gsautiere.bsky.social - Profile | ThreadSky | a Reddit-style client for Bluesky

. @kenklippenstein.bsky.social joined me to break down why TikTok was REALLY banned, according to the very lawmakers who banned it. It was never about China. www.youtube.com/watch?v=Sht8...

submitted 3 days ago • 13 comments

One of the things I’ve come to appreciate being with privacy folks this week is how unique Signal is at this moment. I know it’s maybe a cliche at this point, but it needs to be said.

submitted 3 days ago • 8 comments

PSA: Mozilla released new Firefox ToS terms yesterday, which claim they have the rights to all data you upload to any website using Firefox (presumably to train AI slop). www.mozilla.org/en-US/about/...

submitted 5 days ago • 17 comments

Hello efficient ML folks! My team at Qualcomm AI Research Amsterdam is organizing the 2nd workshop on Efficient Large Vision Models at CPVR 2025! Personally I am quite excited by the long context modeling / linear RNNs side of things, some really exciting stuff going on these days!

submitted 6 days ago • 0 comments

Congratulations to @flashes.blue for their official launch on the App Store 🎉🎉🎉 It’s an instagram alternative built on the AT protocol 🙌 all the visual scrolling with none of the Meta algorithm

submitted 7 days ago • 322 comments

Might be the only silverlining of the current situation: get EU to unify in economy and defense.

submitted 7 days ago • 0 comments

We are hiring on the Generative Media team in London: boards.greenhouse.io/deepmind/job... We work on Imagen, Veo, Lyria and all that good stuff. Come work with us! If you're interested, apply before Feb 28.

submitted 11 days ago • 5 comments

A little story about repression and why I've never treated arrests as something anyone should do: for those of you who don't remember, Occupy Wall Street was a massive movement that was centered around encampments that popped up in city centers across the US (and Canada and UK) from Sept to Nov 2011

submitted 10 days ago • 2 comments

After 6+ months in the making and over a year of GPU compute, we're excited to release the "Ultra-Scale Playbook": hf.co/spaces/nanot... A book to learn all about 5D parallelism, ZeRO, CUDA kernels, how/why overlap compute & coms with theory, motivation, interactive plots and 4000+ experiments!

submitted 13 days ago • 3 comments

Introducing The AI CUDA Engineer: An agentic AI system that automates the production of highly optimized CUDA kernels. sakana.ai/ai-cuda-engi... The AI CUDA Engineer can produce highly optimized CUDA kernels, reaching 10-100x speedup over common machine learning operations in PyTorch. Examples:

submitted 12 days ago • 3 comments

This paper masks out principal components instead of RGB patches because (1) visible pixels may be redundant with masked ones, (2) visible pixels may not be predictive of masked regions. +38% on classification tasks. I wonder how much CroCo & *ST3R might benefit from this. arxiv.org/abs/2502.06314

submitted 15 days ago • 4 comments

fuck sifted.eu/articles/mis...

submitted 20 days ago • 3 comments

i wrote something about This Place for @404media.co. please read it, and then log off. 💗

submitted 27 days ago • 48 comments

👋 hello new followers! - i’m Jan i’m a journalist, electronic musician and visual artist from brooklyn. i’m currently looking for work and available for freelance writing & design commissions. i’m also setting up a storefront to sell prints of my work, TBA. here’s me + some of my recent pieces:

submitted 26 days ago • 5 comments

but surely the AI weapons and surveillance contracts will fix revenue for next quarter ;) www.wired.com/story/google...

submitted 27 days ago • 2 comments

My depresing view on why r1 led to such response is that the market believed OpenAI / NVIDIA would have the monopoly in capturing the value of the workforce they intend to replace with AI, now that it's clear anyone will be able to do it, then they suddenly become less interesting / valuable.

submitted 28 days ago • 0 comments

New video just dropped… and you might actually be able to see it because I’m sharing it on Bluesky! youtu.be/Fv38P3CPglU

submitted 28 days ago • 103 comments

So I hear GenAI is definitely a chill loving place to work no politics at all

submitted 40 days ago • 7 comments

Wow. Bishop Mariann Edgar Budde fearlessly calls out Trump and Vance to their faces. This is heroic. (posting in 3 parts due to Bluesky's 1 minute limit)

submitted 42 days ago • 2214 comments

📢PSA: #NeurIPS2024 recordings are now publicly available! The workshops always have tons of interesting things on at once, so the FOMO is real😵‍💫 Luckily it's all recorded, so I've been catching up on what I missed. Thread below with some personal highlights🧵

submitted 41 days ago • 1 comment

This Stargate project iniated by OpenAI plans to spend 500 billions (!) on AI over the next 4 years. Please someone help me make sense of this. What on Earth. Either they have info I don't have, or I'm terrible at predicting value of AI as it exists, or the market is crazy.

submitted 41 days ago • 0 comments

having succeeded in purchasing positions of influence inside the White House, the billionaires and tech VCs have now set their sights back on Wikipedia

submitted 42 days ago • 250 comments

Finally gave this post a proper read, invaluable information in there! My understanding of generative modeling would be nowhere near as it is now if not for @sedielem.bsky.social blog posts! Thank you again for the time and effort put there.

submitted 46 days ago • 0 comments

I keep coming back to Major Brigg's dream over and over again, such an amazing moment of television. David Lynch, you will me missed. youtu.be/FcsNhIozEPo

submitted 46 days ago • 1 comment

📢 Paper + code release 📃💻 After 2 years of work, I'm excited to announce our newest paper, MatterGen, has been published in Nature! www.nature.com/articles/s41... We are also releasing all the training data, model weights, model code, and evaluation code on GitHub! github.com/microsoft/ma...

submitted 47 days ago • 3 comments

I'm joining forces with Mike Knoop to start ndea.com, a new AI lab. Our focus: deep learning-guided program synthesis. We're betting on a different path to build AI capable of true invention, adaptation, and innovation.

submitted 48 days ago • 12 comments

Transformer²: Self-adaptive LLMs arxiv.org/abs/2501.06252 Check out the new paper from Sakana AI (@sakanaai.bsky.social) paper. We show the power of an LLM that can self-adapt its weights to its environment!

submitted 48 days ago • 1 comment

Relevant to AI climate impact: "Power consumed by U.S. data centers has grown by 50% since last year, & it is expected to grow that much again by the end of 2025, ... Data centers are expected to consume as much as 17% of all U.S. electricity by 2030 — more than quadruple what they consume now."

submitted 55 days ago • 4 comments

You cherry picked the worst song of the movie though, many bangers in there! I loved Emilia Pérez, such an unconventional film.

submitted 56 days ago • 0 comments

i was annoyed at having many chrome tabs with PDF papers having uninformative titles, so i created a small chrome extension to fix it. i'm using it for a while now, works well. today i put it on github. enjoy. github.com/yoavg/pdf-ta...

submitted 57 days ago • 5 comments

The world's richest man has joined a growing chorus of right-wing voices attacking Wikipedia as part of an intensifying campaign against free and open access information. Why do they hate it so much?

submitted 60 days ago • 328 comments

New paper just dropped! How do you combine pre-trained diffusion models without having to train a new one 🤓? Turns out you can use our all new Ito density estimator 🔥 to compute densities under a diffusion model efficiently 🚀!

submitted 66 days ago • 0 comments