Simple tutorials on Pytorch DDP training

Overview

pytorch-distributed-training

Distribute Dataparallel (DDP) Training on Pytorch

Features

Good Notes

分享一些网上优质的笔记

TODO

  • 完成DP和DDP源码解读笔记(当前进度50%)
  • 修改代码细节, 复现实验结果

Quick start

想直接运行查看结果的可以执行以下命令, 注意一定要用--ip--port来指定主机的ip地址以及空闲的端口,否则可能无法运行

$ python dataparallel.py --gpu 0,1,2,3
$ CUDA_VISIBLE_DEVICES=0,1,2,3 python -m torch.distributed.launch --nproc_per_node=4 distributed.py
$ CUDA_VISIBLE_DEVICES=0,1,2,3 python distributed_mp.py
$ CUDA_VISIBLE_DEVICES=0,1,2,3 python distributed_apex.py
  • --ip=str, e.g --ip='10.24.82.10' 来指定主进程的ip地址

  • --port=int, e.g --port=23456 来指定启动端口号

  • --batch_size=int, e.g --batch_size=128 设定训练batch_size

  • distributed_gradient_accumulation.py

$ CUDA_VISIBLE_DEVICES=0,1,2,3 python distributed_apex.py
  • --ip=str, e.g --ip='10.24.82.10' 来指定主进程的ip地址
  • --port=int, e.g --port=23456 来指定启动端口号
  • --grad_accu_steps=int, e.g --grad_accu_steps=4' 来指定gradient_step

Comparison

结果不够准确,GPU状态不同结果可能差异较大

默认情况下都使用SyncBatchNorm, 这会导致执行速度变慢一些,因为需要增加进程之间的通讯来计算BatchNorm, 但有利于保证准确率

Concepts

  • apex
  • DP: DataParallel
  • DDP: DistributedDataParallel

Environments

  • 4 × 2080Ti
model dataset training method time(seconds/epoch) Top-1 accuracy
resnet18 cifar100 DP 20s
resnet18 cifar100 DP+apex 18s
resnet18 cifar100 DDP 16s
resnet18 cifar100 DDP+apex 14.5s

Basic Concept

  • group: 表示进程组,默认情况下只有一个进程组。
  • world size: 全局进程个数
    • 比如16张卡单卡单进程: world size = 16
    • 8卡单进程: world size = 1
    • 只有当连接的进程数等于world size, 程序才会执行
  • rank: 进程序号,用于进程间通讯,表示进程优先级,rank=0表示主进程
  • local_rank: 进程内,GPU编号,非显示参数,由torch.distributed.launch内部指定,rank=3, local_rank=0 表示第3个进程的第1GPU

Usage 单机多卡

1. 获取当前进程的index

pytorch可以通过torch.distributed.lauch启动器,在命令行分布式地执行.py文件, 在执行的过程中会将当前进程的index通过参数传递给python

import argparse
parser = argparse.ArgumentParser()
parser.add_argument('--local_rank', default=-1, type=int,
                    help='node rank for distributed training')
args = parser.parse_args()
print(args.local_rank)

2. 定义 main_worker 函数

主要的训练流程都写在main_worker函数中,main_worker需要接受三个参数(最后一个参数optional):

def main_worker(local_rank, nprocs, args):
    training...
  • local_rank: 接受当前进程的rank值,在一机多卡的情况下对应使用的GPU号
  • nprocs: 进程数量
  • args: 自己定义的额外参数

main_worker,相当于你每个进程需要运行的函数(每个进程执行的函数内容是一致的,只不过传入的local_rank不一样)

3. main_worker函数中的整体流程

main_worker函数中完整的训练流程

import torch
import torch.distributed as dist
import torch.backends.cudnn as cudnn
def main_worker(local_rank, nprocs, args):
    args.local_rank = local_rank
    # 分布式初始化,对于每个进程来说,都需要进行初始化
    cudnn.benchmark = True
    dist.init_process_group(backend='nccl', init_method='tcp://ip:port', world_size=nprocs, rank=local_rank)
    # 模型、损失函数、优化器定义
    model = ...
    criterion = ...
    optimizer = ...
    # 设置进程对应使用的GPU
    torch.cuda.set_device(local_rank)
    model.cuda(local_rank)
    # 使用分布式函数定义模型
    model = model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[local_rank])
    
