Understanding Why Adam Outperforms SGD: Gradient Heterogeneity in Transformers

31 January 2025

Papers citing "Understanding Why Adam Outperforms SGD: Gradient Heterogeneity in Transformers"

50 / 52 papers shown

Title
Understanding Adam Requires Better Rotation Dependent Assumptions Lucas Maes Tianyue H. Zhang Alexia Jolicoeur-Martineau Ioannis Mitliagkas Damien Scieur Simon Lacoste-Julien Charles Guille-Escuret 44 3 0 25 Oct 2024
What Does It Mean to Be a Transformer? Insights from a Theoretical Hessian Analysis Weronika Ormaniec Felix Dangel Sidak Pal Singh 99 7 0 14 Oct 2024
Deconstructing What Makes a Good Optimizer for Language Models Rosie Zhao Depen Morwani David Brandfonbrener Nikhil Vyas Sham Kakade 81 22 0 10 Jul 2024
Adam-mini: Use Fewer Learning Rates To Gain More Yushun Zhang Congliang Chen Ziniu Li Tian Ding Chenwei Wu Yinyu Ye Zhi-Quan Luo Ruoyu Sun 75 52 0 24 Jun 2024
On the Role of Attention Masks and LayerNorm in Transformers Xinyi Wu A. Ajorlou Yifei Wang Stefanie Jegelka Ali Jadbabaie 75 12 0 29 May 2024
Understanding Linear Probing then Fine-tuning Language Models from NTK Perspective Akiyoshi Tomihari Issei Sato 56 4 0 27 May 2024
$Implicit Bias of AdamW: $\ell_\infty$ Norm Constrained Optimization$ Implicit Bias of AdamW: $\ell_\infty$ Norm Constrained Optimization Shuo Xie Zhiyuan Li OffRL 66 20 0 05 Apr 2024
Cherry on Top: Parameter Heterogeneity and Quantization in Large Language Models Wanyun Cui Qianle Wang MQ 62 3 0 03 Apr 2024
Heavy-Tailed Class Imbalance and Why Adam Outperforms Gradient Descent on Language Models Frederik Kunstner Robin Yadav Alan Milligan Mark Schmidt Alberto Bietti 63 32 0 29 Feb 2024
Why Transformers Need Adam: A Hessian Perspective Yushun Zhang Congliang Chen Tian Ding Ziniu Li Ruoyu Sun Zhimin Luo 77 53 0 26 Feb 2024
Self-attention Networks Localize When QK-eigenspectrum Concentrates Han Bao Ryuichiro Hataya Ryo Karakida 32 5 0 03 Feb 2024
Outliers with Opposing Signals Have an Outsized Effect on Neural Network Optimization Elan Rosenfeld Andrej Risteski 44 11 0 07 Nov 2023
Mistral 7B Albert Q. Jiang Alexandre Sablayrolles A. Mensch Chris Bamford Devendra Singh Chaplot ... Teven Le Scao Thibaut Lavril Thomas Wang Timothée Lacroix William El Sayed MoE LRM 55 2,170 0 10 Oct 2023
Lion Secretly Solves Constrained Optimization: As Lyapunov Predicts Lizhang Chen Bo Liu Kaizhao Liang Qian Liu ODL 48 19 0 09 Oct 2023
Linear attention is (maybe) all you need (to understand transformer optimization) Kwangjun Ahn Xiang Cheng Minhak Song Chulhee Yun Ali Jadbabaie S. Sra 67 51 1 02 Oct 2023
Toward Understanding Why Adam Converges Faster Than SGD for Transformers Yan Pan Yuanzhi Li 112 43 0 31 May 2023
Noise Is Not the Main Factor Behind the Gap Between SGD and Adam on Transformers, but Sign Descent Might Be Frederik Kunstner Jacques Chen J. Lavington Mark Schmidt 75 71 0 27 Apr 2023
Stabilizing Transformer Training by Preventing Attention Entropy Collapse Shuangfei Zhai Tatiana Likhomanenko Etai Littwin Dan Busbridge Jason Ramapuram Yizhe Zhang Jiatao Gu J. Susskind AAML 73 73 0 11 Mar 2023
Symbolic Discovery of Optimization Algorithms Xiangning Chen Chen Liang Da Huang Esteban Real Kaiyuan Wang ... Xuanyi Dong Thang Luong Cho-Jui Hsieh Yifeng Lu Quoc V. Le 136 373 0 13 Feb 2023
How Does Adaptive Optimization Impact Local Neural Network Geometry? Kaiqi Jiang Dhruv Malik Yuanzhi Li 89 18 0 04 Nov 2022
Scratching Visual Transformer's Back with Uniform Attention Nam Hyeon-Woo Kim Yu-Ji Byeongho Heo Doonyoon Han Seong Joon Oh Tae-Hyun Oh 448 23 0 16 Oct 2022
Robustness to Unbounded Smoothness of Generalized SignSGD M. Crawshaw Mingrui Liu Francesco Orabona Wei Zhang Zhenxun Zhuang AAML 71 72 0 23 Aug 2022
Adam Can Converge Without Any Modification On Update Rules Yushun Zhang Congliang Chen Naichen Shi Ruoyu Sun Zhimin Luo 36 67 0 20 Aug 2022
Signal Propagation in Transformers: Theoretical Perspectives and the Role of Rank Collapse Lorenzo Noci Sotiris Anagnostidis Luca Biggio Antonio Orvieto Sidak Pal Singh Aurelien Lucchi 76 72 0 07 Jun 2022
