Arithmetic Transformers Can Length-Generalize in Both Operand Length and Count
π₯ Hanseul Cho*, Jaeyoung Cha*, Srinadh Bhojanapalli*, and Chulhee Yun
- π Seoul, South Korea
- π CV
- Github
- Google Scholar
- Semantic Scholar
&edge_flat=false)
Publications by Category
Arxiv
ICLR
SGDA with Shuffling: Faster Convergence For Nonconvex-PΕ Minimax Optimization
π₯ Hanseul Cho and Chulhee Yun
π° ICLR 2023 π π [paper] [arxiv]
π NAVER Outstanding Theory Paper Award at the 7th Joint Conference of Korea Artificial Intelligence Association (JKAIA 2022)
ICLR Workshop
Fundamental Benefit of Alternating Updates in Minimax Optimization
π₯ Jaewook Lee*, Hanseul Cho*, and Chulhee Yun
π° ICML 2024 (Short version at ICLR 2024 Workshop on Bridging the Gap Between Practice and Theory in Deep Learning (BGPT)) π π [paper] [arxiv] [code]
π Spotlight @ ICML 2024 (Top 3.5% among total submissions)
ICML
Fundamental Benefit of Alternating Updates in Minimax Optimization
π₯ Jaewook Lee*, Hanseul Cho*, and Chulhee Yun
π° ICML 2024 (Short version at ICLR 2024 Workshop on Bridging the Gap Between Practice and Theory in Deep Learning (BGPT)) π π [paper] [arxiv] [code]
π Spotlight @ ICML 2024 (Top 3.5% among total submissions)
ICML Workshop
DASH: Warm-Starting Neural Network Training in Stationary Settings without Loss of Plasticity
π₯ Baekrok Shin*, Junsoo Oh*, Hanseul Cho, and Chulhee Yun
π° NeurIPS 2024 (Short version at ICML 2024 Workshop on Advancing Neural Network Training (WANT)) π π [paper] [arxiv] [code]
Position Coupling: Improving Length Generalization of Arithmetic Transformers Using Task Structure
π₯ Hanseul Cho*, Jaeyoung Cha*, Pranjal Awasthi, Srinadh Bhojanapalli, Anupam Gupta, and Chulhee Yun
π° NeurIPS 2024 (Short version at ICML 2024 Workshop on Long-Context Foundation Models (LCFM)) π π [paper] [arxiv] [code]
KAIA
SGDA with Shuffling: Faster Convergence For Nonconvex-PΕ Minimax Optimization
π₯ Hanseul Cho and Chulhee Yun
π° ICLR 2023 π π [paper] [arxiv]
π NAVER Outstanding Theory Paper Award at the 7th Joint Conference of Korea Artificial Intelligence Association (JKAIA 2022)
Convergence and Implicit Bias of Gradient Descent on Continual Linear Classification
π₯ Hyunji Jung*, Hanseul Cho*, and Chulhee Yun
π° Under Review π
π Best Paper Award at the 11th Joint Conference of Korean Artificial Intelligence Association (JKAIA 2024)
NeurIPS
PLASTIC: Improving Input and Label Plasticity for Sample Efficient Reinforcement Learning
π₯ Hojoon Lee*, Hanseul Cho*, Hyunseung Kim*, Daehoon Gwak, Joonkee Kim, Jaegul Choo, Se-Young Yun, and Chulhee Yun
π° NeurIPS 2023 π π [paper] [arxiv] [code]
Fair Streaming Principal Component Analysis: Statistical and Algorithmic Viewpoint
π₯ Junghyun Lee*, Hanseul Cho*, Se-Young Yun, and Chulhee Yun
π° NeurIPS 2023 π π [paper] [arxiv] [code]
DASH: Warm-Starting Neural Network Training in Stationary Settings without Loss of Plasticity
π₯ Baekrok Shin*, Junsoo Oh*, Hanseul Cho, and Chulhee Yun
π° NeurIPS 2024 (Short version at ICML 2024 Workshop on Advancing Neural Network Training (WANT)) π π [paper] [arxiv] [code]
Position Coupling: Improving Length Generalization of Arithmetic Transformers Using Task Structure
π₯ Hanseul Cho*, Jaeyoung Cha*, Pranjal Awasthi, Srinadh Bhojanapalli, Anupam Gupta, and Chulhee Yun
π° NeurIPS 2024 (Short version at ICML 2024 Workshop on Long-Context Foundation Models (LCFM)) π π [paper] [arxiv] [code]