Yifang Chen "陈奕方"

I am Yifang Chen "陈奕方", a student with a strong focus on research in Large Language Models, Machine Learning Theory, Transformers, and Attention Mechanism.

Yifang 'Michael' Chen

Education

  • University of Chicago
    University of ChicagoM.S. in Applied Data Science2024 - Present
  • New York University
    New York UniversityB.S. in Computer Science2020 - 2024
  • North Broward Preparatory School
    North Broward Preparatory SchoolHigh School2016 - 2020

News

May 2025Research

Universal Approximation of VAR accepted at ICML 2025

My paper on Universal Approximation of VAR was accepted at ICML 2025, here is the publication link: https://arxiv.org/abs/2502.06167

May 2025Research

Attend the CPAL 2025 conference at Stanford University

I attended the CPAL 2025 conference at Stanford University, here is the conference link: https://cpal.cc/

January 2025Research

Mamba paper accepted at CPAL Oral 2025

My paper on Mamba was accepted at the CPAL Oral 2025, here is the publication link: https://openreview.net/forum?id=bImlLT3r62

Publications:

Thumbnail for Universal Approximation of Visual Autoregressive Transformers

Universal Approximation of Visual Autoregressive Transformers

Authors: Yifang Chen, Xiaoyu Li, Yingyu Liang, Zhenmei Shi, Zhao Song.

Publication: ICML 2025: Forty-Second International Conference on Machine Learning

View Paper

Thumbnail for The Computational Limits of State-Space Models and Mamba via the Lens of Circuit Complexity

The Computational Limits of State-Space Models and Mamba via the Lens of Circuit Complexity

Authors: Yifang Chen, Xiaoyu Li, Yingyu Liang, Zhenmei Shi, Zhao Song.

Publication: CPAL 2025 (Oral): Second Conference on Parsimony and Learning

View Paper

Manuscripts:

Thumbnail for Limits of KV Cache Compression for Tensor Attention based Autoregressive Transformers

Limits of KV Cache Compression for Tensor Attention based Autoregressive Transformers

Authors: Yifang Chen, Xiaoyu Li, Yingyu Liang, Zhenmei Shi, Zhao Song, Yu Tian

Preprint: arXiv(2025)

View Paper

Thumbnail for Fast Gradient Computation for RoPE Attention in Almost Linear Time

Fast Gradient Computation for RoPE Attention in Almost Linear Time

Authors: Yifang Chen, Jiayan Huo, Xiaoyu Li, Yingyu Liang, Zhenmei Shi, Zhao Song.

Publication: ICLR SCOPE Workshop 2025

View Paper

Misc. Interests

Martial Arts

Outside of academic pursuits, I am passionate about martial arts. I've been actively training in:

  • Brazilian Jiu-JitsuBlue Belt • Ground fighting and submission grappling
  • BoxingDeveloping striking technique and footwork
  • KickboxingExpanding my striking arsenal

BJJ has become my primary focus, and I'm proud to have earned my blue belt through consistent training and dedication.

Brazilian Jiu-Jitsu Training
Blue Belt