paper-with-me

Papers

ZeRO++: Extremely Efficient Collective Communication for Giant Model Training

2023-06-16 · Guanhua Wang, Heyang Qin, Sam Ade Jacobs, Connor Holmes, Samyam Rajbhandari, Olatunji Ruwase, Feng Yan, Lei Yang, Yuxiong He

Zero Redundancy Optimizer (ZeRO) has been used to train a wide range of large language models on massive GPUs clusters due to its ease of use, efficiency, and good scalability. However, when training on low-bandwidth clusters, or at scale which forces batch size per GPU to be small, ZeRO's effective throughput is limited because of high communication volume from gathering weights in forward pass, backward pass, and averaging gradients. This paper introduces three communication volume reduction techniques, which we collectively refer to as ZeRO++, targeting each of the communication collectives in ZeRO. First is block-quantization based all-gather. Second is data remapping that trades-off communication for more memory. Third is a novel all-to-all based quantized gradient averaging paradigm as replacement of reduce-scatter collective, which preserves accuracy despite communicating low precision data. Collectively, ZeRO++ reduces communication volume of ZeRO by 4x, enabling up to 2.16x better throughput at 384 GPU scale.

📄 PDF Abstract BibTeX arXiv:2306.10209

Code (1)

microsoft/DeepSpeed 공식 구현 pytorch

Tasks

GPUQuantization

Methods 이 논문이 사용한 방법론

ZeRO Zero Redundancy Optimizer (ZeRO) is a sharded data parallel method for distributed training. ZeRODP removes the memory state redundancies across data-parallel processes by…

Similar Papers 제목 키워드 기반

Enhancing Stability for Large Language Models Training in Constrained Bandwidth Networks

2024-06-28 · Yun Dai, Tejas Dharamsi, Byron Hsu, Tao Song 외

Training extremely large language models (LLMs) with billions of parameters is a computationally intensive task that pushes the limits of current data parallel training systems. While techniques like ZeRO++ have enabled …

Occluded Oculus: Operationalizing Stylistic Obscurement

2026-07-27 · Robert Dilworth arxiv

What did it take for Hermes, the devout messenger of the Olympian gods, to slay Argus Panoptes, the multi-eyed giant of Greek myth? As the perfect guardian, Panoptes' legion of ever-watchful eyes proved difficult -- but …

GIANT: Globally Improved Approximate Newton Method for Distributed Optimization

2017-09-11 · NeurIPS 2018 12 · Shusen Wang, Farbod Roosta-Khorasani, Peng Xu, Michael W. Mahoney

For distributed computing environment, we consider the empirical risk minimization problem and propose a distributed and communication-efficient Newton-type optimization method. At every iteration, each worker locally fi…

Distributed ComputingDistributed Optimization

S2 Reducer: High-Performance Sparse Communication to Accelerate Distributed Deep Learning

2021-10-05 · Keshi Ge, Yongquan Fu, Zhiquan Lai, Xiaoge Deng 외

Distributed stochastic gradient descent (SGD) approach has been widely used in large-scale deep learning, and the gradient collective method is vital to ensure the training scalability of the distributed deep learning sy…

Deep LearningVocal Bursts Intensity Prediction

PacTrain: Pruning and Adaptive Sparse Gradient Compression for Efficient Collective Communication in Distributed Deep Learning

2025-05-24 · Yisu Wang, Ruilong Wu, Xinjiao Li, Dirk Kutscher

Large-scale deep neural networks (DNN) exhibit excellent performance for various tasks. As DNNs and datasets grow, distributed training becomes extremely time-consuming and demands larger clusters. A main bottleneck is t…