dnddnjs / pytorch-multigpu Goto Github PK
View Code? Open in Web Editor NEWMulti GPU Training Code for Deep Learning with PyTorch
License: MIT License
Multi GPU Training Code for Deep Learning with PyTorch
License: MIT License
hi..^^
I am trying to test your code dist_parallel/train.py
I have 2 computer, and each computer has 1 gpu card.
first computer, i run train.py --gpu_device 0 --rank 0 --batch_size 120
second computer, i run train.py --gpu_device 0 --rank 1 --batch_size 120
But, it is not working... help us..^^
################################################################################
import os
import time
import datetime
import torch
import torch.nn as nn
import torch.optim as optim
from torch.optim import lr_scheduler
import torch.backends.cudnn as cudnn
import torchvision
import torchvision.transforms as transforms
from torchvision.datasets import CIFAR10
from torch.utils.data import DataLoader
import torch.distributed as dist
import torch.multiprocessing as mp
import torch.utils.data.distributed
from model import pyramidnet
import argparse
from tensorboardX import SummaryWriter
parser = argparse.ArgumentParser(description='cifar10 classification models')
parser.add_argument('--lr', default=0.1, help='')
parser.add_argument('--resume', default=None, help='')
parser.add_argument('--batch_size', type=int, default=100, help='')
parser.add_argument('--num_workers', type=int, default=4, help='')
parser.add_argument("--gpu_devices", type=int, nargs='+', default=None, help="")
parser.add_argument('--gpu', default=None, type=int, help='GPU id to use.')
parser.add_argument('--dist-url', default='tcp://192.168.0.179:3456', type=str, help='')
parser.add_argument('--dist-backend', default='nccl', type=str, help='')
parser.add_argument('--rank', default=0, type=int, help='')
parser.add_argument('--world_size', default=1, type=int, help='')
parser.add_argument('--distributed', action='store_true', help='')
args = parser.parse_args()
gpu_devices = ','.join([str(id) for id in args.gpu_devices])
os.environ["CUDA_VISIBLE_DEVICES"] = gpu_devices
def main():
args = parser.parse_args()
ngpus_per_node = torch.cuda.device_count()
args.world_size = ngpus_per_node * args.world_size
mp.spawn(main_worker, nprocs=ngpus_per_node, args=(ngpus_per_node, args))
def main_worker(gpu, ngpus_per_node, args):
args.gpu = gpu
ngpus_per_node = torch.cuda.device_count()
print("Use GPU: {} for training".format(args.gpu))
args.rank = args.rank * ngpus_per_node + gpu
dist.init_process_group(backend=args.dist_backend
,init_method=args.dist_url
,world_size=args.world_size
,rank=args.rank)
print('==> Making model..')
net = pyramidnet()
torch.cuda.set_device(args.gpu)
net.cuda(args.gpu)
args.batch_size = int(args.batch_size / ngpus_per_node)
args.num_workers = int(args.num_workers / ngpus_per_node)
net = torch.nn.parallel.DistributedDataParallel(net, device_ids=[args.gpu])
num_params = sum(p.numel() for p in net.parameters() if p.requires_grad)
print('The number of parameters of model is', num_params)
print('==> Preparing data..')
transforms_train = transforms.Compose([
transforms.RandomCrop(32, padding=4),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010))])
dataset_train = CIFAR10(root='../data',
train=True,
download=True,
transform=transforms_train)
train_sampler = torch.utils.data.distributed.DistributedSampler(dataset_train)
train_loader = DataLoader(dataset_train,
batch_size=args.batch_size,
shuffle=(train_sampler is None),
num_workers=args.num_workers,
sampler=train_sampler)
# there are 10 classes so the dataset name is cifar-10
classes = ('plane', 'car', 'bird', 'cat', 'deer',
'dog', 'frog', 'horse', 'ship', 'truck')
criterion = nn.CrossEntropyLoss()
optimizer = optim.SGD(net.parameters(),
lr=args.lr,
momentum=0.9,
weight_decay=1e-4)
train(net, criterion, optimizer, train_loader, args.gpu)
def train(net, criterion, optimizer, train_loader, device):
net.train()
train_loss = 0
correct = 0
total = 0
epoch_start = time.time()
for batch_idx, (inputs, targets) in enumerate(train_loader):
start = time.time()
inputs = inputs.cuda(device)
targets = targets.cuda(device)
outputs = net(inputs)
loss = criterion(outputs, targets)
optimizer.zero_grad()
loss.backward()
optimizer.step()
train_loss += loss.item()
_, predicted = outputs.max(1)
total += targets.size(0)
correct += predicted.eq(targets).sum().item()
acc = 100 * correct / total
batch_time = time.time() - start
if batch_idx % 20 == 0:
print('Epoch: [{}/{}]| loss: {:.3f} | acc: {:.3f} | batch time: {:.3f}s '.format(
batch_idx, len(train_loader), train_loss/(batch_idx+1), acc, batch_time))
elapse_time = time.time() - epoch_start
elapse_time = datetime.timedelta(seconds=elapse_time)
print("Training time {}".format(elapse_time))
if name=='main':
main()
Hi, I just wonder the difference of train script between "single_gpu" and "data_parallel", since they seem like have the same structure and module, also using the same API.
By the way, would you introduce how to use the distributed one? I am a little bit confuse about how to set the url and how to start using this.
Thx.
Great code! Provides a nice skeleton of how the different multi gpu thing works.
When I use the dist parallel with 8 cards, my ram saturates and my processor maxes out.
What CPU and how much RAM did your setup use?
Everything work with less GPUs, but then I can only use 2 out of the 8.
Hello
How are you?
Thanks for contributing to this project.
I have a question.
I can NOT see loss.mean() in your Data-Parallel implementation.
How should I understand?
Thanks
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.