O(1) Communication for Distributed SGD through Two-Level Gradient Averaging

• URL: http://arxiv.org/abs/2006.07405v2
• Date: Tue, 16 Jun 2020 01:49:28 GMT
• Title: O(1) Communication for Distributed SGD through Two-Level Gradient Averaging
• Authors: Subhadeep Bhattacharya, Weikuan Yu and Fahim Tahmid Chowdhury
• Abstract summary: We introduce a strategy called two-level gradient averaging (A2SGD) to consolidate all gradients down to merely two local averages per worker. Our theoretical analysis shows that A2SGD converges similarly like the default distributed SGD algorithm.
• Score: 0.0
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large neural network models present a hefty communication challenge to distributed Stochastic Gradient Descent (SGD), with a communication complexity of O(n) per worker for a model of n parameters. Many sparsification and quantization techniques have been proposed to compress the gradients, some reducing the communication complexity to O(k), where k << n. In this paper, we introduce a strategy called two-level gradient averaging (A2SGD) to consolidate all gradients down to merely two local averages per worker before the computation of two global averages for an updated model. A2SGD also retains local errors to maintain the variance for fast convergence. Our theoretical analysis shows that A2SGD converges similarly like the default distributed SGD algorithm. Our evaluation validates the theoretical conclusion and demonstrates that A2SGD significantly reduces the communication traffic per worker, and improves the overall training time of LSTM-PTB by 3.2x and 23.2x, respectively, compared to Top-K and QSGD. To the best of our knowledge, A2SGD is the first to achieve O(1) communication complexity per worker for distributed SGD.

Related papers

• FusionLLM: A Decentralized LLM Training System on Geo-distributed GPUs with Adaptive Compression [55.992528247880685]
  Decentralized training faces significant challenges regarding system design and efficiency. We present FusionLLM, a decentralized training system designed and implemented for training large deep neural networks (DNNs) We show that our system and method can achieve 1.45 - 9.39x speedup compared to baseline methods while ensuring convergence.
  arXiv  Detail & Related papers  (2024-10-16T16:13:19Z)
• On the Convergence of (Stochastic) Gradient Descent for Kolmogorov--Arnold Networks [56.78271181959529]
  Kolmogorov--Arnold Networks (KANs) have gained significant attention in the deep learning community. Empirical investigations demonstrate that KANs optimized via gradient descent (SGD) are capable of achieving near-zero training loss.
  arXiv  Detail & Related papers  (2024-10-10T15:34:10Z)
• Scaling up Stochastic Gradient Descent for Non-convex Optimisation [5.908471365011942]
  We propose a novel approach to the problem of shared parallel computation. By combining two strategies into a unified framework, DPSGD is a better trade computation framework. The potential gains can be achieved by DPSGD on a deep learning (DRL) problem (Latent Diletrichal inference) and on a deep learning (DRL) problem (advantage actor - A2C)
  arXiv  Detail & Related papers  (2022-10-06T13:06:08Z)
• A Communication-efficient Algorithm with Linear Convergence for Federated Minimax Learning [1.713291434132985]
  We study a large-scale multi-agent minimax optimization problem, which models Geneimation Adversarial Networks (GANs) The overall objective is a sum of agents' private local objective functions. We show that FedGDA-GT converges linearly with a constant stepsize to global $epsilon GDA solution.
  arXiv  Detail & Related papers  (2022-06-02T16:31:16Z)
• Acceleration in Distributed Optimization Under Similarity [72.54787082152278]
  We study distributed (strongly convex) optimization problems over a network of agents, with no centralized nodes. An $varepsilon$-solution is achieved in $tildemathcalrhoObig(sqrtfracbeta/mu (1-)log1/varepsilonbig)$ number of communications steps. This rate matches (up to poly-log factors) for the first time lower complexity communication bounds of distributed gossip-algorithms applied to the class of problems of interest.
  arXiv  Detail & Related papers  (2021-10-24T04:03:00Z)
• DaSGD: Squeezing SGD Parallelization Performance in Distributed Training Using Delayed Averaging [4.652668321425679]
  Minibatch gradient descent (SGD) algorithm requires workers to halt forward/back propagations. DaSGD parallelizes SGD and forward/back propagations to hide 100% of the communication overhead.
  arXiv  Detail & Related papers  (2020-05-31T05:43:50Z)
• Communication-Efficient Distributed Stochastic AUC Maximization with Deep Neural Networks [50.42141893913188]
  We study a distributed variable for large-scale AUC for a neural network as with a deep neural network. Our model requires a much less number of communication rounds and still a number of communication rounds in theory. Our experiments on several datasets show the effectiveness of our theory and also confirm our theory.
  arXiv  Detail & Related papers  (2020-05-05T18:08:23Z)
• Breaking (Global) Barriers in Parallel Stochastic Optimization with Wait-Avoiding Group Averaging [34.55741812648229]
  We present WAGMA-SGD, a wait-avoiding subgroup that reduces global communication via weight exchange. We train ResNet-50 on ImageNet; Transformer for machine translation; and deep reinforcement learning for navigation at scale. Compared with state-of-the-art decentralized SGD variants, WAGMA-SGD significantly improves training throughput.
  arXiv  Detail & Related papers  (2020-04-30T22:11:53Z)
• Detached Error Feedback for Distributed SGD with Random Sparsification [98.98236187442258]
  Communication bottleneck has been a critical problem in large-scale deep learning. We propose a new distributed error feedback (DEF) algorithm, which shows better convergence than error feedback for non-efficient distributed problems. We also propose DEFA to accelerate the generalization of DEF, which shows better bounds than DEF.
  arXiv  Detail & Related papers  (2020-04-11T03:50:59Z)
• Non-asymptotic Convergence of Adam-type Reinforcement Learning Algorithms under Markovian Sampling [56.394284787780364]
  This paper provides the first theoretical convergence analysis for two fundamental RL algorithms of policy gradient (PG) and temporal difference (TD) learning. Under general nonlinear function approximation, PG-AMSGrad with a constant stepsize converges to a neighborhood of a stationary point at the rate of $mathcalO(log T/sqrtT)$. Under linear function approximation, TD-AMSGrad with a constant stepsize converges to a neighborhood of the global optimum at the rate of $mathcalO(log T/sqrtT
  arXiv  Detail & Related papers  (2020-02-15T00:26:49Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.