I’m using Colab Pro, along with Pytorch.
The GPU allocated for me is: Tesla P100-PCIE-16GB
However, CPU RAM filled in the first epoch, and it doesn’t even reach the validation step, my dataset is a set of images of size 128*128. Training set: 122k. valid set: 21k
Here’s my code:
def train(model, train_loader, validation_loader):
optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate)
model = model.to(device)
loss_train_log = []
loss_val_log = []
epoch_val_log = []
print('START TRAINING...')
for epoch in range(1, num_epochs + 1):
# Training
model.train()
epoch_loss = 0
for batch_idx, batch_samples in enumerate(train_loader):
img = batch_samples[0].view(batch_samples[0].size(0),
batch_samples[0].size(1),
batch_samples[0].size(2),
batch_samples[0].size(3)).to(device)
labels = batch_samples[1].to(device)
optimizer.zero_grad()
prd = model(img)
loss = F.mse_loss(prd.view(-1).float(), labels.view(-1).float())
loss.backward()
optimizer.step()
epoch_loss += loss.detach().item() * len(batch_samples[0]) / len(train_loader.sampler)
del loss
del prd
loss_train_log.append(epoch_loss)
del epoch_loss
print('+ TRAINING \tEpoch: {} \tLoss: {:.6f}'.format(epoch, loss_train_log[-1]))
# Validation
if epoch == 1 or epoch % val_interval == 0:
print("Validation")
model.eval()
loss_val = 0
with torch.no_grad():
for data_sample in validation_loader:
img = data_sample[0].to(device)
labels = data_sample[1].to(device)
img = img.view(img.size(0),img.size(1),img.size(2),img.size(3))
prd = model(img)
prd_flat = prd.view(-1)
true = labels.view(-1)
loss_val += F.mse_loss(prd_flat.float(), true.float()).item() * len(data_sample[0]) / len(validation_loader.sampler)
del prd
loss_val_log.append(loss_val)
epoch_val_log.append(epoch)
del loss_val
print('-------------------------------------------------------------')
print('+ VALIDATE \tEpoch: {} \tLoss: {:.5f}'.format(epoch, loss_val_log[-1]))
print('-------------------------------------------------------------')
print('\nFinished TRAINING.')
train_dataset=dataset(X_train,y_train, transform=transform) # dataset is a class I defined.
valid_dataset=dataset(X_valid,y_valid, transform=transform)
train_loader = torch.utils.data.DataLoader(train_dataset, batch_size=batch_size, shuffle=True)
validation_loader = torch.utils.data.DataLoader(valid_dataset, batch_size=batch_size, shuffle=True)
model = Model()
train(model, train_loader, validation_loader)
Could someone help me with this, please?
Here is my dataset class along with my transform:
transform = T.Compose([
T.ToPILImage(),
T.ToTensor(),
T.Normalize((0.5,0.5,0.5), (0.5,0.5,0.5)),
])
class dataset(Dataset):
def __init__(self, data, label, transform=None):
self.data=data
self.label=label
self.transform=transform
def __getitem__(self, index):
if self.transform:
self.data[index]=self.transform(self.data[index])
return self.data[index],self.label[index]
def __len__(self):
return len(self.data)