paper-with-me

Papers

DeAR: Accelerating Distributed Deep Learning with Fine-Grained All-Reduce Pipelining

2023-02-24 · Lin Zhang, Shaohuai Shi, Xiaowen Chu, Wei Wang, Bo Li, Chengjian Liu

Communication scheduling has been shown to be effective in accelerating distributed training, which enables all-reduce communications to be overlapped with backpropagation computations. This has been commonly adopted in popular distributed deep learning frameworks. However, there exist two fundamental problems: (1) excessive startup latency proportional to the number of workers for each all-reduce operation; (2) it only achieves sub-optimal training performance due to the dependency and synchronization requirement of the feed-forward computation in the next iteration. We propose a novel scheduling algorithm, DeAR, that decouples the all-reduce primitive into two continuous operations, which overlaps with both backpropagation and feed-forward computations without extra communications. We further design a practical tensor fusion algorithm to improve the training performance. Experimental results with five popular models show that DeAR achieves up to 83% and 15% training speedup over the state-of-the-art solutions on a 64-GPU cluster with 10Gb/s Ethernet and 100Gb/s InfiniBand interconnects, respectively.

📄 PDF Abstract BibTeX arXiv:2302.12445

Code (1)

lzhangbv/dear_pytorch 공식 구현 pytorch

Tasks

AllGPUScheduling

Similar Papers 제목 키워드 기반

DeAR: Dual-Stage Document Reranking with Reasoning Agents via LLM Distillation

2025-08-23 · Abdelrahman Abdallah, Jamshid Mozafari, Bhawna Piryani, Adam Jatowt arxiv

Large Language Models (LLMs) have transformed listwise document reranking by enabling global reasoning over candidate sets, yet single models often struggle to balance fine-grained relevance scoring with holistic cross-d…

Natural Questions

DeAR: Fine-Grained VLM Adaptation by Decomposing Attention Head Roles

2026-03-01 · Yiming Ma, Hongkun Yang, Lionel Z. Wang, Bin Chen 외 arxiv

Prompt learning is a dominant paradigm for adapting pre-trained Vision-Language Models (VLMs) to downstream tasks. However, existing methods often rely on a simplistic, layer-centric view, assuming shallow layers capture…

Zero-shot Generalization

BigCodeArena: Unveiling More Reliable Human Preferences in Code Generation via Execution

2025-10-09 · Terry Yue Zhuo, Xiaolong Jin, Hange Liu, Juyong Jiang 외 arxiv

Crowdsourced model evaluation platforms, such as Chatbot Arena, enable real-time evaluation from human perspectives to assess the quality of model responses. In the coding domain, manually examining the quality of LLM-ge…

Code Generation

NAVIG: Natural Language-guided Analysis with Vision Language Models for Image Geo-localization

2025-02-20 · Zheyuan Zhang, Runze Li, Tasnim Kabir, Jordan Boyd-Graber

Image geo-localization is the task of predicting the specific location of an image and requires complex reasoning across visual, geographical, and cultural contexts. While prior Vision Language Models (VLMs) have the bes…

geo-localization

DearFSAC: An Approach to Optimizing Unreliable Federated Learning via Deep Reinforcement Learning

2022-01-30 · Chenghao Huang, Weilong Chen, Yuxi Chen, Shunji Yang 외

In federated learning (FL), model aggregation has been widely adopted for data privacy. In recent years, assigning different weights to local models has been used to alleviate the FL performance degradation caused by dif…

Deep Reinforcement LearningFederated Learningreinforcement-learningReinforcement Learning (RL)