Revisiting Over-smoothing in BERT from the Perspective of Graph Han Shi Jiahui Gao Hang Xu Xiaodan Liang Zhenguo Li Lingpeng Kong Stephen M. S. Lee James T. Kwok 56 74 0 17 Feb 2022
Revisiting Parameter-Efficient Tuning: Are We Really There Yet? Guanzheng Chen Fangyu Liu Zaiqiao Meng Shangsong Liang 45 93 0 16 Feb 2022
How Do Vision Transformers Work? Namuk Park Songkuk Kim ViT 73 477 0 14 Feb 2022
Escaping the Gradient Vanishing: Periodic Alternatives of Softmax in Attention Mechanism Shulun Wang Bin Liu Feng Liu 107 16 0 16 Aug 2021
RealFormer: Transformer Likes Residual Attention Ruining He Anirudh Ravula Bhargav Kanagal Joshua Ainslie 63 109 0 21 Dec 2020
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale Alexey Dosovitskiy Lucas Beyer Alexander Kolesnikov Dirk Weissenborn Xiaohua Zhai ... Matthias Minderer G. Heigold Sylvain Gelly Jakob Uszkoreit N. Houlsby ViT 517 40,739 0 22 Oct 2020
Big Bird: Transformers for Longer Sequences Manzil Zaheer Guru Guruganesh Kumar Avinava Dubey Joshua Ainslie Chris Alberti ... Philip Pham Anirudh Ravula Qifan Wang Li Yang Amr Ahmed VLM 499 2,074 0 28 Jul 2020
Descending through a Crowded Valley - Benchmarking Deep Learning Optimizers Robin M. Schmidt Frank Schneider Philipp Hennig ODL 69 165 0 03 Jul 2020
Longformer: The Long-Document Transformer Iz Beltagy Matthew E. Peters Arman Cohan RALM VLM 124 4,048 0 10 Apr 2020
On Layer Normalization in the Transformer Architecture Ruibin Xiong Yunchang Yang Di He Kai Zheng Shuxin Zheng Chen Xing Huishuai Zhang Yanyan Lan Liwei Wang Tie-Yan Liu AI4CE 112 988 0 12 Feb 2020
PyHessian: Neural Networks Through the Lens of the Hessian Z. Yao A. Gholami Kurt Keutzer Michael W. Mahoney ODL 45 302 0 16 Dec 2019
On Empirical Comparisons of Optimizers for Deep Learning Dami Choi Christopher J. Shallue Zachary Nado Jaehoon Lee Chris J. Maddison George E. Dahl 66 260 0 11 Oct 2019
RoBERTa: A Robustly Optimized BERT Pretraining Approach Yinhan Liu Myle Ott Naman Goyal Jingfei Du Mandar Joshi Danqi Chen Omer Levy M. Lewis Luke Zettlemoyer Veselin Stoyanov AIMat 498 24,351 0 26 Jul 2019
What Does BERT Look At? An Analysis of BERT's Attention Kevin Clark Urvashi Khandelwal Omer Levy Christopher D. Manning MILM 205 1,592 0 11 Jun 2019
Learning Deep Transformer Models for Machine Translation Qiang Wang Bei Li Tong Xiao Jingbo Zhu Changliang Li Derek F. Wong Lidia S. Chao 70 670 0 05 Jun 2019
Why gradient clipping accelerates training: A theoretical justification for adaptivity J.N. Zhang Tianxing He S. Sra Ali Jadbabaie 72 459 0 28 May 2019
BoolQ: Exploring the Surprising Difficulty of Natural Yes/No Questions Christopher Clark Kenton Lee Ming-Wei Chang Tom Kwiatkowski Michael Collins Kristina Toutanova 201 1,511 0 24 May 2019
SuperGLUE: A Stickier Benchmark for General-Purpose Language Understanding Systems Alex Jinpeng Wang Yada Pruksachatkun Nikita Nangia Amanpreet Singh Julian Michael Felix Hill Omer Levy Samuel R. Bowman ELM 222 2,305 0 02 May 2019
WiC: the Word-in-Context Dataset for Evaluating Context-Sensitive Meaning Representations Mohammad Taher Pilehvar Jose Camacho-Collados 152 485 0 28 Aug 2018
Neural Network Acceptability Judgments Alex Warstadt Amanpreet Singh Samuel R. Bowman 203 1,406 0 31 May 2018
GLUE: A Multi-Task Benchmark and Analysis Platform for Natural Language Understanding Alex Jinpeng Wang Amanpreet Singh Julian Michael Felix Hill Omer Levy Samuel R. Bowman ELM 822 7,141 0 20 Apr 2018
signSGD: Compressed Optimisation for Non-Convex Problems Jeremy Bernstein Yu Wang Kamyar Azizzadenesheli Anima Anandkumar FedML ODL 85 1,041 0 13 Feb 2018
Block-diagonal Hessian-free Optimization for Training Neural Networks Huishuai Zhang Caiming Xiong James Bradbury R. Socher ODL 28 22 0 20 Dec 2017
Attention Is All You Need Ashish Vaswani Noam M. Shazeer Niki Parmar Jakob Uszkoreit Llion Jones Aidan Gomez Lukasz Kaiser Illia Polosukhin 3DV 591 130,942 0 12 Jun 2017
Dissecting Adam: The Sign, Magnitude and Variance of Stochastic Gradients Lukas Balles Philipp Hennig 66 169 0 22 May 2017
Deep Residual Learning for Image Recognition Kaiming He Xinming Zhang Shaoqing Ren Jian Sun MedIm 1.8K 193,426 0 10 Dec 2015