    # 数据集的定义,使用 DistributedSampler
    mini_batch_size = batch_size / nprocs # 手动划分 batch_size to mini-batch_size
    train_dataset = ...
    train_sampler = torch.utils.data.distributed.DistributedSampler(train_dataset)
    trainloader = torch.utils.data.DataLoader(train_dataset, batch_size=mini_batch_size, num_workers=..., pin_memory=..., 
                                              sampler=train_sampler)
    
    test_dataset = ...
    test_sampler = torch.utils.data.distributed.DistributedSampler(test_dataset)
    testloader = torch.utils.data.DataLoader(train_dataset, batch_size=mini_batch_size, num_workers=..., pin_memory=..., 
                                             sampler=test_sampler) 
    
    # 正常的 train 流程
    for epoch in range(300):
       model.train()
       for batch_idx, (images, target) in enumerate(trainloader):
          images = images.cuda(non_blocking=True)
          target = target.cuda(non_blocking=True)
          ...
          pred = model(images)
          loss = loss_function(pred, target)
          ...
          optimizer.zero_grad()
          loss.backward()
          optimizer.step()

4. 定义main函数

import argparse
import torch
parser = argparse.ArgumentParser(description='PyTorch ImageNet Training')
parser.add_argument('--local_rank', default=-1, type=int, help='node rank for distributed training')
parser.add_argument('--batch_size','--batch-size', default=256, type=int)
parser.add_argument('--lr', default=0.1, type=float)

def main_worker(local_rank, nprocs, args):
    ...

def main():
    args = parser.parse_args()
    args.nprocs = torch.cuda.device_count()
    # 执行 main_worker
    main_worker(args.local_rank, args.nprocs, args)

if __name__ == '__main__':
    main()

5. Command Line 启动

$ CUDA_VISIBLE_DEVICES=0,1,2,3 python -m torch.distributed.launch --nproc_per_node=4 distributed.py
  • --ip=str, e.g --ip='10.24.82.10' 来指定主进程的ip地址
  • --port=int, e.g --port=23456 来指定启动端口号

参数说明:

  • --nnodes 表示机器的数量
  • --node_rank 表示当前的机器
  • --nproc_per_node 表示每台机器上的进程数量

参考 distributed.py

6. torch.multiprocessing

使用torch.multiprocessing来解决进程自发控制可能产生问题,这种方式比较稳定,推荐使用

import argparse
import torch
import torch.multiprocessing as mp

parser = argparse.ArgumentParser(description='PyTorch ImageNet Training')
parser.add_argument('--local_rank', default=-1, type=int, help='node rank for distributed training')
parser.add_argument('--batch_size','--batch-size', default=256, type=int)
parser.add_argument('--lr', default=0.1, type=float)

def main_worker(local_rank, nprocs, args):
    ...

def main():
    args = parser.parse_args()
    args.nprocs = torch.cuda.device_count()
    # 将 main_worker 放入 mp.spawn 中
    mp.spawn(main_worker, nprocs=args.nprocs, args=(args.nprocs, args))

if __name__ == '__main__':
    main()

参考 distributed_mp.py 启动方式如下:

$ CUDA_VISIBLE_DEVICES=0,1,2,3 python distributed_mp.py
  • --ip=str, e.g --ip='10.24.82.10' 来指定主进程的ip地址
  • --port=int, e.g --port=23456 来指定启动端口号

Implemented Work

参考的文章如下(如果有文章没有引用,但是内容差不多的,可以提issue给我,我会补上,实在抱歉):

Owner
Ren Tianhe
Ren Tianhe
Efficient Speech Processing Tookit for Automatic Speaker Recognition

Sugar Efficient Speech Processing Tookit for Automatic Speaker Recognition | HuggingFace | What's New EfficientTDNN: Efficient Architecture Search for

WangRui 14 Sep 14, 2022
Only a Matter of Style: Age Transformation Using a Style-Based Regression Model

Only a Matter of Style: Age Transformation Using a Style-Based Regression Model The task of age transformation illustrates the change of an individual

444 Dec 30, 2022
Pairwise learning neural link prediction for ogb link prediction

Pairwise Learning for Neural Link Prediction for OGB (PLNLP-OGB) This repository provides evaluation codes of PLNLP for OGB link property prediction t

Zhitao WANG 31 Oct 10, 2022
Fully Adaptive Bayesian Algorithm for Data Analysis (FABADA) is a new approach of noise reduction methods. In this repository is shown the package developed for this new method based on \citepaper.

