'''
Training script for Fashion-MNIST
'''
from __future__ import print_function
import argparse
import os
import shutil
import time
import random
import torch
import torch.nn as nn
import torch.nn.parallel
import torch.backends.cudnn as cudnn
import torch.optim as optim
import torch.utils.data as data
import torchvision.datasets as datasets
import models.fashion as models
import transforms
import numpy as np
from utils import Bar, Logger, AverageMeter, accuracy, mkdir_p, savefig
model_names = sorted(name for name in models.__dict__
if name.islower() and not name.startswith("__")
and callable(models.__dict__[name]))
parser = argparse.ArgumentParser(description='PyTorch Fashion-MNIST Training')
# Datasets
parser.add_argument('-d', '--dataset', default='fashionmnist', type=str)
parser.add_argument('-j', '--workers', default=4, type=int, metavar='N',
help='number of data loading workers (default: 4)')
# Optimization options
parser.add_argument('--epochs', default=300, type=int, metavar='N',
help='number of total epochs to run')
parser.add_argument('--start-epoch', default=0, type=int, metavar='N',
help='manual epoch number (useful on restarts)')
parser.add_argument('--train-batch', default=128, type=int, metavar='N',
help='train batchsize')
parser.add_argument('--test-batch', default=100, type=int, metavar='N',
help='test batchsize')
parser.add_argument('--lr', '--learning-rate', default=0.1, type=float,
metavar='LR', help='initial learning rate')
parser.add_argument('--drop', '--dropout', default=0, type=float,
metavar='Dropout', help='Dropout ratio')
parser.add_argument('--schedule', type=int, nargs='+', default=[150, 225],
help='Decrease learning rate at these epochs.')
parser.add_argument('--gamma', type=float, default=0.1, help='LR is multiplied by gamma on schedule.')
parser.add_argument('--momentum', default=0.9, type=float, metavar='M',
help='momentum')
parser.add_argument('--weight-decay', '--wd', default=5e-4, type=float,
metavar='W', help='weight decay (default: 1e-4)')
# Checkpoints
parser.add_argument('-c', '--checkpoint', default='checkpoint', type=str, metavar='PATH',
help='path to save checkpoint (default: checkpoint)')
parser.add_argument('--resume', default='', type=str, metavar='PATH',
help='path to latest checkpoint (default: none)')
# Architecture
parser.add_argument('--arch', '-a', metavar='ARCH', default='resnet',
choices=model_names,
help='model architecture: ' +
' | '.join(model_names) +
' (default: resnet20)')
parser.add_argument('--depth', type=int, default=20, help='Model depth.')
parser.add_argument('--widen-factor', type=int, default=10, help='Widen factor. 10')
parser.add_argument('--growthRate', type=int, default=12, help='Growth rate for DenseNet.')
parser.add_argument('--compressionRate', type=int, default=2, help='Compression Rate (theta) for DenseNet.')
