Update 0130, add CIFAR and kaggle-dog-cat.

CIFAR/config.py

@@ -0,0 +1,4 @@
+lr = 0.1
+resume = False
+start_epoch = 0
+epochs = 60

CIFAR/main.py

@@ -1,30 +1,24 @@
 '''Train CIFAR10 with PyTorch.'''
 from __future__ import print_function
+import os
 
 import torch
 import torch.nn as nn
 import torch.optim as optim
+from torch.autograd import Variable
 import torch.nn.functional as F
-#import torch.backends.cudnn as cudnn
+import torch.backends.cudnn as cudnn
 
 import torchvision
 import torchvision.transforms as transforms
 
-import os
-import argparse
-
+import config
 from models import *
-from utils import progress_bar
-from torch.autograd import Variable
-
-
-parser = argparse.ArgumentParser(description='PyTorch CIFAR10 Training')
-parser.add_argument('--lr', default=0.1, type=float, help='learning rate')
-parser.add_argument('--resume', '-r', action='store_true', help='resume from checkpoint')
-args = parser.parse_args()
+from utils import *
 
 use_cuda = torch.cuda.is_available()
-best_acc = 0  # best test accuracy
+best_acc = 0  # best val accuracy
+best_epoch = 0
 start_epoch = 0  # start from epoch 0 or last checkpoint epoch
 
 # Data
@@ -36,43 +30,43 @@
     transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010)),
 ])
 
-trainset = torchvision.datasets.CIFAR10(root='./data', train=True, download=False, transform=transform)
+trainset = torchvision.datasets.CIFAR10(root='./data', train=True, download=True, transform=transform)
 trainloader = torch.utils.data.DataLoader(trainset, batch_size=128, shuffle=True, num_workers=2)
 
-testset = torchvision.datasets.CIFAR10(root='./data', train=False, download=False, transform=transform)
-testloader = torch.utils.data.DataLoader(testset, batch_size=100, shuffle=False, num_workers=2)
+valset = torchvision.datasets.CIFAR10(root='./data', train=False, download=True, transform=transform)
+valloader = torch.utils.data.DataLoader(valset, batch_size=100, shuffle=False, num_workers=2)
 
 classes = ('plane', 'car', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck')
 
 # Model
-if args.resume:
+if config.resume:
     # Load checkpoint.
     print('==> Resuming from checkpoint..')
     assert os.path.isdir('checkpoint'), 'Error: no checkpoint directory found!'
     checkpoint = torch.load('./checkpoint/ckpt.t7')
     net = checkpoint['net']
     best_acc = checkpoint['acc']
+    best_epoch = checkpoint['epoch']
     start_epoch = checkpoint['epoch']
 else:
     print('==> Building model..')
-    net = VGG('VGG19')
-    # net = ResNet18()
-    # net = GoogLeNet()
-    # net = DenseNet121()
-    # net = ResNeXt29_2x64d()
-    # net = LeNet()
+    #net = VGG('VGG19')
+    #net = ResNet18()
+    #net = GoogLeNet()
+    net = DenseNet121()
+    #net = ResNeXt29_2x64d()
+    #net = LeNet()
 
 if use_cuda:
     net.cuda()
-    #net = torch.nn.DataParallel(net, device_ids=[0,1,2,3])
-    #cudnn.benchmark = True
+    net = torch.nn.DataParallel(net, device_ids=range(torch.cuda.device_count()))
+    cudnn.benchmark = True
 
 criterion = nn.CrossEntropyLoss()
-optimizer = optim.SGD(net.parameters(), lr=args.lr, momentum=0.9, weight_decay=5e-4)
+optimizer = optim.SGD(net.parameters(), lr=config.lr, momentum=0.9, weight_decay=5e-4)
 
 # Training
 def train(epoch):
-
     print('\nEpoch: %d' % epoch)
     net.train()
     train_loss = 0
@@ -96,30 +90,35 @@ def train(epoch):
         progress_bar(batch_idx, len(trainloader), 'Loss: %.3f | Acc: %.3f%% (%d/%d)'
             % (train_loss/(batch_idx+1), 100.*correct/total, correct, total))
 
-def test(epoch):
+    print("Epoch: ", epoch, "Acc: ", 100.*correct/total, correct, total)
+
+def val(epoch):
 
     global best_acc
+    global best_epoch
+
     net.eval()
-    test_loss = 0
+    val_loss = 0
     correct = 0
     total = 0
-    for batch_idx, (inputs, targets) in enumerate(testloader):
+    for batch_idx, (inputs, targets) in enumerate(valloader):
         if use_cuda:
             inputs, targets = inputs.cuda(), targets.cuda()
         inputs, targets = Variable(inputs, volatile=True), Variable(targets)
         outputs = net(inputs)
         loss = criterion(outputs, targets)
 
