes I consider the question of what could benefit if older generation chip fabs somehow get dramatically cheaper. It is hard to make a case for any logic or memory use on several generations old process even if it was free, due to power and capabilities. One possible use
-
Quantization Technique Reduces LLM Size and Memory Requirements
By
–
While SOTA LLMs are too large to run on laptops, quantization is a technique that reduces LLMs’ computational and memory requirements. Quantization reduces a model’s size and speeds up processing by converting its parameters from 32-bit to lower-precision formats like 16-bit or
-
Drones: Real-World Applications Simplifying Complex Logistics Tasks
By
–
Los drones tienen un montón de aplicaciones en el mundo real y están ayudándonos a simplificar un montón de tareas que hasta hace poco necesitaban de una logística extremadamente compleja.
— Juan Merodio (@juanmerodio) 2 septembre 2024
Pero también sirven para gastar una broma de las que pone los pelos de punta 😅 pic.twitter.com/XuANTASB1lLos drones tienen un montón de aplicaciones en el mundo real y están ayudándonos a simplificar un montón de tareas que hasta hace poco necesitaban de una logística extremadamente compleja. Pero también sirven para gastar una broma de las que pone los pelos de punta
-
Cautious optimism on compute infrastructure and lilypad exploration
By
–
Agree on being cautiously optimistic about compute, but curious nonetheless. Will check lilypad out!