If you're interested in what the LPX team has been up to at NVIDIA you should read this excellent tech blog post we just published. Some wild numbers:
We hit 10,996 OTSU speeds (high AL case) on Gemma 4 at full quality
3,431 median OTSU on Gemma 4 for full 100K input context window workloads
Vera Rubin + LPU drafts for 2T-scale models for best in class tok/watt at 1000 tok/s interactivity
@mikker Is the UI built with https://t.co/KG5lCPU8ZE? I thought it was "inspired by" the Google Design System. If so, thanks for enlightening me :-)
I'm still recovering from working with their workspace admin UI
@stubailo Branded types are a good fit. I use io-ts for this as it also has excellent encoding and decoding functionality which you will need. The learning curve can be steep unless you are already familiar with monads. https://t.co/YMvJuidrsu
@apievangelist Small nit: GraphQL work started in 2012 but was not presented until 2015…. - around the same time Netflix introduced Falcor https://t.co/H1uqx9FTrF
This talk by @briebunge on adopting TypeScript at scale is so good. Great explanation of rolling out a major change across a large codebase with a lot of engineers. Tons of generally-applicable tidbits.
https://t.co/aY5oVUb7MT
@mikker As a one-man-band, you will always lose. It’s all about spending as little time as possible on these “features”. Make it look like it works (when using the account owners IP) and change/redirect the actual links on the profile. Good luck! :-)
Today I’m launching https://t.co/TowCj1GQt6. 🚀🎉 The site provides designers with the best sizes to use for Image and Video content on Social Media. All in one place with downloadable Sketch templates (no more googling for everything). #some#smm#designers#socialmedia#shipit