Chanhyuk Lee

I am Chanhyuk David Lee (이찬혁), a Master’s student at KAIST School of Computing. I’m currently working in VL Lab, which is led by Prof. Seunghoon Hong. Before joining the lab, I received my Bachelor’s degree from KAIST in a double major of Computer Science and Chemistry.

I’m interested in neural network scaling, particularly developing efficient approaches to build large models with limited resources. My current work focuses on model fusion and parameter alignment, exploring how different networks can be combined while preserving their capabilities. For these practical approaches, I am deeply interested in theoretical developments such as infinite-size network frameworks like Neural Tangent Kernel and loss landscape theories like mode connectivity.

Education

Publications

Universal Few-Shot Spatial Control for Diffusion Models

Kiet Nguyen Tuan, Chanhyuk Lee, Donggyun Kim, Donghoon Lee, Seunghoon Hong

Preprint

AdaRank: Adaptive Rank Pruning for Enhanced Model Merging

Chanhyuk Lee, Jiho Choi, Chanryeol Lee, Donggyun Kim, Seunghoon Hong

Preprint

arXiv Code

Revisiting Weight Averaging for Model Merging

Jiho Choi, Donggyun Kim, Chanhyuk Lee, Seunghoon Hong

Preprint

arXiv Code