-
Notifications
You must be signed in to change notification settings - Fork 24
Open
Description
def train_one_epoch(args, net, train_loader, opt, mode, ep):
if mode == 'train':
net.train()
aff_net = Affinity().cuda()
aff_net.train()
elif mode == 'val':
net.eval()
aff_net = Affinity().cuda()
aff_net.eval()
num_examples, total_loss, loss_items, trk_met, seg_met, flow_met = epoch(args, net, train_loader, opt=opt,
mode=mode, ep_num=ep)
total_loss = total_loss * 1.0 / num_examples
When I run the code for VoD dataset, It show me that bug as follows:
“ total_loss = total_loss * 1.0 / num_examples
ZeroDivisionError: float division by zero”
Does it any configuration that I mistake?
Metadata
Metadata
Assignees
Labels
No labels