Peter Richtarik

Boycotting the grocery store until they start buying an equivalent volume of groceries from me

Hiring: richtarik.org/i_join.html

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

papercopilot.com/paper-list/neu…

Place 8.-14. for my Optimization and Machine Learning Lab at KAUST, Saudi Arabia

Super proud of my brilliant and hard-working team!

richtarik.org/i_team.html

Random photo of Saudi Arabia.

Random photo of KAUST.

Random photo of KAUST

Random photo of KAUST.

8.

7.

6.

5.

4.

3.

2.

A random photo of KAUST

Random photo of KAUST

Randomized Iterative Methods for Linear Systems We develop a novel, fundamental and surprisingly simple randomized iterative method for solving consistent linear systems. Our method has six different but equivalent interpretations: sketch-and-proje...

the coordinate-free rand Kaczmarz is not new; see arxiv.org/abs/1506.03296 & follow-up work

Random photo of KAUST

People from my lab attending: Omar Shaikh Omar, Hanmin Li, Kai Yi, Abdurakhmon Sadiev, Kaja Gruntkowska, Egor Shulgin, Artavazd Maranjyan and Marta Pozzi (former intern).

I am not attending NeurIPS this year but many of my students are!

Here are our papers; check them out!

Want to join my lab as an intern, MS/PhD student, PhD student or a postdoc? Apply here: apply.interfolio.com/105097
Plus read this, too: richtarik.org/i_join.html

When 4 out of 5 reviews are "borderline reject", the authors are on the verge of withdrawing the paper, but then, against all odds, the only "accept" reviewer decides to champion the paper.

true

Random photo of KAUST

Simons Institute for the Theory of Computing and SLMath Joint Workshop: AI for Mathematics and Theoretical Computer Science This is an exciting time for mathematics, as new technologies for mathematical reasoning provide novel opportunities for mathematical research, communication, and discovery. Mathlib, a library of form...

*AI for Mathematics and Theoretical Computer Science*, a workshop hosted jointly by the Simons Institute for the Theory of Computing and the Simons Laufer Mathematical Sciences Institute, will be held in Berkeley at April 7-11, 2025.

simons.berkeley.edu/workshops/si...

Random photo of KAUST

Pushing the Limits of Large Language Model Quantization via the Linearity Theorem Quantizing large language models has become a standard way to reduce their memory and computational costs. Typically, existing methods focus on breaking down the problem into individual layer-wise sub...

Paper: Pushing the Limits of Large Language Model Quantization via the Linearity Theorem ( arxiv.org/abs/2411.17525 )
PR to Huggingface Transformers: github.com/huggingface/...

HIGGS - A simple data-free LLM quantization method using Hadamard rotations and MSE-optimal grids, which outperforms all prior data-free approaches such as the extremely popular NF4 quantized format.

One of the author: @richtarik.bsky.social

Random photo of KAUST

Random photo of KAUST.