Mimee // MLforSystems CFP open now
@MimeeXu
what good can I do with my life if I only know math and computers. PhD candidate on weekdays, finishing procrastinating doctorate on ML/AI x security/privacy
That IOI friend getting increasingly impressed with LLM is maybe a sign for AI, but it might also be a sign of resignation that we as a community should probably stop orienting our lives with costly resources just to impress our IOI friends like seriously we aren’t 16 any more
Exciting news! 📣 The call for papers for ML for Systems @NeurIPSConf is now open. Submission deadline: Aug 22 AoE Help spread the word! P.S. Agents and LLM systems are systems, too! mlforsystems.org/call_for_paper… #MLforSystems #MachineLearning #LLMs #Agents #CodeModels #neurips2025
Today, we at @OpenAI achieved a milestone that many considered years away: gold medal-level performance on the 2025 IMO with a general reasoning LLM—under the same time limits as humans, without tools. As remarkable as that sounds, it’s even more significant than the headline 🧵
1/N I’m excited to share that our latest @OpenAI experimental reasoning LLM has achieved a longstanding grand challenge in AI: gold medal-level performance on the world’s most prestigious math competition—the International Math Olympiad (IMO).
Why do people only talk about AI privacy in a “, but then …” context
#MLSys2026 will be led by the general chair @luisceze and PC chairs @JiaZhihao and @achowdhery. The conference will be held in Bellevue on Seattle's east side. Consider submitting and bringing your latest works in AI and systems—more details at mlsys.org.
📢Exciting updates from #MLSys2025! All session recordings are now available and free to watch at mlsys.org. We’re also thrilled to announce that #MLSys2026 will be held in Seattle next May—submissions open next month with a deadline of Oct 30. We look forward to…
On that note, our Machine Learning For Systems workshop videos are now live! SlidesLive: slideslive.com/neurips-2024/w… Website: mlforsystems.org Neurips: neurips.cc/virtual/2024/w…
I'm always realizing we don't have enough "AI meets software" people. Those that study system architecture/design (and programming systems, etc) and the people that study, say, machine learning essentially don't talk? So you'll have to put up with my rants until that's fixed.
:raise your hand if your peak math was actually peak chinese: jokes aside, my most recent linguistic heldup is that i observed chinese to be a pseudo-axiomatic language. Linguistic determinism might be fake, but written chinese reads so imperative, i habitually penalize against
When your LLM starts speaking Chinese during RL to improve math:
i believe the dynamic, which 1️⃣ makes causal identification tricky irl w/network influence 2️⃣ has fueled arguments to NOT talk about ed on socials (and NOT post "thinspo" of yourself): even if done critically, vulnerable brains may 'maladapt' towards novel destructive behaviour.
Rule of thumb: When people deny even the possibility of social contagion, there is a lot of social contagion happening.
Making LLMs run efficiently can feel scary, but scaling isn’t magic, it’s math! We wanted to demystify the “systems view” of LLMs and wrote a little textbook called “How To Scale Your Model” which we’re releasing today. 1/n
This Beijing-born privacy researcher was not expecting to be in this 'Let China Have Our Data' timeline. Misappropriating the word "privacy" will backfire. People know real privacy respects individual liberty. It cannot be about giving up control over their own (data) destiny.