shuffle=True,
num_workers=2)
augloader1 = torch.utils.data.DataLoader(augset1,
batch_size=256,
shuffle=True,
num_workers=2)
augloader2 = torch.utils.data.DataLoader(augset2,
batch_size=256,
shuffle=True,
num_workers=2)
augloader3 = torch.utils.data.DataLoader(augset3,
batch_size=256,
shuffle=True,
num_workers=2)
net = get_model()
checkPointDir = './model.pth'
checkpoint = torch.load(checkPointDir)
net.load_state_dict(checkpoint)
net = net.cuda()
criterion = nn.CrossEntropyLoss()
optimizer = torch.optim.Adam(net.parameters(), lr=0.001)
print('Start Training')
net.train()
for epoch in range(100):
running_loss = 0.0
for i, data in enumerate(trainloader): #Original training set
# get the inputs; data is a list of [inputs, labels]
# ==========================================================================
# ==========================================================================
from submission import get_model, eval_transform, team_id, team_name, email_address
trainset = CustomDataset(root='./dataset',
split="train",
transform=train_transform)
# trainset = addIndexToTrainData(trainset)
trainloader = torch.utils.data.DataLoader(trainset,
batch_size=256,
shuffle=True,
num_workers=2)
net = get_model().cuda()
net = torch.nn.DataParallel(net)
net = net.cuda()
# trainLabeledImage(net, trainloader)
#
unLabeledSet = CustomDataset(root='./dataset',
split="unlabeled",
transform=train_transform)
xxx = addIndexToTrainData(unLabeledSet)
# unLabeledLoader = torch.utils.data.DataLoader(unLabeledSet, batch_size = 10, shuffle=True, num_workers = 2)
#
# net = get_model()
# checkPointDir = './checkPoint/net_demo.pth'
# # checkpoint = torch.load(args.checkpoint_path)
# checkpoint = torch.load(checkPointDir)
# net.load_state_dict(checkpoint)