activity
20182021
most citedAre Transformers universal approximators of sequence-to-sequence functions?

44 citations · 82 across the 5 of their papers we have counts for

collaborators
Showing cs.LGShow all

14 papers · 1 filter

cs.LG2023

Linear attention is (maybe) all you need (to understand transformer optimization)

Kwangjun Ahn, Xiang Cheng, Minhak Song +3

Transformer training is notoriously difficult, requiring a careful design of optimizers and use of various heuristics. We make progress towards understanding the subtleties of trai…

cs.LG2023

Trajectory Alignment: Understanding the Edge of Stability Phenomenon via Bifurcation Theory

Minhak Song, Chulhee Yun

Cohen et al. (2021) empirically study the evolution of the largest eigenvalue of the loss Hessian, also known as sharpness, along the gradient descent (GD) trajectory and observe t…

cs.LG2023

PLASTIC: Improving Input and Label Plasticity for Sample Efficient Reinforcement Learning

Hojoon Lee, Hanseul Cho, Hyunseung Kim +5

In Reinforcement Learning (RL), enhancing sample efficiency is crucial, particularly in scenarios when data acquisition is costly and risky. In principle, off-policy RL algorithms…

cs.LG2023

Practical Sharpness-Aware Minimization Cannot Converge All the Way to Optima

Dongkuk Si, Chulhee Yun

Sharpness-Aware Minimization (SAM) is an optimizer that takes a descent step based on the gradient at a perturbation $y_t = x_t + ρ\frac{\nabla f(x_t)}{\lVert \nabla f(x_t) \rVert}…

cs.LG20212 cited

Can Single-Shuffle SGD be Better than Reshuffling SGD and GD?

Chulhee Yun, Suvrit Sra, Ali Jadbabaie

We propose matrix norm inequalities that extend the Recht-Ré (2012) conjecture on a noncommutative AM-GM inequality by supplementing it with another inequality that accounts for si…

cs.LG2020

Provable Memorization via Deep Neural Networks using Sub-linear Parameters

Sejun Park, Jaeho Lee, Chulhee Yun +1

It is known that parameters are sufficient for neural networks to memorize arbitrary input-label pairs. By exploiting depth, we show that parameters suffice…