
AI Research Scientist @nvidia; PhD; Science is everyone's but opinions are my own
Joined July 2018
- Tweets571
- Following405
- Followers2.3K
- Likes9.8K
Pinned Tweet
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 ๐
Itay Evron retweeted
Are you interested in a PhD in deep learning scaling and optimisation? I'm seeking students to join me at University of Bath, in order to expand the literature on automatic / measurement-based approaches to hyperparameter tuning. See below for motivation.
findaphd.com/phds/project/hyโฆ
Itay Evron retweeted
In nonsmooth nonconvex stochastic optimization, existing results suggested that constraints make the problem harder.
In a new paper with Guy Kornowski, we show the oracle complexity is in fact unchanged.
Itay Evron retweeted
RAG and long context are usually treated as rival approaches, but both pick the evidence that matters, just at different scales.
What if one LLM did both?
UNREAL: one frozen LLM retrieves from a 3B-token corpus and filters its own 128K prompt.
arxiv.org/abs/2610.08463
Itay Evron retweeted
My dream had always been to prove beautiful theorems as the AI agents unloaded my dishwasher. Well, it went the wrong way... ๐
Itay Evron retweeted
Can stochastic contexts make adversarial online learning computationally as easy as statistical learning?
We answer this affirmatively for binary prediction, with an application to contextual bandits.
With @HazanPrinceton
Itay Evron retweeted
Junior scholars: "I feel awkward citing myself"
Senior scholars: "as I cleverly argued (1988; 1991), admirably reiterated (1993; 1995; 1996); and handsomely concluded (2001; 2004; 2007)..."
Weโre sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics.
The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra.
The problem concerns whether the description of smooth three-dimensional fluid motion modeled by the Navier-Stokes equations can break down. It has remained unresolved for roughly 90 years.
Itay Evron retweeted
Iโm looking to recruit students through @ELLISforEurope
Please consider applying and mention my name:
ellis.eu/research/phd-postdoโฆ
Main areas of interest:
- AI Interpretability, Control, Safety, Trustworthiness
- AI for Science
- Multi-agent communication
Itay Evron retweeted
Today we remember chemistry laureate Ada Yonath who passed away yesterday at the age of 87.
At the beginning of the twentieth century, the chemical foundations for life were a mystery. Today we know how many of the most important processes function, all the way down to the atomic level. This is due in part to the work of biochemist, Ada Yonath.
In the 1970s, Yonath began a project that culminated in 2000 with her successful mapping (together with other researchers) of the structure of ribosomes โ the cellโs own protein factory โ by using x-ray crystallography. Among other applications, this has been important in the production of antibiotics.
In 2009, Yonath shared the Nobel Prize in Chemistry with Venkatraman Ramakrishnan and Thomas Steitz for studies of the structure and function of the ribosome.
Read more: bit.ly/4kJscDU
Happy to share that I joined @nvidia as an AI Research Scientist! ๐ฅณ
1/11 Can a stronger coding model rescue a run started by a cheaper one?
Can a cheaper model continue a stronger model's work - retaining quality at lower cost?
The two directions behave very differently...
Check out our new paper: The Handoff Tax ๐งต
๐ arxiv.org/abs/2608.24358
Itay Evron retweeted
ืืืคืฉ ืืช ืืืื\ืช - ืืืื ืืขืืื ืืืชื.
ืืืืก ืจืืฉืื ืฉืื ื. ืืฉืื ืืฉืืชืืคืื!
ื-8 ืืืืฉืื ืืืืจืื ืื ืื ื ืื ืขื ืืชืคืจ ืฉื ืืืจืกื, ืืื ืืืืืื ืAI.
ืื ืืชื:
ืืฆืืื ืื ืืืืืช.
ืื ืฉื ML ืขื ืืืข ืขืืืง ืืืฉืื, ืืืืืื ืืืืืจื ืกืืจืืช ืืื ืื ืืืชืืช.
ืื ืืคืืื ืืชืื ืืืืืช ืืืจืืื ืขื ืืขืจืืืช ืฉ(ืืืจ)ืืืืืืช ืืืืืื ื ืืืืจืื. >>
Itay Evron retweeted
Always stunning to see how much the quality of NeurIPS reviews depends on the subcommunity a paper is sent to. But the worst fate is for papers that do not clearly belong to any community: they end up torn apart in a no-manโs-land of confident misunderstanding and bitterness.
Itay Evron retweeted
We're hosting the workshop on Interpretability for Discovery๐ฌ at NeurIPS 2026, Atlanta!
AI models predict protein folding, forecast weather patterns, and more. They might be learning new patterns we haven't figured out yet. Can we use interpretability to read out these patterns?


























