tl.add(loss.item())
ta.add(acc)
optimizer.zero_grad()
loss.backward()
optimizer.step()
tl = tl.item()
ta = ta.item()
print('epoch {}, loss={:.4f} acc={:.4f}'
.format(epoch, tl, ta))
model.eval()
vl = Averager()
va = Averager()
for i, batch in enumerate(val_loader, 1):
data, _ = [_.cuda() for _ in batch]
p = args.shot * args.test_way
data_shot, data_query = data[:p], data[p:]
if args.shot == 1:
data_shot = torch.cat((data_shot, flip(data_shot, 3)), dim=0)
proto = model(data_shot)
proto = proto.reshape(shot_num, args.train_way, -1)
proto = torch.transpose(proto, 0, 1)
loss = F.cross_entropy(logits, label)
acc = count_acc(logits, label)
print('epoch {}, train {}/{}, loss={:.4f} acc={:.4f}'.format(
epoch, i, len(train_loader), loss.item(), acc))
tl.add(loss.item())
ta.add(acc)
optimizer.zero_grad() #set all graditnes to zero first
loss.backward() # calc gradients
optimizer.step() # run gradient desc
tl = tl.item()
ta = ta.item()
model.eval() # no grad updates with fixed model params for evaluation
vl = Averager()
va = Averager()
for i, batch in enumerate(val_loader, 1):
data, _ = [_.cuda() for _ in batch]
p = args.shot * args.test_way
data_shot, data_query = data[:p], data[p:]
proto = model(data_shot)
proto = proto.reshape(args.shot, args.train_way, -1)
proto = torch.transpose(proto, 0, 1)
hyperplanes, mu = projection_pro.hyperplanes(
proto, args.test_way, args.shot)