Itay Evron
@itayevron
Followers
774
Following
9K
Media
53
Statuses
485
PhD; Research Scientist @Meta (opinions are my own)
Joined July 2018
Why does L1 regularization induce sparse models? Many illustrate this using the least squares problem with a norm constraint. The least squares level sets are drawn next to the different unit "circles". I prepared a cool animation which I believe makes it even clearer 🙂
12
228
1K
@wrotator Elon's physique reflects functional strength from rigorous schedules, including judo and weight training, prioritizing endurance over aesthetics—evident in his sustained high-output lifestyle without visible excess bulk. His mind fuses first-principles reasoning from physics with
54
1K
20K
Congratulations to Professor Nati Srebro, who has been named the new Lenore Blum Professor of Computer Science at TTIC. Read the full announcement: https://t.co/xFhl4Qikqe
0
1
24
I feel like I should balance some of the recent criticism about #ICLR2026 review quality. Our paper received four reviews, all surprisingly high-quality. Even though not all were positive, they were knowledgeable, and pointed out concrete issues that we can now try to address.
5
2
93
The next generation of paper writing LLMs will be trained in a GAN-like fashion
Has anyone run LLM detection on all ICLR papers and reviews? If not I'm willing to offer a bounty of $50 to the first person who does it (well). Happy to have any other people chip in 😃
1
0
4
Such behavior must be penalized, especially for reciprocal reviewers
I wasted 3 hours of my life today 😃 An @iclr_conf review, gave 2/10 only because "these 7 papers already showed what you showed" 3/7 don't exist (also confirmed w an author) 3/7 don't do anything like what we do. 1/7 is similar, but far from the same takeaway. So productive.
0
0
5
I am an AC for ICLR 2026. One of the papers in my batch was just withdrawn. The authors wrote a brief response, explaining why the reviewers failed at their job. I agree with most of their comments. The authors gave up. They are fed up. Just like many of us. I understand. We
34
203
1K
Talking tomorrow at the RSRG/FALCON seminar (4pm, ANB 209) in beautiful @Caltech on our new approach to online control 🤩 come say hi if you’re around!
Excited about new approach for control in the spectral domain, provably via gradient methods, w. my students Anand Brahmbhatt, @gon_buzaglo and Sofiia Druchyna: https://t.co/klZCfgdb1D
0
1
7
Mathematicians handling linear algebra to the machine learning community.
129
2K
21K
ChatGPT-5's level of analysis is mind blowing. Already an enormous force multiplier for theoreticians.
1
0
9
Training Data Attribution (TDA) should account for learning dynamics! The same data can influence model behavior in dramatically different ways at different time points of training. We call for a shift towards stagewise data attribution and the study of influence dynamics. 1/11
1
11
37
Are Greedy Task Orderings Better Than Random in Continual Linear Regression?.
arxiv.org
We analyze task orderings in continual learning for linear regression, assuming joint realizability of training data. We focus on orderings that greedily maximize dissimilarity between consecutive...
0
1
2
Honored to be selected as a 2025 Google PhD Fellow! Deeply grateful to my advisor, Prof. @YishayMansour, for his invaluable guidance and support, and to my collaborators and lab members for their ongoing inspiration.
🎉 We're excited to announce the 2025 Google PhD Fellows! @GoogleOrg is providing over $10 million to support 255 PhD students across 35 countries, fostering the next generation of research talent to strengthen the global scientific landscape. Read more: https://t.co/0Pvuv6hsgP
2
2
16
“How many years of AI research experience do you have?” “I published my first paper at NIPS.” * uncomfortable silence *
14
12
788
Excited to announce that The Workshop on Online Learning and Optimization 2025 will be held on Nov 10 (hybrid, @RIKEN_AIP 🇯🇵)! Invited speakers: 🎓 @NicoloCB (Univ. of Milan & Politecnico di Milano) 🎓 @mehtalearner (Univ. of Victoria) More info 👉
sites.google.com
Date: November 10, 2025 (JST) 9:30-19:30 Format: Hybrid (In-person + Online) In-person registration is now open! Book now (Registration deadline: Nov. 3, Mon.) We closed the in-person registration....
2
21
95
Excited to share our new paper https://t.co/kkPME5PeDm with @GolowichNoah and my advisor @HazanPrinceton. We introduce the Hidden Game Problem, motivated by AI alignment and language games. 🧵 1/
arxiv.org
This paper investigates a class of games with large strategy spaces, motivated by challenges in AI alignment and language games. We introduce the hidden game problem, where for each player, an...
2
20
117
Got yet another Google Scholar alert for a hallucinated citation of a paper I didn't write. This is the second time this has happened. Please don't use LLMs to write your related work section
14
38
398
[1/7] New paper: "Beyond Token Probes: Hallucination Detection via Activation Tensors with ACT-ViT" #NeurIPS2025 [ https://t.co/8aD1n8ejyp] Joint work with: @ffabffrasca (co-first), @yaniv_galron, @YftahZ , @HaggaiMaron
1
3
14
sharing a new paper w Peter Bartlett, @jasondeanlee, @ShamKakade6, Bin Yu ppl talking about implicit regularization, but how good is it? We show its surprisingly effective, that GD dominates ridge for all linear regression, w/ more cool stuff on GD vs SGD https://t.co/oAVKiVgUUQ
10
32
188