-        test_loss += loss.data[0]
+        val_loss += loss.data[0]
         _, predicted = torch.max(outputs.data, 1)
         total += targets.size(0)
         correct += predicted.eq(targets.data).cpu().sum()
 
-        progress_bar(batch_idx, len(testloader), 'Loss: %.3f | Acc: %.3f%% (%d/%d)'
-            % (test_loss/(batch_idx+1), 100.*correct/total, correct, total))
+        progress_bar(batch_idx, len(valloader), 'Loss: %.3f | Acc: %.3f%% (%d/%d)'
+            % (val_loss/(batch_idx+1), 100.*correct/total, correct, total))
 
+    print("Epoch: ", epoch, "Acc: ", 100.*correct/total, correct, total)
     # Save checkpoint.
-    acc = 100.*correct/total
+    acc = 100. * correct / total
     if acc > best_acc:
         print('Saving..')
         state = {
@@ -132,8 +131,20 @@ def test(epoch):
             os.mkdir('checkpoint')
         torch.save(state, './checkpoint/ckpt.t7')
         best_acc = acc
+        best_epoch = epoch
+    print("best epoch: ", best_epoch, " acc: ", best_acc)
+
+for epoch in range(start_epoch, start_epoch + config.epochs):
+    if epoch < 20:
+        train(epoch)
+        val(epoch)
+    elif epoch >= 20 and epoch < 40:
+        optimizer = optim.SGD(net.parameters(), lr=config.lr/10.0, momentum=0.9, weight_decay=5e-4)
+        train(epoch)
+        val(epoch)
+    else:
+        optimizer = optim.SGD(net.parameters(), lr=config.lr/100.0, momentum=0.9, weight_decay=5e-4)
+        train(epoch)
+        val(epoch)
 
 
-for epoch in range(start_epoch, start_epoch+200):
-    train(epoch)
-    test(epoch)

CIFAR/models/resnet.py

@@ -115,5 +115,3 @@ def test_resnet():
     net = ResNet50()
     y = net(Variable(torch.randn(1,3,32,32)))
     print(y.size())
-
-# test_resnet()

CIFAR/utils.py

@@ -30,14 +30,6 @@ def get_mean_and_std(dataset):
     std.div_(len(dataset))
     return mean, std
 
-# import torchvision
-# import torchvision.transforms as transforms
-# transform = transforms.Compose([transforms.ToTensor()])
-# dataset = torchvision.datasets.CIFAR10(root='./data', train=True, download=False, transform=transform)
-# mean, std = get_mean_and_std(dataset)
-# print(mean)
-# print(std)
-
 def init_params(net):
     '''Init layer parameters.'''
     for m in net.modules():
@@ -53,20 +45,20 @@ def init_params(net):
             if m.bias:
                 init.constant(m.bias, 0)
 
-
 _, term_width = os.popen('stty size', 'r').read().split()
 term_width = int(term_width)
 
-TOTAL_BAR_LENGTH = 86.
+TOTAL_BAR_LENGTH = 65.
 last_time = time.time()
 begin_time = last_time
 
 def progress_bar(current, total, msg=None):
+
     global last_time, begin_time
     if current == 0:
         begin_time = time.time()  # Reset for new bar.
 
-    cur_len = int(TOTAL_BAR_LENGTH*current/total)
+    cur_len = int(TOTAL_BAR_LENGTH * current / total)
     rest_len = int(TOTAL_BAR_LENGTH - cur_len) - 1
 
     sys.stdout.write(' [')
@@ -83,8 +75,8 @@ def progress_bar(current, total, msg=None):
     tot_time = cur_time - begin_time
 
     L = []
-    L.append('  Step: %s' % format_time(step_time))
-    L.append(' | Tot: %s' % format_time(tot_time))
+    #L.append('  Step: %d / %d' % current+1, total)
+    L.append('Used Time: %s' % format_time(tot_time))
     if msg:
         L.append(' | ' + msg)
 
@@ -94,9 +86,9 @@ def progress_bar(current, total, msg=None):
         sys.stdout.write(' ')
 
     # Go back to the center of the bar.
-    for i in range(term_width-int(TOTAL_BAR_LENGTH/2)):
-        sys.stdout.write('\b')
-    sys.stdout.write(' %d/%d ' % (current+1, total))
+    #for i in range(term_width-int(TOTAL_BAR_LENGTH/2)):
+     #   sys.stdout.write('\b')
+    #sys.stdout.write(' %d/%d ' % (current+1, total))
 
     if current < total-1:
         sys.stdout.write('\r')

dog_and_cat/config.py

@@ -0,0 +1,5 @@
+lr = 0.01
+resume = False
+start_epoch = 0
+epochs = 50
+pretrained = True