Alert button
Picture for Yair Carmon

Yair Carmon

Alert button

Language models scale reliably with over-training and on downstream tasks

Mar 13, 2024
Samir Yitzhak Gadre, Georgios Smyrnis, Vaishaal Shankar, Suchin Gururangan, Mitchell Wortsman, Rulin Shao, Jean Mercat, Alex Fang, Jeffrey Li, Sedrick Keh, Rui Xin, Marianna Nezhurina, Igor Vasiljevic, Jenia Jitsev, Alexandros G. Dimakis, Gabriel Ilharco, Shuran Song, Thomas Kollar, Yair Carmon, Achal Dave, Reinhard Heckel, Niklas Muennighoff, Ludwig Schmidt

Viaarxiv icon

The Price of Adaptivity in Stochastic Convex Optimization

Feb 16, 2024
Yair Carmon, Oliver Hinder

Viaarxiv icon

A Whole New Ball Game: A Primal Accelerated Method for Matrix Games and Minimizing the Maximum of Smooth Functions

Nov 17, 2023
Yair Carmon, Arun Jambulapati, Yujia Jin, Aaron Sidford

Figure 1 for A Whole New Ball Game: A Primal Accelerated Method for Matrix Games and Minimizing the Maximum of Smooth Functions
Figure 2 for A Whole New Ball Game: A Primal Accelerated Method for Matrix Games and Minimizing the Maximum of Smooth Functions
Viaarxiv icon

Gradient Descent Monotonically Decreases the Sharpness of Gradient Flow Solutions in Scalar Networks and Beyond

May 22, 2023
Itai Kreisler, Mor Shpigel Nacson, Daniel Soudry, Yair Carmon

Figure 1 for Gradient Descent Monotonically Decreases the Sharpness of Gradient Flow Solutions in Scalar Networks and Beyond
Figure 2 for Gradient Descent Monotonically Decreases the Sharpness of Gradient Flow Solutions in Scalar Networks and Beyond
Figure 3 for Gradient Descent Monotonically Decreases the Sharpness of Gradient Flow Solutions in Scalar Networks and Beyond
Figure 4 for Gradient Descent Monotonically Decreases the Sharpness of Gradient Flow Solutions in Scalar Networks and Beyond
Viaarxiv icon

DataComp: In search of the next generation of multimodal datasets

May 03, 2023
Samir Yitzhak Gadre, Gabriel Ilharco, Alex Fang, Jonathan Hayase, Georgios Smyrnis, Thao Nguyen, Ryan Marten, Mitchell Wortsman, Dhruba Ghosh, Jieyu Zhang, Eyal Orgad, Rahim Entezari, Giannis Daras, Sarah Pratt, Vivek Ramanujan, Yonatan Bitton, Kalyani Marathe, Stephen Mussmann, Richard Vencu, Mehdi Cherti, Ranjay Krishna, Pang Wei Koh, Olga Saukh, Alexander Ratner, Shuran Song, Hannaneh Hajishirzi, Ali Farhadi, Romain Beaumont, Sewoong Oh, Alex Dimakis, Jenia Jitsev, Yair Carmon, Vaishaal Shankar, Ludwig Schmidt

Figure 1 for DataComp: In search of the next generation of multimodal datasets
Figure 2 for DataComp: In search of the next generation of multimodal datasets
Figure 3 for DataComp: In search of the next generation of multimodal datasets
Figure 4 for DataComp: In search of the next generation of multimodal datasets
Viaarxiv icon

DoG is SGD's Best Friend: A Parameter-Free Dynamic Step Size Schedule

Feb 08, 2023
Maor Ivgi, Oliver Hinder, Yair Carmon

Figure 1 for DoG is SGD's Best Friend: A Parameter-Free Dynamic Step Size Schedule
Figure 2 for DoG is SGD's Best Friend: A Parameter-Free Dynamic Step Size Schedule
Figure 3 for DoG is SGD's Best Friend: A Parameter-Free Dynamic Step Size Schedule
Figure 4 for DoG is SGD's Best Friend: A Parameter-Free Dynamic Step Size Schedule
Viaarxiv icon

ReSQueing Parallel and Private Stochastic Convex Optimization

Jan 01, 2023
Yair Carmon, Arun Jambulapati, Yujia Jin, Yin Tat Lee, Daogao Liu, Aaron Sidford, Kevin Tian

Figure 1 for ReSQueing Parallel and Private Stochastic Convex Optimization
Figure 2 for ReSQueing Parallel and Private Stochastic Convex Optimization
Figure 3 for ReSQueing Parallel and Private Stochastic Convex Optimization
Viaarxiv icon

Malign Overfitting: Interpolation Can Provably Preclude Invariance

Nov 28, 2022
Yoav Wald, Gal Yona, Uri Shalit, Yair Carmon

Figure 1 for Malign Overfitting: Interpolation Can Provably Preclude Invariance
Figure 2 for Malign Overfitting: Interpolation Can Provably Preclude Invariance
Figure 3 for Malign Overfitting: Interpolation Can Provably Preclude Invariance
Figure 4 for Malign Overfitting: Interpolation Can Provably Preclude Invariance
Viaarxiv icon

RECAPP: Crafting a More Efficient Catalyst for Convex Optimization

Jun 17, 2022
Yair Carmon, Arun Jambulapati, Yujia Jin, Aaron Sidford

Figure 1 for RECAPP: Crafting a More Efficient Catalyst for Convex Optimization
Figure 2 for RECAPP: Crafting a More Efficient Catalyst for Convex Optimization
Viaarxiv icon