Fully Adaptive Bayesian Algorithm for Data Analysis FABADA FABADA is a novel non-parametric noise reduction technique which arise from the point of vi

18 Oct 20, 2022
ChatBot-Pytorch - A GPT-2 ChatBot implemented using Pytorch and Huggingface-transformers

ChatBot-Pytorch A GPT-2 ChatBot implemented using Pytorch and Huggingface-transf

ParZival 42 Dec 09, 2022
Sub-Cluster AdaCos: Learning Representations for Anomalous Sound Detection.

Accompanying code for the paper Sub-Cluster AdaCos: Learning Representations for Anomalous Sound Detection.

Kevin Wilkinghoff 6 Dec 01, 2022
The code for our CVPR paper PISE: Person Image Synthesis and Editing with Decoupled GAN, Project Page, supp.

PISE The code for our CVPR paper PISE: Person Image Synthesis and Editing with Decoupled GAN, Project Page, supp. Requirement conda create -n pise pyt

jinszhang 110 Nov 21, 2022
Official PyTorch repo for JoJoGAN: One Shot Face Stylization

JoJoGAN: One Shot Face Stylization This is the PyTorch implementation of JoJoGAN: One Shot Face Stylization. Abstract: While there have been recent ad

1.3k Dec 29, 2022
An image classification app boilerplate to serve your deep learning models asap!

Image 🖼 Classification App Boilerplate Have you been puzzled by tons of videos, blogs and other resources on the internet and don't know where and ho

Smaranjit Ghose 27 Oct 06, 2022
Point Cloud Registration using Representative Overlapping Points.

Point Cloud Registration using Representative Overlapping Points (ROPNet) Abstract 3D point cloud registration is a fundamental task in robotics and c

ZhuLifa 36 Dec 16, 2022
novel deep learning research works with PaddlePaddle

Research 发布基于飞桨的前沿研究工作,包括CV、NLP、KG、STDM等领域的顶会论文和比赛冠军模型。 目录 计算机视觉(Computer Vision) 自然语言处理(Natrual Language Processing) 知识图谱(Knowledge Graph) 时空数据挖掘(Spa

1.5k Dec 29, 2022
Self-supervised Augmentation Consistency for Adapting Semantic Segmentation (CVPR 2021)

Self-supervised Augmentation Consistency for Adapting Semantic Segmentation This repository contains the official implementation of our paper: Self-su

Visual Inference Lab @TU Darmstadt 132 Dec 21, 2022
[CVPR2021] The source code for our paper 《Removing the Background by Adding the Background: Towards Background Robust Self-supervised Video Representation Learning》.

TBE The source code for our paper "Removing the Background by Adding the Background: Towards Background Robust Self-supervised Video Representation Le

Jinpeng Wang 150 Dec 28, 2022
Does Pretraining for Summarization Reuqire Knowledge Transfer?

Pretraining summarization models using a corpus of nonsense

Approximately Correct Machine Intelligence (ACMI) Lab 12 Dec 19, 2022
MGFN: Multi-Graph Fusion Networks for Urban Region Embedding was accepted by IJCAI-2022.

Multi-Graph Fusion Networks for Urban Region Embedding (IJCAI-22) This is the implementation of Multi-Graph Fusion Networks for Urban Region Embedding

202 Nov 18, 2022
Implementation of a protein autoregressive language model, but with autoregressive infilling objective (editing subsequences capability)

Protein GLM (wip) Implementation of a protein autoregressive language model, but with autoregressive infilling objective (editing subsequences capabil

Phil Wang 17 May 06, 2022
Bayesian Generative Adversarial Networks in Tensorflow

Bayesian Generative Adversarial Networks in Tensorflow This repository contains the Tensorflow implementation of the Bayesian GAN by Yunus Saatchi and

Andrew Gordon Wilson 1k Nov 29, 2022
Dilated Convolution with Learnable Spacings PyTorch

Dilated-Convolution-with-Learnable-Spacings-PyTorch Ismail Khalfaoui Hassani Dilated Convolution with Learnable Spacings (abbreviated to DCLS) is a no

15 Dec 09, 2022
Code release for NeX: Real-time View Synthesis with Neural Basis Expansion

NeX: Real-time View Synthesis with Neural Basis Expansion Project Page | Video | Paper | COLAB | Shiny Dataset We present NeX, a new approach to novel

538 Jan 09, 2023
CVPR 2022 "Online Convolutional Re-parameterization"

OREPA: Online Convolutional Re-parameterization This repo is the PyTorch implementation of our paper to appear in CVPR2022 on "Online Convolutional Re

Mu Hu 121 Dec 21, 2022