Samet Oymak
@SametOymac
Professor @UMich EECS, Research on the Foundations of ML+RL+LLM
📣 We will soon become the James M. and Cathleen D. Stone Center on Inequality and Shaping the Future of Work! Starting in July, we will usher in a significant new era of research, policy, and education, made possible by a gift from the Stone Foundation. news.mit.edu/2025/mit-econo…
We are hiring! Our team at Google Research, NY is seeking a Research Scientist! Our recent research efforts include developing algorithms for improving inference efficiency and alignment of LLMs. If you are interested, then please consider applying! google.com/about/careers/…
Some SAC context for NeurIPS review scores - out of my batch of 100 papers: - 1 paper ≥5.0 - 6 papers ≥4.5 - 11 papers ≥4.0 - 25 papers ≥3.75 - 42 papers ≥3.5 So currently naive accept cutoff would be 3.75.
NSF budgets slashed by 50%, ongoing grants cancelled, NSF staff drastically reduced, all 37 divisions abolished, and grants will now be reviewed by a political kommissar. How will that help technological leadership? linkedin.com/posts/yann-lec…
RL and post-training play a central role in giving language models advanced reasoning capabilities, but many algorithmic and scientific questions remain unanswered. Join us at FoPT @ COLT '25 to explore pressing emerging challenges and opportunities for theory to bring clarity.
Announcing the first workshop on Foundations of Post-Training (FoPT) at COLT 2025! 📝 Soliciting abstracts/posters exploring theoretical & practical aspects of post-training and RL with language models! │ 🗓️ Deadline: May 19, 2025
We’ve been cooking... a new open weights 14B Phi-4 reasoning model, SFT’d on ~1.4M carefully curated reasoning demonstrations from o3-mini and RL’d for a tiny bit. This model is a little beast.
If Trump's tariff formula is indeed ChatGPT generated, it might be the most ingenious way for AI to destroy humanity. No need for Skynet or anything, just generate a dumb equation :) On a serious note, policy-makers should better be careful using AI tools. theverge.com/news/642620/tr…
innocent *after* proven guilty. such a joke
Today I was issued a full and unconditional pardon by @realDonaldTrump himself. He called me personally to tell me. This pardon is not just about me—it’s about every American who has been railroaded by the government, and unfortunately, that’s a lot of people. It is no wonder…
I am happy to announce our new paper "Univariate-guided sparse regression". It's a new lasso that leverages the signs and magnitude of univariate coefficients . Sparser and more interpretable than the lasso. We're excited! arxiv.org/abs/2501.18360 R: github.com/trevorhastie/u…
This is the ultimate dick boss move from Musk - except he isn’t even the boss, he’s just a dick.
Ironic that we got free AI from a hedge fund and $200/month AI from a nonprofit.
📢 ICML Call for Papers is out! Excited to serve as a program chair this year along with Maryam Fazel, @djhsu, and @SimonLacosteJ. See 🧵below for links & a summary of changes this year: