A Quadratic Synchronization Rule for Distributed Deep Learning

• URL: http://arxiv.org/abs/2310.14423v2
• Date: Fri, 12 Apr 2024 13:59:01 GMT
• Title: A Quadratic Synchronization Rule for Distributed Deep Learning
• Authors: Xinran Gu, Kaifeng Lyu, Sanjeev Arora, Jingzhao Zhang, Longbo Huang,
• Abstract summary: This work proposes a theory-grounded method for determining $H$, named the Quadratic Synchronization Rule (QSR) Experiments on ResNet and ViT show that local gradient methods with QSR consistently improve the test accuracy over other synchronization strategies.
• Score: 66.68264684667562
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: In distributed deep learning with data parallelism, synchronizing gradients at each training step can cause a huge communication overhead, especially when many nodes work together to train large models. Local gradient methods, such as Local SGD, address this issue by allowing workers to compute locally for $H$ steps without synchronizing with others, hence reducing communication frequency. While $H$ has been viewed as a hyperparameter to trade optimization efficiency for communication cost, recent research indicates that setting a proper $H$ value can lead to generalization improvement. Yet, selecting a proper $H$ is elusive. This work proposes a theory-grounded method for determining $H$, named the Quadratic Synchronization Rule (QSR), which recommends dynamically setting $H$ in proportion to $\frac{1}{\eta^2}$ as the learning rate $\eta$ decays over time. Extensive ImageNet experiments on ResNet and ViT show that local gradient methods with QSR consistently improve the test accuracy over other synchronization strategies. Compared with the standard data parallel training, QSR enables Local AdamW on ViT-B to cut the training time on 16 or 64 GPUs down from 26.7 to 20.2 hours or from 8.6 to 5.5 hours and, at the same time, achieves $1.16\%$ or $0.84\%$ higher top-1 validation accuracy.

Related papers

• ACCO: Accumulate while you Communicate, Hiding Communications in Distributed LLM Training [16.560270624096706]
  We propose a memory-efficient optimization algorithm tailored for distributed training of Large Language Models. Our method relies on a novel technique to mitigate the one-step delay inherent in parallel execution of gradient computations and communications.
  arXiv  Detail & Related papers  (2024-06-03T08:23:45Z)
• Transfer Q Star: Principled Decoding for LLM Alignment [105.89114186982972]
  Transfer $Q*$ estimates the optimal value function for a target reward $r$ through a baseline model. Our approach significantly reduces the sub-optimality gap observed in prior SoTA methods.
  arXiv  Detail & Related papers  (2024-05-30T21:36:12Z)
• Communication-Efficient Adam-Type Algorithms for Distributed Data Mining [93.50424502011626]
  We propose a class of novel distributed Adam-type algorithms (emphi.e., SketchedAMSGrad) utilizing sketching. Our new algorithm achieves a fast convergence rate of $O(frac1sqrtnT + frac1(k/d)2 T)$ with the communication cost of $O(k log(d))$ at each iteration.
  arXiv  Detail & Related papers  (2022-10-14T01:42:05Z)
• DADAO: Decoupled Accelerated Decentralized Asynchronous Optimization [0.0]
  DADAO is the first decentralized, accelerated, asynchronous, primal, first-order algorithm to minimize a sum of $L$-smooth and $mu$-strongly convex functions distributed over a given network of size $n$. We show that our algorithm requires $mathcalO(nsqrtchisqrtfracLmulog(frac1epsilon)$ local and only $mathcalO(nsqrtchisqrtfracLmulog(
  arXiv  Detail & Related papers  (2022-07-26T08:47:54Z)
• Sharper Convergence Guarantees for Asynchronous SGD for Distributed and Federated Learning [77.22019100456595]
  We show a training algorithm for distributed computation workers with varying communication frequency. In this work, we obtain a tighter convergence rate of $mathcalO!!!(sigma2-2_avg!! . We also show that the heterogeneity term in rate is affected by the average delay within each worker.
  arXiv  Detail & Related papers  (2022-06-16T17:10:57Z)
• Asynchronous Stochastic Optimization Robust to Arbitrary Delays [54.61797739710608]
  We consider optimization with delayed gradients where, at each time stept$, the algorithm makes an update using a stale computation - d_t$ for arbitrary delay $d_t gradient. Our experiments demonstrate the efficacy and robustness of our algorithm in cases where the delay distribution is skewed or heavy-tailed.
  arXiv  Detail & Related papers  (2021-06-22T15:50:45Z)
• Communication-efficient SGD: From Local SGD to One-Shot Averaging [16.00658606157781]
  We consider speeding up gradient descent (SGD) by parallelizing it across multiple workers. We suggest a Local SGD scheme that communicates less overall by communicating less frequently as the number of iterations grows.
  arXiv  Detail & Related papers  (2021-06-09T01:10:34Z)
• Variance Reduced EXTRA and DIGing and Their Optimal Acceleration for Strongly Convex Decentralized Optimization [69.49313819343992]
  We extend the widely used EXTRA and DIGing methods with variance reduction (VR), and propose two methods: VR-EXTRA and VR-DIGing. The proposed VR-EXTRA requires the time of $O(kappa_s+n)logfrac1epsilon)$ gradient evaluations and $O(kappa_b+kappa_c)logfrac1epsilon)$ communication rounds. The proposed VR-DIGing has a little higher communication cost of $O(kappa_b+kappa
  arXiv  Detail & Related papers  (2020-09-09T15:48:44Z)

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.