开启时会模糊预览图，关闭后正常显示
RT @sapinker: Judea Pearl is universally revered as a pioneer of our understanding of causality and how it can be incorporated into artific…
RT @naval: Great Engineers are Also Artists. “I characterize art as something that is done for its own sake, and done well, and often crea…
Founder: @mixpanel Pizzatarian, programmer, music maker
Not sure how this could fit but might also motion for tokenizerless training. Vector representation of ajustable dimensions would be much preferable to fixed token
Artisanal baker of reasoning models @pleiasfr
RT @ilyasut: One point I made that didn’t come across: - Scaling the current thing will keep leading to improvements. In particular, it wo…
GP @ a16z ... questionable heuristics in a grossly underdetermined world
Jokes aside, taking it very seriously. People never worked on super tiny architectures and from this on you could build up extreme MoEs and many other exotic shapes with unclear properties. One possible tech tree for cognitive core.
Not sure how this could fit but might also motion for tokenizerless training. Vector representation of ajustable dimensions would be much preferable to fixed token
Happens every time for me on chatgpt app on my desktop, have to then go to the website to see the answer lol
Cofounder and Head of Post Training @NousResearch, prev @StabilityAI Github: https://t.co/LZwHTUFwPq HuggingFace: https://t.co/sN2FFU8PVE