A conversation with Ani Baddepudi about Gemini's vision capabilities, how we got to SOTA, and where we and the ecosystem go next. Ani is a friend and collaborator so this conversation was a lot of fun : )

Cartwheel now exports directly to the UE5 skeleton. Works with Mannequin and Metahuman control rigs. Native. Fast. Unreal-ready. getcartwheel.com

My latest post: The American DeepSeek Project Build fully open models in the US in the next two years to enable a flourishing, global scientific AI ecosystem to balance China's surge in open-source and an alternative to building products ontop of leading closed models.

I always read PSGD as Pierre St. Germain Descent...

It turns out the Turing Award is actually a silvery bowl from Tiffanys.

Sleep when the baby sleeps, eat when the baby eats, debug cuda kernels when the baby debugs cuda kernels

OpenAI survived a full Anthropic leaving them, surely they survive this too?

Seems like a lot of you don't know the lore

An explanation of Match3 functions and motivation for 2-simplicial attention. You have a shelf of about 30 film which include: Lord of the Rings, Independence Day, Harry Potter series, Idiocracy, Mission: Impossible, The Social Network, Die Hard series, and so on. Now say you

Pretty cool that red bull racing uses q and KDB in their cars

If AGI was achieved, and hosted behind a closed API (lots of assumptions here). Could you generate sufficient data to distill it into an open model?

vLLM runs on free-threaded Python! A group of engineers from Meta’s Python runtime language team has shown that it’s possible to run vLLM on the nogil distribution of Python. We’re incredibly excited to embrace this future technique and be early adopters 😍

🚀 A new way to control AI video generation - by concatenation of control signals and LoRAs trained on just a few samples We’re releasing 3 control LoRAs for LTX-Video: open-pose, depth, and canny edges. Plus: training code so you can build your own types of control. 🧵

This needs to go in the book!!

I didn't see this mentioned but NSA from DeepSeek inspired the efficient computation for the 2-simplical work!! Open science informed advancements that improved the scaling exponent. In the gpt 4.5 retrospective podcast they said the only thing that matters, moving forward,

how is strava the best friend photo sharing app in 2025