import torch
from tqdm import tqdm
from model import VQVAE
args = config.get_args()
transform = config.get_transform()
dataset = datasets.ImageFolder(args.path, transform=transform)
loader = DataLoader(dataset,
batch_size=args.batch,
shuffle=True,
num_workers=0)
model = VQVAE()
model = model.cuda()
criterion = nn.MSELoss()
optimizer = optim.Adam(model.parameters(), lr=args.lr)
from torch.autograd import Variable
for epoch in range(args.epoch):
loader = tqdm(loader)
for i, (img, _) in enumerate(loader):
img = img.cuda()
#generate the attention regions for the images
saliency = utilities.compute_saliency_maps(img, model)
kwargs = {'num_workers': 1, 'pin_memory': True} if args.cuda else {}
train_loader = torch.utils.data.DataLoader(datasets.MNIST(
'../data', train=True, download=True, transform=transforms.ToTensor()),
batch_size=args.batch_size,
shuffle=True,
**kwargs)
test_loader = torch.utils.data.DataLoader(datasets.MNIST(
'../data', train=False, transform=transforms.ToTensor()),
batch_size=args.batch_size,
shuffle=True,
**kwargs)
model = VQVAE(args.input_dim, args.emb_dim, args.emb_num, args.batch_size)
if args.cuda:
model.cuda()
optimizer = optim.Adam(model.parameters(), lr=1e-3)
def train(epoch):
"""run one epoch of model to train with data loader"""
model.train()
train_loss = 0
for batch_idx, (data, _) in enumerate(train_loader):
data = Variable(data).view(-1, 784)
if args.cuda:
data = data.cuda()
# run forward
# compute losses
recon_batch, reconst_loss, embed_loss, commit_loss = model(data)