# Miscs
parser.add_argument('--manualSeed', type=int, help='manual seed')
parser.add_argument('-e', '--evaluate', dest='evaluate', action='store_true',
help='evaluate model on validation set')
# Random Erasing
parser.add_argument('--p', default=0, type=float, help='Random Erasing probability')
parser.add_argument('--sh', default=0.4, type=float, help='max erasing area')
parser.add_argument('--r1', default=0.3, type=float, help='aspect of erasing area')
args = parser.parse_args()
state = {k: v for k, v in args._get_kwargs()}
# Validate dataset
assert args.dataset == 'fashionmnist'
# Use CUDA
use_cuda = torch.cuda.is_available()
# Random seed
if args.manualSeed is None:
args.manualSeed = random.randint(1, 10000)
random.seed(args.manualSeed)
torch.manual_seed(args.manualSeed)
if use_cuda:
torch.cuda.manual_seed_all(args.manualSeed)
best_acc = 0 # best test accuracy
def main():
global best_acc
start_epoch = args.start_epoch # start from epoch 0 or last checkpoint epoch
if not os.path.isdir(args.checkpoint):
mkdir_p(args.checkpoint)
# Data
print('==> Preparing dataset %s' % args.dataset)
transform_train = transforms.Compose([
transforms.RandomCrop(28, padding=4),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
transforms.Normalize((0.1307,), (0.3081,)),
transforms.RandomErasing(probability = args.p, sh = args.sh, r1 = args.r1, mean = [0.4914]),
])
transform_test = transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307,), (0.3081,)),
])
if args.dataset == 'fashionmnist':
dataloader = datasets.FashionMNIST
num_classes = 10
trainset = dataloader(root='./data', train=True, download=True, transform=transform_train)
trainloader = data.DataLoader(trainset, batch_size=args.train_batch, shuffle=True, num_workers=args.workers)
testset = dataloader(root='./data', train=False, download=False, transform=transform_test)
testloader = data.DataLoader(testset, batch_size=args.test_batch, shuffle=False, num_workers=args.workers)
# Model
print("==> creating model '{}'".format(args.arch))
if args.arch.startswith('wrn'):
model = models.__dict__[args.arch](
num_classes=num_classes,
depth=args.depth,
widen_factor=args.widen_factor,
dropRate=args.drop,
)
elif args.arch.endswith('resnet'):
model = models.__dict__[args.arch](
num_classes=num_classes,
depth=args.depth,
)
model = torch.nn.DataParallel(model).cuda()
cudnn.benchmark = True
print(' Total params: %.2fM' % (sum(p.numel() for p in model.parameters())/1000000.0))
criterion = nn.CrossEntropyLoss()
optimizer = optim.SGD(model.parameters(), lr=args.lr, momentum=args.momentum, weight_decay=args.weight_decay)
# Resume
title = 'fashionmnist-' + args.arch
if args.resume:
# Load checkpoint.
print('==> Resuming from checkpoint..')
assert os.path.isfile(args.resume), 'Error: no checkpoint directory found!'
args.checkpoint = os.path.dirname(args.resume)
checkpoint = torch.load(args.resume)
best_acc = checkpoint['best_acc']
start_epoch = checkpoint['epoch']
model.load_state_dict(checkpoint['state_dict'])
optimizer.load_state_dict(checkpoint['optimizer'])
logger = Logger(os.path.join(args.checkpoint, 'log.txt'), title=title, resume=True)
else:
logger = Logger(os.path.join(args.checkpoint, 'log.txt'), title=title)
logger.set_names(['Learning Rate', 'Train Loss', 'Valid Loss', 'Train Acc.', 'Valid Acc.'])
if args.evaluate:
print('\nEvaluation only')
test_loss, test_acc = test(testloader, model, criterion, start_epoch, use_cuda)
print(' Test Loss: %.8f, Test Acc: %.2f' % (test_loss, test_acc))
return
# Train and val
for epoch in range(start_epoch, args.epochs):
adjust_learning_rate(optimizer, epoch)
print('\nEpoch: [%d | %d] LR: %f' % (epoch + 1, args.epochs, state['lr']))
train_loss, train_acc = train(trainloader, model, criterion, optimizer, epoch, use_cuda)
test_loss, test_acc = test(testloader, model, criterion, epoch, use_cuda)
