12 citations · 27 across the 9 of their papers we have counts for
11 papers
A Journey to the Edge of Stability
Jaerin Lee, Kyoung Mu Lee
It has recently been found that deep learning often occurs at the "edge of stability (EoS)," where the maximum Hessian eigenvalue of the model is stabilized at a value reciprocal t…
The Road Taken: The Role of Optimizers at the Edge of Stability
Jaerin Lee, Kyoung Mu Lee
The edge of stability refers to a phenomenon in deep learning with gradient-based optimizers where the Hessian eigenvalues of the loss remain stable above a threshold that the clas…
Vector Scaffolding: Inter-Scale Orchestration for Differentiable Image Vectorization
Jaerin Lee, Kanggeon Lee, Kyoung Mu Lee
Differentiable vector graphics have enabled powerful gradient-based optimization of vector primitives directly from raster images. However, existing frameworks formulate this as a…
Greedy Alignment Principle for Optimizer Selection
Jaerin Lee, Kyoung Mu Lee
Recent works have shown that gradient-update alignment is a powerful signal for modulating optimizer updates, often leading to faster training. We promote this update-wise heuristi…
MARS2 2025 Challenge on Multimodal Reasoning: Datasets, Methods, Results, Discussion, and Outlook
Peng Xu, Shengwu Xiong, Jiajun Zhang +125
This paper reviews the MARS2 2025 Challenge on Multimodal Reasoning. We aim to bring together different approaches in multimodal machine learning and LLMs via a large benchmark. We…
Grokfast: Accelerated Grokking by Amplifying Slow Gradients
Jaerin Lee, Bong Gyun Kang, Kihoon Kim +1
One puzzling artifact in machine learning dubbed grokking is where delayed generalization is achieved tenfolds of iterations after near perfect overfitting to the training data. Fo…
SemanticDraw: Towards Real-Time Interactive Content Creation from Image Diffusion Models
Jaerin Lee, Daniel Sungho Jung, Kanggeon Lee +1
We introduce SemanticDraw, a new paradigm of interactive content creation where high-quality images are generated in near real-time from given multiple hand-drawn regions, each enc…
Rethinking RGB Color Representation for Image Restoration Models
Jaerin Lee, JoonKyu Park, Sungyong Baik +1
Image restoration models are typically trained with a pixel-wise distance loss defined over the RGB color representation space, which is well known to be a source of blurry and unr…
LucidDreamer: Domain-free Generation of 3D Gaussian Splatting Scenes
Jaeyoung Chung, Suyoung Lee, Hyeongjin Nam +2
With the widespread usage of VR devices and contents, demands for 3D scene generation techniques become more popular. Existing 3D scene generation models, however, limit the target…
Clean Images are Hard to Reblur: Exploiting the Ill-Posed Inverse Task for Dynamic Scene Deblurring
Seungjun Nah, Sanghyun Son, Jaerin Lee +1
The goal of dynamic scene deblurring is to remove the motion blur in a given image. Typical learning-based approaches implement their solutions by minimizing the L1 or L2 distance…
AIM 2020 Challenge on Video Temporal Super-Resolution
Sanghyun Son, Jaerin Lee, Seungjun Nah +2
Videos in the real-world contain various dynamics and motions that may look unnaturally discontinuous in time when the recordedframe rate is low. This paper reports the second AIM…