Anton Tsitsulin

Pre-training data for Gemini and Gemma (又菜又爱玩).
past (?) life: graph machine learning

[Anton Tsitsulin @graph_]

@giffmana @OfficialLoganK @zapier @demishassabis

[Anton Tsitsulin @graph_]

Imagine dogpiling on one of most influential researchers on physics:ml intersection and thinking you made such a great point. Don't even want to QT disgusting responses to some sensible points about AI use policy on arXiv. People really have gone insane.

[Lenka Zdeborova @zdeborova]

Occasional errors and oversights are part of science. If we lost our driver’s license for a year every time we exceeded the speed limit by 10 km/h, daily life would become unworkable. Many countries instead use point systems, where trust can be rebuilt through good behavior.

[Timothy Gowers @wtgowers @wtgowers]

AI has now solved a major open problem -- one of the best known Erdos problems called the unit distance problem, one of Erdos's favourite questions and one that many mathematicians had tried.
https://t.co/SD1vVPkrHR

[Vlad Feinberg @FeinbergVlad]

How to land a job at a frontier lab

https://t.co/oHIqLgBMbC

[Vahab Mirrokni @mirrokni]

Happy to share new proofs of simple constructions of general Knuth's recent cycle conjecture for even case based on a new Gemini system for long proofs; one with 40+ pages & one with 70+ pages here: https://t.co/hzgXNURaLd

Excited about the system for long proofs. See Comments

[Anton Tsitsulin @graph_]

This is why we have 4 frontier labs btw

[PsyPost.org @PsyPost]

A new study shows that individuals with similar levels of autistic traits are naturally drawn to one another. Brain scans reveal they use alternative, highly effective neural strategies to connect, challenging traditional deficit-based models of autism. https://t.co/6YgJpV6iHU

[alex wortega @justALEXWORTEGA]

Tired of getting rejects and reading reviewer comments that clearly missed the point of your paper?
I built an peer review skill for Claude Code that actually gives you useful feedback before you submit.
👇

[Anton Tsitsulin @graph_]

Correspondence chess community is the definition of long-termist

[Anton Tsitsulin @graph_]

@hopes_revenge come to the dark side

[Anton Tsitsulin @graph_]

@simonbatzner Congrats!!

[Anton Tsitsulin @graph_]

this is good actually

[Sarath Chandar @apsarathchandar]

This is really bad. As an area chair, I usually accept papers with a reject if the reject review is not valid. I never reject a paper with all accepts.

[Anton Tsitsulin @graph_]

Unstoppable force of AI meets an immovable object

[Jim Cramer @jimcramer]

The bottom line: they are short compute and Codex is on fire...

[Anton Tsitsulin @graph_]

@kyliebytes that episode deserves a microwaved cheese sandwich tbh

[Anton Tsitsulin @graph_]

@deanwball @teortaxesTex My comment only referred to "useful substrate for all manner of upstarts, worldwide. This is pro-social, and I wish the US had an equivalent" part. I strongly believe that it's the case with Gemma – it's why I keep working on the project:)

[Anton Tsitsulin @graph_]

@teortaxesTex @deanwball memes aside a bunch of folks tune and build their business on gemmas, especially in non-English languages

[Anton Tsitsulin @graph_]

@teortaxesTex @deanwball on par in spirit 🥺

[Anton Tsitsulin @graph_]

5.5 pro is a great but weirdly shaped model. Feels like a different interaction mode than previous pro generations somehow

[Keivan Rezaei @RezaeiKeivan]

🔎 If you are interested in machine unlearning, check out our poster tomorrow at #ICLR2026.
(Pavilion 4, P4-#3916 | Apr 24, 10:30 AM – 1:00 PM)

We explore a new direction for machine unlearning, leveraging model checkpoints!

📜 Paper: https://t.co/jIjhQ7j5yg

[Anton Tsitsulin @graph_]

@stevehou 👀

https://t.co/FnyqS3iWOv

[Demis Hassabis @demishassabis]

@Steve_Yegge Maybe tell your buddy to do some actual work and to stop spreading absolute nonsense. This post is completely false and just pure clickbait.