# append logger file
logger.append([state['lr'], train_loss, test_loss, train_acc, test_acc])
# save model
is_best = test_acc > best_acc
best_acc = max(test_acc, best_acc)
save_checkpoint({
'epoch': epoch + 1,
'state_dict': model.state_dict(),
'acc': test_acc,
'best_acc': best_acc,
'optimizer' : optimizer.state_dict(),
}, is_best, checkpoint=args.checkpoint)
logger.close()
logger.plot()
savefig(os.path.join(args.checkpoint, 'log.eps'))
print('Best acc:')
print(best_acc)
def train(trainloader, model, criterion, optimizer, epoch, use_cuda):
# switch to train mode
model.train()
batch_time = AverageMeter()
data_time = AverageMeter()
losses = AverageMeter()
top1 = AverageMeter()
top5 = AverageMeter()
end = time.time()
bar = Bar('Processing', max=len(trainloader))
for batch_idx, (inputs, targets) in enumerate(trainloader):
# measure data loading time
data_time.update(time.time() - end)
if use_cuda:
inputs, targets = inputs.cuda(), targets.cuda(async=True)
inputs, targets = torch.autograd.Variable(inputs), torch.autograd.Variable(targets)
# compute output
outputs = model(inputs)
loss = criterion(outputs, targets)
# measure accuracy and record loss
prec1, prec5 = accuracy(outputs.data, targets.data, topk=(1, 5))
losses.update(loss.item(), inputs.size(0))
top1.update(prec1.item(), inputs.size(0))
top5.update(prec5.item(), inputs.size(0))
# compute gradient and do SGD step
optimizer.zero_grad()
loss.backward()
optimizer.step()
# measure elapsed time
batch_time.update(time.time() - end)
end = time.time()
# plot progress
bar.suffix = '({batch}/{size}) Data: {data:.3f}s | Batch: {bt:.3f}s | Total: {total:} | ETA: {eta:} | Loss: {loss:.4f} | top1: {top1: .4f} | top5: {top5: .4f}'.format(
batch=batch_idx + 1,
size=len(trainloader),
data=data_time.avg,
bt=batch_time.avg,
total=bar.elapsed_td,
eta=bar.eta_td,
loss=losses.avg,
top1=top1.avg,
top5=top5.avg,
)
bar.next()
bar.finish()
return (losses.avg, top1.avg)
def test(testloader, model, criterion, epoch, use_cuda):
global best_acc
batch_time = AverageMeter()
data_time = AverageMeter()
losses = AverageMeter()
top1 = AverageMeter()
top5 = AverageMeter()
# switch to evaluate mode
model.eval()
end = time.time()
bar = Bar('Processing', max=len(testloader))
for batch_idx, (inputs, targets) in enumerate(testloader):
# measure data loading time
data_time.update(time.time() - end)
if use_cuda:
inputs, targets = inputs.cuda(), targets.cuda()
inputs, targets = torch.autograd.Variable(inputs, volatile=True), torch.autograd.Variable(targets)
# compute output
outputs = model(inputs)
loss = criterion(outputs, targets)
# measure accuracy and record loss
prec1, prec5 = accuracy(outputs.data, targets.data, topk=(1, 5))
losses.update(loss.item(), inputs.size(0))
top1.update(prec1.item(), inputs.size(0))
top5.update(prec5.item(), inputs.size(0))
# measure elapsed time
batch_time.update(time.time() - end)
end = time.time()
# plot progress
bar.suffix = '({batch}/{size}) Data: {data:.3f}s | Batch: {bt:.3f}s | Total: {total:} | ETA: {eta:} | Loss: {loss:.4f} | top1: {top1: .4f} | top5: {top5: .4f}'.format(
batch=batch_idx + 1,
size=len(testloader),
data=data_time.avg,
bt=batch_time.avg,
total=bar.elapsed_td,
eta=bar.eta_td,
loss=losses.avg,
top1=top1.avg,
top5=top5.avg,
)
bar.next()
bar.finish()
return (losses.avg, top1.avg)
def save_checkpoint(state, is_best, checkpoint='checkpoint', filename='checkpoint.pth.tar'):
filepath = os.path.join(checkpoint, filename)
torch.save(state, filepath)
if is_best:
shutil.copyfile(filepath, os.path.join(checkpoint, 'model_best.pth.tar'))
def adjust_learning_rate(optimizer, epoch):
global state
if epoch in args.schedule:
state['lr'] *= args.gamma
for param_group in optimizer.param_groups:
param_group['lr'] = state['lr']
if __name__ == '__main__':
main()
