123321

[123321 @w11123321]

@MennDennis @_akhaliq Nice work 👍

[CNN @CNN]

Supreme Court rules that Trump's sweeping emergency tariffs are illegal.
https://t.co/D4cEdbZsCH

[123321 @w11123321]

@yang_ML https://t.co/NQF4fi6jT7

[Peter Richtarik @peter_richtarik]

One of my rejected ICML 2025 (@icmlconf) papers. Can anyone spot any criticism in the metareview? What a joke 🙃

[Peter Richtarik @peter_richtarik]

- Will you get your PhD between May 15, 2025, and June 30, 2026?
- Are you looking for a postdoc position starting in Fall 2026?
- Do you want to work on the latest and coolest stuff related to optimization for machine learning (distributed training, model compression, LLMs, SGD, algorithms, theory, AI applications) in my lab at KAUST?

Apply for a Schmidt Science Fellowship!

110k USD / year for up to 2 years

KAUST deadline: May 1, 2025

[Keller Jordan @kellerjordan0]

This is an exciting moment: The world's first report on successful large-scale training with a super-Adamic optimizer. Congratulations to the @Kimi_Moonshot team and to every Muon contributor: @Yuchenj_UW @bozavlado @YouJiacheng @leloykun L. Newhouse @jxbz
https://t.co/fiu5OeoZGe

[Peter Richtarik @peter_richtarik]

I am hiring people: https://t.co/0ptZQfVyF2

[Peter Richtarik @peter_richtarik]

List of Top 100 NeurIPS 2024 authors in terms of # of accepted papers (kind of):

https://t.co/dT1a3Szt6L

Place 8.-14. for my Optimization and Machine Learning Lab at KAUST, Saudi Arabia (@KAUST_News @AI_KAUST @cemseKAUST @SDAIA_KAUST_AI)

Super proud of my brilliant and hard-working team!

https://t.co/J2OIFKtSoI

[Gilad @giladturok]

Amazingly *clear* monograph on Monte Carlo & sampling: https://t.co/fUeSAnvsGK.

Highly recommend skimming if you want to learn these topics!

[123321 @w11123321]

@AlexShtf Here is a plot of the (nonconvex) landscape: https://t.co/JSEVWhU3PV

[123321 @w11123321]

@AlexShtf If we change the variable as u = - exp(v), the problem becomes min_{w,u<0} {-u exp(f(w,x)) - log(-u)} - 1, which does not seem to be jointly convex in w and u.

[123321 @w11123321]

@AlexShtf We can discuss more in detail if you are interested.

[123321 @w11123321]

@AlexShtf We also thought about the variational form. However, the problem in the variational form is not jointly convex in w and v. As far as I know, there is no existing convergence theory on the min-min problem that is not jointly convex.

[MaLGa @malga_center]

Check out the new #malga_center paper "Iterative #regularization in classification via hinge loss diagonal descent" by V. Apidopoulos, T. Poggio, @lrntzrsc, S. Villa

https://t.co/ZjugeZBuZq

[123321 @w11123321]

@roydanroy Then the TMLR review system will break up in a couple of months/years after people flood in?