Qingyun Li(he/him)

About

I am a Computer Science student at the University of Washington. My current work studies how language models learn, represent, and expose structure through tokenization, decoding, and representation geometry. My earlier published work was in computational photography, which was my first route into computer vision and applied AI.

Language Models Tokenization Decoding Representation Learning Computer Vision Machine Learning

Publications

Language Models Detokenize, But Only for Themselves

Qingyun Li, Ruoyun Li

AACL-IJCNLP 2026 Main track New

Education

2025-present

University of Washington, Seattle

Computer Science, Paul G. Allen School of Computer Science & Engineering.

2023-2025

Everett Community College

Associate of Arts and Sciences in Mathematics, President's List, 4.0 GPA.

2025

High School

Graduated in ninth grade with a 4.0 GPA.

Random Stuff

2026

Kyril Faenov Endowed Scholarship in Computer Science & Engineering

Awarded July 2026.

2025

Desmos Art Contest 2024 Winner

Selected in February 2025 as one of 20 graphs from more than 10,000 submissions.

2023

Perfect ACT score of 36

Earned at age 11 in September 2023.

2021-2023

USACO Gold Division

Reached Silver in Dec 2021 with a perfect score at age 10, Reached Gold in Feb 2023 at 11.

Patents

Patent information forthcoming.